diff --git a/Documentation/devicetree/bindings/arm/sti.yaml b/Documentation/devicetree/bindings/arm/sti.yaml
index b1f28d1..a41cd87 100644
--- a/Documentation/devicetree/bindings/arm/sti.yaml
+++ b/Documentation/devicetree/bindings/arm/sti.yaml
@@ -7,7 +7,7 @@
 title: ST STi Platforms Device Tree Bindings
 
 maintainers:
-  - Patrice Chotard <patrice.chotard@st.com>
+  - Patrice Chotard <patrice.chotard@foss.st.com>
 
 properties:
   $nodename:
diff --git a/Documentation/devicetree/bindings/arm/stm32/st,mlahb.yaml b/Documentation/devicetree/bindings/arm/stm32/st,mlahb.yaml
index 8e711bd..ecb28e9 100644
--- a/Documentation/devicetree/bindings/arm/stm32/st,mlahb.yaml
+++ b/Documentation/devicetree/bindings/arm/stm32/st,mlahb.yaml
@@ -7,8 +7,8 @@
 title: STMicroelectronics STM32 ML-AHB interconnect bindings
 
 maintainers:
-  - Fabien Dessenne <fabien.dessenne@st.com>
-  - Arnaud Pouliquen <arnaud.pouliquen@st.com>
+  - Fabien Dessenne <fabien.dessenne@foss.st.com>
+  - Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
 
 description: |
   These bindings describe the STM32 SoCs ML-AHB interconnect bus which connects
diff --git a/Documentation/devicetree/bindings/arm/stm32/st,stm32-syscon.yaml b/Documentation/devicetree/bindings/arm/stm32/st,stm32-syscon.yaml
index 149afb5..6f846d6 100644
--- a/Documentation/devicetree/bindings/arm/stm32/st,stm32-syscon.yaml
+++ b/Documentation/devicetree/bindings/arm/stm32/st,stm32-syscon.yaml
@@ -7,8 +7,8 @@
 title: STMicroelectronics STM32 Platforms System Controller bindings
 
 maintainers:
-  - Alexandre Torgue <alexandre.torgue@st.com>
-  - Christophe Roullier <christophe.roullier@st.com>
+  - Alexandre Torgue <alexandre.torgue@foss.st.com>
+  - Christophe Roullier <christophe.roullier@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/arm/stm32/stm32.yaml b/Documentation/devicetree/bindings/arm/stm32/stm32.yaml
index 9ac7da0..bcaf7be 100644
--- a/Documentation/devicetree/bindings/arm/stm32/stm32.yaml
+++ b/Documentation/devicetree/bindings/arm/stm32/stm32.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 Platforms Device Tree Bindings
 
 maintainers:
-  - Alexandre Torgue <alexandre.torgue@st.com>
+  - Alexandre Torgue <alexandre.torgue@foss.st.com>
 
 properties:
   $nodename:
diff --git a/Documentation/devicetree/bindings/clock/ingenic,cgu.yaml b/Documentation/devicetree/bindings/clock/ingenic,cgu.yaml
index 6e80dbc..aa1df03 100644
--- a/Documentation/devicetree/bindings/clock/ingenic,cgu.yaml
+++ b/Documentation/devicetree/bindings/clock/ingenic,cgu.yaml
@@ -104,7 +104,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4770-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4770-cgu.h>
     cgu: clock-controller@10000000 {
       compatible = "ingenic,jz4770-cgu", "simple-mfd";
       reg = <0x10000000 0x100>;
diff --git a/Documentation/devicetree/bindings/clock/sifive/fu740-prci.yaml b/Documentation/devicetree/bindings/clock/sifive/fu740-prci.yaml
index e17143c..252085a 100644
--- a/Documentation/devicetree/bindings/clock/sifive/fu740-prci.yaml
+++ b/Documentation/devicetree/bindings/clock/sifive/fu740-prci.yaml
@@ -42,6 +42,9 @@
   "#clock-cells":
     const: 1
 
+  "#reset-cells":
+    const: 1
+
 required:
   - compatible
   - reg
@@ -57,4 +60,5 @@
       reg = <0x10000000 0x1000>;
       clocks = <&hfclk>, <&rtcclk>;
       #clock-cells = <1>;
+      #reset-cells = <1>;
     };
diff --git a/Documentation/devicetree/bindings/clock/silabs,si5351.txt b/Documentation/devicetree/bindings/clock/silabs,si5351.txt
index 8fe6f80..bfda6af 100644
--- a/Documentation/devicetree/bindings/clock/silabs,si5351.txt
+++ b/Documentation/devicetree/bindings/clock/silabs,si5351.txt
@@ -2,7 +2,7 @@
 
 Reference
 [1] Si5351A/B/C Data Sheet
-    https://www.silabs.com/Support%20Documents/TechnicalDocs/Si5351.pdf
+    https://www.skyworksinc.com/-/media/Skyworks/SL/documents/public/data-sheets/Si5351-B.pdf
 
 The Si5351a/b/c are programmable i2c clock generators with up to 8 output
 clocks. Si5351a also has a reduced pin-count package (MSOP10) where only
diff --git a/Documentation/devicetree/bindings/clock/socionext,uniphier-clock.yaml b/Documentation/devicetree/bindings/clock/socionext,uniphier-clock.yaml
index c3930ed..9a0cc73 100644
--- a/Documentation/devicetree/bindings/clock/socionext,uniphier-clock.yaml
+++ b/Documentation/devicetree/bindings/clock/socionext,uniphier-clock.yaml
@@ -23,6 +23,7 @@
           - socionext,uniphier-ld11-clock
           - socionext,uniphier-ld20-clock
           - socionext,uniphier-pxs3-clock
+          - socionext,uniphier-nx1-clock
       - description: Media I/O (MIO) clock, SD clock
         enum:
           - socionext,uniphier-ld4-mio-clock
@@ -33,6 +34,7 @@
           - socionext,uniphier-ld11-mio-clock
           - socionext,uniphier-ld20-sd-clock
           - socionext,uniphier-pxs3-sd-clock
+          - socionext,uniphier-nx1-sd-clock
       - description: Peripheral clock
         enum:
           - socionext,uniphier-ld4-peri-clock
@@ -43,6 +45,10 @@
           - socionext,uniphier-ld11-peri-clock
           - socionext,uniphier-ld20-peri-clock
           - socionext,uniphier-pxs3-peri-clock
+          - socionext,uniphier-nx1-peri-clock
+      - description: SoC-glue clock
+        enum:
+          - socionext,uniphier-pro4-sg-clock
 
   "#clock-cells":
     const: 1
diff --git a/Documentation/devicetree/bindings/clock/st,stm32mp1-rcc.yaml b/Documentation/devicetree/bindings/clock/st,stm32mp1-rcc.yaml
index 8b1ecb2..a0ae486 100644
--- a/Documentation/devicetree/bindings/clock/st,stm32mp1-rcc.yaml
+++ b/Documentation/devicetree/bindings/clock/st,stm32mp1-rcc.yaml
@@ -7,7 +7,7 @@
 title: Reset Clock Controller Binding
 
 maintainers:
-  - Gabriel Fernandez <gabriel.fernandez@st.com>
+  - Gabriel Fernandez <gabriel.fernandez@foss.st.com>
 
 description: |
   The RCC IP is both a reset and a clock controller.
diff --git a/Documentation/devicetree/bindings/crypto/st,stm32-crc.yaml b/Documentation/devicetree/bindings/crypto/st,stm32-crc.yaml
index cee624c..b72e485 100644
--- a/Documentation/devicetree/bindings/crypto/st,stm32-crc.yaml
+++ b/Documentation/devicetree/bindings/crypto/st,stm32-crc.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 CRC bindings
 
 maintainers:
-  - Lionel Debieve <lionel.debieve@st.com>
+  - Lionel Debieve <lionel.debieve@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/crypto/st,stm32-cryp.yaml b/Documentation/devicetree/bindings/crypto/st,stm32-cryp.yaml
index a457455..ed23bf9 100644
--- a/Documentation/devicetree/bindings/crypto/st,stm32-cryp.yaml
+++ b/Documentation/devicetree/bindings/crypto/st,stm32-cryp.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 CRYP bindings
 
 maintainers:
-  - Lionel Debieve <lionel.debieve@st.com>
+  - Lionel Debieve <lionel.debieve@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/crypto/st,stm32-hash.yaml b/Documentation/devicetree/bindings/crypto/st,stm32-hash.yaml
index 6dd658f..10ba947 100644
--- a/Documentation/devicetree/bindings/crypto/st,stm32-hash.yaml
+++ b/Documentation/devicetree/bindings/crypto/st,stm32-hash.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 HASH bindings
 
 maintainers:
-  - Lionel Debieve <lionel.debieve@st.com>
+  - Lionel Debieve <lionel.debieve@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/display/bridge/snps,dw-mipi-dsi.yaml b/Documentation/devicetree/bindings/display/bridge/snps,dw-mipi-dsi.yaml
index 3c3e51a..11fd68a 100644
--- a/Documentation/devicetree/bindings/display/bridge/snps,dw-mipi-dsi.yaml
+++ b/Documentation/devicetree/bindings/display/bridge/snps,dw-mipi-dsi.yaml
@@ -7,7 +7,7 @@
 title: Synopsys DesignWare MIPI DSI host controller
 
 maintainers:
-  - Philippe CORNU <philippe.cornu@st.com>
+  - Philippe CORNU <philippe.cornu@foss.st.com>
 
 description: |
   This document defines device tree properties for the Synopsys DesignWare MIPI
diff --git a/Documentation/devicetree/bindings/display/bridge/toshiba,tc358767.txt b/Documentation/devicetree/bindings/display/bridge/toshiba,tc358767.txt
deleted file mode 100644
index 583c5e9..0000000
--- a/Documentation/devicetree/bindings/display/bridge/toshiba,tc358767.txt
+++ /dev/null
@@ -1,54 +0,0 @@
-Toshiba TC358767 eDP bridge bindings
-
-Required properties:
- - compatible: "toshiba,tc358767"
- - reg: i2c address of the bridge, 0x68 or 0x0f, depending on bootstrap pins
- - clock-names: should be "ref"
- - clocks: OF device-tree clock specification for refclk input. The reference
-   clock rate must be 13 MHz, 19.2 MHz, 26 MHz, or 38.4 MHz.
-
-Optional properties:
- - shutdown-gpios: OF device-tree gpio specification for SD pin
-                   (active high shutdown input)
- - reset-gpios: OF device-tree gpio specification for RSTX pin
-                (active low system reset)
- - toshiba,hpd-pin: TC358767 GPIO pin number to which HPD is connected to (0 or 1)
- - ports: the ports node can contain video interface port nodes to connect
-   to a DPI/DSI source and to an eDP/DP sink according to [1][2]:
-    - port@0: DSI input port
-    - port@1: DPI input port
-    - port@2: eDP/DP output port
-
-[1]: Documentation/devicetree/bindings/graph.txt
-[2]: Documentation/devicetree/bindings/media/video-interfaces.txt
-
-Example:
-	edp-bridge@68 {
-		compatible = "toshiba,tc358767";
-		reg = <0x68>;
-		shutdown-gpios = <&gpio3 23 GPIO_ACTIVE_HIGH>;
-		reset-gpios = <&gpio3 24 GPIO_ACTIVE_LOW>;
-		clock-names = "ref";
-		clocks = <&edp_refclk>;
-
-		ports {
-			#address-cells = <1>;
-			#size-cells = <0>;
-
-			port@1 {
-				reg = <1>;
-
-				bridge_in: endpoint {
-					remote-endpoint = <&dpi_out>;
-				};
-			};
-
-			port@2 {
-				reg = <2>;
-
-				bridge_out: endpoint {
-					remote-endpoint = <&panel_in>;
-				};
-			};
-		};
-	};
diff --git a/Documentation/devicetree/bindings/display/bridge/toshiba,tc358767.yaml b/Documentation/devicetree/bindings/display/bridge/toshiba,tc358767.yaml
new file mode 100644
index 0000000..f1541cc
--- /dev/null
+++ b/Documentation/devicetree/bindings/display/bridge/toshiba,tc358767.yaml
@@ -0,0 +1,158 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/display/bridge/toshiba,tc358767.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Toshiba TC358767 eDP bridge bindings
+
+maintainers:
+  - Andrey Gusakov <andrey.gusakov@cogentembedded.com>
+
+description: The TC358767 is bridge device which converts DSI/DPI to eDP/DP
+
+properties:
+  compatible:
+    const: toshiba,tc358767
+
+  reg:
+    enum:
+      - 0x68
+      - 0x0f
+    description: |
+        i2c address of the bridge, 0x68 or 0x0f, depending on bootstrap pins
+
+  clock-names:
+    const: "ref"
+
+  clocks:
+    maxItems: 1
+    description: |
+        OF device-tree clock specification for refclk input. The reference.
+        clock rate must be 13 MHz, 19.2 MHz, 26 MHz, or 38.4 MHz.
+
+  shutdown-gpios:
+    maxItems: 1
+    description: |
+        OF device-tree gpio specification for SD pin(active high shutdown input)
+
+  reset-gpios:
+    maxItems: 1
+    description: |
+        OF device-tree gpio specification for RSTX pin(active low system reset)
+
+  toshiba,hpd-pin:
+    $ref: /schemas/types.yaml#/definitions/uint32
+    enum:
+      - 0
+      - 1
+    description: TC358767 GPIO pin number to which HPD is connected to (0 or 1)
+
+  ports:
+    $ref: /schemas/graph.yaml#/properties/ports
+
+    properties:
+      port@0:
+        $ref: /schemas/graph.yaml#/properties/port
+        description: |
+            DSI input port. The remote endpoint phandle should be a
+            reference to a valid DSI output endpoint node
+
+      port@1:
+        $ref: /schemas/graph.yaml#/properties/port
+        description: |
+            DPI input port. The remote endpoint phandle should be a
+            reference to a valid DPI output endpoint node
+
+      port@2:
+        $ref: /schemas/graph.yaml#/properties/port
+        description: |
+            eDP/DP output port. The remote endpoint phandle should be a
+            reference to a valid eDP panel input endpoint node. This port is
+            optional, treated as DP panel if not defined
+
+    oneOf:
+      - required:
+          - port@0
+      - required:
+          - port@1
+
+
+required:
+  - compatible
+  - reg
+  - clock-names
+  - clocks
+  - ports
+
+additionalProperties: false
+
+examples:
+  - |
+    #include <dt-bindings/gpio/gpio.h>
+
+    /* DPI input and eDP output */
+
+    i2c {
+        #address-cells = <1>;
+        #size-cells = <0>;
+
+        edp-bridge@68 {
+            compatible = "toshiba,tc358767";
+            reg = <0x68>;
+            shutdown-gpios = <&gpio3 23 GPIO_ACTIVE_HIGH>;
+            reset-gpios = <&gpio3 24 GPIO_ACTIVE_LOW>;
+            clock-names = "ref";
+            clocks = <&edp_refclk>;
+
+            ports {
+                #address-cells = <1>;
+                #size-cells = <0>;
+
+                port@1 {
+                    reg = <1>;
+
+                    bridge_in_0: endpoint {
+                        remote-endpoint = <&dpi_out>;
+                    };
+                };
+
+                port@2 {
+                    reg = <2>;
+
+                    bridge_out: endpoint {
+                        remote-endpoint = <&panel_in>;
+                    };
+                };
+            };
+        };
+    };
+  - |
+    /* DPI input and DP output */
+
+    i2c {
+        #address-cells = <1>;
+        #size-cells = <0>;
+
+        edp-bridge@68 {
+            compatible = "toshiba,tc358767";
+            reg = <0x68>;
+            shutdown-gpios = <&gpio3 23 GPIO_ACTIVE_HIGH>;
+            reset-gpios = <&gpio3 24 GPIO_ACTIVE_LOW>;
+            clock-names = "ref";
+            clocks = <&edp_refclk>;
+
+            ports {
+                #address-cells = <1>;
+                #size-cells = <0>;
+
+                port@1 {
+                    reg = <1>;
+
+                    bridge_in_1: endpoint {
+                        remote-endpoint = <&dpi_out>;
+                    };
+                };
+            };
+        };
+    };
diff --git a/Documentation/devicetree/bindings/display/ingenic,ipu.yaml b/Documentation/devicetree/bindings/display/ingenic,ipu.yaml
index e679f48..3f93def 100644
--- a/Documentation/devicetree/bindings/display/ingenic,ipu.yaml
+++ b/Documentation/devicetree/bindings/display/ingenic,ipu.yaml
@@ -45,7 +45,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4770-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4770-cgu.h>
     ipu@13080000 {
       compatible = "ingenic,jz4770-ipu", "ingenic,jz4760-ipu";
       reg = <0x13080000 0x800>;
diff --git a/Documentation/devicetree/bindings/display/ingenic,lcd.yaml b/Documentation/devicetree/bindings/display/ingenic,lcd.yaml
index 50d2b0a..0049010 100644
--- a/Documentation/devicetree/bindings/display/ingenic,lcd.yaml
+++ b/Documentation/devicetree/bindings/display/ingenic,lcd.yaml
@@ -88,7 +88,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4740-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4740-cgu.h>
     lcd-controller@13050000 {
       compatible = "ingenic,jz4740-lcd";
       reg = <0x13050000 0x1000>;
@@ -107,7 +107,7 @@
     };
 
   - |
-    #include <dt-bindings/clock/jz4725b-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4725b-cgu.h>
     lcd-controller@13050000 {
       compatible = "ingenic,jz4725b-lcd";
       reg = <0x13050000 0x1000>;
diff --git a/Documentation/devicetree/bindings/display/panel/orisetech,otm8009a.yaml b/Documentation/devicetree/bindings/display/panel/orisetech,otm8009a.yaml
index 4b6dda6..17cbd0a 100644
--- a/Documentation/devicetree/bindings/display/panel/orisetech,otm8009a.yaml
+++ b/Documentation/devicetree/bindings/display/panel/orisetech,otm8009a.yaml
@@ -7,7 +7,7 @@
 title: Orise Tech OTM8009A 3.97" 480x800 TFT LCD panel (MIPI-DSI video mode)
 
 maintainers:
-  - Philippe CORNU <philippe.cornu@st.com>
+  - Philippe CORNU <philippe.cornu@foss.st.com>
 
 description: |
              The Orise Tech OTM8009A is a 3.97" 480x800 TFT LCD panel connected using
diff --git a/Documentation/devicetree/bindings/display/panel/raydium,rm68200.yaml b/Documentation/devicetree/bindings/display/panel/raydium,rm68200.yaml
index 3947779..e8ce231 100644
--- a/Documentation/devicetree/bindings/display/panel/raydium,rm68200.yaml
+++ b/Documentation/devicetree/bindings/display/panel/raydium,rm68200.yaml
@@ -7,7 +7,7 @@
 title: Raydium Semiconductor Corporation RM68200 5.5" 720p MIPI-DSI TFT LCD panel
 
 maintainers:
-  - Philippe CORNU <philippe.cornu@st.com>
+  - Philippe CORNU <philippe.cornu@foss.st.com>
 
 description: |
   The Raydium Semiconductor Corporation RM68200 is a 5.5" 720x1280 TFT LCD
diff --git a/Documentation/devicetree/bindings/display/st,stm32-dsi.yaml b/Documentation/devicetree/bindings/display/st,stm32-dsi.yaml
index ed310bb..ce1ef93 100644
--- a/Documentation/devicetree/bindings/display/st,stm32-dsi.yaml
+++ b/Documentation/devicetree/bindings/display/st,stm32-dsi.yaml
@@ -7,8 +7,8 @@
 title: STMicroelectronics STM32 DSI host controller
 
 maintainers:
-  - Philippe Cornu <philippe.cornu@st.com>
-  - Yannick Fertre <yannick.fertre@st.com>
+  - Philippe Cornu <philippe.cornu@foss.st.com>
+  - Yannick Fertre <yannick.fertre@foss.st.com>
 
 description:
   The STMicroelectronics STM32 DSI controller uses the Synopsys DesignWare MIPI-DSI host controller.
diff --git a/Documentation/devicetree/bindings/display/st,stm32-ltdc.yaml b/Documentation/devicetree/bindings/display/st,stm32-ltdc.yaml
index 4ae3d75..01e2da2 100644
--- a/Documentation/devicetree/bindings/display/st,stm32-ltdc.yaml
+++ b/Documentation/devicetree/bindings/display/st,stm32-ltdc.yaml
@@ -7,8 +7,8 @@
 title: STMicroelectronics STM32 lcd-tft display controller
 
 maintainers:
-  - Philippe Cornu <philippe.cornu@st.com>
-  - Yannick Fertre <yannick.fertre@st.com>
+  - Philippe Cornu <philippe.cornu@foss.st.com>
+  - Yannick Fertre <yannick.fertre@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/dma/ingenic,dma.yaml b/Documentation/devicetree/bindings/dma/ingenic,dma.yaml
index ac4d594..dc059d6 100644
--- a/Documentation/devicetree/bindings/dma/ingenic,dma.yaml
+++ b/Documentation/devicetree/bindings/dma/ingenic,dma.yaml
@@ -68,7 +68,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4780-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4780-cgu.h>
     dma: dma-controller@13420000 {
       compatible = "ingenic,jz4780-dma";
       reg = <0x13420000 0x400>, <0x13421000 0x40>;
diff --git a/Documentation/devicetree/bindings/dma/st,stm32-dma.yaml b/Documentation/devicetree/bindings/dma/st,stm32-dma.yaml
index 4bf676f..55faab6 100644
--- a/Documentation/devicetree/bindings/dma/st,stm32-dma.yaml
+++ b/Documentation/devicetree/bindings/dma/st,stm32-dma.yaml
@@ -50,7 +50,7 @@
 
 
 maintainers:
-  - Amelie Delaunay <amelie.delaunay@st.com>
+  - Amelie Delaunay <amelie.delaunay@foss.st.com>
 
 allOf:
   - $ref: "dma-controller.yaml#"
diff --git a/Documentation/devicetree/bindings/dma/st,stm32-dmamux.yaml b/Documentation/devicetree/bindings/dma/st,stm32-dmamux.yaml
index c8d2b51..f751796 100644
--- a/Documentation/devicetree/bindings/dma/st,stm32-dmamux.yaml
+++ b/Documentation/devicetree/bindings/dma/st,stm32-dmamux.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 DMA MUX (DMA request router) bindings
 
 maintainers:
-  - Amelie Delaunay <amelie.delaunay@st.com>
+  - Amelie Delaunay <amelie.delaunay@foss.st.com>
 
 allOf:
   - $ref: "dma-router.yaml#"
diff --git a/Documentation/devicetree/bindings/dma/st,stm32-mdma.yaml b/Documentation/devicetree/bindings/dma/st,stm32-mdma.yaml
index c30be84..87b4afd 100644
--- a/Documentation/devicetree/bindings/dma/st,stm32-mdma.yaml
+++ b/Documentation/devicetree/bindings/dma/st,stm32-mdma.yaml
@@ -50,7 +50,7 @@
        if no HW ack signal is used by the MDMA client
 
 maintainers:
-  - Amelie Delaunay <amelie.delaunay@st.com>
+  - Amelie Delaunay <amelie.delaunay@foss.st.com>
 
 allOf:
   - $ref: "dma-controller.yaml#"
diff --git a/Documentation/devicetree/bindings/gpio/gpio-xlp.txt b/Documentation/devicetree/bindings/gpio/gpio-xlp.txt
deleted file mode 100644
index 47fc649..0000000
--- a/Documentation/devicetree/bindings/gpio/gpio-xlp.txt
+++ /dev/null
@@ -1,49 +0,0 @@
-Netlogic XLP Family GPIO
-========================
-
-This GPIO driver is used for following Netlogic XLP SoCs:
-	XLP832, XLP316, XLP208, XLP980, XLP532
-This GPIO driver is also compatible with GPIO controller found on
-Broadcom Vulcan ARM64.
-
-Required properties:
--------------------
-
-- compatible: Should be one of the following:
-  - "netlogic,xlp832-gpio": For Netlogic XLP832
-  - "netlogic,xlp316-gpio": For Netlogic XLP316
-  - "netlogic,xlp208-gpio": For Netlogic XLP208
-  - "netlogic,xlp980-gpio": For Netlogic XLP980
-  - "netlogic,xlp532-gpio": For Netlogic XLP532
-  - "brcm,vulcan-gpio": For Broadcom Vulcan ARM64
-- reg: Physical base address and length of the controller's registers.
-- #gpio-cells: Should be two. The first cell is the pin number and the second
-  cell is used to specify optional parameters (currently unused).
-- gpio-controller: Marks the device node as a GPIO controller.
-- nr-gpios: Number of GPIO pins supported by the controller.
-- interrupt-cells: Should be two. The first cell is the GPIO Number. The
-  second cell is used to specify flags. The following subset of flags is
-  supported:
-  - trigger type:
-	1 = low to high edge triggered.
-	2 = high to low edge triggered.
-	4 = active high level-sensitive.
-	8 = active low level-sensitive.
-- interrupts: Interrupt number for this device.
-- interrupt-controller: Identifies the node as an interrupt controller.
-
-Example:
-
-	gpio: xlp_gpio@34000 {
-		compatible = "netlogic,xlp316-gpio";
-		reg = <0 0x34100 0x1000
-		       0 0x35100 0x1000>;
-		#gpio-cells = <2>;
-		gpio-controller;
-		nr-gpios = <57>;
-
-		#interrupt-cells = <2>;
-		interrupt-parent = <&pic>;
-		interrupts = <39>;
-		interrupt-controller;
-	};
diff --git a/Documentation/devicetree/bindings/hwlock/st,stm32-hwspinlock.yaml b/Documentation/devicetree/bindings/hwlock/st,stm32-hwspinlock.yaml
index 47cf9c8..b18c616 100644
--- a/Documentation/devicetree/bindings/hwlock/st,stm32-hwspinlock.yaml
+++ b/Documentation/devicetree/bindings/hwlock/st,stm32-hwspinlock.yaml
@@ -7,8 +7,7 @@
 title: STMicroelectronics STM32 Hardware Spinlock bindings
 
 maintainers:
-  - Benjamin Gaignard <benjamin.gaignard@st.com>
-  - Fabien Dessenne <fabien.dessenne@st.com>
+  - Fabien Dessenne <fabien.dessenne@foss.st.com>
 
 properties:
   "#hwlock-cells":
diff --git a/Documentation/devicetree/bindings/i2c/i2c-imx.yaml b/Documentation/devicetree/bindings/i2c/i2c-imx.yaml
index 3592d49..c167958 100644
--- a/Documentation/devicetree/bindings/i2c/i2c-imx.yaml
+++ b/Documentation/devicetree/bindings/i2c/i2c-imx.yaml
@@ -57,7 +57,9 @@
     const: ipg
 
   clock-frequency:
-    enum: [ 100000, 400000 ]
+    minimum: 1
+    default: 100000
+    maximum: 400000
 
   dmas:
     items:
diff --git a/Documentation/devicetree/bindings/i2c/i2c-xlp9xx.txt b/Documentation/devicetree/bindings/i2c/i2c-xlp9xx.txt
deleted file mode 100644
index f818ef5..0000000
--- a/Documentation/devicetree/bindings/i2c/i2c-xlp9xx.txt
+++ /dev/null
@@ -1,22 +0,0 @@
-Device tree configuration for the I2C controller on the XLP9xx/5xx SoC
-
-Required properties:
-- compatible      : should be "netlogic,xlp980-i2c"
-- reg             : bus address start and address range size of device
-- interrupts      : interrupt number
-
-Optional properties:
-- clock-frequency : frequency of bus clock in Hz
-                    Defaults to 100 KHz when the property is not specified
-
-Example:
-
-i2c0: i2c@113100 {
-	compatible = "netlogic,xlp980-i2c";
-	#address-cells = <1>;
-	#size-cells = <0>;
-	reg = <0 0x113100 0x100>;
-	clock-frequency = <400000>;
-	interrupts = <30>;
-	interrupt-parent = <&pic>;
-};
diff --git a/Documentation/devicetree/bindings/i2c/ingenic,i2c.yaml b/Documentation/devicetree/bindings/i2c/ingenic,i2c.yaml
index e1e65eb..febde6cc 100644
--- a/Documentation/devicetree/bindings/i2c/ingenic,i2c.yaml
+++ b/Documentation/devicetree/bindings/i2c/ingenic,i2c.yaml
@@ -60,7 +60,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4780-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4780-cgu.h>
     #include <dt-bindings/dma/jz4780-dma.h>
     #include <dt-bindings/interrupt-controller/irq.h>
     i2c@10054000 {
diff --git a/Documentation/devicetree/bindings/i2c/st,stm32-i2c.yaml b/Documentation/devicetree/bindings/i2c/st,stm32-i2c.yaml
index d747f49..c07289a 100644
--- a/Documentation/devicetree/bindings/i2c/st,stm32-i2c.yaml
+++ b/Documentation/devicetree/bindings/i2c/st,stm32-i2c.yaml
@@ -7,7 +7,7 @@
 title: I2C controller embedded in STMicroelectronics STM32 I2C platform
 
 maintainers:
-  - Pierre-Yves MORDRET <pierre-yves.mordret@st.com>
+  - Pierre-Yves MORDRET <pierre-yves.mordret@foss.st.com>
 
 allOf:
   - $ref: /schemas/i2c/i2c-controller.yaml#
diff --git a/Documentation/devicetree/bindings/iio/adc/ingenic,adc.yaml b/Documentation/devicetree/bindings/iio/adc/ingenic,adc.yaml
index 3eb7aa8..698beb8 100644
--- a/Documentation/devicetree/bindings/iio/adc/ingenic,adc.yaml
+++ b/Documentation/devicetree/bindings/iio/adc/ingenic,adc.yaml
@@ -74,7 +74,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4740-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4740-cgu.h>
     #include <dt-bindings/iio/adc/ingenic,adc.h>
 
     adc@10070000 {
diff --git a/Documentation/devicetree/bindings/iio/adc/sigma-delta-modulator.yaml b/Documentation/devicetree/bindings/iio/adc/sigma-delta-modulator.yaml
index a390343d..2287697 100644
--- a/Documentation/devicetree/bindings/iio/adc/sigma-delta-modulator.yaml
+++ b/Documentation/devicetree/bindings/iio/adc/sigma-delta-modulator.yaml
@@ -7,7 +7,7 @@
 title: Device-Tree bindings for sigma delta modulator
 
 maintainers:
-  - Arnaud Pouliquen <arnaud.pouliquen@st.com>
+  - Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/iio/adc/st,stm32-adc.yaml b/Documentation/devicetree/bindings/iio/adc/st,stm32-adc.yaml
index ec0450d..4d60745 100644
--- a/Documentation/devicetree/bindings/iio/adc/st,stm32-adc.yaml
+++ b/Documentation/devicetree/bindings/iio/adc/st,stm32-adc.yaml
@@ -19,7 +19,7 @@
   Each STM32 ADC block can have up to 3 ADC instances.
 
 maintainers:
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/iio/adc/st,stm32-dfsdm-adc.yaml b/Documentation/devicetree/bindings/iio/adc/st,stm32-dfsdm-adc.yaml
index 733351d..7c260f2 100644
--- a/Documentation/devicetree/bindings/iio/adc/st,stm32-dfsdm-adc.yaml
+++ b/Documentation/devicetree/bindings/iio/adc/st,stm32-dfsdm-adc.yaml
@@ -7,8 +7,8 @@
 title: STMicroelectronics STM32 DFSDM ADC device driver
 
 maintainers:
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
-  - Olivier Moysan <olivier.moysan@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
+  - Olivier Moysan <olivier.moysan@foss.st.com>
 
 description: |
   STM32 DFSDM ADC is a sigma delta analog-to-digital converter dedicated to
diff --git a/Documentation/devicetree/bindings/iio/dac/st,stm32-dac.yaml b/Documentation/devicetree/bindings/iio/dac/st,stm32-dac.yaml
index 393f700..6adeda4 100644
--- a/Documentation/devicetree/bindings/iio/dac/st,stm32-dac.yaml
+++ b/Documentation/devicetree/bindings/iio/dac/st,stm32-dac.yaml
@@ -15,7 +15,7 @@
   current.
 
 maintainers:
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/interrupt-controller/st,stm32-exti.yaml b/Documentation/devicetree/bindings/interrupt-controller/st,stm32-exti.yaml
index 6d3e68e..d19c881 100644
--- a/Documentation/devicetree/bindings/interrupt-controller/st,stm32-exti.yaml
+++ b/Documentation/devicetree/bindings/interrupt-controller/st,stm32-exti.yaml
@@ -7,8 +7,8 @@
 title: STM32 External Interrupt Controller Device Tree Bindings
 
 maintainers:
-  - Alexandre Torgue <alexandre.torgue@st.com>
-  - Ludovic Barre <ludovic.barre@st.com>
+  - Alexandre Torgue <alexandre.torgue@foss.st.com>
+  - Ludovic Barre <ludovic.barre@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/mailbox/st,stm32-ipcc.yaml b/Documentation/devicetree/bindings/mailbox/st,stm32-ipcc.yaml
index b15da9b..8eb4bf5 100644
--- a/Documentation/devicetree/bindings/mailbox/st,stm32-ipcc.yaml
+++ b/Documentation/devicetree/bindings/mailbox/st,stm32-ipcc.yaml
@@ -13,8 +13,8 @@
   channels (N) can be read from a dedicated register.
 
 maintainers:
-  - Fabien Dessenne <fabien.dessenne@st.com>
-  - Arnaud Pouliquen <arnaud.pouliquen@st.com>
+  - Fabien Dessenne <fabien.dessenne@foss.st.com>
+  - Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/media/qcom,sc7280-venus.yaml b/Documentation/devicetree/bindings/media/qcom,sc7280-venus.yaml
index fa54c56..e287468 100644
--- a/Documentation/devicetree/bindings/media/qcom,sc7280-venus.yaml
+++ b/Documentation/devicetree/bindings/media/qcom,sc7280-venus.yaml
@@ -30,7 +30,6 @@
 
   power-domain-names:
     minItems: 2
-    maxItems: 3
     items:
       - const: venus
       - const: vcodec0
diff --git a/Documentation/devicetree/bindings/media/st,stm32-cec.yaml b/Documentation/devicetree/bindings/media/st,stm32-cec.yaml
index d75019c..77144cc 100644
--- a/Documentation/devicetree/bindings/media/st,stm32-cec.yaml
+++ b/Documentation/devicetree/bindings/media/st,stm32-cec.yaml
@@ -7,8 +7,7 @@
 title: STMicroelectronics STM32 CEC bindings
 
 maintainers:
-  - Benjamin Gaignard <benjamin.gaignard@st.com>
-  - Yannick Fertre <yannick.fertre@st.com>
+  - Yannick Fertre <yannick.fertre@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/media/st,stm32-dcmi.yaml b/Documentation/devicetree/bindings/media/st,stm32-dcmi.yaml
index 41e1d0c..9c1262a 100644
--- a/Documentation/devicetree/bindings/media/st,stm32-dcmi.yaml
+++ b/Documentation/devicetree/bindings/media/st,stm32-dcmi.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 Digital Camera Memory Interface (DCMI) binding
 
 maintainers:
-  - Hugues Fruchet <hugues.fruchet@st.com>
+  - Hugues Fruchet <hugues.fruchet@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/memory-controllers/ingenic,nemc.yaml b/Documentation/devicetree/bindings/memory-controllers/ingenic,nemc.yaml
index fe0ce19..24f9e19 100644
--- a/Documentation/devicetree/bindings/memory-controllers/ingenic,nemc.yaml
+++ b/Documentation/devicetree/bindings/memory-controllers/ingenic,nemc.yaml
@@ -84,7 +84,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4780-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4780-cgu.h>
     #include <dt-bindings/gpio/gpio.h>
     nemc: memory-controller@13410000 {
       compatible = "ingenic,jz4780-nemc";
diff --git a/Documentation/devicetree/bindings/memory-controllers/st,stm32-fmc2-ebi.yaml b/Documentation/devicetree/bindings/memory-controllers/st,stm32-fmc2-ebi.yaml
index cba7420..6b516d3 100644
--- a/Documentation/devicetree/bindings/memory-controllers/st,stm32-fmc2-ebi.yaml
+++ b/Documentation/devicetree/bindings/memory-controllers/st,stm32-fmc2-ebi.yaml
@@ -19,7 +19,7 @@
   Select. The FMC2 performs only one access at a time to an external device.
 
 maintainers:
-  - Christophe Kerello <christophe.kerello@st.com>
+  - Christophe Kerello <christophe.kerello@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/mfd/st,stm32-lptimer.yaml b/Documentation/devicetree/bindings/mfd/st,stm32-lptimer.yaml
index 8bcea8d..ec7f019 100644
--- a/Documentation/devicetree/bindings/mfd/st,stm32-lptimer.yaml
+++ b/Documentation/devicetree/bindings/mfd/st,stm32-lptimer.yaml
@@ -17,7 +17,7 @@
      - simple counter from IN1 input signal.
 
 maintainers:
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/mfd/st,stm32-timers.yaml b/Documentation/devicetree/bindings/mfd/st,stm32-timers.yaml
index dace353..10b330d 100644
--- a/Documentation/devicetree/bindings/mfd/st,stm32-timers.yaml
+++ b/Documentation/devicetree/bindings/mfd/st,stm32-timers.yaml
@@ -17,8 +17,7 @@
       programmable prescaler.
 
 maintainers:
-  - Benjamin Gaignard <benjamin.gaignard@st.com>
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/mfd/st,stmfx.yaml b/Documentation/devicetree/bindings/mfd/st,stmfx.yaml
index 19e9afb..b2a4e4a 100644
--- a/Documentation/devicetree/bindings/mfd/st,stmfx.yaml
+++ b/Documentation/devicetree/bindings/mfd/st,stmfx.yaml
@@ -12,7 +12,7 @@
                through VDD) and resistive touchscreen controller.
 
 maintainers:
-  - Amelie Delaunay <amelie.delaunay@st.com>
+  - Amelie Delaunay <amelie.delaunay@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/mfd/st,stpmic1.yaml b/Documentation/devicetree/bindings/mfd/st,stpmic1.yaml
index 305123e7..426658a 100644
--- a/Documentation/devicetree/bindings/mfd/st,stpmic1.yaml
+++ b/Documentation/devicetree/bindings/mfd/st,stpmic1.yaml
@@ -9,7 +9,7 @@
 description: STMicroelectronics STPMIC1 Power Management IC
 
 maintainers:
-  - pascal Paillet <p.paillet@st.com>
+  - pascal Paillet <p.paillet@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/mips/ingenic/ingenic,cpu.yaml b/Documentation/devicetree/bindings/mips/ingenic/ingenic,cpu.yaml
index 6df1a94..b7e7fa71 100644
--- a/Documentation/devicetree/bindings/mips/ingenic/ingenic,cpu.yaml
+++ b/Documentation/devicetree/bindings/mips/ingenic/ingenic,cpu.yaml
@@ -44,7 +44,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4780-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4780-cgu.h>
 
     cpus {
         #address-cells = <1>;
diff --git a/Documentation/devicetree/bindings/mmc/ingenic,mmc.yaml b/Documentation/devicetree/bindings/mmc/ingenic,mmc.yaml
index 546480f..01d5c6d 100644
--- a/Documentation/devicetree/bindings/mmc/ingenic,mmc.yaml
+++ b/Documentation/devicetree/bindings/mmc/ingenic,mmc.yaml
@@ -61,7 +61,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4780-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4780-cgu.h>
     #include <dt-bindings/dma/jz4780-dma.h>
     mmc0: mmc@13450000 {
       compatible = "ingenic,jz4780-mmc";
diff --git a/Documentation/devicetree/bindings/mtd/ingenic,nand.yaml b/Documentation/devicetree/bindings/mtd/ingenic,nand.yaml
index 89aa3ce..9de8ef6 100644
--- a/Documentation/devicetree/bindings/mtd/ingenic,nand.yaml
+++ b/Documentation/devicetree/bindings/mtd/ingenic,nand.yaml
@@ -55,7 +55,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4780-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4780-cgu.h>
     memory-controller@13410000 {
       compatible = "ingenic,jz4780-nemc";
       reg = <0x13410000 0x10000>;
diff --git a/Documentation/devicetree/bindings/mtd/st,stm32-fmc2-nand.yaml b/Documentation/devicetree/bindings/mtd/st,stm32-fmc2-nand.yaml
index 29c5ef2..eab8ea3 100644
--- a/Documentation/devicetree/bindings/mtd/st,stm32-fmc2-nand.yaml
+++ b/Documentation/devicetree/bindings/mtd/st,stm32-fmc2-nand.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics Flexible Memory Controller 2 (FMC2) Bindings
 
 maintainers:
-  - Christophe Kerello <christophe.kerello@st.com>
+  - Christophe Kerello <christophe.kerello@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/net/ingenic,mac.yaml b/Documentation/devicetree/bindings/net/ingenic,mac.yaml
index d08a881..8e52b2e 100644
--- a/Documentation/devicetree/bindings/net/ingenic,mac.yaml
+++ b/Documentation/devicetree/bindings/net/ingenic,mac.yaml
@@ -58,7 +58,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/x1000-cgu.h>
+    #include <dt-bindings/clock/ingenic,x1000-cgu.h>
 
     mac: ethernet@134b0000 {
         compatible = "ingenic,x1000-mac";
diff --git a/Documentation/devicetree/bindings/net/snps,dwmac.yaml b/Documentation/devicetree/bindings/net/snps,dwmac.yaml
index 282d774..7ae70dc 100644
--- a/Documentation/devicetree/bindings/net/snps,dwmac.yaml
+++ b/Documentation/devicetree/bindings/net/snps,dwmac.yaml
@@ -7,7 +7,7 @@
 title: Synopsys DesignWare MAC Device Tree Bindings
 
 maintainers:
-  - Alexandre Torgue <alexandre.torgue@st.com>
+  - Alexandre Torgue <alexandre.torgue@foss.st.com>
   - Giuseppe Cavallaro <peppe.cavallaro@st.com>
   - Jose Abreu <joabreu@synopsys.com>
 
diff --git a/Documentation/devicetree/bindings/net/stm32-dwmac.yaml b/Documentation/devicetree/bindings/net/stm32-dwmac.yaml
index d3f05d5..577f4e2 100644
--- a/Documentation/devicetree/bindings/net/stm32-dwmac.yaml
+++ b/Documentation/devicetree/bindings/net/stm32-dwmac.yaml
@@ -8,8 +8,8 @@
 title: STMicroelectronics STM32 / MCU DWMAC glue layer controller
 
 maintainers:
-  - Alexandre Torgue <alexandre.torgue@st.com>
-  - Christophe Roullier <christophe.roullier@st.com>
+  - Alexandre Torgue <alexandre.torgue@foss.st.com>
+  - Christophe Roullier <christophe.roullier@foss.st.com>
 
 description:
   This file documents platform glue layer for stmmac.
diff --git a/Documentation/devicetree/bindings/nvmem/ingenic,jz4780-efuse.yaml b/Documentation/devicetree/bindings/nvmem/ingenic,jz4780-efuse.yaml
index 1485d3f..bf84768 100644
--- a/Documentation/devicetree/bindings/nvmem/ingenic,jz4780-efuse.yaml
+++ b/Documentation/devicetree/bindings/nvmem/ingenic,jz4780-efuse.yaml
@@ -33,7 +33,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4780-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4780-cgu.h>
 
     efuse@134100d0 {
         compatible = "ingenic,jz4780-efuse";
diff --git a/Documentation/devicetree/bindings/nvmem/st,stm32-romem.yaml b/Documentation/devicetree/bindings/nvmem/st,stm32-romem.yaml
index 0b80ce2..a48c8fa 100644
--- a/Documentation/devicetree/bindings/nvmem/st,stm32-romem.yaml
+++ b/Documentation/devicetree/bindings/nvmem/st,stm32-romem.yaml
@@ -13,7 +13,7 @@
   internal vref (VREFIN_CAL), unique device ID...
 
 maintainers:
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
 
 allOf:
   - $ref: "nvmem.yaml#"
diff --git a/Documentation/devicetree/bindings/phy/ingenic,phy-usb.yaml b/Documentation/devicetree/bindings/phy/ingenic,phy-usb.yaml
index 0fd93d7..5cab216 100644
--- a/Documentation/devicetree/bindings/phy/ingenic,phy-usb.yaml
+++ b/Documentation/devicetree/bindings/phy/ingenic,phy-usb.yaml
@@ -46,7 +46,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4770-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4770-cgu.h>
     otg_phy: usb-phy@3c {
       compatible = "ingenic,jz4770-phy";
       reg = <0x3c 0x10>;
diff --git a/Documentation/devicetree/bindings/phy/phy-stm32-usbphyc.yaml b/Documentation/devicetree/bindings/phy/phy-stm32-usbphyc.yaml
index 2251283..267b695 100644
--- a/Documentation/devicetree/bindings/phy/phy-stm32-usbphyc.yaml
+++ b/Documentation/devicetree/bindings/phy/phy-stm32-usbphyc.yaml
@@ -24,7 +24,7 @@
   |_ UTMI switch_______|          OTG controller
 
 maintainers:
-  - Amelie Delaunay <amelie.delaunay@st.com>
+  - Amelie Delaunay <amelie.delaunay@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/pinctrl/st,stm32-pinctrl.yaml b/Documentation/devicetree/bindings/pinctrl/st,stm32-pinctrl.yaml
index dfee6d3..ac88e01 100644
--- a/Documentation/devicetree/bindings/pinctrl/st,stm32-pinctrl.yaml
+++ b/Documentation/devicetree/bindings/pinctrl/st,stm32-pinctrl.yaml
@@ -8,7 +8,7 @@
 title: STM32 GPIO and Pin Mux/Config controller
 
 maintainers:
-  - Alexandre TORGUE <alexandre.torgue@st.com>
+  - Alexandre TORGUE <alexandre.torgue@foss.st.com>
 
 description: |
   STMicroelectronics's STM32 MCUs intregrate a GPIO and Pin mux/config hardware
diff --git a/Documentation/devicetree/bindings/regulator/st,stm32-booster.yaml b/Documentation/devicetree/bindings/regulator/st,stm32-booster.yaml
index 9f1c703..df0191b 100644
--- a/Documentation/devicetree/bindings/regulator/st,stm32-booster.yaml
+++ b/Documentation/devicetree/bindings/regulator/st,stm32-booster.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 booster for ADC analog input switches bindings
 
 maintainers:
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
 
 description: |
   Some STM32 devices embed a 3.3V booster supplied by Vdda, that can be used
diff --git a/Documentation/devicetree/bindings/regulator/st,stm32-vrefbuf.yaml b/Documentation/devicetree/bindings/regulator/st,stm32-vrefbuf.yaml
index 3cd4a254..836d415 100644
--- a/Documentation/devicetree/bindings/regulator/st,stm32-vrefbuf.yaml
+++ b/Documentation/devicetree/bindings/regulator/st,stm32-vrefbuf.yaml
@@ -12,7 +12,7 @@
   components through the dedicated VREF+ pin.
 
 maintainers:
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
 
 allOf:
   - $ref: "regulator.yaml#"
diff --git a/Documentation/devicetree/bindings/regulator/st,stm32mp1-pwr-reg.yaml b/Documentation/devicetree/bindings/regulator/st,stm32mp1-pwr-reg.yaml
index e6322bc..bd07b9c 100644
--- a/Documentation/devicetree/bindings/regulator/st,stm32mp1-pwr-reg.yaml
+++ b/Documentation/devicetree/bindings/regulator/st,stm32mp1-pwr-reg.yaml
@@ -7,7 +7,7 @@
 title: STM32MP1 PWR voltage regulators
 
 maintainers:
-  - Pascal Paillet <p.paillet@st.com>
+  - Pascal Paillet <p.paillet@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/remoteproc/ingenic,vpu.yaml b/Documentation/devicetree/bindings/remoteproc/ingenic,vpu.yaml
index d0aa91b..aaaaaba 100644
--- a/Documentation/devicetree/bindings/remoteproc/ingenic,vpu.yaml
+++ b/Documentation/devicetree/bindings/remoteproc/ingenic,vpu.yaml
@@ -58,7 +58,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4770-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4770-cgu.h>
 
     vpu: video-decoder@132a0000 {
       compatible = "ingenic,jz4770-vpu-rproc";
diff --git a/Documentation/devicetree/bindings/remoteproc/st,stm32-rproc.yaml b/Documentation/devicetree/bindings/remoteproc/st,stm32-rproc.yaml
index 1e62256..b587c97 100644
--- a/Documentation/devicetree/bindings/remoteproc/st,stm32-rproc.yaml
+++ b/Documentation/devicetree/bindings/remoteproc/st,stm32-rproc.yaml
@@ -11,8 +11,8 @@
   boots firmwares on the ST32MP family chipset.
 
 maintainers:
-  - Fabien Dessenne <fabien.dessenne@st.com>
-  - Arnaud Pouliquen <arnaud.pouliquen@st.com>
+  - Fabien Dessenne <fabien.dessenne@foss.st.com>
+  - Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/rng/ingenic,trng.yaml b/Documentation/devicetree/bindings/rng/ingenic,trng.yaml
index 808f247..044d9a0 100644
--- a/Documentation/devicetree/bindings/rng/ingenic,trng.yaml
+++ b/Documentation/devicetree/bindings/rng/ingenic,trng.yaml
@@ -32,7 +32,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/x1830-cgu.h>
+    #include <dt-bindings/clock/ingenic,x1830-cgu.h>
 
     dtrng: trng@10072000 {
         compatible = "ingenic,x1830-dtrng";
diff --git a/Documentation/devicetree/bindings/rng/st,stm32-rng.yaml b/Documentation/devicetree/bindings/rng/st,stm32-rng.yaml
index 82bb2e9..9a6e4ea 100644
--- a/Documentation/devicetree/bindings/rng/st,stm32-rng.yaml
+++ b/Documentation/devicetree/bindings/rng/st,stm32-rng.yaml
@@ -11,7 +11,7 @@
   IP and is fully separated from other crypto functions.
 
 maintainers:
-  - Lionel Debieve <lionel.debieve@st.com>
+  - Lionel Debieve <lionel.debieve@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/rtc/ingenic,rtc.yaml b/Documentation/devicetree/bindings/rtc/ingenic,rtc.yaml
index 60e93e8..b235b24 100644
--- a/Documentation/devicetree/bindings/rtc/ingenic,rtc.yaml
+++ b/Documentation/devicetree/bindings/rtc/ingenic,rtc.yaml
@@ -72,7 +72,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4740-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4740-cgu.h>
     rtc_dev: rtc@10003000 {
       compatible = "ingenic,jz4740-rtc";
       reg = <0x10003000 0x40>;
diff --git a/Documentation/devicetree/bindings/rtc/st,stm32-rtc.yaml b/Documentation/devicetree/bindings/rtc/st,stm32-rtc.yaml
index 5456604..2359f54 100644
--- a/Documentation/devicetree/bindings/rtc/st,stm32-rtc.yaml
+++ b/Documentation/devicetree/bindings/rtc/st,stm32-rtc.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 Real Time Clock Bindings
 
 maintainers:
-  - Gabriel Fernandez <gabriel.fernandez@st.com>
+  - Gabriel Fernandez <gabriel.fernandez@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/serial/ingenic,uart.yaml b/Documentation/devicetree/bindings/serial/ingenic,uart.yaml
index b432d4d..9ca7a18 100644
--- a/Documentation/devicetree/bindings/serial/ingenic,uart.yaml
+++ b/Documentation/devicetree/bindings/serial/ingenic,uart.yaml
@@ -71,7 +71,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4780-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4780-cgu.h>
     #include <dt-bindings/dma/jz4780-dma.h>
     #include <dt-bindings/gpio/gpio.h>
     serial@10032000 {
diff --git a/Documentation/devicetree/bindings/serial/st,stm32-uart.yaml b/Documentation/devicetree/bindings/serial/st,stm32-uart.yaml
index f50f4ca..333dc42 100644
--- a/Documentation/devicetree/bindings/serial/st,stm32-uart.yaml
+++ b/Documentation/devicetree/bindings/serial/st,stm32-uart.yaml
@@ -5,7 +5,7 @@
 $schema: http://devicetree.org/meta-schemas/core.yaml#
 
 maintainers:
-  - Erwan Le Ray <erwan.leray@st.com>
+  - Erwan Le Ray <erwan.leray@foss.st.com>
 
 title: STMicroelectronics STM32 USART bindings
 
diff --git a/Documentation/devicetree/bindings/sound/cirrus,cs42l51.yaml b/Documentation/devicetree/bindings/sound/cirrus,cs42l51.yaml
index 0d87e2c..963a871 100644
--- a/Documentation/devicetree/bindings/sound/cirrus,cs42l51.yaml
+++ b/Documentation/devicetree/bindings/sound/cirrus,cs42l51.yaml
@@ -7,7 +7,7 @@
 title: CS42L51 audio codec DT bindings
 
 maintainers:
-  - Olivier Moysan <olivier.moysan@st.com>
+  - Olivier Moysan <olivier.moysan@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/sound/ingenic,aic.yaml b/Documentation/devicetree/bindings/sound/ingenic,aic.yaml
index cdc0fda..d607325 100644
--- a/Documentation/devicetree/bindings/sound/ingenic,aic.yaml
+++ b/Documentation/devicetree/bindings/sound/ingenic,aic.yaml
@@ -71,7 +71,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4740-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4740-cgu.h>
     aic: audio-controller@10020000 {
       compatible = "ingenic,jz4740-i2s";
       reg = <0x10020000 0x38>;
diff --git a/Documentation/devicetree/bindings/sound/ingenic,codec.yaml b/Documentation/devicetree/bindings/sound/ingenic,codec.yaml
index 97d5f38..48aae54 100644
--- a/Documentation/devicetree/bindings/sound/ingenic,codec.yaml
+++ b/Documentation/devicetree/bindings/sound/ingenic,codec.yaml
@@ -48,7 +48,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4740-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4740-cgu.h>
     codec: audio-codec@10020080 {
       compatible = "ingenic,jz4740-codec";
       reg = <0x10020080 0x8>;
diff --git a/Documentation/devicetree/bindings/sound/st,stm32-i2s.yaml b/Documentation/devicetree/bindings/sound/st,stm32-i2s.yaml
index 6feb5a0..d3966ae 100644
--- a/Documentation/devicetree/bindings/sound/st,stm32-i2s.yaml
+++ b/Documentation/devicetree/bindings/sound/st,stm32-i2s.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 SPI/I2S Controller
 
 maintainers:
-  - Olivier Moysan <olivier.moysan@st.com>
+  - Olivier Moysan <olivier.moysan@foss.st.com>
 
 description:
   The SPI/I2S block supports I2S/PCM protocols when configured on I2S mode.
diff --git a/Documentation/devicetree/bindings/sound/st,stm32-sai.yaml b/Documentation/devicetree/bindings/sound/st,stm32-sai.yaml
index f971324..1538d11 100644
--- a/Documentation/devicetree/bindings/sound/st,stm32-sai.yaml
+++ b/Documentation/devicetree/bindings/sound/st,stm32-sai.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 Serial Audio Interface (SAI)
 
 maintainers:
-  - Olivier Moysan <olivier.moysan@st.com>
+  - Olivier Moysan <olivier.moysan@foss.st.com>
 
 description:
   The SAI interface (Serial Audio Interface) offers a wide set of audio
diff --git a/Documentation/devicetree/bindings/sound/st,stm32-spdifrx.yaml b/Documentation/devicetree/bindings/sound/st,stm32-spdifrx.yaml
index b7f7dc4..837e830 100644
--- a/Documentation/devicetree/bindings/sound/st,stm32-spdifrx.yaml
+++ b/Documentation/devicetree/bindings/sound/st,stm32-spdifrx.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 S/PDIF receiver (SPDIFRX)
 
 maintainers:
-  - Olivier Moysan <olivier.moysan@st.com>
+  - Olivier Moysan <olivier.moysan@foss.st.com>
 
 description: |
   The SPDIFRX peripheral, is designed to receive an S/PDIF flow compliant with
diff --git a/Documentation/devicetree/bindings/spi/ingenic,spi.yaml b/Documentation/devicetree/bindings/spi/ingenic,spi.yaml
index cf56cc4..5b1c7a2 100644
--- a/Documentation/devicetree/bindings/spi/ingenic,spi.yaml
+++ b/Documentation/devicetree/bindings/spi/ingenic,spi.yaml
@@ -55,7 +55,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4770-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4770-cgu.h>
     spi@10043000 {
       compatible = "ingenic,jz4770-spi", "ingenic,jz4750-spi";
       reg = <0x10043000 0x1c>;
diff --git a/Documentation/devicetree/bindings/spi/spi-xlp.txt b/Documentation/devicetree/bindings/spi/spi-xlp.txt
deleted file mode 100644
index f4925ec..0000000
--- a/Documentation/devicetree/bindings/spi/spi-xlp.txt
+++ /dev/null
@@ -1,38 +0,0 @@
-SPI Master controller for Netlogic XLP MIPS64 SOCs
-==================================================
-
-Currently this SPI controller driver is supported for the following
-Netlogic XLP SoCs:
-	XLP832, XLP316, XLP208, XLP980, XLP532
-
-Required properties:
-- compatible		: Should be "netlogic,xlp832-spi".
-- #address-cells	: Number of cells required to define a chip select address
-			  on the SPI bus.
-- #size-cells		: Should be zero.
-- reg			: Should contain register location and length.
-- clocks		: Phandle of the spi clock
-- interrupts		: Interrupt number used by this controller.
-
-SPI slave nodes must be children of the SPI master node and can contain
-properties described in Documentation/devicetree/bindings/spi/spi-bus.txt.
-
-Example:
-
-	spi: xlp_spi@3a100 {
-		compatible = "netlogic,xlp832-spi";
-		#address-cells = <1>;
-		#size-cells = <0>;
-		reg = <0 0x3a100 0x100>;
-		clocks = <&spi_clk>;
-		interrupts = <34>;
-		interrupt-parent = <&pic>;
-
-		spi_nor@1 {
-			compatible = "spansion,s25sl12801";
-			#address-cells = <1>;
-			#size-cells = <1>;
-			reg = <1>;	/* Chip Select */
-			spi-max-frequency = <40000000>;
-		};
-};
diff --git a/Documentation/devicetree/bindings/spi/st,stm32-qspi.yaml b/Documentation/devicetree/bindings/spi/st,stm32-qspi.yaml
index 983c4e5..6ec6f55 100644
--- a/Documentation/devicetree/bindings/spi/st,stm32-qspi.yaml
+++ b/Documentation/devicetree/bindings/spi/st,stm32-qspi.yaml
@@ -7,8 +7,8 @@
 title: STMicroelectronics STM32 Quad Serial Peripheral Interface (QSPI) bindings
 
 maintainers:
-  - Christophe Kerello <christophe.kerello@st.com>
-  - Patrice Chotard <patrice.chotard@st.com>
+  - Christophe Kerello <christophe.kerello@foss.st.com>
+  - Patrice Chotard <patrice.chotard@foss.st.com>
 
 allOf:
   - $ref: "spi-controller.yaml#"
diff --git a/Documentation/devicetree/bindings/spi/st,stm32-spi.yaml b/Documentation/devicetree/bindings/spi/st,stm32-spi.yaml
index 2d9af4c..3d64bed 100644
--- a/Documentation/devicetree/bindings/spi/st,stm32-spi.yaml
+++ b/Documentation/devicetree/bindings/spi/st,stm32-spi.yaml
@@ -13,8 +13,8 @@
   from 4 to 32-bit data size.
 
 maintainers:
-  - Erwan Leray <erwan.leray@st.com>
-  - Fabrice Gasnier <fabrice.gasnier@st.com>
+  - Erwan Leray <erwan.leray@foss.st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
 
 allOf:
   - $ref: "spi-controller.yaml#"
diff --git a/Documentation/devicetree/bindings/thermal/st,stm32-thermal.yaml b/Documentation/devicetree/bindings/thermal/st,stm32-thermal.yaml
index c0f59c5..bee41cf 100644
--- a/Documentation/devicetree/bindings/thermal/st,stm32-thermal.yaml
+++ b/Documentation/devicetree/bindings/thermal/st,stm32-thermal.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STM32 digital thermal sensor (DTS) binding
 
 maintainers:
-  - David Hernandez Sanchez <david.hernandezsanchez@st.com>
+  - Pascal Paillet <p.paillet@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/timer/ingenic,sysost.yaml b/Documentation/devicetree/bindings/timer/ingenic,sysost.yaml
index df3eb76..98648bf 100644
--- a/Documentation/devicetree/bindings/timer/ingenic,sysost.yaml
+++ b/Documentation/devicetree/bindings/timer/ingenic,sysost.yaml
@@ -46,7 +46,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/x1000-cgu.h>
+    #include <dt-bindings/clock/ingenic,x1000-cgu.h>
 
     ost: timer@12000000 {
         compatible = "ingenic,x1000-ost";
diff --git a/Documentation/devicetree/bindings/timer/ingenic,tcu.yaml b/Documentation/devicetree/bindings/timer/ingenic,tcu.yaml
index 8165df4..7fb37ea 100644
--- a/Documentation/devicetree/bindings/timer/ingenic,tcu.yaml
+++ b/Documentation/devicetree/bindings/timer/ingenic,tcu.yaml
@@ -237,7 +237,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4770-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4770-cgu.h>
     #include <dt-bindings/clock/ingenic,tcu.h>
     tcu: timer@10002000 {
       compatible = "ingenic,jz4770-tcu", "ingenic,jz4760-tcu", "simple-mfd";
diff --git a/Documentation/devicetree/bindings/timer/st,stm32-timer.yaml b/Documentation/devicetree/bindings/timer/st,stm32-timer.yaml
index 176aa3c..937aa8a 100644
--- a/Documentation/devicetree/bindings/timer/st,stm32-timer.yaml
+++ b/Documentation/devicetree/bindings/timer/st,stm32-timer.yaml
@@ -7,7 +7,8 @@
 title: STMicroelectronics STM32 general-purpose 16 and 32 bits timers bindings
 
 maintainers:
-  - Benjamin Gaignard <benjamin.gaignard@st.com>
+  - Fabrice Gasnier <fabrice.gasnier@foss.st.com>
+  - Patrice Chotard <patrice.chotard@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/usb/ingenic,musb.yaml b/Documentation/devicetree/bindings/usb/ingenic,musb.yaml
index f506225a4..5921235 100644
--- a/Documentation/devicetree/bindings/usb/ingenic,musb.yaml
+++ b/Documentation/devicetree/bindings/usb/ingenic,musb.yaml
@@ -58,7 +58,7 @@
 
 examples:
   - |
-    #include <dt-bindings/clock/jz4740-cgu.h>
+    #include <dt-bindings/clock/ingenic,jz4740-cgu.h>
     usb_phy: usb-phy {
       compatible = "usb-nop-xceiv";
       #phy-cells = <0>;
diff --git a/Documentation/devicetree/bindings/usb/st,stusb160x.yaml b/Documentation/devicetree/bindings/usb/st,stusb160x.yaml
index 9a51efa..ead1571 100644
--- a/Documentation/devicetree/bindings/usb/st,stusb160x.yaml
+++ b/Documentation/devicetree/bindings/usb/st,stusb160x.yaml
@@ -7,7 +7,7 @@
 title: STMicroelectronics STUSB160x Type-C controller bindings
 
 maintainers:
-  - Amelie Delaunay <amelie.delaunay@st.com>
+  - Amelie Delaunay <amelie.delaunay@foss.st.com>
 
 properties:
   compatible:
diff --git a/Documentation/devicetree/bindings/watchdog/allwinner,sun4i-a10-wdt.yaml b/Documentation/devicetree/bindings/watchdog/allwinner,sun4i-a10-wdt.yaml
index 44cad94..43afa24 100644
--- a/Documentation/devicetree/bindings/watchdog/allwinner,sun4i-a10-wdt.yaml
+++ b/Documentation/devicetree/bindings/watchdog/allwinner,sun4i-a10-wdt.yaml
@@ -40,14 +40,12 @@
 
   clocks:
     minItems: 1
-    maxItems: 2
     items:
       - description: High-frequency oscillator input, divided internally
       - description: Low-frequency oscillator input, only found on some variants
 
   clock-names:
     minItems: 1
-    maxItems: 2
     items:
       - const: hosc
       - const: losc
diff --git a/Documentation/devicetree/bindings/watchdog/st,stm32-iwdg.yaml b/Documentation/devicetree/bindings/watchdog/st,stm32-iwdg.yaml
index 481bf91f..3973644 100644
--- a/Documentation/devicetree/bindings/watchdog/st,stm32-iwdg.yaml
+++ b/Documentation/devicetree/bindings/watchdog/st,stm32-iwdg.yaml
@@ -7,8 +7,8 @@
 title: STMicroelectronics STM32 Independent WatchDoG (IWDG) bindings
 
 maintainers:
-  - Yannick Fertre <yannick.fertre@st.com>
-  - Christophe Roullier <christophe.roullier@st.com>
+  - Yannick Fertre <yannick.fertre@foss.st.com>
+  - Christophe Roullier <christophe.roullier@foss.st.com>
 
 allOf:
   - $ref: "watchdog.yaml#"
diff --git a/Documentation/x86/xstate.rst b/Documentation/x86/xstate.rst
index 65de3f0..5cec7fb 100644
--- a/Documentation/x86/xstate.rst
+++ b/Documentation/x86/xstate.rst
@@ -63,3 +63,12 @@
 the handler allocates a larger xstate buffer for the task so the large
 state can be context switched. In the unlikely cases that the allocation
 fails, the kernel sends SIGSEGV.
+
+Dynamic features in signal frames
+---------------------------------
+
+Dynamcally enabled features are not written to the signal frame upon signal
+entry if the feature is in its initial configuration.  This differs from
+non-dynamic features which are always written regardless of their
+configuration.  Signal handlers can examine the XSAVE buffer's XSTATE_BV
+field to determine if a features was written.
diff --git a/MAINTAINERS b/MAINTAINERS
index 41ec20b..7a2345c 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -14412,7 +14412,9 @@
 M:	Deep Shah <sdeep@vmware.com>
 M:	"VMware, Inc." <pv-drivers@vmware.com>
 L:	virtualization@lists.linux-foundation.org
+L:	x86@kernel.org
 S:	Supported
+T:	git git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git x86/core
 F:	Documentation/virt/paravirt_ops.rst
 F:	arch/*/include/asm/paravirt*.h
 F:	arch/*/kernel/paravirt*
@@ -21078,6 +21080,18 @@
 F:	include/linux/zsmalloc.h
 F:	mm/zsmalloc.c
 
+ZSTD
+M:	Nick Terrell <terrelln@fb.com>
+S:	Maintained
+B:	https://github.com/facebook/zstd/issues
+T:	git git://github.com/terrelln/linux.git
+F:	include/linux/zstd*
+F:	lib/zstd/
+F:	lib/decompress_unzstd.c
+F:	crypto/zstd.c
+N:	zstd
+K:	zstd
+
 ZSWAP COMPRESSED SWAP CACHING
 M:	Seth Jennings <sjenning@redhat.com>
 M:	Dan Streetman <ddstreet@ieee.org>
diff --git a/arch/arm/Makefile b/arch/arm/Makefile
index a522716..77172d5 100644
--- a/arch/arm/Makefile
+++ b/arch/arm/Makefile
@@ -60,15 +60,15 @@
 # Note that GCC does not numerically define an architecture version
 # macro, but instead defines a whole series of macros which makes
 # testing for a specific architecture or later rather impossible.
-arch-$(CONFIG_CPU_32v7M)	=-D__LINUX_ARM_ARCH__=7 -march=armv7-m -Wa,-march=armv7-m
-arch-$(CONFIG_CPU_32v7)		=-D__LINUX_ARM_ARCH__=7 $(call cc-option,-march=armv7-a,-march=armv5t -Wa$(comma)-march=armv7-a)
-arch-$(CONFIG_CPU_32v6)		=-D__LINUX_ARM_ARCH__=6 $(call cc-option,-march=armv6,-march=armv5t -Wa$(comma)-march=armv6)
+arch-$(CONFIG_CPU_32v7M)	=-D__LINUX_ARM_ARCH__=7 -march=armv7-m
+arch-$(CONFIG_CPU_32v7)		=-D__LINUX_ARM_ARCH__=7 -march=armv7-a
+arch-$(CONFIG_CPU_32v6)		=-D__LINUX_ARM_ARCH__=6 -march=armv6
 # Only override the compiler option if ARMv6. The ARMv6K extensions are
 # always available in ARMv7
 ifeq ($(CONFIG_CPU_32v6),y)
-arch-$(CONFIG_CPU_32v6K)	=-D__LINUX_ARM_ARCH__=6 $(call cc-option,-march=armv6k,-march=armv5t -Wa$(comma)-march=armv6k)
+arch-$(CONFIG_CPU_32v6K)	=-D__LINUX_ARM_ARCH__=6 -march=armv6k
 endif
-arch-$(CONFIG_CPU_32v5)		=-D__LINUX_ARM_ARCH__=5 $(call cc-option,-march=armv5te,-march=armv4t)
+arch-$(CONFIG_CPU_32v5)		=-D__LINUX_ARM_ARCH__=5 -march=armv5te
 arch-$(CONFIG_CPU_32v4T)	=-D__LINUX_ARM_ARCH__=4 -march=armv4t
 arch-$(CONFIG_CPU_32v4)		=-D__LINUX_ARM_ARCH__=4 -march=armv4
 arch-$(CONFIG_CPU_32v3)		=-D__LINUX_ARM_ARCH__=3 -march=armv3m
@@ -82,7 +82,7 @@
 tune-$(CONFIG_CPU_ARM740T)	=-mtune=arm7tdmi
 tune-$(CONFIG_CPU_ARM9TDMI)	=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_ARM940T)	=-mtune=arm9tdmi
-tune-$(CONFIG_CPU_ARM946E)	=$(call cc-option,-mtune=arm9e,-mtune=arm9tdmi)
+tune-$(CONFIG_CPU_ARM946E)	=-mtune=arm9e
 tune-$(CONFIG_CPU_ARM920T)	=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_ARM922T)	=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_ARM925T)	=-mtune=arm9tdmi
@@ -90,11 +90,11 @@
 tune-$(CONFIG_CPU_FA526)	=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_SA110)	=-mtune=strongarm110
 tune-$(CONFIG_CPU_SA1100)	=-mtune=strongarm1100
-tune-$(CONFIG_CPU_XSCALE)	=$(call cc-option,-mtune=xscale,-mtune=strongarm110) -Wa,-mcpu=xscale
-tune-$(CONFIG_CPU_XSC3)		=$(call cc-option,-mtune=xscale,-mtune=strongarm110) -Wa,-mcpu=xscale
-tune-$(CONFIG_CPU_FEROCEON)	=$(call cc-option,-mtune=marvell-f,-mtune=xscale)
-tune-$(CONFIG_CPU_V6)		=$(call cc-option,-mtune=arm1136j-s,-mtune=strongarm)
-tune-$(CONFIG_CPU_V6K)		=$(call cc-option,-mtune=arm1136j-s,-mtune=strongarm)
+tune-$(CONFIG_CPU_XSCALE)	=-mtune=xscale
+tune-$(CONFIG_CPU_XSC3)		=-mtune=xscale
+tune-$(CONFIG_CPU_FEROCEON)	=-mtune=xscale
+tune-$(CONFIG_CPU_V6)		=-mtune=arm1136j-s
+tune-$(CONFIG_CPU_V6K)		=-mtune=arm1136j-s
 
 # Evaluate tune cc-option calls now
 tune-y := $(tune-y)
diff --git a/arch/arm/mm/mmu.c b/arch/arm/mm/mmu.c
index a4e0060..274e4f7 100644
--- a/arch/arm/mm/mmu.c
+++ b/arch/arm/mm/mmu.c
@@ -390,9 +390,9 @@ void __set_fixmap(enum fixed_addresses idx, phys_addr_t phys, pgprot_t prot)
 	BUILD_BUG_ON(__fix_to_virt(__end_of_fixed_addresses) < FIXADDR_START);
 	BUG_ON(idx >= __end_of_fixed_addresses);
 
-	/* we only support device mappings until pgprot_kernel has been set */
+	/* We support only device mappings before pgprot_kernel is set. */
 	if (WARN_ON(pgprot_val(prot) != pgprot_val(FIXMAP_PAGE_IO) &&
-		    pgprot_val(pgprot_kernel) == 0))
+		    pgprot_val(prot) && pgprot_val(pgprot_kernel) == 0))
 		return;
 
 	if (pgprot_val(prot))
diff --git a/arch/mips/boot/dts/ingenic/jz4725b.dtsi b/arch/mips/boot/dts/ingenic/jz4725b.dtsi
index a1f0b71..0c6a5a4 100644
--- a/arch/mips/boot/dts/ingenic/jz4725b.dtsi
+++ b/arch/mips/boot/dts/ingenic/jz4725b.dtsi
@@ -1,5 +1,5 @@
 // SPDX-License-Identifier: GPL-2.0
-#include <dt-bindings/clock/jz4725b-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4725b-cgu.h>
 #include <dt-bindings/clock/ingenic,tcu.h>
 
 / {
diff --git a/arch/mips/boot/dts/ingenic/jz4740.dtsi b/arch/mips/boot/dts/ingenic/jz4740.dtsi
index c1afdfd..772542e 100644
--- a/arch/mips/boot/dts/ingenic/jz4740.dtsi
+++ b/arch/mips/boot/dts/ingenic/jz4740.dtsi
@@ -1,5 +1,5 @@
 // SPDX-License-Identifier: GPL-2.0
-#include <dt-bindings/clock/jz4740-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4740-cgu.h>
 #include <dt-bindings/clock/ingenic,tcu.h>
 
 / {
diff --git a/arch/mips/boot/dts/ingenic/jz4770.dtsi b/arch/mips/boot/dts/ingenic/jz4770.dtsi
index 05c00b9..dfe7432 100644
--- a/arch/mips/boot/dts/ingenic/jz4770.dtsi
+++ b/arch/mips/boot/dts/ingenic/jz4770.dtsi
@@ -1,5 +1,5 @@
 // SPDX-License-Identifier: GPL-2.0
-#include <dt-bindings/clock/jz4770-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4770-cgu.h>
 #include <dt-bindings/clock/ingenic,tcu.h>
 
 / {
diff --git a/arch/mips/boot/dts/ingenic/jz4780.dtsi b/arch/mips/boot/dts/ingenic/jz4780.dtsi
index 28adc3d9..b0a4e2e 100644
--- a/arch/mips/boot/dts/ingenic/jz4780.dtsi
+++ b/arch/mips/boot/dts/ingenic/jz4780.dtsi
@@ -1,5 +1,5 @@
 // SPDX-License-Identifier: GPL-2.0
-#include <dt-bindings/clock/jz4780-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4780-cgu.h>
 #include <dt-bindings/clock/ingenic,tcu.h>
 #include <dt-bindings/dma/jz4780-dma.h>
 
diff --git a/arch/mips/boot/dts/ingenic/x1000.dtsi b/arch/mips/boot/dts/ingenic/x1000.dtsi
index dec7909..8bd27ede 100644
--- a/arch/mips/boot/dts/ingenic/x1000.dtsi
+++ b/arch/mips/boot/dts/ingenic/x1000.dtsi
@@ -1,6 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 #include <dt-bindings/clock/ingenic,tcu.h>
-#include <dt-bindings/clock/x1000-cgu.h>
+#include <dt-bindings/clock/ingenic,x1000-cgu.h>
 #include <dt-bindings/dma/x1000-dma.h>
 
 / {
diff --git a/arch/mips/boot/dts/ingenic/x1830.dtsi b/arch/mips/boot/dts/ingenic/x1830.dtsi
index 215257f..2595df8 100644
--- a/arch/mips/boot/dts/ingenic/x1830.dtsi
+++ b/arch/mips/boot/dts/ingenic/x1830.dtsi
@@ -1,6 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 #include <dt-bindings/clock/ingenic,tcu.h>
-#include <dt-bindings/clock/x1830-cgu.h>
+#include <dt-bindings/clock/ingenic,x1830-cgu.h>
 #include <dt-bindings/dma/x1830-dma.h>
 
 / {
diff --git a/arch/sh/Kconfig b/arch/sh/Kconfig
index 6904f4b..70afb30 100644
--- a/arch/sh/Kconfig
+++ b/arch/sh/Kconfig
@@ -56,7 +56,6 @@
 	select HAVE_STACKPROTECTOR
 	select HAVE_SYSCALL_TRACEPOINTS
 	select IRQ_FORCED_THREADING
-	select MAY_HAVE_SPARSE_IRQ
 	select MODULES_USE_ELF_RELA
 	select NEED_SG_DMA_LENGTH
 	select NO_DMA if !MMU && !DMA_COHERENT
diff --git a/arch/sh/Kconfig.debug b/arch/sh/Kconfig.debug
index 958f790..10290e5 100644
--- a/arch/sh/Kconfig.debug
+++ b/arch/sh/Kconfig.debug
@@ -54,6 +54,7 @@
 
 config DWARF_UNWINDER
 	bool "Enable the DWARF unwinder for stacktraces"
+	depends on DEBUG_KERNEL
 	select FRAME_POINTER
 	default n
 	help
diff --git a/arch/sh/boards/mach-landisk/irq.c b/arch/sh/boards/mach-landisk/irq.c
index 29b8b1f..0b672b8 100644
--- a/arch/sh/boards/mach-landisk/irq.c
+++ b/arch/sh/boards/mach-landisk/irq.c
@@ -26,8 +26,8 @@ enum {
 	PCI_INTD, /* PCI int D */
 	ATA,	  /* ATA */
 	FATA,	  /* CF */
-	POWER,	  /* Power swtich */
-	BUTTON,	  /* Button swtich */
+	POWER,	  /* Power switch */
+	BUTTON,	  /* Button switch */
 };
 
 /* Vectors for LANDISK */
diff --git a/arch/sh/boot/Makefile b/arch/sh/boot/Makefile
index c081e7e..5c123f5 100644
--- a/arch/sh/boot/Makefile
+++ b/arch/sh/boot/Makefile
@@ -27,8 +27,8 @@
 suffix-$(CONFIG_KERNEL_LZO)	:= lzo
 
 targets := zImage vmlinux.srec romImage uImage uImage.srec uImage.gz \
-	   uImage.bz2 uImage.lzma uImage.xz uImage.lzo uImage.bin
-extra-y += vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2 vmlinux.bin.lzma \
+	   uImage.bz2 uImage.lzma uImage.xz uImage.lzo uImage.bin \
+	   vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2 vmlinux.bin.lzma \
 	   vmlinux.bin.xz vmlinux.bin.lzo
 subdir- := compressed romimage
 
diff --git a/arch/sh/boot/compressed/.gitignore b/arch/sh/boot/compressed/.gitignore
index 37aa530..cd16663 100644
--- a/arch/sh/boot/compressed/.gitignore
+++ b/arch/sh/boot/compressed/.gitignore
@@ -1,7 +1,2 @@
 # SPDX-License-Identifier: GPL-2.0-only
-ashiftrt.S
-ashldi3.c
-ashlsi3.S
-ashrsi3.S
-lshrsi3.S
 vmlinux.bin.*
diff --git a/arch/sh/boot/compressed/Makefile b/arch/sh/boot/compressed/Makefile
index 589d2d8..cf3174d 100644
--- a/arch/sh/boot/compressed/Makefile
+++ b/arch/sh/boot/compressed/Makefile
@@ -5,12 +5,18 @@
 # create a compressed vmlinux image from the original vmlinux
 #
 
-targets		:= vmlinux vmlinux.bin vmlinux.bin.gz \
-		   vmlinux.bin.bz2 vmlinux.bin.lzma \
-		   vmlinux.bin.xz vmlinux.bin.lzo \
-		   head_32.o misc.o piggy.o
+OBJECTS := head_32.o misc.o cache.o piggy.o \
+           ashiftrt.o ashldi3.o ashrsi3.o ashlsi3.o lshrsi3.o
 
-OBJECTS = $(obj)/head_32.o $(obj)/misc.o $(obj)/cache.o
+# These were previously generated files. When you are building the kernel
+# with O=, make sure to remove the stale files in the output tree. Otherwise,
+# the build system wrongly compiles the stale ones.
+ifdef building_out_of_srctree
+$(shell rm -f $(addprefix $(obj)/, ashiftrt.S ashldi3.c ashrsi3.S ashlsi3.S lshrsi3.S))
+endif
+
+targets := vmlinux vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2 \
+           vmlinux.bin.lzma vmlinux.bin.xz vmlinux.bin.lzo $(OBJECTS)
 
 GCOV_PROFILE := n
 
@@ -33,21 +39,9 @@
 LDFLAGS_vmlinux := --oformat $(ld-bfd) -Ttext $(IMAGE_OFFSET) -e startup \
 		   -T $(obj)/../../kernel/vmlinux.lds
 
-#
-# Pull in the necessary libgcc bits from the in-kernel implementation.
-#
-lib1funcs-y	:= ashiftrt.S ashldi3.c ashrsi3.S ashlsi3.S lshrsi3.S
-lib1funcs-obj   := \
-	$(addsuffix .o, $(basename $(addprefix $(obj)/, $(lib1funcs-y))))
+KBUILD_CFLAGS += -DDISABLE_BRANCH_PROFILING
 
-lib1funcs-dir		:= $(srctree)/arch/$(SRCARCH)/lib
-
-KBUILD_CFLAGS += -I$(lib1funcs-dir) -DDISABLE_BRANCH_PROFILING
-
-$(addprefix $(obj)/,$(lib1funcs-y)): $(obj)/%: $(lib1funcs-dir)/% FORCE
-	$(call cmd,shipped)
-
-$(obj)/vmlinux: $(OBJECTS) $(obj)/piggy.o $(lib1funcs-obj) FORCE
+$(obj)/vmlinux: $(addprefix $(obj)/, $(OBJECTS)) FORCE
 	$(call if_changed,ld)
 
 $(obj)/vmlinux.bin: vmlinux FORCE
diff --git a/arch/sh/boot/compressed/ashiftrt.S b/arch/sh/boot/compressed/ashiftrt.S
new file mode 100644
index 0000000..0f3b291
--- /dev/null
+++ b/arch/sh/boot/compressed/ashiftrt.S
@@ -0,0 +1,2 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#include "../../lib/ashiftrt.S"
diff --git a/arch/sh/boot/compressed/ashldi3.c b/arch/sh/boot/compressed/ashldi3.c
new file mode 100644
index 0000000..7cebd64
--- /dev/null
+++ b/arch/sh/boot/compressed/ashldi3.c
@@ -0,0 +1,2 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include "../../lib/ashldi3.c"
diff --git a/arch/sh/boot/compressed/ashlsi3.S b/arch/sh/boot/compressed/ashlsi3.S
new file mode 100644
index 0000000..e354262
--- /dev/null
+++ b/arch/sh/boot/compressed/ashlsi3.S
@@ -0,0 +1,2 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#include "../../lib/ashlsi3.S"
diff --git a/arch/sh/boot/compressed/ashrsi3.S b/arch/sh/boot/compressed/ashrsi3.S
new file mode 100644
index 0000000..e564be9
--- /dev/null
+++ b/arch/sh/boot/compressed/ashrsi3.S
@@ -0,0 +1,2 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#include "../../lib/ashrsi3.S"
diff --git a/arch/sh/boot/compressed/lshrsi3.S b/arch/sh/boot/compressed/lshrsi3.S
new file mode 100644
index 0000000..5a8281b
--- /dev/null
+++ b/arch/sh/boot/compressed/lshrsi3.S
@@ -0,0 +1,2 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#include "../../lib/lshrsi3.S"
diff --git a/arch/sh/include/asm/checksum_32.h b/arch/sh/include/asm/checksum_32.h
index 1a391e3..a6501b8 100644
--- a/arch/sh/include/asm/checksum_32.h
+++ b/arch/sh/include/asm/checksum_32.h
@@ -84,7 +84,8 @@ static inline __sum16 csum_fold(__wsum sum)
  */
 static inline __sum16 ip_fast_csum(const void *iph, unsigned int ihl)
 {
-	unsigned int sum, __dummy0, __dummy1;
+	__wsum sum;
+	unsigned int __dummy0, __dummy1;
 
 	__asm__ __volatile__(
 		"mov.l	@%1+, %0\n\t"
@@ -197,6 +198,6 @@ static inline __wsum csum_and_copy_to_user(const void *src,
 {
 	if (!access_ok(dst, len))
 		return 0;
-	return csum_partial_copy_generic((__force const void *)src, dst, len);
+	return csum_partial_copy_generic(src, (__force void *)dst, len);
 }
 #endif /* __ASM_SH_CHECKSUM_H */
diff --git a/arch/sh/include/asm/irq.h b/arch/sh/include/asm/irq.h
index 839551c..1c49235 100644
--- a/arch/sh/include/asm/irq.h
+++ b/arch/sh/include/asm/irq.h
@@ -6,17 +6,6 @@
 #include <asm/machvec.h>
 
 /*
- * Only legacy non-sparseirq platforms have to set a reasonably sane
- * value here. sparseirq platforms allocate their irq_descs on the fly,
- * so will expand automatically based on the number of registered IRQs.
- */
-#ifdef CONFIG_SPARSE_IRQ
-# define NR_IRQS		8
-#else
-# define NR_IRQS		512
-#endif
-
-/*
  * This is a special IRQ number for indicating that no IRQ has been
  * triggered and to simply ignore the IRQ dispatch. This is a special
  * case that can happen with IRQ auto-distribution when multiple CPUs
diff --git a/arch/sh/include/asm/sfp-machine.h b/arch/sh/include/asm/sfp-machine.h
index cbc7cf8c..2d24234 100644
--- a/arch/sh/include/asm/sfp-machine.h
+++ b/arch/sh/include/asm/sfp-machine.h
@@ -13,6 +13,14 @@
 #ifndef _SFP_MACHINE_H
 #define _SFP_MACHINE_H
 
+#ifdef __BIG_ENDIAN__
+#define __BYTE_ORDER __BIG_ENDIAN
+#define __LITTLE_ENDIAN 0
+#else
+#define __BYTE_ORDER __LITTLE_ENDIAN
+#define __BIG_ENDIAN 0
+#endif
+
 #define _FP_W_TYPE_SIZE		32
 #define _FP_W_TYPE		unsigned long
 #define _FP_WS_TYPE		signed long
diff --git a/arch/sh/include/asm/uaccess.h b/arch/sh/include/asm/uaccess.h
index 73f3b48..8867bb0 100644
--- a/arch/sh/include/asm/uaccess.h
+++ b/arch/sh/include/asm/uaccess.h
@@ -68,7 +68,7 @@ struct __large_struct { unsigned long buf[100]; };
 ({									\
 	long __gu_err = -EFAULT;					\
 	unsigned long __gu_val = 0;					\
-	const __typeof__(*(ptr)) *__gu_addr = (ptr);			\
+	const __typeof__(*(ptr)) __user *__gu_addr = (ptr);			\
 	if (likely(access_ok(__gu_addr, (size))))		\
 		__get_user_size(__gu_val, __gu_addr, (size), __gu_err);	\
 	(x) = (__force __typeof__(*(ptr)))__gu_val;			\
@@ -124,7 +124,7 @@ raw_copy_to_user(void __user *to, const void *from, unsigned long n)
  * Clear the area and return remaining number of bytes
  * (on failure.  Usually it's 0.)
  */
-__kernel_size_t __clear_user(void *addr, __kernel_size_t size);
+__kernel_size_t __clear_user(void __user *addr, __kernel_size_t size);
 
 #define clear_user(addr,n)						\
 ({									\
diff --git a/arch/sh/kernel/cpu/sh4a/smp-shx3.c b/arch/sh/kernel/cpu/sh4a/smp-shx3.c
index f8a2bec..1261dc7b 100644
--- a/arch/sh/kernel/cpu/sh4a/smp-shx3.c
+++ b/arch/sh/kernel/cpu/sh4a/smp-shx3.c
@@ -73,8 +73,9 @@ static void shx3_prepare_cpus(unsigned int max_cpus)
 	BUILD_BUG_ON(SMP_MSG_NR >= 8);
 
 	for (i = 0; i < SMP_MSG_NR; i++)
-		request_irq(104 + i, ipi_interrupt_handler,
-			    IRQF_PERCPU, "IPI", (void *)(long)i);
+		if (request_irq(104 + i, ipi_interrupt_handler,
+			    IRQF_PERCPU, "IPI", (void *)(long)i))
+			pr_err("Failed to request irq %d\n", i);
 
 	for (i = 0; i < max_cpus; i++)
 		set_cpu_present(i, true);
diff --git a/arch/sh/kernel/crash_dump.c b/arch/sh/kernel/crash_dump.c
index a908612..5b41b59 100644
--- a/arch/sh/kernel/crash_dump.c
+++ b/arch/sh/kernel/crash_dump.c
@@ -26,7 +26,7 @@
 ssize_t copy_oldmem_page(unsigned long pfn, char *buf,
                                size_t csize, unsigned long offset, int userbuf)
 {
-	void  *vaddr;
+	void  __iomem *vaddr;
 
 	if (!csize)
 		return 0;
@@ -34,7 +34,7 @@ ssize_t copy_oldmem_page(unsigned long pfn, char *buf,
 	vaddr = ioremap(pfn << PAGE_SHIFT, PAGE_SIZE);
 
 	if (userbuf) {
-		if (copy_to_user(buf, (vaddr + offset), csize)) {
+		if (copy_to_user((void __user *)buf, (vaddr + offset), csize)) {
 			iounmap(vaddr);
 			return -EFAULT;
 		}
diff --git a/arch/sh/kernel/traps_32.c b/arch/sh/kernel/traps_32.c
index b62ad0b..b3c715b 100644
--- a/arch/sh/kernel/traps_32.c
+++ b/arch/sh/kernel/traps_32.c
@@ -490,7 +490,7 @@ asmlinkage void do_address_error(struct pt_regs *regs,
 		inc_unaligned_user_access();
 
 		oldfs = force_uaccess_begin();
-		if (copy_from_user(&instruction, (insn_size_t *)(regs->pc & ~1),
+		if (copy_from_user(&instruction, (insn_size_t __user *)(regs->pc & ~1),
 				   sizeof(instruction))) {
 			force_uaccess_end(oldfs);
 			goto uspace_segv;
@@ -614,7 +614,7 @@ asmlinkage void do_reserved_inst(void)
 	unsigned short inst = 0;
 	int err;
 
-	get_user(inst, (unsigned short*)regs->pc);
+	get_user(inst, (unsigned short __user *)regs->pc);
 
 	err = do_fpu_inst(inst, regs);
 	if (!err) {
@@ -699,9 +699,9 @@ asmlinkage void do_illegal_slot_inst(void)
 		return;
 
 #ifdef CONFIG_SH_FPU_EMU
-	get_user(inst, (unsigned short *)regs->pc + 1);
+	get_user(inst, (unsigned short __user *)regs->pc + 1);
 	if (!do_fpu_inst(inst, regs)) {
-		get_user(inst, (unsigned short *)regs->pc);
+		get_user(inst, (unsigned short __user *)regs->pc);
 		if (!emulate_branch(inst, regs))
 			return;
 		/* fault in branch.*/
diff --git a/arch/sh/math-emu/math.c b/arch/sh/math-emu/math.c
index e8be0ec..cdaef65 100644
--- a/arch/sh/math-emu/math.c
+++ b/arch/sh/math-emu/math.c
@@ -51,8 +51,8 @@
 #define Rn	(regs->regs[n])
 #define Rm	(regs->regs[m])
 
-#define WRITE(d,a)	({if(put_user(d, (typeof (d)*)a)) return -EFAULT;})
-#define READ(d,a)	({if(get_user(d, (typeof (d)*)a)) return -EFAULT;})
+#define MWRITE(d,a)	({if(put_user(d, (typeof (d) __user *)a)) return -EFAULT;})
+#define MREAD(d,a)	({if(get_user(d, (typeof (d) __user *)a)) return -EFAULT;})
 
 #define PACK_S(r,f)	FP_PACK_SP(&r,f)
 #define UNPACK_S(f,r)	FP_UNPACK_SP(f,&r)
@@ -157,11 +157,11 @@ fmov_idx_reg(struct sh_fpu_soft_struct *fregs, struct pt_regs *regs, int m,
 {
 	if (FPSCR_SZ) {
 		FMOV_EXT(n);
-		READ(FRn, Rm + R0 + 4);
+		MREAD(FRn, Rm + R0 + 4);
 		n++;
-		READ(FRn, Rm + R0);
+		MREAD(FRn, Rm + R0);
 	} else {
-		READ(FRn, Rm + R0);
+		MREAD(FRn, Rm + R0);
 	}
 
 	return 0;
@@ -173,11 +173,11 @@ fmov_mem_reg(struct sh_fpu_soft_struct *fregs, struct pt_regs *regs, int m,
 {
 	if (FPSCR_SZ) {
 		FMOV_EXT(n);
-		READ(FRn, Rm + 4);
+		MREAD(FRn, Rm + 4);
 		n++;
-		READ(FRn, Rm);
+		MREAD(FRn, Rm);
 	} else {
-		READ(FRn, Rm);
+		MREAD(FRn, Rm);
 	}
 
 	return 0;
@@ -189,12 +189,12 @@ fmov_inc_reg(struct sh_fpu_soft_struct *fregs, struct pt_regs *regs, int m,
 {
 	if (FPSCR_SZ) {
 		FMOV_EXT(n);
-		READ(FRn, Rm + 4);
+		MREAD(FRn, Rm + 4);
 		n++;
-		READ(FRn, Rm);
+		MREAD(FRn, Rm);
 		Rm += 8;
 	} else {
-		READ(FRn, Rm);
+		MREAD(FRn, Rm);
 		Rm += 4;
 	}
 
@@ -207,11 +207,11 @@ fmov_reg_idx(struct sh_fpu_soft_struct *fregs, struct pt_regs *regs, int m,
 {
 	if (FPSCR_SZ) {
 		FMOV_EXT(m);
-		WRITE(FRm, Rn + R0 + 4);
+		MWRITE(FRm, Rn + R0 + 4);
 		m++;
-		WRITE(FRm, Rn + R0);
+		MWRITE(FRm, Rn + R0);
 	} else {
-		WRITE(FRm, Rn + R0);
+		MWRITE(FRm, Rn + R0);
 	}
 
 	return 0;
@@ -223,11 +223,11 @@ fmov_reg_mem(struct sh_fpu_soft_struct *fregs, struct pt_regs *regs, int m,
 {
 	if (FPSCR_SZ) {
 		FMOV_EXT(m);
-		WRITE(FRm, Rn + 4);
+		MWRITE(FRm, Rn + 4);
 		m++;
-		WRITE(FRm, Rn);
+		MWRITE(FRm, Rn);
 	} else {
-		WRITE(FRm, Rn);
+		MWRITE(FRm, Rn);
 	}
 
 	return 0;
@@ -240,12 +240,12 @@ fmov_reg_dec(struct sh_fpu_soft_struct *fregs, struct pt_regs *regs, int m,
 	if (FPSCR_SZ) {
 		FMOV_EXT(m);
 		Rn -= 8;
-		WRITE(FRm, Rn + 4);
+		MWRITE(FRm, Rn + 4);
 		m++;
-		WRITE(FRm, Rn);
+		MWRITE(FRm, Rn);
 	} else {
 		Rn -= 4;
-		WRITE(FRm, Rn);
+		MWRITE(FRm, Rn);
 	}
 
 	return 0;
@@ -445,11 +445,11 @@ id_sys(struct sh_fpu_soft_struct *fregs, struct pt_regs *regs, u16 code)
 	case 0x4052:
 	case 0x4062:
 		Rn -= 4;
-		WRITE(*reg, Rn);
+		MWRITE(*reg, Rn);
 		break;
 	case 0x4056:
 	case 0x4066:
-		READ(*reg, Rn);
+		MREAD(*reg, Rn);
 		Rn += 4;
 		break;
 	default:
@@ -468,109 +468,6 @@ static int fpu_emulate(u16 code, struct sh_fpu_soft_struct *fregs, struct pt_reg
 }
 
 /**
- *	denormal_to_double - Given denormalized float number,
- *	                     store double float
- *
- *	@fpu: Pointer to sh_fpu_soft structure
- *	@n: Index to FP register
- */
-static void denormal_to_double(struct sh_fpu_soft_struct *fpu, int n)
-{
-	unsigned long du, dl;
-	unsigned long x = fpu->fpul;
-	int exp = 1023 - 126;
-
-	if (x != 0 && (x & 0x7f800000) == 0) {
-		du = (x & 0x80000000);
-		while ((x & 0x00800000) == 0) {
-			x <<= 1;
-			exp--;
-		}
-		x &= 0x007fffff;
-		du |= (exp << 20) | (x >> 3);
-		dl = x << 29;
-
-		fpu->fp_regs[n] = du;
-		fpu->fp_regs[n+1] = dl;
-	}
-}
-
-/**
- *	ieee_fpe_handler - Handle denormalized number exception
- *
- *	@regs: Pointer to register structure
- *
- *	Returns 1 when it's handled (should not cause exception).
- */
-static int ieee_fpe_handler(struct pt_regs *regs)
-{
-	unsigned short insn = *(unsigned short *)regs->pc;
-	unsigned short finsn;
-	unsigned long nextpc;
-	int nib[4] = {
-		(insn >> 12) & 0xf,
-		(insn >> 8) & 0xf,
-		(insn >> 4) & 0xf,
-		insn & 0xf};
-
-	if (nib[0] == 0xb ||
-	    (nib[0] == 0x4 && nib[2] == 0x0 && nib[3] == 0xb)) /* bsr & jsr */
-		regs->pr = regs->pc + 4;
-
-	if (nib[0] == 0xa || nib[0] == 0xb) { /* bra & bsr */
-		nextpc = regs->pc + 4 + ((short) ((insn & 0xfff) << 4) >> 3);
-		finsn = *(unsigned short *) (regs->pc + 2);
-	} else if (nib[0] == 0x8 && nib[1] == 0xd) { /* bt/s */
-		if (regs->sr & 1)
-			nextpc = regs->pc + 4 + ((char) (insn & 0xff) << 1);
-		else
-			nextpc = regs->pc + 4;
-		finsn = *(unsigned short *) (regs->pc + 2);
-	} else if (nib[0] == 0x8 && nib[1] == 0xf) { /* bf/s */
-		if (regs->sr & 1)
-			nextpc = regs->pc + 4;
-		else
-			nextpc = regs->pc + 4 + ((char) (insn & 0xff) << 1);
-		finsn = *(unsigned short *) (regs->pc + 2);
-	} else if (nib[0] == 0x4 && nib[3] == 0xb &&
-		 (nib[2] == 0x0 || nib[2] == 0x2)) { /* jmp & jsr */
-		nextpc = regs->regs[nib[1]];
-		finsn = *(unsigned short *) (regs->pc + 2);
-	} else if (nib[0] == 0x0 && nib[3] == 0x3 &&
-		 (nib[2] == 0x0 || nib[2] == 0x2)) { /* braf & bsrf */
-		nextpc = regs->pc + 4 + regs->regs[nib[1]];
-		finsn = *(unsigned short *) (regs->pc + 2);
-	} else if (insn == 0x000b) { /* rts */
-		nextpc = regs->pr;
-		finsn = *(unsigned short *) (regs->pc + 2);
-	} else {
-		nextpc = regs->pc + 2;
-		finsn = insn;
-	}
-
-	if ((finsn & 0xf1ff) == 0xf0ad) { /* fcnvsd */
-		struct task_struct *tsk = current;
-
-		if ((tsk->thread.xstate->softfpu.fpscr & (1 << 17))) {
-			/* FPU error */
-			denormal_to_double (&tsk->thread.xstate->softfpu,
-					    (finsn >> 8) & 0xf);
-			tsk->thread.xstate->softfpu.fpscr &=
-				~(FPSCR_CAUSE_MASK | FPSCR_FLAG_MASK);
-			task_thread_info(tsk)->status |= TS_USEDFPU;
-		} else {
-			force_sig_fault(SIGFPE, FPE_FLTINV,
-					(void __user *)regs->pc);
-		}
-
-		regs->pc = nextpc;
-		return 1;
-	}
-
-	return 0;
-}
-
-/**
  * fpu_init - Initialize FPU registers
  * @fpu: Pointer to software emulated FPU registers.
  */
diff --git a/arch/sh/mm/nommu.c b/arch/sh/mm/nommu.c
index 8b45044..78c4b6e 100644
--- a/arch/sh/mm/nommu.c
+++ b/arch/sh/mm/nommu.c
@@ -28,9 +28,9 @@ __kernel_size_t __copy_user(void *to, const void *from, __kernel_size_t n)
 	return 0;
 }
 
-__kernel_size_t __clear_user(void *to, __kernel_size_t n)
+__kernel_size_t __clear_user(void __user *to, __kernel_size_t n)
 {
-	memset(to, 0, n);
+	memset((__force void *)to, 0, n);
 	return 0;
 }
 
diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
index 6039644..42cf01e 100644
--- a/arch/x86/events/intel/core.c
+++ b/arch/x86/events/intel/core.c
@@ -3048,8 +3048,10 @@ intel_vlbr_constraints(struct perf_event *event)
 {
 	struct event_constraint *c = &vlbr_constraint;
 
-	if (unlikely(constraint_match(c, event->hw.config)))
+	if (unlikely(constraint_match(c, event->hw.config))) {
+		event->hw.flags |= c->flags;
 		return c;
+	}
 
 	return NULL;
 }
diff --git a/arch/x86/events/intel/lbr.c b/arch/x86/events/intel/lbr.c
index 6b72e9b..8043213 100644
--- a/arch/x86/events/intel/lbr.c
+++ b/arch/x86/events/intel/lbr.c
@@ -265,6 +265,8 @@ void intel_pmu_lbr_reset(void)
 
 	cpuc->last_task_ctx = NULL;
 	cpuc->last_log_id = 0;
+	if (!static_cpu_has(X86_FEATURE_ARCH_LBR) && cpuc->lbr_select)
+		wrmsrl(MSR_LBR_SELECT, 0);
 }
 
 /*
diff --git a/arch/x86/include/asm/fpu/xcr.h b/arch/x86/include/asm/fpu/xcr.h
index 79f95d3..9656a5b 100644
--- a/arch/x86/include/asm/fpu/xcr.h
+++ b/arch/x86/include/asm/fpu/xcr.h
@@ -3,6 +3,7 @@
 #define _ASM_X86_FPU_XCR_H
 
 #define XCR_XFEATURE_ENABLED_MASK	0x00000000
+#define XCR_XFEATURE_IN_USE_MASK	0x00000001
 
 static inline u64 xgetbv(u32 index)
 {
@@ -20,4 +21,15 @@ static inline void xsetbv(u32 index, u64 value)
 	asm volatile("xsetbv" :: "a" (eax), "d" (edx), "c" (index));
 }
 
+/*
+ * Return a mask of xfeatures which are currently being tracked
+ * by the processor as being in the initial configuration.
+ *
+ * Callers should check X86_FEATURE_XGETBV1.
+ */
+static inline u64 xfeatures_in_use(void)
+{
+	return xgetbv(XCR_XFEATURE_IN_USE_MASK);
+}
+
 #endif /* _ASM_X86_FPU_XCR_H */
diff --git a/arch/x86/include/asm/fpu/xstate.h b/arch/x86/include/asm/fpu/xstate.h
index 0f8b90a..cd3dd17 100644
--- a/arch/x86/include/asm/fpu/xstate.h
+++ b/arch/x86/include/asm/fpu/xstate.h
@@ -92,6 +92,13 @@
 #define XFEATURE_MASK_FPSTATE	(XFEATURE_MASK_USER_RESTORE | \
 				 XFEATURE_MASK_SUPERVISOR_SUPPORTED)
 
+/*
+ * Features in this mask have space allocated in the signal frame, but may not
+ * have that space initialized when the feature is in its init state.
+ */
+#define XFEATURE_MASK_SIGFRAME_INITOPT	(XFEATURE_MASK_XTILE | \
+					 XFEATURE_MASK_USER_DYNAMIC)
+
 extern u64 xstate_fx_sw_bytes[USER_XSTATE_FX_SW_WORDS];
 
 extern void __init update_regset_xstate_info(unsigned int size,
diff --git a/arch/x86/include/asm/intel-family.h b/arch/x86/include/asm/intel-family.h
index 2715843..5a0bcf8 100644
--- a/arch/x86/include/asm/intel-family.h
+++ b/arch/x86/include/asm/intel-family.h
@@ -108,6 +108,8 @@
 #define INTEL_FAM6_ALDERLAKE		0x97	/* Golden Cove / Gracemont */
 #define INTEL_FAM6_ALDERLAKE_L		0x9A	/* Golden Cove / Gracemont */
 
+#define INTEL_FAM6_RAPTOR_LAKE		0xB7
+
 /* "Small Core" Processors (Atom) */
 
 #define INTEL_FAM6_ATOM_BONNELL		0x1C /* Diamondville, Pineview */
diff --git a/arch/x86/include/asm/smp.h b/arch/x86/include/asm/smp.h
index 08b0e90..81a0211 100644
--- a/arch/x86/include/asm/smp.h
+++ b/arch/x86/include/asm/smp.h
@@ -126,6 +126,7 @@ static inline void arch_send_call_function_ipi_mask(const struct cpumask *mask)
 
 void cpu_disable_common(void);
 void native_smp_prepare_boot_cpu(void);
+void smp_prepare_cpus_common(void);
 void native_smp_prepare_cpus(unsigned int max_cpus);
 void calculate_max_logical_packages(void);
 void native_smp_cpus_done(unsigned int max_cpus);
diff --git a/arch/x86/include/asm/static_call.h b/arch/x86/include/asm/static_call.h
index cbb67b6..39ebe05 100644
--- a/arch/x86/include/asm/static_call.h
+++ b/arch/x86/include/asm/static_call.h
@@ -27,6 +27,7 @@
 	    ".globl " STATIC_CALL_TRAMP_STR(name) "		\n"	\
 	    STATIC_CALL_TRAMP_STR(name) ":			\n"	\
 	    insns "						\n"	\
+	    ".byte 0x53, 0x43, 0x54				\n"	\
 	    ".type " STATIC_CALL_TRAMP_STR(name) ", @function	\n"	\
 	    ".size " STATIC_CALL_TRAMP_STR(name) ", . - " STATIC_CALL_TRAMP_STR(name) " \n" \
 	    ".popsection					\n")
diff --git a/arch/x86/kernel/cpu/cpuid-deps.c b/arch/x86/kernel/cpu/cpuid-deps.c
index cb2fdd13..c881bca 100644
--- a/arch/x86/kernel/cpu/cpuid-deps.c
+++ b/arch/x86/kernel/cpu/cpuid-deps.c
@@ -76,6 +76,7 @@ static const struct cpuid_dep cpuid_deps[] = {
 	{ X86_FEATURE_SGX1,			X86_FEATURE_SGX       },
 	{ X86_FEATURE_SGX2,			X86_FEATURE_SGX1      },
 	{ X86_FEATURE_XFD,			X86_FEATURE_XSAVES    },
+	{ X86_FEATURE_XFD,			X86_FEATURE_XGETBV1   },
 	{ X86_FEATURE_AMX_TILE,			X86_FEATURE_XFD       },
 	{}
 };
diff --git a/arch/x86/kernel/cpu/mce/intel.c b/arch/x86/kernel/cpu/mce/intel.c
index acfd5d9..bb9a46a 100644
--- a/arch/x86/kernel/cpu/mce/intel.c
+++ b/arch/x86/kernel/cpu/mce/intel.c
@@ -547,12 +547,13 @@ bool intel_filter_mce(struct mce *m)
 {
 	struct cpuinfo_x86 *c = &boot_cpu_data;
 
-	/* MCE errata HSD131, HSM142, HSW131, BDM48, and HSM142 */
+	/* MCE errata HSD131, HSM142, HSW131, BDM48, HSM142 and SKX37 */
 	if ((c->x86 == 6) &&
 	    ((c->x86_model == INTEL_FAM6_HASWELL) ||
 	     (c->x86_model == INTEL_FAM6_HASWELL_L) ||
 	     (c->x86_model == INTEL_FAM6_BROADWELL) ||
-	     (c->x86_model == INTEL_FAM6_HASWELL_G)) &&
+	     (c->x86_model == INTEL_FAM6_HASWELL_G) ||
+	     (c->x86_model == INTEL_FAM6_SKYLAKE_X)) &&
 	    (m->bank == 0) &&
 	    ((m->status & 0xa0000000ffffffff) == 0x80000000000f0005))
 		return true;
diff --git a/arch/x86/kernel/fpu/xstate.h b/arch/x86/kernel/fpu/xstate.h
index e18210d..86ea7c0 100644
--- a/arch/x86/kernel/fpu/xstate.h
+++ b/arch/x86/kernel/fpu/xstate.h
@@ -4,6 +4,7 @@
 
 #include <asm/cpufeature.h>
 #include <asm/fpu/xstate.h>
+#include <asm/fpu/xcr.h>
 
 #ifdef CONFIG_X86_64
 DECLARE_PER_CPU(u64, xfd_state);
@@ -199,6 +200,32 @@ static inline void os_xrstor_supervisor(struct fpstate *fpstate)
 }
 
 /*
+ * XSAVE itself always writes all requested xfeatures.  Removing features
+ * from the request bitmap reduces the features which are written.
+ * Generate a mask of features which must be written to a sigframe.  The
+ * unset features can be optimized away and not written.
+ *
+ * This optimization is user-visible.  Only use for states where
+ * uninitialized sigframe contents are tolerable, like dynamic features.
+ *
+ * Users of buffers produced with this optimization must check XSTATE_BV
+ * to determine which features have been optimized out.
+ */
+static inline u64 xfeatures_need_sigframe_write(void)
+{
+	u64 xfeaures_to_write;
+
+	/* In-use features must be written: */
+	xfeaures_to_write = xfeatures_in_use();
+
+	/* Also write all non-optimizable sigframe features: */
+	xfeaures_to_write |= XFEATURE_MASK_USER_SUPPORTED &
+			     ~XFEATURE_MASK_SIGFRAME_INITOPT;
+
+	return xfeaures_to_write;
+}
+
+/*
  * Save xstate to user space xsave area.
  *
  * We don't use modified optimization because xrstor/xrstors might track
@@ -220,10 +247,16 @@ static inline int xsave_to_user_sigframe(struct xregs_state __user *buf)
 	 */
 	struct fpstate *fpstate = current->thread.fpu.fpstate;
 	u64 mask = fpstate->user_xfeatures;
-	u32 lmask = mask;
-	u32 hmask = mask >> 32;
+	u32 lmask;
+	u32 hmask;
 	int err;
 
+	/* Optimize away writing unnecessary xfeatures: */
+	if (fpu_state_size_dynamic())
+		mask &= xfeatures_need_sigframe_write();
+
+	lmask = mask;
+	hmask = mask >> 32;
 	xfd_validate_state(fpstate, mask, false);
 
 	stac();
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 8241927..ac2909f 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -1350,12 +1350,7 @@ static void __init smp_get_logical_apicid(void)
 		cpu0_logical_apicid = GET_APIC_LOGICAL_ID(apic_read(APIC_LDR));
 }
 
-/*
- * Prepare for SMP bootup.
- * @max_cpus: configured maximum number of CPUs, It is a legacy parameter
- *            for common interface support.
- */
-void __init native_smp_prepare_cpus(unsigned int max_cpus)
+void __init smp_prepare_cpus_common(void)
 {
 	unsigned int i;
 
@@ -1386,6 +1381,17 @@ void __init native_smp_prepare_cpus(unsigned int max_cpus)
 	set_sched_topology(x86_topology);
 
 	set_cpu_sibling_map(0);
+}
+
+/*
+ * Prepare for SMP bootup.
+ * @max_cpus: configured maximum number of CPUs, It is a legacy parameter
+ *            for common interface support.
+ */
+void __init native_smp_prepare_cpus(unsigned int max_cpus)
+{
+	smp_prepare_cpus_common();
+
 	init_freq_invariance(false, false);
 	smp_sanity_check();
 
diff --git a/arch/x86/kernel/static_call.c b/arch/x86/kernel/static_call.c
index ea028e7..9c407a3 100644
--- a/arch/x86/kernel/static_call.c
+++ b/arch/x86/kernel/static_call.c
@@ -56,10 +56,15 @@ static void __ref __static_call_transform(void *insn, enum insn_type type, void
 	text_poke_bp(insn, code, size, emulate);
 }
 
-static void __static_call_validate(void *insn, bool tail)
+static void __static_call_validate(void *insn, bool tail, bool tramp)
 {
 	u8 opcode = *(u8 *)insn;
 
+	if (tramp && memcmp(insn+5, "SCT", 3)) {
+		pr_err("trampoline signature fail");
+		BUG();
+	}
+
 	if (tail) {
 		if (opcode == JMP32_INSN_OPCODE ||
 		    opcode == RET_INSN_OPCODE)
@@ -74,7 +79,8 @@ static void __static_call_validate(void *insn, bool tail)
 	/*
 	 * If we ever trigger this, our text is corrupt, we'll probably not live long.
 	 */
-	WARN_ONCE(1, "unexpected static_call insn opcode 0x%x at %pS\n", opcode, insn);
+	pr_err("unexpected static_call insn opcode 0x%x at %pS\n", opcode, insn);
+	BUG();
 }
 
 static inline enum insn_type __sc_insn(bool null, bool tail)
@@ -97,12 +103,12 @@ void arch_static_call_transform(void *site, void *tramp, void *func, bool tail)
 	mutex_lock(&text_mutex);
 
 	if (tramp) {
-		__static_call_validate(tramp, true);
+		__static_call_validate(tramp, true, true);
 		__static_call_transform(tramp, __sc_insn(!func, true), func);
 	}
 
 	if (IS_ENABLED(CONFIG_HAVE_STATIC_CALL_INLINE) && site) {
-		__static_call_validate(site, tail);
+		__static_call_validate(site, tail, false);
 		__static_call_transform(site, __sc_insn(!func, tail), func);
 	}
 
diff --git a/arch/x86/xen/smp_pv.c b/arch/x86/xen/smp_pv.c
index 9e55bcb..6a8f3b5 100644
--- a/arch/x86/xen/smp_pv.c
+++ b/arch/x86/xen/smp_pv.c
@@ -225,7 +225,6 @@ static void __init xen_pv_smp_prepare_boot_cpu(void)
 static void __init xen_pv_smp_prepare_cpus(unsigned int max_cpus)
 {
 	unsigned cpu;
-	unsigned int i;
 
 	if (skip_ioapic_setup) {
 		char *m = (max_cpus == 0) ?
@@ -238,16 +237,9 @@ static void __init xen_pv_smp_prepare_cpus(unsigned int max_cpus)
 	}
 	xen_init_lock_cpu(0);
 
-	smp_store_boot_cpu_info();
-	cpu_data(0).x86_max_cores = 1;
+	smp_prepare_cpus_common();
 
-	for_each_possible_cpu(i) {
-		zalloc_cpumask_var(&per_cpu(cpu_sibling_map, i), GFP_KERNEL);
-		zalloc_cpumask_var(&per_cpu(cpu_core_map, i), GFP_KERNEL);
-		zalloc_cpumask_var(&per_cpu(cpu_die_map, i), GFP_KERNEL);
-		zalloc_cpumask_var(&per_cpu(cpu_llc_shared_map, i), GFP_KERNEL);
-	}
-	set_cpu_sibling_map(0);
+	cpu_data(0).x86_max_cores = 1;
 
 	speculative_store_bypass_ht_init();
 
diff --git a/block/blk-core.c b/block/blk-core.c
index b043de2..9ee32f8 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -809,10 +809,8 @@ noinline_for_stack bool submit_bio_checks(struct bio *bio)
 	if (unlikely(!current->io_context))
 		create_task_io_context(current, GFP_ATOMIC, q->node);
 
-	if (blk_throtl_bio(bio)) {
-		blkcg_bio_issue_init(bio);
+	if (blk_throtl_bio(bio))
 		return false;
-	}
 
 	blk_cgroup_bio_start(bio);
 	blkcg_bio_issue_init(bio);
diff --git a/block/blk-ia-ranges.c b/block/blk-ia-ranges.c
index c246c42..b925f3d 100644
--- a/block/blk-ia-ranges.c
+++ b/block/blk-ia-ranges.c
@@ -104,8 +104,8 @@ static struct kobj_type blk_ia_ranges_ktype = {
 };
 
 /**
- * disk_register_ia_ranges - register with sysfs a set of independent
- *			    access ranges
+ * disk_register_independent_access_ranges - register with sysfs a set of
+ *		independent access ranges
  * @disk:	Target disk
  * @new_iars:	New set of independent access ranges
  *
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index 4be652f..ba21449 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -370,9 +370,6 @@ bool blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio,
 	bool ret = false;
 	enum hctx_type type;
 
-	if (bio_queue_enter(bio))
-		return false;
-
 	if (e && e->type->ops.bio_merge) {
 		ret = e->type->ops.bio_merge(q, bio, nr_segs);
 		goto out_put;
@@ -397,7 +394,6 @@ bool blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio,
 
 	spin_unlock(&ctx->lock);
 out_put:
-	blk_queue_exit(q);
 	return ret;
 }
 
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 629cf42..3ab34c4 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -2495,8 +2495,9 @@ static inline unsigned short blk_plug_max_rq_count(struct blk_plug *plug)
 	return BLK_MAX_REQUEST_COUNT;
 }
 
-static bool blk_attempt_bio_merge(struct request_queue *q, struct bio *bio,
-				  unsigned int nr_segs, bool *same_queue_rq)
+static bool blk_mq_attempt_bio_merge(struct request_queue *q,
+				     struct bio *bio, unsigned int nr_segs,
+				     bool *same_queue_rq)
 {
 	if (!blk_queue_nomerges(q) && bio_mergeable(bio)) {
 		if (blk_attempt_plug_merge(q, bio, nr_segs, same_queue_rq))
@@ -2520,12 +2521,8 @@ static struct request *blk_mq_get_new_requests(struct request_queue *q,
 	};
 	struct request *rq;
 
-	if (unlikely(bio_queue_enter(bio)))
+	if (blk_mq_attempt_bio_merge(q, bio, nsegs, same_queue_rq))
 		return NULL;
-	if (unlikely(!submit_bio_checks(bio)))
-		goto put_exit;
-	if (blk_attempt_bio_merge(q, bio, nsegs, same_queue_rq))
-		goto put_exit;
 
 	rq_qos_throttle(q, bio);
 
@@ -2542,26 +2539,44 @@ static struct request *blk_mq_get_new_requests(struct request_queue *q,
 	rq_qos_cleanup(q, bio);
 	if (bio->bi_opf & REQ_NOWAIT)
 		bio_wouldblock_error(bio);
-put_exit:
-	blk_queue_exit(q);
+
 	return NULL;
 }
 
+static inline bool blk_mq_can_use_cached_rq(struct request *rq,
+		struct bio *bio)
+{
+	if (blk_mq_get_hctx_type(bio->bi_opf) != rq->mq_hctx->type)
+		return false;
+
+	if (op_is_flush(rq->cmd_flags) != op_is_flush(bio->bi_opf))
+		return false;
+
+	return true;
+}
+
 static inline struct request *blk_mq_get_request(struct request_queue *q,
 						 struct blk_plug *plug,
 						 struct bio *bio,
 						 unsigned int nsegs,
 						 bool *same_queue_rq)
 {
+	struct request *rq;
+	bool checked = false;
+
 	if (plug) {
-		struct request *rq;
 
 		rq = rq_list_peek(&plug->cached_rq);
 		if (rq && rq->q == q) {
 			if (unlikely(!submit_bio_checks(bio)))
 				return NULL;
-			if (blk_attempt_bio_merge(q, bio, nsegs, same_queue_rq))
+			if (blk_mq_attempt_bio_merge(q, bio, nsegs,
+						same_queue_rq))
 				return NULL;
+			checked = true;
+			if (!blk_mq_can_use_cached_rq(rq, bio))
+				goto fallback;
+			rq->cmd_flags = bio->bi_opf;
 			plug->cached_rq = rq_list_next(rq);
 			INIT_LIST_HEAD(&rq->queuelist);
 			rq_qos_throttle(q, bio);
@@ -2569,7 +2584,15 @@ static inline struct request *blk_mq_get_request(struct request_queue *q,
 		}
 	}
 
-	return blk_mq_get_new_requests(q, plug, bio, nsegs, same_queue_rq);
+fallback:
+	if (unlikely(bio_queue_enter(bio)))
+		return NULL;
+	if (!checked && !submit_bio_checks(bio))
+		return NULL;
+	rq = blk_mq_get_new_requests(q, plug, bio, nsegs, same_queue_rq);
+	if (!rq)
+		blk_queue_exit(q);
+	return rq;
 }
 
 /**
diff --git a/block/blk-mq.h b/block/blk-mq.h
index cb0b548..8acfa65 100644
--- a/block/blk-mq.h
+++ b/block/blk-mq.h
@@ -89,15 +89,7 @@ static inline struct blk_mq_hw_ctx *blk_mq_map_queue_type(struct request_queue *
 	return q->queue_hw_ctx[q->tag_set->map[type].mq_map[cpu]];
 }
 
-/*
- * blk_mq_map_queue() - map (cmd_flags,type) to hardware queue
- * @q: request queue
- * @flags: request command flags
- * @ctx: software queue cpu ctx
- */
-static inline struct blk_mq_hw_ctx *blk_mq_map_queue(struct request_queue *q,
-						     unsigned int flags,
-						     struct blk_mq_ctx *ctx)
+static inline enum hctx_type blk_mq_get_hctx_type(unsigned int flags)
 {
 	enum hctx_type type = HCTX_TYPE_DEFAULT;
 
@@ -108,8 +100,20 @@ static inline struct blk_mq_hw_ctx *blk_mq_map_queue(struct request_queue *q,
 		type = HCTX_TYPE_POLL;
 	else if ((flags & REQ_OP_MASK) == REQ_OP_READ)
 		type = HCTX_TYPE_READ;
-	
-	return ctx->hctxs[type];
+	return type;
+}
+
+/*
+ * blk_mq_map_queue() - map (cmd_flags,type) to hardware queue
+ * @q: request queue
+ * @flags: request command flags
+ * @ctx: software queue cpu ctx
+ */
+static inline struct blk_mq_hw_ctx *blk_mq_map_queue(struct request_queue *q,
+						     unsigned int flags,
+						     struct blk_mq_ctx *ctx)
+{
+	return ctx->hctxs[blk_mq_get_hctx_type(flags)];
 }
 
 /*
@@ -149,7 +153,7 @@ struct blk_mq_alloc_data {
 	blk_mq_req_flags_t flags;
 	unsigned int shallow_depth;
 	unsigned int cmd_flags;
-	unsigned int rq_flags;
+	req_flags_t rq_flags;
 
 	/* allocate multiple requests/tags in one go */
 	unsigned int nr_tags;
diff --git a/block/blk-zoned.c b/block/blk-zoned.c
index 1d0c76c..774ecc5 100644
--- a/block/blk-zoned.c
+++ b/block/blk-zoned.c
@@ -429,9 +429,10 @@ int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
 		op = REQ_OP_ZONE_RESET;
 
 		/* Invalidate the page cache, including dirty pages. */
+		filemap_invalidate_lock(bdev->bd_inode->i_mapping);
 		ret = blkdev_truncate_zone_range(bdev, mode, &zrange);
 		if (ret)
-			return ret;
+			goto fail;
 		break;
 	case BLKOPENZONE:
 		op = REQ_OP_ZONE_OPEN;
@@ -449,15 +450,9 @@ int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
 	ret = blkdev_zone_mgmt(bdev, op, zrange.sector, zrange.nr_sectors,
 			       GFP_KERNEL);
 
-	/*
-	 * Invalidate the page cache again for zone reset: writes can only be
-	 * direct for zoned devices so concurrent writes would not add any page
-	 * to the page cache after/during reset. The page cache may be filled
-	 * again due to concurrent reads though and dropping the pages for
-	 * these is fine.
-	 */
-	if (!ret && cmd == BLKRESETZONE)
-		ret = blkdev_truncate_zone_range(bdev, mode, &zrange);
+fail:
+	if (cmd == BLKRESETZONE)
+		filemap_invalidate_unlock(bdev->bd_inode->i_mapping);
 
 	return ret;
 }
diff --git a/block/genhd.c b/block/genhd.c
index ca2fbab..c5392cc 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -394,8 +394,8 @@ static void disk_scan_partitions(struct gendisk *disk)
  * This function registers the partitioning information in @disk
  * with the kernel.
  */
-int device_add_disk(struct device *parent, struct gendisk *disk,
-		     const struct attribute_group **groups)
+int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
+				 const struct attribute_group **groups)
 
 {
 	struct device *ddev = disk_to_dev(disk);
@@ -544,7 +544,7 @@ int device_add_disk(struct device *parent, struct gendisk *disk,
 out_free_ext_minor:
 	if (disk->major == BLOCK_EXT_MAJOR)
 		blk_free_ext_minor(disk->first_minor);
-	return WARN_ON_ONCE(ret); /* keep until all callers handle errors */
+	return ret;
 }
 EXPORT_SYMBOL(device_add_disk);
 
diff --git a/block/ioctl.c b/block/ioctl.c
index d6af0ac..0a1d10a 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -113,6 +113,7 @@ static int blk_ioctl_discard(struct block_device *bdev, fmode_t mode,
 	uint64_t range[2];
 	uint64_t start, len;
 	struct request_queue *q = bdev_get_queue(bdev);
+	struct inode *inode = bdev->bd_inode;
 	int err;
 
 	if (!(mode & FMODE_WRITE))
@@ -135,12 +136,17 @@ static int blk_ioctl_discard(struct block_device *bdev, fmode_t mode,
 	if (start + len > bdev_nr_bytes(bdev))
 		return -EINVAL;
 
+	filemap_invalidate_lock(inode->i_mapping);
 	err = truncate_bdev_range(bdev, mode, start, start + len - 1);
 	if (err)
-		return err;
+		goto fail;
 
-	return blkdev_issue_discard(bdev, start >> 9, len >> 9,
-				    GFP_KERNEL, flags);
+	err = blkdev_issue_discard(bdev, start >> 9, len >> 9,
+				   GFP_KERNEL, flags);
+
+fail:
+	filemap_invalidate_unlock(inode->i_mapping);
+	return err;
 }
 
 static int blk_ioctl_zeroout(struct block_device *bdev, fmode_t mode,
@@ -148,6 +154,7 @@ static int blk_ioctl_zeroout(struct block_device *bdev, fmode_t mode,
 {
 	uint64_t range[2];
 	uint64_t start, end, len;
+	struct inode *inode = bdev->bd_inode;
 	int err;
 
 	if (!(mode & FMODE_WRITE))
@@ -170,12 +177,17 @@ static int blk_ioctl_zeroout(struct block_device *bdev, fmode_t mode,
 		return -EINVAL;
 
 	/* Invalidate the page cache, including dirty pages */
+	filemap_invalidate_lock(inode->i_mapping);
 	err = truncate_bdev_range(bdev, mode, start, end);
 	if (err)
-		return err;
+		goto fail;
 
-	return blkdev_issue_zeroout(bdev, start >> 9, len >> 9, GFP_KERNEL,
-			BLKDEV_ZERO_NOUNMAP);
+	err = blkdev_issue_zeroout(bdev, start >> 9, len >> 9, GFP_KERNEL,
+				   BLKDEV_ZERO_NOUNMAP);
+
+fail:
+	filemap_invalidate_unlock(inode->i_mapping);
+	return err;
 }
 
 static int put_ushort(unsigned short __user *argp, unsigned short val)
diff --git a/crypto/zstd.c b/crypto/zstd.c
index 1a3309f..154a969 100644
--- a/crypto/zstd.c
+++ b/crypto/zstd.c
@@ -18,22 +18,22 @@
 #define ZSTD_DEF_LEVEL	3
 
 struct zstd_ctx {
-	ZSTD_CCtx *cctx;
-	ZSTD_DCtx *dctx;
+	zstd_cctx *cctx;
+	zstd_dctx *dctx;
 	void *cwksp;
 	void *dwksp;
 };
 
-static ZSTD_parameters zstd_params(void)
+static zstd_parameters zstd_params(void)
 {
-	return ZSTD_getParams(ZSTD_DEF_LEVEL, 0, 0);
+	return zstd_get_params(ZSTD_DEF_LEVEL, 0);
 }
 
 static int zstd_comp_init(struct zstd_ctx *ctx)
 {
 	int ret = 0;
-	const ZSTD_parameters params = zstd_params();
-	const size_t wksp_size = ZSTD_CCtxWorkspaceBound(params.cParams);
+	const zstd_parameters params = zstd_params();
+	const size_t wksp_size = zstd_cctx_workspace_bound(&params.cParams);
 
 	ctx->cwksp = vzalloc(wksp_size);
 	if (!ctx->cwksp) {
@@ -41,7 +41,7 @@ static int zstd_comp_init(struct zstd_ctx *ctx)
 		goto out;
 	}
 
-	ctx->cctx = ZSTD_initCCtx(ctx->cwksp, wksp_size);
+	ctx->cctx = zstd_init_cctx(ctx->cwksp, wksp_size);
 	if (!ctx->cctx) {
 		ret = -EINVAL;
 		goto out_free;
@@ -56,7 +56,7 @@ static int zstd_comp_init(struct zstd_ctx *ctx)
 static int zstd_decomp_init(struct zstd_ctx *ctx)
 {
 	int ret = 0;
-	const size_t wksp_size = ZSTD_DCtxWorkspaceBound();
+	const size_t wksp_size = zstd_dctx_workspace_bound();
 
 	ctx->dwksp = vzalloc(wksp_size);
 	if (!ctx->dwksp) {
@@ -64,7 +64,7 @@ static int zstd_decomp_init(struct zstd_ctx *ctx)
 		goto out;
 	}
 
-	ctx->dctx = ZSTD_initDCtx(ctx->dwksp, wksp_size);
+	ctx->dctx = zstd_init_dctx(ctx->dwksp, wksp_size);
 	if (!ctx->dctx) {
 		ret = -EINVAL;
 		goto out_free;
@@ -152,10 +152,10 @@ static int __zstd_compress(const u8 *src, unsigned int slen,
 {
 	size_t out_len;
 	struct zstd_ctx *zctx = ctx;
-	const ZSTD_parameters params = zstd_params();
+	const zstd_parameters params = zstd_params();
 
-	out_len = ZSTD_compressCCtx(zctx->cctx, dst, *dlen, src, slen, params);
-	if (ZSTD_isError(out_len))
+	out_len = zstd_compress_cctx(zctx->cctx, dst, *dlen, src, slen, &params);
+	if (zstd_is_error(out_len))
 		return -EINVAL;
 	*dlen = out_len;
 	return 0;
@@ -182,8 +182,8 @@ static int __zstd_decompress(const u8 *src, unsigned int slen,
 	size_t out_len;
 	struct zstd_ctx *zctx = ctx;
 
-	out_len = ZSTD_decompressDCtx(zctx->dctx, dst, *dlen, src, slen);
-	if (ZSTD_isError(out_len))
+	out_len = zstd_decompress_dctx(zctx->dctx, dst, *dlen, src, slen);
+	if (zstd_is_error(out_len))
 		return -EINVAL;
 	*dlen = out_len;
 	return 0;
diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
index 981e72a..ff16a36 100644
--- a/drivers/base/arch_topology.c
+++ b/drivers/base/arch_topology.c
@@ -677,6 +677,8 @@ void remove_cpu_topology(unsigned int cpu)
 		cpumask_clear_cpu(cpu, topology_core_cpumask(sibling));
 	for_each_cpu(sibling, topology_sibling_cpumask(cpu))
 		cpumask_clear_cpu(cpu, topology_sibling_cpumask(sibling));
+	for_each_cpu(sibling, topology_cluster_cpumask(cpu))
+		cpumask_clear_cpu(cpu, topology_cluster_cpumask(sibling));
 	for_each_cpu(sibling, topology_llc_cpumask(cpu))
 		cpumask_clear_cpu(cpu, topology_llc_cpumask(sibling));
 
diff --git a/drivers/clk/actions/owl-factor.c b/drivers/clk/actions/owl-factor.c
index f15e262..64f316c 100644
--- a/drivers/clk/actions/owl-factor.c
+++ b/drivers/clk/actions/owl-factor.c
@@ -10,7 +10,6 @@
 
 #include <linux/clk-provider.h>
 #include <linux/regmap.h>
-#include <linux/slab.h>
 
 #include "owl-factor.h"
 
diff --git a/drivers/clk/clk-ast2600.c b/drivers/clk/clk-ast2600.c
index bc3be5f..24dab23 100644
--- a/drivers/clk/clk-ast2600.c
+++ b/drivers/clk/clk-ast2600.c
@@ -51,6 +51,8 @@ static DEFINE_SPINLOCK(aspeed_g6_clk_lock);
 static struct clk_hw_onecell_data *aspeed_g6_clk_data;
 
 static void __iomem *scu_g6_base;
+/* AST2600 revision: A0, A1, A2, etc */
+static u8 soc_rev;
 
 /*
  * Clocks marked with CLK_IS_CRITICAL:
@@ -191,9 +193,8 @@ static struct clk_hw *ast2600_calc_pll(const char *name, u32 val)
 static struct clk_hw *ast2600_calc_apll(const char *name, u32 val)
 {
 	unsigned int mult, div;
-	u32 chip_id = readl(scu_g6_base + ASPEED_G6_SILICON_REV);
 
-	if (((chip_id & CHIP_REVISION_ID) >> 16) >= 2) {
+	if (soc_rev >= 2) {
 		if (val & BIT(24)) {
 			/* Pass through mode */
 			mult = div = 1;
@@ -707,7 +708,7 @@ static const u32 ast2600_a1_axi_ahb200_tbl[] = {
 static void __init aspeed_g6_cc(struct regmap *map)
 {
 	struct clk_hw *hw;
-	u32 val, div, divbits, chip_id, axi_div, ahb_div;
+	u32 val, div, divbits, axi_div, ahb_div;
 
 	clk_hw_register_fixed_rate(NULL, "clkin", NULL, 0, 25000000);
 
@@ -738,8 +739,7 @@ static void __init aspeed_g6_cc(struct regmap *map)
 		axi_div = 2;
 
 	divbits = (val >> 11) & 0x3;
-	regmap_read(map, ASPEED_G6_SILICON_REV, &chip_id);
-	if (chip_id & BIT(16)) {
+	if (soc_rev >= 1) {
 		if (!divbits) {
 			ahb_div = ast2600_a1_axi_ahb200_tbl[(val >> 8) & 0x3];
 			if (val & BIT(16))
@@ -784,6 +784,8 @@ static void __init aspeed_g6_cc_init(struct device_node *np)
 	if (!scu_g6_base)
 		return;
 
+	soc_rev = (readl(scu_g6_base + ASPEED_G6_SILICON_REV) & CHIP_REVISION_ID) >> 16;
+
 	aspeed_g6_clk_data = kzalloc(struct_size(aspeed_g6_clk_data, hws,
 				      ASPEED_G6_NUM_CLKS), GFP_KERNEL);
 	if (!aspeed_g6_clk_data)
diff --git a/drivers/clk/clk-composite.c b/drivers/clk/clk-composite.c
index c04ae0e..b9c5f90 100644
--- a/drivers/clk/clk-composite.c
+++ b/drivers/clk/clk-composite.c
@@ -97,6 +97,7 @@ static int clk_composite_determine_rate(struct clk_hw *hw,
 				return ret;
 
 			req->rate = tmp_req.rate;
+			req->best_parent_hw = tmp_req.best_parent_hw;
 			req->best_parent_rate = tmp_req.best_parent_rate;
 
 			return 0;
diff --git a/drivers/clk/clk-si5351.c b/drivers/clk/clk-si5351.c
index 57e4597..93fa8c9 100644
--- a/drivers/clk/clk-si5351.c
+++ b/drivers/clk/clk-si5351.c
@@ -1,15 +1,15 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /*
- * clk-si5351.c: Silicon Laboratories Si5351A/B/C I2C Clock Generator
+ * clk-si5351.c: Skyworks / Silicon Labs Si5351A/B/C I2C Clock Generator
  *
  * Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
  * Rabeeh Khoury <rabeeh@solid-run.com>
  *
  * References:
  * [1] "Si5351A/B/C Data Sheet"
- *     https://www.silabs.com/Support%20Documents/TechnicalDocs/Si5351.pdf
- * [2] "Manually Generating an Si5351 Register Map"
- *     https://www.silabs.com/Support%20Documents/TechnicalDocs/AN619.pdf
+ *     https://www.skyworksinc.com/-/media/Skyworks/SL/documents/public/data-sheets/Si5351-B.pdf
+ * [2] "AN619: Manually Generating an Si5351 Register Map"
+ *     https://www.skyworksinc.com/-/media/Skyworks/SL/documents/public/application-notes/AN619.pdf
  */
 
 #include <linux/module.h>
diff --git a/drivers/clk/clk-si5351.h b/drivers/clk/clk-si5351.h
index 73dc8ef..e9e2bfd 100644
--- a/drivers/clk/clk-si5351.h
+++ b/drivers/clk/clk-si5351.h
@@ -1,6 +1,6 @@
 /* SPDX-License-Identifier: GPL-2.0-or-later */
 /*
- * clk-si5351.h: Silicon Laboratories Si5351A/B/C I2C Clock Generator
+ * clk-si5351.h: Skyworks / Silicon Labs Si5351A/B/C I2C Clock Generator
  *
  * Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
  * Rabeeh Khoury <rabeeh@solid-run.com>
diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index c6d3b1a..e7be3e5 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -905,7 +905,7 @@ static int vc5_get_output_config(struct i2c_client *client,
 
 static const struct of_device_id clk_vc5_of_match[];
 
-static int vc5_probe(struct i2c_client *client, const struct i2c_device_id *id)
+static int vc5_probe(struct i2c_client *client)
 {
 	unsigned int oe, sd, src_mask = 0, src_val = 0;
 	struct vc5_driver_data *vc5;
@@ -1244,7 +1244,7 @@ static struct i2c_driver vc5_driver = {
 		.pm	= &vc5_pm_ops,
 		.of_match_table = clk_vc5_of_match,
 	},
-	.probe		= vc5_probe,
+	.probe_new	= vc5_probe,
 	.remove		= vc5_remove,
 	.id_table	= vc5_id,
 };
diff --git a/drivers/clk/imx/clk.h b/drivers/clk/imx/clk.h
index 8199499..7d220a0 100644
--- a/drivers/clk/imx/clk.h
+++ b/drivers/clk/imx/clk.h
@@ -391,11 +391,11 @@ struct clk_hw *__imx8m_clk_hw_composite(const char *name,
 
 #define imx8m_clk_hw_composite(name, parent_names, reg) \
 	_imx8m_clk_hw_composite(name, parent_names, reg, \
-			IMX_COMPOSITE_CORE, IMX_COMPOSITE_CLK_FLAGS_DEFAULT)
+			0, IMX_COMPOSITE_CLK_FLAGS_DEFAULT)
 
 #define imx8m_clk_hw_composite_critical(name, parent_names, reg) \
 	_imx8m_clk_hw_composite(name, parent_names, reg, \
-			IMX_COMPOSITE_CORE, IMX_COMPOSITE_CLK_FLAGS_CRITICAL)
+			0, IMX_COMPOSITE_CLK_FLAGS_CRITICAL)
 
 #define imx8m_clk_hw_composite_bus(name, parent_names, reg)	\
 	_imx8m_clk_hw_composite(name, parent_names, reg, \
diff --git a/drivers/clk/ingenic/cgu.c b/drivers/clk/ingenic/cgu.c
index 266c759..af31633 100644
--- a/drivers/clk/ingenic/cgu.c
+++ b/drivers/clk/ingenic/cgu.c
@@ -453,15 +453,15 @@ ingenic_clk_calc_div(struct clk_hw *hw,
 	}
 
 	/* Impose hardware constraints */
-	div = min_t(unsigned, div, 1 << clk_info->div.bits);
-	div = max_t(unsigned, div, 1);
+	div = clamp_t(unsigned int, div, clk_info->div.div,
+		      clk_info->div.div << clk_info->div.bits);
 
 	/*
 	 * If the divider value itself must be divided before being written to
 	 * the divider register, we must ensure we don't have any bits set that
 	 * would be lost as a result of doing so.
 	 */
-	div /= clk_info->div.div;
+	div = DIV_ROUND_UP(div, clk_info->div.div);
 	div *= clk_info->div.div;
 
 	return div;
diff --git a/drivers/clk/ingenic/jz4725b-cgu.c b/drivers/clk/ingenic/jz4725b-cgu.c
index 5154b0c..744d136 100644
--- a/drivers/clk/ingenic/jz4725b-cgu.c
+++ b/drivers/clk/ingenic/jz4725b-cgu.c
@@ -10,7 +10,7 @@
 #include <linux/delay.h>
 #include <linux/of.h>
 
-#include <dt-bindings/clock/jz4725b-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4725b-cgu.h>
 
 #include "cgu.h"
 #include "pm.h"
diff --git a/drivers/clk/ingenic/jz4740-cgu.c b/drivers/clk/ingenic/jz4740-cgu.c
index cd878f0..43ffb62 100644
--- a/drivers/clk/ingenic/jz4740-cgu.c
+++ b/drivers/clk/ingenic/jz4740-cgu.c
@@ -11,7 +11,7 @@
 #include <linux/io.h>
 #include <linux/of.h>
 
-#include <dt-bindings/clock/jz4740-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4740-cgu.h>
 
 #include "cgu.h"
 #include "pm.h"
diff --git a/drivers/clk/ingenic/jz4760-cgu.c b/drivers/clk/ingenic/jz4760-cgu.c
index 1448379..080d492 100644
--- a/drivers/clk/ingenic/jz4760-cgu.c
+++ b/drivers/clk/ingenic/jz4760-cgu.c
@@ -12,7 +12,7 @@
 
 #include <linux/clk.h>
 
-#include <dt-bindings/clock/jz4760-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4760-cgu.h>
 
 #include "cgu.h"
 #include "pm.h"
diff --git a/drivers/clk/ingenic/jz4770-cgu.c b/drivers/clk/ingenic/jz4770-cgu.c
index 2321742..8c6c120 100644
--- a/drivers/clk/ingenic/jz4770-cgu.c
+++ b/drivers/clk/ingenic/jz4770-cgu.c
@@ -10,7 +10,7 @@
 #include <linux/io.h>
 #include <linux/of.h>
 
-#include <dt-bindings/clock/jz4770-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4770-cgu.h>
 
 #include "cgu.h"
 #include "pm.h"
diff --git a/drivers/clk/ingenic/jz4780-cgu.c b/drivers/clk/ingenic/jz4780-cgu.c
index 0268d23..e357c22 100644
--- a/drivers/clk/ingenic/jz4780-cgu.c
+++ b/drivers/clk/ingenic/jz4780-cgu.c
@@ -13,7 +13,7 @@
 #include <linux/iopoll.h>
 #include <linux/of.h>
 
-#include <dt-bindings/clock/jz4780-cgu.h>
+#include <dt-bindings/clock/ingenic,jz4780-cgu.h>
 
 #include "cgu.h"
 #include "pm.h"
diff --git a/drivers/clk/ingenic/x1000-cgu.c b/drivers/clk/ingenic/x1000-cgu.c
index 9aa20b5..3c4d5a7 100644
--- a/drivers/clk/ingenic/x1000-cgu.c
+++ b/drivers/clk/ingenic/x1000-cgu.c
@@ -9,7 +9,7 @@
 #include <linux/io.h>
 #include <linux/of.h>
 
-#include <dt-bindings/clock/x1000-cgu.h>
+#include <dt-bindings/clock/ingenic,x1000-cgu.h>
 
 #include "cgu.h"
 #include "pm.h"
diff --git a/drivers/clk/ingenic/x1830-cgu.c b/drivers/clk/ingenic/x1830-cgu.c
index 950aee2..e01ec2d 100644
--- a/drivers/clk/ingenic/x1830-cgu.c
+++ b/drivers/clk/ingenic/x1830-cgu.c
@@ -9,7 +9,7 @@
 #include <linux/io.h>
 #include <linux/of.h>
 
-#include <dt-bindings/clock/x1830-cgu.h>
+#include <dt-bindings/clock/ingenic,x1830-cgu.h>
 
 #include "cgu.h"
 #include "pm.h"
diff --git a/drivers/clk/mediatek/clk-mt8195-imp_iic_wrap.c b/drivers/clk/mediatek/clk-mt8195-imp_iic_wrap.c
index 0e2ac0a..4ab312e 100644
--- a/drivers/clk/mediatek/clk-mt8195-imp_iic_wrap.c
+++ b/drivers/clk/mediatek/clk-mt8195-imp_iic_wrap.c
@@ -10,8 +10,6 @@
 #include <linux/clk-provider.h>
 #include <linux/platform_device.h>
 
-#include <dt-bindings/clock/mt8195-clk.h>
-
 static const struct mtk_gate_regs imp_iic_wrap_cg_regs = {
 	.set_ofs = 0xe08,
 	.clr_ofs = 0xe04,
diff --git a/drivers/clk/qcom/gcc-msm8996.c b/drivers/clk/qcom/gcc-msm8996.c
index 3c3a7ff..9b1674b 100644
--- a/drivers/clk/qcom/gcc-msm8996.c
+++ b/drivers/clk/qcom/gcc-msm8996.c
@@ -2937,20 +2937,6 @@ static struct clk_branch gcc_smmu_aggre0_ahb_clk = {
 	},
 };
 
-static struct clk_branch gcc_aggre1_pnoc_ahb_clk = {
-	.halt_reg = 0x82014,
-	.clkr = {
-		.enable_reg = 0x82014,
-		.enable_mask = BIT(0),
-		.hw.init = &(struct clk_init_data){
-			.name = "gcc_aggre1_pnoc_ahb_clk",
-			.parent_names = (const char *[]){ "periph_noc_clk_src" },
-			.num_parents = 1,
-			.ops = &clk_branch2_ops,
-		},
-	},
-};
-
 static struct clk_branch gcc_aggre2_ufs_axi_clk = {
 	.halt_reg = 0x83014,
 	.clkr = {
@@ -3474,7 +3460,6 @@ static struct clk_regmap *gcc_msm8996_clocks[] = {
 	[GCC_AGGRE0_CNOC_AHB_CLK] = &gcc_aggre0_cnoc_ahb_clk.clkr,
 	[GCC_SMMU_AGGRE0_AXI_CLK] = &gcc_smmu_aggre0_axi_clk.clkr,
 	[GCC_SMMU_AGGRE0_AHB_CLK] = &gcc_smmu_aggre0_ahb_clk.clkr,
-	[GCC_AGGRE1_PNOC_AHB_CLK] = &gcc_aggre1_pnoc_ahb_clk.clkr,
 	[GCC_AGGRE2_UFS_AXI_CLK] = &gcc_aggre2_ufs_axi_clk.clkr,
 	[GCC_AGGRE2_USB3_AXI_CLK] = &gcc_aggre2_usb3_axi_clk.clkr,
 	[GCC_QSPI_AHB_CLK] = &gcc_qspi_ahb_clk.clkr,
diff --git a/drivers/clk/rockchip/Kconfig b/drivers/clk/rockchip/Kconfig
index 2dfd6a3..3067bdb 100644
--- a/drivers/clk/rockchip/Kconfig
+++ b/drivers/clk/rockchip/Kconfig
@@ -80,14 +80,14 @@
 	  Build the driver for RK3368 Clock Driver.
 
 config CLK_RK3399
-	tristate "Rockchip RK3399 clock controller support"
+	bool "Rockchip RK3399 clock controller support"
 	depends on ARM64 || COMPILE_TEST
 	default y
 	help
 	  Build the driver for RK3399 Clock Driver.
 
 config CLK_RK3568
-	tristate "Rockchip RK3568 clock controller support"
+	bool "Rockchip RK3568 clock controller support"
 	depends on ARM64 || COMPILE_TEST
 	default y
 	help
diff --git a/drivers/clk/rockchip/clk-rk3399.c b/drivers/clk/rockchip/clk-rk3399.c
index 7924598..306910a 100644
--- a/drivers/clk/rockchip/clk-rk3399.c
+++ b/drivers/clk/rockchip/clk-rk3399.c
@@ -1630,7 +1630,6 @@ static const struct of_device_id clk_rk3399_match_table[] = {
 	},
 	{ }
 };
-MODULE_DEVICE_TABLE(of, clk_rk3399_match_table);
 
 static int __init clk_rk3399_probe(struct platform_device *pdev)
 {
@@ -1656,7 +1655,4 @@ static struct platform_driver clk_rk3399_driver = {
 		.suppress_bind_attrs = true,
 	},
 };
-module_platform_driver_probe(clk_rk3399_driver, clk_rk3399_probe);
-
-MODULE_DESCRIPTION("Rockchip RK3399 Clock Driver");
-MODULE_LICENSE("GPL");
+builtin_platform_driver_probe(clk_rk3399_driver, clk_rk3399_probe);
diff --git a/drivers/clk/rockchip/clk-rk3568.c b/drivers/clk/rockchip/clk-rk3568.c
index 939e707..69a9e80 100644
--- a/drivers/clk/rockchip/clk-rk3568.c
+++ b/drivers/clk/rockchip/clk-rk3568.c
@@ -1693,7 +1693,6 @@ static const struct of_device_id clk_rk3568_match_table[] = {
 	},
 	{ }
 };
-MODULE_DEVICE_TABLE(of, clk_rk3568_match_table);
 
 static int __init clk_rk3568_probe(struct platform_device *pdev)
 {
@@ -1719,7 +1718,4 @@ static struct platform_driver clk_rk3568_driver = {
 		.suppress_bind_attrs = true,
 	},
 };
-module_platform_driver_probe(clk_rk3568_driver, clk_rk3568_probe);
-
-MODULE_DESCRIPTION("Rockchip RK3568 Clock Driver");
-MODULE_LICENSE("GPL");
+builtin_platform_driver_probe(clk_rk3568_driver, clk_rk3568_probe);
diff --git a/drivers/clk/uniphier/clk-uniphier-core.c b/drivers/clk/uniphier/clk-uniphier-core.c
index 1238023..46c66fa 100644
--- a/drivers/clk/uniphier/clk-uniphier-core.c
+++ b/drivers/clk/uniphier/clk-uniphier-core.c
@@ -132,6 +132,10 @@ static const struct of_device_id uniphier_clk_match[] = {
 		.compatible = "socionext,uniphier-pxs3-clock",
 		.data = uniphier_pxs3_sys_clk_data,
 	},
+	{
+		.compatible = "socionext,uniphier-nx1-clock",
+		.data = uniphier_nx1_sys_clk_data,
+	},
 	/* Media I/O clock, SD clock */
 	{
 		.compatible = "socionext,uniphier-ld4-mio-clock",
@@ -165,6 +169,10 @@ static const struct of_device_id uniphier_clk_match[] = {
 		.compatible = "socionext,uniphier-pxs3-sd-clock",
 		.data = uniphier_pro5_sd_clk_data,
 	},
+	{
+		.compatible = "socionext,uniphier-nx1-sd-clock",
+		.data = uniphier_pro5_sd_clk_data,
+	},
 	/* Peripheral clock */
 	{
 		.compatible = "socionext,uniphier-ld4-peri-clock",
@@ -198,6 +206,15 @@ static const struct of_device_id uniphier_clk_match[] = {
 		.compatible = "socionext,uniphier-pxs3-peri-clock",
 		.data = uniphier_pro4_peri_clk_data,
 	},
+	{
+		.compatible = "socionext,uniphier-nx1-peri-clock",
+		.data = uniphier_pro4_peri_clk_data,
+	},
+	/* SoC-glue clock */
+	{
+		.compatible = "socionext,uniphier-pro4-sg-clock",
+		.data = uniphier_pro4_sg_clk_data,
+	},
 	{ /* sentinel */ }
 };
 
diff --git a/drivers/clk/uniphier/clk-uniphier-sys.c b/drivers/clk/uniphier/clk-uniphier-sys.c
index 32b3017..0180470 100644
--- a/drivers/clk/uniphier/clk-uniphier-sys.c
+++ b/drivers/clk/uniphier/clk-uniphier-sys.c
@@ -20,6 +20,10 @@
 	UNIPHIER_CLK_FACTOR("sd-200m", -1, "spll", 1, 10),		\
 	UNIPHIER_CLK_FACTOR("sd-133m", -1, "spll", 1, 15)
 
+#define UNIPHIER_NX1_SYS_CLK_SD						\
+	UNIPHIER_CLK_FACTOR("sd-200m", -1, "spll", 1, 4),		\
+	UNIPHIER_CLK_FACTOR("sd-133m", -1, "spll", 1, 6)
+
 #define UNIPHIER_LD4_SYS_CLK_NAND(idx)					\
 	UNIPHIER_CLK_FACTOR("nand-50m", -1, "spll", 1, 32),		\
 	UNIPHIER_CLK_GATE("nand", (idx), "nand-50m", 0x2104, 2)
@@ -288,6 +292,8 @@ const struct uniphier_clk_data uniphier_pxs3_sys_clk_data[] = {
 	UNIPHIER_CLK_GATE("sata0", 28, NULL, 0x210c, 7),
 	UNIPHIER_CLK_GATE("sata1", 29, NULL, 0x210c, 8),
 	UNIPHIER_CLK_GATE("sata-phy", 30, NULL, 0x210c, 21),
+	UNIPHIER_LD11_SYS_CLK_AIO(40),
+	UNIPHIER_LD11_SYS_CLK_EXIV(42),
 	/* CPU gears */
 	UNIPHIER_CLK_DIV4("cpll", 2, 3, 4, 8),
 	UNIPHIER_CLK_DIV4("spll", 2, 3, 4, 8),
@@ -300,3 +306,44 @@ const struct uniphier_clk_data uniphier_pxs3_sys_clk_data[] = {
 			     "spll/4", "spll/8", "s2pll/4", "s2pll/8"),
 	{ /* sentinel */ }
 };
+
+const struct uniphier_clk_data uniphier_nx1_sys_clk_data[] = {
+	UNIPHIER_CLK_FACTOR("cpll", -1, "ref", 100, 1),		/* ARM: 2500 MHz */
+	UNIPHIER_CLK_FACTOR("spll", -1, "ref", 32, 1),		/* 800 MHz */
+	UNIPHIER_CLK_FACTOR("uart", 0, "spll", 1, 6),
+	UNIPHIER_CLK_FACTOR("i2c", 1, "spll", 1, 16),
+	UNIPHIER_NX1_SYS_CLK_SD,
+	UNIPHIER_CLK_GATE("emmc", 4, NULL, 0x2108, 8),
+	UNIPHIER_CLK_GATE("ether", 6, NULL, 0x210c, 0),
+	UNIPHIER_CLK_GATE("usb30-0", 12, NULL, 0x210c, 16),	/* =GIO */
+	UNIPHIER_CLK_GATE("usb30-1", 13, NULL, 0x210c, 20),	/* =GIO1P */
+	UNIPHIER_CLK_GATE("usb30-hsphy0", 16, NULL, 0x210c, 24),
+	UNIPHIER_CLK_GATE("usb30-ssphy0", 17, NULL, 0x210c, 25),
+	UNIPHIER_CLK_GATE("usb30-ssphy1", 18, NULL, 0x210c, 26),
+	UNIPHIER_CLK_GATE("pcie", 24, NULL, 0x210c, 8),
+	UNIPHIER_CLK_GATE("voc", 52, NULL, 0x2110, 0),
+	UNIPHIER_CLK_GATE("hdmitx", 58, NULL, 0x2110, 8),
+	/* CPU gears */
+	UNIPHIER_CLK_DIV5("cpll", 2, 4, 8, 16, 32),
+	UNIPHIER_CLK_CPUGEAR("cpu-ca53", 33, 0x8080, 0xf, 5,
+			     "cpll/2", "cpll/4", "cpll/8", "cpll/16",
+			     "cpll/32"),
+	{ /* sentinel */ }
+};
+
+const struct uniphier_clk_data uniphier_pro4_sg_clk_data[] = {
+	UNIPHIER_CLK_DIV("gpll", 4),
+	{
+		.name = "sata-ref",
+		.type = UNIPHIER_CLK_TYPE_MUX,
+		.idx = 0,
+		.data.mux = {
+			.parent_names = { "gpll/4", "ref", },
+			.num_parents = 2,
+			.reg = 0x1a28,
+			.masks = { 0x1, 0x1, },
+			.vals  = { 0x0, 0x1, },
+		},
+	},
+	{ /* sentinel */ }
+};
diff --git a/drivers/clk/uniphier/clk-uniphier.h b/drivers/clk/uniphier/clk-uniphier.h
index 9e30362..dea0c78 100644
--- a/drivers/clk/uniphier/clk-uniphier.h
+++ b/drivers/clk/uniphier/clk-uniphier.h
@@ -119,6 +119,10 @@ struct uniphier_clk_data {
 	UNIPHIER_CLK_DIV2(parent, div0, div1),			\
 	UNIPHIER_CLK_DIV2(parent, div2, div3)
 
+#define UNIPHIER_CLK_DIV5(parent, div0, div1, div2, div3, div4)	\
+	UNIPHIER_CLK_DIV4(parent, div0, div1, div2, div3),	\
+	UNIPHIER_CLK_DIV(parent, div4)
+
 struct clk_hw *uniphier_clk_register_cpugear(struct device *dev,
 					     struct regmap *regmap,
 					     const char *name,
@@ -146,9 +150,11 @@ extern const struct uniphier_clk_data uniphier_pxs2_sys_clk_data[];
 extern const struct uniphier_clk_data uniphier_ld11_sys_clk_data[];
 extern const struct uniphier_clk_data uniphier_ld20_sys_clk_data[];
 extern const struct uniphier_clk_data uniphier_pxs3_sys_clk_data[];
+extern const struct uniphier_clk_data uniphier_nx1_sys_clk_data[];
 extern const struct uniphier_clk_data uniphier_ld4_mio_clk_data[];
 extern const struct uniphier_clk_data uniphier_pro5_sd_clk_data[];
 extern const struct uniphier_clk_data uniphier_ld4_peri_clk_data[];
 extern const struct uniphier_clk_data uniphier_pro4_peri_clk_data[];
+extern const struct uniphier_clk_data uniphier_pro4_sg_clk_data[];
 
 #endif /* __CLK_UNIPHIER_H__ */
diff --git a/drivers/clk/versatile/clk-icst.c b/drivers/clk/versatile/clk-icst.c
index 77fd0ec..d52f976 100644
--- a/drivers/clk/versatile/clk-icst.c
+++ b/drivers/clk/versatile/clk-icst.c
@@ -484,7 +484,7 @@ static void __init of_syscon_icst_setup(struct device_node *np)
 	struct device_node *parent;
 	struct regmap *map;
 	struct clk_icst_desc icst_desc;
-	const char *name = np->name;
+	const char *name;
 	const char *parent_name;
 	struct clk *regclk;
 	enum icst_control_type ctype;
@@ -533,15 +533,17 @@ static void __init of_syscon_icst_setup(struct device_node *np)
 		icst_desc.params = &icst525_apcp_cm_params;
 		ctype = ICST_INTEGRATOR_CP_CM_MEM;
 	} else {
-		pr_err("unknown ICST clock %s\n", name);
+		pr_err("unknown ICST clock %pOF\n", np);
 		return;
 	}
 
 	/* Parent clock name is not the same as node parent */
 	parent_name = of_clk_get_parent_name(np, 0);
+	name = kasprintf(GFP_KERNEL, "%pOFP", np);
 
 	regclk = icst_clk_setup(NULL, &icst_desc, name, parent_name, map, ctype);
 	if (IS_ERR(regclk)) {
+		kfree(name);
 		pr_err("error setting up syscon ICST clock %s\n", name);
 		return;
 	}
diff --git a/drivers/irqchip/irq-csky-mpintc.c b/drivers/irqchip/irq-csky-mpintc.c
index cb403c9..4aebd67 100644
--- a/drivers/irqchip/irq-csky-mpintc.c
+++ b/drivers/irqchip/irq-csky-mpintc.c
@@ -78,7 +78,7 @@ static void csky_mpintc_handler(struct pt_regs *regs)
 		readl_relaxed(reg_base + INTCL_RDYIR));
 }
 
-static void csky_mpintc_enable(struct irq_data *d)
+static void csky_mpintc_unmask(struct irq_data *d)
 {
 	void __iomem *reg_base = this_cpu_read(intcl_reg);
 
@@ -87,7 +87,7 @@ static void csky_mpintc_enable(struct irq_data *d)
 	writel_relaxed(d->hwirq, reg_base + INTCL_SENR);
 }
 
-static void csky_mpintc_disable(struct irq_data *d)
+static void csky_mpintc_mask(struct irq_data *d)
 {
 	void __iomem *reg_base = this_cpu_read(intcl_reg);
 
@@ -164,8 +164,8 @@ static int csky_irq_set_affinity(struct irq_data *d,
 static struct irq_chip csky_irq_chip = {
 	.name           = "C-SKY SMP Intc",
 	.irq_eoi	= csky_mpintc_eoi,
-	.irq_enable	= csky_mpintc_enable,
-	.irq_disable	= csky_mpintc_disable,
+	.irq_unmask	= csky_mpintc_unmask,
+	.irq_mask	= csky_mpintc_mask,
 	.irq_set_type	= csky_mpintc_set_type,
 #ifdef CONFIG_SMP
 	.irq_set_affinity = csky_irq_set_affinity,
diff --git a/drivers/irqchip/irq-sifive-plic.c b/drivers/irqchip/irq-sifive-plic.c
index cf74cfa..259065d 100644
--- a/drivers/irqchip/irq-sifive-plic.c
+++ b/drivers/irqchip/irq-sifive-plic.c
@@ -163,7 +163,13 @@ static void plic_irq_eoi(struct irq_data *d)
 {
 	struct plic_handler *handler = this_cpu_ptr(&plic_handlers);
 
-	writel(d->hwirq, handler->hart_base + CONTEXT_CLAIM);
+	if (irqd_irq_masked(d)) {
+		plic_irq_unmask(d);
+		writel(d->hwirq, handler->hart_base + CONTEXT_CLAIM);
+		plic_irq_mask(d);
+	} else {
+		writel(d->hwirq, handler->hart_base + CONTEXT_CLAIM);
+	}
 }
 
 static struct irq_chip plic_chip = {
diff --git a/drivers/of/irq.c b/drivers/of/irq.c
index 32be5a0..b10f015b 100644
--- a/drivers/of/irq.c
+++ b/drivers/of/irq.c
@@ -161,9 +161,10 @@ int of_irq_parse_raw(const __be32 *addr, struct of_phandle_args *out_irq)
 		 * if it is then we are done, unless there is an
 		 * interrupt-map which takes precedence.
 		 */
+		bool intc = of_property_read_bool(ipar, "interrupt-controller");
+
 		imap = of_get_property(ipar, "interrupt-map", &imaplen);
-		if (imap == NULL &&
-		    of_property_read_bool(ipar, "interrupt-controller")) {
+		if (imap == NULL && intc) {
 			pr_debug(" -> got it !\n");
 			return 0;
 		}
@@ -244,8 +245,20 @@ int of_irq_parse_raw(const __be32 *addr, struct of_phandle_args *out_irq)
 
 			pr_debug(" -> imaplen=%d\n", imaplen);
 		}
-		if (!match)
+		if (!match) {
+			if (intc) {
+				/*
+				 * The PASEMI Nemo is a known offender, so
+				 * let's only warn for anyone else.
+				 */
+				WARN(!IS_ENABLED(CONFIG_PPC_PASEMI),
+				     "%pOF interrupt-map failed, using interrupt-controller\n",
+				     ipar);
+				return 0;
+			}
+
 			goto fail;
+		}
 
 		/*
 		 * Successfully parsed an interrrupt-map translation; copy new
diff --git a/drivers/of/platform.c b/drivers/of/platform.c
index 07813fb..b3faf89 100644
--- a/drivers/of/platform.c
+++ b/drivers/of/platform.c
@@ -76,6 +76,7 @@ static void of_device_make_bus_id(struct device *dev)
 	struct device_node *node = dev->of_node;
 	const __be32 *reg;
 	u64 addr;
+	u32 mask;
 
 	/* Construct the name, using parent nodes if necessary to ensure uniqueness */
 	while (node->parent) {
@@ -85,8 +86,13 @@ static void of_device_make_bus_id(struct device *dev)
 		 */
 		reg = of_get_property(node, "reg", NULL);
 		if (reg && (addr = of_translate_address(node, reg)) != OF_BAD_ADDR) {
-			dev_set_name(dev, dev_name(dev) ? "%llx.%pOFn:%s" : "%llx.%pOFn",
-				     addr, node, dev_name(dev));
+			if (!of_property_read_u32(node, "mask", &mask))
+				dev_set_name(dev, dev_name(dev) ? "%llx.%x.%pOFn:%s" : "%llx.%x.%pOFn",
+					     addr, ffs(mask) - 1, node, dev_name(dev));
+
+			else
+				dev_set_name(dev, dev_name(dev) ? "%llx.%pOFn:%s" : "%llx.%pOFn",
+					     addr, node, dev_name(dev));
 			return;
 		}
 
diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c
index 12e296d..48e3f4e 100644
--- a/drivers/pci/msi.c
+++ b/drivers/pci/msi.c
@@ -148,6 +148,9 @@ static noinline void pci_msi_update_mask(struct msi_desc *desc, u32 clear, u32 s
 	raw_spinlock_t *lock = &desc->dev->msi_lock;
 	unsigned long flags;
 
+	if (!desc->msi_attrib.can_mask)
+		return;
+
 	raw_spin_lock_irqsave(lock, flags);
 	desc->msi_mask &= ~clear;
 	desc->msi_mask |= set;
@@ -181,7 +184,8 @@ static void pci_msix_write_vector_ctrl(struct msi_desc *desc, u32 ctrl)
 {
 	void __iomem *desc_addr = pci_msix_desc_addr(desc);
 
-	writel(ctrl, desc_addr + PCI_MSIX_ENTRY_VECTOR_CTRL);
+	if (desc->msi_attrib.can_mask)
+		writel(ctrl, desc_addr + PCI_MSIX_ENTRY_VECTOR_CTRL);
 }
 
 static inline void pci_msix_mask(struct msi_desc *desc)
@@ -200,23 +204,17 @@ static inline void pci_msix_unmask(struct msi_desc *desc)
 
 static void __pci_msi_mask_desc(struct msi_desc *desc, u32 mask)
 {
-	if (pci_msi_ignore_mask || desc->msi_attrib.is_virtual)
-		return;
-
 	if (desc->msi_attrib.is_msix)
 		pci_msix_mask(desc);
-	else if (desc->msi_attrib.maskbit)
+	else
 		pci_msi_mask(desc, mask);
 }
 
 static void __pci_msi_unmask_desc(struct msi_desc *desc, u32 mask)
 {
-	if (pci_msi_ignore_mask || desc->msi_attrib.is_virtual)
-		return;
-
 	if (desc->msi_attrib.is_msix)
 		pci_msix_unmask(desc);
-	else if (desc->msi_attrib.maskbit)
+	else
 		pci_msi_unmask(desc, mask);
 }
 
@@ -370,6 +368,11 @@ static void free_msi_irqs(struct pci_dev *dev)
 			for (i = 0; i < entry->nvec_used; i++)
 				BUG_ON(irq_has_action(entry->irq + i));
 
+	if (dev->msi_irq_groups) {
+		msi_destroy_sysfs(&dev->dev, dev->msi_irq_groups);
+		dev->msi_irq_groups = NULL;
+	}
+
 	pci_msi_teardown_msi_irqs(dev);
 
 	list_for_each_entry_safe(entry, tmp, msi_list, list) {
@@ -381,11 +384,6 @@ static void free_msi_irqs(struct pci_dev *dev)
 		list_del(&entry->list);
 		free_msi_entry(entry);
 	}
-
-	if (dev->msi_irq_groups) {
-		msi_destroy_sysfs(&dev->dev, dev->msi_irq_groups);
-		dev->msi_irq_groups = NULL;
-	}
 }
 
 static void pci_intx_for_msi(struct pci_dev *dev, int enable)
@@ -479,12 +477,16 @@ msi_setup_entry(struct pci_dev *dev, int nvec, struct irq_affinity *affd)
 		goto out;
 
 	pci_read_config_word(dev, dev->msi_cap + PCI_MSI_FLAGS, &control);
+	/* Lies, damned lies, and MSIs */
+	if (dev->dev_flags & PCI_DEV_FLAGS_HAS_MSI_MASKING)
+		control |= PCI_MSI_FLAGS_MASKBIT;
 
 	entry->msi_attrib.is_msix	= 0;
 	entry->msi_attrib.is_64		= !!(control & PCI_MSI_FLAGS_64BIT);
 	entry->msi_attrib.is_virtual    = 0;
 	entry->msi_attrib.entry_nr	= 0;
-	entry->msi_attrib.maskbit	= !!(control & PCI_MSI_FLAGS_MASKBIT);
+	entry->msi_attrib.can_mask	= !pci_msi_ignore_mask &&
+					  !!(control & PCI_MSI_FLAGS_MASKBIT);
 	entry->msi_attrib.default_irq	= dev->irq;	/* Save IOAPIC IRQ */
 	entry->msi_attrib.multi_cap	= (control & PCI_MSI_FLAGS_QMASK) >> 1;
 	entry->msi_attrib.multiple	= ilog2(__roundup_pow_of_two(nvec));
@@ -495,7 +497,7 @@ msi_setup_entry(struct pci_dev *dev, int nvec, struct irq_affinity *affd)
 		entry->mask_pos = dev->msi_cap + PCI_MSI_MASK_32;
 
 	/* Save the initial mask status */
-	if (entry->msi_attrib.maskbit)
+	if (entry->msi_attrib.can_mask)
 		pci_read_config_dword(dev, entry->mask_pos, &entry->msi_mask);
 
 out:
@@ -639,10 +641,13 @@ static int msix_setup_entries(struct pci_dev *dev, void __iomem *base,
 		entry->msi_attrib.is_virtual =
 			entry->msi_attrib.entry_nr >= vec_count;
 
+		entry->msi_attrib.can_mask	= !pci_msi_ignore_mask &&
+						  !entry->msi_attrib.is_virtual;
+
 		entry->msi_attrib.default_irq	= dev->irq;
 		entry->mask_base		= base;
 
-		if (!entry->msi_attrib.is_virtual) {
+		if (entry->msi_attrib.can_mask) {
 			addr = pci_msix_desc_addr(entry);
 			entry->msix_ctrl = readl(addr + PCI_MSIX_ENTRY_VECTOR_CTRL);
 		}
diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index aedb78c..003950c 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -5851,3 +5851,9 @@ DECLARE_PCI_FIXUP_ENABLE(PCI_VENDOR_ID_PERICOM, 0x2303,
 			 pci_fixup_pericom_acs_store_forward);
 DECLARE_PCI_FIXUP_RESUME(PCI_VENDOR_ID_PERICOM, 0x2303,
 			 pci_fixup_pericom_acs_store_forward);
+
+static void nvidia_ion_ahci_fixup(struct pci_dev *pdev)
+{
+	pdev->dev_flags |= PCI_DEV_FLAGS_HAS_MSI_MASKING;
+}
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_NVIDIA, 0x0ab8, nvidia_ion_ahci_fixup);
diff --git a/drivers/sh/maple/maple.c b/drivers/sh/maple/maple.c
index bd0fbcd..e24e220 100644
--- a/drivers/sh/maple/maple.c
+++ b/drivers/sh/maple/maple.c
@@ -834,8 +834,10 @@ static int __init maple_bus_init(void)
 
 	maple_queue_cache = KMEM_CACHE(maple_buffer, SLAB_HWCACHE_ALIGN);
 
-	if (!maple_queue_cache)
+	if (!maple_queue_cache) {
+		retval = -ENOMEM;
 		goto cleanup_bothirqs;
+	}
 
 	INIT_LIST_HEAD(&maple_waitq);
 	INIT_LIST_HEAD(&maple_sentq);
@@ -848,6 +850,7 @@ static int __init maple_bus_init(void)
 		if (!mdev[i]) {
 			while (i-- > 0)
 				maple_free_dev(mdev[i]);
+			retval = -ENOMEM;
 			goto cleanup_cache;
 		}
 		baseunits[i] = mdev[i];
diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index 0da0af2..96e5a87 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -2889,6 +2889,7 @@ static unsigned int virtio_mem_features[] = {
 #if defined(CONFIG_NUMA) && defined(CONFIG_ACPI_NUMA)
 	VIRTIO_MEM_F_ACPI_PXM,
 #endif
+	VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE,
 };
 
 static const struct virtio_device_id virtio_mem_id_table[] = {
diff --git a/fs/btrfs/zstd.c b/fs/btrfs/zstd.c
index f06b680..fc42dd0 100644
--- a/fs/btrfs/zstd.c
+++ b/fs/btrfs/zstd.c
@@ -28,10 +28,10 @@
 /* 307s to avoid pathologically clashing with transaction commit */
 #define ZSTD_BTRFS_RECLAIM_JIFFIES (307 * HZ)
 
-static ZSTD_parameters zstd_get_btrfs_parameters(unsigned int level,
+static zstd_parameters zstd_get_btrfs_parameters(unsigned int level,
 						 size_t src_len)
 {
-	ZSTD_parameters params = ZSTD_getParams(level, src_len, 0);
+	zstd_parameters params = zstd_get_params(level, src_len);
 
 	if (params.cParams.windowLog > ZSTD_BTRFS_MAX_WINDOWLOG)
 		params.cParams.windowLog = ZSTD_BTRFS_MAX_WINDOWLOG;
@@ -48,8 +48,8 @@ struct workspace {
 	unsigned long last_used; /* jiffies */
 	struct list_head list;
 	struct list_head lru_list;
-	ZSTD_inBuffer in_buf;
-	ZSTD_outBuffer out_buf;
+	zstd_in_buffer in_buf;
+	zstd_out_buffer out_buf;
 };
 
 /*
@@ -155,12 +155,12 @@ static void zstd_calc_ws_mem_sizes(void)
 	unsigned int level;
 
 	for (level = 1; level <= ZSTD_BTRFS_MAX_LEVEL; level++) {
-		ZSTD_parameters params =
+		zstd_parameters params =
 			zstd_get_btrfs_parameters(level, ZSTD_BTRFS_MAX_INPUT);
 		size_t level_size =
 			max_t(size_t,
-			      ZSTD_CStreamWorkspaceBound(params.cParams),
-			      ZSTD_DStreamWorkspaceBound(ZSTD_BTRFS_MAX_INPUT));
+			      zstd_cstream_workspace_bound(&params.cParams),
+			      zstd_dstream_workspace_bound(ZSTD_BTRFS_MAX_INPUT));
 
 		max_size = max_t(size_t, max_size, level_size);
 		zstd_ws_mem_sizes[level - 1] = max_size;
@@ -371,7 +371,7 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 		unsigned long *total_in, unsigned long *total_out)
 {
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
-	ZSTD_CStream *stream;
+	zstd_cstream *stream;
 	int ret = 0;
 	int nr_pages = 0;
 	struct page *in_page = NULL;  /* The current page to read */
@@ -381,7 +381,7 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 	unsigned long len = *total_out;
 	const unsigned long nr_dest_pages = *out_pages;
 	unsigned long max_out = nr_dest_pages * PAGE_SIZE;
-	ZSTD_parameters params = zstd_get_btrfs_parameters(workspace->req_level,
+	zstd_parameters params = zstd_get_btrfs_parameters(workspace->req_level,
 							   len);
 
 	*out_pages = 0;
@@ -389,10 +389,10 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 	*total_in = 0;
 
 	/* Initialize the stream */
-	stream = ZSTD_initCStream(params, len, workspace->mem,
+	stream = zstd_init_cstream(&params, len, workspace->mem,
 			workspace->size);
 	if (!stream) {
-		pr_warn("BTRFS: ZSTD_initCStream failed\n");
+		pr_warn("BTRFS: zstd_init_cstream failed\n");
 		ret = -EIO;
 		goto out;
 	}
@@ -418,11 +418,11 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 	while (1) {
 		size_t ret2;
 
-		ret2 = ZSTD_compressStream(stream, &workspace->out_buf,
+		ret2 = zstd_compress_stream(stream, &workspace->out_buf,
 				&workspace->in_buf);
-		if (ZSTD_isError(ret2)) {
-			pr_debug("BTRFS: ZSTD_compressStream returned %d\n",
-					ZSTD_getErrorCode(ret2));
+		if (zstd_is_error(ret2)) {
+			pr_debug("BTRFS: zstd_compress_stream returned %d\n",
+					zstd_get_error_code(ret2));
 			ret = -EIO;
 			goto out;
 		}
@@ -487,10 +487,10 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 	while (1) {
 		size_t ret2;
 
-		ret2 = ZSTD_endStream(stream, &workspace->out_buf);
-		if (ZSTD_isError(ret2)) {
-			pr_debug("BTRFS: ZSTD_endStream returned %d\n",
-					ZSTD_getErrorCode(ret2));
+		ret2 = zstd_end_stream(stream, &workspace->out_buf);
+		if (zstd_is_error(ret2)) {
+			pr_debug("BTRFS: zstd_end_stream returned %d\n",
+					zstd_get_error_code(ret2));
 			ret = -EIO;
 			goto out;
 		}
@@ -548,17 +548,17 @@ int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
 	struct page **pages_in = cb->compressed_pages;
 	size_t srclen = cb->compressed_len;
-	ZSTD_DStream *stream;
+	zstd_dstream *stream;
 	int ret = 0;
 	unsigned long page_in_index = 0;
 	unsigned long total_pages_in = DIV_ROUND_UP(srclen, PAGE_SIZE);
 	unsigned long buf_start;
 	unsigned long total_out = 0;
 
-	stream = ZSTD_initDStream(
+	stream = zstd_init_dstream(
 			ZSTD_BTRFS_MAX_INPUT, workspace->mem, workspace->size);
 	if (!stream) {
-		pr_debug("BTRFS: ZSTD_initDStream failed\n");
+		pr_debug("BTRFS: zstd_init_dstream failed\n");
 		ret = -EIO;
 		goto done;
 	}
@@ -574,11 +574,11 @@ int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 	while (1) {
 		size_t ret2;
 
-		ret2 = ZSTD_decompressStream(stream, &workspace->out_buf,
+		ret2 = zstd_decompress_stream(stream, &workspace->out_buf,
 				&workspace->in_buf);
-		if (ZSTD_isError(ret2)) {
-			pr_debug("BTRFS: ZSTD_decompressStream returned %d\n",
-					ZSTD_getErrorCode(ret2));
+		if (zstd_is_error(ret2)) {
+			pr_debug("BTRFS: zstd_decompress_stream returned %d\n",
+					zstd_get_error_code(ret2));
 			ret = -EIO;
 			goto done;
 		}
@@ -624,16 +624,16 @@ int zstd_decompress(struct list_head *ws, unsigned char *data_in,
 		size_t destlen)
 {
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
-	ZSTD_DStream *stream;
+	zstd_dstream *stream;
 	int ret = 0;
 	size_t ret2;
 	unsigned long total_out = 0;
 	unsigned long pg_offset = 0;
 
-	stream = ZSTD_initDStream(
+	stream = zstd_init_dstream(
 			ZSTD_BTRFS_MAX_INPUT, workspace->mem, workspace->size);
 	if (!stream) {
-		pr_warn("BTRFS: ZSTD_initDStream failed\n");
+		pr_warn("BTRFS: zstd_init_dstream failed\n");
 		ret = -EIO;
 		goto finish;
 	}
@@ -657,15 +657,15 @@ int zstd_decompress(struct list_head *ws, unsigned char *data_in,
 
 		/* Check if the frame is over and we still need more input */
 		if (ret2 == 0) {
-			pr_debug("BTRFS: ZSTD_decompressStream ended early\n");
+			pr_debug("BTRFS: zstd_decompress_stream ended early\n");
 			ret = -EIO;
 			goto finish;
 		}
-		ret2 = ZSTD_decompressStream(stream, &workspace->out_buf,
+		ret2 = zstd_decompress_stream(stream, &workspace->out_buf,
 				&workspace->in_buf);
-		if (ZSTD_isError(ret2)) {
-			pr_debug("BTRFS: ZSTD_decompressStream returned %d\n",
-					ZSTD_getErrorCode(ret2));
+		if (zstd_is_error(ret2)) {
+			pr_debug("BTRFS: zstd_decompress_stream returned %d\n",
+					zstd_get_error_code(ret2));
 			ret = -EIO;
 			goto finish;
 		}
diff --git a/fs/f2fs/compress.c b/fs/f2fs/compress.c
index a0d5cfa..49121a2 100644
--- a/fs/f2fs/compress.c
+++ b/fs/f2fs/compress.c
@@ -336,8 +336,8 @@ static const struct f2fs_compress_ops f2fs_lz4_ops = {
 
 static int zstd_init_compress_ctx(struct compress_ctx *cc)
 {
-	ZSTD_parameters params;
-	ZSTD_CStream *stream;
+	zstd_parameters params;
+	zstd_cstream *stream;
 	void *workspace;
 	unsigned int workspace_size;
 	unsigned char level = F2FS_I(cc->inode)->i_compress_flag >>
@@ -346,17 +346,17 @@ static int zstd_init_compress_ctx(struct compress_ctx *cc)
 	if (!level)
 		level = F2FS_ZSTD_DEFAULT_CLEVEL;
 
-	params = ZSTD_getParams(level, cc->rlen, 0);
-	workspace_size = ZSTD_CStreamWorkspaceBound(params.cParams);
+	params = zstd_get_params(F2FS_ZSTD_DEFAULT_CLEVEL, cc->rlen);
+	workspace_size = zstd_cstream_workspace_bound(&params.cParams);
 
 	workspace = f2fs_kvmalloc(F2FS_I_SB(cc->inode),
 					workspace_size, GFP_NOFS);
 	if (!workspace)
 		return -ENOMEM;
 
-	stream = ZSTD_initCStream(params, 0, workspace, workspace_size);
+	stream = zstd_init_cstream(&params, 0, workspace, workspace_size);
 	if (!stream) {
-		printk_ratelimited("%sF2FS-fs (%s): %s ZSTD_initCStream failed\n",
+		printk_ratelimited("%sF2FS-fs (%s): %s zstd_init_cstream failed\n",
 				KERN_ERR, F2FS_I_SB(cc->inode)->sb->s_id,
 				__func__);
 		kvfree(workspace);
@@ -379,9 +379,9 @@ static void zstd_destroy_compress_ctx(struct compress_ctx *cc)
 
 static int zstd_compress_pages(struct compress_ctx *cc)
 {
-	ZSTD_CStream *stream = cc->private2;
-	ZSTD_inBuffer inbuf;
-	ZSTD_outBuffer outbuf;
+	zstd_cstream *stream = cc->private2;
+	zstd_in_buffer inbuf;
+	zstd_out_buffer outbuf;
 	int src_size = cc->rlen;
 	int dst_size = src_size - PAGE_SIZE - COMPRESS_HEADER_SIZE;
 	int ret;
@@ -394,19 +394,19 @@ static int zstd_compress_pages(struct compress_ctx *cc)
 	outbuf.dst = cc->cbuf->cdata;
 	outbuf.size = dst_size;
 
-	ret = ZSTD_compressStream(stream, &outbuf, &inbuf);
-	if (ZSTD_isError(ret)) {
-		printk_ratelimited("%sF2FS-fs (%s): %s ZSTD_compressStream failed, ret: %d\n",
+	ret = zstd_compress_stream(stream, &outbuf, &inbuf);
+	if (zstd_is_error(ret)) {
+		printk_ratelimited("%sF2FS-fs (%s): %s zstd_compress_stream failed, ret: %d\n",
 				KERN_ERR, F2FS_I_SB(cc->inode)->sb->s_id,
-				__func__, ZSTD_getErrorCode(ret));
+				__func__, zstd_get_error_code(ret));
 		return -EIO;
 	}
 
-	ret = ZSTD_endStream(stream, &outbuf);
-	if (ZSTD_isError(ret)) {
-		printk_ratelimited("%sF2FS-fs (%s): %s ZSTD_endStream returned %d\n",
+	ret = zstd_end_stream(stream, &outbuf);
+	if (zstd_is_error(ret)) {
+		printk_ratelimited("%sF2FS-fs (%s): %s zstd_end_stream returned %d\n",
 				KERN_ERR, F2FS_I_SB(cc->inode)->sb->s_id,
-				__func__, ZSTD_getErrorCode(ret));
+				__func__, zstd_get_error_code(ret));
 		return -EIO;
 	}
 
@@ -423,22 +423,22 @@ static int zstd_compress_pages(struct compress_ctx *cc)
 
 static int zstd_init_decompress_ctx(struct decompress_io_ctx *dic)
 {
-	ZSTD_DStream *stream;
+	zstd_dstream *stream;
 	void *workspace;
 	unsigned int workspace_size;
 	unsigned int max_window_size =
 			MAX_COMPRESS_WINDOW_SIZE(dic->log_cluster_size);
 
-	workspace_size = ZSTD_DStreamWorkspaceBound(max_window_size);
+	workspace_size = zstd_dstream_workspace_bound(max_window_size);
 
 	workspace = f2fs_kvmalloc(F2FS_I_SB(dic->inode),
 					workspace_size, GFP_NOFS);
 	if (!workspace)
 		return -ENOMEM;
 
-	stream = ZSTD_initDStream(max_window_size, workspace, workspace_size);
+	stream = zstd_init_dstream(max_window_size, workspace, workspace_size);
 	if (!stream) {
-		printk_ratelimited("%sF2FS-fs (%s): %s ZSTD_initDStream failed\n",
+		printk_ratelimited("%sF2FS-fs (%s): %s zstd_init_dstream failed\n",
 				KERN_ERR, F2FS_I_SB(dic->inode)->sb->s_id,
 				__func__);
 		kvfree(workspace);
@@ -460,9 +460,9 @@ static void zstd_destroy_decompress_ctx(struct decompress_io_ctx *dic)
 
 static int zstd_decompress_pages(struct decompress_io_ctx *dic)
 {
-	ZSTD_DStream *stream = dic->private2;
-	ZSTD_inBuffer inbuf;
-	ZSTD_outBuffer outbuf;
+	zstd_dstream *stream = dic->private2;
+	zstd_in_buffer inbuf;
+	zstd_out_buffer outbuf;
 	int ret;
 
 	inbuf.pos = 0;
@@ -473,11 +473,11 @@ static int zstd_decompress_pages(struct decompress_io_ctx *dic)
 	outbuf.dst = dic->rbuf;
 	outbuf.size = dic->rlen;
 
-	ret = ZSTD_decompressStream(stream, &outbuf, &inbuf);
-	if (ZSTD_isError(ret)) {
-		printk_ratelimited("%sF2FS-fs (%s): %s ZSTD_compressStream failed, ret: %d\n",
+	ret = zstd_decompress_stream(stream, &outbuf, &inbuf);
+	if (zstd_is_error(ret)) {
+		printk_ratelimited("%sF2FS-fs (%s): %s zstd_decompress_stream failed, ret: %d\n",
 				KERN_ERR, F2FS_I_SB(dic->inode)->sb->s_id,
-				__func__, ZSTD_getErrorCode(ret));
+				__func__, zstd_get_error_code(ret));
 		return -EIO;
 	}
 
diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
index 7960ce0..040b6d0 100644
--- a/fs/f2fs/super.c
+++ b/fs/f2fs/super.c
@@ -593,7 +593,7 @@ static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
 	if (kstrtouint(str + 1, 10, &level))
 		return -EINVAL;
 
-	if (!level || level > ZSTD_maxCLevel()) {
+	if (!level || level > zstd_max_clevel()) {
 		f2fs_info(sbi, "invalid zstd compress level: %d", level);
 		return -EINVAL;
 	}
diff --git a/fs/io-wq.c b/fs/io-wq.c
index afd955d..88202de 100644
--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -423,9 +423,10 @@ static inline unsigned int io_get_work_hash(struct io_wq_work *work)
 	return work->flags >> IO_WQ_HASH_SHIFT;
 }
 
-static void io_wait_on_hash(struct io_wqe *wqe, unsigned int hash)
+static bool io_wait_on_hash(struct io_wqe *wqe, unsigned int hash)
 {
 	struct io_wq *wq = wqe->wq;
+	bool ret = false;
 
 	spin_lock_irq(&wq->hash->wait.lock);
 	if (list_empty(&wqe->wait.entry)) {
@@ -433,9 +434,11 @@ static void io_wait_on_hash(struct io_wqe *wqe, unsigned int hash)
 		if (!test_bit(hash, &wq->hash->map)) {
 			__set_current_state(TASK_RUNNING);
 			list_del_init(&wqe->wait.entry);
+			ret = true;
 		}
 	}
 	spin_unlock_irq(&wq->hash->wait.lock);
+	return ret;
 }
 
 static struct io_wq_work *io_get_next_work(struct io_wqe_acct *acct,
@@ -475,14 +478,21 @@ static struct io_wq_work *io_get_next_work(struct io_wqe_acct *acct,
 	}
 
 	if (stall_hash != -1U) {
+		bool unstalled;
+
 		/*
 		 * Set this before dropping the lock to avoid racing with new
 		 * work being added and clearing the stalled bit.
 		 */
 		set_bit(IO_ACCT_STALLED_BIT, &acct->flags);
 		raw_spin_unlock(&wqe->lock);
-		io_wait_on_hash(wqe, stall_hash);
+		unstalled = io_wait_on_hash(wqe, stall_hash);
 		raw_spin_lock(&wqe->lock);
+		if (unstalled) {
+			clear_bit(IO_ACCT_STALLED_BIT, &acct->flags);
+			if (wq_has_sleeper(&wqe->wq->hash->wait))
+				wake_up(&wqe->wq->hash->wait);
+		}
 	}
 
 	return NULL;
@@ -564,8 +574,11 @@ static void io_worker_handle_work(struct io_worker *worker)
 				io_wqe_enqueue(wqe, linked);
 
 			if (hash != -1U && !next_hashed) {
+				/* serialize hash clear with wake_up() */
+				spin_lock_irq(&wq->hash->wait.lock);
 				clear_bit(hash, &wq->hash->map);
 				clear_bit(IO_ACCT_STALLED_BIT, &acct->flags);
+				spin_unlock_irq(&wq->hash->wait.lock);
 				if (wq_has_sleeper(&wq->hash->wait))
 					wake_up(&wq->hash->wait);
 				raw_spin_lock(&wqe->lock);
diff --git a/fs/pstore/platform.c b/fs/pstore/platform.c
index b9614db..f243cb5 100644
--- a/fs/pstore/platform.c
+++ b/fs/pstore/platform.c
@@ -218,7 +218,7 @@ static int zbufsize_842(size_t size)
 #if IS_ENABLED(CONFIG_PSTORE_ZSTD_COMPRESS)
 static int zbufsize_zstd(size_t size)
 {
-	return ZSTD_compressBound(size);
+	return zstd_compress_bound(size);
 }
 #endif
 
diff --git a/fs/squashfs/zstd_wrapper.c b/fs/squashfs/zstd_wrapper.c
index 0015cf8..c40445d 100644
--- a/fs/squashfs/zstd_wrapper.c
+++ b/fs/squashfs/zstd_wrapper.c
@@ -34,7 +34,7 @@ static void *zstd_init(struct squashfs_sb_info *msblk, void *buff)
 		goto failed;
 	wksp->window_size = max_t(size_t,
 			msblk->block_size, SQUASHFS_METADATA_SIZE);
-	wksp->mem_size = ZSTD_DStreamWorkspaceBound(wksp->window_size);
+	wksp->mem_size = zstd_dstream_workspace_bound(wksp->window_size);
 	wksp->mem = vmalloc(wksp->mem_size);
 	if (wksp->mem == NULL)
 		goto failed;
@@ -63,15 +63,15 @@ static int zstd_uncompress(struct squashfs_sb_info *msblk, void *strm,
 	struct squashfs_page_actor *output)
 {
 	struct workspace *wksp = strm;
-	ZSTD_DStream *stream;
+	zstd_dstream *stream;
 	size_t total_out = 0;
 	int error = 0;
-	ZSTD_inBuffer in_buf = { NULL, 0, 0 };
-	ZSTD_outBuffer out_buf = { NULL, 0, 0 };
+	zstd_in_buffer in_buf = { NULL, 0, 0 };
+	zstd_out_buffer out_buf = { NULL, 0, 0 };
 	struct bvec_iter_all iter_all = {};
 	struct bio_vec *bvec = bvec_init_iter_all(&iter_all);
 
-	stream = ZSTD_initDStream(wksp->window_size, wksp->mem, wksp->mem_size);
+	stream = zstd_init_dstream(wksp->window_size, wksp->mem, wksp->mem_size);
 
 	if (!stream) {
 		ERROR("Failed to initialize zstd decompressor\n");
@@ -116,14 +116,14 @@ static int zstd_uncompress(struct squashfs_sb_info *msblk, void *strm,
 		}
 
 		total_out -= out_buf.pos;
-		zstd_err = ZSTD_decompressStream(stream, &out_buf, &in_buf);
+		zstd_err = zstd_decompress_stream(stream, &out_buf, &in_buf);
 		total_out += out_buf.pos; /* add the additional data produced */
 		if (zstd_err == 0)
 			break;
 
-		if (ZSTD_isError(zstd_err)) {
+		if (zstd_is_error(zstd_err)) {
 			ERROR("zstd decompression error: %d\n",
-					(int)ZSTD_getErrorCode(zstd_err));
+					(int)zstd_get_error_code(zstd_err));
 			error = -EIO;
 			break;
 		}
diff --git a/include/dt-bindings/clock/jz4725b-cgu.h b/include/dt-bindings/clock/ingenic,jz4725b-cgu.h
similarity index 100%
rename from include/dt-bindings/clock/jz4725b-cgu.h
rename to include/dt-bindings/clock/ingenic,jz4725b-cgu.h
diff --git a/include/dt-bindings/clock/jz4740-cgu.h b/include/dt-bindings/clock/ingenic,jz4740-cgu.h
similarity index 100%
rename from include/dt-bindings/clock/jz4740-cgu.h
rename to include/dt-bindings/clock/ingenic,jz4740-cgu.h
diff --git a/include/dt-bindings/clock/jz4760-cgu.h b/include/dt-bindings/clock/ingenic,jz4760-cgu.h
similarity index 100%
rename from include/dt-bindings/clock/jz4760-cgu.h
rename to include/dt-bindings/clock/ingenic,jz4760-cgu.h
diff --git a/include/dt-bindings/clock/jz4770-cgu.h b/include/dt-bindings/clock/ingenic,jz4770-cgu.h
similarity index 100%
rename from include/dt-bindings/clock/jz4770-cgu.h
rename to include/dt-bindings/clock/ingenic,jz4770-cgu.h
diff --git a/include/dt-bindings/clock/jz4780-cgu.h b/include/dt-bindings/clock/ingenic,jz4780-cgu.h
similarity index 100%
rename from include/dt-bindings/clock/jz4780-cgu.h
rename to include/dt-bindings/clock/ingenic,jz4780-cgu.h
diff --git a/include/dt-bindings/clock/x1000-cgu.h b/include/dt-bindings/clock/ingenic,x1000-cgu.h
similarity index 100%
rename from include/dt-bindings/clock/x1000-cgu.h
rename to include/dt-bindings/clock/ingenic,x1000-cgu.h
diff --git a/include/dt-bindings/clock/x1830-cgu.h b/include/dt-bindings/clock/ingenic,x1830-cgu.h
similarity index 100%
rename from include/dt-bindings/clock/x1830-cgu.h
rename to include/dt-bindings/clock/ingenic,x1830-cgu.h
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 462634b..74c4102 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -205,9 +205,9 @@ static inline dev_t disk_devt(struct gendisk *disk)
 void disk_uevent(struct gendisk *disk, enum kobject_action action);
 
 /* block/genhd.c */
-int device_add_disk(struct device *parent, struct gendisk *disk,
-		const struct attribute_group **groups);
-static inline int add_disk(struct gendisk *disk)
+int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
+				 const struct attribute_group **groups);
+static inline int __must_check add_disk(struct gendisk *disk)
 {
 	return device_add_disk(NULL, disk, NULL);
 }
diff --git a/include/linux/kernel.h b/include/linux/kernel.h
index 968b4c4..77755ac 100644
--- a/include/linux/kernel.h
+++ b/include/linux/kernel.h
@@ -85,7 +85,7 @@
 struct completion;
 struct user;
 
-#ifdef CONFIG_PREEMPT_VOLUNTARY
+#ifdef CONFIG_PREEMPT_VOLUNTARY_BUILD
 
 extern int __cond_resched(void);
 # define might_resched() __cond_resched()
diff --git a/include/linux/msi.h b/include/linux/msi.h
index 49cf6eb..e616f94 100644
--- a/include/linux/msi.h
+++ b/include/linux/msi.h
@@ -148,7 +148,7 @@ struct msi_desc {
 				u8	is_msix		: 1;
 				u8	multiple	: 3;
 				u8	multi_cap	: 3;
-				u8	maskbit		: 1;
+				u8	can_mask	: 1;
 				u8	is_64		: 1;
 				u8	is_virtual	: 1;
 				u16	entry_nr;
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 0b9d580..18a75c8e 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -233,6 +233,8 @@ enum pci_dev_flags {
 	PCI_DEV_FLAGS_NO_FLR_RESET = (__force pci_dev_flags_t) (1 << 10),
 	/* Don't use Relaxed Ordering for TLPs directed at this device */
 	PCI_DEV_FLAGS_NO_RELAXED_ORDERING = (__force pci_dev_flags_t) (1 << 11),
+	/* Device does honor MSI masking despite saying otherwise */
+	PCI_DEV_FLAGS_HAS_MSI_MASKING = (__force pci_dev_flags_t) (1 << 12),
 };
 
 enum pci_irq_reroute_variant {
diff --git a/include/linux/posix-timers.h b/include/linux/posix-timers.h
index 00fef00..5bbcd28 100644
--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -184,8 +184,10 @@ static inline void posix_cputimers_group_init(struct posix_cputimers *pct,
 #endif
 
 #ifdef CONFIG_POSIX_CPU_TIMERS_TASK_WORK
+void clear_posix_cputimers_work(struct task_struct *p);
 void posix_cputimers_init_work(void);
 #else
+static inline void clear_posix_cputimers_work(struct task_struct *p) { }
 static inline void posix_cputimers_init_work(void) { }
 #endif
 
diff --git a/include/linux/vermagic.h b/include/linux/vermagic.h
index 1eaaa93..329d63b 100644
--- a/include/linux/vermagic.h
+++ b/include/linux/vermagic.h
@@ -15,7 +15,7 @@
 #else
 #define MODULE_VERMAGIC_SMP ""
 #endif
-#ifdef CONFIG_PREEMPT
+#ifdef CONFIG_PREEMPT_BUILD
 #define MODULE_VERMAGIC_PREEMPT "preempt "
 #elif defined(CONFIG_PREEMPT_RT)
 #define MODULE_VERMAGIC_PREEMPT "preempt_rt "
diff --git a/include/linux/zstd.h b/include/linux/zstd.h
index e87f78c..113408e 100644
--- a/include/linux/zstd.h
+++ b/include/linux/zstd.h
@@ -1,138 +1,97 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd) and
+ * the GPLv2 (found in the COPYING file in the root directory of
+ * https://github.com/facebook/zstd). You may select, at your option, one of the
+ * above-listed licenses.
  */
 
-#ifndef ZSTD_H
-#define ZSTD_H
-
-/* ======   Dependency   ======*/
-#include <linux/types.h>   /* size_t */
-
-
-/*-*****************************************************************************
- * Introduction
- *
- * zstd, short for Zstandard, is a fast lossless compression algorithm,
- * targeting real-time compression scenarios at zlib-level and better
- * compression ratios. The zstd compression library provides in-memory
- * compression and decompression functions. The library supports compression
- * levels from 1 up to ZSTD_maxCLevel() which is 22. Levels >= 20, labeled
- * ultra, should be used with caution, as they require more memory.
- * Compression can be done in:
- *  - a single step, reusing a context (described as Explicit memory management)
- *  - unbounded multiple steps (described as Streaming compression)
- * The compression ratio achievable on small data can be highly improved using
- * compression with a dictionary in:
- *  - a single step (described as Simple dictionary API)
- *  - a single step, reusing a dictionary (described as Fast dictionary API)
- ******************************************************************************/
-
-/*======  Helper functions  ======*/
+#ifndef LINUX_ZSTD_H
+#define LINUX_ZSTD_H
 
 /**
- * enum ZSTD_ErrorCode - zstd error codes
- *
- * Functions that return size_t can be checked for errors using ZSTD_isError()
- * and the ZSTD_ErrorCode can be extracted using ZSTD_getErrorCode().
+ * This is a kernel-style API that wraps the upstream zstd API, which cannot be
+ * used directly because the symbols aren't exported. It exposes the minimal
+ * functionality which is currently required by users of zstd in the kernel.
+ * Expose extra functions from lib/zstd/zstd.h as needed.
  */
-typedef enum {
-	ZSTD_error_no_error,
-	ZSTD_error_GENERIC,
-	ZSTD_error_prefix_unknown,
-	ZSTD_error_version_unsupported,
-	ZSTD_error_parameter_unknown,
-	ZSTD_error_frameParameter_unsupported,
-	ZSTD_error_frameParameter_unsupportedBy32bits,
-	ZSTD_error_frameParameter_windowTooLarge,
-	ZSTD_error_compressionParameter_unsupported,
-	ZSTD_error_init_missing,
-	ZSTD_error_memory_allocation,
-	ZSTD_error_stage_wrong,
-	ZSTD_error_dstSize_tooSmall,
-	ZSTD_error_srcSize_wrong,
-	ZSTD_error_corruption_detected,
-	ZSTD_error_checksum_wrong,
-	ZSTD_error_tableLog_tooLarge,
-	ZSTD_error_maxSymbolValue_tooLarge,
-	ZSTD_error_maxSymbolValue_tooSmall,
-	ZSTD_error_dictionary_corrupted,
-	ZSTD_error_dictionary_wrong,
-	ZSTD_error_dictionaryCreation_failed,
-	ZSTD_error_maxCode
-} ZSTD_ErrorCode;
+
+/* ======   Dependency   ====== */
+#include <linux/types.h>
+#include <linux/zstd_errors.h>
+#include <linux/zstd_lib.h>
+
+/* ======   Helper Functions   ====== */
+/**
+ * zstd_compress_bound() - maximum compressed size in worst case scenario
+ * @src_size: The size of the data to compress.
+ *
+ * Return:    The maximum compressed size in the worst case scenario.
+ */
+size_t zstd_compress_bound(size_t src_size);
 
 /**
- * ZSTD_maxCLevel() - maximum compression level available
- *
- * Return: Maximum compression level available.
- */
-int ZSTD_maxCLevel(void);
-/**
- * ZSTD_compressBound() - maximum compressed size in worst case scenario
- * @srcSize: The size of the data to compress.
- *
- * Return:   The maximum compressed size in the worst case scenario.
- */
-size_t ZSTD_compressBound(size_t srcSize);
-/**
- * ZSTD_isError() - tells if a size_t function result is an error code
+ * zstd_is_error() - tells if a size_t function result is an error code
  * @code:  The function result to check for error.
  *
  * Return: Non-zero iff the code is an error.
  */
-static __attribute__((unused)) unsigned int ZSTD_isError(size_t code)
-{
-	return code > (size_t)-ZSTD_error_maxCode;
-}
-/**
- * ZSTD_getErrorCode() - translates an error function result to a ZSTD_ErrorCode
- * @functionResult: The result of a function for which ZSTD_isError() is true.
- *
- * Return:          The ZSTD_ErrorCode corresponding to the functionResult or 0
- *                  if the functionResult isn't an error.
- */
-static __attribute__((unused)) ZSTD_ErrorCode ZSTD_getErrorCode(
-	size_t functionResult)
-{
-	if (!ZSTD_isError(functionResult))
-		return (ZSTD_ErrorCode)0;
-	return (ZSTD_ErrorCode)(0 - functionResult);
-}
+unsigned int zstd_is_error(size_t code);
 
 /**
- * enum ZSTD_strategy - zstd compression search strategy
- *
- * From faster to stronger.
+ * enum zstd_error_code - zstd error codes
  */
-typedef enum {
-	ZSTD_fast,
-	ZSTD_dfast,
-	ZSTD_greedy,
-	ZSTD_lazy,
-	ZSTD_lazy2,
-	ZSTD_btlazy2,
-	ZSTD_btopt,
-	ZSTD_btopt2
-} ZSTD_strategy;
+typedef ZSTD_ErrorCode zstd_error_code;
 
 /**
- * struct ZSTD_compressionParameters - zstd compression parameters
+ * zstd_get_error_code() - translates an error function result to an error code
+ * @code:  The function result for which zstd_is_error(code) is true.
+ *
+ * Return: A unique error code for this error.
+ */
+zstd_error_code zstd_get_error_code(size_t code);
+
+/**
+ * zstd_get_error_name() - translates an error function result to a string
+ * @code:  The function result for which zstd_is_error(code) is true.
+ *
+ * Return: An error string corresponding to the error code.
+ */
+const char *zstd_get_error_name(size_t code);
+
+/**
+ * zstd_min_clevel() - minimum allowed compression level
+ *
+ * Return: The minimum allowed compression level.
+ */
+int zstd_min_clevel(void);
+
+/**
+ * zstd_max_clevel() - maximum allowed compression level
+ *
+ * Return: The maximum allowed compression level.
+ */
+int zstd_max_clevel(void);
+
+/* ======   Parameter Selection   ====== */
+
+/**
+ * enum zstd_strategy - zstd compression search strategy
+ *
+ * From faster to stronger. See zstd_lib.h.
+ */
+typedef ZSTD_strategy zstd_strategy;
+
+/**
+ * struct zstd_compression_parameters - zstd compression parameters
  * @windowLog:    Log of the largest match distance. Larger means more
  *                compression, and more memory needed during decompression.
- * @chainLog:     Fully searched segment. Larger means more compression, slower,
- *                and more memory (useless for fast).
+ * @chainLog:     Fully searched segment. Larger means more compression,
+ *                slower, and more memory (useless for fast).
  * @hashLog:      Dispatch table. Larger means more compression,
  *                slower, and more memory.
  * @searchLog:    Number of searches. Larger means more compression and slower.
@@ -141,1017 +100,348 @@ typedef enum {
  * @targetLength: Acceptable match size for optimal parser (only). Larger means
  *                more compression, and slower.
  * @strategy:     The zstd compression strategy.
- */
-typedef struct {
-	unsigned int windowLog;
-	unsigned int chainLog;
-	unsigned int hashLog;
-	unsigned int searchLog;
-	unsigned int searchLength;
-	unsigned int targetLength;
-	ZSTD_strategy strategy;
-} ZSTD_compressionParameters;
-
-/**
- * struct ZSTD_frameParameters - zstd frame parameters
- * @contentSizeFlag: Controls whether content size will be present in the frame
- *                   header (when known).
- * @checksumFlag:    Controls whether a 32-bit checksum is generated at the end
- *                   of the frame for error detection.
- * @noDictIDFlag:    Controls whether dictID will be saved into the frame header
- *                   when using dictionary compression.
  *
- * The default value is all fields set to 0.
+ * See zstd_lib.h.
  */
-typedef struct {
-	unsigned int contentSizeFlag;
-	unsigned int checksumFlag;
-	unsigned int noDictIDFlag;
-} ZSTD_frameParameters;
+typedef ZSTD_compressionParameters zstd_compression_parameters;
 
 /**
- * struct ZSTD_parameters - zstd parameters
+ * struct zstd_frame_parameters - zstd frame parameters
+ * @contentSizeFlag: Controls whether content size will be present in the
+ *                   frame header (when known).
+ * @checksumFlag:    Controls whether a 32-bit checksum is generated at the
+ *                   end of the frame for error detection.
+ * @noDictIDFlag:    Controls whether dictID will be saved into the frame
+ *                   header when using dictionary compression.
+ *
+ * The default value is all fields set to 0. See zstd_lib.h.
+ */
+typedef ZSTD_frameParameters zstd_frame_parameters;
+
+/**
+ * struct zstd_parameters - zstd parameters
  * @cParams: The compression parameters.
  * @fParams: The frame parameters.
  */
-typedef struct {
-	ZSTD_compressionParameters cParams;
-	ZSTD_frameParameters fParams;
-} ZSTD_parameters;
+typedef ZSTD_parameters zstd_parameters;
 
 /**
- * ZSTD_getCParams() - returns ZSTD_compressionParameters for selected level
- * @compressionLevel: The compression level from 1 to ZSTD_maxCLevel().
- * @estimatedSrcSize: The estimated source size to compress or 0 if unknown.
- * @dictSize:         The dictionary size or 0 if a dictionary isn't being used.
+ * zstd_get_params() - returns zstd_parameters for selected level
+ * @level:              The compression level
+ * @estimated_src_size: The estimated source size to compress or 0
+ *                      if unknown.
  *
- * Return:            The selected ZSTD_compressionParameters.
+ * Return:              The selected zstd_parameters.
  */
-ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel,
-	unsigned long long estimatedSrcSize, size_t dictSize);
+zstd_parameters zstd_get_params(int level,
+	unsigned long long estimated_src_size);
+
+/* ======   Single-pass Compression   ====== */
+
+typedef ZSTD_CCtx zstd_cctx;
 
 /**
- * ZSTD_getParams() - returns ZSTD_parameters for selected level
- * @compressionLevel: The compression level from 1 to ZSTD_maxCLevel().
- * @estimatedSrcSize: The estimated source size to compress or 0 if unknown.
- * @dictSize:         The dictionary size or 0 if a dictionary isn't being used.
- *
- * The same as ZSTD_getCParams() except also selects the default frame
- * parameters (all zero).
- *
- * Return:            The selected ZSTD_parameters.
- */
-ZSTD_parameters ZSTD_getParams(int compressionLevel,
-	unsigned long long estimatedSrcSize, size_t dictSize);
-
-/*-*************************************
- * Explicit memory management
- **************************************/
-
-/**
- * ZSTD_CCtxWorkspaceBound() - amount of memory needed to initialize a ZSTD_CCtx
- * @cParams: The compression parameters to be used for compression.
+ * zstd_cctx_workspace_bound() - max memory needed to initialize a zstd_cctx
+ * @parameters: The compression parameters to be used.
  *
  * If multiple compression parameters might be used, the caller must call
- * ZSTD_CCtxWorkspaceBound() for each set of parameters and use the maximum
+ * zstd_cctx_workspace_bound() for each set of parameters and use the maximum
  * size.
  *
- * Return:   A lower bound on the size of the workspace that is passed to
- *           ZSTD_initCCtx().
+ * Return:      A lower bound on the size of the workspace that is passed to
+ *              zstd_init_cctx().
  */
-size_t ZSTD_CCtxWorkspaceBound(ZSTD_compressionParameters cParams);
+size_t zstd_cctx_workspace_bound(const zstd_compression_parameters *parameters);
 
 /**
- * struct ZSTD_CCtx - the zstd compression context
+ * zstd_init_cctx() - initialize a zstd compression context
+ * @workspace:      The workspace to emplace the context into. It must outlive
+ *                  the returned context.
+ * @workspace_size: The size of workspace. Use zstd_cctx_workspace_bound() to
+ *                  determine how large the workspace must be.
  *
- * When compressing many times it is recommended to allocate a context just once
- * and reuse it for each successive compression operation.
+ * Return:          A zstd compression context or NULL on error.
  */
-typedef struct ZSTD_CCtx_s ZSTD_CCtx;
-/**
- * ZSTD_initCCtx() - initialize a zstd compression context
- * @workspace:     The workspace to emplace the context into. It must outlive
- *                 the returned context.
- * @workspaceSize: The size of workspace. Use ZSTD_CCtxWorkspaceBound() to
- *                 determine how large the workspace must be.
- *
- * Return:         A compression context emplaced into workspace.
- */
-ZSTD_CCtx *ZSTD_initCCtx(void *workspace, size_t workspaceSize);
+zstd_cctx *zstd_init_cctx(void *workspace, size_t workspace_size);
 
 /**
- * ZSTD_compressCCtx() - compress src into dst
- * @ctx:         The context. Must have been initialized with a workspace at
- *               least as large as ZSTD_CCtxWorkspaceBound(params.cParams).
- * @dst:         The buffer to compress src into.
- * @dstCapacity: The size of the destination buffer. May be any size, but
- *               ZSTD_compressBound(srcSize) is guaranteed to be large enough.
- * @src:         The data to compress.
- * @srcSize:     The size of the data to compress.
- * @params:      The parameters to use for compression. See ZSTD_getParams().
+ * zstd_compress_cctx() - compress src into dst with the initialized parameters
+ * @cctx:         The context. Must have been initialized with zstd_init_cctx().
+ * @dst:          The buffer to compress src into.
+ * @dst_capacity: The size of the destination buffer. May be any size, but
+ *                ZSTD_compressBound(srcSize) is guaranteed to be large enough.
+ * @src:          The data to compress.
+ * @src_size:     The size of the data to compress.
+ * @parameters:   The compression parameters to be used.
  *
- * Return:       The compressed size or an error, which can be checked using
- *               ZSTD_isError().
+ * Return:        The compressed size or an error, which can be checked using
+ *                zstd_is_error().
  */
-size_t ZSTD_compressCCtx(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize, ZSTD_parameters params);
+size_t zstd_compress_cctx(zstd_cctx *cctx, void *dst, size_t dst_capacity,
+	const void *src, size_t src_size, const zstd_parameters *parameters);
+
+/* ======   Single-pass Decompression   ====== */
+
+typedef ZSTD_DCtx zstd_dctx;
 
 /**
- * ZSTD_DCtxWorkspaceBound() - amount of memory needed to initialize a ZSTD_DCtx
+ * zstd_dctx_workspace_bound() - max memory needed to initialize a zstd_dctx
  *
  * Return: A lower bound on the size of the workspace that is passed to
- *         ZSTD_initDCtx().
+ *         zstd_init_dctx().
  */
-size_t ZSTD_DCtxWorkspaceBound(void);
+size_t zstd_dctx_workspace_bound(void);
 
 /**
- * struct ZSTD_DCtx - the zstd decompression context
+ * zstd_init_dctx() - initialize a zstd decompression context
+ * @workspace:      The workspace to emplace the context into. It must outlive
+ *                  the returned context.
+ * @workspace_size: The size of workspace. Use zstd_dctx_workspace_bound() to
+ *                  determine how large the workspace must be.
  *
- * When decompressing many times it is recommended to allocate a context just
- * once and reuse it for each successive decompression operation.
+ * Return:          A zstd decompression context or NULL on error.
  */
-typedef struct ZSTD_DCtx_s ZSTD_DCtx;
-/**
- * ZSTD_initDCtx() - initialize a zstd decompression context
- * @workspace:     The workspace to emplace the context into. It must outlive
- *                 the returned context.
- * @workspaceSize: The size of workspace. Use ZSTD_DCtxWorkspaceBound() to
- *                 determine how large the workspace must be.
- *
- * Return:         A decompression context emplaced into workspace.
- */
-ZSTD_DCtx *ZSTD_initDCtx(void *workspace, size_t workspaceSize);
+zstd_dctx *zstd_init_dctx(void *workspace, size_t workspace_size);
 
 /**
- * ZSTD_decompressDCtx() - decompress zstd compressed src into dst
- * @ctx:         The decompression context.
- * @dst:         The buffer to decompress src into.
- * @dstCapacity: The size of the destination buffer. Must be at least as large
- *               as the decompressed size. If the caller cannot upper bound the
- *               decompressed size, then it's better to use the streaming API.
- * @src:         The zstd compressed data to decompress. Multiple concatenated
- *               frames and skippable frames are allowed.
- * @srcSize:     The exact size of the data to decompress.
+ * zstd_decompress_dctx() - decompress zstd compressed src into dst
+ * @dctx:         The decompression context.
+ * @dst:          The buffer to decompress src into.
+ * @dst_capacity: The size of the destination buffer. Must be at least as large
+ *                as the decompressed size. If the caller cannot upper bound the
+ *                decompressed size, then it's better to use the streaming API.
+ * @src:          The zstd compressed data to decompress. Multiple concatenated
+ *                frames and skippable frames are allowed.
+ * @src_size:     The exact size of the data to decompress.
  *
- * Return:       The decompressed size or an error, which can be checked using
- *               ZSTD_isError().
+ * Return:        The decompressed size or an error, which can be checked using
+ *                zstd_is_error().
  */
-size_t ZSTD_decompressDCtx(ZSTD_DCtx *ctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
+size_t zstd_decompress_dctx(zstd_dctx *dctx, void *dst, size_t dst_capacity,
+	const void *src, size_t src_size);
 
-/*-************************
- * Simple dictionary API
- **************************/
+/* ======   Streaming Buffers   ====== */
 
 /**
- * ZSTD_compress_usingDict() - compress src into dst using a dictionary
- * @ctx:         The context. Must have been initialized with a workspace at
- *               least as large as ZSTD_CCtxWorkspaceBound(params.cParams).
- * @dst:         The buffer to compress src into.
- * @dstCapacity: The size of the destination buffer. May be any size, but
- *               ZSTD_compressBound(srcSize) is guaranteed to be large enough.
- * @src:         The data to compress.
- * @srcSize:     The size of the data to compress.
- * @dict:        The dictionary to use for compression.
- * @dictSize:    The size of the dictionary.
- * @params:      The parameters to use for compression. See ZSTD_getParams().
- *
- * Compression using a predefined dictionary. The same dictionary must be used
- * during decompression.
- *
- * Return:       The compressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_compress_usingDict(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize, const void *dict, size_t dictSize,
-	ZSTD_parameters params);
-
-/**
- * ZSTD_decompress_usingDict() - decompress src into dst using a dictionary
- * @ctx:         The decompression context.
- * @dst:         The buffer to decompress src into.
- * @dstCapacity: The size of the destination buffer. Must be at least as large
- *               as the decompressed size. If the caller cannot upper bound the
- *               decompressed size, then it's better to use the streaming API.
- * @src:         The zstd compressed data to decompress. Multiple concatenated
- *               frames and skippable frames are allowed.
- * @srcSize:     The exact size of the data to decompress.
- * @dict:        The dictionary to use for decompression. The same dictionary
- *               must've been used to compress the data.
- * @dictSize:    The size of the dictionary.
- *
- * Return:       The decompressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_decompress_usingDict(ZSTD_DCtx *ctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize, const void *dict, size_t dictSize);
-
-/*-**************************
- * Fast dictionary API
- ***************************/
-
-/**
- * ZSTD_CDictWorkspaceBound() - memory needed to initialize a ZSTD_CDict
- * @cParams: The compression parameters to be used for compression.
- *
- * Return:   A lower bound on the size of the workspace that is passed to
- *           ZSTD_initCDict().
- */
-size_t ZSTD_CDictWorkspaceBound(ZSTD_compressionParameters cParams);
-
-/**
- * struct ZSTD_CDict - a digested dictionary to be used for compression
- */
-typedef struct ZSTD_CDict_s ZSTD_CDict;
-
-/**
- * ZSTD_initCDict() - initialize a digested dictionary for compression
- * @dictBuffer:    The dictionary to digest. The buffer is referenced by the
- *                 ZSTD_CDict so it must outlive the returned ZSTD_CDict.
- * @dictSize:      The size of the dictionary.
- * @params:        The parameters to use for compression. See ZSTD_getParams().
- * @workspace:     The workspace. It must outlive the returned ZSTD_CDict.
- * @workspaceSize: The workspace size. Must be at least
- *                 ZSTD_CDictWorkspaceBound(params.cParams).
- *
- * When compressing multiple messages / blocks with the same dictionary it is
- * recommended to load it just once. The ZSTD_CDict merely references the
- * dictBuffer, so it must outlive the returned ZSTD_CDict.
- *
- * Return:         The digested dictionary emplaced into workspace.
- */
-ZSTD_CDict *ZSTD_initCDict(const void *dictBuffer, size_t dictSize,
-	ZSTD_parameters params, void *workspace, size_t workspaceSize);
-
-/**
- * ZSTD_compress_usingCDict() - compress src into dst using a ZSTD_CDict
- * @ctx:         The context. Must have been initialized with a workspace at
- *               least as large as ZSTD_CCtxWorkspaceBound(cParams) where
- *               cParams are the compression parameters used to initialize the
- *               cdict.
- * @dst:         The buffer to compress src into.
- * @dstCapacity: The size of the destination buffer. May be any size, but
- *               ZSTD_compressBound(srcSize) is guaranteed to be large enough.
- * @src:         The data to compress.
- * @srcSize:     The size of the data to compress.
- * @cdict:       The digested dictionary to use for compression.
- * @params:      The parameters to use for compression. See ZSTD_getParams().
- *
- * Compression using a digested dictionary. The same dictionary must be used
- * during decompression.
- *
- * Return:       The compressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_compress_usingCDict(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize, const ZSTD_CDict *cdict);
-
-
-/**
- * ZSTD_DDictWorkspaceBound() - memory needed to initialize a ZSTD_DDict
- *
- * Return:  A lower bound on the size of the workspace that is passed to
- *          ZSTD_initDDict().
- */
-size_t ZSTD_DDictWorkspaceBound(void);
-
-/**
- * struct ZSTD_DDict - a digested dictionary to be used for decompression
- */
-typedef struct ZSTD_DDict_s ZSTD_DDict;
-
-/**
- * ZSTD_initDDict() - initialize a digested dictionary for decompression
- * @dictBuffer:    The dictionary to digest. The buffer is referenced by the
- *                 ZSTD_DDict so it must outlive the returned ZSTD_DDict.
- * @dictSize:      The size of the dictionary.
- * @workspace:     The workspace. It must outlive the returned ZSTD_DDict.
- * @workspaceSize: The workspace size. Must be at least
- *                 ZSTD_DDictWorkspaceBound().
- *
- * When decompressing multiple messages / blocks with the same dictionary it is
- * recommended to load it just once. The ZSTD_DDict merely references the
- * dictBuffer, so it must outlive the returned ZSTD_DDict.
- *
- * Return:         The digested dictionary emplaced into workspace.
- */
-ZSTD_DDict *ZSTD_initDDict(const void *dictBuffer, size_t dictSize,
-	void *workspace, size_t workspaceSize);
-
-/**
- * ZSTD_decompress_usingDDict() - decompress src into dst using a ZSTD_DDict
- * @ctx:         The decompression context.
- * @dst:         The buffer to decompress src into.
- * @dstCapacity: The size of the destination buffer. Must be at least as large
- *               as the decompressed size. If the caller cannot upper bound the
- *               decompressed size, then it's better to use the streaming API.
- * @src:         The zstd compressed data to decompress. Multiple concatenated
- *               frames and skippable frames are allowed.
- * @srcSize:     The exact size of the data to decompress.
- * @ddict:       The digested dictionary to use for decompression. The same
- *               dictionary must've been used to compress the data.
- *
- * Return:       The decompressed size or an error, which can be checked using
- *               ZSTD_isError().
- */
-size_t ZSTD_decompress_usingDDict(ZSTD_DCtx *dctx, void *dst,
-	size_t dstCapacity, const void *src, size_t srcSize,
-	const ZSTD_DDict *ddict);
-
-
-/*-**************************
- * Streaming
- ***************************/
-
-/**
- * struct ZSTD_inBuffer - input buffer for streaming
+ * struct zstd_in_buffer - input buffer for streaming
  * @src:  Start of the input buffer.
  * @size: Size of the input buffer.
  * @pos:  Position where reading stopped. Will be updated.
  *        Necessarily 0 <= pos <= size.
+ *
+ * See zstd_lib.h.
  */
-typedef struct ZSTD_inBuffer_s {
-	const void *src;
-	size_t size;
-	size_t pos;
-} ZSTD_inBuffer;
+typedef ZSTD_inBuffer zstd_in_buffer;
 
 /**
- * struct ZSTD_outBuffer - output buffer for streaming
+ * struct zstd_out_buffer - output buffer for streaming
  * @dst:  Start of the output buffer.
  * @size: Size of the output buffer.
  * @pos:  Position where writing stopped. Will be updated.
  *        Necessarily 0 <= pos <= size.
+ *
+ * See zstd_lib.h.
  */
-typedef struct ZSTD_outBuffer_s {
-	void *dst;
-	size_t size;
-	size_t pos;
-} ZSTD_outBuffer;
+typedef ZSTD_outBuffer zstd_out_buffer;
 
+/* ======   Streaming Compression   ====== */
 
-
-/*-*****************************************************************************
- * Streaming compression - HowTo
- *
- * A ZSTD_CStream object is required to track streaming operation.
- * Use ZSTD_initCStream() to initialize a ZSTD_CStream object.
- * ZSTD_CStream objects can be reused multiple times on consecutive compression
- * operations. It is recommended to re-use ZSTD_CStream in situations where many
- * streaming operations will be achieved consecutively. Use one separate
- * ZSTD_CStream per thread for parallel execution.
- *
- * Use ZSTD_compressStream() repetitively to consume input stream.
- * The function will automatically update both `pos` fields.
- * Note that it may not consume the entire input, in which case `pos < size`,
- * and it's up to the caller to present again remaining data.
- * It returns a hint for the preferred number of bytes to use as an input for
- * the next function call.
- *
- * At any moment, it's possible to flush whatever data remains within internal
- * buffer, using ZSTD_flushStream(). `output->pos` will be updated. There might
- * still be some content left within the internal buffer if `output->size` is
- * too small. It returns the number of bytes left in the internal buffer and
- * must be called until it returns 0.
- *
- * ZSTD_endStream() instructs to finish a frame. It will perform a flush and
- * write frame epilogue. The epilogue is required for decoders to consider a
- * frame completed. Similar to ZSTD_flushStream(), it may not be able to flush
- * the full content if `output->size` is too small. In which case, call again
- * ZSTD_endStream() to complete the flush. It returns the number of bytes left
- * in the internal buffer and must be called until it returns 0.
- ******************************************************************************/
+typedef ZSTD_CStream zstd_cstream;
 
 /**
- * ZSTD_CStreamWorkspaceBound() - memory needed to initialize a ZSTD_CStream
- * @cParams: The compression parameters to be used for compression.
+ * zstd_cstream_workspace_bound() - memory needed to initialize a zstd_cstream
+ * @cparams: The compression parameters to be used for compression.
  *
  * Return:   A lower bound on the size of the workspace that is passed to
- *           ZSTD_initCStream() and ZSTD_initCStream_usingCDict().
+ *           zstd_init_cstream().
  */
-size_t ZSTD_CStreamWorkspaceBound(ZSTD_compressionParameters cParams);
+size_t zstd_cstream_workspace_bound(const zstd_compression_parameters *cparams);
 
 /**
- * struct ZSTD_CStream - the zstd streaming compression context
- */
-typedef struct ZSTD_CStream_s ZSTD_CStream;
-
-/*===== ZSTD_CStream management functions =====*/
-/**
- * ZSTD_initCStream() - initialize a zstd streaming compression context
- * @params:         The zstd compression parameters.
- * @pledgedSrcSize: If params.fParams.contentSizeFlag == 1 then the caller must
- *                  pass the source size (zero means empty source). Otherwise,
- *                  the caller may optionally pass the source size, or zero if
- *                  unknown.
- * @workspace:      The workspace to emplace the context into. It must outlive
- *                  the returned context.
- * @workspaceSize:  The size of workspace.
- *                  Use ZSTD_CStreamWorkspaceBound(params.cParams) to determine
- *                  how large the workspace must be.
+ * zstd_init_cstream() - initialize a zstd streaming compression context
+ * @parameters        The zstd parameters to use for compression.
+ * @pledged_src_size: If params.fParams.contentSizeFlag == 1 then the caller
+ *                    must pass the source size (zero means empty source).
+ *                    Otherwise, the caller may optionally pass the source
+ *                    size, or zero if unknown.
+ * @workspace:        The workspace to emplace the context into. It must outlive
+ *                    the returned context.
+ * @workspace_size:   The size of workspace.
+ *                    Use zstd_cstream_workspace_bound(params->cparams) to
+ *                    determine how large the workspace must be.
  *
- * Return:          The zstd streaming compression context.
+ * Return:            The zstd streaming compression context or NULL on error.
  */
-ZSTD_CStream *ZSTD_initCStream(ZSTD_parameters params,
-	unsigned long long pledgedSrcSize, void *workspace,
-	size_t workspaceSize);
+zstd_cstream *zstd_init_cstream(const zstd_parameters *parameters,
+	unsigned long long pledged_src_size, void *workspace, size_t workspace_size);
 
 /**
- * ZSTD_initCStream_usingCDict() - initialize a streaming compression context
- * @cdict:          The digested dictionary to use for compression.
- * @pledgedSrcSize: Optionally the source size, or zero if unknown.
- * @workspace:      The workspace to emplace the context into. It must outlive
- *                  the returned context.
- * @workspaceSize:  The size of workspace. Call ZSTD_CStreamWorkspaceBound()
- *                  with the cParams used to initialize the cdict to determine
- *                  how large the workspace must be.
- *
- * Return:          The zstd streaming compression context.
- */
-ZSTD_CStream *ZSTD_initCStream_usingCDict(const ZSTD_CDict *cdict,
-	unsigned long long pledgedSrcSize, void *workspace,
-	size_t workspaceSize);
-
-/*===== Streaming compression functions =====*/
-/**
- * ZSTD_resetCStream() - reset the context using parameters from creation
- * @zcs:            The zstd streaming compression context to reset.
- * @pledgedSrcSize: Optionally the source size, or zero if unknown.
+ * zstd_reset_cstream() - reset the context using parameters from creation
+ * @cstream:          The zstd streaming compression context to reset.
+ * @pledged_src_size: Optionally the source size, or zero if unknown.
  *
  * Resets the context using the parameters from creation. Skips dictionary
- * loading, since it can be reused. If `pledgedSrcSize` is non-zero the frame
+ * loading, since it can be reused. If `pledged_src_size` is non-zero the frame
  * content size is always written into the frame header.
  *
- * Return:          Zero or an error, which can be checked using ZSTD_isError().
+ * Return:            Zero or an error, which can be checked using
+ *                    zstd_is_error().
  */
-size_t ZSTD_resetCStream(ZSTD_CStream *zcs, unsigned long long pledgedSrcSize);
+size_t zstd_reset_cstream(zstd_cstream *cstream,
+	unsigned long long pledged_src_size);
+
 /**
- * ZSTD_compressStream() - streaming compress some of input into output
- * @zcs:    The zstd streaming compression context.
- * @output: Destination buffer. `output->pos` is updated to indicate how much
- *          compressed data was written.
- * @input:  Source buffer. `input->pos` is updated to indicate how much data was
- *          read. Note that it may not consume the entire input, in which case
- *          `input->pos < input->size`, and it's up to the caller to present
- *          remaining data again.
+ * zstd_compress_stream() - streaming compress some of input into output
+ * @cstream: The zstd streaming compression context.
+ * @output:  Destination buffer. `output->pos` is updated to indicate how much
+ *           compressed data was written.
+ * @input:   Source buffer. `input->pos` is updated to indicate how much data
+ *           was read. Note that it may not consume the entire input, in which
+ *           case `input->pos < input->size`, and it's up to the caller to
+ *           present remaining data again.
  *
  * The `input` and `output` buffers may be any size. Guaranteed to make some
  * forward progress if `input` and `output` are not empty.
  *
- * Return:  A hint for the number of bytes to use as the input for the next
- *          function call or an error, which can be checked using
- *          ZSTD_isError().
+ * Return:   A hint for the number of bytes to use as the input for the next
+ *           function call or an error, which can be checked using
+ *           zstd_is_error().
  */
-size_t ZSTD_compressStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output,
-	ZSTD_inBuffer *input);
+size_t zstd_compress_stream(zstd_cstream *cstream, zstd_out_buffer *output,
+	zstd_in_buffer *input);
+
 /**
- * ZSTD_flushStream() - flush internal buffers into output
- * @zcs:    The zstd streaming compression context.
- * @output: Destination buffer. `output->pos` is updated to indicate how much
- *          compressed data was written.
+ * zstd_flush_stream() - flush internal buffers into output
+ * @cstream: The zstd streaming compression context.
+ * @output:  Destination buffer. `output->pos` is updated to indicate how much
+ *           compressed data was written.
  *
- * ZSTD_flushStream() must be called until it returns 0, meaning all the data
- * has been flushed. Since ZSTD_flushStream() causes a block to be ended,
+ * zstd_flush_stream() must be called until it returns 0, meaning all the data
+ * has been flushed. Since zstd_flush_stream() causes a block to be ended,
  * calling it too often will degrade the compression ratio.
  *
- * Return:  The number of bytes still present within internal buffers or an
- *          error, which can be checked using ZSTD_isError().
+ * Return:   The number of bytes still present within internal buffers or an
+ *           error, which can be checked using zstd_is_error().
  */
-size_t ZSTD_flushStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output);
+size_t zstd_flush_stream(zstd_cstream *cstream, zstd_out_buffer *output);
+
 /**
- * ZSTD_endStream() - flush internal buffers into output and end the frame
- * @zcs:    The zstd streaming compression context.
- * @output: Destination buffer. `output->pos` is updated to indicate how much
- *          compressed data was written.
+ * zstd_end_stream() - flush internal buffers into output and end the frame
+ * @cstream: The zstd streaming compression context.
+ * @output:  Destination buffer. `output->pos` is updated to indicate how much
+ *           compressed data was written.
  *
- * ZSTD_endStream() must be called until it returns 0, meaning all the data has
+ * zstd_end_stream() must be called until it returns 0, meaning all the data has
  * been flushed and the frame epilogue has been written.
  *
- * Return:  The number of bytes still present within internal buffers or an
- *          error, which can be checked using ZSTD_isError().
+ * Return:   The number of bytes still present within internal buffers or an
+ *           error, which can be checked using zstd_is_error().
  */
-size_t ZSTD_endStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output);
+size_t zstd_end_stream(zstd_cstream *cstream, zstd_out_buffer *output);
+
+/* ======   Streaming Decompression   ====== */
+
+typedef ZSTD_DStream zstd_dstream;
 
 /**
- * ZSTD_CStreamInSize() - recommended size for the input buffer
+ * zstd_dstream_workspace_bound() - memory needed to initialize a zstd_dstream
+ * @max_window_size: The maximum window size allowed for compressed frames.
  *
- * Return: The recommended size for the input buffer.
+ * Return:           A lower bound on the size of the workspace that is passed
+ *                   to zstd_init_dstream().
  */
-size_t ZSTD_CStreamInSize(void);
-/**
- * ZSTD_CStreamOutSize() - recommended size for the output buffer
- *
- * When the output buffer is at least this large, it is guaranteed to be large
- * enough to flush at least one complete compressed block.
- *
- * Return: The recommended size for the output buffer.
- */
-size_t ZSTD_CStreamOutSize(void);
-
-
-
-/*-*****************************************************************************
- * Streaming decompression - HowTo
- *
- * A ZSTD_DStream object is required to track streaming operations.
- * Use ZSTD_initDStream() to initialize a ZSTD_DStream object.
- * ZSTD_DStream objects can be re-used multiple times.
- *
- * Use ZSTD_decompressStream() repetitively to consume your input.
- * The function will update both `pos` fields.
- * If `input->pos < input->size`, some input has not been consumed.
- * It's up to the caller to present again remaining data.
- * If `output->pos < output->size`, decoder has flushed everything it could.
- * Returns 0 iff a frame is completely decoded and fully flushed.
- * Otherwise it returns a suggested next input size that will never load more
- * than the current frame.
- ******************************************************************************/
+size_t zstd_dstream_workspace_bound(size_t max_window_size);
 
 /**
- * ZSTD_DStreamWorkspaceBound() - memory needed to initialize a ZSTD_DStream
- * @maxWindowSize: The maximum window size allowed for compressed frames.
+ * zstd_init_dstream() - initialize a zstd streaming decompression context
+ * @max_window_size: The maximum window size allowed for compressed frames.
+ * @workspace:       The workspace to emplace the context into. It must outlive
+ *                   the returned context.
+ * @workspaceSize:   The size of workspace.
+ *                   Use zstd_dstream_workspace_bound(max_window_size) to
+ *                   determine how large the workspace must be.
  *
- * Return:         A lower bound on the size of the workspace that is passed to
- *                 ZSTD_initDStream() and ZSTD_initDStream_usingDDict().
+ * Return:           The zstd streaming decompression context.
  */
-size_t ZSTD_DStreamWorkspaceBound(size_t maxWindowSize);
+zstd_dstream *zstd_init_dstream(size_t max_window_size, void *workspace,
+	size_t workspace_size);
 
 /**
- * struct ZSTD_DStream - the zstd streaming decompression context
- */
-typedef struct ZSTD_DStream_s ZSTD_DStream;
-/*===== ZSTD_DStream management functions =====*/
-/**
- * ZSTD_initDStream() - initialize a zstd streaming decompression context
- * @maxWindowSize: The maximum window size allowed for compressed frames.
- * @workspace:     The workspace to emplace the context into. It must outlive
- *                 the returned context.
- * @workspaceSize: The size of workspace.
- *                 Use ZSTD_DStreamWorkspaceBound(maxWindowSize) to determine
- *                 how large the workspace must be.
- *
- * Return:         The zstd streaming decompression context.
- */
-ZSTD_DStream *ZSTD_initDStream(size_t maxWindowSize, void *workspace,
-	size_t workspaceSize);
-/**
- * ZSTD_initDStream_usingDDict() - initialize streaming decompression context
- * @maxWindowSize: The maximum window size allowed for compressed frames.
- * @ddict:         The digested dictionary to use for decompression.
- * @workspace:     The workspace to emplace the context into. It must outlive
- *                 the returned context.
- * @workspaceSize: The size of workspace.
- *                 Use ZSTD_DStreamWorkspaceBound(maxWindowSize) to determine
- *                 how large the workspace must be.
- *
- * Return:         The zstd streaming decompression context.
- */
-ZSTD_DStream *ZSTD_initDStream_usingDDict(size_t maxWindowSize,
-	const ZSTD_DDict *ddict, void *workspace, size_t workspaceSize);
-
-/*===== Streaming decompression functions =====*/
-/**
- * ZSTD_resetDStream() - reset the context using parameters from creation
- * @zds:   The zstd streaming decompression context to reset.
+ * zstd_reset_dstream() - reset the context using parameters from creation
+ * @dstream: The zstd streaming decompression context to reset.
  *
  * Resets the context using the parameters from creation. Skips dictionary
  * loading, since it can be reused.
  *
- * Return: Zero or an error, which can be checked using ZSTD_isError().
+ * Return:   Zero or an error, which can be checked using zstd_is_error().
  */
-size_t ZSTD_resetDStream(ZSTD_DStream *zds);
+size_t zstd_reset_dstream(zstd_dstream *dstream);
+
 /**
- * ZSTD_decompressStream() - streaming decompress some of input into output
- * @zds:    The zstd streaming decompression context.
- * @output: Destination buffer. `output.pos` is updated to indicate how much
- *          decompressed data was written.
- * @input:  Source buffer. `input.pos` is updated to indicate how much data was
- *          read. Note that it may not consume the entire input, in which case
- *          `input.pos < input.size`, and it's up to the caller to present
- *          remaining data again.
+ * zstd_decompress_stream() - streaming decompress some of input into output
+ * @dstream: The zstd streaming decompression context.
+ * @output:  Destination buffer. `output.pos` is updated to indicate how much
+ *           decompressed data was written.
+ * @input:   Source buffer. `input.pos` is updated to indicate how much data was
+ *           read. Note that it may not consume the entire input, in which case
+ *           `input.pos < input.size`, and it's up to the caller to present
+ *           remaining data again.
  *
  * The `input` and `output` buffers may be any size. Guaranteed to make some
  * forward progress if `input` and `output` are not empty.
- * ZSTD_decompressStream() will not consume the last byte of the frame until
+ * zstd_decompress_stream() will not consume the last byte of the frame until
  * the entire frame is flushed.
  *
- * Return:  Returns 0 iff a frame is completely decoded and fully flushed.
- *          Otherwise returns a hint for the number of bytes to use as the input
- *          for the next function call or an error, which can be checked using
- *          ZSTD_isError(). The size hint will never load more than the frame.
+ * Return:   Returns 0 iff a frame is completely decoded and fully flushed.
+ *           Otherwise returns a hint for the number of bytes to use as the
+ *           input for the next function call or an error, which can be checked
+ *           using zstd_is_error(). The size hint will never load more than the
+ *           frame.
  */
-size_t ZSTD_decompressStream(ZSTD_DStream *zds, ZSTD_outBuffer *output,
-	ZSTD_inBuffer *input);
+size_t zstd_decompress_stream(zstd_dstream *dstream, zstd_out_buffer *output,
+	zstd_in_buffer *input);
+
+/* ======   Frame Inspection Functions ====== */
 
 /**
- * ZSTD_DStreamInSize() - recommended size for the input buffer
+ * zstd_find_frame_compressed_size() - returns the size of a compressed frame
+ * @src:      Source buffer. It should point to the start of a zstd encoded
+ *            frame or a skippable frame.
+ * @src_size: The size of the source buffer. It must be at least as large as the
+ *            size of the frame.
  *
- * Return: The recommended size for the input buffer.
+ * Return:    The compressed size of the frame pointed to by `src` or an error,
+ *            which can be check with zstd_is_error().
+ *            Suitable to pass to ZSTD_decompress() or similar functions.
  */
-size_t ZSTD_DStreamInSize(void);
-/**
- * ZSTD_DStreamOutSize() - recommended size for the output buffer
- *
- * When the output buffer is at least this large, it is guaranteed to be large
- * enough to flush at least one complete decompressed block.
- *
- * Return: The recommended size for the output buffer.
- */
-size_t ZSTD_DStreamOutSize(void);
-
-
-/* --- Constants ---*/
-#define ZSTD_MAGICNUMBER            0xFD2FB528   /* >= v0.8.0 */
-#define ZSTD_MAGIC_SKIPPABLE_START  0x184D2A50U
-
-#define ZSTD_CONTENTSIZE_UNKNOWN (0ULL - 1)
-#define ZSTD_CONTENTSIZE_ERROR   (0ULL - 2)
-
-#define ZSTD_WINDOWLOG_MAX_32  27
-#define ZSTD_WINDOWLOG_MAX_64  27
-#define ZSTD_WINDOWLOG_MAX \
-	((unsigned int)(sizeof(size_t) == 4 \
-		? ZSTD_WINDOWLOG_MAX_32 \
-		: ZSTD_WINDOWLOG_MAX_64))
-#define ZSTD_WINDOWLOG_MIN 10
-#define ZSTD_HASHLOG_MAX ZSTD_WINDOWLOG_MAX
-#define ZSTD_HASHLOG_MIN        6
-#define ZSTD_CHAINLOG_MAX     (ZSTD_WINDOWLOG_MAX+1)
-#define ZSTD_CHAINLOG_MIN      ZSTD_HASHLOG_MIN
-#define ZSTD_HASHLOG3_MAX      17
-#define ZSTD_SEARCHLOG_MAX    (ZSTD_WINDOWLOG_MAX-1)
-#define ZSTD_SEARCHLOG_MIN      1
-/* only for ZSTD_fast, other strategies are limited to 6 */
-#define ZSTD_SEARCHLENGTH_MAX   7
-/* only for ZSTD_btopt, other strategies are limited to 4 */
-#define ZSTD_SEARCHLENGTH_MIN   3
-#define ZSTD_TARGETLENGTH_MIN   4
-#define ZSTD_TARGETLENGTH_MAX 999
-
-/* for static allocation */
-#define ZSTD_FRAMEHEADERSIZE_MAX 18
-#define ZSTD_FRAMEHEADERSIZE_MIN  6
-#define ZSTD_frameHeaderSize_prefix 5
-#define ZSTD_frameHeaderSize_min ZSTD_FRAMEHEADERSIZE_MIN
-#define ZSTD_frameHeaderSize_max ZSTD_FRAMEHEADERSIZE_MAX
-/* magic number + skippable frame length */
-#define ZSTD_skippableHeaderSize 8
-
-
-/*-*************************************
- * Compressed size functions
- **************************************/
+size_t zstd_find_frame_compressed_size(const void *src, size_t src_size);
 
 /**
- * ZSTD_findFrameCompressedSize() - returns the size of a compressed frame
- * @src:     Source buffer. It should point to the start of a zstd encoded frame
- *           or a skippable frame.
- * @srcSize: The size of the source buffer. It must be at least as large as the
- *           size of the frame.
- *
- * Return:   The compressed size of the frame pointed to by `src` or an error,
- *           which can be check with ZSTD_isError().
- *           Suitable to pass to ZSTD_decompress() or similar functions.
- */
-size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize);
-
-/*-*************************************
- * Decompressed size functions
- **************************************/
-/**
- * ZSTD_getFrameContentSize() - returns the content size in a zstd frame header
- * @src:     It should point to the start of a zstd encoded frame.
- * @srcSize: The size of the source buffer. It must be at least as large as the
- *           frame header. `ZSTD_frameHeaderSize_max` is always large enough.
- *
- * Return:   The frame content size stored in the frame header if known.
- *           `ZSTD_CONTENTSIZE_UNKNOWN` if the content size isn't stored in the
- *           frame header. `ZSTD_CONTENTSIZE_ERROR` on invalid input.
- */
-unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize);
-
-/**
- * ZSTD_findDecompressedSize() - returns decompressed size of a series of frames
- * @src:     It should point to the start of a series of zstd encoded and/or
- *           skippable frames.
- * @srcSize: The exact size of the series of frames.
- *
- * If any zstd encoded frame in the series doesn't have the frame content size
- * set, `ZSTD_CONTENTSIZE_UNKNOWN` is returned. But frame content size is always
- * set when using ZSTD_compress(). The decompressed size can be very large.
- * If the source is untrusted, the decompressed size could be wrong or
- * intentionally modified. Always ensure the result fits within the
- * application's authorized limits. ZSTD_findDecompressedSize() handles multiple
- * frames, and so it must traverse the input to read each frame header. This is
- * efficient as most of the data is skipped, however it does mean that all frame
- * data must be present and valid.
- *
- * Return:   Decompressed size of all the data contained in the frames if known.
- *           `ZSTD_CONTENTSIZE_UNKNOWN` if the decompressed size is unknown.
- *           `ZSTD_CONTENTSIZE_ERROR` if an error occurred.
- */
-unsigned long long ZSTD_findDecompressedSize(const void *src, size_t srcSize);
-
-/*-*************************************
- * Advanced compression functions
- **************************************/
-/**
- * ZSTD_checkCParams() - ensure parameter values remain within authorized range
- * @cParams: The zstd compression parameters.
- *
- * Return:   Zero or an error, which can be checked using ZSTD_isError().
- */
-size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams);
-
-/**
- * ZSTD_adjustCParams() - optimize parameters for a given srcSize and dictSize
- * @srcSize:  Optionally the estimated source size, or zero if unknown.
- * @dictSize: Optionally the estimated dictionary size, or zero if unknown.
- *
- * Return:    The optimized parameters.
- */
-ZSTD_compressionParameters ZSTD_adjustCParams(
-	ZSTD_compressionParameters cParams, unsigned long long srcSize,
-	size_t dictSize);
-
-/*--- Advanced decompression functions ---*/
-
-/**
- * ZSTD_isFrame() - returns true iff the buffer starts with a valid frame
- * @buffer: The source buffer to check.
- * @size:   The size of the source buffer, must be at least 4 bytes.
- *
- * Return: True iff the buffer starts with a zstd or skippable frame identifier.
- */
-unsigned int ZSTD_isFrame(const void *buffer, size_t size);
-
-/**
- * ZSTD_getDictID_fromDict() - returns the dictionary id stored in a dictionary
- * @dict:     The dictionary buffer.
- * @dictSize: The size of the dictionary buffer.
- *
- * Return:    The dictionary id stored within the dictionary or 0 if the
- *            dictionary is not a zstd dictionary. If it returns 0 the
- *            dictionary can still be loaded as a content-only dictionary.
- */
-unsigned int ZSTD_getDictID_fromDict(const void *dict, size_t dictSize);
-
-/**
- * ZSTD_getDictID_fromDDict() - returns the dictionary id stored in a ZSTD_DDict
- * @ddict: The ddict to find the id of.
- *
- * Return: The dictionary id stored within `ddict` or 0 if the dictionary is not
- *         a zstd dictionary. If it returns 0 `ddict` will be loaded as a
- *         content-only dictionary.
- */
-unsigned int ZSTD_getDictID_fromDDict(const ZSTD_DDict *ddict);
-
-/**
- * ZSTD_getDictID_fromFrame() - returns the dictionary id stored in a zstd frame
- * @src:     Source buffer. It must be a zstd encoded frame.
- * @srcSize: The size of the source buffer. It must be at least as large as the
- *           frame header. `ZSTD_frameHeaderSize_max` is always large enough.
- *
- * Return:   The dictionary id required to decompress the frame stored within
- *           `src` or 0 if the dictionary id could not be decoded. It can return
- *           0 if the frame does not require a dictionary, the dictionary id
- *           wasn't stored in the frame, `src` is not a zstd frame, or `srcSize`
- *           is too small.
- */
-unsigned int ZSTD_getDictID_fromFrame(const void *src, size_t srcSize);
-
-/**
- * struct ZSTD_frameParams - zstd frame parameters stored in the frame header
- * @frameContentSize: The frame content size, or 0 if not present.
+ * struct zstd_frame_params - zstd frame parameters stored in the frame header
+ * @frameContentSize: The frame content size, or ZSTD_CONTENTSIZE_UNKNOWN if not
+ *                    present.
  * @windowSize:       The window size, or 0 if the frame is a skippable frame.
+ * @blockSizeMax:     The maximum block size.
+ * @frameType:        The frame type (zstd or skippable)
+ * @headerSize:       The size of the frame header.
  * @dictID:           The dictionary id, or 0 if not present.
  * @checksumFlag:     Whether a checksum was used.
+ *
+ * See zstd_lib.h.
  */
-typedef struct {
-	unsigned long long frameContentSize;
-	unsigned int windowSize;
-	unsigned int dictID;
-	unsigned int checksumFlag;
-} ZSTD_frameParams;
+typedef ZSTD_frameHeader zstd_frame_header;
 
 /**
- * ZSTD_getFrameParams() - extracts parameters from a zstd or skippable frame
- * @fparamsPtr: On success the frame parameters are written here.
- * @src:        The source buffer. It must point to a zstd or skippable frame.
- * @srcSize:    The size of the source buffer. `ZSTD_frameHeaderSize_max` is
- *              always large enough to succeed.
+ * zstd_get_frame_header() - extracts parameters from a zstd or skippable frame
+ * @params:   On success the frame parameters are written here.
+ * @src:      The source buffer. It must point to a zstd or skippable frame.
+ * @src_size: The size of the source buffer.
  *
- * Return:      0 on success. If more data is required it returns how many bytes
- *              must be provided to make forward progress. Otherwise it returns
- *              an error, which can be checked using ZSTD_isError().
+ * Return:    0 on success. If more data is required it returns how many bytes
+ *            must be provided to make forward progress. Otherwise it returns
+ *            an error, which can be checked using zstd_is_error().
  */
-size_t ZSTD_getFrameParams(ZSTD_frameParams *fparamsPtr, const void *src,
-	size_t srcSize);
+size_t zstd_get_frame_header(zstd_frame_header *params, const void *src,
+	size_t src_size);
 
-/*-*****************************************************************************
- * Buffer-less and synchronous inner streaming functions
- *
- * This is an advanced API, giving full control over buffer management, for
- * users which need direct control over memory.
- * But it's also a complex one, with many restrictions (documented below).
- * Prefer using normal streaming API for an easier experience
- ******************************************************************************/
-
-/*-*****************************************************************************
- * Buffer-less streaming compression (synchronous mode)
- *
- * A ZSTD_CCtx object is required to track streaming operations.
- * Use ZSTD_initCCtx() to initialize a context.
- * ZSTD_CCtx object can be re-used multiple times within successive compression
- * operations.
- *
- * Start by initializing a context.
- * Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary
- * compression,
- * or ZSTD_compressBegin_advanced(), for finer parameter control.
- * It's also possible to duplicate a reference context which has already been
- * initialized, using ZSTD_copyCCtx()
- *
- * Then, consume your input using ZSTD_compressContinue().
- * There are some important considerations to keep in mind when using this
- * advanced function :
- * - ZSTD_compressContinue() has no internal buffer. It uses externally provided
- *   buffer only.
- * - Interface is synchronous : input is consumed entirely and produce 1+
- *   (or more) compressed blocks.
- * - Caller must ensure there is enough space in `dst` to store compressed data
- *   under worst case scenario. Worst case evaluation is provided by
- *   ZSTD_compressBound().
- *   ZSTD_compressContinue() doesn't guarantee recover after a failed
- *   compression.
- * - ZSTD_compressContinue() presumes prior input ***is still accessible and
- *   unmodified*** (up to maximum distance size, see WindowLog).
- *   It remembers all previous contiguous blocks, plus one separated memory
- *   segment (which can itself consists of multiple contiguous blocks)
- * - ZSTD_compressContinue() detects that prior input has been overwritten when
- *   `src` buffer overlaps. In which case, it will "discard" the relevant memory
- *   section from its history.
- *
- * Finish a frame with ZSTD_compressEnd(), which will write the last block(s)
- * and optional checksum. It's possible to use srcSize==0, in which case, it
- * will write a final empty block to end the frame. Without last block mark,
- * frames will be considered unfinished (corrupted) by decoders.
- *
- * `ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress some new
- * frame.
- ******************************************************************************/
-
-/*=====   Buffer-less streaming compression functions  =====*/
-size_t ZSTD_compressBegin(ZSTD_CCtx *cctx, int compressionLevel);
-size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx *cctx, const void *dict,
-	size_t dictSize, int compressionLevel);
-size_t ZSTD_compressBegin_advanced(ZSTD_CCtx *cctx, const void *dict,
-	size_t dictSize, ZSTD_parameters params,
-	unsigned long long pledgedSrcSize);
-size_t ZSTD_copyCCtx(ZSTD_CCtx *cctx, const ZSTD_CCtx *preparedCCtx,
-	unsigned long long pledgedSrcSize);
-size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx *cctx, const ZSTD_CDict *cdict,
-	unsigned long long pledgedSrcSize);
-size_t ZSTD_compressContinue(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-size_t ZSTD_compressEnd(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-
-
-
-/*-*****************************************************************************
- * Buffer-less streaming decompression (synchronous mode)
- *
- * A ZSTD_DCtx object is required to track streaming operations.
- * Use ZSTD_initDCtx() to initialize a context.
- * A ZSTD_DCtx object can be re-used multiple times.
- *
- * First typical operation is to retrieve frame parameters, using
- * ZSTD_getFrameParams(). It fills a ZSTD_frameParams structure which provide
- * important information to correctly decode the frame, such as the minimum
- * rolling buffer size to allocate to decompress data (`windowSize`), and the
- * dictionary ID used.
- * Note: content size is optional, it may not be present. 0 means unknown.
- * Note that these values could be wrong, either because of data malformation,
- * or because an attacker is spoofing deliberate false information. As a
- * consequence, check that values remain within valid application range,
- * especially `windowSize`, before allocation. Each application can set its own
- * limit, depending on local restrictions. For extended interoperability, it is
- * recommended to support at least 8 MB.
- * Frame parameters are extracted from the beginning of the compressed frame.
- * Data fragment must be large enough to ensure successful decoding, typically
- * `ZSTD_frameHeaderSize_max` bytes.
- * Result: 0: successful decoding, the `ZSTD_frameParams` structure is filled.
- *        >0: `srcSize` is too small, provide at least this many bytes.
- *        errorCode, which can be tested using ZSTD_isError().
- *
- * Start decompression, with ZSTD_decompressBegin() or
- * ZSTD_decompressBegin_usingDict(). Alternatively, you can copy a prepared
- * context, using ZSTD_copyDCtx().
- *
- * Then use ZSTD_nextSrcSizeToDecompress() and ZSTD_decompressContinue()
- * alternatively.
- * ZSTD_nextSrcSizeToDecompress() tells how many bytes to provide as 'srcSize'
- * to ZSTD_decompressContinue().
- * ZSTD_decompressContinue() requires this _exact_ amount of bytes, or it will
- * fail.
- *
- * The result of ZSTD_decompressContinue() is the number of bytes regenerated
- * within 'dst' (necessarily <= dstCapacity). It can be zero, which is not an
- * error; it just means ZSTD_decompressContinue() has decoded some metadata
- * item. It can also be an error code, which can be tested with ZSTD_isError().
- *
- * ZSTD_decompressContinue() needs previous data blocks during decompression, up
- * to `windowSize`. They should preferably be located contiguously, prior to
- * current block. Alternatively, a round buffer of sufficient size is also
- * possible. Sufficient size is determined by frame parameters.
- * ZSTD_decompressContinue() is very sensitive to contiguity, if 2 blocks don't
- * follow each other, make sure that either the compressor breaks contiguity at
- * the same place, or that previous contiguous segment is large enough to
- * properly handle maximum back-reference.
- *
- * A frame is fully decoded when ZSTD_nextSrcSizeToDecompress() returns zero.
- * Context can then be reset to start a new decompression.
- *
- * Note: it's possible to know if next input to present is a header or a block,
- * using ZSTD_nextInputType(). This information is not required to properly
- * decode a frame.
- *
- * == Special case: skippable frames ==
- *
- * Skippable frames allow integration of user-defined data into a flow of
- * concatenated frames. Skippable frames will be ignored (skipped) by a
- * decompressor. The format of skippable frames is as follows:
- * a) Skippable frame ID - 4 Bytes, Little endian format, any value from
- *    0x184D2A50 to 0x184D2A5F
- * b) Frame Size - 4 Bytes, Little endian format, unsigned 32-bits
- * c) Frame Content - any content (User Data) of length equal to Frame Size
- * For skippable frames ZSTD_decompressContinue() always returns 0.
- * For skippable frames ZSTD_getFrameParams() returns fparamsPtr->windowLog==0
- * what means that a frame is skippable.
- * Note: If fparamsPtr->frameContentSize==0, it is ambiguous: the frame might
- *       actually be a zstd encoded frame with no content. For purposes of
- *       decompression, it is valid in both cases to skip the frame using
- *       ZSTD_findFrameCompressedSize() to find its size in bytes.
- * It also returns frame size as fparamsPtr->frameContentSize.
- ******************************************************************************/
-
-/*=====   Buffer-less streaming decompression functions  =====*/
-size_t ZSTD_decompressBegin(ZSTD_DCtx *dctx);
-size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx *dctx, const void *dict,
-	size_t dictSize);
-void   ZSTD_copyDCtx(ZSTD_DCtx *dctx, const ZSTD_DCtx *preparedDCtx);
-size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx *dctx);
-size_t ZSTD_decompressContinue(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-typedef enum {
-	ZSTDnit_frameHeader,
-	ZSTDnit_blockHeader,
-	ZSTDnit_block,
-	ZSTDnit_lastBlock,
-	ZSTDnit_checksum,
-	ZSTDnit_skippableFrame
-} ZSTD_nextInputType_e;
-ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx *dctx);
-
-/*-*****************************************************************************
- * Block functions
- *
- * Block functions produce and decode raw zstd blocks, without frame metadata.
- * Frame metadata cost is typically ~18 bytes, which can be non-negligible for
- * very small blocks (< 100 bytes). User will have to take in charge required
- * information to regenerate data, such as compressed and content sizes.
- *
- * A few rules to respect:
- * - Compressing and decompressing require a context structure
- *   + Use ZSTD_initCCtx() and ZSTD_initDCtx()
- * - It is necessary to init context before starting
- *   + compression : ZSTD_compressBegin()
- *   + decompression : ZSTD_decompressBegin()
- *   + variants _usingDict() are also allowed
- *   + copyCCtx() and copyDCtx() work too
- * - Block size is limited, it must be <= ZSTD_getBlockSizeMax()
- *   + If you need to compress more, cut data into multiple blocks
- *   + Consider using the regular ZSTD_compress() instead, as frame metadata
- *     costs become negligible when source size is large.
- * - When a block is considered not compressible enough, ZSTD_compressBlock()
- *   result will be zero. In which case, nothing is produced into `dst`.
- *   + User must test for such outcome and deal directly with uncompressed data
- *   + ZSTD_decompressBlock() doesn't accept uncompressed data as input!!!
- *   + In case of multiple successive blocks, decoder must be informed of
- *     uncompressed block existence to follow proper history. Use
- *     ZSTD_insertBlock() in such a case.
- ******************************************************************************/
-
-/* Define for static allocation */
-#define ZSTD_BLOCKSIZE_ABSOLUTEMAX (128 * 1024)
-/*=====   Raw zstd block functions  =====*/
-size_t ZSTD_getBlockSizeMax(ZSTD_CCtx *cctx);
-size_t ZSTD_compressBlock(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-size_t ZSTD_decompressBlock(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity,
-	const void *src, size_t srcSize);
-size_t ZSTD_insertBlock(ZSTD_DCtx *dctx, const void *blockStart,
-	size_t blockSize);
-
-#endif  /* ZSTD_H */
+#endif  /* LINUX_ZSTD_H */
diff --git a/include/linux/zstd_errors.h b/include/linux/zstd_errors.h
new file mode 100644
index 0000000..58b6dd4
--- /dev/null
+++ b/include/linux/zstd_errors.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_ERRORS_H_398273423
+#define ZSTD_ERRORS_H_398273423
+
+
+/*===== dependency =====*/
+#include <linux/types.h>   /* size_t */
+
+
+/* =====   ZSTDERRORLIB_API : control library symbols visibility   ===== */
+#define ZSTDERRORLIB_VISIBILITY 
+#define ZSTDERRORLIB_API ZSTDERRORLIB_VISIBILITY
+
+/*-*********************************************
+ *  Error codes list
+ *-*********************************************
+ *  Error codes _values_ are pinned down since v1.3.1 only.
+ *  Therefore, don't rely on values if you may link to any version < v1.3.1.
+ *
+ *  Only values < 100 are considered stable.
+ *
+ *  note 1 : this API shall be used with static linking only.
+ *           dynamic linking is not yet officially supported.
+ *  note 2 : Prefer relying on the enum than on its value whenever possible
+ *           This is the only supported way to use the error list < v1.3.1
+ *  note 3 : ZSTD_isError() is always correct, whatever the library version.
+ **********************************************/
+typedef enum {
+  ZSTD_error_no_error = 0,
+  ZSTD_error_GENERIC  = 1,
+  ZSTD_error_prefix_unknown                = 10,
+  ZSTD_error_version_unsupported           = 12,
+  ZSTD_error_frameParameter_unsupported    = 14,
+  ZSTD_error_frameParameter_windowTooLarge = 16,
+  ZSTD_error_corruption_detected = 20,
+  ZSTD_error_checksum_wrong      = 22,
+  ZSTD_error_dictionary_corrupted      = 30,
+  ZSTD_error_dictionary_wrong          = 32,
+  ZSTD_error_dictionaryCreation_failed = 34,
+  ZSTD_error_parameter_unsupported   = 40,
+  ZSTD_error_parameter_outOfBound    = 42,
+  ZSTD_error_tableLog_tooLarge       = 44,
+  ZSTD_error_maxSymbolValue_tooLarge = 46,
+  ZSTD_error_maxSymbolValue_tooSmall = 48,
+  ZSTD_error_stage_wrong       = 60,
+  ZSTD_error_init_missing      = 62,
+  ZSTD_error_memory_allocation = 64,
+  ZSTD_error_workSpace_tooSmall= 66,
+  ZSTD_error_dstSize_tooSmall = 70,
+  ZSTD_error_srcSize_wrong    = 72,
+  ZSTD_error_dstBuffer_null   = 74,
+  /* following error codes are __NOT STABLE__, they can be removed or changed in future versions */
+  ZSTD_error_frameIndex_tooLarge = 100,
+  ZSTD_error_seekableIO          = 102,
+  ZSTD_error_dstBuffer_wrong     = 104,
+  ZSTD_error_srcBuffer_wrong     = 105,
+  ZSTD_error_maxCode = 120  /* never EVER use this value directly, it can change in future versions! Use ZSTD_isError() instead */
+} ZSTD_ErrorCode;
+
+/*! ZSTD_getErrorCode() :
+    convert a `size_t` function result into a `ZSTD_ErrorCode` enum type,
+    which can be used to compare with enum list published above */
+ZSTDERRORLIB_API ZSTD_ErrorCode ZSTD_getErrorCode(size_t functionResult);
+ZSTDERRORLIB_API const char* ZSTD_getErrorString(ZSTD_ErrorCode code);   /*< Same as ZSTD_getErrorName, but using a `ZSTD_ErrorCode` enum argument */
+
+
+
+#endif /* ZSTD_ERRORS_H_398273423 */
diff --git a/include/linux/zstd_lib.h b/include/linux/zstd_lib.h
new file mode 100644
index 0000000..b8c7dbf9
--- /dev/null
+++ b/include/linux/zstd_lib.h
@@ -0,0 +1,2432 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_H_235446
+#define ZSTD_H_235446
+
+/* ======   Dependency   ======*/
+#include <linux/limits.h>   /* INT_MAX */
+#include <linux/types.h>   /* size_t */
+
+
+/* =====   ZSTDLIB_API : control library symbols visibility   ===== */
+#define ZSTDLIB_VISIBILITY 
+#define ZSTDLIB_API ZSTDLIB_VISIBILITY
+
+
+/* *****************************************************************************
+  Introduction
+
+  zstd, short for Zstandard, is a fast lossless compression algorithm, targeting
+  real-time compression scenarios at zlib-level and better compression ratios.
+  The zstd compression library provides in-memory compression and decompression
+  functions.
+
+  The library supports regular compression levels from 1 up to ZSTD_maxCLevel(),
+  which is currently 22. Levels >= 20, labeled `--ultra`, should be used with
+  caution, as they require more memory. The library also offers negative
+  compression levels, which extend the range of speed vs. ratio preferences.
+  The lower the level, the faster the speed (at the cost of compression).
+
+  Compression can be done in:
+    - a single step (described as Simple API)
+    - a single step, reusing a context (described as Explicit context)
+    - unbounded multiple steps (described as Streaming compression)
+
+  The compression ratio achievable on small data can be highly improved using
+  a dictionary. Dictionary compression can be performed in:
+    - a single step (described as Simple dictionary API)
+    - a single step, reusing a dictionary (described as Bulk-processing
+      dictionary API)
+
+  Advanced experimental functions can be accessed using
+  `#define ZSTD_STATIC_LINKING_ONLY` before including zstd.h.
+
+  Advanced experimental APIs should never be used with a dynamically-linked
+  library. They are not "stable"; their definitions or signatures may change in
+  the future. Only static linking is allowed.
+*******************************************************************************/
+
+/*------   Version   ------*/
+#define ZSTD_VERSION_MAJOR    1
+#define ZSTD_VERSION_MINOR    4
+#define ZSTD_VERSION_RELEASE  10
+#define ZSTD_VERSION_NUMBER  (ZSTD_VERSION_MAJOR *100*100 + ZSTD_VERSION_MINOR *100 + ZSTD_VERSION_RELEASE)
+
+/*! ZSTD_versionNumber() :
+ *  Return runtime library version, the value is (MAJOR*100*100 + MINOR*100 + RELEASE). */
+ZSTDLIB_API unsigned ZSTD_versionNumber(void);
+
+#define ZSTD_LIB_VERSION ZSTD_VERSION_MAJOR.ZSTD_VERSION_MINOR.ZSTD_VERSION_RELEASE
+#define ZSTD_QUOTE(str) #str
+#define ZSTD_EXPAND_AND_QUOTE(str) ZSTD_QUOTE(str)
+#define ZSTD_VERSION_STRING ZSTD_EXPAND_AND_QUOTE(ZSTD_LIB_VERSION)
+
+/*! ZSTD_versionString() :
+ *  Return runtime library version, like "1.4.5". Requires v1.3.0+. */
+ZSTDLIB_API const char* ZSTD_versionString(void);
+
+/* *************************************
+ *  Default constant
+ ***************************************/
+#ifndef ZSTD_CLEVEL_DEFAULT
+#  define ZSTD_CLEVEL_DEFAULT 3
+#endif
+
+/* *************************************
+ *  Constants
+ ***************************************/
+
+/* All magic numbers are supposed read/written to/from files/memory using little-endian convention */
+#define ZSTD_MAGICNUMBER            0xFD2FB528    /* valid since v0.8.0 */
+#define ZSTD_MAGIC_DICTIONARY       0xEC30A437    /* valid since v0.7.0 */
+#define ZSTD_MAGIC_SKIPPABLE_START  0x184D2A50    /* all 16 values, from 0x184D2A50 to 0x184D2A5F, signal the beginning of a skippable frame */
+#define ZSTD_MAGIC_SKIPPABLE_MASK   0xFFFFFFF0
+
+#define ZSTD_BLOCKSIZELOG_MAX  17
+#define ZSTD_BLOCKSIZE_MAX     (1<<ZSTD_BLOCKSIZELOG_MAX)
+
+
+
+/* *************************************
+*  Simple API
+***************************************/
+/*! ZSTD_compress() :
+ *  Compresses `src` content as a single zstd compressed frame into already allocated `dst`.
+ *  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+ *  @return : compressed size written into `dst` (<= `dstCapacity),
+ *            or an error code if it fails (which can be tested using ZSTD_isError()). */
+ZSTDLIB_API size_t ZSTD_compress( void* dst, size_t dstCapacity,
+                            const void* src, size_t srcSize,
+                                  int compressionLevel);
+
+/*! ZSTD_decompress() :
+ *  `compressedSize` : must be the _exact_ size of some number of compressed and/or skippable frames.
+ *  `dstCapacity` is an upper bound of originalSize to regenerate.
+ *  If user cannot imply a maximum upper bound, it's better to use streaming mode to decompress data.
+ *  @return : the number of bytes decompressed into `dst` (<= `dstCapacity`),
+ *            or an errorCode if it fails (which can be tested using ZSTD_isError()). */
+ZSTDLIB_API size_t ZSTD_decompress( void* dst, size_t dstCapacity,
+                              const void* src, size_t compressedSize);
+
+/*! ZSTD_getFrameContentSize() : requires v1.3.0+
+ *  `src` should point to the start of a ZSTD encoded frame.
+ *  `srcSize` must be at least as large as the frame header.
+ *            hint : any size >= `ZSTD_frameHeaderSize_max` is large enough.
+ *  @return : - decompressed size of `src` frame content, if known
+ *            - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
+ *            - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small)
+ *   note 1 : a 0 return value means the frame is valid but "empty".
+ *   note 2 : decompressed size is an optional field, it may not be present, typically in streaming mode.
+ *            When `return==ZSTD_CONTENTSIZE_UNKNOWN`, data to decompress could be any size.
+ *            In which case, it's necessary to use streaming mode to decompress data.
+ *            Optionally, application can rely on some implicit limit,
+ *            as ZSTD_decompress() only needs an upper bound of decompressed size.
+ *            (For example, data could be necessarily cut into blocks <= 16 KB).
+ *   note 3 : decompressed size is always present when compression is completed using single-pass functions,
+ *            such as ZSTD_compress(), ZSTD_compressCCtx() ZSTD_compress_usingDict() or ZSTD_compress_usingCDict().
+ *   note 4 : decompressed size can be very large (64-bits value),
+ *            potentially larger than what local system can handle as a single memory segment.
+ *            In which case, it's necessary to use streaming mode to decompress data.
+ *   note 5 : If source is untrusted, decompressed size could be wrong or intentionally modified.
+ *            Always ensure return value fits within application's authorized limits.
+ *            Each application can set its own limits.
+ *   note 6 : This function replaces ZSTD_getDecompressedSize() */
+#define ZSTD_CONTENTSIZE_UNKNOWN (0ULL - 1)
+#define ZSTD_CONTENTSIZE_ERROR   (0ULL - 2)
+ZSTDLIB_API unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize);
+
+/*! ZSTD_getDecompressedSize() :
+ *  NOTE: This function is now obsolete, in favor of ZSTD_getFrameContentSize().
+ *  Both functions work the same way, but ZSTD_getDecompressedSize() blends
+ *  "empty", "unknown" and "error" results to the same return value (0),
+ *  while ZSTD_getFrameContentSize() gives them separate return values.
+ * @return : decompressed size of `src` frame content _if known and not empty_, 0 otherwise. */
+ZSTDLIB_API unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize);
+
+/*! ZSTD_findFrameCompressedSize() :
+ * `src` should point to the start of a ZSTD frame or skippable frame.
+ * `srcSize` must be >= first frame size
+ * @return : the compressed size of the first frame starting at `src`,
+ *           suitable to pass as `srcSize` to `ZSTD_decompress` or similar,
+ *        or an error code if input is invalid */
+ZSTDLIB_API size_t ZSTD_findFrameCompressedSize(const void* src, size_t srcSize);
+
+
+/*======  Helper functions  ======*/
+#define ZSTD_COMPRESSBOUND(srcSize)   ((srcSize) + ((srcSize)>>8) + (((srcSize) < (128<<10)) ? (((128<<10) - (srcSize)) >> 11) /* margin, from 64 to 0 */ : 0))  /* this formula ensures that bound(A) + bound(B) <= bound(A+B) as long as A and B >= 128 KB */
+ZSTDLIB_API size_t      ZSTD_compressBound(size_t srcSize); /*!< maximum compressed size in worst case single-pass scenario */
+ZSTDLIB_API unsigned    ZSTD_isError(size_t code);          /*!< tells if a `size_t` function result is an error code */
+ZSTDLIB_API const char* ZSTD_getErrorName(size_t code);     /*!< provides readable string from an error code */
+ZSTDLIB_API int         ZSTD_minCLevel(void);               /*!< minimum negative compression level allowed */
+ZSTDLIB_API int         ZSTD_maxCLevel(void);               /*!< maximum compression level available */
+
+
+/* *************************************
+*  Explicit context
+***************************************/
+/*= Compression context
+ *  When compressing many times,
+ *  it is recommended to allocate a context just once,
+ *  and re-use it for each successive compression operation.
+ *  This will make workload friendlier for system's memory.
+ *  Note : re-using context is just a speed / resource optimization.
+ *         It doesn't change the compression ratio, which remains identical.
+ *  Note 2 : In multi-threaded environments,
+ *         use one different context per thread for parallel execution.
+ */
+typedef struct ZSTD_CCtx_s ZSTD_CCtx;
+ZSTDLIB_API ZSTD_CCtx* ZSTD_createCCtx(void);
+ZSTDLIB_API size_t     ZSTD_freeCCtx(ZSTD_CCtx* cctx);  /* accept NULL pointer */
+
+/*! ZSTD_compressCCtx() :
+ *  Same as ZSTD_compress(), using an explicit ZSTD_CCtx.
+ *  Important : in order to behave similarly to `ZSTD_compress()`,
+ *  this function compresses at requested compression level,
+ *  __ignoring any other parameter__ .
+ *  If any advanced parameter was set using the advanced API,
+ *  they will all be reset. Only `compressionLevel` remains.
+ */
+ZSTDLIB_API size_t ZSTD_compressCCtx(ZSTD_CCtx* cctx,
+                                     void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                                     int compressionLevel);
+
+/*= Decompression context
+ *  When decompressing many times,
+ *  it is recommended to allocate a context only once,
+ *  and re-use it for each successive compression operation.
+ *  This will make workload friendlier for system's memory.
+ *  Use one context per thread for parallel execution. */
+typedef struct ZSTD_DCtx_s ZSTD_DCtx;
+ZSTDLIB_API ZSTD_DCtx* ZSTD_createDCtx(void);
+ZSTDLIB_API size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  /* accept NULL pointer */
+
+/*! ZSTD_decompressDCtx() :
+ *  Same as ZSTD_decompress(),
+ *  requires an allocated ZSTD_DCtx.
+ *  Compatible with sticky parameters.
+ */
+ZSTDLIB_API size_t ZSTD_decompressDCtx(ZSTD_DCtx* dctx,
+                                       void* dst, size_t dstCapacity,
+                                 const void* src, size_t srcSize);
+
+
+/* *************************************
+*  Advanced compression API
+***************************************/
+
+/* API design :
+ *   Parameters are pushed one by one into an existing context,
+ *   using ZSTD_CCtx_set*() functions.
+ *   Pushed parameters are sticky : they are valid for next compressed frame, and any subsequent frame.
+ *   "sticky" parameters are applicable to `ZSTD_compress2()` and `ZSTD_compressStream*()` !
+ *   __They do not apply to "simple" one-shot variants such as ZSTD_compressCCtx()__ .
+ *
+ *   It's possible to reset all parameters to "default" using ZSTD_CCtx_reset().
+ *
+ *   This API supercedes all other "advanced" API entry points in the experimental section.
+ *   In the future, we expect to remove from experimental API entry points which are redundant with this API.
+ */
+
+
+/* Compression strategies, listed from fastest to strongest */
+typedef enum { ZSTD_fast=1,
+               ZSTD_dfast=2,
+               ZSTD_greedy=3,
+               ZSTD_lazy=4,
+               ZSTD_lazy2=5,
+               ZSTD_btlazy2=6,
+               ZSTD_btopt=7,
+               ZSTD_btultra=8,
+               ZSTD_btultra2=9
+               /* note : new strategies _might_ be added in the future.
+                         Only the order (from fast to strong) is guaranteed */
+} ZSTD_strategy;
+
+
+typedef enum {
+
+    /* compression parameters
+     * Note: When compressing with a ZSTD_CDict these parameters are superseded
+     * by the parameters used to construct the ZSTD_CDict.
+     * See ZSTD_CCtx_refCDict() for more info (superseded-by-cdict). */
+    ZSTD_c_compressionLevel=100, /* Set compression parameters according to pre-defined cLevel table.
+                              * Note that exact compression parameters are dynamically determined,
+                              * depending on both compression level and srcSize (when known).
+                              * Default level is ZSTD_CLEVEL_DEFAULT==3.
+                              * Special: value 0 means default, which is controlled by ZSTD_CLEVEL_DEFAULT.
+                              * Note 1 : it's possible to pass a negative compression level.
+                              * Note 2 : setting a level does not automatically set all other compression parameters
+                              *   to default. Setting this will however eventually dynamically impact the compression
+                              *   parameters which have not been manually set. The manually set
+                              *   ones will 'stick'. */
+    /* Advanced compression parameters :
+     * It's possible to pin down compression parameters to some specific values.
+     * In which case, these values are no longer dynamically selected by the compressor */
+    ZSTD_c_windowLog=101,    /* Maximum allowed back-reference distance, expressed as power of 2.
+                              * This will set a memory budget for streaming decompression,
+                              * with larger values requiring more memory
+                              * and typically compressing more.
+                              * Must be clamped between ZSTD_WINDOWLOG_MIN and ZSTD_WINDOWLOG_MAX.
+                              * Special: value 0 means "use default windowLog".
+                              * Note: Using a windowLog greater than ZSTD_WINDOWLOG_LIMIT_DEFAULT
+                              *       requires explicitly allowing such size at streaming decompression stage. */
+    ZSTD_c_hashLog=102,      /* Size of the initial probe table, as a power of 2.
+                              * Resulting memory usage is (1 << (hashLog+2)).
+                              * Must be clamped between ZSTD_HASHLOG_MIN and ZSTD_HASHLOG_MAX.
+                              * Larger tables improve compression ratio of strategies <= dFast,
+                              * and improve speed of strategies > dFast.
+                              * Special: value 0 means "use default hashLog". */
+    ZSTD_c_chainLog=103,     /* Size of the multi-probe search table, as a power of 2.
+                              * Resulting memory usage is (1 << (chainLog+2)).
+                              * Must be clamped between ZSTD_CHAINLOG_MIN and ZSTD_CHAINLOG_MAX.
+                              * Larger tables result in better and slower compression.
+                              * This parameter is useless for "fast" strategy.
+                              * It's still useful when using "dfast" strategy,
+                              * in which case it defines a secondary probe table.
+                              * Special: value 0 means "use default chainLog". */
+    ZSTD_c_searchLog=104,    /* Number of search attempts, as a power of 2.
+                              * More attempts result in better and slower compression.
+                              * This parameter is useless for "fast" and "dFast" strategies.
+                              * Special: value 0 means "use default searchLog". */
+    ZSTD_c_minMatch=105,     /* Minimum size of searched matches.
+                              * Note that Zstandard can still find matches of smaller size,
+                              * it just tweaks its search algorithm to look for this size and larger.
+                              * Larger values increase compression and decompression speed, but decrease ratio.
+                              * Must be clamped between ZSTD_MINMATCH_MIN and ZSTD_MINMATCH_MAX.
+                              * Note that currently, for all strategies < btopt, effective minimum is 4.
+                              *                    , for all strategies > fast, effective maximum is 6.
+                              * Special: value 0 means "use default minMatchLength". */
+    ZSTD_c_targetLength=106, /* Impact of this field depends on strategy.
+                              * For strategies btopt, btultra & btultra2:
+                              *     Length of Match considered "good enough" to stop search.
+                              *     Larger values make compression stronger, and slower.
+                              * For strategy fast:
+                              *     Distance between match sampling.
+                              *     Larger values make compression faster, and weaker.
+                              * Special: value 0 means "use default targetLength". */
+    ZSTD_c_strategy=107,     /* See ZSTD_strategy enum definition.
+                              * The higher the value of selected strategy, the more complex it is,
+                              * resulting in stronger and slower compression.
+                              * Special: value 0 means "use default strategy". */
+
+    /* LDM mode parameters */
+    ZSTD_c_enableLongDistanceMatching=160, /* Enable long distance matching.
+                                     * This parameter is designed to improve compression ratio
+                                     * for large inputs, by finding large matches at long distance.
+                                     * It increases memory usage and window size.
+                                     * Note: enabling this parameter increases default ZSTD_c_windowLog to 128 MB
+                                     * except when expressly set to a different value.
+                                     * Note: will be enabled by default if ZSTD_c_windowLog >= 128 MB and
+                                     * compression strategy >= ZSTD_btopt (== compression level 16+) */
+    ZSTD_c_ldmHashLog=161,   /* Size of the table for long distance matching, as a power of 2.
+                              * Larger values increase memory usage and compression ratio,
+                              * but decrease compression speed.
+                              * Must be clamped between ZSTD_HASHLOG_MIN and ZSTD_HASHLOG_MAX
+                              * default: windowlog - 7.
+                              * Special: value 0 means "automatically determine hashlog". */
+    ZSTD_c_ldmMinMatch=162,  /* Minimum match size for long distance matcher.
+                              * Larger/too small values usually decrease compression ratio.
+                              * Must be clamped between ZSTD_LDM_MINMATCH_MIN and ZSTD_LDM_MINMATCH_MAX.
+                              * Special: value 0 means "use default value" (default: 64). */
+    ZSTD_c_ldmBucketSizeLog=163, /* Log size of each bucket in the LDM hash table for collision resolution.
+                              * Larger values improve collision resolution but decrease compression speed.
+                              * The maximum value is ZSTD_LDM_BUCKETSIZELOG_MAX.
+                              * Special: value 0 means "use default value" (default: 3). */
+    ZSTD_c_ldmHashRateLog=164, /* Frequency of inserting/looking up entries into the LDM hash table.
+                              * Must be clamped between 0 and (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN).
+                              * Default is MAX(0, (windowLog - ldmHashLog)), optimizing hash table usage.
+                              * Larger values improve compression speed.
+                              * Deviating far from default value will likely result in a compression ratio decrease.
+                              * Special: value 0 means "automatically determine hashRateLog". */
+
+    /* frame parameters */
+    ZSTD_c_contentSizeFlag=200, /* Content size will be written into frame header _whenever known_ (default:1)
+                              * Content size must be known at the beginning of compression.
+                              * This is automatically the case when using ZSTD_compress2(),
+                              * For streaming scenarios, content size must be provided with ZSTD_CCtx_setPledgedSrcSize() */
+    ZSTD_c_checksumFlag=201, /* A 32-bits checksum of content is written at end of frame (default:0) */
+    ZSTD_c_dictIDFlag=202,   /* When applicable, dictionary's ID is written into frame header (default:1) */
+
+    /* multi-threading parameters */
+    /* These parameters are only active if multi-threading is enabled (compiled with build macro ZSTD_MULTITHREAD).
+     * Otherwise, trying to set any other value than default (0) will be a no-op and return an error.
+     * In a situation where it's unknown if the linked library supports multi-threading or not,
+     * setting ZSTD_c_nbWorkers to any value >= 1 and consulting the return value provides a quick way to check this property.
+     */
+    ZSTD_c_nbWorkers=400,    /* Select how many threads will be spawned to compress in parallel.
+                              * When nbWorkers >= 1, triggers asynchronous mode when invoking ZSTD_compressStream*() :
+                              * ZSTD_compressStream*() consumes input and flush output if possible, but immediately gives back control to caller,
+                              * while compression is performed in parallel, within worker thread(s).
+                              * (note : a strong exception to this rule is when first invocation of ZSTD_compressStream2() sets ZSTD_e_end :
+                              *  in which case, ZSTD_compressStream2() delegates to ZSTD_compress2(), which is always a blocking call).
+                              * More workers improve speed, but also increase memory usage.
+                              * Default value is `0`, aka "single-threaded mode" : no worker is spawned,
+                              * compression is performed inside Caller's thread, and all invocations are blocking */
+    ZSTD_c_jobSize=401,      /* Size of a compression job. This value is enforced only when nbWorkers >= 1.
+                              * Each compression job is completed in parallel, so this value can indirectly impact the nb of active threads.
+                              * 0 means default, which is dynamically determined based on compression parameters.
+                              * Job size must be a minimum of overlap size, or 1 MB, whichever is largest.
+                              * The minimum size is automatically and transparently enforced. */
+    ZSTD_c_overlapLog=402,   /* Control the overlap size, as a fraction of window size.
+                              * The overlap size is an amount of data reloaded from previous job at the beginning of a new job.
+                              * It helps preserve compression ratio, while each job is compressed in parallel.
+                              * This value is enforced only when nbWorkers >= 1.
+                              * Larger values increase compression ratio, but decrease speed.
+                              * Possible values range from 0 to 9 :
+                              * - 0 means "default" : value will be determined by the library, depending on strategy
+                              * - 1 means "no overlap"
+                              * - 9 means "full overlap", using a full window size.
+                              * Each intermediate rank increases/decreases load size by a factor 2 :
+                              * 9: full window;  8: w/2;  7: w/4;  6: w/8;  5:w/16;  4: w/32;  3:w/64;  2:w/128;  1:no overlap;  0:default
+                              * default value varies between 6 and 9, depending on strategy */
+
+    /* note : additional experimental parameters are also available
+     * within the experimental section of the API.
+     * At the time of this writing, they include :
+     * ZSTD_c_rsyncable
+     * ZSTD_c_format
+     * ZSTD_c_forceMaxWindow
+     * ZSTD_c_forceAttachDict
+     * ZSTD_c_literalCompressionMode
+     * ZSTD_c_targetCBlockSize
+     * ZSTD_c_srcSizeHint
+     * ZSTD_c_enableDedicatedDictSearch
+     * ZSTD_c_stableInBuffer
+     * ZSTD_c_stableOutBuffer
+     * ZSTD_c_blockDelimiters
+     * ZSTD_c_validateSequences
+     * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
+     * note : never ever use experimentalParam? names directly;
+     *        also, the enums values themselves are unstable and can still change.
+     */
+     ZSTD_c_experimentalParam1=500,
+     ZSTD_c_experimentalParam2=10,
+     ZSTD_c_experimentalParam3=1000,
+     ZSTD_c_experimentalParam4=1001,
+     ZSTD_c_experimentalParam5=1002,
+     ZSTD_c_experimentalParam6=1003,
+     ZSTD_c_experimentalParam7=1004,
+     ZSTD_c_experimentalParam8=1005,
+     ZSTD_c_experimentalParam9=1006,
+     ZSTD_c_experimentalParam10=1007,
+     ZSTD_c_experimentalParam11=1008,
+     ZSTD_c_experimentalParam12=1009
+} ZSTD_cParameter;
+
+typedef struct {
+    size_t error;
+    int lowerBound;
+    int upperBound;
+} ZSTD_bounds;
+
+/*! ZSTD_cParam_getBounds() :
+ *  All parameters must belong to an interval with lower and upper bounds,
+ *  otherwise they will either trigger an error or be automatically clamped.
+ * @return : a structure, ZSTD_bounds, which contains
+ *         - an error status field, which must be tested using ZSTD_isError()
+ *         - lower and upper bounds, both inclusive
+ */
+ZSTDLIB_API ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter cParam);
+
+/*! ZSTD_CCtx_setParameter() :
+ *  Set one compression parameter, selected by enum ZSTD_cParameter.
+ *  All parameters have valid bounds. Bounds can be queried using ZSTD_cParam_getBounds().
+ *  Providing a value beyond bound will either clamp it, or trigger an error (depending on parameter).
+ *  Setting a parameter is generally only possible during frame initialization (before starting compression).
+ *  Exception : when using multi-threading mode (nbWorkers >= 1),
+ *              the following parameters can be updated _during_ compression (within same frame):
+ *              => compressionLevel, hashLog, chainLog, searchLog, minMatch, targetLength and strategy.
+ *              new parameters will be active for next job only (after a flush()).
+ * @return : an error code (which can be tested using ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value);
+
+/*! ZSTD_CCtx_setPledgedSrcSize() :
+ *  Total input data size to be compressed as a single frame.
+ *  Value will be written in frame header, unless if explicitly forbidden using ZSTD_c_contentSizeFlag.
+ *  This value will also be controlled at end of frame, and trigger an error if not respected.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Note 1 : pledgedSrcSize==0 actually means zero, aka an empty frame.
+ *           In order to mean "unknown content size", pass constant ZSTD_CONTENTSIZE_UNKNOWN.
+ *           ZSTD_CONTENTSIZE_UNKNOWN is default value for any new frame.
+ *  Note 2 : pledgedSrcSize is only valid once, for the next frame.
+ *           It's discarded at the end of the frame, and replaced by ZSTD_CONTENTSIZE_UNKNOWN.
+ *  Note 3 : Whenever all input data is provided and consumed in a single round,
+ *           for example with ZSTD_compress2(),
+ *           or invoking immediately ZSTD_compressStream2(,,,ZSTD_e_end),
+ *           this value is automatically overridden by srcSize instead.
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize);
+
+typedef enum {
+    ZSTD_reset_session_only = 1,
+    ZSTD_reset_parameters = 2,
+    ZSTD_reset_session_and_parameters = 3
+} ZSTD_ResetDirective;
+
+/*! ZSTD_CCtx_reset() :
+ *  There are 2 different things that can be reset, independently or jointly :
+ *  - The session : will stop compressing current frame, and make CCtx ready to start a new one.
+ *                  Useful after an error, or to interrupt any ongoing compression.
+ *                  Any internal data not yet flushed is cancelled.
+ *                  Compression parameters and dictionary remain unchanged.
+ *                  They will be used to compress next frame.
+ *                  Resetting session never fails.
+ *  - The parameters : changes all parameters back to "default".
+ *                  This removes any reference to any dictionary too.
+ *                  Parameters can only be changed between 2 sessions (i.e. no compression is currently ongoing)
+ *                  otherwise the reset fails, and function returns an error value (which can be tested using ZSTD_isError())
+ *  - Both : similar to resetting the session, followed by resetting parameters.
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset);
+
+/*! ZSTD_compress2() :
+ *  Behave the same as ZSTD_compressCCtx(), but compression parameters are set using the advanced API.
+ *  ZSTD_compress2() always starts a new frame.
+ *  Should cctx hold data from a previously unfinished frame, everything about it is forgotten.
+ *  - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
+ *  - The function is always blocking, returns when compression is completed.
+ *  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+ * @return : compressed size written into `dst` (<= `dstCapacity),
+ *           or an error code if it fails (which can be tested using ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_compress2( ZSTD_CCtx* cctx,
+                                   void* dst, size_t dstCapacity,
+                             const void* src, size_t srcSize);
+
+
+/* *************************************
+*  Advanced decompression API
+***************************************/
+
+/* The advanced API pushes parameters one by one into an existing DCtx context.
+ * Parameters are sticky, and remain valid for all following frames
+ * using the same DCtx context.
+ * It's possible to reset parameters to default values using ZSTD_DCtx_reset().
+ * Note : This API is compatible with existing ZSTD_decompressDCtx() and ZSTD_decompressStream().
+ *        Therefore, no new decompression function is necessary.
+ */
+
+typedef enum {
+
+    ZSTD_d_windowLogMax=100, /* Select a size limit (in power of 2) beyond which
+                              * the streaming API will refuse to allocate memory buffer
+                              * in order to protect the host from unreasonable memory requirements.
+                              * This parameter is only useful in streaming mode, since no internal buffer is allocated in single-pass mode.
+                              * By default, a decompression context accepts window sizes <= (1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT).
+                              * Special: value 0 means "use default maximum windowLog". */
+
+    /* note : additional experimental parameters are also available
+     * within the experimental section of the API.
+     * At the time of this writing, they include :
+     * ZSTD_d_format
+     * ZSTD_d_stableOutBuffer
+     * ZSTD_d_forceIgnoreChecksum
+     * ZSTD_d_refMultipleDDicts
+     * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
+     * note : never ever use experimentalParam? names directly
+     */
+     ZSTD_d_experimentalParam1=1000,
+     ZSTD_d_experimentalParam2=1001,
+     ZSTD_d_experimentalParam3=1002,
+     ZSTD_d_experimentalParam4=1003
+
+} ZSTD_dParameter;
+
+/*! ZSTD_dParam_getBounds() :
+ *  All parameters must belong to an interval with lower and upper bounds,
+ *  otherwise they will either trigger an error or be automatically clamped.
+ * @return : a structure, ZSTD_bounds, which contains
+ *         - an error status field, which must be tested using ZSTD_isError()
+ *         - both lower and upper bounds, inclusive
+ */
+ZSTDLIB_API ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam);
+
+/*! ZSTD_DCtx_setParameter() :
+ *  Set one compression parameter, selected by enum ZSTD_dParameter.
+ *  All parameters have valid bounds. Bounds can be queried using ZSTD_dParam_getBounds().
+ *  Providing a value beyond bound will either clamp it, or trigger an error (depending on parameter).
+ *  Setting a parameter is only possible during frame initialization (before starting decompression).
+ * @return : 0, or an error code (which can be tested using ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int value);
+
+/*! ZSTD_DCtx_reset() :
+ *  Return a DCtx to clean state.
+ *  Session and parameters can be reset jointly or separately.
+ *  Parameters can only be reset when no active frame is being decompressed.
+ * @return : 0, or an error code, which can be tested with ZSTD_isError()
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_reset(ZSTD_DCtx* dctx, ZSTD_ResetDirective reset);
+
+
+/* **************************
+*  Streaming
+****************************/
+
+typedef struct ZSTD_inBuffer_s {
+  const void* src;    /*< start of input buffer */
+  size_t size;        /*< size of input buffer */
+  size_t pos;         /*< position where reading stopped. Will be updated. Necessarily 0 <= pos <= size */
+} ZSTD_inBuffer;
+
+typedef struct ZSTD_outBuffer_s {
+  void*  dst;         /*< start of output buffer */
+  size_t size;        /*< size of output buffer */
+  size_t pos;         /*< position where writing stopped. Will be updated. Necessarily 0 <= pos <= size */
+} ZSTD_outBuffer;
+
+
+
+/*-***********************************************************************
+*  Streaming compression - HowTo
+*
+*  A ZSTD_CStream object is required to track streaming operation.
+*  Use ZSTD_createCStream() and ZSTD_freeCStream() to create/release resources.
+*  ZSTD_CStream objects can be reused multiple times on consecutive compression operations.
+*  It is recommended to re-use ZSTD_CStream since it will play nicer with system's memory, by re-using already allocated memory.
+*
+*  For parallel execution, use one separate ZSTD_CStream per thread.
+*
+*  note : since v1.3.0, ZSTD_CStream and ZSTD_CCtx are the same thing.
+*
+*  Parameters are sticky : when starting a new compression on the same context,
+*  it will re-use the same sticky parameters as previous compression session.
+*  When in doubt, it's recommended to fully initialize the context before usage.
+*  Use ZSTD_CCtx_reset() to reset the context and ZSTD_CCtx_setParameter(),
+*  ZSTD_CCtx_setPledgedSrcSize(), or ZSTD_CCtx_loadDictionary() and friends to
+*  set more specific parameters, the pledged source size, or load a dictionary.
+*
+*  Use ZSTD_compressStream2() with ZSTD_e_continue as many times as necessary to
+*  consume input stream. The function will automatically update both `pos`
+*  fields within `input` and `output`.
+*  Note that the function may not consume the entire input, for example, because
+*  the output buffer is already full, in which case `input.pos < input.size`.
+*  The caller must check if input has been entirely consumed.
+*  If not, the caller must make some room to receive more compressed data,
+*  and then present again remaining input data.
+*  note: ZSTD_e_continue is guaranteed to make some forward progress when called,
+*        but doesn't guarantee maximal forward progress. This is especially relevant
+*        when compressing with multiple threads. The call won't block if it can
+*        consume some input, but if it can't it will wait for some, but not all,
+*        output to be flushed.
+* @return : provides a minimum amount of data remaining to be flushed from internal buffers
+*           or an error code, which can be tested using ZSTD_isError().
+*
+*  At any moment, it's possible to flush whatever data might remain stuck within internal buffer,
+*  using ZSTD_compressStream2() with ZSTD_e_flush. `output->pos` will be updated.
+*  Note that, if `output->size` is too small, a single invocation with ZSTD_e_flush might not be enough (return code > 0).
+*  In which case, make some room to receive more compressed data, and call again ZSTD_compressStream2() with ZSTD_e_flush.
+*  You must continue calling ZSTD_compressStream2() with ZSTD_e_flush until it returns 0, at which point you can change the
+*  operation.
+*  note: ZSTD_e_flush will flush as much output as possible, meaning when compressing with multiple threads, it will
+*        block until the flush is complete or the output buffer is full.
+*  @return : 0 if internal buffers are entirely flushed,
+*            >0 if some data still present within internal buffer (the value is minimal estimation of remaining size),
+*            or an error code, which can be tested using ZSTD_isError().
+*
+*  Calling ZSTD_compressStream2() with ZSTD_e_end instructs to finish a frame.
+*  It will perform a flush and write frame epilogue.
+*  The epilogue is required for decoders to consider a frame completed.
+*  flush operation is the same, and follows same rules as calling ZSTD_compressStream2() with ZSTD_e_flush.
+*  You must continue calling ZSTD_compressStream2() with ZSTD_e_end until it returns 0, at which point you are free to
+*  start a new frame.
+*  note: ZSTD_e_end will flush as much output as possible, meaning when compressing with multiple threads, it will
+*        block until the flush is complete or the output buffer is full.
+*  @return : 0 if frame fully completed and fully flushed,
+*            >0 if some data still present within internal buffer (the value is minimal estimation of remaining size),
+*            or an error code, which can be tested using ZSTD_isError().
+*
+* *******************************************************************/
+
+typedef ZSTD_CCtx ZSTD_CStream;  /*< CCtx and CStream are now effectively same object (>= v1.3.0) */
+                                 /* Continue to distinguish them for compatibility with older versions <= v1.2.0 */
+/*===== ZSTD_CStream management functions =====*/
+ZSTDLIB_API ZSTD_CStream* ZSTD_createCStream(void);
+ZSTDLIB_API size_t ZSTD_freeCStream(ZSTD_CStream* zcs);  /* accept NULL pointer */
+
+/*===== Streaming compression functions =====*/
+typedef enum {
+    ZSTD_e_continue=0, /* collect more data, encoder decides when to output compressed result, for optimal compression ratio */
+    ZSTD_e_flush=1,    /* flush any data provided so far,
+                        * it creates (at least) one new block, that can be decoded immediately on reception;
+                        * frame will continue: any future data can still reference previously compressed data, improving compression.
+                        * note : multithreaded compression will block to flush as much output as possible. */
+    ZSTD_e_end=2       /* flush any remaining data _and_ close current frame.
+                        * note that frame is only closed after compressed data is fully flushed (return value == 0).
+                        * After that point, any additional data starts a new frame.
+                        * note : each frame is independent (does not reference any content from previous frame).
+                        : note : multithreaded compression will block to flush as much output as possible. */
+} ZSTD_EndDirective;
+
+/*! ZSTD_compressStream2() :
+ *  Behaves about the same as ZSTD_compressStream, with additional control on end directive.
+ *  - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
+ *  - Compression parameters cannot be changed once compression is started (save a list of exceptions in multi-threading mode)
+ *  - output->pos must be <= dstCapacity, input->pos must be <= srcSize
+ *  - output->pos and input->pos will be updated. They are guaranteed to remain below their respective limit.
+ *  - endOp must be a valid directive
+ *  - When nbWorkers==0 (default), function is blocking : it completes its job before returning to caller.
+ *  - When nbWorkers>=1, function is non-blocking : it copies a portion of input, distributes jobs to internal worker threads, flush to output whatever is available,
+ *                                                  and then immediately returns, just indicating that there is some data remaining to be flushed.
+ *                                                  The function nonetheless guarantees forward progress : it will return only after it reads or write at least 1+ byte.
+ *  - Exception : if the first call requests a ZSTD_e_end directive and provides enough dstCapacity, the function delegates to ZSTD_compress2() which is always blocking.
+ *  - @return provides a minimum amount of data remaining to be flushed from internal buffers
+ *            or an error code, which can be tested using ZSTD_isError().
+ *            if @return != 0, flush is not fully completed, there is still some data left within internal buffers.
+ *            This is useful for ZSTD_e_flush, since in this case more flushes are necessary to empty all buffers.
+ *            For ZSTD_e_end, @return == 0 when internal buffers are fully flushed and frame is completed.
+ *  - after a ZSTD_e_end directive, if internal buffer is not fully flushed (@return != 0),
+ *            only ZSTD_e_end or ZSTD_e_flush operations are allowed.
+ *            Before starting a new compression job, or changing compression parameters,
+ *            it is required to fully flush internal buffers.
+ */
+ZSTDLIB_API size_t ZSTD_compressStream2( ZSTD_CCtx* cctx,
+                                         ZSTD_outBuffer* output,
+                                         ZSTD_inBuffer* input,
+                                         ZSTD_EndDirective endOp);
+
+
+/* These buffer sizes are softly recommended.
+ * They are not required : ZSTD_compressStream*() happily accepts any buffer size, for both input and output.
+ * Respecting the recommended size just makes it a bit easier for ZSTD_compressStream*(),
+ * reducing the amount of memory shuffling and buffering, resulting in minor performance savings.
+ *
+ * However, note that these recommendations are from the perspective of a C caller program.
+ * If the streaming interface is invoked from some other language,
+ * especially managed ones such as Java or Go, through a foreign function interface such as jni or cgo,
+ * a major performance rule is to reduce crossing such interface to an absolute minimum.
+ * It's not rare that performance ends being spent more into the interface, rather than compression itself.
+ * In which cases, prefer using large buffers, as large as practical,
+ * for both input and output, to reduce the nb of roundtrips.
+ */
+ZSTDLIB_API size_t ZSTD_CStreamInSize(void);    /*< recommended size for input buffer */
+ZSTDLIB_API size_t ZSTD_CStreamOutSize(void);   /*< recommended size for output buffer. Guarantee to successfully flush at least one complete compressed block. */
+
+
+/* *****************************************************************************
+ * This following is a legacy streaming API.
+ * It can be replaced by ZSTD_CCtx_reset() and ZSTD_compressStream2().
+ * It is redundant, but remains fully supported.
+ * Advanced parameters and dictionary compression can only be used through the
+ * new API.
+ ******************************************************************************/
+
+/*!
+ * Equivalent to:
+ *
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_refCDict(zcs, NULL); // clear the dictionary (if any)
+ *     ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
+ */
+ZSTDLIB_API size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel);
+/*!
+ * Alternative for ZSTD_compressStream2(zcs, output, input, ZSTD_e_continue).
+ * NOTE: The return value is different. ZSTD_compressStream() returns a hint for
+ * the next read size (if non-zero and not an error). ZSTD_compressStream2()
+ * returns the minimum nb of bytes left to flush (if non-zero and not an error).
+ */
+ZSTDLIB_API size_t ZSTD_compressStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
+/*! Equivalent to ZSTD_compressStream2(zcs, output, &emptyInput, ZSTD_e_flush). */
+ZSTDLIB_API size_t ZSTD_flushStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
+/*! Equivalent to ZSTD_compressStream2(zcs, output, &emptyInput, ZSTD_e_end). */
+ZSTDLIB_API size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
+
+
+/*-***************************************************************************
+*  Streaming decompression - HowTo
+*
+*  A ZSTD_DStream object is required to track streaming operations.
+*  Use ZSTD_createDStream() and ZSTD_freeDStream() to create/release resources.
+*  ZSTD_DStream objects can be re-used multiple times.
+*
+*  Use ZSTD_initDStream() to start a new decompression operation.
+* @return : recommended first input size
+*  Alternatively, use advanced API to set specific properties.
+*
+*  Use ZSTD_decompressStream() repetitively to consume your input.
+*  The function will update both `pos` fields.
+*  If `input.pos < input.size`, some input has not been consumed.
+*  It's up to the caller to present again remaining data.
+*  The function tries to flush all data decoded immediately, respecting output buffer size.
+*  If `output.pos < output.size`, decoder has flushed everything it could.
+*  But if `output.pos == output.size`, there might be some data left within internal buffers.,
+*  In which case, call ZSTD_decompressStream() again to flush whatever remains in the buffer.
+*  Note : with no additional input provided, amount of data flushed is necessarily <= ZSTD_BLOCKSIZE_MAX.
+* @return : 0 when a frame is completely decoded and fully flushed,
+*        or an error code, which can be tested using ZSTD_isError(),
+*        or any other value > 0, which means there is still some decoding or flushing to do to complete current frame :
+*                                the return value is a suggested next input size (just a hint for better latency)
+*                                that will never request more than the remaining frame size.
+* *******************************************************************************/
+
+typedef ZSTD_DCtx ZSTD_DStream;  /*< DCtx and DStream are now effectively same object (>= v1.3.0) */
+                                 /* For compatibility with versions <= v1.2.0, prefer differentiating them. */
+/*===== ZSTD_DStream management functions =====*/
+ZSTDLIB_API ZSTD_DStream* ZSTD_createDStream(void);
+ZSTDLIB_API size_t ZSTD_freeDStream(ZSTD_DStream* zds);  /* accept NULL pointer */
+
+/*===== Streaming decompression functions =====*/
+
+/* This function is redundant with the advanced API and equivalent to:
+ *
+ *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+ *     ZSTD_DCtx_refDDict(zds, NULL);
+ */
+ZSTDLIB_API size_t ZSTD_initDStream(ZSTD_DStream* zds);
+
+ZSTDLIB_API size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
+
+ZSTDLIB_API size_t ZSTD_DStreamInSize(void);    /*!< recommended size for input buffer */
+ZSTDLIB_API size_t ZSTD_DStreamOutSize(void);   /*!< recommended size for output buffer. Guarantee to successfully flush at least one complete block in all circumstances. */
+
+
+/* ************************
+*  Simple dictionary API
+***************************/
+/*! ZSTD_compress_usingDict() :
+ *  Compression at an explicit compression level using a Dictionary.
+ *  A dictionary can be any arbitrary data segment (also called a prefix),
+ *  or a buffer with specified information (see dictBuilder/zdict.h).
+ *  Note : This function loads the dictionary, resulting in significant startup delay.
+ *         It's intended for a dictionary used only once.
+ *  Note 2 : When `dict == NULL || dictSize < 8` no dictionary is used. */
+ZSTDLIB_API size_t ZSTD_compress_usingDict(ZSTD_CCtx* ctx,
+                                           void* dst, size_t dstCapacity,
+                                     const void* src, size_t srcSize,
+                                     const void* dict,size_t dictSize,
+                                           int compressionLevel);
+
+/*! ZSTD_decompress_usingDict() :
+ *  Decompression using a known Dictionary.
+ *  Dictionary must be identical to the one used during compression.
+ *  Note : This function loads the dictionary, resulting in significant startup delay.
+ *         It's intended for a dictionary used only once.
+ *  Note : When `dict == NULL || dictSize < 8` no dictionary is used. */
+ZSTDLIB_API size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx,
+                                             void* dst, size_t dstCapacity,
+                                       const void* src, size_t srcSize,
+                                       const void* dict,size_t dictSize);
+
+
+/* *********************************
+ *  Bulk processing dictionary API
+ **********************************/
+typedef struct ZSTD_CDict_s ZSTD_CDict;
+
+/*! ZSTD_createCDict() :
+ *  When compressing multiple messages or blocks using the same dictionary,
+ *  it's recommended to digest the dictionary only once, since it's a costly operation.
+ *  ZSTD_createCDict() will create a state from digesting a dictionary.
+ *  The resulting state can be used for future compression operations with very limited startup cost.
+ *  ZSTD_CDict can be created once and shared by multiple threads concurrently, since its usage is read-only.
+ * @dictBuffer can be released after ZSTD_CDict creation, because its content is copied within CDict.
+ *  Note 1 : Consider experimental function `ZSTD_createCDict_byReference()` if you prefer to not duplicate @dictBuffer content.
+ *  Note 2 : A ZSTD_CDict can be created from an empty @dictBuffer,
+ *      in which case the only thing that it transports is the @compressionLevel.
+ *      This can be useful in a pipeline featuring ZSTD_compress_usingCDict() exclusively,
+ *      expecting a ZSTD_CDict parameter with any data, including those without a known dictionary. */
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict(const void* dictBuffer, size_t dictSize,
+                                         int compressionLevel);
+
+/*! ZSTD_freeCDict() :
+ *  Function frees memory allocated by ZSTD_createCDict().
+ *  If a NULL pointer is passed, no operation is performed. */
+ZSTDLIB_API size_t      ZSTD_freeCDict(ZSTD_CDict* CDict);
+
+/*! ZSTD_compress_usingCDict() :
+ *  Compression using a digested Dictionary.
+ *  Recommended when same dictionary is used multiple times.
+ *  Note : compression level is _decided at dictionary creation time_,
+ *     and frame parameters are hardcoded (dictID=yes, contentSize=yes, checksum=no) */
+ZSTDLIB_API size_t ZSTD_compress_usingCDict(ZSTD_CCtx* cctx,
+                                            void* dst, size_t dstCapacity,
+                                      const void* src, size_t srcSize,
+                                      const ZSTD_CDict* cdict);
+
+
+typedef struct ZSTD_DDict_s ZSTD_DDict;
+
+/*! ZSTD_createDDict() :
+ *  Create a digested dictionary, ready to start decompression operation without startup delay.
+ *  dictBuffer can be released after DDict creation, as its content is copied inside DDict. */
+ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict(const void* dictBuffer, size_t dictSize);
+
+/*! ZSTD_freeDDict() :
+ *  Function frees memory allocated with ZSTD_createDDict()
+ *  If a NULL pointer is passed, no operation is performed. */
+ZSTDLIB_API size_t      ZSTD_freeDDict(ZSTD_DDict* ddict);
+
+/*! ZSTD_decompress_usingDDict() :
+ *  Decompression using a digested Dictionary.
+ *  Recommended when same dictionary is used multiple times. */
+ZSTDLIB_API size_t ZSTD_decompress_usingDDict(ZSTD_DCtx* dctx,
+                                              void* dst, size_t dstCapacity,
+                                        const void* src, size_t srcSize,
+                                        const ZSTD_DDict* ddict);
+
+
+/* ******************************
+ *  Dictionary helper functions
+ *******************************/
+
+/*! ZSTD_getDictID_fromDict() :
+ *  Provides the dictID stored within dictionary.
+ *  if @return == 0, the dictionary is not conformant with Zstandard specification.
+ *  It can still be loaded, but as a content-only dictionary. */
+ZSTDLIB_API unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize);
+
+/*! ZSTD_getDictID_fromDDict() :
+ *  Provides the dictID of the dictionary loaded into `ddict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+ZSTDLIB_API unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict* ddict);
+
+/*! ZSTD_getDictID_fromFrame() :
+ *  Provides the dictID required to decompressed the frame stored within `src`.
+ *  If @return == 0, the dictID could not be decoded.
+ *  This could for one of the following reasons :
+ *  - The frame does not require a dictionary to be decoded (most common case).
+ *  - The frame was built with dictID intentionally removed. Whatever dictionary is necessary is a hidden information.
+ *    Note : this use case also happens when using a non-conformant dictionary.
+ *  - `srcSize` is too small, and as a result, the frame header could not be decoded (only possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`).
+ *  - This is not a Zstandard frame.
+ *  When identifying the exact failure cause, it's possible to use ZSTD_getFrameHeader(), which will provide a more precise error code. */
+ZSTDLIB_API unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize);
+
+
+/* *****************************************************************************
+ * Advanced dictionary and prefix API
+ *
+ * This API allows dictionaries to be used with ZSTD_compress2(),
+ * ZSTD_compressStream2(), and ZSTD_decompress(). Dictionaries are sticky, and
+ * only reset with the context is reset with ZSTD_reset_parameters or
+ * ZSTD_reset_session_and_parameters. Prefixes are single-use.
+ ******************************************************************************/
+
+
+/*! ZSTD_CCtx_loadDictionary() :
+ *  Create an internal CDict from `dict` buffer.
+ *  Decompression will have to use same dictionary.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Special: Loading a NULL (or 0-size) dictionary invalidates previous dictionary,
+ *           meaning "return to no-dictionary mode".
+ *  Note 1 : Dictionary is sticky, it will be used for all future compressed frames.
+ *           To return to "no-dictionary" situation, load a NULL dictionary (or reset parameters).
+ *  Note 2 : Loading a dictionary involves building tables.
+ *           It's also a CPU consuming operation, with non-negligible impact on latency.
+ *           Tables are dependent on compression parameters, and for this reason,
+ *           compression parameters can no longer be changed after loading a dictionary.
+ *  Note 3 :`dict` content will be copied internally.
+ *           Use experimental ZSTD_CCtx_loadDictionary_byReference() to reference content instead.
+ *           In such a case, dictionary buffer must outlive its users.
+ *  Note 4 : Use ZSTD_CCtx_loadDictionary_advanced()
+ *           to precisely select how dictionary content must be interpreted. */
+ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);
+
+/*! ZSTD_CCtx_refCDict() :
+ *  Reference a prepared dictionary, to be used for all next compressed frames.
+ *  Note that compression parameters are enforced from within CDict,
+ *  and supersede any compression parameter previously set within CCtx.
+ *  The parameters ignored are labelled as "superseded-by-cdict" in the ZSTD_cParameter enum docs.
+ *  The ignored parameters will be used again if the CCtx is returned to no-dictionary mode.
+ *  The dictionary will remain valid for future compressed frames using same CCtx.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Special : Referencing a NULL CDict means "return to no-dictionary mode".
+ *  Note 1 : Currently, only one dictionary can be managed.
+ *           Referencing a new dictionary effectively "discards" any previous one.
+ *  Note 2 : CDict is just referenced, its lifetime must outlive its usage within CCtx. */
+ZSTDLIB_API size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict);
+
+/*! ZSTD_CCtx_refPrefix() :
+ *  Reference a prefix (single-usage dictionary) for next compressed frame.
+ *  A prefix is **only used once**. Tables are discarded at end of frame (ZSTD_e_end).
+ *  Decompression will need same prefix to properly regenerate data.
+ *  Compressing with a prefix is similar in outcome as performing a diff and compressing it,
+ *  but performs much faster, especially during decompression (compression speed is tunable with compression level).
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Special: Adding any prefix (including NULL) invalidates any previous prefix or dictionary
+ *  Note 1 : Prefix buffer is referenced. It **must** outlive compression.
+ *           Its content must remain unmodified during compression.
+ *  Note 2 : If the intention is to diff some large src data blob with some prior version of itself,
+ *           ensure that the window size is large enough to contain the entire source.
+ *           See ZSTD_c_windowLog.
+ *  Note 3 : Referencing a prefix involves building tables, which are dependent on compression parameters.
+ *           It's a CPU consuming operation, with non-negligible impact on latency.
+ *           If there is a need to use the same prefix multiple times, consider loadDictionary instead.
+ *  Note 4 : By default, the prefix is interpreted as raw content (ZSTD_dct_rawContent).
+ *           Use experimental ZSTD_CCtx_refPrefix_advanced() to alter dictionary interpretation. */
+ZSTDLIB_API size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx,
+                                 const void* prefix, size_t prefixSize);
+
+/*! ZSTD_DCtx_loadDictionary() :
+ *  Create an internal DDict from dict buffer,
+ *  to be used to decompress next frames.
+ *  The dictionary remains valid for all future frames, until explicitly invalidated.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Special : Adding a NULL (or 0-size) dictionary invalidates any previous dictionary,
+ *            meaning "return to no-dictionary mode".
+ *  Note 1 : Loading a dictionary involves building tables,
+ *           which has a non-negligible impact on CPU usage and latency.
+ *           It's recommended to "load once, use many times", to amortize the cost
+ *  Note 2 :`dict` content will be copied internally, so `dict` can be released after loading.
+ *           Use ZSTD_DCtx_loadDictionary_byReference() to reference dictionary content instead.
+ *  Note 3 : Use ZSTD_DCtx_loadDictionary_advanced() to take control of
+ *           how dictionary content is loaded and interpreted.
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
+
+/*! ZSTD_DCtx_refDDict() :
+ *  Reference a prepared dictionary, to be used to decompress next frames.
+ *  The dictionary remains active for decompression of future frames using same DCtx.
+ *
+ *  If called with ZSTD_d_refMultipleDDicts enabled, repeated calls of this function
+ *  will store the DDict references in a table, and the DDict used for decompression
+ *  will be determined at decompression time, as per the dict ID in the frame.
+ *  The memory for the table is allocated on the first call to refDDict, and can be
+ *  freed with ZSTD_freeDCtx().
+ *
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Note 1 : Currently, only one dictionary can be managed.
+ *           Referencing a new dictionary effectively "discards" any previous one.
+ *  Special: referencing a NULL DDict means "return to no-dictionary mode".
+ *  Note 2 : DDict is just referenced, its lifetime must outlive its usage from DCtx.
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);
+
+/*! ZSTD_DCtx_refPrefix() :
+ *  Reference a prefix (single-usage dictionary) to decompress next frame.
+ *  This is the reverse operation of ZSTD_CCtx_refPrefix(),
+ *  and must use the same prefix as the one used during compression.
+ *  Prefix is **only used once**. Reference is discarded at end of frame.
+ *  End of frame is reached when ZSTD_decompressStream() returns 0.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ *  Note 1 : Adding any prefix (including NULL) invalidates any previously set prefix or dictionary
+ *  Note 2 : Prefix buffer is referenced. It **must** outlive decompression.
+ *           Prefix buffer must remain unmodified up to the end of frame,
+ *           reached when ZSTD_decompressStream() returns 0.
+ *  Note 3 : By default, the prefix is treated as raw content (ZSTD_dct_rawContent).
+ *           Use ZSTD_CCtx_refPrefix_advanced() to alter dictMode (Experimental section)
+ *  Note 4 : Referencing a raw content prefix has almost no cpu nor memory cost.
+ *           A full dictionary is more costly, as it requires building tables.
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_refPrefix(ZSTD_DCtx* dctx,
+                                 const void* prefix, size_t prefixSize);
+
+/* ===   Memory management   === */
+
+/*! ZSTD_sizeof_*() :
+ *  These functions give the _current_ memory usage of selected object.
+ *  Note that object memory usage can evolve (increase or decrease) over time. */
+ZSTDLIB_API size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx);
+ZSTDLIB_API size_t ZSTD_sizeof_DCtx(const ZSTD_DCtx* dctx);
+ZSTDLIB_API size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs);
+ZSTDLIB_API size_t ZSTD_sizeof_DStream(const ZSTD_DStream* zds);
+ZSTDLIB_API size_t ZSTD_sizeof_CDict(const ZSTD_CDict* cdict);
+ZSTDLIB_API size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
+
+#endif  /* ZSTD_H_235446 */
+
+
+/* **************************************************************************************
+ *   ADVANCED AND EXPERIMENTAL FUNCTIONS
+ ****************************************************************************************
+ * The definitions in the following section are considered experimental.
+ * They are provided for advanced scenarios.
+ * They should never be used with a dynamic library, as prototypes may change in the future.
+ * Use them only in association with static linking.
+ * ***************************************************************************************/
+
+#if !defined(ZSTD_H_ZSTD_STATIC_LINKING_ONLY)
+#define ZSTD_H_ZSTD_STATIC_LINKING_ONLY
+
+/* **************************************************************************************
+ *   experimental API (static linking only)
+ ****************************************************************************************
+ * The following symbols and constants
+ * are not planned to join "stable API" status in the near future.
+ * They can still change in future versions.
+ * Some of them are planned to remain in the static_only section indefinitely.
+ * Some of them might be removed in the future (especially when redundant with existing stable functions)
+ * ***************************************************************************************/
+
+#define ZSTD_FRAMEHEADERSIZE_PREFIX(format) ((format) == ZSTD_f_zstd1 ? 5 : 1)   /* minimum input size required to query frame header size */
+#define ZSTD_FRAMEHEADERSIZE_MIN(format)    ((format) == ZSTD_f_zstd1 ? 6 : 2)
+#define ZSTD_FRAMEHEADERSIZE_MAX   18   /* can be useful for static allocation */
+#define ZSTD_SKIPPABLEHEADERSIZE    8
+
+/* compression parameter bounds */
+#define ZSTD_WINDOWLOG_MAX_32    30
+#define ZSTD_WINDOWLOG_MAX_64    31
+#define ZSTD_WINDOWLOG_MAX     ((int)(sizeof(size_t) == 4 ? ZSTD_WINDOWLOG_MAX_32 : ZSTD_WINDOWLOG_MAX_64))
+#define ZSTD_WINDOWLOG_MIN       10
+#define ZSTD_HASHLOG_MAX       ((ZSTD_WINDOWLOG_MAX < 30) ? ZSTD_WINDOWLOG_MAX : 30)
+#define ZSTD_HASHLOG_MIN          6
+#define ZSTD_CHAINLOG_MAX_32     29
+#define ZSTD_CHAINLOG_MAX_64     30
+#define ZSTD_CHAINLOG_MAX      ((int)(sizeof(size_t) == 4 ? ZSTD_CHAINLOG_MAX_32 : ZSTD_CHAINLOG_MAX_64))
+#define ZSTD_CHAINLOG_MIN        ZSTD_HASHLOG_MIN
+#define ZSTD_SEARCHLOG_MAX      (ZSTD_WINDOWLOG_MAX-1)
+#define ZSTD_SEARCHLOG_MIN        1
+#define ZSTD_MINMATCH_MAX         7   /* only for ZSTD_fast, other strategies are limited to 6 */
+#define ZSTD_MINMATCH_MIN         3   /* only for ZSTD_btopt+, faster strategies are limited to 4 */
+#define ZSTD_TARGETLENGTH_MAX    ZSTD_BLOCKSIZE_MAX
+#define ZSTD_TARGETLENGTH_MIN     0   /* note : comparing this constant to an unsigned results in a tautological test */
+#define ZSTD_STRATEGY_MIN        ZSTD_fast
+#define ZSTD_STRATEGY_MAX        ZSTD_btultra2
+
+
+#define ZSTD_OVERLAPLOG_MIN       0
+#define ZSTD_OVERLAPLOG_MAX       9
+
+#define ZSTD_WINDOWLOG_LIMIT_DEFAULT 27   /* by default, the streaming decoder will refuse any frame
+                                           * requiring larger than (1<<ZSTD_WINDOWLOG_LIMIT_DEFAULT) window size,
+                                           * to preserve host's memory from unreasonable requirements.
+                                           * This limit can be overridden using ZSTD_DCtx_setParameter(,ZSTD_d_windowLogMax,).
+                                           * The limit does not apply for one-pass decoders (such as ZSTD_decompress()), since no additional memory is allocated */
+
+
+/* LDM parameter bounds */
+#define ZSTD_LDM_HASHLOG_MIN      ZSTD_HASHLOG_MIN
+#define ZSTD_LDM_HASHLOG_MAX      ZSTD_HASHLOG_MAX
+#define ZSTD_LDM_MINMATCH_MIN        4
+#define ZSTD_LDM_MINMATCH_MAX     4096
+#define ZSTD_LDM_BUCKETSIZELOG_MIN   1
+#define ZSTD_LDM_BUCKETSIZELOG_MAX   8
+#define ZSTD_LDM_HASHRATELOG_MIN     0
+#define ZSTD_LDM_HASHRATELOG_MAX (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN)
+
+/* Advanced parameter bounds */
+#define ZSTD_TARGETCBLOCKSIZE_MIN   64
+#define ZSTD_TARGETCBLOCKSIZE_MAX   ZSTD_BLOCKSIZE_MAX
+#define ZSTD_SRCSIZEHINT_MIN        0
+#define ZSTD_SRCSIZEHINT_MAX        INT_MAX
+
+/* internal */
+#define ZSTD_HASHLOG3_MAX           17
+
+
+/* ---  Advanced types  --- */
+
+typedef struct ZSTD_CCtx_params_s ZSTD_CCtx_params;
+
+typedef struct {
+    unsigned int offset;      /* The offset of the match. (NOT the same as the offset code)
+                               * If offset == 0 and matchLength == 0, this sequence represents the last
+                               * literals in the block of litLength size.
+                               */
+
+    unsigned int litLength;   /* Literal length of the sequence. */
+    unsigned int matchLength; /* Match length of the sequence. */
+
+                              /* Note: Users of this API may provide a sequence with matchLength == litLength == offset == 0.
+                               * In this case, we will treat the sequence as a marker for a block boundary.
+                               */
+
+    unsigned int rep;         /* Represents which repeat offset is represented by the field 'offset'.
+                               * Ranges from [0, 3].
+                               *
+                               * Repeat offsets are essentially previous offsets from previous sequences sorted in
+                               * recency order. For more detail, see doc/zstd_compression_format.md
+                               *
+                               * If rep == 0, then 'offset' does not contain a repeat offset.
+                               * If rep > 0:
+                               *  If litLength != 0:
+                               *      rep == 1 --> offset == repeat_offset_1
+                               *      rep == 2 --> offset == repeat_offset_2
+                               *      rep == 3 --> offset == repeat_offset_3
+                               *  If litLength == 0:
+                               *      rep == 1 --> offset == repeat_offset_2
+                               *      rep == 2 --> offset == repeat_offset_3
+                               *      rep == 3 --> offset == repeat_offset_1 - 1
+                               *
+                               * Note: This field is optional. ZSTD_generateSequences() will calculate the value of
+                               * 'rep', but repeat offsets do not necessarily need to be calculated from an external
+                               * sequence provider's perspective. For example, ZSTD_compressSequences() does not
+                               * use this 'rep' field at all (as of now).
+                               */
+} ZSTD_Sequence;
+
+typedef struct {
+    unsigned windowLog;       /*< largest match distance : larger == more compression, more memory needed during decompression */
+    unsigned chainLog;        /*< fully searched segment : larger == more compression, slower, more memory (useless for fast) */
+    unsigned hashLog;         /*< dispatch table : larger == faster, more memory */
+    unsigned searchLog;       /*< nb of searches : larger == more compression, slower */
+    unsigned minMatch;        /*< match length searched : larger == faster decompression, sometimes less compression */
+    unsigned targetLength;    /*< acceptable match size for optimal parser (only) : larger == more compression, slower */
+    ZSTD_strategy strategy;   /*< see ZSTD_strategy definition above */
+} ZSTD_compressionParameters;
+
+typedef struct {
+    int contentSizeFlag; /*< 1: content size will be in frame header (when known) */
+    int checksumFlag;    /*< 1: generate a 32-bits checksum using XXH64 algorithm at end of frame, for error detection */
+    int noDictIDFlag;    /*< 1: no dictID will be saved into frame header (dictID is only useful for dictionary compression) */
+} ZSTD_frameParameters;
+
+typedef struct {
+    ZSTD_compressionParameters cParams;
+    ZSTD_frameParameters fParams;
+} ZSTD_parameters;
+
+typedef enum {
+    ZSTD_dct_auto = 0,       /* dictionary is "full" when starting with ZSTD_MAGIC_DICTIONARY, otherwise it is "rawContent" */
+    ZSTD_dct_rawContent = 1, /* ensures dictionary is always loaded as rawContent, even if it starts with ZSTD_MAGIC_DICTIONARY */
+    ZSTD_dct_fullDict = 2    /* refuses to load a dictionary if it does not respect Zstandard's specification, starting with ZSTD_MAGIC_DICTIONARY */
+} ZSTD_dictContentType_e;
+
+typedef enum {
+    ZSTD_dlm_byCopy = 0,  /*< Copy dictionary content internally */
+    ZSTD_dlm_byRef = 1    /*< Reference dictionary content -- the dictionary buffer must outlive its users. */
+} ZSTD_dictLoadMethod_e;
+
+typedef enum {
+    ZSTD_f_zstd1 = 0,           /* zstd frame format, specified in zstd_compression_format.md (default) */
+    ZSTD_f_zstd1_magicless = 1  /* Variant of zstd frame format, without initial 4-bytes magic number.
+                                 * Useful to save 4 bytes per generated frame.
+                                 * Decoder cannot recognise automatically this format, requiring this instruction. */
+} ZSTD_format_e;
+
+typedef enum {
+    /* Note: this enum controls ZSTD_d_forceIgnoreChecksum */
+    ZSTD_d_validateChecksum = 0,
+    ZSTD_d_ignoreChecksum = 1
+} ZSTD_forceIgnoreChecksum_e;
+
+typedef enum {
+    /* Note: this enum controls ZSTD_d_refMultipleDDicts */
+    ZSTD_rmd_refSingleDDict = 0,
+    ZSTD_rmd_refMultipleDDicts = 1
+} ZSTD_refMultipleDDicts_e;
+
+typedef enum {
+    /* Note: this enum and the behavior it controls are effectively internal
+     * implementation details of the compressor. They are expected to continue
+     * to evolve and should be considered only in the context of extremely
+     * advanced performance tuning.
+     *
+     * Zstd currently supports the use of a CDict in three ways:
+     *
+     * - The contents of the CDict can be copied into the working context. This
+     *   means that the compression can search both the dictionary and input
+     *   while operating on a single set of internal tables. This makes
+     *   the compression faster per-byte of input. However, the initial copy of
+     *   the CDict's tables incurs a fixed cost at the beginning of the
+     *   compression. For small compressions (< 8 KB), that copy can dominate
+     *   the cost of the compression.
+     *
+     * - The CDict's tables can be used in-place. In this model, compression is
+     *   slower per input byte, because the compressor has to search two sets of
+     *   tables. However, this model incurs no start-up cost (as long as the
+     *   working context's tables can be reused). For small inputs, this can be
+     *   faster than copying the CDict's tables.
+     *
+     * - The CDict's tables are not used at all, and instead we use the working
+     *   context alone to reload the dictionary and use params based on the source
+     *   size. See ZSTD_compress_insertDictionary() and ZSTD_compress_usingDict().
+     *   This method is effective when the dictionary sizes are very small relative
+     *   to the input size, and the input size is fairly large to begin with.
+     *
+     * Zstd has a simple internal heuristic that selects which strategy to use
+     * at the beginning of a compression. However, if experimentation shows that
+     * Zstd is making poor choices, it is possible to override that choice with
+     * this enum.
+     */
+    ZSTD_dictDefaultAttach = 0, /* Use the default heuristic. */
+    ZSTD_dictForceAttach   = 1, /* Never copy the dictionary. */
+    ZSTD_dictForceCopy     = 2, /* Always copy the dictionary. */
+    ZSTD_dictForceLoad     = 3  /* Always reload the dictionary */
+} ZSTD_dictAttachPref_e;
+
+typedef enum {
+  ZSTD_lcm_auto = 0,          /*< Automatically determine the compression mode based on the compression level.
+                               *   Negative compression levels will be uncompressed, and positive compression
+                               *   levels will be compressed. */
+  ZSTD_lcm_huffman = 1,       /*< Always attempt Huffman compression. Uncompressed literals will still be
+                               *   emitted if Huffman compression is not profitable. */
+  ZSTD_lcm_uncompressed = 2   /*< Always emit uncompressed literals. */
+} ZSTD_literalCompressionMode_e;
+
+
+/* *************************************
+*  Frame size functions
+***************************************/
+
+/*! ZSTD_findDecompressedSize() :
+ *  `src` should point to the start of a series of ZSTD encoded and/or skippable frames
+ *  `srcSize` must be the _exact_ size of this series
+ *       (i.e. there should be a frame boundary at `src + srcSize`)
+ *  @return : - decompressed size of all data in all successive frames
+ *            - if the decompressed size cannot be determined: ZSTD_CONTENTSIZE_UNKNOWN
+ *            - if an error occurred: ZSTD_CONTENTSIZE_ERROR
+ *
+ *   note 1 : decompressed size is an optional field, that may not be present, especially in streaming mode.
+ *            When `return==ZSTD_CONTENTSIZE_UNKNOWN`, data to decompress could be any size.
+ *            In which case, it's necessary to use streaming mode to decompress data.
+ *   note 2 : decompressed size is always present when compression is done with ZSTD_compress()
+ *   note 3 : decompressed size can be very large (64-bits value),
+ *            potentially larger than what local system can handle as a single memory segment.
+ *            In which case, it's necessary to use streaming mode to decompress data.
+ *   note 4 : If source is untrusted, decompressed size could be wrong or intentionally modified.
+ *            Always ensure result fits within application's authorized limits.
+ *            Each application can set its own limits.
+ *   note 5 : ZSTD_findDecompressedSize handles multiple frames, and so it must traverse the input to
+ *            read each contained frame header.  This is fast as most of the data is skipped,
+ *            however it does mean that all frame data must be present and valid. */
+ZSTDLIB_API unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize);
+
+/*! ZSTD_decompressBound() :
+ *  `src` should point to the start of a series of ZSTD encoded and/or skippable frames
+ *  `srcSize` must be the _exact_ size of this series
+ *       (i.e. there should be a frame boundary at `src + srcSize`)
+ *  @return : - upper-bound for the decompressed size of all data in all successive frames
+ *            - if an error occurred: ZSTD_CONTENTSIZE_ERROR
+ *
+ *  note 1  : an error can occur if `src` contains an invalid or incorrectly formatted frame.
+ *  note 2  : the upper-bound is exact when the decompressed size field is available in every ZSTD encoded frame of `src`.
+ *            in this case, `ZSTD_findDecompressedSize` and `ZSTD_decompressBound` return the same value.
+ *  note 3  : when the decompressed size field isn't available, the upper-bound for that frame is calculated by:
+ *              upper-bound = # blocks * min(128 KB, Window_Size)
+ */
+ZSTDLIB_API unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize);
+
+/*! ZSTD_frameHeaderSize() :
+ *  srcSize must be >= ZSTD_FRAMEHEADERSIZE_PREFIX.
+ * @return : size of the Frame Header,
+ *           or an error code (if srcSize is too small) */
+ZSTDLIB_API size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize);
+
+typedef enum {
+  ZSTD_sf_noBlockDelimiters = 0,         /* Representation of ZSTD_Sequence has no block delimiters, sequences only */
+  ZSTD_sf_explicitBlockDelimiters = 1    /* Representation of ZSTD_Sequence contains explicit block delimiters */
+} ZSTD_sequenceFormat_e;
+
+/*! ZSTD_generateSequences() :
+ * Generate sequences using ZSTD_compress2, given a source buffer.
+ *
+ * Each block will end with a dummy sequence
+ * with offset == 0, matchLength == 0, and litLength == length of last literals.
+ * litLength may be == 0, and if so, then the sequence of (of: 0 ml: 0 ll: 0)
+ * simply acts as a block delimiter.
+ *
+ * zc can be used to insert custom compression params.
+ * This function invokes ZSTD_compress2
+ *
+ * The output of this function can be fed into ZSTD_compressSequences() with CCtx
+ * setting of ZSTD_c_blockDelimiters as ZSTD_sf_explicitBlockDelimiters
+ * @return : number of sequences generated
+ */
+
+ZSTDLIB_API size_t ZSTD_generateSequences(ZSTD_CCtx* zc, ZSTD_Sequence* outSeqs,
+                                          size_t outSeqsSize, const void* src, size_t srcSize);
+
+/*! ZSTD_mergeBlockDelimiters() :
+ * Given an array of ZSTD_Sequence, remove all sequences that represent block delimiters/last literals
+ * by merging them into into the literals of the next sequence.
+ *
+ * As such, the final generated result has no explicit representation of block boundaries,
+ * and the final last literals segment is not represented in the sequences.
+ *
+ * The output of this function can be fed into ZSTD_compressSequences() with CCtx
+ * setting of ZSTD_c_blockDelimiters as ZSTD_sf_noBlockDelimiters
+ * @return : number of sequences left after merging
+ */
+ZSTDLIB_API size_t ZSTD_mergeBlockDelimiters(ZSTD_Sequence* sequences, size_t seqsSize);
+
+/*! ZSTD_compressSequences() :
+ * Compress an array of ZSTD_Sequence, generated from the original source buffer, into dst.
+ * If a dictionary is included, then the cctx should reference the dict. (see: ZSTD_CCtx_refCDict(), ZSTD_CCtx_loadDictionary(), etc.)
+ * The entire source is compressed into a single frame.
+ *
+ * The compression behavior changes based on cctx params. In particular:
+ *    If ZSTD_c_blockDelimiters == ZSTD_sf_noBlockDelimiters, the array of ZSTD_Sequence is expected to contain
+ *    no block delimiters (defined in ZSTD_Sequence). Block boundaries are roughly determined based on
+ *    the block size derived from the cctx, and sequences may be split. This is the default setting.
+ *
+ *    If ZSTD_c_blockDelimiters == ZSTD_sf_explicitBlockDelimiters, the array of ZSTD_Sequence is expected to contain
+ *    block delimiters (defined in ZSTD_Sequence). Behavior is undefined if no block delimiters are provided.
+ *
+ *    If ZSTD_c_validateSequences == 0, this function will blindly accept the sequences provided. Invalid sequences cause undefined
+ *    behavior. If ZSTD_c_validateSequences == 1, then if sequence is invalid (see doc/zstd_compression_format.md for
+ *    specifics regarding offset/matchlength requirements) then the function will bail out and return an error.
+ *
+ *    In addition to the two adjustable experimental params, there are other important cctx params.
+ *    - ZSTD_c_minMatch MUST be set as less than or equal to the smallest match generated by the match finder. It has a minimum value of ZSTD_MINMATCH_MIN.
+ *    - ZSTD_c_compressionLevel accordingly adjusts the strength of the entropy coder, as it would in typical compression.
+ *    - ZSTD_c_windowLog affects offset validation: this function will return an error at higher debug levels if a provided offset
+ *      is larger than what the spec allows for a given window log and dictionary (if present). See: doc/zstd_compression_format.md
+ *
+ * Note: Repcodes are, as of now, always re-calculated within this function, so ZSTD_Sequence::rep is unused.
+ * Note 2: Once we integrate ability to ingest repcodes, the explicit block delims mode must respect those repcodes exactly,
+ *         and cannot emit an RLE block that disagrees with the repcode history
+ * @return : final compressed size or a ZSTD error.
+ */
+ZSTDLIB_API size_t ZSTD_compressSequences(ZSTD_CCtx* const cctx, void* dst, size_t dstSize,
+                                  const ZSTD_Sequence* inSeqs, size_t inSeqsSize,
+                                  const void* src, size_t srcSize);
+
+
+/*! ZSTD_writeSkippableFrame() :
+ * Generates a zstd skippable frame containing data given by src, and writes it to dst buffer.
+ *
+ * Skippable frames begin with a a 4-byte magic number. There are 16 possible choices of magic number,
+ * ranging from ZSTD_MAGIC_SKIPPABLE_START to ZSTD_MAGIC_SKIPPABLE_START+15.
+ * As such, the parameter magicVariant controls the exact skippable frame magic number variant used, so
+ * the magic number used will be ZSTD_MAGIC_SKIPPABLE_START + magicVariant.
+ *
+ * Returns an error if destination buffer is not large enough, if the source size is not representable
+ * with a 4-byte unsigned int, or if the parameter magicVariant is greater than 15 (and therefore invalid).
+ *
+ * @return : number of bytes written or a ZSTD error.
+ */
+ZSTDLIB_API size_t ZSTD_writeSkippableFrame(void* dst, size_t dstCapacity,
+                                            const void* src, size_t srcSize, unsigned magicVariant);
+
+
+/* *************************************
+*  Memory management
+***************************************/
+
+/*! ZSTD_estimate*() :
+ *  These functions make it possible to estimate memory usage
+ *  of a future {D,C}Ctx, before its creation.
+ *
+ *  ZSTD_estimateCCtxSize() will provide a memory budget large enough
+ *  for any compression level up to selected one.
+ *  Note : Unlike ZSTD_estimateCStreamSize*(), this estimate
+ *         does not include space for a window buffer.
+ *         Therefore, the estimation is only guaranteed for single-shot compressions, not streaming.
+ *  The estimate will assume the input may be arbitrarily large,
+ *  which is the worst case.
+ *
+ *  When srcSize can be bound by a known and rather "small" value,
+ *  this fact can be used to provide a tighter estimation
+ *  because the CCtx compression context will need less memory.
+ *  This tighter estimation can be provided by more advanced functions
+ *  ZSTD_estimateCCtxSize_usingCParams(), which can be used in tandem with ZSTD_getCParams(),
+ *  and ZSTD_estimateCCtxSize_usingCCtxParams(), which can be used in tandem with ZSTD_CCtxParams_setParameter().
+ *  Both can be used to estimate memory using custom compression parameters and arbitrary srcSize limits.
+ *
+ *  Note 2 : only single-threaded compression is supported.
+ *  ZSTD_estimateCCtxSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
+ */
+ZSTDLIB_API size_t ZSTD_estimateCCtxSize(int compressionLevel);
+ZSTDLIB_API size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams);
+ZSTDLIB_API size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params);
+ZSTDLIB_API size_t ZSTD_estimateDCtxSize(void);
+
+/*! ZSTD_estimateCStreamSize() :
+ *  ZSTD_estimateCStreamSize() will provide a budget large enough for any compression level up to selected one.
+ *  It will also consider src size to be arbitrarily "large", which is worst case.
+ *  If srcSize is known to always be small, ZSTD_estimateCStreamSize_usingCParams() can provide a tighter estimation.
+ *  ZSTD_estimateCStreamSize_usingCParams() can be used in tandem with ZSTD_getCParams() to create cParams from compressionLevel.
+ *  ZSTD_estimateCStreamSize_usingCCtxParams() can be used in tandem with ZSTD_CCtxParams_setParameter(). Only single-threaded compression is supported. This function will return an error code if ZSTD_c_nbWorkers is >= 1.
+ *  Note : CStream size estimation is only correct for single-threaded compression.
+ *  ZSTD_DStream memory budget depends on window Size.
+ *  This information can be passed manually, using ZSTD_estimateDStreamSize,
+ *  or deducted from a valid frame Header, using ZSTD_estimateDStreamSize_fromFrame();
+ *  Note : if streaming is init with function ZSTD_init?Stream_usingDict(),
+ *         an internal ?Dict will be created, which additional size is not estimated here.
+ *         In this case, get total size by adding ZSTD_estimate?DictSize */
+ZSTDLIB_API size_t ZSTD_estimateCStreamSize(int compressionLevel);
+ZSTDLIB_API size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams);
+ZSTDLIB_API size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params);
+ZSTDLIB_API size_t ZSTD_estimateDStreamSize(size_t windowSize);
+ZSTDLIB_API size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, size_t srcSize);
+
+/*! ZSTD_estimate?DictSize() :
+ *  ZSTD_estimateCDictSize() will bet that src size is relatively "small", and content is copied, like ZSTD_createCDict().
+ *  ZSTD_estimateCDictSize_advanced() makes it possible to control compression parameters precisely, like ZSTD_createCDict_advanced().
+ *  Note : dictionaries created by reference (`ZSTD_dlm_byRef`) are logically smaller.
+ */
+ZSTDLIB_API size_t ZSTD_estimateCDictSize(size_t dictSize, int compressionLevel);
+ZSTDLIB_API size_t ZSTD_estimateCDictSize_advanced(size_t dictSize, ZSTD_compressionParameters cParams, ZSTD_dictLoadMethod_e dictLoadMethod);
+ZSTDLIB_API size_t ZSTD_estimateDDictSize(size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod);
+
+/*! ZSTD_initStatic*() :
+ *  Initialize an object using a pre-allocated fixed-size buffer.
+ *  workspace: The memory area to emplace the object into.
+ *             Provided pointer *must be 8-bytes aligned*.
+ *             Buffer must outlive object.
+ *  workspaceSize: Use ZSTD_estimate*Size() to determine
+ *                 how large workspace must be to support target scenario.
+ * @return : pointer to object (same address as workspace, just different type),
+ *           or NULL if error (size too small, incorrect alignment, etc.)
+ *  Note : zstd will never resize nor malloc() when using a static buffer.
+ *         If the object requires more memory than available,
+ *         zstd will just error out (typically ZSTD_error_memory_allocation).
+ *  Note 2 : there is no corresponding "free" function.
+ *           Since workspace is allocated externally, it must be freed externally too.
+ *  Note 3 : cParams : use ZSTD_getCParams() to convert a compression level
+ *           into its associated cParams.
+ *  Limitation 1 : currently not compatible with internal dictionary creation, triggered by
+ *                 ZSTD_CCtx_loadDictionary(), ZSTD_initCStream_usingDict() or ZSTD_initDStream_usingDict().
+ *  Limitation 2 : static cctx currently not compatible with multi-threading.
+ *  Limitation 3 : static dctx is incompatible with legacy support.
+ */
+ZSTDLIB_API ZSTD_CCtx*    ZSTD_initStaticCCtx(void* workspace, size_t workspaceSize);
+ZSTDLIB_API ZSTD_CStream* ZSTD_initStaticCStream(void* workspace, size_t workspaceSize);    /*< same as ZSTD_initStaticCCtx() */
+
+ZSTDLIB_API ZSTD_DCtx*    ZSTD_initStaticDCtx(void* workspace, size_t workspaceSize);
+ZSTDLIB_API ZSTD_DStream* ZSTD_initStaticDStream(void* workspace, size_t workspaceSize);    /*< same as ZSTD_initStaticDCtx() */
+
+ZSTDLIB_API const ZSTD_CDict* ZSTD_initStaticCDict(
+                                        void* workspace, size_t workspaceSize,
+                                        const void* dict, size_t dictSize,
+                                        ZSTD_dictLoadMethod_e dictLoadMethod,
+                                        ZSTD_dictContentType_e dictContentType,
+                                        ZSTD_compressionParameters cParams);
+
+ZSTDLIB_API const ZSTD_DDict* ZSTD_initStaticDDict(
+                                        void* workspace, size_t workspaceSize,
+                                        const void* dict, size_t dictSize,
+                                        ZSTD_dictLoadMethod_e dictLoadMethod,
+                                        ZSTD_dictContentType_e dictContentType);
+
+
+/*! Custom memory allocation :
+ *  These prototypes make it possible to pass your own allocation/free functions.
+ *  ZSTD_customMem is provided at creation time, using ZSTD_create*_advanced() variants listed below.
+ *  All allocation/free operations will be completed using these custom variants instead of regular <stdlib.h> ones.
+ */
+typedef void* (*ZSTD_allocFunction) (void* opaque, size_t size);
+typedef void  (*ZSTD_freeFunction) (void* opaque, void* address);
+typedef struct { ZSTD_allocFunction customAlloc; ZSTD_freeFunction customFree; void* opaque; } ZSTD_customMem;
+static
+__attribute__((__unused__))
+ZSTD_customMem const ZSTD_defaultCMem = { NULL, NULL, NULL };  /*< this constant defers to stdlib's functions */
+
+ZSTDLIB_API ZSTD_CCtx*    ZSTD_createCCtx_advanced(ZSTD_customMem customMem);
+ZSTDLIB_API ZSTD_CStream* ZSTD_createCStream_advanced(ZSTD_customMem customMem);
+ZSTDLIB_API ZSTD_DCtx*    ZSTD_createDCtx_advanced(ZSTD_customMem customMem);
+ZSTDLIB_API ZSTD_DStream* ZSTD_createDStream_advanced(ZSTD_customMem customMem);
+
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_advanced(const void* dict, size_t dictSize,
+                                                  ZSTD_dictLoadMethod_e dictLoadMethod,
+                                                  ZSTD_dictContentType_e dictContentType,
+                                                  ZSTD_compressionParameters cParams,
+                                                  ZSTD_customMem customMem);
+
+/* ! Thread pool :
+ * These prototypes make it possible to share a thread pool among multiple compression contexts.
+ * This can limit resources for applications with multiple threads where each one uses
+ * a threaded compression mode (via ZSTD_c_nbWorkers parameter).
+ * ZSTD_createThreadPool creates a new thread pool with a given number of threads.
+ * Note that the lifetime of such pool must exist while being used.
+ * ZSTD_CCtx_refThreadPool assigns a thread pool to a context (use NULL argument value
+ * to use an internal thread pool).
+ * ZSTD_freeThreadPool frees a thread pool, accepts NULL pointer.
+ */
+typedef struct POOL_ctx_s ZSTD_threadPool;
+ZSTDLIB_API ZSTD_threadPool* ZSTD_createThreadPool(size_t numThreads);
+ZSTDLIB_API void ZSTD_freeThreadPool (ZSTD_threadPool* pool);  /* accept NULL pointer */
+ZSTDLIB_API size_t ZSTD_CCtx_refThreadPool(ZSTD_CCtx* cctx, ZSTD_threadPool* pool);
+
+
+/*
+ * This API is temporary and is expected to change or disappear in the future!
+ */
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_advanced2(
+    const void* dict, size_t dictSize,
+    ZSTD_dictLoadMethod_e dictLoadMethod,
+    ZSTD_dictContentType_e dictContentType,
+    const ZSTD_CCtx_params* cctxParams,
+    ZSTD_customMem customMem);
+
+ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict_advanced(
+    const void* dict, size_t dictSize,
+    ZSTD_dictLoadMethod_e dictLoadMethod,
+    ZSTD_dictContentType_e dictContentType,
+    ZSTD_customMem customMem);
+
+
+/* *************************************
+*  Advanced compression functions
+***************************************/
+
+/*! ZSTD_createCDict_byReference() :
+ *  Create a digested dictionary for compression
+ *  Dictionary content is just referenced, not duplicated.
+ *  As a consequence, `dictBuffer` **must** outlive CDict,
+ *  and its content must remain unmodified throughout the lifetime of CDict.
+ *  note: equivalent to ZSTD_createCDict_advanced(), with dictLoadMethod==ZSTD_dlm_byRef */
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_byReference(const void* dictBuffer, size_t dictSize, int compressionLevel);
+
+/*! ZSTD_getDictID_fromCDict() :
+ *  Provides the dictID of the dictionary loaded into `cdict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+ZSTDLIB_API unsigned ZSTD_getDictID_fromCDict(const ZSTD_CDict* cdict);
+
+/*! ZSTD_getCParams() :
+ * @return ZSTD_compressionParameters structure for a selected compression level and estimated srcSize.
+ * `estimatedSrcSize` value is optional, select 0 if not known */
+ZSTDLIB_API ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long estimatedSrcSize, size_t dictSize);
+
+/*! ZSTD_getParams() :
+ *  same as ZSTD_getCParams(), but @return a full `ZSTD_parameters` object instead of sub-component `ZSTD_compressionParameters`.
+ *  All fields of `ZSTD_frameParameters` are set to default : contentSize=1, checksum=0, noDictID=0 */
+ZSTDLIB_API ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long estimatedSrcSize, size_t dictSize);
+
+/*! ZSTD_checkCParams() :
+ *  Ensure param values remain within authorized range.
+ * @return 0 on success, or an error code (can be checked with ZSTD_isError()) */
+ZSTDLIB_API size_t ZSTD_checkCParams(ZSTD_compressionParameters params);
+
+/*! ZSTD_adjustCParams() :
+ *  optimize params for a given `srcSize` and `dictSize`.
+ * `srcSize` can be unknown, in which case use ZSTD_CONTENTSIZE_UNKNOWN.
+ * `dictSize` must be `0` when there is no dictionary.
+ *  cPar can be invalid : all parameters will be clamped within valid range in the @return struct.
+ *  This function never fails (wide contract) */
+ZSTDLIB_API ZSTD_compressionParameters ZSTD_adjustCParams(ZSTD_compressionParameters cPar, unsigned long long srcSize, size_t dictSize);
+
+/*! ZSTD_compress_advanced() :
+ *  Note : this function is now DEPRECATED.
+ *         It can be replaced by ZSTD_compress2(), in combination with ZSTD_CCtx_setParameter() and other parameter setters.
+ *  This prototype will be marked as deprecated and generate compilation warning on reaching v1.5.x */
+ZSTDLIB_API size_t ZSTD_compress_advanced(ZSTD_CCtx* cctx,
+                                          void* dst, size_t dstCapacity,
+                                    const void* src, size_t srcSize,
+                                    const void* dict,size_t dictSize,
+                                          ZSTD_parameters params);
+
+/*! ZSTD_compress_usingCDict_advanced() :
+ *  Note : this function is now REDUNDANT.
+ *         It can be replaced by ZSTD_compress2(), in combination with ZSTD_CCtx_loadDictionary() and other parameter setters.
+ *  This prototype will be marked as deprecated and generate compilation warning in some future version */
+ZSTDLIB_API size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx,
+                                              void* dst, size_t dstCapacity,
+                                        const void* src, size_t srcSize,
+                                        const ZSTD_CDict* cdict,
+                                              ZSTD_frameParameters fParams);
+
+
+/*! ZSTD_CCtx_loadDictionary_byReference() :
+ *  Same as ZSTD_CCtx_loadDictionary(), but dictionary content is referenced, instead of being copied into CCtx.
+ *  It saves some memory, but also requires that `dict` outlives its usage within `cctx` */
+ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_byReference(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);
+
+/*! ZSTD_CCtx_loadDictionary_advanced() :
+ *  Same as ZSTD_CCtx_loadDictionary(), but gives finer control over
+ *  how to load the dictionary (by copy ? by reference ?)
+ *  and how to interpret it (automatic ? force raw mode ? full mode only ?) */
+ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_advanced(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType);
+
+/*! ZSTD_CCtx_refPrefix_advanced() :
+ *  Same as ZSTD_CCtx_refPrefix(), but gives finer control over
+ *  how to interpret prefix content (automatic ? force raw mode (default) ? full mode only ?) */
+ZSTDLIB_API size_t ZSTD_CCtx_refPrefix_advanced(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType);
+
+/* ===   experimental parameters   === */
+/* these parameters can be used with ZSTD_setParameter()
+ * they are not guaranteed to remain supported in the future */
+
+ /* Enables rsyncable mode,
+  * which makes compressed files more rsync friendly
+  * by adding periodic synchronization points to the compressed data.
+  * The target average block size is ZSTD_c_jobSize / 2.
+  * It's possible to modify the job size to increase or decrease
+  * the granularity of the synchronization point.
+  * Once the jobSize is smaller than the window size,
+  * it will result in compression ratio degradation.
+  * NOTE 1: rsyncable mode only works when multithreading is enabled.
+  * NOTE 2: rsyncable performs poorly in combination with long range mode,
+  * since it will decrease the effectiveness of synchronization points,
+  * though mileage may vary.
+  * NOTE 3: Rsyncable mode limits maximum compression speed to ~400 MB/s.
+  * If the selected compression level is already running significantly slower,
+  * the overall speed won't be significantly impacted.
+  */
+ #define ZSTD_c_rsyncable ZSTD_c_experimentalParam1
+
+/* Select a compression format.
+ * The value must be of type ZSTD_format_e.
+ * See ZSTD_format_e enum definition for details */
+#define ZSTD_c_format ZSTD_c_experimentalParam2
+
+/* Force back-reference distances to remain < windowSize,
+ * even when referencing into Dictionary content (default:0) */
+#define ZSTD_c_forceMaxWindow ZSTD_c_experimentalParam3
+
+/* Controls whether the contents of a CDict
+ * are used in place, or copied into the working context.
+ * Accepts values from the ZSTD_dictAttachPref_e enum.
+ * See the comments on that enum for an explanation of the feature. */
+#define ZSTD_c_forceAttachDict ZSTD_c_experimentalParam4
+
+/* Controls how the literals are compressed (default is auto).
+ * The value must be of type ZSTD_literalCompressionMode_e.
+ * See ZSTD_literalCompressionMode_t enum definition for details.
+ */
+#define ZSTD_c_literalCompressionMode ZSTD_c_experimentalParam5
+
+/* Tries to fit compressed block size to be around targetCBlockSize.
+ * No target when targetCBlockSize == 0.
+ * There is no guarantee on compressed block size (default:0) */
+#define ZSTD_c_targetCBlockSize ZSTD_c_experimentalParam6
+
+/* User's best guess of source size.
+ * Hint is not valid when srcSizeHint == 0.
+ * There is no guarantee that hint is close to actual source size,
+ * but compression ratio may regress significantly if guess considerably underestimates */
+#define ZSTD_c_srcSizeHint ZSTD_c_experimentalParam7
+
+/* Controls whether the new and experimental "dedicated dictionary search
+ * structure" can be used. This feature is still rough around the edges, be
+ * prepared for surprising behavior!
+ *
+ * How to use it:
+ *
+ * When using a CDict, whether to use this feature or not is controlled at
+ * CDict creation, and it must be set in a CCtxParams set passed into that
+ * construction (via ZSTD_createCDict_advanced2()). A compression will then
+ * use the feature or not based on how the CDict was constructed; the value of
+ * this param, set in the CCtx, will have no effect.
+ *
+ * However, when a dictionary buffer is passed into a CCtx, such as via
+ * ZSTD_CCtx_loadDictionary(), this param can be set on the CCtx to control
+ * whether the CDict that is created internally can use the feature or not.
+ *
+ * What it does:
+ *
+ * Normally, the internal data structures of the CDict are analogous to what
+ * would be stored in a CCtx after compressing the contents of a dictionary.
+ * To an approximation, a compression using a dictionary can then use those
+ * data structures to simply continue what is effectively a streaming
+ * compression where the simulated compression of the dictionary left off.
+ * Which is to say, the search structures in the CDict are normally the same
+ * format as in the CCtx.
+ *
+ * It is possible to do better, since the CDict is not like a CCtx: the search
+ * structures are written once during CDict creation, and then are only read
+ * after that, while the search structures in the CCtx are both read and
+ * written as the compression goes along. This means we can choose a search
+ * structure for the dictionary that is read-optimized.
+ *
+ * This feature enables the use of that different structure.
+ *
+ * Note that some of the members of the ZSTD_compressionParameters struct have
+ * different semantics and constraints in the dedicated search structure. It is
+ * highly recommended that you simply set a compression level in the CCtxParams
+ * you pass into the CDict creation call, and avoid messing with the cParams
+ * directly.
+ *
+ * Effects:
+ *
+ * This will only have any effect when the selected ZSTD_strategy
+ * implementation supports this feature. Currently, that's limited to
+ * ZSTD_greedy, ZSTD_lazy, and ZSTD_lazy2.
+ *
+ * Note that this means that the CDict tables can no longer be copied into the
+ * CCtx, so the dict attachment mode ZSTD_dictForceCopy will no longer be
+ * useable. The dictionary can only be attached or reloaded.
+ *
+ * In general, you should expect compression to be faster--sometimes very much
+ * so--and CDict creation to be slightly slower. Eventually, we will probably
+ * make this mode the default.
+ */
+#define ZSTD_c_enableDedicatedDictSearch ZSTD_c_experimentalParam8
+
+/* ZSTD_c_stableInBuffer
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable.
+ *
+ * Tells the compressor that the ZSTD_inBuffer will ALWAYS be the same
+ * between calls, except for the modifications that zstd makes to pos (the
+ * caller must not modify pos). This is checked by the compressor, and
+ * compression will fail if it ever changes. This means the only flush
+ * mode that makes sense is ZSTD_e_end, so zstd will error if ZSTD_e_end
+ * is not used. The data in the ZSTD_inBuffer in the range [src, src + pos)
+ * MUST not be modified during compression or you will get data corruption.
+ *
+ * When this flag is enabled zstd won't allocate an input window buffer,
+ * because the user guarantees it can reference the ZSTD_inBuffer until
+ * the frame is complete. But, it will still allocate an output buffer
+ * large enough to fit a block (see ZSTD_c_stableOutBuffer). This will also
+ * avoid the memcpy() from the input buffer to the input window buffer.
+ *
+ * NOTE: ZSTD_compressStream2() will error if ZSTD_e_end is not used.
+ * That means this flag cannot be used with ZSTD_compressStream().
+ *
+ * NOTE: So long as the ZSTD_inBuffer always points to valid memory, using
+ * this flag is ALWAYS memory safe, and will never access out-of-bounds
+ * memory. However, compression WILL fail if you violate the preconditions.
+ *
+ * WARNING: The data in the ZSTD_inBuffer in the range [dst, dst + pos) MUST
+ * not be modified during compression or you will get data corruption. This
+ * is because zstd needs to reference data in the ZSTD_inBuffer to find
+ * matches. Normally zstd maintains its own window buffer for this purpose,
+ * but passing this flag tells zstd to use the user provided buffer.
+ */
+#define ZSTD_c_stableInBuffer ZSTD_c_experimentalParam9
+
+/* ZSTD_c_stableOutBuffer
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable.
+ *
+ * Tells he compressor that the ZSTD_outBuffer will not be resized between
+ * calls. Specifically: (out.size - out.pos) will never grow. This gives the
+ * compressor the freedom to say: If the compressed data doesn't fit in the
+ * output buffer then return ZSTD_error_dstSizeTooSmall. This allows us to
+ * always decompress directly into the output buffer, instead of decompressing
+ * into an internal buffer and copying to the output buffer.
+ *
+ * When this flag is enabled zstd won't allocate an output buffer, because
+ * it can write directly to the ZSTD_outBuffer. It will still allocate the
+ * input window buffer (see ZSTD_c_stableInBuffer).
+ *
+ * Zstd will check that (out.size - out.pos) never grows and return an error
+ * if it does. While not strictly necessary, this should prevent surprises.
+ */
+#define ZSTD_c_stableOutBuffer ZSTD_c_experimentalParam10
+
+/* ZSTD_c_blockDelimiters
+ * Default is 0 == ZSTD_sf_noBlockDelimiters.
+ *
+ * For use with sequence compression API: ZSTD_compressSequences().
+ *
+ * Designates whether or not the given array of ZSTD_Sequence contains block delimiters
+ * and last literals, which are defined as sequences with offset == 0 and matchLength == 0.
+ * See the definition of ZSTD_Sequence for more specifics.
+ */
+#define ZSTD_c_blockDelimiters ZSTD_c_experimentalParam11
+
+/* ZSTD_c_validateSequences
+ * Default is 0 == disabled. Set to 1 to enable sequence validation.
+ *
+ * For use with sequence compression API: ZSTD_compressSequences().
+ * Designates whether or not we validate sequences provided to ZSTD_compressSequences()
+ * during function execution.
+ *
+ * Without validation, providing a sequence that does not conform to the zstd spec will cause
+ * undefined behavior, and may produce a corrupted block.
+ *
+ * With validation enabled, a if sequence is invalid (see doc/zstd_compression_format.md for
+ * specifics regarding offset/matchlength requirements) then the function will bail out and
+ * return an error.
+ *
+ */
+#define ZSTD_c_validateSequences ZSTD_c_experimentalParam12
+
+/*! ZSTD_CCtx_getParameter() :
+ *  Get the requested compression parameter value, selected by enum ZSTD_cParameter,
+ *  and store it into int* value.
+ * @return : 0, or an error code (which can be tested with ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_getParameter(const ZSTD_CCtx* cctx, ZSTD_cParameter param, int* value);
+
+
+/*! ZSTD_CCtx_params :
+ *  Quick howto :
+ *  - ZSTD_createCCtxParams() : Create a ZSTD_CCtx_params structure
+ *  - ZSTD_CCtxParams_setParameter() : Push parameters one by one into
+ *                                     an existing ZSTD_CCtx_params structure.
+ *                                     This is similar to
+ *                                     ZSTD_CCtx_setParameter().
+ *  - ZSTD_CCtx_setParametersUsingCCtxParams() : Apply parameters to
+ *                                    an existing CCtx.
+ *                                    These parameters will be applied to
+ *                                    all subsequent frames.
+ *  - ZSTD_compressStream2() : Do compression using the CCtx.
+ *  - ZSTD_freeCCtxParams() : Free the memory, accept NULL pointer.
+ *
+ *  This can be used with ZSTD_estimateCCtxSize_advanced_usingCCtxParams()
+ *  for static allocation of CCtx for single-threaded compression.
+ */
+ZSTDLIB_API ZSTD_CCtx_params* ZSTD_createCCtxParams(void);
+ZSTDLIB_API size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params);  /* accept NULL pointer */
+
+/*! ZSTD_CCtxParams_reset() :
+ *  Reset params to default values.
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params);
+
+/*! ZSTD_CCtxParams_init() :
+ *  Initializes the compression parameters of cctxParams according to
+ *  compression level. All other parameters are reset to their default values.
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel);
+
+/*! ZSTD_CCtxParams_init_advanced() :
+ *  Initializes the compression and frame parameters of cctxParams according to
+ *  params. All other parameters are reset to their default values.
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params);
+
+/*! ZSTD_CCtxParams_setParameter() :
+ *  Similar to ZSTD_CCtx_setParameter.
+ *  Set one compression parameter, selected by enum ZSTD_cParameter.
+ *  Parameters must be applied to a ZSTD_CCtx using
+ *  ZSTD_CCtx_setParametersUsingCCtxParams().
+ * @result : a code representing success or failure (which can be tested with
+ *           ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* params, ZSTD_cParameter param, int value);
+
+/*! ZSTD_CCtxParams_getParameter() :
+ * Similar to ZSTD_CCtx_getParameter.
+ * Get the requested value of one compression parameter, selected by enum ZSTD_cParameter.
+ * @result : 0, or an error code (which can be tested with ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_CCtxParams_getParameter(const ZSTD_CCtx_params* params, ZSTD_cParameter param, int* value);
+
+/*! ZSTD_CCtx_setParametersUsingCCtxParams() :
+ *  Apply a set of ZSTD_CCtx_params to the compression context.
+ *  This can be done even after compression is started,
+ *    if nbWorkers==0, this will have no impact until a new compression is started.
+ *    if nbWorkers>=1, new parameters will be picked up at next job,
+ *       with a few restrictions (windowLog, pledgedSrcSize, nbWorkers, jobSize, and overlapLog are not updated).
+ */
+ZSTDLIB_API size_t ZSTD_CCtx_setParametersUsingCCtxParams(
+        ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params);
+
+/*! ZSTD_compressStream2_simpleArgs() :
+ *  Same as ZSTD_compressStream2(),
+ *  but using only integral types as arguments.
+ *  This variant might be helpful for binders from dynamic languages
+ *  which have troubles handling structures containing memory pointers.
+ */
+ZSTDLIB_API size_t ZSTD_compressStream2_simpleArgs (
+                            ZSTD_CCtx* cctx,
+                            void* dst, size_t dstCapacity, size_t* dstPos,
+                      const void* src, size_t srcSize, size_t* srcPos,
+                            ZSTD_EndDirective endOp);
+
+
+/* *************************************
+*  Advanced decompression functions
+***************************************/
+
+/*! ZSTD_isFrame() :
+ *  Tells if the content of `buffer` starts with a valid Frame Identifier.
+ *  Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
+ *  Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
+ *  Note 3 : Skippable Frame Identifiers are considered valid. */
+ZSTDLIB_API unsigned ZSTD_isFrame(const void* buffer, size_t size);
+
+/*! ZSTD_createDDict_byReference() :
+ *  Create a digested dictionary, ready to start decompression operation without startup delay.
+ *  Dictionary content is referenced, and therefore stays in dictBuffer.
+ *  It is important that dictBuffer outlives DDict,
+ *  it must remain read accessible throughout the lifetime of DDict */
+ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize);
+
+/*! ZSTD_DCtx_loadDictionary_byReference() :
+ *  Same as ZSTD_DCtx_loadDictionary(),
+ *  but references `dict` content instead of copying it into `dctx`.
+ *  This saves memory if `dict` remains around.,
+ *  However, it's imperative that `dict` remains accessible (and unmodified) while being used, so it must outlive decompression. */
+ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary_byReference(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
+
+/*! ZSTD_DCtx_loadDictionary_advanced() :
+ *  Same as ZSTD_DCtx_loadDictionary(),
+ *  but gives direct control over
+ *  how to load the dictionary (by copy ? by reference ?)
+ *  and how to interpret it (automatic ? force raw mode ? full mode only ?). */
+ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx, const void* dict, size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType);
+
+/*! ZSTD_DCtx_refPrefix_advanced() :
+ *  Same as ZSTD_DCtx_refPrefix(), but gives finer control over
+ *  how to interpret prefix content (automatic ? force raw mode (default) ? full mode only ?) */
+ZSTDLIB_API size_t ZSTD_DCtx_refPrefix_advanced(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType);
+
+/*! ZSTD_DCtx_setMaxWindowSize() :
+ *  Refuses allocating internal buffers for frames requiring a window size larger than provided limit.
+ *  This protects a decoder context from reserving too much memory for itself (potential attack scenario).
+ *  This parameter is only useful in streaming mode, since no internal buffer is allocated in single-pass mode.
+ *  By default, a decompression context accepts all window sizes <= (1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT)
+ * @return : 0, or an error code (which can be tested using ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx* dctx, size_t maxWindowSize);
+
+/*! ZSTD_DCtx_getParameter() :
+ *  Get the requested decompression parameter value, selected by enum ZSTD_dParameter,
+ *  and store it into int* value.
+ * @return : 0, or an error code (which can be tested with ZSTD_isError()).
+ */
+ZSTDLIB_API size_t ZSTD_DCtx_getParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int* value);
+
+/* ZSTD_d_format
+ * experimental parameter,
+ * allowing selection between ZSTD_format_e input compression formats
+ */
+#define ZSTD_d_format ZSTD_d_experimentalParam1
+/* ZSTD_d_stableOutBuffer
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable.
+ *
+ * Tells the decompressor that the ZSTD_outBuffer will ALWAYS be the same
+ * between calls, except for the modifications that zstd makes to pos (the
+ * caller must not modify pos). This is checked by the decompressor, and
+ * decompression will fail if it ever changes. Therefore the ZSTD_outBuffer
+ * MUST be large enough to fit the entire decompressed frame. This will be
+ * checked when the frame content size is known. The data in the ZSTD_outBuffer
+ * in the range [dst, dst + pos) MUST not be modified during decompression
+ * or you will get data corruption.
+ *
+ * When this flags is enabled zstd won't allocate an output buffer, because
+ * it can write directly to the ZSTD_outBuffer, but it will still allocate
+ * an input buffer large enough to fit any compressed block. This will also
+ * avoid the memcpy() from the internal output buffer to the ZSTD_outBuffer.
+ * If you need to avoid the input buffer allocation use the buffer-less
+ * streaming API.
+ *
+ * NOTE: So long as the ZSTD_outBuffer always points to valid memory, using
+ * this flag is ALWAYS memory safe, and will never access out-of-bounds
+ * memory. However, decompression WILL fail if you violate the preconditions.
+ *
+ * WARNING: The data in the ZSTD_outBuffer in the range [dst, dst + pos) MUST
+ * not be modified during decompression or you will get data corruption. This
+ * is because zstd needs to reference data in the ZSTD_outBuffer to regenerate
+ * matches. Normally zstd maintains its own buffer for this purpose, but passing
+ * this flag tells zstd to use the user provided buffer.
+ */
+#define ZSTD_d_stableOutBuffer ZSTD_d_experimentalParam2
+
+/* ZSTD_d_forceIgnoreChecksum
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable
+ *
+ * Tells the decompressor to skip checksum validation during decompression, regardless
+ * of whether checksumming was specified during compression. This offers some
+ * slight performance benefits, and may be useful for debugging.
+ * Param has values of type ZSTD_forceIgnoreChecksum_e
+ */
+#define ZSTD_d_forceIgnoreChecksum ZSTD_d_experimentalParam3
+
+/* ZSTD_d_refMultipleDDicts
+ * Experimental parameter.
+ * Default is 0 == disabled. Set to 1 to enable
+ *
+ * If enabled and dctx is allocated on the heap, then additional memory will be allocated
+ * to store references to multiple ZSTD_DDict. That is, multiple calls of ZSTD_refDDict()
+ * using a given ZSTD_DCtx, rather than overwriting the previous DDict reference, will instead
+ * store all references. At decompression time, the appropriate dictID is selected
+ * from the set of DDicts based on the dictID in the frame.
+ *
+ * Usage is simply calling ZSTD_refDDict() on multiple dict buffers.
+ *
+ * Param has values of byte ZSTD_refMultipleDDicts_e
+ *
+ * WARNING: Enabling this parameter and calling ZSTD_DCtx_refDDict(), will trigger memory
+ * allocation for the hash table. ZSTD_freeDCtx() also frees this memory.
+ * Memory is allocated as per ZSTD_DCtx::customMem.
+ *
+ * Although this function allocates memory for the table, the user is still responsible for
+ * memory management of the underlying ZSTD_DDict* themselves.
+ */
+#define ZSTD_d_refMultipleDDicts ZSTD_d_experimentalParam4
+
+
+/*! ZSTD_DCtx_setFormat() :
+ *  Instruct the decoder context about what kind of data to decode next.
+ *  This instruction is mandatory to decode data without a fully-formed header,
+ *  such ZSTD_f_zstd1_magicless for example.
+ * @return : 0, or an error code (which can be tested using ZSTD_isError()). */
+ZSTDLIB_API size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format);
+
+/*! ZSTD_decompressStream_simpleArgs() :
+ *  Same as ZSTD_decompressStream(),
+ *  but using only integral types as arguments.
+ *  This can be helpful for binders from dynamic languages
+ *  which have troubles handling structures containing memory pointers.
+ */
+ZSTDLIB_API size_t ZSTD_decompressStream_simpleArgs (
+                            ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity, size_t* dstPos,
+                      const void* src, size_t srcSize, size_t* srcPos);
+
+
+/* ******************************************************************
+*  Advanced streaming functions
+*  Warning : most of these functions are now redundant with the Advanced API.
+*  Once Advanced API reaches "stable" status,
+*  redundant functions will be deprecated, and then at some point removed.
+********************************************************************/
+
+/*=====   Advanced Streaming compression functions  =====*/
+
+/*! ZSTD_initCStream_srcSize() :
+ * This function is deprecated, and equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_refCDict(zcs, NULL); // clear the dictionary (if any)
+ *     ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
+ *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
+ *
+ * pledgedSrcSize must be correct. If it is not known at init time, use
+ * ZSTD_CONTENTSIZE_UNKNOWN. Note that, for compatibility with older programs,
+ * "0" also disables frame content size field. It may be enabled in the future.
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t
+ZSTD_initCStream_srcSize(ZSTD_CStream* zcs,
+                         int compressionLevel,
+                         unsigned long long pledgedSrcSize);
+
+/*! ZSTD_initCStream_usingDict() :
+ * This function is deprecated, and is equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
+ *     ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);
+ *
+ * Creates of an internal CDict (incompatible with static CCtx), except if
+ * dict == NULL or dictSize < 8, in which case no dict is used.
+ * Note: dict is loaded with ZSTD_dct_auto (treated as a full zstd dictionary if
+ * it begins with ZSTD_MAGIC_DICTIONARY, else as raw content) and ZSTD_dlm_byCopy.
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t
+ZSTD_initCStream_usingDict(ZSTD_CStream* zcs,
+                     const void* dict, size_t dictSize,
+                           int compressionLevel);
+
+/*! ZSTD_initCStream_advanced() :
+ * This function is deprecated, and is approximately equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     // Pseudocode: Set each zstd parameter and leave the rest as-is.
+ *     for ((param, value) : params) {
+ *         ZSTD_CCtx_setParameter(zcs, param, value);
+ *     }
+ *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
+ *     ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);
+ *
+ * dict is loaded with ZSTD_dct_auto and ZSTD_dlm_byCopy.
+ * pledgedSrcSize must be correct.
+ * If srcSize is not known at init time, use value ZSTD_CONTENTSIZE_UNKNOWN.
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t
+ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
+                    const void* dict, size_t dictSize,
+                          ZSTD_parameters params,
+                          unsigned long long pledgedSrcSize);
+
+/*! ZSTD_initCStream_usingCDict() :
+ * This function is deprecated, and equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_refCDict(zcs, cdict);
+ *
+ * note : cdict will just be referenced, and must outlive compression session
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict);
+
+/*! ZSTD_initCStream_usingCDict_advanced() :
+ *   This function is DEPRECATED, and is approximately equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     // Pseudocode: Set each zstd frame parameter and leave the rest as-is.
+ *     for ((fParam, value) : fParams) {
+ *         ZSTD_CCtx_setParameter(zcs, fParam, value);
+ *     }
+ *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
+ *     ZSTD_CCtx_refCDict(zcs, cdict);
+ *
+ * same as ZSTD_initCStream_usingCDict(), with control over frame parameters.
+ * pledgedSrcSize must be correct. If srcSize is not known at init time, use
+ * value ZSTD_CONTENTSIZE_UNKNOWN.
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t
+ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
+                               const ZSTD_CDict* cdict,
+                                     ZSTD_frameParameters fParams,
+                                     unsigned long long pledgedSrcSize);
+
+/*! ZSTD_resetCStream() :
+ * This function is deprecated, and is equivalent to:
+ *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+ *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
+ *
+ *  start a new frame, using same parameters from previous frame.
+ *  This is typically useful to skip dictionary loading stage, since it will re-use it in-place.
+ *  Note that zcs must be init at least once before using ZSTD_resetCStream().
+ *  If pledgedSrcSize is not known at reset time, use macro ZSTD_CONTENTSIZE_UNKNOWN.
+ *  If pledgedSrcSize > 0, its value must be correct, as it will be written in header, and controlled at the end.
+ *  For the time being, pledgedSrcSize==0 is interpreted as "srcSize unknown" for compatibility with older programs,
+ *  but it will change to mean "empty" in future version, so use macro ZSTD_CONTENTSIZE_UNKNOWN instead.
+ * @return : 0, or an error code (which can be tested using ZSTD_isError())
+ *  Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);
+
+
+typedef struct {
+    unsigned long long ingested;   /* nb input bytes read and buffered */
+    unsigned long long consumed;   /* nb input bytes actually compressed */
+    unsigned long long produced;   /* nb of compressed bytes generated and buffered */
+    unsigned long long flushed;    /* nb of compressed bytes flushed : not provided; can be tracked from caller side */
+    unsigned currentJobID;         /* MT only : latest started job nb */
+    unsigned nbActiveWorkers;      /* MT only : nb of workers actively compressing at probe time */
+} ZSTD_frameProgression;
+
+/* ZSTD_getFrameProgression() :
+ * tells how much data has been ingested (read from input)
+ * consumed (input actually compressed) and produced (output) for current frame.
+ * Note : (ingested - consumed) is amount of input data buffered internally, not yet compressed.
+ * Aggregates progression inside active worker threads.
+ */
+ZSTDLIB_API ZSTD_frameProgression ZSTD_getFrameProgression(const ZSTD_CCtx* cctx);
+
+/*! ZSTD_toFlushNow() :
+ *  Tell how many bytes are ready to be flushed immediately.
+ *  Useful for multithreading scenarios (nbWorkers >= 1).
+ *  Probe the oldest active job, defined as oldest job not yet entirely flushed,
+ *  and check its output buffer.
+ * @return : amount of data stored in oldest job and ready to be flushed immediately.
+ *  if @return == 0, it means either :
+ *  + there is no active job (could be checked with ZSTD_frameProgression()), or
+ *  + oldest job is still actively compressing data,
+ *    but everything it has produced has also been flushed so far,
+ *    therefore flush speed is limited by production speed of oldest job
+ *    irrespective of the speed of concurrent (and newer) jobs.
+ */
+ZSTDLIB_API size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx);
+
+
+/*=====   Advanced Streaming decompression functions  =====*/
+
+/*!
+ * This function is deprecated, and is equivalent to:
+ *
+ *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+ *     ZSTD_DCtx_loadDictionary(zds, dict, dictSize);
+ *
+ * note: no dictionary will be used if dict == NULL or dictSize < 8
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize);
+
+/*!
+ * This function is deprecated, and is equivalent to:
+ *
+ *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+ *     ZSTD_DCtx_refDDict(zds, ddict);
+ *
+ * note : ddict is referenced, it must outlive decompression session
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* zds, const ZSTD_DDict* ddict);
+
+/*!
+ * This function is deprecated, and is equivalent to:
+ *
+ *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+ *
+ * re-use decompression parameters from previous init; saves dictionary loading
+ * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
+ */
+ZSTDLIB_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);
+
+
+/* *******************************************************************
+*  Buffer-less and synchronous inner streaming functions
+*
+*  This is an advanced API, giving full control over buffer management, for users which need direct control over memory.
+*  But it's also a complex one, with several restrictions, documented below.
+*  Prefer normal streaming API for an easier experience.
+********************************************************************* */
+
+/*
+  Buffer-less streaming compression (synchronous mode)
+
+  A ZSTD_CCtx object is required to track streaming operations.
+  Use ZSTD_createCCtx() / ZSTD_freeCCtx() to manage resource.
+  ZSTD_CCtx object can be re-used multiple times within successive compression operations.
+
+  Start by initializing a context.
+  Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary compression,
+  or ZSTD_compressBegin_advanced(), for finer parameter control.
+  It's also possible to duplicate a reference context which has already been initialized, using ZSTD_copyCCtx()
+
+  Then, consume your input using ZSTD_compressContinue().
+  There are some important considerations to keep in mind when using this advanced function :
+  - ZSTD_compressContinue() has no internal buffer. It uses externally provided buffers only.
+  - Interface is synchronous : input is consumed entirely and produces 1+ compressed blocks.
+  - Caller must ensure there is enough space in `dst` to store compressed data under worst case scenario.
+    Worst case evaluation is provided by ZSTD_compressBound().
+    ZSTD_compressContinue() doesn't guarantee recover after a failed compression.
+  - ZSTD_compressContinue() presumes prior input ***is still accessible and unmodified*** (up to maximum distance size, see WindowLog).
+    It remembers all previous contiguous blocks, plus one separated memory segment (which can itself consists of multiple contiguous blocks)
+  - ZSTD_compressContinue() detects that prior input has been overwritten when `src` buffer overlaps.
+    In which case, it will "discard" the relevant memory section from its history.
+
+  Finish a frame with ZSTD_compressEnd(), which will write the last block(s) and optional checksum.
+  It's possible to use srcSize==0, in which case, it will write a final empty block to end the frame.
+  Without last block mark, frames are considered unfinished (hence corrupted) by compliant decoders.
+
+  `ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress again.
+*/
+
+/*=====   Buffer-less streaming compression functions  =====*/
+ZSTDLIB_API size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
+ZSTDLIB_API size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel);
+ZSTDLIB_API size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize); /*< pledgedSrcSize : If srcSize is not known at init time, use ZSTD_CONTENTSIZE_UNKNOWN */
+ZSTDLIB_API size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict); /*< note: fails if cdict==NULL */
+ZSTDLIB_API size_t ZSTD_compressBegin_usingCDict_advanced(ZSTD_CCtx* const cctx, const ZSTD_CDict* const cdict, ZSTD_frameParameters const fParams, unsigned long long const pledgedSrcSize);   /* compression parameters are already set within cdict. pledgedSrcSize must be correct. If srcSize is not known, use macro ZSTD_CONTENTSIZE_UNKNOWN */
+ZSTDLIB_API size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* preparedCCtx, unsigned long long pledgedSrcSize); /*<  note: if pledgedSrcSize is not known, use ZSTD_CONTENTSIZE_UNKNOWN */
+
+ZSTDLIB_API size_t ZSTD_compressContinue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTDLIB_API size_t ZSTD_compressEnd(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+
+
+/*
+  Buffer-less streaming decompression (synchronous mode)
+
+  A ZSTD_DCtx object is required to track streaming operations.
+  Use ZSTD_createDCtx() / ZSTD_freeDCtx() to manage it.
+  A ZSTD_DCtx object can be re-used multiple times.
+
+  First typical operation is to retrieve frame parameters, using ZSTD_getFrameHeader().
+  Frame header is extracted from the beginning of compressed frame, so providing only the frame's beginning is enough.
+  Data fragment must be large enough to ensure successful decoding.
+ `ZSTD_frameHeaderSize_max` bytes is guaranteed to always be large enough.
+  @result : 0 : successful decoding, the `ZSTD_frameHeader` structure is correctly filled.
+           >0 : `srcSize` is too small, please provide at least @result bytes on next attempt.
+           errorCode, which can be tested using ZSTD_isError().
+
+  It fills a ZSTD_frameHeader structure with important information to correctly decode the frame,
+  such as the dictionary ID, content size, or maximum back-reference distance (`windowSize`).
+  Note that these values could be wrong, either because of data corruption, or because a 3rd party deliberately spoofs false information.
+  As a consequence, check that values remain within valid application range.
+  For example, do not allocate memory blindly, check that `windowSize` is within expectation.
+  Each application can set its own limits, depending on local restrictions.
+  For extended interoperability, it is recommended to support `windowSize` of at least 8 MB.
+
+  ZSTD_decompressContinue() needs previous data blocks during decompression, up to `windowSize` bytes.
+  ZSTD_decompressContinue() is very sensitive to contiguity,
+  if 2 blocks don't follow each other, make sure that either the compressor breaks contiguity at the same place,
+  or that previous contiguous segment is large enough to properly handle maximum back-reference distance.
+  There are multiple ways to guarantee this condition.
+
+  The most memory efficient way is to use a round buffer of sufficient size.
+  Sufficient size is determined by invoking ZSTD_decodingBufferSize_min(),
+  which can @return an error code if required value is too large for current system (in 32-bits mode).
+  In a round buffer methodology, ZSTD_decompressContinue() decompresses each block next to previous one,
+  up to the moment there is not enough room left in the buffer to guarantee decoding another full block,
+  which maximum size is provided in `ZSTD_frameHeader` structure, field `blockSizeMax`.
+  At which point, decoding can resume from the beginning of the buffer.
+  Note that already decoded data stored in the buffer should be flushed before being overwritten.
+
+  There are alternatives possible, for example using two or more buffers of size `windowSize` each, though they consume more memory.
+
+  Finally, if you control the compression process, you can also ignore all buffer size rules,
+  as long as the encoder and decoder progress in "lock-step",
+  aka use exactly the same buffer sizes, break contiguity at the same place, etc.
+
+  Once buffers are setup, start decompression, with ZSTD_decompressBegin().
+  If decompression requires a dictionary, use ZSTD_decompressBegin_usingDict() or ZSTD_decompressBegin_usingDDict().
+
+  Then use ZSTD_nextSrcSizeToDecompress() and ZSTD_decompressContinue() alternatively.
+  ZSTD_nextSrcSizeToDecompress() tells how many bytes to provide as 'srcSize' to ZSTD_decompressContinue().
+  ZSTD_decompressContinue() requires this _exact_ amount of bytes, or it will fail.
+
+ @result of ZSTD_decompressContinue() is the number of bytes regenerated within 'dst' (necessarily <= dstCapacity).
+  It can be zero : it just means ZSTD_decompressContinue() has decoded some metadata item.
+  It can also be an error code, which can be tested with ZSTD_isError().
+
+  A frame is fully decoded when ZSTD_nextSrcSizeToDecompress() returns zero.
+  Context can then be reset to start a new decompression.
+
+  Note : it's possible to know if next input to present is a header or a block, using ZSTD_nextInputType().
+  This information is not required to properly decode a frame.
+
+  == Special case : skippable frames ==
+
+  Skippable frames allow integration of user-defined data into a flow of concatenated frames.
+  Skippable frames will be ignored (skipped) by decompressor.
+  The format of skippable frames is as follows :
+  a) Skippable frame ID - 4 Bytes, Little endian format, any value from 0x184D2A50 to 0x184D2A5F
+  b) Frame Size - 4 Bytes, Little endian format, unsigned 32-bits
+  c) Frame Content - any content (User Data) of length equal to Frame Size
+  For skippable frames ZSTD_getFrameHeader() returns zfhPtr->frameType==ZSTD_skippableFrame.
+  For skippable frames ZSTD_decompressContinue() always returns 0 : it only skips the content.
+*/
+
+/*=====   Buffer-less streaming decompression functions  =====*/
+typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
+typedef struct {
+    unsigned long long frameContentSize; /* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */
+    unsigned long long windowSize;       /* can be very large, up to <= frameContentSize */
+    unsigned blockSizeMax;
+    ZSTD_frameType_e frameType;          /* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */
+    unsigned headerSize;
+    unsigned dictID;
+    unsigned checksumFlag;
+} ZSTD_frameHeader;
+
+/*! ZSTD_getFrameHeader() :
+ *  decode Frame Header, or requires larger `srcSize`.
+ * @return : 0, `zfhPtr` is correctly filled,
+ *          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+ *           or an error code, which can be tested using ZSTD_isError() */
+ZSTDLIB_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize);   /*< doesn't consume input */
+/*! ZSTD_getFrameHeader_advanced() :
+ *  same as ZSTD_getFrameHeader(),
+ *  with added capability to select a format (like ZSTD_f_zstd1_magicless) */
+ZSTDLIB_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
+ZSTDLIB_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize);  /*< when frame content size is not known, pass in frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN */
+
+ZSTDLIB_API size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx);
+ZSTDLIB_API size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
+ZSTDLIB_API size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);
+
+ZSTDLIB_API size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx);
+ZSTDLIB_API size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+
+/* misc */
+ZSTDLIB_API void   ZSTD_copyDCtx(ZSTD_DCtx* dctx, const ZSTD_DCtx* preparedDCtx);
+typedef enum { ZSTDnit_frameHeader, ZSTDnit_blockHeader, ZSTDnit_block, ZSTDnit_lastBlock, ZSTDnit_checksum, ZSTDnit_skippableFrame } ZSTD_nextInputType_e;
+ZSTDLIB_API ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx);
+
+
+
+
+/* ============================ */
+/*       Block level API       */
+/* ============================ */
+
+/*!
+    Block functions produce and decode raw zstd blocks, without frame metadata.
+    Frame metadata cost is typically ~12 bytes, which can be non-negligible for very small blocks (< 100 bytes).
+    But users will have to take in charge needed metadata to regenerate data, such as compressed and content sizes.
+
+    A few rules to respect :
+    - Compressing and decompressing require a context structure
+      + Use ZSTD_createCCtx() and ZSTD_createDCtx()
+    - It is necessary to init context before starting
+      + compression : any ZSTD_compressBegin*() variant, including with dictionary
+      + decompression : any ZSTD_decompressBegin*() variant, including with dictionary
+      + copyCCtx() and copyDCtx() can be used too
+    - Block size is limited, it must be <= ZSTD_getBlockSize() <= ZSTD_BLOCKSIZE_MAX == 128 KB
+      + If input is larger than a block size, it's necessary to split input data into multiple blocks
+      + For inputs larger than a single block, consider using regular ZSTD_compress() instead.
+        Frame metadata is not that costly, and quickly becomes negligible as source size grows larger than a block.
+    - When a block is considered not compressible enough, ZSTD_compressBlock() result will be 0 (zero) !
+      ===> In which case, nothing is produced into `dst` !
+      + User __must__ test for such outcome and deal directly with uncompressed data
+      + A block cannot be declared incompressible if ZSTD_compressBlock() return value was != 0.
+        Doing so would mess up with statistics history, leading to potential data corruption.
+      + ZSTD_decompressBlock() _doesn't accept uncompressed data as input_ !!
+      + In case of multiple successive blocks, should some of them be uncompressed,
+        decoder must be informed of their existence in order to follow proper history.
+        Use ZSTD_insertBlock() for such a case.
+*/
+
+/*=====   Raw zstd block functions  =====*/
+ZSTDLIB_API size_t ZSTD_getBlockSize   (const ZSTD_CCtx* cctx);
+ZSTDLIB_API size_t ZSTD_compressBlock  (ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTDLIB_API size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTDLIB_API size_t ZSTD_insertBlock    (ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize);  /*< insert uncompressed block into `dctx` history. Useful for multi-blocks decompression. */
+
+
+#endif   /* ZSTD_H_ZSTD_STATIC_LINKING_ONLY */
+
diff --git a/include/uapi/linux/virtio_mem.h b/include/uapi/linux/virtio_mem.h
index 70e01c6..e9122f1 100644
--- a/include/uapi/linux/virtio_mem.h
+++ b/include/uapi/linux/virtio_mem.h
@@ -68,9 +68,10 @@
  * explicitly triggered (VIRTIO_MEM_REQ_UNPLUG).
  *
  * There are no guarantees what will happen if unplugged memory is
- * read/written. Such memory should, in general, not be touched. E.g.,
- * even writing might succeed, but the values will simply be discarded at
- * random points in time.
+ * read/written. In general, unplugged memory should not be touched, because
+ * the resulting action is undefined. There is one exception: without
+ * VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE, unplugged memory inside the usable
+ * region can be read, to simplify creation of memory dumps.
  *
  * It can happen that the device cannot process a request, because it is
  * busy. The device driver has to retry later.
@@ -87,6 +88,8 @@
 
 /* node_id is an ACPI PXM and is valid */
 #define VIRTIO_MEM_F_ACPI_PXM		0
+/* unplugged memory must not be accessed */
+#define VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE	1
 
 
 /* --- virtio-mem: guest -> host requests --- */
diff --git a/init/Makefile b/init/Makefile
index 2846113..04eeee12 100644
--- a/init/Makefile
+++ b/init/Makefile
@@ -30,7 +30,7 @@
 quiet_cmd_compile.h = CHK     $@
       cmd_compile.h = \
 	$(CONFIG_SHELL) $(srctree)/scripts/mkcompile_h $@	\
-	"$(UTS_MACHINE)" "$(CONFIG_SMP)" "$(CONFIG_PREEMPT)"	\
+	"$(UTS_MACHINE)" "$(CONFIG_SMP)" "$(CONFIG_PREEMPT_BUILD)"	\
 	"$(CONFIG_PREEMPT_RT)" $(CONFIG_CC_VERSION_TEXT) "$(LD)"
 
 include/generated/compile.h: FORCE
diff --git a/kernel/Kconfig.preempt b/kernel/Kconfig.preempt
index 60f1bfc..ce77f02 100644
--- a/kernel/Kconfig.preempt
+++ b/kernel/Kconfig.preempt
@@ -1,12 +1,23 @@
 # SPDX-License-Identifier: GPL-2.0-only
 
+config PREEMPT_NONE_BUILD
+	bool
+
+config PREEMPT_VOLUNTARY_BUILD
+	bool
+
+config PREEMPT_BUILD
+	bool
+	select PREEMPTION
+	select UNINLINE_SPIN_UNLOCK if !ARCH_INLINE_SPIN_UNLOCK
+
 choice
 	prompt "Preemption Model"
-	default PREEMPT_NONE_BEHAVIOUR
+	default PREEMPT_NONE
 
-config PREEMPT_NONE_BEHAVIOUR
+config PREEMPT_NONE
 	bool "No Forced Preemption (Server)"
-	select PREEMPT_NONE if !PREEMPT_DYNAMIC
+	select PREEMPT_NONE_BUILD if !PREEMPT_DYNAMIC
 	help
 	  This is the traditional Linux preemption model, geared towards
 	  throughput. It will still provide good latencies most of the
@@ -18,10 +29,10 @@
 	  raw processing power of the kernel, irrespective of scheduling
 	  latencies.
 
-config PREEMPT_VOLUNTARY_BEHAVIOUR
+config PREEMPT_VOLUNTARY
 	bool "Voluntary Kernel Preemption (Desktop)"
 	depends on !ARCH_NO_PREEMPT
-	select PREEMPT_VOLUNTARY if !PREEMPT_DYNAMIC
+	select PREEMPT_VOLUNTARY_BUILD if !PREEMPT_DYNAMIC
 	help
 	  This option reduces the latency of the kernel by adding more
 	  "explicit preemption points" to the kernel code. These new
@@ -37,10 +48,10 @@
 
 	  Select this if you are building a kernel for a desktop system.
 
-config PREEMPT_BEHAVIOUR
+config PREEMPT
 	bool "Preemptible Kernel (Low-Latency Desktop)"
 	depends on !ARCH_NO_PREEMPT
-	select PREEMPT
+	select PREEMPT_BUILD
 	help
 	  This option reduces the latency of the kernel by making
 	  all kernel code (that is not executing in a critical section)
@@ -58,7 +69,7 @@
 
 config PREEMPT_RT
 	bool "Fully Preemptible Kernel (Real-Time)"
-	depends on EXPERT && ARCH_SUPPORTS_RT && !PREEMPT_DYNAMIC
+	depends on EXPERT && ARCH_SUPPORTS_RT
 	select PREEMPTION
 	help
 	  This option turns the kernel into a real-time kernel by replacing
@@ -75,17 +86,6 @@
 
 endchoice
 
-config PREEMPT_NONE
-	bool
-
-config PREEMPT_VOLUNTARY
-	bool
-
-config PREEMPT
-	bool
-	select PREEMPTION
-	select UNINLINE_SPIN_UNLOCK if !ARCH_INLINE_SPIN_UNLOCK
-
 config PREEMPT_COUNT
        bool
 
@@ -95,8 +95,8 @@
 
 config PREEMPT_DYNAMIC
 	bool "Preemption behaviour defined on boot"
-	depends on HAVE_PREEMPT_DYNAMIC
-	select PREEMPT
+	depends on HAVE_PREEMPT_DYNAMIC && !PREEMPT_RT
+	select PREEMPT_BUILD
 	default y
 	help
 	  This option allows to define the preemption model on the kernel
diff --git a/kernel/events/core.c b/kernel/events/core.c
index f2253ea..523106a 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -7154,7 +7154,6 @@ void perf_output_sample(struct perf_output_handle *handle,
 static u64 perf_virt_to_phys(u64 virt)
 {
 	u64 phys_addr = 0;
-	struct page *p = NULL;
 
 	if (!virt)
 		return 0;
@@ -7173,14 +7172,15 @@ static u64 perf_virt_to_phys(u64 virt)
 		 * If failed, leave phys_addr as 0.
 		 */
 		if (current->mm != NULL) {
+			struct page *p;
+
 			pagefault_disable();
-			if (get_user_page_fast_only(virt, 0, &p))
+			if (get_user_page_fast_only(virt, 0, &p)) {
 				phys_addr = page_to_phys(p) + virt % PAGE_SIZE;
+				put_page(p);
+			}
 			pagefault_enable();
 		}
-
-		if (p)
-			put_page(p);
 	}
 
 	return phys_addr;
diff --git a/kernel/fork.c b/kernel/fork.c
index 5de23f3..3244cc5 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -2277,6 +2277,7 @@ static __latent_entropy struct task_struct *copy_process(
 	p->pdeath_signal = 0;
 	INIT_LIST_HEAD(&p->thread_group);
 	p->task_works = NULL;
+	clear_posix_cputimers_work(p);
 
 #ifdef CONFIG_KRETPROBES
 	p->kretprobe_instances.first = NULL;
diff --git a/kernel/irq/msi.c b/kernel/irq/msi.c
index 6a5ecee..7f350ae 100644
--- a/kernel/irq/msi.c
+++ b/kernel/irq/msi.c
@@ -529,10 +529,10 @@ static bool msi_check_reservation_mode(struct irq_domain *domain,
 
 	/*
 	 * Checking the first MSI descriptor is sufficient. MSIX supports
-	 * masking and MSI does so when the maskbit is set.
+	 * masking and MSI does so when the can_mask attribute is set.
 	 */
 	desc = first_msi_entry(dev);
-	return desc->msi_attrib.is_msix || desc->msi_attrib.maskbit;
+	return desc->msi_attrib.is_msix || desc->msi_attrib.can_mask;
 }
 
 int __msi_domain_alloc_irqs(struct irq_domain *domain, struct device *dev,
diff --git a/kernel/sched/autogroup.c b/kernel/sched/autogroup.c
index 2067080..8629b37 100644
--- a/kernel/sched/autogroup.c
+++ b/kernel/sched/autogroup.c
@@ -31,7 +31,7 @@ static inline void autogroup_destroy(struct kref *kref)
 	ag->tg->rt_se = NULL;
 	ag->tg->rt_rq = NULL;
 #endif
-	sched_offline_group(ag->tg);
+	sched_release_group(ag->tg);
 	sched_destroy_group(ag->tg);
 }
 
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 523fd60..3c9b0fd 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -3726,6 +3726,9 @@ void wake_up_if_idle(int cpu)
 
 bool cpus_share_cache(int this_cpu, int that_cpu)
 {
+	if (this_cpu == that_cpu)
+		return true;
+
 	return per_cpu(sd_llc_id, this_cpu) == per_cpu(sd_llc_id, that_cpu);
 }
 
@@ -6625,13 +6628,13 @@ __setup("preempt=", setup_preempt_mode);
 static void __init preempt_dynamic_init(void)
 {
 	if (preempt_dynamic_mode == preempt_dynamic_undefined) {
-		if (IS_ENABLED(CONFIG_PREEMPT_NONE_BEHAVIOUR)) {
+		if (IS_ENABLED(CONFIG_PREEMPT_NONE)) {
 			sched_dynamic_update(preempt_dynamic_none);
-		} else if (IS_ENABLED(CONFIG_PREEMPT_VOLUNTARY_BEHAVIOUR)) {
+		} else if (IS_ENABLED(CONFIG_PREEMPT_VOLUNTARY)) {
 			sched_dynamic_update(preempt_dynamic_voluntary);
 		} else {
 			/* Default static call setting, nothing to do */
-			WARN_ON_ONCE(!IS_ENABLED(CONFIG_PREEMPT_BEHAVIOUR));
+			WARN_ON_ONCE(!IS_ENABLED(CONFIG_PREEMPT));
 			preempt_dynamic_mode = preempt_dynamic_full;
 			pr_info("Dynamic Preempt: full\n");
 		}
@@ -9716,6 +9719,22 @@ static void sched_free_group(struct task_group *tg)
 	kmem_cache_free(task_group_cache, tg);
 }
 
+static void sched_free_group_rcu(struct rcu_head *rcu)
+{
+	sched_free_group(container_of(rcu, struct task_group, rcu));
+}
+
+static void sched_unregister_group(struct task_group *tg)
+{
+	unregister_fair_sched_group(tg);
+	unregister_rt_sched_group(tg);
+	/*
+	 * We have to wait for yet another RCU grace period to expire, as
+	 * print_cfs_stats() might run concurrently.
+	 */
+	call_rcu(&tg->rcu, sched_free_group_rcu);
+}
+
 /* allocate runqueue etc for a new task group */
 struct task_group *sched_create_group(struct task_group *parent)
 {
@@ -9759,25 +9778,35 @@ void sched_online_group(struct task_group *tg, struct task_group *parent)
 }
 
 /* rcu callback to free various structures associated with a task group */
-static void sched_free_group_rcu(struct rcu_head *rhp)
+static void sched_unregister_group_rcu(struct rcu_head *rhp)
 {
 	/* Now it should be safe to free those cfs_rqs: */
-	sched_free_group(container_of(rhp, struct task_group, rcu));
+	sched_unregister_group(container_of(rhp, struct task_group, rcu));
 }
 
 void sched_destroy_group(struct task_group *tg)
 {
 	/* Wait for possible concurrent references to cfs_rqs complete: */
-	call_rcu(&tg->rcu, sched_free_group_rcu);
+	call_rcu(&tg->rcu, sched_unregister_group_rcu);
 }
 
-void sched_offline_group(struct task_group *tg)
+void sched_release_group(struct task_group *tg)
 {
 	unsigned long flags;
 
-	/* End participation in shares distribution: */
-	unregister_fair_sched_group(tg);
-
+	/*
+	 * Unlink first, to avoid walk_tg_tree_from() from finding us (via
+	 * sched_cfs_period_timer()).
+	 *
+	 * For this to be effective, we have to wait for all pending users of
+	 * this task group to leave their RCU critical section to ensure no new
+	 * user will see our dying task group any more. Specifically ensure
+	 * that tg_unthrottle_up() won't add decayed cfs_rq's to it.
+	 *
+	 * We therefore defer calling unregister_fair_sched_group() to
+	 * sched_unregister_group() which is guarantied to get called only after the
+	 * current RCU grace period has expired.
+	 */
 	spin_lock_irqsave(&task_group_lock, flags);
 	list_del_rcu(&tg->list);
 	list_del_rcu(&tg->siblings);
@@ -9896,7 +9925,7 @@ static void cpu_cgroup_css_released(struct cgroup_subsys_state *css)
 {
 	struct task_group *tg = css_tg(css);
 
-	sched_offline_group(tg);
+	sched_release_group(tg);
 }
 
 static void cpu_cgroup_css_free(struct cgroup_subsys_state *css)
@@ -9906,7 +9935,7 @@ static void cpu_cgroup_css_free(struct cgroup_subsys_state *css)
 	/*
 	 * Relies on the RCU grace period between css_released() and this.
 	 */
-	sched_free_group(tg);
+	sched_unregister_group(tg);
 }
 
 /*
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 13950be..6e476f6 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -11456,8 +11456,6 @@ void free_fair_sched_group(struct task_group *tg)
 {
 	int i;
 
-	destroy_cfs_bandwidth(tg_cfs_bandwidth(tg));
-
 	for_each_possible_cpu(i) {
 		if (tg->cfs_rq)
 			kfree(tg->cfs_rq[i]);
@@ -11534,6 +11532,8 @@ void unregister_fair_sched_group(struct task_group *tg)
 	struct rq *rq;
 	int cpu;
 
+	destroy_cfs_bandwidth(tg_cfs_bandwidth(tg));
+
 	for_each_possible_cpu(cpu) {
 		if (tg->se[cpu])
 			remove_entity_load_avg(tg->se[cpu]);
diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index bb945f8..b48baab 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -137,13 +137,17 @@ static inline struct rq *rq_of_rt_se(struct sched_rt_entity *rt_se)
 	return rt_rq->rq;
 }
 
+void unregister_rt_sched_group(struct task_group *tg)
+{
+	if (tg->rt_se)
+		destroy_rt_bandwidth(&tg->rt_bandwidth);
+
+}
+
 void free_rt_sched_group(struct task_group *tg)
 {
 	int i;
 
-	if (tg->rt_se)
-		destroy_rt_bandwidth(&tg->rt_bandwidth);
-
 	for_each_possible_cpu(i) {
 		if (tg->rt_rq)
 			kfree(tg->rt_rq[i]);
@@ -250,6 +254,8 @@ static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se)
 	return &rq->rt;
 }
 
+void unregister_rt_sched_group(struct task_group *tg) { }
+
 void free_rt_sched_group(struct task_group *tg) { }
 
 int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent)
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 7f1612d..0e66749 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -488,6 +488,7 @@ extern void __refill_cfs_bandwidth_runtime(struct cfs_bandwidth *cfs_b);
 extern void start_cfs_bandwidth(struct cfs_bandwidth *cfs_b);
 extern void unthrottle_cfs_rq(struct cfs_rq *cfs_rq);
 
+extern void unregister_rt_sched_group(struct task_group *tg);
 extern void free_rt_sched_group(struct task_group *tg);
 extern int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent);
 extern void init_tg_rt_entry(struct task_group *tg, struct rt_rq *rt_rq,
@@ -503,7 +504,7 @@ extern struct task_group *sched_create_group(struct task_group *parent);
 extern void sched_online_group(struct task_group *tg,
 			       struct task_group *parent);
 extern void sched_destroy_group(struct task_group *tg);
-extern void sched_offline_group(struct task_group *tg);
+extern void sched_release_group(struct task_group *tg);
 
 extern void sched_move_task(struct task_struct *tsk);
 
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index 643d412..96b4e78 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -1159,13 +1159,28 @@ static void posix_cpu_timers_work(struct callback_head *work)
 }
 
 /*
+ * Clear existing posix CPU timers task work.
+ */
+void clear_posix_cputimers_work(struct task_struct *p)
+{
+	/*
+	 * A copied work entry from the old task is not meaningful, clear it.
+	 * N.B. init_task_work will not do this.
+	 */
+	memset(&p->posix_cputimers_work.work, 0,
+	       sizeof(p->posix_cputimers_work.work));
+	init_task_work(&p->posix_cputimers_work.work,
+		       posix_cpu_timers_work);
+	p->posix_cputimers_work.scheduled = false;
+}
+
+/*
  * Initialize posix CPU timers task work in init task. Out of line to
  * keep the callback static and to avoid header recursion hell.
  */
 void __init posix_cputimers_init_work(void)
 {
-	init_task_work(&current->posix_cputimers_work.work,
-		       posix_cpu_timers_work);
+	clear_posix_cputimers_work(current);
 }
 
 /*
diff --git a/lib/decompress_unzstd.c b/lib/decompress_unzstd.c
index 6b629ab..a512b99 100644
--- a/lib/decompress_unzstd.c
+++ b/lib/decompress_unzstd.c
@@ -68,11 +68,7 @@
 #ifdef STATIC
 # define UNZSTD_PREBOOT
 # include "xxhash.c"
-# include "zstd/entropy_common.c"
-# include "zstd/fse_decompress.c"
-# include "zstd/huf_decompress.c"
-# include "zstd/zstd_common.c"
-# include "zstd/decompress.c"
+# include "zstd/decompress_sources.h"
 #endif
 
 #include <linux/decompress/mm.h>
@@ -91,11 +87,15 @@
 
 static int INIT handle_zstd_error(size_t ret, void (*error)(char *x))
 {
-	const int err = ZSTD_getErrorCode(ret);
+	const zstd_error_code err = zstd_get_error_code(ret);
 
-	if (!ZSTD_isError(ret))
+	if (!zstd_is_error(ret))
 		return 0;
 
+	/*
+	 * zstd_get_error_name() cannot be used because error takes a char *
+	 * not a const char *
+	 */
 	switch (err) {
 	case ZSTD_error_memory_allocation:
 		error("ZSTD decompressor ran out of memory");
@@ -124,28 +124,28 @@ static int INIT decompress_single(const u8 *in_buf, long in_len, u8 *out_buf,
 				  long out_len, long *in_pos,
 				  void (*error)(char *x))
 {
-	const size_t wksp_size = ZSTD_DCtxWorkspaceBound();
+	const size_t wksp_size = zstd_dctx_workspace_bound();
 	void *wksp = large_malloc(wksp_size);
-	ZSTD_DCtx *dctx = ZSTD_initDCtx(wksp, wksp_size);
+	zstd_dctx *dctx = zstd_init_dctx(wksp, wksp_size);
 	int err;
 	size_t ret;
 
 	if (dctx == NULL) {
-		error("Out of memory while allocating ZSTD_DCtx");
+		error("Out of memory while allocating zstd_dctx");
 		err = -1;
 		goto out;
 	}
 	/*
 	 * Find out how large the frame actually is, there may be junk at
-	 * the end of the frame that ZSTD_decompressDCtx() can't handle.
+	 * the end of the frame that zstd_decompress_dctx() can't handle.
 	 */
-	ret = ZSTD_findFrameCompressedSize(in_buf, in_len);
+	ret = zstd_find_frame_compressed_size(in_buf, in_len);
 	err = handle_zstd_error(ret, error);
 	if (err)
 		goto out;
 	in_len = (long)ret;
 
-	ret = ZSTD_decompressDCtx(dctx, out_buf, out_len, in_buf, in_len);
+	ret = zstd_decompress_dctx(dctx, out_buf, out_len, in_buf, in_len);
 	err = handle_zstd_error(ret, error);
 	if (err)
 		goto out;
@@ -167,14 +167,14 @@ static int INIT __unzstd(unsigned char *in_buf, long in_len,
 			 long *in_pos,
 			 void (*error)(char *x))
 {
-	ZSTD_inBuffer in;
-	ZSTD_outBuffer out;
-	ZSTD_frameParams params;
+	zstd_in_buffer in;
+	zstd_out_buffer out;
+	zstd_frame_header header;
 	void *in_allocated = NULL;
 	void *out_allocated = NULL;
 	void *wksp = NULL;
 	size_t wksp_size;
-	ZSTD_DStream *dstream;
+	zstd_dstream *dstream;
 	int err;
 	size_t ret;
 
@@ -238,13 +238,13 @@ static int INIT __unzstd(unsigned char *in_buf, long in_len,
 	out.size = out_len;
 
 	/*
-	 * We need to know the window size to allocate the ZSTD_DStream.
+	 * We need to know the window size to allocate the zstd_dstream.
 	 * Since we are streaming, we need to allocate a buffer for the sliding
 	 * window. The window size varies from 1 KB to ZSTD_WINDOWSIZE_MAX
 	 * (8 MB), so it is important to use the actual value so as not to
 	 * waste memory when it is smaller.
 	 */
-	ret = ZSTD_getFrameParams(&params, in.src, in.size);
+	ret = zstd_get_frame_header(&header, in.src, in.size);
 	err = handle_zstd_error(ret, error);
 	if (err)
 		goto out;
@@ -253,19 +253,19 @@ static int INIT __unzstd(unsigned char *in_buf, long in_len,
 		err = -1;
 		goto out;
 	}
-	if (params.windowSize > ZSTD_WINDOWSIZE_MAX) {
+	if (header.windowSize > ZSTD_WINDOWSIZE_MAX) {
 		error("ZSTD-compressed data has too large a window size");
 		err = -1;
 		goto out;
 	}
 
 	/*
-	 * Allocate the ZSTD_DStream now that we know how much memory is
+	 * Allocate the zstd_dstream now that we know how much memory is
 	 * required.
 	 */
-	wksp_size = ZSTD_DStreamWorkspaceBound(params.windowSize);
+	wksp_size = zstd_dstream_workspace_bound(header.windowSize);
 	wksp = large_malloc(wksp_size);
-	dstream = ZSTD_initDStream(params.windowSize, wksp, wksp_size);
+	dstream = zstd_init_dstream(header.windowSize, wksp, wksp_size);
 	if (dstream == NULL) {
 		error("Out of memory while allocating ZSTD_DStream");
 		err = -1;
@@ -298,7 +298,7 @@ static int INIT __unzstd(unsigned char *in_buf, long in_len,
 			in.size = in_len;
 		}
 		/* Returns zero when the frame is complete. */
-		ret = ZSTD_decompressStream(dstream, &out, &in);
+		ret = zstd_decompress_stream(dstream, &out, &in);
 		err = handle_zstd_error(ret, error);
 		if (err)
 			goto out;
diff --git a/lib/zstd/Makefile b/lib/zstd/Makefile
index f5d778e..65218ec 100644
--- a/lib/zstd/Makefile
+++ b/lib/zstd/Makefile
@@ -1,10 +1,46 @@
-# SPDX-License-Identifier: GPL-2.0-only
+# SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause
+# ################################################################
+# Copyright (c) Facebook, Inc.
+# All rights reserved.
+#
+# This source code is licensed under both the BSD-style license (found in the
+# LICENSE file in the root directory of this source tree) and the GPLv2 (found
+# in the COPYING file in the root directory of this source tree).
+# You may select, at your option, one of the above-listed licenses.
+# ################################################################
 obj-$(CONFIG_ZSTD_COMPRESS) += zstd_compress.o
 obj-$(CONFIG_ZSTD_DECOMPRESS) += zstd_decompress.o
 
 ccflags-y += -O3
 
-zstd_compress-y := fse_compress.o huf_compress.o compress.o \
-		   entropy_common.o fse_decompress.o zstd_common.o
-zstd_decompress-y := huf_decompress.o decompress.o \
-		     entropy_common.o fse_decompress.o zstd_common.o
+zstd_compress-y := \
+		zstd_compress_module.o \
+		common/debug.o \
+		common/entropy_common.o \
+		common/error_private.o \
+		common/fse_decompress.o \
+		common/zstd_common.o \
+		compress/fse_compress.o \
+		compress/hist.o \
+		compress/huf_compress.o \
+		compress/zstd_compress.o \
+		compress/zstd_compress_literals.o \
+		compress/zstd_compress_sequences.o \
+		compress/zstd_compress_superblock.o \
+		compress/zstd_double_fast.o \
+		compress/zstd_fast.o \
+		compress/zstd_lazy.o \
+		compress/zstd_ldm.o \
+		compress/zstd_opt.o \
+
+zstd_decompress-y := \
+		zstd_decompress_module.o \
+		common/debug.o \
+		common/entropy_common.o \
+		common/error_private.o \
+		common/fse_decompress.o \
+		common/zstd_common.o \
+		decompress/huf_decompress.o \
+		decompress/zstd_ddict.o \
+		decompress/zstd_decompress.o \
+		decompress/zstd_decompress_block.o \
diff --git a/lib/zstd/bitstream.h b/lib/zstd/bitstream.h
deleted file mode 100644
index 5d6343c..0000000
--- a/lib/zstd/bitstream.h
+++ /dev/null
@@ -1,380 +0,0 @@
-/*
- * bitstream
- * Part of FSE library
- * header file (to include)
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-#ifndef BITSTREAM_H_MODULE
-#define BITSTREAM_H_MODULE
-
-/*
-*  This API consists of small unitary functions, which must be inlined for best performance.
-*  Since link-time-optimization is not available for all compilers,
-*  these functions are defined into a .h to be included.
-*/
-
-/*-****************************************
-*  Dependencies
-******************************************/
-#include "error_private.h" /* error codes and messages */
-#include "mem.h"	   /* unaligned access routines */
-
-/*=========================================
-*  Target specific
-=========================================*/
-#define STREAM_ACCUMULATOR_MIN_32 25
-#define STREAM_ACCUMULATOR_MIN_64 57
-#define STREAM_ACCUMULATOR_MIN ((U32)(ZSTD_32bits() ? STREAM_ACCUMULATOR_MIN_32 : STREAM_ACCUMULATOR_MIN_64))
-
-/*-******************************************
-*  bitStream encoding API (write forward)
-********************************************/
-/* bitStream can mix input from multiple sources.
-*  A critical property of these streams is that they encode and decode in **reverse** direction.
-*  So the first bit sequence you add will be the last to be read, like a LIFO stack.
-*/
-typedef struct {
-	size_t bitContainer;
-	int bitPos;
-	char *startPtr;
-	char *ptr;
-	char *endPtr;
-} BIT_CStream_t;
-
-ZSTD_STATIC size_t BIT_initCStream(BIT_CStream_t *bitC, void *dstBuffer, size_t dstCapacity);
-ZSTD_STATIC void BIT_addBits(BIT_CStream_t *bitC, size_t value, unsigned nbBits);
-ZSTD_STATIC void BIT_flushBits(BIT_CStream_t *bitC);
-ZSTD_STATIC size_t BIT_closeCStream(BIT_CStream_t *bitC);
-
-/* Start with initCStream, providing the size of buffer to write into.
-*  bitStream will never write outside of this buffer.
-*  `dstCapacity` must be >= sizeof(bitD->bitContainer), otherwise @return will be an error code.
-*
-*  bits are first added to a local register.
-*  Local register is size_t, hence 64-bits on 64-bits systems, or 32-bits on 32-bits systems.
-*  Writing data into memory is an explicit operation, performed by the flushBits function.
-*  Hence keep track how many bits are potentially stored into local register to avoid register overflow.
-*  After a flushBits, a maximum of 7 bits might still be stored into local register.
-*
-*  Avoid storing elements of more than 24 bits if you want compatibility with 32-bits bitstream readers.
-*
-*  Last operation is to close the bitStream.
-*  The function returns the final size of CStream in bytes.
-*  If data couldn't fit into `dstBuffer`, it will return a 0 ( == not storable)
-*/
-
-/*-********************************************
-*  bitStream decoding API (read backward)
-**********************************************/
-typedef struct {
-	size_t bitContainer;
-	unsigned bitsConsumed;
-	const char *ptr;
-	const char *start;
-} BIT_DStream_t;
-
-typedef enum {
-	BIT_DStream_unfinished = 0,
-	BIT_DStream_endOfBuffer = 1,
-	BIT_DStream_completed = 2,
-	BIT_DStream_overflow = 3
-} BIT_DStream_status; /* result of BIT_reloadDStream() */
-/* 1,2,4,8 would be better for bitmap combinations, but slows down performance a bit ... :( */
-
-ZSTD_STATIC size_t BIT_initDStream(BIT_DStream_t *bitD, const void *srcBuffer, size_t srcSize);
-ZSTD_STATIC size_t BIT_readBits(BIT_DStream_t *bitD, unsigned nbBits);
-ZSTD_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t *bitD);
-ZSTD_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t *bitD);
-
-/* Start by invoking BIT_initDStream().
-*  A chunk of the bitStream is then stored into a local register.
-*  Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
-*  You can then retrieve bitFields stored into the local register, **in reverse order**.
-*  Local register is explicitly reloaded from memory by the BIT_reloadDStream() method.
-*  A reload guarantee a minimum of ((8*sizeof(bitD->bitContainer))-7) bits when its result is BIT_DStream_unfinished.
-*  Otherwise, it can be less than that, so proceed accordingly.
-*  Checking if DStream has reached its end can be performed with BIT_endOfDStream().
-*/
-
-/*-****************************************
-*  unsafe API
-******************************************/
-ZSTD_STATIC void BIT_addBitsFast(BIT_CStream_t *bitC, size_t value, unsigned nbBits);
-/* faster, but works only if value is "clean", meaning all high bits above nbBits are 0 */
-
-ZSTD_STATIC void BIT_flushBitsFast(BIT_CStream_t *bitC);
-/* unsafe version; does not check buffer overflow */
-
-ZSTD_STATIC size_t BIT_readBitsFast(BIT_DStream_t *bitD, unsigned nbBits);
-/* faster, but works only if nbBits >= 1 */
-
-/*-**************************************************************
-*  Internal functions
-****************************************************************/
-ZSTD_STATIC unsigned BIT_highbit32(register U32 val) { return 31 - __builtin_clz(val); }
-
-/*=====    Local Constants   =====*/
-static const unsigned BIT_mask[] = {0,       1,       3,       7,	0xF,      0x1F,     0x3F,     0x7F,      0xFF,
-				    0x1FF,   0x3FF,   0x7FF,   0xFFF,    0x1FFF,   0x3FFF,   0x7FFF,   0xFFFF,    0x1FFFF,
-				    0x3FFFF, 0x7FFFF, 0xFFFFF, 0x1FFFFF, 0x3FFFFF, 0x7FFFFF, 0xFFFFFF, 0x1FFFFFF, 0x3FFFFFF}; /* up to 26 bits */
-
-/*-**************************************************************
-*  bitStream encoding
-****************************************************************/
-/*! BIT_initCStream() :
- *  `dstCapacity` must be > sizeof(void*)
- *  @return : 0 if success,
-			  otherwise an error code (can be tested using ERR_isError() ) */
-ZSTD_STATIC size_t BIT_initCStream(BIT_CStream_t *bitC, void *startPtr, size_t dstCapacity)
-{
-	bitC->bitContainer = 0;
-	bitC->bitPos = 0;
-	bitC->startPtr = (char *)startPtr;
-	bitC->ptr = bitC->startPtr;
-	bitC->endPtr = bitC->startPtr + dstCapacity - sizeof(bitC->ptr);
-	if (dstCapacity <= sizeof(bitC->ptr))
-		return ERROR(dstSize_tooSmall);
-	return 0;
-}
-
-/*! BIT_addBits() :
-	can add up to 26 bits into `bitC`.
-	Does not check for register overflow ! */
-ZSTD_STATIC void BIT_addBits(BIT_CStream_t *bitC, size_t value, unsigned nbBits)
-{
-	bitC->bitContainer |= (value & BIT_mask[nbBits]) << bitC->bitPos;
-	bitC->bitPos += nbBits;
-}
-
-/*! BIT_addBitsFast() :
- *  works only if `value` is _clean_, meaning all high bits above nbBits are 0 */
-ZSTD_STATIC void BIT_addBitsFast(BIT_CStream_t *bitC, size_t value, unsigned nbBits)
-{
-	bitC->bitContainer |= value << bitC->bitPos;
-	bitC->bitPos += nbBits;
-}
-
-/*! BIT_flushBitsFast() :
- *  unsafe version; does not check buffer overflow */
-ZSTD_STATIC void BIT_flushBitsFast(BIT_CStream_t *bitC)
-{
-	size_t const nbBytes = bitC->bitPos >> 3;
-	ZSTD_writeLEST(bitC->ptr, bitC->bitContainer);
-	bitC->ptr += nbBytes;
-	bitC->bitPos &= 7;
-	bitC->bitContainer >>= nbBytes * 8; /* if bitPos >= sizeof(bitContainer)*8 --> undefined behavior */
-}
-
-/*! BIT_flushBits() :
- *  safe version; check for buffer overflow, and prevents it.
- *  note : does not signal buffer overflow. This will be revealed later on using BIT_closeCStream() */
-ZSTD_STATIC void BIT_flushBits(BIT_CStream_t *bitC)
-{
-	size_t const nbBytes = bitC->bitPos >> 3;
-	ZSTD_writeLEST(bitC->ptr, bitC->bitContainer);
-	bitC->ptr += nbBytes;
-	if (bitC->ptr > bitC->endPtr)
-		bitC->ptr = bitC->endPtr;
-	bitC->bitPos &= 7;
-	bitC->bitContainer >>= nbBytes * 8; /* if bitPos >= sizeof(bitContainer)*8 --> undefined behavior */
-}
-
-/*! BIT_closeCStream() :
- *  @return : size of CStream, in bytes,
-			  or 0 if it could not fit into dstBuffer */
-ZSTD_STATIC size_t BIT_closeCStream(BIT_CStream_t *bitC)
-{
-	BIT_addBitsFast(bitC, 1, 1); /* endMark */
-	BIT_flushBits(bitC);
-
-	if (bitC->ptr >= bitC->endPtr)
-		return 0; /* doesn't fit within authorized budget : cancel */
-
-	return (bitC->ptr - bitC->startPtr) + (bitC->bitPos > 0);
-}
-
-/*-********************************************************
-* bitStream decoding
-**********************************************************/
-/*! BIT_initDStream() :
-*   Initialize a BIT_DStream_t.
-*   `bitD` : a pointer to an already allocated BIT_DStream_t structure.
-*   `srcSize` must be the *exact* size of the bitStream, in bytes.
-*   @return : size of stream (== srcSize) or an errorCode if a problem is detected
-*/
-ZSTD_STATIC size_t BIT_initDStream(BIT_DStream_t *bitD, const void *srcBuffer, size_t srcSize)
-{
-	if (srcSize < 1) {
-		memset(bitD, 0, sizeof(*bitD));
-		return ERROR(srcSize_wrong);
-	}
-
-	if (srcSize >= sizeof(bitD->bitContainer)) { /* normal case */
-		bitD->start = (const char *)srcBuffer;
-		bitD->ptr = (const char *)srcBuffer + srcSize - sizeof(bitD->bitContainer);
-		bitD->bitContainer = ZSTD_readLEST(bitD->ptr);
-		{
-			BYTE const lastByte = ((const BYTE *)srcBuffer)[srcSize - 1];
-			bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0; /* ensures bitsConsumed is always set */
-			if (lastByte == 0)
-				return ERROR(GENERIC); /* endMark not present */
-		}
-	} else {
-		bitD->start = (const char *)srcBuffer;
-		bitD->ptr = bitD->start;
-		bitD->bitContainer = *(const BYTE *)(bitD->start);
-		switch (srcSize) {
-		case 7: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[6]) << (sizeof(bitD->bitContainer) * 8 - 16);
-			fallthrough;
-		case 6: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[5]) << (sizeof(bitD->bitContainer) * 8 - 24);
-			fallthrough;
-		case 5: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[4]) << (sizeof(bitD->bitContainer) * 8 - 32);
-			fallthrough;
-		case 4: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[3]) << 24;
-			fallthrough;
-		case 3: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[2]) << 16;
-			fallthrough;
-		case 2: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[1]) << 8;
-			fallthrough;
-		default:;
-		}
-		{
-			BYTE const lastByte = ((const BYTE *)srcBuffer)[srcSize - 1];
-			bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;
-			if (lastByte == 0)
-				return ERROR(GENERIC); /* endMark not present */
-		}
-		bitD->bitsConsumed += (U32)(sizeof(bitD->bitContainer) - srcSize) * 8;
-	}
-
-	return srcSize;
-}
-
-ZSTD_STATIC size_t BIT_getUpperBits(size_t bitContainer, U32 const start) { return bitContainer >> start; }
-
-ZSTD_STATIC size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits) { return (bitContainer >> start) & BIT_mask[nbBits]; }
-
-ZSTD_STATIC size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits) { return bitContainer & BIT_mask[nbBits]; }
-
-/*! BIT_lookBits() :
- *  Provides next n bits from local register.
- *  local register is not modified.
- *  On 32-bits, maxNbBits==24.
- *  On 64-bits, maxNbBits==56.
- *  @return : value extracted
- */
-ZSTD_STATIC size_t BIT_lookBits(const BIT_DStream_t *bitD, U32 nbBits)
-{
-	U32 const bitMask = sizeof(bitD->bitContainer) * 8 - 1;
-	return ((bitD->bitContainer << (bitD->bitsConsumed & bitMask)) >> 1) >> ((bitMask - nbBits) & bitMask);
-}
-
-/*! BIT_lookBitsFast() :
-*   unsafe version; only works only if nbBits >= 1 */
-ZSTD_STATIC size_t BIT_lookBitsFast(const BIT_DStream_t *bitD, U32 nbBits)
-{
-	U32 const bitMask = sizeof(bitD->bitContainer) * 8 - 1;
-	return (bitD->bitContainer << (bitD->bitsConsumed & bitMask)) >> (((bitMask + 1) - nbBits) & bitMask);
-}
-
-ZSTD_STATIC void BIT_skipBits(BIT_DStream_t *bitD, U32 nbBits) { bitD->bitsConsumed += nbBits; }
-
-/*! BIT_readBits() :
- *  Read (consume) next n bits from local register and update.
- *  Pay attention to not read more than nbBits contained into local register.
- *  @return : extracted value.
- */
-ZSTD_STATIC size_t BIT_readBits(BIT_DStream_t *bitD, U32 nbBits)
-{
-	size_t const value = BIT_lookBits(bitD, nbBits);
-	BIT_skipBits(bitD, nbBits);
-	return value;
-}
-
-/*! BIT_readBitsFast() :
-*   unsafe version; only works only if nbBits >= 1 */
-ZSTD_STATIC size_t BIT_readBitsFast(BIT_DStream_t *bitD, U32 nbBits)
-{
-	size_t const value = BIT_lookBitsFast(bitD, nbBits);
-	BIT_skipBits(bitD, nbBits);
-	return value;
-}
-
-/*! BIT_reloadDStream() :
-*   Refill `bitD` from buffer previously set in BIT_initDStream() .
-*   This function is safe, it guarantees it will not read beyond src buffer.
-*   @return : status of `BIT_DStream_t` internal register.
-			  if status == BIT_DStream_unfinished, internal register is filled with >= (sizeof(bitD->bitContainer)*8 - 7) bits */
-ZSTD_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t *bitD)
-{
-	if (bitD->bitsConsumed > (sizeof(bitD->bitContainer) * 8)) /* should not happen => corruption detected */
-		return BIT_DStream_overflow;
-
-	if (bitD->ptr >= bitD->start + sizeof(bitD->bitContainer)) {
-		bitD->ptr -= bitD->bitsConsumed >> 3;
-		bitD->bitsConsumed &= 7;
-		bitD->bitContainer = ZSTD_readLEST(bitD->ptr);
-		return BIT_DStream_unfinished;
-	}
-	if (bitD->ptr == bitD->start) {
-		if (bitD->bitsConsumed < sizeof(bitD->bitContainer) * 8)
-			return BIT_DStream_endOfBuffer;
-		return BIT_DStream_completed;
-	}
-	{
-		U32 nbBytes = bitD->bitsConsumed >> 3;
-		BIT_DStream_status result = BIT_DStream_unfinished;
-		if (bitD->ptr - nbBytes < bitD->start) {
-			nbBytes = (U32)(bitD->ptr - bitD->start); /* ptr > start */
-			result = BIT_DStream_endOfBuffer;
-		}
-		bitD->ptr -= nbBytes;
-		bitD->bitsConsumed -= nbBytes * 8;
-		bitD->bitContainer = ZSTD_readLEST(bitD->ptr); /* reminder : srcSize > sizeof(bitD) */
-		return result;
-	}
-}
-
-/*! BIT_endOfDStream() :
-*   @return Tells if DStream has exactly reached its end (all bits consumed).
-*/
-ZSTD_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t *DStream)
-{
-	return ((DStream->ptr == DStream->start) && (DStream->bitsConsumed == sizeof(DStream->bitContainer) * 8));
-}
-
-#endif /* BITSTREAM_H_MODULE */
diff --git a/lib/zstd/common/bitstream.h b/lib/zstd/common/bitstream.h
new file mode 100644
index 0000000..28248ab
--- /dev/null
+++ b/lib/zstd/common/bitstream.h
@@ -0,0 +1,437 @@
+/* ******************************************************************
+ * bitstream
+ * Part of FSE library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+#ifndef BITSTREAM_H_MODULE
+#define BITSTREAM_H_MODULE
+
+/*
+*  This API consists of small unitary functions, which must be inlined for best performance.
+*  Since link-time-optimization is not available for all compilers,
+*  these functions are defined into a .h to be included.
+*/
+
+/*-****************************************
+*  Dependencies
+******************************************/
+#include "mem.h"            /* unaligned access routines */
+#include "compiler.h"       /* UNLIKELY() */
+#include "debug.h"          /* assert(), DEBUGLOG(), RAWLOG() */
+#include "error_private.h"  /* error codes and messages */
+
+
+/*=========================================
+*  Target specific
+=========================================*/
+
+#define STREAM_ACCUMULATOR_MIN_32  25
+#define STREAM_ACCUMULATOR_MIN_64  57
+#define STREAM_ACCUMULATOR_MIN    ((U32)(MEM_32bits() ? STREAM_ACCUMULATOR_MIN_32 : STREAM_ACCUMULATOR_MIN_64))
+
+
+/*-******************************************
+*  bitStream encoding API (write forward)
+********************************************/
+/* bitStream can mix input from multiple sources.
+ * A critical property of these streams is that they encode and decode in **reverse** direction.
+ * So the first bit sequence you add will be the last to be read, like a LIFO stack.
+ */
+typedef struct {
+    size_t bitContainer;
+    unsigned bitPos;
+    char*  startPtr;
+    char*  ptr;
+    char*  endPtr;
+} BIT_CStream_t;
+
+MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC, void* dstBuffer, size_t dstCapacity);
+MEM_STATIC void   BIT_addBits(BIT_CStream_t* bitC, size_t value, unsigned nbBits);
+MEM_STATIC void   BIT_flushBits(BIT_CStream_t* bitC);
+MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC);
+
+/* Start with initCStream, providing the size of buffer to write into.
+*  bitStream will never write outside of this buffer.
+*  `dstCapacity` must be >= sizeof(bitD->bitContainer), otherwise @return will be an error code.
+*
+*  bits are first added to a local register.
+*  Local register is size_t, hence 64-bits on 64-bits systems, or 32-bits on 32-bits systems.
+*  Writing data into memory is an explicit operation, performed by the flushBits function.
+*  Hence keep track how many bits are potentially stored into local register to avoid register overflow.
+*  After a flushBits, a maximum of 7 bits might still be stored into local register.
+*
+*  Avoid storing elements of more than 24 bits if you want compatibility with 32-bits bitstream readers.
+*
+*  Last operation is to close the bitStream.
+*  The function returns the final size of CStream in bytes.
+*  If data couldn't fit into `dstBuffer`, it will return a 0 ( == not storable)
+*/
+
+
+/*-********************************************
+*  bitStream decoding API (read backward)
+**********************************************/
+typedef struct {
+    size_t   bitContainer;
+    unsigned bitsConsumed;
+    const char* ptr;
+    const char* start;
+    const char* limitPtr;
+} BIT_DStream_t;
+
+typedef enum { BIT_DStream_unfinished = 0,
+               BIT_DStream_endOfBuffer = 1,
+               BIT_DStream_completed = 2,
+               BIT_DStream_overflow = 3 } BIT_DStream_status;  /* result of BIT_reloadDStream() */
+               /* 1,2,4,8 would be better for bitmap combinations, but slows down performance a bit ... :( */
+
+MEM_STATIC size_t   BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize);
+MEM_STATIC size_t   BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits);
+MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD);
+MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* bitD);
+
+
+/* Start by invoking BIT_initDStream().
+*  A chunk of the bitStream is then stored into a local register.
+*  Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
+*  You can then retrieve bitFields stored into the local register, **in reverse order**.
+*  Local register is explicitly reloaded from memory by the BIT_reloadDStream() method.
+*  A reload guarantee a minimum of ((8*sizeof(bitD->bitContainer))-7) bits when its result is BIT_DStream_unfinished.
+*  Otherwise, it can be less than that, so proceed accordingly.
+*  Checking if DStream has reached its end can be performed with BIT_endOfDStream().
+*/
+
+
+/*-****************************************
+*  unsafe API
+******************************************/
+MEM_STATIC void BIT_addBitsFast(BIT_CStream_t* bitC, size_t value, unsigned nbBits);
+/* faster, but works only if value is "clean", meaning all high bits above nbBits are 0 */
+
+MEM_STATIC void BIT_flushBitsFast(BIT_CStream_t* bitC);
+/* unsafe version; does not check buffer overflow */
+
+MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits);
+/* faster, but works only if nbBits >= 1 */
+
+
+
+/*-**************************************************************
+*  Internal functions
+****************************************************************/
+MEM_STATIC unsigned BIT_highbit32 (U32 val)
+{
+    assert(val != 0);
+    {
+#   if (__GNUC__ >= 3)   /* Use GCC Intrinsic */
+        return __builtin_clz (val) ^ 31;
+#   else   /* Software version */
+        static const unsigned DeBruijnClz[32] = { 0,  9,  1, 10, 13, 21,  2, 29,
+                                                 11, 14, 16, 18, 22, 25,  3, 30,
+                                                  8, 12, 20, 28, 15, 17, 24,  7,
+                                                 19, 27, 23,  6, 26,  5,  4, 31 };
+        U32 v = val;
+        v |= v >> 1;
+        v |= v >> 2;
+        v |= v >> 4;
+        v |= v >> 8;
+        v |= v >> 16;
+        return DeBruijnClz[ (U32) (v * 0x07C4ACDDU) >> 27];
+#   endif
+    }
+}
+
+/*=====    Local Constants   =====*/
+static const unsigned BIT_mask[] = {
+    0,          1,         3,         7,         0xF,       0x1F,
+    0x3F,       0x7F,      0xFF,      0x1FF,     0x3FF,     0x7FF,
+    0xFFF,      0x1FFF,    0x3FFF,    0x7FFF,    0xFFFF,    0x1FFFF,
+    0x3FFFF,    0x7FFFF,   0xFFFFF,   0x1FFFFF,  0x3FFFFF,  0x7FFFFF,
+    0xFFFFFF,   0x1FFFFFF, 0x3FFFFFF, 0x7FFFFFF, 0xFFFFFFF, 0x1FFFFFFF,
+    0x3FFFFFFF, 0x7FFFFFFF}; /* up to 31 bits */
+#define BIT_MASK_SIZE (sizeof(BIT_mask) / sizeof(BIT_mask[0]))
+
+/*-**************************************************************
+*  bitStream encoding
+****************************************************************/
+/*! BIT_initCStream() :
+ *  `dstCapacity` must be > sizeof(size_t)
+ *  @return : 0 if success,
+ *            otherwise an error code (can be tested using ERR_isError()) */
+MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC,
+                                  void* startPtr, size_t dstCapacity)
+{
+    bitC->bitContainer = 0;
+    bitC->bitPos = 0;
+    bitC->startPtr = (char*)startPtr;
+    bitC->ptr = bitC->startPtr;
+    bitC->endPtr = bitC->startPtr + dstCapacity - sizeof(bitC->bitContainer);
+    if (dstCapacity <= sizeof(bitC->bitContainer)) return ERROR(dstSize_tooSmall);
+    return 0;
+}
+
+/*! BIT_addBits() :
+ *  can add up to 31 bits into `bitC`.
+ *  Note : does not check for register overflow ! */
+MEM_STATIC void BIT_addBits(BIT_CStream_t* bitC,
+                            size_t value, unsigned nbBits)
+{
+    DEBUG_STATIC_ASSERT(BIT_MASK_SIZE == 32);
+    assert(nbBits < BIT_MASK_SIZE);
+    assert(nbBits + bitC->bitPos < sizeof(bitC->bitContainer) * 8);
+    bitC->bitContainer |= (value & BIT_mask[nbBits]) << bitC->bitPos;
+    bitC->bitPos += nbBits;
+}
+
+/*! BIT_addBitsFast() :
+ *  works only if `value` is _clean_,
+ *  meaning all high bits above nbBits are 0 */
+MEM_STATIC void BIT_addBitsFast(BIT_CStream_t* bitC,
+                                size_t value, unsigned nbBits)
+{
+    assert((value>>nbBits) == 0);
+    assert(nbBits + bitC->bitPos < sizeof(bitC->bitContainer) * 8);
+    bitC->bitContainer |= value << bitC->bitPos;
+    bitC->bitPos += nbBits;
+}
+
+/*! BIT_flushBitsFast() :
+ *  assumption : bitContainer has not overflowed
+ *  unsafe version; does not check buffer overflow */
+MEM_STATIC void BIT_flushBitsFast(BIT_CStream_t* bitC)
+{
+    size_t const nbBytes = bitC->bitPos >> 3;
+    assert(bitC->bitPos < sizeof(bitC->bitContainer) * 8);
+    assert(bitC->ptr <= bitC->endPtr);
+    MEM_writeLEST(bitC->ptr, bitC->bitContainer);
+    bitC->ptr += nbBytes;
+    bitC->bitPos &= 7;
+    bitC->bitContainer >>= nbBytes*8;
+}
+
+/*! BIT_flushBits() :
+ *  assumption : bitContainer has not overflowed
+ *  safe version; check for buffer overflow, and prevents it.
+ *  note : does not signal buffer overflow.
+ *  overflow will be revealed later on using BIT_closeCStream() */
+MEM_STATIC void BIT_flushBits(BIT_CStream_t* bitC)
+{
+    size_t const nbBytes = bitC->bitPos >> 3;
+    assert(bitC->bitPos < sizeof(bitC->bitContainer) * 8);
+    assert(bitC->ptr <= bitC->endPtr);
+    MEM_writeLEST(bitC->ptr, bitC->bitContainer);
+    bitC->ptr += nbBytes;
+    if (bitC->ptr > bitC->endPtr) bitC->ptr = bitC->endPtr;
+    bitC->bitPos &= 7;
+    bitC->bitContainer >>= nbBytes*8;
+}
+
+/*! BIT_closeCStream() :
+ *  @return : size of CStream, in bytes,
+ *            or 0 if it could not fit into dstBuffer */
+MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC)
+{
+    BIT_addBitsFast(bitC, 1, 1);   /* endMark */
+    BIT_flushBits(bitC);
+    if (bitC->ptr >= bitC->endPtr) return 0; /* overflow detected */
+    return (bitC->ptr - bitC->startPtr) + (bitC->bitPos > 0);
+}
+
+
+/*-********************************************************
+*  bitStream decoding
+**********************************************************/
+/*! BIT_initDStream() :
+ *  Initialize a BIT_DStream_t.
+ * `bitD` : a pointer to an already allocated BIT_DStream_t structure.
+ * `srcSize` must be the *exact* size of the bitStream, in bytes.
+ * @return : size of stream (== srcSize), or an errorCode if a problem is detected
+ */
+MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize)
+{
+    if (srcSize < 1) { ZSTD_memset(bitD, 0, sizeof(*bitD)); return ERROR(srcSize_wrong); }
+
+    bitD->start = (const char*)srcBuffer;
+    bitD->limitPtr = bitD->start + sizeof(bitD->bitContainer);
+
+    if (srcSize >=  sizeof(bitD->bitContainer)) {  /* normal case */
+        bitD->ptr   = (const char*)srcBuffer + srcSize - sizeof(bitD->bitContainer);
+        bitD->bitContainer = MEM_readLEST(bitD->ptr);
+        { BYTE const lastByte = ((const BYTE*)srcBuffer)[srcSize-1];
+          bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;  /* ensures bitsConsumed is always set */
+          if (lastByte == 0) return ERROR(GENERIC); /* endMark not present */ }
+    } else {
+        bitD->ptr   = bitD->start;
+        bitD->bitContainer = *(const BYTE*)(bitD->start);
+        switch(srcSize)
+        {
+        case 7: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[6]) << (sizeof(bitD->bitContainer)*8 - 16);
+                ZSTD_FALLTHROUGH;
+
+        case 6: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[5]) << (sizeof(bitD->bitContainer)*8 - 24);
+                ZSTD_FALLTHROUGH;
+
+        case 5: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[4]) << (sizeof(bitD->bitContainer)*8 - 32);
+                ZSTD_FALLTHROUGH;
+
+        case 4: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[3]) << 24;
+                ZSTD_FALLTHROUGH;
+
+        case 3: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[2]) << 16;
+                ZSTD_FALLTHROUGH;
+
+        case 2: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[1]) <<  8;
+                ZSTD_FALLTHROUGH;
+
+        default: break;
+        }
+        {   BYTE const lastByte = ((const BYTE*)srcBuffer)[srcSize-1];
+            bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;
+            if (lastByte == 0) return ERROR(corruption_detected);  /* endMark not present */
+        }
+        bitD->bitsConsumed += (U32)(sizeof(bitD->bitContainer) - srcSize)*8;
+    }
+
+    return srcSize;
+}
+
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getUpperBits(size_t bitContainer, U32 const start)
+{
+    return bitContainer >> start;
+}
+
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits)
+{
+    U32 const regMask = sizeof(bitContainer)*8 - 1;
+    /* if start > regMask, bitstream is corrupted, and result is undefined */
+    assert(nbBits < BIT_MASK_SIZE);
+    return (bitContainer >> (start & regMask)) & BIT_mask[nbBits];
+}
+
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
+{
+    assert(nbBits < BIT_MASK_SIZE);
+    return bitContainer & BIT_mask[nbBits];
+}
+
+/*! BIT_lookBits() :
+ *  Provides next n bits from local register.
+ *  local register is not modified.
+ *  On 32-bits, maxNbBits==24.
+ *  On 64-bits, maxNbBits==56.
+ * @return : value extracted */
+MEM_STATIC  FORCE_INLINE_ATTR size_t BIT_lookBits(const BIT_DStream_t*  bitD, U32 nbBits)
+{
+    /* arbitrate between double-shift and shift+mask */
+#if 1
+    /* if bitD->bitsConsumed + nbBits > sizeof(bitD->bitContainer)*8,
+     * bitstream is likely corrupted, and result is undefined */
+    return BIT_getMiddleBits(bitD->bitContainer, (sizeof(bitD->bitContainer)*8) - bitD->bitsConsumed - nbBits, nbBits);
+#else
+    /* this code path is slower on my os-x laptop */
+    U32 const regMask = sizeof(bitD->bitContainer)*8 - 1;
+    return ((bitD->bitContainer << (bitD->bitsConsumed & regMask)) >> 1) >> ((regMask-nbBits) & regMask);
+#endif
+}
+
+/*! BIT_lookBitsFast() :
+ *  unsafe version; only works if nbBits >= 1 */
+MEM_STATIC size_t BIT_lookBitsFast(const BIT_DStream_t* bitD, U32 nbBits)
+{
+    U32 const regMask = sizeof(bitD->bitContainer)*8 - 1;
+    assert(nbBits >= 1);
+    return (bitD->bitContainer << (bitD->bitsConsumed & regMask)) >> (((regMask+1)-nbBits) & regMask);
+}
+
+MEM_STATIC FORCE_INLINE_ATTR void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
+{
+    bitD->bitsConsumed += nbBits;
+}
+
+/*! BIT_readBits() :
+ *  Read (consume) next n bits from local register and update.
+ *  Pay attention to not read more than nbBits contained into local register.
+ * @return : extracted value. */
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits)
+{
+    size_t const value = BIT_lookBits(bitD, nbBits);
+    BIT_skipBits(bitD, nbBits);
+    return value;
+}
+
+/*! BIT_readBitsFast() :
+ *  unsafe version; only works only if nbBits >= 1 */
+MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits)
+{
+    size_t const value = BIT_lookBitsFast(bitD, nbBits);
+    assert(nbBits >= 1);
+    BIT_skipBits(bitD, nbBits);
+    return value;
+}
+
+/*! BIT_reloadDStreamFast() :
+ *  Similar to BIT_reloadDStream(), but with two differences:
+ *  1. bitsConsumed <= sizeof(bitD->bitContainer)*8 must hold!
+ *  2. Returns BIT_DStream_overflow when bitD->ptr < bitD->limitPtr, at this
+ *     point you must use BIT_reloadDStream() to reload.
+ */
+MEM_STATIC BIT_DStream_status BIT_reloadDStreamFast(BIT_DStream_t* bitD)
+{
+    if (UNLIKELY(bitD->ptr < bitD->limitPtr))
+        return BIT_DStream_overflow;
+    assert(bitD->bitsConsumed <= sizeof(bitD->bitContainer)*8);
+    bitD->ptr -= bitD->bitsConsumed >> 3;
+    bitD->bitsConsumed &= 7;
+    bitD->bitContainer = MEM_readLEST(bitD->ptr);
+    return BIT_DStream_unfinished;
+}
+
+/*! BIT_reloadDStream() :
+ *  Refill `bitD` from buffer previously set in BIT_initDStream() .
+ *  This function is safe, it guarantees it will not read beyond src buffer.
+ * @return : status of `BIT_DStream_t` internal register.
+ *           when status == BIT_DStream_unfinished, internal register is filled with at least 25 or 57 bits */
+MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
+{
+    if (bitD->bitsConsumed > (sizeof(bitD->bitContainer)*8))  /* overflow detected, like end of stream */
+        return BIT_DStream_overflow;
+
+    if (bitD->ptr >= bitD->limitPtr) {
+        return BIT_reloadDStreamFast(bitD);
+    }
+    if (bitD->ptr == bitD->start) {
+        if (bitD->bitsConsumed < sizeof(bitD->bitContainer)*8) return BIT_DStream_endOfBuffer;
+        return BIT_DStream_completed;
+    }
+    /* start < ptr < limitPtr */
+    {   U32 nbBytes = bitD->bitsConsumed >> 3;
+        BIT_DStream_status result = BIT_DStream_unfinished;
+        if (bitD->ptr - nbBytes < bitD->start) {
+            nbBytes = (U32)(bitD->ptr - bitD->start);  /* ptr > start */
+            result = BIT_DStream_endOfBuffer;
+        }
+        bitD->ptr -= nbBytes;
+        bitD->bitsConsumed -= nbBytes*8;
+        bitD->bitContainer = MEM_readLEST(bitD->ptr);   /* reminder : srcSize > sizeof(bitD->bitContainer), otherwise bitD->ptr == bitD->start */
+        return result;
+    }
+}
+
+/*! BIT_endOfDStream() :
+ * @return : 1 if DStream has _exactly_ reached its end (all bits consumed).
+ */
+MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* DStream)
+{
+    return ((DStream->ptr == DStream->start) && (DStream->bitsConsumed == sizeof(DStream->bitContainer)*8));
+}
+
+
+#endif /* BITSTREAM_H_MODULE */
diff --git a/lib/zstd/common/compiler.h b/lib/zstd/common/compiler.h
new file mode 100644
index 0000000..a1a051e
--- /dev/null
+++ b/lib/zstd/common/compiler.h
@@ -0,0 +1,170 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_COMPILER_H
+#define ZSTD_COMPILER_H
+
+/*-*******************************************************
+*  Compiler specifics
+*********************************************************/
+/* force inlining */
+
+#if (defined(__GNUC__) && !defined(__STRICT_ANSI__)) || defined(__cplusplus) || defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L   /* C99 */
+#  define INLINE_KEYWORD inline
+#else
+#  define INLINE_KEYWORD
+#endif
+
+#define FORCE_INLINE_ATTR __attribute__((always_inline))
+
+
+/*
+  On MSVC qsort requires that functions passed into it use the __cdecl calling conversion(CC).
+  This explictly marks such functions as __cdecl so that the code will still compile
+  if a CC other than __cdecl has been made the default.
+*/
+#define WIN_CDECL
+
+/*
+ * FORCE_INLINE_TEMPLATE is used to define C "templates", which take constant
+ * parameters. They must be inlined for the compiler to eliminate the constant
+ * branches.
+ */
+#define FORCE_INLINE_TEMPLATE static INLINE_KEYWORD FORCE_INLINE_ATTR
+/*
+ * HINT_INLINE is used to help the compiler generate better code. It is *not*
+ * used for "templates", so it can be tweaked based on the compilers
+ * performance.
+ *
+ * gcc-4.8 and gcc-4.9 have been shown to benefit from leaving off the
+ * always_inline attribute.
+ *
+ * clang up to 5.0.0 (trunk) benefit tremendously from the always_inline
+ * attribute.
+ */
+#if !defined(__clang__) && defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 8 && __GNUC__ < 5
+#  define HINT_INLINE static INLINE_KEYWORD
+#else
+#  define HINT_INLINE static INLINE_KEYWORD FORCE_INLINE_ATTR
+#endif
+
+/* UNUSED_ATTR tells the compiler it is okay if the function is unused. */
+#define UNUSED_ATTR __attribute__((unused))
+
+/* force no inlining */
+#define FORCE_NOINLINE static __attribute__((__noinline__))
+
+
+/* target attribute */
+#ifndef __has_attribute
+  #define __has_attribute(x) 0  /* Compatibility with non-clang compilers. */
+#endif
+#define TARGET_ATTRIBUTE(target) __attribute__((__target__(target)))
+
+/* Enable runtime BMI2 dispatch based on the CPU.
+ * Enabled for clang & gcc >=4.8 on x86 when BMI2 isn't enabled by default.
+ */
+#ifndef DYNAMIC_BMI2
+  #if ((defined(__clang__) && __has_attribute(__target__)) \
+      || (defined(__GNUC__) \
+          && (__GNUC__ >= 5 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)))) \
+      && (defined(__x86_64__) || defined(_M_X86)) \
+      && !defined(__BMI2__)
+  #  define DYNAMIC_BMI2 1
+  #else
+  #  define DYNAMIC_BMI2 0
+  #endif
+#endif
+
+/* prefetch
+ * can be disabled, by declaring NO_PREFETCH build macro */
+#if ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) )
+#  define PREFETCH_L1(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */)
+#  define PREFETCH_L2(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */)
+#elif defined(__aarch64__)
+#  define PREFETCH_L1(ptr)  __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr)))
+#  define PREFETCH_L2(ptr)  __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr)))
+#else
+#  define PREFETCH_L1(ptr) (void)(ptr)  /* disabled */
+#  define PREFETCH_L2(ptr) (void)(ptr)  /* disabled */
+#endif  /* NO_PREFETCH */
+
+#define CACHELINE_SIZE 64
+
+#define PREFETCH_AREA(p, s)  {            \
+    const char* const _ptr = (const char*)(p);  \
+    size_t const _size = (size_t)(s);     \
+    size_t _pos;                          \
+    for (_pos=0; _pos<_size; _pos+=CACHELINE_SIZE) {  \
+        PREFETCH_L2(_ptr + _pos);         \
+    }                                     \
+}
+
+/* vectorization
+ * older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
+#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
+#  if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) || (__GNUC__ >= 5)
+#    define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
+#  else
+#    define DONT_VECTORIZE _Pragma("GCC optimize(\"no-tree-vectorize\")")
+#  endif
+#else
+#  define DONT_VECTORIZE
+#endif
+
+/* Tell the compiler that a branch is likely or unlikely.
+ * Only use these macros if it causes the compiler to generate better code.
+ * If you can remove a LIKELY/UNLIKELY annotation without speed changes in gcc
+ * and clang, please do.
+ */
+#define LIKELY(x) (__builtin_expect((x), 1))
+#define UNLIKELY(x) (__builtin_expect((x), 0))
+
+/* disable warnings */
+
+/*Like DYNAMIC_BMI2 but for compile time determination of BMI2 support*/
+
+
+/* compat. with non-clang compilers */
+#ifndef __has_builtin
+#  define __has_builtin(x) 0
+#endif
+
+/* compat. with non-clang compilers */
+#ifndef __has_feature
+#  define __has_feature(x) 0
+#endif
+
+/* C-language Attributes are added in C23. */
+#if defined(__STDC_VERSION__) && (__STDC_VERSION__ > 201710L) && defined(__has_c_attribute)
+# define ZSTD_HAS_C_ATTRIBUTE(x) __has_c_attribute(x)
+#else
+# define ZSTD_HAS_C_ATTRIBUTE(x) 0
+#endif
+
+/* Only use C++ attributes in C++. Some compilers report support for C++
+ * attributes when compiling with C.
+ */
+#define ZSTD_HAS_CPP_ATTRIBUTE(x) 0
+
+/* Define ZSTD_FALLTHROUGH macro for annotating switch case with the 'fallthrough' attribute.
+ * - C23: https://en.cppreference.com/w/c/language/attributes/fallthrough
+ * - CPP17: https://en.cppreference.com/w/cpp/language/attributes/fallthrough
+ * - Else: __attribute__((__fallthrough__))
+ */
+#define ZSTD_FALLTHROUGH fallthrough
+
+/* detects whether we are being compiled under msan */
+
+
+/* detects whether we are being compiled under asan */
+
+
+#endif /* ZSTD_COMPILER_H */
diff --git a/lib/zstd/common/cpu.h b/lib/zstd/common/cpu.h
new file mode 100644
index 0000000..0db7b42
--- /dev/null
+++ b/lib/zstd/common/cpu.h
@@ -0,0 +1,194 @@
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_COMMON_CPU_H
+#define ZSTD_COMMON_CPU_H
+
+/*
+ * Implementation taken from folly/CpuId.h
+ * https://github.com/facebook/folly/blob/master/folly/CpuId.h
+ */
+
+#include "mem.h"
+
+
+typedef struct {
+    U32 f1c;
+    U32 f1d;
+    U32 f7b;
+    U32 f7c;
+} ZSTD_cpuid_t;
+
+MEM_STATIC ZSTD_cpuid_t ZSTD_cpuid(void) {
+    U32 f1c = 0;
+    U32 f1d = 0;
+    U32 f7b = 0;
+    U32 f7c = 0;
+#if defined(__i386__) && defined(__PIC__) && !defined(__clang__) && defined(__GNUC__)
+    /* The following block like the normal cpuid branch below, but gcc
+     * reserves ebx for use of its pic register so we must specially
+     * handle the save and restore to avoid clobbering the register
+     */
+    U32 n;
+    __asm__(
+        "pushl %%ebx\n\t"
+        "cpuid\n\t"
+        "popl %%ebx\n\t"
+        : "=a"(n)
+        : "a"(0)
+        : "ecx", "edx");
+    if (n >= 1) {
+      U32 f1a;
+      __asm__(
+          "pushl %%ebx\n\t"
+          "cpuid\n\t"
+          "popl %%ebx\n\t"
+          : "=a"(f1a), "=c"(f1c), "=d"(f1d)
+          : "a"(1));
+    }
+    if (n >= 7) {
+      __asm__(
+          "pushl %%ebx\n\t"
+          "cpuid\n\t"
+          "movl %%ebx, %%eax\n\t"
+          "popl %%ebx"
+          : "=a"(f7b), "=c"(f7c)
+          : "a"(7), "c"(0)
+          : "edx");
+    }
+#elif defined(__x86_64__) || defined(_M_X64) || defined(__i386__)
+    U32 n;
+    __asm__("cpuid" : "=a"(n) : "a"(0) : "ebx", "ecx", "edx");
+    if (n >= 1) {
+      U32 f1a;
+      __asm__("cpuid" : "=a"(f1a), "=c"(f1c), "=d"(f1d) : "a"(1) : "ebx");
+    }
+    if (n >= 7) {
+      U32 f7a;
+      __asm__("cpuid"
+              : "=a"(f7a), "=b"(f7b), "=c"(f7c)
+              : "a"(7), "c"(0)
+              : "edx");
+    }
+#endif
+    {
+        ZSTD_cpuid_t cpuid;
+        cpuid.f1c = f1c;
+        cpuid.f1d = f1d;
+        cpuid.f7b = f7b;
+        cpuid.f7c = f7c;
+        return cpuid;
+    }
+}
+
+#define X(name, r, bit)                                                        \
+  MEM_STATIC int ZSTD_cpuid_##name(ZSTD_cpuid_t const cpuid) {                 \
+    return ((cpuid.r) & (1U << bit)) != 0;                                     \
+  }
+
+/* cpuid(1): Processor Info and Feature Bits. */
+#define C(name, bit) X(name, f1c, bit)
+  C(sse3, 0)
+  C(pclmuldq, 1)
+  C(dtes64, 2)
+  C(monitor, 3)
+  C(dscpl, 4)
+  C(vmx, 5)
+  C(smx, 6)
+  C(eist, 7)
+  C(tm2, 8)
+  C(ssse3, 9)
+  C(cnxtid, 10)
+  C(fma, 12)
+  C(cx16, 13)
+  C(xtpr, 14)
+  C(pdcm, 15)
+  C(pcid, 17)
+  C(dca, 18)
+  C(sse41, 19)
+  C(sse42, 20)
+  C(x2apic, 21)
+  C(movbe, 22)
+  C(popcnt, 23)
+  C(tscdeadline, 24)
+  C(aes, 25)
+  C(xsave, 26)
+  C(osxsave, 27)
+  C(avx, 28)
+  C(f16c, 29)
+  C(rdrand, 30)
+#undef C
+#define D(name, bit) X(name, f1d, bit)
+  D(fpu, 0)
+  D(vme, 1)
+  D(de, 2)
+  D(pse, 3)
+  D(tsc, 4)
+  D(msr, 5)
+  D(pae, 6)
+  D(mce, 7)
+  D(cx8, 8)
+  D(apic, 9)
+  D(sep, 11)
+  D(mtrr, 12)
+  D(pge, 13)
+  D(mca, 14)
+  D(cmov, 15)
+  D(pat, 16)
+  D(pse36, 17)
+  D(psn, 18)
+  D(clfsh, 19)
+  D(ds, 21)
+  D(acpi, 22)
+  D(mmx, 23)
+  D(fxsr, 24)
+  D(sse, 25)
+  D(sse2, 26)
+  D(ss, 27)
+  D(htt, 28)
+  D(tm, 29)
+  D(pbe, 31)
+#undef D
+
+/* cpuid(7): Extended Features. */
+#define B(name, bit) X(name, f7b, bit)
+  B(bmi1, 3)
+  B(hle, 4)
+  B(avx2, 5)
+  B(smep, 7)
+  B(bmi2, 8)
+  B(erms, 9)
+  B(invpcid, 10)
+  B(rtm, 11)
+  B(mpx, 14)
+  B(avx512f, 16)
+  B(avx512dq, 17)
+  B(rdseed, 18)
+  B(adx, 19)
+  B(smap, 20)
+  B(avx512ifma, 21)
+  B(pcommit, 22)
+  B(clflushopt, 23)
+  B(clwb, 24)
+  B(avx512pf, 26)
+  B(avx512er, 27)
+  B(avx512cd, 28)
+  B(sha, 29)
+  B(avx512bw, 30)
+  B(avx512vl, 31)
+#undef B
+#define C(name, bit) X(name, f7c, bit)
+  C(prefetchwt1, 0)
+  C(avx512vbmi, 1)
+#undef C
+
+#undef X
+
+#endif /* ZSTD_COMMON_CPU_H */
diff --git a/lib/zstd/common/debug.c b/lib/zstd/common/debug.c
new file mode 100644
index 0000000..bb863c9
--- /dev/null
+++ b/lib/zstd/common/debug.c
@@ -0,0 +1,24 @@
+/* ******************************************************************
+ * debug
+ * Part of FSE library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+/*
+ * This module only hosts one global variable
+ * which can be used to dynamically influence the verbosity of traces,
+ * such as DEBUGLOG and RAWLOG
+ */
+
+#include "debug.h"
+
+int g_debuglevel = DEBUGLEVEL;
diff --git a/lib/zstd/common/debug.h b/lib/zstd/common/debug.h
new file mode 100644
index 0000000..6dd88d1
--- /dev/null
+++ b/lib/zstd/common/debug.h
@@ -0,0 +1,101 @@
+/* ******************************************************************
+ * debug
+ * Part of FSE library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+/*
+ * The purpose of this header is to enable debug functions.
+ * They regroup assert(), DEBUGLOG() and RAWLOG() for run-time,
+ * and DEBUG_STATIC_ASSERT() for compile-time.
+ *
+ * By default, DEBUGLEVEL==0, which means run-time debug is disabled.
+ *
+ * Level 1 enables assert() only.
+ * Starting level 2, traces can be generated and pushed to stderr.
+ * The higher the level, the more verbose the traces.
+ *
+ * It's possible to dynamically adjust level using variable g_debug_level,
+ * which is only declared if DEBUGLEVEL>=2,
+ * and is a global variable, not multi-thread protected (use with care)
+ */
+
+#ifndef DEBUG_H_12987983217
+#define DEBUG_H_12987983217
+
+
+
+/* static assert is triggered at compile time, leaving no runtime artefact.
+ * static assert only works with compile-time constants.
+ * Also, this variant can only be used inside a function. */
+#define DEBUG_STATIC_ASSERT(c) (void)sizeof(char[(c) ? 1 : -1])
+
+
+/* DEBUGLEVEL is expected to be defined externally,
+ * typically through compiler command line.
+ * Value must be a number. */
+#ifndef DEBUGLEVEL
+#  define DEBUGLEVEL 0
+#endif
+
+
+/* recommended values for DEBUGLEVEL :
+ * 0 : release mode, no debug, all run-time checks disabled
+ * 1 : enables assert() only, no display
+ * 2 : reserved, for currently active debug path
+ * 3 : events once per object lifetime (CCtx, CDict, etc.)
+ * 4 : events once per frame
+ * 5 : events once per block
+ * 6 : events once per sequence (verbose)
+ * 7+: events at every position (*very* verbose)
+ *
+ * It's generally inconvenient to output traces > 5.
+ * In which case, it's possible to selectively trigger high verbosity levels
+ * by modifying g_debug_level.
+ */
+
+#if (DEBUGLEVEL>=1)
+#  define ZSTD_DEPS_NEED_ASSERT
+#  include "zstd_deps.h"
+#else
+#  ifndef assert   /* assert may be already defined, due to prior #include <assert.h> */
+#    define assert(condition) ((void)0)   /* disable assert (default) */
+#  endif
+#endif
+
+#if (DEBUGLEVEL>=2)
+#  define ZSTD_DEPS_NEED_IO
+#  include "zstd_deps.h"
+extern int g_debuglevel; /* the variable is only declared,
+                            it actually lives in debug.c,
+                            and is shared by the whole process.
+                            It's not thread-safe.
+                            It's useful when enabling very verbose levels
+                            on selective conditions (such as position in src) */
+
+#  define RAWLOG(l, ...) {                                       \
+                if (l<=g_debuglevel) {                           \
+                    ZSTD_DEBUG_PRINT(__VA_ARGS__);               \
+            }   }
+#  define DEBUGLOG(l, ...) {                                     \
+                if (l<=g_debuglevel) {                           \
+                    ZSTD_DEBUG_PRINT(__FILE__ ": " __VA_ARGS__); \
+                    ZSTD_DEBUG_PRINT(" \n");                     \
+            }   }
+#else
+#  define RAWLOG(l, ...)      {}    /* disabled */
+#  define DEBUGLOG(l, ...)    {}    /* disabled */
+#endif
+
+
+
+#endif /* DEBUG_H_12987983217 */
diff --git a/lib/zstd/common/entropy_common.c b/lib/zstd/common/entropy_common.c
new file mode 100644
index 0000000..53b47a2
--- /dev/null
+++ b/lib/zstd/common/entropy_common.c
@@ -0,0 +1,357 @@
+/* ******************************************************************
+ * Common functions of New Generation Entropy library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ *  You can contact the author at :
+ *  - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+/* *************************************
+*  Dependencies
+***************************************/
+#include "mem.h"
+#include "error_private.h"       /* ERR_*, ERROR */
+#define FSE_STATIC_LINKING_ONLY  /* FSE_MIN_TABLELOG */
+#include "fse.h"
+#define HUF_STATIC_LINKING_ONLY  /* HUF_TABLELOG_ABSOLUTEMAX */
+#include "huf.h"
+
+
+/*===   Version   ===*/
+unsigned FSE_versionNumber(void) { return FSE_VERSION_NUMBER; }
+
+
+/*===   Error Management   ===*/
+unsigned FSE_isError(size_t code) { return ERR_isError(code); }
+const char* FSE_getErrorName(size_t code) { return ERR_getErrorName(code); }
+
+unsigned HUF_isError(size_t code) { return ERR_isError(code); }
+const char* HUF_getErrorName(size_t code) { return ERR_getErrorName(code); }
+
+
+/*-**************************************************************
+*  FSE NCount encoding-decoding
+****************************************************************/
+static U32 FSE_ctz(U32 val)
+{
+    assert(val != 0);
+    {
+#   if (__GNUC__ >= 3)   /* GCC Intrinsic */
+        return __builtin_ctz(val);
+#   else   /* Software version */
+        U32 count = 0;
+        while ((val & 1) == 0) {
+            val >>= 1;
+            ++count;
+        }
+        return count;
+#   endif
+    }
+}
+
+FORCE_INLINE_TEMPLATE
+size_t FSE_readNCount_body(short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+                           const void* headerBuffer, size_t hbSize)
+{
+    const BYTE* const istart = (const BYTE*) headerBuffer;
+    const BYTE* const iend = istart + hbSize;
+    const BYTE* ip = istart;
+    int nbBits;
+    int remaining;
+    int threshold;
+    U32 bitStream;
+    int bitCount;
+    unsigned charnum = 0;
+    unsigned const maxSV1 = *maxSVPtr + 1;
+    int previous0 = 0;
+
+    if (hbSize < 8) {
+        /* This function only works when hbSize >= 8 */
+        char buffer[8] = {0};
+        ZSTD_memcpy(buffer, headerBuffer, hbSize);
+        {   size_t const countSize = FSE_readNCount(normalizedCounter, maxSVPtr, tableLogPtr,
+                                                    buffer, sizeof(buffer));
+            if (FSE_isError(countSize)) return countSize;
+            if (countSize > hbSize) return ERROR(corruption_detected);
+            return countSize;
+    }   }
+    assert(hbSize >= 8);
+
+    /* init */
+    ZSTD_memset(normalizedCounter, 0, (*maxSVPtr+1) * sizeof(normalizedCounter[0]));   /* all symbols not present in NCount have a frequency of 0 */
+    bitStream = MEM_readLE32(ip);
+    nbBits = (bitStream & 0xF) + FSE_MIN_TABLELOG;   /* extract tableLog */
+    if (nbBits > FSE_TABLELOG_ABSOLUTE_MAX) return ERROR(tableLog_tooLarge);
+    bitStream >>= 4;
+    bitCount = 4;
+    *tableLogPtr = nbBits;
+    remaining = (1<<nbBits)+1;
+    threshold = 1<<nbBits;
+    nbBits++;
+
+    for (;;) {
+        if (previous0) {
+            /* Count the number of repeats. Each time the
+             * 2-bit repeat code is 0b11 there is another
+             * repeat.
+             * Avoid UB by setting the high bit to 1.
+             */
+            int repeats = FSE_ctz(~bitStream | 0x80000000) >> 1;
+            while (repeats >= 12) {
+                charnum += 3 * 12;
+                if (LIKELY(ip <= iend-7)) {
+                    ip += 3;
+                } else {
+                    bitCount -= (int)(8 * (iend - 7 - ip));
+                    bitCount &= 31;
+                    ip = iend - 4;
+                }
+                bitStream = MEM_readLE32(ip) >> bitCount;
+                repeats = FSE_ctz(~bitStream | 0x80000000) >> 1;
+            }
+            charnum += 3 * repeats;
+            bitStream >>= 2 * repeats;
+            bitCount += 2 * repeats;
+
+            /* Add the final repeat which isn't 0b11. */
+            assert((bitStream & 3) < 3);
+            charnum += bitStream & 3;
+            bitCount += 2;
+
+            /* This is an error, but break and return an error
+             * at the end, because returning out of a loop makes
+             * it harder for the compiler to optimize.
+             */
+            if (charnum >= maxSV1) break;
+
+            /* We don't need to set the normalized count to 0
+             * because we already memset the whole buffer to 0.
+             */
+
+            if (LIKELY(ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) {
+                assert((bitCount >> 3) <= 3); /* For first condition to work */
+                ip += bitCount>>3;
+                bitCount &= 7;
+            } else {
+                bitCount -= (int)(8 * (iend - 4 - ip));
+                bitCount &= 31;
+                ip = iend - 4;
+            }
+            bitStream = MEM_readLE32(ip) >> bitCount;
+        }
+        {
+            int const max = (2*threshold-1) - remaining;
+            int count;
+
+            if ((bitStream & (threshold-1)) < (U32)max) {
+                count = bitStream & (threshold-1);
+                bitCount += nbBits-1;
+            } else {
+                count = bitStream & (2*threshold-1);
+                if (count >= threshold) count -= max;
+                bitCount += nbBits;
+            }
+
+            count--;   /* extra accuracy */
+            /* When it matters (small blocks), this is a
+             * predictable branch, because we don't use -1.
+             */
+            if (count >= 0) {
+                remaining -= count;
+            } else {
+                assert(count == -1);
+                remaining += count;
+            }
+            normalizedCounter[charnum++] = (short)count;
+            previous0 = !count;
+
+            assert(threshold > 1);
+            if (remaining < threshold) {
+                /* This branch can be folded into the
+                 * threshold update condition because we
+                 * know that threshold > 1.
+                 */
+                if (remaining <= 1) break;
+                nbBits = BIT_highbit32(remaining) + 1;
+                threshold = 1 << (nbBits - 1);
+            }
+            if (charnum >= maxSV1) break;
+
+            if (LIKELY(ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) {
+                ip += bitCount>>3;
+                bitCount &= 7;
+            } else {
+                bitCount -= (int)(8 * (iend - 4 - ip));
+                bitCount &= 31;
+                ip = iend - 4;
+            }
+            bitStream = MEM_readLE32(ip) >> bitCount;
+    }   }
+    if (remaining != 1) return ERROR(corruption_detected);
+    /* Only possible when there are too many zeros. */
+    if (charnum > maxSV1) return ERROR(maxSymbolValue_tooSmall);
+    if (bitCount > 32) return ERROR(corruption_detected);
+    *maxSVPtr = charnum-1;
+
+    ip += (bitCount+7)>>3;
+    return ip-istart;
+}
+
+/* Avoids the FORCE_INLINE of the _body() function. */
+static size_t FSE_readNCount_body_default(
+        short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+        const void* headerBuffer, size_t hbSize)
+{
+    return FSE_readNCount_body(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
+}
+
+#if DYNAMIC_BMI2
+TARGET_ATTRIBUTE("bmi2") static size_t FSE_readNCount_body_bmi2(
+        short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+        const void* headerBuffer, size_t hbSize)
+{
+    return FSE_readNCount_body(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
+}
+#endif
+
+size_t FSE_readNCount_bmi2(
+        short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+        const void* headerBuffer, size_t hbSize, int bmi2)
+{
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        return FSE_readNCount_body_bmi2(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
+    }
+#endif
+    (void)bmi2;
+    return FSE_readNCount_body_default(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize);
+}
+
+size_t FSE_readNCount(
+        short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
+        const void* headerBuffer, size_t hbSize)
+{
+    return FSE_readNCount_bmi2(normalizedCounter, maxSVPtr, tableLogPtr, headerBuffer, hbSize, /* bmi2 */ 0);
+}
+
+
+/*! HUF_readStats() :
+    Read compact Huffman tree, saved by HUF_writeCTable().
+    `huffWeight` is destination buffer.
+    `rankStats` is assumed to be a table of at least HUF_TABLELOG_MAX U32.
+    @return : size read from `src` , or an error Code .
+    Note : Needed by HUF_readCTable() and HUF_readDTableX?() .
+*/
+size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                     U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize)
+{
+    U32 wksp[HUF_READ_STATS_WORKSPACE_SIZE_U32];
+    return HUF_readStats_wksp(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, wksp, sizeof(wksp), /* bmi2 */ 0);
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_readStats_body(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                   U32* nbSymbolsPtr, U32* tableLogPtr,
+                   const void* src, size_t srcSize,
+                   void* workSpace, size_t wkspSize,
+                   int bmi2)
+{
+    U32 weightTotal;
+    const BYTE* ip = (const BYTE*) src;
+    size_t iSize;
+    size_t oSize;
+
+    if (!srcSize) return ERROR(srcSize_wrong);
+    iSize = ip[0];
+    /* ZSTD_memset(huffWeight, 0, hwSize);   *//* is not necessary, even though some analyzer complain ... */
+
+    if (iSize >= 128) {  /* special header */
+        oSize = iSize - 127;
+        iSize = ((oSize+1)/2);
+        if (iSize+1 > srcSize) return ERROR(srcSize_wrong);
+        if (oSize >= hwSize) return ERROR(corruption_detected);
+        ip += 1;
+        {   U32 n;
+            for (n=0; n<oSize; n+=2) {
+                huffWeight[n]   = ip[n/2] >> 4;
+                huffWeight[n+1] = ip[n/2] & 15;
+    }   }   }
+    else  {   /* header compressed with FSE (normal case) */
+        if (iSize+1 > srcSize) return ERROR(srcSize_wrong);
+        /* max (hwSize-1) values decoded, as last one is implied */
+        oSize = FSE_decompress_wksp_bmi2(huffWeight, hwSize-1, ip+1, iSize, 6, workSpace, wkspSize, bmi2);
+        if (FSE_isError(oSize)) return oSize;
+    }
+
+    /* collect weight stats */
+    ZSTD_memset(rankStats, 0, (HUF_TABLELOG_MAX + 1) * sizeof(U32));
+    weightTotal = 0;
+    {   U32 n; for (n=0; n<oSize; n++) {
+            if (huffWeight[n] >= HUF_TABLELOG_MAX) return ERROR(corruption_detected);
+            rankStats[huffWeight[n]]++;
+            weightTotal += (1 << huffWeight[n]) >> 1;
+    }   }
+    if (weightTotal == 0) return ERROR(corruption_detected);
+
+    /* get last non-null symbol weight (implied, total must be 2^n) */
+    {   U32 const tableLog = BIT_highbit32(weightTotal) + 1;
+        if (tableLog > HUF_TABLELOG_MAX) return ERROR(corruption_detected);
+        *tableLogPtr = tableLog;
+        /* determine last weight */
+        {   U32 const total = 1 << tableLog;
+            U32 const rest = total - weightTotal;
+            U32 const verif = 1 << BIT_highbit32(rest);
+            U32 const lastWeight = BIT_highbit32(rest) + 1;
+            if (verif != rest) return ERROR(corruption_detected);    /* last value must be a clean power of 2 */
+            huffWeight[oSize] = (BYTE)lastWeight;
+            rankStats[lastWeight]++;
+    }   }
+
+    /* check tree construction validity */
+    if ((rankStats[1] < 2) || (rankStats[1] & 1)) return ERROR(corruption_detected);   /* by construction : at least 2 elts of rank 1, must be even */
+
+    /* results */
+    *nbSymbolsPtr = (U32)(oSize+1);
+    return iSize+1;
+}
+
+/* Avoids the FORCE_INLINE of the _body() function. */
+static size_t HUF_readStats_body_default(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                     U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize,
+                     void* workSpace, size_t wkspSize)
+{
+    return HUF_readStats_body(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize, 0);
+}
+
+#if DYNAMIC_BMI2
+static TARGET_ATTRIBUTE("bmi2") size_t HUF_readStats_body_bmi2(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                     U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize,
+                     void* workSpace, size_t wkspSize)
+{
+    return HUF_readStats_body(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize, 1);
+}
+#endif
+
+size_t HUF_readStats_wksp(BYTE* huffWeight, size_t hwSize, U32* rankStats,
+                     U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize,
+                     void* workSpace, size_t wkspSize,
+                     int bmi2)
+{
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        return HUF_readStats_body_bmi2(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize);
+    }
+#endif
+    (void)bmi2;
+    return HUF_readStats_body_default(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize);
+}
diff --git a/lib/zstd/common/error_private.c b/lib/zstd/common/error_private.c
new file mode 100644
index 0000000..6d1135f
--- /dev/null
+++ b/lib/zstd/common/error_private.c
@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* The purpose of this file is to have a single list of error strings embedded in binary */
+
+#include "error_private.h"
+
+const char* ERR_getErrorString(ERR_enum code)
+{
+#ifdef ZSTD_STRIP_ERROR_STRINGS
+    (void)code;
+    return "Error strings stripped";
+#else
+    static const char* const notErrorCode = "Unspecified error code";
+    switch( code )
+    {
+    case PREFIX(no_error): return "No error detected";
+    case PREFIX(GENERIC):  return "Error (generic)";
+    case PREFIX(prefix_unknown): return "Unknown frame descriptor";
+    case PREFIX(version_unsupported): return "Version not supported";
+    case PREFIX(frameParameter_unsupported): return "Unsupported frame parameter";
+    case PREFIX(frameParameter_windowTooLarge): return "Frame requires too much memory for decoding";
+    case PREFIX(corruption_detected): return "Corrupted block detected";
+    case PREFIX(checksum_wrong): return "Restored data doesn't match checksum";
+    case PREFIX(parameter_unsupported): return "Unsupported parameter";
+    case PREFIX(parameter_outOfBound): return "Parameter is out of bound";
+    case PREFIX(init_missing): return "Context should be init first";
+    case PREFIX(memory_allocation): return "Allocation error : not enough memory";
+    case PREFIX(workSpace_tooSmall): return "workSpace buffer is not large enough";
+    case PREFIX(stage_wrong): return "Operation not authorized at current processing stage";
+    case PREFIX(tableLog_tooLarge): return "tableLog requires too much memory : unsupported";
+    case PREFIX(maxSymbolValue_tooLarge): return "Unsupported max Symbol Value : too large";
+    case PREFIX(maxSymbolValue_tooSmall): return "Specified maxSymbolValue is too small";
+    case PREFIX(dictionary_corrupted): return "Dictionary is corrupted";
+    case PREFIX(dictionary_wrong): return "Dictionary mismatch";
+    case PREFIX(dictionaryCreation_failed): return "Cannot create Dictionary from provided samples";
+    case PREFIX(dstSize_tooSmall): return "Destination buffer is too small";
+    case PREFIX(srcSize_wrong): return "Src size is incorrect";
+    case PREFIX(dstBuffer_null): return "Operation on NULL destination buffer";
+        /* following error codes are not stable and may be removed or changed in a future version */
+    case PREFIX(frameIndex_tooLarge): return "Frame index is too large";
+    case PREFIX(seekableIO): return "An I/O error occurred when reading/seeking";
+    case PREFIX(dstBuffer_wrong): return "Destination buffer is wrong";
+    case PREFIX(srcBuffer_wrong): return "Source buffer is wrong";
+    case PREFIX(maxCode):
+    default: return notErrorCode;
+    }
+#endif
+}
diff --git a/lib/zstd/common/error_private.h b/lib/zstd/common/error_private.h
new file mode 100644
index 0000000..d14e686
--- /dev/null
+++ b/lib/zstd/common/error_private.h
@@ -0,0 +1,66 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* Note : this module is expected to remain private, do not expose it */
+
+#ifndef ERROR_H_MODULE
+#define ERROR_H_MODULE
+
+
+
+/* ****************************************
+*  Dependencies
+******************************************/
+#include "zstd_deps.h"    /* size_t */
+#include <linux/zstd_errors.h>  /* enum list */
+
+
+/* ****************************************
+*  Compiler-specific
+******************************************/
+#define ERR_STATIC static __attribute__((unused))
+
+
+/*-****************************************
+*  Customization (error_public.h)
+******************************************/
+typedef ZSTD_ErrorCode ERR_enum;
+#define PREFIX(name) ZSTD_error_##name
+
+
+/*-****************************************
+*  Error codes handling
+******************************************/
+#undef ERROR   /* already defined on Visual Studio */
+#define ERROR(name) ZSTD_ERROR(name)
+#define ZSTD_ERROR(name) ((size_t)-PREFIX(name))
+
+ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); }
+
+ERR_STATIC ERR_enum ERR_getErrorCode(size_t code) { if (!ERR_isError(code)) return (ERR_enum)0; return (ERR_enum) (0-code); }
+
+/* check and forward error code */
+#define CHECK_V_F(e, f) size_t const e = f; if (ERR_isError(e)) return e
+#define CHECK_F(f)   { CHECK_V_F(_var_err__, f); }
+
+
+/*-****************************************
+*  Error Strings
+******************************************/
+
+const char* ERR_getErrorString(ERR_enum code);   /* error_private.c */
+
+ERR_STATIC const char* ERR_getErrorName(size_t code)
+{
+    return ERR_getErrorString(ERR_getErrorCode(code));
+}
+
+
+#endif /* ERROR_H_MODULE */
diff --git a/lib/zstd/common/fse.h b/lib/zstd/common/fse.h
new file mode 100644
index 0000000..0bb174c
--- /dev/null
+++ b/lib/zstd/common/fse.h
@@ -0,0 +1,710 @@
+/* ******************************************************************
+ * FSE : Finite State Entropy codec
+ * Public Prototypes declaration
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+#ifndef FSE_H
+#define FSE_H
+
+
+/*-*****************************************
+*  Dependencies
+******************************************/
+#include "zstd_deps.h"    /* size_t, ptrdiff_t */
+
+
+/*-*****************************************
+*  FSE_PUBLIC_API : control library symbols visibility
+******************************************/
+#if defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) && defined(__GNUC__) && (__GNUC__ >= 4)
+#  define FSE_PUBLIC_API __attribute__ ((visibility ("default")))
+#elif defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1)   /* Visual expected */
+#  define FSE_PUBLIC_API __declspec(dllexport)
+#elif defined(FSE_DLL_IMPORT) && (FSE_DLL_IMPORT==1)
+#  define FSE_PUBLIC_API __declspec(dllimport) /* It isn't required but allows to generate better code, saving a function pointer load from the IAT and an indirect jump.*/
+#else
+#  define FSE_PUBLIC_API
+#endif
+
+/*------   Version   ------*/
+#define FSE_VERSION_MAJOR    0
+#define FSE_VERSION_MINOR    9
+#define FSE_VERSION_RELEASE  0
+
+#define FSE_LIB_VERSION FSE_VERSION_MAJOR.FSE_VERSION_MINOR.FSE_VERSION_RELEASE
+#define FSE_QUOTE(str) #str
+#define FSE_EXPAND_AND_QUOTE(str) FSE_QUOTE(str)
+#define FSE_VERSION_STRING FSE_EXPAND_AND_QUOTE(FSE_LIB_VERSION)
+
+#define FSE_VERSION_NUMBER  (FSE_VERSION_MAJOR *100*100 + FSE_VERSION_MINOR *100 + FSE_VERSION_RELEASE)
+FSE_PUBLIC_API unsigned FSE_versionNumber(void);   /*< library version number; to be used when checking dll version */
+
+
+/*-****************************************
+*  FSE simple functions
+******************************************/
+/*! FSE_compress() :
+    Compress content of buffer 'src', of size 'srcSize', into destination buffer 'dst'.
+    'dst' buffer must be already allocated. Compression runs faster is dstCapacity >= FSE_compressBound(srcSize).
+    @return : size of compressed data (<= dstCapacity).
+    Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
+                     if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression instead.
+                     if FSE_isError(return), compression failed (more details using FSE_getErrorName())
+*/
+FSE_PUBLIC_API size_t FSE_compress(void* dst, size_t dstCapacity,
+                             const void* src, size_t srcSize);
+
+/*! FSE_decompress():
+    Decompress FSE data from buffer 'cSrc', of size 'cSrcSize',
+    into already allocated destination buffer 'dst', of size 'dstCapacity'.
+    @return : size of regenerated data (<= maxDstSize),
+              or an error code, which can be tested using FSE_isError() .
+
+    ** Important ** : FSE_decompress() does not decompress non-compressible nor RLE data !!!
+    Why ? : making this distinction requires a header.
+    Header management is intentionally delegated to the user layer, which can better manage special cases.
+*/
+FSE_PUBLIC_API size_t FSE_decompress(void* dst,  size_t dstCapacity,
+                               const void* cSrc, size_t cSrcSize);
+
+
+/*-*****************************************
+*  Tool functions
+******************************************/
+FSE_PUBLIC_API size_t FSE_compressBound(size_t size);       /* maximum compressed size */
+
+/* Error Management */
+FSE_PUBLIC_API unsigned    FSE_isError(size_t code);        /* tells if a return value is an error code */
+FSE_PUBLIC_API const char* FSE_getErrorName(size_t code);   /* provides error code string (useful for debugging) */
+
+
+/*-*****************************************
+*  FSE advanced functions
+******************************************/
+/*! FSE_compress2() :
+    Same as FSE_compress(), but allows the selection of 'maxSymbolValue' and 'tableLog'
+    Both parameters can be defined as '0' to mean : use default value
+    @return : size of compressed data
+    Special values : if return == 0, srcData is not compressible => Nothing is stored within cSrc !!!
+                     if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression.
+                     if FSE_isError(return), it's an error code.
+*/
+FSE_PUBLIC_API size_t FSE_compress2 (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
+
+
+/*-*****************************************
+*  FSE detailed API
+******************************************/
+/*!
+FSE_compress() does the following:
+1. count symbol occurrence from source[] into table count[] (see hist.h)
+2. normalize counters so that sum(count[]) == Power_of_2 (2^tableLog)
+3. save normalized counters to memory buffer using writeNCount()
+4. build encoding table 'CTable' from normalized counters
+5. encode the data stream using encoding table 'CTable'
+
+FSE_decompress() does the following:
+1. read normalized counters with readNCount()
+2. build decoding table 'DTable' from normalized counters
+3. decode the data stream using decoding table 'DTable'
+
+The following API allows targeting specific sub-functions for advanced tasks.
+For example, it's possible to compress several blocks using the same 'CTable',
+or to save and provide normalized distribution using external method.
+*/
+
+/* *** COMPRESSION *** */
+
+/*! FSE_optimalTableLog():
+    dynamically downsize 'tableLog' when conditions are met.
+    It saves CPU time, by using smaller tables, while preserving or even improving compression ratio.
+    @return : recommended tableLog (necessarily <= 'maxTableLog') */
+FSE_PUBLIC_API unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
+
+/*! FSE_normalizeCount():
+    normalize counts so that sum(count[]) == Power_of_2 (2^tableLog)
+    'normalizedCounter' is a table of short, of minimum size (maxSymbolValue+1).
+    useLowProbCount is a boolean parameter which trades off compressed size for
+    faster header decoding. When it is set to 1, the compressed data will be slightly
+    smaller. And when it is set to 0, FSE_readNCount() and FSE_buildDTable() will be
+    faster. If you are compressing a small amount of data (< 2 KB) then useLowProbCount=0
+    is a good default, since header deserialization makes a big speed difference.
+    Otherwise, useLowProbCount=1 is a good default, since the speed difference is small.
+    @return : tableLog,
+              or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_normalizeCount(short* normalizedCounter, unsigned tableLog,
+                    const unsigned* count, size_t srcSize, unsigned maxSymbolValue, unsigned useLowProbCount);
+
+/*! FSE_NCountWriteBound():
+    Provides the maximum possible size of an FSE normalized table, given 'maxSymbolValue' and 'tableLog'.
+    Typically useful for allocation purpose. */
+FSE_PUBLIC_API size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog);
+
+/*! FSE_writeNCount():
+    Compactly save 'normalizedCounter' into 'buffer'.
+    @return : size of the compressed table,
+              or an errorCode, which can be tested using FSE_isError(). */
+FSE_PUBLIC_API size_t FSE_writeNCount (void* buffer, size_t bufferSize,
+                                 const short* normalizedCounter,
+                                 unsigned maxSymbolValue, unsigned tableLog);
+
+/*! Constructor and Destructor of FSE_CTable.
+    Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */
+typedef unsigned FSE_CTable;   /* don't allocate that. It's only meant to be more restrictive than void* */
+FSE_PUBLIC_API FSE_CTable* FSE_createCTable (unsigned maxSymbolValue, unsigned tableLog);
+FSE_PUBLIC_API void        FSE_freeCTable (FSE_CTable* ct);
+
+/*! FSE_buildCTable():
+    Builds `ct`, which must be already allocated, using FSE_createCTable().
+    @return : 0, or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_buildCTable(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
+
+/*! FSE_compress_usingCTable():
+    Compress `src` using `ct` into `dst` which must be already allocated.
+    @return : size of compressed data (<= `dstCapacity`),
+              or 0 if compressed data could not fit into `dst`,
+              or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_compress_usingCTable (void* dst, size_t dstCapacity, const void* src, size_t srcSize, const FSE_CTable* ct);
+
+/*!
+Tutorial :
+----------
+The first step is to count all symbols. FSE_count() does this job very fast.
+Result will be saved into 'count', a table of unsigned int, which must be already allocated, and have 'maxSymbolValuePtr[0]+1' cells.
+'src' is a table of bytes of size 'srcSize'. All values within 'src' MUST be <= maxSymbolValuePtr[0]
+maxSymbolValuePtr[0] will be updated, with its real value (necessarily <= original value)
+FSE_count() will return the number of occurrence of the most frequent symbol.
+This can be used to know if there is a single symbol within 'src', and to quickly evaluate its compressibility.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
+
+The next step is to normalize the frequencies.
+FSE_normalizeCount() will ensure that sum of frequencies is == 2 ^'tableLog'.
+It also guarantees a minimum of 1 to any Symbol with frequency >= 1.
+You can use 'tableLog'==0 to mean "use default tableLog value".
+If you are unsure of which tableLog value to use, you can ask FSE_optimalTableLog(),
+which will provide the optimal valid tableLog given sourceSize, maxSymbolValue, and a user-defined maximum (0 means "default").
+
+The result of FSE_normalizeCount() will be saved into a table,
+called 'normalizedCounter', which is a table of signed short.
+'normalizedCounter' must be already allocated, and have at least 'maxSymbolValue+1' cells.
+The return value is tableLog if everything proceeded as expected.
+It is 0 if there is a single symbol within distribution.
+If there is an error (ex: invalid tableLog value), the function will return an ErrorCode (which can be tested using FSE_isError()).
+
+'normalizedCounter' can be saved in a compact manner to a memory area using FSE_writeNCount().
+'buffer' must be already allocated.
+For guaranteed success, buffer size must be at least FSE_headerBound().
+The result of the function is the number of bytes written into 'buffer'.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError(); ex : buffer size too small).
+
+'normalizedCounter' can then be used to create the compression table 'CTable'.
+The space required by 'CTable' must be already allocated, using FSE_createCTable().
+You can then use FSE_buildCTable() to fill 'CTable'.
+If there is an error, both functions will return an ErrorCode (which can be tested using FSE_isError()).
+
+'CTable' can then be used to compress 'src', with FSE_compress_usingCTable().
+Similar to FSE_count(), the convention is that 'src' is assumed to be a table of char of size 'srcSize'
+The function returns the size of compressed data (without header), necessarily <= `dstCapacity`.
+If it returns '0', compressed data could not fit into 'dst'.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
+*/
+
+
+/* *** DECOMPRESSION *** */
+
+/*! FSE_readNCount():
+    Read compactly saved 'normalizedCounter' from 'rBuffer'.
+    @return : size read from 'rBuffer',
+              or an errorCode, which can be tested using FSE_isError().
+              maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */
+FSE_PUBLIC_API size_t FSE_readNCount (short* normalizedCounter,
+                           unsigned* maxSymbolValuePtr, unsigned* tableLogPtr,
+                           const void* rBuffer, size_t rBuffSize);
+
+/*! FSE_readNCount_bmi2():
+ * Same as FSE_readNCount() but pass bmi2=1 when your CPU supports BMI2 and 0 otherwise.
+ */
+FSE_PUBLIC_API size_t FSE_readNCount_bmi2(short* normalizedCounter,
+                           unsigned* maxSymbolValuePtr, unsigned* tableLogPtr,
+                           const void* rBuffer, size_t rBuffSize, int bmi2);
+
+/*! Constructor and Destructor of FSE_DTable.
+    Note that its size depends on 'tableLog' */
+typedef unsigned FSE_DTable;   /* don't allocate that. It's just a way to be more restrictive than void* */
+FSE_PUBLIC_API FSE_DTable* FSE_createDTable(unsigned tableLog);
+FSE_PUBLIC_API void        FSE_freeDTable(FSE_DTable* dt);
+
+/*! FSE_buildDTable():
+    Builds 'dt', which must be already allocated, using FSE_createDTable().
+    return : 0, or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_buildDTable (FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
+
+/*! FSE_decompress_usingDTable():
+    Decompress compressed source `cSrc` of size `cSrcSize` using `dt`
+    into `dst` which must be already allocated.
+    @return : size of regenerated data (necessarily <= `dstCapacity`),
+              or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_decompress_usingDTable(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt);
+
+/*!
+Tutorial :
+----------
+(Note : these functions only decompress FSE-compressed blocks.
+ If block is uncompressed, use memcpy() instead
+ If block is a single repeated byte, use memset() instead )
+
+The first step is to obtain the normalized frequencies of symbols.
+This can be performed by FSE_readNCount() if it was saved using FSE_writeNCount().
+'normalizedCounter' must be already allocated, and have at least 'maxSymbolValuePtr[0]+1' cells of signed short.
+In practice, that means it's necessary to know 'maxSymbolValue' beforehand,
+or size the table to handle worst case situations (typically 256).
+FSE_readNCount() will provide 'tableLog' and 'maxSymbolValue'.
+The result of FSE_readNCount() is the number of bytes read from 'rBuffer'.
+Note that 'rBufferSize' must be at least 4 bytes, even if useful information is less than that.
+If there is an error, the function will return an error code, which can be tested using FSE_isError().
+
+The next step is to build the decompression tables 'FSE_DTable' from 'normalizedCounter'.
+This is performed by the function FSE_buildDTable().
+The space required by 'FSE_DTable' must be already allocated using FSE_createDTable().
+If there is an error, the function will return an error code, which can be tested using FSE_isError().
+
+`FSE_DTable` can then be used to decompress `cSrc`, with FSE_decompress_usingDTable().
+`cSrcSize` must be strictly correct, otherwise decompression will fail.
+FSE_decompress_usingDTable() result will tell how many bytes were regenerated (<=`dstCapacity`).
+If there is an error, the function will return an error code, which can be tested using FSE_isError(). (ex: dst buffer too small)
+*/
+
+#endif  /* FSE_H */
+
+#if !defined(FSE_H_FSE_STATIC_LINKING_ONLY)
+#define FSE_H_FSE_STATIC_LINKING_ONLY
+
+/* *** Dependency *** */
+#include "bitstream.h"
+
+
+/* *****************************************
+*  Static allocation
+*******************************************/
+/* FSE buffer bounds */
+#define FSE_NCOUNTBOUND 512
+#define FSE_BLOCKBOUND(size) ((size) + ((size)>>7) + 4 /* fse states */ + sizeof(size_t) /* bitContainer */)
+#define FSE_COMPRESSBOUND(size) (FSE_NCOUNTBOUND + FSE_BLOCKBOUND(size))   /* Macro version, useful for static allocation */
+
+/* It is possible to statically allocate FSE CTable/DTable as a table of FSE_CTable/FSE_DTable using below macros */
+#define FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue)   (1 + (1<<((maxTableLog)-1)) + (((maxSymbolValue)+1)*2))
+#define FSE_DTABLE_SIZE_U32(maxTableLog)                   (1 + (1<<(maxTableLog)))
+
+/* or use the size to malloc() space directly. Pay attention to alignment restrictions though */
+#define FSE_CTABLE_SIZE(maxTableLog, maxSymbolValue)   (FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) * sizeof(FSE_CTable))
+#define FSE_DTABLE_SIZE(maxTableLog)                   (FSE_DTABLE_SIZE_U32(maxTableLog) * sizeof(FSE_DTable))
+
+
+/* *****************************************
+ *  FSE advanced API
+ ***************************************** */
+
+unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus);
+/*< same as FSE_optimalTableLog(), which used `minus==2` */
+
+/* FSE_compress_wksp() :
+ * Same as FSE_compress2(), but using an externally allocated scratch buffer (`workSpace`).
+ * FSE_COMPRESS_WKSP_SIZE_U32() provides the minimum size required for `workSpace` as a table of FSE_CTable.
+ */
+#define FSE_COMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue)   ( FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) + ((maxTableLog > 12) ? (1 << (maxTableLog - 2)) : 1024) )
+size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
+
+size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits);
+/*< build a fake FSE_CTable, designed for a flat distribution, where each symbol uses nbBits */
+
+size_t FSE_buildCTable_rle (FSE_CTable* ct, unsigned char symbolValue);
+/*< build a fake FSE_CTable, designed to compress always the same symbolValue */
+
+/* FSE_buildCTable_wksp() :
+ * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
+ * `wkspSize` must be >= `FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog)` of `unsigned`.
+ */
+#define FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog) (maxSymbolValue + 2 + (1ull << (tableLog - 2)))
+#define FSE_BUILD_CTABLE_WORKSPACE_SIZE(maxSymbolValue, tableLog) (sizeof(unsigned) * FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog))
+size_t FSE_buildCTable_wksp(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
+
+#define FSE_BUILD_DTABLE_WKSP_SIZE(maxTableLog, maxSymbolValue) (sizeof(short) * (maxSymbolValue + 1) + (1ULL << maxTableLog) + 8)
+#define FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) ((FSE_BUILD_DTABLE_WKSP_SIZE(maxTableLog, maxSymbolValue) + sizeof(unsigned) - 1) / sizeof(unsigned))
+FSE_PUBLIC_API size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
+/*< Same as FSE_buildDTable(), using an externally allocated `workspace` produced with `FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxSymbolValue)` */
+
+size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits);
+/*< build a fake FSE_DTable, designed to read a flat distribution where each symbol uses nbBits */
+
+size_t FSE_buildDTable_rle (FSE_DTable* dt, unsigned char symbolValue);
+/*< build a fake FSE_DTable, designed to always generate the same symbolValue */
+
+#define FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) (FSE_DTABLE_SIZE_U32(maxTableLog) + FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) + (FSE_MAX_SYMBOL_VALUE + 1) / 2 + 1)
+#define FSE_DECOMPRESS_WKSP_SIZE(maxTableLog, maxSymbolValue) (FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) * sizeof(unsigned))
+size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize);
+/*< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DECOMPRESS_WKSP_SIZE_U32(maxLog, maxSymbolValue)` */
+
+size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize, int bmi2);
+/*< Same as FSE_decompress_wksp() but with dynamic BMI2 support. Pass 1 if your CPU supports BMI2 or 0 if it doesn't. */
+
+typedef enum {
+   FSE_repeat_none,  /*< Cannot use the previous table */
+   FSE_repeat_check, /*< Can use the previous table but it must be checked */
+   FSE_repeat_valid  /*< Can use the previous table and it is assumed to be valid */
+ } FSE_repeat;
+
+/* *****************************************
+*  FSE symbol compression API
+*******************************************/
+/*!
+   This API consists of small unitary functions, which highly benefit from being inlined.
+   Hence their body are included in next section.
+*/
+typedef struct {
+    ptrdiff_t   value;
+    const void* stateTable;
+    const void* symbolTT;
+    unsigned    stateLog;
+} FSE_CState_t;
+
+static void FSE_initCState(FSE_CState_t* CStatePtr, const FSE_CTable* ct);
+
+static void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* CStatePtr, unsigned symbol);
+
+static void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* CStatePtr);
+
+/*<
+These functions are inner components of FSE_compress_usingCTable().
+They allow the creation of custom streams, mixing multiple tables and bit sources.
+
+A key property to keep in mind is that encoding and decoding are done **in reverse direction**.
+So the first symbol you will encode is the last you will decode, like a LIFO stack.
+
+You will need a few variables to track your CStream. They are :
+
+FSE_CTable    ct;         // Provided by FSE_buildCTable()
+BIT_CStream_t bitStream;  // bitStream tracking structure
+FSE_CState_t  state;      // State tracking structure (can have several)
+
+
+The first thing to do is to init bitStream and state.
+    size_t errorCode = BIT_initCStream(&bitStream, dstBuffer, maxDstSize);
+    FSE_initCState(&state, ct);
+
+Note that BIT_initCStream() can produce an error code, so its result should be tested, using FSE_isError();
+You can then encode your input data, byte after byte.
+FSE_encodeSymbol() outputs a maximum of 'tableLog' bits at a time.
+Remember decoding will be done in reverse direction.
+    FSE_encodeByte(&bitStream, &state, symbol);
+
+At any time, you can also add any bit sequence.
+Note : maximum allowed nbBits is 25, for compatibility with 32-bits decoders
+    BIT_addBits(&bitStream, bitField, nbBits);
+
+The above methods don't commit data to memory, they just store it into local register, for speed.
+Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
+Writing data to memory is a manual operation, performed by the flushBits function.
+    BIT_flushBits(&bitStream);
+
+Your last FSE encoding operation shall be to flush your last state value(s).
+    FSE_flushState(&bitStream, &state);
+
+Finally, you must close the bitStream.
+The function returns the size of CStream in bytes.
+If data couldn't fit into dstBuffer, it will return a 0 ( == not compressible)
+If there is an error, it returns an errorCode (which can be tested using FSE_isError()).
+    size_t size = BIT_closeCStream(&bitStream);
+*/
+
+
+/* *****************************************
+*  FSE symbol decompression API
+*******************************************/
+typedef struct {
+    size_t      state;
+    const void* table;   /* precise table may vary, depending on U16 */
+} FSE_DState_t;
+
+
+static void     FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt);
+
+static unsigned char FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD);
+
+static unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr);
+
+/*<
+Let's now decompose FSE_decompress_usingDTable() into its unitary components.
+You will decode FSE-encoded symbols from the bitStream,
+and also any other bitFields you put in, **in reverse order**.
+
+You will need a few variables to track your bitStream. They are :
+
+BIT_DStream_t DStream;    // Stream context
+FSE_DState_t  DState;     // State context. Multiple ones are possible
+FSE_DTable*   DTablePtr;  // Decoding table, provided by FSE_buildDTable()
+
+The first thing to do is to init the bitStream.
+    errorCode = BIT_initDStream(&DStream, srcBuffer, srcSize);
+
+You should then retrieve your initial state(s)
+(in reverse flushing order if you have several ones) :
+    errorCode = FSE_initDState(&DState, &DStream, DTablePtr);
+
+You can then decode your data, symbol after symbol.
+For information the maximum number of bits read by FSE_decodeSymbol() is 'tableLog'.
+Keep in mind that symbols are decoded in reverse order, like a LIFO stack (last in, first out).
+    unsigned char symbol = FSE_decodeSymbol(&DState, &DStream);
+
+You can retrieve any bitfield you eventually stored into the bitStream (in reverse order)
+Note : maximum allowed nbBits is 25, for 32-bits compatibility
+    size_t bitField = BIT_readBits(&DStream, nbBits);
+
+All above operations only read from local register (which size depends on size_t).
+Refueling the register from memory is manually performed by the reload method.
+    endSignal = FSE_reloadDStream(&DStream);
+
+BIT_reloadDStream() result tells if there is still some more data to read from DStream.
+BIT_DStream_unfinished : there is still some data left into the DStream.
+BIT_DStream_endOfBuffer : Dstream reached end of buffer. Its container may no longer be completely filled.
+BIT_DStream_completed : Dstream reached its exact end, corresponding in general to decompression completed.
+BIT_DStream_tooFar : Dstream went too far. Decompression result is corrupted.
+
+When reaching end of buffer (BIT_DStream_endOfBuffer), progress slowly, notably if you decode multiple symbols per loop,
+to properly detect the exact end of stream.
+After each decoded symbol, check if DStream is fully consumed using this simple test :
+    BIT_reloadDStream(&DStream) >= BIT_DStream_completed
+
+When it's done, verify decompression is fully completed, by checking both DStream and the relevant states.
+Checking if DStream has reached its end is performed by :
+    BIT_endOfDStream(&DStream);
+Check also the states. There might be some symbols left there, if some high probability ones (>50%) are possible.
+    FSE_endOfDState(&DState);
+*/
+
+
+/* *****************************************
+*  FSE unsafe API
+*******************************************/
+static unsigned char FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD);
+/* faster, but works only if nbBits is always >= 1 (otherwise, result will be corrupted) */
+
+
+/* *****************************************
+*  Implementation of inlined functions
+*******************************************/
+typedef struct {
+    int deltaFindState;
+    U32 deltaNbBits;
+} FSE_symbolCompressionTransform; /* total 8 bytes */
+
+MEM_STATIC void FSE_initCState(FSE_CState_t* statePtr, const FSE_CTable* ct)
+{
+    const void* ptr = ct;
+    const U16* u16ptr = (const U16*) ptr;
+    const U32 tableLog = MEM_read16(ptr);
+    statePtr->value = (ptrdiff_t)1<<tableLog;
+    statePtr->stateTable = u16ptr+2;
+    statePtr->symbolTT = ct + 1 + (tableLog ? (1<<(tableLog-1)) : 1);
+    statePtr->stateLog = tableLog;
+}
+
+
+/*! FSE_initCState2() :
+*   Same as FSE_initCState(), but the first symbol to include (which will be the last to be read)
+*   uses the smallest state value possible, saving the cost of this symbol */
+MEM_STATIC void FSE_initCState2(FSE_CState_t* statePtr, const FSE_CTable* ct, U32 symbol)
+{
+    FSE_initCState(statePtr, ct);
+    {   const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
+        const U16* stateTable = (const U16*)(statePtr->stateTable);
+        U32 nbBitsOut  = (U32)((symbolTT.deltaNbBits + (1<<15)) >> 16);
+        statePtr->value = (nbBitsOut << 16) - symbolTT.deltaNbBits;
+        statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
+    }
+}
+
+MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, unsigned symbol)
+{
+    FSE_symbolCompressionTransform const symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
+    const U16* const stateTable = (const U16*)(statePtr->stateTable);
+    U32 const nbBitsOut  = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16);
+    BIT_addBits(bitC, statePtr->value, nbBitsOut);
+    statePtr->value = stateTable[ (statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
+}
+
+MEM_STATIC void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* statePtr)
+{
+    BIT_addBits(bitC, statePtr->value, statePtr->stateLog);
+    BIT_flushBits(bitC);
+}
+
+
+/* FSE_getMaxNbBits() :
+ * Approximate maximum cost of a symbol, in bits.
+ * Fractional get rounded up (i.e : a symbol with a normalized frequency of 3 gives the same result as a frequency of 2)
+ * note 1 : assume symbolValue is valid (<= maxSymbolValue)
+ * note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
+MEM_STATIC U32 FSE_getMaxNbBits(const void* symbolTTPtr, U32 symbolValue)
+{
+    const FSE_symbolCompressionTransform* symbolTT = (const FSE_symbolCompressionTransform*) symbolTTPtr;
+    return (symbolTT[symbolValue].deltaNbBits + ((1<<16)-1)) >> 16;
+}
+
+/* FSE_bitCost() :
+ * Approximate symbol cost, as fractional value, using fixed-point format (accuracyLog fractional bits)
+ * note 1 : assume symbolValue is valid (<= maxSymbolValue)
+ * note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
+MEM_STATIC U32 FSE_bitCost(const void* symbolTTPtr, U32 tableLog, U32 symbolValue, U32 accuracyLog)
+{
+    const FSE_symbolCompressionTransform* symbolTT = (const FSE_symbolCompressionTransform*) symbolTTPtr;
+    U32 const minNbBits = symbolTT[symbolValue].deltaNbBits >> 16;
+    U32 const threshold = (minNbBits+1) << 16;
+    assert(tableLog < 16);
+    assert(accuracyLog < 31-tableLog);  /* ensure enough room for renormalization double shift */
+    {   U32 const tableSize = 1 << tableLog;
+        U32 const deltaFromThreshold = threshold - (symbolTT[symbolValue].deltaNbBits + tableSize);
+        U32 const normalizedDeltaFromThreshold = (deltaFromThreshold << accuracyLog) >> tableLog;   /* linear interpolation (very approximate) */
+        U32 const bitMultiplier = 1 << accuracyLog;
+        assert(symbolTT[symbolValue].deltaNbBits + tableSize <= threshold);
+        assert(normalizedDeltaFromThreshold <= bitMultiplier);
+        return (minNbBits+1)*bitMultiplier - normalizedDeltaFromThreshold;
+    }
+}
+
+
+/* ======    Decompression    ====== */
+
+typedef struct {
+    U16 tableLog;
+    U16 fastMode;
+} FSE_DTableHeader;   /* sizeof U32 */
+
+typedef struct
+{
+    unsigned short newState;
+    unsigned char  symbol;
+    unsigned char  nbBits;
+} FSE_decode_t;   /* size == U32 */
+
+MEM_STATIC void FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt)
+{
+    const void* ptr = dt;
+    const FSE_DTableHeader* const DTableH = (const FSE_DTableHeader*)ptr;
+    DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
+    BIT_reloadDStream(bitD);
+    DStatePtr->table = dt + 1;
+}
+
+MEM_STATIC BYTE FSE_peekSymbol(const FSE_DState_t* DStatePtr)
+{
+    FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
+    return DInfo.symbol;
+}
+
+MEM_STATIC void FSE_updateState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
+{
+    FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
+    U32 const nbBits = DInfo.nbBits;
+    size_t const lowBits = BIT_readBits(bitD, nbBits);
+    DStatePtr->state = DInfo.newState + lowBits;
+}
+
+MEM_STATIC BYTE FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
+{
+    FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
+    U32 const nbBits = DInfo.nbBits;
+    BYTE const symbol = DInfo.symbol;
+    size_t const lowBits = BIT_readBits(bitD, nbBits);
+
+    DStatePtr->state = DInfo.newState + lowBits;
+    return symbol;
+}
+
+/*! FSE_decodeSymbolFast() :
+    unsafe, only works if no symbol has a probability > 50% */
+MEM_STATIC BYTE FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
+{
+    FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
+    U32 const nbBits = DInfo.nbBits;
+    BYTE const symbol = DInfo.symbol;
+    size_t const lowBits = BIT_readBitsFast(bitD, nbBits);
+
+    DStatePtr->state = DInfo.newState + lowBits;
+    return symbol;
+}
+
+MEM_STATIC unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr)
+{
+    return DStatePtr->state == 0;
+}
+
+
+
+#ifndef FSE_COMMONDEFS_ONLY
+
+/* **************************************************************
+*  Tuning parameters
+****************************************************************/
+/*!MEMORY_USAGE :
+*  Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; etc.)
+*  Increasing memory usage improves compression ratio
+*  Reduced memory usage can improve speed, due to cache effect
+*  Recommended max value is 14, for 16KB, which nicely fits into Intel x86 L1 cache */
+#ifndef FSE_MAX_MEMORY_USAGE
+#  define FSE_MAX_MEMORY_USAGE 14
+#endif
+#ifndef FSE_DEFAULT_MEMORY_USAGE
+#  define FSE_DEFAULT_MEMORY_USAGE 13
+#endif
+#if (FSE_DEFAULT_MEMORY_USAGE > FSE_MAX_MEMORY_USAGE)
+#  error "FSE_DEFAULT_MEMORY_USAGE must be <= FSE_MAX_MEMORY_USAGE"
+#endif
+
+/*!FSE_MAX_SYMBOL_VALUE :
+*  Maximum symbol value authorized.
+*  Required for proper stack allocation */
+#ifndef FSE_MAX_SYMBOL_VALUE
+#  define FSE_MAX_SYMBOL_VALUE 255
+#endif
+
+/* **************************************************************
+*  template functions type & suffix
+****************************************************************/
+#define FSE_FUNCTION_TYPE BYTE
+#define FSE_FUNCTION_EXTENSION
+#define FSE_DECODE_TYPE FSE_decode_t
+
+
+#endif   /* !FSE_COMMONDEFS_ONLY */
+
+
+/* ***************************************************************
+*  Constants
+*****************************************************************/
+#define FSE_MAX_TABLELOG  (FSE_MAX_MEMORY_USAGE-2)
+#define FSE_MAX_TABLESIZE (1U<<FSE_MAX_TABLELOG)
+#define FSE_MAXTABLESIZE_MASK (FSE_MAX_TABLESIZE-1)
+#define FSE_DEFAULT_TABLELOG (FSE_DEFAULT_MEMORY_USAGE-2)
+#define FSE_MIN_TABLELOG 5
+
+#define FSE_TABLELOG_ABSOLUTE_MAX 15
+#if FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX
+#  error "FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX is not supported"
+#endif
+
+#define FSE_TABLESTEP(tableSize) (((tableSize)>>1) + ((tableSize)>>3) + 3)
+
+
+#endif /* FSE_STATIC_LINKING_ONLY */
+
+
diff --git a/lib/zstd/common/fse_decompress.c b/lib/zstd/common/fse_decompress.c
new file mode 100644
index 0000000..2c8bbe3
--- /dev/null
+++ b/lib/zstd/common/fse_decompress.c
@@ -0,0 +1,390 @@
+/* ******************************************************************
+ * FSE : Finite State Entropy decoder
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ *  You can contact the author at :
+ *  - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+/* **************************************************************
+*  Includes
+****************************************************************/
+#include "debug.h"      /* assert */
+#include "bitstream.h"
+#include "compiler.h"
+#define FSE_STATIC_LINKING_ONLY
+#include "fse.h"
+#include "error_private.h"
+#define ZSTD_DEPS_NEED_MALLOC
+#include "zstd_deps.h"
+
+
+/* **************************************************************
+*  Error Management
+****************************************************************/
+#define FSE_isError ERR_isError
+#define FSE_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c)   /* use only *after* variable declarations */
+
+
+/* **************************************************************
+*  Templates
+****************************************************************/
+/*
+  designed to be included
+  for type-specific functions (template emulation in C)
+  Objective is to write these functions only once, for improved maintenance
+*/
+
+/* safety checks */
+#ifndef FSE_FUNCTION_EXTENSION
+#  error "FSE_FUNCTION_EXTENSION must be defined"
+#endif
+#ifndef FSE_FUNCTION_TYPE
+#  error "FSE_FUNCTION_TYPE must be defined"
+#endif
+
+/* Function names */
+#define FSE_CAT(X,Y) X##Y
+#define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
+#define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)
+
+
+/* Function templates */
+FSE_DTable* FSE_createDTable (unsigned tableLog)
+{
+    if (tableLog > FSE_TABLELOG_ABSOLUTE_MAX) tableLog = FSE_TABLELOG_ABSOLUTE_MAX;
+    return (FSE_DTable*)ZSTD_malloc( FSE_DTABLE_SIZE_U32(tableLog) * sizeof (U32) );
+}
+
+void FSE_freeDTable (FSE_DTable* dt)
+{
+    ZSTD_free(dt);
+}
+
+static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
+{
+    void* const tdPtr = dt+1;   /* because *dt is unsigned, 32-bits aligned on 32-bits */
+    FSE_DECODE_TYPE* const tableDecode = (FSE_DECODE_TYPE*) (tdPtr);
+    U16* symbolNext = (U16*)workSpace;
+    BYTE* spread = (BYTE*)(symbolNext + maxSymbolValue + 1);
+
+    U32 const maxSV1 = maxSymbolValue + 1;
+    U32 const tableSize = 1 << tableLog;
+    U32 highThreshold = tableSize-1;
+
+    /* Sanity Checks */
+    if (FSE_BUILD_DTABLE_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(maxSymbolValue_tooLarge);
+    if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE) return ERROR(maxSymbolValue_tooLarge);
+    if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);
+
+    /* Init, lay down lowprob symbols */
+    {   FSE_DTableHeader DTableH;
+        DTableH.tableLog = (U16)tableLog;
+        DTableH.fastMode = 1;
+        {   S16 const largeLimit= (S16)(1 << (tableLog-1));
+            U32 s;
+            for (s=0; s<maxSV1; s++) {
+                if (normalizedCounter[s]==-1) {
+                    tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s;
+                    symbolNext[s] = 1;
+                } else {
+                    if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
+                    symbolNext[s] = normalizedCounter[s];
+        }   }   }
+        ZSTD_memcpy(dt, &DTableH, sizeof(DTableH));
+    }
+
+    /* Spread symbols */
+    if (highThreshold == tableSize - 1) {
+        size_t const tableMask = tableSize-1;
+        size_t const step = FSE_TABLESTEP(tableSize);
+        /* First lay down the symbols in order.
+         * We use a uint64_t to lay down 8 bytes at a time. This reduces branch
+         * misses since small blocks generally have small table logs, so nearly
+         * all symbols have counts <= 8. We ensure we have 8 bytes at the end of
+         * our buffer to handle the over-write.
+         */
+        {
+            U64 const add = 0x0101010101010101ull;
+            size_t pos = 0;
+            U64 sv = 0;
+            U32 s;
+            for (s=0; s<maxSV1; ++s, sv += add) {
+                int i;
+                int const n = normalizedCounter[s];
+                MEM_write64(spread + pos, sv);
+                for (i = 8; i < n; i += 8) {
+                    MEM_write64(spread + pos + i, sv);
+                }
+                pos += n;
+            }
+        }
+        /* Now we spread those positions across the table.
+         * The benefit of doing it in two stages is that we avoid the the
+         * variable size inner loop, which caused lots of branch misses.
+         * Now we can run through all the positions without any branch misses.
+         * We unroll the loop twice, since that is what emperically worked best.
+         */
+        {
+            size_t position = 0;
+            size_t s;
+            size_t const unroll = 2;
+            assert(tableSize % unroll == 0); /* FSE_MIN_TABLELOG is 5 */
+            for (s = 0; s < (size_t)tableSize; s += unroll) {
+                size_t u;
+                for (u = 0; u < unroll; ++u) {
+                    size_t const uPosition = (position + (u * step)) & tableMask;
+                    tableDecode[uPosition].symbol = spread[s + u];
+                }
+                position = (position + (unroll * step)) & tableMask;
+            }
+            assert(position == 0);
+        }
+    } else {
+        U32 const tableMask = tableSize-1;
+        U32 const step = FSE_TABLESTEP(tableSize);
+        U32 s, position = 0;
+        for (s=0; s<maxSV1; s++) {
+            int i;
+            for (i=0; i<normalizedCounter[s]; i++) {
+                tableDecode[position].symbol = (FSE_FUNCTION_TYPE)s;
+                position = (position + step) & tableMask;
+                while (position > highThreshold) position = (position + step) & tableMask;   /* lowprob area */
+        }   }
+        if (position!=0) return ERROR(GENERIC);   /* position must reach all cells once, otherwise normalizedCounter is incorrect */
+    }
+
+    /* Build Decoding table */
+    {   U32 u;
+        for (u=0; u<tableSize; u++) {
+            FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
+            U32 const nextState = symbolNext[symbol]++;
+            tableDecode[u].nbBits = (BYTE) (tableLog - BIT_highbit32(nextState) );
+            tableDecode[u].newState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
+    }   }
+
+    return 0;
+}
+
+size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
+{
+    return FSE_buildDTable_internal(dt, normalizedCounter, maxSymbolValue, tableLog, workSpace, wkspSize);
+}
+
+
+#ifndef FSE_COMMONDEFS_ONLY
+
+/*-*******************************************************
+*  Decompression (Byte symbols)
+*********************************************************/
+size_t FSE_buildDTable_rle (FSE_DTable* dt, BYTE symbolValue)
+{
+    void* ptr = dt;
+    FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
+    void* dPtr = dt + 1;
+    FSE_decode_t* const cell = (FSE_decode_t*)dPtr;
+
+    DTableH->tableLog = 0;
+    DTableH->fastMode = 0;
+
+    cell->newState = 0;
+    cell->symbol = symbolValue;
+    cell->nbBits = 0;
+
+    return 0;
+}
+
+
+size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits)
+{
+    void* ptr = dt;
+    FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
+    void* dPtr = dt + 1;
+    FSE_decode_t* const dinfo = (FSE_decode_t*)dPtr;
+    const unsigned tableSize = 1 << nbBits;
+    const unsigned tableMask = tableSize - 1;
+    const unsigned maxSV1 = tableMask+1;
+    unsigned s;
+
+    /* Sanity checks */
+    if (nbBits < 1) return ERROR(GENERIC);         /* min size */
+
+    /* Build Decoding Table */
+    DTableH->tableLog = (U16)nbBits;
+    DTableH->fastMode = 1;
+    for (s=0; s<maxSV1; s++) {
+        dinfo[s].newState = 0;
+        dinfo[s].symbol = (BYTE)s;
+        dinfo[s].nbBits = (BYTE)nbBits;
+    }
+
+    return 0;
+}
+
+FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
+          void* dst, size_t maxDstSize,
+    const void* cSrc, size_t cSrcSize,
+    const FSE_DTable* dt, const unsigned fast)
+{
+    BYTE* const ostart = (BYTE*) dst;
+    BYTE* op = ostart;
+    BYTE* const omax = op + maxDstSize;
+    BYTE* const olimit = omax-3;
+
+    BIT_DStream_t bitD;
+    FSE_DState_t state1;
+    FSE_DState_t state2;
+
+    /* Init */
+    CHECK_F(BIT_initDStream(&bitD, cSrc, cSrcSize));
+
+    FSE_initDState(&state1, &bitD, dt);
+    FSE_initDState(&state2, &bitD, dt);
+
+#define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD)
+
+    /* 4 symbols per loop */
+    for ( ; (BIT_reloadDStream(&bitD)==BIT_DStream_unfinished) & (op<olimit) ; op+=4) {
+        op[0] = FSE_GETSYMBOL(&state1);
+
+        if (FSE_MAX_TABLELOG*2+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
+            BIT_reloadDStream(&bitD);
+
+        op[1] = FSE_GETSYMBOL(&state2);
+
+        if (FSE_MAX_TABLELOG*4+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
+            { if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) { op+=2; break; } }
+
+        op[2] = FSE_GETSYMBOL(&state1);
+
+        if (FSE_MAX_TABLELOG*2+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
+            BIT_reloadDStream(&bitD);
+
+        op[3] = FSE_GETSYMBOL(&state2);
+    }
+
+    /* tail */
+    /* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */
+    while (1) {
+        if (op>(omax-2)) return ERROR(dstSize_tooSmall);
+        *op++ = FSE_GETSYMBOL(&state1);
+        if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
+            *op++ = FSE_GETSYMBOL(&state2);
+            break;
+        }
+
+        if (op>(omax-2)) return ERROR(dstSize_tooSmall);
+        *op++ = FSE_GETSYMBOL(&state2);
+        if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
+            *op++ = FSE_GETSYMBOL(&state1);
+            break;
+    }   }
+
+    return op-ostart;
+}
+
+
+size_t FSE_decompress_usingDTable(void* dst, size_t originalSize,
+                            const void* cSrc, size_t cSrcSize,
+                            const FSE_DTable* dt)
+{
+    const void* ptr = dt;
+    const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
+    const U32 fastMode = DTableH->fastMode;
+
+    /* select fast mode (static) */
+    if (fastMode) return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1);
+    return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0);
+}
+
+
+size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
+{
+    return FSE_decompress_wksp_bmi2(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, /* bmi2 */ 0);
+}
+
+typedef struct {
+    short ncount[FSE_MAX_SYMBOL_VALUE + 1];
+    FSE_DTable dtable[1]; /* Dynamically sized */
+} FSE_DecompressWksp;
+
+
+FORCE_INLINE_TEMPLATE size_t FSE_decompress_wksp_body(
+        void* dst, size_t dstCapacity,
+        const void* cSrc, size_t cSrcSize,
+        unsigned maxLog, void* workSpace, size_t wkspSize,
+        int bmi2)
+{
+    const BYTE* const istart = (const BYTE*)cSrc;
+    const BYTE* ip = istart;
+    unsigned tableLog;
+    unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
+    FSE_DecompressWksp* const wksp = (FSE_DecompressWksp*)workSpace;
+
+    DEBUG_STATIC_ASSERT((FSE_MAX_SYMBOL_VALUE + 1) % 2 == 0);
+    if (wkspSize < sizeof(*wksp)) return ERROR(GENERIC);
+
+    /* normal FSE decoding mode */
+    {
+        size_t const NCountLength = FSE_readNCount_bmi2(wksp->ncount, &maxSymbolValue, &tableLog, istart, cSrcSize, bmi2);
+        if (FSE_isError(NCountLength)) return NCountLength;
+        if (tableLog > maxLog) return ERROR(tableLog_tooLarge);
+        assert(NCountLength <= cSrcSize);
+        ip += NCountLength;
+        cSrcSize -= NCountLength;
+    }
+
+    if (FSE_DECOMPRESS_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(tableLog_tooLarge);
+    workSpace = wksp->dtable + FSE_DTABLE_SIZE_U32(tableLog);
+    wkspSize -= sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);
+
+    CHECK_F( FSE_buildDTable_internal(wksp->dtable, wksp->ncount, maxSymbolValue, tableLog, workSpace, wkspSize) );
+
+    {
+        const void* ptr = wksp->dtable;
+        const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
+        const U32 fastMode = DTableH->fastMode;
+
+        /* select fast mode (static) */
+        if (fastMode) return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, wksp->dtable, 1);
+        return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, wksp->dtable, 0);
+    }
+}
+
+/* Avoids the FORCE_INLINE of the _body() function. */
+static size_t FSE_decompress_wksp_body_default(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
+{
+    return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 0);
+}
+
+#if DYNAMIC_BMI2
+TARGET_ATTRIBUTE("bmi2") static size_t FSE_decompress_wksp_body_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
+{
+    return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 1);
+}
+#endif
+
+size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize, int bmi2)
+{
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        return FSE_decompress_wksp_body_bmi2(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
+    }
+#endif
+    (void)bmi2;
+    return FSE_decompress_wksp_body_default(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
+}
+
+
+typedef FSE_DTable DTable_max_t[FSE_DTABLE_SIZE_U32(FSE_MAX_TABLELOG)];
+
+
+
+#endif   /* FSE_COMMONDEFS_ONLY */
diff --git a/lib/zstd/common/huf.h b/lib/zstd/common/huf.h
new file mode 100644
index 0000000..88c5586
--- /dev/null
+++ b/lib/zstd/common/huf.h
@@ -0,0 +1,356 @@
+/* ******************************************************************
+ * huff0 huffman codec,
+ * part of Finite State Entropy library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+
+#ifndef HUF_H_298734234
+#define HUF_H_298734234
+
+/* *** Dependencies *** */
+#include "zstd_deps.h"    /* size_t */
+
+
+/* *** library symbols visibility *** */
+/* Note : when linking with -fvisibility=hidden on gcc, or by default on Visual,
+ *        HUF symbols remain "private" (internal symbols for library only).
+ *        Set macro FSE_DLL_EXPORT to 1 if you want HUF symbols visible on DLL interface */
+#if defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) && defined(__GNUC__) && (__GNUC__ >= 4)
+#  define HUF_PUBLIC_API __attribute__ ((visibility ("default")))
+#elif defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1)   /* Visual expected */
+#  define HUF_PUBLIC_API __declspec(dllexport)
+#elif defined(FSE_DLL_IMPORT) && (FSE_DLL_IMPORT==1)
+#  define HUF_PUBLIC_API __declspec(dllimport)  /* not required, just to generate faster code (saves a function pointer load from IAT and an indirect jump) */
+#else
+#  define HUF_PUBLIC_API
+#endif
+
+
+/* ========================== */
+/* ***  simple functions  *** */
+/* ========================== */
+
+/* HUF_compress() :
+ *  Compress content from buffer 'src', of size 'srcSize', into buffer 'dst'.
+ * 'dst' buffer must be already allocated.
+ *  Compression runs faster if `dstCapacity` >= HUF_compressBound(srcSize).
+ * `srcSize` must be <= `HUF_BLOCKSIZE_MAX` == 128 KB.
+ * @return : size of compressed data (<= `dstCapacity`).
+ *  Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
+ *                   if HUF_isError(return), compression failed (more details using HUF_getErrorName())
+ */
+HUF_PUBLIC_API size_t HUF_compress(void* dst, size_t dstCapacity,
+                             const void* src, size_t srcSize);
+
+/* HUF_decompress() :
+ *  Decompress HUF data from buffer 'cSrc', of size 'cSrcSize',
+ *  into already allocated buffer 'dst', of minimum size 'dstSize'.
+ * `originalSize` : **must** be the ***exact*** size of original (uncompressed) data.
+ *  Note : in contrast with FSE, HUF_decompress can regenerate
+ *         RLE (cSrcSize==1) and uncompressed (cSrcSize==dstSize) data,
+ *         because it knows size to regenerate (originalSize).
+ * @return : size of regenerated data (== originalSize),
+ *           or an error code, which can be tested using HUF_isError()
+ */
+HUF_PUBLIC_API size_t HUF_decompress(void* dst,  size_t originalSize,
+                               const void* cSrc, size_t cSrcSize);
+
+
+/* ***   Tool functions *** */
+#define HUF_BLOCKSIZE_MAX (128 * 1024)                  /*< maximum input size for a single block compressed with HUF_compress */
+HUF_PUBLIC_API size_t HUF_compressBound(size_t size);   /*< maximum compressed size (worst case) */
+
+/* Error Management */
+HUF_PUBLIC_API unsigned    HUF_isError(size_t code);       /*< tells if a return value is an error code */
+HUF_PUBLIC_API const char* HUF_getErrorName(size_t code);  /*< provides error code string (useful for debugging) */
+
+
+/* ***   Advanced function   *** */
+
+/* HUF_compress2() :
+ *  Same as HUF_compress(), but offers control over `maxSymbolValue` and `tableLog`.
+ * `maxSymbolValue` must be <= HUF_SYMBOLVALUE_MAX .
+ * `tableLog` must be `<= HUF_TABLELOG_MAX` . */
+HUF_PUBLIC_API size_t HUF_compress2 (void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                               unsigned maxSymbolValue, unsigned tableLog);
+
+/* HUF_compress4X_wksp() :
+ *  Same as HUF_compress2(), but uses externally allocated `workSpace`.
+ * `workspace` must have minimum alignment of 4, and be at least as large as HUF_WORKSPACE_SIZE */
+#define HUF_WORKSPACE_SIZE ((6 << 10) + 256)
+#define HUF_WORKSPACE_SIZE_U32 (HUF_WORKSPACE_SIZE / sizeof(U32))
+HUF_PUBLIC_API size_t HUF_compress4X_wksp (void* dst, size_t dstCapacity,
+                                     const void* src, size_t srcSize,
+                                     unsigned maxSymbolValue, unsigned tableLog,
+                                     void* workSpace, size_t wkspSize);
+
+#endif   /* HUF_H_298734234 */
+
+/* ******************************************************************
+ *  WARNING !!
+ *  The following section contains advanced and experimental definitions
+ *  which shall never be used in the context of a dynamic library,
+ *  because they are not guaranteed to remain stable in the future.
+ *  Only consider them in association with static linking.
+ * *****************************************************************/
+#if !defined(HUF_H_HUF_STATIC_LINKING_ONLY)
+#define HUF_H_HUF_STATIC_LINKING_ONLY
+
+/* *** Dependencies *** */
+#include "mem.h"   /* U32 */
+#define FSE_STATIC_LINKING_ONLY
+#include "fse.h"
+
+
+/* *** Constants *** */
+#define HUF_TABLELOG_MAX      12      /* max runtime value of tableLog (due to static allocation); can be modified up to HUF_ABSOLUTEMAX_TABLELOG */
+#define HUF_TABLELOG_DEFAULT  11      /* default tableLog value when none specified */
+#define HUF_SYMBOLVALUE_MAX  255
+
+#define HUF_TABLELOG_ABSOLUTEMAX  15  /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */
+#if (HUF_TABLELOG_MAX > HUF_TABLELOG_ABSOLUTEMAX)
+#  error "HUF_TABLELOG_MAX is too large !"
+#endif
+
+
+/* ****************************************
+*  Static allocation
+******************************************/
+/* HUF buffer bounds */
+#define HUF_CTABLEBOUND 129
+#define HUF_BLOCKBOUND(size) (size + (size>>8) + 8)   /* only true when incompressible is pre-filtered with fast heuristic */
+#define HUF_COMPRESSBOUND(size) (HUF_CTABLEBOUND + HUF_BLOCKBOUND(size))   /* Macro version, useful for static allocation */
+
+/* static allocation of HUF's Compression Table */
+/* this is a private definition, just exposed for allocation and strict aliasing purpose. never EVER access its members directly */
+struct HUF_CElt_s {
+  U16  val;
+  BYTE nbBits;
+};   /* typedef'd to HUF_CElt */
+typedef struct HUF_CElt_s HUF_CElt;   /* consider it an incomplete type */
+#define HUF_CTABLE_SIZE_U32(maxSymbolValue)   ((maxSymbolValue)+1)   /* Use tables of U32, for proper alignment */
+#define HUF_CTABLE_SIZE(maxSymbolValue)       (HUF_CTABLE_SIZE_U32(maxSymbolValue) * sizeof(U32))
+#define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \
+    HUF_CElt name[HUF_CTABLE_SIZE_U32(maxSymbolValue)] /* no final ; */
+
+/* static allocation of HUF's DTable */
+typedef U32 HUF_DTable;
+#define HUF_DTABLE_SIZE(maxTableLog)   (1 + (1<<(maxTableLog)))
+#define HUF_CREATE_STATIC_DTABLEX1(DTable, maxTableLog) \
+        HUF_DTable DTable[HUF_DTABLE_SIZE((maxTableLog)-1)] = { ((U32)((maxTableLog)-1) * 0x01000001) }
+#define HUF_CREATE_STATIC_DTABLEX2(DTable, maxTableLog) \
+        HUF_DTable DTable[HUF_DTABLE_SIZE(maxTableLog)] = { ((U32)(maxTableLog) * 0x01000001) }
+
+
+/* ****************************************
+*  Advanced decompression functions
+******************************************/
+size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< double-symbols decoder */
+#endif
+
+size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< decodes RLE and uncompressed */
+size_t HUF_decompress4X_hufOnly(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< considers RLE and uncompressed as errors */
+size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /*< considers RLE and uncompressed as errors */
+size_t HUF_decompress4X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< single-symbol decoder */
+size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /*< single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< double-symbols decoder */
+size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /*< double-symbols decoder */
+#endif
+
+
+/* ****************************************
+ *  HUF detailed API
+ * ****************************************/
+
+/*! HUF_compress() does the following:
+ *  1. count symbol occurrence from source[] into table count[] using FSE_count() (exposed within "fse.h")
+ *  2. (optional) refine tableLog using HUF_optimalTableLog()
+ *  3. build Huffman table from count using HUF_buildCTable()
+ *  4. save Huffman table to memory buffer using HUF_writeCTable()
+ *  5. encode the data stream using HUF_compress4X_usingCTable()
+ *
+ *  The following API allows targeting specific sub-functions for advanced tasks.
+ *  For example, it's possible to compress several blocks using the same 'CTable',
+ *  or to save and regenerate 'CTable' using external methods.
+ */
+unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
+size_t HUF_buildCTable (HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits);   /* @return : maxNbBits; CTable and count can overlap. In which case, CTable will overwrite count content */
+size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog);
+size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog, void* workspace, size_t workspaceSize);
+size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
+size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
+int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
+
+typedef enum {
+   HUF_repeat_none,  /*< Cannot use the previous table */
+   HUF_repeat_check, /*< Can use the previous table but it must be checked. Note : The previous table must have been constructed by HUF_compress{1, 4}X_repeat */
+   HUF_repeat_valid  /*< Can use the previous table and it is assumed to be valid */
+ } HUF_repeat;
+/* HUF_compress4X_repeat() :
+ *  Same as HUF_compress4X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
+ *  If it uses hufTable it does not modify hufTable or repeat.
+ *  If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
+ *  If preferRepeat then the old table will always be used if valid. */
+size_t HUF_compress4X_repeat(void* dst, size_t dstSize,
+                       const void* src, size_t srcSize,
+                       unsigned maxSymbolValue, unsigned tableLog,
+                       void* workSpace, size_t wkspSize,    /*< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
+                       HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);
+
+/* HUF_buildCTable_wksp() :
+ *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
+ * `workSpace` must be aligned on 4-bytes boundaries, and its size must be >= HUF_CTABLE_WORKSPACE_SIZE.
+ */
+#define HUF_CTABLE_WORKSPACE_SIZE_U32 (2*HUF_SYMBOLVALUE_MAX +1 +1)
+#define HUF_CTABLE_WORKSPACE_SIZE (HUF_CTABLE_WORKSPACE_SIZE_U32 * sizeof(unsigned))
+size_t HUF_buildCTable_wksp (HUF_CElt* tree,
+                       const unsigned* count, U32 maxSymbolValue, U32 maxNbBits,
+                             void* workSpace, size_t wkspSize);
+
+/*! HUF_readStats() :
+ *  Read compact Huffman tree, saved by HUF_writeCTable().
+ * `huffWeight` is destination buffer.
+ * @return : size read from `src` , or an error Code .
+ *  Note : Needed by HUF_readCTable() and HUF_readDTableXn() . */
+size_t HUF_readStats(BYTE* huffWeight, size_t hwSize,
+                     U32* rankStats, U32* nbSymbolsPtr, U32* tableLogPtr,
+                     const void* src, size_t srcSize);
+
+/*! HUF_readStats_wksp() :
+ * Same as HUF_readStats() but takes an external workspace which must be
+ * 4-byte aligned and its size must be >= HUF_READ_STATS_WORKSPACE_SIZE.
+ * If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
+ */
+#define HUF_READ_STATS_WORKSPACE_SIZE_U32 FSE_DECOMPRESS_WKSP_SIZE_U32(6, HUF_TABLELOG_MAX-1)
+#define HUF_READ_STATS_WORKSPACE_SIZE (HUF_READ_STATS_WORKSPACE_SIZE_U32 * sizeof(unsigned))
+size_t HUF_readStats_wksp(BYTE* huffWeight, size_t hwSize,
+                          U32* rankStats, U32* nbSymbolsPtr, U32* tableLogPtr,
+                          const void* src, size_t srcSize,
+                          void* workspace, size_t wkspSize,
+                          int bmi2);
+
+/* HUF_readCTable() :
+ *  Loading a CTable saved with HUF_writeCTable() */
+size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize, unsigned *hasZeroWeights);
+
+/* HUF_getNbBits() :
+ *  Read nbBits from CTable symbolTable, for symbol `symbolValue` presumed <= HUF_SYMBOLVALUE_MAX
+ *  Note 1 : is not inlined, as HUF_CElt definition is private
+ *  Note 2 : const void* used, so that it can provide a statically allocated table as argument (which uses type U32) */
+U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue);
+
+/*
+ * HUF_decompress() does the following:
+ * 1. select the decompression algorithm (X1, X2) based on pre-computed heuristics
+ * 2. build Huffman table from save, using HUF_readDTableX?()
+ * 3. decode 1 or 4 segments in parallel using HUF_decompress?X?_usingDTable()
+ */
+
+/* HUF_selectDecoder() :
+ *  Tells which decoder is likely to decode faster,
+ *  based on a set of pre-computed metrics.
+ * @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 .
+ *  Assumption : 0 < dstSize <= 128 KB */
+U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize);
+
+/*
+ *  The minimum workspace size for the `workSpace` used in
+ *  HUF_readDTableX1_wksp() and HUF_readDTableX2_wksp().
+ *
+ *  The space used depends on HUF_TABLELOG_MAX, ranging from ~1500 bytes when
+ *  HUF_TABLE_LOG_MAX=12 to ~1850 bytes when HUF_TABLE_LOG_MAX=15.
+ *  Buffer overflow errors may potentially occur if code modifications result in
+ *  a required workspace size greater than that specified in the following
+ *  macro.
+ */
+#define HUF_DECOMPRESS_WORKSPACE_SIZE ((2 << 10) + (1 << 9))
+#define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))
+
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_readDTableX1 (HUF_DTable* DTable, const void* src, size_t srcSize);
+size_t HUF_readDTableX1_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_readDTableX2 (HUF_DTable* DTable, const void* src, size_t srcSize);
+size_t HUF_readDTableX2_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
+#endif
+
+size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress4X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress4X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+
+
+/* ====================== */
+/* single stream variants */
+/* ====================== */
+
+size_t HUF_compress1X (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
+size_t HUF_compress1X_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);  /*< `workSpace` must be a table of at least HUF_WORKSPACE_SIZE_U32 unsigned */
+size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
+/* HUF_compress1X_repeat() :
+ *  Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
+ *  If it uses hufTable it does not modify hufTable or repeat.
+ *  If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
+ *  If preferRepeat then the old table will always be used if valid. */
+size_t HUF_compress1X_repeat(void* dst, size_t dstSize,
+                       const void* src, size_t srcSize,
+                       unsigned maxSymbolValue, unsigned tableLog,
+                       void* workSpace, size_t wkspSize,   /*< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
+                       HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);
+
+size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /* single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /* double-symbol decoder */
+#endif
+
+size_t HUF_decompress1X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
+size_t HUF_decompress1X_DCtx_wksp (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress1X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< single-symbol decoder */
+size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /*< single-symbol decoder */
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress1X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /*< double-symbols decoder */
+size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /*< double-symbols decoder */
+#endif
+
+size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);   /*< automatic selection of sing or double symbol decoder, based on DTable */
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress1X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
+size_t HUF_decompress1X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+
+/* BMI2 variants.
+ * If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
+ */
+size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
+#endif
+size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
+size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_readDTableX1_wksp_bmi2(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int bmi2);
+#endif
+
+#endif /* HUF_STATIC_LINKING_ONLY */
+
diff --git a/lib/zstd/common/mem.h b/lib/zstd/common/mem.h
new file mode 100644
index 0000000..dcdd586
--- /dev/null
+++ b/lib/zstd/common/mem.h
@@ -0,0 +1,259 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef MEM_H_MODULE
+#define MEM_H_MODULE
+
+/*-****************************************
+*  Dependencies
+******************************************/
+#include <asm/unaligned.h>  /* get_unaligned, put_unaligned* */
+#include <linux/compiler.h>  /* inline */
+#include <linux/swab.h>  /* swab32, swab64 */
+#include <linux/types.h>  /* size_t, ptrdiff_t */
+#include "debug.h"  /* DEBUG_STATIC_ASSERT */
+
+/*-****************************************
+*  Compiler specifics
+******************************************/
+#define MEM_STATIC static inline
+
+/*-**************************************************************
+*  Basic Types
+*****************************************************************/
+typedef uint8_t  BYTE;
+typedef uint16_t U16;
+typedef int16_t  S16;
+typedef uint32_t U32;
+typedef int32_t  S32;
+typedef uint64_t U64;
+typedef int64_t  S64;
+
+/*-**************************************************************
+*  Memory I/O API
+*****************************************************************/
+/*=== Static platform detection ===*/
+MEM_STATIC unsigned MEM_32bits(void);
+MEM_STATIC unsigned MEM_64bits(void);
+MEM_STATIC unsigned MEM_isLittleEndian(void);
+
+/*=== Native unaligned read/write ===*/
+MEM_STATIC U16 MEM_read16(const void* memPtr);
+MEM_STATIC U32 MEM_read32(const void* memPtr);
+MEM_STATIC U64 MEM_read64(const void* memPtr);
+MEM_STATIC size_t MEM_readST(const void* memPtr);
+
+MEM_STATIC void MEM_write16(void* memPtr, U16 value);
+MEM_STATIC void MEM_write32(void* memPtr, U32 value);
+MEM_STATIC void MEM_write64(void* memPtr, U64 value);
+
+/*=== Little endian unaligned read/write ===*/
+MEM_STATIC U16 MEM_readLE16(const void* memPtr);
+MEM_STATIC U32 MEM_readLE24(const void* memPtr);
+MEM_STATIC U32 MEM_readLE32(const void* memPtr);
+MEM_STATIC U64 MEM_readLE64(const void* memPtr);
+MEM_STATIC size_t MEM_readLEST(const void* memPtr);
+
+MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val);
+MEM_STATIC void MEM_writeLE24(void* memPtr, U32 val);
+MEM_STATIC void MEM_writeLE32(void* memPtr, U32 val32);
+MEM_STATIC void MEM_writeLE64(void* memPtr, U64 val64);
+MEM_STATIC void MEM_writeLEST(void* memPtr, size_t val);
+
+/*=== Big endian unaligned read/write ===*/
+MEM_STATIC U32 MEM_readBE32(const void* memPtr);
+MEM_STATIC U64 MEM_readBE64(const void* memPtr);
+MEM_STATIC size_t MEM_readBEST(const void* memPtr);
+
+MEM_STATIC void MEM_writeBE32(void* memPtr, U32 val32);
+MEM_STATIC void MEM_writeBE64(void* memPtr, U64 val64);
+MEM_STATIC void MEM_writeBEST(void* memPtr, size_t val);
+
+/*=== Byteswap ===*/
+MEM_STATIC U32 MEM_swap32(U32 in);
+MEM_STATIC U64 MEM_swap64(U64 in);
+MEM_STATIC size_t MEM_swapST(size_t in);
+
+/*-**************************************************************
+*  Memory I/O Implementation
+*****************************************************************/
+MEM_STATIC unsigned MEM_32bits(void)
+{
+    return sizeof(size_t) == 4;
+}
+
+MEM_STATIC unsigned MEM_64bits(void)
+{
+    return sizeof(size_t) == 8;
+}
+
+#if defined(__LITTLE_ENDIAN)
+#define MEM_LITTLE_ENDIAN 1
+#else
+#define MEM_LITTLE_ENDIAN 0
+#endif
+
+MEM_STATIC unsigned MEM_isLittleEndian(void)
+{
+    return MEM_LITTLE_ENDIAN;
+}
+
+MEM_STATIC U16 MEM_read16(const void *memPtr)
+{
+    return get_unaligned((const U16 *)memPtr);
+}
+
+MEM_STATIC U32 MEM_read32(const void *memPtr)
+{
+    return get_unaligned((const U32 *)memPtr);
+}
+
+MEM_STATIC U64 MEM_read64(const void *memPtr)
+{
+    return get_unaligned((const U64 *)memPtr);
+}
+
+MEM_STATIC size_t MEM_readST(const void *memPtr)
+{
+    return get_unaligned((const size_t *)memPtr);
+}
+
+MEM_STATIC void MEM_write16(void *memPtr, U16 value)
+{
+    put_unaligned(value, (U16 *)memPtr);
+}
+
+MEM_STATIC void MEM_write32(void *memPtr, U32 value)
+{
+    put_unaligned(value, (U32 *)memPtr);
+}
+
+MEM_STATIC void MEM_write64(void *memPtr, U64 value)
+{
+    put_unaligned(value, (U64 *)memPtr);
+}
+
+/*=== Little endian r/w ===*/
+
+MEM_STATIC U16 MEM_readLE16(const void *memPtr)
+{
+    return get_unaligned_le16(memPtr);
+}
+
+MEM_STATIC void MEM_writeLE16(void *memPtr, U16 val)
+{
+    put_unaligned_le16(val, memPtr);
+}
+
+MEM_STATIC U32 MEM_readLE24(const void *memPtr)
+{
+    return MEM_readLE16(memPtr) + (((const BYTE *)memPtr)[2] << 16);
+}
+
+MEM_STATIC void MEM_writeLE24(void *memPtr, U32 val)
+{
+	MEM_writeLE16(memPtr, (U16)val);
+	((BYTE *)memPtr)[2] = (BYTE)(val >> 16);
+}
+
+MEM_STATIC U32 MEM_readLE32(const void *memPtr)
+{
+    return get_unaligned_le32(memPtr);
+}
+
+MEM_STATIC void MEM_writeLE32(void *memPtr, U32 val32)
+{
+    put_unaligned_le32(val32, memPtr);
+}
+
+MEM_STATIC U64 MEM_readLE64(const void *memPtr)
+{
+    return get_unaligned_le64(memPtr);
+}
+
+MEM_STATIC void MEM_writeLE64(void *memPtr, U64 val64)
+{
+    put_unaligned_le64(val64, memPtr);
+}
+
+MEM_STATIC size_t MEM_readLEST(const void *memPtr)
+{
+	if (MEM_32bits())
+		return (size_t)MEM_readLE32(memPtr);
+	else
+		return (size_t)MEM_readLE64(memPtr);
+}
+
+MEM_STATIC void MEM_writeLEST(void *memPtr, size_t val)
+{
+	if (MEM_32bits())
+		MEM_writeLE32(memPtr, (U32)val);
+	else
+		MEM_writeLE64(memPtr, (U64)val);
+}
+
+/*=== Big endian r/w ===*/
+
+MEM_STATIC U32 MEM_readBE32(const void *memPtr)
+{
+    return get_unaligned_be32(memPtr);
+}
+
+MEM_STATIC void MEM_writeBE32(void *memPtr, U32 val32)
+{
+    put_unaligned_be32(val32, memPtr);
+}
+
+MEM_STATIC U64 MEM_readBE64(const void *memPtr)
+{
+    return get_unaligned_be64(memPtr);
+}
+
+MEM_STATIC void MEM_writeBE64(void *memPtr, U64 val64)
+{
+    put_unaligned_be64(val64, memPtr);
+}
+
+MEM_STATIC size_t MEM_readBEST(const void *memPtr)
+{
+	if (MEM_32bits())
+		return (size_t)MEM_readBE32(memPtr);
+	else
+		return (size_t)MEM_readBE64(memPtr);
+}
+
+MEM_STATIC void MEM_writeBEST(void *memPtr, size_t val)
+{
+	if (MEM_32bits())
+		MEM_writeBE32(memPtr, (U32)val);
+	else
+		MEM_writeBE64(memPtr, (U64)val);
+}
+
+MEM_STATIC U32 MEM_swap32(U32 in)
+{
+    return swab32(in);
+}
+
+MEM_STATIC U64 MEM_swap64(U64 in)
+{
+    return swab64(in);
+}
+
+MEM_STATIC size_t MEM_swapST(size_t in)
+{
+    if (MEM_32bits())
+        return (size_t)MEM_swap32((U32)in);
+    else
+        return (size_t)MEM_swap64((U64)in);
+}
+
+#endif /* MEM_H_MODULE */
diff --git a/lib/zstd/common/zstd_common.c b/lib/zstd/common/zstd_common.c
new file mode 100644
index 0000000..3d7e35b
--- /dev/null
+++ b/lib/zstd/common/zstd_common.c
@@ -0,0 +1,83 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#define ZSTD_DEPS_NEED_MALLOC
+#include "zstd_deps.h"   /* ZSTD_malloc, ZSTD_calloc, ZSTD_free, ZSTD_memset */
+#include "error_private.h"
+#include "zstd_internal.h"
+
+
+/*-****************************************
+*  Version
+******************************************/
+unsigned ZSTD_versionNumber(void) { return ZSTD_VERSION_NUMBER; }
+
+const char* ZSTD_versionString(void) { return ZSTD_VERSION_STRING; }
+
+
+/*-****************************************
+*  ZSTD Error Management
+******************************************/
+#undef ZSTD_isError   /* defined within zstd_internal.h */
+/*! ZSTD_isError() :
+ *  tells if a return value is an error code
+ *  symbol is required for external callers */
+unsigned ZSTD_isError(size_t code) { return ERR_isError(code); }
+
+/*! ZSTD_getErrorName() :
+ *  provides error code string from function result (useful for debugging) */
+const char* ZSTD_getErrorName(size_t code) { return ERR_getErrorName(code); }
+
+/*! ZSTD_getError() :
+ *  convert a `size_t` function result into a proper ZSTD_errorCode enum */
+ZSTD_ErrorCode ZSTD_getErrorCode(size_t code) { return ERR_getErrorCode(code); }
+
+/*! ZSTD_getErrorString() :
+ *  provides error code string from enum */
+const char* ZSTD_getErrorString(ZSTD_ErrorCode code) { return ERR_getErrorString(code); }
+
+
+
+/*=**************************************************************
+*  Custom allocator
+****************************************************************/
+void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem)
+{
+    if (customMem.customAlloc)
+        return customMem.customAlloc(customMem.opaque, size);
+    return ZSTD_malloc(size);
+}
+
+void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem)
+{
+    if (customMem.customAlloc) {
+        /* calloc implemented as malloc+memset;
+         * not as efficient as calloc, but next best guess for custom malloc */
+        void* const ptr = customMem.customAlloc(customMem.opaque, size);
+        ZSTD_memset(ptr, 0, size);
+        return ptr;
+    }
+    return ZSTD_calloc(1, size);
+}
+
+void ZSTD_customFree(void* ptr, ZSTD_customMem customMem)
+{
+    if (ptr!=NULL) {
+        if (customMem.customFree)
+            customMem.customFree(customMem.opaque, ptr);
+        else
+            ZSTD_free(ptr);
+    }
+}
diff --git a/lib/zstd/common/zstd_deps.h b/lib/zstd/common/zstd_deps.h
new file mode 100644
index 0000000..7a5bf44
--- /dev/null
+++ b/lib/zstd/common/zstd_deps.h
@@ -0,0 +1,125 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/*
+ * This file provides common libc dependencies that zstd requires.
+ * The purpose is to allow replacing this file with a custom implementation
+ * to compile zstd without libc support.
+ */
+
+/* Need:
+ * NULL
+ * INT_MAX
+ * UINT_MAX
+ * ZSTD_memcpy()
+ * ZSTD_memset()
+ * ZSTD_memmove()
+ */
+#ifndef ZSTD_DEPS_COMMON
+#define ZSTD_DEPS_COMMON
+
+#include <linux/limits.h>
+#include <linux/stddef.h>
+
+#define ZSTD_memcpy(d,s,n) __builtin_memcpy((d),(s),(n))
+#define ZSTD_memmove(d,s,n) __builtin_memmove((d),(s),(n))
+#define ZSTD_memset(d,s,n) __builtin_memset((d),(s),(n))
+
+#endif /* ZSTD_DEPS_COMMON */
+
+/*
+ * Define malloc as always failing. That means the user must
+ * either use ZSTD_customMem or statically allocate memory.
+ * Need:
+ * ZSTD_malloc()
+ * ZSTD_free()
+ * ZSTD_calloc()
+ */
+#ifdef ZSTD_DEPS_NEED_MALLOC
+#ifndef ZSTD_DEPS_MALLOC
+#define ZSTD_DEPS_MALLOC
+
+#define ZSTD_malloc(s) ({ (void)(s); NULL; })
+#define ZSTD_free(p) ((void)(p))
+#define ZSTD_calloc(n,s) ({ (void)(n); (void)(s); NULL; })
+
+#endif /* ZSTD_DEPS_MALLOC */
+#endif /* ZSTD_DEPS_NEED_MALLOC */
+
+/*
+ * Provides 64-bit math support.
+ * Need:
+ * U64 ZSTD_div64(U64 dividend, U32 divisor)
+ */
+#ifdef ZSTD_DEPS_NEED_MATH64
+#ifndef ZSTD_DEPS_MATH64
+#define ZSTD_DEPS_MATH64
+
+#include <linux/math64.h>
+
+static uint64_t ZSTD_div64(uint64_t dividend, uint32_t divisor) {
+  return div_u64(dividend, divisor);
+}
+
+#endif /* ZSTD_DEPS_MATH64 */
+#endif /* ZSTD_DEPS_NEED_MATH64 */
+
+/*
+ * This is only requested when DEBUGLEVEL >= 1, meaning
+ * it is disabled in production.
+ * Need:
+ * assert()
+ */
+#ifdef ZSTD_DEPS_NEED_ASSERT
+#ifndef ZSTD_DEPS_ASSERT
+#define ZSTD_DEPS_ASSERT
+
+#include <linux/kernel.h>
+
+#define assert(x) WARN_ON((x))
+
+#endif /* ZSTD_DEPS_ASSERT */
+#endif /* ZSTD_DEPS_NEED_ASSERT */
+
+/*
+ * This is only requested when DEBUGLEVEL >= 2, meaning
+ * it is disabled in production.
+ * Need:
+ * ZSTD_DEBUG_PRINT()
+ */
+#ifdef ZSTD_DEPS_NEED_IO
+#ifndef ZSTD_DEPS_IO
+#define ZSTD_DEPS_IO
+
+#include <linux/printk.h>
+
+#define ZSTD_DEBUG_PRINT(...) pr_debug(__VA_ARGS__)
+
+#endif /* ZSTD_DEPS_IO */
+#endif /* ZSTD_DEPS_NEED_IO */
+
+/*
+ * Only requested when MSAN is enabled.
+ * Need:
+ * intptr_t
+ */
+#ifdef ZSTD_DEPS_NEED_STDINT
+#ifndef ZSTD_DEPS_STDINT
+#define ZSTD_DEPS_STDINT
+
+/*
+ * The Linux Kernel doesn't provide intptr_t, only uintptr_t, which
+ * is an unsigned long.
+ */
+typedef long intptr_t;
+
+#endif /* ZSTD_DEPS_STDINT */
+#endif /* ZSTD_DEPS_NEED_STDINT */
diff --git a/lib/zstd/common/zstd_internal.h b/lib/zstd/common/zstd_internal.h
new file mode 100644
index 0000000..fc6f3a9
--- /dev/null
+++ b/lib/zstd/common/zstd_internal.h
@@ -0,0 +1,450 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_CCOMMON_H_MODULE
+#define ZSTD_CCOMMON_H_MODULE
+
+/* this module contains definitions which must be identical
+ * across compression, decompression and dictBuilder.
+ * It also contains a few functions useful to at least 2 of them
+ * and which benefit from being inlined */
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#include "compiler.h"
+#include "mem.h"
+#include "debug.h"                 /* assert, DEBUGLOG, RAWLOG, g_debuglevel */
+#include "error_private.h"
+#define ZSTD_STATIC_LINKING_ONLY
+#include <linux/zstd.h>
+#define FSE_STATIC_LINKING_ONLY
+#include "fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "huf.h"
+#include <linux/xxhash.h>                /* XXH_reset, update, digest */
+#define ZSTD_TRACE 0
+
+
+/* ---- static assert (debug) --- */
+#define ZSTD_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c)
+#define ZSTD_isError ERR_isError   /* for inlining */
+#define FSE_isError  ERR_isError
+#define HUF_isError  ERR_isError
+
+
+/*-*************************************
+*  shared macros
+***************************************/
+#undef MIN
+#undef MAX
+#define MIN(a,b) ((a)<(b) ? (a) : (b))
+#define MAX(a,b) ((a)>(b) ? (a) : (b))
+
+/*
+ * Ignore: this is an internal helper.
+ *
+ * This is a helper function to help force C99-correctness during compilation.
+ * Under strict compilation modes, variadic macro arguments can't be empty.
+ * However, variadic function arguments can be. Using a function therefore lets
+ * us statically check that at least one (string) argument was passed,
+ * independent of the compilation flags.
+ */
+static INLINE_KEYWORD UNUSED_ATTR
+void _force_has_format_string(const char *format, ...) {
+  (void)format;
+}
+
+/*
+ * Ignore: this is an internal helper.
+ *
+ * We want to force this function invocation to be syntactically correct, but
+ * we don't want to force runtime evaluation of its arguments.
+ */
+#define _FORCE_HAS_FORMAT_STRING(...) \
+  if (0) { \
+    _force_has_format_string(__VA_ARGS__); \
+  }
+
+/*
+ * Return the specified error if the condition evaluates to true.
+ *
+ * In debug modes, prints additional information.
+ * In order to do that (particularly, printing the conditional that failed),
+ * this can't just wrap RETURN_ERROR().
+ */
+#define RETURN_ERROR_IF(cond, err, ...) \
+  if (cond) { \
+    RAWLOG(3, "%s:%d: ERROR!: check %s failed, returning %s", \
+           __FILE__, __LINE__, ZSTD_QUOTE(cond), ZSTD_QUOTE(ERROR(err))); \
+    _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
+    RAWLOG(3, ": " __VA_ARGS__); \
+    RAWLOG(3, "\n"); \
+    return ERROR(err); \
+  }
+
+/*
+ * Unconditionally return the specified error.
+ *
+ * In debug modes, prints additional information.
+ */
+#define RETURN_ERROR(err, ...) \
+  do { \
+    RAWLOG(3, "%s:%d: ERROR!: unconditional check failed, returning %s", \
+           __FILE__, __LINE__, ZSTD_QUOTE(ERROR(err))); \
+    _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
+    RAWLOG(3, ": " __VA_ARGS__); \
+    RAWLOG(3, "\n"); \
+    return ERROR(err); \
+  } while(0);
+
+/*
+ * If the provided expression evaluates to an error code, returns that error code.
+ *
+ * In debug modes, prints additional information.
+ */
+#define FORWARD_IF_ERROR(err, ...) \
+  do { \
+    size_t const err_code = (err); \
+    if (ERR_isError(err_code)) { \
+      RAWLOG(3, "%s:%d: ERROR!: forwarding error in %s: %s", \
+             __FILE__, __LINE__, ZSTD_QUOTE(err), ERR_getErrorName(err_code)); \
+      _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
+      RAWLOG(3, ": " __VA_ARGS__); \
+      RAWLOG(3, "\n"); \
+      return err_code; \
+    } \
+  } while(0);
+
+
+/*-*************************************
+*  Common constants
+***************************************/
+#define ZSTD_OPT_NUM    (1<<12)
+
+#define ZSTD_REP_NUM      3                 /* number of repcodes */
+#define ZSTD_REP_MOVE     (ZSTD_REP_NUM-1)
+static UNUSED_ATTR const U32 repStartValue[ZSTD_REP_NUM] = { 1, 4, 8 };
+
+#define KB *(1 <<10)
+#define MB *(1 <<20)
+#define GB *(1U<<30)
+
+#define BIT7 128
+#define BIT6  64
+#define BIT5  32
+#define BIT4  16
+#define BIT1   2
+#define BIT0   1
+
+#define ZSTD_WINDOWLOG_ABSOLUTEMIN 10
+static UNUSED_ATTR const size_t ZSTD_fcs_fieldSize[4] = { 0, 2, 4, 8 };
+static UNUSED_ATTR const size_t ZSTD_did_fieldSize[4] = { 0, 1, 2, 4 };
+
+#define ZSTD_FRAMEIDSIZE 4   /* magic number size */
+
+#define ZSTD_BLOCKHEADERSIZE 3   /* C standard doesn't allow `static const` variable to be init using another `static const` variable */
+static UNUSED_ATTR const size_t ZSTD_blockHeaderSize = ZSTD_BLOCKHEADERSIZE;
+typedef enum { bt_raw, bt_rle, bt_compressed, bt_reserved } blockType_e;
+
+#define ZSTD_FRAMECHECKSUMSIZE 4
+
+#define MIN_SEQUENCES_SIZE 1 /* nbSeq==0 */
+#define MIN_CBLOCK_SIZE (1 /*litCSize*/ + 1 /* RLE or RAW */ + MIN_SEQUENCES_SIZE /* nbSeq==0 */)   /* for a non-null block */
+
+#define HufLog 12
+typedef enum { set_basic, set_rle, set_compressed, set_repeat } symbolEncodingType_e;
+
+#define LONGNBSEQ 0x7F00
+
+#define MINMATCH 3
+
+#define Litbits  8
+#define MaxLit ((1<<Litbits) - 1)
+#define MaxML   52
+#define MaxLL   35
+#define DefaultMaxOff 28
+#define MaxOff  31
+#define MaxSeq MAX(MaxLL, MaxML)   /* Assumption : MaxOff < MaxLL,MaxML */
+#define MLFSELog    9
+#define LLFSELog    9
+#define OffFSELog   8
+#define MaxFSELog  MAX(MAX(MLFSELog, LLFSELog), OffFSELog)
+
+#define ZSTD_MAX_HUF_HEADER_SIZE 128 /* header + <= 127 byte tree description */
+/* Each table cannot take more than #symbols * FSELog bits */
+#define ZSTD_MAX_FSE_HEADERS_SIZE (((MaxML + 1) * MLFSELog + (MaxLL + 1) * LLFSELog + (MaxOff + 1) * OffFSELog + 7) / 8)
+
+static UNUSED_ATTR const U32 LL_bits[MaxLL+1] = {
+     0, 0, 0, 0, 0, 0, 0, 0,
+     0, 0, 0, 0, 0, 0, 0, 0,
+     1, 1, 1, 1, 2, 2, 3, 3,
+     4, 6, 7, 8, 9,10,11,12,
+    13,14,15,16
+};
+static UNUSED_ATTR const S16 LL_defaultNorm[MaxLL+1] = {
+     4, 3, 2, 2, 2, 2, 2, 2,
+     2, 2, 2, 2, 2, 1, 1, 1,
+     2, 2, 2, 2, 2, 2, 2, 2,
+     2, 3, 2, 1, 1, 1, 1, 1,
+    -1,-1,-1,-1
+};
+#define LL_DEFAULTNORMLOG 6  /* for static allocation */
+static UNUSED_ATTR const U32 LL_defaultNormLog = LL_DEFAULTNORMLOG;
+
+static UNUSED_ATTR const U32 ML_bits[MaxML+1] = {
+     0, 0, 0, 0, 0, 0, 0, 0,
+     0, 0, 0, 0, 0, 0, 0, 0,
+     0, 0, 0, 0, 0, 0, 0, 0,
+     0, 0, 0, 0, 0, 0, 0, 0,
+     1, 1, 1, 1, 2, 2, 3, 3,
+     4, 4, 5, 7, 8, 9,10,11,
+    12,13,14,15,16
+};
+static UNUSED_ATTR const S16 ML_defaultNorm[MaxML+1] = {
+     1, 4, 3, 2, 2, 2, 2, 2,
+     2, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1,-1,-1,
+    -1,-1,-1,-1,-1
+};
+#define ML_DEFAULTNORMLOG 6  /* for static allocation */
+static UNUSED_ATTR const U32 ML_defaultNormLog = ML_DEFAULTNORMLOG;
+
+static UNUSED_ATTR const S16 OF_defaultNorm[DefaultMaxOff+1] = {
+     1, 1, 1, 1, 1, 1, 2, 2,
+     2, 1, 1, 1, 1, 1, 1, 1,
+     1, 1, 1, 1, 1, 1, 1, 1,
+    -1,-1,-1,-1,-1
+};
+#define OF_DEFAULTNORMLOG 5  /* for static allocation */
+static UNUSED_ATTR const U32 OF_defaultNormLog = OF_DEFAULTNORMLOG;
+
+
+/*-*******************************************
+*  Shared functions to include for inlining
+*********************************************/
+static void ZSTD_copy8(void* dst, const void* src) {
+    ZSTD_memcpy(dst, src, 8);
+}
+
+#define COPY8(d,s) { ZSTD_copy8(d,s); d+=8; s+=8; }
+static void ZSTD_copy16(void* dst, const void* src) {
+    ZSTD_memcpy(dst, src, 16);
+}
+#define COPY16(d,s) { ZSTD_copy16(d,s); d+=16; s+=16; }
+
+#define WILDCOPY_OVERLENGTH 32
+#define WILDCOPY_VECLEN 16
+
+typedef enum {
+    ZSTD_no_overlap,
+    ZSTD_overlap_src_before_dst
+    /*  ZSTD_overlap_dst_before_src, */
+} ZSTD_overlap_e;
+
+/*! ZSTD_wildcopy() :
+ *  Custom version of ZSTD_memcpy(), can over read/write up to WILDCOPY_OVERLENGTH bytes (if length==0)
+ *  @param ovtype controls the overlap detection
+ *         - ZSTD_no_overlap: The source and destination are guaranteed to be at least WILDCOPY_VECLEN bytes apart.
+ *         - ZSTD_overlap_src_before_dst: The src and dst may overlap, but they MUST be at least 8 bytes apart.
+ *           The src buffer must be before the dst buffer.
+ */
+MEM_STATIC FORCE_INLINE_ATTR
+void ZSTD_wildcopy(void* dst, const void* src, ptrdiff_t length, ZSTD_overlap_e const ovtype)
+{
+    ptrdiff_t diff = (BYTE*)dst - (const BYTE*)src;
+    const BYTE* ip = (const BYTE*)src;
+    BYTE* op = (BYTE*)dst;
+    BYTE* const oend = op + length;
+
+    assert(diff >= 8 || (ovtype == ZSTD_no_overlap && diff <= -WILDCOPY_VECLEN));
+
+    if (ovtype == ZSTD_overlap_src_before_dst && diff < WILDCOPY_VECLEN) {
+        /* Handle short offset copies. */
+        do {
+            COPY8(op, ip)
+        } while (op < oend);
+    } else {
+        assert(diff >= WILDCOPY_VECLEN || diff <= -WILDCOPY_VECLEN);
+        /* Separate out the first COPY16() call because the copy length is
+         * almost certain to be short, so the branches have different
+         * probabilities. Since it is almost certain to be short, only do
+         * one COPY16() in the first call. Then, do two calls per loop since
+         * at that point it is more likely to have a high trip count.
+         */
+#ifdef __aarch64__
+        do {
+            COPY16(op, ip);
+        }
+        while (op < oend);
+#else
+        ZSTD_copy16(op, ip);
+        if (16 >= length) return;
+        op += 16;
+        ip += 16;
+        do {
+            COPY16(op, ip);
+            COPY16(op, ip);
+        }
+        while (op < oend);
+#endif
+    }
+}
+
+MEM_STATIC size_t ZSTD_limitCopy(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    size_t const length = MIN(dstCapacity, srcSize);
+    if (length > 0) {
+        ZSTD_memcpy(dst, src, length);
+    }
+    return length;
+}
+
+/* define "workspace is too large" as this number of times larger than needed */
+#define ZSTD_WORKSPACETOOLARGE_FACTOR 3
+
+/* when workspace is continuously too large
+ * during at least this number of times,
+ * context's memory usage is considered wasteful,
+ * because it's sized to handle a worst case scenario which rarely happens.
+ * In which case, resize it down to free some memory */
+#define ZSTD_WORKSPACETOOLARGE_MAXDURATION 128
+
+/* Controls whether the input/output buffer is buffered or stable. */
+typedef enum {
+    ZSTD_bm_buffered = 0,  /* Buffer the input/output */
+    ZSTD_bm_stable = 1     /* ZSTD_inBuffer/ZSTD_outBuffer is stable */
+} ZSTD_bufferMode_e;
+
+
+/*-*******************************************
+*  Private declarations
+*********************************************/
+typedef struct seqDef_s {
+    U32 offset;         /* Offset code of the sequence */
+    U16 litLength;
+    U16 matchLength;
+} seqDef;
+
+typedef struct {
+    seqDef* sequencesStart;
+    seqDef* sequences;      /* ptr to end of sequences */
+    BYTE* litStart;
+    BYTE* lit;              /* ptr to end of literals */
+    BYTE* llCode;
+    BYTE* mlCode;
+    BYTE* ofCode;
+    size_t maxNbSeq;
+    size_t maxNbLit;
+
+    /* longLengthPos and longLengthID to allow us to represent either a single litLength or matchLength
+     * in the seqStore that has a value larger than U16 (if it exists). To do so, we increment
+     * the existing value of the litLength or matchLength by 0x10000.
+     */
+    U32   longLengthID;   /* 0 == no longLength; 1 == Represent the long literal; 2 == Represent the long match; */
+    U32   longLengthPos;  /* Index of the sequence to apply long length modification to */
+} seqStore_t;
+
+typedef struct {
+    U32 litLength;
+    U32 matchLength;
+} ZSTD_sequenceLength;
+
+/*
+ * Returns the ZSTD_sequenceLength for the given sequences. It handles the decoding of long sequences
+ * indicated by longLengthPos and longLengthID, and adds MINMATCH back to matchLength.
+ */
+MEM_STATIC ZSTD_sequenceLength ZSTD_getSequenceLength(seqStore_t const* seqStore, seqDef const* seq)
+{
+    ZSTD_sequenceLength seqLen;
+    seqLen.litLength = seq->litLength;
+    seqLen.matchLength = seq->matchLength + MINMATCH;
+    if (seqStore->longLengthPos == (U32)(seq - seqStore->sequencesStart)) {
+        if (seqStore->longLengthID == 1) {
+            seqLen.litLength += 0xFFFF;
+        }
+        if (seqStore->longLengthID == 2) {
+            seqLen.matchLength += 0xFFFF;
+        }
+    }
+    return seqLen;
+}
+
+/*
+ * Contains the compressed frame size and an upper-bound for the decompressed frame size.
+ * Note: before using `compressedSize`, check for errors using ZSTD_isError().
+ *       similarly, before using `decompressedBound`, check for errors using:
+ *          `decompressedBound != ZSTD_CONTENTSIZE_ERROR`
+ */
+typedef struct {
+    size_t compressedSize;
+    unsigned long long decompressedBound;
+} ZSTD_frameSizeInfo;   /* decompress & legacy */
+
+const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx);   /* compress & dictBuilder */
+void ZSTD_seqToCodes(const seqStore_t* seqStorePtr);   /* compress, dictBuilder, decodeCorpus (shouldn't get its definition from here) */
+
+/* custom memory allocation functions */
+void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem);
+void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem);
+void ZSTD_customFree(void* ptr, ZSTD_customMem customMem);
+
+
+MEM_STATIC U32 ZSTD_highbit32(U32 val)   /* compress, dictBuilder, decodeCorpus */
+{
+    assert(val != 0);
+    {
+#   if (__GNUC__ >= 3)   /* GCC Intrinsic */
+        return __builtin_clz (val) ^ 31;
+#   else   /* Software version */
+        static const U32 DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 };
+        U32 v = val;
+        v |= v >> 1;
+        v |= v >> 2;
+        v |= v >> 4;
+        v |= v >> 8;
+        v |= v >> 16;
+        return DeBruijnClz[(v * 0x07C4ACDDU) >> 27];
+#   endif
+    }
+}
+
+
+/* ZSTD_invalidateRepCodes() :
+ * ensures next compression will not use repcodes from previous block.
+ * Note : only works with regular variant;
+ *        do not use with extDict variant ! */
+void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx);   /* zstdmt, adaptive_compression (shouldn't get this definition from here) */
+
+
+typedef struct {
+    blockType_e blockType;
+    U32 lastBlock;
+    U32 origSize;
+} blockProperties_t;   /* declared here for decompress and fullbench */
+
+/*! ZSTD_getcBlockSize() :
+ *  Provides the size of compressed block from block header `src` */
+/* Used by: decompress, fullbench (does not get its definition from here) */
+size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
+                          blockProperties_t* bpPtr);
+
+/*! ZSTD_decodeSeqHeaders() :
+ *  decode sequence header from src */
+/* Used by: decompress, fullbench (does not get its definition from here) */
+size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
+                       const void* src, size_t srcSize);
+
+
+
+#endif   /* ZSTD_CCOMMON_H_MODULE */
diff --git a/lib/zstd/compress.c b/lib/zstd/compress.c
deleted file mode 100644
index b080264..0000000
--- a/lib/zstd/compress.c
+++ /dev/null
@@ -1,3485 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-/*-*************************************
-*  Dependencies
-***************************************/
-#include "fse.h"
-#include "huf.h"
-#include "mem.h"
-#include "zstd_internal.h" /* includes zstd.h */
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/string.h> /* memset */
-
-/*-*************************************
-*  Constants
-***************************************/
-static const U32 g_searchStrength = 8; /* control skip over incompressible data */
-#define HASH_READ_SIZE 8
-typedef enum { ZSTDcs_created = 0, ZSTDcs_init, ZSTDcs_ongoing, ZSTDcs_ending } ZSTD_compressionStage_e;
-
-/*-*************************************
-*  Helper functions
-***************************************/
-size_t ZSTD_compressBound(size_t srcSize) { return FSE_compressBound(srcSize) + 12; }
-
-/*-*************************************
-*  Sequence storage
-***************************************/
-static void ZSTD_resetSeqStore(seqStore_t *ssPtr)
-{
-	ssPtr->lit = ssPtr->litStart;
-	ssPtr->sequences = ssPtr->sequencesStart;
-	ssPtr->longLengthID = 0;
-}
-
-/*-*************************************
-*  Context memory management
-***************************************/
-struct ZSTD_CCtx_s {
-	const BYTE *nextSrc;  /* next block here to continue on curr prefix */
-	const BYTE *base;     /* All regular indexes relative to this position */
-	const BYTE *dictBase; /* extDict indexes relative to this position */
-	U32 dictLimit;	/* below that point, need extDict */
-	U32 lowLimit;	 /* below that point, no more data */
-	U32 nextToUpdate;     /* index from which to continue dictionary update */
-	U32 nextToUpdate3;    /* index from which to continue dictionary update */
-	U32 hashLog3;	 /* dispatch table : larger == faster, more memory */
-	U32 loadedDictEnd;    /* index of end of dictionary */
-	U32 forceWindow;      /* force back-references to respect limit of 1<<wLog, even for dictionary */
-	U32 forceRawDict;     /* Force loading dictionary in "content-only" mode (no header analysis) */
-	ZSTD_compressionStage_e stage;
-	U32 rep[ZSTD_REP_NUM];
-	U32 repToConfirm[ZSTD_REP_NUM];
-	U32 dictID;
-	ZSTD_parameters params;
-	void *workSpace;
-	size_t workSpaceSize;
-	size_t blockSize;
-	U64 frameContentSize;
-	struct xxh64_state xxhState;
-	ZSTD_customMem customMem;
-
-	seqStore_t seqStore; /* sequences storage ptrs */
-	U32 *hashTable;
-	U32 *hashTable3;
-	U32 *chainTable;
-	HUF_CElt *hufTable;
-	U32 flagStaticTables;
-	HUF_repeat flagStaticHufTable;
-	FSE_CTable offcodeCTable[FSE_CTABLE_SIZE_U32(OffFSELog, MaxOff)];
-	FSE_CTable matchlengthCTable[FSE_CTABLE_SIZE_U32(MLFSELog, MaxML)];
-	FSE_CTable litlengthCTable[FSE_CTABLE_SIZE_U32(LLFSELog, MaxLL)];
-	unsigned tmpCounters[HUF_COMPRESS_WORKSPACE_SIZE_U32];
-};
-
-size_t ZSTD_CCtxWorkspaceBound(ZSTD_compressionParameters cParams)
-{
-	size_t const blockSize = MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, (size_t)1 << cParams.windowLog);
-	U32 const divider = (cParams.searchLength == 3) ? 3 : 4;
-	size_t const maxNbSeq = blockSize / divider;
-	size_t const tokenSpace = blockSize + 11 * maxNbSeq;
-	size_t const chainSize = (cParams.strategy == ZSTD_fast) ? 0 : (1 << cParams.chainLog);
-	size_t const hSize = ((size_t)1) << cParams.hashLog;
-	U32 const hashLog3 = (cParams.searchLength > 3) ? 0 : MIN(ZSTD_HASHLOG3_MAX, cParams.windowLog);
-	size_t const h3Size = ((size_t)1) << hashLog3;
-	size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
-	size_t const optSpace =
-	    ((MaxML + 1) + (MaxLL + 1) + (MaxOff + 1) + (1 << Litbits)) * sizeof(U32) + (ZSTD_OPT_NUM + 1) * (sizeof(ZSTD_match_t) + sizeof(ZSTD_optimal_t));
-	size_t const workspaceSize = tableSpace + (256 * sizeof(U32)) /* huffTable */ + tokenSpace +
-				     (((cParams.strategy == ZSTD_btopt) || (cParams.strategy == ZSTD_btopt2)) ? optSpace : 0);
-
-	return ZSTD_ALIGN(sizeof(ZSTD_stack)) + ZSTD_ALIGN(sizeof(ZSTD_CCtx)) + ZSTD_ALIGN(workspaceSize);
-}
-
-static ZSTD_CCtx *ZSTD_createCCtx_advanced(ZSTD_customMem customMem)
-{
-	ZSTD_CCtx *cctx;
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-	cctx = (ZSTD_CCtx *)ZSTD_malloc(sizeof(ZSTD_CCtx), customMem);
-	if (!cctx)
-		return NULL;
-	memset(cctx, 0, sizeof(ZSTD_CCtx));
-	cctx->customMem = customMem;
-	return cctx;
-}
-
-ZSTD_CCtx *ZSTD_initCCtx(void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	ZSTD_CCtx *cctx = ZSTD_createCCtx_advanced(stackMem);
-	if (cctx) {
-		cctx->workSpace = ZSTD_stackAllocAll(cctx->customMem.opaque, &cctx->workSpaceSize);
-	}
-	return cctx;
-}
-
-size_t ZSTD_freeCCtx(ZSTD_CCtx *cctx)
-{
-	if (cctx == NULL)
-		return 0; /* support free on NULL */
-	ZSTD_free(cctx->workSpace, cctx->customMem);
-	ZSTD_free(cctx, cctx->customMem);
-	return 0; /* reserved as a potential error code in the future */
-}
-
-const seqStore_t *ZSTD_getSeqStore(const ZSTD_CCtx *ctx) /* hidden interface */ { return &(ctx->seqStore); }
-
-static ZSTD_parameters ZSTD_getParamsFromCCtx(const ZSTD_CCtx *cctx) { return cctx->params; }
-
-/** ZSTD_checkParams() :
-	ensure param values remain within authorized range.
-	@return : 0, or an error code if one value is beyond authorized range */
-size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams)
-{
-#define CLAMPCHECK(val, min, max)                                       \
-	{                                                               \
-		if ((val < min) | (val > max))                          \
-			return ERROR(compressionParameter_unsupported); \
-	}
-	CLAMPCHECK(cParams.windowLog, ZSTD_WINDOWLOG_MIN, ZSTD_WINDOWLOG_MAX);
-	CLAMPCHECK(cParams.chainLog, ZSTD_CHAINLOG_MIN, ZSTD_CHAINLOG_MAX);
-	CLAMPCHECK(cParams.hashLog, ZSTD_HASHLOG_MIN, ZSTD_HASHLOG_MAX);
-	CLAMPCHECK(cParams.searchLog, ZSTD_SEARCHLOG_MIN, ZSTD_SEARCHLOG_MAX);
-	CLAMPCHECK(cParams.searchLength, ZSTD_SEARCHLENGTH_MIN, ZSTD_SEARCHLENGTH_MAX);
-	CLAMPCHECK(cParams.targetLength, ZSTD_TARGETLENGTH_MIN, ZSTD_TARGETLENGTH_MAX);
-	if ((U32)(cParams.strategy) > (U32)ZSTD_btopt2)
-		return ERROR(compressionParameter_unsupported);
-	return 0;
-}
-
-/** ZSTD_cycleLog() :
- *  condition for correct operation : hashLog > 1 */
-static U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat)
-{
-	U32 const btScale = ((U32)strat >= (U32)ZSTD_btlazy2);
-	return hashLog - btScale;
-}
-
-/** ZSTD_adjustCParams() :
-	optimize `cPar` for a given input (`srcSize` and `dictSize`).
-	mostly downsizing to reduce memory consumption and initialization.
-	Both `srcSize` and `dictSize` are optional (use 0 if unknown),
-	but if both are 0, no optimization can be done.
-	Note : cPar is considered validated at this stage. Use ZSTD_checkParams() to ensure that. */
-ZSTD_compressionParameters ZSTD_adjustCParams(ZSTD_compressionParameters cPar, unsigned long long srcSize, size_t dictSize)
-{
-	if (srcSize + dictSize == 0)
-		return cPar; /* no size information available : no adjustment */
-
-	/* resize params, to use less memory when necessary */
-	{
-		U32 const minSrcSize = (srcSize == 0) ? 500 : 0;
-		U64 const rSize = srcSize + dictSize + minSrcSize;
-		if (rSize < ((U64)1 << ZSTD_WINDOWLOG_MAX)) {
-			U32 const srcLog = MAX(ZSTD_HASHLOG_MIN, ZSTD_highbit32((U32)(rSize)-1) + 1);
-			if (cPar.windowLog > srcLog)
-				cPar.windowLog = srcLog;
-		}
-	}
-	if (cPar.hashLog > cPar.windowLog)
-		cPar.hashLog = cPar.windowLog;
-	{
-		U32 const cycleLog = ZSTD_cycleLog(cPar.chainLog, cPar.strategy);
-		if (cycleLog > cPar.windowLog)
-			cPar.chainLog -= (cycleLog - cPar.windowLog);
-	}
-
-	if (cPar.windowLog < ZSTD_WINDOWLOG_ABSOLUTEMIN)
-		cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN; /* required for frame header */
-
-	return cPar;
-}
-
-static U32 ZSTD_equivalentParams(ZSTD_parameters param1, ZSTD_parameters param2)
-{
-	return (param1.cParams.hashLog == param2.cParams.hashLog) & (param1.cParams.chainLog == param2.cParams.chainLog) &
-	       (param1.cParams.strategy == param2.cParams.strategy) & ((param1.cParams.searchLength == 3) == (param2.cParams.searchLength == 3));
-}
-
-/*! ZSTD_continueCCtx() :
-	reuse CCtx without reset (note : requires no dictionary) */
-static size_t ZSTD_continueCCtx(ZSTD_CCtx *cctx, ZSTD_parameters params, U64 frameContentSize)
-{
-	U32 const end = (U32)(cctx->nextSrc - cctx->base);
-	cctx->params = params;
-	cctx->frameContentSize = frameContentSize;
-	cctx->lowLimit = end;
-	cctx->dictLimit = end;
-	cctx->nextToUpdate = end + 1;
-	cctx->stage = ZSTDcs_init;
-	cctx->dictID = 0;
-	cctx->loadedDictEnd = 0;
-	{
-		int i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			cctx->rep[i] = repStartValue[i];
-	}
-	cctx->seqStore.litLengthSum = 0; /* force reset of btopt stats */
-	xxh64_reset(&cctx->xxhState, 0);
-	return 0;
-}
-
-typedef enum { ZSTDcrp_continue, ZSTDcrp_noMemset, ZSTDcrp_fullReset } ZSTD_compResetPolicy_e;
-
-/*! ZSTD_resetCCtx_advanced() :
-	note : `params` must be validated */
-static size_t ZSTD_resetCCtx_advanced(ZSTD_CCtx *zc, ZSTD_parameters params, U64 frameContentSize, ZSTD_compResetPolicy_e const crp)
-{
-	if (crp == ZSTDcrp_continue)
-		if (ZSTD_equivalentParams(params, zc->params)) {
-			zc->flagStaticTables = 0;
-			zc->flagStaticHufTable = HUF_repeat_none;
-			return ZSTD_continueCCtx(zc, params, frameContentSize);
-		}
-
-	{
-		size_t const blockSize = MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, (size_t)1 << params.cParams.windowLog);
-		U32 const divider = (params.cParams.searchLength == 3) ? 3 : 4;
-		size_t const maxNbSeq = blockSize / divider;
-		size_t const tokenSpace = blockSize + 11 * maxNbSeq;
-		size_t const chainSize = (params.cParams.strategy == ZSTD_fast) ? 0 : (1 << params.cParams.chainLog);
-		size_t const hSize = ((size_t)1) << params.cParams.hashLog;
-		U32 const hashLog3 = (params.cParams.searchLength > 3) ? 0 : MIN(ZSTD_HASHLOG3_MAX, params.cParams.windowLog);
-		size_t const h3Size = ((size_t)1) << hashLog3;
-		size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
-		void *ptr;
-
-		/* Check if workSpace is large enough, alloc a new one if needed */
-		{
-			size_t const optSpace = ((MaxML + 1) + (MaxLL + 1) + (MaxOff + 1) + (1 << Litbits)) * sizeof(U32) +
-						(ZSTD_OPT_NUM + 1) * (sizeof(ZSTD_match_t) + sizeof(ZSTD_optimal_t));
-			size_t const neededSpace = tableSpace + (256 * sizeof(U32)) /* huffTable */ + tokenSpace +
-						   (((params.cParams.strategy == ZSTD_btopt) || (params.cParams.strategy == ZSTD_btopt2)) ? optSpace : 0);
-			if (zc->workSpaceSize < neededSpace) {
-				ZSTD_free(zc->workSpace, zc->customMem);
-				zc->workSpace = ZSTD_malloc(neededSpace, zc->customMem);
-				if (zc->workSpace == NULL)
-					return ERROR(memory_allocation);
-				zc->workSpaceSize = neededSpace;
-			}
-		}
-
-		if (crp != ZSTDcrp_noMemset)
-			memset(zc->workSpace, 0, tableSpace); /* reset tables only */
-		xxh64_reset(&zc->xxhState, 0);
-		zc->hashLog3 = hashLog3;
-		zc->hashTable = (U32 *)(zc->workSpace);
-		zc->chainTable = zc->hashTable + hSize;
-		zc->hashTable3 = zc->chainTable + chainSize;
-		ptr = zc->hashTable3 + h3Size;
-		zc->hufTable = (HUF_CElt *)ptr;
-		zc->flagStaticTables = 0;
-		zc->flagStaticHufTable = HUF_repeat_none;
-		ptr = ((U32 *)ptr) + 256; /* note : HUF_CElt* is incomplete type, size is simulated using U32 */
-
-		zc->nextToUpdate = 1;
-		zc->nextSrc = NULL;
-		zc->base = NULL;
-		zc->dictBase = NULL;
-		zc->dictLimit = 0;
-		zc->lowLimit = 0;
-		zc->params = params;
-		zc->blockSize = blockSize;
-		zc->frameContentSize = frameContentSize;
-		{
-			int i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				zc->rep[i] = repStartValue[i];
-		}
-
-		if ((params.cParams.strategy == ZSTD_btopt) || (params.cParams.strategy == ZSTD_btopt2)) {
-			zc->seqStore.litFreq = (U32 *)ptr;
-			zc->seqStore.litLengthFreq = zc->seqStore.litFreq + (1 << Litbits);
-			zc->seqStore.matchLengthFreq = zc->seqStore.litLengthFreq + (MaxLL + 1);
-			zc->seqStore.offCodeFreq = zc->seqStore.matchLengthFreq + (MaxML + 1);
-			ptr = zc->seqStore.offCodeFreq + (MaxOff + 1);
-			zc->seqStore.matchTable = (ZSTD_match_t *)ptr;
-			ptr = zc->seqStore.matchTable + ZSTD_OPT_NUM + 1;
-			zc->seqStore.priceTable = (ZSTD_optimal_t *)ptr;
-			ptr = zc->seqStore.priceTable + ZSTD_OPT_NUM + 1;
-			zc->seqStore.litLengthSum = 0;
-		}
-		zc->seqStore.sequencesStart = (seqDef *)ptr;
-		ptr = zc->seqStore.sequencesStart + maxNbSeq;
-		zc->seqStore.llCode = (BYTE *)ptr;
-		zc->seqStore.mlCode = zc->seqStore.llCode + maxNbSeq;
-		zc->seqStore.ofCode = zc->seqStore.mlCode + maxNbSeq;
-		zc->seqStore.litStart = zc->seqStore.ofCode + maxNbSeq;
-
-		zc->stage = ZSTDcs_init;
-		zc->dictID = 0;
-		zc->loadedDictEnd = 0;
-
-		return 0;
-	}
-}
-
-/* ZSTD_invalidateRepCodes() :
- * ensures next compression will not use repcodes from previous block.
- * Note : only works with regular variant;
- *        do not use with extDict variant ! */
-void ZSTD_invalidateRepCodes(ZSTD_CCtx *cctx)
-{
-	int i;
-	for (i = 0; i < ZSTD_REP_NUM; i++)
-		cctx->rep[i] = 0;
-}
-
-/*! ZSTD_copyCCtx() :
-*   Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
-*   Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
-*   @return : 0, or an error code */
-size_t ZSTD_copyCCtx(ZSTD_CCtx *dstCCtx, const ZSTD_CCtx *srcCCtx, unsigned long long pledgedSrcSize)
-{
-	if (srcCCtx->stage != ZSTDcs_init)
-		return ERROR(stage_wrong);
-
-	memcpy(&dstCCtx->customMem, &srcCCtx->customMem, sizeof(ZSTD_customMem));
-	{
-		ZSTD_parameters params = srcCCtx->params;
-		params.fParams.contentSizeFlag = (pledgedSrcSize > 0);
-		ZSTD_resetCCtx_advanced(dstCCtx, params, pledgedSrcSize, ZSTDcrp_noMemset);
-	}
-
-	/* copy tables */
-	{
-		size_t const chainSize = (srcCCtx->params.cParams.strategy == ZSTD_fast) ? 0 : (1 << srcCCtx->params.cParams.chainLog);
-		size_t const hSize = ((size_t)1) << srcCCtx->params.cParams.hashLog;
-		size_t const h3Size = (size_t)1 << srcCCtx->hashLog3;
-		size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
-		memcpy(dstCCtx->workSpace, srcCCtx->workSpace, tableSpace);
-	}
-
-	/* copy dictionary offsets */
-	dstCCtx->nextToUpdate = srcCCtx->nextToUpdate;
-	dstCCtx->nextToUpdate3 = srcCCtx->nextToUpdate3;
-	dstCCtx->nextSrc = srcCCtx->nextSrc;
-	dstCCtx->base = srcCCtx->base;
-	dstCCtx->dictBase = srcCCtx->dictBase;
-	dstCCtx->dictLimit = srcCCtx->dictLimit;
-	dstCCtx->lowLimit = srcCCtx->lowLimit;
-	dstCCtx->loadedDictEnd = srcCCtx->loadedDictEnd;
-	dstCCtx->dictID = srcCCtx->dictID;
-
-	/* copy entropy tables */
-	dstCCtx->flagStaticTables = srcCCtx->flagStaticTables;
-	dstCCtx->flagStaticHufTable = srcCCtx->flagStaticHufTable;
-	if (srcCCtx->flagStaticTables) {
-		memcpy(dstCCtx->litlengthCTable, srcCCtx->litlengthCTable, sizeof(dstCCtx->litlengthCTable));
-		memcpy(dstCCtx->matchlengthCTable, srcCCtx->matchlengthCTable, sizeof(dstCCtx->matchlengthCTable));
-		memcpy(dstCCtx->offcodeCTable, srcCCtx->offcodeCTable, sizeof(dstCCtx->offcodeCTable));
-	}
-	if (srcCCtx->flagStaticHufTable) {
-		memcpy(dstCCtx->hufTable, srcCCtx->hufTable, 256 * 4);
-	}
-
-	return 0;
-}
-
-/*! ZSTD_reduceTable() :
-*   reduce table indexes by `reducerValue` */
-static void ZSTD_reduceTable(U32 *const table, U32 const size, U32 const reducerValue)
-{
-	U32 u;
-	for (u = 0; u < size; u++) {
-		if (table[u] < reducerValue)
-			table[u] = 0;
-		else
-			table[u] -= reducerValue;
-	}
-}
-
-/*! ZSTD_reduceIndex() :
-*   rescale all indexes to avoid future overflow (indexes are U32) */
-static void ZSTD_reduceIndex(ZSTD_CCtx *zc, const U32 reducerValue)
-{
-	{
-		U32 const hSize = 1 << zc->params.cParams.hashLog;
-		ZSTD_reduceTable(zc->hashTable, hSize, reducerValue);
-	}
-
-	{
-		U32 const chainSize = (zc->params.cParams.strategy == ZSTD_fast) ? 0 : (1 << zc->params.cParams.chainLog);
-		ZSTD_reduceTable(zc->chainTable, chainSize, reducerValue);
-	}
-
-	{
-		U32 const h3Size = (zc->hashLog3) ? 1 << zc->hashLog3 : 0;
-		ZSTD_reduceTable(zc->hashTable3, h3Size, reducerValue);
-	}
-}
-
-/*-*******************************************************
-*  Block entropic compression
-*********************************************************/
-
-/* See doc/zstd_compression_format.md for detailed format description */
-
-size_t ZSTD_noCompressBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	if (srcSize + ZSTD_blockHeaderSize > dstCapacity)
-		return ERROR(dstSize_tooSmall);
-	memcpy((BYTE *)dst + ZSTD_blockHeaderSize, src, srcSize);
-	ZSTD_writeLE24(dst, (U32)(srcSize << 2) + (U32)bt_raw);
-	return ZSTD_blockHeaderSize + srcSize;
-}
-
-static size_t ZSTD_noCompressLiterals(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	BYTE *const ostart = (BYTE * const)dst;
-	U32 const flSize = 1 + (srcSize > 31) + (srcSize > 4095);
-
-	if (srcSize + flSize > dstCapacity)
-		return ERROR(dstSize_tooSmall);
-
-	switch (flSize) {
-	case 1: /* 2 - 1 - 5 */ ostart[0] = (BYTE)((U32)set_basic + (srcSize << 3)); break;
-	case 2: /* 2 - 2 - 12 */ ZSTD_writeLE16(ostart, (U16)((U32)set_basic + (1 << 2) + (srcSize << 4))); break;
-	default: /*note : should not be necessary : flSize is within {1,2,3} */
-	case 3: /* 2 - 2 - 20 */ ZSTD_writeLE32(ostart, (U32)((U32)set_basic + (3 << 2) + (srcSize << 4))); break;
-	}
-
-	memcpy(ostart + flSize, src, srcSize);
-	return srcSize + flSize;
-}
-
-static size_t ZSTD_compressRleLiteralsBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	BYTE *const ostart = (BYTE * const)dst;
-	U32 const flSize = 1 + (srcSize > 31) + (srcSize > 4095);
-
-	(void)dstCapacity; /* dstCapacity already guaranteed to be >=4, hence large enough */
-
-	switch (flSize) {
-	case 1: /* 2 - 1 - 5 */ ostart[0] = (BYTE)((U32)set_rle + (srcSize << 3)); break;
-	case 2: /* 2 - 2 - 12 */ ZSTD_writeLE16(ostart, (U16)((U32)set_rle + (1 << 2) + (srcSize << 4))); break;
-	default: /*note : should not be necessary : flSize is necessarily within {1,2,3} */
-	case 3: /* 2 - 2 - 20 */ ZSTD_writeLE32(ostart, (U32)((U32)set_rle + (3 << 2) + (srcSize << 4))); break;
-	}
-
-	ostart[flSize] = *(const BYTE *)src;
-	return flSize + 1;
-}
-
-static size_t ZSTD_minGain(size_t srcSize) { return (srcSize >> 6) + 2; }
-
-static size_t ZSTD_compressLiterals(ZSTD_CCtx *zc, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	size_t const minGain = ZSTD_minGain(srcSize);
-	size_t const lhSize = 3 + (srcSize >= 1 KB) + (srcSize >= 16 KB);
-	BYTE *const ostart = (BYTE *)dst;
-	U32 singleStream = srcSize < 256;
-	symbolEncodingType_e hType = set_compressed;
-	size_t cLitSize;
-
-/* small ? don't even attempt compression (speed opt) */
-#define LITERAL_NOENTROPY 63
-	{
-		size_t const minLitSize = zc->flagStaticHufTable == HUF_repeat_valid ? 6 : LITERAL_NOENTROPY;
-		if (srcSize <= minLitSize)
-			return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
-	}
-
-	if (dstCapacity < lhSize + 1)
-		return ERROR(dstSize_tooSmall); /* not enough space for compression */
-	{
-		HUF_repeat repeat = zc->flagStaticHufTable;
-		int const preferRepeat = zc->params.cParams.strategy < ZSTD_lazy ? srcSize <= 1024 : 0;
-		if (repeat == HUF_repeat_valid && lhSize == 3)
-			singleStream = 1;
-		cLitSize = singleStream ? HUF_compress1X_repeat(ostart + lhSize, dstCapacity - lhSize, src, srcSize, 255, 11, zc->tmpCounters,
-								sizeof(zc->tmpCounters), zc->hufTable, &repeat, preferRepeat)
-					: HUF_compress4X_repeat(ostart + lhSize, dstCapacity - lhSize, src, srcSize, 255, 11, zc->tmpCounters,
-								sizeof(zc->tmpCounters), zc->hufTable, &repeat, preferRepeat);
-		if (repeat != HUF_repeat_none) {
-			hType = set_repeat;
-		} /* reused the existing table */
-		else {
-			zc->flagStaticHufTable = HUF_repeat_check;
-		} /* now have a table to reuse */
-	}
-
-	if ((cLitSize == 0) | (cLitSize >= srcSize - minGain)) {
-		zc->flagStaticHufTable = HUF_repeat_none;
-		return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
-	}
-	if (cLitSize == 1) {
-		zc->flagStaticHufTable = HUF_repeat_none;
-		return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
-	}
-
-	/* Build header */
-	switch (lhSize) {
-	case 3: /* 2 - 2 - 10 - 10 */
-	{
-		U32 const lhc = hType + ((!singleStream) << 2) + ((U32)srcSize << 4) + ((U32)cLitSize << 14);
-		ZSTD_writeLE24(ostart, lhc);
-		break;
-	}
-	case 4: /* 2 - 2 - 14 - 14 */
-	{
-		U32 const lhc = hType + (2 << 2) + ((U32)srcSize << 4) + ((U32)cLitSize << 18);
-		ZSTD_writeLE32(ostart, lhc);
-		break;
-	}
-	default: /* should not be necessary, lhSize is only {3,4,5} */
-	case 5:  /* 2 - 2 - 18 - 18 */
-	{
-		U32 const lhc = hType + (3 << 2) + ((U32)srcSize << 4) + ((U32)cLitSize << 22);
-		ZSTD_writeLE32(ostart, lhc);
-		ostart[4] = (BYTE)(cLitSize >> 10);
-		break;
-	}
-	}
-	return lhSize + cLitSize;
-}
-
-static const BYTE LL_Code[64] = {0,  1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 16, 16, 17, 17, 18, 18,
-				 19, 19, 20, 20, 20, 20, 21, 21, 21, 21, 22, 22, 22, 22, 22, 22, 22, 22, 23, 23, 23, 23,
-				 23, 23, 23, 23, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24};
-
-static const BYTE ML_Code[128] = {0,  1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,
-				  26, 27, 28, 29, 30, 31, 32, 32, 33, 33, 34, 34, 35, 35, 36, 36, 36, 36, 37, 37, 37, 37, 38, 38, 38, 38,
-				  38, 38, 38, 38, 39, 39, 39, 39, 39, 39, 39, 39, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
-				  40, 40, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 42, 42, 42, 42, 42, 42, 42, 42,
-				  42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42};
-
-void ZSTD_seqToCodes(const seqStore_t *seqStorePtr)
-{
-	BYTE const LL_deltaCode = 19;
-	BYTE const ML_deltaCode = 36;
-	const seqDef *const sequences = seqStorePtr->sequencesStart;
-	BYTE *const llCodeTable = seqStorePtr->llCode;
-	BYTE *const ofCodeTable = seqStorePtr->ofCode;
-	BYTE *const mlCodeTable = seqStorePtr->mlCode;
-	U32 const nbSeq = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
-	U32 u;
-	for (u = 0; u < nbSeq; u++) {
-		U32 const llv = sequences[u].litLength;
-		U32 const mlv = sequences[u].matchLength;
-		llCodeTable[u] = (llv > 63) ? (BYTE)ZSTD_highbit32(llv) + LL_deltaCode : LL_Code[llv];
-		ofCodeTable[u] = (BYTE)ZSTD_highbit32(sequences[u].offset);
-		mlCodeTable[u] = (mlv > 127) ? (BYTE)ZSTD_highbit32(mlv) + ML_deltaCode : ML_Code[mlv];
-	}
-	if (seqStorePtr->longLengthID == 1)
-		llCodeTable[seqStorePtr->longLengthPos] = MaxLL;
-	if (seqStorePtr->longLengthID == 2)
-		mlCodeTable[seqStorePtr->longLengthPos] = MaxML;
-}
-
-ZSTD_STATIC size_t ZSTD_compressSequences_internal(ZSTD_CCtx *zc, void *dst, size_t dstCapacity)
-{
-	const int longOffsets = zc->params.cParams.windowLog > STREAM_ACCUMULATOR_MIN;
-	const seqStore_t *seqStorePtr = &(zc->seqStore);
-	FSE_CTable *CTable_LitLength = zc->litlengthCTable;
-	FSE_CTable *CTable_OffsetBits = zc->offcodeCTable;
-	FSE_CTable *CTable_MatchLength = zc->matchlengthCTable;
-	U32 LLtype, Offtype, MLtype; /* compressed, raw or rle */
-	const seqDef *const sequences = seqStorePtr->sequencesStart;
-	const BYTE *const ofCodeTable = seqStorePtr->ofCode;
-	const BYTE *const llCodeTable = seqStorePtr->llCode;
-	const BYTE *const mlCodeTable = seqStorePtr->mlCode;
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *const oend = ostart + dstCapacity;
-	BYTE *op = ostart;
-	size_t const nbSeq = seqStorePtr->sequences - seqStorePtr->sequencesStart;
-	BYTE *seqHead;
-
-	U32 *count;
-	S16 *norm;
-	U32 *workspace;
-	size_t workspaceSize = sizeof(zc->tmpCounters);
-	{
-		size_t spaceUsed32 = 0;
-		count = (U32 *)zc->tmpCounters + spaceUsed32;
-		spaceUsed32 += MaxSeq + 1;
-		norm = (S16 *)((U32 *)zc->tmpCounters + spaceUsed32);
-		spaceUsed32 += ALIGN(sizeof(S16) * (MaxSeq + 1), sizeof(U32)) >> 2;
-
-		workspace = (U32 *)zc->tmpCounters + spaceUsed32;
-		workspaceSize -= (spaceUsed32 << 2);
-	}
-
-	/* Compress literals */
-	{
-		const BYTE *const literals = seqStorePtr->litStart;
-		size_t const litSize = seqStorePtr->lit - literals;
-		size_t const cSize = ZSTD_compressLiterals(zc, op, dstCapacity, literals, litSize);
-		if (ZSTD_isError(cSize))
-			return cSize;
-		op += cSize;
-	}
-
-	/* Sequences Header */
-	if ((oend - op) < 3 /*max nbSeq Size*/ + 1 /*seqHead */)
-		return ERROR(dstSize_tooSmall);
-	if (nbSeq < 0x7F)
-		*op++ = (BYTE)nbSeq;
-	else if (nbSeq < LONGNBSEQ)
-		op[0] = (BYTE)((nbSeq >> 8) + 0x80), op[1] = (BYTE)nbSeq, op += 2;
-	else
-		op[0] = 0xFF, ZSTD_writeLE16(op + 1, (U16)(nbSeq - LONGNBSEQ)), op += 3;
-	if (nbSeq == 0)
-		return op - ostart;
-
-	/* seqHead : flags for FSE encoding type */
-	seqHead = op++;
-
-#define MIN_SEQ_FOR_DYNAMIC_FSE 64
-#define MAX_SEQ_FOR_STATIC_FSE 1000
-
-	/* convert length/distances into codes */
-	ZSTD_seqToCodes(seqStorePtr);
-
-	/* CTable for Literal Lengths */
-	{
-		U32 max = MaxLL;
-		size_t const mostFrequent = FSE_countFast_wksp(count, &max, llCodeTable, nbSeq, workspace);
-		if ((mostFrequent == nbSeq) && (nbSeq > 2)) {
-			*op++ = llCodeTable[0];
-			FSE_buildCTable_rle(CTable_LitLength, (BYTE)max);
-			LLtype = set_rle;
-		} else if ((zc->flagStaticTables) && (nbSeq < MAX_SEQ_FOR_STATIC_FSE)) {
-			LLtype = set_repeat;
-		} else if ((nbSeq < MIN_SEQ_FOR_DYNAMIC_FSE) || (mostFrequent < (nbSeq >> (LL_defaultNormLog - 1)))) {
-			FSE_buildCTable_wksp(CTable_LitLength, LL_defaultNorm, MaxLL, LL_defaultNormLog, workspace, workspaceSize);
-			LLtype = set_basic;
-		} else {
-			size_t nbSeq_1 = nbSeq;
-			const U32 tableLog = FSE_optimalTableLog(LLFSELog, nbSeq, max);
-			if (count[llCodeTable[nbSeq - 1]] > 1) {
-				count[llCodeTable[nbSeq - 1]]--;
-				nbSeq_1--;
-			}
-			FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max);
-			{
-				size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog); /* overflow protected */
-				if (FSE_isError(NCountSize))
-					return NCountSize;
-				op += NCountSize;
-			}
-			FSE_buildCTable_wksp(CTable_LitLength, norm, max, tableLog, workspace, workspaceSize);
-			LLtype = set_compressed;
-		}
-	}
-
-	/* CTable for Offsets */
-	{
-		U32 max = MaxOff;
-		size_t const mostFrequent = FSE_countFast_wksp(count, &max, ofCodeTable, nbSeq, workspace);
-		if ((mostFrequent == nbSeq) && (nbSeq > 2)) {
-			*op++ = ofCodeTable[0];
-			FSE_buildCTable_rle(CTable_OffsetBits, (BYTE)max);
-			Offtype = set_rle;
-		} else if ((zc->flagStaticTables) && (nbSeq < MAX_SEQ_FOR_STATIC_FSE)) {
-			Offtype = set_repeat;
-		} else if ((nbSeq < MIN_SEQ_FOR_DYNAMIC_FSE) || (mostFrequent < (nbSeq >> (OF_defaultNormLog - 1)))) {
-			FSE_buildCTable_wksp(CTable_OffsetBits, OF_defaultNorm, MaxOff, OF_defaultNormLog, workspace, workspaceSize);
-			Offtype = set_basic;
-		} else {
-			size_t nbSeq_1 = nbSeq;
-			const U32 tableLog = FSE_optimalTableLog(OffFSELog, nbSeq, max);
-			if (count[ofCodeTable[nbSeq - 1]] > 1) {
-				count[ofCodeTable[nbSeq - 1]]--;
-				nbSeq_1--;
-			}
-			FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max);
-			{
-				size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog); /* overflow protected */
-				if (FSE_isError(NCountSize))
-					return NCountSize;
-				op += NCountSize;
-			}
-			FSE_buildCTable_wksp(CTable_OffsetBits, norm, max, tableLog, workspace, workspaceSize);
-			Offtype = set_compressed;
-		}
-	}
-
-	/* CTable for MatchLengths */
-	{
-		U32 max = MaxML;
-		size_t const mostFrequent = FSE_countFast_wksp(count, &max, mlCodeTable, nbSeq, workspace);
-		if ((mostFrequent == nbSeq) && (nbSeq > 2)) {
-			*op++ = *mlCodeTable;
-			FSE_buildCTable_rle(CTable_MatchLength, (BYTE)max);
-			MLtype = set_rle;
-		} else if ((zc->flagStaticTables) && (nbSeq < MAX_SEQ_FOR_STATIC_FSE)) {
-			MLtype = set_repeat;
-		} else if ((nbSeq < MIN_SEQ_FOR_DYNAMIC_FSE) || (mostFrequent < (nbSeq >> (ML_defaultNormLog - 1)))) {
-			FSE_buildCTable_wksp(CTable_MatchLength, ML_defaultNorm, MaxML, ML_defaultNormLog, workspace, workspaceSize);
-			MLtype = set_basic;
-		} else {
-			size_t nbSeq_1 = nbSeq;
-			const U32 tableLog = FSE_optimalTableLog(MLFSELog, nbSeq, max);
-			if (count[mlCodeTable[nbSeq - 1]] > 1) {
-				count[mlCodeTable[nbSeq - 1]]--;
-				nbSeq_1--;
-			}
-			FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max);
-			{
-				size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog); /* overflow protected */
-				if (FSE_isError(NCountSize))
-					return NCountSize;
-				op += NCountSize;
-			}
-			FSE_buildCTable_wksp(CTable_MatchLength, norm, max, tableLog, workspace, workspaceSize);
-			MLtype = set_compressed;
-		}
-	}
-
-	*seqHead = (BYTE)((LLtype << 6) + (Offtype << 4) + (MLtype << 2));
-	zc->flagStaticTables = 0;
-
-	/* Encoding Sequences */
-	{
-		BIT_CStream_t blockStream;
-		FSE_CState_t stateMatchLength;
-		FSE_CState_t stateOffsetBits;
-		FSE_CState_t stateLitLength;
-
-		CHECK_E(BIT_initCStream(&blockStream, op, oend - op), dstSize_tooSmall); /* not enough space remaining */
-
-		/* first symbols */
-		FSE_initCState2(&stateMatchLength, CTable_MatchLength, mlCodeTable[nbSeq - 1]);
-		FSE_initCState2(&stateOffsetBits, CTable_OffsetBits, ofCodeTable[nbSeq - 1]);
-		FSE_initCState2(&stateLitLength, CTable_LitLength, llCodeTable[nbSeq - 1]);
-		BIT_addBits(&blockStream, sequences[nbSeq - 1].litLength, LL_bits[llCodeTable[nbSeq - 1]]);
-		if (ZSTD_32bits())
-			BIT_flushBits(&blockStream);
-		BIT_addBits(&blockStream, sequences[nbSeq - 1].matchLength, ML_bits[mlCodeTable[nbSeq - 1]]);
-		if (ZSTD_32bits())
-			BIT_flushBits(&blockStream);
-		if (longOffsets) {
-			U32 const ofBits = ofCodeTable[nbSeq - 1];
-			int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN - 1);
-			if (extraBits) {
-				BIT_addBits(&blockStream, sequences[nbSeq - 1].offset, extraBits);
-				BIT_flushBits(&blockStream);
-			}
-			BIT_addBits(&blockStream, sequences[nbSeq - 1].offset >> extraBits, ofBits - extraBits);
-		} else {
-			BIT_addBits(&blockStream, sequences[nbSeq - 1].offset, ofCodeTable[nbSeq - 1]);
-		}
-		BIT_flushBits(&blockStream);
-
-		{
-			size_t n;
-			for (n = nbSeq - 2; n < nbSeq; n--) { /* intentional underflow */
-				BYTE const llCode = llCodeTable[n];
-				BYTE const ofCode = ofCodeTable[n];
-				BYTE const mlCode = mlCodeTable[n];
-				U32 const llBits = LL_bits[llCode];
-				U32 const ofBits = ofCode; /* 32b*/ /* 64b*/
-				U32 const mlBits = ML_bits[mlCode];
-				/* (7)*/							    /* (7)*/
-				FSE_encodeSymbol(&blockStream, &stateOffsetBits, ofCode); /* 15 */  /* 15 */
-				FSE_encodeSymbol(&blockStream, &stateMatchLength, mlCode); /* 24 */ /* 24 */
-				if (ZSTD_32bits())
-					BIT_flushBits(&blockStream);				  /* (7)*/
-				FSE_encodeSymbol(&blockStream, &stateLitLength, llCode); /* 16 */ /* 33 */
-				if (ZSTD_32bits() || (ofBits + mlBits + llBits >= 64 - 7 - (LLFSELog + MLFSELog + OffFSELog)))
-					BIT_flushBits(&blockStream); /* (7)*/
-				BIT_addBits(&blockStream, sequences[n].litLength, llBits);
-				if (ZSTD_32bits() && ((llBits + mlBits) > 24))
-					BIT_flushBits(&blockStream);
-				BIT_addBits(&blockStream, sequences[n].matchLength, mlBits);
-				if (ZSTD_32bits())
-					BIT_flushBits(&blockStream); /* (7)*/
-				if (longOffsets) {
-					int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN - 1);
-					if (extraBits) {
-						BIT_addBits(&blockStream, sequences[n].offset, extraBits);
-						BIT_flushBits(&blockStream); /* (7)*/
-					}
-					BIT_addBits(&blockStream, sequences[n].offset >> extraBits, ofBits - extraBits); /* 31 */
-				} else {
-					BIT_addBits(&blockStream, sequences[n].offset, ofBits); /* 31 */
-				}
-				BIT_flushBits(&blockStream); /* (7)*/
-			}
-		}
-
-		FSE_flushCState(&blockStream, &stateMatchLength);
-		FSE_flushCState(&blockStream, &stateOffsetBits);
-		FSE_flushCState(&blockStream, &stateLitLength);
-
-		{
-			size_t const streamSize = BIT_closeCStream(&blockStream);
-			if (streamSize == 0)
-				return ERROR(dstSize_tooSmall); /* not enough space */
-			op += streamSize;
-		}
-	}
-	return op - ostart;
-}
-
-ZSTD_STATIC size_t ZSTD_compressSequences(ZSTD_CCtx *zc, void *dst, size_t dstCapacity, size_t srcSize)
-{
-	size_t const cSize = ZSTD_compressSequences_internal(zc, dst, dstCapacity);
-	size_t const minGain = ZSTD_minGain(srcSize);
-	size_t const maxCSize = srcSize - minGain;
-	/* If the srcSize <= dstCapacity, then there is enough space to write a
-	 * raw uncompressed block. Since we ran out of space, the block must not
-	 * be compressible, so fall back to a raw uncompressed block.
-	 */
-	int const uncompressibleError = cSize == ERROR(dstSize_tooSmall) && srcSize <= dstCapacity;
-	int i;
-
-	if (ZSTD_isError(cSize) && !uncompressibleError)
-		return cSize;
-	if (cSize >= maxCSize || uncompressibleError) {
-		zc->flagStaticHufTable = HUF_repeat_none;
-		return 0;
-	}
-	/* confirm repcodes */
-	for (i = 0; i < ZSTD_REP_NUM; i++)
-		zc->rep[i] = zc->repToConfirm[i];
-	return cSize;
-}
-
-/*! ZSTD_storeSeq() :
-	Store a sequence (literal length, literals, offset code and match length code) into seqStore_t.
-	`offsetCode` : distance to match, or 0 == repCode.
-	`matchCode` : matchLength - MINMATCH
-*/
-ZSTD_STATIC void ZSTD_storeSeq(seqStore_t *seqStorePtr, size_t litLength, const void *literals, U32 offsetCode, size_t matchCode)
-{
-	/* copy Literals */
-	ZSTD_wildcopy(seqStorePtr->lit, literals, litLength);
-	seqStorePtr->lit += litLength;
-
-	/* literal Length */
-	if (litLength > 0xFFFF) {
-		seqStorePtr->longLengthID = 1;
-		seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
-	}
-	seqStorePtr->sequences[0].litLength = (U16)litLength;
-
-	/* match offset */
-	seqStorePtr->sequences[0].offset = offsetCode + 1;
-
-	/* match Length */
-	if (matchCode > 0xFFFF) {
-		seqStorePtr->longLengthID = 2;
-		seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
-	}
-	seqStorePtr->sequences[0].matchLength = (U16)matchCode;
-
-	seqStorePtr->sequences++;
-}
-
-/*-*************************************
-*  Match length counter
-***************************************/
-static unsigned ZSTD_NbCommonBytes(register size_t val)
-{
-	if (ZSTD_isLittleEndian()) {
-		if (ZSTD_64bits()) {
-			return (__builtin_ctzll((U64)val) >> 3);
-		} else { /* 32 bits */
-			return (__builtin_ctz((U32)val) >> 3);
-		}
-	} else { /* Big Endian CPU */
-		if (ZSTD_64bits()) {
-			return (__builtin_clzll(val) >> 3);
-		} else { /* 32 bits */
-			return (__builtin_clz((U32)val) >> 3);
-		}
-	}
-}
-
-static size_t ZSTD_count(const BYTE *pIn, const BYTE *pMatch, const BYTE *const pInLimit)
-{
-	const BYTE *const pStart = pIn;
-	const BYTE *const pInLoopLimit = pInLimit - (sizeof(size_t) - 1);
-
-	while (pIn < pInLoopLimit) {
-		size_t const diff = ZSTD_readST(pMatch) ^ ZSTD_readST(pIn);
-		if (!diff) {
-			pIn += sizeof(size_t);
-			pMatch += sizeof(size_t);
-			continue;
-		}
-		pIn += ZSTD_NbCommonBytes(diff);
-		return (size_t)(pIn - pStart);
-	}
-	if (ZSTD_64bits())
-		if ((pIn < (pInLimit - 3)) && (ZSTD_read32(pMatch) == ZSTD_read32(pIn))) {
-			pIn += 4;
-			pMatch += 4;
-		}
-	if ((pIn < (pInLimit - 1)) && (ZSTD_read16(pMatch) == ZSTD_read16(pIn))) {
-		pIn += 2;
-		pMatch += 2;
-	}
-	if ((pIn < pInLimit) && (*pMatch == *pIn))
-		pIn++;
-	return (size_t)(pIn - pStart);
-}
-
-/** ZSTD_count_2segments() :
-*   can count match length with `ip` & `match` in 2 different segments.
-*   convention : on reaching mEnd, match count continue starting from iStart
-*/
-static size_t ZSTD_count_2segments(const BYTE *ip, const BYTE *match, const BYTE *iEnd, const BYTE *mEnd, const BYTE *iStart)
-{
-	const BYTE *const vEnd = MIN(ip + (mEnd - match), iEnd);
-	size_t const matchLength = ZSTD_count(ip, match, vEnd);
-	if (match + matchLength != mEnd)
-		return matchLength;
-	return matchLength + ZSTD_count(ip + matchLength, iStart, iEnd);
-}
-
-/*-*************************************
-*  Hashes
-***************************************/
-static const U32 prime3bytes = 506832829U;
-static U32 ZSTD_hash3(U32 u, U32 h) { return ((u << (32 - 24)) * prime3bytes) >> (32 - h); }
-ZSTD_STATIC size_t ZSTD_hash3Ptr(const void *ptr, U32 h) { return ZSTD_hash3(ZSTD_readLE32(ptr), h); } /* only in zstd_opt.h */
-
-static const U32 prime4bytes = 2654435761U;
-static U32 ZSTD_hash4(U32 u, U32 h) { return (u * prime4bytes) >> (32 - h); }
-static size_t ZSTD_hash4Ptr(const void *ptr, U32 h) { return ZSTD_hash4(ZSTD_read32(ptr), h); }
-
-static const U64 prime5bytes = 889523592379ULL;
-static size_t ZSTD_hash5(U64 u, U32 h) { return (size_t)(((u << (64 - 40)) * prime5bytes) >> (64 - h)); }
-static size_t ZSTD_hash5Ptr(const void *p, U32 h) { return ZSTD_hash5(ZSTD_readLE64(p), h); }
-
-static const U64 prime6bytes = 227718039650203ULL;
-static size_t ZSTD_hash6(U64 u, U32 h) { return (size_t)(((u << (64 - 48)) * prime6bytes) >> (64 - h)); }
-static size_t ZSTD_hash6Ptr(const void *p, U32 h) { return ZSTD_hash6(ZSTD_readLE64(p), h); }
-
-static const U64 prime7bytes = 58295818150454627ULL;
-static size_t ZSTD_hash7(U64 u, U32 h) { return (size_t)(((u << (64 - 56)) * prime7bytes) >> (64 - h)); }
-static size_t ZSTD_hash7Ptr(const void *p, U32 h) { return ZSTD_hash7(ZSTD_readLE64(p), h); }
-
-static const U64 prime8bytes = 0xCF1BBCDCB7A56463ULL;
-static size_t ZSTD_hash8(U64 u, U32 h) { return (size_t)(((u)*prime8bytes) >> (64 - h)); }
-static size_t ZSTD_hash8Ptr(const void *p, U32 h) { return ZSTD_hash8(ZSTD_readLE64(p), h); }
-
-static size_t ZSTD_hashPtr(const void *p, U32 hBits, U32 mls)
-{
-	switch (mls) {
-	// case 3: return ZSTD_hash3Ptr(p, hBits);
-	default:
-	case 4: return ZSTD_hash4Ptr(p, hBits);
-	case 5: return ZSTD_hash5Ptr(p, hBits);
-	case 6: return ZSTD_hash6Ptr(p, hBits);
-	case 7: return ZSTD_hash7Ptr(p, hBits);
-	case 8: return ZSTD_hash8Ptr(p, hBits);
-	}
-}
-
-/*-*************************************
-*  Fast Scan
-***************************************/
-static void ZSTD_fillHashTable(ZSTD_CCtx *zc, const void *end, const U32 mls)
-{
-	U32 *const hashTable = zc->hashTable;
-	U32 const hBits = zc->params.cParams.hashLog;
-	const BYTE *const base = zc->base;
-	const BYTE *ip = base + zc->nextToUpdate;
-	const BYTE *const iend = ((const BYTE *)end) - HASH_READ_SIZE;
-	const size_t fastHashFillStep = 3;
-
-	while (ip <= iend) {
-		hashTable[ZSTD_hashPtr(ip, hBits, mls)] = (U32)(ip - base);
-		ip += fastHashFillStep;
-	}
-}
-
-FORCE_INLINE
-void ZSTD_compressBlock_fast_generic(ZSTD_CCtx *cctx, const void *src, size_t srcSize, const U32 mls)
-{
-	U32 *const hashTable = cctx->hashTable;
-	U32 const hBits = cctx->params.cParams.hashLog;
-	seqStore_t *seqStorePtr = &(cctx->seqStore);
-	const BYTE *const base = cctx->base;
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const U32 lowestIndex = cctx->dictLimit;
-	const BYTE *const lowest = base + lowestIndex;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - HASH_READ_SIZE;
-	U32 offset_1 = cctx->rep[0], offset_2 = cctx->rep[1];
-	U32 offsetSaved = 0;
-
-	/* init */
-	ip += (ip == lowest);
-	{
-		U32 const maxRep = (U32)(ip - lowest);
-		if (offset_2 > maxRep)
-			offsetSaved = offset_2, offset_2 = 0;
-		if (offset_1 > maxRep)
-			offsetSaved = offset_1, offset_1 = 0;
-	}
-
-	/* Main Search Loop */
-	while (ip < ilimit) { /* < instead of <=, because repcode check at (ip+1) */
-		size_t mLength;
-		size_t const h = ZSTD_hashPtr(ip, hBits, mls);
-		U32 const curr = (U32)(ip - base);
-		U32 const matchIndex = hashTable[h];
-		const BYTE *match = base + matchIndex;
-		hashTable[h] = curr; /* update hash table */
-
-		if ((offset_1 > 0) & (ZSTD_read32(ip + 1 - offset_1) == ZSTD_read32(ip + 1))) {
-			mLength = ZSTD_count(ip + 1 + 4, ip + 1 + 4 - offset_1, iend) + 4;
-			ip++;
-			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, 0, mLength - MINMATCH);
-		} else {
-			U32 offset;
-			if ((matchIndex <= lowestIndex) || (ZSTD_read32(match) != ZSTD_read32(ip))) {
-				ip += ((ip - anchor) >> g_searchStrength) + 1;
-				continue;
-			}
-			mLength = ZSTD_count(ip + 4, match + 4, iend) + 4;
-			offset = (U32)(ip - match);
-			while (((ip > anchor) & (match > lowest)) && (ip[-1] == match[-1])) {
-				ip--;
-				match--;
-				mLength++;
-			} /* catch up */
-			offset_2 = offset_1;
-			offset_1 = offset;
-
-			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
-		}
-
-		/* match found */
-		ip += mLength;
-		anchor = ip;
-
-		if (ip <= ilimit) {
-			/* Fill Table */
-			hashTable[ZSTD_hashPtr(base + curr + 2, hBits, mls)] = curr + 2; /* here because curr+2 could be > iend-8 */
-			hashTable[ZSTD_hashPtr(ip - 2, hBits, mls)] = (U32)(ip - 2 - base);
-			/* check immediate repcode */
-			while ((ip <= ilimit) && ((offset_2 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_2)))) {
-				/* store sequence */
-				size_t const rLength = ZSTD_count(ip + 4, ip + 4 - offset_2, iend) + 4;
-				{
-					U32 const tmpOff = offset_2;
-					offset_2 = offset_1;
-					offset_1 = tmpOff;
-				} /* swap offset_2 <=> offset_1 */
-				hashTable[ZSTD_hashPtr(ip, hBits, mls)] = (U32)(ip - base);
-				ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, rLength - MINMATCH);
-				ip += rLength;
-				anchor = ip;
-				continue; /* faster when present ... (?) */
-			}
-		}
-	}
-
-	/* save reps for next block */
-	cctx->repToConfirm[0] = offset_1 ? offset_1 : offsetSaved;
-	cctx->repToConfirm[1] = offset_2 ? offset_2 : offsetSaved;
-
-	/* Last Literals */
-	{
-		size_t const lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-static void ZSTD_compressBlock_fast(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-	const U32 mls = ctx->params.cParams.searchLength;
-	switch (mls) {
-	default: /* includes case 3 */
-	case 4: ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 4); return;
-	case 5: ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 5); return;
-	case 6: ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 6); return;
-	case 7: ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 7); return;
-	}
-}
-
-static void ZSTD_compressBlock_fast_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const U32 mls)
-{
-	U32 *hashTable = ctx->hashTable;
-	const U32 hBits = ctx->params.cParams.hashLog;
-	seqStore_t *seqStorePtr = &(ctx->seqStore);
-	const BYTE *const base = ctx->base;
-	const BYTE *const dictBase = ctx->dictBase;
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const U32 lowestIndex = ctx->lowLimit;
-	const BYTE *const dictStart = dictBase + lowestIndex;
-	const U32 dictLimit = ctx->dictLimit;
-	const BYTE *const lowPrefixPtr = base + dictLimit;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - 8;
-	U32 offset_1 = ctx->rep[0], offset_2 = ctx->rep[1];
-
-	/* Search Loop */
-	while (ip < ilimit) { /* < instead of <=, because (ip+1) */
-		const size_t h = ZSTD_hashPtr(ip, hBits, mls);
-		const U32 matchIndex = hashTable[h];
-		const BYTE *matchBase = matchIndex < dictLimit ? dictBase : base;
-		const BYTE *match = matchBase + matchIndex;
-		const U32 curr = (U32)(ip - base);
-		const U32 repIndex = curr + 1 - offset_1; /* offset_1 expected <= curr +1 */
-		const BYTE *repBase = repIndex < dictLimit ? dictBase : base;
-		const BYTE *repMatch = repBase + repIndex;
-		size_t mLength;
-		hashTable[h] = curr; /* update hash table */
-
-		if ((((U32)((dictLimit - 1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > lowestIndex)) &&
-		    (ZSTD_read32(repMatch) == ZSTD_read32(ip + 1))) {
-			const BYTE *repMatchEnd = repIndex < dictLimit ? dictEnd : iend;
-			mLength = ZSTD_count_2segments(ip + 1 + EQUAL_READ32, repMatch + EQUAL_READ32, iend, repMatchEnd, lowPrefixPtr) + EQUAL_READ32;
-			ip++;
-			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, 0, mLength - MINMATCH);
-		} else {
-			if ((matchIndex < lowestIndex) || (ZSTD_read32(match) != ZSTD_read32(ip))) {
-				ip += ((ip - anchor) >> g_searchStrength) + 1;
-				continue;
-			}
-			{
-				const BYTE *matchEnd = matchIndex < dictLimit ? dictEnd : iend;
-				const BYTE *lowMatchPtr = matchIndex < dictLimit ? dictStart : lowPrefixPtr;
-				U32 offset;
-				mLength = ZSTD_count_2segments(ip + EQUAL_READ32, match + EQUAL_READ32, iend, matchEnd, lowPrefixPtr) + EQUAL_READ32;
-				while (((ip > anchor) & (match > lowMatchPtr)) && (ip[-1] == match[-1])) {
-					ip--;
-					match--;
-					mLength++;
-				} /* catch up */
-				offset = curr - matchIndex;
-				offset_2 = offset_1;
-				offset_1 = offset;
-				ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
-			}
-		}
-
-		/* found a match : store it */
-		ip += mLength;
-		anchor = ip;
-
-		if (ip <= ilimit) {
-			/* Fill Table */
-			hashTable[ZSTD_hashPtr(base + curr + 2, hBits, mls)] = curr + 2;
-			hashTable[ZSTD_hashPtr(ip - 2, hBits, mls)] = (U32)(ip - 2 - base);
-			/* check immediate repcode */
-			while (ip <= ilimit) {
-				U32 const curr2 = (U32)(ip - base);
-				U32 const repIndex2 = curr2 - offset_2;
-				const BYTE *repMatch2 = repIndex2 < dictLimit ? dictBase + repIndex2 : base + repIndex2;
-				if ((((U32)((dictLimit - 1) - repIndex2) >= 3) & (repIndex2 > lowestIndex)) /* intentional overflow */
-				    && (ZSTD_read32(repMatch2) == ZSTD_read32(ip))) {
-					const BYTE *const repEnd2 = repIndex2 < dictLimit ? dictEnd : iend;
-					size_t repLength2 =
-					    ZSTD_count_2segments(ip + EQUAL_READ32, repMatch2 + EQUAL_READ32, iend, repEnd2, lowPrefixPtr) + EQUAL_READ32;
-					U32 tmpOffset = offset_2;
-					offset_2 = offset_1;
-					offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
-					ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, repLength2 - MINMATCH);
-					hashTable[ZSTD_hashPtr(ip, hBits, mls)] = curr2;
-					ip += repLength2;
-					anchor = ip;
-					continue;
-				}
-				break;
-			}
-		}
-	}
-
-	/* save reps for next block */
-	ctx->repToConfirm[0] = offset_1;
-	ctx->repToConfirm[1] = offset_2;
-
-	/* Last Literals */
-	{
-		size_t const lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-static void ZSTD_compressBlock_fast_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-	U32 const mls = ctx->params.cParams.searchLength;
-	switch (mls) {
-	default: /* includes case 3 */
-	case 4: ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 4); return;
-	case 5: ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 5); return;
-	case 6: ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 6); return;
-	case 7: ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 7); return;
-	}
-}
-
-/*-*************************************
-*  Double Fast
-***************************************/
-static void ZSTD_fillDoubleHashTable(ZSTD_CCtx *cctx, const void *end, const U32 mls)
-{
-	U32 *const hashLarge = cctx->hashTable;
-	U32 const hBitsL = cctx->params.cParams.hashLog;
-	U32 *const hashSmall = cctx->chainTable;
-	U32 const hBitsS = cctx->params.cParams.chainLog;
-	const BYTE *const base = cctx->base;
-	const BYTE *ip = base + cctx->nextToUpdate;
-	const BYTE *const iend = ((const BYTE *)end) - HASH_READ_SIZE;
-	const size_t fastHashFillStep = 3;
-
-	while (ip <= iend) {
-		hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip - base);
-		hashLarge[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip - base);
-		ip += fastHashFillStep;
-	}
-}
-
-FORCE_INLINE
-void ZSTD_compressBlock_doubleFast_generic(ZSTD_CCtx *cctx, const void *src, size_t srcSize, const U32 mls)
-{
-	U32 *const hashLong = cctx->hashTable;
-	const U32 hBitsL = cctx->params.cParams.hashLog;
-	U32 *const hashSmall = cctx->chainTable;
-	const U32 hBitsS = cctx->params.cParams.chainLog;
-	seqStore_t *seqStorePtr = &(cctx->seqStore);
-	const BYTE *const base = cctx->base;
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const U32 lowestIndex = cctx->dictLimit;
-	const BYTE *const lowest = base + lowestIndex;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - HASH_READ_SIZE;
-	U32 offset_1 = cctx->rep[0], offset_2 = cctx->rep[1];
-	U32 offsetSaved = 0;
-
-	/* init */
-	ip += (ip == lowest);
-	{
-		U32 const maxRep = (U32)(ip - lowest);
-		if (offset_2 > maxRep)
-			offsetSaved = offset_2, offset_2 = 0;
-		if (offset_1 > maxRep)
-			offsetSaved = offset_1, offset_1 = 0;
-	}
-
-	/* Main Search Loop */
-	while (ip < ilimit) { /* < instead of <=, because repcode check at (ip+1) */
-		size_t mLength;
-		size_t const h2 = ZSTD_hashPtr(ip, hBitsL, 8);
-		size_t const h = ZSTD_hashPtr(ip, hBitsS, mls);
-		U32 const curr = (U32)(ip - base);
-		U32 const matchIndexL = hashLong[h2];
-		U32 const matchIndexS = hashSmall[h];
-		const BYTE *matchLong = base + matchIndexL;
-		const BYTE *match = base + matchIndexS;
-		hashLong[h2] = hashSmall[h] = curr; /* update hash tables */
-
-		if ((offset_1 > 0) & (ZSTD_read32(ip + 1 - offset_1) == ZSTD_read32(ip + 1))) { /* note : by construction, offset_1 <= curr */
-			mLength = ZSTD_count(ip + 1 + 4, ip + 1 + 4 - offset_1, iend) + 4;
-			ip++;
-			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, 0, mLength - MINMATCH);
-		} else {
-			U32 offset;
-			if ((matchIndexL > lowestIndex) && (ZSTD_read64(matchLong) == ZSTD_read64(ip))) {
-				mLength = ZSTD_count(ip + 8, matchLong + 8, iend) + 8;
-				offset = (U32)(ip - matchLong);
-				while (((ip > anchor) & (matchLong > lowest)) && (ip[-1] == matchLong[-1])) {
-					ip--;
-					matchLong--;
-					mLength++;
-				} /* catch up */
-			} else if ((matchIndexS > lowestIndex) && (ZSTD_read32(match) == ZSTD_read32(ip))) {
-				size_t const h3 = ZSTD_hashPtr(ip + 1, hBitsL, 8);
-				U32 const matchIndex3 = hashLong[h3];
-				const BYTE *match3 = base + matchIndex3;
-				hashLong[h3] = curr + 1;
-				if ((matchIndex3 > lowestIndex) && (ZSTD_read64(match3) == ZSTD_read64(ip + 1))) {
-					mLength = ZSTD_count(ip + 9, match3 + 8, iend) + 8;
-					ip++;
-					offset = (U32)(ip - match3);
-					while (((ip > anchor) & (match3 > lowest)) && (ip[-1] == match3[-1])) {
-						ip--;
-						match3--;
-						mLength++;
-					} /* catch up */
-				} else {
-					mLength = ZSTD_count(ip + 4, match + 4, iend) + 4;
-					offset = (U32)(ip - match);
-					while (((ip > anchor) & (match > lowest)) && (ip[-1] == match[-1])) {
-						ip--;
-						match--;
-						mLength++;
-					} /* catch up */
-				}
-			} else {
-				ip += ((ip - anchor) >> g_searchStrength) + 1;
-				continue;
-			}
-
-			offset_2 = offset_1;
-			offset_1 = offset;
-
-			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
-		}
-
-		/* match found */
-		ip += mLength;
-		anchor = ip;
-
-		if (ip <= ilimit) {
-			/* Fill Table */
-			hashLong[ZSTD_hashPtr(base + curr + 2, hBitsL, 8)] = hashSmall[ZSTD_hashPtr(base + curr + 2, hBitsS, mls)] =
-			    curr + 2; /* here because curr+2 could be > iend-8 */
-			hashLong[ZSTD_hashPtr(ip - 2, hBitsL, 8)] = hashSmall[ZSTD_hashPtr(ip - 2, hBitsS, mls)] = (U32)(ip - 2 - base);
-
-			/* check immediate repcode */
-			while ((ip <= ilimit) && ((offset_2 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_2)))) {
-				/* store sequence */
-				size_t const rLength = ZSTD_count(ip + 4, ip + 4 - offset_2, iend) + 4;
-				{
-					U32 const tmpOff = offset_2;
-					offset_2 = offset_1;
-					offset_1 = tmpOff;
-				} /* swap offset_2 <=> offset_1 */
-				hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip - base);
-				hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip - base);
-				ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, rLength - MINMATCH);
-				ip += rLength;
-				anchor = ip;
-				continue; /* faster when present ... (?) */
-			}
-		}
-	}
-
-	/* save reps for next block */
-	cctx->repToConfirm[0] = offset_1 ? offset_1 : offsetSaved;
-	cctx->repToConfirm[1] = offset_2 ? offset_2 : offsetSaved;
-
-	/* Last Literals */
-	{
-		size_t const lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-static void ZSTD_compressBlock_doubleFast(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-	const U32 mls = ctx->params.cParams.searchLength;
-	switch (mls) {
-	default: /* includes case 3 */
-	case 4: ZSTD_compressBlock_doubleFast_generic(ctx, src, srcSize, 4); return;
-	case 5: ZSTD_compressBlock_doubleFast_generic(ctx, src, srcSize, 5); return;
-	case 6: ZSTD_compressBlock_doubleFast_generic(ctx, src, srcSize, 6); return;
-	case 7: ZSTD_compressBlock_doubleFast_generic(ctx, src, srcSize, 7); return;
-	}
-}
-
-static void ZSTD_compressBlock_doubleFast_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const U32 mls)
-{
-	U32 *const hashLong = ctx->hashTable;
-	U32 const hBitsL = ctx->params.cParams.hashLog;
-	U32 *const hashSmall = ctx->chainTable;
-	U32 const hBitsS = ctx->params.cParams.chainLog;
-	seqStore_t *seqStorePtr = &(ctx->seqStore);
-	const BYTE *const base = ctx->base;
-	const BYTE *const dictBase = ctx->dictBase;
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const U32 lowestIndex = ctx->lowLimit;
-	const BYTE *const dictStart = dictBase + lowestIndex;
-	const U32 dictLimit = ctx->dictLimit;
-	const BYTE *const lowPrefixPtr = base + dictLimit;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - 8;
-	U32 offset_1 = ctx->rep[0], offset_2 = ctx->rep[1];
-
-	/* Search Loop */
-	while (ip < ilimit) { /* < instead of <=, because (ip+1) */
-		const size_t hSmall = ZSTD_hashPtr(ip, hBitsS, mls);
-		const U32 matchIndex = hashSmall[hSmall];
-		const BYTE *matchBase = matchIndex < dictLimit ? dictBase : base;
-		const BYTE *match = matchBase + matchIndex;
-
-		const size_t hLong = ZSTD_hashPtr(ip, hBitsL, 8);
-		const U32 matchLongIndex = hashLong[hLong];
-		const BYTE *matchLongBase = matchLongIndex < dictLimit ? dictBase : base;
-		const BYTE *matchLong = matchLongBase + matchLongIndex;
-
-		const U32 curr = (U32)(ip - base);
-		const U32 repIndex = curr + 1 - offset_1; /* offset_1 expected <= curr +1 */
-		const BYTE *repBase = repIndex < dictLimit ? dictBase : base;
-		const BYTE *repMatch = repBase + repIndex;
-		size_t mLength;
-		hashSmall[hSmall] = hashLong[hLong] = curr; /* update hash table */
-
-		if ((((U32)((dictLimit - 1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > lowestIndex)) &&
-		    (ZSTD_read32(repMatch) == ZSTD_read32(ip + 1))) {
-			const BYTE *repMatchEnd = repIndex < dictLimit ? dictEnd : iend;
-			mLength = ZSTD_count_2segments(ip + 1 + 4, repMatch + 4, iend, repMatchEnd, lowPrefixPtr) + 4;
-			ip++;
-			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, 0, mLength - MINMATCH);
-		} else {
-			if ((matchLongIndex > lowestIndex) && (ZSTD_read64(matchLong) == ZSTD_read64(ip))) {
-				const BYTE *matchEnd = matchLongIndex < dictLimit ? dictEnd : iend;
-				const BYTE *lowMatchPtr = matchLongIndex < dictLimit ? dictStart : lowPrefixPtr;
-				U32 offset;
-				mLength = ZSTD_count_2segments(ip + 8, matchLong + 8, iend, matchEnd, lowPrefixPtr) + 8;
-				offset = curr - matchLongIndex;
-				while (((ip > anchor) & (matchLong > lowMatchPtr)) && (ip[-1] == matchLong[-1])) {
-					ip--;
-					matchLong--;
-					mLength++;
-				} /* catch up */
-				offset_2 = offset_1;
-				offset_1 = offset;
-				ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
-
-			} else if ((matchIndex > lowestIndex) && (ZSTD_read32(match) == ZSTD_read32(ip))) {
-				size_t const h3 = ZSTD_hashPtr(ip + 1, hBitsL, 8);
-				U32 const matchIndex3 = hashLong[h3];
-				const BYTE *const match3Base = matchIndex3 < dictLimit ? dictBase : base;
-				const BYTE *match3 = match3Base + matchIndex3;
-				U32 offset;
-				hashLong[h3] = curr + 1;
-				if ((matchIndex3 > lowestIndex) && (ZSTD_read64(match3) == ZSTD_read64(ip + 1))) {
-					const BYTE *matchEnd = matchIndex3 < dictLimit ? dictEnd : iend;
-					const BYTE *lowMatchPtr = matchIndex3 < dictLimit ? dictStart : lowPrefixPtr;
-					mLength = ZSTD_count_2segments(ip + 9, match3 + 8, iend, matchEnd, lowPrefixPtr) + 8;
-					ip++;
-					offset = curr + 1 - matchIndex3;
-					while (((ip > anchor) & (match3 > lowMatchPtr)) && (ip[-1] == match3[-1])) {
-						ip--;
-						match3--;
-						mLength++;
-					} /* catch up */
-				} else {
-					const BYTE *matchEnd = matchIndex < dictLimit ? dictEnd : iend;
-					const BYTE *lowMatchPtr = matchIndex < dictLimit ? dictStart : lowPrefixPtr;
-					mLength = ZSTD_count_2segments(ip + 4, match + 4, iend, matchEnd, lowPrefixPtr) + 4;
-					offset = curr - matchIndex;
-					while (((ip > anchor) & (match > lowMatchPtr)) && (ip[-1] == match[-1])) {
-						ip--;
-						match--;
-						mLength++;
-					} /* catch up */
-				}
-				offset_2 = offset_1;
-				offset_1 = offset;
-				ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
-
-			} else {
-				ip += ((ip - anchor) >> g_searchStrength) + 1;
-				continue;
-			}
-		}
-
-		/* found a match : store it */
-		ip += mLength;
-		anchor = ip;
-
-		if (ip <= ilimit) {
-			/* Fill Table */
-			hashSmall[ZSTD_hashPtr(base + curr + 2, hBitsS, mls)] = curr + 2;
-			hashLong[ZSTD_hashPtr(base + curr + 2, hBitsL, 8)] = curr + 2;
-			hashSmall[ZSTD_hashPtr(ip - 2, hBitsS, mls)] = (U32)(ip - 2 - base);
-			hashLong[ZSTD_hashPtr(ip - 2, hBitsL, 8)] = (U32)(ip - 2 - base);
-			/* check immediate repcode */
-			while (ip <= ilimit) {
-				U32 const curr2 = (U32)(ip - base);
-				U32 const repIndex2 = curr2 - offset_2;
-				const BYTE *repMatch2 = repIndex2 < dictLimit ? dictBase + repIndex2 : base + repIndex2;
-				if ((((U32)((dictLimit - 1) - repIndex2) >= 3) & (repIndex2 > lowestIndex)) /* intentional overflow */
-				    && (ZSTD_read32(repMatch2) == ZSTD_read32(ip))) {
-					const BYTE *const repEnd2 = repIndex2 < dictLimit ? dictEnd : iend;
-					size_t const repLength2 =
-					    ZSTD_count_2segments(ip + EQUAL_READ32, repMatch2 + EQUAL_READ32, iend, repEnd2, lowPrefixPtr) + EQUAL_READ32;
-					U32 tmpOffset = offset_2;
-					offset_2 = offset_1;
-					offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
-					ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, repLength2 - MINMATCH);
-					hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = curr2;
-					hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = curr2;
-					ip += repLength2;
-					anchor = ip;
-					continue;
-				}
-				break;
-			}
-		}
-	}
-
-	/* save reps for next block */
-	ctx->repToConfirm[0] = offset_1;
-	ctx->repToConfirm[1] = offset_2;
-
-	/* Last Literals */
-	{
-		size_t const lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-static void ZSTD_compressBlock_doubleFast_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-	U32 const mls = ctx->params.cParams.searchLength;
-	switch (mls) {
-	default: /* includes case 3 */
-	case 4: ZSTD_compressBlock_doubleFast_extDict_generic(ctx, src, srcSize, 4); return;
-	case 5: ZSTD_compressBlock_doubleFast_extDict_generic(ctx, src, srcSize, 5); return;
-	case 6: ZSTD_compressBlock_doubleFast_extDict_generic(ctx, src, srcSize, 6); return;
-	case 7: ZSTD_compressBlock_doubleFast_extDict_generic(ctx, src, srcSize, 7); return;
-	}
-}
-
-/*-*************************************
-*  Binary Tree search
-***************************************/
-/** ZSTD_insertBt1() : add one or multiple positions to tree.
-*   ip : assumed <= iend-8 .
-*   @return : nb of positions added */
-static U32 ZSTD_insertBt1(ZSTD_CCtx *zc, const BYTE *const ip, const U32 mls, const BYTE *const iend, U32 nbCompares, U32 extDict)
-{
-	U32 *const hashTable = zc->hashTable;
-	U32 const hashLog = zc->params.cParams.hashLog;
-	size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
-	U32 *const bt = zc->chainTable;
-	U32 const btLog = zc->params.cParams.chainLog - 1;
-	U32 const btMask = (1 << btLog) - 1;
-	U32 matchIndex = hashTable[h];
-	size_t commonLengthSmaller = 0, commonLengthLarger = 0;
-	const BYTE *const base = zc->base;
-	const BYTE *const dictBase = zc->dictBase;
-	const U32 dictLimit = zc->dictLimit;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-	const BYTE *const prefixStart = base + dictLimit;
-	const BYTE *match;
-	const U32 curr = (U32)(ip - base);
-	const U32 btLow = btMask >= curr ? 0 : curr - btMask;
-	U32 *smallerPtr = bt + 2 * (curr & btMask);
-	U32 *largerPtr = smallerPtr + 1;
-	U32 dummy32; /* to be nullified at the end */
-	U32 const windowLow = zc->lowLimit;
-	U32 matchEndIdx = curr + 8;
-	size_t bestLength = 8;
-
-	hashTable[h] = curr; /* Update Hash Table */
-
-	while (nbCompares-- && (matchIndex > windowLow)) {
-		U32 *const nextPtr = bt + 2 * (matchIndex & btMask);
-		size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
-
-		if ((!extDict) || (matchIndex + matchLength >= dictLimit)) {
-			match = base + matchIndex;
-			if (match[matchLength] == ip[matchLength])
-				matchLength += ZSTD_count(ip + matchLength + 1, match + matchLength + 1, iend) + 1;
-		} else {
-			match = dictBase + matchIndex;
-			matchLength += ZSTD_count_2segments(ip + matchLength, match + matchLength, iend, dictEnd, prefixStart);
-			if (matchIndex + matchLength >= dictLimit)
-				match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
-		}
-
-		if (matchLength > bestLength) {
-			bestLength = matchLength;
-			if (matchLength > matchEndIdx - matchIndex)
-				matchEndIdx = matchIndex + (U32)matchLength;
-		}
-
-		if (ip + matchLength == iend) /* equal : no way to know if inf or sup */
-			break;		      /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt the tree */
-
-		if (match[matchLength] < ip[matchLength]) { /* necessarily within correct buffer */
-			/* match is smaller than curr */
-			*smallerPtr = matchIndex;	  /* update smaller idx */
-			commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
-			if (matchIndex <= btLow) {
-				smallerPtr = &dummy32;
-				break;
-			}			  /* beyond tree size, stop the search */
-			smallerPtr = nextPtr + 1; /* new "smaller" => larger of match */
-			matchIndex = nextPtr[1];  /* new matchIndex larger than previous (closer to curr) */
-		} else {
-			/* match is larger than curr */
-			*largerPtr = matchIndex;
-			commonLengthLarger = matchLength;
-			if (matchIndex <= btLow) {
-				largerPtr = &dummy32;
-				break;
-			} /* beyond tree size, stop the search */
-			largerPtr = nextPtr;
-			matchIndex = nextPtr[0];
-		}
-	}
-
-	*smallerPtr = *largerPtr = 0;
-	if (bestLength > 384)
-		return MIN(192, (U32)(bestLength - 384)); /* speed optimization */
-	if (matchEndIdx > curr + 8)
-		return matchEndIdx - curr - 8;
-	return 1;
-}
-
-static size_t ZSTD_insertBtAndFindBestMatch(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iend, size_t *offsetPtr, U32 nbCompares, const U32 mls,
-					    U32 extDict)
-{
-	U32 *const hashTable = zc->hashTable;
-	U32 const hashLog = zc->params.cParams.hashLog;
-	size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
-	U32 *const bt = zc->chainTable;
-	U32 const btLog = zc->params.cParams.chainLog - 1;
-	U32 const btMask = (1 << btLog) - 1;
-	U32 matchIndex = hashTable[h];
-	size_t commonLengthSmaller = 0, commonLengthLarger = 0;
-	const BYTE *const base = zc->base;
-	const BYTE *const dictBase = zc->dictBase;
-	const U32 dictLimit = zc->dictLimit;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-	const BYTE *const prefixStart = base + dictLimit;
-	const U32 curr = (U32)(ip - base);
-	const U32 btLow = btMask >= curr ? 0 : curr - btMask;
-	const U32 windowLow = zc->lowLimit;
-	U32 *smallerPtr = bt + 2 * (curr & btMask);
-	U32 *largerPtr = bt + 2 * (curr & btMask) + 1;
-	U32 matchEndIdx = curr + 8;
-	U32 dummy32; /* to be nullified at the end */
-	size_t bestLength = 0;
-
-	hashTable[h] = curr; /* Update Hash Table */
-
-	while (nbCompares-- && (matchIndex > windowLow)) {
-		U32 *const nextPtr = bt + 2 * (matchIndex & btMask);
-		size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
-		const BYTE *match;
-
-		if ((!extDict) || (matchIndex + matchLength >= dictLimit)) {
-			match = base + matchIndex;
-			if (match[matchLength] == ip[matchLength])
-				matchLength += ZSTD_count(ip + matchLength + 1, match + matchLength + 1, iend) + 1;
-		} else {
-			match = dictBase + matchIndex;
-			matchLength += ZSTD_count_2segments(ip + matchLength, match + matchLength, iend, dictEnd, prefixStart);
-			if (matchIndex + matchLength >= dictLimit)
-				match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
-		}
-
-		if (matchLength > bestLength) {
-			if (matchLength > matchEndIdx - matchIndex)
-				matchEndIdx = matchIndex + (U32)matchLength;
-			if ((4 * (int)(matchLength - bestLength)) > (int)(ZSTD_highbit32(curr - matchIndex + 1) - ZSTD_highbit32((U32)offsetPtr[0] + 1)))
-				bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + curr - matchIndex;
-			if (ip + matchLength == iend) /* equal : no way to know if inf or sup */
-				break;		      /* drop, to guarantee consistency (miss a little bit of compression) */
-		}
-
-		if (match[matchLength] < ip[matchLength]) {
-			/* match is smaller than curr */
-			*smallerPtr = matchIndex;	  /* update smaller idx */
-			commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
-			if (matchIndex <= btLow) {
-				smallerPtr = &dummy32;
-				break;
-			}			  /* beyond tree size, stop the search */
-			smallerPtr = nextPtr + 1; /* new "smaller" => larger of match */
-			matchIndex = nextPtr[1];  /* new matchIndex larger than previous (closer to curr) */
-		} else {
-			/* match is larger than curr */
-			*largerPtr = matchIndex;
-			commonLengthLarger = matchLength;
-			if (matchIndex <= btLow) {
-				largerPtr = &dummy32;
-				break;
-			} /* beyond tree size, stop the search */
-			largerPtr = nextPtr;
-			matchIndex = nextPtr[0];
-		}
-	}
-
-	*smallerPtr = *largerPtr = 0;
-
-	zc->nextToUpdate = (matchEndIdx > curr + 8) ? matchEndIdx - 8 : curr + 1;
-	return bestLength;
-}
-
-static void ZSTD_updateTree(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iend, const U32 nbCompares, const U32 mls)
-{
-	const BYTE *const base = zc->base;
-	const U32 target = (U32)(ip - base);
-	U32 idx = zc->nextToUpdate;
-
-	while (idx < target)
-		idx += ZSTD_insertBt1(zc, base + idx, mls, iend, nbCompares, 0);
-}
-
-/** ZSTD_BtFindBestMatch() : Tree updater, providing best match */
-static size_t ZSTD_BtFindBestMatch(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts, const U32 mls)
-{
-	if (ip < zc->base + zc->nextToUpdate)
-		return 0; /* skipped area */
-	ZSTD_updateTree(zc, ip, iLimit, maxNbAttempts, mls);
-	return ZSTD_insertBtAndFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, mls, 0);
-}
-
-static size_t ZSTD_BtFindBestMatch_selectMLS(ZSTD_CCtx *zc, /* Index table will be updated */
-					     const BYTE *ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts, const U32 matchLengthSearch)
-{
-	switch (matchLengthSearch) {
-	default: /* includes case 3 */
-	case 4: return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4);
-	case 5: return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5);
-	case 7:
-	case 6: return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6);
-	}
-}
-
-static void ZSTD_updateTree_extDict(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iend, const U32 nbCompares, const U32 mls)
-{
-	const BYTE *const base = zc->base;
-	const U32 target = (U32)(ip - base);
-	U32 idx = zc->nextToUpdate;
-
-	while (idx < target)
-		idx += ZSTD_insertBt1(zc, base + idx, mls, iend, nbCompares, 1);
-}
-
-/** Tree updater, providing best match */
-static size_t ZSTD_BtFindBestMatch_extDict(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts,
-					   const U32 mls)
-{
-	if (ip < zc->base + zc->nextToUpdate)
-		return 0; /* skipped area */
-	ZSTD_updateTree_extDict(zc, ip, iLimit, maxNbAttempts, mls);
-	return ZSTD_insertBtAndFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, mls, 1);
-}
-
-static size_t ZSTD_BtFindBestMatch_selectMLS_extDict(ZSTD_CCtx *zc, /* Index table will be updated */
-						     const BYTE *ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts,
-						     const U32 matchLengthSearch)
-{
-	switch (matchLengthSearch) {
-	default: /* includes case 3 */
-	case 4: return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4);
-	case 5: return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5);
-	case 7:
-	case 6: return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6);
-	}
-}
-
-/* *********************************
-*  Hash Chain
-***********************************/
-#define NEXT_IN_CHAIN(d, mask) chainTable[(d)&mask]
-
-/* Update chains up to ip (excluded)
-   Assumption : always within prefix (i.e. not within extDict) */
-FORCE_INLINE
-U32 ZSTD_insertAndFindFirstIndex(ZSTD_CCtx *zc, const BYTE *ip, U32 mls)
-{
-	U32 *const hashTable = zc->hashTable;
-	const U32 hashLog = zc->params.cParams.hashLog;
-	U32 *const chainTable = zc->chainTable;
-	const U32 chainMask = (1 << zc->params.cParams.chainLog) - 1;
-	const BYTE *const base = zc->base;
-	const U32 target = (U32)(ip - base);
-	U32 idx = zc->nextToUpdate;
-
-	while (idx < target) { /* catch up */
-		size_t const h = ZSTD_hashPtr(base + idx, hashLog, mls);
-		NEXT_IN_CHAIN(idx, chainMask) = hashTable[h];
-		hashTable[h] = idx;
-		idx++;
-	}
-
-	zc->nextToUpdate = target;
-	return hashTable[ZSTD_hashPtr(ip, hashLog, mls)];
-}
-
-/* inlining is important to hardwire a hot branch (template emulation) */
-FORCE_INLINE
-size_t ZSTD_HcFindBestMatch_generic(ZSTD_CCtx *zc, /* Index table will be updated */
-				    const BYTE *const ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts, const U32 mls,
-				    const U32 extDict)
-{
-	U32 *const chainTable = zc->chainTable;
-	const U32 chainSize = (1 << zc->params.cParams.chainLog);
-	const U32 chainMask = chainSize - 1;
-	const BYTE *const base = zc->base;
-	const BYTE *const dictBase = zc->dictBase;
-	const U32 dictLimit = zc->dictLimit;
-	const BYTE *const prefixStart = base + dictLimit;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-	const U32 lowLimit = zc->lowLimit;
-	const U32 curr = (U32)(ip - base);
-	const U32 minChain = curr > chainSize ? curr - chainSize : 0;
-	int nbAttempts = maxNbAttempts;
-	size_t ml = EQUAL_READ32 - 1;
-
-	/* HC4 match finder */
-	U32 matchIndex = ZSTD_insertAndFindFirstIndex(zc, ip, mls);
-
-	for (; (matchIndex > lowLimit) & (nbAttempts > 0); nbAttempts--) {
-		const BYTE *match;
-		size_t currMl = 0;
-		if ((!extDict) || matchIndex >= dictLimit) {
-			match = base + matchIndex;
-			if (match[ml] == ip[ml]) /* potentially better */
-				currMl = ZSTD_count(ip, match, iLimit);
-		} else {
-			match = dictBase + matchIndex;
-			if (ZSTD_read32(match) == ZSTD_read32(ip)) /* assumption : matchIndex <= dictLimit-4 (by table construction) */
-				currMl = ZSTD_count_2segments(ip + EQUAL_READ32, match + EQUAL_READ32, iLimit, dictEnd, prefixStart) + EQUAL_READ32;
-		}
-
-		/* save best solution */
-		if (currMl > ml) {
-			ml = currMl;
-			*offsetPtr = curr - matchIndex + ZSTD_REP_MOVE;
-			if (ip + currMl == iLimit)
-				break; /* best possible, and avoid read overflow*/
-		}
-
-		if (matchIndex <= minChain)
-			break;
-		matchIndex = NEXT_IN_CHAIN(matchIndex, chainMask);
-	}
-
-	return ml;
-}
-
-FORCE_INLINE size_t ZSTD_HcFindBestMatch_selectMLS(ZSTD_CCtx *zc, const BYTE *ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts,
-						   const U32 matchLengthSearch)
-{
-	switch (matchLengthSearch) {
-	default: /* includes case 3 */
-	case 4: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4, 0);
-	case 5: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5, 0);
-	case 7:
-	case 6: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6, 0);
-	}
-}
-
-FORCE_INLINE size_t ZSTD_HcFindBestMatch_extDict_selectMLS(ZSTD_CCtx *zc, const BYTE *ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts,
-							   const U32 matchLengthSearch)
-{
-	switch (matchLengthSearch) {
-	default: /* includes case 3 */
-	case 4: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4, 1);
-	case 5: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5, 1);
-	case 7:
-	case 6: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6, 1);
-	}
-}
-
-/* *******************************
-*  Common parser - lazy strategy
-*********************************/
-FORCE_INLINE
-void ZSTD_compressBlock_lazy_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const U32 searchMethod, const U32 depth)
-{
-	seqStore_t *seqStorePtr = &(ctx->seqStore);
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - 8;
-	const BYTE *const base = ctx->base + ctx->dictLimit;
-
-	U32 const maxSearches = 1 << ctx->params.cParams.searchLog;
-	U32 const mls = ctx->params.cParams.searchLength;
-
-	typedef size_t (*searchMax_f)(ZSTD_CCtx * zc, const BYTE *ip, const BYTE *iLimit, size_t *offsetPtr, U32 maxNbAttempts, U32 matchLengthSearch);
-	searchMax_f const searchMax = searchMethod ? ZSTD_BtFindBestMatch_selectMLS : ZSTD_HcFindBestMatch_selectMLS;
-	U32 offset_1 = ctx->rep[0], offset_2 = ctx->rep[1], savedOffset = 0;
-
-	/* init */
-	ip += (ip == base);
-	ctx->nextToUpdate3 = ctx->nextToUpdate;
-	{
-		U32 const maxRep = (U32)(ip - base);
-		if (offset_2 > maxRep)
-			savedOffset = offset_2, offset_2 = 0;
-		if (offset_1 > maxRep)
-			savedOffset = offset_1, offset_1 = 0;
-	}
-
-	/* Match Loop */
-	while (ip < ilimit) {
-		size_t matchLength = 0;
-		size_t offset = 0;
-		const BYTE *start = ip + 1;
-
-		/* check repCode */
-		if ((offset_1 > 0) & (ZSTD_read32(ip + 1) == ZSTD_read32(ip + 1 - offset_1))) {
-			/* repcode : we take it */
-			matchLength = ZSTD_count(ip + 1 + EQUAL_READ32, ip + 1 + EQUAL_READ32 - offset_1, iend) + EQUAL_READ32;
-			if (depth == 0)
-				goto _storeSequence;
-		}
-
-		/* first search (depth 0) */
-		{
-			size_t offsetFound = 99999999;
-			size_t const ml2 = searchMax(ctx, ip, iend, &offsetFound, maxSearches, mls);
-			if (ml2 > matchLength)
-				matchLength = ml2, start = ip, offset = offsetFound;
-		}
-
-		if (matchLength < EQUAL_READ32) {
-			ip += ((ip - anchor) >> g_searchStrength) + 1; /* jump faster over incompressible sections */
-			continue;
-		}
-
-		/* let's try to find a better solution */
-		if (depth >= 1)
-			while (ip < ilimit) {
-				ip++;
-				if ((offset) && ((offset_1 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_1)))) {
-					size_t const mlRep = ZSTD_count(ip + EQUAL_READ32, ip + EQUAL_READ32 - offset_1, iend) + EQUAL_READ32;
-					int const gain2 = (int)(mlRep * 3);
-					int const gain1 = (int)(matchLength * 3 - ZSTD_highbit32((U32)offset + 1) + 1);
-					if ((mlRep >= EQUAL_READ32) && (gain2 > gain1))
-						matchLength = mlRep, offset = 0, start = ip;
-				}
-				{
-					size_t offset2 = 99999999;
-					size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
-					int const gain2 = (int)(ml2 * 4 - ZSTD_highbit32((U32)offset2 + 1)); /* raw approx */
-					int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 4);
-					if ((ml2 >= EQUAL_READ32) && (gain2 > gain1)) {
-						matchLength = ml2, offset = offset2, start = ip;
-						continue; /* search a better one */
-					}
-				}
-
-				/* let's find an even better one */
-				if ((depth == 2) && (ip < ilimit)) {
-					ip++;
-					if ((offset) && ((offset_1 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_1)))) {
-						size_t const ml2 = ZSTD_count(ip + EQUAL_READ32, ip + EQUAL_READ32 - offset_1, iend) + EQUAL_READ32;
-						int const gain2 = (int)(ml2 * 4);
-						int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 1);
-						if ((ml2 >= EQUAL_READ32) && (gain2 > gain1))
-							matchLength = ml2, offset = 0, start = ip;
-					}
-					{
-						size_t offset2 = 99999999;
-						size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
-						int const gain2 = (int)(ml2 * 4 - ZSTD_highbit32((U32)offset2 + 1)); /* raw approx */
-						int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 7);
-						if ((ml2 >= EQUAL_READ32) && (gain2 > gain1)) {
-							matchLength = ml2, offset = offset2, start = ip;
-							continue;
-						}
-					}
-				}
-				break; /* nothing found : store previous solution */
-			}
-
-		/* NOTE:
-		 * start[-offset+ZSTD_REP_MOVE-1] is undefined behavior.
-		 * (-offset+ZSTD_REP_MOVE-1) is unsigned, and is added to start, which
-		 * overflows the pointer, which is undefined behavior.
-		 */
-		/* catch up */
-		if (offset) {
-			while ((start > anchor) && (start > base + offset - ZSTD_REP_MOVE) &&
-			       (start[-1] == (start-offset+ZSTD_REP_MOVE)[-1])) /* only search for offset within prefix */
-			{
-				start--;
-				matchLength++;
-			}
-			offset_2 = offset_1;
-			offset_1 = (U32)(offset - ZSTD_REP_MOVE);
-		}
-
-	/* store sequence */
-_storeSequence:
-		{
-			size_t const litLength = start - anchor;
-			ZSTD_storeSeq(seqStorePtr, litLength, anchor, (U32)offset, matchLength - MINMATCH);
-			anchor = ip = start + matchLength;
-		}
-
-		/* check immediate repcode */
-		while ((ip <= ilimit) && ((offset_2 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_2)))) {
-			/* store sequence */
-			matchLength = ZSTD_count(ip + EQUAL_READ32, ip + EQUAL_READ32 - offset_2, iend) + EQUAL_READ32;
-			offset = offset_2;
-			offset_2 = offset_1;
-			offset_1 = (U32)offset; /* swap repcodes */
-			ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, matchLength - MINMATCH);
-			ip += matchLength;
-			anchor = ip;
-			continue; /* faster when present ... (?) */
-		}
-	}
-
-	/* Save reps for next block */
-	ctx->repToConfirm[0] = offset_1 ? offset_1 : savedOffset;
-	ctx->repToConfirm[1] = offset_2 ? offset_2 : savedOffset;
-
-	/* Last Literals */
-	{
-		size_t const lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-static void ZSTD_compressBlock_btlazy2(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 1, 2); }
-
-static void ZSTD_compressBlock_lazy2(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 2); }
-
-static void ZSTD_compressBlock_lazy(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 1); }
-
-static void ZSTD_compressBlock_greedy(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 0); }
-
-FORCE_INLINE
-void ZSTD_compressBlock_lazy_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const U32 searchMethod, const U32 depth)
-{
-	seqStore_t *seqStorePtr = &(ctx->seqStore);
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - 8;
-	const BYTE *const base = ctx->base;
-	const U32 dictLimit = ctx->dictLimit;
-	const U32 lowestIndex = ctx->lowLimit;
-	const BYTE *const prefixStart = base + dictLimit;
-	const BYTE *const dictBase = ctx->dictBase;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-	const BYTE *const dictStart = dictBase + ctx->lowLimit;
-
-	const U32 maxSearches = 1 << ctx->params.cParams.searchLog;
-	const U32 mls = ctx->params.cParams.searchLength;
-
-	typedef size_t (*searchMax_f)(ZSTD_CCtx * zc, const BYTE *ip, const BYTE *iLimit, size_t *offsetPtr, U32 maxNbAttempts, U32 matchLengthSearch);
-	searchMax_f searchMax = searchMethod ? ZSTD_BtFindBestMatch_selectMLS_extDict : ZSTD_HcFindBestMatch_extDict_selectMLS;
-
-	U32 offset_1 = ctx->rep[0], offset_2 = ctx->rep[1];
-
-	/* init */
-	ctx->nextToUpdate3 = ctx->nextToUpdate;
-	ip += (ip == prefixStart);
-
-	/* Match Loop */
-	while (ip < ilimit) {
-		size_t matchLength = 0;
-		size_t offset = 0;
-		const BYTE *start = ip + 1;
-		U32 curr = (U32)(ip - base);
-
-		/* check repCode */
-		{
-			const U32 repIndex = (U32)(curr + 1 - offset_1);
-			const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
-			const BYTE *const repMatch = repBase + repIndex;
-			if (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
-				if (ZSTD_read32(ip + 1) == ZSTD_read32(repMatch)) {
-					/* repcode detected we should take it */
-					const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
-					matchLength =
-					    ZSTD_count_2segments(ip + 1 + EQUAL_READ32, repMatch + EQUAL_READ32, iend, repEnd, prefixStart) + EQUAL_READ32;
-					if (depth == 0)
-						goto _storeSequence;
-				}
-		}
-
-		/* first search (depth 0) */
-		{
-			size_t offsetFound = 99999999;
-			size_t const ml2 = searchMax(ctx, ip, iend, &offsetFound, maxSearches, mls);
-			if (ml2 > matchLength)
-				matchLength = ml2, start = ip, offset = offsetFound;
-		}
-
-		if (matchLength < EQUAL_READ32) {
-			ip += ((ip - anchor) >> g_searchStrength) + 1; /* jump faster over incompressible sections */
-			continue;
-		}
-
-		/* let's try to find a better solution */
-		if (depth >= 1)
-			while (ip < ilimit) {
-				ip++;
-				curr++;
-				/* check repCode */
-				if (offset) {
-					const U32 repIndex = (U32)(curr - offset_1);
-					const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
-					const BYTE *const repMatch = repBase + repIndex;
-					if (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
-						if (ZSTD_read32(ip) == ZSTD_read32(repMatch)) {
-							/* repcode detected */
-							const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
-							size_t const repLength =
-							    ZSTD_count_2segments(ip + EQUAL_READ32, repMatch + EQUAL_READ32, iend, repEnd, prefixStart) +
-							    EQUAL_READ32;
-							int const gain2 = (int)(repLength * 3);
-							int const gain1 = (int)(matchLength * 3 - ZSTD_highbit32((U32)offset + 1) + 1);
-							if ((repLength >= EQUAL_READ32) && (gain2 > gain1))
-								matchLength = repLength, offset = 0, start = ip;
-						}
-				}
-
-				/* search match, depth 1 */
-				{
-					size_t offset2 = 99999999;
-					size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
-					int const gain2 = (int)(ml2 * 4 - ZSTD_highbit32((U32)offset2 + 1)); /* raw approx */
-					int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 4);
-					if ((ml2 >= EQUAL_READ32) && (gain2 > gain1)) {
-						matchLength = ml2, offset = offset2, start = ip;
-						continue; /* search a better one */
-					}
-				}
-
-				/* let's find an even better one */
-				if ((depth == 2) && (ip < ilimit)) {
-					ip++;
-					curr++;
-					/* check repCode */
-					if (offset) {
-						const U32 repIndex = (U32)(curr - offset_1);
-						const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
-						const BYTE *const repMatch = repBase + repIndex;
-						if (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
-							if (ZSTD_read32(ip) == ZSTD_read32(repMatch)) {
-								/* repcode detected */
-								const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
-								size_t repLength = ZSTD_count_2segments(ip + EQUAL_READ32, repMatch + EQUAL_READ32, iend,
-													repEnd, prefixStart) +
-										   EQUAL_READ32;
-								int gain2 = (int)(repLength * 4);
-								int gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 1);
-								if ((repLength >= EQUAL_READ32) && (gain2 > gain1))
-									matchLength = repLength, offset = 0, start = ip;
-							}
-					}
-
-					/* search match, depth 2 */
-					{
-						size_t offset2 = 99999999;
-						size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
-						int const gain2 = (int)(ml2 * 4 - ZSTD_highbit32((U32)offset2 + 1)); /* raw approx */
-						int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 7);
-						if ((ml2 >= EQUAL_READ32) && (gain2 > gain1)) {
-							matchLength = ml2, offset = offset2, start = ip;
-							continue;
-						}
-					}
-				}
-				break; /* nothing found : store previous solution */
-			}
-
-		/* catch up */
-		if (offset) {
-			U32 const matchIndex = (U32)((start - base) - (offset - ZSTD_REP_MOVE));
-			const BYTE *match = (matchIndex < dictLimit) ? dictBase + matchIndex : base + matchIndex;
-			const BYTE *const mStart = (matchIndex < dictLimit) ? dictStart : prefixStart;
-			while ((start > anchor) && (match > mStart) && (start[-1] == match[-1])) {
-				start--;
-				match--;
-				matchLength++;
-			} /* catch up */
-			offset_2 = offset_1;
-			offset_1 = (U32)(offset - ZSTD_REP_MOVE);
-		}
-
-	/* store sequence */
-	_storeSequence : {
-		size_t const litLength = start - anchor;
-		ZSTD_storeSeq(seqStorePtr, litLength, anchor, (U32)offset, matchLength - MINMATCH);
-		anchor = ip = start + matchLength;
-	}
-
-		/* check immediate repcode */
-		while (ip <= ilimit) {
-			const U32 repIndex = (U32)((ip - base) - offset_2);
-			const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
-			const BYTE *const repMatch = repBase + repIndex;
-			if (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
-				if (ZSTD_read32(ip) == ZSTD_read32(repMatch)) {
-					/* repcode detected we should take it */
-					const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
-					matchLength =
-					    ZSTD_count_2segments(ip + EQUAL_READ32, repMatch + EQUAL_READ32, iend, repEnd, prefixStart) + EQUAL_READ32;
-					offset = offset_2;
-					offset_2 = offset_1;
-					offset_1 = (U32)offset; /* swap offset history */
-					ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, matchLength - MINMATCH);
-					ip += matchLength;
-					anchor = ip;
-					continue; /* faster when present ... (?) */
-				}
-			break;
-		}
-	}
-
-	/* Save reps for next block */
-	ctx->repToConfirm[0] = offset_1;
-	ctx->repToConfirm[1] = offset_2;
-
-	/* Last Literals */
-	{
-		size_t const lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-void ZSTD_compressBlock_greedy_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 0); }
-
-static void ZSTD_compressBlock_lazy_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-	ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 1);
-}
-
-static void ZSTD_compressBlock_lazy2_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-	ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 2);
-}
-
-static void ZSTD_compressBlock_btlazy2_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-	ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 1, 2);
-}
-
-/* The optimal parser */
-#include "zstd_opt.h"
-
-static void ZSTD_compressBlock_btopt(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-#ifdef ZSTD_OPT_H_91842398743
-	ZSTD_compressBlock_opt_generic(ctx, src, srcSize, 0);
-#else
-	(void)ctx;
-	(void)src;
-	(void)srcSize;
-	return;
-#endif
-}
-
-static void ZSTD_compressBlock_btopt2(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-#ifdef ZSTD_OPT_H_91842398743
-	ZSTD_compressBlock_opt_generic(ctx, src, srcSize, 1);
-#else
-	(void)ctx;
-	(void)src;
-	(void)srcSize;
-	return;
-#endif
-}
-
-static void ZSTD_compressBlock_btopt_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-#ifdef ZSTD_OPT_H_91842398743
-	ZSTD_compressBlock_opt_extDict_generic(ctx, src, srcSize, 0);
-#else
-	(void)ctx;
-	(void)src;
-	(void)srcSize;
-	return;
-#endif
-}
-
-static void ZSTD_compressBlock_btopt2_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
-{
-#ifdef ZSTD_OPT_H_91842398743
-	ZSTD_compressBlock_opt_extDict_generic(ctx, src, srcSize, 1);
-#else
-	(void)ctx;
-	(void)src;
-	(void)srcSize;
-	return;
-#endif
-}
-
-typedef void (*ZSTD_blockCompressor)(ZSTD_CCtx *ctx, const void *src, size_t srcSize);
-
-static ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, int extDict)
-{
-	static const ZSTD_blockCompressor blockCompressor[2][8] = {
-	    {ZSTD_compressBlock_fast, ZSTD_compressBlock_doubleFast, ZSTD_compressBlock_greedy, ZSTD_compressBlock_lazy, ZSTD_compressBlock_lazy2,
-	     ZSTD_compressBlock_btlazy2, ZSTD_compressBlock_btopt, ZSTD_compressBlock_btopt2},
-	    {ZSTD_compressBlock_fast_extDict, ZSTD_compressBlock_doubleFast_extDict, ZSTD_compressBlock_greedy_extDict, ZSTD_compressBlock_lazy_extDict,
-	     ZSTD_compressBlock_lazy2_extDict, ZSTD_compressBlock_btlazy2_extDict, ZSTD_compressBlock_btopt_extDict, ZSTD_compressBlock_btopt2_extDict}};
-
-	return blockCompressor[extDict][(U32)strat];
-}
-
-static size_t ZSTD_compressBlock_internal(ZSTD_CCtx *zc, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	ZSTD_blockCompressor const blockCompressor = ZSTD_selectBlockCompressor(zc->params.cParams.strategy, zc->lowLimit < zc->dictLimit);
-	const BYTE *const base = zc->base;
-	const BYTE *const istart = (const BYTE *)src;
-	const U32 curr = (U32)(istart - base);
-	if (srcSize < MIN_CBLOCK_SIZE + ZSTD_blockHeaderSize + 1)
-		return 0; /* don't even attempt compression below a certain srcSize */
-	ZSTD_resetSeqStore(&(zc->seqStore));
-	if (curr > zc->nextToUpdate + 384)
-		zc->nextToUpdate = curr - MIN(192, (U32)(curr - zc->nextToUpdate - 384)); /* update tree not updated after finding very long rep matches */
-	blockCompressor(zc, src, srcSize);
-	return ZSTD_compressSequences(zc, dst, dstCapacity, srcSize);
-}
-
-/*! ZSTD_compress_generic() :
-*   Compress a chunk of data into one or multiple blocks.
-*   All blocks will be terminated, all input will be consumed.
-*   Function will issue an error if there is not enough `dstCapacity` to hold the compressed content.
-*   Frame is supposed already started (header already produced)
-*   @return : compressed size, or an error code
-*/
-static size_t ZSTD_compress_generic(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, U32 lastFrameChunk)
-{
-	size_t blockSize = cctx->blockSize;
-	size_t remaining = srcSize;
-	const BYTE *ip = (const BYTE *)src;
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *op = ostart;
-	U32 const maxDist = 1 << cctx->params.cParams.windowLog;
-
-	if (cctx->params.fParams.checksumFlag && srcSize)
-		xxh64_update(&cctx->xxhState, src, srcSize);
-
-	while (remaining) {
-		U32 const lastBlock = lastFrameChunk & (blockSize >= remaining);
-		size_t cSize;
-
-		if (dstCapacity < ZSTD_blockHeaderSize + MIN_CBLOCK_SIZE)
-			return ERROR(dstSize_tooSmall); /* not enough space to store compressed block */
-		if (remaining < blockSize)
-			blockSize = remaining;
-
-		/* preemptive overflow correction */
-		if (cctx->lowLimit > (3U << 29)) {
-			U32 const cycleMask = (1 << ZSTD_cycleLog(cctx->params.cParams.hashLog, cctx->params.cParams.strategy)) - 1;
-			U32 const curr = (U32)(ip - cctx->base);
-			U32 const newCurr = (curr & cycleMask) + (1 << cctx->params.cParams.windowLog);
-			U32 const correction = curr - newCurr;
-			ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX_64 <= 30);
-			ZSTD_reduceIndex(cctx, correction);
-			cctx->base += correction;
-			cctx->dictBase += correction;
-			cctx->lowLimit -= correction;
-			cctx->dictLimit -= correction;
-			if (cctx->nextToUpdate < correction)
-				cctx->nextToUpdate = 0;
-			else
-				cctx->nextToUpdate -= correction;
-		}
-
-		if ((U32)(ip + blockSize - cctx->base) > cctx->loadedDictEnd + maxDist) {
-			/* enforce maxDist */
-			U32 const newLowLimit = (U32)(ip + blockSize - cctx->base) - maxDist;
-			if (cctx->lowLimit < newLowLimit)
-				cctx->lowLimit = newLowLimit;
-			if (cctx->dictLimit < cctx->lowLimit)
-				cctx->dictLimit = cctx->lowLimit;
-		}
-
-		cSize = ZSTD_compressBlock_internal(cctx, op + ZSTD_blockHeaderSize, dstCapacity - ZSTD_blockHeaderSize, ip, blockSize);
-		if (ZSTD_isError(cSize))
-			return cSize;
-
-		if (cSize == 0) { /* block is not compressible */
-			U32 const cBlockHeader24 = lastBlock + (((U32)bt_raw) << 1) + (U32)(blockSize << 3);
-			if (blockSize + ZSTD_blockHeaderSize > dstCapacity)
-				return ERROR(dstSize_tooSmall);
-			ZSTD_writeLE32(op, cBlockHeader24); /* no pb, 4th byte will be overwritten */
-			memcpy(op + ZSTD_blockHeaderSize, ip, blockSize);
-			cSize = ZSTD_blockHeaderSize + blockSize;
-		} else {
-			U32 const cBlockHeader24 = lastBlock + (((U32)bt_compressed) << 1) + (U32)(cSize << 3);
-			ZSTD_writeLE24(op, cBlockHeader24);
-			cSize += ZSTD_blockHeaderSize;
-		}
-
-		remaining -= blockSize;
-		dstCapacity -= cSize;
-		ip += blockSize;
-		op += cSize;
-	}
-
-	if (lastFrameChunk && (op > ostart))
-		cctx->stage = ZSTDcs_ending;
-	return op - ostart;
-}
-
-static size_t ZSTD_writeFrameHeader(void *dst, size_t dstCapacity, ZSTD_parameters params, U64 pledgedSrcSize, U32 dictID)
-{
-	BYTE *const op = (BYTE *)dst;
-	U32 const dictIDSizeCode = (dictID > 0) + (dictID >= 256) + (dictID >= 65536); /* 0-3 */
-	U32 const checksumFlag = params.fParams.checksumFlag > 0;
-	U32 const windowSize = 1U << params.cParams.windowLog;
-	U32 const singleSegment = params.fParams.contentSizeFlag && (windowSize >= pledgedSrcSize);
-	BYTE const windowLogByte = (BYTE)((params.cParams.windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN) << 3);
-	U32 const fcsCode =
-	    params.fParams.contentSizeFlag ? (pledgedSrcSize >= 256) + (pledgedSrcSize >= 65536 + 256) + (pledgedSrcSize >= 0xFFFFFFFFU) : 0; /* 0-3 */
-	BYTE const frameHeaderDecriptionByte = (BYTE)(dictIDSizeCode + (checksumFlag << 2) + (singleSegment << 5) + (fcsCode << 6));
-	size_t pos;
-
-	if (dstCapacity < ZSTD_frameHeaderSize_max)
-		return ERROR(dstSize_tooSmall);
-
-	ZSTD_writeLE32(dst, ZSTD_MAGICNUMBER);
-	op[4] = frameHeaderDecriptionByte;
-	pos = 5;
-	if (!singleSegment)
-		op[pos++] = windowLogByte;
-	switch (dictIDSizeCode) {
-	default: /* impossible */
-	case 0: break;
-	case 1:
-		op[pos] = (BYTE)(dictID);
-		pos++;
-		break;
-	case 2:
-		ZSTD_writeLE16(op + pos, (U16)dictID);
-		pos += 2;
-		break;
-	case 3:
-		ZSTD_writeLE32(op + pos, dictID);
-		pos += 4;
-		break;
-	}
-	switch (fcsCode) {
-	default: /* impossible */
-	case 0:
-		if (singleSegment)
-			op[pos++] = (BYTE)(pledgedSrcSize);
-		break;
-	case 1:
-		ZSTD_writeLE16(op + pos, (U16)(pledgedSrcSize - 256));
-		pos += 2;
-		break;
-	case 2:
-		ZSTD_writeLE32(op + pos, (U32)(pledgedSrcSize));
-		pos += 4;
-		break;
-	case 3:
-		ZSTD_writeLE64(op + pos, (U64)(pledgedSrcSize));
-		pos += 8;
-		break;
-	}
-	return pos;
-}
-
-static size_t ZSTD_compressContinue_internal(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, U32 frame, U32 lastFrameChunk)
-{
-	const BYTE *const ip = (const BYTE *)src;
-	size_t fhSize = 0;
-
-	if (cctx->stage == ZSTDcs_created)
-		return ERROR(stage_wrong); /* missing init (ZSTD_compressBegin) */
-
-	if (frame && (cctx->stage == ZSTDcs_init)) {
-		fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, cctx->params, cctx->frameContentSize, cctx->dictID);
-		if (ZSTD_isError(fhSize))
-			return fhSize;
-		dstCapacity -= fhSize;
-		dst = (char *)dst + fhSize;
-		cctx->stage = ZSTDcs_ongoing;
-	}
-
-	/* Check if blocks follow each other */
-	if (src != cctx->nextSrc) {
-		/* not contiguous */
-		ptrdiff_t const delta = cctx->nextSrc - ip;
-		cctx->lowLimit = cctx->dictLimit;
-		cctx->dictLimit = (U32)(cctx->nextSrc - cctx->base);
-		cctx->dictBase = cctx->base;
-		cctx->base -= delta;
-		cctx->nextToUpdate = cctx->dictLimit;
-		if (cctx->dictLimit - cctx->lowLimit < HASH_READ_SIZE)
-			cctx->lowLimit = cctx->dictLimit; /* too small extDict */
-	}
-
-	/* if input and dictionary overlap : reduce dictionary (area presumed modified by input) */
-	if ((ip + srcSize > cctx->dictBase + cctx->lowLimit) & (ip < cctx->dictBase + cctx->dictLimit)) {
-		ptrdiff_t const highInputIdx = (ip + srcSize) - cctx->dictBase;
-		U32 const lowLimitMax = (highInputIdx > (ptrdiff_t)cctx->dictLimit) ? cctx->dictLimit : (U32)highInputIdx;
-		cctx->lowLimit = lowLimitMax;
-	}
-
-	cctx->nextSrc = ip + srcSize;
-
-	if (srcSize) {
-		size_t const cSize = frame ? ZSTD_compress_generic(cctx, dst, dstCapacity, src, srcSize, lastFrameChunk)
-					   : ZSTD_compressBlock_internal(cctx, dst, dstCapacity, src, srcSize);
-		if (ZSTD_isError(cSize))
-			return cSize;
-		return cSize + fhSize;
-	} else
-		return fhSize;
-}
-
-size_t ZSTD_compressContinue(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 1, 0);
-}
-
-size_t ZSTD_getBlockSizeMax(ZSTD_CCtx *cctx) { return MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, 1 << cctx->params.cParams.windowLog); }
-
-size_t ZSTD_compressBlock(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	size_t const blockSizeMax = ZSTD_getBlockSizeMax(cctx);
-	if (srcSize > blockSizeMax)
-		return ERROR(srcSize_wrong);
-	return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 0, 0);
-}
-
-/*! ZSTD_loadDictionaryContent() :
- *  @return : 0, or an error code
- */
-static size_t ZSTD_loadDictionaryContent(ZSTD_CCtx *zc, const void *src, size_t srcSize)
-{
-	const BYTE *const ip = (const BYTE *)src;
-	const BYTE *const iend = ip + srcSize;
-
-	/* input becomes curr prefix */
-	zc->lowLimit = zc->dictLimit;
-	zc->dictLimit = (U32)(zc->nextSrc - zc->base);
-	zc->dictBase = zc->base;
-	zc->base += ip - zc->nextSrc;
-	zc->nextToUpdate = zc->dictLimit;
-	zc->loadedDictEnd = zc->forceWindow ? 0 : (U32)(iend - zc->base);
-
-	zc->nextSrc = iend;
-	if (srcSize <= HASH_READ_SIZE)
-		return 0;
-
-	switch (zc->params.cParams.strategy) {
-	case ZSTD_fast: ZSTD_fillHashTable(zc, iend, zc->params.cParams.searchLength); break;
-
-	case ZSTD_dfast: ZSTD_fillDoubleHashTable(zc, iend, zc->params.cParams.searchLength); break;
-
-	case ZSTD_greedy:
-	case ZSTD_lazy:
-	case ZSTD_lazy2:
-		if (srcSize >= HASH_READ_SIZE)
-			ZSTD_insertAndFindFirstIndex(zc, iend - HASH_READ_SIZE, zc->params.cParams.searchLength);
-		break;
-
-	case ZSTD_btlazy2:
-	case ZSTD_btopt:
-	case ZSTD_btopt2:
-		if (srcSize >= HASH_READ_SIZE)
-			ZSTD_updateTree(zc, iend - HASH_READ_SIZE, iend, 1 << zc->params.cParams.searchLog, zc->params.cParams.searchLength);
-		break;
-
-	default:
-		return ERROR(GENERIC); /* strategy doesn't exist; impossible */
-	}
-
-	zc->nextToUpdate = (U32)(iend - zc->base);
-	return 0;
-}
-
-/* Dictionaries that assign zero probability to symbols that show up causes problems
-   when FSE encoding.  Refuse dictionaries that assign zero probability to symbols
-   that we may encounter during compression.
-   NOTE: This behavior is not standard and could be improved in the future. */
-static size_t ZSTD_checkDictNCount(short *normalizedCounter, unsigned dictMaxSymbolValue, unsigned maxSymbolValue)
-{
-	U32 s;
-	if (dictMaxSymbolValue < maxSymbolValue)
-		return ERROR(dictionary_corrupted);
-	for (s = 0; s <= maxSymbolValue; ++s) {
-		if (normalizedCounter[s] == 0)
-			return ERROR(dictionary_corrupted);
-	}
-	return 0;
-}
-
-/* Dictionary format :
- * See :
- * https://github.com/facebook/zstd/blob/master/doc/zstd_compression_format.md#dictionary-format
- */
-/*! ZSTD_loadZstdDictionary() :
- * @return : 0, or an error code
- *  assumptions : magic number supposed already checked
- *                dictSize supposed > 8
- */
-static size_t ZSTD_loadZstdDictionary(ZSTD_CCtx *cctx, const void *dict, size_t dictSize)
-{
-	const BYTE *dictPtr = (const BYTE *)dict;
-	const BYTE *const dictEnd = dictPtr + dictSize;
-	short offcodeNCount[MaxOff + 1];
-	unsigned offcodeMaxValue = MaxOff;
-
-	dictPtr += 4; /* skip magic number */
-	cctx->dictID = cctx->params.fParams.noDictIDFlag ? 0 : ZSTD_readLE32(dictPtr);
-	dictPtr += 4;
-
-	{
-		size_t const hufHeaderSize = HUF_readCTable_wksp(cctx->hufTable, 255, dictPtr, dictEnd - dictPtr, cctx->tmpCounters, sizeof(cctx->tmpCounters));
-		if (HUF_isError(hufHeaderSize))
-			return ERROR(dictionary_corrupted);
-		dictPtr += hufHeaderSize;
-	}
-
-	{
-		unsigned offcodeLog;
-		size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(offcodeHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (offcodeLog > OffFSELog)
-			return ERROR(dictionary_corrupted);
-		/* Defer checking offcodeMaxValue because we need to know the size of the dictionary content */
-		CHECK_E(FSE_buildCTable_wksp(cctx->offcodeCTable, offcodeNCount, offcodeMaxValue, offcodeLog, cctx->tmpCounters, sizeof(cctx->tmpCounters)),
-			dictionary_corrupted);
-		dictPtr += offcodeHeaderSize;
-	}
-
-	{
-		short matchlengthNCount[MaxML + 1];
-		unsigned matchlengthMaxValue = MaxML, matchlengthLog;
-		size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(matchlengthHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (matchlengthLog > MLFSELog)
-			return ERROR(dictionary_corrupted);
-		/* Every match length code must have non-zero probability */
-		CHECK_F(ZSTD_checkDictNCount(matchlengthNCount, matchlengthMaxValue, MaxML));
-		CHECK_E(
-		    FSE_buildCTable_wksp(cctx->matchlengthCTable, matchlengthNCount, matchlengthMaxValue, matchlengthLog, cctx->tmpCounters, sizeof(cctx->tmpCounters)),
-		    dictionary_corrupted);
-		dictPtr += matchlengthHeaderSize;
-	}
-
-	{
-		short litlengthNCount[MaxLL + 1];
-		unsigned litlengthMaxValue = MaxLL, litlengthLog;
-		size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(litlengthHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (litlengthLog > LLFSELog)
-			return ERROR(dictionary_corrupted);
-		/* Every literal length code must have non-zero probability */
-		CHECK_F(ZSTD_checkDictNCount(litlengthNCount, litlengthMaxValue, MaxLL));
-		CHECK_E(FSE_buildCTable_wksp(cctx->litlengthCTable, litlengthNCount, litlengthMaxValue, litlengthLog, cctx->tmpCounters, sizeof(cctx->tmpCounters)),
-			dictionary_corrupted);
-		dictPtr += litlengthHeaderSize;
-	}
-
-	if (dictPtr + 12 > dictEnd)
-		return ERROR(dictionary_corrupted);
-	cctx->rep[0] = ZSTD_readLE32(dictPtr + 0);
-	cctx->rep[1] = ZSTD_readLE32(dictPtr + 4);
-	cctx->rep[2] = ZSTD_readLE32(dictPtr + 8);
-	dictPtr += 12;
-
-	{
-		size_t const dictContentSize = (size_t)(dictEnd - dictPtr);
-		U32 offcodeMax = MaxOff;
-		if (dictContentSize <= ((U32)-1) - 128 KB) {
-			U32 const maxOffset = (U32)dictContentSize + 128 KB; /* The maximum offset that must be supported */
-			offcodeMax = ZSTD_highbit32(maxOffset);		     /* Calculate minimum offset code required to represent maxOffset */
-		}
-		/* All offset values <= dictContentSize + 128 KB must be representable */
-		CHECK_F(ZSTD_checkDictNCount(offcodeNCount, offcodeMaxValue, MIN(offcodeMax, MaxOff)));
-		/* All repCodes must be <= dictContentSize and != 0*/
-		{
-			U32 u;
-			for (u = 0; u < 3; u++) {
-				if (cctx->rep[u] == 0)
-					return ERROR(dictionary_corrupted);
-				if (cctx->rep[u] > dictContentSize)
-					return ERROR(dictionary_corrupted);
-			}
-		}
-
-		cctx->flagStaticTables = 1;
-		cctx->flagStaticHufTable = HUF_repeat_valid;
-		return ZSTD_loadDictionaryContent(cctx, dictPtr, dictContentSize);
-	}
-}
-
-/** ZSTD_compress_insertDictionary() :
-*   @return : 0, or an error code */
-static size_t ZSTD_compress_insertDictionary(ZSTD_CCtx *cctx, const void *dict, size_t dictSize)
-{
-	if ((dict == NULL) || (dictSize <= 8))
-		return 0;
-
-	/* dict as pure content */
-	if ((ZSTD_readLE32(dict) != ZSTD_DICT_MAGIC) || (cctx->forceRawDict))
-		return ZSTD_loadDictionaryContent(cctx, dict, dictSize);
-
-	/* dict as zstd dictionary */
-	return ZSTD_loadZstdDictionary(cctx, dict, dictSize);
-}
-
-/*! ZSTD_compressBegin_internal() :
-*   @return : 0, or an error code */
-static size_t ZSTD_compressBegin_internal(ZSTD_CCtx *cctx, const void *dict, size_t dictSize, ZSTD_parameters params, U64 pledgedSrcSize)
-{
-	ZSTD_compResetPolicy_e const crp = dictSize ? ZSTDcrp_fullReset : ZSTDcrp_continue;
-	CHECK_F(ZSTD_resetCCtx_advanced(cctx, params, pledgedSrcSize, crp));
-	return ZSTD_compress_insertDictionary(cctx, dict, dictSize);
-}
-
-/*! ZSTD_compressBegin_advanced() :
-*   @return : 0, or an error code */
-size_t ZSTD_compressBegin_advanced(ZSTD_CCtx *cctx, const void *dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize)
-{
-	/* compression parameters verification and optimization */
-	CHECK_F(ZSTD_checkCParams(params.cParams));
-	return ZSTD_compressBegin_internal(cctx, dict, dictSize, params, pledgedSrcSize);
-}
-
-size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx *cctx, const void *dict, size_t dictSize, int compressionLevel)
-{
-	ZSTD_parameters const params = ZSTD_getParams(compressionLevel, 0, dictSize);
-	return ZSTD_compressBegin_internal(cctx, dict, dictSize, params, 0);
-}
-
-size_t ZSTD_compressBegin(ZSTD_CCtx *cctx, int compressionLevel) { return ZSTD_compressBegin_usingDict(cctx, NULL, 0, compressionLevel); }
-
-/*! ZSTD_writeEpilogue() :
-*   Ends a frame.
-*   @return : nb of bytes written into dst (or an error code) */
-static size_t ZSTD_writeEpilogue(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity)
-{
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *op = ostart;
-	size_t fhSize = 0;
-
-	if (cctx->stage == ZSTDcs_created)
-		return ERROR(stage_wrong); /* init missing */
-
-	/* special case : empty frame */
-	if (cctx->stage == ZSTDcs_init) {
-		fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, cctx->params, 0, 0);
-		if (ZSTD_isError(fhSize))
-			return fhSize;
-		dstCapacity -= fhSize;
-		op += fhSize;
-		cctx->stage = ZSTDcs_ongoing;
-	}
-
-	if (cctx->stage != ZSTDcs_ending) {
-		/* write one last empty block, make it the "last" block */
-		U32 const cBlockHeader24 = 1 /* last block */ + (((U32)bt_raw) << 1) + 0;
-		if (dstCapacity < 4)
-			return ERROR(dstSize_tooSmall);
-		ZSTD_writeLE32(op, cBlockHeader24);
-		op += ZSTD_blockHeaderSize;
-		dstCapacity -= ZSTD_blockHeaderSize;
-	}
-
-	if (cctx->params.fParams.checksumFlag) {
-		U32 const checksum = (U32)xxh64_digest(&cctx->xxhState);
-		if (dstCapacity < 4)
-			return ERROR(dstSize_tooSmall);
-		ZSTD_writeLE32(op, checksum);
-		op += 4;
-	}
-
-	cctx->stage = ZSTDcs_created; /* return to "created but no init" status */
-	return op - ostart;
-}
-
-size_t ZSTD_compressEnd(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	size_t endResult;
-	size_t const cSize = ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 1, 1);
-	if (ZSTD_isError(cSize))
-		return cSize;
-	endResult = ZSTD_writeEpilogue(cctx, (char *)dst + cSize, dstCapacity - cSize);
-	if (ZSTD_isError(endResult))
-		return endResult;
-	return cSize + endResult;
-}
-
-static size_t ZSTD_compress_internal(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize,
-				     ZSTD_parameters params)
-{
-	CHECK_F(ZSTD_compressBegin_internal(cctx, dict, dictSize, params, srcSize));
-	return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
-}
-
-size_t ZSTD_compress_usingDict(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize,
-			       ZSTD_parameters params)
-{
-	return ZSTD_compress_internal(ctx, dst, dstCapacity, src, srcSize, dict, dictSize, params);
-}
-
-size_t ZSTD_compressCCtx(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, ZSTD_parameters params)
-{
-	return ZSTD_compress_internal(ctx, dst, dstCapacity, src, srcSize, NULL, 0, params);
-}
-
-/* =====  Dictionary API  ===== */
-
-struct ZSTD_CDict_s {
-	void *dictBuffer;
-	const void *dictContent;
-	size_t dictContentSize;
-	ZSTD_CCtx *refContext;
-}; /* typedef'd tp ZSTD_CDict within "zstd.h" */
-
-size_t ZSTD_CDictWorkspaceBound(ZSTD_compressionParameters cParams) { return ZSTD_CCtxWorkspaceBound(cParams) + ZSTD_ALIGN(sizeof(ZSTD_CDict)); }
-
-static ZSTD_CDict *ZSTD_createCDict_advanced(const void *dictBuffer, size_t dictSize, unsigned byReference, ZSTD_parameters params, ZSTD_customMem customMem)
-{
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-
-	{
-		ZSTD_CDict *const cdict = (ZSTD_CDict *)ZSTD_malloc(sizeof(ZSTD_CDict), customMem);
-		ZSTD_CCtx *const cctx = ZSTD_createCCtx_advanced(customMem);
-
-		if (!cdict || !cctx) {
-			ZSTD_free(cdict, customMem);
-			ZSTD_freeCCtx(cctx);
-			return NULL;
-		}
-
-		if ((byReference) || (!dictBuffer) || (!dictSize)) {
-			cdict->dictBuffer = NULL;
-			cdict->dictContent = dictBuffer;
-		} else {
-			void *const internalBuffer = ZSTD_malloc(dictSize, customMem);
-			if (!internalBuffer) {
-				ZSTD_free(cctx, customMem);
-				ZSTD_free(cdict, customMem);
-				return NULL;
-			}
-			memcpy(internalBuffer, dictBuffer, dictSize);
-			cdict->dictBuffer = internalBuffer;
-			cdict->dictContent = internalBuffer;
-		}
-
-		{
-			size_t const errorCode = ZSTD_compressBegin_advanced(cctx, cdict->dictContent, dictSize, params, 0);
-			if (ZSTD_isError(errorCode)) {
-				ZSTD_free(cdict->dictBuffer, customMem);
-				ZSTD_free(cdict, customMem);
-				ZSTD_freeCCtx(cctx);
-				return NULL;
-			}
-		}
-
-		cdict->refContext = cctx;
-		cdict->dictContentSize = dictSize;
-		return cdict;
-	}
-}
-
-ZSTD_CDict *ZSTD_initCDict(const void *dict, size_t dictSize, ZSTD_parameters params, void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	return ZSTD_createCDict_advanced(dict, dictSize, 1, params, stackMem);
-}
-
-size_t ZSTD_freeCDict(ZSTD_CDict *cdict)
-{
-	if (cdict == NULL)
-		return 0; /* support free on NULL */
-	{
-		ZSTD_customMem const cMem = cdict->refContext->customMem;
-		ZSTD_freeCCtx(cdict->refContext);
-		ZSTD_free(cdict->dictBuffer, cMem);
-		ZSTD_free(cdict, cMem);
-		return 0;
-	}
-}
-
-static ZSTD_parameters ZSTD_getParamsFromCDict(const ZSTD_CDict *cdict) { return ZSTD_getParamsFromCCtx(cdict->refContext); }
-
-size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx *cctx, const ZSTD_CDict *cdict, unsigned long long pledgedSrcSize)
-{
-	if (cdict->dictContentSize)
-		CHECK_F(ZSTD_copyCCtx(cctx, cdict->refContext, pledgedSrcSize))
-	else {
-		ZSTD_parameters params = cdict->refContext->params;
-		params.fParams.contentSizeFlag = (pledgedSrcSize > 0);
-		CHECK_F(ZSTD_compressBegin_advanced(cctx, NULL, 0, params, pledgedSrcSize));
-	}
-	return 0;
-}
-
-/*! ZSTD_compress_usingCDict() :
-*   Compression using a digested Dictionary.
-*   Faster startup than ZSTD_compress_usingDict(), recommended when same dictionary is used multiple times.
-*   Note that compression level is decided during dictionary creation */
-size_t ZSTD_compress_usingCDict(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const ZSTD_CDict *cdict)
-{
-	CHECK_F(ZSTD_compressBegin_usingCDict(cctx, cdict, srcSize));
-
-	if (cdict->refContext->params.fParams.contentSizeFlag == 1) {
-		cctx->params.fParams.contentSizeFlag = 1;
-		cctx->frameContentSize = srcSize;
-	} else {
-		cctx->params.fParams.contentSizeFlag = 0;
-	}
-
-	return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
-}
-
-/* ******************************************************************
-*  Streaming
-********************************************************************/
-
-typedef enum { zcss_init, zcss_load, zcss_flush, zcss_final } ZSTD_cStreamStage;
-
-struct ZSTD_CStream_s {
-	ZSTD_CCtx *cctx;
-	ZSTD_CDict *cdictLocal;
-	const ZSTD_CDict *cdict;
-	char *inBuff;
-	size_t inBuffSize;
-	size_t inToCompress;
-	size_t inBuffPos;
-	size_t inBuffTarget;
-	size_t blockSize;
-	char *outBuff;
-	size_t outBuffSize;
-	size_t outBuffContentSize;
-	size_t outBuffFlushedSize;
-	ZSTD_cStreamStage stage;
-	U32 checksum;
-	U32 frameEnded;
-	U64 pledgedSrcSize;
-	U64 inputProcessed;
-	ZSTD_parameters params;
-	ZSTD_customMem customMem;
-}; /* typedef'd to ZSTD_CStream within "zstd.h" */
-
-size_t ZSTD_CStreamWorkspaceBound(ZSTD_compressionParameters cParams)
-{
-	size_t const inBuffSize = (size_t)1 << cParams.windowLog;
-	size_t const blockSize = MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, inBuffSize);
-	size_t const outBuffSize = ZSTD_compressBound(blockSize) + 1;
-
-	return ZSTD_CCtxWorkspaceBound(cParams) + ZSTD_ALIGN(sizeof(ZSTD_CStream)) + ZSTD_ALIGN(inBuffSize) + ZSTD_ALIGN(outBuffSize);
-}
-
-ZSTD_CStream *ZSTD_createCStream_advanced(ZSTD_customMem customMem)
-{
-	ZSTD_CStream *zcs;
-
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-
-	zcs = (ZSTD_CStream *)ZSTD_malloc(sizeof(ZSTD_CStream), customMem);
-	if (zcs == NULL)
-		return NULL;
-	memset(zcs, 0, sizeof(ZSTD_CStream));
-	memcpy(&zcs->customMem, &customMem, sizeof(ZSTD_customMem));
-	zcs->cctx = ZSTD_createCCtx_advanced(customMem);
-	if (zcs->cctx == NULL) {
-		ZSTD_freeCStream(zcs);
-		return NULL;
-	}
-	return zcs;
-}
-
-size_t ZSTD_freeCStream(ZSTD_CStream *zcs)
-{
-	if (zcs == NULL)
-		return 0; /* support free on NULL */
-	{
-		ZSTD_customMem const cMem = zcs->customMem;
-		ZSTD_freeCCtx(zcs->cctx);
-		zcs->cctx = NULL;
-		ZSTD_freeCDict(zcs->cdictLocal);
-		zcs->cdictLocal = NULL;
-		ZSTD_free(zcs->inBuff, cMem);
-		zcs->inBuff = NULL;
-		ZSTD_free(zcs->outBuff, cMem);
-		zcs->outBuff = NULL;
-		ZSTD_free(zcs, cMem);
-		return 0;
-	}
-}
-
-/*======   Initialization   ======*/
-
-size_t ZSTD_CStreamInSize(void) { return ZSTD_BLOCKSIZE_ABSOLUTEMAX; }
-size_t ZSTD_CStreamOutSize(void) { return ZSTD_compressBound(ZSTD_BLOCKSIZE_ABSOLUTEMAX) + ZSTD_blockHeaderSize + 4 /* 32-bits hash */; }
-
-static size_t ZSTD_resetCStream_internal(ZSTD_CStream *zcs, unsigned long long pledgedSrcSize)
-{
-	if (zcs->inBuffSize == 0)
-		return ERROR(stage_wrong); /* zcs has not been init at least once => can't reset */
-
-	if (zcs->cdict)
-		CHECK_F(ZSTD_compressBegin_usingCDict(zcs->cctx, zcs->cdict, pledgedSrcSize))
-	else
-		CHECK_F(ZSTD_compressBegin_advanced(zcs->cctx, NULL, 0, zcs->params, pledgedSrcSize));
-
-	zcs->inToCompress = 0;
-	zcs->inBuffPos = 0;
-	zcs->inBuffTarget = zcs->blockSize;
-	zcs->outBuffContentSize = zcs->outBuffFlushedSize = 0;
-	zcs->stage = zcss_load;
-	zcs->frameEnded = 0;
-	zcs->pledgedSrcSize = pledgedSrcSize;
-	zcs->inputProcessed = 0;
-	return 0; /* ready to go */
-}
-
-size_t ZSTD_resetCStream(ZSTD_CStream *zcs, unsigned long long pledgedSrcSize)
-{
-
-	zcs->params.fParams.contentSizeFlag = (pledgedSrcSize > 0);
-
-	return ZSTD_resetCStream_internal(zcs, pledgedSrcSize);
-}
-
-static size_t ZSTD_initCStream_advanced(ZSTD_CStream *zcs, const void *dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize)
-{
-	/* allocate buffers */
-	{
-		size_t const neededInBuffSize = (size_t)1 << params.cParams.windowLog;
-		if (zcs->inBuffSize < neededInBuffSize) {
-			zcs->inBuffSize = neededInBuffSize;
-			ZSTD_free(zcs->inBuff, zcs->customMem);
-			zcs->inBuff = (char *)ZSTD_malloc(neededInBuffSize, zcs->customMem);
-			if (zcs->inBuff == NULL)
-				return ERROR(memory_allocation);
-		}
-		zcs->blockSize = MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, neededInBuffSize);
-	}
-	if (zcs->outBuffSize < ZSTD_compressBound(zcs->blockSize) + 1) {
-		zcs->outBuffSize = ZSTD_compressBound(zcs->blockSize) + 1;
-		ZSTD_free(zcs->outBuff, zcs->customMem);
-		zcs->outBuff = (char *)ZSTD_malloc(zcs->outBuffSize, zcs->customMem);
-		if (zcs->outBuff == NULL)
-			return ERROR(memory_allocation);
-	}
-
-	if (dict && dictSize >= 8) {
-		ZSTD_freeCDict(zcs->cdictLocal);
-		zcs->cdictLocal = ZSTD_createCDict_advanced(dict, dictSize, 0, params, zcs->customMem);
-		if (zcs->cdictLocal == NULL)
-			return ERROR(memory_allocation);
-		zcs->cdict = zcs->cdictLocal;
-	} else
-		zcs->cdict = NULL;
-
-	zcs->checksum = params.fParams.checksumFlag > 0;
-	zcs->params = params;
-
-	return ZSTD_resetCStream_internal(zcs, pledgedSrcSize);
-}
-
-ZSTD_CStream *ZSTD_initCStream(ZSTD_parameters params, unsigned long long pledgedSrcSize, void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	ZSTD_CStream *const zcs = ZSTD_createCStream_advanced(stackMem);
-	if (zcs) {
-		size_t const code = ZSTD_initCStream_advanced(zcs, NULL, 0, params, pledgedSrcSize);
-		if (ZSTD_isError(code)) {
-			return NULL;
-		}
-	}
-	return zcs;
-}
-
-ZSTD_CStream *ZSTD_initCStream_usingCDict(const ZSTD_CDict *cdict, unsigned long long pledgedSrcSize, void *workspace, size_t workspaceSize)
-{
-	ZSTD_parameters const params = ZSTD_getParamsFromCDict(cdict);
-	ZSTD_CStream *const zcs = ZSTD_initCStream(params, pledgedSrcSize, workspace, workspaceSize);
-	if (zcs) {
-		zcs->cdict = cdict;
-		if (ZSTD_isError(ZSTD_resetCStream_internal(zcs, pledgedSrcSize))) {
-			return NULL;
-		}
-	}
-	return zcs;
-}
-
-/*======   Compression   ======*/
-
-typedef enum { zsf_gather, zsf_flush, zsf_end } ZSTD_flush_e;
-
-ZSTD_STATIC size_t ZSTD_limitCopy(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	size_t const length = MIN(dstCapacity, srcSize);
-	memcpy(dst, src, length);
-	return length;
-}
-
-static size_t ZSTD_compressStream_generic(ZSTD_CStream *zcs, void *dst, size_t *dstCapacityPtr, const void *src, size_t *srcSizePtr, ZSTD_flush_e const flush)
-{
-	U32 someMoreWork = 1;
-	const char *const istart = (const char *)src;
-	const char *const iend = istart + *srcSizePtr;
-	const char *ip = istart;
-	char *const ostart = (char *)dst;
-	char *const oend = ostart + *dstCapacityPtr;
-	char *op = ostart;
-
-	while (someMoreWork) {
-		switch (zcs->stage) {
-		case zcss_init:
-			return ERROR(init_missing); /* call ZBUFF_compressInit() first ! */
-
-		case zcss_load:
-			/* complete inBuffer */
-			{
-				size_t const toLoad = zcs->inBuffTarget - zcs->inBuffPos;
-				size_t const loaded = ZSTD_limitCopy(zcs->inBuff + zcs->inBuffPos, toLoad, ip, iend - ip);
-				zcs->inBuffPos += loaded;
-				ip += loaded;
-				if ((zcs->inBuffPos == zcs->inToCompress) || (!flush && (toLoad != loaded))) {
-					someMoreWork = 0;
-					break; /* not enough input to get a full block : stop there, wait for more */
-				}
-			}
-			/* compress curr block (note : this stage cannot be stopped in the middle) */
-			{
-				void *cDst;
-				size_t cSize;
-				size_t const iSize = zcs->inBuffPos - zcs->inToCompress;
-				size_t oSize = oend - op;
-				if (oSize >= ZSTD_compressBound(iSize))
-					cDst = op; /* compress directly into output buffer (avoid flush stage) */
-				else
-					cDst = zcs->outBuff, oSize = zcs->outBuffSize;
-				cSize = (flush == zsf_end) ? ZSTD_compressEnd(zcs->cctx, cDst, oSize, zcs->inBuff + zcs->inToCompress, iSize)
-							   : ZSTD_compressContinue(zcs->cctx, cDst, oSize, zcs->inBuff + zcs->inToCompress, iSize);
-				if (ZSTD_isError(cSize))
-					return cSize;
-				if (flush == zsf_end)
-					zcs->frameEnded = 1;
-				/* prepare next block */
-				zcs->inBuffTarget = zcs->inBuffPos + zcs->blockSize;
-				if (zcs->inBuffTarget > zcs->inBuffSize)
-					zcs->inBuffPos = 0, zcs->inBuffTarget = zcs->blockSize; /* note : inBuffSize >= blockSize */
-				zcs->inToCompress = zcs->inBuffPos;
-				if (cDst == op) {
-					op += cSize;
-					break;
-				} /* no need to flush */
-				zcs->outBuffContentSize = cSize;
-				zcs->outBuffFlushedSize = 0;
-				zcs->stage = zcss_flush; /* pass-through to flush stage */
-			}
-			fallthrough;
-
-		case zcss_flush: {
-			size_t const toFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
-			size_t const flushed = ZSTD_limitCopy(op, oend - op, zcs->outBuff + zcs->outBuffFlushedSize, toFlush);
-			op += flushed;
-			zcs->outBuffFlushedSize += flushed;
-			if (toFlush != flushed) {
-				someMoreWork = 0;
-				break;
-			} /* dst too small to store flushed data : stop there */
-			zcs->outBuffContentSize = zcs->outBuffFlushedSize = 0;
-			zcs->stage = zcss_load;
-			break;
-		}
-
-		case zcss_final:
-			someMoreWork = 0; /* do nothing */
-			break;
-
-		default:
-			return ERROR(GENERIC); /* impossible */
-		}
-	}
-
-	*srcSizePtr = ip - istart;
-	*dstCapacityPtr = op - ostart;
-	zcs->inputProcessed += *srcSizePtr;
-	if (zcs->frameEnded)
-		return 0;
-	{
-		size_t hintInSize = zcs->inBuffTarget - zcs->inBuffPos;
-		if (hintInSize == 0)
-			hintInSize = zcs->blockSize;
-		return hintInSize;
-	}
-}
-
-size_t ZSTD_compressStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output, ZSTD_inBuffer *input)
-{
-	size_t sizeRead = input->size - input->pos;
-	size_t sizeWritten = output->size - output->pos;
-	size_t const result =
-	    ZSTD_compressStream_generic(zcs, (char *)(output->dst) + output->pos, &sizeWritten, (const char *)(input->src) + input->pos, &sizeRead, zsf_gather);
-	input->pos += sizeRead;
-	output->pos += sizeWritten;
-	return result;
-}
-
-/*======   Finalize   ======*/
-
-/*! ZSTD_flushStream() :
-*   @return : amount of data remaining to flush */
-size_t ZSTD_flushStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output)
-{
-	size_t srcSize = 0;
-	size_t sizeWritten = output->size - output->pos;
-	size_t const result = ZSTD_compressStream_generic(zcs, (char *)(output->dst) + output->pos, &sizeWritten, &srcSize,
-							  &srcSize, /* use a valid src address instead of NULL */
-							  zsf_flush);
-	output->pos += sizeWritten;
-	if (ZSTD_isError(result))
-		return result;
-	return zcs->outBuffContentSize - zcs->outBuffFlushedSize; /* remaining to flush */
-}
-
-size_t ZSTD_endStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output)
-{
-	BYTE *const ostart = (BYTE *)(output->dst) + output->pos;
-	BYTE *const oend = (BYTE *)(output->dst) + output->size;
-	BYTE *op = ostart;
-
-	if ((zcs->pledgedSrcSize) && (zcs->inputProcessed != zcs->pledgedSrcSize))
-		return ERROR(srcSize_wrong); /* pledgedSrcSize not respected */
-
-	if (zcs->stage != zcss_final) {
-		/* flush whatever remains */
-		size_t srcSize = 0;
-		size_t sizeWritten = output->size - output->pos;
-		size_t const notEnded =
-		    ZSTD_compressStream_generic(zcs, ostart, &sizeWritten, &srcSize, &srcSize, zsf_end); /* use a valid src address instead of NULL */
-		size_t const remainingToFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
-		op += sizeWritten;
-		if (remainingToFlush) {
-			output->pos += sizeWritten;
-			return remainingToFlush + ZSTD_BLOCKHEADERSIZE /* final empty block */ + (zcs->checksum * 4);
-		}
-		/* create epilogue */
-		zcs->stage = zcss_final;
-		zcs->outBuffContentSize = !notEnded ? 0 : ZSTD_compressEnd(zcs->cctx, zcs->outBuff, zcs->outBuffSize, NULL,
-									   0); /* write epilogue, including final empty block, into outBuff */
-	}
-
-	/* flush epilogue */
-	{
-		size_t const toFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
-		size_t const flushed = ZSTD_limitCopy(op, oend - op, zcs->outBuff + zcs->outBuffFlushedSize, toFlush);
-		op += flushed;
-		zcs->outBuffFlushedSize += flushed;
-		output->pos += op - ostart;
-		if (toFlush == flushed)
-			zcs->stage = zcss_init; /* end reached */
-		return toFlush - flushed;
-	}
-}
-
-/*-=====  Pre-defined compression levels  =====-*/
-
-#define ZSTD_DEFAULT_CLEVEL 1
-#define ZSTD_MAX_CLEVEL 22
-int ZSTD_maxCLevel(void) { return ZSTD_MAX_CLEVEL; }
-
-static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEVEL + 1] = {
-    {
-	/* "default" */
-	/* W,  C,  H,  S,  L, TL, strat */
-	{18, 12, 12, 1, 7, 16, ZSTD_fast},    /* level  0 - never used */
-	{19, 13, 14, 1, 7, 16, ZSTD_fast},    /* level  1 */
-	{19, 15, 16, 1, 6, 16, ZSTD_fast},    /* level  2 */
-	{20, 16, 17, 1, 5, 16, ZSTD_dfast},   /* level  3.*/
-	{20, 18, 18, 1, 5, 16, ZSTD_dfast},   /* level  4.*/
-	{20, 15, 18, 3, 5, 16, ZSTD_greedy},  /* level  5 */
-	{21, 16, 19, 2, 5, 16, ZSTD_lazy},    /* level  6 */
-	{21, 17, 20, 3, 5, 16, ZSTD_lazy},    /* level  7 */
-	{21, 18, 20, 3, 5, 16, ZSTD_lazy2},   /* level  8 */
-	{21, 20, 20, 3, 5, 16, ZSTD_lazy2},   /* level  9 */
-	{21, 19, 21, 4, 5, 16, ZSTD_lazy2},   /* level 10 */
-	{22, 20, 22, 4, 5, 16, ZSTD_lazy2},   /* level 11 */
-	{22, 20, 22, 5, 5, 16, ZSTD_lazy2},   /* level 12 */
-	{22, 21, 22, 5, 5, 16, ZSTD_lazy2},   /* level 13 */
-	{22, 21, 22, 6, 5, 16, ZSTD_lazy2},   /* level 14 */
-	{22, 21, 21, 5, 5, 16, ZSTD_btlazy2}, /* level 15 */
-	{23, 22, 22, 5, 5, 16, ZSTD_btlazy2}, /* level 16 */
-	{23, 21, 22, 4, 5, 24, ZSTD_btopt},   /* level 17 */
-	{23, 23, 22, 6, 5, 32, ZSTD_btopt},   /* level 18 */
-	{23, 23, 22, 6, 3, 48, ZSTD_btopt},   /* level 19 */
-	{25, 25, 23, 7, 3, 64, ZSTD_btopt2},  /* level 20 */
-	{26, 26, 23, 7, 3, 256, ZSTD_btopt2}, /* level 21 */
-	{27, 27, 25, 9, 3, 512, ZSTD_btopt2}, /* level 22 */
-    },
-    {
-	/* for srcSize <= 256 KB */
-	/* W,  C,  H,  S,  L,  T, strat */
-	{0, 0, 0, 0, 0, 0, ZSTD_fast},	 /* level  0 - not used */
-	{18, 13, 14, 1, 6, 8, ZSTD_fast},      /* level  1 */
-	{18, 14, 13, 1, 5, 8, ZSTD_dfast},     /* level  2 */
-	{18, 16, 15, 1, 5, 8, ZSTD_dfast},     /* level  3 */
-	{18, 15, 17, 1, 5, 8, ZSTD_greedy},    /* level  4.*/
-	{18, 16, 17, 4, 5, 8, ZSTD_greedy},    /* level  5.*/
-	{18, 16, 17, 3, 5, 8, ZSTD_lazy},      /* level  6.*/
-	{18, 17, 17, 4, 4, 8, ZSTD_lazy},      /* level  7 */
-	{18, 17, 17, 4, 4, 8, ZSTD_lazy2},     /* level  8 */
-	{18, 17, 17, 5, 4, 8, ZSTD_lazy2},     /* level  9 */
-	{18, 17, 17, 6, 4, 8, ZSTD_lazy2},     /* level 10 */
-	{18, 18, 17, 6, 4, 8, ZSTD_lazy2},     /* level 11.*/
-	{18, 18, 17, 7, 4, 8, ZSTD_lazy2},     /* level 12.*/
-	{18, 19, 17, 6, 4, 8, ZSTD_btlazy2},   /* level 13 */
-	{18, 18, 18, 4, 4, 16, ZSTD_btopt},    /* level 14.*/
-	{18, 18, 18, 4, 3, 16, ZSTD_btopt},    /* level 15.*/
-	{18, 19, 18, 6, 3, 32, ZSTD_btopt},    /* level 16.*/
-	{18, 19, 18, 8, 3, 64, ZSTD_btopt},    /* level 17.*/
-	{18, 19, 18, 9, 3, 128, ZSTD_btopt},   /* level 18.*/
-	{18, 19, 18, 10, 3, 256, ZSTD_btopt},  /* level 19.*/
-	{18, 19, 18, 11, 3, 512, ZSTD_btopt2}, /* level 20.*/
-	{18, 19, 18, 12, 3, 512, ZSTD_btopt2}, /* level 21.*/
-	{18, 19, 18, 13, 3, 512, ZSTD_btopt2}, /* level 22.*/
-    },
-    {
-	/* for srcSize <= 128 KB */
-	/* W,  C,  H,  S,  L,  T, strat */
-	{17, 12, 12, 1, 7, 8, ZSTD_fast},      /* level  0 - not used */
-	{17, 12, 13, 1, 6, 8, ZSTD_fast},      /* level  1 */
-	{17, 13, 16, 1, 5, 8, ZSTD_fast},      /* level  2 */
-	{17, 16, 16, 2, 5, 8, ZSTD_dfast},     /* level  3 */
-	{17, 13, 15, 3, 4, 8, ZSTD_greedy},    /* level  4 */
-	{17, 15, 17, 4, 4, 8, ZSTD_greedy},    /* level  5 */
-	{17, 16, 17, 3, 4, 8, ZSTD_lazy},      /* level  6 */
-	{17, 15, 17, 4, 4, 8, ZSTD_lazy2},     /* level  7 */
-	{17, 17, 17, 4, 4, 8, ZSTD_lazy2},     /* level  8 */
-	{17, 17, 17, 5, 4, 8, ZSTD_lazy2},     /* level  9 */
-	{17, 17, 17, 6, 4, 8, ZSTD_lazy2},     /* level 10 */
-	{17, 17, 17, 7, 4, 8, ZSTD_lazy2},     /* level 11 */
-	{17, 17, 17, 8, 4, 8, ZSTD_lazy2},     /* level 12 */
-	{17, 18, 17, 6, 4, 8, ZSTD_btlazy2},   /* level 13.*/
-	{17, 17, 17, 7, 3, 8, ZSTD_btopt},     /* level 14.*/
-	{17, 17, 17, 7, 3, 16, ZSTD_btopt},    /* level 15.*/
-	{17, 18, 17, 7, 3, 32, ZSTD_btopt},    /* level 16.*/
-	{17, 18, 17, 7, 3, 64, ZSTD_btopt},    /* level 17.*/
-	{17, 18, 17, 7, 3, 256, ZSTD_btopt},   /* level 18.*/
-	{17, 18, 17, 8, 3, 256, ZSTD_btopt},   /* level 19.*/
-	{17, 18, 17, 9, 3, 256, ZSTD_btopt2},  /* level 20.*/
-	{17, 18, 17, 10, 3, 256, ZSTD_btopt2}, /* level 21.*/
-	{17, 18, 17, 11, 3, 512, ZSTD_btopt2}, /* level 22.*/
-    },
-    {
-	/* for srcSize <= 16 KB */
-	/* W,  C,  H,  S,  L,  T, strat */
-	{14, 12, 12, 1, 7, 6, ZSTD_fast},      /* level  0 - not used */
-	{14, 14, 14, 1, 6, 6, ZSTD_fast},      /* level  1 */
-	{14, 14, 14, 1, 4, 6, ZSTD_fast},      /* level  2 */
-	{14, 14, 14, 1, 4, 6, ZSTD_dfast},     /* level  3.*/
-	{14, 14, 14, 4, 4, 6, ZSTD_greedy},    /* level  4.*/
-	{14, 14, 14, 3, 4, 6, ZSTD_lazy},      /* level  5.*/
-	{14, 14, 14, 4, 4, 6, ZSTD_lazy2},     /* level  6 */
-	{14, 14, 14, 5, 4, 6, ZSTD_lazy2},     /* level  7 */
-	{14, 14, 14, 6, 4, 6, ZSTD_lazy2},     /* level  8.*/
-	{14, 15, 14, 6, 4, 6, ZSTD_btlazy2},   /* level  9.*/
-	{14, 15, 14, 3, 3, 6, ZSTD_btopt},     /* level 10.*/
-	{14, 15, 14, 6, 3, 8, ZSTD_btopt},     /* level 11.*/
-	{14, 15, 14, 6, 3, 16, ZSTD_btopt},    /* level 12.*/
-	{14, 15, 14, 6, 3, 24, ZSTD_btopt},    /* level 13.*/
-	{14, 15, 15, 6, 3, 48, ZSTD_btopt},    /* level 14.*/
-	{14, 15, 15, 6, 3, 64, ZSTD_btopt},    /* level 15.*/
-	{14, 15, 15, 6, 3, 96, ZSTD_btopt},    /* level 16.*/
-	{14, 15, 15, 6, 3, 128, ZSTD_btopt},   /* level 17.*/
-	{14, 15, 15, 6, 3, 256, ZSTD_btopt},   /* level 18.*/
-	{14, 15, 15, 7, 3, 256, ZSTD_btopt},   /* level 19.*/
-	{14, 15, 15, 8, 3, 256, ZSTD_btopt2},  /* level 20.*/
-	{14, 15, 15, 9, 3, 256, ZSTD_btopt2},  /* level 21.*/
-	{14, 15, 15, 10, 3, 256, ZSTD_btopt2}, /* level 22.*/
-    },
-};
-
-/*! ZSTD_getCParams() :
-*   @return ZSTD_compressionParameters structure for a selected compression level, `srcSize` and `dictSize`.
-*   Size values are optional, provide 0 if not known or unused */
-ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long srcSize, size_t dictSize)
-{
-	ZSTD_compressionParameters cp;
-	size_t const addedSize = srcSize ? 0 : 500;
-	U64 const rSize = srcSize + dictSize ? srcSize + dictSize + addedSize : (U64)-1;
-	U32 const tableID = (rSize <= 256 KB) + (rSize <= 128 KB) + (rSize <= 16 KB); /* intentional underflow for srcSizeHint == 0 */
-	if (compressionLevel <= 0)
-		compressionLevel = ZSTD_DEFAULT_CLEVEL; /* 0 == default; no negative compressionLevel yet */
-	if (compressionLevel > ZSTD_MAX_CLEVEL)
-		compressionLevel = ZSTD_MAX_CLEVEL;
-	cp = ZSTD_defaultCParameters[tableID][compressionLevel];
-	if (ZSTD_32bits()) { /* auto-correction, for 32-bits mode */
-		if (cp.windowLog > ZSTD_WINDOWLOG_MAX)
-			cp.windowLog = ZSTD_WINDOWLOG_MAX;
-		if (cp.chainLog > ZSTD_CHAINLOG_MAX)
-			cp.chainLog = ZSTD_CHAINLOG_MAX;
-		if (cp.hashLog > ZSTD_HASHLOG_MAX)
-			cp.hashLog = ZSTD_HASHLOG_MAX;
-	}
-	cp = ZSTD_adjustCParams(cp, srcSize, dictSize);
-	return cp;
-}
-
-/*! ZSTD_getParams() :
-*   same as ZSTD_getCParams(), but @return a `ZSTD_parameters` object (instead of `ZSTD_compressionParameters`).
-*   All fields of `ZSTD_frameParameters` are set to default (0) */
-ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long srcSize, size_t dictSize)
-{
-	ZSTD_parameters params;
-	ZSTD_compressionParameters const cParams = ZSTD_getCParams(compressionLevel, srcSize, dictSize);
-	memset(&params, 0, sizeof(params));
-	params.cParams = cParams;
-	return params;
-}
-
-EXPORT_SYMBOL(ZSTD_maxCLevel);
-EXPORT_SYMBOL(ZSTD_compressBound);
-
-EXPORT_SYMBOL(ZSTD_CCtxWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initCCtx);
-EXPORT_SYMBOL(ZSTD_compressCCtx);
-EXPORT_SYMBOL(ZSTD_compress_usingDict);
-
-EXPORT_SYMBOL(ZSTD_CDictWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initCDict);
-EXPORT_SYMBOL(ZSTD_compress_usingCDict);
-
-EXPORT_SYMBOL(ZSTD_CStreamWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initCStream);
-EXPORT_SYMBOL(ZSTD_initCStream_usingCDict);
-EXPORT_SYMBOL(ZSTD_resetCStream);
-EXPORT_SYMBOL(ZSTD_compressStream);
-EXPORT_SYMBOL(ZSTD_flushStream);
-EXPORT_SYMBOL(ZSTD_endStream);
-EXPORT_SYMBOL(ZSTD_CStreamInSize);
-EXPORT_SYMBOL(ZSTD_CStreamOutSize);
-
-EXPORT_SYMBOL(ZSTD_getCParams);
-EXPORT_SYMBOL(ZSTD_getParams);
-EXPORT_SYMBOL(ZSTD_checkCParams);
-EXPORT_SYMBOL(ZSTD_adjustCParams);
-
-EXPORT_SYMBOL(ZSTD_compressBegin);
-EXPORT_SYMBOL(ZSTD_compressBegin_usingDict);
-EXPORT_SYMBOL(ZSTD_compressBegin_advanced);
-EXPORT_SYMBOL(ZSTD_copyCCtx);
-EXPORT_SYMBOL(ZSTD_compressBegin_usingCDict);
-EXPORT_SYMBOL(ZSTD_compressContinue);
-EXPORT_SYMBOL(ZSTD_compressEnd);
-
-EXPORT_SYMBOL(ZSTD_getBlockSizeMax);
-EXPORT_SYMBOL(ZSTD_compressBlock);
-
-MODULE_LICENSE("Dual BSD/GPL");
-MODULE_DESCRIPTION("Zstd Compressor");
diff --git a/lib/zstd/compress/fse_compress.c b/lib/zstd/compress/fse_compress.c
new file mode 100644
index 0000000..436985b
--- /dev/null
+++ b/lib/zstd/compress/fse_compress.c
@@ -0,0 +1,625 @@
+/* ******************************************************************
+ * FSE : Finite State Entropy encoder
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ *  You can contact the author at :
+ *  - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+/* **************************************************************
+*  Includes
+****************************************************************/
+#include "../common/compiler.h"
+#include "../common/mem.h"        /* U32, U16, etc. */
+#include "../common/debug.h"      /* assert, DEBUGLOG */
+#include "hist.h"       /* HIST_count_wksp */
+#include "../common/bitstream.h"
+#define FSE_STATIC_LINKING_ONLY
+#include "../common/fse.h"
+#include "../common/error_private.h"
+#define ZSTD_DEPS_NEED_MALLOC
+#define ZSTD_DEPS_NEED_MATH64
+#include "../common/zstd_deps.h"  /* ZSTD_malloc, ZSTD_free, ZSTD_memcpy, ZSTD_memset */
+
+
+/* **************************************************************
+*  Error Management
+****************************************************************/
+#define FSE_isError ERR_isError
+
+
+/* **************************************************************
+*  Templates
+****************************************************************/
+/*
+  designed to be included
+  for type-specific functions (template emulation in C)
+  Objective is to write these functions only once, for improved maintenance
+*/
+
+/* safety checks */
+#ifndef FSE_FUNCTION_EXTENSION
+#  error "FSE_FUNCTION_EXTENSION must be defined"
+#endif
+#ifndef FSE_FUNCTION_TYPE
+#  error "FSE_FUNCTION_TYPE must be defined"
+#endif
+
+/* Function names */
+#define FSE_CAT(X,Y) X##Y
+#define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
+#define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)
+
+
+/* Function templates */
+
+/* FSE_buildCTable_wksp() :
+ * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
+ * wkspSize should be sized to handle worst case situation, which is `1<<max_tableLog * sizeof(FSE_FUNCTION_TYPE)`
+ * workSpace must also be properly aligned with FSE_FUNCTION_TYPE requirements
+ */
+size_t FSE_buildCTable_wksp(FSE_CTable* ct,
+                      const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog,
+                            void* workSpace, size_t wkspSize)
+{
+    U32 const tableSize = 1 << tableLog;
+    U32 const tableMask = tableSize - 1;
+    void* const ptr = ct;
+    U16* const tableU16 = ( (U16*) ptr) + 2;
+    void* const FSCT = ((U32*)ptr) + 1 /* header */ + (tableLog ? tableSize>>1 : 1) ;
+    FSE_symbolCompressionTransform* const symbolTT = (FSE_symbolCompressionTransform*) (FSCT);
+    U32 const step = FSE_TABLESTEP(tableSize);
+
+    U32* cumul = (U32*)workSpace;
+    FSE_FUNCTION_TYPE* tableSymbol = (FSE_FUNCTION_TYPE*)(cumul + (maxSymbolValue + 2));
+
+    U32 highThreshold = tableSize-1;
+
+    if ((size_t)workSpace & 3) return ERROR(GENERIC); /* Must be 4 byte aligned */
+    if (FSE_BUILD_CTABLE_WORKSPACE_SIZE(maxSymbolValue, tableLog) > wkspSize) return ERROR(tableLog_tooLarge);
+    /* CTable header */
+    tableU16[-2] = (U16) tableLog;
+    tableU16[-1] = (U16) maxSymbolValue;
+    assert(tableLog < 16);   /* required for threshold strategy to work */
+
+    /* For explanations on how to distribute symbol values over the table :
+     * http://fastcompression.blogspot.fr/2014/02/fse-distributing-symbol-values.html */
+
+     #ifdef __clang_analyzer__
+     ZSTD_memset(tableSymbol, 0, sizeof(*tableSymbol) * tableSize);   /* useless initialization, just to keep scan-build happy */
+     #endif
+
+    /* symbol start positions */
+    {   U32 u;
+        cumul[0] = 0;
+        for (u=1; u <= maxSymbolValue+1; u++) {
+            if (normalizedCounter[u-1]==-1) {  /* Low proba symbol */
+                cumul[u] = cumul[u-1] + 1;
+                tableSymbol[highThreshold--] = (FSE_FUNCTION_TYPE)(u-1);
+            } else {
+                cumul[u] = cumul[u-1] + normalizedCounter[u-1];
+        }   }
+        cumul[maxSymbolValue+1] = tableSize+1;
+    }
+
+    /* Spread symbols */
+    {   U32 position = 0;
+        U32 symbol;
+        for (symbol=0; symbol<=maxSymbolValue; symbol++) {
+            int nbOccurrences;
+            int const freq = normalizedCounter[symbol];
+            for (nbOccurrences=0; nbOccurrences<freq; nbOccurrences++) {
+                tableSymbol[position] = (FSE_FUNCTION_TYPE)symbol;
+                position = (position + step) & tableMask;
+                while (position > highThreshold)
+                    position = (position + step) & tableMask;   /* Low proba area */
+        }   }
+
+        assert(position==0);  /* Must have initialized all positions */
+    }
+
+    /* Build table */
+    {   U32 u; for (u=0; u<tableSize; u++) {
+        FSE_FUNCTION_TYPE s = tableSymbol[u];   /* note : static analyzer may not understand tableSymbol is properly initialized */
+        tableU16[cumul[s]++] = (U16) (tableSize+u);   /* TableU16 : sorted by symbol order; gives next state value */
+    }   }
+
+    /* Build Symbol Transformation Table */
+    {   unsigned total = 0;
+        unsigned s;
+        for (s=0; s<=maxSymbolValue; s++) {
+            switch (normalizedCounter[s])
+            {
+            case  0:
+                /* filling nonetheless, for compatibility with FSE_getMaxNbBits() */
+                symbolTT[s].deltaNbBits = ((tableLog+1) << 16) - (1<<tableLog);
+                break;
+
+            case -1:
+            case  1:
+                symbolTT[s].deltaNbBits = (tableLog << 16) - (1<<tableLog);
+                symbolTT[s].deltaFindState = total - 1;
+                total ++;
+                break;
+            default :
+                {
+                    U32 const maxBitsOut = tableLog - BIT_highbit32 (normalizedCounter[s]-1);
+                    U32 const minStatePlus = normalizedCounter[s] << maxBitsOut;
+                    symbolTT[s].deltaNbBits = (maxBitsOut << 16) - minStatePlus;
+                    symbolTT[s].deltaFindState = total - normalizedCounter[s];
+                    total +=  normalizedCounter[s];
+    }   }   }   }
+
+#if 0  /* debug : symbol costs */
+    DEBUGLOG(5, "\n --- table statistics : ");
+    {   U32 symbol;
+        for (symbol=0; symbol<=maxSymbolValue; symbol++) {
+            DEBUGLOG(5, "%3u: w=%3i,   maxBits=%u, fracBits=%.2f",
+                symbol, normalizedCounter[symbol],
+                FSE_getMaxNbBits(symbolTT, symbol),
+                (double)FSE_bitCost(symbolTT, tableLog, symbol, 8) / 256);
+        }
+    }
+#endif
+
+    return 0;
+}
+
+
+
+
+#ifndef FSE_COMMONDEFS_ONLY
+
+
+/*-**************************************************************
+*  FSE NCount encoding
+****************************************************************/
+size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog)
+{
+    size_t const maxHeaderSize = (((maxSymbolValue+1) * tableLog) >> 3) + 3;
+    return maxSymbolValue ? maxHeaderSize : FSE_NCOUNTBOUND;  /* maxSymbolValue==0 ? use default */
+}
+
+static size_t
+FSE_writeNCount_generic (void* header, size_t headerBufferSize,
+                   const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog,
+                         unsigned writeIsSafe)
+{
+    BYTE* const ostart = (BYTE*) header;
+    BYTE* out = ostart;
+    BYTE* const oend = ostart + headerBufferSize;
+    int nbBits;
+    const int tableSize = 1 << tableLog;
+    int remaining;
+    int threshold;
+    U32 bitStream = 0;
+    int bitCount = 0;
+    unsigned symbol = 0;
+    unsigned const alphabetSize = maxSymbolValue + 1;
+    int previousIs0 = 0;
+
+    /* Table Size */
+    bitStream += (tableLog-FSE_MIN_TABLELOG) << bitCount;
+    bitCount  += 4;
+
+    /* Init */
+    remaining = tableSize+1;   /* +1 for extra accuracy */
+    threshold = tableSize;
+    nbBits = tableLog+1;
+
+    while ((symbol < alphabetSize) && (remaining>1)) {  /* stops at 1 */
+        if (previousIs0) {
+            unsigned start = symbol;
+            while ((symbol < alphabetSize) && !normalizedCounter[symbol]) symbol++;
+            if (symbol == alphabetSize) break;   /* incorrect distribution */
+            while (symbol >= start+24) {
+                start+=24;
+                bitStream += 0xFFFFU << bitCount;
+                if ((!writeIsSafe) && (out > oend-2))
+                    return ERROR(dstSize_tooSmall);   /* Buffer overflow */
+                out[0] = (BYTE) bitStream;
+                out[1] = (BYTE)(bitStream>>8);
+                out+=2;
+                bitStream>>=16;
+            }
+            while (symbol >= start+3) {
+                start+=3;
+                bitStream += 3 << bitCount;
+                bitCount += 2;
+            }
+            bitStream += (symbol-start) << bitCount;
+            bitCount += 2;
+            if (bitCount>16) {
+                if ((!writeIsSafe) && (out > oend - 2))
+                    return ERROR(dstSize_tooSmall);   /* Buffer overflow */
+                out[0] = (BYTE)bitStream;
+                out[1] = (BYTE)(bitStream>>8);
+                out += 2;
+                bitStream >>= 16;
+                bitCount -= 16;
+        }   }
+        {   int count = normalizedCounter[symbol++];
+            int const max = (2*threshold-1) - remaining;
+            remaining -= count < 0 ? -count : count;
+            count++;   /* +1 for extra accuracy */
+            if (count>=threshold)
+                count += max;   /* [0..max[ [max..threshold[ (...) [threshold+max 2*threshold[ */
+            bitStream += count << bitCount;
+            bitCount  += nbBits;
+            bitCount  -= (count<max);
+            previousIs0  = (count==1);
+            if (remaining<1) return ERROR(GENERIC);
+            while (remaining<threshold) { nbBits--; threshold>>=1; }
+        }
+        if (bitCount>16) {
+            if ((!writeIsSafe) && (out > oend - 2))
+                return ERROR(dstSize_tooSmall);   /* Buffer overflow */
+            out[0] = (BYTE)bitStream;
+            out[1] = (BYTE)(bitStream>>8);
+            out += 2;
+            bitStream >>= 16;
+            bitCount -= 16;
+    }   }
+
+    if (remaining != 1)
+        return ERROR(GENERIC);  /* incorrect normalized distribution */
+    assert(symbol <= alphabetSize);
+
+    /* flush remaining bitStream */
+    if ((!writeIsSafe) && (out > oend - 2))
+        return ERROR(dstSize_tooSmall);   /* Buffer overflow */
+    out[0] = (BYTE)bitStream;
+    out[1] = (BYTE)(bitStream>>8);
+    out+= (bitCount+7) /8;
+
+    return (out-ostart);
+}
+
+
+size_t FSE_writeNCount (void* buffer, size_t bufferSize,
+                  const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog)
+{
+    if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);   /* Unsupported */
+    if (tableLog < FSE_MIN_TABLELOG) return ERROR(GENERIC);   /* Unsupported */
+
+    if (bufferSize < FSE_NCountWriteBound(maxSymbolValue, tableLog))
+        return FSE_writeNCount_generic(buffer, bufferSize, normalizedCounter, maxSymbolValue, tableLog, 0);
+
+    return FSE_writeNCount_generic(buffer, bufferSize, normalizedCounter, maxSymbolValue, tableLog, 1 /* write in buffer is safe */);
+}
+
+
+/*-**************************************************************
+*  FSE Compression Code
+****************************************************************/
+
+FSE_CTable* FSE_createCTable (unsigned maxSymbolValue, unsigned tableLog)
+{
+    size_t size;
+    if (tableLog > FSE_TABLELOG_ABSOLUTE_MAX) tableLog = FSE_TABLELOG_ABSOLUTE_MAX;
+    size = FSE_CTABLE_SIZE_U32 (tableLog, maxSymbolValue) * sizeof(U32);
+    return (FSE_CTable*)ZSTD_malloc(size);
+}
+
+void FSE_freeCTable (FSE_CTable* ct) { ZSTD_free(ct); }
+
+/* provides the minimum logSize to safely represent a distribution */
+static unsigned FSE_minTableLog(size_t srcSize, unsigned maxSymbolValue)
+{
+    U32 minBitsSrc = BIT_highbit32((U32)(srcSize)) + 1;
+    U32 minBitsSymbols = BIT_highbit32(maxSymbolValue) + 2;
+    U32 minBits = minBitsSrc < minBitsSymbols ? minBitsSrc : minBitsSymbols;
+    assert(srcSize > 1); /* Not supported, RLE should be used instead */
+    return minBits;
+}
+
+unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus)
+{
+    U32 maxBitsSrc = BIT_highbit32((U32)(srcSize - 1)) - minus;
+    U32 tableLog = maxTableLog;
+    U32 minBits = FSE_minTableLog(srcSize, maxSymbolValue);
+    assert(srcSize > 1); /* Not supported, RLE should be used instead */
+    if (tableLog==0) tableLog = FSE_DEFAULT_TABLELOG;
+    if (maxBitsSrc < tableLog) tableLog = maxBitsSrc;   /* Accuracy can be reduced */
+    if (minBits > tableLog) tableLog = minBits;   /* Need a minimum to safely represent all symbol values */
+    if (tableLog < FSE_MIN_TABLELOG) tableLog = FSE_MIN_TABLELOG;
+    if (tableLog > FSE_MAX_TABLELOG) tableLog = FSE_MAX_TABLELOG;
+    return tableLog;
+}
+
+unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
+{
+    return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 2);
+}
+
+/* Secondary normalization method.
+   To be used when primary method fails. */
+
+static size_t FSE_normalizeM2(short* norm, U32 tableLog, const unsigned* count, size_t total, U32 maxSymbolValue, short lowProbCount)
+{
+    short const NOT_YET_ASSIGNED = -2;
+    U32 s;
+    U32 distributed = 0;
+    U32 ToDistribute;
+
+    /* Init */
+    U32 const lowThreshold = (U32)(total >> tableLog);
+    U32 lowOne = (U32)((total * 3) >> (tableLog + 1));
+
+    for (s=0; s<=maxSymbolValue; s++) {
+        if (count[s] == 0) {
+            norm[s]=0;
+            continue;
+        }
+        if (count[s] <= lowThreshold) {
+            norm[s] = lowProbCount;
+            distributed++;
+            total -= count[s];
+            continue;
+        }
+        if (count[s] <= lowOne) {
+            norm[s] = 1;
+            distributed++;
+            total -= count[s];
+            continue;
+        }
+
+        norm[s]=NOT_YET_ASSIGNED;
+    }
+    ToDistribute = (1 << tableLog) - distributed;
+
+    if (ToDistribute == 0)
+        return 0;
+
+    if ((total / ToDistribute) > lowOne) {
+        /* risk of rounding to zero */
+        lowOne = (U32)((total * 3) / (ToDistribute * 2));
+        for (s=0; s<=maxSymbolValue; s++) {
+            if ((norm[s] == NOT_YET_ASSIGNED) && (count[s] <= lowOne)) {
+                norm[s] = 1;
+                distributed++;
+                total -= count[s];
+                continue;
+        }   }
+        ToDistribute = (1 << tableLog) - distributed;
+    }
+
+    if (distributed == maxSymbolValue+1) {
+        /* all values are pretty poor;
+           probably incompressible data (should have already been detected);
+           find max, then give all remaining points to max */
+        U32 maxV = 0, maxC = 0;
+        for (s=0; s<=maxSymbolValue; s++)
+            if (count[s] > maxC) { maxV=s; maxC=count[s]; }
+        norm[maxV] += (short)ToDistribute;
+        return 0;
+    }
+
+    if (total == 0) {
+        /* all of the symbols were low enough for the lowOne or lowThreshold */
+        for (s=0; ToDistribute > 0; s = (s+1)%(maxSymbolValue+1))
+            if (norm[s] > 0) { ToDistribute--; norm[s]++; }
+        return 0;
+    }
+
+    {   U64 const vStepLog = 62 - tableLog;
+        U64 const mid = (1ULL << (vStepLog-1)) - 1;
+        U64 const rStep = ZSTD_div64((((U64)1<<vStepLog) * ToDistribute) + mid, (U32)total);   /* scale on remaining */
+        U64 tmpTotal = mid;
+        for (s=0; s<=maxSymbolValue; s++) {
+            if (norm[s]==NOT_YET_ASSIGNED) {
+                U64 const end = tmpTotal + (count[s] * rStep);
+                U32 const sStart = (U32)(tmpTotal >> vStepLog);
+                U32 const sEnd = (U32)(end >> vStepLog);
+                U32 const weight = sEnd - sStart;
+                if (weight < 1)
+                    return ERROR(GENERIC);
+                norm[s] = (short)weight;
+                tmpTotal = end;
+    }   }   }
+
+    return 0;
+}
+
+size_t FSE_normalizeCount (short* normalizedCounter, unsigned tableLog,
+                           const unsigned* count, size_t total,
+                           unsigned maxSymbolValue, unsigned useLowProbCount)
+{
+    /* Sanity checks */
+    if (tableLog==0) tableLog = FSE_DEFAULT_TABLELOG;
+    if (tableLog < FSE_MIN_TABLELOG) return ERROR(GENERIC);   /* Unsupported size */
+    if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);   /* Unsupported size */
+    if (tableLog < FSE_minTableLog(total, maxSymbolValue)) return ERROR(GENERIC);   /* Too small tableLog, compression potentially impossible */
+
+    {   static U32 const rtbTable[] = {     0, 473195, 504333, 520860, 550000, 700000, 750000, 830000 };
+        short const lowProbCount = useLowProbCount ? -1 : 1;
+        U64 const scale = 62 - tableLog;
+        U64 const step = ZSTD_div64((U64)1<<62, (U32)total);   /* <== here, one division ! */
+        U64 const vStep = 1ULL<<(scale-20);
+        int stillToDistribute = 1<<tableLog;
+        unsigned s;
+        unsigned largest=0;
+        short largestP=0;
+        U32 lowThreshold = (U32)(total >> tableLog);
+
+        for (s=0; s<=maxSymbolValue; s++) {
+            if (count[s] == total) return 0;   /* rle special case */
+            if (count[s] == 0) { normalizedCounter[s]=0; continue; }
+            if (count[s] <= lowThreshold) {
+                normalizedCounter[s] = lowProbCount;
+                stillToDistribute--;
+            } else {
+                short proba = (short)((count[s]*step) >> scale);
+                if (proba<8) {
+                    U64 restToBeat = vStep * rtbTable[proba];
+                    proba += (count[s]*step) - ((U64)proba<<scale) > restToBeat;
+                }
+                if (proba > largestP) { largestP=proba; largest=s; }
+                normalizedCounter[s] = proba;
+                stillToDistribute -= proba;
+        }   }
+        if (-stillToDistribute >= (normalizedCounter[largest] >> 1)) {
+            /* corner case, need another normalization method */
+            size_t const errorCode = FSE_normalizeM2(normalizedCounter, tableLog, count, total, maxSymbolValue, lowProbCount);
+            if (FSE_isError(errorCode)) return errorCode;
+        }
+        else normalizedCounter[largest] += (short)stillToDistribute;
+    }
+
+#if 0
+    {   /* Print Table (debug) */
+        U32 s;
+        U32 nTotal = 0;
+        for (s=0; s<=maxSymbolValue; s++)
+            RAWLOG(2, "%3i: %4i \n", s, normalizedCounter[s]);
+        for (s=0; s<=maxSymbolValue; s++)
+            nTotal += abs(normalizedCounter[s]);
+        if (nTotal != (1U<<tableLog))
+            RAWLOG(2, "Warning !!! Total == %u != %u !!!", nTotal, 1U<<tableLog);
+        getchar();
+    }
+#endif
+
+    return tableLog;
+}
+
+
+/* fake FSE_CTable, for raw (uncompressed) input */
+size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits)
+{
+    const unsigned tableSize = 1 << nbBits;
+    const unsigned tableMask = tableSize - 1;
+    const unsigned maxSymbolValue = tableMask;
+    void* const ptr = ct;
+    U16* const tableU16 = ( (U16*) ptr) + 2;
+    void* const FSCT = ((U32*)ptr) + 1 /* header */ + (tableSize>>1);   /* assumption : tableLog >= 1 */
+    FSE_symbolCompressionTransform* const symbolTT = (FSE_symbolCompressionTransform*) (FSCT);
+    unsigned s;
+
+    /* Sanity checks */
+    if (nbBits < 1) return ERROR(GENERIC);             /* min size */
+
+    /* header */
+    tableU16[-2] = (U16) nbBits;
+    tableU16[-1] = (U16) maxSymbolValue;
+
+    /* Build table */
+    for (s=0; s<tableSize; s++)
+        tableU16[s] = (U16)(tableSize + s);
+
+    /* Build Symbol Transformation Table */
+    {   const U32 deltaNbBits = (nbBits << 16) - (1 << nbBits);
+        for (s=0; s<=maxSymbolValue; s++) {
+            symbolTT[s].deltaNbBits = deltaNbBits;
+            symbolTT[s].deltaFindState = s-1;
+    }   }
+
+    return 0;
+}
+
+/* fake FSE_CTable, for rle input (always same symbol) */
+size_t FSE_buildCTable_rle (FSE_CTable* ct, BYTE symbolValue)
+{
+    void* ptr = ct;
+    U16* tableU16 = ( (U16*) ptr) + 2;
+    void* FSCTptr = (U32*)ptr + 2;
+    FSE_symbolCompressionTransform* symbolTT = (FSE_symbolCompressionTransform*) FSCTptr;
+
+    /* header */
+    tableU16[-2] = (U16) 0;
+    tableU16[-1] = (U16) symbolValue;
+
+    /* Build table */
+    tableU16[0] = 0;
+    tableU16[1] = 0;   /* just in case */
+
+    /* Build Symbol Transformation Table */
+    symbolTT[symbolValue].deltaNbBits = 0;
+    symbolTT[symbolValue].deltaFindState = 0;
+
+    return 0;
+}
+
+
+static size_t FSE_compress_usingCTable_generic (void* dst, size_t dstSize,
+                           const void* src, size_t srcSize,
+                           const FSE_CTable* ct, const unsigned fast)
+{
+    const BYTE* const istart = (const BYTE*) src;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* ip=iend;
+
+    BIT_CStream_t bitC;
+    FSE_CState_t CState1, CState2;
+
+    /* init */
+    if (srcSize <= 2) return 0;
+    { size_t const initError = BIT_initCStream(&bitC, dst, dstSize);
+      if (FSE_isError(initError)) return 0; /* not enough space available to write a bitstream */ }
+
+#define FSE_FLUSHBITS(s)  (fast ? BIT_flushBitsFast(s) : BIT_flushBits(s))
+
+    if (srcSize & 1) {
+        FSE_initCState2(&CState1, ct, *--ip);
+        FSE_initCState2(&CState2, ct, *--ip);
+        FSE_encodeSymbol(&bitC, &CState1, *--ip);
+        FSE_FLUSHBITS(&bitC);
+    } else {
+        FSE_initCState2(&CState2, ct, *--ip);
+        FSE_initCState2(&CState1, ct, *--ip);
+    }
+
+    /* join to mod 4 */
+    srcSize -= 2;
+    if ((sizeof(bitC.bitContainer)*8 > FSE_MAX_TABLELOG*4+7 ) && (srcSize & 2)) {  /* test bit 2 */
+        FSE_encodeSymbol(&bitC, &CState2, *--ip);
+        FSE_encodeSymbol(&bitC, &CState1, *--ip);
+        FSE_FLUSHBITS(&bitC);
+    }
+
+    /* 2 or 4 encoding per loop */
+    while ( ip>istart ) {
+
+        FSE_encodeSymbol(&bitC, &CState2, *--ip);
+
+        if (sizeof(bitC.bitContainer)*8 < FSE_MAX_TABLELOG*2+7 )   /* this test must be static */
+            FSE_FLUSHBITS(&bitC);
+
+        FSE_encodeSymbol(&bitC, &CState1, *--ip);
+
+        if (sizeof(bitC.bitContainer)*8 > FSE_MAX_TABLELOG*4+7 ) {  /* this test must be static */
+            FSE_encodeSymbol(&bitC, &CState2, *--ip);
+            FSE_encodeSymbol(&bitC, &CState1, *--ip);
+        }
+
+        FSE_FLUSHBITS(&bitC);
+    }
+
+    FSE_flushCState(&bitC, &CState2);
+    FSE_flushCState(&bitC, &CState1);
+    return BIT_closeCStream(&bitC);
+}
+
+size_t FSE_compress_usingCTable (void* dst, size_t dstSize,
+                           const void* src, size_t srcSize,
+                           const FSE_CTable* ct)
+{
+    unsigned const fast = (dstSize >= FSE_BLOCKBOUND(srcSize));
+
+    if (fast)
+        return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 1);
+    else
+        return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 0);
+}
+
+
+size_t FSE_compressBound(size_t size) { return FSE_COMPRESSBOUND(size); }
+
+
+#endif   /* FSE_COMMONDEFS_ONLY */
diff --git a/lib/zstd/compress/hist.c b/lib/zstd/compress/hist.c
new file mode 100644
index 0000000..3ddc6df
--- /dev/null
+++ b/lib/zstd/compress/hist.c
@@ -0,0 +1,165 @@
+/* ******************************************************************
+ * hist : Histogram functions
+ * part of Finite State Entropy project
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ *  You can contact the author at :
+ *  - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+/* --- dependencies --- */
+#include "../common/mem.h"             /* U32, BYTE, etc. */
+#include "../common/debug.h"           /* assert, DEBUGLOG */
+#include "../common/error_private.h"   /* ERROR */
+#include "hist.h"
+
+
+/* --- Error management --- */
+unsigned HIST_isError(size_t code) { return ERR_isError(code); }
+
+/*-**************************************************************
+ *  Histogram functions
+ ****************************************************************/
+unsigned HIST_count_simple(unsigned* count, unsigned* maxSymbolValuePtr,
+                           const void* src, size_t srcSize)
+{
+    const BYTE* ip = (const BYTE*)src;
+    const BYTE* const end = ip + srcSize;
+    unsigned maxSymbolValue = *maxSymbolValuePtr;
+    unsigned largestCount=0;
+
+    ZSTD_memset(count, 0, (maxSymbolValue+1) * sizeof(*count));
+    if (srcSize==0) { *maxSymbolValuePtr = 0; return 0; }
+
+    while (ip<end) {
+        assert(*ip <= maxSymbolValue);
+        count[*ip++]++;
+    }
+
+    while (!count[maxSymbolValue]) maxSymbolValue--;
+    *maxSymbolValuePtr = maxSymbolValue;
+
+    {   U32 s;
+        for (s=0; s<=maxSymbolValue; s++)
+            if (count[s] > largestCount) largestCount = count[s];
+    }
+
+    return largestCount;
+}
+
+typedef enum { trustInput, checkMaxSymbolValue } HIST_checkInput_e;
+
+/* HIST_count_parallel_wksp() :
+ * store histogram into 4 intermediate tables, recombined at the end.
+ * this design makes better use of OoO cpus,
+ * and is noticeably faster when some values are heavily repeated.
+ * But it needs some additional workspace for intermediate tables.
+ * `workSpace` must be a U32 table of size >= HIST_WKSP_SIZE_U32.
+ * @return : largest histogram frequency,
+ *           or an error code (notably when histogram's alphabet is larger than *maxSymbolValuePtr) */
+static size_t HIST_count_parallel_wksp(
+                                unsigned* count, unsigned* maxSymbolValuePtr,
+                                const void* source, size_t sourceSize,
+                                HIST_checkInput_e check,
+                                U32* const workSpace)
+{
+    const BYTE* ip = (const BYTE*)source;
+    const BYTE* const iend = ip+sourceSize;
+    size_t const countSize = (*maxSymbolValuePtr + 1) * sizeof(*count);
+    unsigned max=0;
+    U32* const Counting1 = workSpace;
+    U32* const Counting2 = Counting1 + 256;
+    U32* const Counting3 = Counting2 + 256;
+    U32* const Counting4 = Counting3 + 256;
+
+    /* safety checks */
+    assert(*maxSymbolValuePtr <= 255);
+    if (!sourceSize) {
+        ZSTD_memset(count, 0, countSize);
+        *maxSymbolValuePtr = 0;
+        return 0;
+    }
+    ZSTD_memset(workSpace, 0, 4*256*sizeof(unsigned));
+
+    /* by stripes of 16 bytes */
+    {   U32 cached = MEM_read32(ip); ip += 4;
+        while (ip < iend-15) {
+            U32 c = cached; cached = MEM_read32(ip); ip += 4;
+            Counting1[(BYTE) c     ]++;
+            Counting2[(BYTE)(c>>8) ]++;
+            Counting3[(BYTE)(c>>16)]++;
+            Counting4[       c>>24 ]++;
+            c = cached; cached = MEM_read32(ip); ip += 4;
+            Counting1[(BYTE) c     ]++;
+            Counting2[(BYTE)(c>>8) ]++;
+            Counting3[(BYTE)(c>>16)]++;
+            Counting4[       c>>24 ]++;
+            c = cached; cached = MEM_read32(ip); ip += 4;
+            Counting1[(BYTE) c     ]++;
+            Counting2[(BYTE)(c>>8) ]++;
+            Counting3[(BYTE)(c>>16)]++;
+            Counting4[       c>>24 ]++;
+            c = cached; cached = MEM_read32(ip); ip += 4;
+            Counting1[(BYTE) c     ]++;
+            Counting2[(BYTE)(c>>8) ]++;
+            Counting3[(BYTE)(c>>16)]++;
+            Counting4[       c>>24 ]++;
+        }
+        ip-=4;
+    }
+
+    /* finish last symbols */
+    while (ip<iend) Counting1[*ip++]++;
+
+    {   U32 s;
+        for (s=0; s<256; s++) {
+            Counting1[s] += Counting2[s] + Counting3[s] + Counting4[s];
+            if (Counting1[s] > max) max = Counting1[s];
+    }   }
+
+    {   unsigned maxSymbolValue = 255;
+        while (!Counting1[maxSymbolValue]) maxSymbolValue--;
+        if (check && maxSymbolValue > *maxSymbolValuePtr) return ERROR(maxSymbolValue_tooSmall);
+        *maxSymbolValuePtr = maxSymbolValue;
+        ZSTD_memmove(count, Counting1, countSize);   /* in case count & Counting1 are overlapping */
+    }
+    return (size_t)max;
+}
+
+/* HIST_countFast_wksp() :
+ * Same as HIST_countFast(), but using an externally provided scratch buffer.
+ * `workSpace` is a writable buffer which must be 4-bytes aligned,
+ * `workSpaceSize` must be >= HIST_WKSP_SIZE
+ */
+size_t HIST_countFast_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
+                          const void* source, size_t sourceSize,
+                          void* workSpace, size_t workSpaceSize)
+{
+    if (sourceSize < 1500) /* heuristic threshold */
+        return HIST_count_simple(count, maxSymbolValuePtr, source, sourceSize);
+    if ((size_t)workSpace & 3) return ERROR(GENERIC);  /* must be aligned on 4-bytes boundaries */
+    if (workSpaceSize < HIST_WKSP_SIZE) return ERROR(workSpace_tooSmall);
+    return HIST_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, trustInput, (U32*)workSpace);
+}
+
+/* HIST_count_wksp() :
+ * Same as HIST_count(), but using an externally provided scratch buffer.
+ * `workSpace` size must be table of >= HIST_WKSP_SIZE_U32 unsigned */
+size_t HIST_count_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
+                       const void* source, size_t sourceSize,
+                       void* workSpace, size_t workSpaceSize)
+{
+    if ((size_t)workSpace & 3) return ERROR(GENERIC);  /* must be aligned on 4-bytes boundaries */
+    if (workSpaceSize < HIST_WKSP_SIZE) return ERROR(workSpace_tooSmall);
+    if (*maxSymbolValuePtr < 255)
+        return HIST_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, checkMaxSymbolValue, (U32*)workSpace);
+    *maxSymbolValuePtr = 255;
+    return HIST_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, workSpace, workSpaceSize);
+}
+
diff --git a/lib/zstd/compress/hist.h b/lib/zstd/compress/hist.h
new file mode 100644
index 0000000..fc1830a
--- /dev/null
+++ b/lib/zstd/compress/hist.h
@@ -0,0 +1,75 @@
+/* ******************************************************************
+ * hist : Histogram functions
+ * part of Finite State Entropy project
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ *  You can contact the author at :
+ *  - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+/* --- dependencies --- */
+#include "../common/zstd_deps.h"   /* size_t */
+
+
+/* --- simple histogram functions --- */
+
+/*! HIST_count():
+ *  Provides the precise count of each byte within a table 'count'.
+ * 'count' is a table of unsigned int, of minimum size (*maxSymbolValuePtr+1).
+ *  Updates *maxSymbolValuePtr with actual largest symbol value detected.
+ * @return : count of the most frequent symbol (which isn't identified).
+ *           or an error code, which can be tested using HIST_isError().
+ *           note : if return == srcSize, there is only one symbol.
+ */
+size_t HIST_count(unsigned* count, unsigned* maxSymbolValuePtr,
+                  const void* src, size_t srcSize);
+
+unsigned HIST_isError(size_t code);  /*< tells if a return value is an error code */
+
+
+/* --- advanced histogram functions --- */
+
+#define HIST_WKSP_SIZE_U32 1024
+#define HIST_WKSP_SIZE    (HIST_WKSP_SIZE_U32 * sizeof(unsigned))
+/* HIST_count_wksp() :
+ *  Same as HIST_count(), but using an externally provided scratch buffer.
+ *  Benefit is this function will use very little stack space.
+ * `workSpace` is a writable buffer which must be 4-bytes aligned,
+ * `workSpaceSize` must be >= HIST_WKSP_SIZE
+ */
+size_t HIST_count_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
+                       const void* src, size_t srcSize,
+                       void* workSpace, size_t workSpaceSize);
+
+/* HIST_countFast() :
+ *  same as HIST_count(), but blindly trusts that all byte values within src are <= *maxSymbolValuePtr.
+ *  This function is unsafe, and will segfault if any value within `src` is `> *maxSymbolValuePtr`
+ */
+size_t HIST_countFast(unsigned* count, unsigned* maxSymbolValuePtr,
+                      const void* src, size_t srcSize);
+
+/* HIST_countFast_wksp() :
+ *  Same as HIST_countFast(), but using an externally provided scratch buffer.
+ * `workSpace` is a writable buffer which must be 4-bytes aligned,
+ * `workSpaceSize` must be >= HIST_WKSP_SIZE
+ */
+size_t HIST_countFast_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
+                           const void* src, size_t srcSize,
+                           void* workSpace, size_t workSpaceSize);
+
+/*! HIST_count_simple() :
+ *  Same as HIST_countFast(), this function is unsafe,
+ *  and will segfault if any value within `src` is `> *maxSymbolValuePtr`.
+ *  It is also a bit slower for large inputs.
+ *  However, it does not need any additional memory (not even on stack).
+ * @return : count of the most frequent symbol.
+ *  Note this function doesn't produce any error (i.e. it must succeed).
+ */
+unsigned HIST_count_simple(unsigned* count, unsigned* maxSymbolValuePtr,
+                           const void* src, size_t srcSize);
diff --git a/lib/zstd/compress/huf_compress.c b/lib/zstd/compress/huf_compress.c
new file mode 100644
index 0000000..f76a526
--- /dev/null
+++ b/lib/zstd/compress/huf_compress.c
@@ -0,0 +1,905 @@
+/* ******************************************************************
+ * Huffman encoder, part of New Generation Entropy library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ *  You can contact the author at :
+ *  - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+/* **************************************************************
+*  Compiler specifics
+****************************************************************/
+
+
+/* **************************************************************
+*  Includes
+****************************************************************/
+#include "../common/zstd_deps.h"     /* ZSTD_memcpy, ZSTD_memset */
+#include "../common/compiler.h"
+#include "../common/bitstream.h"
+#include "hist.h"
+#define FSE_STATIC_LINKING_ONLY   /* FSE_optimalTableLog_internal */
+#include "../common/fse.h"        /* header compression */
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include "../common/error_private.h"
+
+
+/* **************************************************************
+*  Error Management
+****************************************************************/
+#define HUF_isError ERR_isError
+#define HUF_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c)   /* use only *after* variable declarations */
+
+
+/* **************************************************************
+*  Utils
+****************************************************************/
+unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
+{
+    return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 1);
+}
+
+
+/* *******************************************************
+*  HUF : Huffman block compression
+*********************************************************/
+/* HUF_compressWeights() :
+ * Same as FSE_compress(), but dedicated to huff0's weights compression.
+ * The use case needs much less stack memory.
+ * Note : all elements within weightTable are supposed to be <= HUF_TABLELOG_MAX.
+ */
+#define MAX_FSE_TABLELOG_FOR_HUFF_HEADER 6
+
+typedef struct {
+    FSE_CTable CTable[FSE_CTABLE_SIZE_U32(MAX_FSE_TABLELOG_FOR_HUFF_HEADER, HUF_TABLELOG_MAX)];
+    U32 scratchBuffer[FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(HUF_TABLELOG_MAX, MAX_FSE_TABLELOG_FOR_HUFF_HEADER)];
+    unsigned count[HUF_TABLELOG_MAX+1];
+    S16 norm[HUF_TABLELOG_MAX+1];
+} HUF_CompressWeightsWksp;
+
+static size_t HUF_compressWeights(void* dst, size_t dstSize, const void* weightTable, size_t wtSize, void* workspace, size_t workspaceSize)
+{
+    BYTE* const ostart = (BYTE*) dst;
+    BYTE* op = ostart;
+    BYTE* const oend = ostart + dstSize;
+
+    unsigned maxSymbolValue = HUF_TABLELOG_MAX;
+    U32 tableLog = MAX_FSE_TABLELOG_FOR_HUFF_HEADER;
+    HUF_CompressWeightsWksp* wksp = (HUF_CompressWeightsWksp*)workspace;
+
+    if (workspaceSize < sizeof(HUF_CompressWeightsWksp)) return ERROR(GENERIC);
+
+    /* init conditions */
+    if (wtSize <= 1) return 0;  /* Not compressible */
+
+    /* Scan input and build symbol stats */
+    {   unsigned const maxCount = HIST_count_simple(wksp->count, &maxSymbolValue, weightTable, wtSize);   /* never fails */
+        if (maxCount == wtSize) return 1;   /* only a single symbol in src : rle */
+        if (maxCount == 1) return 0;        /* each symbol present maximum once => not compressible */
+    }
+
+    tableLog = FSE_optimalTableLog(tableLog, wtSize, maxSymbolValue);
+    CHECK_F( FSE_normalizeCount(wksp->norm, tableLog, wksp->count, wtSize, maxSymbolValue, /* useLowProbCount */ 0) );
+
+    /* Write table description header */
+    {   CHECK_V_F(hSize, FSE_writeNCount(op, (size_t)(oend-op), wksp->norm, maxSymbolValue, tableLog) );
+        op += hSize;
+    }
+
+    /* Compress */
+    CHECK_F( FSE_buildCTable_wksp(wksp->CTable, wksp->norm, maxSymbolValue, tableLog, wksp->scratchBuffer, sizeof(wksp->scratchBuffer)) );
+    {   CHECK_V_F(cSize, FSE_compress_usingCTable(op, (size_t)(oend - op), weightTable, wtSize, wksp->CTable) );
+        if (cSize == 0) return 0;   /* not enough space for compressed data */
+        op += cSize;
+    }
+
+    return (size_t)(op-ostart);
+}
+
+
+typedef struct {
+    HUF_CompressWeightsWksp wksp;
+    BYTE bitsToWeight[HUF_TABLELOG_MAX + 1];   /* precomputed conversion table */
+    BYTE huffWeight[HUF_SYMBOLVALUE_MAX];
+} HUF_WriteCTableWksp;
+
+size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize,
+                            const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog,
+                            void* workspace, size_t workspaceSize)
+{
+    BYTE* op = (BYTE*)dst;
+    U32 n;
+    HUF_WriteCTableWksp* wksp = (HUF_WriteCTableWksp*)workspace;
+
+    /* check conditions */
+    if (workspaceSize < sizeof(HUF_WriteCTableWksp)) return ERROR(GENERIC);
+    if (maxSymbolValue > HUF_SYMBOLVALUE_MAX) return ERROR(maxSymbolValue_tooLarge);
+
+    /* convert to weight */
+    wksp->bitsToWeight[0] = 0;
+    for (n=1; n<huffLog+1; n++)
+        wksp->bitsToWeight[n] = (BYTE)(huffLog + 1 - n);
+    for (n=0; n<maxSymbolValue; n++)
+        wksp->huffWeight[n] = wksp->bitsToWeight[CTable[n].nbBits];
+
+    /* attempt weights compression by FSE */
+    {   CHECK_V_F(hSize, HUF_compressWeights(op+1, maxDstSize-1, wksp->huffWeight, maxSymbolValue, &wksp->wksp, sizeof(wksp->wksp)) );
+        if ((hSize>1) & (hSize < maxSymbolValue/2)) {   /* FSE compressed */
+            op[0] = (BYTE)hSize;
+            return hSize+1;
+    }   }
+
+    /* write raw values as 4-bits (max : 15) */
+    if (maxSymbolValue > (256-128)) return ERROR(GENERIC);   /* should not happen : likely means source cannot be compressed */
+    if (((maxSymbolValue+1)/2) + 1 > maxDstSize) return ERROR(dstSize_tooSmall);   /* not enough space within dst buffer */
+    op[0] = (BYTE)(128 /*special case*/ + (maxSymbolValue-1));
+    wksp->huffWeight[maxSymbolValue] = 0;   /* to be sure it doesn't cause msan issue in final combination */
+    for (n=0; n<maxSymbolValue; n+=2)
+        op[(n/2)+1] = (BYTE)((wksp->huffWeight[n] << 4) + wksp->huffWeight[n+1]);
+    return ((maxSymbolValue+1)/2) + 1;
+}
+
+/*! HUF_writeCTable() :
+    `CTable` : Huffman tree to save, using huf representation.
+    @return : size of saved CTable */
+size_t HUF_writeCTable (void* dst, size_t maxDstSize,
+                        const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog)
+{
+    HUF_WriteCTableWksp wksp;
+    return HUF_writeCTable_wksp(dst, maxDstSize, CTable, maxSymbolValue, huffLog, &wksp, sizeof(wksp));
+}
+
+
+size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize, unsigned* hasZeroWeights)
+{
+    BYTE huffWeight[HUF_SYMBOLVALUE_MAX + 1];   /* init not required, even though some static analyzer may complain */
+    U32 rankVal[HUF_TABLELOG_ABSOLUTEMAX + 1];   /* large enough for values from 0 to 16 */
+    U32 tableLog = 0;
+    U32 nbSymbols = 0;
+
+    /* get symbol weights */
+    CHECK_V_F(readSize, HUF_readStats(huffWeight, HUF_SYMBOLVALUE_MAX+1, rankVal, &nbSymbols, &tableLog, src, srcSize));
+    *hasZeroWeights = (rankVal[0] > 0);
+
+    /* check result */
+    if (tableLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
+    if (nbSymbols > *maxSymbolValuePtr+1) return ERROR(maxSymbolValue_tooSmall);
+
+    /* Prepare base value per rank */
+    {   U32 n, nextRankStart = 0;
+        for (n=1; n<=tableLog; n++) {
+            U32 curr = nextRankStart;
+            nextRankStart += (rankVal[n] << (n-1));
+            rankVal[n] = curr;
+    }   }
+
+    /* fill nbBits */
+    {   U32 n; for (n=0; n<nbSymbols; n++) {
+            const U32 w = huffWeight[n];
+            CTable[n].nbBits = (BYTE)(tableLog + 1 - w) & -(w != 0);
+    }   }
+
+    /* fill val */
+    {   U16 nbPerRank[HUF_TABLELOG_MAX+2]  = {0};  /* support w=0=>n=tableLog+1 */
+        U16 valPerRank[HUF_TABLELOG_MAX+2] = {0};
+        { U32 n; for (n=0; n<nbSymbols; n++) nbPerRank[CTable[n].nbBits]++; }
+        /* determine stating value per rank */
+        valPerRank[tableLog+1] = 0;   /* for w==0 */
+        {   U16 min = 0;
+            U32 n; for (n=tableLog; n>0; n--) {  /* start at n=tablelog <-> w=1 */
+                valPerRank[n] = min;     /* get starting value within each rank */
+                min += nbPerRank[n];
+                min >>= 1;
+        }   }
+        /* assign value within rank, symbol order */
+        { U32 n; for (n=0; n<nbSymbols; n++) CTable[n].val = valPerRank[CTable[n].nbBits]++; }
+    }
+
+    *maxSymbolValuePtr = nbSymbols - 1;
+    return readSize;
+}
+
+U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue)
+{
+    const HUF_CElt* table = (const HUF_CElt*)symbolTable;
+    assert(symbolValue <= HUF_SYMBOLVALUE_MAX);
+    return table[symbolValue].nbBits;
+}
+
+
+typedef struct nodeElt_s {
+    U32 count;
+    U16 parent;
+    BYTE byte;
+    BYTE nbBits;
+} nodeElt;
+
+/*
+ * HUF_setMaxHeight():
+ * Enforces maxNbBits on the Huffman tree described in huffNode.
+ *
+ * It sets all nodes with nbBits > maxNbBits to be maxNbBits. Then it adjusts
+ * the tree to so that it is a valid canonical Huffman tree.
+ *
+ * @pre               The sum of the ranks of each symbol == 2^largestBits,
+ *                    where largestBits == huffNode[lastNonNull].nbBits.
+ * @post              The sum of the ranks of each symbol == 2^largestBits,
+ *                    where largestBits is the return value <= maxNbBits.
+ *
+ * @param huffNode    The Huffman tree modified in place to enforce maxNbBits.
+ * @param lastNonNull The symbol with the lowest count in the Huffman tree.
+ * @param maxNbBits   The maximum allowed number of bits, which the Huffman tree
+ *                    may not respect. After this function the Huffman tree will
+ *                    respect maxNbBits.
+ * @return            The maximum number of bits of the Huffman tree after adjustment,
+ *                    necessarily no more than maxNbBits.
+ */
+static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 maxNbBits)
+{
+    const U32 largestBits = huffNode[lastNonNull].nbBits;
+    /* early exit : no elt > maxNbBits, so the tree is already valid. */
+    if (largestBits <= maxNbBits) return largestBits;
+
+    /* there are several too large elements (at least >= 2) */
+    {   int totalCost = 0;
+        const U32 baseCost = 1 << (largestBits - maxNbBits);
+        int n = (int)lastNonNull;
+
+        /* Adjust any ranks > maxNbBits to maxNbBits.
+         * Compute totalCost, which is how far the sum of the ranks is
+         * we are over 2^largestBits after adjust the offending ranks.
+         */
+        while (huffNode[n].nbBits > maxNbBits) {
+            totalCost += baseCost - (1 << (largestBits - huffNode[n].nbBits));
+            huffNode[n].nbBits = (BYTE)maxNbBits;
+            n--;
+        }
+        /* n stops at huffNode[n].nbBits <= maxNbBits */
+        assert(huffNode[n].nbBits <= maxNbBits);
+        /* n end at index of smallest symbol using < maxNbBits */
+        while (huffNode[n].nbBits == maxNbBits) --n;
+
+        /* renorm totalCost from 2^largestBits to 2^maxNbBits
+         * note : totalCost is necessarily a multiple of baseCost */
+        assert((totalCost & (baseCost - 1)) == 0);
+        totalCost >>= (largestBits - maxNbBits);
+        assert(totalCost > 0);
+
+        /* repay normalized cost */
+        {   U32 const noSymbol = 0xF0F0F0F0;
+            U32 rankLast[HUF_TABLELOG_MAX+2];
+
+            /* Get pos of last (smallest = lowest cum. count) symbol per rank */
+            ZSTD_memset(rankLast, 0xF0, sizeof(rankLast));
+            {   U32 currentNbBits = maxNbBits;
+                int pos;
+                for (pos=n ; pos >= 0; pos--) {
+                    if (huffNode[pos].nbBits >= currentNbBits) continue;
+                    currentNbBits = huffNode[pos].nbBits;   /* < maxNbBits */
+                    rankLast[maxNbBits-currentNbBits] = (U32)pos;
+            }   }
+
+            while (totalCost > 0) {
+                /* Try to reduce the next power of 2 above totalCost because we
+                 * gain back half the rank.
+                 */
+                U32 nBitsToDecrease = BIT_highbit32((U32)totalCost) + 1;
+                for ( ; nBitsToDecrease > 1; nBitsToDecrease--) {
+                    U32 const highPos = rankLast[nBitsToDecrease];
+                    U32 const lowPos = rankLast[nBitsToDecrease-1];
+                    if (highPos == noSymbol) continue;
+                    /* Decrease highPos if no symbols of lowPos or if it is
+                     * not cheaper to remove 2 lowPos than highPos.
+                     */
+                    if (lowPos == noSymbol) break;
+                    {   U32 const highTotal = huffNode[highPos].count;
+                        U32 const lowTotal = 2 * huffNode[lowPos].count;
+                        if (highTotal <= lowTotal) break;
+                }   }
+                /* only triggered when no more rank 1 symbol left => find closest one (note : there is necessarily at least one !) */
+                assert(rankLast[nBitsToDecrease] != noSymbol || nBitsToDecrease == 1);
+                /* HUF_MAX_TABLELOG test just to please gcc 5+; but it should not be necessary */
+                while ((nBitsToDecrease<=HUF_TABLELOG_MAX) && (rankLast[nBitsToDecrease] == noSymbol))
+                    nBitsToDecrease++;
+                assert(rankLast[nBitsToDecrease] != noSymbol);
+                /* Increase the number of bits to gain back half the rank cost. */
+                totalCost -= 1 << (nBitsToDecrease-1);
+                huffNode[rankLast[nBitsToDecrease]].nbBits++;
+
+                /* Fix up the new rank.
+                 * If the new rank was empty, this symbol is now its smallest.
+                 * Otherwise, this symbol will be the largest in the new rank so no adjustment.
+                 */
+                if (rankLast[nBitsToDecrease-1] == noSymbol)
+                    rankLast[nBitsToDecrease-1] = rankLast[nBitsToDecrease];
+                /* Fix up the old rank.
+                 * If the symbol was at position 0, meaning it was the highest weight symbol in the tree,
+                 * it must be the only symbol in its rank, so the old rank now has no symbols.
+                 * Otherwise, since the Huffman nodes are sorted by count, the previous position is now
+                 * the smallest node in the rank. If the previous position belongs to a different rank,
+                 * then the rank is now empty.
+                 */
+                if (rankLast[nBitsToDecrease] == 0)    /* special case, reached largest symbol */
+                    rankLast[nBitsToDecrease] = noSymbol;
+                else {
+                    rankLast[nBitsToDecrease]--;
+                    if (huffNode[rankLast[nBitsToDecrease]].nbBits != maxNbBits-nBitsToDecrease)
+                        rankLast[nBitsToDecrease] = noSymbol;   /* this rank is now empty */
+                }
+            }   /* while (totalCost > 0) */
+
+            /* If we've removed too much weight, then we have to add it back.
+             * To avoid overshooting again, we only adjust the smallest rank.
+             * We take the largest nodes from the lowest rank 0 and move them
+             * to rank 1. There's guaranteed to be enough rank 0 symbols because
+             * TODO.
+             */
+            while (totalCost < 0) {  /* Sometimes, cost correction overshoot */
+                /* special case : no rank 1 symbol (using maxNbBits-1);
+                 * let's create one from largest rank 0 (using maxNbBits).
+                 */
+                if (rankLast[1] == noSymbol) {
+                    while (huffNode[n].nbBits == maxNbBits) n--;
+                    huffNode[n+1].nbBits--;
+                    assert(n >= 0);
+                    rankLast[1] = (U32)(n+1);
+                    totalCost++;
+                    continue;
+                }
+                huffNode[ rankLast[1] + 1 ].nbBits--;
+                rankLast[1]++;
+                totalCost ++;
+            }
+        }   /* repay normalized cost */
+    }   /* there are several too large elements (at least >= 2) */
+
+    return maxNbBits;
+}
+
+typedef struct {
+    U32 base;
+    U32 curr;
+} rankPos;
+
+typedef nodeElt huffNodeTable[HUF_CTABLE_WORKSPACE_SIZE_U32];
+
+#define RANK_POSITION_TABLE_SIZE 32
+
+typedef struct {
+  huffNodeTable huffNodeTbl;
+  rankPos rankPosition[RANK_POSITION_TABLE_SIZE];
+} HUF_buildCTable_wksp_tables;
+
+/*
+ * HUF_sort():
+ * Sorts the symbols [0, maxSymbolValue] by count[symbol] in decreasing order.
+ *
+ * @param[out] huffNode       Sorted symbols by decreasing count. Only members `.count` and `.byte` are filled.
+ *                            Must have (maxSymbolValue + 1) entries.
+ * @param[in]  count          Histogram of the symbols.
+ * @param[in]  maxSymbolValue Maximum symbol value.
+ * @param      rankPosition   This is a scratch workspace. Must have RANK_POSITION_TABLE_SIZE entries.
+ */
+static void HUF_sort(nodeElt* huffNode, const unsigned* count, U32 maxSymbolValue, rankPos* rankPosition)
+{
+    int n;
+    int const maxSymbolValue1 = (int)maxSymbolValue + 1;
+
+    /* Compute base and set curr to base.
+     * For symbol s let lowerRank = BIT_highbit32(count[n]+1) and rank = lowerRank + 1.
+     * Then 2^lowerRank <= count[n]+1 <= 2^rank.
+     * We attribute each symbol to lowerRank's base value, because we want to know where
+     * each rank begins in the output, so for rank R we want to count ranks R+1 and above.
+     */
+    ZSTD_memset(rankPosition, 0, sizeof(*rankPosition) * RANK_POSITION_TABLE_SIZE);
+    for (n = 0; n < maxSymbolValue1; ++n) {
+        U32 lowerRank = BIT_highbit32(count[n] + 1);
+        rankPosition[lowerRank].base++;
+    }
+    assert(rankPosition[RANK_POSITION_TABLE_SIZE - 1].base == 0);
+    for (n = RANK_POSITION_TABLE_SIZE - 1; n > 0; --n) {
+        rankPosition[n-1].base += rankPosition[n].base;
+        rankPosition[n-1].curr = rankPosition[n-1].base;
+    }
+    /* Sort */
+    for (n = 0; n < maxSymbolValue1; ++n) {
+        U32 const c = count[n];
+        U32 const r = BIT_highbit32(c+1) + 1;
+        U32 pos = rankPosition[r].curr++;
+        /* Insert into the correct position in the rank.
+         * We have at most 256 symbols, so this insertion should be fine.
+         */
+        while ((pos > rankPosition[r].base) && (c > huffNode[pos-1].count)) {
+            huffNode[pos] = huffNode[pos-1];
+            pos--;
+        }
+        huffNode[pos].count = c;
+        huffNode[pos].byte  = (BYTE)n;
+    }
+}
+
+
+/* HUF_buildCTable_wksp() :
+ *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
+ *  `workSpace` must be aligned on 4-bytes boundaries, and be at least as large as sizeof(HUF_buildCTable_wksp_tables).
+ */
+#define STARTNODE (HUF_SYMBOLVALUE_MAX+1)
+
+/* HUF_buildTree():
+ * Takes the huffNode array sorted by HUF_sort() and builds an unlimited-depth Huffman tree.
+ *
+ * @param huffNode        The array sorted by HUF_sort(). Builds the Huffman tree in this array.
+ * @param maxSymbolValue  The maximum symbol value.
+ * @return                The smallest node in the Huffman tree (by count).
+ */
+static int HUF_buildTree(nodeElt* huffNode, U32 maxSymbolValue)
+{
+    nodeElt* const huffNode0 = huffNode - 1;
+    int nonNullRank;
+    int lowS, lowN;
+    int nodeNb = STARTNODE;
+    int n, nodeRoot;
+    /* init for parents */
+    nonNullRank = (int)maxSymbolValue;
+    while(huffNode[nonNullRank].count == 0) nonNullRank--;
+    lowS = nonNullRank; nodeRoot = nodeNb + lowS - 1; lowN = nodeNb;
+    huffNode[nodeNb].count = huffNode[lowS].count + huffNode[lowS-1].count;
+    huffNode[lowS].parent = huffNode[lowS-1].parent = (U16)nodeNb;
+    nodeNb++; lowS-=2;
+    for (n=nodeNb; n<=nodeRoot; n++) huffNode[n].count = (U32)(1U<<30);
+    huffNode0[0].count = (U32)(1U<<31);  /* fake entry, strong barrier */
+
+    /* create parents */
+    while (nodeNb <= nodeRoot) {
+        int const n1 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++;
+        int const n2 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++;
+        huffNode[nodeNb].count = huffNode[n1].count + huffNode[n2].count;
+        huffNode[n1].parent = huffNode[n2].parent = (U16)nodeNb;
+        nodeNb++;
+    }
+
+    /* distribute weights (unlimited tree height) */
+    huffNode[nodeRoot].nbBits = 0;
+    for (n=nodeRoot-1; n>=STARTNODE; n--)
+        huffNode[n].nbBits = huffNode[ huffNode[n].parent ].nbBits + 1;
+    for (n=0; n<=nonNullRank; n++)
+        huffNode[n].nbBits = huffNode[ huffNode[n].parent ].nbBits + 1;
+
+    return nonNullRank;
+}
+
+/*
+ * HUF_buildCTableFromTree():
+ * Build the CTable given the Huffman tree in huffNode.
+ *
+ * @param[out] CTable         The output Huffman CTable.
+ * @param      huffNode       The Huffman tree.
+ * @param      nonNullRank    The last and smallest node in the Huffman tree.
+ * @param      maxSymbolValue The maximum symbol value.
+ * @param      maxNbBits      The exact maximum number of bits used in the Huffman tree.
+ */
+static void HUF_buildCTableFromTree(HUF_CElt* CTable, nodeElt const* huffNode, int nonNullRank, U32 maxSymbolValue, U32 maxNbBits)
+{
+    /* fill result into ctable (val, nbBits) */
+    int n;
+    U16 nbPerRank[HUF_TABLELOG_MAX+1] = {0};
+    U16 valPerRank[HUF_TABLELOG_MAX+1] = {0};
+    int const alphabetSize = (int)(maxSymbolValue + 1);
+    for (n=0; n<=nonNullRank; n++)
+        nbPerRank[huffNode[n].nbBits]++;
+    /* determine starting value per rank */
+    {   U16 min = 0;
+        for (n=(int)maxNbBits; n>0; n--) {
+            valPerRank[n] = min;      /* get starting value within each rank */
+            min += nbPerRank[n];
+            min >>= 1;
+    }   }
+    for (n=0; n<alphabetSize; n++)
+        CTable[huffNode[n].byte].nbBits = huffNode[n].nbBits;   /* push nbBits per symbol, symbol order */
+    for (n=0; n<alphabetSize; n++)
+        CTable[n].val = valPerRank[CTable[n].nbBits]++;   /* assign value within rank, symbol order */
+}
+
+size_t HUF_buildCTable_wksp (HUF_CElt* tree, const unsigned* count, U32 maxSymbolValue, U32 maxNbBits, void* workSpace, size_t wkspSize)
+{
+    HUF_buildCTable_wksp_tables* const wksp_tables = (HUF_buildCTable_wksp_tables*)workSpace;
+    nodeElt* const huffNode0 = wksp_tables->huffNodeTbl;
+    nodeElt* const huffNode = huffNode0+1;
+    int nonNullRank;
+
+    /* safety checks */
+    if (((size_t)workSpace & 3) != 0) return ERROR(GENERIC);  /* must be aligned on 4-bytes boundaries */
+    if (wkspSize < sizeof(HUF_buildCTable_wksp_tables))
+      return ERROR(workSpace_tooSmall);
+    if (maxNbBits == 0) maxNbBits = HUF_TABLELOG_DEFAULT;
+    if (maxSymbolValue > HUF_SYMBOLVALUE_MAX)
+      return ERROR(maxSymbolValue_tooLarge);
+    ZSTD_memset(huffNode0, 0, sizeof(huffNodeTable));
+
+    /* sort, decreasing order */
+    HUF_sort(huffNode, count, maxSymbolValue, wksp_tables->rankPosition);
+
+    /* build tree */
+    nonNullRank = HUF_buildTree(huffNode, maxSymbolValue);
+
+    /* enforce maxTableLog */
+    maxNbBits = HUF_setMaxHeight(huffNode, (U32)nonNullRank, maxNbBits);
+    if (maxNbBits > HUF_TABLELOG_MAX) return ERROR(GENERIC);   /* check fit into table */
+
+    HUF_buildCTableFromTree(tree, huffNode, nonNullRank, maxSymbolValue, maxNbBits);
+
+    return maxNbBits;
+}
+
+size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue)
+{
+    size_t nbBits = 0;
+    int s;
+    for (s = 0; s <= (int)maxSymbolValue; ++s) {
+        nbBits += CTable[s].nbBits * count[s];
+    }
+    return nbBits >> 3;
+}
+
+int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue) {
+  int bad = 0;
+  int s;
+  for (s = 0; s <= (int)maxSymbolValue; ++s) {
+    bad |= (count[s] != 0) & (CTable[s].nbBits == 0);
+  }
+  return !bad;
+}
+
+size_t HUF_compressBound(size_t size) { return HUF_COMPRESSBOUND(size); }
+
+FORCE_INLINE_TEMPLATE void
+HUF_encodeSymbol(BIT_CStream_t* bitCPtr, U32 symbol, const HUF_CElt* CTable)
+{
+    BIT_addBitsFast(bitCPtr, CTable[symbol].val, CTable[symbol].nbBits);
+}
+
+#define HUF_FLUSHBITS(s)  BIT_flushBits(s)
+
+#define HUF_FLUSHBITS_1(stream) \
+    if (sizeof((stream)->bitContainer)*8 < HUF_TABLELOG_MAX*2+7) HUF_FLUSHBITS(stream)
+
+#define HUF_FLUSHBITS_2(stream) \
+    if (sizeof((stream)->bitContainer)*8 < HUF_TABLELOG_MAX*4+7) HUF_FLUSHBITS(stream)
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_compress1X_usingCTable_internal_body(void* dst, size_t dstSize,
+                                   const void* src, size_t srcSize,
+                                   const HUF_CElt* CTable)
+{
+    const BYTE* ip = (const BYTE*) src;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + dstSize;
+    BYTE* op = ostart;
+    size_t n;
+    BIT_CStream_t bitC;
+
+    /* init */
+    if (dstSize < 8) return 0;   /* not enough space to compress */
+    { size_t const initErr = BIT_initCStream(&bitC, op, (size_t)(oend-op));
+      if (HUF_isError(initErr)) return 0; }
+
+    n = srcSize & ~3;  /* join to mod 4 */
+    switch (srcSize & 3)
+    {
+        case 3:
+            HUF_encodeSymbol(&bitC, ip[n+ 2], CTable);
+            HUF_FLUSHBITS_2(&bitC);
+            ZSTD_FALLTHROUGH;
+        case 2:
+            HUF_encodeSymbol(&bitC, ip[n+ 1], CTable);
+            HUF_FLUSHBITS_1(&bitC);
+            ZSTD_FALLTHROUGH;
+        case 1:
+            HUF_encodeSymbol(&bitC, ip[n+ 0], CTable);
+            HUF_FLUSHBITS(&bitC);
+            ZSTD_FALLTHROUGH;
+        case 0: ZSTD_FALLTHROUGH;
+        default: break;
+    }
+
+    for (; n>0; n-=4) {  /* note : n&3==0 at this stage */
+        HUF_encodeSymbol(&bitC, ip[n- 1], CTable);
+        HUF_FLUSHBITS_1(&bitC);
+        HUF_encodeSymbol(&bitC, ip[n- 2], CTable);
+        HUF_FLUSHBITS_2(&bitC);
+        HUF_encodeSymbol(&bitC, ip[n- 3], CTable);
+        HUF_FLUSHBITS_1(&bitC);
+        HUF_encodeSymbol(&bitC, ip[n- 4], CTable);
+        HUF_FLUSHBITS(&bitC);
+    }
+
+    return BIT_closeCStream(&bitC);
+}
+
+#if DYNAMIC_BMI2
+
+static TARGET_ATTRIBUTE("bmi2") size_t
+HUF_compress1X_usingCTable_internal_bmi2(void* dst, size_t dstSize,
+                                   const void* src, size_t srcSize,
+                                   const HUF_CElt* CTable)
+{
+    return HUF_compress1X_usingCTable_internal_body(dst, dstSize, src, srcSize, CTable);
+}
+
+static size_t
+HUF_compress1X_usingCTable_internal_default(void* dst, size_t dstSize,
+                                      const void* src, size_t srcSize,
+                                      const HUF_CElt* CTable)
+{
+    return HUF_compress1X_usingCTable_internal_body(dst, dstSize, src, srcSize, CTable);
+}
+
+static size_t
+HUF_compress1X_usingCTable_internal(void* dst, size_t dstSize,
+                              const void* src, size_t srcSize,
+                              const HUF_CElt* CTable, const int bmi2)
+{
+    if (bmi2) {
+        return HUF_compress1X_usingCTable_internal_bmi2(dst, dstSize, src, srcSize, CTable);
+    }
+    return HUF_compress1X_usingCTable_internal_default(dst, dstSize, src, srcSize, CTable);
+}
+
+#else
+
+static size_t
+HUF_compress1X_usingCTable_internal(void* dst, size_t dstSize,
+                              const void* src, size_t srcSize,
+                              const HUF_CElt* CTable, const int bmi2)
+{
+    (void)bmi2;
+    return HUF_compress1X_usingCTable_internal_body(dst, dstSize, src, srcSize, CTable);
+}
+
+#endif
+
+size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable)
+{
+    return HUF_compress1X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, /* bmi2 */ 0);
+}
+
+
+static size_t
+HUF_compress4X_usingCTable_internal(void* dst, size_t dstSize,
+                              const void* src, size_t srcSize,
+                              const HUF_CElt* CTable, int bmi2)
+{
+    size_t const segmentSize = (srcSize+3)/4;   /* first 3 segments */
+    const BYTE* ip = (const BYTE*) src;
+    const BYTE* const iend = ip + srcSize;
+    BYTE* const ostart = (BYTE*) dst;
+    BYTE* const oend = ostart + dstSize;
+    BYTE* op = ostart;
+
+    if (dstSize < 6 + 1 + 1 + 1 + 8) return 0;   /* minimum space to compress successfully */
+    if (srcSize < 12) return 0;   /* no saving possible : too small input */
+    op += 6;   /* jumpTable */
+
+    assert(op <= oend);
+    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
+        if (cSize==0) return 0;
+        assert(cSize <= 65535);
+        MEM_writeLE16(ostart, (U16)cSize);
+        op += cSize;
+    }
+
+    ip += segmentSize;
+    assert(op <= oend);
+    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
+        if (cSize==0) return 0;
+        assert(cSize <= 65535);
+        MEM_writeLE16(ostart+2, (U16)cSize);
+        op += cSize;
+    }
+
+    ip += segmentSize;
+    assert(op <= oend);
+    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
+        if (cSize==0) return 0;
+        assert(cSize <= 65535);
+        MEM_writeLE16(ostart+4, (U16)cSize);
+        op += cSize;
+    }
+
+    ip += segmentSize;
+    assert(op <= oend);
+    assert(ip <= iend);
+    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, (size_t)(iend-ip), CTable, bmi2) );
+        if (cSize==0) return 0;
+        op += cSize;
+    }
+
+    return (size_t)(op-ostart);
+}
+
+size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable)
+{
+    return HUF_compress4X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, /* bmi2 */ 0);
+}
+
+typedef enum { HUF_singleStream, HUF_fourStreams } HUF_nbStreams_e;
+
+static size_t HUF_compressCTable_internal(
+                BYTE* const ostart, BYTE* op, BYTE* const oend,
+                const void* src, size_t srcSize,
+                HUF_nbStreams_e nbStreams, const HUF_CElt* CTable, const int bmi2)
+{
+    size_t const cSize = (nbStreams==HUF_singleStream) ?
+                         HUF_compress1X_usingCTable_internal(op, (size_t)(oend - op), src, srcSize, CTable, bmi2) :
+                         HUF_compress4X_usingCTable_internal(op, (size_t)(oend - op), src, srcSize, CTable, bmi2);
+    if (HUF_isError(cSize)) { return cSize; }
+    if (cSize==0) { return 0; }   /* uncompressible */
+    op += cSize;
+    /* check compressibility */
+    assert(op >= ostart);
+    if ((size_t)(op-ostart) >= srcSize-1) { return 0; }
+    return (size_t)(op-ostart);
+}
+
+typedef struct {
+    unsigned count[HUF_SYMBOLVALUE_MAX + 1];
+    HUF_CElt CTable[HUF_SYMBOLVALUE_MAX + 1];
+    union {
+        HUF_buildCTable_wksp_tables buildCTable_wksp;
+        HUF_WriteCTableWksp writeCTable_wksp;
+    } wksps;
+} HUF_compress_tables_t;
+
+/* HUF_compress_internal() :
+ * `workSpace_align4` must be aligned on 4-bytes boundaries,
+ * and occupies the same space as a table of HUF_WORKSPACE_SIZE_U32 unsigned */
+static size_t
+HUF_compress_internal (void* dst, size_t dstSize,
+                 const void* src, size_t srcSize,
+                       unsigned maxSymbolValue, unsigned huffLog,
+                       HUF_nbStreams_e nbStreams,
+                       void* workSpace_align4, size_t wkspSize,
+                       HUF_CElt* oldHufTable, HUF_repeat* repeat, int preferRepeat,
+                 const int bmi2)
+{
+    HUF_compress_tables_t* const table = (HUF_compress_tables_t*)workSpace_align4;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + dstSize;
+    BYTE* op = ostart;
+
+    HUF_STATIC_ASSERT(sizeof(*table) <= HUF_WORKSPACE_SIZE);
+    assert(((size_t)workSpace_align4 & 3) == 0);   /* must be aligned on 4-bytes boundaries */
+
+    /* checks & inits */
+    if (wkspSize < HUF_WORKSPACE_SIZE) return ERROR(workSpace_tooSmall);
+    if (!srcSize) return 0;  /* Uncompressed */
+    if (!dstSize) return 0;  /* cannot fit anything within dst budget */
+    if (srcSize > HUF_BLOCKSIZE_MAX) return ERROR(srcSize_wrong);   /* current block size limit */
+    if (huffLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
+    if (maxSymbolValue > HUF_SYMBOLVALUE_MAX) return ERROR(maxSymbolValue_tooLarge);
+    if (!maxSymbolValue) maxSymbolValue = HUF_SYMBOLVALUE_MAX;
+    if (!huffLog) huffLog = HUF_TABLELOG_DEFAULT;
+
+    /* Heuristic : If old table is valid, use it for small inputs */
+    if (preferRepeat && repeat && *repeat == HUF_repeat_valid) {
+        return HUF_compressCTable_internal(ostart, op, oend,
+                                           src, srcSize,
+                                           nbStreams, oldHufTable, bmi2);
+    }
+
+    /* Scan input and build symbol stats */
+    {   CHECK_V_F(largest, HIST_count_wksp (table->count, &maxSymbolValue, (const BYTE*)src, srcSize, workSpace_align4, wkspSize) );
+        if (largest == srcSize) { *ostart = ((const BYTE*)src)[0]; return 1; }   /* single symbol, rle */
+        if (largest <= (srcSize >> 7)+4) return 0;   /* heuristic : probably not compressible enough */
+    }
+
+    /* Check validity of previous table */
+    if ( repeat
+      && *repeat == HUF_repeat_check
+      && !HUF_validateCTable(oldHufTable, table->count, maxSymbolValue)) {
+        *repeat = HUF_repeat_none;
+    }
+    /* Heuristic : use existing table for small inputs */
+    if (preferRepeat && repeat && *repeat != HUF_repeat_none) {
+        return HUF_compressCTable_internal(ostart, op, oend,
+                                           src, srcSize,
+                                           nbStreams, oldHufTable, bmi2);
+    }
+
+    /* Build Huffman Tree */
+    huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
+    {   size_t const maxBits = HUF_buildCTable_wksp(table->CTable, table->count,
+                                            maxSymbolValue, huffLog,
+                                            &table->wksps.buildCTable_wksp, sizeof(table->wksps.buildCTable_wksp));
+        CHECK_F(maxBits);
+        huffLog = (U32)maxBits;
+        /* Zero unused symbols in CTable, so we can check it for validity */
+        ZSTD_memset(table->CTable + (maxSymbolValue + 1), 0,
+               sizeof(table->CTable) - ((maxSymbolValue + 1) * sizeof(HUF_CElt)));
+    }
+
+    /* Write table description header */
+    {   CHECK_V_F(hSize, HUF_writeCTable_wksp(op, dstSize, table->CTable, maxSymbolValue, huffLog,
+                                              &table->wksps.writeCTable_wksp, sizeof(table->wksps.writeCTable_wksp)) );
+        /* Check if using previous huffman table is beneficial */
+        if (repeat && *repeat != HUF_repeat_none) {
+            size_t const oldSize = HUF_estimateCompressedSize(oldHufTable, table->count, maxSymbolValue);
+            size_t const newSize = HUF_estimateCompressedSize(table->CTable, table->count, maxSymbolValue);
+            if (oldSize <= hSize + newSize || hSize + 12 >= srcSize) {
+                return HUF_compressCTable_internal(ostart, op, oend,
+                                                   src, srcSize,
+                                                   nbStreams, oldHufTable, bmi2);
+        }   }
+
+        /* Use the new huffman table */
+        if (hSize + 12ul >= srcSize) { return 0; }
+        op += hSize;
+        if (repeat) { *repeat = HUF_repeat_none; }
+        if (oldHufTable)
+            ZSTD_memcpy(oldHufTable, table->CTable, sizeof(table->CTable));  /* Save new table */
+    }
+    return HUF_compressCTable_internal(ostart, op, oend,
+                                       src, srcSize,
+                                       nbStreams, table->CTable, bmi2);
+}
+
+
+size_t HUF_compress1X_wksp (void* dst, size_t dstSize,
+                      const void* src, size_t srcSize,
+                      unsigned maxSymbolValue, unsigned huffLog,
+                      void* workSpace, size_t wkspSize)
+{
+    return HUF_compress_internal(dst, dstSize, src, srcSize,
+                                 maxSymbolValue, huffLog, HUF_singleStream,
+                                 workSpace, wkspSize,
+                                 NULL, NULL, 0, 0 /*bmi2*/);
+}
+
+size_t HUF_compress1X_repeat (void* dst, size_t dstSize,
+                      const void* src, size_t srcSize,
+                      unsigned maxSymbolValue, unsigned huffLog,
+                      void* workSpace, size_t wkspSize,
+                      HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2)
+{
+    return HUF_compress_internal(dst, dstSize, src, srcSize,
+                                 maxSymbolValue, huffLog, HUF_singleStream,
+                                 workSpace, wkspSize, hufTable,
+                                 repeat, preferRepeat, bmi2);
+}
+
+/* HUF_compress4X_repeat():
+ * compress input using 4 streams.
+ * provide workspace to generate compression tables */
+size_t HUF_compress4X_wksp (void* dst, size_t dstSize,
+                      const void* src, size_t srcSize,
+                      unsigned maxSymbolValue, unsigned huffLog,
+                      void* workSpace, size_t wkspSize)
+{
+    return HUF_compress_internal(dst, dstSize, src, srcSize,
+                                 maxSymbolValue, huffLog, HUF_fourStreams,
+                                 workSpace, wkspSize,
+                                 NULL, NULL, 0, 0 /*bmi2*/);
+}
+
+/* HUF_compress4X_repeat():
+ * compress input using 4 streams.
+ * re-use an existing huffman compression table */
+size_t HUF_compress4X_repeat (void* dst, size_t dstSize,
+                      const void* src, size_t srcSize,
+                      unsigned maxSymbolValue, unsigned huffLog,
+                      void* workSpace, size_t wkspSize,
+                      HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2)
+{
+    return HUF_compress_internal(dst, dstSize, src, srcSize,
+                                 maxSymbolValue, huffLog, HUF_fourStreams,
+                                 workSpace, wkspSize,
+                                 hufTable, repeat, preferRepeat, bmi2);
+}
+
diff --git a/lib/zstd/compress/zstd_compress.c b/lib/zstd/compress/zstd_compress.c
new file mode 100644
index 0000000..a4e9160
--- /dev/null
+++ b/lib/zstd/compress/zstd_compress.c
@@ -0,0 +1,5109 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#include "../common/zstd_deps.h"  /* INT_MAX, ZSTD_memset, ZSTD_memcpy */
+#include "../common/cpu.h"
+#include "../common/mem.h"
+#include "hist.h"           /* HIST_countFast_wksp */
+#define FSE_STATIC_LINKING_ONLY   /* FSE_encodeSymbol */
+#include "../common/fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include "zstd_compress_internal.h"
+#include "zstd_compress_sequences.h"
+#include "zstd_compress_literals.h"
+#include "zstd_fast.h"
+#include "zstd_double_fast.h"
+#include "zstd_lazy.h"
+#include "zstd_opt.h"
+#include "zstd_ldm.h"
+#include "zstd_compress_superblock.h"
+
+/* ***************************************************************
+*  Tuning parameters
+*****************************************************************/
+/*!
+ * COMPRESS_HEAPMODE :
+ * Select how default decompression function ZSTD_compress() allocates its context,
+ * on stack (0, default), or into heap (1).
+ * Note that functions with explicit context such as ZSTD_compressCCtx() are unaffected.
+ */
+
+
+/*-*************************************
+*  Helper functions
+***************************************/
+/* ZSTD_compressBound()
+ * Note that the result from this function is only compatible with the "normal"
+ * full-block strategy.
+ * When there are a lot of small blocks due to frequent flush in streaming mode
+ * the overhead of headers can make the compressed data to be larger than the
+ * return value of ZSTD_compressBound().
+ */
+size_t ZSTD_compressBound(size_t srcSize) {
+    return ZSTD_COMPRESSBOUND(srcSize);
+}
+
+
+/*-*************************************
+*  Context memory management
+***************************************/
+struct ZSTD_CDict_s {
+    const void* dictContent;
+    size_t dictContentSize;
+    ZSTD_dictContentType_e dictContentType; /* The dictContentType the CDict was created with */
+    U32* entropyWorkspace; /* entropy workspace of HUF_WORKSPACE_SIZE bytes */
+    ZSTD_cwksp workspace;
+    ZSTD_matchState_t matchState;
+    ZSTD_compressedBlockState_t cBlockState;
+    ZSTD_customMem customMem;
+    U32 dictID;
+    int compressionLevel; /* 0 indicates that advanced API was used to select CDict params */
+};  /* typedef'd to ZSTD_CDict within "zstd.h" */
+
+ZSTD_CCtx* ZSTD_createCCtx(void)
+{
+    return ZSTD_createCCtx_advanced(ZSTD_defaultCMem);
+}
+
+static void ZSTD_initCCtx(ZSTD_CCtx* cctx, ZSTD_customMem memManager)
+{
+    assert(cctx != NULL);
+    ZSTD_memset(cctx, 0, sizeof(*cctx));
+    cctx->customMem = memManager;
+    cctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
+    {   size_t const err = ZSTD_CCtx_reset(cctx, ZSTD_reset_parameters);
+        assert(!ZSTD_isError(err));
+        (void)err;
+    }
+}
+
+ZSTD_CCtx* ZSTD_createCCtx_advanced(ZSTD_customMem customMem)
+{
+    ZSTD_STATIC_ASSERT(zcss_init==0);
+    ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN==(0ULL - 1));
+    if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
+    {   ZSTD_CCtx* const cctx = (ZSTD_CCtx*)ZSTD_customMalloc(sizeof(ZSTD_CCtx), customMem);
+        if (!cctx) return NULL;
+        ZSTD_initCCtx(cctx, customMem);
+        return cctx;
+    }
+}
+
+ZSTD_CCtx* ZSTD_initStaticCCtx(void* workspace, size_t workspaceSize)
+{
+    ZSTD_cwksp ws;
+    ZSTD_CCtx* cctx;
+    if (workspaceSize <= sizeof(ZSTD_CCtx)) return NULL;  /* minimum size */
+    if ((size_t)workspace & 7) return NULL;  /* must be 8-aligned */
+    ZSTD_cwksp_init(&ws, workspace, workspaceSize, ZSTD_cwksp_static_alloc);
+
+    cctx = (ZSTD_CCtx*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CCtx));
+    if (cctx == NULL) return NULL;
+
+    ZSTD_memset(cctx, 0, sizeof(ZSTD_CCtx));
+    ZSTD_cwksp_move(&cctx->workspace, &ws);
+    cctx->staticSize = workspaceSize;
+
+    /* statically sized space. entropyWorkspace never moves (but prev/next block swap places) */
+    if (!ZSTD_cwksp_check_available(&cctx->workspace, ENTROPY_WORKSPACE_SIZE + 2 * sizeof(ZSTD_compressedBlockState_t))) return NULL;
+    cctx->blockState.prevCBlock = (ZSTD_compressedBlockState_t*)ZSTD_cwksp_reserve_object(&cctx->workspace, sizeof(ZSTD_compressedBlockState_t));
+    cctx->blockState.nextCBlock = (ZSTD_compressedBlockState_t*)ZSTD_cwksp_reserve_object(&cctx->workspace, sizeof(ZSTD_compressedBlockState_t));
+    cctx->entropyWorkspace = (U32*)ZSTD_cwksp_reserve_object(&cctx->workspace, ENTROPY_WORKSPACE_SIZE);
+    cctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
+    return cctx;
+}
+
+/*
+ * Clears and frees all of the dictionaries in the CCtx.
+ */
+static void ZSTD_clearAllDicts(ZSTD_CCtx* cctx)
+{
+    ZSTD_customFree(cctx->localDict.dictBuffer, cctx->customMem);
+    ZSTD_freeCDict(cctx->localDict.cdict);
+    ZSTD_memset(&cctx->localDict, 0, sizeof(cctx->localDict));
+    ZSTD_memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict));
+    cctx->cdict = NULL;
+}
+
+static size_t ZSTD_sizeof_localDict(ZSTD_localDict dict)
+{
+    size_t const bufferSize = dict.dictBuffer != NULL ? dict.dictSize : 0;
+    size_t const cdictSize = ZSTD_sizeof_CDict(dict.cdict);
+    return bufferSize + cdictSize;
+}
+
+static void ZSTD_freeCCtxContent(ZSTD_CCtx* cctx)
+{
+    assert(cctx != NULL);
+    assert(cctx->staticSize == 0);
+    ZSTD_clearAllDicts(cctx);
+    ZSTD_cwksp_free(&cctx->workspace, cctx->customMem);
+}
+
+size_t ZSTD_freeCCtx(ZSTD_CCtx* cctx)
+{
+    if (cctx==NULL) return 0;   /* support free on NULL */
+    RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
+                    "not compatible with static CCtx");
+    {
+        int cctxInWorkspace = ZSTD_cwksp_owns_buffer(&cctx->workspace, cctx);
+        ZSTD_freeCCtxContent(cctx);
+        if (!cctxInWorkspace) {
+            ZSTD_customFree(cctx, cctx->customMem);
+        }
+    }
+    return 0;
+}
+
+
+static size_t ZSTD_sizeof_mtctx(const ZSTD_CCtx* cctx)
+{
+    (void)cctx;
+    return 0;
+}
+
+
+size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx)
+{
+    if (cctx==NULL) return 0;   /* support sizeof on NULL */
+    /* cctx may be in the workspace */
+    return (cctx->workspace.workspace == cctx ? 0 : sizeof(*cctx))
+           + ZSTD_cwksp_sizeof(&cctx->workspace)
+           + ZSTD_sizeof_localDict(cctx->localDict)
+           + ZSTD_sizeof_mtctx(cctx);
+}
+
+size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs)
+{
+    return ZSTD_sizeof_CCtx(zcs);  /* same object */
+}
+
+/* private API call, for dictBuilder only */
+const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx) { return &(ctx->seqStore); }
+
+/* Returns 1 if compression parameters are such that we should
+ * enable long distance matching (wlog >= 27, strategy >= btopt).
+ * Returns 0 otherwise.
+ */
+static U32 ZSTD_CParams_shouldEnableLdm(const ZSTD_compressionParameters* const cParams) {
+    return cParams->strategy >= ZSTD_btopt && cParams->windowLog >= 27;
+}
+
+static ZSTD_CCtx_params ZSTD_makeCCtxParamsFromCParams(
+        ZSTD_compressionParameters cParams)
+{
+    ZSTD_CCtx_params cctxParams;
+    /* should not matter, as all cParams are presumed properly defined */
+    ZSTD_CCtxParams_init(&cctxParams, ZSTD_CLEVEL_DEFAULT);
+    cctxParams.cParams = cParams;
+
+    if (ZSTD_CParams_shouldEnableLdm(&cParams)) {
+        DEBUGLOG(4, "ZSTD_makeCCtxParamsFromCParams(): Including LDM into cctx params");
+        cctxParams.ldmParams.enableLdm = 1;
+        /* LDM is enabled by default for optimal parser and window size >= 128MB */
+        ZSTD_ldm_adjustParameters(&cctxParams.ldmParams, &cParams);
+        assert(cctxParams.ldmParams.hashLog >= cctxParams.ldmParams.bucketSizeLog);
+        assert(cctxParams.ldmParams.hashRateLog < 32);
+    }
+
+    assert(!ZSTD_checkCParams(cParams));
+    return cctxParams;
+}
+
+static ZSTD_CCtx_params* ZSTD_createCCtxParams_advanced(
+        ZSTD_customMem customMem)
+{
+    ZSTD_CCtx_params* params;
+    if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
+    params = (ZSTD_CCtx_params*)ZSTD_customCalloc(
+            sizeof(ZSTD_CCtx_params), customMem);
+    if (!params) { return NULL; }
+    ZSTD_CCtxParams_init(params, ZSTD_CLEVEL_DEFAULT);
+    params->customMem = customMem;
+    return params;
+}
+
+ZSTD_CCtx_params* ZSTD_createCCtxParams(void)
+{
+    return ZSTD_createCCtxParams_advanced(ZSTD_defaultCMem);
+}
+
+size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params)
+{
+    if (params == NULL) { return 0; }
+    ZSTD_customFree(params, params->customMem);
+    return 0;
+}
+
+size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params)
+{
+    return ZSTD_CCtxParams_init(params, ZSTD_CLEVEL_DEFAULT);
+}
+
+size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel) {
+    RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!");
+    ZSTD_memset(cctxParams, 0, sizeof(*cctxParams));
+    cctxParams->compressionLevel = compressionLevel;
+    cctxParams->fParams.contentSizeFlag = 1;
+    return 0;
+}
+
+#define ZSTD_NO_CLEVEL 0
+
+/*
+ * Initializes the cctxParams from params and compressionLevel.
+ * @param compressionLevel If params are derived from a compression level then that compression level, otherwise ZSTD_NO_CLEVEL.
+ */
+static void ZSTD_CCtxParams_init_internal(ZSTD_CCtx_params* cctxParams, ZSTD_parameters const* params, int compressionLevel)
+{
+    assert(!ZSTD_checkCParams(params->cParams));
+    ZSTD_memset(cctxParams, 0, sizeof(*cctxParams));
+    cctxParams->cParams = params->cParams;
+    cctxParams->fParams = params->fParams;
+    /* Should not matter, as all cParams are presumed properly defined.
+     * But, set it for tracing anyway.
+     */
+    cctxParams->compressionLevel = compressionLevel;
+}
+
+size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params)
+{
+    RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!");
+    FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , "");
+    ZSTD_CCtxParams_init_internal(cctxParams, &params, ZSTD_NO_CLEVEL);
+    return 0;
+}
+
+/*
+ * Sets cctxParams' cParams and fParams from params, but otherwise leaves them alone.
+ * @param param Validated zstd parameters.
+ */
+static void ZSTD_CCtxParams_setZstdParams(
+        ZSTD_CCtx_params* cctxParams, const ZSTD_parameters* params)
+{
+    assert(!ZSTD_checkCParams(params->cParams));
+    cctxParams->cParams = params->cParams;
+    cctxParams->fParams = params->fParams;
+    /* Should not matter, as all cParams are presumed properly defined.
+     * But, set it for tracing anyway.
+     */
+    cctxParams->compressionLevel = ZSTD_NO_CLEVEL;
+}
+
+ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter param)
+{
+    ZSTD_bounds bounds = { 0, 0, 0 };
+
+    switch(param)
+    {
+    case ZSTD_c_compressionLevel:
+        bounds.lowerBound = ZSTD_minCLevel();
+        bounds.upperBound = ZSTD_maxCLevel();
+        return bounds;
+
+    case ZSTD_c_windowLog:
+        bounds.lowerBound = ZSTD_WINDOWLOG_MIN;
+        bounds.upperBound = ZSTD_WINDOWLOG_MAX;
+        return bounds;
+
+    case ZSTD_c_hashLog:
+        bounds.lowerBound = ZSTD_HASHLOG_MIN;
+        bounds.upperBound = ZSTD_HASHLOG_MAX;
+        return bounds;
+
+    case ZSTD_c_chainLog:
+        bounds.lowerBound = ZSTD_CHAINLOG_MIN;
+        bounds.upperBound = ZSTD_CHAINLOG_MAX;
+        return bounds;
+
+    case ZSTD_c_searchLog:
+        bounds.lowerBound = ZSTD_SEARCHLOG_MIN;
+        bounds.upperBound = ZSTD_SEARCHLOG_MAX;
+        return bounds;
+
+    case ZSTD_c_minMatch:
+        bounds.lowerBound = ZSTD_MINMATCH_MIN;
+        bounds.upperBound = ZSTD_MINMATCH_MAX;
+        return bounds;
+
+    case ZSTD_c_targetLength:
+        bounds.lowerBound = ZSTD_TARGETLENGTH_MIN;
+        bounds.upperBound = ZSTD_TARGETLENGTH_MAX;
+        return bounds;
+
+    case ZSTD_c_strategy:
+        bounds.lowerBound = ZSTD_STRATEGY_MIN;
+        bounds.upperBound = ZSTD_STRATEGY_MAX;
+        return bounds;
+
+    case ZSTD_c_contentSizeFlag:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 1;
+        return bounds;
+
+    case ZSTD_c_checksumFlag:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 1;
+        return bounds;
+
+    case ZSTD_c_dictIDFlag:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 1;
+        return bounds;
+
+    case ZSTD_c_nbWorkers:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 0;
+        return bounds;
+
+    case ZSTD_c_jobSize:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 0;
+        return bounds;
+
+    case ZSTD_c_overlapLog:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 0;
+        return bounds;
+
+    case ZSTD_c_enableDedicatedDictSearch:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 1;
+        return bounds;
+
+    case ZSTD_c_enableLongDistanceMatching:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 1;
+        return bounds;
+
+    case ZSTD_c_ldmHashLog:
+        bounds.lowerBound = ZSTD_LDM_HASHLOG_MIN;
+        bounds.upperBound = ZSTD_LDM_HASHLOG_MAX;
+        return bounds;
+
+    case ZSTD_c_ldmMinMatch:
+        bounds.lowerBound = ZSTD_LDM_MINMATCH_MIN;
+        bounds.upperBound = ZSTD_LDM_MINMATCH_MAX;
+        return bounds;
+
+    case ZSTD_c_ldmBucketSizeLog:
+        bounds.lowerBound = ZSTD_LDM_BUCKETSIZELOG_MIN;
+        bounds.upperBound = ZSTD_LDM_BUCKETSIZELOG_MAX;
+        return bounds;
+
+    case ZSTD_c_ldmHashRateLog:
+        bounds.lowerBound = ZSTD_LDM_HASHRATELOG_MIN;
+        bounds.upperBound = ZSTD_LDM_HASHRATELOG_MAX;
+        return bounds;
+
+    /* experimental parameters */
+    case ZSTD_c_rsyncable:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 1;
+        return bounds;
+
+    case ZSTD_c_forceMaxWindow :
+        bounds.lowerBound = 0;
+        bounds.upperBound = 1;
+        return bounds;
+
+    case ZSTD_c_format:
+        ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless);
+        bounds.lowerBound = ZSTD_f_zstd1;
+        bounds.upperBound = ZSTD_f_zstd1_magicless;   /* note : how to ensure at compile time that this is the highest value enum ? */
+        return bounds;
+
+    case ZSTD_c_forceAttachDict:
+        ZSTD_STATIC_ASSERT(ZSTD_dictDefaultAttach < ZSTD_dictForceLoad);
+        bounds.lowerBound = ZSTD_dictDefaultAttach;
+        bounds.upperBound = ZSTD_dictForceLoad;       /* note : how to ensure at compile time that this is the highest value enum ? */
+        return bounds;
+
+    case ZSTD_c_literalCompressionMode:
+        ZSTD_STATIC_ASSERT(ZSTD_lcm_auto < ZSTD_lcm_huffman && ZSTD_lcm_huffman < ZSTD_lcm_uncompressed);
+        bounds.lowerBound = ZSTD_lcm_auto;
+        bounds.upperBound = ZSTD_lcm_uncompressed;
+        return bounds;
+
+    case ZSTD_c_targetCBlockSize:
+        bounds.lowerBound = ZSTD_TARGETCBLOCKSIZE_MIN;
+        bounds.upperBound = ZSTD_TARGETCBLOCKSIZE_MAX;
+        return bounds;
+
+    case ZSTD_c_srcSizeHint:
+        bounds.lowerBound = ZSTD_SRCSIZEHINT_MIN;
+        bounds.upperBound = ZSTD_SRCSIZEHINT_MAX;
+        return bounds;
+
+    case ZSTD_c_stableInBuffer:
+    case ZSTD_c_stableOutBuffer:
+        bounds.lowerBound = (int)ZSTD_bm_buffered;
+        bounds.upperBound = (int)ZSTD_bm_stable;
+        return bounds;
+
+    case ZSTD_c_blockDelimiters:
+        bounds.lowerBound = (int)ZSTD_sf_noBlockDelimiters;
+        bounds.upperBound = (int)ZSTD_sf_explicitBlockDelimiters;
+        return bounds;
+
+    case ZSTD_c_validateSequences:
+        bounds.lowerBound = 0;
+        bounds.upperBound = 1;
+        return bounds;
+
+    default:
+        bounds.error = ERROR(parameter_unsupported);
+        return bounds;
+    }
+}
+
+/* ZSTD_cParam_clampBounds:
+ * Clamps the value into the bounded range.
+ */
+static size_t ZSTD_cParam_clampBounds(ZSTD_cParameter cParam, int* value)
+{
+    ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);
+    if (ZSTD_isError(bounds.error)) return bounds.error;
+    if (*value < bounds.lowerBound) *value = bounds.lowerBound;
+    if (*value > bounds.upperBound) *value = bounds.upperBound;
+    return 0;
+}
+
+#define BOUNDCHECK(cParam, val) { \
+    RETURN_ERROR_IF(!ZSTD_cParam_withinBounds(cParam,val), \
+                    parameter_outOfBound, "Param out of bounds"); \
+}
+
+
+static int ZSTD_isUpdateAuthorized(ZSTD_cParameter param)
+{
+    switch(param)
+    {
+    case ZSTD_c_compressionLevel:
+    case ZSTD_c_hashLog:
+    case ZSTD_c_chainLog:
+    case ZSTD_c_searchLog:
+    case ZSTD_c_minMatch:
+    case ZSTD_c_targetLength:
+    case ZSTD_c_strategy:
+        return 1;
+
+    case ZSTD_c_format:
+    case ZSTD_c_windowLog:
+    case ZSTD_c_contentSizeFlag:
+    case ZSTD_c_checksumFlag:
+    case ZSTD_c_dictIDFlag:
+    case ZSTD_c_forceMaxWindow :
+    case ZSTD_c_nbWorkers:
+    case ZSTD_c_jobSize:
+    case ZSTD_c_overlapLog:
+    case ZSTD_c_rsyncable:
+    case ZSTD_c_enableDedicatedDictSearch:
+    case ZSTD_c_enableLongDistanceMatching:
+    case ZSTD_c_ldmHashLog:
+    case ZSTD_c_ldmMinMatch:
+    case ZSTD_c_ldmBucketSizeLog:
+    case ZSTD_c_ldmHashRateLog:
+    case ZSTD_c_forceAttachDict:
+    case ZSTD_c_literalCompressionMode:
+    case ZSTD_c_targetCBlockSize:
+    case ZSTD_c_srcSizeHint:
+    case ZSTD_c_stableInBuffer:
+    case ZSTD_c_stableOutBuffer:
+    case ZSTD_c_blockDelimiters:
+    case ZSTD_c_validateSequences:
+    default:
+        return 0;
+    }
+}
+
+size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value)
+{
+    DEBUGLOG(4, "ZSTD_CCtx_setParameter (%i, %i)", (int)param, value);
+    if (cctx->streamStage != zcss_init) {
+        if (ZSTD_isUpdateAuthorized(param)) {
+            cctx->cParamsChanged = 1;
+        } else {
+            RETURN_ERROR(stage_wrong, "can only set params in ctx init stage");
+    }   }
+
+    switch(param)
+    {
+    case ZSTD_c_nbWorkers:
+        RETURN_ERROR_IF((value!=0) && cctx->staticSize, parameter_unsupported,
+                        "MT not compatible with static alloc");
+        break;
+
+    case ZSTD_c_compressionLevel:
+    case ZSTD_c_windowLog:
+    case ZSTD_c_hashLog:
+    case ZSTD_c_chainLog:
+    case ZSTD_c_searchLog:
+    case ZSTD_c_minMatch:
+    case ZSTD_c_targetLength:
+    case ZSTD_c_strategy:
+    case ZSTD_c_ldmHashRateLog:
+    case ZSTD_c_format:
+    case ZSTD_c_contentSizeFlag:
+    case ZSTD_c_checksumFlag:
+    case ZSTD_c_dictIDFlag:
+    case ZSTD_c_forceMaxWindow:
+    case ZSTD_c_forceAttachDict:
+    case ZSTD_c_literalCompressionMode:
+    case ZSTD_c_jobSize:
+    case ZSTD_c_overlapLog:
+    case ZSTD_c_rsyncable:
+    case ZSTD_c_enableDedicatedDictSearch:
+    case ZSTD_c_enableLongDistanceMatching:
+    case ZSTD_c_ldmHashLog:
+    case ZSTD_c_ldmMinMatch:
+    case ZSTD_c_ldmBucketSizeLog:
+    case ZSTD_c_targetCBlockSize:
+    case ZSTD_c_srcSizeHint:
+    case ZSTD_c_stableInBuffer:
+    case ZSTD_c_stableOutBuffer:
+    case ZSTD_c_blockDelimiters:
+    case ZSTD_c_validateSequences:
+        break;
+
+    default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
+    }
+    return ZSTD_CCtxParams_setParameter(&cctx->requestedParams, param, value);
+}
+
+size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* CCtxParams,
+                                    ZSTD_cParameter param, int value)
+{
+    DEBUGLOG(4, "ZSTD_CCtxParams_setParameter (%i, %i)", (int)param, value);
+    switch(param)
+    {
+    case ZSTD_c_format :
+        BOUNDCHECK(ZSTD_c_format, value);
+        CCtxParams->format = (ZSTD_format_e)value;
+        return (size_t)CCtxParams->format;
+
+    case ZSTD_c_compressionLevel : {
+        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
+        if (value == 0)
+            CCtxParams->compressionLevel = ZSTD_CLEVEL_DEFAULT; /* 0 == default */
+        else
+            CCtxParams->compressionLevel = value;
+        if (CCtxParams->compressionLevel >= 0) return (size_t)CCtxParams->compressionLevel;
+        return 0;  /* return type (size_t) cannot represent negative values */
+    }
+
+    case ZSTD_c_windowLog :
+        if (value!=0)   /* 0 => use default */
+            BOUNDCHECK(ZSTD_c_windowLog, value);
+        CCtxParams->cParams.windowLog = (U32)value;
+        return CCtxParams->cParams.windowLog;
+
+    case ZSTD_c_hashLog :
+        if (value!=0)   /* 0 => use default */
+            BOUNDCHECK(ZSTD_c_hashLog, value);
+        CCtxParams->cParams.hashLog = (U32)value;
+        return CCtxParams->cParams.hashLog;
+
+    case ZSTD_c_chainLog :
+        if (value!=0)   /* 0 => use default */
+            BOUNDCHECK(ZSTD_c_chainLog, value);
+        CCtxParams->cParams.chainLog = (U32)value;
+        return CCtxParams->cParams.chainLog;
+
+    case ZSTD_c_searchLog :
+        if (value!=0)   /* 0 => use default */
+            BOUNDCHECK(ZSTD_c_searchLog, value);
+        CCtxParams->cParams.searchLog = (U32)value;
+        return (size_t)value;
+
+    case ZSTD_c_minMatch :
+        if (value!=0)   /* 0 => use default */
+            BOUNDCHECK(ZSTD_c_minMatch, value);
+        CCtxParams->cParams.minMatch = value;
+        return CCtxParams->cParams.minMatch;
+
+    case ZSTD_c_targetLength :
+        BOUNDCHECK(ZSTD_c_targetLength, value);
+        CCtxParams->cParams.targetLength = value;
+        return CCtxParams->cParams.targetLength;
+
+    case ZSTD_c_strategy :
+        if (value!=0)   /* 0 => use default */
+            BOUNDCHECK(ZSTD_c_strategy, value);
+        CCtxParams->cParams.strategy = (ZSTD_strategy)value;
+        return (size_t)CCtxParams->cParams.strategy;
+
+    case ZSTD_c_contentSizeFlag :
+        /* Content size written in frame header _when known_ (default:1) */
+        DEBUGLOG(4, "set content size flag = %u", (value!=0));
+        CCtxParams->fParams.contentSizeFlag = value != 0;
+        return CCtxParams->fParams.contentSizeFlag;
+
+    case ZSTD_c_checksumFlag :
+        /* A 32-bits content checksum will be calculated and written at end of frame (default:0) */
+        CCtxParams->fParams.checksumFlag = value != 0;
+        return CCtxParams->fParams.checksumFlag;
+
+    case ZSTD_c_dictIDFlag : /* When applicable, dictionary's dictID is provided in frame header (default:1) */
+        DEBUGLOG(4, "set dictIDFlag = %u", (value!=0));
+        CCtxParams->fParams.noDictIDFlag = !value;
+        return !CCtxParams->fParams.noDictIDFlag;
+
+    case ZSTD_c_forceMaxWindow :
+        CCtxParams->forceWindow = (value != 0);
+        return CCtxParams->forceWindow;
+
+    case ZSTD_c_forceAttachDict : {
+        const ZSTD_dictAttachPref_e pref = (ZSTD_dictAttachPref_e)value;
+        BOUNDCHECK(ZSTD_c_forceAttachDict, pref);
+        CCtxParams->attachDictPref = pref;
+        return CCtxParams->attachDictPref;
+    }
+
+    case ZSTD_c_literalCompressionMode : {
+        const ZSTD_literalCompressionMode_e lcm = (ZSTD_literalCompressionMode_e)value;
+        BOUNDCHECK(ZSTD_c_literalCompressionMode, lcm);
+        CCtxParams->literalCompressionMode = lcm;
+        return CCtxParams->literalCompressionMode;
+    }
+
+    case ZSTD_c_nbWorkers :
+        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
+        return 0;
+
+    case ZSTD_c_jobSize :
+        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
+        return 0;
+
+    case ZSTD_c_overlapLog :
+        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
+        return 0;
+
+    case ZSTD_c_rsyncable :
+        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
+        return 0;
+
+    case ZSTD_c_enableDedicatedDictSearch :
+        CCtxParams->enableDedicatedDictSearch = (value!=0);
+        return CCtxParams->enableDedicatedDictSearch;
+
+    case ZSTD_c_enableLongDistanceMatching :
+        CCtxParams->ldmParams.enableLdm = (value!=0);
+        return CCtxParams->ldmParams.enableLdm;
+
+    case ZSTD_c_ldmHashLog :
+        if (value!=0)   /* 0 ==> auto */
+            BOUNDCHECK(ZSTD_c_ldmHashLog, value);
+        CCtxParams->ldmParams.hashLog = value;
+        return CCtxParams->ldmParams.hashLog;
+
+    case ZSTD_c_ldmMinMatch :
+        if (value!=0)   /* 0 ==> default */
+            BOUNDCHECK(ZSTD_c_ldmMinMatch, value);
+        CCtxParams->ldmParams.minMatchLength = value;
+        return CCtxParams->ldmParams.minMatchLength;
+
+    case ZSTD_c_ldmBucketSizeLog :
+        if (value!=0)   /* 0 ==> default */
+            BOUNDCHECK(ZSTD_c_ldmBucketSizeLog, value);
+        CCtxParams->ldmParams.bucketSizeLog = value;
+        return CCtxParams->ldmParams.bucketSizeLog;
+
+    case ZSTD_c_ldmHashRateLog :
+        if (value!=0)   /* 0 ==> default */
+            BOUNDCHECK(ZSTD_c_ldmHashRateLog, value);
+        CCtxParams->ldmParams.hashRateLog = value;
+        return CCtxParams->ldmParams.hashRateLog;
+
+    case ZSTD_c_targetCBlockSize :
+        if (value!=0)   /* 0 ==> default */
+            BOUNDCHECK(ZSTD_c_targetCBlockSize, value);
+        CCtxParams->targetCBlockSize = value;
+        return CCtxParams->targetCBlockSize;
+
+    case ZSTD_c_srcSizeHint :
+        if (value!=0)    /* 0 ==> default */
+            BOUNDCHECK(ZSTD_c_srcSizeHint, value);
+        CCtxParams->srcSizeHint = value;
+        return CCtxParams->srcSizeHint;
+
+    case ZSTD_c_stableInBuffer:
+        BOUNDCHECK(ZSTD_c_stableInBuffer, value);
+        CCtxParams->inBufferMode = (ZSTD_bufferMode_e)value;
+        return CCtxParams->inBufferMode;
+
+    case ZSTD_c_stableOutBuffer:
+        BOUNDCHECK(ZSTD_c_stableOutBuffer, value);
+        CCtxParams->outBufferMode = (ZSTD_bufferMode_e)value;
+        return CCtxParams->outBufferMode;
+
+    case ZSTD_c_blockDelimiters:
+        BOUNDCHECK(ZSTD_c_blockDelimiters, value);
+        CCtxParams->blockDelimiters = (ZSTD_sequenceFormat_e)value;
+        return CCtxParams->blockDelimiters;
+
+    case ZSTD_c_validateSequences:
+        BOUNDCHECK(ZSTD_c_validateSequences, value);
+        CCtxParams->validateSequences = value;
+        return CCtxParams->validateSequences;
+
+    default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
+    }
+}
+
+size_t ZSTD_CCtx_getParameter(ZSTD_CCtx const* cctx, ZSTD_cParameter param, int* value)
+{
+    return ZSTD_CCtxParams_getParameter(&cctx->requestedParams, param, value);
+}
+
+size_t ZSTD_CCtxParams_getParameter(
+        ZSTD_CCtx_params const* CCtxParams, ZSTD_cParameter param, int* value)
+{
+    switch(param)
+    {
+    case ZSTD_c_format :
+        *value = CCtxParams->format;
+        break;
+    case ZSTD_c_compressionLevel :
+        *value = CCtxParams->compressionLevel;
+        break;
+    case ZSTD_c_windowLog :
+        *value = (int)CCtxParams->cParams.windowLog;
+        break;
+    case ZSTD_c_hashLog :
+        *value = (int)CCtxParams->cParams.hashLog;
+        break;
+    case ZSTD_c_chainLog :
+        *value = (int)CCtxParams->cParams.chainLog;
+        break;
+    case ZSTD_c_searchLog :
+        *value = CCtxParams->cParams.searchLog;
+        break;
+    case ZSTD_c_minMatch :
+        *value = CCtxParams->cParams.minMatch;
+        break;
+    case ZSTD_c_targetLength :
+        *value = CCtxParams->cParams.targetLength;
+        break;
+    case ZSTD_c_strategy :
+        *value = (unsigned)CCtxParams->cParams.strategy;
+        break;
+    case ZSTD_c_contentSizeFlag :
+        *value = CCtxParams->fParams.contentSizeFlag;
+        break;
+    case ZSTD_c_checksumFlag :
+        *value = CCtxParams->fParams.checksumFlag;
+        break;
+    case ZSTD_c_dictIDFlag :
+        *value = !CCtxParams->fParams.noDictIDFlag;
+        break;
+    case ZSTD_c_forceMaxWindow :
+        *value = CCtxParams->forceWindow;
+        break;
+    case ZSTD_c_forceAttachDict :
+        *value = CCtxParams->attachDictPref;
+        break;
+    case ZSTD_c_literalCompressionMode :
+        *value = CCtxParams->literalCompressionMode;
+        break;
+    case ZSTD_c_nbWorkers :
+        assert(CCtxParams->nbWorkers == 0);
+        *value = CCtxParams->nbWorkers;
+        break;
+    case ZSTD_c_jobSize :
+        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
+    case ZSTD_c_overlapLog :
+        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
+    case ZSTD_c_rsyncable :
+        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
+    case ZSTD_c_enableDedicatedDictSearch :
+        *value = CCtxParams->enableDedicatedDictSearch;
+        break;
+    case ZSTD_c_enableLongDistanceMatching :
+        *value = CCtxParams->ldmParams.enableLdm;
+        break;
+    case ZSTD_c_ldmHashLog :
+        *value = CCtxParams->ldmParams.hashLog;
+        break;
+    case ZSTD_c_ldmMinMatch :
+        *value = CCtxParams->ldmParams.minMatchLength;
+        break;
+    case ZSTD_c_ldmBucketSizeLog :
+        *value = CCtxParams->ldmParams.bucketSizeLog;
+        break;
+    case ZSTD_c_ldmHashRateLog :
+        *value = CCtxParams->ldmParams.hashRateLog;
+        break;
+    case ZSTD_c_targetCBlockSize :
+        *value = (int)CCtxParams->targetCBlockSize;
+        break;
+    case ZSTD_c_srcSizeHint :
+        *value = (int)CCtxParams->srcSizeHint;
+        break;
+    case ZSTD_c_stableInBuffer :
+        *value = (int)CCtxParams->inBufferMode;
+        break;
+    case ZSTD_c_stableOutBuffer :
+        *value = (int)CCtxParams->outBufferMode;
+        break;
+    case ZSTD_c_blockDelimiters :
+        *value = (int)CCtxParams->blockDelimiters;
+        break;
+    case ZSTD_c_validateSequences :
+        *value = (int)CCtxParams->validateSequences;
+        break;
+    default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
+    }
+    return 0;
+}
+
+/* ZSTD_CCtx_setParametersUsingCCtxParams() :
+ *  just applies `params` into `cctx`
+ *  no action is performed, parameters are merely stored.
+ *  If ZSTDMT is enabled, parameters are pushed to cctx->mtctx.
+ *    This is possible even if a compression is ongoing.
+ *    In which case, new parameters will be applied on the fly, starting with next compression job.
+ */
+size_t ZSTD_CCtx_setParametersUsingCCtxParams(
+        ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params)
+{
+    DEBUGLOG(4, "ZSTD_CCtx_setParametersUsingCCtxParams");
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "The context is in the wrong stage!");
+    RETURN_ERROR_IF(cctx->cdict, stage_wrong,
+                    "Can't override parameters with cdict attached (some must "
+                    "be inherited from the cdict).");
+
+    cctx->requestedParams = *params;
+    return 0;
+}
+
+ZSTDLIB_API size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize)
+{
+    DEBUGLOG(4, "ZSTD_CCtx_setPledgedSrcSize to %u bytes", (U32)pledgedSrcSize);
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't set pledgedSrcSize when not in init stage.");
+    cctx->pledgedSrcSizePlusOne = pledgedSrcSize+1;
+    return 0;
+}
+
+static ZSTD_compressionParameters ZSTD_dedicatedDictSearch_getCParams(
+        int const compressionLevel,
+        size_t const dictSize);
+static int ZSTD_dedicatedDictSearch_isSupported(
+        const ZSTD_compressionParameters* cParams);
+static void ZSTD_dedicatedDictSearch_revertCParams(
+        ZSTD_compressionParameters* cParams);
+
+/*
+ * Initializes the local dict using the requested parameters.
+ * NOTE: This does not use the pledged src size, because it may be used for more
+ * than one compression.
+ */
+static size_t ZSTD_initLocalDict(ZSTD_CCtx* cctx)
+{
+    ZSTD_localDict* const dl = &cctx->localDict;
+    if (dl->dict == NULL) {
+        /* No local dictionary. */
+        assert(dl->dictBuffer == NULL);
+        assert(dl->cdict == NULL);
+        assert(dl->dictSize == 0);
+        return 0;
+    }
+    if (dl->cdict != NULL) {
+        assert(cctx->cdict == dl->cdict);
+        /* Local dictionary already initialized. */
+        return 0;
+    }
+    assert(dl->dictSize > 0);
+    assert(cctx->cdict == NULL);
+    assert(cctx->prefixDict.dict == NULL);
+
+    dl->cdict = ZSTD_createCDict_advanced2(
+            dl->dict,
+            dl->dictSize,
+            ZSTD_dlm_byRef,
+            dl->dictContentType,
+            &cctx->requestedParams,
+            cctx->customMem);
+    RETURN_ERROR_IF(!dl->cdict, memory_allocation, "ZSTD_createCDict_advanced failed");
+    cctx->cdict = dl->cdict;
+    return 0;
+}
+
+size_t ZSTD_CCtx_loadDictionary_advanced(
+        ZSTD_CCtx* cctx, const void* dict, size_t dictSize,
+        ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType)
+{
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't load a dictionary when ctx is not in init stage.");
+    DEBUGLOG(4, "ZSTD_CCtx_loadDictionary_advanced (size: %u)", (U32)dictSize);
+    ZSTD_clearAllDicts(cctx);  /* in case one already exists */
+    if (dict == NULL || dictSize == 0)  /* no dictionary mode */
+        return 0;
+    if (dictLoadMethod == ZSTD_dlm_byRef) {
+        cctx->localDict.dict = dict;
+    } else {
+        void* dictBuffer;
+        RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
+                        "no malloc for static CCtx");
+        dictBuffer = ZSTD_customMalloc(dictSize, cctx->customMem);
+        RETURN_ERROR_IF(!dictBuffer, memory_allocation, "NULL pointer!");
+        ZSTD_memcpy(dictBuffer, dict, dictSize);
+        cctx->localDict.dictBuffer = dictBuffer;
+        cctx->localDict.dict = dictBuffer;
+    }
+    cctx->localDict.dictSize = dictSize;
+    cctx->localDict.dictContentType = dictContentType;
+    return 0;
+}
+
+ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_byReference(
+      ZSTD_CCtx* cctx, const void* dict, size_t dictSize)
+{
+    return ZSTD_CCtx_loadDictionary_advanced(
+            cctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto);
+}
+
+ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize)
+{
+    return ZSTD_CCtx_loadDictionary_advanced(
+            cctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto);
+}
+
+
+size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict)
+{
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't ref a dict when ctx not in init stage.");
+    /* Free the existing local cdict (if any) to save memory. */
+    ZSTD_clearAllDicts(cctx);
+    cctx->cdict = cdict;
+    return 0;
+}
+
+size_t ZSTD_CCtx_refThreadPool(ZSTD_CCtx* cctx, ZSTD_threadPool* pool)
+{
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't ref a pool when ctx not in init stage.");
+    cctx->pool = pool;
+    return 0;
+}
+
+size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize)
+{
+    return ZSTD_CCtx_refPrefix_advanced(cctx, prefix, prefixSize, ZSTD_dct_rawContent);
+}
+
+size_t ZSTD_CCtx_refPrefix_advanced(
+        ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType)
+{
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't ref a prefix when ctx not in init stage.");
+    ZSTD_clearAllDicts(cctx);
+    if (prefix != NULL && prefixSize > 0) {
+        cctx->prefixDict.dict = prefix;
+        cctx->prefixDict.dictSize = prefixSize;
+        cctx->prefixDict.dictContentType = dictContentType;
+    }
+    return 0;
+}
+
+/*! ZSTD_CCtx_reset() :
+ *  Also dumps dictionary */
+size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset)
+{
+    if ( (reset == ZSTD_reset_session_only)
+      || (reset == ZSTD_reset_session_and_parameters) ) {
+        cctx->streamStage = zcss_init;
+        cctx->pledgedSrcSizePlusOne = 0;
+    }
+    if ( (reset == ZSTD_reset_parameters)
+      || (reset == ZSTD_reset_session_and_parameters) ) {
+        RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                        "Can't reset parameters only when not in init stage.");
+        ZSTD_clearAllDicts(cctx);
+        return ZSTD_CCtxParams_reset(&cctx->requestedParams);
+    }
+    return 0;
+}
+
+
+/* ZSTD_checkCParams() :
+    control CParam values remain within authorized range.
+    @return : 0, or an error code if one value is beyond authorized range */
+size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams)
+{
+    BOUNDCHECK(ZSTD_c_windowLog, (int)cParams.windowLog);
+    BOUNDCHECK(ZSTD_c_chainLog,  (int)cParams.chainLog);
+    BOUNDCHECK(ZSTD_c_hashLog,   (int)cParams.hashLog);
+    BOUNDCHECK(ZSTD_c_searchLog, (int)cParams.searchLog);
+    BOUNDCHECK(ZSTD_c_minMatch,  (int)cParams.minMatch);
+    BOUNDCHECK(ZSTD_c_targetLength,(int)cParams.targetLength);
+    BOUNDCHECK(ZSTD_c_strategy,  cParams.strategy);
+    return 0;
+}
+
+/* ZSTD_clampCParams() :
+ *  make CParam values within valid range.
+ *  @return : valid CParams */
+static ZSTD_compressionParameters
+ZSTD_clampCParams(ZSTD_compressionParameters cParams)
+{
+#   define CLAMP_TYPE(cParam, val, type) {                                \
+        ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);         \
+        if ((int)val<bounds.lowerBound) val=(type)bounds.lowerBound;      \
+        else if ((int)val>bounds.upperBound) val=(type)bounds.upperBound; \
+    }
+#   define CLAMP(cParam, val) CLAMP_TYPE(cParam, val, unsigned)
+    CLAMP(ZSTD_c_windowLog, cParams.windowLog);
+    CLAMP(ZSTD_c_chainLog,  cParams.chainLog);
+    CLAMP(ZSTD_c_hashLog,   cParams.hashLog);
+    CLAMP(ZSTD_c_searchLog, cParams.searchLog);
+    CLAMP(ZSTD_c_minMatch,  cParams.minMatch);
+    CLAMP(ZSTD_c_targetLength,cParams.targetLength);
+    CLAMP_TYPE(ZSTD_c_strategy,cParams.strategy, ZSTD_strategy);
+    return cParams;
+}
+
+/* ZSTD_cycleLog() :
+ *  condition for correct operation : hashLog > 1 */
+U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat)
+{
+    U32 const btScale = ((U32)strat >= (U32)ZSTD_btlazy2);
+    return hashLog - btScale;
+}
+
+/* ZSTD_dictAndWindowLog() :
+ * Returns an adjusted window log that is large enough to fit the source and the dictionary.
+ * The zstd format says that the entire dictionary is valid if one byte of the dictionary
+ * is within the window. So the hashLog and chainLog should be large enough to reference both
+ * the dictionary and the window. So we must use this adjusted dictAndWindowLog when downsizing
+ * the hashLog and windowLog.
+ * NOTE: srcSize must not be ZSTD_CONTENTSIZE_UNKNOWN.
+ */
+static U32 ZSTD_dictAndWindowLog(U32 windowLog, U64 srcSize, U64 dictSize)
+{
+    const U64 maxWindowSize = 1ULL << ZSTD_WINDOWLOG_MAX;
+    /* No dictionary ==> No change */
+    if (dictSize == 0) {
+        return windowLog;
+    }
+    assert(windowLog <= ZSTD_WINDOWLOG_MAX);
+    assert(srcSize != ZSTD_CONTENTSIZE_UNKNOWN); /* Handled in ZSTD_adjustCParams_internal() */
+    {
+        U64 const windowSize = 1ULL << windowLog;
+        U64 const dictAndWindowSize = dictSize + windowSize;
+        /* If the window size is already large enough to fit both the source and the dictionary
+         * then just use the window size. Otherwise adjust so that it fits the dictionary and
+         * the window.
+         */
+        if (windowSize >= dictSize + srcSize) {
+            return windowLog; /* Window size large enough already */
+        } else if (dictAndWindowSize >= maxWindowSize) {
+            return ZSTD_WINDOWLOG_MAX; /* Larger than max window log */
+        } else  {
+            return ZSTD_highbit32((U32)dictAndWindowSize - 1) + 1;
+        }
+    }
+}
+
+/* ZSTD_adjustCParams_internal() :
+ *  optimize `cPar` for a specified input (`srcSize` and `dictSize`).
+ *  mostly downsize to reduce memory consumption and initialization latency.
+ * `srcSize` can be ZSTD_CONTENTSIZE_UNKNOWN when not known.
+ * `mode` is the mode for parameter adjustment. See docs for `ZSTD_cParamMode_e`.
+ *  note : `srcSize==0` means 0!
+ *  condition : cPar is presumed validated (can be checked using ZSTD_checkCParams()). */
+static ZSTD_compressionParameters
+ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar,
+                            unsigned long long srcSize,
+                            size_t dictSize,
+                            ZSTD_cParamMode_e mode)
+{
+    const U64 minSrcSize = 513; /* (1<<9) + 1 */
+    const U64 maxWindowResize = 1ULL << (ZSTD_WINDOWLOG_MAX-1);
+    assert(ZSTD_checkCParams(cPar)==0);
+
+    switch (mode) {
+    case ZSTD_cpm_unknown:
+    case ZSTD_cpm_noAttachDict:
+        /* If we don't know the source size, don't make any
+         * assumptions about it. We will already have selected
+         * smaller parameters if a dictionary is in use.
+         */
+        break;
+    case ZSTD_cpm_createCDict:
+        /* Assume a small source size when creating a dictionary
+         * with an unkown source size.
+         */
+        if (dictSize && srcSize == ZSTD_CONTENTSIZE_UNKNOWN)
+            srcSize = minSrcSize;
+        break;
+    case ZSTD_cpm_attachDict:
+        /* Dictionary has its own dedicated parameters which have
+         * already been selected. We are selecting parameters
+         * for only the source.
+         */
+        dictSize = 0;
+        break;
+    default:
+        assert(0);
+        break;
+    }
+
+    /* resize windowLog if input is small enough, to use less memory */
+    if ( (srcSize < maxWindowResize)
+      && (dictSize < maxWindowResize) )  {
+        U32 const tSize = (U32)(srcSize + dictSize);
+        static U32 const hashSizeMin = 1 << ZSTD_HASHLOG_MIN;
+        U32 const srcLog = (tSize < hashSizeMin) ? ZSTD_HASHLOG_MIN :
+                            ZSTD_highbit32(tSize-1) + 1;
+        if (cPar.windowLog > srcLog) cPar.windowLog = srcLog;
+    }
+    if (srcSize != ZSTD_CONTENTSIZE_UNKNOWN) {
+        U32 const dictAndWindowLog = ZSTD_dictAndWindowLog(cPar.windowLog, (U64)srcSize, (U64)dictSize);
+        U32 const cycleLog = ZSTD_cycleLog(cPar.chainLog, cPar.strategy);
+        if (cPar.hashLog > dictAndWindowLog+1) cPar.hashLog = dictAndWindowLog+1;
+        if (cycleLog > dictAndWindowLog)
+            cPar.chainLog -= (cycleLog - dictAndWindowLog);
+    }
+
+    if (cPar.windowLog < ZSTD_WINDOWLOG_ABSOLUTEMIN)
+        cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN;  /* minimum wlog required for valid frame header */
+
+    return cPar;
+}
+
+ZSTD_compressionParameters
+ZSTD_adjustCParams(ZSTD_compressionParameters cPar,
+                   unsigned long long srcSize,
+                   size_t dictSize)
+{
+    cPar = ZSTD_clampCParams(cPar);   /* resulting cPar is necessarily valid (all parameters within range) */
+    if (srcSize == 0) srcSize = ZSTD_CONTENTSIZE_UNKNOWN;
+    return ZSTD_adjustCParams_internal(cPar, srcSize, dictSize, ZSTD_cpm_unknown);
+}
+
+static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize, ZSTD_cParamMode_e mode);
+static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize, ZSTD_cParamMode_e mode);
+
+static void ZSTD_overrideCParams(
+              ZSTD_compressionParameters* cParams,
+        const ZSTD_compressionParameters* overrides)
+{
+    if (overrides->windowLog)    cParams->windowLog    = overrides->windowLog;
+    if (overrides->hashLog)      cParams->hashLog      = overrides->hashLog;
+    if (overrides->chainLog)     cParams->chainLog     = overrides->chainLog;
+    if (overrides->searchLog)    cParams->searchLog    = overrides->searchLog;
+    if (overrides->minMatch)     cParams->minMatch     = overrides->minMatch;
+    if (overrides->targetLength) cParams->targetLength = overrides->targetLength;
+    if (overrides->strategy)     cParams->strategy     = overrides->strategy;
+}
+
+ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
+        const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize, ZSTD_cParamMode_e mode)
+{
+    ZSTD_compressionParameters cParams;
+    if (srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN && CCtxParams->srcSizeHint > 0) {
+      srcSizeHint = CCtxParams->srcSizeHint;
+    }
+    cParams = ZSTD_getCParams_internal(CCtxParams->compressionLevel, srcSizeHint, dictSize, mode);
+    if (CCtxParams->ldmParams.enableLdm) cParams.windowLog = ZSTD_LDM_DEFAULT_WINDOW_LOG;
+    ZSTD_overrideCParams(&cParams, &CCtxParams->cParams);
+    assert(!ZSTD_checkCParams(cParams));
+    /* srcSizeHint == 0 means 0 */
+    return ZSTD_adjustCParams_internal(cParams, srcSizeHint, dictSize, mode);
+}
+
+static size_t
+ZSTD_sizeof_matchState(const ZSTD_compressionParameters* const cParams,
+                       const U32 forCCtx)
+{
+    size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog);
+    size_t const hSize = ((size_t)1) << cParams->hashLog;
+    U32    const hashLog3 = (forCCtx && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
+    size_t const h3Size = hashLog3 ? ((size_t)1) << hashLog3 : 0;
+    /* We don't use ZSTD_cwksp_alloc_size() here because the tables aren't
+     * surrounded by redzones in ASAN. */
+    size_t const tableSpace = chainSize * sizeof(U32)
+                            + hSize * sizeof(U32)
+                            + h3Size * sizeof(U32);
+    size_t const optPotentialSpace =
+        ZSTD_cwksp_alloc_size((MaxML+1) * sizeof(U32))
+      + ZSTD_cwksp_alloc_size((MaxLL+1) * sizeof(U32))
+      + ZSTD_cwksp_alloc_size((MaxOff+1) * sizeof(U32))
+      + ZSTD_cwksp_alloc_size((1<<Litbits) * sizeof(U32))
+      + ZSTD_cwksp_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_match_t))
+      + ZSTD_cwksp_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_optimal_t));
+    size_t const optSpace = (forCCtx && (cParams->strategy >= ZSTD_btopt))
+                                ? optPotentialSpace
+                                : 0;
+    DEBUGLOG(4, "chainSize: %u - hSize: %u - h3Size: %u",
+                (U32)chainSize, (U32)hSize, (U32)h3Size);
+    return tableSpace + optSpace;
+}
+
+static size_t ZSTD_estimateCCtxSize_usingCCtxParams_internal(
+        const ZSTD_compressionParameters* cParams,
+        const ldmParams_t* ldmParams,
+        const int isStatic,
+        const size_t buffInSize,
+        const size_t buffOutSize,
+        const U64 pledgedSrcSize)
+{
+    size_t const windowSize = MAX(1, (size_t)MIN(((U64)1 << cParams->windowLog), pledgedSrcSize));
+    size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, windowSize);
+    U32    const divider = (cParams->minMatch==3) ? 3 : 4;
+    size_t const maxNbSeq = blockSize / divider;
+    size_t const tokenSpace = ZSTD_cwksp_alloc_size(WILDCOPY_OVERLENGTH + blockSize)
+                            + ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(seqDef))
+                            + 3 * ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(BYTE));
+    size_t const entropySpace = ZSTD_cwksp_alloc_size(ENTROPY_WORKSPACE_SIZE);
+    size_t const blockStateSpace = 2 * ZSTD_cwksp_alloc_size(sizeof(ZSTD_compressedBlockState_t));
+    size_t const matchStateSize = ZSTD_sizeof_matchState(cParams, /* forCCtx */ 1);
+
+    size_t const ldmSpace = ZSTD_ldm_getTableSize(*ldmParams);
+    size_t const maxNbLdmSeq = ZSTD_ldm_getMaxNbSeq(*ldmParams, blockSize);
+    size_t const ldmSeqSpace = ldmParams->enableLdm ?
+        ZSTD_cwksp_alloc_size(maxNbLdmSeq * sizeof(rawSeq)) : 0;
+
+
+    size_t const bufferSpace = ZSTD_cwksp_alloc_size(buffInSize)
+                             + ZSTD_cwksp_alloc_size(buffOutSize);
+
+    size_t const cctxSpace = isStatic ? ZSTD_cwksp_alloc_size(sizeof(ZSTD_CCtx)) : 0;
+
+    size_t const neededSpace =
+        cctxSpace +
+        entropySpace +
+        blockStateSpace +
+        ldmSpace +
+        ldmSeqSpace +
+        matchStateSize +
+        tokenSpace +
+        bufferSpace;
+
+    DEBUGLOG(5, "estimate workspace : %u", (U32)neededSpace);
+    return neededSpace;
+}
+
+size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params)
+{
+    ZSTD_compressionParameters const cParams =
+                ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0, ZSTD_cpm_noAttachDict);
+
+    RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
+    /* estimateCCtxSize is for one-shot compression. So no buffers should
+     * be needed. However, we still allocate two 0-sized buffers, which can
+     * take space under ASAN. */
+    return ZSTD_estimateCCtxSize_usingCCtxParams_internal(
+        &cParams, &params->ldmParams, 1, 0, 0, ZSTD_CONTENTSIZE_UNKNOWN);
+}
+
+size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams)
+{
+    ZSTD_CCtx_params const params = ZSTD_makeCCtxParamsFromCParams(cParams);
+    return ZSTD_estimateCCtxSize_usingCCtxParams(&params);
+}
+
+static size_t ZSTD_estimateCCtxSize_internal(int compressionLevel)
+{
+    int tier = 0;
+    size_t largestSize = 0;
+    static const unsigned long long srcSizeTiers[4] = {16 KB, 128 KB, 256 KB, ZSTD_CONTENTSIZE_UNKNOWN};
+    for (; tier < 4; ++tier) {
+        /* Choose the set of cParams for a given level across all srcSizes that give the largest cctxSize */
+        ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, srcSizeTiers[tier], 0, ZSTD_cpm_noAttachDict);
+        largestSize = MAX(ZSTD_estimateCCtxSize_usingCParams(cParams), largestSize);
+    }
+    return largestSize;
+}
+
+size_t ZSTD_estimateCCtxSize(int compressionLevel)
+{
+    int level;
+    size_t memBudget = 0;
+    for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) {
+        /* Ensure monotonically increasing memory usage as compression level increases */
+        size_t const newMB = ZSTD_estimateCCtxSize_internal(level);
+        if (newMB > memBudget) memBudget = newMB;
+    }
+    return memBudget;
+}
+
+size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params)
+{
+    RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
+    {   ZSTD_compressionParameters const cParams =
+                ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0, ZSTD_cpm_noAttachDict);
+        size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog);
+        size_t const inBuffSize = (params->inBufferMode == ZSTD_bm_buffered)
+                ? ((size_t)1 << cParams.windowLog) + blockSize
+                : 0;
+        size_t const outBuffSize = (params->outBufferMode == ZSTD_bm_buffered)
+                ? ZSTD_compressBound(blockSize) + 1
+                : 0;
+
+        return ZSTD_estimateCCtxSize_usingCCtxParams_internal(
+            &cParams, &params->ldmParams, 1, inBuffSize, outBuffSize,
+            ZSTD_CONTENTSIZE_UNKNOWN);
+    }
+}
+
+size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams)
+{
+    ZSTD_CCtx_params const params = ZSTD_makeCCtxParamsFromCParams(cParams);
+    return ZSTD_estimateCStreamSize_usingCCtxParams(&params);
+}
+
+static size_t ZSTD_estimateCStreamSize_internal(int compressionLevel)
+{
+    ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, 0, ZSTD_cpm_noAttachDict);
+    return ZSTD_estimateCStreamSize_usingCParams(cParams);
+}
+
+size_t ZSTD_estimateCStreamSize(int compressionLevel)
+{
+    int level;
+    size_t memBudget = 0;
+    for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) {
+        size_t const newMB = ZSTD_estimateCStreamSize_internal(level);
+        if (newMB > memBudget) memBudget = newMB;
+    }
+    return memBudget;
+}
+
+/* ZSTD_getFrameProgression():
+ * tells how much data has been consumed (input) and produced (output) for current frame.
+ * able to count progression inside worker threads (non-blocking mode).
+ */
+ZSTD_frameProgression ZSTD_getFrameProgression(const ZSTD_CCtx* cctx)
+{
+    {   ZSTD_frameProgression fp;
+        size_t const buffered = (cctx->inBuff == NULL) ? 0 :
+                                cctx->inBuffPos - cctx->inToCompress;
+        if (buffered) assert(cctx->inBuffPos >= cctx->inToCompress);
+        assert(buffered <= ZSTD_BLOCKSIZE_MAX);
+        fp.ingested = cctx->consumedSrcSize + buffered;
+        fp.consumed = cctx->consumedSrcSize;
+        fp.produced = cctx->producedCSize;
+        fp.flushed  = cctx->producedCSize;   /* simplified; some data might still be left within streaming output buffer */
+        fp.currentJobID = 0;
+        fp.nbActiveWorkers = 0;
+        return fp;
+}   }
+
+/*! ZSTD_toFlushNow()
+ *  Only useful for multithreading scenarios currently (nbWorkers >= 1).
+ */
+size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx)
+{
+    (void)cctx;
+    return 0;   /* over-simplification; could also check if context is currently running in streaming mode, and in which case, report how many bytes are left to be flushed within output buffer */
+}
+
+static void ZSTD_assertEqualCParams(ZSTD_compressionParameters cParams1,
+                                    ZSTD_compressionParameters cParams2)
+{
+    (void)cParams1;
+    (void)cParams2;
+    assert(cParams1.windowLog    == cParams2.windowLog);
+    assert(cParams1.chainLog     == cParams2.chainLog);
+    assert(cParams1.hashLog      == cParams2.hashLog);
+    assert(cParams1.searchLog    == cParams2.searchLog);
+    assert(cParams1.minMatch     == cParams2.minMatch);
+    assert(cParams1.targetLength == cParams2.targetLength);
+    assert(cParams1.strategy     == cParams2.strategy);
+}
+
+void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs)
+{
+    int i;
+    for (i = 0; i < ZSTD_REP_NUM; ++i)
+        bs->rep[i] = repStartValue[i];
+    bs->entropy.huf.repeatMode = HUF_repeat_none;
+    bs->entropy.fse.offcode_repeatMode = FSE_repeat_none;
+    bs->entropy.fse.matchlength_repeatMode = FSE_repeat_none;
+    bs->entropy.fse.litlength_repeatMode = FSE_repeat_none;
+}
+
+/*! ZSTD_invalidateMatchState()
+ *  Invalidate all the matches in the match finder tables.
+ *  Requires nextSrc and base to be set (can be NULL).
+ */
+static void ZSTD_invalidateMatchState(ZSTD_matchState_t* ms)
+{
+    ZSTD_window_clear(&ms->window);
+
+    ms->nextToUpdate = ms->window.dictLimit;
+    ms->loadedDictEnd = 0;
+    ms->opt.litLengthSum = 0;  /* force reset of btopt stats */
+    ms->dictMatchState = NULL;
+}
+
+/*
+ * Controls, for this matchState reset, whether the tables need to be cleared /
+ * prepared for the coming compression (ZSTDcrp_makeClean), or whether the
+ * tables can be left unclean (ZSTDcrp_leaveDirty), because we know that a
+ * subsequent operation will overwrite the table space anyways (e.g., copying
+ * the matchState contents in from a CDict).
+ */
+typedef enum {
+    ZSTDcrp_makeClean,
+    ZSTDcrp_leaveDirty
+} ZSTD_compResetPolicy_e;
+
+/*
+ * Controls, for this matchState reset, whether indexing can continue where it
+ * left off (ZSTDirp_continue), or whether it needs to be restarted from zero
+ * (ZSTDirp_reset).
+ */
+typedef enum {
+    ZSTDirp_continue,
+    ZSTDirp_reset
+} ZSTD_indexResetPolicy_e;
+
+typedef enum {
+    ZSTD_resetTarget_CDict,
+    ZSTD_resetTarget_CCtx
+} ZSTD_resetTarget_e;
+
+static size_t
+ZSTD_reset_matchState(ZSTD_matchState_t* ms,
+                      ZSTD_cwksp* ws,
+                const ZSTD_compressionParameters* cParams,
+                const ZSTD_compResetPolicy_e crp,
+                const ZSTD_indexResetPolicy_e forceResetIndex,
+                const ZSTD_resetTarget_e forWho)
+{
+    size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog);
+    size_t const hSize = ((size_t)1) << cParams->hashLog;
+    U32    const hashLog3 = ((forWho == ZSTD_resetTarget_CCtx) && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
+    size_t const h3Size = hashLog3 ? ((size_t)1) << hashLog3 : 0;
+
+    DEBUGLOG(4, "reset indices : %u", forceResetIndex == ZSTDirp_reset);
+    if (forceResetIndex == ZSTDirp_reset) {
+        ZSTD_window_init(&ms->window);
+        ZSTD_cwksp_mark_tables_dirty(ws);
+    }
+
+    ms->hashLog3 = hashLog3;
+
+    ZSTD_invalidateMatchState(ms);
+
+    assert(!ZSTD_cwksp_reserve_failed(ws)); /* check that allocation hasn't already failed */
+
+    ZSTD_cwksp_clear_tables(ws);
+
+    DEBUGLOG(5, "reserving table space");
+    /* table Space */
+    ms->hashTable = (U32*)ZSTD_cwksp_reserve_table(ws, hSize * sizeof(U32));
+    ms->chainTable = (U32*)ZSTD_cwksp_reserve_table(ws, chainSize * sizeof(U32));
+    ms->hashTable3 = (U32*)ZSTD_cwksp_reserve_table(ws, h3Size * sizeof(U32));
+    RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation,
+                    "failed a workspace allocation in ZSTD_reset_matchState");
+
+    DEBUGLOG(4, "reset table : %u", crp!=ZSTDcrp_leaveDirty);
+    if (crp!=ZSTDcrp_leaveDirty) {
+        /* reset tables only */
+        ZSTD_cwksp_clean_tables(ws);
+    }
+
+    /* opt parser space */
+    if ((forWho == ZSTD_resetTarget_CCtx) && (cParams->strategy >= ZSTD_btopt)) {
+        DEBUGLOG(4, "reserving optimal parser space");
+        ms->opt.litFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (1<<Litbits) * sizeof(unsigned));
+        ms->opt.litLengthFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxLL+1) * sizeof(unsigned));
+        ms->opt.matchLengthFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxML+1) * sizeof(unsigned));
+        ms->opt.offCodeFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxOff+1) * sizeof(unsigned));
+        ms->opt.matchTable = (ZSTD_match_t*)ZSTD_cwksp_reserve_aligned(ws, (ZSTD_OPT_NUM+1) * sizeof(ZSTD_match_t));
+        ms->opt.priceTable = (ZSTD_optimal_t*)ZSTD_cwksp_reserve_aligned(ws, (ZSTD_OPT_NUM+1) * sizeof(ZSTD_optimal_t));
+    }
+
+    ms->cParams = *cParams;
+
+    RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation,
+                    "failed a workspace allocation in ZSTD_reset_matchState");
+
+    return 0;
+}
+
+/* ZSTD_indexTooCloseToMax() :
+ * minor optimization : prefer memset() rather than reduceIndex()
+ * which is measurably slow in some circumstances (reported for Visual Studio).
+ * Works when re-using a context for a lot of smallish inputs :
+ * if all inputs are smaller than ZSTD_INDEXOVERFLOW_MARGIN,
+ * memset() will be triggered before reduceIndex().
+ */
+#define ZSTD_INDEXOVERFLOW_MARGIN (16 MB)
+static int ZSTD_indexTooCloseToMax(ZSTD_window_t w)
+{
+    return (size_t)(w.nextSrc - w.base) > (ZSTD_CURRENT_MAX - ZSTD_INDEXOVERFLOW_MARGIN);
+}
+
+/*! ZSTD_resetCCtx_internal() :
+    note : `params` are assumed fully validated at this stage */
+static size_t ZSTD_resetCCtx_internal(ZSTD_CCtx* zc,
+                                      ZSTD_CCtx_params params,
+                                      U64 const pledgedSrcSize,
+                                      ZSTD_compResetPolicy_e const crp,
+                                      ZSTD_buffered_policy_e const zbuff)
+{
+    ZSTD_cwksp* const ws = &zc->workspace;
+    DEBUGLOG(4, "ZSTD_resetCCtx_internal: pledgedSrcSize=%u, wlog=%u",
+                (U32)pledgedSrcSize, params.cParams.windowLog);
+    assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));
+
+    zc->isFirstBlock = 1;
+
+    if (params.ldmParams.enableLdm) {
+        /* Adjust long distance matching parameters */
+        ZSTD_ldm_adjustParameters(&params.ldmParams, &params.cParams);
+        assert(params.ldmParams.hashLog >= params.ldmParams.bucketSizeLog);
+        assert(params.ldmParams.hashRateLog < 32);
+    }
+
+    {   size_t const windowSize = MAX(1, (size_t)MIN(((U64)1 << params.cParams.windowLog), pledgedSrcSize));
+        size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, windowSize);
+        U32    const divider = (params.cParams.minMatch==3) ? 3 : 4;
+        size_t const maxNbSeq = blockSize / divider;
+        size_t const buffOutSize = (zbuff == ZSTDb_buffered && params.outBufferMode == ZSTD_bm_buffered)
+                ? ZSTD_compressBound(blockSize) + 1
+                : 0;
+        size_t const buffInSize = (zbuff == ZSTDb_buffered && params.inBufferMode == ZSTD_bm_buffered)
+                ? windowSize + blockSize
+                : 0;
+        size_t const maxNbLdmSeq = ZSTD_ldm_getMaxNbSeq(params.ldmParams, blockSize);
+
+        int const indexTooClose = ZSTD_indexTooCloseToMax(zc->blockState.matchState.window);
+        ZSTD_indexResetPolicy_e needsIndexReset =
+            (!indexTooClose && zc->initialized) ? ZSTDirp_continue : ZSTDirp_reset;
+
+        size_t const neededSpace =
+            ZSTD_estimateCCtxSize_usingCCtxParams_internal(
+                &params.cParams, &params.ldmParams, zc->staticSize != 0,
+                buffInSize, buffOutSize, pledgedSrcSize);
+        FORWARD_IF_ERROR(neededSpace, "cctx size estimate failed!");
+
+        if (!zc->staticSize) ZSTD_cwksp_bump_oversized_duration(ws, 0);
+
+        /* Check if workspace is large enough, alloc a new one if needed */
+        {
+            int const workspaceTooSmall = ZSTD_cwksp_sizeof(ws) < neededSpace;
+            int const workspaceWasteful = ZSTD_cwksp_check_wasteful(ws, neededSpace);
+
+            DEBUGLOG(4, "Need %zu B workspace", neededSpace);
+            DEBUGLOG(4, "windowSize: %zu - blockSize: %zu", windowSize, blockSize);
+
+            if (workspaceTooSmall || workspaceWasteful) {
+                DEBUGLOG(4, "Resize workspaceSize from %zuKB to %zuKB",
+                            ZSTD_cwksp_sizeof(ws) >> 10,
+                            neededSpace >> 10);
+
+                RETURN_ERROR_IF(zc->staticSize, memory_allocation, "static cctx : no resize");
+
+                needsIndexReset = ZSTDirp_reset;
+
+                ZSTD_cwksp_free(ws, zc->customMem);
+                FORWARD_IF_ERROR(ZSTD_cwksp_create(ws, neededSpace, zc->customMem), "");
+
+                DEBUGLOG(5, "reserving object space");
+                /* Statically sized space.
+                 * entropyWorkspace never moves,
+                 * though prev/next block swap places */
+                assert(ZSTD_cwksp_check_available(ws, 2 * sizeof(ZSTD_compressedBlockState_t)));
+                zc->blockState.prevCBlock = (ZSTD_compressedBlockState_t*) ZSTD_cwksp_reserve_object(ws, sizeof(ZSTD_compressedBlockState_t));
+                RETURN_ERROR_IF(zc->blockState.prevCBlock == NULL, memory_allocation, "couldn't allocate prevCBlock");
+                zc->blockState.nextCBlock = (ZSTD_compressedBlockState_t*) ZSTD_cwksp_reserve_object(ws, sizeof(ZSTD_compressedBlockState_t));
+                RETURN_ERROR_IF(zc->blockState.nextCBlock == NULL, memory_allocation, "couldn't allocate nextCBlock");
+                zc->entropyWorkspace = (U32*) ZSTD_cwksp_reserve_object(ws, ENTROPY_WORKSPACE_SIZE);
+                RETURN_ERROR_IF(zc->blockState.nextCBlock == NULL, memory_allocation, "couldn't allocate entropyWorkspace");
+        }   }
+
+        ZSTD_cwksp_clear(ws);
+
+        /* init params */
+        zc->appliedParams = params;
+        zc->blockState.matchState.cParams = params.cParams;
+        zc->pledgedSrcSizePlusOne = pledgedSrcSize+1;
+        zc->consumedSrcSize = 0;
+        zc->producedCSize = 0;
+        if (pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN)
+            zc->appliedParams.fParams.contentSizeFlag = 0;
+        DEBUGLOG(4, "pledged content size : %u ; flag : %u",
+            (unsigned)pledgedSrcSize, zc->appliedParams.fParams.contentSizeFlag);
+        zc->blockSize = blockSize;
+
+        xxh64_reset(&zc->xxhState, 0);
+        zc->stage = ZSTDcs_init;
+        zc->dictID = 0;
+        zc->dictContentSize = 0;
+
+        ZSTD_reset_compressedBlockState(zc->blockState.prevCBlock);
+
+        /* ZSTD_wildcopy() is used to copy into the literals buffer,
+         * so we have to oversize the buffer by WILDCOPY_OVERLENGTH bytes.
+         */
+        zc->seqStore.litStart = ZSTD_cwksp_reserve_buffer(ws, blockSize + WILDCOPY_OVERLENGTH);
+        zc->seqStore.maxNbLit = blockSize;
+
+        /* buffers */
+        zc->bufferedPolicy = zbuff;
+        zc->inBuffSize = buffInSize;
+        zc->inBuff = (char*)ZSTD_cwksp_reserve_buffer(ws, buffInSize);
+        zc->outBuffSize = buffOutSize;
+        zc->outBuff = (char*)ZSTD_cwksp_reserve_buffer(ws, buffOutSize);
+
+        /* ldm bucketOffsets table */
+        if (params.ldmParams.enableLdm) {
+            /* TODO: avoid memset? */
+            size_t const numBuckets =
+                  ((size_t)1) << (params.ldmParams.hashLog -
+                                  params.ldmParams.bucketSizeLog);
+            zc->ldmState.bucketOffsets = ZSTD_cwksp_reserve_buffer(ws, numBuckets);
+            ZSTD_memset(zc->ldmState.bucketOffsets, 0, numBuckets);
+        }
+
+        /* sequences storage */
+        ZSTD_referenceExternalSequences(zc, NULL, 0);
+        zc->seqStore.maxNbSeq = maxNbSeq;
+        zc->seqStore.llCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
+        zc->seqStore.mlCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
+        zc->seqStore.ofCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
+        zc->seqStore.sequencesStart = (seqDef*)ZSTD_cwksp_reserve_aligned(ws, maxNbSeq * sizeof(seqDef));
+
+        FORWARD_IF_ERROR(ZSTD_reset_matchState(
+            &zc->blockState.matchState,
+            ws,
+            &params.cParams,
+            crp,
+            needsIndexReset,
+            ZSTD_resetTarget_CCtx), "");
+
+        /* ldm hash table */
+        if (params.ldmParams.enableLdm) {
+            /* TODO: avoid memset? */
+            size_t const ldmHSize = ((size_t)1) << params.ldmParams.hashLog;
+            zc->ldmState.hashTable = (ldmEntry_t*)ZSTD_cwksp_reserve_aligned(ws, ldmHSize * sizeof(ldmEntry_t));
+            ZSTD_memset(zc->ldmState.hashTable, 0, ldmHSize * sizeof(ldmEntry_t));
+            zc->ldmSequences = (rawSeq*)ZSTD_cwksp_reserve_aligned(ws, maxNbLdmSeq * sizeof(rawSeq));
+            zc->maxNbLdmSequences = maxNbLdmSeq;
+
+            ZSTD_window_init(&zc->ldmState.window);
+            ZSTD_window_clear(&zc->ldmState.window);
+            zc->ldmState.loadedDictEnd = 0;
+        }
+
+        /* Due to alignment, when reusing a workspace, we can actually consume
+         * up to 3 extra bytes for alignment. See the comments in zstd_cwksp.h
+         */
+        assert(ZSTD_cwksp_used(ws) >= neededSpace &&
+               ZSTD_cwksp_used(ws) <= neededSpace + 3);
+
+        DEBUGLOG(3, "wksp: finished allocating, %zd bytes remain available", ZSTD_cwksp_available_space(ws));
+        zc->initialized = 1;
+
+        return 0;
+    }
+}
+
+/* ZSTD_invalidateRepCodes() :
+ * ensures next compression will not use repcodes from previous block.
+ * Note : only works with regular variant;
+ *        do not use with extDict variant ! */
+void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx) {
+    int i;
+    for (i=0; i<ZSTD_REP_NUM; i++) cctx->blockState.prevCBlock->rep[i] = 0;
+    assert(!ZSTD_window_hasExtDict(cctx->blockState.matchState.window));
+}
+
+/* These are the approximate sizes for each strategy past which copying the
+ * dictionary tables into the working context is faster than using them
+ * in-place.
+ */
+static const size_t attachDictSizeCutoffs[ZSTD_STRATEGY_MAX+1] = {
+    8 KB,  /* unused */
+    8 KB,  /* ZSTD_fast */
+    16 KB, /* ZSTD_dfast */
+    32 KB, /* ZSTD_greedy */
+    32 KB, /* ZSTD_lazy */
+    32 KB, /* ZSTD_lazy2 */
+    32 KB, /* ZSTD_btlazy2 */
+    32 KB, /* ZSTD_btopt */
+    8 KB,  /* ZSTD_btultra */
+    8 KB   /* ZSTD_btultra2 */
+};
+
+static int ZSTD_shouldAttachDict(const ZSTD_CDict* cdict,
+                                 const ZSTD_CCtx_params* params,
+                                 U64 pledgedSrcSize)
+{
+    size_t cutoff = attachDictSizeCutoffs[cdict->matchState.cParams.strategy];
+    int const dedicatedDictSearch = cdict->matchState.dedicatedDictSearch;
+    return dedicatedDictSearch
+        || ( ( pledgedSrcSize <= cutoff
+            || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
+            || params->attachDictPref == ZSTD_dictForceAttach )
+          && params->attachDictPref != ZSTD_dictForceCopy
+          && !params->forceWindow ); /* dictMatchState isn't correctly
+                                      * handled in _enforceMaxDist */
+}
+
+static size_t
+ZSTD_resetCCtx_byAttachingCDict(ZSTD_CCtx* cctx,
+                        const ZSTD_CDict* cdict,
+                        ZSTD_CCtx_params params,
+                        U64 pledgedSrcSize,
+                        ZSTD_buffered_policy_e zbuff)
+{
+    {
+        ZSTD_compressionParameters adjusted_cdict_cParams = cdict->matchState.cParams;
+        unsigned const windowLog = params.cParams.windowLog;
+        assert(windowLog != 0);
+        /* Resize working context table params for input only, since the dict
+         * has its own tables. */
+        /* pledgedSrcSize == 0 means 0! */
+
+        if (cdict->matchState.dedicatedDictSearch) {
+            ZSTD_dedicatedDictSearch_revertCParams(&adjusted_cdict_cParams);
+        }
+
+        params.cParams = ZSTD_adjustCParams_internal(adjusted_cdict_cParams, pledgedSrcSize,
+                                                     cdict->dictContentSize, ZSTD_cpm_attachDict);
+        params.cParams.windowLog = windowLog;
+        FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
+                                                 ZSTDcrp_makeClean, zbuff), "");
+        assert(cctx->appliedParams.cParams.strategy == adjusted_cdict_cParams.strategy);
+    }
+
+    {   const U32 cdictEnd = (U32)( cdict->matchState.window.nextSrc
+                                  - cdict->matchState.window.base);
+        const U32 cdictLen = cdictEnd - cdict->matchState.window.dictLimit;
+        if (cdictLen == 0) {
+            /* don't even attach dictionaries with no contents */
+            DEBUGLOG(4, "skipping attaching empty dictionary");
+        } else {
+            DEBUGLOG(4, "attaching dictionary into context");
+            cctx->blockState.matchState.dictMatchState = &cdict->matchState;
+
+            /* prep working match state so dict matches never have negative indices
+             * when they are translated to the working context's index space. */
+            if (cctx->blockState.matchState.window.dictLimit < cdictEnd) {
+                cctx->blockState.matchState.window.nextSrc =
+                    cctx->blockState.matchState.window.base + cdictEnd;
+                ZSTD_window_clear(&cctx->blockState.matchState.window);
+            }
+            /* loadedDictEnd is expressed within the referential of the active context */
+            cctx->blockState.matchState.loadedDictEnd = cctx->blockState.matchState.window.dictLimit;
+    }   }
+
+    cctx->dictID = cdict->dictID;
+    cctx->dictContentSize = cdict->dictContentSize;
+
+    /* copy block state */
+    ZSTD_memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState));
+
+    return 0;
+}
+
+static size_t ZSTD_resetCCtx_byCopyingCDict(ZSTD_CCtx* cctx,
+                            const ZSTD_CDict* cdict,
+                            ZSTD_CCtx_params params,
+                            U64 pledgedSrcSize,
+                            ZSTD_buffered_policy_e zbuff)
+{
+    const ZSTD_compressionParameters *cdict_cParams = &cdict->matchState.cParams;
+
+    assert(!cdict->matchState.dedicatedDictSearch);
+
+    DEBUGLOG(4, "copying dictionary into context");
+
+    {   unsigned const windowLog = params.cParams.windowLog;
+        assert(windowLog != 0);
+        /* Copy only compression parameters related to tables. */
+        params.cParams = *cdict_cParams;
+        params.cParams.windowLog = windowLog;
+        FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
+                                                 ZSTDcrp_leaveDirty, zbuff), "");
+        assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy);
+        assert(cctx->appliedParams.cParams.hashLog == cdict_cParams->hashLog);
+        assert(cctx->appliedParams.cParams.chainLog == cdict_cParams->chainLog);
+    }
+
+    ZSTD_cwksp_mark_tables_dirty(&cctx->workspace);
+
+    /* copy tables */
+    {   size_t const chainSize = (cdict_cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cdict_cParams->chainLog);
+        size_t const hSize =  (size_t)1 << cdict_cParams->hashLog;
+
+        ZSTD_memcpy(cctx->blockState.matchState.hashTable,
+               cdict->matchState.hashTable,
+               hSize * sizeof(U32));
+        ZSTD_memcpy(cctx->blockState.matchState.chainTable,
+               cdict->matchState.chainTable,
+               chainSize * sizeof(U32));
+    }
+
+    /* Zero the hashTable3, since the cdict never fills it */
+    {   int const h3log = cctx->blockState.matchState.hashLog3;
+        size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0;
+        assert(cdict->matchState.hashLog3 == 0);
+        ZSTD_memset(cctx->blockState.matchState.hashTable3, 0, h3Size * sizeof(U32));
+    }
+
+    ZSTD_cwksp_mark_tables_clean(&cctx->workspace);
+
+    /* copy dictionary offsets */
+    {   ZSTD_matchState_t const* srcMatchState = &cdict->matchState;
+        ZSTD_matchState_t* dstMatchState = &cctx->blockState.matchState;
+        dstMatchState->window       = srcMatchState->window;
+        dstMatchState->nextToUpdate = srcMatchState->nextToUpdate;
+        dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd;
+    }
+
+    cctx->dictID = cdict->dictID;
+    cctx->dictContentSize = cdict->dictContentSize;
+
+    /* copy block state */
+    ZSTD_memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState));
+
+    return 0;
+}
+
+/* We have a choice between copying the dictionary context into the working
+ * context, or referencing the dictionary context from the working context
+ * in-place. We decide here which strategy to use. */
+static size_t ZSTD_resetCCtx_usingCDict(ZSTD_CCtx* cctx,
+                            const ZSTD_CDict* cdict,
+                            const ZSTD_CCtx_params* params,
+                            U64 pledgedSrcSize,
+                            ZSTD_buffered_policy_e zbuff)
+{
+
+    DEBUGLOG(4, "ZSTD_resetCCtx_usingCDict (pledgedSrcSize=%u)",
+                (unsigned)pledgedSrcSize);
+
+    if (ZSTD_shouldAttachDict(cdict, params, pledgedSrcSize)) {
+        return ZSTD_resetCCtx_byAttachingCDict(
+            cctx, cdict, *params, pledgedSrcSize, zbuff);
+    } else {
+        return ZSTD_resetCCtx_byCopyingCDict(
+            cctx, cdict, *params, pledgedSrcSize, zbuff);
+    }
+}
+
+/*! ZSTD_copyCCtx_internal() :
+ *  Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
+ *  Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
+ *  The "context", in this case, refers to the hash and chain tables,
+ *  entropy tables, and dictionary references.
+ * `windowLog` value is enforced if != 0, otherwise value is copied from srcCCtx.
+ * @return : 0, or an error code */
+static size_t ZSTD_copyCCtx_internal(ZSTD_CCtx* dstCCtx,
+                            const ZSTD_CCtx* srcCCtx,
+                            ZSTD_frameParameters fParams,
+                            U64 pledgedSrcSize,
+                            ZSTD_buffered_policy_e zbuff)
+{
+    DEBUGLOG(5, "ZSTD_copyCCtx_internal");
+    RETURN_ERROR_IF(srcCCtx->stage!=ZSTDcs_init, stage_wrong,
+                    "Can't copy a ctx that's not in init stage.");
+
+    ZSTD_memcpy(&dstCCtx->customMem, &srcCCtx->customMem, sizeof(ZSTD_customMem));
+    {   ZSTD_CCtx_params params = dstCCtx->requestedParams;
+        /* Copy only compression parameters related to tables. */
+        params.cParams = srcCCtx->appliedParams.cParams;
+        params.fParams = fParams;
+        ZSTD_resetCCtx_internal(dstCCtx, params, pledgedSrcSize,
+                                ZSTDcrp_leaveDirty, zbuff);
+        assert(dstCCtx->appliedParams.cParams.windowLog == srcCCtx->appliedParams.cParams.windowLog);
+        assert(dstCCtx->appliedParams.cParams.strategy == srcCCtx->appliedParams.cParams.strategy);
+        assert(dstCCtx->appliedParams.cParams.hashLog == srcCCtx->appliedParams.cParams.hashLog);
+        assert(dstCCtx->appliedParams.cParams.chainLog == srcCCtx->appliedParams.cParams.chainLog);
+        assert(dstCCtx->blockState.matchState.hashLog3 == srcCCtx->blockState.matchState.hashLog3);
+    }
+
+    ZSTD_cwksp_mark_tables_dirty(&dstCCtx->workspace);
+
+    /* copy tables */
+    {   size_t const chainSize = (srcCCtx->appliedParams.cParams.strategy == ZSTD_fast) ? 0 : ((size_t)1 << srcCCtx->appliedParams.cParams.chainLog);
+        size_t const hSize =  (size_t)1 << srcCCtx->appliedParams.cParams.hashLog;
+        int const h3log = srcCCtx->blockState.matchState.hashLog3;
+        size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0;
+
+        ZSTD_memcpy(dstCCtx->blockState.matchState.hashTable,
+               srcCCtx->blockState.matchState.hashTable,
+               hSize * sizeof(U32));
+        ZSTD_memcpy(dstCCtx->blockState.matchState.chainTable,
+               srcCCtx->blockState.matchState.chainTable,
+               chainSize * sizeof(U32));
+        ZSTD_memcpy(dstCCtx->blockState.matchState.hashTable3,
+               srcCCtx->blockState.matchState.hashTable3,
+               h3Size * sizeof(U32));
+    }
+
+    ZSTD_cwksp_mark_tables_clean(&dstCCtx->workspace);
+
+    /* copy dictionary offsets */
+    {
+        const ZSTD_matchState_t* srcMatchState = &srcCCtx->blockState.matchState;
+        ZSTD_matchState_t* dstMatchState = &dstCCtx->blockState.matchState;
+        dstMatchState->window       = srcMatchState->window;
+        dstMatchState->nextToUpdate = srcMatchState->nextToUpdate;
+        dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd;
+    }
+    dstCCtx->dictID = srcCCtx->dictID;
+    dstCCtx->dictContentSize = srcCCtx->dictContentSize;
+
+    /* copy block state */
+    ZSTD_memcpy(dstCCtx->blockState.prevCBlock, srcCCtx->blockState.prevCBlock, sizeof(*srcCCtx->blockState.prevCBlock));
+
+    return 0;
+}
+
+/*! ZSTD_copyCCtx() :
+ *  Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
+ *  Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
+ *  pledgedSrcSize==0 means "unknown".
+*   @return : 0, or an error code */
+size_t ZSTD_copyCCtx(ZSTD_CCtx* dstCCtx, const ZSTD_CCtx* srcCCtx, unsigned long long pledgedSrcSize)
+{
+    ZSTD_frameParameters fParams = { 1 /*content*/, 0 /*checksum*/, 0 /*noDictID*/ };
+    ZSTD_buffered_policy_e const zbuff = srcCCtx->bufferedPolicy;
+    ZSTD_STATIC_ASSERT((U32)ZSTDb_buffered==1);
+    if (pledgedSrcSize==0) pledgedSrcSize = ZSTD_CONTENTSIZE_UNKNOWN;
+    fParams.contentSizeFlag = (pledgedSrcSize != ZSTD_CONTENTSIZE_UNKNOWN);
+
+    return ZSTD_copyCCtx_internal(dstCCtx, srcCCtx,
+                                fParams, pledgedSrcSize,
+                                zbuff);
+}
+
+
+#define ZSTD_ROWSIZE 16
+/*! ZSTD_reduceTable() :
+ *  reduce table indexes by `reducerValue`, or squash to zero.
+ *  PreserveMark preserves "unsorted mark" for btlazy2 strategy.
+ *  It must be set to a clear 0/1 value, to remove branch during inlining.
+ *  Presume table size is a multiple of ZSTD_ROWSIZE
+ *  to help auto-vectorization */
+FORCE_INLINE_TEMPLATE void
+ZSTD_reduceTable_internal (U32* const table, U32 const size, U32 const reducerValue, int const preserveMark)
+{
+    int const nbRows = (int)size / ZSTD_ROWSIZE;
+    int cellNb = 0;
+    int rowNb;
+    assert((size & (ZSTD_ROWSIZE-1)) == 0);  /* multiple of ZSTD_ROWSIZE */
+    assert(size < (1U<<31));   /* can be casted to int */
+
+
+    for (rowNb=0 ; rowNb < nbRows ; rowNb++) {
+        int column;
+        for (column=0; column<ZSTD_ROWSIZE; column++) {
+            if (preserveMark) {
+                U32 const adder = (table[cellNb] == ZSTD_DUBT_UNSORTED_MARK) ? reducerValue : 0;
+                table[cellNb] += adder;
+            }
+            if (table[cellNb] < reducerValue) table[cellNb] = 0;
+            else table[cellNb] -= reducerValue;
+            cellNb++;
+    }   }
+}
+
+static void ZSTD_reduceTable(U32* const table, U32 const size, U32 const reducerValue)
+{
+    ZSTD_reduceTable_internal(table, size, reducerValue, 0);
+}
+
+static void ZSTD_reduceTable_btlazy2(U32* const table, U32 const size, U32 const reducerValue)
+{
+    ZSTD_reduceTable_internal(table, size, reducerValue, 1);
+}
+
+/*! ZSTD_reduceIndex() :
+*   rescale all indexes to avoid future overflow (indexes are U32) */
+static void ZSTD_reduceIndex (ZSTD_matchState_t* ms, ZSTD_CCtx_params const* params, const U32 reducerValue)
+{
+    {   U32 const hSize = (U32)1 << params->cParams.hashLog;
+        ZSTD_reduceTable(ms->hashTable, hSize, reducerValue);
+    }
+
+    if (params->cParams.strategy != ZSTD_fast) {
+        U32 const chainSize = (U32)1 << params->cParams.chainLog;
+        if (params->cParams.strategy == ZSTD_btlazy2)
+            ZSTD_reduceTable_btlazy2(ms->chainTable, chainSize, reducerValue);
+        else
+            ZSTD_reduceTable(ms->chainTable, chainSize, reducerValue);
+    }
+
+    if (ms->hashLog3) {
+        U32 const h3Size = (U32)1 << ms->hashLog3;
+        ZSTD_reduceTable(ms->hashTable3, h3Size, reducerValue);
+    }
+}
+
+
+/*-*******************************************************
+*  Block entropic compression
+*********************************************************/
+
+/* See doc/zstd_compression_format.md for detailed format description */
+
+void ZSTD_seqToCodes(const seqStore_t* seqStorePtr)
+{
+    const seqDef* const sequences = seqStorePtr->sequencesStart;
+    BYTE* const llCodeTable = seqStorePtr->llCode;
+    BYTE* const ofCodeTable = seqStorePtr->ofCode;
+    BYTE* const mlCodeTable = seqStorePtr->mlCode;
+    U32 const nbSeq = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
+    U32 u;
+    assert(nbSeq <= seqStorePtr->maxNbSeq);
+    for (u=0; u<nbSeq; u++) {
+        U32 const llv = sequences[u].litLength;
+        U32 const mlv = sequences[u].matchLength;
+        llCodeTable[u] = (BYTE)ZSTD_LLcode(llv);
+        ofCodeTable[u] = (BYTE)ZSTD_highbit32(sequences[u].offset);
+        mlCodeTable[u] = (BYTE)ZSTD_MLcode(mlv);
+    }
+    if (seqStorePtr->longLengthID==1)
+        llCodeTable[seqStorePtr->longLengthPos] = MaxLL;
+    if (seqStorePtr->longLengthID==2)
+        mlCodeTable[seqStorePtr->longLengthPos] = MaxML;
+}
+
+/* ZSTD_useTargetCBlockSize():
+ * Returns if target compressed block size param is being used.
+ * If used, compression will do best effort to make a compressed block size to be around targetCBlockSize.
+ * Returns 1 if true, 0 otherwise. */
+static int ZSTD_useTargetCBlockSize(const ZSTD_CCtx_params* cctxParams)
+{
+    DEBUGLOG(5, "ZSTD_useTargetCBlockSize (targetCBlockSize=%zu)", cctxParams->targetCBlockSize);
+    return (cctxParams->targetCBlockSize != 0);
+}
+
+/* ZSTD_entropyCompressSequences_internal():
+ * actually compresses both literals and sequences */
+MEM_STATIC size_t
+ZSTD_entropyCompressSequences_internal(seqStore_t* seqStorePtr,
+                          const ZSTD_entropyCTables_t* prevEntropy,
+                                ZSTD_entropyCTables_t* nextEntropy,
+                          const ZSTD_CCtx_params* cctxParams,
+                                void* dst, size_t dstCapacity,
+                                void* entropyWorkspace, size_t entropyWkspSize,
+                          const int bmi2)
+{
+    const int longOffsets = cctxParams->cParams.windowLog > STREAM_ACCUMULATOR_MIN;
+    ZSTD_strategy const strategy = cctxParams->cParams.strategy;
+    unsigned* count = (unsigned*)entropyWorkspace;
+    FSE_CTable* CTable_LitLength = nextEntropy->fse.litlengthCTable;
+    FSE_CTable* CTable_OffsetBits = nextEntropy->fse.offcodeCTable;
+    FSE_CTable* CTable_MatchLength = nextEntropy->fse.matchlengthCTable;
+    U32 LLtype, Offtype, MLtype;   /* compressed, raw or rle */
+    const seqDef* const sequences = seqStorePtr->sequencesStart;
+    const BYTE* const ofCodeTable = seqStorePtr->ofCode;
+    const BYTE* const llCodeTable = seqStorePtr->llCode;
+    const BYTE* const mlCodeTable = seqStorePtr->mlCode;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + dstCapacity;
+    BYTE* op = ostart;
+    size_t const nbSeq = (size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
+    BYTE* seqHead;
+    BYTE* lastNCount = NULL;
+
+    entropyWorkspace = count + (MaxSeq + 1);
+    entropyWkspSize -= (MaxSeq + 1) * sizeof(*count);
+
+    DEBUGLOG(4, "ZSTD_entropyCompressSequences_internal (nbSeq=%zu)", nbSeq);
+    ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog)));
+    assert(entropyWkspSize >= HUF_WORKSPACE_SIZE);
+
+    /* Compress literals */
+    {   const BYTE* const literals = seqStorePtr->litStart;
+        size_t const litSize = (size_t)(seqStorePtr->lit - literals);
+        size_t const cSize = ZSTD_compressLiterals(
+                                    &prevEntropy->huf, &nextEntropy->huf,
+                                    cctxParams->cParams.strategy,
+                                    ZSTD_disableLiteralsCompression(cctxParams),
+                                    op, dstCapacity,
+                                    literals, litSize,
+                                    entropyWorkspace, entropyWkspSize,
+                                    bmi2);
+        FORWARD_IF_ERROR(cSize, "ZSTD_compressLiterals failed");
+        assert(cSize <= dstCapacity);
+        op += cSize;
+    }
+
+    /* Sequences Header */
+    RETURN_ERROR_IF((oend-op) < 3 /*max nbSeq Size*/ + 1 /*seqHead*/,
+                    dstSize_tooSmall, "Can't fit seq hdr in output buf!");
+    if (nbSeq < 128) {
+        *op++ = (BYTE)nbSeq;
+    } else if (nbSeq < LONGNBSEQ) {
+        op[0] = (BYTE)((nbSeq>>8) + 0x80);
+        op[1] = (BYTE)nbSeq;
+        op+=2;
+    } else {
+        op[0]=0xFF;
+        MEM_writeLE16(op+1, (U16)(nbSeq - LONGNBSEQ));
+        op+=3;
+    }
+    assert(op <= oend);
+    if (nbSeq==0) {
+        /* Copy the old tables over as if we repeated them */
+        ZSTD_memcpy(&nextEntropy->fse, &prevEntropy->fse, sizeof(prevEntropy->fse));
+        return (size_t)(op - ostart);
+    }
+
+    /* seqHead : flags for FSE encoding type */
+    seqHead = op++;
+    assert(op <= oend);
+
+    /* convert length/distances into codes */
+    ZSTD_seqToCodes(seqStorePtr);
+    /* build CTable for Literal Lengths */
+    {   unsigned max = MaxLL;
+        size_t const mostFrequent = HIST_countFast_wksp(count, &max, llCodeTable, nbSeq, entropyWorkspace, entropyWkspSize);   /* can't fail */
+        DEBUGLOG(5, "Building LL table");
+        nextEntropy->fse.litlength_repeatMode = prevEntropy->fse.litlength_repeatMode;
+        LLtype = ZSTD_selectEncodingType(&nextEntropy->fse.litlength_repeatMode,
+                                        count, max, mostFrequent, nbSeq,
+                                        LLFSELog, prevEntropy->fse.litlengthCTable,
+                                        LL_defaultNorm, LL_defaultNormLog,
+                                        ZSTD_defaultAllowed, strategy);
+        assert(set_basic < set_compressed && set_rle < set_compressed);
+        assert(!(LLtype < set_compressed && nextEntropy->fse.litlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
+        {   size_t const countSize = ZSTD_buildCTable(
+                op, (size_t)(oend - op),
+                CTable_LitLength, LLFSELog, (symbolEncodingType_e)LLtype,
+                count, max, llCodeTable, nbSeq,
+                LL_defaultNorm, LL_defaultNormLog, MaxLL,
+                prevEntropy->fse.litlengthCTable,
+                sizeof(prevEntropy->fse.litlengthCTable),
+                entropyWorkspace, entropyWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for LitLens failed");
+            if (LLtype == set_compressed)
+                lastNCount = op;
+            op += countSize;
+            assert(op <= oend);
+    }   }
+    /* build CTable for Offsets */
+    {   unsigned max = MaxOff;
+        size_t const mostFrequent = HIST_countFast_wksp(
+            count, &max, ofCodeTable, nbSeq, entropyWorkspace, entropyWkspSize);  /* can't fail */
+        /* We can only use the basic table if max <= DefaultMaxOff, otherwise the offsets are too large */
+        ZSTD_defaultPolicy_e const defaultPolicy = (max <= DefaultMaxOff) ? ZSTD_defaultAllowed : ZSTD_defaultDisallowed;
+        DEBUGLOG(5, "Building OF table");
+        nextEntropy->fse.offcode_repeatMode = prevEntropy->fse.offcode_repeatMode;
+        Offtype = ZSTD_selectEncodingType(&nextEntropy->fse.offcode_repeatMode,
+                                        count, max, mostFrequent, nbSeq,
+                                        OffFSELog, prevEntropy->fse.offcodeCTable,
+                                        OF_defaultNorm, OF_defaultNormLog,
+                                        defaultPolicy, strategy);
+        assert(!(Offtype < set_compressed && nextEntropy->fse.offcode_repeatMode != FSE_repeat_none)); /* We don't copy tables */
+        {   size_t const countSize = ZSTD_buildCTable(
+                op, (size_t)(oend - op),
+                CTable_OffsetBits, OffFSELog, (symbolEncodingType_e)Offtype,
+                count, max, ofCodeTable, nbSeq,
+                OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
+                prevEntropy->fse.offcodeCTable,
+                sizeof(prevEntropy->fse.offcodeCTable),
+                entropyWorkspace, entropyWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for Offsets failed");
+            if (Offtype == set_compressed)
+                lastNCount = op;
+            op += countSize;
+            assert(op <= oend);
+    }   }
+    /* build CTable for MatchLengths */
+    {   unsigned max = MaxML;
+        size_t const mostFrequent = HIST_countFast_wksp(
+            count, &max, mlCodeTable, nbSeq, entropyWorkspace, entropyWkspSize);   /* can't fail */
+        DEBUGLOG(5, "Building ML table (remaining space : %i)", (int)(oend-op));
+        nextEntropy->fse.matchlength_repeatMode = prevEntropy->fse.matchlength_repeatMode;
+        MLtype = ZSTD_selectEncodingType(&nextEntropy->fse.matchlength_repeatMode,
+                                        count, max, mostFrequent, nbSeq,
+                                        MLFSELog, prevEntropy->fse.matchlengthCTable,
+                                        ML_defaultNorm, ML_defaultNormLog,
+                                        ZSTD_defaultAllowed, strategy);
+        assert(!(MLtype < set_compressed && nextEntropy->fse.matchlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
+        {   size_t const countSize = ZSTD_buildCTable(
+                op, (size_t)(oend - op),
+                CTable_MatchLength, MLFSELog, (symbolEncodingType_e)MLtype,
+                count, max, mlCodeTable, nbSeq,
+                ML_defaultNorm, ML_defaultNormLog, MaxML,
+                prevEntropy->fse.matchlengthCTable,
+                sizeof(prevEntropy->fse.matchlengthCTable),
+                entropyWorkspace, entropyWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for MatchLengths failed");
+            if (MLtype == set_compressed)
+                lastNCount = op;
+            op += countSize;
+            assert(op <= oend);
+    }   }
+
+    *seqHead = (BYTE)((LLtype<<6) + (Offtype<<4) + (MLtype<<2));
+
+    {   size_t const bitstreamSize = ZSTD_encodeSequences(
+                                        op, (size_t)(oend - op),
+                                        CTable_MatchLength, mlCodeTable,
+                                        CTable_OffsetBits, ofCodeTable,
+                                        CTable_LitLength, llCodeTable,
+                                        sequences, nbSeq,
+                                        longOffsets, bmi2);
+        FORWARD_IF_ERROR(bitstreamSize, "ZSTD_encodeSequences failed");
+        op += bitstreamSize;
+        assert(op <= oend);
+        /* zstd versions <= 1.3.4 mistakenly report corruption when
+         * FSE_readNCount() receives a buffer < 4 bytes.
+         * Fixed by https://github.com/facebook/zstd/pull/1146.
+         * This can happen when the last set_compressed table present is 2
+         * bytes and the bitstream is only one byte.
+         * In this exceedingly rare case, we will simply emit an uncompressed
+         * block, since it isn't worth optimizing.
+         */
+        if (lastNCount && (op - lastNCount) < 4) {
+            /* NCountSize >= 2 && bitstreamSize > 0 ==> lastCountSize == 3 */
+            assert(op - lastNCount == 3);
+            DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.3.4 by "
+                        "emitting an uncompressed block.");
+            return 0;
+        }
+    }
+
+    DEBUGLOG(5, "compressed block size : %u", (unsigned)(op - ostart));
+    return (size_t)(op - ostart);
+}
+
+MEM_STATIC size_t
+ZSTD_entropyCompressSequences(seqStore_t* seqStorePtr,
+                       const ZSTD_entropyCTables_t* prevEntropy,
+                             ZSTD_entropyCTables_t* nextEntropy,
+                       const ZSTD_CCtx_params* cctxParams,
+                             void* dst, size_t dstCapacity,
+                             size_t srcSize,
+                             void* entropyWorkspace, size_t entropyWkspSize,
+                             int bmi2)
+{
+    size_t const cSize = ZSTD_entropyCompressSequences_internal(
+                            seqStorePtr, prevEntropy, nextEntropy, cctxParams,
+                            dst, dstCapacity,
+                            entropyWorkspace, entropyWkspSize, bmi2);
+    if (cSize == 0) return 0;
+    /* When srcSize <= dstCapacity, there is enough space to write a raw uncompressed block.
+     * Since we ran out of space, block must be not compressible, so fall back to raw uncompressed block.
+     */
+    if ((cSize == ERROR(dstSize_tooSmall)) & (srcSize <= dstCapacity))
+        return 0;  /* block not compressed */
+    FORWARD_IF_ERROR(cSize, "ZSTD_entropyCompressSequences_internal failed");
+
+    /* Check compressibility */
+    {   size_t const maxCSize = srcSize - ZSTD_minGain(srcSize, cctxParams->cParams.strategy);
+        if (cSize >= maxCSize) return 0;  /* block not compressed */
+    }
+    DEBUGLOG(4, "ZSTD_entropyCompressSequences() cSize: %zu\n", cSize);
+    return cSize;
+}
+
+/* ZSTD_selectBlockCompressor() :
+ * Not static, but internal use only (used by long distance matcher)
+ * assumption : strat is a valid strategy */
+ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, ZSTD_dictMode_e dictMode)
+{
+    static const ZSTD_blockCompressor blockCompressor[4][ZSTD_STRATEGY_MAX+1] = {
+        { ZSTD_compressBlock_fast  /* default for 0 */,
+          ZSTD_compressBlock_fast,
+          ZSTD_compressBlock_doubleFast,
+          ZSTD_compressBlock_greedy,
+          ZSTD_compressBlock_lazy,
+          ZSTD_compressBlock_lazy2,
+          ZSTD_compressBlock_btlazy2,
+          ZSTD_compressBlock_btopt,
+          ZSTD_compressBlock_btultra,
+          ZSTD_compressBlock_btultra2 },
+        { ZSTD_compressBlock_fast_extDict  /* default for 0 */,
+          ZSTD_compressBlock_fast_extDict,
+          ZSTD_compressBlock_doubleFast_extDict,
+          ZSTD_compressBlock_greedy_extDict,
+          ZSTD_compressBlock_lazy_extDict,
+          ZSTD_compressBlock_lazy2_extDict,
+          ZSTD_compressBlock_btlazy2_extDict,
+          ZSTD_compressBlock_btopt_extDict,
+          ZSTD_compressBlock_btultra_extDict,
+          ZSTD_compressBlock_btultra_extDict },
+        { ZSTD_compressBlock_fast_dictMatchState  /* default for 0 */,
+          ZSTD_compressBlock_fast_dictMatchState,
+          ZSTD_compressBlock_doubleFast_dictMatchState,
+          ZSTD_compressBlock_greedy_dictMatchState,
+          ZSTD_compressBlock_lazy_dictMatchState,
+          ZSTD_compressBlock_lazy2_dictMatchState,
+          ZSTD_compressBlock_btlazy2_dictMatchState,
+          ZSTD_compressBlock_btopt_dictMatchState,
+          ZSTD_compressBlock_btultra_dictMatchState,
+          ZSTD_compressBlock_btultra_dictMatchState },
+        { NULL  /* default for 0 */,
+          NULL,
+          NULL,
+          ZSTD_compressBlock_greedy_dedicatedDictSearch,
+          ZSTD_compressBlock_lazy_dedicatedDictSearch,
+          ZSTD_compressBlock_lazy2_dedicatedDictSearch,
+          NULL,
+          NULL,
+          NULL,
+          NULL }
+    };
+    ZSTD_blockCompressor selectedCompressor;
+    ZSTD_STATIC_ASSERT((unsigned)ZSTD_fast == 1);
+
+    assert(ZSTD_cParam_withinBounds(ZSTD_c_strategy, strat));
+    selectedCompressor = blockCompressor[(int)dictMode][(int)strat];
+    assert(selectedCompressor != NULL);
+    return selectedCompressor;
+}
+
+static void ZSTD_storeLastLiterals(seqStore_t* seqStorePtr,
+                                   const BYTE* anchor, size_t lastLLSize)
+{
+    ZSTD_memcpy(seqStorePtr->lit, anchor, lastLLSize);
+    seqStorePtr->lit += lastLLSize;
+}
+
+void ZSTD_resetSeqStore(seqStore_t* ssPtr)
+{
+    ssPtr->lit = ssPtr->litStart;
+    ssPtr->sequences = ssPtr->sequencesStart;
+    ssPtr->longLengthID = 0;
+}
+
+typedef enum { ZSTDbss_compress, ZSTDbss_noCompress } ZSTD_buildSeqStore_e;
+
+static size_t ZSTD_buildSeqStore(ZSTD_CCtx* zc, const void* src, size_t srcSize)
+{
+    ZSTD_matchState_t* const ms = &zc->blockState.matchState;
+    DEBUGLOG(5, "ZSTD_buildSeqStore (srcSize=%zu)", srcSize);
+    assert(srcSize <= ZSTD_BLOCKSIZE_MAX);
+    /* Assert that we have correctly flushed the ctx params into the ms's copy */
+    ZSTD_assertEqualCParams(zc->appliedParams.cParams, ms->cParams);
+    if (srcSize < MIN_CBLOCK_SIZE+ZSTD_blockHeaderSize+1) {
+        if (zc->appliedParams.cParams.strategy >= ZSTD_btopt) {
+            ZSTD_ldm_skipRawSeqStoreBytes(&zc->externSeqStore, srcSize);
+        } else {
+            ZSTD_ldm_skipSequences(&zc->externSeqStore, srcSize, zc->appliedParams.cParams.minMatch);
+        }
+        return ZSTDbss_noCompress; /* don't even attempt compression below a certain srcSize */
+    }
+    ZSTD_resetSeqStore(&(zc->seqStore));
+    /* required for optimal parser to read stats from dictionary */
+    ms->opt.symbolCosts = &zc->blockState.prevCBlock->entropy;
+    /* tell the optimal parser how we expect to compress literals */
+    ms->opt.literalCompressionMode = zc->appliedParams.literalCompressionMode;
+    /* a gap between an attached dict and the current window is not safe,
+     * they must remain adjacent,
+     * and when that stops being the case, the dict must be unset */
+    assert(ms->dictMatchState == NULL || ms->loadedDictEnd == ms->window.dictLimit);
+
+    /* limited update after a very long match */
+    {   const BYTE* const base = ms->window.base;
+        const BYTE* const istart = (const BYTE*)src;
+        const U32 curr = (U32)(istart-base);
+        if (sizeof(ptrdiff_t)==8) assert(istart - base < (ptrdiff_t)(U32)(-1));   /* ensure no overflow */
+        if (curr > ms->nextToUpdate + 384)
+            ms->nextToUpdate = curr - MIN(192, (U32)(curr - ms->nextToUpdate - 384));
+    }
+
+    /* select and store sequences */
+    {   ZSTD_dictMode_e const dictMode = ZSTD_matchState_dictMode(ms);
+        size_t lastLLSize;
+        {   int i;
+            for (i = 0; i < ZSTD_REP_NUM; ++i)
+                zc->blockState.nextCBlock->rep[i] = zc->blockState.prevCBlock->rep[i];
+        }
+        if (zc->externSeqStore.pos < zc->externSeqStore.size) {
+            assert(!zc->appliedParams.ldmParams.enableLdm);
+            /* Updates ldmSeqStore.pos */
+            lastLLSize =
+                ZSTD_ldm_blockCompress(&zc->externSeqStore,
+                                       ms, &zc->seqStore,
+                                       zc->blockState.nextCBlock->rep,
+                                       src, srcSize);
+            assert(zc->externSeqStore.pos <= zc->externSeqStore.size);
+        } else if (zc->appliedParams.ldmParams.enableLdm) {
+            rawSeqStore_t ldmSeqStore = kNullRawSeqStore;
+
+            ldmSeqStore.seq = zc->ldmSequences;
+            ldmSeqStore.capacity = zc->maxNbLdmSequences;
+            /* Updates ldmSeqStore.size */
+            FORWARD_IF_ERROR(ZSTD_ldm_generateSequences(&zc->ldmState, &ldmSeqStore,
+                                               &zc->appliedParams.ldmParams,
+                                               src, srcSize), "");
+            /* Updates ldmSeqStore.pos */
+            lastLLSize =
+                ZSTD_ldm_blockCompress(&ldmSeqStore,
+                                       ms, &zc->seqStore,
+                                       zc->blockState.nextCBlock->rep,
+                                       src, srcSize);
+            assert(ldmSeqStore.pos == ldmSeqStore.size);
+        } else {   /* not long range mode */
+            ZSTD_blockCompressor const blockCompressor = ZSTD_selectBlockCompressor(zc->appliedParams.cParams.strategy, dictMode);
+            ms->ldmSeqStore = NULL;
+            lastLLSize = blockCompressor(ms, &zc->seqStore, zc->blockState.nextCBlock->rep, src, srcSize);
+        }
+        {   const BYTE* const lastLiterals = (const BYTE*)src + srcSize - lastLLSize;
+            ZSTD_storeLastLiterals(&zc->seqStore, lastLiterals, lastLLSize);
+    }   }
+    return ZSTDbss_compress;
+}
+
+static void ZSTD_copyBlockSequences(ZSTD_CCtx* zc)
+{
+    const seqStore_t* seqStore = ZSTD_getSeqStore(zc);
+    const seqDef* seqStoreSeqs = seqStore->sequencesStart;
+    size_t seqStoreSeqSize = seqStore->sequences - seqStoreSeqs;
+    size_t seqStoreLiteralsSize = (size_t)(seqStore->lit - seqStore->litStart);
+    size_t literalsRead = 0;
+    size_t lastLLSize;
+
+    ZSTD_Sequence* outSeqs = &zc->seqCollector.seqStart[zc->seqCollector.seqIndex];
+    size_t i;
+    repcodes_t updatedRepcodes;
+
+    assert(zc->seqCollector.seqIndex + 1 < zc->seqCollector.maxSequences);
+    /* Ensure we have enough space for last literals "sequence" */
+    assert(zc->seqCollector.maxSequences >= seqStoreSeqSize + 1);
+    ZSTD_memcpy(updatedRepcodes.rep, zc->blockState.prevCBlock->rep, sizeof(repcodes_t));
+    for (i = 0; i < seqStoreSeqSize; ++i) {
+        U32 rawOffset = seqStoreSeqs[i].offset - ZSTD_REP_NUM;
+        outSeqs[i].litLength = seqStoreSeqs[i].litLength;
+        outSeqs[i].matchLength = seqStoreSeqs[i].matchLength + MINMATCH;
+        outSeqs[i].rep = 0;
+
+        if (i == seqStore->longLengthPos) {
+            if (seqStore->longLengthID == 1) {
+                outSeqs[i].litLength += 0x10000;
+            } else if (seqStore->longLengthID == 2) {
+                outSeqs[i].matchLength += 0x10000;
+            }
+        }
+
+        if (seqStoreSeqs[i].offset <= ZSTD_REP_NUM) {
+            /* Derive the correct offset corresponding to a repcode */
+            outSeqs[i].rep = seqStoreSeqs[i].offset;
+            if (outSeqs[i].litLength != 0) {
+                rawOffset = updatedRepcodes.rep[outSeqs[i].rep - 1];
+            } else {
+                if (outSeqs[i].rep == 3) {
+                    rawOffset = updatedRepcodes.rep[0] - 1;
+                } else {
+                    rawOffset = updatedRepcodes.rep[outSeqs[i].rep];
+                }
+            }
+        }
+        outSeqs[i].offset = rawOffset;
+        /* seqStoreSeqs[i].offset == offCode+1, and ZSTD_updateRep() expects offCode
+           so we provide seqStoreSeqs[i].offset - 1 */
+        updatedRepcodes = ZSTD_updateRep(updatedRepcodes.rep,
+                                         seqStoreSeqs[i].offset - 1,
+                                         seqStoreSeqs[i].litLength == 0);
+        literalsRead += outSeqs[i].litLength;
+    }
+    /* Insert last literals (if any exist) in the block as a sequence with ml == off == 0.
+     * If there are no last literals, then we'll emit (of: 0, ml: 0, ll: 0), which is a marker
+     * for the block boundary, according to the API.
+     */
+    assert(seqStoreLiteralsSize >= literalsRead);
+    lastLLSize = seqStoreLiteralsSize - literalsRead;
+    outSeqs[i].litLength = (U32)lastLLSize;
+    outSeqs[i].matchLength = outSeqs[i].offset = outSeqs[i].rep = 0;
+    seqStoreSeqSize++;
+    zc->seqCollector.seqIndex += seqStoreSeqSize;
+}
+
+size_t ZSTD_generateSequences(ZSTD_CCtx* zc, ZSTD_Sequence* outSeqs,
+                              size_t outSeqsSize, const void* src, size_t srcSize)
+{
+    const size_t dstCapacity = ZSTD_compressBound(srcSize);
+    void* dst = ZSTD_customMalloc(dstCapacity, ZSTD_defaultCMem);
+    SeqCollector seqCollector;
+
+    RETURN_ERROR_IF(dst == NULL, memory_allocation, "NULL pointer!");
+
+    seqCollector.collectSequences = 1;
+    seqCollector.seqStart = outSeqs;
+    seqCollector.seqIndex = 0;
+    seqCollector.maxSequences = outSeqsSize;
+    zc->seqCollector = seqCollector;
+
+    ZSTD_compress2(zc, dst, dstCapacity, src, srcSize);
+    ZSTD_customFree(dst, ZSTD_defaultCMem);
+    return zc->seqCollector.seqIndex;
+}
+
+size_t ZSTD_mergeBlockDelimiters(ZSTD_Sequence* sequences, size_t seqsSize) {
+    size_t in = 0;
+    size_t out = 0;
+    for (; in < seqsSize; ++in) {
+        if (sequences[in].offset == 0 && sequences[in].matchLength == 0) {
+            if (in != seqsSize - 1) {
+                sequences[in+1].litLength += sequences[in].litLength;
+            }
+        } else {
+            sequences[out] = sequences[in];
+            ++out;
+        }
+    }
+    return out;
+}
+
+/* Unrolled loop to read four size_ts of input at a time. Returns 1 if is RLE, 0 if not. */
+static int ZSTD_isRLE(const BYTE* src, size_t length) {
+    const BYTE* ip = src;
+    const BYTE value = ip[0];
+    const size_t valueST = (size_t)((U64)value * 0x0101010101010101ULL);
+    const size_t unrollSize = sizeof(size_t) * 4;
+    const size_t unrollMask = unrollSize - 1;
+    const size_t prefixLength = length & unrollMask;
+    size_t i;
+    size_t u;
+    if (length == 1) return 1;
+    /* Check if prefix is RLE first before using unrolled loop */
+    if (prefixLength && ZSTD_count(ip+1, ip, ip+prefixLength) != prefixLength-1) {
+        return 0;
+    }
+    for (i = prefixLength; i != length; i += unrollSize) {
+        for (u = 0; u < unrollSize; u += sizeof(size_t)) {
+            if (MEM_readST(ip + i + u) != valueST) {
+                return 0;
+            }
+        }
+    }
+    return 1;
+}
+
+/* Returns true if the given block may be RLE.
+ * This is just a heuristic based on the compressibility.
+ * It may return both false positives and false negatives.
+ */
+static int ZSTD_maybeRLE(seqStore_t const* seqStore)
+{
+    size_t const nbSeqs = (size_t)(seqStore->sequences - seqStore->sequencesStart);
+    size_t const nbLits = (size_t)(seqStore->lit - seqStore->litStart);
+
+    return nbSeqs < 4 && nbLits < 10;
+}
+
+static void ZSTD_confirmRepcodesAndEntropyTables(ZSTD_CCtx* zc)
+{
+    ZSTD_compressedBlockState_t* const tmp = zc->blockState.prevCBlock;
+    zc->blockState.prevCBlock = zc->blockState.nextCBlock;
+    zc->blockState.nextCBlock = tmp;
+}
+
+static size_t ZSTD_compressBlock_internal(ZSTD_CCtx* zc,
+                                        void* dst, size_t dstCapacity,
+                                        const void* src, size_t srcSize, U32 frame)
+{
+    /* This the upper bound for the length of an rle block.
+     * This isn't the actual upper bound. Finding the real threshold
+     * needs further investigation.
+     */
+    const U32 rleMaxLength = 25;
+    size_t cSize;
+    const BYTE* ip = (const BYTE*)src;
+    BYTE* op = (BYTE*)dst;
+    DEBUGLOG(5, "ZSTD_compressBlock_internal (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u)",
+                (unsigned)dstCapacity, (unsigned)zc->blockState.matchState.window.dictLimit,
+                (unsigned)zc->blockState.matchState.nextToUpdate);
+
+    {   const size_t bss = ZSTD_buildSeqStore(zc, src, srcSize);
+        FORWARD_IF_ERROR(bss, "ZSTD_buildSeqStore failed");
+        if (bss == ZSTDbss_noCompress) { cSize = 0; goto out; }
+    }
+
+    if (zc->seqCollector.collectSequences) {
+        ZSTD_copyBlockSequences(zc);
+        ZSTD_confirmRepcodesAndEntropyTables(zc);
+        return 0;
+    }
+
+    /* encode sequences and literals */
+    cSize = ZSTD_entropyCompressSequences(&zc->seqStore,
+            &zc->blockState.prevCBlock->entropy, &zc->blockState.nextCBlock->entropy,
+            &zc->appliedParams,
+            dst, dstCapacity,
+            srcSize,
+            zc->entropyWorkspace, ENTROPY_WORKSPACE_SIZE /* statically allocated in resetCCtx */,
+            zc->bmi2);
+
+    if (zc->seqCollector.collectSequences) {
+        ZSTD_copyBlockSequences(zc);
+        return 0;
+    }
+
+
+    if (frame &&
+        /* We don't want to emit our first block as a RLE even if it qualifies because
+         * doing so will cause the decoder (cli only) to throw a "should consume all input error."
+         * This is only an issue for zstd <= v1.4.3
+         */
+        !zc->isFirstBlock &&
+        cSize < rleMaxLength &&
+        ZSTD_isRLE(ip, srcSize))
+    {
+        cSize = 1;
+        op[0] = ip[0];
+    }
+
+out:
+    if (!ZSTD_isError(cSize) && cSize > 1) {
+        ZSTD_confirmRepcodesAndEntropyTables(zc);
+    }
+    /* We check that dictionaries have offset codes available for the first
+     * block. After the first block, the offcode table might not have large
+     * enough codes to represent the offsets in the data.
+     */
+    if (zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode == FSE_repeat_valid)
+        zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode = FSE_repeat_check;
+
+    return cSize;
+}
+
+static size_t ZSTD_compressBlock_targetCBlockSize_body(ZSTD_CCtx* zc,
+                               void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                               const size_t bss, U32 lastBlock)
+{
+    DEBUGLOG(6, "Attempting ZSTD_compressSuperBlock()");
+    if (bss == ZSTDbss_compress) {
+        if (/* We don't want to emit our first block as a RLE even if it qualifies because
+            * doing so will cause the decoder (cli only) to throw a "should consume all input error."
+            * This is only an issue for zstd <= v1.4.3
+            */
+            !zc->isFirstBlock &&
+            ZSTD_maybeRLE(&zc->seqStore) &&
+            ZSTD_isRLE((BYTE const*)src, srcSize))
+        {
+            return ZSTD_rleCompressBlock(dst, dstCapacity, *(BYTE const*)src, srcSize, lastBlock);
+        }
+        /* Attempt superblock compression.
+         *
+         * Note that compressed size of ZSTD_compressSuperBlock() is not bound by the
+         * standard ZSTD_compressBound(). This is a problem, because even if we have
+         * space now, taking an extra byte now could cause us to run out of space later
+         * and violate ZSTD_compressBound().
+         *
+         * Define blockBound(blockSize) = blockSize + ZSTD_blockHeaderSize.
+         *
+         * In order to respect ZSTD_compressBound() we must attempt to emit a raw
+         * uncompressed block in these cases:
+         *   * cSize == 0: Return code for an uncompressed block.
+         *   * cSize == dstSize_tooSmall: We may have expanded beyond blockBound(srcSize).
+         *     ZSTD_noCompressBlock() will return dstSize_tooSmall if we are really out of
+         *     output space.
+         *   * cSize >= blockBound(srcSize): We have expanded the block too much so
+         *     emit an uncompressed block.
+         */
+        {
+            size_t const cSize = ZSTD_compressSuperBlock(zc, dst, dstCapacity, src, srcSize, lastBlock);
+            if (cSize != ERROR(dstSize_tooSmall)) {
+                size_t const maxCSize = srcSize - ZSTD_minGain(srcSize, zc->appliedParams.cParams.strategy);
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressSuperBlock failed");
+                if (cSize != 0 && cSize < maxCSize + ZSTD_blockHeaderSize) {
+                    ZSTD_confirmRepcodesAndEntropyTables(zc);
+                    return cSize;
+                }
+            }
+        }
+    }
+
+    DEBUGLOG(6, "Resorting to ZSTD_noCompressBlock()");
+    /* Superblock compression failed, attempt to emit a single no compress block.
+     * The decoder will be able to stream this block since it is uncompressed.
+     */
+    return ZSTD_noCompressBlock(dst, dstCapacity, src, srcSize, lastBlock);
+}
+
+static size_t ZSTD_compressBlock_targetCBlockSize(ZSTD_CCtx* zc,
+                               void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                               U32 lastBlock)
+{
+    size_t cSize = 0;
+    const size_t bss = ZSTD_buildSeqStore(zc, src, srcSize);
+    DEBUGLOG(5, "ZSTD_compressBlock_targetCBlockSize (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u, srcSize=%zu)",
+                (unsigned)dstCapacity, (unsigned)zc->blockState.matchState.window.dictLimit, (unsigned)zc->blockState.matchState.nextToUpdate, srcSize);
+    FORWARD_IF_ERROR(bss, "ZSTD_buildSeqStore failed");
+
+    cSize = ZSTD_compressBlock_targetCBlockSize_body(zc, dst, dstCapacity, src, srcSize, bss, lastBlock);
+    FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_targetCBlockSize_body failed");
+
+    if (zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode == FSE_repeat_valid)
+        zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode = FSE_repeat_check;
+
+    return cSize;
+}
+
+static void ZSTD_overflowCorrectIfNeeded(ZSTD_matchState_t* ms,
+                                         ZSTD_cwksp* ws,
+                                         ZSTD_CCtx_params const* params,
+                                         void const* ip,
+                                         void const* iend)
+{
+    if (ZSTD_window_needOverflowCorrection(ms->window, iend)) {
+        U32 const maxDist = (U32)1 << params->cParams.windowLog;
+        U32 const cycleLog = ZSTD_cycleLog(params->cParams.chainLog, params->cParams.strategy);
+        U32 const correction = ZSTD_window_correctOverflow(&ms->window, cycleLog, maxDist, ip);
+        ZSTD_STATIC_ASSERT(ZSTD_CHAINLOG_MAX <= 30);
+        ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX_32 <= 30);
+        ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX <= 31);
+        ZSTD_cwksp_mark_tables_dirty(ws);
+        ZSTD_reduceIndex(ms, params, correction);
+        ZSTD_cwksp_mark_tables_clean(ws);
+        if (ms->nextToUpdate < correction) ms->nextToUpdate = 0;
+        else ms->nextToUpdate -= correction;
+        /* invalidate dictionaries on overflow correction */
+        ms->loadedDictEnd = 0;
+        ms->dictMatchState = NULL;
+    }
+}
+
+/*! ZSTD_compress_frameChunk() :
+*   Compress a chunk of data into one or multiple blocks.
+*   All blocks will be terminated, all input will be consumed.
+*   Function will issue an error if there is not enough `dstCapacity` to hold the compressed content.
+*   Frame is supposed already started (header already produced)
+*   @return : compressed size, or an error code
+*/
+static size_t ZSTD_compress_frameChunk (ZSTD_CCtx* cctx,
+                                     void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                                     U32 lastFrameChunk)
+{
+    size_t blockSize = cctx->blockSize;
+    size_t remaining = srcSize;
+    const BYTE* ip = (const BYTE*)src;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* op = ostart;
+    U32 const maxDist = (U32)1 << cctx->appliedParams.cParams.windowLog;
+
+    assert(cctx->appliedParams.cParams.windowLog <= ZSTD_WINDOWLOG_MAX);
+
+    DEBUGLOG(4, "ZSTD_compress_frameChunk (blockSize=%u)", (unsigned)blockSize);
+    if (cctx->appliedParams.fParams.checksumFlag && srcSize)
+        xxh64_update(&cctx->xxhState, src, srcSize);
+
+    while (remaining) {
+        ZSTD_matchState_t* const ms = &cctx->blockState.matchState;
+        U32 const lastBlock = lastFrameChunk & (blockSize >= remaining);
+
+        RETURN_ERROR_IF(dstCapacity < ZSTD_blockHeaderSize + MIN_CBLOCK_SIZE,
+                        dstSize_tooSmall,
+                        "not enough space to store compressed block");
+        if (remaining < blockSize) blockSize = remaining;
+
+        ZSTD_overflowCorrectIfNeeded(
+            ms, &cctx->workspace, &cctx->appliedParams, ip, ip + blockSize);
+        ZSTD_checkDictValidity(&ms->window, ip + blockSize, maxDist, &ms->loadedDictEnd, &ms->dictMatchState);
+
+        /* Ensure hash/chain table insertion resumes no sooner than lowlimit */
+        if (ms->nextToUpdate < ms->window.lowLimit) ms->nextToUpdate = ms->window.lowLimit;
+
+        {   size_t cSize;
+            if (ZSTD_useTargetCBlockSize(&cctx->appliedParams)) {
+                cSize = ZSTD_compressBlock_targetCBlockSize(cctx, op, dstCapacity, ip, blockSize, lastBlock);
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_targetCBlockSize failed");
+                assert(cSize > 0);
+                assert(cSize <= blockSize + ZSTD_blockHeaderSize);
+            } else {
+                cSize = ZSTD_compressBlock_internal(cctx,
+                                        op+ZSTD_blockHeaderSize, dstCapacity-ZSTD_blockHeaderSize,
+                                        ip, blockSize, 1 /* frame */);
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_internal failed");
+
+                if (cSize == 0) {  /* block is not compressible */
+                    cSize = ZSTD_noCompressBlock(op, dstCapacity, ip, blockSize, lastBlock);
+                    FORWARD_IF_ERROR(cSize, "ZSTD_noCompressBlock failed");
+                } else {
+                    U32 const cBlockHeader = cSize == 1 ?
+                        lastBlock + (((U32)bt_rle)<<1) + (U32)(blockSize << 3) :
+                        lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3);
+                    MEM_writeLE24(op, cBlockHeader);
+                    cSize += ZSTD_blockHeaderSize;
+                }
+            }
+
+
+            ip += blockSize;
+            assert(remaining >= blockSize);
+            remaining -= blockSize;
+            op += cSize;
+            assert(dstCapacity >= cSize);
+            dstCapacity -= cSize;
+            cctx->isFirstBlock = 0;
+            DEBUGLOG(5, "ZSTD_compress_frameChunk: adding a block of size %u",
+                        (unsigned)cSize);
+    }   }
+
+    if (lastFrameChunk && (op>ostart)) cctx->stage = ZSTDcs_ending;
+    return (size_t)(op-ostart);
+}
+
+
+static size_t ZSTD_writeFrameHeader(void* dst, size_t dstCapacity,
+                                    const ZSTD_CCtx_params* params, U64 pledgedSrcSize, U32 dictID)
+{   BYTE* const op = (BYTE*)dst;
+    U32   const dictIDSizeCodeLength = (dictID>0) + (dictID>=256) + (dictID>=65536);   /* 0-3 */
+    U32   const dictIDSizeCode = params->fParams.noDictIDFlag ? 0 : dictIDSizeCodeLength;   /* 0-3 */
+    U32   const checksumFlag = params->fParams.checksumFlag>0;
+    U32   const windowSize = (U32)1 << params->cParams.windowLog;
+    U32   const singleSegment = params->fParams.contentSizeFlag && (windowSize >= pledgedSrcSize);
+    BYTE  const windowLogByte = (BYTE)((params->cParams.windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN) << 3);
+    U32   const fcsCode = params->fParams.contentSizeFlag ?
+                     (pledgedSrcSize>=256) + (pledgedSrcSize>=65536+256) + (pledgedSrcSize>=0xFFFFFFFFU) : 0;  /* 0-3 */
+    BYTE  const frameHeaderDescriptionByte = (BYTE)(dictIDSizeCode + (checksumFlag<<2) + (singleSegment<<5) + (fcsCode<<6) );
+    size_t pos=0;
+
+    assert(!(params->fParams.contentSizeFlag && pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN));
+    RETURN_ERROR_IF(dstCapacity < ZSTD_FRAMEHEADERSIZE_MAX, dstSize_tooSmall,
+                    "dst buf is too small to fit worst-case frame header size.");
+    DEBUGLOG(4, "ZSTD_writeFrameHeader : dictIDFlag : %u ; dictID : %u ; dictIDSizeCode : %u",
+                !params->fParams.noDictIDFlag, (unsigned)dictID, (unsigned)dictIDSizeCode);
+    if (params->format == ZSTD_f_zstd1) {
+        MEM_writeLE32(dst, ZSTD_MAGICNUMBER);
+        pos = 4;
+    }
+    op[pos++] = frameHeaderDescriptionByte;
+    if (!singleSegment) op[pos++] = windowLogByte;
+    switch(dictIDSizeCode)
+    {
+        default:
+            assert(0); /* impossible */
+            ZSTD_FALLTHROUGH;
+        case 0 : break;
+        case 1 : op[pos] = (BYTE)(dictID); pos++; break;
+        case 2 : MEM_writeLE16(op+pos, (U16)dictID); pos+=2; break;
+        case 3 : MEM_writeLE32(op+pos, dictID); pos+=4; break;
+    }
+    switch(fcsCode)
+    {
+        default:
+            assert(0); /* impossible */
+            ZSTD_FALLTHROUGH;
+        case 0 : if (singleSegment) op[pos++] = (BYTE)(pledgedSrcSize); break;
+        case 1 : MEM_writeLE16(op+pos, (U16)(pledgedSrcSize-256)); pos+=2; break;
+        case 2 : MEM_writeLE32(op+pos, (U32)(pledgedSrcSize)); pos+=4; break;
+        case 3 : MEM_writeLE64(op+pos, (U64)(pledgedSrcSize)); pos+=8; break;
+    }
+    return pos;
+}
+
+/* ZSTD_writeSkippableFrame_advanced() :
+ * Writes out a skippable frame with the specified magic number variant (16 are supported),
+ * from ZSTD_MAGIC_SKIPPABLE_START to ZSTD_MAGIC_SKIPPABLE_START+15, and the desired source data.
+ *
+ * Returns the total number of bytes written, or a ZSTD error code.
+ */
+size_t ZSTD_writeSkippableFrame(void* dst, size_t dstCapacity,
+                                const void* src, size_t srcSize, unsigned magicVariant) {
+    BYTE* op = (BYTE*)dst;
+    RETURN_ERROR_IF(dstCapacity < srcSize + ZSTD_SKIPPABLEHEADERSIZE /* Skippable frame overhead */,
+                    dstSize_tooSmall, "Not enough room for skippable frame");
+    RETURN_ERROR_IF(srcSize > (unsigned)0xFFFFFFFF, srcSize_wrong, "Src size too large for skippable frame");
+    RETURN_ERROR_IF(magicVariant > 15, parameter_outOfBound, "Skippable frame magic number variant not supported");
+
+    MEM_writeLE32(op, (U32)(ZSTD_MAGIC_SKIPPABLE_START + magicVariant));
+    MEM_writeLE32(op+4, (U32)srcSize);
+    ZSTD_memcpy(op+8, src, srcSize);
+    return srcSize + ZSTD_SKIPPABLEHEADERSIZE;
+}
+
+/* ZSTD_writeLastEmptyBlock() :
+ * output an empty Block with end-of-frame mark to complete a frame
+ * @return : size of data written into `dst` (== ZSTD_blockHeaderSize (defined in zstd_internal.h))
+ *           or an error code if `dstCapacity` is too small (<ZSTD_blockHeaderSize)
+ */
+size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity)
+{
+    RETURN_ERROR_IF(dstCapacity < ZSTD_blockHeaderSize, dstSize_tooSmall,
+                    "dst buf is too small to write frame trailer empty block.");
+    {   U32 const cBlockHeader24 = 1 /*lastBlock*/ + (((U32)bt_raw)<<1);  /* 0 size */
+        MEM_writeLE24(dst, cBlockHeader24);
+        return ZSTD_blockHeaderSize;
+    }
+}
+
+size_t ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq)
+{
+    RETURN_ERROR_IF(cctx->stage != ZSTDcs_init, stage_wrong,
+                    "wrong cctx stage");
+    RETURN_ERROR_IF(cctx->appliedParams.ldmParams.enableLdm,
+                    parameter_unsupported,
+                    "incompatible with ldm");
+    cctx->externSeqStore.seq = seq;
+    cctx->externSeqStore.size = nbSeq;
+    cctx->externSeqStore.capacity = nbSeq;
+    cctx->externSeqStore.pos = 0;
+    cctx->externSeqStore.posInSequence = 0;
+    return 0;
+}
+
+
+static size_t ZSTD_compressContinue_internal (ZSTD_CCtx* cctx,
+                              void* dst, size_t dstCapacity,
+                        const void* src, size_t srcSize,
+                               U32 frame, U32 lastFrameChunk)
+{
+    ZSTD_matchState_t* const ms = &cctx->blockState.matchState;
+    size_t fhSize = 0;
+
+    DEBUGLOG(5, "ZSTD_compressContinue_internal, stage: %u, srcSize: %u",
+                cctx->stage, (unsigned)srcSize);
+    RETURN_ERROR_IF(cctx->stage==ZSTDcs_created, stage_wrong,
+                    "missing init (ZSTD_compressBegin)");
+
+    if (frame && (cctx->stage==ZSTDcs_init)) {
+        fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, &cctx->appliedParams,
+                                       cctx->pledgedSrcSizePlusOne-1, cctx->dictID);
+        FORWARD_IF_ERROR(fhSize, "ZSTD_writeFrameHeader failed");
+        assert(fhSize <= dstCapacity);
+        dstCapacity -= fhSize;
+        dst = (char*)dst + fhSize;
+        cctx->stage = ZSTDcs_ongoing;
+    }
+
+    if (!srcSize) return fhSize;  /* do not generate an empty block if no input */
+
+    if (!ZSTD_window_update(&ms->window, src, srcSize)) {
+        ms->nextToUpdate = ms->window.dictLimit;
+    }
+    if (cctx->appliedParams.ldmParams.enableLdm) {
+        ZSTD_window_update(&cctx->ldmState.window, src, srcSize);
+    }
+
+    if (!frame) {
+        /* overflow check and correction for block mode */
+        ZSTD_overflowCorrectIfNeeded(
+            ms, &cctx->workspace, &cctx->appliedParams,
+            src, (BYTE const*)src + srcSize);
+    }
+
+    DEBUGLOG(5, "ZSTD_compressContinue_internal (blockSize=%u)", (unsigned)cctx->blockSize);
+    {   size_t const cSize = frame ?
+                             ZSTD_compress_frameChunk (cctx, dst, dstCapacity, src, srcSize, lastFrameChunk) :
+                             ZSTD_compressBlock_internal (cctx, dst, dstCapacity, src, srcSize, 0 /* frame */);
+        FORWARD_IF_ERROR(cSize, "%s", frame ? "ZSTD_compress_frameChunk failed" : "ZSTD_compressBlock_internal failed");
+        cctx->consumedSrcSize += srcSize;
+        cctx->producedCSize += (cSize + fhSize);
+        assert(!(cctx->appliedParams.fParams.contentSizeFlag && cctx->pledgedSrcSizePlusOne == 0));
+        if (cctx->pledgedSrcSizePlusOne != 0) {  /* control src size */
+            ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN == (unsigned long long)-1);
+            RETURN_ERROR_IF(
+                cctx->consumedSrcSize+1 > cctx->pledgedSrcSizePlusOne,
+                srcSize_wrong,
+                "error : pledgedSrcSize = %u, while realSrcSize >= %u",
+                (unsigned)cctx->pledgedSrcSizePlusOne-1,
+                (unsigned)cctx->consumedSrcSize);
+        }
+        return cSize + fhSize;
+    }
+}
+
+size_t ZSTD_compressContinue (ZSTD_CCtx* cctx,
+                              void* dst, size_t dstCapacity,
+                        const void* src, size_t srcSize)
+{
+    DEBUGLOG(5, "ZSTD_compressContinue (srcSize=%u)", (unsigned)srcSize);
+    return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 1 /* frame mode */, 0 /* last chunk */);
+}
+
+
+size_t ZSTD_getBlockSize(const ZSTD_CCtx* cctx)
+{
+    ZSTD_compressionParameters const cParams = cctx->appliedParams.cParams;
+    assert(!ZSTD_checkCParams(cParams));
+    return MIN (ZSTD_BLOCKSIZE_MAX, (U32)1 << cParams.windowLog);
+}
+
+size_t ZSTD_compressBlock(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    DEBUGLOG(5, "ZSTD_compressBlock: srcSize = %u", (unsigned)srcSize);
+    { size_t const blockSizeMax = ZSTD_getBlockSize(cctx);
+      RETURN_ERROR_IF(srcSize > blockSizeMax, srcSize_wrong, "input is larger than a block"); }
+
+    return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 0 /* frame mode */, 0 /* last chunk */);
+}
+
+/*! ZSTD_loadDictionaryContent() :
+ *  @return : 0, or an error code
+ */
+static size_t ZSTD_loadDictionaryContent(ZSTD_matchState_t* ms,
+                                         ldmState_t* ls,
+                                         ZSTD_cwksp* ws,
+                                         ZSTD_CCtx_params const* params,
+                                         const void* src, size_t srcSize,
+                                         ZSTD_dictTableLoadMethod_e dtlm)
+{
+    const BYTE* ip = (const BYTE*) src;
+    const BYTE* const iend = ip + srcSize;
+
+    ZSTD_window_update(&ms->window, src, srcSize);
+    ms->loadedDictEnd = params->forceWindow ? 0 : (U32)(iend - ms->window.base);
+
+    if (params->ldmParams.enableLdm && ls != NULL) {
+        ZSTD_window_update(&ls->window, src, srcSize);
+        ls->loadedDictEnd = params->forceWindow ? 0 : (U32)(iend - ls->window.base);
+    }
+
+    /* Assert that we the ms params match the params we're being given */
+    ZSTD_assertEqualCParams(params->cParams, ms->cParams);
+
+    if (srcSize <= HASH_READ_SIZE) return 0;
+
+    while (iend - ip > HASH_READ_SIZE) {
+        size_t const remaining = (size_t)(iend - ip);
+        size_t const chunk = MIN(remaining, ZSTD_CHUNKSIZE_MAX);
+        const BYTE* const ichunk = ip + chunk;
+
+        ZSTD_overflowCorrectIfNeeded(ms, ws, params, ip, ichunk);
+
+        if (params->ldmParams.enableLdm && ls != NULL)
+            ZSTD_ldm_fillHashTable(ls, (const BYTE*)src, (const BYTE*)src + srcSize, &params->ldmParams);
+
+        switch(params->cParams.strategy)
+        {
+        case ZSTD_fast:
+            ZSTD_fillHashTable(ms, ichunk, dtlm);
+            break;
+        case ZSTD_dfast:
+            ZSTD_fillDoubleHashTable(ms, ichunk, dtlm);
+            break;
+
+        case ZSTD_greedy:
+        case ZSTD_lazy:
+        case ZSTD_lazy2:
+            if (chunk >= HASH_READ_SIZE && ms->dedicatedDictSearch) {
+                assert(chunk == remaining); /* must load everything in one go */
+                ZSTD_dedicatedDictSearch_lazy_loadDictionary(ms, ichunk-HASH_READ_SIZE);
+            } else if (chunk >= HASH_READ_SIZE) {
+                ZSTD_insertAndFindFirstIndex(ms, ichunk-HASH_READ_SIZE);
+            }
+            break;
+
+        case ZSTD_btlazy2:   /* we want the dictionary table fully sorted */
+        case ZSTD_btopt:
+        case ZSTD_btultra:
+        case ZSTD_btultra2:
+            if (chunk >= HASH_READ_SIZE)
+                ZSTD_updateTree(ms, ichunk-HASH_READ_SIZE, ichunk);
+            break;
+
+        default:
+            assert(0);  /* not possible : not a valid strategy id */
+        }
+
+        ip = ichunk;
+    }
+
+    ms->nextToUpdate = (U32)(iend - ms->window.base);
+    return 0;
+}
+
+
+/* Dictionaries that assign zero probability to symbols that show up causes problems
+ * when FSE encoding. Mark dictionaries with zero probability symbols as FSE_repeat_check
+ * and only dictionaries with 100% valid symbols can be assumed valid.
+ */
+static FSE_repeat ZSTD_dictNCountRepeat(short* normalizedCounter, unsigned dictMaxSymbolValue, unsigned maxSymbolValue)
+{
+    U32 s;
+    if (dictMaxSymbolValue < maxSymbolValue) {
+        return FSE_repeat_check;
+    }
+    for (s = 0; s <= maxSymbolValue; ++s) {
+        if (normalizedCounter[s] == 0) {
+            return FSE_repeat_check;
+        }
+    }
+    return FSE_repeat_valid;
+}
+
+size_t ZSTD_loadCEntropy(ZSTD_compressedBlockState_t* bs, void* workspace,
+                         const void* const dict, size_t dictSize)
+{
+    short offcodeNCount[MaxOff+1];
+    unsigned offcodeMaxValue = MaxOff;
+    const BYTE* dictPtr = (const BYTE*)dict;    /* skip magic num and dict ID */
+    const BYTE* const dictEnd = dictPtr + dictSize;
+    dictPtr += 8;
+    bs->entropy.huf.repeatMode = HUF_repeat_check;
+
+    {   unsigned maxSymbolValue = 255;
+        unsigned hasZeroWeights = 1;
+        size_t const hufHeaderSize = HUF_readCTable((HUF_CElt*)bs->entropy.huf.CTable, &maxSymbolValue, dictPtr,
+            dictEnd-dictPtr, &hasZeroWeights);
+
+        /* We only set the loaded table as valid if it contains all non-zero
+         * weights. Otherwise, we set it to check */
+        if (!hasZeroWeights)
+            bs->entropy.huf.repeatMode = HUF_repeat_valid;
+
+        RETURN_ERROR_IF(HUF_isError(hufHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(maxSymbolValue < 255, dictionary_corrupted, "");
+        dictPtr += hufHeaderSize;
+    }
+
+    {   unsigned offcodeLog;
+        size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, dictEnd-dictPtr);
+        RETURN_ERROR_IF(FSE_isError(offcodeHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(offcodeLog > OffFSELog, dictionary_corrupted, "");
+        /* fill all offset symbols to avoid garbage at end of table */
+        RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
+                bs->entropy.fse.offcodeCTable,
+                offcodeNCount, MaxOff, offcodeLog,
+                workspace, HUF_WORKSPACE_SIZE)),
+            dictionary_corrupted, "");
+        /* Defer checking offcodeMaxValue because we need to know the size of the dictionary content */
+        dictPtr += offcodeHeaderSize;
+    }
+
+    {   short matchlengthNCount[MaxML+1];
+        unsigned matchlengthMaxValue = MaxML, matchlengthLog;
+        size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd-dictPtr);
+        RETURN_ERROR_IF(FSE_isError(matchlengthHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(matchlengthLog > MLFSELog, dictionary_corrupted, "");
+        RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
+                bs->entropy.fse.matchlengthCTable,
+                matchlengthNCount, matchlengthMaxValue, matchlengthLog,
+                workspace, HUF_WORKSPACE_SIZE)),
+            dictionary_corrupted, "");
+        bs->entropy.fse.matchlength_repeatMode = ZSTD_dictNCountRepeat(matchlengthNCount, matchlengthMaxValue, MaxML);
+        dictPtr += matchlengthHeaderSize;
+    }
+
+    {   short litlengthNCount[MaxLL+1];
+        unsigned litlengthMaxValue = MaxLL, litlengthLog;
+        size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd-dictPtr);
+        RETURN_ERROR_IF(FSE_isError(litlengthHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(litlengthLog > LLFSELog, dictionary_corrupted, "");
+        RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
+                bs->entropy.fse.litlengthCTable,
+                litlengthNCount, litlengthMaxValue, litlengthLog,
+                workspace, HUF_WORKSPACE_SIZE)),
+            dictionary_corrupted, "");
+        bs->entropy.fse.litlength_repeatMode = ZSTD_dictNCountRepeat(litlengthNCount, litlengthMaxValue, MaxLL);
+        dictPtr += litlengthHeaderSize;
+    }
+
+    RETURN_ERROR_IF(dictPtr+12 > dictEnd, dictionary_corrupted, "");
+    bs->rep[0] = MEM_readLE32(dictPtr+0);
+    bs->rep[1] = MEM_readLE32(dictPtr+4);
+    bs->rep[2] = MEM_readLE32(dictPtr+8);
+    dictPtr += 12;
+
+    {   size_t const dictContentSize = (size_t)(dictEnd - dictPtr);
+        U32 offcodeMax = MaxOff;
+        if (dictContentSize <= ((U32)-1) - 128 KB) {
+            U32 const maxOffset = (U32)dictContentSize + 128 KB; /* The maximum offset that must be supported */
+            offcodeMax = ZSTD_highbit32(maxOffset); /* Calculate minimum offset code required to represent maxOffset */
+        }
+        /* All offset values <= dictContentSize + 128 KB must be representable for a valid table */
+        bs->entropy.fse.offcode_repeatMode = ZSTD_dictNCountRepeat(offcodeNCount, offcodeMaxValue, MIN(offcodeMax, MaxOff));
+
+        /* All repCodes must be <= dictContentSize and != 0 */
+        {   U32 u;
+            for (u=0; u<3; u++) {
+                RETURN_ERROR_IF(bs->rep[u] == 0, dictionary_corrupted, "");
+                RETURN_ERROR_IF(bs->rep[u] > dictContentSize, dictionary_corrupted, "");
+    }   }   }
+
+    return dictPtr - (const BYTE*)dict;
+}
+
+/* Dictionary format :
+ * See :
+ * https://github.com/facebook/zstd/blob/release/doc/zstd_compression_format.md#dictionary-format
+ */
+/*! ZSTD_loadZstdDictionary() :
+ * @return : dictID, or an error code
+ *  assumptions : magic number supposed already checked
+ *                dictSize supposed >= 8
+ */
+static size_t ZSTD_loadZstdDictionary(ZSTD_compressedBlockState_t* bs,
+                                      ZSTD_matchState_t* ms,
+                                      ZSTD_cwksp* ws,
+                                      ZSTD_CCtx_params const* params,
+                                      const void* dict, size_t dictSize,
+                                      ZSTD_dictTableLoadMethod_e dtlm,
+                                      void* workspace)
+{
+    const BYTE* dictPtr = (const BYTE*)dict;
+    const BYTE* const dictEnd = dictPtr + dictSize;
+    size_t dictID;
+    size_t eSize;
+
+    ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog)));
+    assert(dictSize >= 8);
+    assert(MEM_readLE32(dictPtr) == ZSTD_MAGIC_DICTIONARY);
+
+    dictID = params->fParams.noDictIDFlag ? 0 :  MEM_readLE32(dictPtr + 4 /* skip magic number */ );
+    eSize = ZSTD_loadCEntropy(bs, workspace, dict, dictSize);
+    FORWARD_IF_ERROR(eSize, "ZSTD_loadCEntropy failed");
+    dictPtr += eSize;
+
+    {
+        size_t const dictContentSize = (size_t)(dictEnd - dictPtr);
+        FORWARD_IF_ERROR(ZSTD_loadDictionaryContent(
+            ms, NULL, ws, params, dictPtr, dictContentSize, dtlm), "");
+    }
+    return dictID;
+}
+
+/* ZSTD_compress_insertDictionary() :
+*   @return : dictID, or an error code */
+static size_t
+ZSTD_compress_insertDictionary(ZSTD_compressedBlockState_t* bs,
+                               ZSTD_matchState_t* ms,
+                               ldmState_t* ls,
+                               ZSTD_cwksp* ws,
+                         const ZSTD_CCtx_params* params,
+                         const void* dict, size_t dictSize,
+                               ZSTD_dictContentType_e dictContentType,
+                               ZSTD_dictTableLoadMethod_e dtlm,
+                               void* workspace)
+{
+    DEBUGLOG(4, "ZSTD_compress_insertDictionary (dictSize=%u)", (U32)dictSize);
+    if ((dict==NULL) || (dictSize<8)) {
+        RETURN_ERROR_IF(dictContentType == ZSTD_dct_fullDict, dictionary_wrong, "");
+        return 0;
+    }
+
+    ZSTD_reset_compressedBlockState(bs);
+
+    /* dict restricted modes */
+    if (dictContentType == ZSTD_dct_rawContent)
+        return ZSTD_loadDictionaryContent(ms, ls, ws, params, dict, dictSize, dtlm);
+
+    if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) {
+        if (dictContentType == ZSTD_dct_auto) {
+            DEBUGLOG(4, "raw content dictionary detected");
+            return ZSTD_loadDictionaryContent(
+                ms, ls, ws, params, dict, dictSize, dtlm);
+        }
+        RETURN_ERROR_IF(dictContentType == ZSTD_dct_fullDict, dictionary_wrong, "");
+        assert(0);   /* impossible */
+    }
+
+    /* dict as full zstd dictionary */
+    return ZSTD_loadZstdDictionary(
+        bs, ms, ws, params, dict, dictSize, dtlm, workspace);
+}
+
+#define ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF (128 KB)
+#define ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER (6ULL)
+
+/*! ZSTD_compressBegin_internal() :
+ * @return : 0, or an error code */
+static size_t ZSTD_compressBegin_internal(ZSTD_CCtx* cctx,
+                                    const void* dict, size_t dictSize,
+                                    ZSTD_dictContentType_e dictContentType,
+                                    ZSTD_dictTableLoadMethod_e dtlm,
+                                    const ZSTD_CDict* cdict,
+                                    const ZSTD_CCtx_params* params, U64 pledgedSrcSize,
+                                    ZSTD_buffered_policy_e zbuff)
+{
+    DEBUGLOG(4, "ZSTD_compressBegin_internal: wlog=%u", params->cParams.windowLog);
+    /* params are supposed to be fully validated at this point */
+    assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams)));
+    assert(!((dict) && (cdict)));  /* either dict or cdict, not both */
+    if ( (cdict)
+      && (cdict->dictContentSize > 0)
+      && ( pledgedSrcSize < ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF
+        || pledgedSrcSize < cdict->dictContentSize * ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER
+        || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
+        || cdict->compressionLevel == 0)
+      && (params->attachDictPref != ZSTD_dictForceLoad) ) {
+        return ZSTD_resetCCtx_usingCDict(cctx, cdict, params, pledgedSrcSize, zbuff);
+    }
+
+    FORWARD_IF_ERROR( ZSTD_resetCCtx_internal(cctx, *params, pledgedSrcSize,
+                                     ZSTDcrp_makeClean, zbuff) , "");
+    {   size_t const dictID = cdict ?
+                ZSTD_compress_insertDictionary(
+                        cctx->blockState.prevCBlock, &cctx->blockState.matchState,
+                        &cctx->ldmState, &cctx->workspace, &cctx->appliedParams, cdict->dictContent,
+                        cdict->dictContentSize, cdict->dictContentType, dtlm,
+                        cctx->entropyWorkspace)
+              : ZSTD_compress_insertDictionary(
+                        cctx->blockState.prevCBlock, &cctx->blockState.matchState,
+                        &cctx->ldmState, &cctx->workspace, &cctx->appliedParams, dict, dictSize,
+                        dictContentType, dtlm, cctx->entropyWorkspace);
+        FORWARD_IF_ERROR(dictID, "ZSTD_compress_insertDictionary failed");
+        assert(dictID <= UINT_MAX);
+        cctx->dictID = (U32)dictID;
+        cctx->dictContentSize = cdict ? cdict->dictContentSize : dictSize;
+    }
+    return 0;
+}
+
+size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx,
+                                    const void* dict, size_t dictSize,
+                                    ZSTD_dictContentType_e dictContentType,
+                                    ZSTD_dictTableLoadMethod_e dtlm,
+                                    const ZSTD_CDict* cdict,
+                                    const ZSTD_CCtx_params* params,
+                                    unsigned long long pledgedSrcSize)
+{
+    DEBUGLOG(4, "ZSTD_compressBegin_advanced_internal: wlog=%u", params->cParams.windowLog);
+    /* compression parameters verification and optimization */
+    FORWARD_IF_ERROR( ZSTD_checkCParams(params->cParams) , "");
+    return ZSTD_compressBegin_internal(cctx,
+                                       dict, dictSize, dictContentType, dtlm,
+                                       cdict,
+                                       params, pledgedSrcSize,
+                                       ZSTDb_not_buffered);
+}
+
+/*! ZSTD_compressBegin_advanced() :
+*   @return : 0, or an error code */
+size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx,
+                             const void* dict, size_t dictSize,
+                                   ZSTD_parameters params, unsigned long long pledgedSrcSize)
+{
+    ZSTD_CCtx_params cctxParams;
+    ZSTD_CCtxParams_init_internal(&cctxParams, &params, ZSTD_NO_CLEVEL);
+    return ZSTD_compressBegin_advanced_internal(cctx,
+                                            dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast,
+                                            NULL /*cdict*/,
+                                            &cctxParams, pledgedSrcSize);
+}
+
+size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel)
+{
+    ZSTD_CCtx_params cctxParams;
+    {
+        ZSTD_parameters const params = ZSTD_getParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize, ZSTD_cpm_noAttachDict);
+        ZSTD_CCtxParams_init_internal(&cctxParams, &params, (compressionLevel == 0) ? ZSTD_CLEVEL_DEFAULT : compressionLevel);
+    }
+    DEBUGLOG(4, "ZSTD_compressBegin_usingDict (dictSize=%u)", (unsigned)dictSize);
+    return ZSTD_compressBegin_internal(cctx, dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, NULL,
+                                       &cctxParams, ZSTD_CONTENTSIZE_UNKNOWN, ZSTDb_not_buffered);
+}
+
+size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel)
+{
+    return ZSTD_compressBegin_usingDict(cctx, NULL, 0, compressionLevel);
+}
+
+
+/*! ZSTD_writeEpilogue() :
+*   Ends a frame.
+*   @return : nb of bytes written into dst (or an error code) */
+static size_t ZSTD_writeEpilogue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity)
+{
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* op = ostart;
+    size_t fhSize = 0;
+
+    DEBUGLOG(4, "ZSTD_writeEpilogue");
+    RETURN_ERROR_IF(cctx->stage == ZSTDcs_created, stage_wrong, "init missing");
+
+    /* special case : empty frame */
+    if (cctx->stage == ZSTDcs_init) {
+        fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, &cctx->appliedParams, 0, 0);
+        FORWARD_IF_ERROR(fhSize, "ZSTD_writeFrameHeader failed");
+        dstCapacity -= fhSize;
+        op += fhSize;
+        cctx->stage = ZSTDcs_ongoing;
+    }
+
+    if (cctx->stage != ZSTDcs_ending) {
+        /* write one last empty block, make it the "last" block */
+        U32 const cBlockHeader24 = 1 /* last block */ + (((U32)bt_raw)<<1) + 0;
+        RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for epilogue");
+        MEM_writeLE32(op, cBlockHeader24);
+        op += ZSTD_blockHeaderSize;
+        dstCapacity -= ZSTD_blockHeaderSize;
+    }
+
+    if (cctx->appliedParams.fParams.checksumFlag) {
+        U32 const checksum = (U32) xxh64_digest(&cctx->xxhState);
+        RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for checksum");
+        DEBUGLOG(4, "ZSTD_writeEpilogue: write checksum : %08X", (unsigned)checksum);
+        MEM_writeLE32(op, checksum);
+        op += 4;
+    }
+
+    cctx->stage = ZSTDcs_created;  /* return to "created but no init" status */
+    return op-ostart;
+}
+
+void ZSTD_CCtx_trace(ZSTD_CCtx* cctx, size_t extraCSize)
+{
+    (void)cctx;
+    (void)extraCSize;
+}
+
+size_t ZSTD_compressEnd (ZSTD_CCtx* cctx,
+                         void* dst, size_t dstCapacity,
+                   const void* src, size_t srcSize)
+{
+    size_t endResult;
+    size_t const cSize = ZSTD_compressContinue_internal(cctx,
+                                dst, dstCapacity, src, srcSize,
+                                1 /* frame mode */, 1 /* last chunk */);
+    FORWARD_IF_ERROR(cSize, "ZSTD_compressContinue_internal failed");
+    endResult = ZSTD_writeEpilogue(cctx, (char*)dst + cSize, dstCapacity-cSize);
+    FORWARD_IF_ERROR(endResult, "ZSTD_writeEpilogue failed");
+    assert(!(cctx->appliedParams.fParams.contentSizeFlag && cctx->pledgedSrcSizePlusOne == 0));
+    if (cctx->pledgedSrcSizePlusOne != 0) {  /* control src size */
+        ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN == (unsigned long long)-1);
+        DEBUGLOG(4, "end of frame : controlling src size");
+        RETURN_ERROR_IF(
+            cctx->pledgedSrcSizePlusOne != cctx->consumedSrcSize+1,
+            srcSize_wrong,
+             "error : pledgedSrcSize = %u, while realSrcSize = %u",
+            (unsigned)cctx->pledgedSrcSizePlusOne-1,
+            (unsigned)cctx->consumedSrcSize);
+    }
+    ZSTD_CCtx_trace(cctx, endResult);
+    return cSize + endResult;
+}
+
+size_t ZSTD_compress_advanced (ZSTD_CCtx* cctx,
+                               void* dst, size_t dstCapacity,
+                         const void* src, size_t srcSize,
+                         const void* dict,size_t dictSize,
+                               ZSTD_parameters params)
+{
+    ZSTD_CCtx_params cctxParams;
+    DEBUGLOG(4, "ZSTD_compress_advanced");
+    FORWARD_IF_ERROR(ZSTD_checkCParams(params.cParams), "");
+    ZSTD_CCtxParams_init_internal(&cctxParams, &params, ZSTD_NO_CLEVEL);
+    return ZSTD_compress_advanced_internal(cctx,
+                                           dst, dstCapacity,
+                                           src, srcSize,
+                                           dict, dictSize,
+                                           &cctxParams);
+}
+
+/* Internal */
+size_t ZSTD_compress_advanced_internal(
+        ZSTD_CCtx* cctx,
+        void* dst, size_t dstCapacity,
+        const void* src, size_t srcSize,
+        const void* dict,size_t dictSize,
+        const ZSTD_CCtx_params* params)
+{
+    DEBUGLOG(4, "ZSTD_compress_advanced_internal (srcSize:%u)", (unsigned)srcSize);
+    FORWARD_IF_ERROR( ZSTD_compressBegin_internal(cctx,
+                         dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, NULL,
+                         params, srcSize, ZSTDb_not_buffered) , "");
+    return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
+}
+
+size_t ZSTD_compress_usingDict(ZSTD_CCtx* cctx,
+                               void* dst, size_t dstCapacity,
+                         const void* src, size_t srcSize,
+                         const void* dict, size_t dictSize,
+                               int compressionLevel)
+{
+    ZSTD_CCtx_params cctxParams;
+    {
+        ZSTD_parameters const params = ZSTD_getParams_internal(compressionLevel, srcSize, dict ? dictSize : 0, ZSTD_cpm_noAttachDict);
+        assert(params.fParams.contentSizeFlag == 1);
+        ZSTD_CCtxParams_init_internal(&cctxParams, &params, (compressionLevel == 0) ? ZSTD_CLEVEL_DEFAULT: compressionLevel);
+    }
+    DEBUGLOG(4, "ZSTD_compress_usingDict (srcSize=%u)", (unsigned)srcSize);
+    return ZSTD_compress_advanced_internal(cctx, dst, dstCapacity, src, srcSize, dict, dictSize, &cctxParams);
+}
+
+size_t ZSTD_compressCCtx(ZSTD_CCtx* cctx,
+                         void* dst, size_t dstCapacity,
+                   const void* src, size_t srcSize,
+                         int compressionLevel)
+{
+    DEBUGLOG(4, "ZSTD_compressCCtx (srcSize=%u)", (unsigned)srcSize);
+    assert(cctx != NULL);
+    return ZSTD_compress_usingDict(cctx, dst, dstCapacity, src, srcSize, NULL, 0, compressionLevel);
+}
+
+size_t ZSTD_compress(void* dst, size_t dstCapacity,
+               const void* src, size_t srcSize,
+                     int compressionLevel)
+{
+    size_t result;
+    ZSTD_CCtx* cctx = ZSTD_createCCtx();
+    RETURN_ERROR_IF(!cctx, memory_allocation, "ZSTD_createCCtx failed");
+    result = ZSTD_compressCCtx(cctx, dst, dstCapacity, src, srcSize, compressionLevel);
+    ZSTD_freeCCtx(cctx);
+    return result;
+}
+
+
+/* =====  Dictionary API  ===== */
+
+/*! ZSTD_estimateCDictSize_advanced() :
+ *  Estimate amount of memory that will be needed to create a dictionary with following arguments */
+size_t ZSTD_estimateCDictSize_advanced(
+        size_t dictSize, ZSTD_compressionParameters cParams,
+        ZSTD_dictLoadMethod_e dictLoadMethod)
+{
+    DEBUGLOG(5, "sizeof(ZSTD_CDict) : %u", (unsigned)sizeof(ZSTD_CDict));
+    return ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict))
+         + ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE)
+         + ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0)
+         + (dictLoadMethod == ZSTD_dlm_byRef ? 0
+            : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void *))));
+}
+
+size_t ZSTD_estimateCDictSize(size_t dictSize, int compressionLevel)
+{
+    ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize, ZSTD_cpm_createCDict);
+    return ZSTD_estimateCDictSize_advanced(dictSize, cParams, ZSTD_dlm_byCopy);
+}
+
+size_t ZSTD_sizeof_CDict(const ZSTD_CDict* cdict)
+{
+    if (cdict==NULL) return 0;   /* support sizeof on NULL */
+    DEBUGLOG(5, "sizeof(*cdict) : %u", (unsigned)sizeof(*cdict));
+    /* cdict may be in the workspace */
+    return (cdict->workspace.workspace == cdict ? 0 : sizeof(*cdict))
+        + ZSTD_cwksp_sizeof(&cdict->workspace);
+}
+
+static size_t ZSTD_initCDict_internal(
+                    ZSTD_CDict* cdict,
+              const void* dictBuffer, size_t dictSize,
+                    ZSTD_dictLoadMethod_e dictLoadMethod,
+                    ZSTD_dictContentType_e dictContentType,
+                    ZSTD_CCtx_params params)
+{
+    DEBUGLOG(3, "ZSTD_initCDict_internal (dictContentType:%u)", (unsigned)dictContentType);
+    assert(!ZSTD_checkCParams(params.cParams));
+    cdict->matchState.cParams = params.cParams;
+    cdict->matchState.dedicatedDictSearch = params.enableDedicatedDictSearch;
+    if (cdict->matchState.dedicatedDictSearch && dictSize > ZSTD_CHUNKSIZE_MAX) {
+        cdict->matchState.dedicatedDictSearch = 0;
+    }
+    if ((dictLoadMethod == ZSTD_dlm_byRef) || (!dictBuffer) || (!dictSize)) {
+        cdict->dictContent = dictBuffer;
+    } else {
+         void *internalBuffer = ZSTD_cwksp_reserve_object(&cdict->workspace, ZSTD_cwksp_align(dictSize, sizeof(void*)));
+        RETURN_ERROR_IF(!internalBuffer, memory_allocation, "NULL pointer!");
+        cdict->dictContent = internalBuffer;
+        ZSTD_memcpy(internalBuffer, dictBuffer, dictSize);
+    }
+    cdict->dictContentSize = dictSize;
+    cdict->dictContentType = dictContentType;
+
+    cdict->entropyWorkspace = (U32*)ZSTD_cwksp_reserve_object(&cdict->workspace, HUF_WORKSPACE_SIZE);
+
+
+    /* Reset the state to no dictionary */
+    ZSTD_reset_compressedBlockState(&cdict->cBlockState);
+    FORWARD_IF_ERROR(ZSTD_reset_matchState(
+        &cdict->matchState,
+        &cdict->workspace,
+        &params.cParams,
+        ZSTDcrp_makeClean,
+        ZSTDirp_reset,
+        ZSTD_resetTarget_CDict), "");
+    /* (Maybe) load the dictionary
+     * Skips loading the dictionary if it is < 8 bytes.
+     */
+    {   params.compressionLevel = ZSTD_CLEVEL_DEFAULT;
+        params.fParams.contentSizeFlag = 1;
+        {   size_t const dictID = ZSTD_compress_insertDictionary(
+                    &cdict->cBlockState, &cdict->matchState, NULL, &cdict->workspace,
+                    &params, cdict->dictContent, cdict->dictContentSize,
+                    dictContentType, ZSTD_dtlm_full, cdict->entropyWorkspace);
+            FORWARD_IF_ERROR(dictID, "ZSTD_compress_insertDictionary failed");
+            assert(dictID <= (size_t)(U32)-1);
+            cdict->dictID = (U32)dictID;
+        }
+    }
+
+    return 0;
+}
+
+static ZSTD_CDict* ZSTD_createCDict_advanced_internal(size_t dictSize,
+                                      ZSTD_dictLoadMethod_e dictLoadMethod,
+                                      ZSTD_compressionParameters cParams, ZSTD_customMem customMem)
+{
+    if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
+
+    {   size_t const workspaceSize =
+            ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict)) +
+            ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE) +
+            ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0) +
+            (dictLoadMethod == ZSTD_dlm_byRef ? 0
+             : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void*))));
+        void* const workspace = ZSTD_customMalloc(workspaceSize, customMem);
+        ZSTD_cwksp ws;
+        ZSTD_CDict* cdict;
+
+        if (!workspace) {
+            ZSTD_customFree(workspace, customMem);
+            return NULL;
+        }
+
+        ZSTD_cwksp_init(&ws, workspace, workspaceSize, ZSTD_cwksp_dynamic_alloc);
+
+        cdict = (ZSTD_CDict*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CDict));
+        assert(cdict != NULL);
+        ZSTD_cwksp_move(&cdict->workspace, &ws);
+        cdict->customMem = customMem;
+        cdict->compressionLevel = ZSTD_NO_CLEVEL; /* signals advanced API usage */
+
+        return cdict;
+    }
+}
+
+ZSTD_CDict* ZSTD_createCDict_advanced(const void* dictBuffer, size_t dictSize,
+                                      ZSTD_dictLoadMethod_e dictLoadMethod,
+                                      ZSTD_dictContentType_e dictContentType,
+                                      ZSTD_compressionParameters cParams,
+                                      ZSTD_customMem customMem)
+{
+    ZSTD_CCtx_params cctxParams;
+    ZSTD_memset(&cctxParams, 0, sizeof(cctxParams));
+    ZSTD_CCtxParams_init(&cctxParams, 0);
+    cctxParams.cParams = cParams;
+    cctxParams.customMem = customMem;
+    return ZSTD_createCDict_advanced2(
+        dictBuffer, dictSize,
+        dictLoadMethod, dictContentType,
+        &cctxParams, customMem);
+}
+
+ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_advanced2(
+        const void* dict, size_t dictSize,
+        ZSTD_dictLoadMethod_e dictLoadMethod,
+        ZSTD_dictContentType_e dictContentType,
+        const ZSTD_CCtx_params* originalCctxParams,
+        ZSTD_customMem customMem)
+{
+    ZSTD_CCtx_params cctxParams = *originalCctxParams;
+    ZSTD_compressionParameters cParams;
+    ZSTD_CDict* cdict;
+
+    DEBUGLOG(3, "ZSTD_createCDict_advanced2, mode %u", (unsigned)dictContentType);
+    if (!customMem.customAlloc ^ !customMem.customFree) return NULL;
+
+    if (cctxParams.enableDedicatedDictSearch) {
+        cParams = ZSTD_dedicatedDictSearch_getCParams(
+            cctxParams.compressionLevel, dictSize);
+        ZSTD_overrideCParams(&cParams, &cctxParams.cParams);
+    } else {
+        cParams = ZSTD_getCParamsFromCCtxParams(
+            &cctxParams, ZSTD_CONTENTSIZE_UNKNOWN, dictSize, ZSTD_cpm_createCDict);
+    }
+
+    if (!ZSTD_dedicatedDictSearch_isSupported(&cParams)) {
+        /* Fall back to non-DDSS params */
+        cctxParams.enableDedicatedDictSearch = 0;
+        cParams = ZSTD_getCParamsFromCCtxParams(
+            &cctxParams, ZSTD_CONTENTSIZE_UNKNOWN, dictSize, ZSTD_cpm_createCDict);
+    }
+
+    cctxParams.cParams = cParams;
+
+    cdict = ZSTD_createCDict_advanced_internal(dictSize,
+                        dictLoadMethod, cctxParams.cParams,
+                        customMem);
+
+    if (ZSTD_isError( ZSTD_initCDict_internal(cdict,
+                                    dict, dictSize,
+                                    dictLoadMethod, dictContentType,
+                                    cctxParams) )) {
+        ZSTD_freeCDict(cdict);
+        return NULL;
+    }
+
+    return cdict;
+}
+
+ZSTD_CDict* ZSTD_createCDict(const void* dict, size_t dictSize, int compressionLevel)
+{
+    ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize, ZSTD_cpm_createCDict);
+    ZSTD_CDict* const cdict = ZSTD_createCDict_advanced(dict, dictSize,
+                                                  ZSTD_dlm_byCopy, ZSTD_dct_auto,
+                                                  cParams, ZSTD_defaultCMem);
+    if (cdict)
+        cdict->compressionLevel = (compressionLevel == 0) ? ZSTD_CLEVEL_DEFAULT : compressionLevel;
+    return cdict;
+}
+
+ZSTD_CDict* ZSTD_createCDict_byReference(const void* dict, size_t dictSize, int compressionLevel)
+{
+    ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize, ZSTD_cpm_createCDict);
+    ZSTD_CDict* const cdict = ZSTD_createCDict_advanced(dict, dictSize,
+                                     ZSTD_dlm_byRef, ZSTD_dct_auto,
+                                     cParams, ZSTD_defaultCMem);
+    if (cdict)
+        cdict->compressionLevel = (compressionLevel == 0) ? ZSTD_CLEVEL_DEFAULT : compressionLevel;
+    return cdict;
+}
+
+size_t ZSTD_freeCDict(ZSTD_CDict* cdict)
+{
+    if (cdict==NULL) return 0;   /* support free on NULL */
+    {   ZSTD_customMem const cMem = cdict->customMem;
+        int cdictInWorkspace = ZSTD_cwksp_owns_buffer(&cdict->workspace, cdict);
+        ZSTD_cwksp_free(&cdict->workspace, cMem);
+        if (!cdictInWorkspace) {
+            ZSTD_customFree(cdict, cMem);
+        }
+        return 0;
+    }
+}
+
+/*! ZSTD_initStaticCDict_advanced() :
+ *  Generate a digested dictionary in provided memory area.
+ *  workspace: The memory area to emplace the dictionary into.
+ *             Provided pointer must 8-bytes aligned.
+ *             It must outlive dictionary usage.
+ *  workspaceSize: Use ZSTD_estimateCDictSize()
+ *                 to determine how large workspace must be.
+ *  cParams : use ZSTD_getCParams() to transform a compression level
+ *            into its relevants cParams.
+ * @return : pointer to ZSTD_CDict*, or NULL if error (size too small)
+ *  Note : there is no corresponding "free" function.
+ *         Since workspace was allocated externally, it must be freed externally.
+ */
+const ZSTD_CDict* ZSTD_initStaticCDict(
+                                 void* workspace, size_t workspaceSize,
+                           const void* dict, size_t dictSize,
+                                 ZSTD_dictLoadMethod_e dictLoadMethod,
+                                 ZSTD_dictContentType_e dictContentType,
+                                 ZSTD_compressionParameters cParams)
+{
+    size_t const matchStateSize = ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0);
+    size_t const neededSize = ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict))
+                            + (dictLoadMethod == ZSTD_dlm_byRef ? 0
+                               : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void*))))
+                            + ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE)
+                            + matchStateSize;
+    ZSTD_CDict* cdict;
+    ZSTD_CCtx_params params;
+
+    if ((size_t)workspace & 7) return NULL;  /* 8-aligned */
+
+    {
+        ZSTD_cwksp ws;
+        ZSTD_cwksp_init(&ws, workspace, workspaceSize, ZSTD_cwksp_static_alloc);
+        cdict = (ZSTD_CDict*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CDict));
+        if (cdict == NULL) return NULL;
+        ZSTD_cwksp_move(&cdict->workspace, &ws);
+    }
+
+    DEBUGLOG(4, "(workspaceSize < neededSize) : (%u < %u) => %u",
+        (unsigned)workspaceSize, (unsigned)neededSize, (unsigned)(workspaceSize < neededSize));
+    if (workspaceSize < neededSize) return NULL;
+
+    ZSTD_CCtxParams_init(&params, 0);
+    params.cParams = cParams;
+
+    if (ZSTD_isError( ZSTD_initCDict_internal(cdict,
+                                              dict, dictSize,
+                                              dictLoadMethod, dictContentType,
+                                              params) ))
+        return NULL;
+
+    return cdict;
+}
+
+ZSTD_compressionParameters ZSTD_getCParamsFromCDict(const ZSTD_CDict* cdict)
+{
+    assert(cdict != NULL);
+    return cdict->matchState.cParams;
+}
+
+/*! ZSTD_getDictID_fromCDict() :
+ *  Provides the dictID of the dictionary loaded into `cdict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+unsigned ZSTD_getDictID_fromCDict(const ZSTD_CDict* cdict)
+{
+    if (cdict==NULL) return 0;
+    return cdict->dictID;
+}
+
+
+/* ZSTD_compressBegin_usingCDict_advanced() :
+ * cdict must be != NULL */
+size_t ZSTD_compressBegin_usingCDict_advanced(
+    ZSTD_CCtx* const cctx, const ZSTD_CDict* const cdict,
+    ZSTD_frameParameters const fParams, unsigned long long const pledgedSrcSize)
+{
+    ZSTD_CCtx_params cctxParams;
+    DEBUGLOG(4, "ZSTD_compressBegin_usingCDict_advanced");
+    RETURN_ERROR_IF(cdict==NULL, dictionary_wrong, "NULL pointer!");
+    /* Initialize the cctxParams from the cdict */
+    {
+        ZSTD_parameters params;
+        params.fParams = fParams;
+        params.cParams = ( pledgedSrcSize < ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF
+                        || pledgedSrcSize < cdict->dictContentSize * ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER
+                        || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
+                        || cdict->compressionLevel == 0 ) ?
+                ZSTD_getCParamsFromCDict(cdict)
+              : ZSTD_getCParams(cdict->compressionLevel,
+                                pledgedSrcSize,
+                                cdict->dictContentSize);
+        ZSTD_CCtxParams_init_internal(&cctxParams, &params, cdict->compressionLevel);
+    }
+    /* Increase window log to fit the entire dictionary and source if the
+     * source size is known. Limit the increase to 19, which is the
+     * window log for compression level 1 with the largest source size.
+     */
+    if (pledgedSrcSize != ZSTD_CONTENTSIZE_UNKNOWN) {
+        U32 const limitedSrcSize = (U32)MIN(pledgedSrcSize, 1U << 19);
+        U32 const limitedSrcLog = limitedSrcSize > 1 ? ZSTD_highbit32(limitedSrcSize - 1) + 1 : 1;
+        cctxParams.cParams.windowLog = MAX(cctxParams.cParams.windowLog, limitedSrcLog);
+    }
+    return ZSTD_compressBegin_internal(cctx,
+                                        NULL, 0, ZSTD_dct_auto, ZSTD_dtlm_fast,
+                                        cdict,
+                                        &cctxParams, pledgedSrcSize,
+                                        ZSTDb_not_buffered);
+}
+
+/* ZSTD_compressBegin_usingCDict() :
+ * pledgedSrcSize=0 means "unknown"
+ * if pledgedSrcSize>0, it will enable contentSizeFlag */
+size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict)
+{
+    ZSTD_frameParameters const fParams = { 0 /*content*/, 0 /*checksum*/, 0 /*noDictID*/ };
+    DEBUGLOG(4, "ZSTD_compressBegin_usingCDict : dictIDFlag == %u", !fParams.noDictIDFlag);
+    return ZSTD_compressBegin_usingCDict_advanced(cctx, cdict, fParams, ZSTD_CONTENTSIZE_UNKNOWN);
+}
+
+size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx,
+                                void* dst, size_t dstCapacity,
+                                const void* src, size_t srcSize,
+                                const ZSTD_CDict* cdict, ZSTD_frameParameters fParams)
+{
+    FORWARD_IF_ERROR(ZSTD_compressBegin_usingCDict_advanced(cctx, cdict, fParams, srcSize), "");   /* will check if cdict != NULL */
+    return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
+}
+
+/*! ZSTD_compress_usingCDict() :
+ *  Compression using a digested Dictionary.
+ *  Faster startup than ZSTD_compress_usingDict(), recommended when same dictionary is used multiple times.
+ *  Note that compression parameters are decided at CDict creation time
+ *  while frame parameters are hardcoded */
+size_t ZSTD_compress_usingCDict(ZSTD_CCtx* cctx,
+                                void* dst, size_t dstCapacity,
+                                const void* src, size_t srcSize,
+                                const ZSTD_CDict* cdict)
+{
+    ZSTD_frameParameters const fParams = { 1 /*content*/, 0 /*checksum*/, 0 /*noDictID*/ };
+    return ZSTD_compress_usingCDict_advanced(cctx, dst, dstCapacity, src, srcSize, cdict, fParams);
+}
+
+
+
+/* ******************************************************************
+*  Streaming
+********************************************************************/
+
+ZSTD_CStream* ZSTD_createCStream(void)
+{
+    DEBUGLOG(3, "ZSTD_createCStream");
+    return ZSTD_createCStream_advanced(ZSTD_defaultCMem);
+}
+
+ZSTD_CStream* ZSTD_initStaticCStream(void *workspace, size_t workspaceSize)
+{
+    return ZSTD_initStaticCCtx(workspace, workspaceSize);
+}
+
+ZSTD_CStream* ZSTD_createCStream_advanced(ZSTD_customMem customMem)
+{   /* CStream and CCtx are now same object */
+    return ZSTD_createCCtx_advanced(customMem);
+}
+
+size_t ZSTD_freeCStream(ZSTD_CStream* zcs)
+{
+    return ZSTD_freeCCtx(zcs);   /* same object */
+}
+
+
+
+/*======   Initialization   ======*/
+
+size_t ZSTD_CStreamInSize(void)  { return ZSTD_BLOCKSIZE_MAX; }
+
+size_t ZSTD_CStreamOutSize(void)
+{
+    return ZSTD_compressBound(ZSTD_BLOCKSIZE_MAX) + ZSTD_blockHeaderSize + 4 /* 32-bits hash */ ;
+}
+
+static ZSTD_cParamMode_e ZSTD_getCParamMode(ZSTD_CDict const* cdict, ZSTD_CCtx_params const* params, U64 pledgedSrcSize)
+{
+    if (cdict != NULL && ZSTD_shouldAttachDict(cdict, params, pledgedSrcSize))
+        return ZSTD_cpm_attachDict;
+    else
+        return ZSTD_cpm_noAttachDict;
+}
+
+/* ZSTD_resetCStream():
+ * pledgedSrcSize == 0 means "unknown" */
+size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pss)
+{
+    /* temporary : 0 interpreted as "unknown" during transition period.
+     * Users willing to specify "unknown" **must** use ZSTD_CONTENTSIZE_UNKNOWN.
+     * 0 will be interpreted as "empty" in the future.
+     */
+    U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
+    DEBUGLOG(4, "ZSTD_resetCStream: pledgedSrcSize = %u", (unsigned)pledgedSrcSize);
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    return 0;
+}
+
+/*! ZSTD_initCStream_internal() :
+ *  Note : for lib/compress only. Used by zstdmt_compress.c.
+ *  Assumption 1 : params are valid
+ *  Assumption 2 : either dict, or cdict, is defined, not both */
+size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs,
+                    const void* dict, size_t dictSize, const ZSTD_CDict* cdict,
+                    const ZSTD_CCtx_params* params,
+                    unsigned long long pledgedSrcSize)
+{
+    DEBUGLOG(4, "ZSTD_initCStream_internal");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams)));
+    zcs->requestedParams = *params;
+    assert(!((dict) && (cdict)));  /* either dict or cdict, not both */
+    if (dict) {
+        FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
+    } else {
+        /* Dictionary is cleared if !cdict */
+        FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
+    }
+    return 0;
+}
+
+/* ZSTD_initCStream_usingCDict_advanced() :
+ * same as ZSTD_initCStream_usingCDict(), with control over frame parameters */
+size_t ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
+                                            const ZSTD_CDict* cdict,
+                                            ZSTD_frameParameters fParams,
+                                            unsigned long long pledgedSrcSize)
+{
+    DEBUGLOG(4, "ZSTD_initCStream_usingCDict_advanced");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    zcs->requestedParams.fParams = fParams;
+    FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
+    return 0;
+}
+
+/* note : cdict must outlive compression session */
+size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict)
+{
+    DEBUGLOG(4, "ZSTD_initCStream_usingCDict");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
+    return 0;
+}
+
+
+/* ZSTD_initCStream_advanced() :
+ * pledgedSrcSize must be exact.
+ * if srcSize is not known at init time, use value ZSTD_CONTENTSIZE_UNKNOWN.
+ * dict is loaded with default parameters ZSTD_dct_auto and ZSTD_dlm_byCopy. */
+size_t ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
+                                 const void* dict, size_t dictSize,
+                                 ZSTD_parameters params, unsigned long long pss)
+{
+    /* for compatibility with older programs relying on this behavior.
+     * Users should now specify ZSTD_CONTENTSIZE_UNKNOWN.
+     * This line will be removed in the future.
+     */
+    U64 const pledgedSrcSize = (pss==0 && params.fParams.contentSizeFlag==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
+    DEBUGLOG(4, "ZSTD_initCStream_advanced");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , "");
+    ZSTD_CCtxParams_setZstdParams(&zcs->requestedParams, &params);
+    FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
+    return 0;
+}
+
+size_t ZSTD_initCStream_usingDict(ZSTD_CStream* zcs, const void* dict, size_t dictSize, int compressionLevel)
+{
+    DEBUGLOG(4, "ZSTD_initCStream_usingDict");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
+    return 0;
+}
+
+size_t ZSTD_initCStream_srcSize(ZSTD_CStream* zcs, int compressionLevel, unsigned long long pss)
+{
+    /* temporary : 0 interpreted as "unknown" during transition period.
+     * Users willing to specify "unknown" **must** use ZSTD_CONTENTSIZE_UNKNOWN.
+     * 0 will be interpreted as "empty" in the future.
+     */
+    U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
+    DEBUGLOG(4, "ZSTD_initCStream_srcSize");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, NULL) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    return 0;
+}
+
+size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel)
+{
+    DEBUGLOG(4, "ZSTD_initCStream");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, NULL) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
+    return 0;
+}
+
+/*======   Compression   ======*/
+
+static size_t ZSTD_nextInputSizeHint(const ZSTD_CCtx* cctx)
+{
+    size_t hintInSize = cctx->inBuffTarget - cctx->inBuffPos;
+    if (hintInSize==0) hintInSize = cctx->blockSize;
+    return hintInSize;
+}
+
+/* ZSTD_compressStream_generic():
+ *  internal function for all *compressStream*() variants
+ *  non-static, because can be called from zstdmt_compress.c
+ * @return : hint size for next input */
+static size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
+                                          ZSTD_outBuffer* output,
+                                          ZSTD_inBuffer* input,
+                                          ZSTD_EndDirective const flushMode)
+{
+    const char* const istart = (const char*)input->src;
+    const char* const iend = input->size != 0 ? istart + input->size : istart;
+    const char* ip = input->pos != 0 ? istart + input->pos : istart;
+    char* const ostart = (char*)output->dst;
+    char* const oend = output->size != 0 ? ostart + output->size : ostart;
+    char* op = output->pos != 0 ? ostart + output->pos : ostart;
+    U32 someMoreWork = 1;
+
+    /* check expectations */
+    DEBUGLOG(5, "ZSTD_compressStream_generic, flush=%u", (unsigned)flushMode);
+    if (zcs->appliedParams.inBufferMode == ZSTD_bm_buffered) {
+        assert(zcs->inBuff != NULL);
+        assert(zcs->inBuffSize > 0);
+    }
+    if (zcs->appliedParams.outBufferMode == ZSTD_bm_buffered) {
+        assert(zcs->outBuff !=  NULL);
+        assert(zcs->outBuffSize > 0);
+    }
+    assert(output->pos <= output->size);
+    assert(input->pos <= input->size);
+    assert((U32)flushMode <= (U32)ZSTD_e_end);
+
+    while (someMoreWork) {
+        switch(zcs->streamStage)
+        {
+        case zcss_init:
+            RETURN_ERROR(init_missing, "call ZSTD_initCStream() first!");
+
+        case zcss_load:
+            if ( (flushMode == ZSTD_e_end)
+              && ( (size_t)(oend-op) >= ZSTD_compressBound(iend-ip)     /* Enough output space */
+                || zcs->appliedParams.outBufferMode == ZSTD_bm_stable)  /* OR we are allowed to return dstSizeTooSmall */
+              && (zcs->inBuffPos == 0) ) {
+                /* shortcut to compression pass directly into output buffer */
+                size_t const cSize = ZSTD_compressEnd(zcs,
+                                                op, oend-op, ip, iend-ip);
+                DEBUGLOG(4, "ZSTD_compressEnd : cSize=%u", (unsigned)cSize);
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressEnd failed");
+                ip = iend;
+                op += cSize;
+                zcs->frameEnded = 1;
+                ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+                someMoreWork = 0; break;
+            }
+            /* complete loading into inBuffer in buffered mode */
+            if (zcs->appliedParams.inBufferMode == ZSTD_bm_buffered) {
+                size_t const toLoad = zcs->inBuffTarget - zcs->inBuffPos;
+                size_t const loaded = ZSTD_limitCopy(
+                                        zcs->inBuff + zcs->inBuffPos, toLoad,
+                                        ip, iend-ip);
+                zcs->inBuffPos += loaded;
+                if (loaded != 0)
+                    ip += loaded;
+                if ( (flushMode == ZSTD_e_continue)
+                  && (zcs->inBuffPos < zcs->inBuffTarget) ) {
+                    /* not enough input to fill full block : stop here */
+                    someMoreWork = 0; break;
+                }
+                if ( (flushMode == ZSTD_e_flush)
+                  && (zcs->inBuffPos == zcs->inToCompress) ) {
+                    /* empty */
+                    someMoreWork = 0; break;
+                }
+            }
+            /* compress current block (note : this stage cannot be stopped in the middle) */
+            DEBUGLOG(5, "stream compression stage (flushMode==%u)", flushMode);
+            {   int const inputBuffered = (zcs->appliedParams.inBufferMode == ZSTD_bm_buffered);
+                void* cDst;
+                size_t cSize;
+                size_t oSize = oend-op;
+                size_t const iSize = inputBuffered
+                    ? zcs->inBuffPos - zcs->inToCompress
+                    : MIN((size_t)(iend - ip), zcs->blockSize);
+                if (oSize >= ZSTD_compressBound(iSize) || zcs->appliedParams.outBufferMode == ZSTD_bm_stable)
+                    cDst = op;   /* compress into output buffer, to skip flush stage */
+                else
+                    cDst = zcs->outBuff, oSize = zcs->outBuffSize;
+                if (inputBuffered) {
+                    unsigned const lastBlock = (flushMode == ZSTD_e_end) && (ip==iend);
+                    cSize = lastBlock ?
+                            ZSTD_compressEnd(zcs, cDst, oSize,
+                                        zcs->inBuff + zcs->inToCompress, iSize) :
+                            ZSTD_compressContinue(zcs, cDst, oSize,
+                                        zcs->inBuff + zcs->inToCompress, iSize);
+                    FORWARD_IF_ERROR(cSize, "%s", lastBlock ? "ZSTD_compressEnd failed" : "ZSTD_compressContinue failed");
+                    zcs->frameEnded = lastBlock;
+                    /* prepare next block */
+                    zcs->inBuffTarget = zcs->inBuffPos + zcs->blockSize;
+                    if (zcs->inBuffTarget > zcs->inBuffSize)
+                        zcs->inBuffPos = 0, zcs->inBuffTarget = zcs->blockSize;
+                    DEBUGLOG(5, "inBuffTarget:%u / inBuffSize:%u",
+                            (unsigned)zcs->inBuffTarget, (unsigned)zcs->inBuffSize);
+                    if (!lastBlock)
+                        assert(zcs->inBuffTarget <= zcs->inBuffSize);
+                    zcs->inToCompress = zcs->inBuffPos;
+                } else {
+                    unsigned const lastBlock = (ip + iSize == iend);
+                    assert(flushMode == ZSTD_e_end /* Already validated */);
+                    cSize = lastBlock ?
+                            ZSTD_compressEnd(zcs, cDst, oSize, ip, iSize) :
+                            ZSTD_compressContinue(zcs, cDst, oSize, ip, iSize);
+                    /* Consume the input prior to error checking to mirror buffered mode. */
+                    if (iSize > 0)
+                        ip += iSize;
+                    FORWARD_IF_ERROR(cSize, "%s", lastBlock ? "ZSTD_compressEnd failed" : "ZSTD_compressContinue failed");
+                    zcs->frameEnded = lastBlock;
+                    if (lastBlock)
+                        assert(ip == iend);
+                }
+                if (cDst == op) {  /* no need to flush */
+                    op += cSize;
+                    if (zcs->frameEnded) {
+                        DEBUGLOG(5, "Frame completed directly in outBuffer");
+                        someMoreWork = 0;
+                        ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+                    }
+                    break;
+                }
+                zcs->outBuffContentSize = cSize;
+                zcs->outBuffFlushedSize = 0;
+                zcs->streamStage = zcss_flush; /* pass-through to flush stage */
+            }
+	    ZSTD_FALLTHROUGH;
+        case zcss_flush:
+            DEBUGLOG(5, "flush stage");
+            assert(zcs->appliedParams.outBufferMode == ZSTD_bm_buffered);
+            {   size_t const toFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
+                size_t const flushed = ZSTD_limitCopy(op, (size_t)(oend-op),
+                            zcs->outBuff + zcs->outBuffFlushedSize, toFlush);
+                DEBUGLOG(5, "toFlush: %u into %u ==> flushed: %u",
+                            (unsigned)toFlush, (unsigned)(oend-op), (unsigned)flushed);
+                if (flushed)
+                    op += flushed;
+                zcs->outBuffFlushedSize += flushed;
+                if (toFlush!=flushed) {
+                    /* flush not fully completed, presumably because dst is too small */
+                    assert(op==oend);
+                    someMoreWork = 0;
+                    break;
+                }
+                zcs->outBuffContentSize = zcs->outBuffFlushedSize = 0;
+                if (zcs->frameEnded) {
+                    DEBUGLOG(5, "Frame completed on flush");
+                    someMoreWork = 0;
+                    ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
+                    break;
+                }
+                zcs->streamStage = zcss_load;
+                break;
+            }
+
+        default: /* impossible */
+            assert(0);
+        }
+    }
+
+    input->pos = ip - istart;
+    output->pos = op - ostart;
+    if (zcs->frameEnded) return 0;
+    return ZSTD_nextInputSizeHint(zcs);
+}
+
+static size_t ZSTD_nextInputSizeHint_MTorST(const ZSTD_CCtx* cctx)
+{
+    return ZSTD_nextInputSizeHint(cctx);
+
+}
+
+size_t ZSTD_compressStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input)
+{
+    FORWARD_IF_ERROR( ZSTD_compressStream2(zcs, output, input, ZSTD_e_continue) , "");
+    return ZSTD_nextInputSizeHint_MTorST(zcs);
+}
+
+/* After a compression call set the expected input/output buffer.
+ * This is validated at the start of the next compression call.
+ */
+static void ZSTD_setBufferExpectations(ZSTD_CCtx* cctx, ZSTD_outBuffer const* output, ZSTD_inBuffer const* input)
+{
+    if (cctx->appliedParams.inBufferMode == ZSTD_bm_stable) {
+        cctx->expectedInBuffer = *input;
+    }
+    if (cctx->appliedParams.outBufferMode == ZSTD_bm_stable) {
+        cctx->expectedOutBufferSize = output->size - output->pos;
+    }
+}
+
+/* Validate that the input/output buffers match the expectations set by
+ * ZSTD_setBufferExpectations.
+ */
+static size_t ZSTD_checkBufferStability(ZSTD_CCtx const* cctx,
+                                        ZSTD_outBuffer const* output,
+                                        ZSTD_inBuffer const* input,
+                                        ZSTD_EndDirective endOp)
+{
+    if (cctx->appliedParams.inBufferMode == ZSTD_bm_stable) {
+        ZSTD_inBuffer const expect = cctx->expectedInBuffer;
+        if (expect.src != input->src || expect.pos != input->pos || expect.size != input->size)
+            RETURN_ERROR(srcBuffer_wrong, "ZSTD_c_stableInBuffer enabled but input differs!");
+        if (endOp != ZSTD_e_end)
+            RETURN_ERROR(srcBuffer_wrong, "ZSTD_c_stableInBuffer can only be used with ZSTD_e_end!");
+    }
+    if (cctx->appliedParams.outBufferMode == ZSTD_bm_stable) {
+        size_t const outBufferSize = output->size - output->pos;
+        if (cctx->expectedOutBufferSize != outBufferSize)
+            RETURN_ERROR(dstBuffer_wrong, "ZSTD_c_stableOutBuffer enabled but output size differs!");
+    }
+    return 0;
+}
+
+static size_t ZSTD_CCtx_init_compressStream2(ZSTD_CCtx* cctx,
+                                             ZSTD_EndDirective endOp,
+                                             size_t inSize) {
+    ZSTD_CCtx_params params = cctx->requestedParams;
+    ZSTD_prefixDict const prefixDict = cctx->prefixDict;
+    FORWARD_IF_ERROR( ZSTD_initLocalDict(cctx) , ""); /* Init the local dict if present. */
+    ZSTD_memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict));   /* single usage */
+    assert(prefixDict.dict==NULL || cctx->cdict==NULL);    /* only one can be set */
+    if (cctx->cdict)
+        params.compressionLevel = cctx->cdict->compressionLevel; /* let cdict take priority in terms of compression level */
+    DEBUGLOG(4, "ZSTD_compressStream2 : transparent init stage");
+    if (endOp == ZSTD_e_end) cctx->pledgedSrcSizePlusOne = inSize + 1;  /* auto-fix pledgedSrcSize */
+    {
+        size_t const dictSize = prefixDict.dict
+                ? prefixDict.dictSize
+                : (cctx->cdict ? cctx->cdict->dictContentSize : 0);
+        ZSTD_cParamMode_e const mode = ZSTD_getCParamMode(cctx->cdict, &params, cctx->pledgedSrcSizePlusOne - 1);
+        params.cParams = ZSTD_getCParamsFromCCtxParams(
+                &params, cctx->pledgedSrcSizePlusOne-1,
+                dictSize, mode);
+    }
+
+    if (ZSTD_CParams_shouldEnableLdm(&params.cParams)) {
+        /* Enable LDM by default for optimal parser and window size >= 128MB */
+        DEBUGLOG(4, "LDM enabled by default (window size >= 128MB, strategy >= btopt)");
+        params.ldmParams.enableLdm = 1;
+    }
+
+    {   U64 const pledgedSrcSize = cctx->pledgedSrcSizePlusOne - 1;
+        assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));
+        FORWARD_IF_ERROR( ZSTD_compressBegin_internal(cctx,
+                prefixDict.dict, prefixDict.dictSize, prefixDict.dictContentType, ZSTD_dtlm_fast,
+                cctx->cdict,
+                &params, pledgedSrcSize,
+                ZSTDb_buffered) , "");
+        assert(cctx->appliedParams.nbWorkers == 0);
+        cctx->inToCompress = 0;
+        cctx->inBuffPos = 0;
+        if (cctx->appliedParams.inBufferMode == ZSTD_bm_buffered) {
+            /* for small input: avoid automatic flush on reaching end of block, since
+            * it would require to add a 3-bytes null block to end frame
+            */
+            cctx->inBuffTarget = cctx->blockSize + (cctx->blockSize == pledgedSrcSize);
+        } else {
+            cctx->inBuffTarget = 0;
+        }
+        cctx->outBuffContentSize = cctx->outBuffFlushedSize = 0;
+        cctx->streamStage = zcss_load;
+        cctx->frameEnded = 0;
+    }
+    return 0;
+}
+
+size_t ZSTD_compressStream2( ZSTD_CCtx* cctx,
+                             ZSTD_outBuffer* output,
+                             ZSTD_inBuffer* input,
+                             ZSTD_EndDirective endOp)
+{
+    DEBUGLOG(5, "ZSTD_compressStream2, endOp=%u ", (unsigned)endOp);
+    /* check conditions */
+    RETURN_ERROR_IF(output->pos > output->size, dstSize_tooSmall, "invalid output buffer");
+    RETURN_ERROR_IF(input->pos  > input->size, srcSize_wrong, "invalid input buffer");
+    RETURN_ERROR_IF((U32)endOp > (U32)ZSTD_e_end, parameter_outOfBound, "invalid endDirective");
+    assert(cctx != NULL);
+
+    /* transparent initialization stage */
+    if (cctx->streamStage == zcss_init) {
+        FORWARD_IF_ERROR(ZSTD_CCtx_init_compressStream2(cctx, endOp, input->size), "CompressStream2 initialization failed");
+        ZSTD_setBufferExpectations(cctx, output, input);    /* Set initial buffer expectations now that we've initialized */
+    }
+    /* end of transparent initialization stage */
+
+    FORWARD_IF_ERROR(ZSTD_checkBufferStability(cctx, output, input, endOp), "invalid buffers");
+    /* compression stage */
+    FORWARD_IF_ERROR( ZSTD_compressStream_generic(cctx, output, input, endOp) , "");
+    DEBUGLOG(5, "completed ZSTD_compressStream2");
+    ZSTD_setBufferExpectations(cctx, output, input);
+    return cctx->outBuffContentSize - cctx->outBuffFlushedSize; /* remaining to flush */
+}
+
+size_t ZSTD_compressStream2_simpleArgs (
+                            ZSTD_CCtx* cctx,
+                            void* dst, size_t dstCapacity, size_t* dstPos,
+                      const void* src, size_t srcSize, size_t* srcPos,
+                            ZSTD_EndDirective endOp)
+{
+    ZSTD_outBuffer output = { dst, dstCapacity, *dstPos };
+    ZSTD_inBuffer  input  = { src, srcSize, *srcPos };
+    /* ZSTD_compressStream2() will check validity of dstPos and srcPos */
+    size_t const cErr = ZSTD_compressStream2(cctx, &output, &input, endOp);
+    *dstPos = output.pos;
+    *srcPos = input.pos;
+    return cErr;
+}
+
+size_t ZSTD_compress2(ZSTD_CCtx* cctx,
+                      void* dst, size_t dstCapacity,
+                      const void* src, size_t srcSize)
+{
+    ZSTD_bufferMode_e const originalInBufferMode = cctx->requestedParams.inBufferMode;
+    ZSTD_bufferMode_e const originalOutBufferMode = cctx->requestedParams.outBufferMode;
+    DEBUGLOG(4, "ZSTD_compress2 (srcSize=%u)", (unsigned)srcSize);
+    ZSTD_CCtx_reset(cctx, ZSTD_reset_session_only);
+    /* Enable stable input/output buffers. */
+    cctx->requestedParams.inBufferMode = ZSTD_bm_stable;
+    cctx->requestedParams.outBufferMode = ZSTD_bm_stable;
+    {   size_t oPos = 0;
+        size_t iPos = 0;
+        size_t const result = ZSTD_compressStream2_simpleArgs(cctx,
+                                        dst, dstCapacity, &oPos,
+                                        src, srcSize, &iPos,
+                                        ZSTD_e_end);
+        /* Reset to the original values. */
+        cctx->requestedParams.inBufferMode = originalInBufferMode;
+        cctx->requestedParams.outBufferMode = originalOutBufferMode;
+        FORWARD_IF_ERROR(result, "ZSTD_compressStream2_simpleArgs failed");
+        if (result != 0) {  /* compression not completed, due to lack of output space */
+            assert(oPos == dstCapacity);
+            RETURN_ERROR(dstSize_tooSmall, "");
+        }
+        assert(iPos == srcSize);   /* all input is expected consumed */
+        return oPos;
+    }
+}
+
+typedef struct {
+    U32 idx;             /* Index in array of ZSTD_Sequence */
+    U32 posInSequence;   /* Position within sequence at idx */
+    size_t posInSrc;        /* Number of bytes given by sequences provided so far */
+} ZSTD_sequencePosition;
+
+/* Returns a ZSTD error code if sequence is not valid */
+static size_t ZSTD_validateSequence(U32 offCode, U32 matchLength,
+                                    size_t posInSrc, U32 windowLog, size_t dictSize, U32 minMatch) {
+    size_t offsetBound;
+    U32 windowSize = 1 << windowLog;
+    /* posInSrc represents the amount of data the the decoder would decode up to this point.
+     * As long as the amount of data decoded is less than or equal to window size, offsets may be
+     * larger than the total length of output decoded in order to reference the dict, even larger than
+     * window size. After output surpasses windowSize, we're limited to windowSize offsets again.
+     */
+    offsetBound = posInSrc > windowSize ? (size_t)windowSize : posInSrc + (size_t)dictSize;
+    RETURN_ERROR_IF(offCode > offsetBound + ZSTD_REP_MOVE, corruption_detected, "Offset too large!");
+    RETURN_ERROR_IF(matchLength < minMatch, corruption_detected, "Matchlength too small");
+    return 0;
+}
+
+/* Returns an offset code, given a sequence's raw offset, the ongoing repcode array, and whether litLength == 0 */
+static U32 ZSTD_finalizeOffCode(U32 rawOffset, const U32 rep[ZSTD_REP_NUM], U32 ll0) {
+    U32 offCode = rawOffset + ZSTD_REP_MOVE;
+    U32 repCode = 0;
+
+    if (!ll0 && rawOffset == rep[0]) {
+        repCode = 1;
+    } else if (rawOffset == rep[1]) {
+        repCode = 2 - ll0;
+    } else if (rawOffset == rep[2]) {
+        repCode = 3 - ll0;
+    } else if (ll0 && rawOffset == rep[0] - 1) {
+        repCode = 3;
+    }
+    if (repCode) {
+        /* ZSTD_storeSeq expects a number in the range [0, 2] to represent a repcode */
+        offCode = repCode - 1;
+    }
+    return offCode;
+}
+
+/* Returns 0 on success, and a ZSTD_error otherwise. This function scans through an array of
+ * ZSTD_Sequence, storing the sequences it finds, until it reaches a block delimiter.
+ */
+static size_t ZSTD_copySequencesToSeqStoreExplicitBlockDelim(ZSTD_CCtx* cctx, ZSTD_sequencePosition* seqPos,
+                                                             const ZSTD_Sequence* const inSeqs, size_t inSeqsSize,
+                                                             const void* src, size_t blockSize) {
+    U32 idx = seqPos->idx;
+    BYTE const* ip = (BYTE const*)(src);
+    const BYTE* const iend = ip + blockSize;
+    repcodes_t updatedRepcodes;
+    U32 dictSize;
+    U32 litLength;
+    U32 matchLength;
+    U32 ll0;
+    U32 offCode;
+
+    if (cctx->cdict) {
+        dictSize = (U32)cctx->cdict->dictContentSize;
+    } else if (cctx->prefixDict.dict) {
+        dictSize = (U32)cctx->prefixDict.dictSize;
+    } else {
+        dictSize = 0;
+    }
+    ZSTD_memcpy(updatedRepcodes.rep, cctx->blockState.prevCBlock->rep, sizeof(repcodes_t));
+    for (; (inSeqs[idx].matchLength != 0 || inSeqs[idx].offset != 0) && idx < inSeqsSize; ++idx) {
+        litLength = inSeqs[idx].litLength;
+        matchLength = inSeqs[idx].matchLength;
+        ll0 = litLength == 0;
+        offCode = ZSTD_finalizeOffCode(inSeqs[idx].offset, updatedRepcodes.rep, ll0);
+        updatedRepcodes = ZSTD_updateRep(updatedRepcodes.rep, offCode, ll0);
+
+        DEBUGLOG(6, "Storing sequence: (of: %u, ml: %u, ll: %u)", offCode, matchLength, litLength);
+        if (cctx->appliedParams.validateSequences) {
+            seqPos->posInSrc += litLength + matchLength;
+            FORWARD_IF_ERROR(ZSTD_validateSequence(offCode, matchLength, seqPos->posInSrc,
+                                                cctx->appliedParams.cParams.windowLog, dictSize,
+                                                cctx->appliedParams.cParams.minMatch),
+                                                "Sequence validation failed");
+        }
+        RETURN_ERROR_IF(idx - seqPos->idx > cctx->seqStore.maxNbSeq, memory_allocation,
+                        "Not enough memory allocated. Try adjusting ZSTD_c_minMatch.");
+        ZSTD_storeSeq(&cctx->seqStore, litLength, ip, iend, offCode, matchLength - MINMATCH);
+        ip += matchLength + litLength;
+    }
+    ZSTD_memcpy(cctx->blockState.nextCBlock->rep, updatedRepcodes.rep, sizeof(repcodes_t));
+
+    if (inSeqs[idx].litLength) {
+        DEBUGLOG(6, "Storing last literals of size: %u", inSeqs[idx].litLength);
+        ZSTD_storeLastLiterals(&cctx->seqStore, ip, inSeqs[idx].litLength);
+        ip += inSeqs[idx].litLength;
+        seqPos->posInSrc += inSeqs[idx].litLength;
+    }
+    RETURN_ERROR_IF(ip != iend, corruption_detected, "Blocksize doesn't agree with block delimiter!");
+    seqPos->idx = idx+1;
+    return 0;
+}
+
+/* Returns the number of bytes to move the current read position back by. Only non-zero
+ * if we ended up splitting a sequence. Otherwise, it may return a ZSTD error if something
+ * went wrong.
+ *
+ * This function will attempt to scan through blockSize bytes represented by the sequences
+ * in inSeqs, storing any (partial) sequences.
+ *
+ * Occasionally, we may want to change the actual number of bytes we consumed from inSeqs to
+ * avoid splitting a match, or to avoid splitting a match such that it would produce a match
+ * smaller than MINMATCH. In this case, we return the number of bytes that we didn't read from this block.
+ */
+static size_t ZSTD_copySequencesToSeqStoreNoBlockDelim(ZSTD_CCtx* cctx, ZSTD_sequencePosition* seqPos,
+                                                       const ZSTD_Sequence* const inSeqs, size_t inSeqsSize,
+                                                       const void* src, size_t blockSize) {
+    U32 idx = seqPos->idx;
+    U32 startPosInSequence = seqPos->posInSequence;
+    U32 endPosInSequence = seqPos->posInSequence + (U32)blockSize;
+    size_t dictSize;
+    BYTE const* ip = (BYTE const*)(src);
+    BYTE const* iend = ip + blockSize;  /* May be adjusted if we decide to process fewer than blockSize bytes */
+    repcodes_t updatedRepcodes;
+    U32 bytesAdjustment = 0;
+    U32 finalMatchSplit = 0;
+    U32 litLength;
+    U32 matchLength;
+    U32 rawOffset;
+    U32 offCode;
+
+    if (cctx->cdict) {
+        dictSize = cctx->cdict->dictContentSize;
+    } else if (cctx->prefixDict.dict) {
+        dictSize = cctx->prefixDict.dictSize;
+    } else {
+        dictSize = 0;
+    }
+    DEBUGLOG(5, "ZSTD_copySequencesToSeqStore: idx: %u PIS: %u blockSize: %zu", idx, startPosInSequence, blockSize);
+    DEBUGLOG(5, "Start seq: idx: %u (of: %u ml: %u ll: %u)", idx, inSeqs[idx].offset, inSeqs[idx].matchLength, inSeqs[idx].litLength);
+    ZSTD_memcpy(updatedRepcodes.rep, cctx->blockState.prevCBlock->rep, sizeof(repcodes_t));
+    while (endPosInSequence && idx < inSeqsSize && !finalMatchSplit) {
+        const ZSTD_Sequence currSeq = inSeqs[idx];
+        litLength = currSeq.litLength;
+        matchLength = currSeq.matchLength;
+        rawOffset = currSeq.offset;
+
+        /* Modify the sequence depending on where endPosInSequence lies */
+        if (endPosInSequence >= currSeq.litLength + currSeq.matchLength) {
+            if (startPosInSequence >= litLength) {
+                startPosInSequence -= litLength;
+                litLength = 0;
+                matchLength -= startPosInSequence;
+            } else {
+                litLength -= startPosInSequence;
+            }
+            /* Move to the next sequence */
+            endPosInSequence -= currSeq.litLength + currSeq.matchLength;
+            startPosInSequence = 0;
+            idx++;
+        } else {
+            /* This is the final (partial) sequence we're adding from inSeqs, and endPosInSequence
+               does not reach the end of the match. So, we have to split the sequence */
+            DEBUGLOG(6, "Require a split: diff: %u, idx: %u PIS: %u",
+                     currSeq.litLength + currSeq.matchLength - endPosInSequence, idx, endPosInSequence);
+            if (endPosInSequence > litLength) {
+                U32 firstHalfMatchLength;
+                litLength = startPosInSequence >= litLength ? 0 : litLength - startPosInSequence;
+                firstHalfMatchLength = endPosInSequence - startPosInSequence - litLength;
+                if (matchLength > blockSize && firstHalfMatchLength >= cctx->appliedParams.cParams.minMatch) {
+                    /* Only ever split the match if it is larger than the block size */
+                    U32 secondHalfMatchLength = currSeq.matchLength + currSeq.litLength - endPosInSequence;
+                    if (secondHalfMatchLength < cctx->appliedParams.cParams.minMatch) {
+                        /* Move the endPosInSequence backward so that it creates match of minMatch length */
+                        endPosInSequence -= cctx->appliedParams.cParams.minMatch - secondHalfMatchLength;
+                        bytesAdjustment = cctx->appliedParams.cParams.minMatch - secondHalfMatchLength;
+                        firstHalfMatchLength -= bytesAdjustment;
+                    }
+                    matchLength = firstHalfMatchLength;
+                    /* Flag that we split the last match - after storing the sequence, exit the loop,
+                       but keep the value of endPosInSequence */
+                    finalMatchSplit = 1;
+                } else {
+                    /* Move the position in sequence backwards so that we don't split match, and break to store
+                     * the last literals. We use the original currSeq.litLength as a marker for where endPosInSequence
+                     * should go. We prefer to do this whenever it is not necessary to split the match, or if doing so
+                     * would cause the first half of the match to be too small
+                     */
+                    bytesAdjustment = endPosInSequence - currSeq.litLength;
+                    endPosInSequence = currSeq.litLength;
+                    break;
+                }
+            } else {
+                /* This sequence ends inside the literals, break to store the last literals */
+                break;
+            }
+        }
+        /* Check if this offset can be represented with a repcode */
+        {   U32 ll0 = (litLength == 0);
+            offCode = ZSTD_finalizeOffCode(rawOffset, updatedRepcodes.rep, ll0);
+            updatedRepcodes = ZSTD_updateRep(updatedRepcodes.rep, offCode, ll0);
+        }
+
+        if (cctx->appliedParams.validateSequences) {
+            seqPos->posInSrc += litLength + matchLength;
+            FORWARD_IF_ERROR(ZSTD_validateSequence(offCode, matchLength, seqPos->posInSrc,
+                                                   cctx->appliedParams.cParams.windowLog, dictSize,
+                                                   cctx->appliedParams.cParams.minMatch),
+                                                   "Sequence validation failed");
+        }
+        DEBUGLOG(6, "Storing sequence: (of: %u, ml: %u, ll: %u)", offCode, matchLength, litLength);
+        RETURN_ERROR_IF(idx - seqPos->idx > cctx->seqStore.maxNbSeq, memory_allocation,
+                        "Not enough memory allocated. Try adjusting ZSTD_c_minMatch.");
+        ZSTD_storeSeq(&cctx->seqStore, litLength, ip, iend, offCode, matchLength - MINMATCH);
+        ip += matchLength + litLength;
+    }
+    DEBUGLOG(5, "Ending seq: idx: %u (of: %u ml: %u ll: %u)", idx, inSeqs[idx].offset, inSeqs[idx].matchLength, inSeqs[idx].litLength);
+    assert(idx == inSeqsSize || endPosInSequence <= inSeqs[idx].litLength + inSeqs[idx].matchLength);
+    seqPos->idx = idx;
+    seqPos->posInSequence = endPosInSequence;
+    ZSTD_memcpy(cctx->blockState.nextCBlock->rep, updatedRepcodes.rep, sizeof(repcodes_t));
+
+    iend -= bytesAdjustment;
+    if (ip != iend) {
+        /* Store any last literals */
+        U32 lastLLSize = (U32)(iend - ip);
+        assert(ip <= iend);
+        DEBUGLOG(6, "Storing last literals of size: %u", lastLLSize);
+        ZSTD_storeLastLiterals(&cctx->seqStore, ip, lastLLSize);
+        seqPos->posInSrc += lastLLSize;
+    }
+
+    return bytesAdjustment;
+}
+
+typedef size_t (*ZSTD_sequenceCopier) (ZSTD_CCtx* cctx, ZSTD_sequencePosition* seqPos,
+                                       const ZSTD_Sequence* const inSeqs, size_t inSeqsSize,
+                                       const void* src, size_t blockSize);
+static ZSTD_sequenceCopier ZSTD_selectSequenceCopier(ZSTD_sequenceFormat_e mode) {
+    ZSTD_sequenceCopier sequenceCopier = NULL;
+    assert(ZSTD_cParam_withinBounds(ZSTD_c_blockDelimiters, mode));
+    if (mode == ZSTD_sf_explicitBlockDelimiters) {
+        return ZSTD_copySequencesToSeqStoreExplicitBlockDelim;
+    } else if (mode == ZSTD_sf_noBlockDelimiters) {
+        return ZSTD_copySequencesToSeqStoreNoBlockDelim;
+    }
+    assert(sequenceCopier != NULL);
+    return sequenceCopier;
+}
+
+/* Compress, block-by-block, all of the sequences given.
+ *
+ * Returns the cumulative size of all compressed blocks (including their headers), otherwise a ZSTD error.
+ */
+static size_t ZSTD_compressSequences_internal(ZSTD_CCtx* cctx,
+                                              void* dst, size_t dstCapacity,
+                                              const ZSTD_Sequence* inSeqs, size_t inSeqsSize,
+                                              const void* src, size_t srcSize) {
+    size_t cSize = 0;
+    U32 lastBlock;
+    size_t blockSize;
+    size_t compressedSeqsSize;
+    size_t remaining = srcSize;
+    ZSTD_sequencePosition seqPos = {0, 0, 0};
+
+    BYTE const* ip = (BYTE const*)src;
+    BYTE* op = (BYTE*)dst;
+    ZSTD_sequenceCopier sequenceCopier = ZSTD_selectSequenceCopier(cctx->appliedParams.blockDelimiters);
+
+    DEBUGLOG(4, "ZSTD_compressSequences_internal srcSize: %zu, inSeqsSize: %zu", srcSize, inSeqsSize);
+    /* Special case: empty frame */
+    if (remaining == 0) {
+        U32 const cBlockHeader24 = 1 /* last block */ + (((U32)bt_raw)<<1);
+        RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "No room for empty frame block header");
+        MEM_writeLE32(op, cBlockHeader24);
+        op += ZSTD_blockHeaderSize;
+        dstCapacity -= ZSTD_blockHeaderSize;
+        cSize += ZSTD_blockHeaderSize;
+    }
+
+    while (remaining) {
+        size_t cBlockSize;
+        size_t additionalByteAdjustment;
+        lastBlock = remaining <= cctx->blockSize;
+        blockSize = lastBlock ? (U32)remaining : (U32)cctx->blockSize;
+        ZSTD_resetSeqStore(&cctx->seqStore);
+        DEBUGLOG(4, "Working on new block. Blocksize: %zu", blockSize);
+
+        additionalByteAdjustment = sequenceCopier(cctx, &seqPos, inSeqs, inSeqsSize, ip, blockSize);
+        FORWARD_IF_ERROR(additionalByteAdjustment, "Bad sequence copy");
+        blockSize -= additionalByteAdjustment;
+
+        /* If blocks are too small, emit as a nocompress block */
+        if (blockSize < MIN_CBLOCK_SIZE+ZSTD_blockHeaderSize+1) {
+            cBlockSize = ZSTD_noCompressBlock(op, dstCapacity, ip, blockSize, lastBlock);
+            FORWARD_IF_ERROR(cBlockSize, "Nocompress block failed");
+            DEBUGLOG(4, "Block too small, writing out nocompress block: cSize: %zu", cBlockSize);
+            cSize += cBlockSize;
+            ip += blockSize;
+            op += cBlockSize;
+            remaining -= blockSize;
+            dstCapacity -= cBlockSize;
+            continue;
+        }
+
+        compressedSeqsSize = ZSTD_entropyCompressSequences(&cctx->seqStore,
+                                &cctx->blockState.prevCBlock->entropy, &cctx->blockState.nextCBlock->entropy,
+                                &cctx->appliedParams,
+                                op + ZSTD_blockHeaderSize /* Leave space for block header */, dstCapacity - ZSTD_blockHeaderSize,
+                                blockSize,
+                                cctx->entropyWorkspace, ENTROPY_WORKSPACE_SIZE /* statically allocated in resetCCtx */,
+                                cctx->bmi2);
+        FORWARD_IF_ERROR(compressedSeqsSize, "Compressing sequences of block failed");
+        DEBUGLOG(4, "Compressed sequences size: %zu", compressedSeqsSize);
+
+        if (!cctx->isFirstBlock &&
+            ZSTD_maybeRLE(&cctx->seqStore) &&
+            ZSTD_isRLE((BYTE const*)src, srcSize)) {
+            /* We don't want to emit our first block as a RLE even if it qualifies because
+            * doing so will cause the decoder (cli only) to throw a "should consume all input error."
+            * This is only an issue for zstd <= v1.4.3
+            */
+            compressedSeqsSize = 1;
+        }
+
+        if (compressedSeqsSize == 0) {
+            /* ZSTD_noCompressBlock writes the block header as well */
+            cBlockSize = ZSTD_noCompressBlock(op, dstCapacity, ip, blockSize, lastBlock);
+            FORWARD_IF_ERROR(cBlockSize, "Nocompress block failed");
+            DEBUGLOG(4, "Writing out nocompress block, size: %zu", cBlockSize);
+        } else if (compressedSeqsSize == 1) {
+            cBlockSize = ZSTD_rleCompressBlock(op, dstCapacity, *ip, blockSize, lastBlock);
+            FORWARD_IF_ERROR(cBlockSize, "RLE compress block failed");
+            DEBUGLOG(4, "Writing out RLE block, size: %zu", cBlockSize);
+        } else {
+            U32 cBlockHeader;
+            /* Error checking and repcodes update */
+            ZSTD_confirmRepcodesAndEntropyTables(cctx);
+            if (cctx->blockState.prevCBlock->entropy.fse.offcode_repeatMode == FSE_repeat_valid)
+                cctx->blockState.prevCBlock->entropy.fse.offcode_repeatMode = FSE_repeat_check;
+
+            /* Write block header into beginning of block*/
+            cBlockHeader = lastBlock + (((U32)bt_compressed)<<1) + (U32)(compressedSeqsSize << 3);
+            MEM_writeLE24(op, cBlockHeader);
+            cBlockSize = ZSTD_blockHeaderSize + compressedSeqsSize;
+            DEBUGLOG(4, "Writing out compressed block, size: %zu", cBlockSize);
+        }
+
+        cSize += cBlockSize;
+        DEBUGLOG(4, "cSize running total: %zu", cSize);
+
+        if (lastBlock) {
+            break;
+        } else {
+            ip += blockSize;
+            op += cBlockSize;
+            remaining -= blockSize;
+            dstCapacity -= cBlockSize;
+            cctx->isFirstBlock = 0;
+        }
+    }
+
+    return cSize;
+}
+
+size_t ZSTD_compressSequences(ZSTD_CCtx* const cctx, void* dst, size_t dstCapacity,
+                              const ZSTD_Sequence* inSeqs, size_t inSeqsSize,
+                              const void* src, size_t srcSize) {
+    BYTE* op = (BYTE*)dst;
+    size_t cSize = 0;
+    size_t compressedBlocksSize = 0;
+    size_t frameHeaderSize = 0;
+
+    /* Transparent initialization stage, same as compressStream2() */
+    DEBUGLOG(3, "ZSTD_compressSequences()");
+    assert(cctx != NULL);
+    FORWARD_IF_ERROR(ZSTD_CCtx_init_compressStream2(cctx, ZSTD_e_end, srcSize), "CCtx initialization failed");
+    /* Begin writing output, starting with frame header */
+    frameHeaderSize = ZSTD_writeFrameHeader(op, dstCapacity, &cctx->appliedParams, srcSize, cctx->dictID);
+    op += frameHeaderSize;
+    dstCapacity -= frameHeaderSize;
+    cSize += frameHeaderSize;
+    if (cctx->appliedParams.fParams.checksumFlag && srcSize) {
+        xxh64_update(&cctx->xxhState, src, srcSize);
+    }
+    /* cSize includes block header size and compressed sequences size */
+    compressedBlocksSize = ZSTD_compressSequences_internal(cctx,
+                                                           op, dstCapacity,
+                                                           inSeqs, inSeqsSize,
+                                                           src, srcSize);
+    FORWARD_IF_ERROR(compressedBlocksSize, "Compressing blocks failed!");
+    cSize += compressedBlocksSize;
+    dstCapacity -= compressedBlocksSize;
+
+    if (cctx->appliedParams.fParams.checksumFlag) {
+        U32 const checksum = (U32) xxh64_digest(&cctx->xxhState);
+        RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for checksum");
+        DEBUGLOG(4, "Write checksum : %08X", (unsigned)checksum);
+        MEM_writeLE32((char*)dst + cSize, checksum);
+        cSize += 4;
+    }
+
+    DEBUGLOG(3, "Final compressed size: %zu", cSize);
+    return cSize;
+}
+
+/*======   Finalize   ======*/
+
+/*! ZSTD_flushStream() :
+ * @return : amount of data remaining to flush */
+size_t ZSTD_flushStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output)
+{
+    ZSTD_inBuffer input = { NULL, 0, 0 };
+    return ZSTD_compressStream2(zcs, output, &input, ZSTD_e_flush);
+}
+
+
+size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output)
+{
+    ZSTD_inBuffer input = { NULL, 0, 0 };
+    size_t const remainingToFlush = ZSTD_compressStream2(zcs, output, &input, ZSTD_e_end);
+    FORWARD_IF_ERROR( remainingToFlush , "ZSTD_compressStream2 failed");
+    if (zcs->appliedParams.nbWorkers > 0) return remainingToFlush;   /* minimal estimation */
+    /* single thread mode : attempt to calculate remaining to flush more precisely */
+    {   size_t const lastBlockSize = zcs->frameEnded ? 0 : ZSTD_BLOCKHEADERSIZE;
+        size_t const checksumSize = (size_t)(zcs->frameEnded ? 0 : zcs->appliedParams.fParams.checksumFlag * 4);
+        size_t const toFlush = remainingToFlush + lastBlockSize + checksumSize;
+        DEBUGLOG(4, "ZSTD_endStream : remaining to flush : %u", (unsigned)toFlush);
+        return toFlush;
+    }
+}
+
+
+/*-=====  Pre-defined compression levels  =====-*/
+
+#define ZSTD_MAX_CLEVEL     22
+int ZSTD_maxCLevel(void) { return ZSTD_MAX_CLEVEL; }
+int ZSTD_minCLevel(void) { return (int)-ZSTD_TARGETLENGTH_MAX; }
+
+static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEVEL+1] = {
+{   /* "default" - for any srcSize > 256 KB */
+    /* W,  C,  H,  S,  L, TL, strat */
+    { 19, 12, 13,  1,  6,  1, ZSTD_fast    },  /* base for negative levels */
+    { 19, 13, 14,  1,  7,  0, ZSTD_fast    },  /* level  1 */
+    { 20, 15, 16,  1,  6,  0, ZSTD_fast    },  /* level  2 */
+    { 21, 16, 17,  1,  5,  0, ZSTD_dfast   },  /* level  3 */
+    { 21, 18, 18,  1,  5,  0, ZSTD_dfast   },  /* level  4 */
+    { 21, 18, 19,  2,  5,  2, ZSTD_greedy  },  /* level  5 */
+    { 21, 19, 19,  3,  5,  4, ZSTD_greedy  },  /* level  6 */
+    { 21, 19, 19,  3,  5,  8, ZSTD_lazy    },  /* level  7 */
+    { 21, 19, 19,  3,  5, 16, ZSTD_lazy2   },  /* level  8 */
+    { 21, 19, 20,  4,  5, 16, ZSTD_lazy2   },  /* level  9 */
+    { 22, 20, 21,  4,  5, 16, ZSTD_lazy2   },  /* level 10 */
+    { 22, 21, 22,  4,  5, 16, ZSTD_lazy2   },  /* level 11 */
+    { 22, 21, 22,  5,  5, 16, ZSTD_lazy2   },  /* level 12 */
+    { 22, 21, 22,  5,  5, 32, ZSTD_btlazy2 },  /* level 13 */
+    { 22, 22, 23,  5,  5, 32, ZSTD_btlazy2 },  /* level 14 */
+    { 22, 23, 23,  6,  5, 32, ZSTD_btlazy2 },  /* level 15 */
+    { 22, 22, 22,  5,  5, 48, ZSTD_btopt   },  /* level 16 */
+    { 23, 23, 22,  5,  4, 64, ZSTD_btopt   },  /* level 17 */
+    { 23, 23, 22,  6,  3, 64, ZSTD_btultra },  /* level 18 */
+    { 23, 24, 22,  7,  3,256, ZSTD_btultra2},  /* level 19 */
+    { 25, 25, 23,  7,  3,256, ZSTD_btultra2},  /* level 20 */
+    { 26, 26, 24,  7,  3,512, ZSTD_btultra2},  /* level 21 */
+    { 27, 27, 25,  9,  3,999, ZSTD_btultra2},  /* level 22 */
+},
+{   /* for srcSize <= 256 KB */
+    /* W,  C,  H,  S,  L,  T, strat */
+    { 18, 12, 13,  1,  5,  1, ZSTD_fast    },  /* base for negative levels */
+    { 18, 13, 14,  1,  6,  0, ZSTD_fast    },  /* level  1 */
+    { 18, 14, 14,  1,  5,  0, ZSTD_dfast   },  /* level  2 */
+    { 18, 16, 16,  1,  4,  0, ZSTD_dfast   },  /* level  3 */
+    { 18, 16, 17,  2,  5,  2, ZSTD_greedy  },  /* level  4.*/
+    { 18, 18, 18,  3,  5,  2, ZSTD_greedy  },  /* level  5.*/
+    { 18, 18, 19,  3,  5,  4, ZSTD_lazy    },  /* level  6.*/
+    { 18, 18, 19,  4,  4,  4, ZSTD_lazy    },  /* level  7 */
+    { 18, 18, 19,  4,  4,  8, ZSTD_lazy2   },  /* level  8 */
+    { 18, 18, 19,  5,  4,  8, ZSTD_lazy2   },  /* level  9 */
+    { 18, 18, 19,  6,  4,  8, ZSTD_lazy2   },  /* level 10 */
+    { 18, 18, 19,  5,  4, 12, ZSTD_btlazy2 },  /* level 11.*/
+    { 18, 19, 19,  7,  4, 12, ZSTD_btlazy2 },  /* level 12.*/
+    { 18, 18, 19,  4,  4, 16, ZSTD_btopt   },  /* level 13 */
+    { 18, 18, 19,  4,  3, 32, ZSTD_btopt   },  /* level 14.*/
+    { 18, 18, 19,  6,  3,128, ZSTD_btopt   },  /* level 15.*/
+    { 18, 19, 19,  6,  3,128, ZSTD_btultra },  /* level 16.*/
+    { 18, 19, 19,  8,  3,256, ZSTD_btultra },  /* level 17.*/
+    { 18, 19, 19,  6,  3,128, ZSTD_btultra2},  /* level 18.*/
+    { 18, 19, 19,  8,  3,256, ZSTD_btultra2},  /* level 19.*/
+    { 18, 19, 19, 10,  3,512, ZSTD_btultra2},  /* level 20.*/
+    { 18, 19, 19, 12,  3,512, ZSTD_btultra2},  /* level 21.*/
+    { 18, 19, 19, 13,  3,999, ZSTD_btultra2},  /* level 22.*/
+},
+{   /* for srcSize <= 128 KB */
+    /* W,  C,  H,  S,  L,  T, strat */
+    { 17, 12, 12,  1,  5,  1, ZSTD_fast    },  /* base for negative levels */
+    { 17, 12, 13,  1,  6,  0, ZSTD_fast    },  /* level  1 */
+    { 17, 13, 15,  1,  5,  0, ZSTD_fast    },  /* level  2 */
+    { 17, 15, 16,  2,  5,  0, ZSTD_dfast   },  /* level  3 */
+    { 17, 17, 17,  2,  4,  0, ZSTD_dfast   },  /* level  4 */
+    { 17, 16, 17,  3,  4,  2, ZSTD_greedy  },  /* level  5 */
+    { 17, 17, 17,  3,  4,  4, ZSTD_lazy    },  /* level  6 */
+    { 17, 17, 17,  3,  4,  8, ZSTD_lazy2   },  /* level  7 */
+    { 17, 17, 17,  4,  4,  8, ZSTD_lazy2   },  /* level  8 */
+    { 17, 17, 17,  5,  4,  8, ZSTD_lazy2   },  /* level  9 */
+    { 17, 17, 17,  6,  4,  8, ZSTD_lazy2   },  /* level 10 */
+    { 17, 17, 17,  5,  4,  8, ZSTD_btlazy2 },  /* level 11 */
+    { 17, 18, 17,  7,  4, 12, ZSTD_btlazy2 },  /* level 12 */
+    { 17, 18, 17,  3,  4, 12, ZSTD_btopt   },  /* level 13.*/
+    { 17, 18, 17,  4,  3, 32, ZSTD_btopt   },  /* level 14.*/
+    { 17, 18, 17,  6,  3,256, ZSTD_btopt   },  /* level 15.*/
+    { 17, 18, 17,  6,  3,128, ZSTD_btultra },  /* level 16.*/
+    { 17, 18, 17,  8,  3,256, ZSTD_btultra },  /* level 17.*/
+    { 17, 18, 17, 10,  3,512, ZSTD_btultra },  /* level 18.*/
+    { 17, 18, 17,  5,  3,256, ZSTD_btultra2},  /* level 19.*/
+    { 17, 18, 17,  7,  3,512, ZSTD_btultra2},  /* level 20.*/
+    { 17, 18, 17,  9,  3,512, ZSTD_btultra2},  /* level 21.*/
+    { 17, 18, 17, 11,  3,999, ZSTD_btultra2},  /* level 22.*/
+},
+{   /* for srcSize <= 16 KB */
+    /* W,  C,  H,  S,  L,  T, strat */
+    { 14, 12, 13,  1,  5,  1, ZSTD_fast    },  /* base for negative levels */
+    { 14, 14, 15,  1,  5,  0, ZSTD_fast    },  /* level  1 */
+    { 14, 14, 15,  1,  4,  0, ZSTD_fast    },  /* level  2 */
+    { 14, 14, 15,  2,  4,  0, ZSTD_dfast   },  /* level  3 */
+    { 14, 14, 14,  4,  4,  2, ZSTD_greedy  },  /* level  4 */
+    { 14, 14, 14,  3,  4,  4, ZSTD_lazy    },  /* level  5.*/
+    { 14, 14, 14,  4,  4,  8, ZSTD_lazy2   },  /* level  6 */
+    { 14, 14, 14,  6,  4,  8, ZSTD_lazy2   },  /* level  7 */
+    { 14, 14, 14,  8,  4,  8, ZSTD_lazy2   },  /* level  8.*/
+    { 14, 15, 14,  5,  4,  8, ZSTD_btlazy2 },  /* level  9.*/
+    { 14, 15, 14,  9,  4,  8, ZSTD_btlazy2 },  /* level 10.*/
+    { 14, 15, 14,  3,  4, 12, ZSTD_btopt   },  /* level 11.*/
+    { 14, 15, 14,  4,  3, 24, ZSTD_btopt   },  /* level 12.*/
+    { 14, 15, 14,  5,  3, 32, ZSTD_btultra },  /* level 13.*/
+    { 14, 15, 15,  6,  3, 64, ZSTD_btultra },  /* level 14.*/
+    { 14, 15, 15,  7,  3,256, ZSTD_btultra },  /* level 15.*/
+    { 14, 15, 15,  5,  3, 48, ZSTD_btultra2},  /* level 16.*/
+    { 14, 15, 15,  6,  3,128, ZSTD_btultra2},  /* level 17.*/
+    { 14, 15, 15,  7,  3,256, ZSTD_btultra2},  /* level 18.*/
+    { 14, 15, 15,  8,  3,256, ZSTD_btultra2},  /* level 19.*/
+    { 14, 15, 15,  8,  3,512, ZSTD_btultra2},  /* level 20.*/
+    { 14, 15, 15,  9,  3,512, ZSTD_btultra2},  /* level 21.*/
+    { 14, 15, 15, 10,  3,999, ZSTD_btultra2},  /* level 22.*/
+},
+};
+
+static ZSTD_compressionParameters ZSTD_dedicatedDictSearch_getCParams(int const compressionLevel, size_t const dictSize)
+{
+    ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, 0, dictSize, ZSTD_cpm_createCDict);
+    switch (cParams.strategy) {
+        case ZSTD_fast:
+        case ZSTD_dfast:
+            break;
+        case ZSTD_greedy:
+        case ZSTD_lazy:
+        case ZSTD_lazy2:
+            cParams.hashLog += ZSTD_LAZY_DDSS_BUCKET_LOG;
+            break;
+        case ZSTD_btlazy2:
+        case ZSTD_btopt:
+        case ZSTD_btultra:
+        case ZSTD_btultra2:
+            break;
+    }
+    return cParams;
+}
+
+static int ZSTD_dedicatedDictSearch_isSupported(
+        ZSTD_compressionParameters const* cParams)
+{
+    return (cParams->strategy >= ZSTD_greedy)
+        && (cParams->strategy <= ZSTD_lazy2)
+        && (cParams->hashLog >= cParams->chainLog)
+        && (cParams->chainLog <= 24);
+}
+
+/*
+ * Reverses the adjustment applied to cparams when enabling dedicated dict
+ * search. This is used to recover the params set to be used in the working
+ * context. (Otherwise, those tables would also grow.)
+ */
+static void ZSTD_dedicatedDictSearch_revertCParams(
+        ZSTD_compressionParameters* cParams) {
+    switch (cParams->strategy) {
+        case ZSTD_fast:
+        case ZSTD_dfast:
+            break;
+        case ZSTD_greedy:
+        case ZSTD_lazy:
+        case ZSTD_lazy2:
+            cParams->hashLog -= ZSTD_LAZY_DDSS_BUCKET_LOG;
+            break;
+        case ZSTD_btlazy2:
+        case ZSTD_btopt:
+        case ZSTD_btultra:
+        case ZSTD_btultra2:
+            break;
+    }
+}
+
+static U64 ZSTD_getCParamRowSize(U64 srcSizeHint, size_t dictSize, ZSTD_cParamMode_e mode)
+{
+    switch (mode) {
+    case ZSTD_cpm_unknown:
+    case ZSTD_cpm_noAttachDict:
+    case ZSTD_cpm_createCDict:
+        break;
+    case ZSTD_cpm_attachDict:
+        dictSize = 0;
+        break;
+    default:
+        assert(0);
+        break;
+    }
+    {   int const unknown = srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN;
+        size_t const addedSize = unknown && dictSize > 0 ? 500 : 0;
+        return unknown && dictSize == 0 ? ZSTD_CONTENTSIZE_UNKNOWN : srcSizeHint+dictSize+addedSize;
+    }
+}
+
+/*! ZSTD_getCParams_internal() :
+ * @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize.
+ *  Note: srcSizeHint 0 means 0, use ZSTD_CONTENTSIZE_UNKNOWN for unknown.
+ *        Use dictSize == 0 for unknown or unused.
+ *  Note: `mode` controls how we treat the `dictSize`. See docs for `ZSTD_cParamMode_e`. */
+static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize, ZSTD_cParamMode_e mode)
+{
+    U64 const rSize = ZSTD_getCParamRowSize(srcSizeHint, dictSize, mode);
+    U32 const tableID = (rSize <= 256 KB) + (rSize <= 128 KB) + (rSize <= 16 KB);
+    int row;
+    DEBUGLOG(5, "ZSTD_getCParams_internal (cLevel=%i)", compressionLevel);
+
+    /* row */
+    if (compressionLevel == 0) row = ZSTD_CLEVEL_DEFAULT;   /* 0 == default */
+    else if (compressionLevel < 0) row = 0;   /* entry 0 is baseline for fast mode */
+    else if (compressionLevel > ZSTD_MAX_CLEVEL) row = ZSTD_MAX_CLEVEL;
+    else row = compressionLevel;
+
+    {   ZSTD_compressionParameters cp = ZSTD_defaultCParameters[tableID][row];
+        /* acceleration factor */
+        if (compressionLevel < 0) {
+            int const clampedCompressionLevel = MAX(ZSTD_minCLevel(), compressionLevel);
+            cp.targetLength = (unsigned)(-clampedCompressionLevel);
+        }
+        /* refine parameters based on srcSize & dictSize */
+        return ZSTD_adjustCParams_internal(cp, srcSizeHint, dictSize, mode);
+    }
+}
+
+/*! ZSTD_getCParams() :
+ * @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize.
+ *  Size values are optional, provide 0 if not known or unused */
+ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize)
+{
+    if (srcSizeHint == 0) srcSizeHint = ZSTD_CONTENTSIZE_UNKNOWN;
+    return ZSTD_getCParams_internal(compressionLevel, srcSizeHint, dictSize, ZSTD_cpm_unknown);
+}
+
+/*! ZSTD_getParams() :
+ *  same idea as ZSTD_getCParams()
+ * @return a `ZSTD_parameters` structure (instead of `ZSTD_compressionParameters`).
+ *  Fields of `ZSTD_frameParameters` are set to default values */
+static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize, ZSTD_cParamMode_e mode) {
+    ZSTD_parameters params;
+    ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, srcSizeHint, dictSize, mode);
+    DEBUGLOG(5, "ZSTD_getParams (cLevel=%i)", compressionLevel);
+    ZSTD_memset(&params, 0, sizeof(params));
+    params.cParams = cParams;
+    params.fParams.contentSizeFlag = 1;
+    return params;
+}
+
+/*! ZSTD_getParams() :
+ *  same idea as ZSTD_getCParams()
+ * @return a `ZSTD_parameters` structure (instead of `ZSTD_compressionParameters`).
+ *  Fields of `ZSTD_frameParameters` are set to default values */
+ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) {
+    if (srcSizeHint == 0) srcSizeHint = ZSTD_CONTENTSIZE_UNKNOWN;
+    return ZSTD_getParams_internal(compressionLevel, srcSizeHint, dictSize, ZSTD_cpm_unknown);
+}
diff --git a/lib/zstd/compress/zstd_compress_internal.h b/lib/zstd/compress/zstd_compress_internal.h
new file mode 100644
index 0000000..685d2f9
--- /dev/null
+++ b/lib/zstd/compress/zstd_compress_internal.h
@@ -0,0 +1,1188 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* This header contains definitions
+ * that shall **only** be used by modules within lib/compress.
+ */
+
+#ifndef ZSTD_COMPRESS_H
+#define ZSTD_COMPRESS_H
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#include "../common/zstd_internal.h"
+#include "zstd_cwksp.h"
+
+
+/*-*************************************
+*  Constants
+***************************************/
+#define kSearchStrength      8
+#define HASH_READ_SIZE       8
+#define ZSTD_DUBT_UNSORTED_MARK 1   /* For btlazy2 strategy, index ZSTD_DUBT_UNSORTED_MARK==1 means "unsorted".
+                                       It could be confused for a real successor at index "1", if sorted as larger than its predecessor.
+                                       It's not a big deal though : candidate will just be sorted again.
+                                       Additionally, candidate position 1 will be lost.
+                                       But candidate 1 cannot hide a large tree of candidates, so it's a minimal loss.
+                                       The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be mishandled after table re-use with a different strategy.
+                                       This constant is required by ZSTD_compressBlock_btlazy2() and ZSTD_reduceTable_internal() */
+
+
+/*-*************************************
+*  Context memory management
+***************************************/
+typedef enum { ZSTDcs_created=0, ZSTDcs_init, ZSTDcs_ongoing, ZSTDcs_ending } ZSTD_compressionStage_e;
+typedef enum { zcss_init=0, zcss_load, zcss_flush } ZSTD_cStreamStage;
+
+typedef struct ZSTD_prefixDict_s {
+    const void* dict;
+    size_t dictSize;
+    ZSTD_dictContentType_e dictContentType;
+} ZSTD_prefixDict;
+
+typedef struct {
+    void* dictBuffer;
+    void const* dict;
+    size_t dictSize;
+    ZSTD_dictContentType_e dictContentType;
+    ZSTD_CDict* cdict;
+} ZSTD_localDict;
+
+typedef struct {
+    HUF_CElt CTable[HUF_CTABLE_SIZE_U32(255)];
+    HUF_repeat repeatMode;
+} ZSTD_hufCTables_t;
+
+typedef struct {
+    FSE_CTable offcodeCTable[FSE_CTABLE_SIZE_U32(OffFSELog, MaxOff)];
+    FSE_CTable matchlengthCTable[FSE_CTABLE_SIZE_U32(MLFSELog, MaxML)];
+    FSE_CTable litlengthCTable[FSE_CTABLE_SIZE_U32(LLFSELog, MaxLL)];
+    FSE_repeat offcode_repeatMode;
+    FSE_repeat matchlength_repeatMode;
+    FSE_repeat litlength_repeatMode;
+} ZSTD_fseCTables_t;
+
+typedef struct {
+    ZSTD_hufCTables_t huf;
+    ZSTD_fseCTables_t fse;
+} ZSTD_entropyCTables_t;
+
+typedef struct {
+    U32 off;            /* Offset code (offset + ZSTD_REP_MOVE) for the match */
+    U32 len;            /* Raw length of match */
+} ZSTD_match_t;
+
+typedef struct {
+    U32 offset;         /* Offset of sequence */
+    U32 litLength;      /* Length of literals prior to match */
+    U32 matchLength;    /* Raw length of match */
+} rawSeq;
+
+typedef struct {
+  rawSeq* seq;          /* The start of the sequences */
+  size_t pos;           /* The index in seq where reading stopped. pos <= size. */
+  size_t posInSequence; /* The position within the sequence at seq[pos] where reading
+                           stopped. posInSequence <= seq[pos].litLength + seq[pos].matchLength */
+  size_t size;          /* The number of sequences. <= capacity. */
+  size_t capacity;      /* The capacity starting from `seq` pointer */
+} rawSeqStore_t;
+
+UNUSED_ATTR static const rawSeqStore_t kNullRawSeqStore = {NULL, 0, 0, 0, 0};
+
+typedef struct {
+    int price;
+    U32 off;
+    U32 mlen;
+    U32 litlen;
+    U32 rep[ZSTD_REP_NUM];
+} ZSTD_optimal_t;
+
+typedef enum { zop_dynamic=0, zop_predef } ZSTD_OptPrice_e;
+
+typedef struct {
+    /* All tables are allocated inside cctx->workspace by ZSTD_resetCCtx_internal() */
+    unsigned* litFreq;           /* table of literals statistics, of size 256 */
+    unsigned* litLengthFreq;     /* table of litLength statistics, of size (MaxLL+1) */
+    unsigned* matchLengthFreq;   /* table of matchLength statistics, of size (MaxML+1) */
+    unsigned* offCodeFreq;       /* table of offCode statistics, of size (MaxOff+1) */
+    ZSTD_match_t* matchTable;    /* list of found matches, of size ZSTD_OPT_NUM+1 */
+    ZSTD_optimal_t* priceTable;  /* All positions tracked by optimal parser, of size ZSTD_OPT_NUM+1 */
+
+    U32  litSum;                 /* nb of literals */
+    U32  litLengthSum;           /* nb of litLength codes */
+    U32  matchLengthSum;         /* nb of matchLength codes */
+    U32  offCodeSum;             /* nb of offset codes */
+    U32  litSumBasePrice;        /* to compare to log2(litfreq) */
+    U32  litLengthSumBasePrice;  /* to compare to log2(llfreq)  */
+    U32  matchLengthSumBasePrice;/* to compare to log2(mlfreq)  */
+    U32  offCodeSumBasePrice;    /* to compare to log2(offreq)  */
+    ZSTD_OptPrice_e priceType;   /* prices can be determined dynamically, or follow a pre-defined cost structure */
+    const ZSTD_entropyCTables_t* symbolCosts;  /* pre-calculated dictionary statistics */
+    ZSTD_literalCompressionMode_e literalCompressionMode;
+} optState_t;
+
+typedef struct {
+  ZSTD_entropyCTables_t entropy;
+  U32 rep[ZSTD_REP_NUM];
+} ZSTD_compressedBlockState_t;
+
+typedef struct {
+    BYTE const* nextSrc;    /* next block here to continue on current prefix */
+    BYTE const* base;       /* All regular indexes relative to this position */
+    BYTE const* dictBase;   /* extDict indexes relative to this position */
+    U32 dictLimit;          /* below that point, need extDict */
+    U32 lowLimit;           /* below that point, no more valid data */
+} ZSTD_window_t;
+
+typedef struct ZSTD_matchState_t ZSTD_matchState_t;
+struct ZSTD_matchState_t {
+    ZSTD_window_t window;   /* State for window round buffer management */
+    U32 loadedDictEnd;      /* index of end of dictionary, within context's referential.
+                             * When loadedDictEnd != 0, a dictionary is in use, and still valid.
+                             * This relies on a mechanism to set loadedDictEnd=0 when dictionary is no longer within distance.
+                             * Such mechanism is provided within ZSTD_window_enforceMaxDist() and ZSTD_checkDictValidity().
+                             * When dict referential is copied into active context (i.e. not attached),
+                             * loadedDictEnd == dictSize, since referential starts from zero.
+                             */
+    U32 nextToUpdate;       /* index from which to continue table update */
+    U32 hashLog3;           /* dispatch table for matches of len==3 : larger == faster, more memory */
+    U32* hashTable;
+    U32* hashTable3;
+    U32* chainTable;
+    int dedicatedDictSearch;  /* Indicates whether this matchState is using the
+                               * dedicated dictionary search structure.
+                               */
+    optState_t opt;         /* optimal parser state */
+    const ZSTD_matchState_t* dictMatchState;
+    ZSTD_compressionParameters cParams;
+    const rawSeqStore_t* ldmSeqStore;
+};
+
+typedef struct {
+    ZSTD_compressedBlockState_t* prevCBlock;
+    ZSTD_compressedBlockState_t* nextCBlock;
+    ZSTD_matchState_t matchState;
+} ZSTD_blockState_t;
+
+typedef struct {
+    U32 offset;
+    U32 checksum;
+} ldmEntry_t;
+
+typedef struct {
+    BYTE const* split;
+    U32 hash;
+    U32 checksum;
+    ldmEntry_t* bucket;
+} ldmMatchCandidate_t;
+
+#define LDM_BATCH_SIZE 64
+
+typedef struct {
+    ZSTD_window_t window;   /* State for the window round buffer management */
+    ldmEntry_t* hashTable;
+    U32 loadedDictEnd;
+    BYTE* bucketOffsets;    /* Next position in bucket to insert entry */
+    size_t splitIndices[LDM_BATCH_SIZE];
+    ldmMatchCandidate_t matchCandidates[LDM_BATCH_SIZE];
+} ldmState_t;
+
+typedef struct {
+    U32 enableLdm;          /* 1 if enable long distance matching */
+    U32 hashLog;            /* Log size of hashTable */
+    U32 bucketSizeLog;      /* Log bucket size for collision resolution, at most 8 */
+    U32 minMatchLength;     /* Minimum match length */
+    U32 hashRateLog;       /* Log number of entries to skip */
+    U32 windowLog;          /* Window log for the LDM */
+} ldmParams_t;
+
+typedef struct {
+    int collectSequences;
+    ZSTD_Sequence* seqStart;
+    size_t seqIndex;
+    size_t maxSequences;
+} SeqCollector;
+
+struct ZSTD_CCtx_params_s {
+    ZSTD_format_e format;
+    ZSTD_compressionParameters cParams;
+    ZSTD_frameParameters fParams;
+
+    int compressionLevel;
+    int forceWindow;           /* force back-references to respect limit of
+                                * 1<<wLog, even for dictionary */
+    size_t targetCBlockSize;   /* Tries to fit compressed block size to be around targetCBlockSize.
+                                * No target when targetCBlockSize == 0.
+                                * There is no guarantee on compressed block size */
+    int srcSizeHint;           /* User's best guess of source size.
+                                * Hint is not valid when srcSizeHint == 0.
+                                * There is no guarantee that hint is close to actual source size */
+
+    ZSTD_dictAttachPref_e attachDictPref;
+    ZSTD_literalCompressionMode_e literalCompressionMode;
+
+    /* Multithreading: used to pass parameters to mtctx */
+    int nbWorkers;
+    size_t jobSize;
+    int overlapLog;
+    int rsyncable;
+
+    /* Long distance matching parameters */
+    ldmParams_t ldmParams;
+
+    /* Dedicated dict search algorithm trigger */
+    int enableDedicatedDictSearch;
+
+    /* Input/output buffer modes */
+    ZSTD_bufferMode_e inBufferMode;
+    ZSTD_bufferMode_e outBufferMode;
+
+    /* Sequence compression API */
+    ZSTD_sequenceFormat_e blockDelimiters;
+    int validateSequences;
+
+    /* Internal use, for createCCtxParams() and freeCCtxParams() only */
+    ZSTD_customMem customMem;
+};  /* typedef'd to ZSTD_CCtx_params within "zstd.h" */
+
+#define COMPRESS_SEQUENCES_WORKSPACE_SIZE (sizeof(unsigned) * (MaxSeq + 2))
+#define ENTROPY_WORKSPACE_SIZE (HUF_WORKSPACE_SIZE + COMPRESS_SEQUENCES_WORKSPACE_SIZE)
+
+/*
+ * Indicates whether this compression proceeds directly from user-provided
+ * source buffer to user-provided destination buffer (ZSTDb_not_buffered), or
+ * whether the context needs to buffer the input/output (ZSTDb_buffered).
+ */
+typedef enum {
+    ZSTDb_not_buffered,
+    ZSTDb_buffered
+} ZSTD_buffered_policy_e;
+
+struct ZSTD_CCtx_s {
+    ZSTD_compressionStage_e stage;
+    int cParamsChanged;                  /* == 1 if cParams(except wlog) or compression level are changed in requestedParams. Triggers transmission of new params to ZSTDMT (if available) then reset to 0. */
+    int bmi2;                            /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */
+    ZSTD_CCtx_params requestedParams;
+    ZSTD_CCtx_params appliedParams;
+    U32   dictID;
+    size_t dictContentSize;
+
+    ZSTD_cwksp workspace; /* manages buffer for dynamic allocations */
+    size_t blockSize;
+    unsigned long long pledgedSrcSizePlusOne;  /* this way, 0 (default) == unknown */
+    unsigned long long consumedSrcSize;
+    unsigned long long producedCSize;
+    struct xxh64_state xxhState;
+    ZSTD_customMem customMem;
+    ZSTD_threadPool* pool;
+    size_t staticSize;
+    SeqCollector seqCollector;
+    int isFirstBlock;
+    int initialized;
+
+    seqStore_t seqStore;      /* sequences storage ptrs */
+    ldmState_t ldmState;      /* long distance matching state */
+    rawSeq* ldmSequences;     /* Storage for the ldm output sequences */
+    size_t maxNbLdmSequences;
+    rawSeqStore_t externSeqStore; /* Mutable reference to external sequences */
+    ZSTD_blockState_t blockState;
+    U32* entropyWorkspace;  /* entropy workspace of ENTROPY_WORKSPACE_SIZE bytes */
+
+    /* Wether we are streaming or not */
+    ZSTD_buffered_policy_e bufferedPolicy;
+
+    /* streaming */
+    char*  inBuff;
+    size_t inBuffSize;
+    size_t inToCompress;
+    size_t inBuffPos;
+    size_t inBuffTarget;
+    char*  outBuff;
+    size_t outBuffSize;
+    size_t outBuffContentSize;
+    size_t outBuffFlushedSize;
+    ZSTD_cStreamStage streamStage;
+    U32    frameEnded;
+
+    /* Stable in/out buffer verification */
+    ZSTD_inBuffer expectedInBuffer;
+    size_t expectedOutBufferSize;
+
+    /* Dictionary */
+    ZSTD_localDict localDict;
+    const ZSTD_CDict* cdict;
+    ZSTD_prefixDict prefixDict;   /* single-usage dictionary */
+
+    /* Multi-threading */
+
+    /* Tracing */
+};
+
+typedef enum { ZSTD_dtlm_fast, ZSTD_dtlm_full } ZSTD_dictTableLoadMethod_e;
+
+typedef enum {
+    ZSTD_noDict = 0,
+    ZSTD_extDict = 1,
+    ZSTD_dictMatchState = 2,
+    ZSTD_dedicatedDictSearch = 3
+} ZSTD_dictMode_e;
+
+typedef enum {
+    ZSTD_cpm_noAttachDict = 0,  /* Compression with ZSTD_noDict or ZSTD_extDict.
+                                 * In this mode we use both the srcSize and the dictSize
+                                 * when selecting and adjusting parameters.
+                                 */
+    ZSTD_cpm_attachDict = 1,    /* Compression with ZSTD_dictMatchState or ZSTD_dedicatedDictSearch.
+                                 * In this mode we only take the srcSize into account when selecting
+                                 * and adjusting parameters.
+                                 */
+    ZSTD_cpm_createCDict = 2,   /* Creating a CDict.
+                                 * In this mode we take both the source size and the dictionary size
+                                 * into account when selecting and adjusting the parameters.
+                                 */
+    ZSTD_cpm_unknown = 3,       /* ZSTD_getCParams, ZSTD_getParams, ZSTD_adjustParams.
+                                 * We don't know what these parameters are for. We default to the legacy
+                                 * behavior of taking both the source size and the dict size into account
+                                 * when selecting and adjusting parameters.
+                                 */
+} ZSTD_cParamMode_e;
+
+typedef size_t (*ZSTD_blockCompressor) (
+        ZSTD_matchState_t* bs, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, ZSTD_dictMode_e dictMode);
+
+
+MEM_STATIC U32 ZSTD_LLcode(U32 litLength)
+{
+    static const BYTE LL_Code[64] = {  0,  1,  2,  3,  4,  5,  6,  7,
+                                       8,  9, 10, 11, 12, 13, 14, 15,
+                                      16, 16, 17, 17, 18, 18, 19, 19,
+                                      20, 20, 20, 20, 21, 21, 21, 21,
+                                      22, 22, 22, 22, 22, 22, 22, 22,
+                                      23, 23, 23, 23, 23, 23, 23, 23,
+                                      24, 24, 24, 24, 24, 24, 24, 24,
+                                      24, 24, 24, 24, 24, 24, 24, 24 };
+    static const U32 LL_deltaCode = 19;
+    return (litLength > 63) ? ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength];
+}
+
+/* ZSTD_MLcode() :
+ * note : mlBase = matchLength - MINMATCH;
+ *        because it's the format it's stored in seqStore->sequences */
+MEM_STATIC U32 ZSTD_MLcode(U32 mlBase)
+{
+    static const BYTE ML_Code[128] = { 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15,
+                                      16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31,
+                                      32, 32, 33, 33, 34, 34, 35, 35, 36, 36, 36, 36, 37, 37, 37, 37,
+                                      38, 38, 38, 38, 38, 38, 38, 38, 39, 39, 39, 39, 39, 39, 39, 39,
+                                      40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
+                                      41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41,
+                                      42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42,
+                                      42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42 };
+    static const U32 ML_deltaCode = 36;
+    return (mlBase > 127) ? ZSTD_highbit32(mlBase) + ML_deltaCode : ML_Code[mlBase];
+}
+
+typedef struct repcodes_s {
+    U32 rep[3];
+} repcodes_t;
+
+MEM_STATIC repcodes_t ZSTD_updateRep(U32 const rep[3], U32 const offset, U32 const ll0)
+{
+    repcodes_t newReps;
+    if (offset >= ZSTD_REP_NUM) {  /* full offset */
+        newReps.rep[2] = rep[1];
+        newReps.rep[1] = rep[0];
+        newReps.rep[0] = offset - ZSTD_REP_MOVE;
+    } else {   /* repcode */
+        U32 const repCode = offset + ll0;
+        if (repCode > 0) {  /* note : if repCode==0, no change */
+            U32 const currentOffset = (repCode==ZSTD_REP_NUM) ? (rep[0] - 1) : rep[repCode];
+            newReps.rep[2] = (repCode >= 2) ? rep[1] : rep[2];
+            newReps.rep[1] = rep[0];
+            newReps.rep[0] = currentOffset;
+        } else {   /* repCode == 0 */
+            ZSTD_memcpy(&newReps, rep, sizeof(newReps));
+        }
+    }
+    return newReps;
+}
+
+/* ZSTD_cParam_withinBounds:
+ * @return 1 if value is within cParam bounds,
+ * 0 otherwise */
+MEM_STATIC int ZSTD_cParam_withinBounds(ZSTD_cParameter cParam, int value)
+{
+    ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);
+    if (ZSTD_isError(bounds.error)) return 0;
+    if (value < bounds.lowerBound) return 0;
+    if (value > bounds.upperBound) return 0;
+    return 1;
+}
+
+/* ZSTD_noCompressBlock() :
+ * Writes uncompressed block to dst buffer from given src.
+ * Returns the size of the block */
+MEM_STATIC size_t ZSTD_noCompressBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize, U32 lastBlock)
+{
+    U32 const cBlockHeader24 = lastBlock + (((U32)bt_raw)<<1) + (U32)(srcSize << 3);
+    RETURN_ERROR_IF(srcSize + ZSTD_blockHeaderSize > dstCapacity,
+                    dstSize_tooSmall, "dst buf too small for uncompressed block");
+    MEM_writeLE24(dst, cBlockHeader24);
+    ZSTD_memcpy((BYTE*)dst + ZSTD_blockHeaderSize, src, srcSize);
+    return ZSTD_blockHeaderSize + srcSize;
+}
+
+MEM_STATIC size_t ZSTD_rleCompressBlock (void* dst, size_t dstCapacity, BYTE src, size_t srcSize, U32 lastBlock)
+{
+    BYTE* const op = (BYTE*)dst;
+    U32 const cBlockHeader = lastBlock + (((U32)bt_rle)<<1) + (U32)(srcSize << 3);
+    RETURN_ERROR_IF(dstCapacity < 4, dstSize_tooSmall, "");
+    MEM_writeLE24(op, cBlockHeader);
+    op[3] = src;
+    return 4;
+}
+
+
+/* ZSTD_minGain() :
+ * minimum compression required
+ * to generate a compress block or a compressed literals section.
+ * note : use same formula for both situations */
+MEM_STATIC size_t ZSTD_minGain(size_t srcSize, ZSTD_strategy strat)
+{
+    U32 const minlog = (strat>=ZSTD_btultra) ? (U32)(strat) - 1 : 6;
+    ZSTD_STATIC_ASSERT(ZSTD_btultra == 8);
+    assert(ZSTD_cParam_withinBounds(ZSTD_c_strategy, strat));
+    return (srcSize >> minlog) + 2;
+}
+
+MEM_STATIC int ZSTD_disableLiteralsCompression(const ZSTD_CCtx_params* cctxParams)
+{
+    switch (cctxParams->literalCompressionMode) {
+    case ZSTD_lcm_huffman:
+        return 0;
+    case ZSTD_lcm_uncompressed:
+        return 1;
+    default:
+        assert(0 /* impossible: pre-validated */);
+        ZSTD_FALLTHROUGH;
+    case ZSTD_lcm_auto:
+        return (cctxParams->cParams.strategy == ZSTD_fast) && (cctxParams->cParams.targetLength > 0);
+    }
+}
+
+/*! ZSTD_safecopyLiterals() :
+ *  memcpy() function that won't read beyond more than WILDCOPY_OVERLENGTH bytes past ilimit_w.
+ *  Only called when the sequence ends past ilimit_w, so it only needs to be optimized for single
+ *  large copies.
+ */
+static void ZSTD_safecopyLiterals(BYTE* op, BYTE const* ip, BYTE const* const iend, BYTE const* ilimit_w) {
+    assert(iend > ilimit_w);
+    if (ip <= ilimit_w) {
+        ZSTD_wildcopy(op, ip, ilimit_w - ip, ZSTD_no_overlap);
+        op += ilimit_w - ip;
+        ip = ilimit_w;
+    }
+    while (ip < iend) *op++ = *ip++;
+}
+
+/*! ZSTD_storeSeq() :
+ *  Store a sequence (litlen, litPtr, offCode and mlBase) into seqStore_t.
+ *  `offCode` : distance to match + ZSTD_REP_MOVE (values <= ZSTD_REP_MOVE are repCodes).
+ *  `mlBase` : matchLength - MINMATCH
+ *  Allowed to overread literals up to litLimit.
+*/
+HINT_INLINE UNUSED_ATTR
+void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const BYTE* literals, const BYTE* litLimit, U32 offCode, size_t mlBase)
+{
+    BYTE const* const litLimit_w = litLimit - WILDCOPY_OVERLENGTH;
+    BYTE const* const litEnd = literals + litLength;
+#if defined(DEBUGLEVEL) && (DEBUGLEVEL >= 6)
+    static const BYTE* g_start = NULL;
+    if (g_start==NULL) g_start = (const BYTE*)literals;  /* note : index only works for compression within a single segment */
+    {   U32 const pos = (U32)((const BYTE*)literals - g_start);
+        DEBUGLOG(6, "Cpos%7u :%3u literals, match%4u bytes at offCode%7u",
+               pos, (U32)litLength, (U32)mlBase+MINMATCH, (U32)offCode);
+    }
+#endif
+    assert((size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart) < seqStorePtr->maxNbSeq);
+    /* copy Literals */
+    assert(seqStorePtr->maxNbLit <= 128 KB);
+    assert(seqStorePtr->lit + litLength <= seqStorePtr->litStart + seqStorePtr->maxNbLit);
+    assert(literals + litLength <= litLimit);
+    if (litEnd <= litLimit_w) {
+        /* Common case we can use wildcopy.
+	 * First copy 16 bytes, because literals are likely short.
+	 */
+        assert(WILDCOPY_OVERLENGTH >= 16);
+        ZSTD_copy16(seqStorePtr->lit, literals);
+        if (litLength > 16) {
+            ZSTD_wildcopy(seqStorePtr->lit+16, literals+16, (ptrdiff_t)litLength-16, ZSTD_no_overlap);
+        }
+    } else {
+        ZSTD_safecopyLiterals(seqStorePtr->lit, literals, litEnd, litLimit_w);
+    }
+    seqStorePtr->lit += litLength;
+
+    /* literal Length */
+    if (litLength>0xFFFF) {
+        assert(seqStorePtr->longLengthID == 0); /* there can only be a single long length */
+        seqStorePtr->longLengthID = 1;
+        seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
+    }
+    seqStorePtr->sequences[0].litLength = (U16)litLength;
+
+    /* match offset */
+    seqStorePtr->sequences[0].offset = offCode + 1;
+
+    /* match Length */
+    if (mlBase>0xFFFF) {
+        assert(seqStorePtr->longLengthID == 0); /* there can only be a single long length */
+        seqStorePtr->longLengthID = 2;
+        seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
+    }
+    seqStorePtr->sequences[0].matchLength = (U16)mlBase;
+
+    seqStorePtr->sequences++;
+}
+
+
+/*-*************************************
+*  Match length counter
+***************************************/
+static unsigned ZSTD_NbCommonBytes (size_t val)
+{
+    if (MEM_isLittleEndian()) {
+        if (MEM_64bits()) {
+#       if (__GNUC__ >= 4)
+            return (__builtin_ctzll((U64)val) >> 3);
+#       else
+            static const int DeBruijnBytePos[64] = { 0, 0, 0, 0, 0, 1, 1, 2,
+                                                     0, 3, 1, 3, 1, 4, 2, 7,
+                                                     0, 2, 3, 6, 1, 5, 3, 5,
+                                                     1, 3, 4, 4, 2, 5, 6, 7,
+                                                     7, 0, 1, 2, 3, 3, 4, 6,
+                                                     2, 6, 5, 5, 3, 4, 5, 6,
+                                                     7, 1, 2, 4, 6, 4, 4, 5,
+                                                     7, 2, 6, 5, 7, 6, 7, 7 };
+            return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58];
+#       endif
+        } else { /* 32 bits */
+#       if (__GNUC__ >= 3)
+            return (__builtin_ctz((U32)val) >> 3);
+#       else
+            static const int DeBruijnBytePos[32] = { 0, 0, 3, 0, 3, 1, 3, 0,
+                                                     3, 2, 2, 1, 3, 2, 0, 1,
+                                                     3, 3, 1, 2, 2, 2, 2, 0,
+                                                     3, 1, 2, 0, 1, 0, 1, 1 };
+            return DeBruijnBytePos[((U32)((val & -(S32)val) * 0x077CB531U)) >> 27];
+#       endif
+        }
+    } else {  /* Big Endian CPU */
+        if (MEM_64bits()) {
+#       if (__GNUC__ >= 4)
+            return (__builtin_clzll(val) >> 3);
+#       else
+            unsigned r;
+            const unsigned n32 = sizeof(size_t)*4;   /* calculate this way due to compiler complaining in 32-bits mode */
+            if (!(val>>n32)) { r=4; } else { r=0; val>>=n32; }
+            if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; }
+            r += (!val);
+            return r;
+#       endif
+        } else { /* 32 bits */
+#       if (__GNUC__ >= 3)
+            return (__builtin_clz((U32)val) >> 3);
+#       else
+            unsigned r;
+            if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; }
+            r += (!val);
+            return r;
+#       endif
+    }   }
+}
+
+
+MEM_STATIC size_t ZSTD_count(const BYTE* pIn, const BYTE* pMatch, const BYTE* const pInLimit)
+{
+    const BYTE* const pStart = pIn;
+    const BYTE* const pInLoopLimit = pInLimit - (sizeof(size_t)-1);
+
+    if (pIn < pInLoopLimit) {
+        { size_t const diff = MEM_readST(pMatch) ^ MEM_readST(pIn);
+          if (diff) return ZSTD_NbCommonBytes(diff); }
+        pIn+=sizeof(size_t); pMatch+=sizeof(size_t);
+        while (pIn < pInLoopLimit) {
+            size_t const diff = MEM_readST(pMatch) ^ MEM_readST(pIn);
+            if (!diff) { pIn+=sizeof(size_t); pMatch+=sizeof(size_t); continue; }
+            pIn += ZSTD_NbCommonBytes(diff);
+            return (size_t)(pIn - pStart);
+    }   }
+    if (MEM_64bits() && (pIn<(pInLimit-3)) && (MEM_read32(pMatch) == MEM_read32(pIn))) { pIn+=4; pMatch+=4; }
+    if ((pIn<(pInLimit-1)) && (MEM_read16(pMatch) == MEM_read16(pIn))) { pIn+=2; pMatch+=2; }
+    if ((pIn<pInLimit) && (*pMatch == *pIn)) pIn++;
+    return (size_t)(pIn - pStart);
+}
+
+/* ZSTD_count_2segments() :
+ *  can count match length with `ip` & `match` in 2 different segments.
+ *  convention : on reaching mEnd, match count continue starting from iStart
+ */
+MEM_STATIC size_t
+ZSTD_count_2segments(const BYTE* ip, const BYTE* match,
+                     const BYTE* iEnd, const BYTE* mEnd, const BYTE* iStart)
+{
+    const BYTE* const vEnd = MIN( ip + (mEnd - match), iEnd);
+    size_t const matchLength = ZSTD_count(ip, match, vEnd);
+    if (match + matchLength != mEnd) return matchLength;
+    DEBUGLOG(7, "ZSTD_count_2segments: found a 2-parts match (current length==%zu)", matchLength);
+    DEBUGLOG(7, "distance from match beginning to end dictionary = %zi", mEnd - match);
+    DEBUGLOG(7, "distance from current pos to end buffer = %zi", iEnd - ip);
+    DEBUGLOG(7, "next byte : ip==%02X, istart==%02X", ip[matchLength], *iStart);
+    DEBUGLOG(7, "final match length = %zu", matchLength + ZSTD_count(ip+matchLength, iStart, iEnd));
+    return matchLength + ZSTD_count(ip+matchLength, iStart, iEnd);
+}
+
+
+/*-*************************************
+ *  Hashes
+ ***************************************/
+static const U32 prime3bytes = 506832829U;
+static U32    ZSTD_hash3(U32 u, U32 h) { return ((u << (32-24)) * prime3bytes)  >> (32-h) ; }
+MEM_STATIC size_t ZSTD_hash3Ptr(const void* ptr, U32 h) { return ZSTD_hash3(MEM_readLE32(ptr), h); } /* only in zstd_opt.h */
+
+static const U32 prime4bytes = 2654435761U;
+static U32    ZSTD_hash4(U32 u, U32 h) { return (u * prime4bytes) >> (32-h) ; }
+static size_t ZSTD_hash4Ptr(const void* ptr, U32 h) { return ZSTD_hash4(MEM_read32(ptr), h); }
+
+static const U64 prime5bytes = 889523592379ULL;
+static size_t ZSTD_hash5(U64 u, U32 h) { return (size_t)(((u  << (64-40)) * prime5bytes) >> (64-h)) ; }
+static size_t ZSTD_hash5Ptr(const void* p, U32 h) { return ZSTD_hash5(MEM_readLE64(p), h); }
+
+static const U64 prime6bytes = 227718039650203ULL;
+static size_t ZSTD_hash6(U64 u, U32 h) { return (size_t)(((u  << (64-48)) * prime6bytes) >> (64-h)) ; }
+static size_t ZSTD_hash6Ptr(const void* p, U32 h) { return ZSTD_hash6(MEM_readLE64(p), h); }
+
+static const U64 prime7bytes = 58295818150454627ULL;
+static size_t ZSTD_hash7(U64 u, U32 h) { return (size_t)(((u  << (64-56)) * prime7bytes) >> (64-h)) ; }
+static size_t ZSTD_hash7Ptr(const void* p, U32 h) { return ZSTD_hash7(MEM_readLE64(p), h); }
+
+static const U64 prime8bytes = 0xCF1BBCDCB7A56463ULL;
+static size_t ZSTD_hash8(U64 u, U32 h) { return (size_t)(((u) * prime8bytes) >> (64-h)) ; }
+static size_t ZSTD_hash8Ptr(const void* p, U32 h) { return ZSTD_hash8(MEM_readLE64(p), h); }
+
+MEM_STATIC FORCE_INLINE_ATTR
+size_t ZSTD_hashPtr(const void* p, U32 hBits, U32 mls)
+{
+    switch(mls)
+    {
+    default:
+    case 4: return ZSTD_hash4Ptr(p, hBits);
+    case 5: return ZSTD_hash5Ptr(p, hBits);
+    case 6: return ZSTD_hash6Ptr(p, hBits);
+    case 7: return ZSTD_hash7Ptr(p, hBits);
+    case 8: return ZSTD_hash8Ptr(p, hBits);
+    }
+}
+
+/* ZSTD_ipow() :
+ * Return base^exponent.
+ */
+static U64 ZSTD_ipow(U64 base, U64 exponent)
+{
+    U64 power = 1;
+    while (exponent) {
+      if (exponent & 1) power *= base;
+      exponent >>= 1;
+      base *= base;
+    }
+    return power;
+}
+
+#define ZSTD_ROLL_HASH_CHAR_OFFSET 10
+
+/* ZSTD_rollingHash_append() :
+ * Add the buffer to the hash value.
+ */
+static U64 ZSTD_rollingHash_append(U64 hash, void const* buf, size_t size)
+{
+    BYTE const* istart = (BYTE const*)buf;
+    size_t pos;
+    for (pos = 0; pos < size; ++pos) {
+        hash *= prime8bytes;
+        hash += istart[pos] + ZSTD_ROLL_HASH_CHAR_OFFSET;
+    }
+    return hash;
+}
+
+/* ZSTD_rollingHash_compute() :
+ * Compute the rolling hash value of the buffer.
+ */
+MEM_STATIC U64 ZSTD_rollingHash_compute(void const* buf, size_t size)
+{
+    return ZSTD_rollingHash_append(0, buf, size);
+}
+
+/* ZSTD_rollingHash_primePower() :
+ * Compute the primePower to be passed to ZSTD_rollingHash_rotate() for a hash
+ * over a window of length bytes.
+ */
+MEM_STATIC U64 ZSTD_rollingHash_primePower(U32 length)
+{
+    return ZSTD_ipow(prime8bytes, length - 1);
+}
+
+/* ZSTD_rollingHash_rotate() :
+ * Rotate the rolling hash by one byte.
+ */
+MEM_STATIC U64 ZSTD_rollingHash_rotate(U64 hash, BYTE toRemove, BYTE toAdd, U64 primePower)
+{
+    hash -= (toRemove + ZSTD_ROLL_HASH_CHAR_OFFSET) * primePower;
+    hash *= prime8bytes;
+    hash += toAdd + ZSTD_ROLL_HASH_CHAR_OFFSET;
+    return hash;
+}
+
+/*-*************************************
+*  Round buffer management
+***************************************/
+#if (ZSTD_WINDOWLOG_MAX_64 > 31)
+# error "ZSTD_WINDOWLOG_MAX is too large : would overflow ZSTD_CURRENT_MAX"
+#endif
+/* Max current allowed */
+#define ZSTD_CURRENT_MAX ((3U << 29) + (1U << ZSTD_WINDOWLOG_MAX))
+/* Maximum chunk size before overflow correction needs to be called again */
+#define ZSTD_CHUNKSIZE_MAX                                                     \
+    ( ((U32)-1)                  /* Maximum ending current index */            \
+    - ZSTD_CURRENT_MAX)          /* Maximum beginning lowLimit */
+
+/*
+ * ZSTD_window_clear():
+ * Clears the window containing the history by simply setting it to empty.
+ */
+MEM_STATIC void ZSTD_window_clear(ZSTD_window_t* window)
+{
+    size_t const endT = (size_t)(window->nextSrc - window->base);
+    U32 const end = (U32)endT;
+
+    window->lowLimit = end;
+    window->dictLimit = end;
+}
+
+/*
+ * ZSTD_window_hasExtDict():
+ * Returns non-zero if the window has a non-empty extDict.
+ */
+MEM_STATIC U32 ZSTD_window_hasExtDict(ZSTD_window_t const window)
+{
+    return window.lowLimit < window.dictLimit;
+}
+
+/*
+ * ZSTD_matchState_dictMode():
+ * Inspects the provided matchState and figures out what dictMode should be
+ * passed to the compressor.
+ */
+MEM_STATIC ZSTD_dictMode_e ZSTD_matchState_dictMode(const ZSTD_matchState_t *ms)
+{
+    return ZSTD_window_hasExtDict(ms->window) ?
+        ZSTD_extDict :
+        ms->dictMatchState != NULL ?
+            (ms->dictMatchState->dedicatedDictSearch ? ZSTD_dedicatedDictSearch : ZSTD_dictMatchState) :
+            ZSTD_noDict;
+}
+
+/*
+ * ZSTD_window_needOverflowCorrection():
+ * Returns non-zero if the indices are getting too large and need overflow
+ * protection.
+ */
+MEM_STATIC U32 ZSTD_window_needOverflowCorrection(ZSTD_window_t const window,
+                                                  void const* srcEnd)
+{
+    U32 const curr = (U32)((BYTE const*)srcEnd - window.base);
+    return curr > ZSTD_CURRENT_MAX;
+}
+
+/*
+ * ZSTD_window_correctOverflow():
+ * Reduces the indices to protect from index overflow.
+ * Returns the correction made to the indices, which must be applied to every
+ * stored index.
+ *
+ * The least significant cycleLog bits of the indices must remain the same,
+ * which may be 0. Every index up to maxDist in the past must be valid.
+ * NOTE: (maxDist & cycleMask) must be zero.
+ */
+MEM_STATIC U32 ZSTD_window_correctOverflow(ZSTD_window_t* window, U32 cycleLog,
+                                           U32 maxDist, void const* src)
+{
+    /* preemptive overflow correction:
+     * 1. correction is large enough:
+     *    lowLimit > (3<<29) ==> current > 3<<29 + 1<<windowLog
+     *    1<<windowLog <= newCurrent < 1<<chainLog + 1<<windowLog
+     *
+     *    current - newCurrent
+     *    > (3<<29 + 1<<windowLog) - (1<<windowLog + 1<<chainLog)
+     *    > (3<<29) - (1<<chainLog)
+     *    > (3<<29) - (1<<30)             (NOTE: chainLog <= 30)
+     *    > 1<<29
+     *
+     * 2. (ip+ZSTD_CHUNKSIZE_MAX - cctx->base) doesn't overflow:
+     *    After correction, current is less than (1<<chainLog + 1<<windowLog).
+     *    In 64-bit mode we are safe, because we have 64-bit ptrdiff_t.
+     *    In 32-bit mode we are safe, because (chainLog <= 29), so
+     *    ip+ZSTD_CHUNKSIZE_MAX - cctx->base < 1<<32.
+     * 3. (cctx->lowLimit + 1<<windowLog) < 1<<32:
+     *    windowLog <= 31 ==> 3<<29 + 1<<windowLog < 7<<29 < 1<<32.
+     */
+    U32 const cycleMask = (1U << cycleLog) - 1;
+    U32 const curr = (U32)((BYTE const*)src - window->base);
+    U32 const currentCycle0 = curr & cycleMask;
+    /* Exclude zero so that newCurrent - maxDist >= 1. */
+    U32 const currentCycle1 = currentCycle0 == 0 ? (1U << cycleLog) : currentCycle0;
+    U32 const newCurrent = currentCycle1 + maxDist;
+    U32 const correction = curr - newCurrent;
+    assert((maxDist & cycleMask) == 0);
+    assert(curr > newCurrent);
+    /* Loose bound, should be around 1<<29 (see above) */
+    assert(correction > 1<<28);
+
+    window->base += correction;
+    window->dictBase += correction;
+    if (window->lowLimit <= correction) window->lowLimit = 1;
+    else window->lowLimit -= correction;
+    if (window->dictLimit <= correction) window->dictLimit = 1;
+    else window->dictLimit -= correction;
+
+    /* Ensure we can still reference the full window. */
+    assert(newCurrent >= maxDist);
+    assert(newCurrent - maxDist >= 1);
+    /* Ensure that lowLimit and dictLimit didn't underflow. */
+    assert(window->lowLimit <= newCurrent);
+    assert(window->dictLimit <= newCurrent);
+
+    DEBUGLOG(4, "Correction of 0x%x bytes to lowLimit=0x%x", correction,
+             window->lowLimit);
+    return correction;
+}
+
+/*
+ * ZSTD_window_enforceMaxDist():
+ * Updates lowLimit so that:
+ *    (srcEnd - base) - lowLimit == maxDist + loadedDictEnd
+ *
+ * It ensures index is valid as long as index >= lowLimit.
+ * This must be called before a block compression call.
+ *
+ * loadedDictEnd is only defined if a dictionary is in use for current compression.
+ * As the name implies, loadedDictEnd represents the index at end of dictionary.
+ * The value lies within context's referential, it can be directly compared to blockEndIdx.
+ *
+ * If loadedDictEndPtr is NULL, no dictionary is in use, and we use loadedDictEnd == 0.
+ * If loadedDictEndPtr is not NULL, we set it to zero after updating lowLimit.
+ * This is because dictionaries are allowed to be referenced fully
+ * as long as the last byte of the dictionary is in the window.
+ * Once input has progressed beyond window size, dictionary cannot be referenced anymore.
+ *
+ * In normal dict mode, the dictionary lies between lowLimit and dictLimit.
+ * In dictMatchState mode, lowLimit and dictLimit are the same,
+ * and the dictionary is below them.
+ * forceWindow and dictMatchState are therefore incompatible.
+ */
+MEM_STATIC void
+ZSTD_window_enforceMaxDist(ZSTD_window_t* window,
+                     const void* blockEnd,
+                           U32   maxDist,
+                           U32*  loadedDictEndPtr,
+                     const ZSTD_matchState_t** dictMatchStatePtr)
+{
+    U32 const blockEndIdx = (U32)((BYTE const*)blockEnd - window->base);
+    U32 const loadedDictEnd = (loadedDictEndPtr != NULL) ? *loadedDictEndPtr : 0;
+    DEBUGLOG(5, "ZSTD_window_enforceMaxDist: blockEndIdx=%u, maxDist=%u, loadedDictEnd=%u",
+                (unsigned)blockEndIdx, (unsigned)maxDist, (unsigned)loadedDictEnd);
+
+    /* - When there is no dictionary : loadedDictEnd == 0.
+         In which case, the test (blockEndIdx > maxDist) is merely to avoid
+         overflowing next operation `newLowLimit = blockEndIdx - maxDist`.
+       - When there is a standard dictionary :
+         Index referential is copied from the dictionary,
+         which means it starts from 0.
+         In which case, loadedDictEnd == dictSize,
+         and it makes sense to compare `blockEndIdx > maxDist + dictSize`
+         since `blockEndIdx` also starts from zero.
+       - When there is an attached dictionary :
+         loadedDictEnd is expressed within the referential of the context,
+         so it can be directly compared against blockEndIdx.
+    */
+    if (blockEndIdx > maxDist + loadedDictEnd) {
+        U32 const newLowLimit = blockEndIdx - maxDist;
+        if (window->lowLimit < newLowLimit) window->lowLimit = newLowLimit;
+        if (window->dictLimit < window->lowLimit) {
+            DEBUGLOG(5, "Update dictLimit to match lowLimit, from %u to %u",
+                        (unsigned)window->dictLimit, (unsigned)window->lowLimit);
+            window->dictLimit = window->lowLimit;
+        }
+        /* On reaching window size, dictionaries are invalidated */
+        if (loadedDictEndPtr) *loadedDictEndPtr = 0;
+        if (dictMatchStatePtr) *dictMatchStatePtr = NULL;
+    }
+}
+
+/* Similar to ZSTD_window_enforceMaxDist(),
+ * but only invalidates dictionary
+ * when input progresses beyond window size.
+ * assumption : loadedDictEndPtr and dictMatchStatePtr are valid (non NULL)
+ *              loadedDictEnd uses same referential as window->base
+ *              maxDist is the window size */
+MEM_STATIC void
+ZSTD_checkDictValidity(const ZSTD_window_t* window,
+                       const void* blockEnd,
+                             U32   maxDist,
+                             U32*  loadedDictEndPtr,
+                       const ZSTD_matchState_t** dictMatchStatePtr)
+{
+    assert(loadedDictEndPtr != NULL);
+    assert(dictMatchStatePtr != NULL);
+    {   U32 const blockEndIdx = (U32)((BYTE const*)blockEnd - window->base);
+        U32 const loadedDictEnd = *loadedDictEndPtr;
+        DEBUGLOG(5, "ZSTD_checkDictValidity: blockEndIdx=%u, maxDist=%u, loadedDictEnd=%u",
+                    (unsigned)blockEndIdx, (unsigned)maxDist, (unsigned)loadedDictEnd);
+        assert(blockEndIdx >= loadedDictEnd);
+
+        if (blockEndIdx > loadedDictEnd + maxDist) {
+            /* On reaching window size, dictionaries are invalidated.
+             * For simplification, if window size is reached anywhere within next block,
+             * the dictionary is invalidated for the full block.
+             */
+            DEBUGLOG(6, "invalidating dictionary for current block (distance > windowSize)");
+            *loadedDictEndPtr = 0;
+            *dictMatchStatePtr = NULL;
+        } else {
+            if (*loadedDictEndPtr != 0) {
+                DEBUGLOG(6, "dictionary considered valid for current block");
+    }   }   }
+}
+
+MEM_STATIC void ZSTD_window_init(ZSTD_window_t* window) {
+    ZSTD_memset(window, 0, sizeof(*window));
+    window->base = (BYTE const*)"";
+    window->dictBase = (BYTE const*)"";
+    window->dictLimit = 1;    /* start from 1, so that 1st position is valid */
+    window->lowLimit = 1;     /* it ensures first and later CCtx usages compress the same */
+    window->nextSrc = window->base + 1;   /* see issue #1241 */
+}
+
+/*
+ * ZSTD_window_update():
+ * Updates the window by appending [src, src + srcSize) to the window.
+ * If it is not contiguous, the current prefix becomes the extDict, and we
+ * forget about the extDict. Handles overlap of the prefix and extDict.
+ * Returns non-zero if the segment is contiguous.
+ */
+MEM_STATIC U32 ZSTD_window_update(ZSTD_window_t* window,
+                                  void const* src, size_t srcSize)
+{
+    BYTE const* const ip = (BYTE const*)src;
+    U32 contiguous = 1;
+    DEBUGLOG(5, "ZSTD_window_update");
+    if (srcSize == 0)
+        return contiguous;
+    assert(window->base != NULL);
+    assert(window->dictBase != NULL);
+    /* Check if blocks follow each other */
+    if (src != window->nextSrc) {
+        /* not contiguous */
+        size_t const distanceFromBase = (size_t)(window->nextSrc - window->base);
+        DEBUGLOG(5, "Non contiguous blocks, new segment starts at %u", window->dictLimit);
+        window->lowLimit = window->dictLimit;
+        assert(distanceFromBase == (size_t)(U32)distanceFromBase);  /* should never overflow */
+        window->dictLimit = (U32)distanceFromBase;
+        window->dictBase = window->base;
+        window->base = ip - distanceFromBase;
+        /* ms->nextToUpdate = window->dictLimit; */
+        if (window->dictLimit - window->lowLimit < HASH_READ_SIZE) window->lowLimit = window->dictLimit;   /* too small extDict */
+        contiguous = 0;
+    }
+    window->nextSrc = ip + srcSize;
+    /* if input and dictionary overlap : reduce dictionary (area presumed modified by input) */
+    if ( (ip+srcSize > window->dictBase + window->lowLimit)
+       & (ip < window->dictBase + window->dictLimit)) {
+        ptrdiff_t const highInputIdx = (ip + srcSize) - window->dictBase;
+        U32 const lowLimitMax = (highInputIdx > (ptrdiff_t)window->dictLimit) ? window->dictLimit : (U32)highInputIdx;
+        window->lowLimit = lowLimitMax;
+        DEBUGLOG(5, "Overlapping extDict and input : new lowLimit = %u", window->lowLimit);
+    }
+    return contiguous;
+}
+
+/*
+ * Returns the lowest allowed match index. It may either be in the ext-dict or the prefix.
+ */
+MEM_STATIC U32 ZSTD_getLowestMatchIndex(const ZSTD_matchState_t* ms, U32 curr, unsigned windowLog)
+{
+    U32    const maxDistance = 1U << windowLog;
+    U32    const lowestValid = ms->window.lowLimit;
+    U32    const withinWindow = (curr - lowestValid > maxDistance) ? curr - maxDistance : lowestValid;
+    U32    const isDictionary = (ms->loadedDictEnd != 0);
+    /* When using a dictionary the entire dictionary is valid if a single byte of the dictionary
+     * is within the window. We invalidate the dictionary (and set loadedDictEnd to 0) when it isn't
+     * valid for the entire block. So this check is sufficient to find the lowest valid match index.
+     */
+    U32    const matchLowest = isDictionary ? lowestValid : withinWindow;
+    return matchLowest;
+}
+
+/*
+ * Returns the lowest allowed match index in the prefix.
+ */
+MEM_STATIC U32 ZSTD_getLowestPrefixIndex(const ZSTD_matchState_t* ms, U32 curr, unsigned windowLog)
+{
+    U32    const maxDistance = 1U << windowLog;
+    U32    const lowestValid = ms->window.dictLimit;
+    U32    const withinWindow = (curr - lowestValid > maxDistance) ? curr - maxDistance : lowestValid;
+    U32    const isDictionary = (ms->loadedDictEnd != 0);
+    /* When computing the lowest prefix index we need to take the dictionary into account to handle
+     * the edge case where the dictionary and the source are contiguous in memory.
+     */
+    U32    const matchLowest = isDictionary ? lowestValid : withinWindow;
+    return matchLowest;
+}
+
+
+
+/* debug functions */
+#if (DEBUGLEVEL>=2)
+
+MEM_STATIC double ZSTD_fWeight(U32 rawStat)
+{
+    U32 const fp_accuracy = 8;
+    U32 const fp_multiplier = (1 << fp_accuracy);
+    U32 const newStat = rawStat + 1;
+    U32 const hb = ZSTD_highbit32(newStat);
+    U32 const BWeight = hb * fp_multiplier;
+    U32 const FWeight = (newStat << fp_accuracy) >> hb;
+    U32 const weight = BWeight + FWeight;
+    assert(hb + fp_accuracy < 31);
+    return (double)weight / fp_multiplier;
+}
+
+/* display a table content,
+ * listing each element, its frequency, and its predicted bit cost */
+MEM_STATIC void ZSTD_debugTable(const U32* table, U32 max)
+{
+    unsigned u, sum;
+    for (u=0, sum=0; u<=max; u++) sum += table[u];
+    DEBUGLOG(2, "total nb elts: %u", sum);
+    for (u=0; u<=max; u++) {
+        DEBUGLOG(2, "%2u: %5u  (%.2f)",
+                u, table[u], ZSTD_fWeight(sum) - ZSTD_fWeight(table[u]) );
+    }
+}
+
+#endif
+
+
+
+/* ===============================================================
+ * Shared internal declarations
+ * These prototypes may be called from sources not in lib/compress
+ * =============================================================== */
+
+/* ZSTD_loadCEntropy() :
+ * dict : must point at beginning of a valid zstd dictionary.
+ * return : size of dictionary header (size of magic number + dict ID + entropy tables)
+ * assumptions : magic number supposed already checked
+ *               and dictSize >= 8 */
+size_t ZSTD_loadCEntropy(ZSTD_compressedBlockState_t* bs, void* workspace,
+                         const void* const dict, size_t dictSize);
+
+void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs);
+
+/* ==============================================================
+ * Private declarations
+ * These prototypes shall only be called from within lib/compress
+ * ============================================================== */
+
+/* ZSTD_getCParamsFromCCtxParams() :
+ * cParams are built depending on compressionLevel, src size hints,
+ * LDM and manually set compression parameters.
+ * Note: srcSizeHint == 0 means 0!
+ */
+ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
+        const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize, ZSTD_cParamMode_e mode);
+
+/*! ZSTD_initCStream_internal() :
+ *  Private use only. Init streaming operation.
+ *  expects params to be valid.
+ *  must receive dict, or cdict, or none, but not both.
+ *  @return : 0, or an error code */
+size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs,
+                     const void* dict, size_t dictSize,
+                     const ZSTD_CDict* cdict,
+                     const ZSTD_CCtx_params* params, unsigned long long pledgedSrcSize);
+
+void ZSTD_resetSeqStore(seqStore_t* ssPtr);
+
+/*! ZSTD_getCParamsFromCDict() :
+ *  as the name implies */
+ZSTD_compressionParameters ZSTD_getCParamsFromCDict(const ZSTD_CDict* cdict);
+
+/* ZSTD_compressBegin_advanced_internal() :
+ * Private use only. To be called from zstdmt_compress.c. */
+size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx,
+                                    const void* dict, size_t dictSize,
+                                    ZSTD_dictContentType_e dictContentType,
+                                    ZSTD_dictTableLoadMethod_e dtlm,
+                                    const ZSTD_CDict* cdict,
+                                    const ZSTD_CCtx_params* params,
+                                    unsigned long long pledgedSrcSize);
+
+/* ZSTD_compress_advanced_internal() :
+ * Private use only. To be called from zstdmt_compress.c. */
+size_t ZSTD_compress_advanced_internal(ZSTD_CCtx* cctx,
+                                       void* dst, size_t dstCapacity,
+                                 const void* src, size_t srcSize,
+                                 const void* dict,size_t dictSize,
+                                 const ZSTD_CCtx_params* params);
+
+
+/* ZSTD_writeLastEmptyBlock() :
+ * output an empty Block with end-of-frame mark to complete a frame
+ * @return : size of data written into `dst` (== ZSTD_blockHeaderSize (defined in zstd_internal.h))
+ *           or an error code if `dstCapacity` is too small (<ZSTD_blockHeaderSize)
+ */
+size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity);
+
+
+/* ZSTD_referenceExternalSequences() :
+ * Must be called before starting a compression operation.
+ * seqs must parse a prefix of the source.
+ * This cannot be used when long range matching is enabled.
+ * Zstd will use these sequences, and pass the literals to a secondary block
+ * compressor.
+ * @return : An error code on failure.
+ * NOTE: seqs are not verified! Invalid sequences can cause out-of-bounds memory
+ * access and data corruption.
+ */
+size_t ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq);
+
+/* ZSTD_cycleLog() :
+ *  condition for correct operation : hashLog > 1 */
+U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat);
+
+/* ZSTD_CCtx_trace() :
+ *  Trace the end of a compression call.
+ */
+void ZSTD_CCtx_trace(ZSTD_CCtx* cctx, size_t extraCSize);
+
+#endif /* ZSTD_COMPRESS_H */
diff --git a/lib/zstd/compress/zstd_compress_literals.c b/lib/zstd/compress/zstd_compress_literals.c
new file mode 100644
index 0000000..655bcda
--- /dev/null
+++ b/lib/zstd/compress/zstd_compress_literals.c
@@ -0,0 +1,158 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+ /*-*************************************
+ *  Dependencies
+ ***************************************/
+#include "zstd_compress_literals.h"
+
+size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    BYTE* const ostart = (BYTE*)dst;
+    U32   const flSize = 1 + (srcSize>31) + (srcSize>4095);
+
+    RETURN_ERROR_IF(srcSize + flSize > dstCapacity, dstSize_tooSmall, "");
+
+    switch(flSize)
+    {
+        case 1: /* 2 - 1 - 5 */
+            ostart[0] = (BYTE)((U32)set_basic + (srcSize<<3));
+            break;
+        case 2: /* 2 - 2 - 12 */
+            MEM_writeLE16(ostart, (U16)((U32)set_basic + (1<<2) + (srcSize<<4)));
+            break;
+        case 3: /* 2 - 2 - 20 */
+            MEM_writeLE32(ostart, (U32)((U32)set_basic + (3<<2) + (srcSize<<4)));
+            break;
+        default:   /* not necessary : flSize is {1,2,3} */
+            assert(0);
+    }
+
+    ZSTD_memcpy(ostart + flSize, src, srcSize);
+    DEBUGLOG(5, "Raw literals: %u -> %u", (U32)srcSize, (U32)(srcSize + flSize));
+    return srcSize + flSize;
+}
+
+size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    BYTE* const ostart = (BYTE*)dst;
+    U32   const flSize = 1 + (srcSize>31) + (srcSize>4095);
+
+    (void)dstCapacity;  /* dstCapacity already guaranteed to be >=4, hence large enough */
+
+    switch(flSize)
+    {
+        case 1: /* 2 - 1 - 5 */
+            ostart[0] = (BYTE)((U32)set_rle + (srcSize<<3));
+            break;
+        case 2: /* 2 - 2 - 12 */
+            MEM_writeLE16(ostart, (U16)((U32)set_rle + (1<<2) + (srcSize<<4)));
+            break;
+        case 3: /* 2 - 2 - 20 */
+            MEM_writeLE32(ostart, (U32)((U32)set_rle + (3<<2) + (srcSize<<4)));
+            break;
+        default:   /* not necessary : flSize is {1,2,3} */
+            assert(0);
+    }
+
+    ostart[flSize] = *(const BYTE*)src;
+    DEBUGLOG(5, "RLE literals: %u -> %u", (U32)srcSize, (U32)flSize + 1);
+    return flSize+1;
+}
+
+size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
+                              ZSTD_hufCTables_t* nextHuf,
+                              ZSTD_strategy strategy, int disableLiteralCompression,
+                              void* dst, size_t dstCapacity,
+                        const void* src, size_t srcSize,
+                              void* entropyWorkspace, size_t entropyWorkspaceSize,
+                        const int bmi2)
+{
+    size_t const minGain = ZSTD_minGain(srcSize, strategy);
+    size_t const lhSize = 3 + (srcSize >= 1 KB) + (srcSize >= 16 KB);
+    BYTE*  const ostart = (BYTE*)dst;
+    U32 singleStream = srcSize < 256;
+    symbolEncodingType_e hType = set_compressed;
+    size_t cLitSize;
+
+    DEBUGLOG(5,"ZSTD_compressLiterals (disableLiteralCompression=%i srcSize=%u)",
+                disableLiteralCompression, (U32)srcSize);
+
+    /* Prepare nextEntropy assuming reusing the existing table */
+    ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
+
+    if (disableLiteralCompression)
+        return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
+
+    /* small ? don't even attempt compression (speed opt) */
+#   define COMPRESS_LITERALS_SIZE_MIN 63
+    {   size_t const minLitSize = (prevHuf->repeatMode == HUF_repeat_valid) ? 6 : COMPRESS_LITERALS_SIZE_MIN;
+        if (srcSize <= minLitSize) return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
+    }
+
+    RETURN_ERROR_IF(dstCapacity < lhSize+1, dstSize_tooSmall, "not enough space for compression");
+    {   HUF_repeat repeat = prevHuf->repeatMode;
+        int const preferRepeat = strategy < ZSTD_lazy ? srcSize <= 1024 : 0;
+        if (repeat == HUF_repeat_valid && lhSize == 3) singleStream = 1;
+        cLitSize = singleStream ?
+            HUF_compress1X_repeat(
+                ostart+lhSize, dstCapacity-lhSize, src, srcSize,
+                HUF_SYMBOLVALUE_MAX, HUF_TABLELOG_DEFAULT, entropyWorkspace, entropyWorkspaceSize,
+                (HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2) :
+            HUF_compress4X_repeat(
+                ostart+lhSize, dstCapacity-lhSize, src, srcSize,
+                HUF_SYMBOLVALUE_MAX, HUF_TABLELOG_DEFAULT, entropyWorkspace, entropyWorkspaceSize,
+                (HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2);
+        if (repeat != HUF_repeat_none) {
+            /* reused the existing table */
+            DEBUGLOG(5, "Reusing previous huffman table");
+            hType = set_repeat;
+        }
+    }
+
+    if ((cLitSize==0) | (cLitSize >= srcSize - minGain) | ERR_isError(cLitSize)) {
+        ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
+        return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
+    }
+    if (cLitSize==1) {
+        ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
+        return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
+    }
+
+    if (hType == set_compressed) {
+        /* using a newly constructed table */
+        nextHuf->repeatMode = HUF_repeat_check;
+    }
+
+    /* Build header */
+    switch(lhSize)
+    {
+    case 3: /* 2 - 2 - 10 - 10 */
+        {   U32 const lhc = hType + ((!singleStream) << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<14);
+            MEM_writeLE24(ostart, lhc);
+            break;
+        }
+    case 4: /* 2 - 2 - 14 - 14 */
+        {   U32 const lhc = hType + (2 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<18);
+            MEM_writeLE32(ostart, lhc);
+            break;
+        }
+    case 5: /* 2 - 2 - 18 - 18 */
+        {   U32 const lhc = hType + (3 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<22);
+            MEM_writeLE32(ostart, lhc);
+            ostart[4] = (BYTE)(cLitSize >> 10);
+            break;
+        }
+    default:  /* not possible : lhSize is {3,4,5} */
+        assert(0);
+    }
+    DEBUGLOG(5, "Compressed literals: %u -> %u", (U32)srcSize, (U32)(lhSize+cLitSize));
+    return lhSize+cLitSize;
+}
diff --git a/lib/zstd/compress/zstd_compress_literals.h b/lib/zstd/compress/zstd_compress_literals.h
new file mode 100644
index 0000000..9904c0c
--- /dev/null
+++ b/lib/zstd/compress/zstd_compress_literals.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_COMPRESS_LITERALS_H
+#define ZSTD_COMPRESS_LITERALS_H
+
+#include "zstd_compress_internal.h" /* ZSTD_hufCTables_t, ZSTD_minGain() */
+
+
+size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+
+size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+
+size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
+                              ZSTD_hufCTables_t* nextHuf,
+                              ZSTD_strategy strategy, int disableLiteralCompression,
+                              void* dst, size_t dstCapacity,
+                        const void* src, size_t srcSize,
+                              void* entropyWorkspace, size_t entropyWorkspaceSize,
+                        const int bmi2);
+
+#endif /* ZSTD_COMPRESS_LITERALS_H */
diff --git a/lib/zstd/compress/zstd_compress_sequences.c b/lib/zstd/compress/zstd_compress_sequences.c
new file mode 100644
index 0000000..dcfcdc9
--- /dev/null
+++ b/lib/zstd/compress/zstd_compress_sequences.c
@@ -0,0 +1,439 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+ /*-*************************************
+ *  Dependencies
+ ***************************************/
+#include "zstd_compress_sequences.h"
+
+/*
+ * -log2(x / 256) lookup table for x in [0, 256).
+ * If x == 0: Return 0
+ * Else: Return floor(-log2(x / 256) * 256)
+ */
+static unsigned const kInverseProbabilityLog256[256] = {
+    0,    2048, 1792, 1642, 1536, 1453, 1386, 1329, 1280, 1236, 1197, 1162,
+    1130, 1100, 1073, 1047, 1024, 1001, 980,  960,  941,  923,  906,  889,
+    874,  859,  844,  830,  817,  804,  791,  779,  768,  756,  745,  734,
+    724,  714,  704,  694,  685,  676,  667,  658,  650,  642,  633,  626,
+    618,  610,  603,  595,  588,  581,  574,  567,  561,  554,  548,  542,
+    535,  529,  523,  517,  512,  506,  500,  495,  489,  484,  478,  473,
+    468,  463,  458,  453,  448,  443,  438,  434,  429,  424,  420,  415,
+    411,  407,  402,  398,  394,  390,  386,  382,  377,  373,  370,  366,
+    362,  358,  354,  350,  347,  343,  339,  336,  332,  329,  325,  322,
+    318,  315,  311,  308,  305,  302,  298,  295,  292,  289,  286,  282,
+    279,  276,  273,  270,  267,  264,  261,  258,  256,  253,  250,  247,
+    244,  241,  239,  236,  233,  230,  228,  225,  222,  220,  217,  215,
+    212,  209,  207,  204,  202,  199,  197,  194,  192,  190,  187,  185,
+    182,  180,  178,  175,  173,  171,  168,  166,  164,  162,  159,  157,
+    155,  153,  151,  149,  146,  144,  142,  140,  138,  136,  134,  132,
+    130,  128,  126,  123,  121,  119,  117,  115,  114,  112,  110,  108,
+    106,  104,  102,  100,  98,   96,   94,   93,   91,   89,   87,   85,
+    83,   82,   80,   78,   76,   74,   73,   71,   69,   67,   66,   64,
+    62,   61,   59,   57,   55,   54,   52,   50,   49,   47,   46,   44,
+    42,   41,   39,   37,   36,   34,   33,   31,   30,   28,   26,   25,
+    23,   22,   20,   19,   17,   16,   14,   13,   11,   10,   8,    7,
+    5,    4,    2,    1,
+};
+
+static unsigned ZSTD_getFSEMaxSymbolValue(FSE_CTable const* ctable) {
+  void const* ptr = ctable;
+  U16 const* u16ptr = (U16 const*)ptr;
+  U32 const maxSymbolValue = MEM_read16(u16ptr + 1);
+  return maxSymbolValue;
+}
+
+/*
+ * Returns true if we should use ncount=-1 else we should
+ * use ncount=1 for low probability symbols instead.
+ */
+static unsigned ZSTD_useLowProbCount(size_t const nbSeq)
+{
+    /* Heuristic: This should cover most blocks <= 16K and
+     * start to fade out after 16K to about 32K depending on
+     * comprssibility.
+     */
+    return nbSeq >= 2048;
+}
+
+/*
+ * Returns the cost in bytes of encoding the normalized count header.
+ * Returns an error if any of the helper functions return an error.
+ */
+static size_t ZSTD_NCountCost(unsigned const* count, unsigned const max,
+                              size_t const nbSeq, unsigned const FSELog)
+{
+    BYTE wksp[FSE_NCOUNTBOUND];
+    S16 norm[MaxSeq + 1];
+    const U32 tableLog = FSE_optimalTableLog(FSELog, nbSeq, max);
+    FORWARD_IF_ERROR(FSE_normalizeCount(norm, tableLog, count, nbSeq, max, ZSTD_useLowProbCount(nbSeq)), "");
+    return FSE_writeNCount(wksp, sizeof(wksp), norm, max, tableLog);
+}
+
+/*
+ * Returns the cost in bits of encoding the distribution described by count
+ * using the entropy bound.
+ */
+static size_t ZSTD_entropyCost(unsigned const* count, unsigned const max, size_t const total)
+{
+    unsigned cost = 0;
+    unsigned s;
+    for (s = 0; s <= max; ++s) {
+        unsigned norm = (unsigned)((256 * count[s]) / total);
+        if (count[s] != 0 && norm == 0)
+            norm = 1;
+        assert(count[s] < total);
+        cost += count[s] * kInverseProbabilityLog256[norm];
+    }
+    return cost >> 8;
+}
+
+/*
+ * Returns the cost in bits of encoding the distribution in count using ctable.
+ * Returns an error if ctable cannot represent all the symbols in count.
+ */
+size_t ZSTD_fseBitCost(
+    FSE_CTable const* ctable,
+    unsigned const* count,
+    unsigned const max)
+{
+    unsigned const kAccuracyLog = 8;
+    size_t cost = 0;
+    unsigned s;
+    FSE_CState_t cstate;
+    FSE_initCState(&cstate, ctable);
+    if (ZSTD_getFSEMaxSymbolValue(ctable) < max) {
+        DEBUGLOG(5, "Repeat FSE_CTable has maxSymbolValue %u < %u",
+                    ZSTD_getFSEMaxSymbolValue(ctable), max);
+        return ERROR(GENERIC);
+    }
+    for (s = 0; s <= max; ++s) {
+        unsigned const tableLog = cstate.stateLog;
+        unsigned const badCost = (tableLog + 1) << kAccuracyLog;
+        unsigned const bitCost = FSE_bitCost(cstate.symbolTT, tableLog, s, kAccuracyLog);
+        if (count[s] == 0)
+            continue;
+        if (bitCost >= badCost) {
+            DEBUGLOG(5, "Repeat FSE_CTable has Prob[%u] == 0", s);
+            return ERROR(GENERIC);
+        }
+        cost += (size_t)count[s] * bitCost;
+    }
+    return cost >> kAccuracyLog;
+}
+
+/*
+ * Returns the cost in bits of encoding the distribution in count using the
+ * table described by norm. The max symbol support by norm is assumed >= max.
+ * norm must be valid for every symbol with non-zero probability in count.
+ */
+size_t ZSTD_crossEntropyCost(short const* norm, unsigned accuracyLog,
+                             unsigned const* count, unsigned const max)
+{
+    unsigned const shift = 8 - accuracyLog;
+    size_t cost = 0;
+    unsigned s;
+    assert(accuracyLog <= 8);
+    for (s = 0; s <= max; ++s) {
+        unsigned const normAcc = (norm[s] != -1) ? (unsigned)norm[s] : 1;
+        unsigned const norm256 = normAcc << shift;
+        assert(norm256 > 0);
+        assert(norm256 < 256);
+        cost += count[s] * kInverseProbabilityLog256[norm256];
+    }
+    return cost >> 8;
+}
+
+symbolEncodingType_e
+ZSTD_selectEncodingType(
+        FSE_repeat* repeatMode, unsigned const* count, unsigned const max,
+        size_t const mostFrequent, size_t nbSeq, unsigned const FSELog,
+        FSE_CTable const* prevCTable,
+        short const* defaultNorm, U32 defaultNormLog,
+        ZSTD_defaultPolicy_e const isDefaultAllowed,
+        ZSTD_strategy const strategy)
+{
+    ZSTD_STATIC_ASSERT(ZSTD_defaultDisallowed == 0 && ZSTD_defaultAllowed != 0);
+    if (mostFrequent == nbSeq) {
+        *repeatMode = FSE_repeat_none;
+        if (isDefaultAllowed && nbSeq <= 2) {
+            /* Prefer set_basic over set_rle when there are 2 or less symbols,
+             * since RLE uses 1 byte, but set_basic uses 5-6 bits per symbol.
+             * If basic encoding isn't possible, always choose RLE.
+             */
+            DEBUGLOG(5, "Selected set_basic");
+            return set_basic;
+        }
+        DEBUGLOG(5, "Selected set_rle");
+        return set_rle;
+    }
+    if (strategy < ZSTD_lazy) {
+        if (isDefaultAllowed) {
+            size_t const staticFse_nbSeq_max = 1000;
+            size_t const mult = 10 - strategy;
+            size_t const baseLog = 3;
+            size_t const dynamicFse_nbSeq_min = (((size_t)1 << defaultNormLog) * mult) >> baseLog;  /* 28-36 for offset, 56-72 for lengths */
+            assert(defaultNormLog >= 5 && defaultNormLog <= 6);  /* xx_DEFAULTNORMLOG */
+            assert(mult <= 9 && mult >= 7);
+            if ( (*repeatMode == FSE_repeat_valid)
+              && (nbSeq < staticFse_nbSeq_max) ) {
+                DEBUGLOG(5, "Selected set_repeat");
+                return set_repeat;
+            }
+            if ( (nbSeq < dynamicFse_nbSeq_min)
+              || (mostFrequent < (nbSeq >> (defaultNormLog-1))) ) {
+                DEBUGLOG(5, "Selected set_basic");
+                /* The format allows default tables to be repeated, but it isn't useful.
+                 * When using simple heuristics to select encoding type, we don't want
+                 * to confuse these tables with dictionaries. When running more careful
+                 * analysis, we don't need to waste time checking both repeating tables
+                 * and default tables.
+                 */
+                *repeatMode = FSE_repeat_none;
+                return set_basic;
+            }
+        }
+    } else {
+        size_t const basicCost = isDefaultAllowed ? ZSTD_crossEntropyCost(defaultNorm, defaultNormLog, count, max) : ERROR(GENERIC);
+        size_t const repeatCost = *repeatMode != FSE_repeat_none ? ZSTD_fseBitCost(prevCTable, count, max) : ERROR(GENERIC);
+        size_t const NCountCost = ZSTD_NCountCost(count, max, nbSeq, FSELog);
+        size_t const compressedCost = (NCountCost << 3) + ZSTD_entropyCost(count, max, nbSeq);
+
+        if (isDefaultAllowed) {
+            assert(!ZSTD_isError(basicCost));
+            assert(!(*repeatMode == FSE_repeat_valid && ZSTD_isError(repeatCost)));
+        }
+        assert(!ZSTD_isError(NCountCost));
+        assert(compressedCost < ERROR(maxCode));
+        DEBUGLOG(5, "Estimated bit costs: basic=%u\trepeat=%u\tcompressed=%u",
+                    (unsigned)basicCost, (unsigned)repeatCost, (unsigned)compressedCost);
+        if (basicCost <= repeatCost && basicCost <= compressedCost) {
+            DEBUGLOG(5, "Selected set_basic");
+            assert(isDefaultAllowed);
+            *repeatMode = FSE_repeat_none;
+            return set_basic;
+        }
+        if (repeatCost <= compressedCost) {
+            DEBUGLOG(5, "Selected set_repeat");
+            assert(!ZSTD_isError(repeatCost));
+            return set_repeat;
+        }
+        assert(compressedCost < basicCost && compressedCost < repeatCost);
+    }
+    DEBUGLOG(5, "Selected set_compressed");
+    *repeatMode = FSE_repeat_check;
+    return set_compressed;
+}
+
+typedef struct {
+    S16 norm[MaxSeq + 1];
+    U32 wksp[FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(MaxSeq, MaxFSELog)];
+} ZSTD_BuildCTableWksp;
+
+size_t
+ZSTD_buildCTable(void* dst, size_t dstCapacity,
+                FSE_CTable* nextCTable, U32 FSELog, symbolEncodingType_e type,
+                unsigned* count, U32 max,
+                const BYTE* codeTable, size_t nbSeq,
+                const S16* defaultNorm, U32 defaultNormLog, U32 defaultMax,
+                const FSE_CTable* prevCTable, size_t prevCTableSize,
+                void* entropyWorkspace, size_t entropyWorkspaceSize)
+{
+    BYTE* op = (BYTE*)dst;
+    const BYTE* const oend = op + dstCapacity;
+    DEBUGLOG(6, "ZSTD_buildCTable (dstCapacity=%u)", (unsigned)dstCapacity);
+
+    switch (type) {
+    case set_rle:
+        FORWARD_IF_ERROR(FSE_buildCTable_rle(nextCTable, (BYTE)max), "");
+        RETURN_ERROR_IF(dstCapacity==0, dstSize_tooSmall, "not enough space");
+        *op = codeTable[0];
+        return 1;
+    case set_repeat:
+        ZSTD_memcpy(nextCTable, prevCTable, prevCTableSize);
+        return 0;
+    case set_basic:
+        FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, defaultNorm, defaultMax, defaultNormLog, entropyWorkspace, entropyWorkspaceSize), "");  /* note : could be pre-calculated */
+        return 0;
+    case set_compressed: {
+        ZSTD_BuildCTableWksp* wksp = (ZSTD_BuildCTableWksp*)entropyWorkspace;
+        size_t nbSeq_1 = nbSeq;
+        const U32 tableLog = FSE_optimalTableLog(FSELog, nbSeq, max);
+        if (count[codeTable[nbSeq-1]] > 1) {
+            count[codeTable[nbSeq-1]]--;
+            nbSeq_1--;
+        }
+        assert(nbSeq_1 > 1);
+        assert(entropyWorkspaceSize >= sizeof(ZSTD_BuildCTableWksp));
+        (void)entropyWorkspaceSize;
+        FORWARD_IF_ERROR(FSE_normalizeCount(wksp->norm, tableLog, count, nbSeq_1, max, ZSTD_useLowProbCount(nbSeq_1)), "");
+        {   size_t const NCountSize = FSE_writeNCount(op, oend - op, wksp->norm, max, tableLog);   /* overflow protected */
+            FORWARD_IF_ERROR(NCountSize, "FSE_writeNCount failed");
+            FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, wksp->norm, max, tableLog, wksp->wksp, sizeof(wksp->wksp)), "");
+            return NCountSize;
+        }
+    }
+    default: assert(0); RETURN_ERROR(GENERIC, "impossible to reach");
+    }
+}
+
+FORCE_INLINE_TEMPLATE size_t
+ZSTD_encodeSequences_body(
+            void* dst, size_t dstCapacity,
+            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
+            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
+            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
+            seqDef const* sequences, size_t nbSeq, int longOffsets)
+{
+    BIT_CStream_t blockStream;
+    FSE_CState_t  stateMatchLength;
+    FSE_CState_t  stateOffsetBits;
+    FSE_CState_t  stateLitLength;
+
+    RETURN_ERROR_IF(
+        ERR_isError(BIT_initCStream(&blockStream, dst, dstCapacity)),
+        dstSize_tooSmall, "not enough space remaining");
+    DEBUGLOG(6, "available space for bitstream : %i  (dstCapacity=%u)",
+                (int)(blockStream.endPtr - blockStream.startPtr),
+                (unsigned)dstCapacity);
+
+    /* first symbols */
+    FSE_initCState2(&stateMatchLength, CTable_MatchLength, mlCodeTable[nbSeq-1]);
+    FSE_initCState2(&stateOffsetBits,  CTable_OffsetBits,  ofCodeTable[nbSeq-1]);
+    FSE_initCState2(&stateLitLength,   CTable_LitLength,   llCodeTable[nbSeq-1]);
+    BIT_addBits(&blockStream, sequences[nbSeq-1].litLength, LL_bits[llCodeTable[nbSeq-1]]);
+    if (MEM_32bits()) BIT_flushBits(&blockStream);
+    BIT_addBits(&blockStream, sequences[nbSeq-1].matchLength, ML_bits[mlCodeTable[nbSeq-1]]);
+    if (MEM_32bits()) BIT_flushBits(&blockStream);
+    if (longOffsets) {
+        U32 const ofBits = ofCodeTable[nbSeq-1];
+        unsigned const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN-1);
+        if (extraBits) {
+            BIT_addBits(&blockStream, sequences[nbSeq-1].offset, extraBits);
+            BIT_flushBits(&blockStream);
+        }
+        BIT_addBits(&blockStream, sequences[nbSeq-1].offset >> extraBits,
+                    ofBits - extraBits);
+    } else {
+        BIT_addBits(&blockStream, sequences[nbSeq-1].offset, ofCodeTable[nbSeq-1]);
+    }
+    BIT_flushBits(&blockStream);
+
+    {   size_t n;
+        for (n=nbSeq-2 ; n<nbSeq ; n--) {      /* intentional underflow */
+            BYTE const llCode = llCodeTable[n];
+            BYTE const ofCode = ofCodeTable[n];
+            BYTE const mlCode = mlCodeTable[n];
+            U32  const llBits = LL_bits[llCode];
+            U32  const ofBits = ofCode;
+            U32  const mlBits = ML_bits[mlCode];
+            DEBUGLOG(6, "encoding: litlen:%2u - matchlen:%2u - offCode:%7u",
+                        (unsigned)sequences[n].litLength,
+                        (unsigned)sequences[n].matchLength + MINMATCH,
+                        (unsigned)sequences[n].offset);
+                                                                            /* 32b*/  /* 64b*/
+                                                                            /* (7)*/  /* (7)*/
+            FSE_encodeSymbol(&blockStream, &stateOffsetBits, ofCode);       /* 15 */  /* 15 */
+            FSE_encodeSymbol(&blockStream, &stateMatchLength, mlCode);      /* 24 */  /* 24 */
+            if (MEM_32bits()) BIT_flushBits(&blockStream);                  /* (7)*/
+            FSE_encodeSymbol(&blockStream, &stateLitLength, llCode);        /* 16 */  /* 33 */
+            if (MEM_32bits() || (ofBits+mlBits+llBits >= 64-7-(LLFSELog+MLFSELog+OffFSELog)))
+                BIT_flushBits(&blockStream);                                /* (7)*/
+            BIT_addBits(&blockStream, sequences[n].litLength, llBits);
+            if (MEM_32bits() && ((llBits+mlBits)>24)) BIT_flushBits(&blockStream);
+            BIT_addBits(&blockStream, sequences[n].matchLength, mlBits);
+            if (MEM_32bits() || (ofBits+mlBits+llBits > 56)) BIT_flushBits(&blockStream);
+            if (longOffsets) {
+                unsigned const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN-1);
+                if (extraBits) {
+                    BIT_addBits(&blockStream, sequences[n].offset, extraBits);
+                    BIT_flushBits(&blockStream);                            /* (7)*/
+                }
+                BIT_addBits(&blockStream, sequences[n].offset >> extraBits,
+                            ofBits - extraBits);                            /* 31 */
+            } else {
+                BIT_addBits(&blockStream, sequences[n].offset, ofBits);     /* 31 */
+            }
+            BIT_flushBits(&blockStream);                                    /* (7)*/
+            DEBUGLOG(7, "remaining space : %i", (int)(blockStream.endPtr - blockStream.ptr));
+    }   }
+
+    DEBUGLOG(6, "ZSTD_encodeSequences: flushing ML state with %u bits", stateMatchLength.stateLog);
+    FSE_flushCState(&blockStream, &stateMatchLength);
+    DEBUGLOG(6, "ZSTD_encodeSequences: flushing Off state with %u bits", stateOffsetBits.stateLog);
+    FSE_flushCState(&blockStream, &stateOffsetBits);
+    DEBUGLOG(6, "ZSTD_encodeSequences: flushing LL state with %u bits", stateLitLength.stateLog);
+    FSE_flushCState(&blockStream, &stateLitLength);
+
+    {   size_t const streamSize = BIT_closeCStream(&blockStream);
+        RETURN_ERROR_IF(streamSize==0, dstSize_tooSmall, "not enough space");
+        return streamSize;
+    }
+}
+
+static size_t
+ZSTD_encodeSequences_default(
+            void* dst, size_t dstCapacity,
+            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
+            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
+            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
+            seqDef const* sequences, size_t nbSeq, int longOffsets)
+{
+    return ZSTD_encodeSequences_body(dst, dstCapacity,
+                                    CTable_MatchLength, mlCodeTable,
+                                    CTable_OffsetBits, ofCodeTable,
+                                    CTable_LitLength, llCodeTable,
+                                    sequences, nbSeq, longOffsets);
+}
+
+
+#if DYNAMIC_BMI2
+
+static TARGET_ATTRIBUTE("bmi2") size_t
+ZSTD_encodeSequences_bmi2(
+            void* dst, size_t dstCapacity,
+            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
+            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
+            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
+            seqDef const* sequences, size_t nbSeq, int longOffsets)
+{
+    return ZSTD_encodeSequences_body(dst, dstCapacity,
+                                    CTable_MatchLength, mlCodeTable,
+                                    CTable_OffsetBits, ofCodeTable,
+                                    CTable_LitLength, llCodeTable,
+                                    sequences, nbSeq, longOffsets);
+}
+
+#endif
+
+size_t ZSTD_encodeSequences(
+            void* dst, size_t dstCapacity,
+            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
+            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
+            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
+            seqDef const* sequences, size_t nbSeq, int longOffsets, int bmi2)
+{
+    DEBUGLOG(5, "ZSTD_encodeSequences: dstCapacity = %u", (unsigned)dstCapacity);
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        return ZSTD_encodeSequences_bmi2(dst, dstCapacity,
+                                         CTable_MatchLength, mlCodeTable,
+                                         CTable_OffsetBits, ofCodeTable,
+                                         CTable_LitLength, llCodeTable,
+                                         sequences, nbSeq, longOffsets);
+    }
+#endif
+    (void)bmi2;
+    return ZSTD_encodeSequences_default(dst, dstCapacity,
+                                        CTable_MatchLength, mlCodeTable,
+                                        CTable_OffsetBits, ofCodeTable,
+                                        CTable_LitLength, llCodeTable,
+                                        sequences, nbSeq, longOffsets);
+}
diff --git a/lib/zstd/compress/zstd_compress_sequences.h b/lib/zstd/compress/zstd_compress_sequences.h
new file mode 100644
index 0000000..7991364
--- /dev/null
+++ b/lib/zstd/compress/zstd_compress_sequences.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_COMPRESS_SEQUENCES_H
+#define ZSTD_COMPRESS_SEQUENCES_H
+
+#include "../common/fse.h" /* FSE_repeat, FSE_CTable */
+#include "../common/zstd_internal.h" /* symbolEncodingType_e, ZSTD_strategy */
+
+typedef enum {
+    ZSTD_defaultDisallowed = 0,
+    ZSTD_defaultAllowed = 1
+} ZSTD_defaultPolicy_e;
+
+symbolEncodingType_e
+ZSTD_selectEncodingType(
+        FSE_repeat* repeatMode, unsigned const* count, unsigned const max,
+        size_t const mostFrequent, size_t nbSeq, unsigned const FSELog,
+        FSE_CTable const* prevCTable,
+        short const* defaultNorm, U32 defaultNormLog,
+        ZSTD_defaultPolicy_e const isDefaultAllowed,
+        ZSTD_strategy const strategy);
+
+size_t
+ZSTD_buildCTable(void* dst, size_t dstCapacity,
+                FSE_CTable* nextCTable, U32 FSELog, symbolEncodingType_e type,
+                unsigned* count, U32 max,
+                const BYTE* codeTable, size_t nbSeq,
+                const S16* defaultNorm, U32 defaultNormLog, U32 defaultMax,
+                const FSE_CTable* prevCTable, size_t prevCTableSize,
+                void* entropyWorkspace, size_t entropyWorkspaceSize);
+
+size_t ZSTD_encodeSequences(
+            void* dst, size_t dstCapacity,
+            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
+            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
+            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
+            seqDef const* sequences, size_t nbSeq, int longOffsets, int bmi2);
+
+size_t ZSTD_fseBitCost(
+    FSE_CTable const* ctable,
+    unsigned const* count,
+    unsigned const max);
+
+size_t ZSTD_crossEntropyCost(short const* norm, unsigned accuracyLog,
+                             unsigned const* count, unsigned const max);
+#endif /* ZSTD_COMPRESS_SEQUENCES_H */
diff --git a/lib/zstd/compress/zstd_compress_superblock.c b/lib/zstd/compress/zstd_compress_superblock.c
new file mode 100644
index 0000000..ee03e0a
--- /dev/null
+++ b/lib/zstd/compress/zstd_compress_superblock.c
@@ -0,0 +1,850 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+ /*-*************************************
+ *  Dependencies
+ ***************************************/
+#include "zstd_compress_superblock.h"
+
+#include "../common/zstd_internal.h"  /* ZSTD_getSequenceLength */
+#include "hist.h"                     /* HIST_countFast_wksp */
+#include "zstd_compress_internal.h"
+#include "zstd_compress_sequences.h"
+#include "zstd_compress_literals.h"
+
+/*-*************************************
+*  Superblock entropy buffer structs
+***************************************/
+/* ZSTD_hufCTablesMetadata_t :
+ *  Stores Literals Block Type for a super-block in hType, and
+ *  huffman tree description in hufDesBuffer.
+ *  hufDesSize refers to the size of huffman tree description in bytes.
+ *  This metadata is populated in ZSTD_buildSuperBlockEntropy_literal() */
+typedef struct {
+    symbolEncodingType_e hType;
+    BYTE hufDesBuffer[ZSTD_MAX_HUF_HEADER_SIZE];
+    size_t hufDesSize;
+} ZSTD_hufCTablesMetadata_t;
+
+/* ZSTD_fseCTablesMetadata_t :
+ *  Stores symbol compression modes for a super-block in {ll, ol, ml}Type, and
+ *  fse tables in fseTablesBuffer.
+ *  fseTablesSize refers to the size of fse tables in bytes.
+ *  This metadata is populated in ZSTD_buildSuperBlockEntropy_sequences() */
+typedef struct {
+    symbolEncodingType_e llType;
+    symbolEncodingType_e ofType;
+    symbolEncodingType_e mlType;
+    BYTE fseTablesBuffer[ZSTD_MAX_FSE_HEADERS_SIZE];
+    size_t fseTablesSize;
+    size_t lastCountSize; /* This is to account for bug in 1.3.4. More detail in ZSTD_compressSubBlock_sequences() */
+} ZSTD_fseCTablesMetadata_t;
+
+typedef struct {
+    ZSTD_hufCTablesMetadata_t hufMetadata;
+    ZSTD_fseCTablesMetadata_t fseMetadata;
+} ZSTD_entropyCTablesMetadata_t;
+
+
+/* ZSTD_buildSuperBlockEntropy_literal() :
+ *  Builds entropy for the super-block literals.
+ *  Stores literals block type (raw, rle, compressed, repeat) and
+ *  huffman description table to hufMetadata.
+ *  @return : size of huffman description table or error code */
+static size_t ZSTD_buildSuperBlockEntropy_literal(void* const src, size_t srcSize,
+                                            const ZSTD_hufCTables_t* prevHuf,
+                                                  ZSTD_hufCTables_t* nextHuf,
+                                                  ZSTD_hufCTablesMetadata_t* hufMetadata,
+                                                  const int disableLiteralsCompression,
+                                                  void* workspace, size_t wkspSize)
+{
+    BYTE* const wkspStart = (BYTE*)workspace;
+    BYTE* const wkspEnd = wkspStart + wkspSize;
+    BYTE* const countWkspStart = wkspStart;
+    unsigned* const countWksp = (unsigned*)workspace;
+    const size_t countWkspSize = (HUF_SYMBOLVALUE_MAX + 1) * sizeof(unsigned);
+    BYTE* const nodeWksp = countWkspStart + countWkspSize;
+    const size_t nodeWkspSize = wkspEnd-nodeWksp;
+    unsigned maxSymbolValue = 255;
+    unsigned huffLog = HUF_TABLELOG_DEFAULT;
+    HUF_repeat repeat = prevHuf->repeatMode;
+
+    DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy_literal (srcSize=%zu)", srcSize);
+
+    /* Prepare nextEntropy assuming reusing the existing table */
+    ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
+
+    if (disableLiteralsCompression) {
+        DEBUGLOG(5, "set_basic - disabled");
+        hufMetadata->hType = set_basic;
+        return 0;
+    }
+
+    /* small ? don't even attempt compression (speed opt) */
+#   define COMPRESS_LITERALS_SIZE_MIN 63
+    {   size_t const minLitSize = (prevHuf->repeatMode == HUF_repeat_valid) ? 6 : COMPRESS_LITERALS_SIZE_MIN;
+        if (srcSize <= minLitSize) {
+            DEBUGLOG(5, "set_basic - too small");
+            hufMetadata->hType = set_basic;
+            return 0;
+        }
+    }
+
+    /* Scan input and build symbol stats */
+    {   size_t const largest = HIST_count_wksp (countWksp, &maxSymbolValue, (const BYTE*)src, srcSize, workspace, wkspSize);
+        FORWARD_IF_ERROR(largest, "HIST_count_wksp failed");
+        if (largest == srcSize) {
+            DEBUGLOG(5, "set_rle");
+            hufMetadata->hType = set_rle;
+            return 0;
+        }
+        if (largest <= (srcSize >> 7)+4) {
+            DEBUGLOG(5, "set_basic - no gain");
+            hufMetadata->hType = set_basic;
+            return 0;
+        }
+    }
+
+    /* Validate the previous Huffman table */
+    if (repeat == HUF_repeat_check && !HUF_validateCTable((HUF_CElt const*)prevHuf->CTable, countWksp, maxSymbolValue)) {
+        repeat = HUF_repeat_none;
+    }
+
+    /* Build Huffman Tree */
+    ZSTD_memset(nextHuf->CTable, 0, sizeof(nextHuf->CTable));
+    huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
+    {   size_t const maxBits = HUF_buildCTable_wksp((HUF_CElt*)nextHuf->CTable, countWksp,
+                                                    maxSymbolValue, huffLog,
+                                                    nodeWksp, nodeWkspSize);
+        FORWARD_IF_ERROR(maxBits, "HUF_buildCTable_wksp");
+        huffLog = (U32)maxBits;
+        {   /* Build and write the CTable */
+            size_t const newCSize = HUF_estimateCompressedSize(
+                    (HUF_CElt*)nextHuf->CTable, countWksp, maxSymbolValue);
+            size_t const hSize = HUF_writeCTable_wksp(
+                    hufMetadata->hufDesBuffer, sizeof(hufMetadata->hufDesBuffer),
+                    (HUF_CElt*)nextHuf->CTable, maxSymbolValue, huffLog,
+                    nodeWksp, nodeWkspSize);
+            /* Check against repeating the previous CTable */
+            if (repeat != HUF_repeat_none) {
+                size_t const oldCSize = HUF_estimateCompressedSize(
+                        (HUF_CElt const*)prevHuf->CTable, countWksp, maxSymbolValue);
+                if (oldCSize < srcSize && (oldCSize <= hSize + newCSize || hSize + 12 >= srcSize)) {
+                    DEBUGLOG(5, "set_repeat - smaller");
+                    ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
+                    hufMetadata->hType = set_repeat;
+                    return 0;
+                }
+            }
+            if (newCSize + hSize >= srcSize) {
+                DEBUGLOG(5, "set_basic - no gains");
+                ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
+                hufMetadata->hType = set_basic;
+                return 0;
+            }
+            DEBUGLOG(5, "set_compressed (hSize=%u)", (U32)hSize);
+            hufMetadata->hType = set_compressed;
+            nextHuf->repeatMode = HUF_repeat_check;
+            return hSize;
+        }
+    }
+}
+
+/* ZSTD_buildSuperBlockEntropy_sequences() :
+ *  Builds entropy for the super-block sequences.
+ *  Stores symbol compression modes and fse table to fseMetadata.
+ *  @return : size of fse tables or error code */
+static size_t ZSTD_buildSuperBlockEntropy_sequences(seqStore_t* seqStorePtr,
+                                              const ZSTD_fseCTables_t* prevEntropy,
+                                                    ZSTD_fseCTables_t* nextEntropy,
+                                              const ZSTD_CCtx_params* cctxParams,
+                                                    ZSTD_fseCTablesMetadata_t* fseMetadata,
+                                                    void* workspace, size_t wkspSize)
+{
+    BYTE* const wkspStart = (BYTE*)workspace;
+    BYTE* const wkspEnd = wkspStart + wkspSize;
+    BYTE* const countWkspStart = wkspStart;
+    unsigned* const countWksp = (unsigned*)workspace;
+    const size_t countWkspSize = (MaxSeq + 1) * sizeof(unsigned);
+    BYTE* const cTableWksp = countWkspStart + countWkspSize;
+    const size_t cTableWkspSize = wkspEnd-cTableWksp;
+    ZSTD_strategy const strategy = cctxParams->cParams.strategy;
+    FSE_CTable* CTable_LitLength = nextEntropy->litlengthCTable;
+    FSE_CTable* CTable_OffsetBits = nextEntropy->offcodeCTable;
+    FSE_CTable* CTable_MatchLength = nextEntropy->matchlengthCTable;
+    const BYTE* const ofCodeTable = seqStorePtr->ofCode;
+    const BYTE* const llCodeTable = seqStorePtr->llCode;
+    const BYTE* const mlCodeTable = seqStorePtr->mlCode;
+    size_t const nbSeq = seqStorePtr->sequences - seqStorePtr->sequencesStart;
+    BYTE* const ostart = fseMetadata->fseTablesBuffer;
+    BYTE* const oend = ostart + sizeof(fseMetadata->fseTablesBuffer);
+    BYTE* op = ostart;
+
+    assert(cTableWkspSize >= (1 << MaxFSELog) * sizeof(FSE_FUNCTION_TYPE));
+    DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy_sequences (nbSeq=%zu)", nbSeq);
+    ZSTD_memset(workspace, 0, wkspSize);
+
+    fseMetadata->lastCountSize = 0;
+    /* convert length/distances into codes */
+    ZSTD_seqToCodes(seqStorePtr);
+    /* build CTable for Literal Lengths */
+    {   U32 LLtype;
+        unsigned max = MaxLL;
+        size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, llCodeTable, nbSeq, workspace, wkspSize);  /* can't fail */
+        DEBUGLOG(5, "Building LL table");
+        nextEntropy->litlength_repeatMode = prevEntropy->litlength_repeatMode;
+        LLtype = ZSTD_selectEncodingType(&nextEntropy->litlength_repeatMode,
+                                        countWksp, max, mostFrequent, nbSeq,
+                                        LLFSELog, prevEntropy->litlengthCTable,
+                                        LL_defaultNorm, LL_defaultNormLog,
+                                        ZSTD_defaultAllowed, strategy);
+        assert(set_basic < set_compressed && set_rle < set_compressed);
+        assert(!(LLtype < set_compressed && nextEntropy->litlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
+        {   size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_LitLength, LLFSELog, (symbolEncodingType_e)LLtype,
+                                                    countWksp, max, llCodeTable, nbSeq, LL_defaultNorm, LL_defaultNormLog, MaxLL,
+                                                    prevEntropy->litlengthCTable, sizeof(prevEntropy->litlengthCTable),
+                                                    cTableWksp, cTableWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for LitLens failed");
+            if (LLtype == set_compressed)
+                fseMetadata->lastCountSize = countSize;
+            op += countSize;
+            fseMetadata->llType = (symbolEncodingType_e) LLtype;
+    }   }
+    /* build CTable for Offsets */
+    {   U32 Offtype;
+        unsigned max = MaxOff;
+        size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, ofCodeTable, nbSeq, workspace, wkspSize);  /* can't fail */
+        /* We can only use the basic table if max <= DefaultMaxOff, otherwise the offsets are too large */
+        ZSTD_defaultPolicy_e const defaultPolicy = (max <= DefaultMaxOff) ? ZSTD_defaultAllowed : ZSTD_defaultDisallowed;
+        DEBUGLOG(5, "Building OF table");
+        nextEntropy->offcode_repeatMode = prevEntropy->offcode_repeatMode;
+        Offtype = ZSTD_selectEncodingType(&nextEntropy->offcode_repeatMode,
+                                        countWksp, max, mostFrequent, nbSeq,
+                                        OffFSELog, prevEntropy->offcodeCTable,
+                                        OF_defaultNorm, OF_defaultNormLog,
+                                        defaultPolicy, strategy);
+        assert(!(Offtype < set_compressed && nextEntropy->offcode_repeatMode != FSE_repeat_none)); /* We don't copy tables */
+        {   size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_OffsetBits, OffFSELog, (symbolEncodingType_e)Offtype,
+                                                    countWksp, max, ofCodeTable, nbSeq, OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
+                                                    prevEntropy->offcodeCTable, sizeof(prevEntropy->offcodeCTable),
+                                                    cTableWksp, cTableWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for Offsets failed");
+            if (Offtype == set_compressed)
+                fseMetadata->lastCountSize = countSize;
+            op += countSize;
+            fseMetadata->ofType = (symbolEncodingType_e) Offtype;
+    }   }
+    /* build CTable for MatchLengths */
+    {   U32 MLtype;
+        unsigned max = MaxML;
+        size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, mlCodeTable, nbSeq, workspace, wkspSize);   /* can't fail */
+        DEBUGLOG(5, "Building ML table (remaining space : %i)", (int)(oend-op));
+        nextEntropy->matchlength_repeatMode = prevEntropy->matchlength_repeatMode;
+        MLtype = ZSTD_selectEncodingType(&nextEntropy->matchlength_repeatMode,
+                                        countWksp, max, mostFrequent, nbSeq,
+                                        MLFSELog, prevEntropy->matchlengthCTable,
+                                        ML_defaultNorm, ML_defaultNormLog,
+                                        ZSTD_defaultAllowed, strategy);
+        assert(!(MLtype < set_compressed && nextEntropy->matchlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
+        {   size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_MatchLength, MLFSELog, (symbolEncodingType_e)MLtype,
+                                                    countWksp, max, mlCodeTable, nbSeq, ML_defaultNorm, ML_defaultNormLog, MaxML,
+                                                    prevEntropy->matchlengthCTable, sizeof(prevEntropy->matchlengthCTable),
+                                                    cTableWksp, cTableWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for MatchLengths failed");
+            if (MLtype == set_compressed)
+                fseMetadata->lastCountSize = countSize;
+            op += countSize;
+            fseMetadata->mlType = (symbolEncodingType_e) MLtype;
+    }   }
+    assert((size_t) (op-ostart) <= sizeof(fseMetadata->fseTablesBuffer));
+    return op-ostart;
+}
+
+
+/* ZSTD_buildSuperBlockEntropy() :
+ *  Builds entropy for the super-block.
+ *  @return : 0 on success or error code */
+static size_t
+ZSTD_buildSuperBlockEntropy(seqStore_t* seqStorePtr,
+                      const ZSTD_entropyCTables_t* prevEntropy,
+                            ZSTD_entropyCTables_t* nextEntropy,
+                      const ZSTD_CCtx_params* cctxParams,
+                            ZSTD_entropyCTablesMetadata_t* entropyMetadata,
+                            void* workspace, size_t wkspSize)
+{
+    size_t const litSize = seqStorePtr->lit - seqStorePtr->litStart;
+    DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy");
+    entropyMetadata->hufMetadata.hufDesSize =
+        ZSTD_buildSuperBlockEntropy_literal(seqStorePtr->litStart, litSize,
+                                            &prevEntropy->huf, &nextEntropy->huf,
+                                            &entropyMetadata->hufMetadata,
+                                            ZSTD_disableLiteralsCompression(cctxParams),
+                                            workspace, wkspSize);
+    FORWARD_IF_ERROR(entropyMetadata->hufMetadata.hufDesSize, "ZSTD_buildSuperBlockEntropy_literal failed");
+    entropyMetadata->fseMetadata.fseTablesSize =
+        ZSTD_buildSuperBlockEntropy_sequences(seqStorePtr,
+                                              &prevEntropy->fse, &nextEntropy->fse,
+                                              cctxParams,
+                                              &entropyMetadata->fseMetadata,
+                                              workspace, wkspSize);
+    FORWARD_IF_ERROR(entropyMetadata->fseMetadata.fseTablesSize, "ZSTD_buildSuperBlockEntropy_sequences failed");
+    return 0;
+}
+
+/* ZSTD_compressSubBlock_literal() :
+ *  Compresses literals section for a sub-block.
+ *  When we have to write the Huffman table we will sometimes choose a header
+ *  size larger than necessary. This is because we have to pick the header size
+ *  before we know the table size + compressed size, so we have a bound on the
+ *  table size. If we guessed incorrectly, we fall back to uncompressed literals.
+ *
+ *  We write the header when writeEntropy=1 and set entropyWritten=1 when we succeeded
+ *  in writing the header, otherwise it is set to 0.
+ *
+ *  hufMetadata->hType has literals block type info.
+ *      If it is set_basic, all sub-blocks literals section will be Raw_Literals_Block.
+ *      If it is set_rle, all sub-blocks literals section will be RLE_Literals_Block.
+ *      If it is set_compressed, first sub-block's literals section will be Compressed_Literals_Block
+ *      If it is set_compressed, first sub-block's literals section will be Treeless_Literals_Block
+ *      and the following sub-blocks' literals sections will be Treeless_Literals_Block.
+ *  @return : compressed size of literals section of a sub-block
+ *            Or 0 if it unable to compress.
+ *            Or error code */
+static size_t ZSTD_compressSubBlock_literal(const HUF_CElt* hufTable,
+                                    const ZSTD_hufCTablesMetadata_t* hufMetadata,
+                                    const BYTE* literals, size_t litSize,
+                                    void* dst, size_t dstSize,
+                                    const int bmi2, int writeEntropy, int* entropyWritten)
+{
+    size_t const header = writeEntropy ? 200 : 0;
+    size_t const lhSize = 3 + (litSize >= (1 KB - header)) + (litSize >= (16 KB - header));
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + dstSize;
+    BYTE* op = ostart + lhSize;
+    U32 const singleStream = lhSize == 3;
+    symbolEncodingType_e hType = writeEntropy ? hufMetadata->hType : set_repeat;
+    size_t cLitSize = 0;
+
+    (void)bmi2; /* TODO bmi2... */
+
+    DEBUGLOG(5, "ZSTD_compressSubBlock_literal (litSize=%zu, lhSize=%zu, writeEntropy=%d)", litSize, lhSize, writeEntropy);
+
+    *entropyWritten = 0;
+    if (litSize == 0 || hufMetadata->hType == set_basic) {
+      DEBUGLOG(5, "ZSTD_compressSubBlock_literal using raw literal");
+      return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
+    } else if (hufMetadata->hType == set_rle) {
+      DEBUGLOG(5, "ZSTD_compressSubBlock_literal using rle literal");
+      return ZSTD_compressRleLiteralsBlock(dst, dstSize, literals, litSize);
+    }
+
+    assert(litSize > 0);
+    assert(hufMetadata->hType == set_compressed || hufMetadata->hType == set_repeat);
+
+    if (writeEntropy && hufMetadata->hType == set_compressed) {
+        ZSTD_memcpy(op, hufMetadata->hufDesBuffer, hufMetadata->hufDesSize);
+        op += hufMetadata->hufDesSize;
+        cLitSize += hufMetadata->hufDesSize;
+        DEBUGLOG(5, "ZSTD_compressSubBlock_literal (hSize=%zu)", hufMetadata->hufDesSize);
+    }
+
+    /* TODO bmi2 */
+    {   const size_t cSize = singleStream ? HUF_compress1X_usingCTable(op, oend-op, literals, litSize, hufTable)
+                                          : HUF_compress4X_usingCTable(op, oend-op, literals, litSize, hufTable);
+        op += cSize;
+        cLitSize += cSize;
+        if (cSize == 0 || ERR_isError(cSize)) {
+            DEBUGLOG(5, "Failed to write entropy tables %s", ZSTD_getErrorName(cSize));
+            return 0;
+        }
+        /* If we expand and we aren't writing a header then emit uncompressed */
+        if (!writeEntropy && cLitSize >= litSize) {
+            DEBUGLOG(5, "ZSTD_compressSubBlock_literal using raw literal because uncompressible");
+            return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
+        }
+        /* If we are writing headers then allow expansion that doesn't change our header size. */
+        if (lhSize < (size_t)(3 + (cLitSize >= 1 KB) + (cLitSize >= 16 KB))) {
+            assert(cLitSize > litSize);
+            DEBUGLOG(5, "Literals expanded beyond allowed header size");
+            return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
+        }
+        DEBUGLOG(5, "ZSTD_compressSubBlock_literal (cSize=%zu)", cSize);
+    }
+
+    /* Build header */
+    switch(lhSize)
+    {
+    case 3: /* 2 - 2 - 10 - 10 */
+        {   U32 const lhc = hType + ((!singleStream) << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<14);
+            MEM_writeLE24(ostart, lhc);
+            break;
+        }
+    case 4: /* 2 - 2 - 14 - 14 */
+        {   U32 const lhc = hType + (2 << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<18);
+            MEM_writeLE32(ostart, lhc);
+            break;
+        }
+    case 5: /* 2 - 2 - 18 - 18 */
+        {   U32 const lhc = hType + (3 << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<22);
+            MEM_writeLE32(ostart, lhc);
+            ostart[4] = (BYTE)(cLitSize >> 10);
+            break;
+        }
+    default:  /* not possible : lhSize is {3,4,5} */
+        assert(0);
+    }
+    *entropyWritten = 1;
+    DEBUGLOG(5, "Compressed literals: %u -> %u", (U32)litSize, (U32)(op-ostart));
+    return op-ostart;
+}
+
+static size_t ZSTD_seqDecompressedSize(seqStore_t const* seqStore, const seqDef* sequences, size_t nbSeq, size_t litSize, int lastSequence) {
+    const seqDef* const sstart = sequences;
+    const seqDef* const send = sequences + nbSeq;
+    const seqDef* sp = sstart;
+    size_t matchLengthSum = 0;
+    size_t litLengthSum = 0;
+    while (send-sp > 0) {
+        ZSTD_sequenceLength const seqLen = ZSTD_getSequenceLength(seqStore, sp);
+        litLengthSum += seqLen.litLength;
+        matchLengthSum += seqLen.matchLength;
+        sp++;
+    }
+    assert(litLengthSum <= litSize);
+    if (!lastSequence) {
+        assert(litLengthSum == litSize);
+    }
+    return matchLengthSum + litSize;
+}
+
+/* ZSTD_compressSubBlock_sequences() :
+ *  Compresses sequences section for a sub-block.
+ *  fseMetadata->llType, fseMetadata->ofType, and fseMetadata->mlType have
+ *  symbol compression modes for the super-block.
+ *  The first successfully compressed block will have these in its header.
+ *  We set entropyWritten=1 when we succeed in compressing the sequences.
+ *  The following sub-blocks will always have repeat mode.
+ *  @return : compressed size of sequences section of a sub-block
+ *            Or 0 if it is unable to compress
+ *            Or error code. */
+static size_t ZSTD_compressSubBlock_sequences(const ZSTD_fseCTables_t* fseTables,
+                                              const ZSTD_fseCTablesMetadata_t* fseMetadata,
+                                              const seqDef* sequences, size_t nbSeq,
+                                              const BYTE* llCode, const BYTE* mlCode, const BYTE* ofCode,
+                                              const ZSTD_CCtx_params* cctxParams,
+                                              void* dst, size_t dstCapacity,
+                                              const int bmi2, int writeEntropy, int* entropyWritten)
+{
+    const int longOffsets = cctxParams->cParams.windowLog > STREAM_ACCUMULATOR_MIN;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + dstCapacity;
+    BYTE* op = ostart;
+    BYTE* seqHead;
+
+    DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (nbSeq=%zu, writeEntropy=%d, longOffsets=%d)", nbSeq, writeEntropy, longOffsets);
+
+    *entropyWritten = 0;
+    /* Sequences Header */
+    RETURN_ERROR_IF((oend-op) < 3 /*max nbSeq Size*/ + 1 /*seqHead*/,
+                    dstSize_tooSmall, "");
+    if (nbSeq < 0x7F)
+        *op++ = (BYTE)nbSeq;
+    else if (nbSeq < LONGNBSEQ)
+        op[0] = (BYTE)((nbSeq>>8) + 0x80), op[1] = (BYTE)nbSeq, op+=2;
+    else
+        op[0]=0xFF, MEM_writeLE16(op+1, (U16)(nbSeq - LONGNBSEQ)), op+=3;
+    if (nbSeq==0) {
+        return op - ostart;
+    }
+
+    /* seqHead : flags for FSE encoding type */
+    seqHead = op++;
+
+    DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (seqHeadSize=%u)", (unsigned)(op-ostart));
+
+    if (writeEntropy) {
+        const U32 LLtype = fseMetadata->llType;
+        const U32 Offtype = fseMetadata->ofType;
+        const U32 MLtype = fseMetadata->mlType;
+        DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (fseTablesSize=%zu)", fseMetadata->fseTablesSize);
+        *seqHead = (BYTE)((LLtype<<6) + (Offtype<<4) + (MLtype<<2));
+        ZSTD_memcpy(op, fseMetadata->fseTablesBuffer, fseMetadata->fseTablesSize);
+        op += fseMetadata->fseTablesSize;
+    } else {
+        const U32 repeat = set_repeat;
+        *seqHead = (BYTE)((repeat<<6) + (repeat<<4) + (repeat<<2));
+    }
+
+    {   size_t const bitstreamSize = ZSTD_encodeSequences(
+                                        op, oend - op,
+                                        fseTables->matchlengthCTable, mlCode,
+                                        fseTables->offcodeCTable, ofCode,
+                                        fseTables->litlengthCTable, llCode,
+                                        sequences, nbSeq,
+                                        longOffsets, bmi2);
+        FORWARD_IF_ERROR(bitstreamSize, "ZSTD_encodeSequences failed");
+        op += bitstreamSize;
+        /* zstd versions <= 1.3.4 mistakenly report corruption when
+         * FSE_readNCount() receives a buffer < 4 bytes.
+         * Fixed by https://github.com/facebook/zstd/pull/1146.
+         * This can happen when the last set_compressed table present is 2
+         * bytes and the bitstream is only one byte.
+         * In this exceedingly rare case, we will simply emit an uncompressed
+         * block, since it isn't worth optimizing.
+         */
+#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+        if (writeEntropy && fseMetadata->lastCountSize && fseMetadata->lastCountSize + bitstreamSize < 4) {
+            /* NCountSize >= 2 && bitstreamSize > 0 ==> lastCountSize == 3 */
+            assert(fseMetadata->lastCountSize + bitstreamSize == 3);
+            DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.3.4 by "
+                        "emitting an uncompressed block.");
+            return 0;
+        }
+#endif
+        DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (bitstreamSize=%zu)", bitstreamSize);
+    }
+
+    /* zstd versions <= 1.4.0 mistakenly report error when
+     * sequences section body size is less than 3 bytes.
+     * Fixed by https://github.com/facebook/zstd/pull/1664.
+     * This can happen when the previous sequences section block is compressed
+     * with rle mode and the current block's sequences section is compressed
+     * with repeat mode where sequences section body size can be 1 byte.
+     */
+#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    if (op-seqHead < 4) {
+        DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.4.0 by emitting "
+                    "an uncompressed block when sequences are < 4 bytes");
+        return 0;
+    }
+#endif
+
+    *entropyWritten = 1;
+    return op - ostart;
+}
+
+/* ZSTD_compressSubBlock() :
+ *  Compresses a single sub-block.
+ *  @return : compressed size of the sub-block
+ *            Or 0 if it failed to compress. */
+static size_t ZSTD_compressSubBlock(const ZSTD_entropyCTables_t* entropy,
+                                    const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
+                                    const seqDef* sequences, size_t nbSeq,
+                                    const BYTE* literals, size_t litSize,
+                                    const BYTE* llCode, const BYTE* mlCode, const BYTE* ofCode,
+                                    const ZSTD_CCtx_params* cctxParams,
+                                    void* dst, size_t dstCapacity,
+                                    const int bmi2,
+                                    int writeLitEntropy, int writeSeqEntropy,
+                                    int* litEntropyWritten, int* seqEntropyWritten,
+                                    U32 lastBlock)
+{
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + dstCapacity;
+    BYTE* op = ostart + ZSTD_blockHeaderSize;
+    DEBUGLOG(5, "ZSTD_compressSubBlock (litSize=%zu, nbSeq=%zu, writeLitEntropy=%d, writeSeqEntropy=%d, lastBlock=%d)",
+                litSize, nbSeq, writeLitEntropy, writeSeqEntropy, lastBlock);
+    {   size_t cLitSize = ZSTD_compressSubBlock_literal((const HUF_CElt*)entropy->huf.CTable,
+                                                        &entropyMetadata->hufMetadata, literals, litSize,
+                                                        op, oend-op, bmi2, writeLitEntropy, litEntropyWritten);
+        FORWARD_IF_ERROR(cLitSize, "ZSTD_compressSubBlock_literal failed");
+        if (cLitSize == 0) return 0;
+        op += cLitSize;
+    }
+    {   size_t cSeqSize = ZSTD_compressSubBlock_sequences(&entropy->fse,
+                                                  &entropyMetadata->fseMetadata,
+                                                  sequences, nbSeq,
+                                                  llCode, mlCode, ofCode,
+                                                  cctxParams,
+                                                  op, oend-op,
+                                                  bmi2, writeSeqEntropy, seqEntropyWritten);
+        FORWARD_IF_ERROR(cSeqSize, "ZSTD_compressSubBlock_sequences failed");
+        if (cSeqSize == 0) return 0;
+        op += cSeqSize;
+    }
+    /* Write block header */
+    {   size_t cSize = (op-ostart)-ZSTD_blockHeaderSize;
+        U32 const cBlockHeader24 = lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3);
+        MEM_writeLE24(ostart, cBlockHeader24);
+    }
+    return op-ostart;
+}
+
+static size_t ZSTD_estimateSubBlockSize_literal(const BYTE* literals, size_t litSize,
+                                                const ZSTD_hufCTables_t* huf,
+                                                const ZSTD_hufCTablesMetadata_t* hufMetadata,
+                                                void* workspace, size_t wkspSize,
+                                                int writeEntropy)
+{
+    unsigned* const countWksp = (unsigned*)workspace;
+    unsigned maxSymbolValue = 255;
+    size_t literalSectionHeaderSize = 3; /* Use hard coded size of 3 bytes */
+
+    if (hufMetadata->hType == set_basic) return litSize;
+    else if (hufMetadata->hType == set_rle) return 1;
+    else if (hufMetadata->hType == set_compressed || hufMetadata->hType == set_repeat) {
+        size_t const largest = HIST_count_wksp (countWksp, &maxSymbolValue, (const BYTE*)literals, litSize, workspace, wkspSize);
+        if (ZSTD_isError(largest)) return litSize;
+        {   size_t cLitSizeEstimate = HUF_estimateCompressedSize((const HUF_CElt*)huf->CTable, countWksp, maxSymbolValue);
+            if (writeEntropy) cLitSizeEstimate += hufMetadata->hufDesSize;
+            return cLitSizeEstimate + literalSectionHeaderSize;
+    }   }
+    assert(0); /* impossible */
+    return 0;
+}
+
+static size_t ZSTD_estimateSubBlockSize_symbolType(symbolEncodingType_e type,
+                        const BYTE* codeTable, unsigned maxCode,
+                        size_t nbSeq, const FSE_CTable* fseCTable,
+                        const U32* additionalBits,
+                        short const* defaultNorm, U32 defaultNormLog, U32 defaultMax,
+                        void* workspace, size_t wkspSize)
+{
+    unsigned* const countWksp = (unsigned*)workspace;
+    const BYTE* ctp = codeTable;
+    const BYTE* const ctStart = ctp;
+    const BYTE* const ctEnd = ctStart + nbSeq;
+    size_t cSymbolTypeSizeEstimateInBits = 0;
+    unsigned max = maxCode;
+
+    HIST_countFast_wksp(countWksp, &max, codeTable, nbSeq, workspace, wkspSize);  /* can't fail */
+    if (type == set_basic) {
+        /* We selected this encoding type, so it must be valid. */
+        assert(max <= defaultMax);
+        cSymbolTypeSizeEstimateInBits = max <= defaultMax
+                ? ZSTD_crossEntropyCost(defaultNorm, defaultNormLog, countWksp, max)
+                : ERROR(GENERIC);
+    } else if (type == set_rle) {
+        cSymbolTypeSizeEstimateInBits = 0;
+    } else if (type == set_compressed || type == set_repeat) {
+        cSymbolTypeSizeEstimateInBits = ZSTD_fseBitCost(fseCTable, countWksp, max);
+    }
+    if (ZSTD_isError(cSymbolTypeSizeEstimateInBits)) return nbSeq * 10;
+    while (ctp < ctEnd) {
+        if (additionalBits) cSymbolTypeSizeEstimateInBits += additionalBits[*ctp];
+        else cSymbolTypeSizeEstimateInBits += *ctp; /* for offset, offset code is also the number of additional bits */
+        ctp++;
+    }
+    return cSymbolTypeSizeEstimateInBits / 8;
+}
+
+static size_t ZSTD_estimateSubBlockSize_sequences(const BYTE* ofCodeTable,
+                                                  const BYTE* llCodeTable,
+                                                  const BYTE* mlCodeTable,
+                                                  size_t nbSeq,
+                                                  const ZSTD_fseCTables_t* fseTables,
+                                                  const ZSTD_fseCTablesMetadata_t* fseMetadata,
+                                                  void* workspace, size_t wkspSize,
+                                                  int writeEntropy)
+{
+    size_t sequencesSectionHeaderSize = 3; /* Use hard coded size of 3 bytes */
+    size_t cSeqSizeEstimate = 0;
+    cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->ofType, ofCodeTable, MaxOff,
+                                         nbSeq, fseTables->offcodeCTable, NULL,
+                                         OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
+                                         workspace, wkspSize);
+    cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->llType, llCodeTable, MaxLL,
+                                         nbSeq, fseTables->litlengthCTable, LL_bits,
+                                         LL_defaultNorm, LL_defaultNormLog, MaxLL,
+                                         workspace, wkspSize);
+    cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->mlType, mlCodeTable, MaxML,
+                                         nbSeq, fseTables->matchlengthCTable, ML_bits,
+                                         ML_defaultNorm, ML_defaultNormLog, MaxML,
+                                         workspace, wkspSize);
+    if (writeEntropy) cSeqSizeEstimate += fseMetadata->fseTablesSize;
+    return cSeqSizeEstimate + sequencesSectionHeaderSize;
+}
+
+static size_t ZSTD_estimateSubBlockSize(const BYTE* literals, size_t litSize,
+                                        const BYTE* ofCodeTable,
+                                        const BYTE* llCodeTable,
+                                        const BYTE* mlCodeTable,
+                                        size_t nbSeq,
+                                        const ZSTD_entropyCTables_t* entropy,
+                                        const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
+                                        void* workspace, size_t wkspSize,
+                                        int writeLitEntropy, int writeSeqEntropy) {
+    size_t cSizeEstimate = 0;
+    cSizeEstimate += ZSTD_estimateSubBlockSize_literal(literals, litSize,
+                                                         &entropy->huf, &entropyMetadata->hufMetadata,
+                                                         workspace, wkspSize, writeLitEntropy);
+    cSizeEstimate += ZSTD_estimateSubBlockSize_sequences(ofCodeTable, llCodeTable, mlCodeTable,
+                                                         nbSeq, &entropy->fse, &entropyMetadata->fseMetadata,
+                                                         workspace, wkspSize, writeSeqEntropy);
+    return cSizeEstimate + ZSTD_blockHeaderSize;
+}
+
+static int ZSTD_needSequenceEntropyTables(ZSTD_fseCTablesMetadata_t const* fseMetadata)
+{
+    if (fseMetadata->llType == set_compressed || fseMetadata->llType == set_rle)
+        return 1;
+    if (fseMetadata->mlType == set_compressed || fseMetadata->mlType == set_rle)
+        return 1;
+    if (fseMetadata->ofType == set_compressed || fseMetadata->ofType == set_rle)
+        return 1;
+    return 0;
+}
+
+/* ZSTD_compressSubBlock_multi() :
+ *  Breaks super-block into multiple sub-blocks and compresses them.
+ *  Entropy will be written to the first block.
+ *  The following blocks will use repeat mode to compress.
+ *  All sub-blocks are compressed blocks (no raw or rle blocks).
+ *  @return : compressed size of the super block (which is multiple ZSTD blocks)
+ *            Or 0 if it failed to compress. */
+static size_t ZSTD_compressSubBlock_multi(const seqStore_t* seqStorePtr,
+                            const ZSTD_compressedBlockState_t* prevCBlock,
+                            ZSTD_compressedBlockState_t* nextCBlock,
+                            const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
+                            const ZSTD_CCtx_params* cctxParams,
+                                  void* dst, size_t dstCapacity,
+                            const void* src, size_t srcSize,
+                            const int bmi2, U32 lastBlock,
+                            void* workspace, size_t wkspSize)
+{
+    const seqDef* const sstart = seqStorePtr->sequencesStart;
+    const seqDef* const send = seqStorePtr->sequences;
+    const seqDef* sp = sstart;
+    const BYTE* const lstart = seqStorePtr->litStart;
+    const BYTE* const lend = seqStorePtr->lit;
+    const BYTE* lp = lstart;
+    BYTE const* ip = (BYTE const*)src;
+    BYTE const* const iend = ip + srcSize;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + dstCapacity;
+    BYTE* op = ostart;
+    const BYTE* llCodePtr = seqStorePtr->llCode;
+    const BYTE* mlCodePtr = seqStorePtr->mlCode;
+    const BYTE* ofCodePtr = seqStorePtr->ofCode;
+    size_t targetCBlockSize = cctxParams->targetCBlockSize;
+    size_t litSize, seqCount;
+    int writeLitEntropy = entropyMetadata->hufMetadata.hType == set_compressed;
+    int writeSeqEntropy = 1;
+    int lastSequence = 0;
+
+    DEBUGLOG(5, "ZSTD_compressSubBlock_multi (litSize=%u, nbSeq=%u)",
+                (unsigned)(lend-lp), (unsigned)(send-sstart));
+
+    litSize = 0;
+    seqCount = 0;
+    do {
+        size_t cBlockSizeEstimate = 0;
+        if (sstart == send) {
+            lastSequence = 1;
+        } else {
+            const seqDef* const sequence = sp + seqCount;
+            lastSequence = sequence == send - 1;
+            litSize += ZSTD_getSequenceLength(seqStorePtr, sequence).litLength;
+            seqCount++;
+        }
+        if (lastSequence) {
+            assert(lp <= lend);
+            assert(litSize <= (size_t)(lend - lp));
+            litSize = (size_t)(lend - lp);
+        }
+        /* I think there is an optimization opportunity here.
+         * Calling ZSTD_estimateSubBlockSize for every sequence can be wasteful
+         * since it recalculates estimate from scratch.
+         * For example, it would recount literal distribution and symbol codes everytime.
+         */
+        cBlockSizeEstimate = ZSTD_estimateSubBlockSize(lp, litSize, ofCodePtr, llCodePtr, mlCodePtr, seqCount,
+                                                       &nextCBlock->entropy, entropyMetadata,
+                                                       workspace, wkspSize, writeLitEntropy, writeSeqEntropy);
+        if (cBlockSizeEstimate > targetCBlockSize || lastSequence) {
+            int litEntropyWritten = 0;
+            int seqEntropyWritten = 0;
+            const size_t decompressedSize = ZSTD_seqDecompressedSize(seqStorePtr, sp, seqCount, litSize, lastSequence);
+            const size_t cSize = ZSTD_compressSubBlock(&nextCBlock->entropy, entropyMetadata,
+                                                       sp, seqCount,
+                                                       lp, litSize,
+                                                       llCodePtr, mlCodePtr, ofCodePtr,
+                                                       cctxParams,
+                                                       op, oend-op,
+                                                       bmi2, writeLitEntropy, writeSeqEntropy,
+                                                       &litEntropyWritten, &seqEntropyWritten,
+                                                       lastBlock && lastSequence);
+            FORWARD_IF_ERROR(cSize, "ZSTD_compressSubBlock failed");
+            if (cSize > 0 && cSize < decompressedSize) {
+                DEBUGLOG(5, "Committed the sub-block");
+                assert(ip + decompressedSize <= iend);
+                ip += decompressedSize;
+                sp += seqCount;
+                lp += litSize;
+                op += cSize;
+                llCodePtr += seqCount;
+                mlCodePtr += seqCount;
+                ofCodePtr += seqCount;
+                litSize = 0;
+                seqCount = 0;
+                /* Entropy only needs to be written once */
+                if (litEntropyWritten) {
+                    writeLitEntropy = 0;
+                }
+                if (seqEntropyWritten) {
+                    writeSeqEntropy = 0;
+                }
+            }
+        }
+    } while (!lastSequence);
+    if (writeLitEntropy) {
+        DEBUGLOG(5, "ZSTD_compressSubBlock_multi has literal entropy tables unwritten");
+        ZSTD_memcpy(&nextCBlock->entropy.huf, &prevCBlock->entropy.huf, sizeof(prevCBlock->entropy.huf));
+    }
+    if (writeSeqEntropy && ZSTD_needSequenceEntropyTables(&entropyMetadata->fseMetadata)) {
+        /* If we haven't written our entropy tables, then we've violated our contract and
+         * must emit an uncompressed block.
+         */
+        DEBUGLOG(5, "ZSTD_compressSubBlock_multi has sequence entropy tables unwritten");
+        return 0;
+    }
+    if (ip < iend) {
+        size_t const cSize = ZSTD_noCompressBlock(op, oend - op, ip, iend - ip, lastBlock);
+        DEBUGLOG(5, "ZSTD_compressSubBlock_multi last sub-block uncompressed, %zu bytes", (size_t)(iend - ip));
+        FORWARD_IF_ERROR(cSize, "ZSTD_noCompressBlock failed");
+        assert(cSize != 0);
+        op += cSize;
+        /* We have to regenerate the repcodes because we've skipped some sequences */
+        if (sp < send) {
+            seqDef const* seq;
+            repcodes_t rep;
+            ZSTD_memcpy(&rep, prevCBlock->rep, sizeof(rep));
+            for (seq = sstart; seq < sp; ++seq) {
+                rep = ZSTD_updateRep(rep.rep, seq->offset - 1, ZSTD_getSequenceLength(seqStorePtr, seq).litLength == 0);
+            }
+            ZSTD_memcpy(nextCBlock->rep, &rep, sizeof(rep));
+        }
+    }
+    DEBUGLOG(5, "ZSTD_compressSubBlock_multi compressed");
+    return op-ostart;
+}
+
+size_t ZSTD_compressSuperBlock(ZSTD_CCtx* zc,
+                               void* dst, size_t dstCapacity,
+                               void const* src, size_t srcSize,
+                               unsigned lastBlock) {
+    ZSTD_entropyCTablesMetadata_t entropyMetadata;
+
+    FORWARD_IF_ERROR(ZSTD_buildSuperBlockEntropy(&zc->seqStore,
+          &zc->blockState.prevCBlock->entropy,
+          &zc->blockState.nextCBlock->entropy,
+          &zc->appliedParams,
+          &entropyMetadata,
+          zc->entropyWorkspace, ENTROPY_WORKSPACE_SIZE /* statically allocated in resetCCtx */), "");
+
+    return ZSTD_compressSubBlock_multi(&zc->seqStore,
+            zc->blockState.prevCBlock,
+            zc->blockState.nextCBlock,
+            &entropyMetadata,
+            &zc->appliedParams,
+            dst, dstCapacity,
+            src, srcSize,
+            zc->bmi2, lastBlock,
+            zc->entropyWorkspace, ENTROPY_WORKSPACE_SIZE /* statically allocated in resetCCtx */);
+}
diff --git a/lib/zstd/compress/zstd_compress_superblock.h b/lib/zstd/compress/zstd_compress_superblock.h
new file mode 100644
index 0000000..224ece7
--- /dev/null
+++ b/lib/zstd/compress/zstd_compress_superblock.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_COMPRESS_ADVANCED_H
+#define ZSTD_COMPRESS_ADVANCED_H
+
+/*-*************************************
+*  Dependencies
+***************************************/
+
+#include <linux/zstd.h> /* ZSTD_CCtx */
+
+/*-*************************************
+*  Target Compressed Block Size
+***************************************/
+
+/* ZSTD_compressSuperBlock() :
+ * Used to compress a super block when targetCBlockSize is being used.
+ * The given block will be compressed into multiple sub blocks that are around targetCBlockSize. */
+size_t ZSTD_compressSuperBlock(ZSTD_CCtx* zc,
+                               void* dst, size_t dstCapacity,
+                               void const* src, size_t srcSize,
+                               unsigned lastBlock);
+
+#endif /* ZSTD_COMPRESS_ADVANCED_H */
diff --git a/lib/zstd/compress/zstd_cwksp.h b/lib/zstd/compress/zstd_cwksp.h
new file mode 100644
index 0000000..98e359a
--- /dev/null
+++ b/lib/zstd/compress/zstd_cwksp.h
@@ -0,0 +1,482 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_CWKSP_H
+#define ZSTD_CWKSP_H
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#include "../common/zstd_internal.h"
+
+
+/*-*************************************
+*  Constants
+***************************************/
+
+/* Since the workspace is effectively its own little malloc implementation /
+ * arena, when we run under ASAN, we should similarly insert redzones between
+ * each internal element of the workspace, so ASAN will catch overruns that
+ * reach outside an object but that stay inside the workspace.
+ *
+ * This defines the size of that redzone.
+ */
+#ifndef ZSTD_CWKSP_ASAN_REDZONE_SIZE
+#define ZSTD_CWKSP_ASAN_REDZONE_SIZE 128
+#endif
+
+/*-*************************************
+*  Structures
+***************************************/
+typedef enum {
+    ZSTD_cwksp_alloc_objects,
+    ZSTD_cwksp_alloc_buffers,
+    ZSTD_cwksp_alloc_aligned
+} ZSTD_cwksp_alloc_phase_e;
+
+/*
+ * Used to describe whether the workspace is statically allocated (and will not
+ * necessarily ever be freed), or if it's dynamically allocated and we can
+ * expect a well-formed caller to free this.
+ */
+typedef enum {
+    ZSTD_cwksp_dynamic_alloc,
+    ZSTD_cwksp_static_alloc
+} ZSTD_cwksp_static_alloc_e;
+
+/*
+ * Zstd fits all its internal datastructures into a single continuous buffer,
+ * so that it only needs to perform a single OS allocation (or so that a buffer
+ * can be provided to it and it can perform no allocations at all). This buffer
+ * is called the workspace.
+ *
+ * Several optimizations complicate that process of allocating memory ranges
+ * from this workspace for each internal datastructure:
+ *
+ * - These different internal datastructures have different setup requirements:
+ *
+ *   - The static objects need to be cleared once and can then be trivially
+ *     reused for each compression.
+ *
+ *   - Various buffers don't need to be initialized at all--they are always
+ *     written into before they're read.
+ *
+ *   - The matchstate tables have a unique requirement that they don't need
+ *     their memory to be totally cleared, but they do need the memory to have
+ *     some bound, i.e., a guarantee that all values in the memory they've been
+ *     allocated is less than some maximum value (which is the starting value
+ *     for the indices that they will then use for compression). When this
+ *     guarantee is provided to them, they can use the memory without any setup
+ *     work. When it can't, they have to clear the area.
+ *
+ * - These buffers also have different alignment requirements.
+ *
+ * - We would like to reuse the objects in the workspace for multiple
+ *   compressions without having to perform any expensive reallocation or
+ *   reinitialization work.
+ *
+ * - We would like to be able to efficiently reuse the workspace across
+ *   multiple compressions **even when the compression parameters change** and
+ *   we need to resize some of the objects (where possible).
+ *
+ * To attempt to manage this buffer, given these constraints, the ZSTD_cwksp
+ * abstraction was created. It works as follows:
+ *
+ * Workspace Layout:
+ *
+ * [                        ... workspace ...                         ]
+ * [objects][tables ... ->] free space [<- ... aligned][<- ... buffers]
+ *
+ * The various objects that live in the workspace are divided into the
+ * following categories, and are allocated separately:
+ *
+ * - Static objects: this is optionally the enclosing ZSTD_CCtx or ZSTD_CDict,
+ *   so that literally everything fits in a single buffer. Note: if present,
+ *   this must be the first object in the workspace, since ZSTD_customFree{CCtx,
+ *   CDict}() rely on a pointer comparison to see whether one or two frees are
+ *   required.
+ *
+ * - Fixed size objects: these are fixed-size, fixed-count objects that are
+ *   nonetheless "dynamically" allocated in the workspace so that we can
+ *   control how they're initialized separately from the broader ZSTD_CCtx.
+ *   Examples:
+ *   - Entropy Workspace
+ *   - 2 x ZSTD_compressedBlockState_t
+ *   - CDict dictionary contents
+ *
+ * - Tables: these are any of several different datastructures (hash tables,
+ *   chain tables, binary trees) that all respect a common format: they are
+ *   uint32_t arrays, all of whose values are between 0 and (nextSrc - base).
+ *   Their sizes depend on the cparams.
+ *
+ * - Aligned: these buffers are used for various purposes that require 4 byte
+ *   alignment, but don't require any initialization before they're used.
+ *
+ * - Buffers: these buffers are used for various purposes that don't require
+ *   any alignment or initialization before they're used. This means they can
+ *   be moved around at no cost for a new compression.
+ *
+ * Allocating Memory:
+ *
+ * The various types of objects must be allocated in order, so they can be
+ * correctly packed into the workspace buffer. That order is:
+ *
+ * 1. Objects
+ * 2. Buffers
+ * 3. Aligned
+ * 4. Tables
+ *
+ * Attempts to reserve objects of different types out of order will fail.
+ */
+typedef struct {
+    void* workspace;
+    void* workspaceEnd;
+
+    void* objectEnd;
+    void* tableEnd;
+    void* tableValidEnd;
+    void* allocStart;
+
+    BYTE allocFailed;
+    int workspaceOversizedDuration;
+    ZSTD_cwksp_alloc_phase_e phase;
+    ZSTD_cwksp_static_alloc_e isStatic;
+} ZSTD_cwksp;
+
+/*-*************************************
+*  Functions
+***************************************/
+
+MEM_STATIC size_t ZSTD_cwksp_available_space(ZSTD_cwksp* ws);
+
+MEM_STATIC void ZSTD_cwksp_assert_internal_consistency(ZSTD_cwksp* ws) {
+    (void)ws;
+    assert(ws->workspace <= ws->objectEnd);
+    assert(ws->objectEnd <= ws->tableEnd);
+    assert(ws->objectEnd <= ws->tableValidEnd);
+    assert(ws->tableEnd <= ws->allocStart);
+    assert(ws->tableValidEnd <= ws->allocStart);
+    assert(ws->allocStart <= ws->workspaceEnd);
+}
+
+/*
+ * Align must be a power of 2.
+ */
+MEM_STATIC size_t ZSTD_cwksp_align(size_t size, size_t const align) {
+    size_t const mask = align - 1;
+    assert((align & mask) == 0);
+    return (size + mask) & ~mask;
+}
+
+/*
+ * Use this to determine how much space in the workspace we will consume to
+ * allocate this object. (Normally it should be exactly the size of the object,
+ * but under special conditions, like ASAN, where we pad each object, it might
+ * be larger.)
+ *
+ * Since tables aren't currently redzoned, you don't need to call through this
+ * to figure out how much space you need for the matchState tables. Everything
+ * else is though.
+ */
+MEM_STATIC size_t ZSTD_cwksp_alloc_size(size_t size) {
+    if (size == 0)
+        return 0;
+    return size;
+}
+
+MEM_STATIC void ZSTD_cwksp_internal_advance_phase(
+        ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase) {
+    assert(phase >= ws->phase);
+    if (phase > ws->phase) {
+        if (ws->phase < ZSTD_cwksp_alloc_buffers &&
+                phase >= ZSTD_cwksp_alloc_buffers) {
+            ws->tableValidEnd = ws->objectEnd;
+        }
+        if (ws->phase < ZSTD_cwksp_alloc_aligned &&
+                phase >= ZSTD_cwksp_alloc_aligned) {
+            /* If unaligned allocations down from a too-large top have left us
+             * unaligned, we need to realign our alloc ptr. Technically, this
+             * can consume space that is unaccounted for in the neededSpace
+             * calculation. However, I believe this can only happen when the
+             * workspace is too large, and specifically when it is too large
+             * by a larger margin than the space that will be consumed. */
+            /* TODO: cleaner, compiler warning friendly way to do this??? */
+            ws->allocStart = (BYTE*)ws->allocStart - ((size_t)ws->allocStart & (sizeof(U32)-1));
+            if (ws->allocStart < ws->tableValidEnd) {
+                ws->tableValidEnd = ws->allocStart;
+            }
+        }
+        ws->phase = phase;
+    }
+}
+
+/*
+ * Returns whether this object/buffer/etc was allocated in this workspace.
+ */
+MEM_STATIC int ZSTD_cwksp_owns_buffer(const ZSTD_cwksp* ws, const void* ptr) {
+    return (ptr != NULL) && (ws->workspace <= ptr) && (ptr <= ws->workspaceEnd);
+}
+
+/*
+ * Internal function. Do not use directly.
+ */
+MEM_STATIC void* ZSTD_cwksp_reserve_internal(
+        ZSTD_cwksp* ws, size_t bytes, ZSTD_cwksp_alloc_phase_e phase) {
+    void* alloc;
+    void* bottom = ws->tableEnd;
+    ZSTD_cwksp_internal_advance_phase(ws, phase);
+    alloc = (BYTE *)ws->allocStart - bytes;
+
+    if (bytes == 0)
+        return NULL;
+
+
+    DEBUGLOG(5, "cwksp: reserving %p %zd bytes, %zd bytes remaining",
+        alloc, bytes, ZSTD_cwksp_available_space(ws) - bytes);
+    ZSTD_cwksp_assert_internal_consistency(ws);
+    assert(alloc >= bottom);
+    if (alloc < bottom) {
+        DEBUGLOG(4, "cwksp: alloc failed!");
+        ws->allocFailed = 1;
+        return NULL;
+    }
+    if (alloc < ws->tableValidEnd) {
+        ws->tableValidEnd = alloc;
+    }
+    ws->allocStart = alloc;
+
+
+    return alloc;
+}
+
+/*
+ * Reserves and returns unaligned memory.
+ */
+MEM_STATIC BYTE* ZSTD_cwksp_reserve_buffer(ZSTD_cwksp* ws, size_t bytes) {
+    return (BYTE*)ZSTD_cwksp_reserve_internal(ws, bytes, ZSTD_cwksp_alloc_buffers);
+}
+
+/*
+ * Reserves and returns memory sized on and aligned on sizeof(unsigned).
+ */
+MEM_STATIC void* ZSTD_cwksp_reserve_aligned(ZSTD_cwksp* ws, size_t bytes) {
+    assert((bytes & (sizeof(U32)-1)) == 0);
+    return ZSTD_cwksp_reserve_internal(ws, ZSTD_cwksp_align(bytes, sizeof(U32)), ZSTD_cwksp_alloc_aligned);
+}
+
+/*
+ * Aligned on sizeof(unsigned). These buffers have the special property that
+ * their values remain constrained, allowing us to re-use them without
+ * memset()-ing them.
+ */
+MEM_STATIC void* ZSTD_cwksp_reserve_table(ZSTD_cwksp* ws, size_t bytes) {
+    const ZSTD_cwksp_alloc_phase_e phase = ZSTD_cwksp_alloc_aligned;
+    void* alloc = ws->tableEnd;
+    void* end = (BYTE *)alloc + bytes;
+    void* top = ws->allocStart;
+
+    DEBUGLOG(5, "cwksp: reserving %p table %zd bytes, %zd bytes remaining",
+        alloc, bytes, ZSTD_cwksp_available_space(ws) - bytes);
+    assert((bytes & (sizeof(U32)-1)) == 0);
+    ZSTD_cwksp_internal_advance_phase(ws, phase);
+    ZSTD_cwksp_assert_internal_consistency(ws);
+    assert(end <= top);
+    if (end > top) {
+        DEBUGLOG(4, "cwksp: table alloc failed!");
+        ws->allocFailed = 1;
+        return NULL;
+    }
+    ws->tableEnd = end;
+
+
+    return alloc;
+}
+
+/*
+ * Aligned on sizeof(void*).
+ */
+MEM_STATIC void* ZSTD_cwksp_reserve_object(ZSTD_cwksp* ws, size_t bytes) {
+    size_t roundedBytes = ZSTD_cwksp_align(bytes, sizeof(void*));
+    void* alloc = ws->objectEnd;
+    void* end = (BYTE*)alloc + roundedBytes;
+
+
+    DEBUGLOG(5,
+        "cwksp: reserving %p object %zd bytes (rounded to %zd), %zd bytes remaining",
+        alloc, bytes, roundedBytes, ZSTD_cwksp_available_space(ws) - roundedBytes);
+    assert(((size_t)alloc & (sizeof(void*)-1)) == 0);
+    assert((bytes & (sizeof(void*)-1)) == 0);
+    ZSTD_cwksp_assert_internal_consistency(ws);
+    /* we must be in the first phase, no advance is possible */
+    if (ws->phase != ZSTD_cwksp_alloc_objects || end > ws->workspaceEnd) {
+        DEBUGLOG(4, "cwksp: object alloc failed!");
+        ws->allocFailed = 1;
+        return NULL;
+    }
+    ws->objectEnd = end;
+    ws->tableEnd = end;
+    ws->tableValidEnd = end;
+
+
+    return alloc;
+}
+
+MEM_STATIC void ZSTD_cwksp_mark_tables_dirty(ZSTD_cwksp* ws) {
+    DEBUGLOG(4, "cwksp: ZSTD_cwksp_mark_tables_dirty");
+
+
+    assert(ws->tableValidEnd >= ws->objectEnd);
+    assert(ws->tableValidEnd <= ws->allocStart);
+    ws->tableValidEnd = ws->objectEnd;
+    ZSTD_cwksp_assert_internal_consistency(ws);
+}
+
+MEM_STATIC void ZSTD_cwksp_mark_tables_clean(ZSTD_cwksp* ws) {
+    DEBUGLOG(4, "cwksp: ZSTD_cwksp_mark_tables_clean");
+    assert(ws->tableValidEnd >= ws->objectEnd);
+    assert(ws->tableValidEnd <= ws->allocStart);
+    if (ws->tableValidEnd < ws->tableEnd) {
+        ws->tableValidEnd = ws->tableEnd;
+    }
+    ZSTD_cwksp_assert_internal_consistency(ws);
+}
+
+/*
+ * Zero the part of the allocated tables not already marked clean.
+ */
+MEM_STATIC void ZSTD_cwksp_clean_tables(ZSTD_cwksp* ws) {
+    DEBUGLOG(4, "cwksp: ZSTD_cwksp_clean_tables");
+    assert(ws->tableValidEnd >= ws->objectEnd);
+    assert(ws->tableValidEnd <= ws->allocStart);
+    if (ws->tableValidEnd < ws->tableEnd) {
+        ZSTD_memset(ws->tableValidEnd, 0, (BYTE*)ws->tableEnd - (BYTE*)ws->tableValidEnd);
+    }
+    ZSTD_cwksp_mark_tables_clean(ws);
+}
+
+/*
+ * Invalidates table allocations.
+ * All other allocations remain valid.
+ */
+MEM_STATIC void ZSTD_cwksp_clear_tables(ZSTD_cwksp* ws) {
+    DEBUGLOG(4, "cwksp: clearing tables!");
+
+
+    ws->tableEnd = ws->objectEnd;
+    ZSTD_cwksp_assert_internal_consistency(ws);
+}
+
+/*
+ * Invalidates all buffer, aligned, and table allocations.
+ * Object allocations remain valid.
+ */
+MEM_STATIC void ZSTD_cwksp_clear(ZSTD_cwksp* ws) {
+    DEBUGLOG(4, "cwksp: clearing!");
+
+
+
+    ws->tableEnd = ws->objectEnd;
+    ws->allocStart = ws->workspaceEnd;
+    ws->allocFailed = 0;
+    if (ws->phase > ZSTD_cwksp_alloc_buffers) {
+        ws->phase = ZSTD_cwksp_alloc_buffers;
+    }
+    ZSTD_cwksp_assert_internal_consistency(ws);
+}
+
+/*
+ * The provided workspace takes ownership of the buffer [start, start+size).
+ * Any existing values in the workspace are ignored (the previously managed
+ * buffer, if present, must be separately freed).
+ */
+MEM_STATIC void ZSTD_cwksp_init(ZSTD_cwksp* ws, void* start, size_t size, ZSTD_cwksp_static_alloc_e isStatic) {
+    DEBUGLOG(4, "cwksp: init'ing workspace with %zd bytes", size);
+    assert(((size_t)start & (sizeof(void*)-1)) == 0); /* ensure correct alignment */
+    ws->workspace = start;
+    ws->workspaceEnd = (BYTE*)start + size;
+    ws->objectEnd = ws->workspace;
+    ws->tableValidEnd = ws->objectEnd;
+    ws->phase = ZSTD_cwksp_alloc_objects;
+    ws->isStatic = isStatic;
+    ZSTD_cwksp_clear(ws);
+    ws->workspaceOversizedDuration = 0;
+    ZSTD_cwksp_assert_internal_consistency(ws);
+}
+
+MEM_STATIC size_t ZSTD_cwksp_create(ZSTD_cwksp* ws, size_t size, ZSTD_customMem customMem) {
+    void* workspace = ZSTD_customMalloc(size, customMem);
+    DEBUGLOG(4, "cwksp: creating new workspace with %zd bytes", size);
+    RETURN_ERROR_IF(workspace == NULL, memory_allocation, "NULL pointer!");
+    ZSTD_cwksp_init(ws, workspace, size, ZSTD_cwksp_dynamic_alloc);
+    return 0;
+}
+
+MEM_STATIC void ZSTD_cwksp_free(ZSTD_cwksp* ws, ZSTD_customMem customMem) {
+    void *ptr = ws->workspace;
+    DEBUGLOG(4, "cwksp: freeing workspace");
+    ZSTD_memset(ws, 0, sizeof(ZSTD_cwksp));
+    ZSTD_customFree(ptr, customMem);
+}
+
+/*
+ * Moves the management of a workspace from one cwksp to another. The src cwksp
+ * is left in an invalid state (src must be re-init()'ed before it's used again).
+ */
+MEM_STATIC void ZSTD_cwksp_move(ZSTD_cwksp* dst, ZSTD_cwksp* src) {
+    *dst = *src;
+    ZSTD_memset(src, 0, sizeof(ZSTD_cwksp));
+}
+
+MEM_STATIC size_t ZSTD_cwksp_sizeof(const ZSTD_cwksp* ws) {
+    return (size_t)((BYTE*)ws->workspaceEnd - (BYTE*)ws->workspace);
+}
+
+MEM_STATIC size_t ZSTD_cwksp_used(const ZSTD_cwksp* ws) {
+    return (size_t)((BYTE*)ws->tableEnd - (BYTE*)ws->workspace)
+         + (size_t)((BYTE*)ws->workspaceEnd - (BYTE*)ws->allocStart);
+}
+
+MEM_STATIC int ZSTD_cwksp_reserve_failed(const ZSTD_cwksp* ws) {
+    return ws->allocFailed;
+}
+
+/*-*************************************
+*  Functions Checking Free Space
+***************************************/
+
+MEM_STATIC size_t ZSTD_cwksp_available_space(ZSTD_cwksp* ws) {
+    return (size_t)((BYTE*)ws->allocStart - (BYTE*)ws->tableEnd);
+}
+
+MEM_STATIC int ZSTD_cwksp_check_available(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
+    return ZSTD_cwksp_available_space(ws) >= additionalNeededSpace;
+}
+
+MEM_STATIC int ZSTD_cwksp_check_too_large(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
+    return ZSTD_cwksp_check_available(
+        ws, additionalNeededSpace * ZSTD_WORKSPACETOOLARGE_FACTOR);
+}
+
+MEM_STATIC int ZSTD_cwksp_check_wasteful(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
+    return ZSTD_cwksp_check_too_large(ws, additionalNeededSpace)
+        && ws->workspaceOversizedDuration > ZSTD_WORKSPACETOOLARGE_MAXDURATION;
+}
+
+MEM_STATIC void ZSTD_cwksp_bump_oversized_duration(
+        ZSTD_cwksp* ws, size_t additionalNeededSpace) {
+    if (ZSTD_cwksp_check_too_large(ws, additionalNeededSpace)) {
+        ws->workspaceOversizedDuration++;
+    } else {
+        ws->workspaceOversizedDuration = 0;
+    }
+}
+
+
+#endif /* ZSTD_CWKSP_H */
diff --git a/lib/zstd/compress/zstd_double_fast.c b/lib/zstd/compress/zstd_double_fast.c
new file mode 100644
index 0000000..b0424d2
--- /dev/null
+++ b/lib/zstd/compress/zstd_double_fast.c
@@ -0,0 +1,519 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#include "zstd_compress_internal.h"
+#include "zstd_double_fast.h"
+
+
+void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
+                              void const* end, ZSTD_dictTableLoadMethod_e dtlm)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32* const hashLarge = ms->hashTable;
+    U32  const hBitsL = cParams->hashLog;
+    U32  const mls = cParams->minMatch;
+    U32* const hashSmall = ms->chainTable;
+    U32  const hBitsS = cParams->chainLog;
+    const BYTE* const base = ms->window.base;
+    const BYTE* ip = base + ms->nextToUpdate;
+    const BYTE* const iend = ((const BYTE*)end) - HASH_READ_SIZE;
+    const U32 fastHashFillStep = 3;
+
+    /* Always insert every fastHashFillStep position into the hash tables.
+     * Insert the other positions into the large hash table if their entry
+     * is empty.
+     */
+    for (; ip + fastHashFillStep - 1 <= iend; ip += fastHashFillStep) {
+        U32 const curr = (U32)(ip - base);
+        U32 i;
+        for (i = 0; i < fastHashFillStep; ++i) {
+            size_t const smHash = ZSTD_hashPtr(ip + i, hBitsS, mls);
+            size_t const lgHash = ZSTD_hashPtr(ip + i, hBitsL, 8);
+            if (i == 0)
+                hashSmall[smHash] = curr + i;
+            if (i == 0 || hashLarge[lgHash] == 0)
+                hashLarge[lgHash] = curr + i;
+            /* Only load extra positions for ZSTD_dtlm_full */
+            if (dtlm == ZSTD_dtlm_fast)
+                break;
+    }   }
+}
+
+
+FORCE_INLINE_TEMPLATE
+size_t ZSTD_compressBlock_doubleFast_generic(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize,
+        U32 const mls /* template */, ZSTD_dictMode_e const dictMode)
+{
+    ZSTD_compressionParameters const* cParams = &ms->cParams;
+    U32* const hashLong = ms->hashTable;
+    const U32 hBitsL = cParams->hashLog;
+    U32* const hashSmall = ms->chainTable;
+    const U32 hBitsS = cParams->chainLog;
+    const BYTE* const base = ms->window.base;
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* ip = istart;
+    const BYTE* anchor = istart;
+    const U32 endIndex = (U32)((size_t)(istart - base) + srcSize);
+    /* presumes that, if there is a dictionary, it must be using Attach mode */
+    const U32 prefixLowestIndex = ZSTD_getLowestPrefixIndex(ms, endIndex, cParams->windowLog);
+    const BYTE* const prefixLowest = base + prefixLowestIndex;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* const ilimit = iend - HASH_READ_SIZE;
+    U32 offset_1=rep[0], offset_2=rep[1];
+    U32 offsetSaved = 0;
+
+    const ZSTD_matchState_t* const dms = ms->dictMatchState;
+    const ZSTD_compressionParameters* const dictCParams =
+                                     dictMode == ZSTD_dictMatchState ?
+                                     &dms->cParams : NULL;
+    const U32* const dictHashLong  = dictMode == ZSTD_dictMatchState ?
+                                     dms->hashTable : NULL;
+    const U32* const dictHashSmall = dictMode == ZSTD_dictMatchState ?
+                                     dms->chainTable : NULL;
+    const U32 dictStartIndex       = dictMode == ZSTD_dictMatchState ?
+                                     dms->window.dictLimit : 0;
+    const BYTE* const dictBase     = dictMode == ZSTD_dictMatchState ?
+                                     dms->window.base : NULL;
+    const BYTE* const dictStart    = dictMode == ZSTD_dictMatchState ?
+                                     dictBase + dictStartIndex : NULL;
+    const BYTE* const dictEnd      = dictMode == ZSTD_dictMatchState ?
+                                     dms->window.nextSrc : NULL;
+    const U32 dictIndexDelta       = dictMode == ZSTD_dictMatchState ?
+                                     prefixLowestIndex - (U32)(dictEnd - dictBase) :
+                                     0;
+    const U32 dictHBitsL           = dictMode == ZSTD_dictMatchState ?
+                                     dictCParams->hashLog : hBitsL;
+    const U32 dictHBitsS           = dictMode == ZSTD_dictMatchState ?
+                                     dictCParams->chainLog : hBitsS;
+    const U32 dictAndPrefixLength  = (U32)((ip - prefixLowest) + (dictEnd - dictStart));
+
+    DEBUGLOG(5, "ZSTD_compressBlock_doubleFast_generic");
+
+    assert(dictMode == ZSTD_noDict || dictMode == ZSTD_dictMatchState);
+
+    /* if a dictionary is attached, it must be within window range */
+    if (dictMode == ZSTD_dictMatchState) {
+        assert(ms->window.dictLimit + (1U << cParams->windowLog) >= endIndex);
+    }
+
+    /* init */
+    ip += (dictAndPrefixLength == 0);
+    if (dictMode == ZSTD_noDict) {
+        U32 const curr = (U32)(ip - base);
+        U32 const windowLow = ZSTD_getLowestPrefixIndex(ms, curr, cParams->windowLog);
+        U32 const maxRep = curr - windowLow;
+        if (offset_2 > maxRep) offsetSaved = offset_2, offset_2 = 0;
+        if (offset_1 > maxRep) offsetSaved = offset_1, offset_1 = 0;
+    }
+    if (dictMode == ZSTD_dictMatchState) {
+        /* dictMatchState repCode checks don't currently handle repCode == 0
+         * disabling. */
+        assert(offset_1 <= dictAndPrefixLength);
+        assert(offset_2 <= dictAndPrefixLength);
+    }
+
+    /* Main Search Loop */
+    while (ip < ilimit) {   /* < instead of <=, because repcode check at (ip+1) */
+        size_t mLength;
+        U32 offset;
+        size_t const h2 = ZSTD_hashPtr(ip, hBitsL, 8);
+        size_t const h = ZSTD_hashPtr(ip, hBitsS, mls);
+        size_t const dictHL = ZSTD_hashPtr(ip, dictHBitsL, 8);
+        size_t const dictHS = ZSTD_hashPtr(ip, dictHBitsS, mls);
+        U32 const curr = (U32)(ip-base);
+        U32 const matchIndexL = hashLong[h2];
+        U32 matchIndexS = hashSmall[h];
+        const BYTE* matchLong = base + matchIndexL;
+        const BYTE* match = base + matchIndexS;
+        const U32 repIndex = curr + 1 - offset_1;
+        const BYTE* repMatch = (dictMode == ZSTD_dictMatchState
+                            && repIndex < prefixLowestIndex) ?
+                               dictBase + (repIndex - dictIndexDelta) :
+                               base + repIndex;
+        hashLong[h2] = hashSmall[h] = curr;   /* update hash tables */
+
+        /* check dictMatchState repcode */
+        if (dictMode == ZSTD_dictMatchState
+            && ((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
+            && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
+            const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
+            mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
+            ip++;
+            ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
+            goto _match_stored;
+        }
+
+        /* check noDict repcode */
+        if ( dictMode == ZSTD_noDict
+          && ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) {
+            mLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
+            ip++;
+            ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
+            goto _match_stored;
+        }
+
+        if (matchIndexL > prefixLowestIndex) {
+            /* check prefix long match */
+            if (MEM_read64(matchLong) == MEM_read64(ip)) {
+                mLength = ZSTD_count(ip+8, matchLong+8, iend) + 8;
+                offset = (U32)(ip-matchLong);
+                while (((ip>anchor) & (matchLong>prefixLowest)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; } /* catch up */
+                goto _match_found;
+            }
+        } else if (dictMode == ZSTD_dictMatchState) {
+            /* check dictMatchState long match */
+            U32 const dictMatchIndexL = dictHashLong[dictHL];
+            const BYTE* dictMatchL = dictBase + dictMatchIndexL;
+            assert(dictMatchL < dictEnd);
+
+            if (dictMatchL > dictStart && MEM_read64(dictMatchL) == MEM_read64(ip)) {
+                mLength = ZSTD_count_2segments(ip+8, dictMatchL+8, iend, dictEnd, prefixLowest) + 8;
+                offset = (U32)(curr - dictMatchIndexL - dictIndexDelta);
+                while (((ip>anchor) & (dictMatchL>dictStart)) && (ip[-1] == dictMatchL[-1])) { ip--; dictMatchL--; mLength++; } /* catch up */
+                goto _match_found;
+        }   }
+
+        if (matchIndexS > prefixLowestIndex) {
+            /* check prefix short match */
+            if (MEM_read32(match) == MEM_read32(ip)) {
+                goto _search_next_long;
+            }
+        } else if (dictMode == ZSTD_dictMatchState) {
+            /* check dictMatchState short match */
+            U32 const dictMatchIndexS = dictHashSmall[dictHS];
+            match = dictBase + dictMatchIndexS;
+            matchIndexS = dictMatchIndexS + dictIndexDelta;
+
+            if (match > dictStart && MEM_read32(match) == MEM_read32(ip)) {
+                goto _search_next_long;
+        }   }
+
+        ip += ((ip-anchor) >> kSearchStrength) + 1;
+#if defined(__aarch64__)
+        PREFETCH_L1(ip+256);
+#endif
+        continue;
+
+_search_next_long:
+
+        {   size_t const hl3 = ZSTD_hashPtr(ip+1, hBitsL, 8);
+            size_t const dictHLNext = ZSTD_hashPtr(ip+1, dictHBitsL, 8);
+            U32 const matchIndexL3 = hashLong[hl3];
+            const BYTE* matchL3 = base + matchIndexL3;
+            hashLong[hl3] = curr + 1;
+
+            /* check prefix long +1 match */
+            if (matchIndexL3 > prefixLowestIndex) {
+                if (MEM_read64(matchL3) == MEM_read64(ip+1)) {
+                    mLength = ZSTD_count(ip+9, matchL3+8, iend) + 8;
+                    ip++;
+                    offset = (U32)(ip-matchL3);
+                    while (((ip>anchor) & (matchL3>prefixLowest)) && (ip[-1] == matchL3[-1])) { ip--; matchL3--; mLength++; } /* catch up */
+                    goto _match_found;
+                }
+            } else if (dictMode == ZSTD_dictMatchState) {
+                /* check dict long +1 match */
+                U32 const dictMatchIndexL3 = dictHashLong[dictHLNext];
+                const BYTE* dictMatchL3 = dictBase + dictMatchIndexL3;
+                assert(dictMatchL3 < dictEnd);
+                if (dictMatchL3 > dictStart && MEM_read64(dictMatchL3) == MEM_read64(ip+1)) {
+                    mLength = ZSTD_count_2segments(ip+1+8, dictMatchL3+8, iend, dictEnd, prefixLowest) + 8;
+                    ip++;
+                    offset = (U32)(curr + 1 - dictMatchIndexL3 - dictIndexDelta);
+                    while (((ip>anchor) & (dictMatchL3>dictStart)) && (ip[-1] == dictMatchL3[-1])) { ip--; dictMatchL3--; mLength++; } /* catch up */
+                    goto _match_found;
+        }   }   }
+
+        /* if no long +1 match, explore the short match we found */
+        if (dictMode == ZSTD_dictMatchState && matchIndexS < prefixLowestIndex) {
+            mLength = ZSTD_count_2segments(ip+4, match+4, iend, dictEnd, prefixLowest) + 4;
+            offset = (U32)(curr - matchIndexS);
+            while (((ip>anchor) & (match>dictStart)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
+        } else {
+            mLength = ZSTD_count(ip+4, match+4, iend) + 4;
+            offset = (U32)(ip - match);
+            while (((ip>anchor) & (match>prefixLowest)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
+        }
+
+_match_found:
+        offset_2 = offset_1;
+        offset_1 = offset;
+
+        ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
+
+_match_stored:
+        /* match found */
+        ip += mLength;
+        anchor = ip;
+
+        if (ip <= ilimit) {
+            /* Complementary insertion */
+            /* done after iLimit test, as candidates could be > iend-8 */
+            {   U32 const indexToInsert = curr+2;
+                hashLong[ZSTD_hashPtr(base+indexToInsert, hBitsL, 8)] = indexToInsert;
+                hashLong[ZSTD_hashPtr(ip-2, hBitsL, 8)] = (U32)(ip-2-base);
+                hashSmall[ZSTD_hashPtr(base+indexToInsert, hBitsS, mls)] = indexToInsert;
+                hashSmall[ZSTD_hashPtr(ip-1, hBitsS, mls)] = (U32)(ip-1-base);
+            }
+
+            /* check immediate repcode */
+            if (dictMode == ZSTD_dictMatchState) {
+                while (ip <= ilimit) {
+                    U32 const current2 = (U32)(ip-base);
+                    U32 const repIndex2 = current2 - offset_2;
+                    const BYTE* repMatch2 = dictMode == ZSTD_dictMatchState
+                        && repIndex2 < prefixLowestIndex ?
+                            dictBase + repIndex2 - dictIndexDelta :
+                            base + repIndex2;
+                    if ( ((U32)((prefixLowestIndex-1) - (U32)repIndex2) >= 3 /* intentional overflow */)
+                       && (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
+                        const BYTE* const repEnd2 = repIndex2 < prefixLowestIndex ? dictEnd : iend;
+                        size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixLowest) + 4;
+                        U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset;   /* swap offset_2 <=> offset_1 */
+                        ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
+                        hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2;
+                        hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2;
+                        ip += repLength2;
+                        anchor = ip;
+                        continue;
+                    }
+                    break;
+            }   }
+
+            if (dictMode == ZSTD_noDict) {
+                while ( (ip <= ilimit)
+                     && ( (offset_2>0)
+                        & (MEM_read32(ip) == MEM_read32(ip - offset_2)) )) {
+                    /* store sequence */
+                    size_t const rLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
+                    U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff;  /* swap offset_2 <=> offset_1 */
+                    hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip-base);
+                    hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip-base);
+                    ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, rLength-MINMATCH);
+                    ip += rLength;
+                    anchor = ip;
+                    continue;   /* faster when present ... (?) */
+        }   }   }
+    }   /* while (ip < ilimit) */
+
+    /* save reps for next block */
+    rep[0] = offset_1 ? offset_1 : offsetSaved;
+    rep[1] = offset_2 ? offset_2 : offsetSaved;
+
+    /* Return the last literals size */
+    return (size_t)(iend - anchor);
+}
+
+
+size_t ZSTD_compressBlock_doubleFast(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    const U32 mls = ms->cParams.minMatch;
+    switch(mls)
+    {
+    default: /* includes case 3 */
+    case 4 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 4, ZSTD_noDict);
+    case 5 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 5, ZSTD_noDict);
+    case 6 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 6, ZSTD_noDict);
+    case 7 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 7, ZSTD_noDict);
+    }
+}
+
+
+size_t ZSTD_compressBlock_doubleFast_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    const U32 mls = ms->cParams.minMatch;
+    switch(mls)
+    {
+    default: /* includes case 3 */
+    case 4 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 4, ZSTD_dictMatchState);
+    case 5 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 5, ZSTD_dictMatchState);
+    case 6 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 6, ZSTD_dictMatchState);
+    case 7 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 7, ZSTD_dictMatchState);
+    }
+}
+
+
+static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize,
+        U32 const mls /* template */)
+{
+    ZSTD_compressionParameters const* cParams = &ms->cParams;
+    U32* const hashLong = ms->hashTable;
+    U32  const hBitsL = cParams->hashLog;
+    U32* const hashSmall = ms->chainTable;
+    U32  const hBitsS = cParams->chainLog;
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* ip = istart;
+    const BYTE* anchor = istart;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* const ilimit = iend - 8;
+    const BYTE* const base = ms->window.base;
+    const U32   endIndex = (U32)((size_t)(istart - base) + srcSize);
+    const U32   lowLimit = ZSTD_getLowestMatchIndex(ms, endIndex, cParams->windowLog);
+    const U32   dictStartIndex = lowLimit;
+    const U32   dictLimit = ms->window.dictLimit;
+    const U32   prefixStartIndex = (dictLimit > lowLimit) ? dictLimit : lowLimit;
+    const BYTE* const prefixStart = base + prefixStartIndex;
+    const BYTE* const dictBase = ms->window.dictBase;
+    const BYTE* const dictStart = dictBase + dictStartIndex;
+    const BYTE* const dictEnd = dictBase + prefixStartIndex;
+    U32 offset_1=rep[0], offset_2=rep[1];
+
+    DEBUGLOG(5, "ZSTD_compressBlock_doubleFast_extDict_generic (srcSize=%zu)", srcSize);
+
+    /* if extDict is invalidated due to maxDistance, switch to "regular" variant */
+    if (prefixStartIndex == dictStartIndex)
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, mls, ZSTD_noDict);
+
+    /* Search Loop */
+    while (ip < ilimit) {  /* < instead of <=, because (ip+1) */
+        const size_t hSmall = ZSTD_hashPtr(ip, hBitsS, mls);
+        const U32 matchIndex = hashSmall[hSmall];
+        const BYTE* const matchBase = matchIndex < prefixStartIndex ? dictBase : base;
+        const BYTE* match = matchBase + matchIndex;
+
+        const size_t hLong = ZSTD_hashPtr(ip, hBitsL, 8);
+        const U32 matchLongIndex = hashLong[hLong];
+        const BYTE* const matchLongBase = matchLongIndex < prefixStartIndex ? dictBase : base;
+        const BYTE* matchLong = matchLongBase + matchLongIndex;
+
+        const U32 curr = (U32)(ip-base);
+        const U32 repIndex = curr + 1 - offset_1;   /* offset_1 expected <= curr +1 */
+        const BYTE* const repBase = repIndex < prefixStartIndex ? dictBase : base;
+        const BYTE* const repMatch = repBase + repIndex;
+        size_t mLength;
+        hashSmall[hSmall] = hashLong[hLong] = curr;   /* update hash table */
+
+        if ((((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow : ensure repIndex doesn't overlap dict + prefix */
+            & (repIndex > dictStartIndex))
+          && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
+            const BYTE* repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
+            mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
+            ip++;
+            ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
+        } else {
+            if ((matchLongIndex > dictStartIndex) && (MEM_read64(matchLong) == MEM_read64(ip))) {
+                const BYTE* const matchEnd = matchLongIndex < prefixStartIndex ? dictEnd : iend;
+                const BYTE* const lowMatchPtr = matchLongIndex < prefixStartIndex ? dictStart : prefixStart;
+                U32 offset;
+                mLength = ZSTD_count_2segments(ip+8, matchLong+8, iend, matchEnd, prefixStart) + 8;
+                offset = curr - matchLongIndex;
+                while (((ip>anchor) & (matchLong>lowMatchPtr)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; }   /* catch up */
+                offset_2 = offset_1;
+                offset_1 = offset;
+                ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
+
+            } else if ((matchIndex > dictStartIndex) && (MEM_read32(match) == MEM_read32(ip))) {
+                size_t const h3 = ZSTD_hashPtr(ip+1, hBitsL, 8);
+                U32 const matchIndex3 = hashLong[h3];
+                const BYTE* const match3Base = matchIndex3 < prefixStartIndex ? dictBase : base;
+                const BYTE* match3 = match3Base + matchIndex3;
+                U32 offset;
+                hashLong[h3] = curr + 1;
+                if ( (matchIndex3 > dictStartIndex) && (MEM_read64(match3) == MEM_read64(ip+1)) ) {
+                    const BYTE* const matchEnd = matchIndex3 < prefixStartIndex ? dictEnd : iend;
+                    const BYTE* const lowMatchPtr = matchIndex3 < prefixStartIndex ? dictStart : prefixStart;
+                    mLength = ZSTD_count_2segments(ip+9, match3+8, iend, matchEnd, prefixStart) + 8;
+                    ip++;
+                    offset = curr+1 - matchIndex3;
+                    while (((ip>anchor) & (match3>lowMatchPtr)) && (ip[-1] == match3[-1])) { ip--; match3--; mLength++; } /* catch up */
+                } else {
+                    const BYTE* const matchEnd = matchIndex < prefixStartIndex ? dictEnd : iend;
+                    const BYTE* const lowMatchPtr = matchIndex < prefixStartIndex ? dictStart : prefixStart;
+                    mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4;
+                    offset = curr - matchIndex;
+                    while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; }   /* catch up */
+                }
+                offset_2 = offset_1;
+                offset_1 = offset;
+                ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
+
+            } else {
+                ip += ((ip-anchor) >> kSearchStrength) + 1;
+                continue;
+        }   }
+
+        /* move to next sequence start */
+        ip += mLength;
+        anchor = ip;
+
+        if (ip <= ilimit) {
+            /* Complementary insertion */
+            /* done after iLimit test, as candidates could be > iend-8 */
+            {   U32 const indexToInsert = curr+2;
+                hashLong[ZSTD_hashPtr(base+indexToInsert, hBitsL, 8)] = indexToInsert;
+                hashLong[ZSTD_hashPtr(ip-2, hBitsL, 8)] = (U32)(ip-2-base);
+                hashSmall[ZSTD_hashPtr(base+indexToInsert, hBitsS, mls)] = indexToInsert;
+                hashSmall[ZSTD_hashPtr(ip-1, hBitsS, mls)] = (U32)(ip-1-base);
+            }
+
+            /* check immediate repcode */
+            while (ip <= ilimit) {
+                U32 const current2 = (U32)(ip-base);
+                U32 const repIndex2 = current2 - offset_2;
+                const BYTE* repMatch2 = repIndex2 < prefixStartIndex ? dictBase + repIndex2 : base + repIndex2;
+                if ( (((U32)((prefixStartIndex-1) - repIndex2) >= 3)   /* intentional overflow : ensure repIndex2 doesn't overlap dict + prefix */
+                    & (repIndex2 > dictStartIndex))
+                  && (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
+                    const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
+                    size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
+                    U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset;   /* swap offset_2 <=> offset_1 */
+                    ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
+                    hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2;
+                    hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2;
+                    ip += repLength2;
+                    anchor = ip;
+                    continue;
+                }
+                break;
+    }   }   }
+
+    /* save reps for next block */
+    rep[0] = offset_1;
+    rep[1] = offset_2;
+
+    /* Return the last literals size */
+    return (size_t)(iend - anchor);
+}
+
+
+size_t ZSTD_compressBlock_doubleFast_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    U32 const mls = ms->cParams.minMatch;
+    switch(mls)
+    {
+    default: /* includes case 3 */
+    case 4 :
+        return ZSTD_compressBlock_doubleFast_extDict_generic(ms, seqStore, rep, src, srcSize, 4);
+    case 5 :
+        return ZSTD_compressBlock_doubleFast_extDict_generic(ms, seqStore, rep, src, srcSize, 5);
+    case 6 :
+        return ZSTD_compressBlock_doubleFast_extDict_generic(ms, seqStore, rep, src, srcSize, 6);
+    case 7 :
+        return ZSTD_compressBlock_doubleFast_extDict_generic(ms, seqStore, rep, src, srcSize, 7);
+    }
+}
diff --git a/lib/zstd/compress/zstd_double_fast.h b/lib/zstd/compress/zstd_double_fast.h
new file mode 100644
index 0000000..6822bde
--- /dev/null
+++ b/lib/zstd/compress/zstd_double_fast.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_DOUBLE_FAST_H
+#define ZSTD_DOUBLE_FAST_H
+
+
+#include "../common/mem.h"      /* U32 */
+#include "zstd_compress_internal.h"     /* ZSTD_CCtx, size_t */
+
+void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
+                              void const* end, ZSTD_dictTableLoadMethod_e dtlm);
+size_t ZSTD_compressBlock_doubleFast(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_doubleFast_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_doubleFast_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+
+
+#endif /* ZSTD_DOUBLE_FAST_H */
diff --git a/lib/zstd/compress/zstd_fast.c b/lib/zstd/compress/zstd_fast.c
new file mode 100644
index 0000000..96b7d48
--- /dev/null
+++ b/lib/zstd/compress/zstd_fast.c
@@ -0,0 +1,496 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#include "zstd_compress_internal.h"  /* ZSTD_hashPtr, ZSTD_count, ZSTD_storeSeq */
+#include "zstd_fast.h"
+
+
+void ZSTD_fillHashTable(ZSTD_matchState_t* ms,
+                        const void* const end,
+                        ZSTD_dictTableLoadMethod_e dtlm)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32* const hashTable = ms->hashTable;
+    U32  const hBits = cParams->hashLog;
+    U32  const mls = cParams->minMatch;
+    const BYTE* const base = ms->window.base;
+    const BYTE* ip = base + ms->nextToUpdate;
+    const BYTE* const iend = ((const BYTE*)end) - HASH_READ_SIZE;
+    const U32 fastHashFillStep = 3;
+
+    /* Always insert every fastHashFillStep position into the hash table.
+     * Insert the other positions if their hash entry is empty.
+     */
+    for ( ; ip + fastHashFillStep < iend + 2; ip += fastHashFillStep) {
+        U32 const curr = (U32)(ip - base);
+        size_t const hash0 = ZSTD_hashPtr(ip, hBits, mls);
+        hashTable[hash0] = curr;
+        if (dtlm == ZSTD_dtlm_fast) continue;
+        /* Only load extra positions for ZSTD_dtlm_full */
+        {   U32 p;
+            for (p = 1; p < fastHashFillStep; ++p) {
+                size_t const hash = ZSTD_hashPtr(ip + p, hBits, mls);
+                if (hashTable[hash] == 0) {  /* not yet filled */
+                    hashTable[hash] = curr + p;
+    }   }   }   }
+}
+
+
+FORCE_INLINE_TEMPLATE size_t
+ZSTD_compressBlock_fast_generic(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize,
+        U32 const mls)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32* const hashTable = ms->hashTable;
+    U32 const hlog = cParams->hashLog;
+    /* support stepSize of 0 */
+    size_t const stepSize = cParams->targetLength + !(cParams->targetLength) + 1;
+    const BYTE* const base = ms->window.base;
+    const BYTE* const istart = (const BYTE*)src;
+    /* We check ip0 (ip + 0) and ip1 (ip + 1) each loop */
+    const BYTE* ip0 = istart;
+    const BYTE* ip1;
+    const BYTE* anchor = istart;
+    const U32   endIndex = (U32)((size_t)(istart - base) + srcSize);
+    const U32   prefixStartIndex = ZSTD_getLowestPrefixIndex(ms, endIndex, cParams->windowLog);
+    const BYTE* const prefixStart = base + prefixStartIndex;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* const ilimit = iend - HASH_READ_SIZE;
+    U32 offset_1=rep[0], offset_2=rep[1];
+    U32 offsetSaved = 0;
+
+    /* init */
+    DEBUGLOG(5, "ZSTD_compressBlock_fast_generic");
+    ip0 += (ip0 == prefixStart);
+    ip1 = ip0 + 1;
+    {   U32 const curr = (U32)(ip0 - base);
+        U32 const windowLow = ZSTD_getLowestPrefixIndex(ms, curr, cParams->windowLog);
+        U32 const maxRep = curr - windowLow;
+        if (offset_2 > maxRep) offsetSaved = offset_2, offset_2 = 0;
+        if (offset_1 > maxRep) offsetSaved = offset_1, offset_1 = 0;
+    }
+
+    /* Main Search Loop */
+#ifdef __INTEL_COMPILER
+    /* From intel 'The vector pragma indicates that the loop should be
+     * vectorized if it is legal to do so'. Can be used together with
+     * #pragma ivdep (but have opted to exclude that because intel
+     * warns against using it).*/
+    #pragma vector always
+#endif
+    while (ip1 < ilimit) {   /* < instead of <=, because check at ip0+2 */
+        size_t mLength;
+        BYTE const* ip2 = ip0 + 2;
+        size_t const h0 = ZSTD_hashPtr(ip0, hlog, mls);
+        U32 const val0 = MEM_read32(ip0);
+        size_t const h1 = ZSTD_hashPtr(ip1, hlog, mls);
+        U32 const val1 = MEM_read32(ip1);
+        U32 const current0 = (U32)(ip0-base);
+        U32 const current1 = (U32)(ip1-base);
+        U32 const matchIndex0 = hashTable[h0];
+        U32 const matchIndex1 = hashTable[h1];
+        BYTE const* repMatch = ip2 - offset_1;
+        const BYTE* match0 = base + matchIndex0;
+        const BYTE* match1 = base + matchIndex1;
+        U32 offcode;
+
+#if defined(__aarch64__)
+        PREFETCH_L1(ip0+256);
+#endif
+
+        hashTable[h0] = current0;   /* update hash table */
+        hashTable[h1] = current1;   /* update hash table */
+
+        assert(ip0 + 1 == ip1);
+
+        if ((offset_1 > 0) & (MEM_read32(repMatch) == MEM_read32(ip2))) {
+            mLength = (ip2[-1] == repMatch[-1]) ? 1 : 0;
+            ip0 = ip2 - mLength;
+            match0 = repMatch - mLength;
+            mLength += 4;
+            offcode = 0;
+            goto _match;
+        }
+        if ((matchIndex0 > prefixStartIndex) && MEM_read32(match0) == val0) {
+            /* found a regular match */
+            goto _offset;
+        }
+        if ((matchIndex1 > prefixStartIndex) && MEM_read32(match1) == val1) {
+            /* found a regular match after one literal */
+            ip0 = ip1;
+            match0 = match1;
+            goto _offset;
+        }
+        {   size_t const step = ((size_t)(ip0-anchor) >> (kSearchStrength - 1)) + stepSize;
+            assert(step >= 2);
+            ip0 += step;
+            ip1 += step;
+            continue;
+        }
+_offset: /* Requires: ip0, match0 */
+        /* Compute the offset code */
+        offset_2 = offset_1;
+        offset_1 = (U32)(ip0-match0);
+        offcode = offset_1 + ZSTD_REP_MOVE;
+        mLength = 4;
+        /* Count the backwards match length */
+        while (((ip0>anchor) & (match0>prefixStart))
+             && (ip0[-1] == match0[-1])) { ip0--; match0--; mLength++; } /* catch up */
+
+_match: /* Requires: ip0, match0, offcode */
+        /* Count the forward length */
+        mLength += ZSTD_count(ip0+mLength, match0+mLength, iend);
+        ZSTD_storeSeq(seqStore, (size_t)(ip0-anchor), anchor, iend, offcode, mLength-MINMATCH);
+        /* match found */
+        ip0 += mLength;
+        anchor = ip0;
+
+        if (ip0 <= ilimit) {
+            /* Fill Table */
+            assert(base+current0+2 > istart);  /* check base overflow */
+            hashTable[ZSTD_hashPtr(base+current0+2, hlog, mls)] = current0+2;  /* here because current+2 could be > iend-8 */
+            hashTable[ZSTD_hashPtr(ip0-2, hlog, mls)] = (U32)(ip0-2-base);
+
+            if (offset_2 > 0) { /* offset_2==0 means offset_2 is invalidated */
+                while ( (ip0 <= ilimit) && (MEM_read32(ip0) == MEM_read32(ip0 - offset_2)) ) {
+                    /* store sequence */
+                    size_t const rLength = ZSTD_count(ip0+4, ip0+4-offset_2, iend) + 4;
+                    { U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; } /* swap offset_2 <=> offset_1 */
+                    hashTable[ZSTD_hashPtr(ip0, hlog, mls)] = (U32)(ip0-base);
+                    ip0 += rLength;
+                    ZSTD_storeSeq(seqStore, 0 /*litLen*/, anchor, iend, 0 /*offCode*/, rLength-MINMATCH);
+                    anchor = ip0;
+                    continue;   /* faster when present (confirmed on gcc-8) ... (?) */
+        }   }   }
+        ip1 = ip0 + 1;
+    }
+
+    /* save reps for next block */
+    rep[0] = offset_1 ? offset_1 : offsetSaved;
+    rep[1] = offset_2 ? offset_2 : offsetSaved;
+
+    /* Return the last literals size */
+    return (size_t)(iend - anchor);
+}
+
+
+size_t ZSTD_compressBlock_fast(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    U32 const mls = ms->cParams.minMatch;
+    assert(ms->dictMatchState == NULL);
+    switch(mls)
+    {
+    default: /* includes case 3 */
+    case 4 :
+        return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 4);
+    case 5 :
+        return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 5);
+    case 6 :
+        return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 6);
+    case 7 :
+        return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 7);
+    }
+}
+
+FORCE_INLINE_TEMPLATE
+size_t ZSTD_compressBlock_fast_dictMatchState_generic(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize, U32 const mls)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32* const hashTable = ms->hashTable;
+    U32 const hlog = cParams->hashLog;
+    /* support stepSize of 0 */
+    U32 const stepSize = cParams->targetLength + !(cParams->targetLength);
+    const BYTE* const base = ms->window.base;
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* ip = istart;
+    const BYTE* anchor = istart;
+    const U32   prefixStartIndex = ms->window.dictLimit;
+    const BYTE* const prefixStart = base + prefixStartIndex;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* const ilimit = iend - HASH_READ_SIZE;
+    U32 offset_1=rep[0], offset_2=rep[1];
+    U32 offsetSaved = 0;
+
+    const ZSTD_matchState_t* const dms = ms->dictMatchState;
+    const ZSTD_compressionParameters* const dictCParams = &dms->cParams ;
+    const U32* const dictHashTable = dms->hashTable;
+    const U32 dictStartIndex       = dms->window.dictLimit;
+    const BYTE* const dictBase     = dms->window.base;
+    const BYTE* const dictStart    = dictBase + dictStartIndex;
+    const BYTE* const dictEnd      = dms->window.nextSrc;
+    const U32 dictIndexDelta       = prefixStartIndex - (U32)(dictEnd - dictBase);
+    const U32 dictAndPrefixLength  = (U32)(ip - prefixStart + dictEnd - dictStart);
+    const U32 dictHLog             = dictCParams->hashLog;
+
+    /* if a dictionary is still attached, it necessarily means that
+     * it is within window size. So we just check it. */
+    const U32 maxDistance = 1U << cParams->windowLog;
+    const U32 endIndex = (U32)((size_t)(ip - base) + srcSize);
+    assert(endIndex - prefixStartIndex <= maxDistance);
+    (void)maxDistance; (void)endIndex;   /* these variables are not used when assert() is disabled */
+
+    /* ensure there will be no underflow
+     * when translating a dict index into a local index */
+    assert(prefixStartIndex >= (U32)(dictEnd - dictBase));
+
+    /* init */
+    DEBUGLOG(5, "ZSTD_compressBlock_fast_dictMatchState_generic");
+    ip += (dictAndPrefixLength == 0);
+    /* dictMatchState repCode checks don't currently handle repCode == 0
+     * disabling. */
+    assert(offset_1 <= dictAndPrefixLength);
+    assert(offset_2 <= dictAndPrefixLength);
+
+    /* Main Search Loop */
+    while (ip < ilimit) {   /* < instead of <=, because repcode check at (ip+1) */
+        size_t mLength;
+        size_t const h = ZSTD_hashPtr(ip, hlog, mls);
+        U32 const curr = (U32)(ip-base);
+        U32 const matchIndex = hashTable[h];
+        const BYTE* match = base + matchIndex;
+        const U32 repIndex = curr + 1 - offset_1;
+        const BYTE* repMatch = (repIndex < prefixStartIndex) ?
+                               dictBase + (repIndex - dictIndexDelta) :
+                               base + repIndex;
+        hashTable[h] = curr;   /* update hash table */
+
+        if ( ((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow : ensure repIndex isn't overlapping dict + prefix */
+          && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
+            const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
+            mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
+            ip++;
+            ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
+        } else if ( (matchIndex <= prefixStartIndex) ) {
+            size_t const dictHash = ZSTD_hashPtr(ip, dictHLog, mls);
+            U32 const dictMatchIndex = dictHashTable[dictHash];
+            const BYTE* dictMatch = dictBase + dictMatchIndex;
+            if (dictMatchIndex <= dictStartIndex ||
+                MEM_read32(dictMatch) != MEM_read32(ip)) {
+                assert(stepSize >= 1);
+                ip += ((ip-anchor) >> kSearchStrength) + stepSize;
+                continue;
+            } else {
+                /* found a dict match */
+                U32 const offset = (U32)(curr-dictMatchIndex-dictIndexDelta);
+                mLength = ZSTD_count_2segments(ip+4, dictMatch+4, iend, dictEnd, prefixStart) + 4;
+                while (((ip>anchor) & (dictMatch>dictStart))
+                     && (ip[-1] == dictMatch[-1])) {
+                    ip--; dictMatch--; mLength++;
+                } /* catch up */
+                offset_2 = offset_1;
+                offset_1 = offset;
+                ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
+            }
+        } else if (MEM_read32(match) != MEM_read32(ip)) {
+            /* it's not a match, and we're not going to check the dictionary */
+            assert(stepSize >= 1);
+            ip += ((ip-anchor) >> kSearchStrength) + stepSize;
+            continue;
+        } else {
+            /* found a regular match */
+            U32 const offset = (U32)(ip-match);
+            mLength = ZSTD_count(ip+4, match+4, iend) + 4;
+            while (((ip>anchor) & (match>prefixStart))
+                 && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
+            offset_2 = offset_1;
+            offset_1 = offset;
+            ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
+        }
+
+        /* match found */
+        ip += mLength;
+        anchor = ip;
+
+        if (ip <= ilimit) {
+            /* Fill Table */
+            assert(base+curr+2 > istart);  /* check base overflow */
+            hashTable[ZSTD_hashPtr(base+curr+2, hlog, mls)] = curr+2;  /* here because curr+2 could be > iend-8 */
+            hashTable[ZSTD_hashPtr(ip-2, hlog, mls)] = (U32)(ip-2-base);
+
+            /* check immediate repcode */
+            while (ip <= ilimit) {
+                U32 const current2 = (U32)(ip-base);
+                U32 const repIndex2 = current2 - offset_2;
+                const BYTE* repMatch2 = repIndex2 < prefixStartIndex ?
+                        dictBase - dictIndexDelta + repIndex2 :
+                        base + repIndex2;
+                if ( ((U32)((prefixStartIndex-1) - (U32)repIndex2) >= 3 /* intentional overflow */)
+                   && (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
+                    const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
+                    size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
+                    U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset;   /* swap offset_2 <=> offset_1 */
+                    ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
+                    hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
+                    ip += repLength2;
+                    anchor = ip;
+                    continue;
+                }
+                break;
+            }
+        }
+    }
+
+    /* save reps for next block */
+    rep[0] = offset_1 ? offset_1 : offsetSaved;
+    rep[1] = offset_2 ? offset_2 : offsetSaved;
+
+    /* Return the last literals size */
+    return (size_t)(iend - anchor);
+}
+
+size_t ZSTD_compressBlock_fast_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    U32 const mls = ms->cParams.minMatch;
+    assert(ms->dictMatchState != NULL);
+    switch(mls)
+    {
+    default: /* includes case 3 */
+    case 4 :
+        return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 4);
+    case 5 :
+        return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 5);
+    case 6 :
+        return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 6);
+    case 7 :
+        return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 7);
+    }
+}
+
+
+static size_t ZSTD_compressBlock_fast_extDict_generic(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize, U32 const mls)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32* const hashTable = ms->hashTable;
+    U32 const hlog = cParams->hashLog;
+    /* support stepSize of 0 */
+    U32 const stepSize = cParams->targetLength + !(cParams->targetLength);
+    const BYTE* const base = ms->window.base;
+    const BYTE* const dictBase = ms->window.dictBase;
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* ip = istart;
+    const BYTE* anchor = istart;
+    const U32   endIndex = (U32)((size_t)(istart - base) + srcSize);
+    const U32   lowLimit = ZSTD_getLowestMatchIndex(ms, endIndex, cParams->windowLog);
+    const U32   dictStartIndex = lowLimit;
+    const BYTE* const dictStart = dictBase + dictStartIndex;
+    const U32   dictLimit = ms->window.dictLimit;
+    const U32   prefixStartIndex = dictLimit < lowLimit ? lowLimit : dictLimit;
+    const BYTE* const prefixStart = base + prefixStartIndex;
+    const BYTE* const dictEnd = dictBase + prefixStartIndex;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* const ilimit = iend - 8;
+    U32 offset_1=rep[0], offset_2=rep[1];
+
+    DEBUGLOG(5, "ZSTD_compressBlock_fast_extDict_generic (offset_1=%u)", offset_1);
+
+    /* switch to "regular" variant if extDict is invalidated due to maxDistance */
+    if (prefixStartIndex == dictStartIndex)
+        return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, mls);
+
+    /* Search Loop */
+    while (ip < ilimit) {  /* < instead of <=, because (ip+1) */
+        const size_t h = ZSTD_hashPtr(ip, hlog, mls);
+        const U32    matchIndex = hashTable[h];
+        const BYTE* const matchBase = matchIndex < prefixStartIndex ? dictBase : base;
+        const BYTE*  match = matchBase + matchIndex;
+        const U32    curr = (U32)(ip-base);
+        const U32    repIndex = curr + 1 - offset_1;
+        const BYTE* const repBase = repIndex < prefixStartIndex ? dictBase : base;
+        const BYTE* const repMatch = repBase + repIndex;
+        hashTable[h] = curr;   /* update hash table */
+        DEBUGLOG(7, "offset_1 = %u , curr = %u", offset_1, curr);
+        assert(offset_1 <= curr +1);   /* check repIndex */
+
+        if ( (((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > dictStartIndex))
+           && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
+            const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
+            size_t const rLength = ZSTD_count_2segments(ip+1 +4, repMatch +4, iend, repMatchEnd, prefixStart) + 4;
+            ip++;
+            ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, rLength-MINMATCH);
+            ip += rLength;
+            anchor = ip;
+        } else {
+            if ( (matchIndex < dictStartIndex) ||
+                 (MEM_read32(match) != MEM_read32(ip)) ) {
+                assert(stepSize >= 1);
+                ip += ((ip-anchor) >> kSearchStrength) + stepSize;
+                continue;
+            }
+            {   const BYTE* const matchEnd = matchIndex < prefixStartIndex ? dictEnd : iend;
+                const BYTE* const lowMatchPtr = matchIndex < prefixStartIndex ? dictStart : prefixStart;
+                U32 const offset = curr - matchIndex;
+                size_t mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4;
+                while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; }   /* catch up */
+                offset_2 = offset_1; offset_1 = offset;  /* update offset history */
+                ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
+                ip += mLength;
+                anchor = ip;
+        }   }
+
+        if (ip <= ilimit) {
+            /* Fill Table */
+            hashTable[ZSTD_hashPtr(base+curr+2, hlog, mls)] = curr+2;
+            hashTable[ZSTD_hashPtr(ip-2, hlog, mls)] = (U32)(ip-2-base);
+            /* check immediate repcode */
+            while (ip <= ilimit) {
+                U32 const current2 = (U32)(ip-base);
+                U32 const repIndex2 = current2 - offset_2;
+                const BYTE* const repMatch2 = repIndex2 < prefixStartIndex ? dictBase + repIndex2 : base + repIndex2;
+                if ( (((U32)((prefixStartIndex-1) - repIndex2) >= 3) & (repIndex2 > dictStartIndex))  /* intentional overflow */
+                   && (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
+                    const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
+                    size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
+                    { U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; }  /* swap offset_2 <=> offset_1 */
+                    ZSTD_storeSeq(seqStore, 0 /*litlen*/, anchor, iend, 0 /*offcode*/, repLength2-MINMATCH);
+                    hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
+                    ip += repLength2;
+                    anchor = ip;
+                    continue;
+                }
+                break;
+    }   }   }
+
+    /* save reps for next block */
+    rep[0] = offset_1;
+    rep[1] = offset_2;
+
+    /* Return the last literals size */
+    return (size_t)(iend - anchor);
+}
+
+
+size_t ZSTD_compressBlock_fast_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    U32 const mls = ms->cParams.minMatch;
+    switch(mls)
+    {
+    default: /* includes case 3 */
+    case 4 :
+        return ZSTD_compressBlock_fast_extDict_generic(ms, seqStore, rep, src, srcSize, 4);
+    case 5 :
+        return ZSTD_compressBlock_fast_extDict_generic(ms, seqStore, rep, src, srcSize, 5);
+    case 6 :
+        return ZSTD_compressBlock_fast_extDict_generic(ms, seqStore, rep, src, srcSize, 6);
+    case 7 :
+        return ZSTD_compressBlock_fast_extDict_generic(ms, seqStore, rep, src, srcSize, 7);
+    }
+}
diff --git a/lib/zstd/compress/zstd_fast.h b/lib/zstd/compress/zstd_fast.h
new file mode 100644
index 0000000..fddc2f5
--- /dev/null
+++ b/lib/zstd/compress/zstd_fast.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_FAST_H
+#define ZSTD_FAST_H
+
+
+#include "../common/mem.h"      /* U32 */
+#include "zstd_compress_internal.h"
+
+void ZSTD_fillHashTable(ZSTD_matchState_t* ms,
+                        void const* end, ZSTD_dictTableLoadMethod_e dtlm);
+size_t ZSTD_compressBlock_fast(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_fast_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_fast_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+
+#endif /* ZSTD_FAST_H */
diff --git a/lib/zstd/compress/zstd_lazy.c b/lib/zstd/compress/zstd_lazy.c
new file mode 100644
index 0000000..fb54d4e
--- /dev/null
+++ b/lib/zstd/compress/zstd_lazy.c
@@ -0,0 +1,1414 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#include "zstd_compress_internal.h"
+#include "zstd_lazy.h"
+
+
+/*-*************************************
+*  Binary Tree search
+***************************************/
+
+static void
+ZSTD_updateDUBT(ZSTD_matchState_t* ms,
+                const BYTE* ip, const BYTE* iend,
+                U32 mls)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32* const hashTable = ms->hashTable;
+    U32  const hashLog = cParams->hashLog;
+
+    U32* const bt = ms->chainTable;
+    U32  const btLog  = cParams->chainLog - 1;
+    U32  const btMask = (1 << btLog) - 1;
+
+    const BYTE* const base = ms->window.base;
+    U32 const target = (U32)(ip - base);
+    U32 idx = ms->nextToUpdate;
+
+    if (idx != target)
+        DEBUGLOG(7, "ZSTD_updateDUBT, from %u to %u (dictLimit:%u)",
+                    idx, target, ms->window.dictLimit);
+    assert(ip + 8 <= iend);   /* condition for ZSTD_hashPtr */
+    (void)iend;
+
+    assert(idx >= ms->window.dictLimit);   /* condition for valid base+idx */
+    for ( ; idx < target ; idx++) {
+        size_t const h  = ZSTD_hashPtr(base + idx, hashLog, mls);   /* assumption : ip + 8 <= iend */
+        U32    const matchIndex = hashTable[h];
+
+        U32*   const nextCandidatePtr = bt + 2*(idx&btMask);
+        U32*   const sortMarkPtr  = nextCandidatePtr + 1;
+
+        DEBUGLOG(8, "ZSTD_updateDUBT: insert %u", idx);
+        hashTable[h] = idx;   /* Update Hash Table */
+        *nextCandidatePtr = matchIndex;   /* update BT like a chain */
+        *sortMarkPtr = ZSTD_DUBT_UNSORTED_MARK;
+    }
+    ms->nextToUpdate = target;
+}
+
+
+/* ZSTD_insertDUBT1() :
+ *  sort one already inserted but unsorted position
+ *  assumption : curr >= btlow == (curr - btmask)
+ *  doesn't fail */
+static void
+ZSTD_insertDUBT1(ZSTD_matchState_t* ms,
+                 U32 curr, const BYTE* inputEnd,
+                 U32 nbCompares, U32 btLow,
+                 const ZSTD_dictMode_e dictMode)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32* const bt = ms->chainTable;
+    U32  const btLog  = cParams->chainLog - 1;
+    U32  const btMask = (1 << btLog) - 1;
+    size_t commonLengthSmaller=0, commonLengthLarger=0;
+    const BYTE* const base = ms->window.base;
+    const BYTE* const dictBase = ms->window.dictBase;
+    const U32 dictLimit = ms->window.dictLimit;
+    const BYTE* const ip = (curr>=dictLimit) ? base + curr : dictBase + curr;
+    const BYTE* const iend = (curr>=dictLimit) ? inputEnd : dictBase + dictLimit;
+    const BYTE* const dictEnd = dictBase + dictLimit;
+    const BYTE* const prefixStart = base + dictLimit;
+    const BYTE* match;
+    U32* smallerPtr = bt + 2*(curr&btMask);
+    U32* largerPtr  = smallerPtr + 1;
+    U32 matchIndex = *smallerPtr;   /* this candidate is unsorted : next sorted candidate is reached through *smallerPtr, while *largerPtr contains previous unsorted candidate (which is already saved and can be overwritten) */
+    U32 dummy32;   /* to be nullified at the end */
+    U32 const windowValid = ms->window.lowLimit;
+    U32 const maxDistance = 1U << cParams->windowLog;
+    U32 const windowLow = (curr - windowValid > maxDistance) ? curr - maxDistance : windowValid;
+
+
+    DEBUGLOG(8, "ZSTD_insertDUBT1(%u) (dictLimit=%u, lowLimit=%u)",
+                curr, dictLimit, windowLow);
+    assert(curr >= btLow);
+    assert(ip < iend);   /* condition for ZSTD_count */
+
+    for (; nbCompares && (matchIndex > windowLow); --nbCompares) {
+        U32* const nextPtr = bt + 2*(matchIndex & btMask);
+        size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger);   /* guaranteed minimum nb of common bytes */
+        assert(matchIndex < curr);
+        /* note : all candidates are now supposed sorted,
+         * but it's still possible to have nextPtr[1] == ZSTD_DUBT_UNSORTED_MARK
+         * when a real index has the same value as ZSTD_DUBT_UNSORTED_MARK */
+
+        if ( (dictMode != ZSTD_extDict)
+          || (matchIndex+matchLength >= dictLimit)  /* both in current segment*/
+          || (curr < dictLimit) /* both in extDict */) {
+            const BYTE* const mBase = ( (dictMode != ZSTD_extDict)
+                                     || (matchIndex+matchLength >= dictLimit)) ?
+                                        base : dictBase;
+            assert( (matchIndex+matchLength >= dictLimit)   /* might be wrong if extDict is incorrectly set to 0 */
+                 || (curr < dictLimit) );
+            match = mBase + matchIndex;
+            matchLength += ZSTD_count(ip+matchLength, match+matchLength, iend);
+        } else {
+            match = dictBase + matchIndex;
+            matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
+            if (matchIndex+matchLength >= dictLimit)
+                match = base + matchIndex;   /* preparation for next read of match[matchLength] */
+        }
+
+        DEBUGLOG(8, "ZSTD_insertDUBT1: comparing %u with %u : found %u common bytes ",
+                    curr, matchIndex, (U32)matchLength);
+
+        if (ip+matchLength == iend) {   /* equal : no way to know if inf or sup */
+            break;   /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt tree */
+        }
+
+        if (match[matchLength] < ip[matchLength]) {  /* necessarily within buffer */
+            /* match is smaller than current */
+            *smallerPtr = matchIndex;             /* update smaller idx */
+            commonLengthSmaller = matchLength;    /* all smaller will now have at least this guaranteed common length */
+            if (matchIndex <= btLow) { smallerPtr=&dummy32; break; }   /* beyond tree size, stop searching */
+            DEBUGLOG(8, "ZSTD_insertDUBT1: %u (>btLow=%u) is smaller : next => %u",
+                        matchIndex, btLow, nextPtr[1]);
+            smallerPtr = nextPtr+1;               /* new "candidate" => larger than match, which was smaller than target */
+            matchIndex = nextPtr[1];              /* new matchIndex, larger than previous and closer to current */
+        } else {
+            /* match is larger than current */
+            *largerPtr = matchIndex;
+            commonLengthLarger = matchLength;
+            if (matchIndex <= btLow) { largerPtr=&dummy32; break; }   /* beyond tree size, stop searching */
+            DEBUGLOG(8, "ZSTD_insertDUBT1: %u (>btLow=%u) is larger => %u",
+                        matchIndex, btLow, nextPtr[0]);
+            largerPtr = nextPtr;
+            matchIndex = nextPtr[0];
+    }   }
+
+    *smallerPtr = *largerPtr = 0;
+}
+
+
+static size_t
+ZSTD_DUBT_findBetterDictMatch (
+        ZSTD_matchState_t* ms,
+        const BYTE* const ip, const BYTE* const iend,
+        size_t* offsetPtr,
+        size_t bestLength,
+        U32 nbCompares,
+        U32 const mls,
+        const ZSTD_dictMode_e dictMode)
+{
+    const ZSTD_matchState_t * const dms = ms->dictMatchState;
+    const ZSTD_compressionParameters* const dmsCParams = &dms->cParams;
+    const U32 * const dictHashTable = dms->hashTable;
+    U32         const hashLog = dmsCParams->hashLog;
+    size_t      const h  = ZSTD_hashPtr(ip, hashLog, mls);
+    U32               dictMatchIndex = dictHashTable[h];
+
+    const BYTE* const base = ms->window.base;
+    const BYTE* const prefixStart = base + ms->window.dictLimit;
+    U32         const curr = (U32)(ip-base);
+    const BYTE* const dictBase = dms->window.base;
+    const BYTE* const dictEnd = dms->window.nextSrc;
+    U32         const dictHighLimit = (U32)(dms->window.nextSrc - dms->window.base);
+    U32         const dictLowLimit = dms->window.lowLimit;
+    U32         const dictIndexDelta = ms->window.lowLimit - dictHighLimit;
+
+    U32*        const dictBt = dms->chainTable;
+    U32         const btLog  = dmsCParams->chainLog - 1;
+    U32         const btMask = (1 << btLog) - 1;
+    U32         const btLow = (btMask >= dictHighLimit - dictLowLimit) ? dictLowLimit : dictHighLimit - btMask;
+
+    size_t commonLengthSmaller=0, commonLengthLarger=0;
+
+    (void)dictMode;
+    assert(dictMode == ZSTD_dictMatchState);
+
+    for (; nbCompares && (dictMatchIndex > dictLowLimit); --nbCompares) {
+        U32* const nextPtr = dictBt + 2*(dictMatchIndex & btMask);
+        size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger);   /* guaranteed minimum nb of common bytes */
+        const BYTE* match = dictBase + dictMatchIndex;
+        matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
+        if (dictMatchIndex+matchLength >= dictHighLimit)
+            match = base + dictMatchIndex + dictIndexDelta;   /* to prepare for next usage of match[matchLength] */
+
+        if (matchLength > bestLength) {
+            U32 matchIndex = dictMatchIndex + dictIndexDelta;
+            if ( (4*(int)(matchLength-bestLength)) > (int)(ZSTD_highbit32(curr-matchIndex+1) - ZSTD_highbit32((U32)offsetPtr[0]+1)) ) {
+                DEBUGLOG(9, "ZSTD_DUBT_findBetterDictMatch(%u) : found better match length %u -> %u and offsetCode %u -> %u (dictMatchIndex %u, matchIndex %u)",
+                    curr, (U32)bestLength, (U32)matchLength, (U32)*offsetPtr, ZSTD_REP_MOVE + curr - matchIndex, dictMatchIndex, matchIndex);
+                bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + curr - matchIndex;
+            }
+            if (ip+matchLength == iend) {   /* reached end of input : ip[matchLength] is not valid, no way to know if it's larger or smaller than match */
+                break;   /* drop, to guarantee consistency (miss a little bit of compression) */
+            }
+        }
+
+        if (match[matchLength] < ip[matchLength]) {
+            if (dictMatchIndex <= btLow) { break; }   /* beyond tree size, stop the search */
+            commonLengthSmaller = matchLength;    /* all smaller will now have at least this guaranteed common length */
+            dictMatchIndex = nextPtr[1];              /* new matchIndex larger than previous (closer to current) */
+        } else {
+            /* match is larger than current */
+            if (dictMatchIndex <= btLow) { break; }   /* beyond tree size, stop the search */
+            commonLengthLarger = matchLength;
+            dictMatchIndex = nextPtr[0];
+        }
+    }
+
+    if (bestLength >= MINMATCH) {
+        U32 const mIndex = curr - ((U32)*offsetPtr - ZSTD_REP_MOVE); (void)mIndex;
+        DEBUGLOG(8, "ZSTD_DUBT_findBetterDictMatch(%u) : found match of length %u and offsetCode %u (pos %u)",
+                    curr, (U32)bestLength, (U32)*offsetPtr, mIndex);
+    }
+    return bestLength;
+
+}
+
+
+static size_t
+ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms,
+                        const BYTE* const ip, const BYTE* const iend,
+                        size_t* offsetPtr,
+                        U32 const mls,
+                        const ZSTD_dictMode_e dictMode)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32*   const hashTable = ms->hashTable;
+    U32    const hashLog = cParams->hashLog;
+    size_t const h  = ZSTD_hashPtr(ip, hashLog, mls);
+    U32          matchIndex  = hashTable[h];
+
+    const BYTE* const base = ms->window.base;
+    U32    const curr = (U32)(ip-base);
+    U32    const windowLow = ZSTD_getLowestMatchIndex(ms, curr, cParams->windowLog);
+
+    U32*   const bt = ms->chainTable;
+    U32    const btLog  = cParams->chainLog - 1;
+    U32    const btMask = (1 << btLog) - 1;
+    U32    const btLow = (btMask >= curr) ? 0 : curr - btMask;
+    U32    const unsortLimit = MAX(btLow, windowLow);
+
+    U32*         nextCandidate = bt + 2*(matchIndex&btMask);
+    U32*         unsortedMark = bt + 2*(matchIndex&btMask) + 1;
+    U32          nbCompares = 1U << cParams->searchLog;
+    U32          nbCandidates = nbCompares;
+    U32          previousCandidate = 0;
+
+    DEBUGLOG(7, "ZSTD_DUBT_findBestMatch (%u) ", curr);
+    assert(ip <= iend-8);   /* required for h calculation */
+    assert(dictMode != ZSTD_dedicatedDictSearch);
+
+    /* reach end of unsorted candidates list */
+    while ( (matchIndex > unsortLimit)
+         && (*unsortedMark == ZSTD_DUBT_UNSORTED_MARK)
+         && (nbCandidates > 1) ) {
+        DEBUGLOG(8, "ZSTD_DUBT_findBestMatch: candidate %u is unsorted",
+                    matchIndex);
+        *unsortedMark = previousCandidate;  /* the unsortedMark becomes a reversed chain, to move up back to original position */
+        previousCandidate = matchIndex;
+        matchIndex = *nextCandidate;
+        nextCandidate = bt + 2*(matchIndex&btMask);
+        unsortedMark = bt + 2*(matchIndex&btMask) + 1;
+        nbCandidates --;
+    }
+
+    /* nullify last candidate if it's still unsorted
+     * simplification, detrimental to compression ratio, beneficial for speed */
+    if ( (matchIndex > unsortLimit)
+      && (*unsortedMark==ZSTD_DUBT_UNSORTED_MARK) ) {
+        DEBUGLOG(7, "ZSTD_DUBT_findBestMatch: nullify last unsorted candidate %u",
+                    matchIndex);
+        *nextCandidate = *unsortedMark = 0;
+    }
+
+    /* batch sort stacked candidates */
+    matchIndex = previousCandidate;
+    while (matchIndex) {  /* will end on matchIndex == 0 */
+        U32* const nextCandidateIdxPtr = bt + 2*(matchIndex&btMask) + 1;
+        U32 const nextCandidateIdx = *nextCandidateIdxPtr;
+        ZSTD_insertDUBT1(ms, matchIndex, iend,
+                         nbCandidates, unsortLimit, dictMode);
+        matchIndex = nextCandidateIdx;
+        nbCandidates++;
+    }
+
+    /* find longest match */
+    {   size_t commonLengthSmaller = 0, commonLengthLarger = 0;
+        const BYTE* const dictBase = ms->window.dictBase;
+        const U32 dictLimit = ms->window.dictLimit;
+        const BYTE* const dictEnd = dictBase + dictLimit;
+        const BYTE* const prefixStart = base + dictLimit;
+        U32* smallerPtr = bt + 2*(curr&btMask);
+        U32* largerPtr  = bt + 2*(curr&btMask) + 1;
+        U32 matchEndIdx = curr + 8 + 1;
+        U32 dummy32;   /* to be nullified at the end */
+        size_t bestLength = 0;
+
+        matchIndex  = hashTable[h];
+        hashTable[h] = curr;   /* Update Hash Table */
+
+        for (; nbCompares && (matchIndex > windowLow); --nbCompares) {
+            U32* const nextPtr = bt + 2*(matchIndex & btMask);
+            size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger);   /* guaranteed minimum nb of common bytes */
+            const BYTE* match;
+
+            if ((dictMode != ZSTD_extDict) || (matchIndex+matchLength >= dictLimit)) {
+                match = base + matchIndex;
+                matchLength += ZSTD_count(ip+matchLength, match+matchLength, iend);
+            } else {
+                match = dictBase + matchIndex;
+                matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
+                if (matchIndex+matchLength >= dictLimit)
+                    match = base + matchIndex;   /* to prepare for next usage of match[matchLength] */
+            }
+
+            if (matchLength > bestLength) {
+                if (matchLength > matchEndIdx - matchIndex)
+                    matchEndIdx = matchIndex + (U32)matchLength;
+                if ( (4*(int)(matchLength-bestLength)) > (int)(ZSTD_highbit32(curr-matchIndex+1) - ZSTD_highbit32((U32)offsetPtr[0]+1)) )
+                    bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + curr - matchIndex;
+                if (ip+matchLength == iend) {   /* equal : no way to know if inf or sup */
+                    if (dictMode == ZSTD_dictMatchState) {
+                        nbCompares = 0; /* in addition to avoiding checking any
+                                         * further in this loop, make sure we
+                                         * skip checking in the dictionary. */
+                    }
+                    break;   /* drop, to guarantee consistency (miss a little bit of compression) */
+                }
+            }
+
+            if (match[matchLength] < ip[matchLength]) {
+                /* match is smaller than current */
+                *smallerPtr = matchIndex;             /* update smaller idx */
+                commonLengthSmaller = matchLength;    /* all smaller will now have at least this guaranteed common length */
+                if (matchIndex <= btLow) { smallerPtr=&dummy32; break; }   /* beyond tree size, stop the search */
+                smallerPtr = nextPtr+1;               /* new "smaller" => larger of match */
+                matchIndex = nextPtr[1];              /* new matchIndex larger than previous (closer to current) */
+            } else {
+                /* match is larger than current */
+                *largerPtr = matchIndex;
+                commonLengthLarger = matchLength;
+                if (matchIndex <= btLow) { largerPtr=&dummy32; break; }   /* beyond tree size, stop the search */
+                largerPtr = nextPtr;
+                matchIndex = nextPtr[0];
+        }   }
+
+        *smallerPtr = *largerPtr = 0;
+
+        assert(nbCompares <= (1U << ZSTD_SEARCHLOG_MAX)); /* Check we haven't underflowed. */
+        if (dictMode == ZSTD_dictMatchState && nbCompares) {
+            bestLength = ZSTD_DUBT_findBetterDictMatch(
+                    ms, ip, iend,
+                    offsetPtr, bestLength, nbCompares,
+                    mls, dictMode);
+        }
+
+        assert(matchEndIdx > curr+8); /* ensure nextToUpdate is increased */
+        ms->nextToUpdate = matchEndIdx - 8;   /* skip repetitive patterns */
+        if (bestLength >= MINMATCH) {
+            U32 const mIndex = curr - ((U32)*offsetPtr - ZSTD_REP_MOVE); (void)mIndex;
+            DEBUGLOG(8, "ZSTD_DUBT_findBestMatch(%u) : found match of length %u and offsetCode %u (pos %u)",
+                        curr, (U32)bestLength, (U32)*offsetPtr, mIndex);
+        }
+        return bestLength;
+    }
+}
+
+
+/* ZSTD_BtFindBestMatch() : Tree updater, providing best match */
+FORCE_INLINE_TEMPLATE size_t
+ZSTD_BtFindBestMatch( ZSTD_matchState_t* ms,
+                const BYTE* const ip, const BYTE* const iLimit,
+                      size_t* offsetPtr,
+                const U32 mls /* template */,
+                const ZSTD_dictMode_e dictMode)
+{
+    DEBUGLOG(7, "ZSTD_BtFindBestMatch");
+    if (ip < ms->window.base + ms->nextToUpdate) return 0;   /* skipped area */
+    ZSTD_updateDUBT(ms, ip, iLimit, mls);
+    return ZSTD_DUBT_findBestMatch(ms, ip, iLimit, offsetPtr, mls, dictMode);
+}
+
+
+static size_t
+ZSTD_BtFindBestMatch_selectMLS (  ZSTD_matchState_t* ms,
+                            const BYTE* ip, const BYTE* const iLimit,
+                                  size_t* offsetPtr)
+{
+    switch(ms->cParams.minMatch)
+    {
+    default : /* includes case 3 */
+    case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_noDict);
+    case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_noDict);
+    case 7 :
+    case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_noDict);
+    }
+}
+
+
+static size_t ZSTD_BtFindBestMatch_dictMatchState_selectMLS (
+                        ZSTD_matchState_t* ms,
+                        const BYTE* ip, const BYTE* const iLimit,
+                        size_t* offsetPtr)
+{
+    switch(ms->cParams.minMatch)
+    {
+    default : /* includes case 3 */
+    case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_dictMatchState);
+    case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_dictMatchState);
+    case 7 :
+    case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_dictMatchState);
+    }
+}
+
+
+static size_t ZSTD_BtFindBestMatch_extDict_selectMLS (
+                        ZSTD_matchState_t* ms,
+                        const BYTE* ip, const BYTE* const iLimit,
+                        size_t* offsetPtr)
+{
+    switch(ms->cParams.minMatch)
+    {
+    default : /* includes case 3 */
+    case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_extDict);
+    case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_extDict);
+    case 7 :
+    case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_extDict);
+    }
+}
+
+
+
+/* *********************************
+*  Hash Chain
+***********************************/
+#define NEXT_IN_CHAIN(d, mask)   chainTable[(d) & (mask)]
+
+/* Update chains up to ip (excluded)
+   Assumption : always within prefix (i.e. not within extDict) */
+FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal(
+                        ZSTD_matchState_t* ms,
+                        const ZSTD_compressionParameters* const cParams,
+                        const BYTE* ip, U32 const mls)
+{
+    U32* const hashTable  = ms->hashTable;
+    const U32 hashLog = cParams->hashLog;
+    U32* const chainTable = ms->chainTable;
+    const U32 chainMask = (1 << cParams->chainLog) - 1;
+    const BYTE* const base = ms->window.base;
+    const U32 target = (U32)(ip - base);
+    U32 idx = ms->nextToUpdate;
+
+    while(idx < target) { /* catch up */
+        size_t const h = ZSTD_hashPtr(base+idx, hashLog, mls);
+        NEXT_IN_CHAIN(idx, chainMask) = hashTable[h];
+        hashTable[h] = idx;
+        idx++;
+    }
+
+    ms->nextToUpdate = target;
+    return hashTable[ZSTD_hashPtr(ip, hashLog, mls)];
+}
+
+U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip) {
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch);
+}
+
+void ZSTD_dedicatedDictSearch_lazy_loadDictionary(ZSTD_matchState_t* ms, const BYTE* const ip)
+{
+    const BYTE* const base = ms->window.base;
+    U32 const target = (U32)(ip - base);
+    U32* const hashTable = ms->hashTable;
+    U32* const chainTable = ms->chainTable;
+    U32 const chainSize = 1 << ms->cParams.chainLog;
+    U32 idx = ms->nextToUpdate;
+    U32 const minChain = chainSize < target ? target - chainSize : idx;
+    U32 const bucketSize = 1 << ZSTD_LAZY_DDSS_BUCKET_LOG;
+    U32 const cacheSize = bucketSize - 1;
+    U32 const chainAttempts = (1 << ms->cParams.searchLog) - cacheSize;
+    U32 const chainLimit = chainAttempts > 255 ? 255 : chainAttempts;
+
+    /* We know the hashtable is oversized by a factor of `bucketSize`.
+     * We are going to temporarily pretend `bucketSize == 1`, keeping only a
+     * single entry. We will use the rest of the space to construct a temporary
+     * chaintable.
+     */
+    U32 const hashLog = ms->cParams.hashLog - ZSTD_LAZY_DDSS_BUCKET_LOG;
+    U32* const tmpHashTable = hashTable;
+    U32* const tmpChainTable = hashTable + ((size_t)1 << hashLog);
+    U32 const tmpChainSize = ((1 << ZSTD_LAZY_DDSS_BUCKET_LOG) - 1) << hashLog;
+    U32 const tmpMinChain = tmpChainSize < target ? target - tmpChainSize : idx;
+
+    U32 hashIdx;
+
+    assert(ms->cParams.chainLog <= 24);
+    assert(ms->cParams.hashLog >= ms->cParams.chainLog);
+    assert(idx != 0);
+    assert(tmpMinChain <= minChain);
+
+    /* fill conventional hash table and conventional chain table */
+    for ( ; idx < target; idx++) {
+        U32 const h = (U32)ZSTD_hashPtr(base + idx, hashLog, ms->cParams.minMatch);
+        if (idx >= tmpMinChain) {
+            tmpChainTable[idx - tmpMinChain] = hashTable[h];
+        }
+        tmpHashTable[h] = idx;
+    }
+
+    /* sort chains into ddss chain table */
+    {
+        U32 chainPos = 0;
+        for (hashIdx = 0; hashIdx < (1U << hashLog); hashIdx++) {
+            U32 count;
+            U32 countBeyondMinChain = 0;
+            U32 i = tmpHashTable[hashIdx];
+            for (count = 0; i >= tmpMinChain && count < cacheSize; count++) {
+                /* skip through the chain to the first position that won't be
+                 * in the hash cache bucket */
+                if (i < minChain) {
+                    countBeyondMinChain++;
+                }
+                i = tmpChainTable[i - tmpMinChain];
+            }
+            if (count == cacheSize) {
+                for (count = 0; count < chainLimit;) {
+                    if (i < minChain) {
+                        if (!i || countBeyondMinChain++ > cacheSize) {
+                            /* only allow pulling `cacheSize` number of entries
+                             * into the cache or chainTable beyond `minChain`,
+                             * to replace the entries pulled out of the
+                             * chainTable into the cache. This lets us reach
+                             * back further without increasing the total number
+                             * of entries in the chainTable, guaranteeing the
+                             * DDSS chain table will fit into the space
+                             * allocated for the regular one. */
+                            break;
+                        }
+                    }
+                    chainTable[chainPos++] = i;
+                    count++;
+                    if (i < tmpMinChain) {
+                        break;
+                    }
+                    i = tmpChainTable[i - tmpMinChain];
+                }
+            } else {
+                count = 0;
+            }
+            if (count) {
+                tmpHashTable[hashIdx] = ((chainPos - count) << 8) + count;
+            } else {
+                tmpHashTable[hashIdx] = 0;
+            }
+        }
+        assert(chainPos <= chainSize); /* I believe this is guaranteed... */
+    }
+
+    /* move chain pointers into the last entry of each hash bucket */
+    for (hashIdx = (1 << hashLog); hashIdx; ) {
+        U32 const bucketIdx = --hashIdx << ZSTD_LAZY_DDSS_BUCKET_LOG;
+        U32 const chainPackedPointer = tmpHashTable[hashIdx];
+        U32 i;
+        for (i = 0; i < cacheSize; i++) {
+            hashTable[bucketIdx + i] = 0;
+        }
+        hashTable[bucketIdx + bucketSize - 1] = chainPackedPointer;
+    }
+
+    /* fill the buckets of the hash table */
+    for (idx = ms->nextToUpdate; idx < target; idx++) {
+        U32 const h = (U32)ZSTD_hashPtr(base + idx, hashLog, ms->cParams.minMatch)
+                   << ZSTD_LAZY_DDSS_BUCKET_LOG;
+        U32 i;
+        /* Shift hash cache down 1. */
+        for (i = cacheSize - 1; i; i--)
+            hashTable[h + i] = hashTable[h + i - 1];
+        hashTable[h] = idx;
+    }
+
+    ms->nextToUpdate = target;
+}
+
+
+/* inlining is important to hardwire a hot branch (template emulation) */
+FORCE_INLINE_TEMPLATE
+size_t ZSTD_HcFindBestMatch_generic (
+                        ZSTD_matchState_t* ms,
+                        const BYTE* const ip, const BYTE* const iLimit,
+                        size_t* offsetPtr,
+                        const U32 mls, const ZSTD_dictMode_e dictMode)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32* const chainTable = ms->chainTable;
+    const U32 chainSize = (1 << cParams->chainLog);
+    const U32 chainMask = chainSize-1;
+    const BYTE* const base = ms->window.base;
+    const BYTE* const dictBase = ms->window.dictBase;
+    const U32 dictLimit = ms->window.dictLimit;
+    const BYTE* const prefixStart = base + dictLimit;
+    const BYTE* const dictEnd = dictBase + dictLimit;
+    const U32 curr = (U32)(ip-base);
+    const U32 maxDistance = 1U << cParams->windowLog;
+    const U32 lowestValid = ms->window.lowLimit;
+    const U32 withinMaxDistance = (curr - lowestValid > maxDistance) ? curr - maxDistance : lowestValid;
+    const U32 isDictionary = (ms->loadedDictEnd != 0);
+    const U32 lowLimit = isDictionary ? lowestValid : withinMaxDistance;
+    const U32 minChain = curr > chainSize ? curr - chainSize : 0;
+    U32 nbAttempts = 1U << cParams->searchLog;
+    size_t ml=4-1;
+
+    const ZSTD_matchState_t* const dms = ms->dictMatchState;
+    const U32 ddsHashLog = dictMode == ZSTD_dedicatedDictSearch
+                         ? dms->cParams.hashLog - ZSTD_LAZY_DDSS_BUCKET_LOG : 0;
+    const size_t ddsIdx = dictMode == ZSTD_dedicatedDictSearch
+                        ? ZSTD_hashPtr(ip, ddsHashLog, mls) << ZSTD_LAZY_DDSS_BUCKET_LOG : 0;
+
+    U32 matchIndex;
+
+    if (dictMode == ZSTD_dedicatedDictSearch) {
+        const U32* entry = &dms->hashTable[ddsIdx];
+        PREFETCH_L1(entry);
+    }
+
+    /* HC4 match finder */
+    matchIndex = ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, mls);
+
+    for ( ; (matchIndex>=lowLimit) & (nbAttempts>0) ; nbAttempts--) {
+        size_t currentMl=0;
+        if ((dictMode != ZSTD_extDict) || matchIndex >= dictLimit) {
+            const BYTE* const match = base + matchIndex;
+            assert(matchIndex >= dictLimit);   /* ensures this is true if dictMode != ZSTD_extDict */
+            if (match[ml] == ip[ml])   /* potentially better */
+                currentMl = ZSTD_count(ip, match, iLimit);
+        } else {
+            const BYTE* const match = dictBase + matchIndex;
+            assert(match+4 <= dictEnd);
+            if (MEM_read32(match) == MEM_read32(ip))   /* assumption : matchIndex <= dictLimit-4 (by table construction) */
+                currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dictEnd, prefixStart) + 4;
+        }
+
+        /* save best solution */
+        if (currentMl > ml) {
+            ml = currentMl;
+            *offsetPtr = curr - matchIndex + ZSTD_REP_MOVE;
+            if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
+        }
+
+        if (matchIndex <= minChain) break;
+        matchIndex = NEXT_IN_CHAIN(matchIndex, chainMask);
+    }
+
+    assert(nbAttempts <= (1U << ZSTD_SEARCHLOG_MAX)); /* Check we haven't underflowed. */
+    if (dictMode == ZSTD_dedicatedDictSearch) {
+        const U32 ddsLowestIndex  = dms->window.dictLimit;
+        const BYTE* const ddsBase = dms->window.base;
+        const BYTE* const ddsEnd  = dms->window.nextSrc;
+        const U32 ddsSize         = (U32)(ddsEnd - ddsBase);
+        const U32 ddsIndexDelta   = dictLimit - ddsSize;
+        const U32 bucketSize      = (1 << ZSTD_LAZY_DDSS_BUCKET_LOG);
+        const U32 bucketLimit     = nbAttempts < bucketSize - 1 ? nbAttempts : bucketSize - 1;
+        U32 ddsAttempt;
+
+        for (ddsAttempt = 0; ddsAttempt < bucketSize - 1; ddsAttempt++) {
+            PREFETCH_L1(ddsBase + dms->hashTable[ddsIdx + ddsAttempt]);
+        }
+
+        {
+            U32 const chainPackedPointer = dms->hashTable[ddsIdx + bucketSize - 1];
+            U32 const chainIndex = chainPackedPointer >> 8;
+
+            PREFETCH_L1(&dms->chainTable[chainIndex]);
+        }
+
+        for (ddsAttempt = 0; ddsAttempt < bucketLimit; ddsAttempt++) {
+            size_t currentMl=0;
+            const BYTE* match;
+            matchIndex = dms->hashTable[ddsIdx + ddsAttempt];
+            match = ddsBase + matchIndex;
+
+            if (!matchIndex) {
+                return ml;
+            }
+
+            /* guaranteed by table construction */
+            (void)ddsLowestIndex;
+            assert(matchIndex >= ddsLowestIndex);
+            assert(match+4 <= ddsEnd);
+            if (MEM_read32(match) == MEM_read32(ip)) {
+                /* assumption : matchIndex <= dictLimit-4 (by table construction) */
+                currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, ddsEnd, prefixStart) + 4;
+            }
+
+            /* save best solution */
+            if (currentMl > ml) {
+                ml = currentMl;
+                *offsetPtr = curr - (matchIndex + ddsIndexDelta) + ZSTD_REP_MOVE;
+                if (ip+currentMl == iLimit) {
+                    /* best possible, avoids read overflow on next attempt */
+                    return ml;
+                }
+            }
+        }
+
+        {
+            U32 const chainPackedPointer = dms->hashTable[ddsIdx + bucketSize - 1];
+            U32 chainIndex = chainPackedPointer >> 8;
+            U32 const chainLength = chainPackedPointer & 0xFF;
+            U32 const chainAttempts = nbAttempts - ddsAttempt;
+            U32 const chainLimit = chainAttempts > chainLength ? chainLength : chainAttempts;
+            U32 chainAttempt;
+
+            for (chainAttempt = 0 ; chainAttempt < chainLimit; chainAttempt++) {
+                PREFETCH_L1(ddsBase + dms->chainTable[chainIndex + chainAttempt]);
+            }
+
+            for (chainAttempt = 0 ; chainAttempt < chainLimit; chainAttempt++, chainIndex++) {
+                size_t currentMl=0;
+                const BYTE* match;
+                matchIndex = dms->chainTable[chainIndex];
+                match = ddsBase + matchIndex;
+
+                /* guaranteed by table construction */
+                assert(matchIndex >= ddsLowestIndex);
+                assert(match+4 <= ddsEnd);
+                if (MEM_read32(match) == MEM_read32(ip)) {
+                    /* assumption : matchIndex <= dictLimit-4 (by table construction) */
+                    currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, ddsEnd, prefixStart) + 4;
+                }
+
+                /* save best solution */
+                if (currentMl > ml) {
+                    ml = currentMl;
+                    *offsetPtr = curr - (matchIndex + ddsIndexDelta) + ZSTD_REP_MOVE;
+                    if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
+                }
+            }
+        }
+    } else if (dictMode == ZSTD_dictMatchState) {
+        const U32* const dmsChainTable = dms->chainTable;
+        const U32 dmsChainSize         = (1 << dms->cParams.chainLog);
+        const U32 dmsChainMask         = dmsChainSize - 1;
+        const U32 dmsLowestIndex       = dms->window.dictLimit;
+        const BYTE* const dmsBase      = dms->window.base;
+        const BYTE* const dmsEnd       = dms->window.nextSrc;
+        const U32 dmsSize              = (U32)(dmsEnd - dmsBase);
+        const U32 dmsIndexDelta        = dictLimit - dmsSize;
+        const U32 dmsMinChain = dmsSize > dmsChainSize ? dmsSize - dmsChainSize : 0;
+
+        matchIndex = dms->hashTable[ZSTD_hashPtr(ip, dms->cParams.hashLog, mls)];
+
+        for ( ; (matchIndex>=dmsLowestIndex) & (nbAttempts>0) ; nbAttempts--) {
+            size_t currentMl=0;
+            const BYTE* const match = dmsBase + matchIndex;
+            assert(match+4 <= dmsEnd);
+            if (MEM_read32(match) == MEM_read32(ip))   /* assumption : matchIndex <= dictLimit-4 (by table construction) */
+                currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dmsEnd, prefixStart) + 4;
+
+            /* save best solution */
+            if (currentMl > ml) {
+                ml = currentMl;
+                *offsetPtr = curr - (matchIndex + dmsIndexDelta) + ZSTD_REP_MOVE;
+                if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
+            }
+
+            if (matchIndex <= dmsMinChain) break;
+
+            matchIndex = dmsChainTable[matchIndex & dmsChainMask];
+        }
+    }
+
+    return ml;
+}
+
+
+FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_selectMLS (
+                        ZSTD_matchState_t* ms,
+                        const BYTE* ip, const BYTE* const iLimit,
+                        size_t* offsetPtr)
+{
+    switch(ms->cParams.minMatch)
+    {
+    default : /* includes case 3 */
+    case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_noDict);
+    case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_noDict);
+    case 7 :
+    case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_noDict);
+    }
+}
+
+
+static size_t ZSTD_HcFindBestMatch_dictMatchState_selectMLS (
+                        ZSTD_matchState_t* ms,
+                        const BYTE* ip, const BYTE* const iLimit,
+                        size_t* offsetPtr)
+{
+    switch(ms->cParams.minMatch)
+    {
+    default : /* includes case 3 */
+    case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_dictMatchState);
+    case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_dictMatchState);
+    case 7 :
+    case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_dictMatchState);
+    }
+}
+
+
+static size_t ZSTD_HcFindBestMatch_dedicatedDictSearch_selectMLS (
+                        ZSTD_matchState_t* ms,
+                        const BYTE* ip, const BYTE* const iLimit,
+                        size_t* offsetPtr)
+{
+    switch(ms->cParams.minMatch)
+    {
+    default : /* includes case 3 */
+    case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_dedicatedDictSearch);
+    case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_dedicatedDictSearch);
+    case 7 :
+    case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_dedicatedDictSearch);
+    }
+}
+
+
+FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_extDict_selectMLS (
+                        ZSTD_matchState_t* ms,
+                        const BYTE* ip, const BYTE* const iLimit,
+                        size_t* offsetPtr)
+{
+    switch(ms->cParams.minMatch)
+    {
+    default : /* includes case 3 */
+    case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_extDict);
+    case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_extDict);
+    case 7 :
+    case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_extDict);
+    }
+}
+
+
+/* *******************************
+*  Common parser - lazy strategy
+*********************************/
+typedef enum { search_hashChain, search_binaryTree } searchMethod_e;
+
+FORCE_INLINE_TEMPLATE size_t
+ZSTD_compressBlock_lazy_generic(
+                        ZSTD_matchState_t* ms, seqStore_t* seqStore,
+                        U32 rep[ZSTD_REP_NUM],
+                        const void* src, size_t srcSize,
+                        const searchMethod_e searchMethod, const U32 depth,
+                        ZSTD_dictMode_e const dictMode)
+{
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* ip = istart;
+    const BYTE* anchor = istart;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* const ilimit = iend - 8;
+    const BYTE* const base = ms->window.base;
+    const U32 prefixLowestIndex = ms->window.dictLimit;
+    const BYTE* const prefixLowest = base + prefixLowestIndex;
+
+    typedef size_t (*searchMax_f)(
+                        ZSTD_matchState_t* ms,
+                        const BYTE* ip, const BYTE* iLimit, size_t* offsetPtr);
+
+    /*
+     * This table is indexed first by the four ZSTD_dictMode_e values, and then
+     * by the two searchMethod_e values. NULLs are placed for configurations
+     * that should never occur (extDict modes go to the other implementation
+     * below and there is no DDSS for binary tree search yet).
+     */
+    const searchMax_f searchFuncs[4][2] = {
+        {
+            ZSTD_HcFindBestMatch_selectMLS,
+            ZSTD_BtFindBestMatch_selectMLS
+        },
+        {
+            NULL,
+            NULL
+        },
+        {
+            ZSTD_HcFindBestMatch_dictMatchState_selectMLS,
+            ZSTD_BtFindBestMatch_dictMatchState_selectMLS
+        },
+        {
+            ZSTD_HcFindBestMatch_dedicatedDictSearch_selectMLS,
+            NULL
+        }
+    };
+
+    searchMax_f const searchMax = searchFuncs[dictMode][searchMethod == search_binaryTree];
+    U32 offset_1 = rep[0], offset_2 = rep[1], savedOffset=0;
+
+    const int isDMS = dictMode == ZSTD_dictMatchState;
+    const int isDDS = dictMode == ZSTD_dedicatedDictSearch;
+    const int isDxS = isDMS || isDDS;
+    const ZSTD_matchState_t* const dms = ms->dictMatchState;
+    const U32 dictLowestIndex      = isDxS ? dms->window.dictLimit : 0;
+    const BYTE* const dictBase     = isDxS ? dms->window.base : NULL;
+    const BYTE* const dictLowest   = isDxS ? dictBase + dictLowestIndex : NULL;
+    const BYTE* const dictEnd      = isDxS ? dms->window.nextSrc : NULL;
+    const U32 dictIndexDelta       = isDxS ?
+                                     prefixLowestIndex - (U32)(dictEnd - dictBase) :
+                                     0;
+    const U32 dictAndPrefixLength = (U32)((ip - prefixLowest) + (dictEnd - dictLowest));
+
+    assert(searchMax != NULL);
+
+    DEBUGLOG(5, "ZSTD_compressBlock_lazy_generic (dictMode=%u)", (U32)dictMode);
+
+    /* init */
+    ip += (dictAndPrefixLength == 0);
+    if (dictMode == ZSTD_noDict) {
+        U32 const curr = (U32)(ip - base);
+        U32 const windowLow = ZSTD_getLowestPrefixIndex(ms, curr, ms->cParams.windowLog);
+        U32 const maxRep = curr - windowLow;
+        if (offset_2 > maxRep) savedOffset = offset_2, offset_2 = 0;
+        if (offset_1 > maxRep) savedOffset = offset_1, offset_1 = 0;
+    }
+    if (isDxS) {
+        /* dictMatchState repCode checks don't currently handle repCode == 0
+         * disabling. */
+        assert(offset_1 <= dictAndPrefixLength);
+        assert(offset_2 <= dictAndPrefixLength);
+    }
+
+    /* Match Loop */
+#if defined(__x86_64__)
+    /* I've measured random a 5% speed loss on levels 5 & 6 (greedy) when the
+     * code alignment is perturbed. To fix the instability align the loop on 32-bytes.
+     */
+    __asm__(".p2align 5");
+#endif
+    while (ip < ilimit) {
+        size_t matchLength=0;
+        size_t offset=0;
+        const BYTE* start=ip+1;
+
+        /* check repCode */
+        if (isDxS) {
+            const U32 repIndex = (U32)(ip - base) + 1 - offset_1;
+            const BYTE* repMatch = ((dictMode == ZSTD_dictMatchState || dictMode == ZSTD_dedicatedDictSearch)
+                                && repIndex < prefixLowestIndex) ?
+                                   dictBase + (repIndex - dictIndexDelta) :
+                                   base + repIndex;
+            if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
+                && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
+                const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
+                matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
+                if (depth==0) goto _storeSequence;
+            }
+        }
+        if ( dictMode == ZSTD_noDict
+          && ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) {
+            matchLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
+            if (depth==0) goto _storeSequence;
+        }
+
+        /* first search (depth 0) */
+        {   size_t offsetFound = 999999999;
+            size_t const ml2 = searchMax(ms, ip, iend, &offsetFound);
+            if (ml2 > matchLength)
+                matchLength = ml2, start = ip, offset=offsetFound;
+        }
+
+        if (matchLength < 4) {
+            ip += ((ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */
+            continue;
+        }
+
+        /* let's try to find a better solution */
+        if (depth>=1)
+        while (ip<ilimit) {
+            ip ++;
+            if ( (dictMode == ZSTD_noDict)
+              && (offset) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) {
+                size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4;
+                int const gain2 = (int)(mlRep * 3);
+                int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1);
+                if ((mlRep >= 4) && (gain2 > gain1))
+                    matchLength = mlRep, offset = 0, start = ip;
+            }
+            if (isDxS) {
+                const U32 repIndex = (U32)(ip - base) - offset_1;
+                const BYTE* repMatch = repIndex < prefixLowestIndex ?
+                               dictBase + (repIndex - dictIndexDelta) :
+                               base + repIndex;
+                if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
+                    && (MEM_read32(repMatch) == MEM_read32(ip)) ) {
+                    const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
+                    size_t const mlRep = ZSTD_count_2segments(ip+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
+                    int const gain2 = (int)(mlRep * 3);
+                    int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1);
+                    if ((mlRep >= 4) && (gain2 > gain1))
+                        matchLength = mlRep, offset = 0, start = ip;
+                }
+            }
+            {   size_t offset2=999999999;
+                size_t const ml2 = searchMax(ms, ip, iend, &offset2);
+                int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1));   /* raw approx */
+                int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 4);
+                if ((ml2 >= 4) && (gain2 > gain1)) {
+                    matchLength = ml2, offset = offset2, start = ip;
+                    continue;   /* search a better one */
+            }   }
+
+            /* let's find an even better one */
+            if ((depth==2) && (ip<ilimit)) {
+                ip ++;
+                if ( (dictMode == ZSTD_noDict)
+                  && (offset) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) {
+                    size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4;
+                    int const gain2 = (int)(mlRep * 4);
+                    int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1);
+                    if ((mlRep >= 4) && (gain2 > gain1))
+                        matchLength = mlRep, offset = 0, start = ip;
+                }
+                if (isDxS) {
+                    const U32 repIndex = (U32)(ip - base) - offset_1;
+                    const BYTE* repMatch = repIndex < prefixLowestIndex ?
+                                   dictBase + (repIndex - dictIndexDelta) :
+                                   base + repIndex;
+                    if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
+                        && (MEM_read32(repMatch) == MEM_read32(ip)) ) {
+                        const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
+                        size_t const mlRep = ZSTD_count_2segments(ip+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
+                        int const gain2 = (int)(mlRep * 4);
+                        int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1);
+                        if ((mlRep >= 4) && (gain2 > gain1))
+                            matchLength = mlRep, offset = 0, start = ip;
+                    }
+                }
+                {   size_t offset2=999999999;
+                    size_t const ml2 = searchMax(ms, ip, iend, &offset2);
+                    int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1));   /* raw approx */
+                    int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 7);
+                    if ((ml2 >= 4) && (gain2 > gain1)) {
+                        matchLength = ml2, offset = offset2, start = ip;
+                        continue;
+            }   }   }
+            break;  /* nothing found : store previous solution */
+        }
+
+        /* NOTE:
+         * start[-offset+ZSTD_REP_MOVE-1] is undefined behavior.
+         * (-offset+ZSTD_REP_MOVE-1) is unsigned, and is added to start, which
+         * overflows the pointer, which is undefined behavior.
+         */
+        /* catch up */
+        if (offset) {
+            if (dictMode == ZSTD_noDict) {
+                while ( ((start > anchor) & (start - (offset-ZSTD_REP_MOVE) > prefixLowest))
+                     && (start[-1] == (start-(offset-ZSTD_REP_MOVE))[-1]) )  /* only search for offset within prefix */
+                    { start--; matchLength++; }
+            }
+            if (isDxS) {
+                U32 const matchIndex = (U32)((start-base) - (offset - ZSTD_REP_MOVE));
+                const BYTE* match = (matchIndex < prefixLowestIndex) ? dictBase + matchIndex - dictIndexDelta : base + matchIndex;
+                const BYTE* const mStart = (matchIndex < prefixLowestIndex) ? dictLowest : prefixLowest;
+                while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; }  /* catch up */
+            }
+            offset_2 = offset_1; offset_1 = (U32)(offset - ZSTD_REP_MOVE);
+        }
+        /* store sequence */
+_storeSequence:
+        {   size_t const litLength = start - anchor;
+            ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH);
+            anchor = ip = start + matchLength;
+        }
+
+        /* check immediate repcode */
+        if (isDxS) {
+            while (ip <= ilimit) {
+                U32 const current2 = (U32)(ip-base);
+                U32 const repIndex = current2 - offset_2;
+                const BYTE* repMatch = repIndex < prefixLowestIndex ?
+                        dictBase - dictIndexDelta + repIndex :
+                        base + repIndex;
+                if ( ((U32)((prefixLowestIndex-1) - (U32)repIndex) >= 3 /* intentional overflow */)
+                   && (MEM_read32(repMatch) == MEM_read32(ip)) ) {
+                    const BYTE* const repEnd2 = repIndex < prefixLowestIndex ? dictEnd : iend;
+                    matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd2, prefixLowest) + 4;
+                    offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset;   /* swap offset_2 <=> offset_1 */
+                    ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
+                    ip += matchLength;
+                    anchor = ip;
+                    continue;
+                }
+                break;
+            }
+        }
+
+        if (dictMode == ZSTD_noDict) {
+            while ( ((ip <= ilimit) & (offset_2>0))
+                 && (MEM_read32(ip) == MEM_read32(ip - offset_2)) ) {
+                /* store sequence */
+                matchLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
+                offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap repcodes */
+                ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
+                ip += matchLength;
+                anchor = ip;
+                continue;   /* faster when present ... (?) */
+    }   }   }
+
+    /* Save reps for next block */
+    rep[0] = offset_1 ? offset_1 : savedOffset;
+    rep[1] = offset_2 ? offset_2 : savedOffset;
+
+    /* Return the last literals size */
+    return (size_t)(iend - anchor);
+}
+
+
+size_t ZSTD_compressBlock_btlazy2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_noDict);
+}
+
+size_t ZSTD_compressBlock_lazy2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_noDict);
+}
+
+size_t ZSTD_compressBlock_lazy(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_noDict);
+}
+
+size_t ZSTD_compressBlock_greedy(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_noDict);
+}
+
+size_t ZSTD_compressBlock_btlazy2_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_dictMatchState);
+}
+
+size_t ZSTD_compressBlock_lazy2_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dictMatchState);
+}
+
+size_t ZSTD_compressBlock_lazy_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dictMatchState);
+}
+
+size_t ZSTD_compressBlock_greedy_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dictMatchState);
+}
+
+
+size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dedicatedDictSearch);
+}
+
+size_t ZSTD_compressBlock_lazy_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dedicatedDictSearch);
+}
+
+size_t ZSTD_compressBlock_greedy_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dedicatedDictSearch);
+}
+
+
+FORCE_INLINE_TEMPLATE
+size_t ZSTD_compressBlock_lazy_extDict_generic(
+                        ZSTD_matchState_t* ms, seqStore_t* seqStore,
+                        U32 rep[ZSTD_REP_NUM],
+                        const void* src, size_t srcSize,
+                        const searchMethod_e searchMethod, const U32 depth)
+{
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* ip = istart;
+    const BYTE* anchor = istart;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* const ilimit = iend - 8;
+    const BYTE* const base = ms->window.base;
+    const U32 dictLimit = ms->window.dictLimit;
+    const BYTE* const prefixStart = base + dictLimit;
+    const BYTE* const dictBase = ms->window.dictBase;
+    const BYTE* const dictEnd  = dictBase + dictLimit;
+    const BYTE* const dictStart  = dictBase + ms->window.lowLimit;
+    const U32 windowLog = ms->cParams.windowLog;
+
+    typedef size_t (*searchMax_f)(
+                        ZSTD_matchState_t* ms,
+                        const BYTE* ip, const BYTE* iLimit, size_t* offsetPtr);
+    searchMax_f searchMax = searchMethod==search_binaryTree ? ZSTD_BtFindBestMatch_extDict_selectMLS : ZSTD_HcFindBestMatch_extDict_selectMLS;
+
+    U32 offset_1 = rep[0], offset_2 = rep[1];
+
+    DEBUGLOG(5, "ZSTD_compressBlock_lazy_extDict_generic");
+
+    /* init */
+    ip += (ip == prefixStart);
+
+    /* Match Loop */
+#if defined(__x86_64__)
+    /* I've measured random a 5% speed loss on levels 5 & 6 (greedy) when the
+     * code alignment is perturbed. To fix the instability align the loop on 32-bytes.
+     */
+    __asm__(".p2align 5");
+#endif
+    while (ip < ilimit) {
+        size_t matchLength=0;
+        size_t offset=0;
+        const BYTE* start=ip+1;
+        U32 curr = (U32)(ip-base);
+
+        /* check repCode */
+        {   const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr+1, windowLog);
+            const U32 repIndex = (U32)(curr+1 - offset_1);
+            const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
+            const BYTE* const repMatch = repBase + repIndex;
+            if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow))   /* intentional overflow */
+            if (MEM_read32(ip+1) == MEM_read32(repMatch)) {
+                /* repcode detected we should take it */
+                const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
+                matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repEnd, prefixStart) + 4;
+                if (depth==0) goto _storeSequence;
+        }   }
+
+        /* first search (depth 0) */
+        {   size_t offsetFound = 999999999;
+            size_t const ml2 = searchMax(ms, ip, iend, &offsetFound);
+            if (ml2 > matchLength)
+                matchLength = ml2, start = ip, offset=offsetFound;
+        }
+
+        if (matchLength < 4) {
+            ip += ((ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */
+            continue;
+        }
+
+        /* let's try to find a better solution */
+        if (depth>=1)
+        while (ip<ilimit) {
+            ip ++;
+            curr++;
+            /* check repCode */
+            if (offset) {
+                const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr, windowLog);
+                const U32 repIndex = (U32)(curr - offset_1);
+                const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
+                const BYTE* const repMatch = repBase + repIndex;
+                if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow))  /* intentional overflow */
+                if (MEM_read32(ip) == MEM_read32(repMatch)) {
+                    /* repcode detected */
+                    const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
+                    size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
+                    int const gain2 = (int)(repLength * 3);
+                    int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1);
+                    if ((repLength >= 4) && (gain2 > gain1))
+                        matchLength = repLength, offset = 0, start = ip;
+            }   }
+
+            /* search match, depth 1 */
+            {   size_t offset2=999999999;
+                size_t const ml2 = searchMax(ms, ip, iend, &offset2);
+                int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1));   /* raw approx */
+                int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 4);
+                if ((ml2 >= 4) && (gain2 > gain1)) {
+                    matchLength = ml2, offset = offset2, start = ip;
+                    continue;   /* search a better one */
+            }   }
+
+            /* let's find an even better one */
+            if ((depth==2) && (ip<ilimit)) {
+                ip ++;
+                curr++;
+                /* check repCode */
+                if (offset) {
+                    const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr, windowLog);
+                    const U32 repIndex = (U32)(curr - offset_1);
+                    const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
+                    const BYTE* const repMatch = repBase + repIndex;
+                    if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow))  /* intentional overflow */
+                    if (MEM_read32(ip) == MEM_read32(repMatch)) {
+                        /* repcode detected */
+                        const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
+                        size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
+                        int const gain2 = (int)(repLength * 4);
+                        int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1);
+                        if ((repLength >= 4) && (gain2 > gain1))
+                            matchLength = repLength, offset = 0, start = ip;
+                }   }
+
+                /* search match, depth 2 */
+                {   size_t offset2=999999999;
+                    size_t const ml2 = searchMax(ms, ip, iend, &offset2);
+                    int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1));   /* raw approx */
+                    int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 7);
+                    if ((ml2 >= 4) && (gain2 > gain1)) {
+                        matchLength = ml2, offset = offset2, start = ip;
+                        continue;
+            }   }   }
+            break;  /* nothing found : store previous solution */
+        }
+
+        /* catch up */
+        if (offset) {
+            U32 const matchIndex = (U32)((start-base) - (offset - ZSTD_REP_MOVE));
+            const BYTE* match = (matchIndex < dictLimit) ? dictBase + matchIndex : base + matchIndex;
+            const BYTE* const mStart = (matchIndex < dictLimit) ? dictStart : prefixStart;
+            while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; }  /* catch up */
+            offset_2 = offset_1; offset_1 = (U32)(offset - ZSTD_REP_MOVE);
+        }
+
+        /* store sequence */
+_storeSequence:
+        {   size_t const litLength = start - anchor;
+            ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH);
+            anchor = ip = start + matchLength;
+        }
+
+        /* check immediate repcode */
+        while (ip <= ilimit) {
+            const U32 repCurrent = (U32)(ip-base);
+            const U32 windowLow = ZSTD_getLowestMatchIndex(ms, repCurrent, windowLog);
+            const U32 repIndex = repCurrent - offset_2;
+            const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
+            const BYTE* const repMatch = repBase + repIndex;
+            if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow))  /* intentional overflow */
+            if (MEM_read32(ip) == MEM_read32(repMatch)) {
+                /* repcode detected we should take it */
+                const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
+                matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
+                offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset;   /* swap offset history */
+                ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
+                ip += matchLength;
+                anchor = ip;
+                continue;   /* faster when present ... (?) */
+            }
+            break;
+    }   }
+
+    /* Save reps for next block */
+    rep[0] = offset_1;
+    rep[1] = offset_2;
+
+    /* Return the last literals size */
+    return (size_t)(iend - anchor);
+}
+
+
+size_t ZSTD_compressBlock_greedy_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0);
+}
+
+size_t ZSTD_compressBlock_lazy_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+
+{
+    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1);
+}
+
+size_t ZSTD_compressBlock_lazy2_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+
+{
+    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2);
+}
+
+size_t ZSTD_compressBlock_btlazy2_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+
+{
+    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2);
+}
diff --git a/lib/zstd/compress/zstd_lazy.h b/lib/zstd/compress/zstd_lazy.h
new file mode 100644
index 0000000..2fc5a61
--- /dev/null
+++ b/lib/zstd/compress/zstd_lazy.h
@@ -0,0 +1,81 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_LAZY_H
+#define ZSTD_LAZY_H
+
+
+#include "zstd_compress_internal.h"
+
+/*
+ * Dedicated Dictionary Search Structure bucket log. In the
+ * ZSTD_dedicatedDictSearch mode, the hashTable has
+ * 2 ** ZSTD_LAZY_DDSS_BUCKET_LOG entries in each bucket, rather than just
+ * one.
+ */
+#define ZSTD_LAZY_DDSS_BUCKET_LOG 2
+
+U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip);
+
+void ZSTD_dedicatedDictSearch_lazy_loadDictionary(ZSTD_matchState_t* ms, const BYTE* const ip);
+
+void ZSTD_preserveUnsortedMark (U32* const table, U32 const size, U32 const reducerValue);  /*! used in ZSTD_reduceIndex(). preemptively increase value of ZSTD_DUBT_UNSORTED_MARK */
+
+size_t ZSTD_compressBlock_btlazy2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_greedy(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+size_t ZSTD_compressBlock_btlazy2_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_greedy_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_greedy_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+size_t ZSTD_compressBlock_greedy_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_btlazy2_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+
+#endif /* ZSTD_LAZY_H */
diff --git a/lib/zstd/compress/zstd_ldm.c b/lib/zstd/compress/zstd_ldm.c
new file mode 100644
index 0000000..8ef7e88
--- /dev/null
+++ b/lib/zstd/compress/zstd_ldm.c
@@ -0,0 +1,686 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#include "zstd_ldm.h"
+
+#include "../common/debug.h"
+#include <linux/xxhash.h>
+#include "zstd_fast.h"          /* ZSTD_fillHashTable() */
+#include "zstd_double_fast.h"   /* ZSTD_fillDoubleHashTable() */
+#include "zstd_ldm_geartab.h"
+
+#define LDM_BUCKET_SIZE_LOG 3
+#define LDM_MIN_MATCH_LENGTH 64
+#define LDM_HASH_RLOG 7
+
+typedef struct {
+    U64 rolling;
+    U64 stopMask;
+} ldmRollingHashState_t;
+
+/* ZSTD_ldm_gear_init():
+ *
+ * Initializes the rolling hash state such that it will honor the
+ * settings in params. */
+static void ZSTD_ldm_gear_init(ldmRollingHashState_t* state, ldmParams_t const* params)
+{
+    unsigned maxBitsInMask = MIN(params->minMatchLength, 64);
+    unsigned hashRateLog = params->hashRateLog;
+
+    state->rolling = ~(U32)0;
+
+    /* The choice of the splitting criterion is subject to two conditions:
+     *   1. it has to trigger on average every 2^(hashRateLog) bytes;
+     *   2. ideally, it has to depend on a window of minMatchLength bytes.
+     *
+     * In the gear hash algorithm, bit n depends on the last n bytes;
+     * so in order to obtain a good quality splitting criterion it is
+     * preferable to use bits with high weight.
+     *
+     * To match condition 1 we use a mask with hashRateLog bits set
+     * and, because of the previous remark, we make sure these bits
+     * have the highest possible weight while still respecting
+     * condition 2.
+     */
+    if (hashRateLog > 0 && hashRateLog <= maxBitsInMask) {
+        state->stopMask = (((U64)1 << hashRateLog) - 1) << (maxBitsInMask - hashRateLog);
+    } else {
+        /* In this degenerate case we simply honor the hash rate. */
+        state->stopMask = ((U64)1 << hashRateLog) - 1;
+    }
+}
+
+/* ZSTD_ldm_gear_feed():
+ *
+ * Registers in the splits array all the split points found in the first
+ * size bytes following the data pointer. This function terminates when
+ * either all the data has been processed or LDM_BATCH_SIZE splits are
+ * present in the splits array.
+ *
+ * Precondition: The splits array must not be full.
+ * Returns: The number of bytes processed. */
+static size_t ZSTD_ldm_gear_feed(ldmRollingHashState_t* state,
+                                 BYTE const* data, size_t size,
+                                 size_t* splits, unsigned* numSplits)
+{
+    size_t n;
+    U64 hash, mask;
+
+    hash = state->rolling;
+    mask = state->stopMask;
+    n = 0;
+
+#define GEAR_ITER_ONCE() do { \
+        hash = (hash << 1) + ZSTD_ldm_gearTab[data[n] & 0xff]; \
+        n += 1; \
+        if (UNLIKELY((hash & mask) == 0)) { \
+            splits[*numSplits] = n; \
+            *numSplits += 1; \
+            if (*numSplits == LDM_BATCH_SIZE) \
+                goto done; \
+        } \
+    } while (0)
+
+    while (n + 3 < size) {
+        GEAR_ITER_ONCE();
+        GEAR_ITER_ONCE();
+        GEAR_ITER_ONCE();
+        GEAR_ITER_ONCE();
+    }
+    while (n < size) {
+        GEAR_ITER_ONCE();
+    }
+
+#undef GEAR_ITER_ONCE
+
+done:
+    state->rolling = hash;
+    return n;
+}
+
+void ZSTD_ldm_adjustParameters(ldmParams_t* params,
+                               ZSTD_compressionParameters const* cParams)
+{
+    params->windowLog = cParams->windowLog;
+    ZSTD_STATIC_ASSERT(LDM_BUCKET_SIZE_LOG <= ZSTD_LDM_BUCKETSIZELOG_MAX);
+    DEBUGLOG(4, "ZSTD_ldm_adjustParameters");
+    if (!params->bucketSizeLog) params->bucketSizeLog = LDM_BUCKET_SIZE_LOG;
+    if (!params->minMatchLength) params->minMatchLength = LDM_MIN_MATCH_LENGTH;
+    if (params->hashLog == 0) {
+        params->hashLog = MAX(ZSTD_HASHLOG_MIN, params->windowLog - LDM_HASH_RLOG);
+        assert(params->hashLog <= ZSTD_HASHLOG_MAX);
+    }
+    if (params->hashRateLog == 0) {
+        params->hashRateLog = params->windowLog < params->hashLog
+                                   ? 0
+                                   : params->windowLog - params->hashLog;
+    }
+    params->bucketSizeLog = MIN(params->bucketSizeLog, params->hashLog);
+}
+
+size_t ZSTD_ldm_getTableSize(ldmParams_t params)
+{
+    size_t const ldmHSize = ((size_t)1) << params.hashLog;
+    size_t const ldmBucketSizeLog = MIN(params.bucketSizeLog, params.hashLog);
+    size_t const ldmBucketSize = ((size_t)1) << (params.hashLog - ldmBucketSizeLog);
+    size_t const totalSize = ZSTD_cwksp_alloc_size(ldmBucketSize)
+                           + ZSTD_cwksp_alloc_size(ldmHSize * sizeof(ldmEntry_t));
+    return params.enableLdm ? totalSize : 0;
+}
+
+size_t ZSTD_ldm_getMaxNbSeq(ldmParams_t params, size_t maxChunkSize)
+{
+    return params.enableLdm ? (maxChunkSize / params.minMatchLength) : 0;
+}
+
+/* ZSTD_ldm_getBucket() :
+ *  Returns a pointer to the start of the bucket associated with hash. */
+static ldmEntry_t* ZSTD_ldm_getBucket(
+        ldmState_t* ldmState, size_t hash, ldmParams_t const ldmParams)
+{
+    return ldmState->hashTable + (hash << ldmParams.bucketSizeLog);
+}
+
+/* ZSTD_ldm_insertEntry() :
+ *  Insert the entry with corresponding hash into the hash table */
+static void ZSTD_ldm_insertEntry(ldmState_t* ldmState,
+                                 size_t const hash, const ldmEntry_t entry,
+                                 ldmParams_t const ldmParams)
+{
+    BYTE* const pOffset = ldmState->bucketOffsets + hash;
+    unsigned const offset = *pOffset;
+
+    *(ZSTD_ldm_getBucket(ldmState, hash, ldmParams) + offset) = entry;
+    *pOffset = (BYTE)((offset + 1) & ((1u << ldmParams.bucketSizeLog) - 1));
+
+}
+
+/* ZSTD_ldm_countBackwardsMatch() :
+ *  Returns the number of bytes that match backwards before pIn and pMatch.
+ *
+ *  We count only bytes where pMatch >= pBase and pIn >= pAnchor. */
+static size_t ZSTD_ldm_countBackwardsMatch(
+            const BYTE* pIn, const BYTE* pAnchor,
+            const BYTE* pMatch, const BYTE* pMatchBase)
+{
+    size_t matchLength = 0;
+    while (pIn > pAnchor && pMatch > pMatchBase && pIn[-1] == pMatch[-1]) {
+        pIn--;
+        pMatch--;
+        matchLength++;
+    }
+    return matchLength;
+}
+
+/* ZSTD_ldm_countBackwardsMatch_2segments() :
+ *  Returns the number of bytes that match backwards from pMatch,
+ *  even with the backwards match spanning 2 different segments.
+ *
+ *  On reaching `pMatchBase`, start counting from mEnd */
+static size_t ZSTD_ldm_countBackwardsMatch_2segments(
+                    const BYTE* pIn, const BYTE* pAnchor,
+                    const BYTE* pMatch, const BYTE* pMatchBase,
+                    const BYTE* pExtDictStart, const BYTE* pExtDictEnd)
+{
+    size_t matchLength = ZSTD_ldm_countBackwardsMatch(pIn, pAnchor, pMatch, pMatchBase);
+    if (pMatch - matchLength != pMatchBase || pMatchBase == pExtDictStart) {
+        /* If backwards match is entirely in the extDict or prefix, immediately return */
+        return matchLength;
+    }
+    DEBUGLOG(7, "ZSTD_ldm_countBackwardsMatch_2segments: found 2-parts backwards match (length in prefix==%zu)", matchLength);
+    matchLength += ZSTD_ldm_countBackwardsMatch(pIn - matchLength, pAnchor, pExtDictEnd, pExtDictStart);
+    DEBUGLOG(7, "final backwards match length = %zu", matchLength);
+    return matchLength;
+}
+
+/* ZSTD_ldm_fillFastTables() :
+ *
+ *  Fills the relevant tables for the ZSTD_fast and ZSTD_dfast strategies.
+ *  This is similar to ZSTD_loadDictionaryContent.
+ *
+ *  The tables for the other strategies are filled within their
+ *  block compressors. */
+static size_t ZSTD_ldm_fillFastTables(ZSTD_matchState_t* ms,
+                                      void const* end)
+{
+    const BYTE* const iend = (const BYTE*)end;
+
+    switch(ms->cParams.strategy)
+    {
+    case ZSTD_fast:
+        ZSTD_fillHashTable(ms, iend, ZSTD_dtlm_fast);
+        break;
+
+    case ZSTD_dfast:
+        ZSTD_fillDoubleHashTable(ms, iend, ZSTD_dtlm_fast);
+        break;
+
+    case ZSTD_greedy:
+    case ZSTD_lazy:
+    case ZSTD_lazy2:
+    case ZSTD_btlazy2:
+    case ZSTD_btopt:
+    case ZSTD_btultra:
+    case ZSTD_btultra2:
+        break;
+    default:
+        assert(0);  /* not possible : not a valid strategy id */
+    }
+
+    return 0;
+}
+
+void ZSTD_ldm_fillHashTable(
+            ldmState_t* ldmState, const BYTE* ip,
+            const BYTE* iend, ldmParams_t const* params)
+{
+    U32 const minMatchLength = params->minMatchLength;
+    U32 const hBits = params->hashLog - params->bucketSizeLog;
+    BYTE const* const base = ldmState->window.base;
+    BYTE const* const istart = ip;
+    ldmRollingHashState_t hashState;
+    size_t* const splits = ldmState->splitIndices;
+    unsigned numSplits;
+
+    DEBUGLOG(5, "ZSTD_ldm_fillHashTable");
+
+    ZSTD_ldm_gear_init(&hashState, params);
+    while (ip < iend) {
+        size_t hashed;
+        unsigned n;
+        
+        numSplits = 0;
+        hashed = ZSTD_ldm_gear_feed(&hashState, ip, iend - ip, splits, &numSplits);
+
+        for (n = 0; n < numSplits; n++) {
+            if (ip + splits[n] >= istart + minMatchLength) {
+                BYTE const* const split = ip + splits[n] - minMatchLength;
+                U64 const xxhash = xxh64(split, minMatchLength, 0);
+                U32 const hash = (U32)(xxhash & (((U32)1 << hBits) - 1));
+                ldmEntry_t entry;
+
+                entry.offset = (U32)(split - base);
+                entry.checksum = (U32)(xxhash >> 32);
+                ZSTD_ldm_insertEntry(ldmState, hash, entry, *params);
+            }
+        }
+
+        ip += hashed;
+    }
+}
+
+
+/* ZSTD_ldm_limitTableUpdate() :
+ *
+ *  Sets cctx->nextToUpdate to a position corresponding closer to anchor
+ *  if it is far way
+ *  (after a long match, only update tables a limited amount). */
+static void ZSTD_ldm_limitTableUpdate(ZSTD_matchState_t* ms, const BYTE* anchor)
+{
+    U32 const curr = (U32)(anchor - ms->window.base);
+    if (curr > ms->nextToUpdate + 1024) {
+        ms->nextToUpdate =
+            curr - MIN(512, curr - ms->nextToUpdate - 1024);
+    }
+}
+
+static size_t ZSTD_ldm_generateSequences_internal(
+        ldmState_t* ldmState, rawSeqStore_t* rawSeqStore,
+        ldmParams_t const* params, void const* src, size_t srcSize)
+{
+    /* LDM parameters */
+    int const extDict = ZSTD_window_hasExtDict(ldmState->window);
+    U32 const minMatchLength = params->minMatchLength;
+    U32 const entsPerBucket = 1U << params->bucketSizeLog;
+    U32 const hBits = params->hashLog - params->bucketSizeLog;
+    /* Prefix and extDict parameters */
+    U32 const dictLimit = ldmState->window.dictLimit;
+    U32 const lowestIndex = extDict ? ldmState->window.lowLimit : dictLimit;
+    BYTE const* const base = ldmState->window.base;
+    BYTE const* const dictBase = extDict ? ldmState->window.dictBase : NULL;
+    BYTE const* const dictStart = extDict ? dictBase + lowestIndex : NULL;
+    BYTE const* const dictEnd = extDict ? dictBase + dictLimit : NULL;
+    BYTE const* const lowPrefixPtr = base + dictLimit;
+    /* Input bounds */
+    BYTE const* const istart = (BYTE const*)src;
+    BYTE const* const iend = istart + srcSize;
+    BYTE const* const ilimit = iend - HASH_READ_SIZE;
+    /* Input positions */
+    BYTE const* anchor = istart;
+    BYTE const* ip = istart;
+    /* Rolling hash state */
+    ldmRollingHashState_t hashState;
+    /* Arrays for staged-processing */
+    size_t* const splits = ldmState->splitIndices;
+    ldmMatchCandidate_t* const candidates = ldmState->matchCandidates;
+    unsigned numSplits;
+
+    if (srcSize < minMatchLength)
+        return iend - anchor;
+
+    /* Initialize the rolling hash state with the first minMatchLength bytes */
+    ZSTD_ldm_gear_init(&hashState, params);
+    {
+        size_t n = 0;
+
+        while (n < minMatchLength) {
+            numSplits = 0;
+            n += ZSTD_ldm_gear_feed(&hashState, ip + n, minMatchLength - n,
+                                    splits, &numSplits);
+        }
+        ip += minMatchLength;
+    }
+
+    while (ip < ilimit) {
+        size_t hashed;
+        unsigned n;
+
+        numSplits = 0;
+        hashed = ZSTD_ldm_gear_feed(&hashState, ip, ilimit - ip,
+                                    splits, &numSplits);
+
+        for (n = 0; n < numSplits; n++) {
+            BYTE const* const split = ip + splits[n] - minMatchLength;
+            U64 const xxhash = xxh64(split, minMatchLength, 0);
+            U32 const hash = (U32)(xxhash & (((U32)1 << hBits) - 1));
+
+            candidates[n].split = split;
+            candidates[n].hash = hash;
+            candidates[n].checksum = (U32)(xxhash >> 32);
+            candidates[n].bucket = ZSTD_ldm_getBucket(ldmState, hash, *params);
+            PREFETCH_L1(candidates[n].bucket);
+        }
+
+        for (n = 0; n < numSplits; n++) {
+            size_t forwardMatchLength = 0, backwardMatchLength = 0,
+                   bestMatchLength = 0, mLength;
+            BYTE const* const split = candidates[n].split;
+            U32 const checksum = candidates[n].checksum;
+            U32 const hash = candidates[n].hash;
+            ldmEntry_t* const bucket = candidates[n].bucket;
+            ldmEntry_t const* cur;
+            ldmEntry_t const* bestEntry = NULL;
+            ldmEntry_t newEntry;
+
+            newEntry.offset = (U32)(split - base);
+            newEntry.checksum = checksum;
+
+            /* If a split point would generate a sequence overlapping with
+             * the previous one, we merely register it in the hash table and
+             * move on */
+            if (split < anchor) {
+                ZSTD_ldm_insertEntry(ldmState, hash, newEntry, *params);
+                continue;
+            }
+
+            for (cur = bucket; cur < bucket + entsPerBucket; cur++) {
+                size_t curForwardMatchLength, curBackwardMatchLength,
+                       curTotalMatchLength;
+                if (cur->checksum != checksum || cur->offset <= lowestIndex) {
+                    continue;
+                }
+                if (extDict) {
+                    BYTE const* const curMatchBase =
+                        cur->offset < dictLimit ? dictBase : base;
+                    BYTE const* const pMatch = curMatchBase + cur->offset;
+                    BYTE const* const matchEnd =
+                        cur->offset < dictLimit ? dictEnd : iend;
+                    BYTE const* const lowMatchPtr =
+                        cur->offset < dictLimit ? dictStart : lowPrefixPtr;
+                    curForwardMatchLength =
+                        ZSTD_count_2segments(split, pMatch, iend, matchEnd, lowPrefixPtr);
+                    if (curForwardMatchLength < minMatchLength) {
+                        continue;
+                    }
+                    curBackwardMatchLength = ZSTD_ldm_countBackwardsMatch_2segments(
+                            split, anchor, pMatch, lowMatchPtr, dictStart, dictEnd);
+                } else { /* !extDict */
+                    BYTE const* const pMatch = base + cur->offset;
+                    curForwardMatchLength = ZSTD_count(split, pMatch, iend);
+                    if (curForwardMatchLength < minMatchLength) {
+                        continue;
+                    }
+                    curBackwardMatchLength =
+                        ZSTD_ldm_countBackwardsMatch(split, anchor, pMatch, lowPrefixPtr);
+                }
+                curTotalMatchLength = curForwardMatchLength + curBackwardMatchLength;
+
+                if (curTotalMatchLength > bestMatchLength) {
+                    bestMatchLength = curTotalMatchLength;
+                    forwardMatchLength = curForwardMatchLength;
+                    backwardMatchLength = curBackwardMatchLength;
+                    bestEntry = cur;
+                }
+            }
+
+            /* No match found -- insert an entry into the hash table
+             * and process the next candidate match */
+            if (bestEntry == NULL) {
+                ZSTD_ldm_insertEntry(ldmState, hash, newEntry, *params);
+                continue;
+            }
+
+            /* Match found */
+            mLength = forwardMatchLength + backwardMatchLength;
+            {
+                U32 const offset = (U32)(split - base) - bestEntry->offset;
+                rawSeq* const seq = rawSeqStore->seq + rawSeqStore->size;
+
+                /* Out of sequence storage */
+                if (rawSeqStore->size == rawSeqStore->capacity)
+                    return ERROR(dstSize_tooSmall);
+                seq->litLength = (U32)(split - backwardMatchLength - anchor);
+                seq->matchLength = (U32)mLength;
+                seq->offset = offset;
+                rawSeqStore->size++;
+            }
+
+            /* Insert the current entry into the hash table --- it must be
+             * done after the previous block to avoid clobbering bestEntry */
+            ZSTD_ldm_insertEntry(ldmState, hash, newEntry, *params);
+
+            anchor = split + forwardMatchLength;
+        }
+
+        ip += hashed;
+    }
+
+    return iend - anchor;
+}
+
+/*! ZSTD_ldm_reduceTable() :
+ *  reduce table indexes by `reducerValue` */
+static void ZSTD_ldm_reduceTable(ldmEntry_t* const table, U32 const size,
+                                 U32 const reducerValue)
+{
+    U32 u;
+    for (u = 0; u < size; u++) {
+        if (table[u].offset < reducerValue) table[u].offset = 0;
+        else table[u].offset -= reducerValue;
+    }
+}
+
+size_t ZSTD_ldm_generateSequences(
+        ldmState_t* ldmState, rawSeqStore_t* sequences,
+        ldmParams_t const* params, void const* src, size_t srcSize)
+{
+    U32 const maxDist = 1U << params->windowLog;
+    BYTE const* const istart = (BYTE const*)src;
+    BYTE const* const iend = istart + srcSize;
+    size_t const kMaxChunkSize = 1 << 20;
+    size_t const nbChunks = (srcSize / kMaxChunkSize) + ((srcSize % kMaxChunkSize) != 0);
+    size_t chunk;
+    size_t leftoverSize = 0;
+
+    assert(ZSTD_CHUNKSIZE_MAX >= kMaxChunkSize);
+    /* Check that ZSTD_window_update() has been called for this chunk prior
+     * to passing it to this function.
+     */
+    assert(ldmState->window.nextSrc >= (BYTE const*)src + srcSize);
+    /* The input could be very large (in zstdmt), so it must be broken up into
+     * chunks to enforce the maximum distance and handle overflow correction.
+     */
+    assert(sequences->pos <= sequences->size);
+    assert(sequences->size <= sequences->capacity);
+    for (chunk = 0; chunk < nbChunks && sequences->size < sequences->capacity; ++chunk) {
+        BYTE const* const chunkStart = istart + chunk * kMaxChunkSize;
+        size_t const remaining = (size_t)(iend - chunkStart);
+        BYTE const *const chunkEnd =
+            (remaining < kMaxChunkSize) ? iend : chunkStart + kMaxChunkSize;
+        size_t const chunkSize = chunkEnd - chunkStart;
+        size_t newLeftoverSize;
+        size_t const prevSize = sequences->size;
+
+        assert(chunkStart < iend);
+        /* 1. Perform overflow correction if necessary. */
+        if (ZSTD_window_needOverflowCorrection(ldmState->window, chunkEnd)) {
+            U32 const ldmHSize = 1U << params->hashLog;
+            U32 const correction = ZSTD_window_correctOverflow(
+                &ldmState->window, /* cycleLog */ 0, maxDist, chunkStart);
+            ZSTD_ldm_reduceTable(ldmState->hashTable, ldmHSize, correction);
+            /* invalidate dictionaries on overflow correction */
+            ldmState->loadedDictEnd = 0;
+        }
+        /* 2. We enforce the maximum offset allowed.
+         *
+         * kMaxChunkSize should be small enough that we don't lose too much of
+         * the window through early invalidation.
+         * TODO: * Test the chunk size.
+         *       * Try invalidation after the sequence generation and test the
+         *         the offset against maxDist directly.
+         *
+         * NOTE: Because of dictionaries + sequence splitting we MUST make sure
+         * that any offset used is valid at the END of the sequence, since it may
+         * be split into two sequences. This condition holds when using
+         * ZSTD_window_enforceMaxDist(), but if we move to checking offsets
+         * against maxDist directly, we'll have to carefully handle that case.
+         */
+        ZSTD_window_enforceMaxDist(&ldmState->window, chunkEnd, maxDist, &ldmState->loadedDictEnd, NULL);
+        /* 3. Generate the sequences for the chunk, and get newLeftoverSize. */
+        newLeftoverSize = ZSTD_ldm_generateSequences_internal(
+            ldmState, sequences, params, chunkStart, chunkSize);
+        if (ZSTD_isError(newLeftoverSize))
+            return newLeftoverSize;
+        /* 4. We add the leftover literals from previous iterations to the first
+         *    newly generated sequence, or add the `newLeftoverSize` if none are
+         *    generated.
+         */
+        /* Prepend the leftover literals from the last call */
+        if (prevSize < sequences->size) {
+            sequences->seq[prevSize].litLength += (U32)leftoverSize;
+            leftoverSize = newLeftoverSize;
+        } else {
+            assert(newLeftoverSize == chunkSize);
+            leftoverSize += chunkSize;
+        }
+    }
+    return 0;
+}
+
+void ZSTD_ldm_skipSequences(rawSeqStore_t* rawSeqStore, size_t srcSize, U32 const minMatch) {
+    while (srcSize > 0 && rawSeqStore->pos < rawSeqStore->size) {
+        rawSeq* seq = rawSeqStore->seq + rawSeqStore->pos;
+        if (srcSize <= seq->litLength) {
+            /* Skip past srcSize literals */
+            seq->litLength -= (U32)srcSize;
+            return;
+        }
+        srcSize -= seq->litLength;
+        seq->litLength = 0;
+        if (srcSize < seq->matchLength) {
+            /* Skip past the first srcSize of the match */
+            seq->matchLength -= (U32)srcSize;
+            if (seq->matchLength < minMatch) {
+                /* The match is too short, omit it */
+                if (rawSeqStore->pos + 1 < rawSeqStore->size) {
+                    seq[1].litLength += seq[0].matchLength;
+                }
+                rawSeqStore->pos++;
+            }
+            return;
+        }
+        srcSize -= seq->matchLength;
+        seq->matchLength = 0;
+        rawSeqStore->pos++;
+    }
+}
+
+/*
+ * If the sequence length is longer than remaining then the sequence is split
+ * between this block and the next.
+ *
+ * Returns the current sequence to handle, or if the rest of the block should
+ * be literals, it returns a sequence with offset == 0.
+ */
+static rawSeq maybeSplitSequence(rawSeqStore_t* rawSeqStore,
+                                 U32 const remaining, U32 const minMatch)
+{
+    rawSeq sequence = rawSeqStore->seq[rawSeqStore->pos];
+    assert(sequence.offset > 0);
+    /* Likely: No partial sequence */
+    if (remaining >= sequence.litLength + sequence.matchLength) {
+        rawSeqStore->pos++;
+        return sequence;
+    }
+    /* Cut the sequence short (offset == 0 ==> rest is literals). */
+    if (remaining <= sequence.litLength) {
+        sequence.offset = 0;
+    } else if (remaining < sequence.litLength + sequence.matchLength) {
+        sequence.matchLength = remaining - sequence.litLength;
+        if (sequence.matchLength < minMatch) {
+            sequence.offset = 0;
+        }
+    }
+    /* Skip past `remaining` bytes for the future sequences. */
+    ZSTD_ldm_skipSequences(rawSeqStore, remaining, minMatch);
+    return sequence;
+}
+
+void ZSTD_ldm_skipRawSeqStoreBytes(rawSeqStore_t* rawSeqStore, size_t nbBytes) {
+    U32 currPos = (U32)(rawSeqStore->posInSequence + nbBytes);
+    while (currPos && rawSeqStore->pos < rawSeqStore->size) {
+        rawSeq currSeq = rawSeqStore->seq[rawSeqStore->pos];
+        if (currPos >= currSeq.litLength + currSeq.matchLength) {
+            currPos -= currSeq.litLength + currSeq.matchLength;
+            rawSeqStore->pos++;
+        } else {
+            rawSeqStore->posInSequence = currPos;
+            break;
+        }
+    }
+    if (currPos == 0 || rawSeqStore->pos == rawSeqStore->size) {
+        rawSeqStore->posInSequence = 0;
+    }
+}
+
+size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
+    ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+    void const* src, size_t srcSize)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    unsigned const minMatch = cParams->minMatch;
+    ZSTD_blockCompressor const blockCompressor =
+        ZSTD_selectBlockCompressor(cParams->strategy, ZSTD_matchState_dictMode(ms));
+    /* Input bounds */
+    BYTE const* const istart = (BYTE const*)src;
+    BYTE const* const iend = istart + srcSize;
+    /* Input positions */
+    BYTE const* ip = istart;
+
+    DEBUGLOG(5, "ZSTD_ldm_blockCompress: srcSize=%zu", srcSize);
+    /* If using opt parser, use LDMs only as candidates rather than always accepting them */
+    if (cParams->strategy >= ZSTD_btopt) {
+        size_t lastLLSize;
+        ms->ldmSeqStore = rawSeqStore;
+        lastLLSize = blockCompressor(ms, seqStore, rep, src, srcSize);
+        ZSTD_ldm_skipRawSeqStoreBytes(rawSeqStore, srcSize);
+        return lastLLSize;
+    }
+
+    assert(rawSeqStore->pos <= rawSeqStore->size);
+    assert(rawSeqStore->size <= rawSeqStore->capacity);
+    /* Loop through each sequence and apply the block compressor to the literals */
+    while (rawSeqStore->pos < rawSeqStore->size && ip < iend) {
+        /* maybeSplitSequence updates rawSeqStore->pos */
+        rawSeq const sequence = maybeSplitSequence(rawSeqStore,
+                                                   (U32)(iend - ip), minMatch);
+        int i;
+        /* End signal */
+        if (sequence.offset == 0)
+            break;
+
+        assert(ip + sequence.litLength + sequence.matchLength <= iend);
+
+        /* Fill tables for block compressor */
+        ZSTD_ldm_limitTableUpdate(ms, ip);
+        ZSTD_ldm_fillFastTables(ms, ip);
+        /* Run the block compressor */
+        DEBUGLOG(5, "pos %u : calling block compressor on segment of size %u", (unsigned)(ip-istart), sequence.litLength);
+        {
+            size_t const newLitLength =
+                blockCompressor(ms, seqStore, rep, ip, sequence.litLength);
+            ip += sequence.litLength;
+            /* Update the repcodes */
+            for (i = ZSTD_REP_NUM - 1; i > 0; i--)
+                rep[i] = rep[i-1];
+            rep[0] = sequence.offset;
+            /* Store the sequence */
+            ZSTD_storeSeq(seqStore, newLitLength, ip - newLitLength, iend,
+                          sequence.offset + ZSTD_REP_MOVE,
+                          sequence.matchLength - MINMATCH);
+            ip += sequence.matchLength;
+        }
+    }
+    /* Fill the tables for the block compressor */
+    ZSTD_ldm_limitTableUpdate(ms, ip);
+    ZSTD_ldm_fillFastTables(ms, ip);
+    /* Compress the last literals */
+    return blockCompressor(ms, seqStore, rep, ip, iend - ip);
+}
diff --git a/lib/zstd/compress/zstd_ldm.h b/lib/zstd/compress/zstd_ldm.h
new file mode 100644
index 0000000..25b2527
--- /dev/null
+++ b/lib/zstd/compress/zstd_ldm.h
@@ -0,0 +1,110 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_LDM_H
+#define ZSTD_LDM_H
+
+
+#include "zstd_compress_internal.h"   /* ldmParams_t, U32 */
+#include <linux/zstd.h>   /* ZSTD_CCtx, size_t */
+
+/*-*************************************
+*  Long distance matching
+***************************************/
+
+#define ZSTD_LDM_DEFAULT_WINDOW_LOG ZSTD_WINDOWLOG_LIMIT_DEFAULT
+
+void ZSTD_ldm_fillHashTable(
+            ldmState_t* state, const BYTE* ip,
+            const BYTE* iend, ldmParams_t const* params);
+
+/*
+ * ZSTD_ldm_generateSequences():
+ *
+ * Generates the sequences using the long distance match finder.
+ * Generates long range matching sequences in `sequences`, which parse a prefix
+ * of the source. `sequences` must be large enough to store every sequence,
+ * which can be checked with `ZSTD_ldm_getMaxNbSeq()`.
+ * @returns 0 or an error code.
+ *
+ * NOTE: The user must have called ZSTD_window_update() for all of the input
+ * they have, even if they pass it to ZSTD_ldm_generateSequences() in chunks.
+ * NOTE: This function returns an error if it runs out of space to store
+ *       sequences.
+ */
+size_t ZSTD_ldm_generateSequences(
+            ldmState_t* ldms, rawSeqStore_t* sequences,
+            ldmParams_t const* params, void const* src, size_t srcSize);
+
+/*
+ * ZSTD_ldm_blockCompress():
+ *
+ * Compresses a block using the predefined sequences, along with a secondary
+ * block compressor. The literals section of every sequence is passed to the
+ * secondary block compressor, and those sequences are interspersed with the
+ * predefined sequences. Returns the length of the last literals.
+ * Updates `rawSeqStore.pos` to indicate how many sequences have been consumed.
+ * `rawSeqStore.seq` may also be updated to split the last sequence between two
+ * blocks.
+ * @return The length of the last literals.
+ *
+ * NOTE: The source must be at most the maximum block size, but the predefined
+ * sequences can be any size, and may be longer than the block. In the case that
+ * they are longer than the block, the last sequences may need to be split into
+ * two. We handle that case correctly, and update `rawSeqStore` appropriately.
+ * NOTE: This function does not return any errors.
+ */
+size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
+            ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+            void const* src, size_t srcSize);
+
+/*
+ * ZSTD_ldm_skipSequences():
+ *
+ * Skip past `srcSize` bytes worth of sequences in `rawSeqStore`.
+ * Avoids emitting matches less than `minMatch` bytes.
+ * Must be called for data that is not passed to ZSTD_ldm_blockCompress().
+ */
+void ZSTD_ldm_skipSequences(rawSeqStore_t* rawSeqStore, size_t srcSize,
+    U32 const minMatch);
+
+/* ZSTD_ldm_skipRawSeqStoreBytes():
+ * Moves forward in rawSeqStore by nbBytes, updating fields 'pos' and 'posInSequence'.
+ * Not to be used in conjunction with ZSTD_ldm_skipSequences().
+ * Must be called for data with is not passed to ZSTD_ldm_blockCompress().
+ */
+void ZSTD_ldm_skipRawSeqStoreBytes(rawSeqStore_t* rawSeqStore, size_t nbBytes);
+
+/* ZSTD_ldm_getTableSize() :
+ *  Estimate the space needed for long distance matching tables or 0 if LDM is
+ *  disabled.
+ */
+size_t ZSTD_ldm_getTableSize(ldmParams_t params);
+
+/* ZSTD_ldm_getSeqSpace() :
+ *  Return an upper bound on the number of sequences that can be produced by
+ *  the long distance matcher, or 0 if LDM is disabled.
+ */
+size_t ZSTD_ldm_getMaxNbSeq(ldmParams_t params, size_t maxChunkSize);
+
+/* ZSTD_ldm_adjustParameters() :
+ *  If the params->hashRateLog is not set, set it to its default value based on
+ *  windowLog and params->hashLog.
+ *
+ *  Ensures that params->bucketSizeLog is <= params->hashLog (setting it to
+ *  params->hashLog if it is not).
+ *
+ *  Ensures that the minMatchLength >= targetLength during optimal parsing.
+ */
+void ZSTD_ldm_adjustParameters(ldmParams_t* params,
+                               ZSTD_compressionParameters const* cParams);
+
+
+#endif /* ZSTD_FAST_H */
diff --git a/lib/zstd/compress/zstd_ldm_geartab.h b/lib/zstd/compress/zstd_ldm_geartab.h
new file mode 100644
index 0000000..e5c24d8
--- /dev/null
+++ b/lib/zstd/compress/zstd_ldm_geartab.h
@@ -0,0 +1,103 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_LDM_GEARTAB_H
+#define ZSTD_LDM_GEARTAB_H
+
+static U64 ZSTD_ldm_gearTab[256] = {
+    0xf5b8f72c5f77775c, 0x84935f266b7ac412, 0xb647ada9ca730ccc,
+    0xb065bb4b114fb1de, 0x34584e7e8c3a9fd0, 0x4e97e17c6ae26b05,
+    0x3a03d743bc99a604, 0xcecd042422c4044f, 0x76de76c58524259e,
+    0x9c8528f65badeaca, 0x86563706e2097529, 0x2902475fa375d889,
+    0xafb32a9739a5ebe6, 0xce2714da3883e639, 0x21eaf821722e69e,
+    0x37b628620b628,    0x49a8d455d88caf5,  0x8556d711e6958140,
+    0x4f7ae74fc605c1f,  0x829f0c3468bd3a20, 0x4ffdc885c625179e,
+    0x8473de048a3daf1b, 0x51008822b05646b2, 0x69d75d12b2d1cc5f,
+    0x8c9d4a19159154bc, 0xc3cc10f4abbd4003, 0xd06ddc1cecb97391,
+    0xbe48e6e7ed80302e, 0x3481db31cee03547, 0xacc3f67cdaa1d210,
+    0x65cb771d8c7f96cc, 0x8eb27177055723dd, 0xc789950d44cd94be,
+    0x934feadc3700b12b, 0x5e485f11edbdf182, 0x1e2e2a46fd64767a,
+    0x2969ca71d82efa7c, 0x9d46e9935ebbba2e, 0xe056b67e05e6822b,
+    0x94d73f55739d03a0, 0xcd7010bdb69b5a03, 0x455ef9fcd79b82f4,
+    0x869cb54a8749c161, 0x38d1a4fa6185d225, 0xb475166f94bbe9bb,
+    0xa4143548720959f1, 0x7aed4780ba6b26ba, 0xd0ce264439e02312,
+    0x84366d746078d508, 0xa8ce973c72ed17be, 0x21c323a29a430b01,
+    0x9962d617e3af80ee, 0xab0ce91d9c8cf75b, 0x530e8ee6d19a4dbc,
+    0x2ef68c0cf53f5d72, 0xc03a681640a85506, 0x496e4e9f9c310967,
+    0x78580472b59b14a0, 0x273824c23b388577, 0x66bf923ad45cb553,
+    0x47ae1a5a2492ba86, 0x35e304569e229659, 0x4765182a46870b6f,
+    0x6cbab625e9099412, 0xddac9a2e598522c1, 0x7172086e666624f2,
+    0xdf5003ca503b7837, 0x88c0c1db78563d09, 0x58d51865acfc289d,
+    0x177671aec65224f1, 0xfb79d8a241e967d7, 0x2be1e101cad9a49a,
+    0x6625682f6e29186b, 0x399553457ac06e50, 0x35dffb4c23abb74,
+    0x429db2591f54aade, 0xc52802a8037d1009, 0x6acb27381f0b25f3,
+    0xf45e2551ee4f823b, 0x8b0ea2d99580c2f7, 0x3bed519cbcb4e1e1,
+    0xff452823dbb010a,  0x9d42ed614f3dd267, 0x5b9313c06257c57b,
+    0xa114b8008b5e1442, 0xc1fe311c11c13d4b, 0x66e8763ea34c5568,
+    0x8b982af1c262f05d, 0xee8876faaa75fbb7, 0x8a62a4d0d172bb2a,
+    0xc13d94a3b7449a97, 0x6dbbba9dc15d037c, 0xc786101f1d92e0f1,
+    0xd78681a907a0b79b, 0xf61aaf2962c9abb9, 0x2cfd16fcd3cb7ad9,
+    0x868c5b6744624d21, 0x25e650899c74ddd7, 0xba042af4a7c37463,
+    0x4eb1a539465a3eca, 0xbe09dbf03b05d5ca, 0x774e5a362b5472ba,
+    0x47a1221229d183cd, 0x504b0ca18ef5a2df, 0xdffbdfbde2456eb9,
+    0x46cd2b2fbee34634, 0xf2aef8fe819d98c3, 0x357f5276d4599d61,
+    0x24a5483879c453e3, 0x88026889192b4b9,  0x28da96671782dbec,
+    0x4ef37c40588e9aaa, 0x8837b90651bc9fb3, 0xc164f741d3f0e5d6,
+    0xbc135a0a704b70ba, 0x69cd868f7622ada,  0xbc37ba89e0b9c0ab,
+    0x47c14a01323552f6, 0x4f00794bacee98bb, 0x7107de7d637a69d5,
+    0x88af793bb6f2255e, 0xf3c6466b8799b598, 0xc288c616aa7f3b59,
+    0x81ca63cf42fca3fd, 0x88d85ace36a2674b, 0xd056bd3792389e7,
+    0xe55c396c4e9dd32d, 0xbefb504571e6c0a6, 0x96ab32115e91e8cc,
+    0xbf8acb18de8f38d1, 0x66dae58801672606, 0x833b6017872317fb,
+    0xb87c16f2d1c92864, 0xdb766a74e58b669c, 0x89659f85c61417be,
+    0xc8daad856011ea0c, 0x76a4b565b6fe7eae, 0xa469d085f6237312,
+    0xaaf0365683a3e96c, 0x4dbb746f8424f7b8, 0x638755af4e4acc1,
+    0x3d7807f5bde64486, 0x17be6d8f5bbb7639, 0x903f0cd44dc35dc,
+    0x67b672eafdf1196c, 0xa676ff93ed4c82f1, 0x521d1004c5053d9d,
+    0x37ba9ad09ccc9202, 0x84e54d297aacfb51, 0xa0b4b776a143445,
+    0x820d471e20b348e,  0x1874383cb83d46dc, 0x97edeec7a1efe11c,
+    0xb330e50b1bdc42aa, 0x1dd91955ce70e032, 0xa514cdb88f2939d5,
+    0x2791233fd90db9d3, 0x7b670a4cc50f7a9b, 0x77c07d2a05c6dfa5,
+    0xe3778b6646d0a6fa, 0xb39c8eda47b56749, 0x933ed448addbef28,
+    0xaf846af6ab7d0bf4, 0xe5af208eb666e49,  0x5e6622f73534cd6a,
+    0x297daeca42ef5b6e, 0x862daef3d35539a6, 0xe68722498f8e1ea9,
+    0x981c53093dc0d572, 0xfa09b0bfbf86fbf5, 0x30b1e96166219f15,
+    0x70e7d466bdc4fb83, 0x5a66736e35f2a8e9, 0xcddb59d2b7c1baef,
+    0xd6c7d247d26d8996, 0xea4e39eac8de1ba3, 0x539c8bb19fa3aff2,
+    0x9f90e4c5fd508d8,  0xa34e5956fbaf3385, 0x2e2f8e151d3ef375,
+    0x173691e9b83faec1, 0xb85a8d56bf016379, 0x8382381267408ae3,
+    0xb90f901bbdc0096d, 0x7c6ad32933bcec65, 0x76bb5e2f2c8ad595,
+    0x390f851a6cf46d28, 0xc3e6064da1c2da72, 0xc52a0c101cfa5389,
+    0xd78eaf84a3fbc530, 0x3781b9e2288b997e, 0x73c2f6dea83d05c4,
+    0x4228e364c5b5ed7,  0x9d7a3edf0da43911, 0x8edcfeda24686756,
+    0x5e7667a7b7a9b3a1, 0x4c4f389fa143791d, 0xb08bc1023da7cddc,
+    0x7ab4be3ae529b1cc, 0x754e6132dbe74ff9, 0x71635442a839df45,
+    0x2f6fb1643fbe52de, 0x961e0a42cf7a8177, 0xf3b45d83d89ef2ea,
+    0xee3de4cf4a6e3e9b, 0xcd6848542c3295e7, 0xe4cee1664c78662f,
+    0x9947548b474c68c4, 0x25d73777a5ed8b0b, 0xc915b1d636b7fc,
+    0x21c2ba75d9b0d2da, 0x5f6b5dcf608a64a1, 0xdcf333255ff9570c,
+    0x633b922418ced4ee, 0xc136dde0b004b34a, 0x58cc83b05d4b2f5a,
+    0x5eb424dda28e42d2, 0x62df47369739cd98, 0xb4e0b42485e4ce17,
+    0x16e1f0c1f9a8d1e7, 0x8ec3916707560ebf, 0x62ba6e2df2cc9db3,
+    0xcbf9f4ff77d83a16, 0x78d9d7d07d2bbcc4, 0xef554ce1e02c41f4,
+    0x8d7581127eccf94d, 0xa9b53336cb3c8a05, 0x38c42c0bf45c4f91,
+    0x640893cdf4488863, 0x80ec34bc575ea568, 0x39f324f5b48eaa40,
+    0xe9d9ed1f8eff527f, 0x9224fc058cc5a214, 0xbaba00b04cfe7741,
+    0x309a9f120fcf52af, 0xa558f3ec65626212, 0x424bec8b7adabe2f,
+    0x41622513a6aea433, 0xb88da2d5324ca798, 0xd287733b245528a4,
+    0x9a44697e6d68aec3, 0x7b1093be2f49bb28, 0x50bbec632e3d8aad,
+    0x6cd90723e1ea8283, 0x897b9e7431b02bf3, 0x219efdcb338a7047,
+    0x3b0311f0a27c0656, 0xdb17bf91c0db96e7, 0x8cd4fd6b4e85a5b2,
+    0xfab071054ba6409d, 0x40d6fe831fa9dfd9, 0xaf358debad7d791e,
+    0xeb8d0e25a65e3e58, 0xbbcbd3df14e08580, 0xcf751f27ecdab2b,
+    0x2b4da14f2613d8f4
+};
+
+#endif /* ZSTD_LDM_GEARTAB_H */
diff --git a/lib/zstd/compress/zstd_opt.c b/lib/zstd/compress/zstd_opt.c
new file mode 100644
index 0000000..0433705
--- /dev/null
+++ b/lib/zstd/compress/zstd_opt.c
@@ -0,0 +1,1346 @@
+/*
+ * Copyright (c) Przemyslaw Skibinski, Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#include "zstd_compress_internal.h"
+#include "hist.h"
+#include "zstd_opt.h"
+
+
+#define ZSTD_LITFREQ_ADD    2   /* scaling factor for litFreq, so that frequencies adapt faster to new stats */
+#define ZSTD_FREQ_DIV       4   /* log factor when using previous stats to init next stats */
+#define ZSTD_MAX_PRICE     (1<<30)
+
+#define ZSTD_PREDEF_THRESHOLD 1024   /* if srcSize < ZSTD_PREDEF_THRESHOLD, symbols' cost is assumed static, directly determined by pre-defined distributions */
+
+
+/*-*************************************
+*  Price functions for optimal parser
+***************************************/
+
+#if 0    /* approximation at bit level */
+#  define BITCOST_ACCURACY 0
+#  define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
+#  define WEIGHT(stat)  ((void)opt, ZSTD_bitWeight(stat))
+#elif 0  /* fractional bit accuracy */
+#  define BITCOST_ACCURACY 8
+#  define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
+#  define WEIGHT(stat,opt) ((void)opt, ZSTD_fracWeight(stat))
+#else    /* opt==approx, ultra==accurate */
+#  define BITCOST_ACCURACY 8
+#  define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
+#  define WEIGHT(stat,opt) (opt ? ZSTD_fracWeight(stat) : ZSTD_bitWeight(stat))
+#endif
+
+MEM_STATIC U32 ZSTD_bitWeight(U32 stat)
+{
+    return (ZSTD_highbit32(stat+1) * BITCOST_MULTIPLIER);
+}
+
+MEM_STATIC U32 ZSTD_fracWeight(U32 rawStat)
+{
+    U32 const stat = rawStat + 1;
+    U32 const hb = ZSTD_highbit32(stat);
+    U32 const BWeight = hb * BITCOST_MULTIPLIER;
+    U32 const FWeight = (stat << BITCOST_ACCURACY) >> hb;
+    U32 const weight = BWeight + FWeight;
+    assert(hb + BITCOST_ACCURACY < 31);
+    return weight;
+}
+
+#if (DEBUGLEVEL>=2)
+/* debugging function,
+ * @return price in bytes as fractional value
+ * for debug messages only */
+MEM_STATIC double ZSTD_fCost(U32 price)
+{
+    return (double)price / (BITCOST_MULTIPLIER*8);
+}
+#endif
+
+static int ZSTD_compressedLiterals(optState_t const* const optPtr)
+{
+    return optPtr->literalCompressionMode != ZSTD_lcm_uncompressed;
+}
+
+static void ZSTD_setBasePrices(optState_t* optPtr, int optLevel)
+{
+    if (ZSTD_compressedLiterals(optPtr))
+        optPtr->litSumBasePrice = WEIGHT(optPtr->litSum, optLevel);
+    optPtr->litLengthSumBasePrice = WEIGHT(optPtr->litLengthSum, optLevel);
+    optPtr->matchLengthSumBasePrice = WEIGHT(optPtr->matchLengthSum, optLevel);
+    optPtr->offCodeSumBasePrice = WEIGHT(optPtr->offCodeSum, optLevel);
+}
+
+
+/* ZSTD_downscaleStat() :
+ * reduce all elements in table by a factor 2^(ZSTD_FREQ_DIV+malus)
+ * return the resulting sum of elements */
+static U32 ZSTD_downscaleStat(unsigned* table, U32 lastEltIndex, int malus)
+{
+    U32 s, sum=0;
+    DEBUGLOG(5, "ZSTD_downscaleStat (nbElts=%u)", (unsigned)lastEltIndex+1);
+    assert(ZSTD_FREQ_DIV+malus > 0 && ZSTD_FREQ_DIV+malus < 31);
+    for (s=0; s<lastEltIndex+1; s++) {
+        table[s] = 1 + (table[s] >> (ZSTD_FREQ_DIV+malus));
+        sum += table[s];
+    }
+    return sum;
+}
+
+/* ZSTD_rescaleFreqs() :
+ * if first block (detected by optPtr->litLengthSum == 0) : init statistics
+ *    take hints from dictionary if there is one
+ *    or init from zero, using src for literals stats, or flat 1 for match symbols
+ * otherwise downscale existing stats, to be used as seed for next block.
+ */
+static void
+ZSTD_rescaleFreqs(optState_t* const optPtr,
+            const BYTE* const src, size_t const srcSize,
+                  int const optLevel)
+{
+    int const compressedLiterals = ZSTD_compressedLiterals(optPtr);
+    DEBUGLOG(5, "ZSTD_rescaleFreqs (srcSize=%u)", (unsigned)srcSize);
+    optPtr->priceType = zop_dynamic;
+
+    if (optPtr->litLengthSum == 0) {  /* first block : init */
+        if (srcSize <= ZSTD_PREDEF_THRESHOLD) {  /* heuristic */
+            DEBUGLOG(5, "(srcSize <= ZSTD_PREDEF_THRESHOLD) => zop_predef");
+            optPtr->priceType = zop_predef;
+        }
+
+        assert(optPtr->symbolCosts != NULL);
+        if (optPtr->symbolCosts->huf.repeatMode == HUF_repeat_valid) {
+            /* huffman table presumed generated by dictionary */
+            optPtr->priceType = zop_dynamic;
+
+            if (compressedLiterals) {
+                unsigned lit;
+                assert(optPtr->litFreq != NULL);
+                optPtr->litSum = 0;
+                for (lit=0; lit<=MaxLit; lit++) {
+                    U32 const scaleLog = 11;   /* scale to 2K */
+                    U32 const bitCost = HUF_getNbBits(optPtr->symbolCosts->huf.CTable, lit);
+                    assert(bitCost <= scaleLog);
+                    optPtr->litFreq[lit] = bitCost ? 1 << (scaleLog-bitCost) : 1 /*minimum to calculate cost*/;
+                    optPtr->litSum += optPtr->litFreq[lit];
+            }   }
+
+            {   unsigned ll;
+                FSE_CState_t llstate;
+                FSE_initCState(&llstate, optPtr->symbolCosts->fse.litlengthCTable);
+                optPtr->litLengthSum = 0;
+                for (ll=0; ll<=MaxLL; ll++) {
+                    U32 const scaleLog = 10;   /* scale to 1K */
+                    U32 const bitCost = FSE_getMaxNbBits(llstate.symbolTT, ll);
+                    assert(bitCost < scaleLog);
+                    optPtr->litLengthFreq[ll] = bitCost ? 1 << (scaleLog-bitCost) : 1 /*minimum to calculate cost*/;
+                    optPtr->litLengthSum += optPtr->litLengthFreq[ll];
+            }   }
+
+            {   unsigned ml;
+                FSE_CState_t mlstate;
+                FSE_initCState(&mlstate, optPtr->symbolCosts->fse.matchlengthCTable);
+                optPtr->matchLengthSum = 0;
+                for (ml=0; ml<=MaxML; ml++) {
+                    U32 const scaleLog = 10;
+                    U32 const bitCost = FSE_getMaxNbBits(mlstate.symbolTT, ml);
+                    assert(bitCost < scaleLog);
+                    optPtr->matchLengthFreq[ml] = bitCost ? 1 << (scaleLog-bitCost) : 1 /*minimum to calculate cost*/;
+                    optPtr->matchLengthSum += optPtr->matchLengthFreq[ml];
+            }   }
+
+            {   unsigned of;
+                FSE_CState_t ofstate;
+                FSE_initCState(&ofstate, optPtr->symbolCosts->fse.offcodeCTable);
+                optPtr->offCodeSum = 0;
+                for (of=0; of<=MaxOff; of++) {
+                    U32 const scaleLog = 10;
+                    U32 const bitCost = FSE_getMaxNbBits(ofstate.symbolTT, of);
+                    assert(bitCost < scaleLog);
+                    optPtr->offCodeFreq[of] = bitCost ? 1 << (scaleLog-bitCost) : 1 /*minimum to calculate cost*/;
+                    optPtr->offCodeSum += optPtr->offCodeFreq[of];
+            }   }
+
+        } else {  /* not a dictionary */
+
+            assert(optPtr->litFreq != NULL);
+            if (compressedLiterals) {
+                unsigned lit = MaxLit;
+                HIST_count_simple(optPtr->litFreq, &lit, src, srcSize);   /* use raw first block to init statistics */
+                optPtr->litSum = ZSTD_downscaleStat(optPtr->litFreq, MaxLit, 1);
+            }
+
+            {   unsigned ll;
+                for (ll=0; ll<=MaxLL; ll++)
+                    optPtr->litLengthFreq[ll] = 1;
+            }
+            optPtr->litLengthSum = MaxLL+1;
+
+            {   unsigned ml;
+                for (ml=0; ml<=MaxML; ml++)
+                    optPtr->matchLengthFreq[ml] = 1;
+            }
+            optPtr->matchLengthSum = MaxML+1;
+
+            {   unsigned of;
+                for (of=0; of<=MaxOff; of++)
+                    optPtr->offCodeFreq[of] = 1;
+            }
+            optPtr->offCodeSum = MaxOff+1;
+
+        }
+
+    } else {   /* new block : re-use previous statistics, scaled down */
+
+        if (compressedLiterals)
+            optPtr->litSum = ZSTD_downscaleStat(optPtr->litFreq, MaxLit, 1);
+        optPtr->litLengthSum = ZSTD_downscaleStat(optPtr->litLengthFreq, MaxLL, 0);
+        optPtr->matchLengthSum = ZSTD_downscaleStat(optPtr->matchLengthFreq, MaxML, 0);
+        optPtr->offCodeSum = ZSTD_downscaleStat(optPtr->offCodeFreq, MaxOff, 0);
+    }
+
+    ZSTD_setBasePrices(optPtr, optLevel);
+}
+
+/* ZSTD_rawLiteralsCost() :
+ * price of literals (only) in specified segment (which length can be 0).
+ * does not include price of literalLength symbol */
+static U32 ZSTD_rawLiteralsCost(const BYTE* const literals, U32 const litLength,
+                                const optState_t* const optPtr,
+                                int optLevel)
+{
+    if (litLength == 0) return 0;
+
+    if (!ZSTD_compressedLiterals(optPtr))
+        return (litLength << 3) * BITCOST_MULTIPLIER;  /* Uncompressed - 8 bytes per literal. */
+
+    if (optPtr->priceType == zop_predef)
+        return (litLength*6) * BITCOST_MULTIPLIER;  /* 6 bit per literal - no statistic used */
+
+    /* dynamic statistics */
+    {   U32 price = litLength * optPtr->litSumBasePrice;
+        U32 u;
+        for (u=0; u < litLength; u++) {
+            assert(WEIGHT(optPtr->litFreq[literals[u]], optLevel) <= optPtr->litSumBasePrice);   /* literal cost should never be negative */
+            price -= WEIGHT(optPtr->litFreq[literals[u]], optLevel);
+        }
+        return price;
+    }
+}
+
+/* ZSTD_litLengthPrice() :
+ * cost of literalLength symbol */
+static U32 ZSTD_litLengthPrice(U32 const litLength, const optState_t* const optPtr, int optLevel)
+{
+    if (optPtr->priceType == zop_predef) return WEIGHT(litLength, optLevel);
+
+    /* dynamic statistics */
+    {   U32 const llCode = ZSTD_LLcode(litLength);
+        return (LL_bits[llCode] * BITCOST_MULTIPLIER)
+             + optPtr->litLengthSumBasePrice
+             - WEIGHT(optPtr->litLengthFreq[llCode], optLevel);
+    }
+}
+
+/* ZSTD_getMatchPrice() :
+ * Provides the cost of the match part (offset + matchLength) of a sequence
+ * Must be combined with ZSTD_fullLiteralsCost() to get the full cost of a sequence.
+ * optLevel: when <2, favors small offset for decompression speed (improved cache efficiency) */
+FORCE_INLINE_TEMPLATE U32
+ZSTD_getMatchPrice(U32 const offset,
+                   U32 const matchLength,
+             const optState_t* const optPtr,
+                   int const optLevel)
+{
+    U32 price;
+    U32 const offCode = ZSTD_highbit32(offset+1);
+    U32 const mlBase = matchLength - MINMATCH;
+    assert(matchLength >= MINMATCH);
+
+    if (optPtr->priceType == zop_predef)  /* fixed scheme, do not use statistics */
+        return WEIGHT(mlBase, optLevel) + ((16 + offCode) * BITCOST_MULTIPLIER);
+
+    /* dynamic statistics */
+    price = (offCode * BITCOST_MULTIPLIER) + (optPtr->offCodeSumBasePrice - WEIGHT(optPtr->offCodeFreq[offCode], optLevel));
+    if ((optLevel<2) /*static*/ && offCode >= 20)
+        price += (offCode-19)*2 * BITCOST_MULTIPLIER; /* handicap for long distance offsets, favor decompression speed */
+
+    /* match Length */
+    {   U32 const mlCode = ZSTD_MLcode(mlBase);
+        price += (ML_bits[mlCode] * BITCOST_MULTIPLIER) + (optPtr->matchLengthSumBasePrice - WEIGHT(optPtr->matchLengthFreq[mlCode], optLevel));
+    }
+
+    price += BITCOST_MULTIPLIER / 5;   /* heuristic : make matches a bit more costly to favor less sequences -> faster decompression speed */
+
+    DEBUGLOG(8, "ZSTD_getMatchPrice(ml:%u) = %u", matchLength, price);
+    return price;
+}
+
+/* ZSTD_updateStats() :
+ * assumption : literals + litLengtn <= iend */
+static void ZSTD_updateStats(optState_t* const optPtr,
+                             U32 litLength, const BYTE* literals,
+                             U32 offsetCode, U32 matchLength)
+{
+    /* literals */
+    if (ZSTD_compressedLiterals(optPtr)) {
+        U32 u;
+        for (u=0; u < litLength; u++)
+            optPtr->litFreq[literals[u]] += ZSTD_LITFREQ_ADD;
+        optPtr->litSum += litLength*ZSTD_LITFREQ_ADD;
+    }
+
+    /* literal Length */
+    {   U32 const llCode = ZSTD_LLcode(litLength);
+        optPtr->litLengthFreq[llCode]++;
+        optPtr->litLengthSum++;
+    }
+
+    /* match offset code (0-2=>repCode; 3+=>offset+2) */
+    {   U32 const offCode = ZSTD_highbit32(offsetCode+1);
+        assert(offCode <= MaxOff);
+        optPtr->offCodeFreq[offCode]++;
+        optPtr->offCodeSum++;
+    }
+
+    /* match Length */
+    {   U32 const mlBase = matchLength - MINMATCH;
+        U32 const mlCode = ZSTD_MLcode(mlBase);
+        optPtr->matchLengthFreq[mlCode]++;
+        optPtr->matchLengthSum++;
+    }
+}
+
+
+/* ZSTD_readMINMATCH() :
+ * function safe only for comparisons
+ * assumption : memPtr must be at least 4 bytes before end of buffer */
+MEM_STATIC U32 ZSTD_readMINMATCH(const void* memPtr, U32 length)
+{
+    switch (length)
+    {
+    default :
+    case 4 : return MEM_read32(memPtr);
+    case 3 : if (MEM_isLittleEndian())
+                return MEM_read32(memPtr)<<8;
+             else
+                return MEM_read32(memPtr)>>8;
+    }
+}
+
+
+/* Update hashTable3 up to ip (excluded)
+   Assumption : always within prefix (i.e. not within extDict) */
+static U32 ZSTD_insertAndFindFirstIndexHash3 (ZSTD_matchState_t* ms,
+                                              U32* nextToUpdate3,
+                                              const BYTE* const ip)
+{
+    U32* const hashTable3 = ms->hashTable3;
+    U32 const hashLog3 = ms->hashLog3;
+    const BYTE* const base = ms->window.base;
+    U32 idx = *nextToUpdate3;
+    U32 const target = (U32)(ip - base);
+    size_t const hash3 = ZSTD_hash3Ptr(ip, hashLog3);
+    assert(hashLog3 > 0);
+
+    while(idx < target) {
+        hashTable3[ZSTD_hash3Ptr(base+idx, hashLog3)] = idx;
+        idx++;
+    }
+
+    *nextToUpdate3 = target;
+    return hashTable3[hash3];
+}
+
+
+/*-*************************************
+*  Binary Tree search
+***************************************/
+/* ZSTD_insertBt1() : add one or multiple positions to tree.
+ *  ip : assumed <= iend-8 .
+ * @return : nb of positions added */
+static U32 ZSTD_insertBt1(
+                ZSTD_matchState_t* ms,
+                const BYTE* const ip, const BYTE* const iend,
+                U32 const mls, const int extDict)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32*   const hashTable = ms->hashTable;
+    U32    const hashLog = cParams->hashLog;
+    size_t const h  = ZSTD_hashPtr(ip, hashLog, mls);
+    U32*   const bt = ms->chainTable;
+    U32    const btLog  = cParams->chainLog - 1;
+    U32    const btMask = (1 << btLog) - 1;
+    U32 matchIndex = hashTable[h];
+    size_t commonLengthSmaller=0, commonLengthLarger=0;
+    const BYTE* const base = ms->window.base;
+    const BYTE* const dictBase = ms->window.dictBase;
+    const U32 dictLimit = ms->window.dictLimit;
+    const BYTE* const dictEnd = dictBase + dictLimit;
+    const BYTE* const prefixStart = base + dictLimit;
+    const BYTE* match;
+    const U32 curr = (U32)(ip-base);
+    const U32 btLow = btMask >= curr ? 0 : curr - btMask;
+    U32* smallerPtr = bt + 2*(curr&btMask);
+    U32* largerPtr  = smallerPtr + 1;
+    U32 dummy32;   /* to be nullified at the end */
+    U32 const windowLow = ms->window.lowLimit;
+    U32 matchEndIdx = curr+8+1;
+    size_t bestLength = 8;
+    U32 nbCompares = 1U << cParams->searchLog;
+#ifdef ZSTD_C_PREDICT
+    U32 predictedSmall = *(bt + 2*((curr-1)&btMask) + 0);
+    U32 predictedLarge = *(bt + 2*((curr-1)&btMask) + 1);
+    predictedSmall += (predictedSmall>0);
+    predictedLarge += (predictedLarge>0);
+#endif /* ZSTD_C_PREDICT */
+
+    DEBUGLOG(8, "ZSTD_insertBt1 (%u)", curr);
+
+    assert(ip <= iend-8);   /* required for h calculation */
+    hashTable[h] = curr;   /* Update Hash Table */
+
+    assert(windowLow > 0);
+    for (; nbCompares && (matchIndex >= windowLow); --nbCompares) {
+        U32* const nextPtr = bt + 2*(matchIndex & btMask);
+        size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger);   /* guaranteed minimum nb of common bytes */
+        assert(matchIndex < curr);
+
+#ifdef ZSTD_C_PREDICT   /* note : can create issues when hlog small <= 11 */
+        const U32* predictPtr = bt + 2*((matchIndex-1) & btMask);   /* written this way, as bt is a roll buffer */
+        if (matchIndex == predictedSmall) {
+            /* no need to check length, result known */
+            *smallerPtr = matchIndex;
+            if (matchIndex <= btLow) { smallerPtr=&dummy32; break; }   /* beyond tree size, stop the search */
+            smallerPtr = nextPtr+1;               /* new "smaller" => larger of match */
+            matchIndex = nextPtr[1];              /* new matchIndex larger than previous (closer to current) */
+            predictedSmall = predictPtr[1] + (predictPtr[1]>0);
+            continue;
+        }
+        if (matchIndex == predictedLarge) {
+            *largerPtr = matchIndex;
+            if (matchIndex <= btLow) { largerPtr=&dummy32; break; }   /* beyond tree size, stop the search */
+            largerPtr = nextPtr;
+            matchIndex = nextPtr[0];
+            predictedLarge = predictPtr[0] + (predictPtr[0]>0);
+            continue;
+        }
+#endif
+
+        if (!extDict || (matchIndex+matchLength >= dictLimit)) {
+            assert(matchIndex+matchLength >= dictLimit);   /* might be wrong if actually extDict */
+            match = base + matchIndex;
+            matchLength += ZSTD_count(ip+matchLength, match+matchLength, iend);
+        } else {
+            match = dictBase + matchIndex;
+            matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
+            if (matchIndex+matchLength >= dictLimit)
+                match = base + matchIndex;   /* to prepare for next usage of match[matchLength] */
+        }
+
+        if (matchLength > bestLength) {
+            bestLength = matchLength;
+            if (matchLength > matchEndIdx - matchIndex)
+                matchEndIdx = matchIndex + (U32)matchLength;
+        }
+
+        if (ip+matchLength == iend) {   /* equal : no way to know if inf or sup */
+            break;   /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt tree */
+        }
+
+        if (match[matchLength] < ip[matchLength]) {  /* necessarily within buffer */
+            /* match is smaller than current */
+            *smallerPtr = matchIndex;             /* update smaller idx */
+            commonLengthSmaller = matchLength;    /* all smaller will now have at least this guaranteed common length */
+            if (matchIndex <= btLow) { smallerPtr=&dummy32; break; }   /* beyond tree size, stop searching */
+            smallerPtr = nextPtr+1;               /* new "candidate" => larger than match, which was smaller than target */
+            matchIndex = nextPtr[1];              /* new matchIndex, larger than previous and closer to current */
+        } else {
+            /* match is larger than current */
+            *largerPtr = matchIndex;
+            commonLengthLarger = matchLength;
+            if (matchIndex <= btLow) { largerPtr=&dummy32; break; }   /* beyond tree size, stop searching */
+            largerPtr = nextPtr;
+            matchIndex = nextPtr[0];
+    }   }
+
+    *smallerPtr = *largerPtr = 0;
+    {   U32 positions = 0;
+        if (bestLength > 384) positions = MIN(192, (U32)(bestLength - 384));   /* speed optimization */
+        assert(matchEndIdx > curr + 8);
+        return MAX(positions, matchEndIdx - (curr + 8));
+    }
+}
+
+FORCE_INLINE_TEMPLATE
+void ZSTD_updateTree_internal(
+                ZSTD_matchState_t* ms,
+                const BYTE* const ip, const BYTE* const iend,
+                const U32 mls, const ZSTD_dictMode_e dictMode)
+{
+    const BYTE* const base = ms->window.base;
+    U32 const target = (U32)(ip - base);
+    U32 idx = ms->nextToUpdate;
+    DEBUGLOG(6, "ZSTD_updateTree_internal, from %u to %u  (dictMode:%u)",
+                idx, target, dictMode);
+
+    while(idx < target) {
+        U32 const forward = ZSTD_insertBt1(ms, base+idx, iend, mls, dictMode == ZSTD_extDict);
+        assert(idx < (U32)(idx + forward));
+        idx += forward;
+    }
+    assert((size_t)(ip - base) <= (size_t)(U32)(-1));
+    assert((size_t)(iend - base) <= (size_t)(U32)(-1));
+    ms->nextToUpdate = target;
+}
+
+void ZSTD_updateTree(ZSTD_matchState_t* ms, const BYTE* ip, const BYTE* iend) {
+    ZSTD_updateTree_internal(ms, ip, iend, ms->cParams.minMatch, ZSTD_noDict);
+}
+
+FORCE_INLINE_TEMPLATE
+U32 ZSTD_insertBtAndGetAllMatches (
+                    ZSTD_match_t* matches,   /* store result (found matches) in this table (presumed large enough) */
+                    ZSTD_matchState_t* ms,
+                    U32* nextToUpdate3,
+                    const BYTE* const ip, const BYTE* const iLimit, const ZSTD_dictMode_e dictMode,
+                    const U32 rep[ZSTD_REP_NUM],
+                    U32 const ll0,   /* tells if associated literal length is 0 or not. This value must be 0 or 1 */
+                    const U32 lengthToBeat,
+                    U32 const mls /* template */)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32 const sufficient_len = MIN(cParams->targetLength, ZSTD_OPT_NUM -1);
+    const BYTE* const base = ms->window.base;
+    U32 const curr = (U32)(ip-base);
+    U32 const hashLog = cParams->hashLog;
+    U32 const minMatch = (mls==3) ? 3 : 4;
+    U32* const hashTable = ms->hashTable;
+    size_t const h  = ZSTD_hashPtr(ip, hashLog, mls);
+    U32 matchIndex  = hashTable[h];
+    U32* const bt   = ms->chainTable;
+    U32 const btLog = cParams->chainLog - 1;
+    U32 const btMask= (1U << btLog) - 1;
+    size_t commonLengthSmaller=0, commonLengthLarger=0;
+    const BYTE* const dictBase = ms->window.dictBase;
+    U32 const dictLimit = ms->window.dictLimit;
+    const BYTE* const dictEnd = dictBase + dictLimit;
+    const BYTE* const prefixStart = base + dictLimit;
+    U32 const btLow = (btMask >= curr) ? 0 : curr - btMask;
+    U32 const windowLow = ZSTD_getLowestMatchIndex(ms, curr, cParams->windowLog);
+    U32 const matchLow = windowLow ? windowLow : 1;
+    U32* smallerPtr = bt + 2*(curr&btMask);
+    U32* largerPtr  = bt + 2*(curr&btMask) + 1;
+    U32 matchEndIdx = curr+8+1;   /* farthest referenced position of any match => detects repetitive patterns */
+    U32 dummy32;   /* to be nullified at the end */
+    U32 mnum = 0;
+    U32 nbCompares = 1U << cParams->searchLog;
+
+    const ZSTD_matchState_t* dms    = dictMode == ZSTD_dictMatchState ? ms->dictMatchState : NULL;
+    const ZSTD_compressionParameters* const dmsCParams =
+                                      dictMode == ZSTD_dictMatchState ? &dms->cParams : NULL;
+    const BYTE* const dmsBase       = dictMode == ZSTD_dictMatchState ? dms->window.base : NULL;
+    const BYTE* const dmsEnd        = dictMode == ZSTD_dictMatchState ? dms->window.nextSrc : NULL;
+    U32         const dmsHighLimit  = dictMode == ZSTD_dictMatchState ? (U32)(dmsEnd - dmsBase) : 0;
+    U32         const dmsLowLimit   = dictMode == ZSTD_dictMatchState ? dms->window.lowLimit : 0;
+    U32         const dmsIndexDelta = dictMode == ZSTD_dictMatchState ? windowLow - dmsHighLimit : 0;
+    U32         const dmsHashLog    = dictMode == ZSTD_dictMatchState ? dmsCParams->hashLog : hashLog;
+    U32         const dmsBtLog      = dictMode == ZSTD_dictMatchState ? dmsCParams->chainLog - 1 : btLog;
+    U32         const dmsBtMask     = dictMode == ZSTD_dictMatchState ? (1U << dmsBtLog) - 1 : 0;
+    U32         const dmsBtLow      = dictMode == ZSTD_dictMatchState && dmsBtMask < dmsHighLimit - dmsLowLimit ? dmsHighLimit - dmsBtMask : dmsLowLimit;
+
+    size_t bestLength = lengthToBeat-1;
+    DEBUGLOG(8, "ZSTD_insertBtAndGetAllMatches: current=%u", curr);
+
+    /* check repCode */
+    assert(ll0 <= 1);   /* necessarily 1 or 0 */
+    {   U32 const lastR = ZSTD_REP_NUM + ll0;
+        U32 repCode;
+        for (repCode = ll0; repCode < lastR; repCode++) {
+            U32 const repOffset = (repCode==ZSTD_REP_NUM) ? (rep[0] - 1) : rep[repCode];
+            U32 const repIndex = curr - repOffset;
+            U32 repLen = 0;
+            assert(curr >= dictLimit);
+            if (repOffset-1 /* intentional overflow, discards 0 and -1 */ < curr-dictLimit) {  /* equivalent to `curr > repIndex >= dictLimit` */
+                /* We must validate the repcode offset because when we're using a dictionary the
+                 * valid offset range shrinks when the dictionary goes out of bounds.
+                 */
+                if ((repIndex >= windowLow) & (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(ip - repOffset, minMatch))) {
+                    repLen = (U32)ZSTD_count(ip+minMatch, ip+minMatch-repOffset, iLimit) + minMatch;
+                }
+            } else {  /* repIndex < dictLimit || repIndex >= curr */
+                const BYTE* const repMatch = dictMode == ZSTD_dictMatchState ?
+                                             dmsBase + repIndex - dmsIndexDelta :
+                                             dictBase + repIndex;
+                assert(curr >= windowLow);
+                if ( dictMode == ZSTD_extDict
+                  && ( ((repOffset-1) /*intentional overflow*/ < curr - windowLow)  /* equivalent to `curr > repIndex >= windowLow` */
+                     & (((U32)((dictLimit-1) - repIndex) >= 3) ) /* intentional overflow : do not test positions overlapping 2 memory segments */)
+                  && (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch)) ) {
+                    repLen = (U32)ZSTD_count_2segments(ip+minMatch, repMatch+minMatch, iLimit, dictEnd, prefixStart) + minMatch;
+                }
+                if (dictMode == ZSTD_dictMatchState
+                  && ( ((repOffset-1) /*intentional overflow*/ < curr - (dmsLowLimit + dmsIndexDelta))  /* equivalent to `curr > repIndex >= dmsLowLimit` */
+                     & ((U32)((dictLimit-1) - repIndex) >= 3) ) /* intentional overflow : do not test positions overlapping 2 memory segments */
+                  && (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch)) ) {
+                    repLen = (U32)ZSTD_count_2segments(ip+minMatch, repMatch+minMatch, iLimit, dmsEnd, prefixStart) + minMatch;
+            }   }
+            /* save longer solution */
+            if (repLen > bestLength) {
+                DEBUGLOG(8, "found repCode %u (ll0:%u, offset:%u) of length %u",
+                            repCode, ll0, repOffset, repLen);
+                bestLength = repLen;
+                matches[mnum].off = repCode - ll0;
+                matches[mnum].len = (U32)repLen;
+                mnum++;
+                if ( (repLen > sufficient_len)
+                   | (ip+repLen == iLimit) ) {  /* best possible */
+                    return mnum;
+    }   }   }   }
+
+    /* HC3 match finder */
+    if ((mls == 3) /*static*/ && (bestLength < mls)) {
+        U32 const matchIndex3 = ZSTD_insertAndFindFirstIndexHash3(ms, nextToUpdate3, ip);
+        if ((matchIndex3 >= matchLow)
+          & (curr - matchIndex3 < (1<<18)) /*heuristic : longer distance likely too expensive*/ ) {
+            size_t mlen;
+            if ((dictMode == ZSTD_noDict) /*static*/ || (dictMode == ZSTD_dictMatchState) /*static*/ || (matchIndex3 >= dictLimit)) {
+                const BYTE* const match = base + matchIndex3;
+                mlen = ZSTD_count(ip, match, iLimit);
+            } else {
+                const BYTE* const match = dictBase + matchIndex3;
+                mlen = ZSTD_count_2segments(ip, match, iLimit, dictEnd, prefixStart);
+            }
+
+            /* save best solution */
+            if (mlen >= mls /* == 3 > bestLength */) {
+                DEBUGLOG(8, "found small match with hlog3, of length %u",
+                            (U32)mlen);
+                bestLength = mlen;
+                assert(curr > matchIndex3);
+                assert(mnum==0);  /* no prior solution */
+                matches[0].off = (curr - matchIndex3) + ZSTD_REP_MOVE;
+                matches[0].len = (U32)mlen;
+                mnum = 1;
+                if ( (mlen > sufficient_len) |
+                     (ip+mlen == iLimit) ) {  /* best possible length */
+                    ms->nextToUpdate = curr+1;  /* skip insertion */
+                    return 1;
+        }   }   }
+        /* no dictMatchState lookup: dicts don't have a populated HC3 table */
+    }
+
+    hashTable[h] = curr;   /* Update Hash Table */
+
+    for (; nbCompares && (matchIndex >= matchLow); --nbCompares) {
+        U32* const nextPtr = bt + 2*(matchIndex & btMask);
+        const BYTE* match;
+        size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger);   /* guaranteed minimum nb of common bytes */
+        assert(curr > matchIndex);
+
+        if ((dictMode == ZSTD_noDict) || (dictMode == ZSTD_dictMatchState) || (matchIndex+matchLength >= dictLimit)) {
+            assert(matchIndex+matchLength >= dictLimit);  /* ensure the condition is correct when !extDict */
+            match = base + matchIndex;
+            if (matchIndex >= dictLimit) assert(memcmp(match, ip, matchLength) == 0);  /* ensure early section of match is equal as expected */
+            matchLength += ZSTD_count(ip+matchLength, match+matchLength, iLimit);
+        } else {
+            match = dictBase + matchIndex;
+            assert(memcmp(match, ip, matchLength) == 0);  /* ensure early section of match is equal as expected */
+            matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iLimit, dictEnd, prefixStart);
+            if (matchIndex+matchLength >= dictLimit)
+                match = base + matchIndex;   /* prepare for match[matchLength] read */
+        }
+
+        if (matchLength > bestLength) {
+            DEBUGLOG(8, "found match of length %u at distance %u (offCode=%u)",
+                    (U32)matchLength, curr - matchIndex, curr - matchIndex + ZSTD_REP_MOVE);
+            assert(matchEndIdx > matchIndex);
+            if (matchLength > matchEndIdx - matchIndex)
+                matchEndIdx = matchIndex + (U32)matchLength;
+            bestLength = matchLength;
+            matches[mnum].off = (curr - matchIndex) + ZSTD_REP_MOVE;
+            matches[mnum].len = (U32)matchLength;
+            mnum++;
+            if ( (matchLength > ZSTD_OPT_NUM)
+               | (ip+matchLength == iLimit) /* equal : no way to know if inf or sup */) {
+                if (dictMode == ZSTD_dictMatchState) nbCompares = 0; /* break should also skip searching dms */
+                break; /* drop, to preserve bt consistency (miss a little bit of compression) */
+            }
+        }
+
+        if (match[matchLength] < ip[matchLength]) {
+            /* match smaller than current */
+            *smallerPtr = matchIndex;             /* update smaller idx */
+            commonLengthSmaller = matchLength;    /* all smaller will now have at least this guaranteed common length */
+            if (matchIndex <= btLow) { smallerPtr=&dummy32; break; }   /* beyond tree size, stop the search */
+            smallerPtr = nextPtr+1;               /* new candidate => larger than match, which was smaller than current */
+            matchIndex = nextPtr[1];              /* new matchIndex, larger than previous, closer to current */
+        } else {
+            *largerPtr = matchIndex;
+            commonLengthLarger = matchLength;
+            if (matchIndex <= btLow) { largerPtr=&dummy32; break; }   /* beyond tree size, stop the search */
+            largerPtr = nextPtr;
+            matchIndex = nextPtr[0];
+    }   }
+
+    *smallerPtr = *largerPtr = 0;
+
+    assert(nbCompares <= (1U << ZSTD_SEARCHLOG_MAX)); /* Check we haven't underflowed. */
+    if (dictMode == ZSTD_dictMatchState && nbCompares) {
+        size_t const dmsH = ZSTD_hashPtr(ip, dmsHashLog, mls);
+        U32 dictMatchIndex = dms->hashTable[dmsH];
+        const U32* const dmsBt = dms->chainTable;
+        commonLengthSmaller = commonLengthLarger = 0;
+        for (; nbCompares && (dictMatchIndex > dmsLowLimit); --nbCompares) {
+            const U32* const nextPtr = dmsBt + 2*(dictMatchIndex & dmsBtMask);
+            size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger);   /* guaranteed minimum nb of common bytes */
+            const BYTE* match = dmsBase + dictMatchIndex;
+            matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iLimit, dmsEnd, prefixStart);
+            if (dictMatchIndex+matchLength >= dmsHighLimit)
+                match = base + dictMatchIndex + dmsIndexDelta;   /* to prepare for next usage of match[matchLength] */
+
+            if (matchLength > bestLength) {
+                matchIndex = dictMatchIndex + dmsIndexDelta;
+                DEBUGLOG(8, "found dms match of length %u at distance %u (offCode=%u)",
+                        (U32)matchLength, curr - matchIndex, curr - matchIndex + ZSTD_REP_MOVE);
+                if (matchLength > matchEndIdx - matchIndex)
+                    matchEndIdx = matchIndex + (U32)matchLength;
+                bestLength = matchLength;
+                matches[mnum].off = (curr - matchIndex) + ZSTD_REP_MOVE;
+                matches[mnum].len = (U32)matchLength;
+                mnum++;
+                if ( (matchLength > ZSTD_OPT_NUM)
+                   | (ip+matchLength == iLimit) /* equal : no way to know if inf or sup */) {
+                    break;   /* drop, to guarantee consistency (miss a little bit of compression) */
+                }
+            }
+
+            if (dictMatchIndex <= dmsBtLow) { break; }   /* beyond tree size, stop the search */
+            if (match[matchLength] < ip[matchLength]) {
+                commonLengthSmaller = matchLength;    /* all smaller will now have at least this guaranteed common length */
+                dictMatchIndex = nextPtr[1];              /* new matchIndex larger than previous (closer to current) */
+            } else {
+                /* match is larger than current */
+                commonLengthLarger = matchLength;
+                dictMatchIndex = nextPtr[0];
+            }
+        }
+    }
+
+    assert(matchEndIdx > curr+8);
+    ms->nextToUpdate = matchEndIdx - 8;  /* skip repetitive patterns */
+    return mnum;
+}
+
+
+FORCE_INLINE_TEMPLATE U32 ZSTD_BtGetAllMatches (
+                        ZSTD_match_t* matches,   /* store result (match found, increasing size) in this table */
+                        ZSTD_matchState_t* ms,
+                        U32* nextToUpdate3,
+                        const BYTE* ip, const BYTE* const iHighLimit, const ZSTD_dictMode_e dictMode,
+                        const U32 rep[ZSTD_REP_NUM],
+                        U32 const ll0,
+                        U32 const lengthToBeat)
+{
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+    U32 const matchLengthSearch = cParams->minMatch;
+    DEBUGLOG(8, "ZSTD_BtGetAllMatches");
+    if (ip < ms->window.base + ms->nextToUpdate) return 0;   /* skipped area */
+    ZSTD_updateTree_internal(ms, ip, iHighLimit, matchLengthSearch, dictMode);
+    switch(matchLengthSearch)
+    {
+    case 3 : return ZSTD_insertBtAndGetAllMatches(matches, ms, nextToUpdate3, ip, iHighLimit, dictMode, rep, ll0, lengthToBeat, 3);
+    default :
+    case 4 : return ZSTD_insertBtAndGetAllMatches(matches, ms, nextToUpdate3, ip, iHighLimit, dictMode, rep, ll0, lengthToBeat, 4);
+    case 5 : return ZSTD_insertBtAndGetAllMatches(matches, ms, nextToUpdate3, ip, iHighLimit, dictMode, rep, ll0, lengthToBeat, 5);
+    case 7 :
+    case 6 : return ZSTD_insertBtAndGetAllMatches(matches, ms, nextToUpdate3, ip, iHighLimit, dictMode, rep, ll0, lengthToBeat, 6);
+    }
+}
+
+/* ***********************
+*  LDM helper functions  *
+*************************/
+
+/* Struct containing info needed to make decision about ldm inclusion */
+typedef struct {
+    rawSeqStore_t seqStore;         /* External match candidates store for this block */
+    U32 startPosInBlock;            /* Start position of the current match candidate */
+    U32 endPosInBlock;              /* End position of the current match candidate */
+    U32 offset;                     /* Offset of the match candidate */
+} ZSTD_optLdm_t;
+
+/* ZSTD_optLdm_skipRawSeqStoreBytes():
+ * Moves forward in rawSeqStore by nbBytes, which will update the fields 'pos' and 'posInSequence'.
+ */
+static void ZSTD_optLdm_skipRawSeqStoreBytes(rawSeqStore_t* rawSeqStore, size_t nbBytes) {
+    U32 currPos = (U32)(rawSeqStore->posInSequence + nbBytes);
+    while (currPos && rawSeqStore->pos < rawSeqStore->size) {
+        rawSeq currSeq = rawSeqStore->seq[rawSeqStore->pos];
+        if (currPos >= currSeq.litLength + currSeq.matchLength) {
+            currPos -= currSeq.litLength + currSeq.matchLength;
+            rawSeqStore->pos++;
+        } else {
+            rawSeqStore->posInSequence = currPos;
+            break;
+        }
+    }
+    if (currPos == 0 || rawSeqStore->pos == rawSeqStore->size) {
+        rawSeqStore->posInSequence = 0;
+    }
+}
+
+/* ZSTD_opt_getNextMatchAndUpdateSeqStore():
+ * Calculates the beginning and end of the next match in the current block.
+ * Updates 'pos' and 'posInSequence' of the ldmSeqStore.
+ */
+static void ZSTD_opt_getNextMatchAndUpdateSeqStore(ZSTD_optLdm_t* optLdm, U32 currPosInBlock,
+                                                   U32 blockBytesRemaining) {
+    rawSeq currSeq;
+    U32 currBlockEndPos;
+    U32 literalsBytesRemaining;
+    U32 matchBytesRemaining;
+
+    /* Setting match end position to MAX to ensure we never use an LDM during this block */
+    if (optLdm->seqStore.size == 0 || optLdm->seqStore.pos >= optLdm->seqStore.size) {
+        optLdm->startPosInBlock = UINT_MAX;
+        optLdm->endPosInBlock = UINT_MAX;
+        return;
+    }
+    /* Calculate appropriate bytes left in matchLength and litLength after adjusting
+       based on ldmSeqStore->posInSequence */
+    currSeq = optLdm->seqStore.seq[optLdm->seqStore.pos];
+    assert(optLdm->seqStore.posInSequence <= currSeq.litLength + currSeq.matchLength);
+    currBlockEndPos = currPosInBlock + blockBytesRemaining;
+    literalsBytesRemaining = (optLdm->seqStore.posInSequence < currSeq.litLength) ?
+            currSeq.litLength - (U32)optLdm->seqStore.posInSequence :
+            0;
+    matchBytesRemaining = (literalsBytesRemaining == 0) ?
+            currSeq.matchLength - ((U32)optLdm->seqStore.posInSequence - currSeq.litLength) :
+            currSeq.matchLength;
+
+    /* If there are more literal bytes than bytes remaining in block, no ldm is possible */
+    if (literalsBytesRemaining >= blockBytesRemaining) {
+        optLdm->startPosInBlock = UINT_MAX;
+        optLdm->endPosInBlock = UINT_MAX;
+        ZSTD_optLdm_skipRawSeqStoreBytes(&optLdm->seqStore, blockBytesRemaining);
+        return;
+    }
+
+    /* Matches may be < MINMATCH by this process. In that case, we will reject them
+       when we are deciding whether or not to add the ldm */
+    optLdm->startPosInBlock = currPosInBlock + literalsBytesRemaining;
+    optLdm->endPosInBlock = optLdm->startPosInBlock + matchBytesRemaining;
+    optLdm->offset = currSeq.offset;
+
+    if (optLdm->endPosInBlock > currBlockEndPos) {
+        /* Match ends after the block ends, we can't use the whole match */
+        optLdm->endPosInBlock = currBlockEndPos;
+        ZSTD_optLdm_skipRawSeqStoreBytes(&optLdm->seqStore, currBlockEndPos - currPosInBlock);
+    } else {
+        /* Consume nb of bytes equal to size of sequence left */
+        ZSTD_optLdm_skipRawSeqStoreBytes(&optLdm->seqStore, literalsBytesRemaining + matchBytesRemaining);
+    }
+}
+
+/* ZSTD_optLdm_maybeAddMatch():
+ * Adds a match if it's long enough, based on it's 'matchStartPosInBlock'
+ * and 'matchEndPosInBlock', into 'matches'. Maintains the correct ordering of 'matches'
+ */
+static void ZSTD_optLdm_maybeAddMatch(ZSTD_match_t* matches, U32* nbMatches,
+                                      ZSTD_optLdm_t* optLdm, U32 currPosInBlock) {
+    U32 posDiff = currPosInBlock - optLdm->startPosInBlock;
+    /* Note: ZSTD_match_t actually contains offCode and matchLength (before subtracting MINMATCH) */
+    U32 candidateMatchLength = optLdm->endPosInBlock - optLdm->startPosInBlock - posDiff;
+    U32 candidateOffCode = optLdm->offset + ZSTD_REP_MOVE;
+
+    /* Ensure that current block position is not outside of the match */
+    if (currPosInBlock < optLdm->startPosInBlock
+      || currPosInBlock >= optLdm->endPosInBlock
+      || candidateMatchLength < MINMATCH) {
+        return;
+    }
+
+    if (*nbMatches == 0 || ((candidateMatchLength > matches[*nbMatches-1].len) && *nbMatches < ZSTD_OPT_NUM)) {
+        DEBUGLOG(6, "ZSTD_optLdm_maybeAddMatch(): Adding ldm candidate match (offCode: %u matchLength %u) at block position=%u",
+                 candidateOffCode, candidateMatchLength, currPosInBlock);
+        matches[*nbMatches].len = candidateMatchLength;
+        matches[*nbMatches].off = candidateOffCode;
+        (*nbMatches)++;
+    }
+}
+
+/* ZSTD_optLdm_processMatchCandidate():
+ * Wrapper function to update ldm seq store and call ldm functions as necessary.
+ */
+static void ZSTD_optLdm_processMatchCandidate(ZSTD_optLdm_t* optLdm, ZSTD_match_t* matches, U32* nbMatches,
+                                              U32 currPosInBlock, U32 remainingBytes) {
+    if (optLdm->seqStore.size == 0 || optLdm->seqStore.pos >= optLdm->seqStore.size) {
+        return;
+    }
+
+    if (currPosInBlock >= optLdm->endPosInBlock) {
+        if (currPosInBlock > optLdm->endPosInBlock) {
+            /* The position at which ZSTD_optLdm_processMatchCandidate() is called is not necessarily
+             * at the end of a match from the ldm seq store, and will often be some bytes
+             * over beyond matchEndPosInBlock. As such, we need to correct for these "overshoots"
+             */
+            U32 posOvershoot = currPosInBlock - optLdm->endPosInBlock;
+            ZSTD_optLdm_skipRawSeqStoreBytes(&optLdm->seqStore, posOvershoot);
+        } 
+        ZSTD_opt_getNextMatchAndUpdateSeqStore(optLdm, currPosInBlock, remainingBytes);
+    }
+    ZSTD_optLdm_maybeAddMatch(matches, nbMatches, optLdm, currPosInBlock);
+}
+
+/*-*******************************
+*  Optimal parser
+*********************************/
+
+
+static U32 ZSTD_totalLen(ZSTD_optimal_t sol)
+{
+    return sol.litlen + sol.mlen;
+}
+
+#if 0 /* debug */
+
+static void
+listStats(const U32* table, int lastEltID)
+{
+    int const nbElts = lastEltID + 1;
+    int enb;
+    for (enb=0; enb < nbElts; enb++) {
+        (void)table;
+        /* RAWLOG(2, "%3i:%3i,  ", enb, table[enb]); */
+        RAWLOG(2, "%4i,", table[enb]);
+    }
+    RAWLOG(2, " \n");
+}
+
+#endif
+
+FORCE_INLINE_TEMPLATE size_t
+ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
+                               seqStore_t* seqStore,
+                               U32 rep[ZSTD_REP_NUM],
+                         const void* src, size_t srcSize,
+                         const int optLevel,
+                         const ZSTD_dictMode_e dictMode)
+{
+    optState_t* const optStatePtr = &ms->opt;
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* ip = istart;
+    const BYTE* anchor = istart;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* const ilimit = iend - 8;
+    const BYTE* const base = ms->window.base;
+    const BYTE* const prefixStart = base + ms->window.dictLimit;
+    const ZSTD_compressionParameters* const cParams = &ms->cParams;
+
+    U32 const sufficient_len = MIN(cParams->targetLength, ZSTD_OPT_NUM -1);
+    U32 const minMatch = (cParams->minMatch == 3) ? 3 : 4;
+    U32 nextToUpdate3 = ms->nextToUpdate;
+
+    ZSTD_optimal_t* const opt = optStatePtr->priceTable;
+    ZSTD_match_t* const matches = optStatePtr->matchTable;
+    ZSTD_optimal_t lastSequence;
+    ZSTD_optLdm_t optLdm;
+
+    optLdm.seqStore = ms->ldmSeqStore ? *ms->ldmSeqStore : kNullRawSeqStore;
+    optLdm.endPosInBlock = optLdm.startPosInBlock = optLdm.offset = 0;
+    ZSTD_opt_getNextMatchAndUpdateSeqStore(&optLdm, (U32)(ip-istart), (U32)(iend-ip));
+
+    /* init */
+    DEBUGLOG(5, "ZSTD_compressBlock_opt_generic: current=%u, prefix=%u, nextToUpdate=%u",
+                (U32)(ip - base), ms->window.dictLimit, ms->nextToUpdate);
+    assert(optLevel <= 2);
+    ZSTD_rescaleFreqs(optStatePtr, (const BYTE*)src, srcSize, optLevel);
+    ip += (ip==prefixStart);
+
+    /* Match Loop */
+    while (ip < ilimit) {
+        U32 cur, last_pos = 0;
+
+        /* find first match */
+        {   U32 const litlen = (U32)(ip - anchor);
+            U32 const ll0 = !litlen;
+            U32 nbMatches = ZSTD_BtGetAllMatches(matches, ms, &nextToUpdate3, ip, iend, dictMode, rep, ll0, minMatch);
+            ZSTD_optLdm_processMatchCandidate(&optLdm, matches, &nbMatches,
+                                              (U32)(ip-istart), (U32)(iend - ip));
+            if (!nbMatches) { ip++; continue; }
+
+            /* initialize opt[0] */
+            { U32 i ; for (i=0; i<ZSTD_REP_NUM; i++) opt[0].rep[i] = rep[i]; }
+            opt[0].mlen = 0;  /* means is_a_literal */
+            opt[0].litlen = litlen;
+            /* We don't need to include the actual price of the literals because
+             * it is static for the duration of the forward pass, and is included
+             * in every price. We include the literal length to avoid negative
+             * prices when we subtract the previous literal length.
+             */
+            opt[0].price = ZSTD_litLengthPrice(litlen, optStatePtr, optLevel);
+
+            /* large match -> immediate encoding */
+            {   U32 const maxML = matches[nbMatches-1].len;
+                U32 const maxOffset = matches[nbMatches-1].off;
+                DEBUGLOG(6, "found %u matches of maxLength=%u and maxOffCode=%u at cPos=%u => start new series",
+                            nbMatches, maxML, maxOffset, (U32)(ip-prefixStart));
+
+                if (maxML > sufficient_len) {
+                    lastSequence.litlen = litlen;
+                    lastSequence.mlen = maxML;
+                    lastSequence.off = maxOffset;
+                    DEBUGLOG(6, "large match (%u>%u), immediate encoding",
+                                maxML, sufficient_len);
+                    cur = 0;
+                    last_pos = ZSTD_totalLen(lastSequence);
+                    goto _shortestPath;
+            }   }
+
+            /* set prices for first matches starting position == 0 */
+            {   U32 const literalsPrice = opt[0].price + ZSTD_litLengthPrice(0, optStatePtr, optLevel);
+                U32 pos;
+                U32 matchNb;
+                for (pos = 1; pos < minMatch; pos++) {
+                    opt[pos].price = ZSTD_MAX_PRICE;   /* mlen, litlen and price will be fixed during forward scanning */
+                }
+                for (matchNb = 0; matchNb < nbMatches; matchNb++) {
+                    U32 const offset = matches[matchNb].off;
+                    U32 const end = matches[matchNb].len;
+                    for ( ; pos <= end ; pos++ ) {
+                        U32 const matchPrice = ZSTD_getMatchPrice(offset, pos, optStatePtr, optLevel);
+                        U32 const sequencePrice = literalsPrice + matchPrice;
+                        DEBUGLOG(7, "rPos:%u => set initial price : %.2f",
+                                    pos, ZSTD_fCost(sequencePrice));
+                        opt[pos].mlen = pos;
+                        opt[pos].off = offset;
+                        opt[pos].litlen = litlen;
+                        opt[pos].price = sequencePrice;
+                }   }
+                last_pos = pos-1;
+            }
+        }
+
+        /* check further positions */
+        for (cur = 1; cur <= last_pos; cur++) {
+            const BYTE* const inr = ip + cur;
+            assert(cur < ZSTD_OPT_NUM);
+            DEBUGLOG(7, "cPos:%zi==rPos:%u", inr-istart, cur)
+
+            /* Fix current position with one literal if cheaper */
+            {   U32 const litlen = (opt[cur-1].mlen == 0) ? opt[cur-1].litlen + 1 : 1;
+                int const price = opt[cur-1].price
+                                + ZSTD_rawLiteralsCost(ip+cur-1, 1, optStatePtr, optLevel)
+                                + ZSTD_litLengthPrice(litlen, optStatePtr, optLevel)
+                                - ZSTD_litLengthPrice(litlen-1, optStatePtr, optLevel);
+                assert(price < 1000000000); /* overflow check */
+                if (price <= opt[cur].price) {
+                    DEBUGLOG(7, "cPos:%zi==rPos:%u : better price (%.2f<=%.2f) using literal (ll==%u) (hist:%u,%u,%u)",
+                                inr-istart, cur, ZSTD_fCost(price), ZSTD_fCost(opt[cur].price), litlen,
+                                opt[cur-1].rep[0], opt[cur-1].rep[1], opt[cur-1].rep[2]);
+                    opt[cur].mlen = 0;
+                    opt[cur].off = 0;
+                    opt[cur].litlen = litlen;
+                    opt[cur].price = price;
+                } else {
+                    DEBUGLOG(7, "cPos:%zi==rPos:%u : literal would cost more (%.2f>%.2f) (hist:%u,%u,%u)",
+                                inr-istart, cur, ZSTD_fCost(price), ZSTD_fCost(opt[cur].price),
+                                opt[cur].rep[0], opt[cur].rep[1], opt[cur].rep[2]);
+                }
+            }
+
+            /* Set the repcodes of the current position. We must do it here
+             * because we rely on the repcodes of the 2nd to last sequence being
+             * correct to set the next chunks repcodes during the backward
+             * traversal.
+             */
+            ZSTD_STATIC_ASSERT(sizeof(opt[cur].rep) == sizeof(repcodes_t));
+            assert(cur >= opt[cur].mlen);
+            if (opt[cur].mlen != 0) {
+                U32 const prev = cur - opt[cur].mlen;
+                repcodes_t newReps = ZSTD_updateRep(opt[prev].rep, opt[cur].off, opt[cur].litlen==0);
+                ZSTD_memcpy(opt[cur].rep, &newReps, sizeof(repcodes_t));
+            } else {
+                ZSTD_memcpy(opt[cur].rep, opt[cur - 1].rep, sizeof(repcodes_t));
+            }
+
+            /* last match must start at a minimum distance of 8 from oend */
+            if (inr > ilimit) continue;
+
+            if (cur == last_pos) break;
+
+            if ( (optLevel==0) /*static_test*/
+              && (opt[cur+1].price <= opt[cur].price + (BITCOST_MULTIPLIER/2)) ) {
+                DEBUGLOG(7, "move to next rPos:%u : price is <=", cur+1);
+                continue;  /* skip unpromising positions; about ~+6% speed, -0.01 ratio */
+            }
+
+            {   U32 const ll0 = (opt[cur].mlen != 0);
+                U32 const litlen = (opt[cur].mlen == 0) ? opt[cur].litlen : 0;
+                U32 const previousPrice = opt[cur].price;
+                U32 const basePrice = previousPrice + ZSTD_litLengthPrice(0, optStatePtr, optLevel);
+                U32 nbMatches = ZSTD_BtGetAllMatches(matches, ms, &nextToUpdate3, inr, iend, dictMode, opt[cur].rep, ll0, minMatch);
+                U32 matchNb;
+
+                ZSTD_optLdm_processMatchCandidate(&optLdm, matches, &nbMatches,
+                                                  (U32)(inr-istart), (U32)(iend-inr));
+
+                if (!nbMatches) {
+                    DEBUGLOG(7, "rPos:%u : no match found", cur);
+                    continue;
+                }
+
+                {   U32 const maxML = matches[nbMatches-1].len;
+                    DEBUGLOG(7, "cPos:%zi==rPos:%u, found %u matches, of maxLength=%u",
+                                inr-istart, cur, nbMatches, maxML);
+
+                    if ( (maxML > sufficient_len)
+                      || (cur + maxML >= ZSTD_OPT_NUM) ) {
+                        lastSequence.mlen = maxML;
+                        lastSequence.off = matches[nbMatches-1].off;
+                        lastSequence.litlen = litlen;
+                        cur -= (opt[cur].mlen==0) ? opt[cur].litlen : 0;  /* last sequence is actually only literals, fix cur to last match - note : may underflow, in which case, it's first sequence, and it's okay */
+                        last_pos = cur + ZSTD_totalLen(lastSequence);
+                        if (cur > ZSTD_OPT_NUM) cur = 0;   /* underflow => first match */
+                        goto _shortestPath;
+                }   }
+
+                /* set prices using matches found at position == cur */
+                for (matchNb = 0; matchNb < nbMatches; matchNb++) {
+                    U32 const offset = matches[matchNb].off;
+                    U32 const lastML = matches[matchNb].len;
+                    U32 const startML = (matchNb>0) ? matches[matchNb-1].len+1 : minMatch;
+                    U32 mlen;
+
+                    DEBUGLOG(7, "testing match %u => offCode=%4u, mlen=%2u, llen=%2u",
+                                matchNb, matches[matchNb].off, lastML, litlen);
+
+                    for (mlen = lastML; mlen >= startML; mlen--) {  /* scan downward */
+                        U32 const pos = cur + mlen;
+                        int const price = basePrice + ZSTD_getMatchPrice(offset, mlen, optStatePtr, optLevel);
+
+                        if ((pos > last_pos) || (price < opt[pos].price)) {
+                            DEBUGLOG(7, "rPos:%u (ml=%2u) => new better price (%.2f<%.2f)",
+                                        pos, mlen, ZSTD_fCost(price), ZSTD_fCost(opt[pos].price));
+                            while (last_pos < pos) { opt[last_pos+1].price = ZSTD_MAX_PRICE; last_pos++; }   /* fill empty positions */
+                            opt[pos].mlen = mlen;
+                            opt[pos].off = offset;
+                            opt[pos].litlen = litlen;
+                            opt[pos].price = price;
+                        } else {
+                            DEBUGLOG(7, "rPos:%u (ml=%2u) => new price is worse (%.2f>=%.2f)",
+                                        pos, mlen, ZSTD_fCost(price), ZSTD_fCost(opt[pos].price));
+                            if (optLevel==0) break;  /* early update abort; gets ~+10% speed for about -0.01 ratio loss */
+                        }
+            }   }   }
+        }  /* for (cur = 1; cur <= last_pos; cur++) */
+
+        lastSequence = opt[last_pos];
+        cur = last_pos > ZSTD_totalLen(lastSequence) ? last_pos - ZSTD_totalLen(lastSequence) : 0;  /* single sequence, and it starts before `ip` */
+        assert(cur < ZSTD_OPT_NUM);  /* control overflow*/
+
+_shortestPath:   /* cur, last_pos, best_mlen, best_off have to be set */
+        assert(opt[0].mlen == 0);
+
+        /* Set the next chunk's repcodes based on the repcodes of the beginning
+         * of the last match, and the last sequence. This avoids us having to
+         * update them while traversing the sequences.
+         */
+        if (lastSequence.mlen != 0) {
+            repcodes_t reps = ZSTD_updateRep(opt[cur].rep, lastSequence.off, lastSequence.litlen==0);
+            ZSTD_memcpy(rep, &reps, sizeof(reps));
+        } else {
+            ZSTD_memcpy(rep, opt[cur].rep, sizeof(repcodes_t));
+        }
+
+        {   U32 const storeEnd = cur + 1;
+            U32 storeStart = storeEnd;
+            U32 seqPos = cur;
+
+            DEBUGLOG(6, "start reverse traversal (last_pos:%u, cur:%u)",
+                        last_pos, cur); (void)last_pos;
+            assert(storeEnd < ZSTD_OPT_NUM);
+            DEBUGLOG(6, "last sequence copied into pos=%u (llen=%u,mlen=%u,ofc=%u)",
+                        storeEnd, lastSequence.litlen, lastSequence.mlen, lastSequence.off);
+            opt[storeEnd] = lastSequence;
+            while (seqPos > 0) {
+                U32 const backDist = ZSTD_totalLen(opt[seqPos]);
+                storeStart--;
+                DEBUGLOG(6, "sequence from rPos=%u copied into pos=%u (llen=%u,mlen=%u,ofc=%u)",
+                            seqPos, storeStart, opt[seqPos].litlen, opt[seqPos].mlen, opt[seqPos].off);
+                opt[storeStart] = opt[seqPos];
+                seqPos = (seqPos > backDist) ? seqPos - backDist : 0;
+            }
+
+            /* save sequences */
+            DEBUGLOG(6, "sending selected sequences into seqStore")
+            {   U32 storePos;
+                for (storePos=storeStart; storePos <= storeEnd; storePos++) {
+                    U32 const llen = opt[storePos].litlen;
+                    U32 const mlen = opt[storePos].mlen;
+                    U32 const offCode = opt[storePos].off;
+                    U32 const advance = llen + mlen;
+                    DEBUGLOG(6, "considering seq starting at %zi, llen=%u, mlen=%u",
+                                anchor - istart, (unsigned)llen, (unsigned)mlen);
+
+                    if (mlen==0) {  /* only literals => must be last "sequence", actually starting a new stream of sequences */
+                        assert(storePos == storeEnd);   /* must be last sequence */
+                        ip = anchor + llen;     /* last "sequence" is a bunch of literals => don't progress anchor */
+                        continue;   /* will finish */
+                    }
+
+                    assert(anchor + llen <= iend);
+                    ZSTD_updateStats(optStatePtr, llen, anchor, offCode, mlen);
+                    ZSTD_storeSeq(seqStore, llen, anchor, iend, offCode, mlen-MINMATCH);
+                    anchor += advance;
+                    ip = anchor;
+            }   }
+            ZSTD_setBasePrices(optStatePtr, optLevel);
+        }
+    }   /* while (ip < ilimit) */
+
+    /* Return the last literals size */
+    return (size_t)(iend - anchor);
+}
+
+
+size_t ZSTD_compressBlock_btopt(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    DEBUGLOG(5, "ZSTD_compressBlock_btopt");
+    return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 0 /*optLevel*/, ZSTD_noDict);
+}
+
+
+/* used in 2-pass strategy */
+static U32 ZSTD_upscaleStat(unsigned* table, U32 lastEltIndex, int bonus)
+{
+    U32 s, sum=0;
+    assert(ZSTD_FREQ_DIV+bonus >= 0);
+    for (s=0; s<lastEltIndex+1; s++) {
+        table[s] <<= ZSTD_FREQ_DIV+bonus;
+        table[s]--;
+        sum += table[s];
+    }
+    return sum;
+}
+
+/* used in 2-pass strategy */
+MEM_STATIC void ZSTD_upscaleStats(optState_t* optPtr)
+{
+    if (ZSTD_compressedLiterals(optPtr))
+        optPtr->litSum = ZSTD_upscaleStat(optPtr->litFreq, MaxLit, 0);
+    optPtr->litLengthSum = ZSTD_upscaleStat(optPtr->litLengthFreq, MaxLL, 0);
+    optPtr->matchLengthSum = ZSTD_upscaleStat(optPtr->matchLengthFreq, MaxML, 0);
+    optPtr->offCodeSum = ZSTD_upscaleStat(optPtr->offCodeFreq, MaxOff, 0);
+}
+
+/* ZSTD_initStats_ultra():
+ * make a first compression pass, just to seed stats with more accurate starting values.
+ * only works on first block, with no dictionary and no ldm.
+ * this function cannot error, hence its contract must be respected.
+ */
+static void
+ZSTD_initStats_ultra(ZSTD_matchState_t* ms,
+                     seqStore_t* seqStore,
+                     U32 rep[ZSTD_REP_NUM],
+               const void* src, size_t srcSize)
+{
+    U32 tmpRep[ZSTD_REP_NUM];  /* updated rep codes will sink here */
+    ZSTD_memcpy(tmpRep, rep, sizeof(tmpRep));
+
+    DEBUGLOG(4, "ZSTD_initStats_ultra (srcSize=%zu)", srcSize);
+    assert(ms->opt.litLengthSum == 0);    /* first block */
+    assert(seqStore->sequences == seqStore->sequencesStart);   /* no ldm */
+    assert(ms->window.dictLimit == ms->window.lowLimit);   /* no dictionary */
+    assert(ms->window.dictLimit - ms->nextToUpdate <= 1);  /* no prefix (note: intentional overflow, defined as 2-complement) */
+
+    ZSTD_compressBlock_opt_generic(ms, seqStore, tmpRep, src, srcSize, 2 /*optLevel*/, ZSTD_noDict);   /* generate stats into ms->opt*/
+
+    /* invalidate first scan from history */
+    ZSTD_resetSeqStore(seqStore);
+    ms->window.base -= srcSize;
+    ms->window.dictLimit += (U32)srcSize;
+    ms->window.lowLimit = ms->window.dictLimit;
+    ms->nextToUpdate = ms->window.dictLimit;
+
+    /* re-inforce weight of collected statistics */
+    ZSTD_upscaleStats(&ms->opt);
+}
+
+size_t ZSTD_compressBlock_btultra(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    DEBUGLOG(5, "ZSTD_compressBlock_btultra (srcSize=%zu)", srcSize);
+    return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /*optLevel*/, ZSTD_noDict);
+}
+
+size_t ZSTD_compressBlock_btultra2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    U32 const curr = (U32)((const BYTE*)src - ms->window.base);
+    DEBUGLOG(5, "ZSTD_compressBlock_btultra2 (srcSize=%zu)", srcSize);
+
+    /* 2-pass strategy:
+     * this strategy makes a first pass over first block to collect statistics
+     * and seed next round's statistics with it.
+     * After 1st pass, function forgets everything, and starts a new block.
+     * Consequently, this can only work if no data has been previously loaded in tables,
+     * aka, no dictionary, no prefix, no ldm preprocessing.
+     * The compression ratio gain is generally small (~0.5% on first block),
+     * the cost is 2x cpu time on first block. */
+    assert(srcSize <= ZSTD_BLOCKSIZE_MAX);
+    if ( (ms->opt.litLengthSum==0)   /* first block */
+      && (seqStore->sequences == seqStore->sequencesStart)  /* no ldm */
+      && (ms->window.dictLimit == ms->window.lowLimit)   /* no dictionary */
+      && (curr == ms->window.dictLimit)   /* start of frame, nothing already loaded nor skipped */
+      && (srcSize > ZSTD_PREDEF_THRESHOLD)
+      ) {
+        ZSTD_initStats_ultra(ms, seqStore, rep, src, srcSize);
+    }
+
+    return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /*optLevel*/, ZSTD_noDict);
+}
+
+size_t ZSTD_compressBlock_btopt_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 0 /*optLevel*/, ZSTD_dictMatchState);
+}
+
+size_t ZSTD_compressBlock_btultra_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /*optLevel*/, ZSTD_dictMatchState);
+}
+
+size_t ZSTD_compressBlock_btopt_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 0 /*optLevel*/, ZSTD_extDict);
+}
+
+size_t ZSTD_compressBlock_btultra_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /*optLevel*/, ZSTD_extDict);
+}
+
+/* note : no btultra2 variant for extDict nor dictMatchState,
+ * because btultra2 is not meant to work with dictionaries
+ * and is only specific for the first block (no prefix) */
diff --git a/lib/zstd/compress/zstd_opt.h b/lib/zstd/compress/zstd_opt.h
new file mode 100644
index 0000000..22b8628
--- /dev/null
+++ b/lib/zstd/compress/zstd_opt.h
@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef ZSTD_OPT_H
+#define ZSTD_OPT_H
+
+
+#include "zstd_compress_internal.h"
+
+/* used in ZSTD_loadDictionaryContent() */
+void ZSTD_updateTree(ZSTD_matchState_t* ms, const BYTE* ip, const BYTE* iend);
+
+size_t ZSTD_compressBlock_btopt(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_btultra(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_btultra2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+
+size_t ZSTD_compressBlock_btopt_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_btultra_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+size_t ZSTD_compressBlock_btopt_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_btultra_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+        /* note : no btultra2 variant for extDict nor dictMatchState,
+         * because btultra2 is not meant to work with dictionaries
+         * and is only specific for the first block (no prefix) */
+
+
+#endif /* ZSTD_OPT_H */
diff --git a/lib/zstd/decompress.c b/lib/zstd/decompress.c
deleted file mode 100644
index 66cd487..0000000
--- a/lib/zstd/decompress.c
+++ /dev/null
@@ -1,2531 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-/* ***************************************************************
-*  Tuning parameters
-*****************************************************************/
-/*!
-*  MAXWINDOWSIZE_DEFAULT :
-*  maximum window size accepted by DStream, by default.
-*  Frames requiring more memory will be rejected.
-*/
-#ifndef ZSTD_MAXWINDOWSIZE_DEFAULT
-#define ZSTD_MAXWINDOWSIZE_DEFAULT ((1 << ZSTD_WINDOWLOG_MAX) + 1) /* defined within zstd.h */
-#endif
-
-/*-*******************************************************
-*  Dependencies
-*********************************************************/
-#include "fse.h"
-#include "huf.h"
-#include "mem.h" /* low level memory routines */
-#include "zstd_internal.h"
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/string.h> /* memcpy, memmove, memset */
-
-#define ZSTD_PREFETCH(ptr) __builtin_prefetch(ptr, 0, 0)
-
-/*-*************************************
-*  Macros
-***************************************/
-#define ZSTD_isError ERR_isError /* for inlining */
-#define FSE_isError ERR_isError
-#define HUF_isError ERR_isError
-
-/*_*******************************************************
-*  Memory operations
-**********************************************************/
-static void ZSTD_copy4(void *dst, const void *src) { memcpy(dst, src, 4); }
-
-/*-*************************************************************
-*   Context management
-***************************************************************/
-typedef enum {
-	ZSTDds_getFrameHeaderSize,
-	ZSTDds_decodeFrameHeader,
-	ZSTDds_decodeBlockHeader,
-	ZSTDds_decompressBlock,
-	ZSTDds_decompressLastBlock,
-	ZSTDds_checkChecksum,
-	ZSTDds_decodeSkippableHeader,
-	ZSTDds_skipFrame
-} ZSTD_dStage;
-
-typedef struct {
-	FSE_DTable LLTable[FSE_DTABLE_SIZE_U32(LLFSELog)];
-	FSE_DTable OFTable[FSE_DTABLE_SIZE_U32(OffFSELog)];
-	FSE_DTable MLTable[FSE_DTABLE_SIZE_U32(MLFSELog)];
-	HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)]; /* can accommodate HUF_decompress4X */
-	U64 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32 / 2];
-	U32 rep[ZSTD_REP_NUM];
-} ZSTD_entropyTables_t;
-
-struct ZSTD_DCtx_s {
-	const FSE_DTable *LLTptr;
-	const FSE_DTable *MLTptr;
-	const FSE_DTable *OFTptr;
-	const HUF_DTable *HUFptr;
-	ZSTD_entropyTables_t entropy;
-	const void *previousDstEnd; /* detect continuity */
-	const void *base;	   /* start of curr segment */
-	const void *vBase;	  /* virtual start of previous segment if it was just before curr one */
-	const void *dictEnd;	/* end of previous segment */
-	size_t expected;
-	ZSTD_frameParams fParams;
-	blockType_e bType; /* used in ZSTD_decompressContinue(), to transfer blockType between header decoding and block decoding stages */
-	ZSTD_dStage stage;
-	U32 litEntropy;
-	U32 fseEntropy;
-	struct xxh64_state xxhState;
-	size_t headerSize;
-	U32 dictID;
-	const BYTE *litPtr;
-	ZSTD_customMem customMem;
-	size_t litSize;
-	size_t rleSize;
-	BYTE litBuffer[ZSTD_BLOCKSIZE_ABSOLUTEMAX + WILDCOPY_OVERLENGTH];
-	BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];
-}; /* typedef'd to ZSTD_DCtx within "zstd.h" */
-
-size_t ZSTD_DCtxWorkspaceBound(void) { return ZSTD_ALIGN(sizeof(ZSTD_stack)) + ZSTD_ALIGN(sizeof(ZSTD_DCtx)); }
-
-size_t ZSTD_decompressBegin(ZSTD_DCtx *dctx)
-{
-	dctx->expected = ZSTD_frameHeaderSize_prefix;
-	dctx->stage = ZSTDds_getFrameHeaderSize;
-	dctx->previousDstEnd = NULL;
-	dctx->base = NULL;
-	dctx->vBase = NULL;
-	dctx->dictEnd = NULL;
-	dctx->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001); /* cover both little and big endian */
-	dctx->litEntropy = dctx->fseEntropy = 0;
-	dctx->dictID = 0;
-	ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.rep) == sizeof(repStartValue));
-	memcpy(dctx->entropy.rep, repStartValue, sizeof(repStartValue)); /* initial repcodes */
-	dctx->LLTptr = dctx->entropy.LLTable;
-	dctx->MLTptr = dctx->entropy.MLTable;
-	dctx->OFTptr = dctx->entropy.OFTable;
-	dctx->HUFptr = dctx->entropy.hufTable;
-	return 0;
-}
-
-ZSTD_DCtx *ZSTD_createDCtx_advanced(ZSTD_customMem customMem)
-{
-	ZSTD_DCtx *dctx;
-
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-
-	dctx = (ZSTD_DCtx *)ZSTD_malloc(sizeof(ZSTD_DCtx), customMem);
-	if (!dctx)
-		return NULL;
-	memcpy(&dctx->customMem, &customMem, sizeof(customMem));
-	ZSTD_decompressBegin(dctx);
-	return dctx;
-}
-
-ZSTD_DCtx *ZSTD_initDCtx(void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	return ZSTD_createDCtx_advanced(stackMem);
-}
-
-size_t ZSTD_freeDCtx(ZSTD_DCtx *dctx)
-{
-	if (dctx == NULL)
-		return 0; /* support free on NULL */
-	ZSTD_free(dctx, dctx->customMem);
-	return 0; /* reserved as a potential error code in the future */
-}
-
-void ZSTD_copyDCtx(ZSTD_DCtx *dstDCtx, const ZSTD_DCtx *srcDCtx)
-{
-	size_t const workSpaceSize = (ZSTD_BLOCKSIZE_ABSOLUTEMAX + WILDCOPY_OVERLENGTH) + ZSTD_frameHeaderSize_max;
-	memcpy(dstDCtx, srcDCtx, sizeof(ZSTD_DCtx) - workSpaceSize); /* no need to copy workspace */
-}
-
-static void ZSTD_refDDict(ZSTD_DCtx *dstDCtx, const ZSTD_DDict *ddict);
-
-/*-*************************************************************
-*   Decompression section
-***************************************************************/
-
-/*! ZSTD_isFrame() :
- *  Tells if the content of `buffer` starts with a valid Frame Identifier.
- *  Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
- *  Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
- *  Note 3 : Skippable Frame Identifiers are considered valid. */
-unsigned ZSTD_isFrame(const void *buffer, size_t size)
-{
-	if (size < 4)
-		return 0;
-	{
-		U32 const magic = ZSTD_readLE32(buffer);
-		if (magic == ZSTD_MAGICNUMBER)
-			return 1;
-		if ((magic & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START)
-			return 1;
-	}
-	return 0;
-}
-
-/** ZSTD_frameHeaderSize() :
-*   srcSize must be >= ZSTD_frameHeaderSize_prefix.
-*   @return : size of the Frame Header */
-static size_t ZSTD_frameHeaderSize(const void *src, size_t srcSize)
-{
-	if (srcSize < ZSTD_frameHeaderSize_prefix)
-		return ERROR(srcSize_wrong);
-	{
-		BYTE const fhd = ((const BYTE *)src)[4];
-		U32 const dictID = fhd & 3;
-		U32 const singleSegment = (fhd >> 5) & 1;
-		U32 const fcsId = fhd >> 6;
-		return ZSTD_frameHeaderSize_prefix + !singleSegment + ZSTD_did_fieldSize[dictID] + ZSTD_fcs_fieldSize[fcsId] + (singleSegment && !fcsId);
-	}
-}
-
-/** ZSTD_getFrameParams() :
-*   decode Frame Header, or require larger `srcSize`.
-*   @return : 0, `fparamsPtr` is correctly filled,
-*            >0, `srcSize` is too small, result is expected `srcSize`,
-*             or an error code, which can be tested using ZSTD_isError() */
-size_t ZSTD_getFrameParams(ZSTD_frameParams *fparamsPtr, const void *src, size_t srcSize)
-{
-	const BYTE *ip = (const BYTE *)src;
-
-	if (srcSize < ZSTD_frameHeaderSize_prefix)
-		return ZSTD_frameHeaderSize_prefix;
-	if (ZSTD_readLE32(src) != ZSTD_MAGICNUMBER) {
-		if ((ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
-			if (srcSize < ZSTD_skippableHeaderSize)
-				return ZSTD_skippableHeaderSize; /* magic number + skippable frame length */
-			memset(fparamsPtr, 0, sizeof(*fparamsPtr));
-			fparamsPtr->frameContentSize = ZSTD_readLE32((const char *)src + 4);
-			fparamsPtr->windowSize = 0; /* windowSize==0 means a frame is skippable */
-			return 0;
-		}
-		return ERROR(prefix_unknown);
-	}
-
-	/* ensure there is enough `srcSize` to fully read/decode frame header */
-	{
-		size_t const fhsize = ZSTD_frameHeaderSize(src, srcSize);
-		if (srcSize < fhsize)
-			return fhsize;
-	}
-
-	{
-		BYTE const fhdByte = ip[4];
-		size_t pos = 5;
-		U32 const dictIDSizeCode = fhdByte & 3;
-		U32 const checksumFlag = (fhdByte >> 2) & 1;
-		U32 const singleSegment = (fhdByte >> 5) & 1;
-		U32 const fcsID = fhdByte >> 6;
-		U32 const windowSizeMax = 1U << ZSTD_WINDOWLOG_MAX;
-		U32 windowSize = 0;
-		U32 dictID = 0;
-		U64 frameContentSize = 0;
-		if ((fhdByte & 0x08) != 0)
-			return ERROR(frameParameter_unsupported); /* reserved bits, which must be zero */
-		if (!singleSegment) {
-			BYTE const wlByte = ip[pos++];
-			U32 const windowLog = (wlByte >> 3) + ZSTD_WINDOWLOG_ABSOLUTEMIN;
-			if (windowLog > ZSTD_WINDOWLOG_MAX)
-				return ERROR(frameParameter_windowTooLarge); /* avoids issue with 1 << windowLog */
-			windowSize = (1U << windowLog);
-			windowSize += (windowSize >> 3) * (wlByte & 7);
-		}
-
-		switch (dictIDSizeCode) {
-		default: /* impossible */
-		case 0: break;
-		case 1:
-			dictID = ip[pos];
-			pos++;
-			break;
-		case 2:
-			dictID = ZSTD_readLE16(ip + pos);
-			pos += 2;
-			break;
-		case 3:
-			dictID = ZSTD_readLE32(ip + pos);
-			pos += 4;
-			break;
-		}
-		switch (fcsID) {
-		default: /* impossible */
-		case 0:
-			if (singleSegment)
-				frameContentSize = ip[pos];
-			break;
-		case 1: frameContentSize = ZSTD_readLE16(ip + pos) + 256; break;
-		case 2: frameContentSize = ZSTD_readLE32(ip + pos); break;
-		case 3: frameContentSize = ZSTD_readLE64(ip + pos); break;
-		}
-		if (!windowSize)
-			windowSize = (U32)frameContentSize;
-		if (windowSize > windowSizeMax)
-			return ERROR(frameParameter_windowTooLarge);
-		fparamsPtr->frameContentSize = frameContentSize;
-		fparamsPtr->windowSize = windowSize;
-		fparamsPtr->dictID = dictID;
-		fparamsPtr->checksumFlag = checksumFlag;
-	}
-	return 0;
-}
-
-/** ZSTD_getFrameContentSize() :
-*   compatible with legacy mode
-*   @return : decompressed size of the single frame pointed to be `src` if known, otherwise
-*             - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
-*             - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small) */
-unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize)
-{
-	{
-		ZSTD_frameParams fParams;
-		if (ZSTD_getFrameParams(&fParams, src, srcSize) != 0)
-			return ZSTD_CONTENTSIZE_ERROR;
-		if (fParams.windowSize == 0) {
-			/* Either skippable or empty frame, size == 0 either way */
-			return 0;
-		} else if (fParams.frameContentSize != 0) {
-			return fParams.frameContentSize;
-		} else {
-			return ZSTD_CONTENTSIZE_UNKNOWN;
-		}
-	}
-}
-
-/** ZSTD_findDecompressedSize() :
- *  compatible with legacy mode
- *  `srcSize` must be the exact length of some number of ZSTD compressed and/or
- *      skippable frames
- *  @return : decompressed size of the frames contained */
-unsigned long long ZSTD_findDecompressedSize(const void *src, size_t srcSize)
-{
-	{
-		unsigned long long totalDstSize = 0;
-		while (srcSize >= ZSTD_frameHeaderSize_prefix) {
-			const U32 magicNumber = ZSTD_readLE32(src);
-
-			if ((magicNumber & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
-				size_t skippableSize;
-				if (srcSize < ZSTD_skippableHeaderSize)
-					return ERROR(srcSize_wrong);
-				skippableSize = ZSTD_readLE32((const BYTE *)src + 4) + ZSTD_skippableHeaderSize;
-				if (srcSize < skippableSize) {
-					return ZSTD_CONTENTSIZE_ERROR;
-				}
-
-				src = (const BYTE *)src + skippableSize;
-				srcSize -= skippableSize;
-				continue;
-			}
-
-			{
-				unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
-				if (ret >= ZSTD_CONTENTSIZE_ERROR)
-					return ret;
-
-				/* check for overflow */
-				if (totalDstSize + ret < totalDstSize)
-					return ZSTD_CONTENTSIZE_ERROR;
-				totalDstSize += ret;
-			}
-			{
-				size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
-				if (ZSTD_isError(frameSrcSize)) {
-					return ZSTD_CONTENTSIZE_ERROR;
-				}
-
-				src = (const BYTE *)src + frameSrcSize;
-				srcSize -= frameSrcSize;
-			}
-		}
-
-		if (srcSize) {
-			return ZSTD_CONTENTSIZE_ERROR;
-		}
-
-		return totalDstSize;
-	}
-}
-
-/** ZSTD_decodeFrameHeader() :
-*   `headerSize` must be the size provided by ZSTD_frameHeaderSize().
-*   @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */
-static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx *dctx, const void *src, size_t headerSize)
-{
-	size_t const result = ZSTD_getFrameParams(&(dctx->fParams), src, headerSize);
-	if (ZSTD_isError(result))
-		return result; /* invalid header */
-	if (result > 0)
-		return ERROR(srcSize_wrong); /* headerSize too small */
-	if (dctx->fParams.dictID && (dctx->dictID != dctx->fParams.dictID))
-		return ERROR(dictionary_wrong);
-	if (dctx->fParams.checksumFlag)
-		xxh64_reset(&dctx->xxhState, 0);
-	return 0;
-}
-
-typedef struct {
-	blockType_e blockType;
-	U32 lastBlock;
-	U32 origSize;
-} blockProperties_t;
-
-/*! ZSTD_getcBlockSize() :
-*   Provides the size of compressed block from block header `src` */
-size_t ZSTD_getcBlockSize(const void *src, size_t srcSize, blockProperties_t *bpPtr)
-{
-	if (srcSize < ZSTD_blockHeaderSize)
-		return ERROR(srcSize_wrong);
-	{
-		U32 const cBlockHeader = ZSTD_readLE24(src);
-		U32 const cSize = cBlockHeader >> 3;
-		bpPtr->lastBlock = cBlockHeader & 1;
-		bpPtr->blockType = (blockType_e)((cBlockHeader >> 1) & 3);
-		bpPtr->origSize = cSize; /* only useful for RLE */
-		if (bpPtr->blockType == bt_rle)
-			return 1;
-		if (bpPtr->blockType == bt_reserved)
-			return ERROR(corruption_detected);
-		return cSize;
-	}
-}
-
-static size_t ZSTD_copyRawBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	if (srcSize > dstCapacity)
-		return ERROR(dstSize_tooSmall);
-	memcpy(dst, src, srcSize);
-	return srcSize;
-}
-
-static size_t ZSTD_setRleBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize, size_t regenSize)
-{
-	if (srcSize != 1)
-		return ERROR(srcSize_wrong);
-	if (regenSize > dstCapacity)
-		return ERROR(dstSize_tooSmall);
-	memset(dst, *(const BYTE *)src, regenSize);
-	return regenSize;
-}
-
-/*! ZSTD_decodeLiteralsBlock() :
-	@return : nb of bytes read from src (< srcSize ) */
-size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx *dctx, const void *src, size_t srcSize) /* note : srcSize < BLOCKSIZE */
-{
-	if (srcSize < MIN_CBLOCK_SIZE)
-		return ERROR(corruption_detected);
-
-	{
-		const BYTE *const istart = (const BYTE *)src;
-		symbolEncodingType_e const litEncType = (symbolEncodingType_e)(istart[0] & 3);
-
-		switch (litEncType) {
-		case set_repeat:
-			if (dctx->litEntropy == 0)
-				return ERROR(dictionary_corrupted);
-			fallthrough;
-		case set_compressed:
-			if (srcSize < 5)
-				return ERROR(corruption_detected); /* srcSize >= MIN_CBLOCK_SIZE == 3; here we need up to 5 for case 3 */
-			{
-				size_t lhSize, litSize, litCSize;
-				U32 singleStream = 0;
-				U32 const lhlCode = (istart[0] >> 2) & 3;
-				U32 const lhc = ZSTD_readLE32(istart);
-				switch (lhlCode) {
-				case 0:
-				case 1:
-				default: /* note : default is impossible, since lhlCode into [0..3] */
-					/* 2 - 2 - 10 - 10 */
-					singleStream = !lhlCode;
-					lhSize = 3;
-					litSize = (lhc >> 4) & 0x3FF;
-					litCSize = (lhc >> 14) & 0x3FF;
-					break;
-				case 2:
-					/* 2 - 2 - 14 - 14 */
-					lhSize = 4;
-					litSize = (lhc >> 4) & 0x3FFF;
-					litCSize = lhc >> 18;
-					break;
-				case 3:
-					/* 2 - 2 - 18 - 18 */
-					lhSize = 5;
-					litSize = (lhc >> 4) & 0x3FFFF;
-					litCSize = (lhc >> 22) + (istart[4] << 10);
-					break;
-				}
-				if (litSize > ZSTD_BLOCKSIZE_ABSOLUTEMAX)
-					return ERROR(corruption_detected);
-				if (litCSize + lhSize > srcSize)
-					return ERROR(corruption_detected);
-
-				if (HUF_isError(
-					(litEncType == set_repeat)
-					    ? (singleStream ? HUF_decompress1X_usingDTable(dctx->litBuffer, litSize, istart + lhSize, litCSize, dctx->HUFptr)
-							    : HUF_decompress4X_usingDTable(dctx->litBuffer, litSize, istart + lhSize, litCSize, dctx->HUFptr))
-					    : (singleStream
-						   ? HUF_decompress1X2_DCtx_wksp(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart + lhSize, litCSize,
-										 dctx->entropy.workspace, sizeof(dctx->entropy.workspace))
-						   : HUF_decompress4X_hufOnly_wksp(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart + lhSize, litCSize,
-										   dctx->entropy.workspace, sizeof(dctx->entropy.workspace)))))
-					return ERROR(corruption_detected);
-
-				dctx->litPtr = dctx->litBuffer;
-				dctx->litSize = litSize;
-				dctx->litEntropy = 1;
-				if (litEncType == set_compressed)
-					dctx->HUFptr = dctx->entropy.hufTable;
-				memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
-				return litCSize + lhSize;
-			}
-
-		case set_basic: {
-			size_t litSize, lhSize;
-			U32 const lhlCode = ((istart[0]) >> 2) & 3;
-			switch (lhlCode) {
-			case 0:
-			case 2:
-			default: /* note : default is impossible, since lhlCode into [0..3] */
-				lhSize = 1;
-				litSize = istart[0] >> 3;
-				break;
-			case 1:
-				lhSize = 2;
-				litSize = ZSTD_readLE16(istart) >> 4;
-				break;
-			case 3:
-				lhSize = 3;
-				litSize = ZSTD_readLE24(istart) >> 4;
-				break;
-			}
-
-			if (lhSize + litSize + WILDCOPY_OVERLENGTH > srcSize) { /* risk reading beyond src buffer with wildcopy */
-				if (litSize + lhSize > srcSize)
-					return ERROR(corruption_detected);
-				memcpy(dctx->litBuffer, istart + lhSize, litSize);
-				dctx->litPtr = dctx->litBuffer;
-				dctx->litSize = litSize;
-				memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
-				return lhSize + litSize;
-			}
-			/* direct reference into compressed stream */
-			dctx->litPtr = istart + lhSize;
-			dctx->litSize = litSize;
-			return lhSize + litSize;
-		}
-
-		case set_rle: {
-			U32 const lhlCode = ((istart[0]) >> 2) & 3;
-			size_t litSize, lhSize;
-			switch (lhlCode) {
-			case 0:
-			case 2:
-			default: /* note : default is impossible, since lhlCode into [0..3] */
-				lhSize = 1;
-				litSize = istart[0] >> 3;
-				break;
-			case 1:
-				lhSize = 2;
-				litSize = ZSTD_readLE16(istart) >> 4;
-				break;
-			case 3:
-				lhSize = 3;
-				litSize = ZSTD_readLE24(istart) >> 4;
-				if (srcSize < 4)
-					return ERROR(corruption_detected); /* srcSize >= MIN_CBLOCK_SIZE == 3; here we need lhSize+1 = 4 */
-				break;
-			}
-			if (litSize > ZSTD_BLOCKSIZE_ABSOLUTEMAX)
-				return ERROR(corruption_detected);
-			memset(dctx->litBuffer, istart[lhSize], litSize + WILDCOPY_OVERLENGTH);
-			dctx->litPtr = dctx->litBuffer;
-			dctx->litSize = litSize;
-			return lhSize + 1;
-		}
-		default:
-			return ERROR(corruption_detected); /* impossible */
-		}
-	}
-}
-
-typedef union {
-	FSE_decode_t realData;
-	U32 alignedBy4;
-} FSE_decode_t4;
-
-static const FSE_decode_t4 LL_defaultDTable[(1 << LL_DEFAULTNORMLOG) + 1] = {
-    {{LL_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
-    {{0, 0, 4}},		 /* 0 : base, symbol, bits */
-    {{16, 0, 4}},
-    {{32, 1, 5}},
-    {{0, 3, 5}},
-    {{0, 4, 5}},
-    {{0, 6, 5}},
-    {{0, 7, 5}},
-    {{0, 9, 5}},
-    {{0, 10, 5}},
-    {{0, 12, 5}},
-    {{0, 14, 6}},
-    {{0, 16, 5}},
-    {{0, 18, 5}},
-    {{0, 19, 5}},
-    {{0, 21, 5}},
-    {{0, 22, 5}},
-    {{0, 24, 5}},
-    {{32, 25, 5}},
-    {{0, 26, 5}},
-    {{0, 27, 6}},
-    {{0, 29, 6}},
-    {{0, 31, 6}},
-    {{32, 0, 4}},
-    {{0, 1, 4}},
-    {{0, 2, 5}},
-    {{32, 4, 5}},
-    {{0, 5, 5}},
-    {{32, 7, 5}},
-    {{0, 8, 5}},
-    {{32, 10, 5}},
-    {{0, 11, 5}},
-    {{0, 13, 6}},
-    {{32, 16, 5}},
-    {{0, 17, 5}},
-    {{32, 19, 5}},
-    {{0, 20, 5}},
-    {{32, 22, 5}},
-    {{0, 23, 5}},
-    {{0, 25, 4}},
-    {{16, 25, 4}},
-    {{32, 26, 5}},
-    {{0, 28, 6}},
-    {{0, 30, 6}},
-    {{48, 0, 4}},
-    {{16, 1, 4}},
-    {{32, 2, 5}},
-    {{32, 3, 5}},
-    {{32, 5, 5}},
-    {{32, 6, 5}},
-    {{32, 8, 5}},
-    {{32, 9, 5}},
-    {{32, 11, 5}},
-    {{32, 12, 5}},
-    {{0, 15, 6}},
-    {{32, 17, 5}},
-    {{32, 18, 5}},
-    {{32, 20, 5}},
-    {{32, 21, 5}},
-    {{32, 23, 5}},
-    {{32, 24, 5}},
-    {{0, 35, 6}},
-    {{0, 34, 6}},
-    {{0, 33, 6}},
-    {{0, 32, 6}},
-}; /* LL_defaultDTable */
-
-static const FSE_decode_t4 ML_defaultDTable[(1 << ML_DEFAULTNORMLOG) + 1] = {
-    {{ML_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
-    {{0, 0, 6}},		 /* 0 : base, symbol, bits */
-    {{0, 1, 4}},
-    {{32, 2, 5}},
-    {{0, 3, 5}},
-    {{0, 5, 5}},
-    {{0, 6, 5}},
-    {{0, 8, 5}},
-    {{0, 10, 6}},
-    {{0, 13, 6}},
-    {{0, 16, 6}},
-    {{0, 19, 6}},
-    {{0, 22, 6}},
-    {{0, 25, 6}},
-    {{0, 28, 6}},
-    {{0, 31, 6}},
-    {{0, 33, 6}},
-    {{0, 35, 6}},
-    {{0, 37, 6}},
-    {{0, 39, 6}},
-    {{0, 41, 6}},
-    {{0, 43, 6}},
-    {{0, 45, 6}},
-    {{16, 1, 4}},
-    {{0, 2, 4}},
-    {{32, 3, 5}},
-    {{0, 4, 5}},
-    {{32, 6, 5}},
-    {{0, 7, 5}},
-    {{0, 9, 6}},
-    {{0, 12, 6}},
-    {{0, 15, 6}},
-    {{0, 18, 6}},
-    {{0, 21, 6}},
-    {{0, 24, 6}},
-    {{0, 27, 6}},
-    {{0, 30, 6}},
-    {{0, 32, 6}},
-    {{0, 34, 6}},
-    {{0, 36, 6}},
-    {{0, 38, 6}},
-    {{0, 40, 6}},
-    {{0, 42, 6}},
-    {{0, 44, 6}},
-    {{32, 1, 4}},
-    {{48, 1, 4}},
-    {{16, 2, 4}},
-    {{32, 4, 5}},
-    {{32, 5, 5}},
-    {{32, 7, 5}},
-    {{32, 8, 5}},
-    {{0, 11, 6}},
-    {{0, 14, 6}},
-    {{0, 17, 6}},
-    {{0, 20, 6}},
-    {{0, 23, 6}},
-    {{0, 26, 6}},
-    {{0, 29, 6}},
-    {{0, 52, 6}},
-    {{0, 51, 6}},
-    {{0, 50, 6}},
-    {{0, 49, 6}},
-    {{0, 48, 6}},
-    {{0, 47, 6}},
-    {{0, 46, 6}},
-}; /* ML_defaultDTable */
-
-static const FSE_decode_t4 OF_defaultDTable[(1 << OF_DEFAULTNORMLOG) + 1] = {
-    {{OF_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
-    {{0, 0, 5}},		 /* 0 : base, symbol, bits */
-    {{0, 6, 4}},
-    {{0, 9, 5}},
-    {{0, 15, 5}},
-    {{0, 21, 5}},
-    {{0, 3, 5}},
-    {{0, 7, 4}},
-    {{0, 12, 5}},
-    {{0, 18, 5}},
-    {{0, 23, 5}},
-    {{0, 5, 5}},
-    {{0, 8, 4}},
-    {{0, 14, 5}},
-    {{0, 20, 5}},
-    {{0, 2, 5}},
-    {{16, 7, 4}},
-    {{0, 11, 5}},
-    {{0, 17, 5}},
-    {{0, 22, 5}},
-    {{0, 4, 5}},
-    {{16, 8, 4}},
-    {{0, 13, 5}},
-    {{0, 19, 5}},
-    {{0, 1, 5}},
-    {{16, 6, 4}},
-    {{0, 10, 5}},
-    {{0, 16, 5}},
-    {{0, 28, 5}},
-    {{0, 27, 5}},
-    {{0, 26, 5}},
-    {{0, 25, 5}},
-    {{0, 24, 5}},
-}; /* OF_defaultDTable */
-
-/*! ZSTD_buildSeqTable() :
-	@return : nb bytes read from src,
-			  or an error code if it fails, testable with ZSTD_isError()
-*/
-static size_t ZSTD_buildSeqTable(FSE_DTable *DTableSpace, const FSE_DTable **DTablePtr, symbolEncodingType_e type, U32 max, U32 maxLog, const void *src,
-				 size_t srcSize, const FSE_decode_t4 *defaultTable, U32 flagRepeatTable, void *workspace, size_t workspaceSize)
-{
-	const void *const tmpPtr = defaultTable; /* bypass strict aliasing */
-	switch (type) {
-	case set_rle:
-		if (!srcSize)
-			return ERROR(srcSize_wrong);
-		if ((*(const BYTE *)src) > max)
-			return ERROR(corruption_detected);
-		FSE_buildDTable_rle(DTableSpace, *(const BYTE *)src);
-		*DTablePtr = DTableSpace;
-		return 1;
-	case set_basic: *DTablePtr = (const FSE_DTable *)tmpPtr; return 0;
-	case set_repeat:
-		if (!flagRepeatTable)
-			return ERROR(corruption_detected);
-		return 0;
-	default: /* impossible */
-	case set_compressed: {
-		U32 tableLog;
-		S16 *norm = (S16 *)workspace;
-		size_t const spaceUsed32 = ALIGN(sizeof(S16) * (MaxSeq + 1), sizeof(U32)) >> 2;
-
-		if ((spaceUsed32 << 2) > workspaceSize)
-			return ERROR(GENERIC);
-		workspace = (U32 *)workspace + spaceUsed32;
-		workspaceSize -= (spaceUsed32 << 2);
-		{
-			size_t const headerSize = FSE_readNCount(norm, &max, &tableLog, src, srcSize);
-			if (FSE_isError(headerSize))
-				return ERROR(corruption_detected);
-			if (tableLog > maxLog)
-				return ERROR(corruption_detected);
-			FSE_buildDTable_wksp(DTableSpace, norm, max, tableLog, workspace, workspaceSize);
-			*DTablePtr = DTableSpace;
-			return headerSize;
-		}
-	}
-	}
-}
-
-size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx *dctx, int *nbSeqPtr, const void *src, size_t srcSize)
-{
-	const BYTE *const istart = (const BYTE *const)src;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *ip = istart;
-
-	/* check */
-	if (srcSize < MIN_SEQUENCES_SIZE)
-		return ERROR(srcSize_wrong);
-
-	/* SeqHead */
-	{
-		int nbSeq = *ip++;
-		if (!nbSeq) {
-			*nbSeqPtr = 0;
-			return 1;
-		}
-		if (nbSeq > 0x7F) {
-			if (nbSeq == 0xFF) {
-				if (ip + 2 > iend)
-					return ERROR(srcSize_wrong);
-				nbSeq = ZSTD_readLE16(ip) + LONGNBSEQ, ip += 2;
-			} else {
-				if (ip >= iend)
-					return ERROR(srcSize_wrong);
-				nbSeq = ((nbSeq - 0x80) << 8) + *ip++;
-			}
-		}
-		*nbSeqPtr = nbSeq;
-	}
-
-	/* FSE table descriptors */
-	if (ip + 4 > iend)
-		return ERROR(srcSize_wrong); /* minimum possible size */
-	{
-		symbolEncodingType_e const LLtype = (symbolEncodingType_e)(*ip >> 6);
-		symbolEncodingType_e const OFtype = (symbolEncodingType_e)((*ip >> 4) & 3);
-		symbolEncodingType_e const MLtype = (symbolEncodingType_e)((*ip >> 2) & 3);
-		ip++;
-
-		/* Build DTables */
-		{
-			size_t const llhSize = ZSTD_buildSeqTable(dctx->entropy.LLTable, &dctx->LLTptr, LLtype, MaxLL, LLFSELog, ip, iend - ip,
-								  LL_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
-			if (ZSTD_isError(llhSize))
-				return ERROR(corruption_detected);
-			ip += llhSize;
-		}
-		{
-			size_t const ofhSize = ZSTD_buildSeqTable(dctx->entropy.OFTable, &dctx->OFTptr, OFtype, MaxOff, OffFSELog, ip, iend - ip,
-								  OF_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
-			if (ZSTD_isError(ofhSize))
-				return ERROR(corruption_detected);
-			ip += ofhSize;
-		}
-		{
-			size_t const mlhSize = ZSTD_buildSeqTable(dctx->entropy.MLTable, &dctx->MLTptr, MLtype, MaxML, MLFSELog, ip, iend - ip,
-								  ML_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
-			if (ZSTD_isError(mlhSize))
-				return ERROR(corruption_detected);
-			ip += mlhSize;
-		}
-	}
-
-	return ip - istart;
-}
-
-typedef struct {
-	size_t litLength;
-	size_t matchLength;
-	size_t offset;
-	const BYTE *match;
-} seq_t;
-
-typedef struct {
-	BIT_DStream_t DStream;
-	FSE_DState_t stateLL;
-	FSE_DState_t stateOffb;
-	FSE_DState_t stateML;
-	size_t prevOffset[ZSTD_REP_NUM];
-	const BYTE *base;
-	size_t pos;
-	uPtrDiff gotoDict;
-} seqState_t;
-
-FORCE_NOINLINE
-size_t ZSTD_execSequenceLast7(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
-			      const BYTE *const vBase, const BYTE *const dictEnd)
-{
-	BYTE *const oLitEnd = op + sequence.litLength;
-	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
-	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
-	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
-	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
-	const BYTE *match = oLitEnd - sequence.offset;
-
-	/* check */
-	if (oMatchEnd > oend)
-		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
-	if (iLitEnd > litLimit)
-		return ERROR(corruption_detected); /* over-read beyond lit buffer */
-	if (oLitEnd <= oend_w)
-		return ERROR(GENERIC); /* Precondition */
-
-	/* copy literals */
-	if (op < oend_w) {
-		ZSTD_wildcopy(op, *litPtr, oend_w - op);
-		*litPtr += oend_w - op;
-		op = oend_w;
-	}
-	while (op < oLitEnd)
-		*op++ = *(*litPtr)++;
-
-	/* copy Match */
-	if (sequence.offset > (size_t)(oLitEnd - base)) {
-		/* offset beyond prefix */
-		if (sequence.offset > (size_t)(oLitEnd - vBase))
-			return ERROR(corruption_detected);
-		match = dictEnd - (base - match);
-		if (match + sequence.matchLength <= dictEnd) {
-			memmove(oLitEnd, match, sequence.matchLength);
-			return sequenceLength;
-		}
-		/* span extDict & currPrefixSegment */
-		{
-			size_t const length1 = dictEnd - match;
-			memmove(oLitEnd, match, length1);
-			op = oLitEnd + length1;
-			sequence.matchLength -= length1;
-			match = base;
-		}
-	}
-	while (op < oMatchEnd)
-		*op++ = *match++;
-	return sequenceLength;
-}
-
-static seq_t ZSTD_decodeSequence(seqState_t *seqState)
-{
-	seq_t seq;
-
-	U32 const llCode = FSE_peekSymbol(&seqState->stateLL);
-	U32 const mlCode = FSE_peekSymbol(&seqState->stateML);
-	U32 const ofCode = FSE_peekSymbol(&seqState->stateOffb); /* <= maxOff, by table construction */
-
-	U32 const llBits = LL_bits[llCode];
-	U32 const mlBits = ML_bits[mlCode];
-	U32 const ofBits = ofCode;
-	U32 const totalBits = llBits + mlBits + ofBits;
-
-	static const U32 LL_base[MaxLL + 1] = {0,  1,  2,  3,  4,  5,  6,  7,  8,    9,     10,    11,    12,    13,     14,     15,     16,     18,
-					       20, 22, 24, 28, 32, 40, 48, 64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000, 0x8000, 0x10000};
-
-	static const U32 ML_base[MaxML + 1] = {3,  4,  5,  6,  7,  8,  9,  10,   11,    12,    13,    14,    15,     16,     17,     18,     19,     20,
-					       21, 22, 23, 24, 25, 26, 27, 28,   29,    30,    31,    32,    33,     34,     35,     37,     39,     41,
-					       43, 47, 51, 59, 67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803, 0x1003, 0x2003, 0x4003, 0x8003, 0x10003};
-
-	static const U32 OF_base[MaxOff + 1] = {0,       1,	1,	5,	0xD,      0x1D,      0x3D,      0x7D,      0xFD,     0x1FD,
-						0x3FD,   0x7FD,    0xFFD,    0x1FFD,   0x3FFD,   0x7FFD,    0xFFFD,    0x1FFFD,   0x3FFFD,  0x7FFFD,
-						0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD, 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD};
-
-	/* sequence */
-	{
-		size_t offset;
-		if (!ofCode)
-			offset = 0;
-		else {
-			offset = OF_base[ofCode] + BIT_readBitsFast(&seqState->DStream, ofBits); /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
-			if (ZSTD_32bits())
-				BIT_reloadDStream(&seqState->DStream);
-		}
-
-		if (ofCode <= 1) {
-			offset += (llCode == 0);
-			if (offset) {
-				size_t temp = (offset == 3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
-				temp += !temp; /* 0 is not valid; input is corrupted; force offset to 1 */
-				if (offset != 1)
-					seqState->prevOffset[2] = seqState->prevOffset[1];
-				seqState->prevOffset[1] = seqState->prevOffset[0];
-				seqState->prevOffset[0] = offset = temp;
-			} else {
-				offset = seqState->prevOffset[0];
-			}
-		} else {
-			seqState->prevOffset[2] = seqState->prevOffset[1];
-			seqState->prevOffset[1] = seqState->prevOffset[0];
-			seqState->prevOffset[0] = offset;
-		}
-		seq.offset = offset;
-	}
-
-	seq.matchLength = ML_base[mlCode] + ((mlCode > 31) ? BIT_readBitsFast(&seqState->DStream, mlBits) : 0); /* <=  16 bits */
-	if (ZSTD_32bits() && (mlBits + llBits > 24))
-		BIT_reloadDStream(&seqState->DStream);
-
-	seq.litLength = LL_base[llCode] + ((llCode > 15) ? BIT_readBitsFast(&seqState->DStream, llBits) : 0); /* <=  16 bits */
-	if (ZSTD_32bits() || (totalBits > 64 - 7 - (LLFSELog + MLFSELog + OffFSELog)))
-		BIT_reloadDStream(&seqState->DStream);
-
-	/* ANS state update */
-	FSE_updateState(&seqState->stateLL, &seqState->DStream); /* <=  9 bits */
-	FSE_updateState(&seqState->stateML, &seqState->DStream); /* <=  9 bits */
-	if (ZSTD_32bits())
-		BIT_reloadDStream(&seqState->DStream);		   /* <= 18 bits */
-	FSE_updateState(&seqState->stateOffb, &seqState->DStream); /* <=  8 bits */
-
-	seq.match = NULL;
-
-	return seq;
-}
-
-FORCE_INLINE
-size_t ZSTD_execSequence(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
-			 const BYTE *const vBase, const BYTE *const dictEnd)
-{
-	BYTE *const oLitEnd = op + sequence.litLength;
-	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
-	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
-	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
-	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
-	const BYTE *match = oLitEnd - sequence.offset;
-
-	/* check */
-	if (oMatchEnd > oend)
-		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
-	if (iLitEnd > litLimit)
-		return ERROR(corruption_detected); /* over-read beyond lit buffer */
-	if (oLitEnd > oend_w)
-		return ZSTD_execSequenceLast7(op, oend, sequence, litPtr, litLimit, base, vBase, dictEnd);
-
-	/* copy Literals */
-	ZSTD_copy8(op, *litPtr);
-	if (sequence.litLength > 8)
-		ZSTD_wildcopy(op + 8, (*litPtr) + 8,
-			      sequence.litLength - 8); /* note : since oLitEnd <= oend-WILDCOPY_OVERLENGTH, no risk of overwrite beyond oend */
-	op = oLitEnd;
-	*litPtr = iLitEnd; /* update for next sequence */
-
-	/* copy Match */
-	if (sequence.offset > (size_t)(oLitEnd - base)) {
-		/* offset beyond prefix */
-		if (sequence.offset > (size_t)(oLitEnd - vBase))
-			return ERROR(corruption_detected);
-		match = dictEnd + (match - base);
-		if (match + sequence.matchLength <= dictEnd) {
-			memmove(oLitEnd, match, sequence.matchLength);
-			return sequenceLength;
-		}
-		/* span extDict & currPrefixSegment */
-		{
-			size_t const length1 = dictEnd - match;
-			memmove(oLitEnd, match, length1);
-			op = oLitEnd + length1;
-			sequence.matchLength -= length1;
-			match = base;
-			if (op > oend_w || sequence.matchLength < MINMATCH) {
-				U32 i;
-				for (i = 0; i < sequence.matchLength; ++i)
-					op[i] = match[i];
-				return sequenceLength;
-			}
-		}
-	}
-	/* Requirement: op <= oend_w && sequence.matchLength >= MINMATCH */
-
-	/* match within prefix */
-	if (sequence.offset < 8) {
-		/* close range match, overlap */
-		static const U32 dec32table[] = {0, 1, 2, 1, 4, 4, 4, 4};   /* added */
-		static const int dec64table[] = {8, 8, 8, 7, 8, 9, 10, 11}; /* subtracted */
-		int const sub2 = dec64table[sequence.offset];
-		op[0] = match[0];
-		op[1] = match[1];
-		op[2] = match[2];
-		op[3] = match[3];
-		match += dec32table[sequence.offset];
-		ZSTD_copy4(op + 4, match);
-		match -= sub2;
-	} else {
-		ZSTD_copy8(op, match);
-	}
-	op += 8;
-	match += 8;
-
-	if (oMatchEnd > oend - (16 - MINMATCH)) {
-		if (op < oend_w) {
-			ZSTD_wildcopy(op, match, oend_w - op);
-			match += oend_w - op;
-			op = oend_w;
-		}
-		while (op < oMatchEnd)
-			*op++ = *match++;
-	} else {
-		ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength - 8); /* works even if matchLength < 8 */
-	}
-	return sequenceLength;
-}
-
-static size_t ZSTD_decompressSequences(ZSTD_DCtx *dctx, void *dst, size_t maxDstSize, const void *seqStart, size_t seqSize)
-{
-	const BYTE *ip = (const BYTE *)seqStart;
-	const BYTE *const iend = ip + seqSize;
-	BYTE *const ostart = (BYTE * const)dst;
-	BYTE *const oend = ostart + maxDstSize;
-	BYTE *op = ostart;
-	const BYTE *litPtr = dctx->litPtr;
-	const BYTE *const litEnd = litPtr + dctx->litSize;
-	const BYTE *const base = (const BYTE *)(dctx->base);
-	const BYTE *const vBase = (const BYTE *)(dctx->vBase);
-	const BYTE *const dictEnd = (const BYTE *)(dctx->dictEnd);
-	int nbSeq;
-
-	/* Build Decoding Tables */
-	{
-		size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, seqSize);
-		if (ZSTD_isError(seqHSize))
-			return seqHSize;
-		ip += seqHSize;
-	}
-
-	/* Regen sequences */
-	if (nbSeq) {
-		seqState_t seqState;
-		dctx->fseEntropy = 1;
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				seqState.prevOffset[i] = dctx->entropy.rep[i];
-		}
-		CHECK_E(BIT_initDStream(&seqState.DStream, ip, iend - ip), corruption_detected);
-		FSE_initDState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
-		FSE_initDState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
-		FSE_initDState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
-
-		for (; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && nbSeq;) {
-			nbSeq--;
-			{
-				seq_t const sequence = ZSTD_decodeSequence(&seqState);
-				size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, base, vBase, dictEnd);
-				if (ZSTD_isError(oneSeqSize))
-					return oneSeqSize;
-				op += oneSeqSize;
-			}
-		}
-
-		/* check if reached exact end */
-		if (nbSeq)
-			return ERROR(corruption_detected);
-		/* save reps for next block */
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]);
-		}
-	}
-
-	/* last literal segment */
-	{
-		size_t const lastLLSize = litEnd - litPtr;
-		if (lastLLSize > (size_t)(oend - op))
-			return ERROR(dstSize_tooSmall);
-		memcpy(op, litPtr, lastLLSize);
-		op += lastLLSize;
-	}
-
-	return op - ostart;
-}
-
-FORCE_INLINE seq_t ZSTD_decodeSequenceLong_generic(seqState_t *seqState, int const longOffsets)
-{
-	seq_t seq;
-
-	U32 const llCode = FSE_peekSymbol(&seqState->stateLL);
-	U32 const mlCode = FSE_peekSymbol(&seqState->stateML);
-	U32 const ofCode = FSE_peekSymbol(&seqState->stateOffb); /* <= maxOff, by table construction */
-
-	U32 const llBits = LL_bits[llCode];
-	U32 const mlBits = ML_bits[mlCode];
-	U32 const ofBits = ofCode;
-	U32 const totalBits = llBits + mlBits + ofBits;
-
-	static const U32 LL_base[MaxLL + 1] = {0,  1,  2,  3,  4,  5,  6,  7,  8,    9,     10,    11,    12,    13,     14,     15,     16,     18,
-					       20, 22, 24, 28, 32, 40, 48, 64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000, 0x8000, 0x10000};
-
-	static const U32 ML_base[MaxML + 1] = {3,  4,  5,  6,  7,  8,  9,  10,   11,    12,    13,    14,    15,     16,     17,     18,     19,     20,
-					       21, 22, 23, 24, 25, 26, 27, 28,   29,    30,    31,    32,    33,     34,     35,     37,     39,     41,
-					       43, 47, 51, 59, 67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803, 0x1003, 0x2003, 0x4003, 0x8003, 0x10003};
-
-	static const U32 OF_base[MaxOff + 1] = {0,       1,	1,	5,	0xD,      0x1D,      0x3D,      0x7D,      0xFD,     0x1FD,
-						0x3FD,   0x7FD,    0xFFD,    0x1FFD,   0x3FFD,   0x7FFD,    0xFFFD,    0x1FFFD,   0x3FFFD,  0x7FFFD,
-						0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD, 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD};
-
-	/* sequence */
-	{
-		size_t offset;
-		if (!ofCode)
-			offset = 0;
-		else {
-			if (longOffsets) {
-				int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN);
-				offset = OF_base[ofCode] + (BIT_readBitsFast(&seqState->DStream, ofBits - extraBits) << extraBits);
-				if (ZSTD_32bits() || extraBits)
-					BIT_reloadDStream(&seqState->DStream);
-				if (extraBits)
-					offset += BIT_readBitsFast(&seqState->DStream, extraBits);
-			} else {
-				offset = OF_base[ofCode] + BIT_readBitsFast(&seqState->DStream, ofBits); /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
-				if (ZSTD_32bits())
-					BIT_reloadDStream(&seqState->DStream);
-			}
-		}
-
-		if (ofCode <= 1) {
-			offset += (llCode == 0);
-			if (offset) {
-				size_t temp = (offset == 3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
-				temp += !temp; /* 0 is not valid; input is corrupted; force offset to 1 */
-				if (offset != 1)
-					seqState->prevOffset[2] = seqState->prevOffset[1];
-				seqState->prevOffset[1] = seqState->prevOffset[0];
-				seqState->prevOffset[0] = offset = temp;
-			} else {
-				offset = seqState->prevOffset[0];
-			}
-		} else {
-			seqState->prevOffset[2] = seqState->prevOffset[1];
-			seqState->prevOffset[1] = seqState->prevOffset[0];
-			seqState->prevOffset[0] = offset;
-		}
-		seq.offset = offset;
-	}
-
-	seq.matchLength = ML_base[mlCode] + ((mlCode > 31) ? BIT_readBitsFast(&seqState->DStream, mlBits) : 0); /* <=  16 bits */
-	if (ZSTD_32bits() && (mlBits + llBits > 24))
-		BIT_reloadDStream(&seqState->DStream);
-
-	seq.litLength = LL_base[llCode] + ((llCode > 15) ? BIT_readBitsFast(&seqState->DStream, llBits) : 0); /* <=  16 bits */
-	if (ZSTD_32bits() || (totalBits > 64 - 7 - (LLFSELog + MLFSELog + OffFSELog)))
-		BIT_reloadDStream(&seqState->DStream);
-
-	{
-		size_t const pos = seqState->pos + seq.litLength;
-		seq.match = seqState->base + pos - seq.offset; /* single memory segment */
-		if (seq.offset > pos)
-			seq.match += seqState->gotoDict; /* separate memory segment */
-		seqState->pos = pos + seq.matchLength;
-	}
-
-	/* ANS state update */
-	FSE_updateState(&seqState->stateLL, &seqState->DStream); /* <=  9 bits */
-	FSE_updateState(&seqState->stateML, &seqState->DStream); /* <=  9 bits */
-	if (ZSTD_32bits())
-		BIT_reloadDStream(&seqState->DStream);		   /* <= 18 bits */
-	FSE_updateState(&seqState->stateOffb, &seqState->DStream); /* <=  8 bits */
-
-	return seq;
-}
-
-static seq_t ZSTD_decodeSequenceLong(seqState_t *seqState, unsigned const windowSize)
-{
-	if (ZSTD_highbit32(windowSize) > STREAM_ACCUMULATOR_MIN) {
-		return ZSTD_decodeSequenceLong_generic(seqState, 1);
-	} else {
-		return ZSTD_decodeSequenceLong_generic(seqState, 0);
-	}
-}
-
-FORCE_INLINE
-size_t ZSTD_execSequenceLong(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
-			     const BYTE *const vBase, const BYTE *const dictEnd)
-{
-	BYTE *const oLitEnd = op + sequence.litLength;
-	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
-	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
-	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
-	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
-	const BYTE *match = sequence.match;
-
-	/* check */
-	if (oMatchEnd > oend)
-		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
-	if (iLitEnd > litLimit)
-		return ERROR(corruption_detected); /* over-read beyond lit buffer */
-	if (oLitEnd > oend_w)
-		return ZSTD_execSequenceLast7(op, oend, sequence, litPtr, litLimit, base, vBase, dictEnd);
-
-	/* copy Literals */
-	ZSTD_copy8(op, *litPtr);
-	if (sequence.litLength > 8)
-		ZSTD_wildcopy(op + 8, (*litPtr) + 8,
-			      sequence.litLength - 8); /* note : since oLitEnd <= oend-WILDCOPY_OVERLENGTH, no risk of overwrite beyond oend */
-	op = oLitEnd;
-	*litPtr = iLitEnd; /* update for next sequence */
-
-	/* copy Match */
-	if (sequence.offset > (size_t)(oLitEnd - base)) {
-		/* offset beyond prefix */
-		if (sequence.offset > (size_t)(oLitEnd - vBase))
-			return ERROR(corruption_detected);
-		if (match + sequence.matchLength <= dictEnd) {
-			memmove(oLitEnd, match, sequence.matchLength);
-			return sequenceLength;
-		}
-		/* span extDict & currPrefixSegment */
-		{
-			size_t const length1 = dictEnd - match;
-			memmove(oLitEnd, match, length1);
-			op = oLitEnd + length1;
-			sequence.matchLength -= length1;
-			match = base;
-			if (op > oend_w || sequence.matchLength < MINMATCH) {
-				U32 i;
-				for (i = 0; i < sequence.matchLength; ++i)
-					op[i] = match[i];
-				return sequenceLength;
-			}
-		}
-	}
-	/* Requirement: op <= oend_w && sequence.matchLength >= MINMATCH */
-
-	/* match within prefix */
-	if (sequence.offset < 8) {
-		/* close range match, overlap */
-		static const U32 dec32table[] = {0, 1, 2, 1, 4, 4, 4, 4};   /* added */
-		static const int dec64table[] = {8, 8, 8, 7, 8, 9, 10, 11}; /* subtracted */
-		int const sub2 = dec64table[sequence.offset];
-		op[0] = match[0];
-		op[1] = match[1];
-		op[2] = match[2];
-		op[3] = match[3];
-		match += dec32table[sequence.offset];
-		ZSTD_copy4(op + 4, match);
-		match -= sub2;
-	} else {
-		ZSTD_copy8(op, match);
-	}
-	op += 8;
-	match += 8;
-
-	if (oMatchEnd > oend - (16 - MINMATCH)) {
-		if (op < oend_w) {
-			ZSTD_wildcopy(op, match, oend_w - op);
-			match += oend_w - op;
-			op = oend_w;
-		}
-		while (op < oMatchEnd)
-			*op++ = *match++;
-	} else {
-		ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength - 8); /* works even if matchLength < 8 */
-	}
-	return sequenceLength;
-}
-
-static size_t ZSTD_decompressSequencesLong(ZSTD_DCtx *dctx, void *dst, size_t maxDstSize, const void *seqStart, size_t seqSize)
-{
-	const BYTE *ip = (const BYTE *)seqStart;
-	const BYTE *const iend = ip + seqSize;
-	BYTE *const ostart = (BYTE * const)dst;
-	BYTE *const oend = ostart + maxDstSize;
-	BYTE *op = ostart;
-	const BYTE *litPtr = dctx->litPtr;
-	const BYTE *const litEnd = litPtr + dctx->litSize;
-	const BYTE *const base = (const BYTE *)(dctx->base);
-	const BYTE *const vBase = (const BYTE *)(dctx->vBase);
-	const BYTE *const dictEnd = (const BYTE *)(dctx->dictEnd);
-	unsigned const windowSize = dctx->fParams.windowSize;
-	int nbSeq;
-
-	/* Build Decoding Tables */
-	{
-		size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, seqSize);
-		if (ZSTD_isError(seqHSize))
-			return seqHSize;
-		ip += seqHSize;
-	}
-
-	/* Regen sequences */
-	if (nbSeq) {
-#define STORED_SEQS 4
-#define STOSEQ_MASK (STORED_SEQS - 1)
-#define ADVANCED_SEQS 4
-		seq_t *sequences = (seq_t *)dctx->entropy.workspace;
-		int const seqAdvance = MIN(nbSeq, ADVANCED_SEQS);
-		seqState_t seqState;
-		int seqNb;
-		ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.workspace) >= sizeof(seq_t) * STORED_SEQS);
-		dctx->fseEntropy = 1;
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				seqState.prevOffset[i] = dctx->entropy.rep[i];
-		}
-		seqState.base = base;
-		seqState.pos = (size_t)(op - base);
-		seqState.gotoDict = (uPtrDiff)dictEnd - (uPtrDiff)base; /* cast to avoid undefined behaviour */
-		CHECK_E(BIT_initDStream(&seqState.DStream, ip, iend - ip), corruption_detected);
-		FSE_initDState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
-		FSE_initDState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
-		FSE_initDState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
-
-		/* prepare in advance */
-		for (seqNb = 0; (BIT_reloadDStream(&seqState.DStream) <= BIT_DStream_completed) && seqNb < seqAdvance; seqNb++) {
-			sequences[seqNb] = ZSTD_decodeSequenceLong(&seqState, windowSize);
-		}
-		if (seqNb < seqAdvance)
-			return ERROR(corruption_detected);
-
-		/* decode and decompress */
-		for (; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && seqNb < nbSeq; seqNb++) {
-			seq_t const sequence = ZSTD_decodeSequenceLong(&seqState, windowSize);
-			size_t const oneSeqSize =
-			    ZSTD_execSequenceLong(op, oend, sequences[(seqNb - ADVANCED_SEQS) & STOSEQ_MASK], &litPtr, litEnd, base, vBase, dictEnd);
-			if (ZSTD_isError(oneSeqSize))
-				return oneSeqSize;
-			ZSTD_PREFETCH(sequence.match);
-			sequences[seqNb & STOSEQ_MASK] = sequence;
-			op += oneSeqSize;
-		}
-		if (seqNb < nbSeq)
-			return ERROR(corruption_detected);
-
-		/* finish queue */
-		seqNb -= seqAdvance;
-		for (; seqNb < nbSeq; seqNb++) {
-			size_t const oneSeqSize = ZSTD_execSequenceLong(op, oend, sequences[seqNb & STOSEQ_MASK], &litPtr, litEnd, base, vBase, dictEnd);
-			if (ZSTD_isError(oneSeqSize))
-				return oneSeqSize;
-			op += oneSeqSize;
-		}
-
-		/* save reps for next block */
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]);
-		}
-	}
-
-	/* last literal segment */
-	{
-		size_t const lastLLSize = litEnd - litPtr;
-		if (lastLLSize > (size_t)(oend - op))
-			return ERROR(dstSize_tooSmall);
-		memcpy(op, litPtr, lastLLSize);
-		op += lastLLSize;
-	}
-
-	return op - ostart;
-}
-
-static size_t ZSTD_decompressBlock_internal(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{ /* blockType == blockCompressed */
-	const BYTE *ip = (const BYTE *)src;
-
-	if (srcSize >= ZSTD_BLOCKSIZE_ABSOLUTEMAX)
-		return ERROR(srcSize_wrong);
-
-	/* Decode literals section */
-	{
-		size_t const litCSize = ZSTD_decodeLiteralsBlock(dctx, src, srcSize);
-		if (ZSTD_isError(litCSize))
-			return litCSize;
-		ip += litCSize;
-		srcSize -= litCSize;
-	}
-	if (sizeof(size_t) > 4) /* do not enable prefetching on 32-bits x86, as it's performance detrimental */
-				/* likely because of register pressure */
-				/* if that's the correct cause, then 32-bits ARM should be affected differently */
-				/* it would be good to test this on ARM real hardware, to see if prefetch version improves speed */
-		if (dctx->fParams.windowSize > (1 << 23))
-			return ZSTD_decompressSequencesLong(dctx, dst, dstCapacity, ip, srcSize);
-	return ZSTD_decompressSequences(dctx, dst, dstCapacity, ip, srcSize);
-}
-
-static void ZSTD_checkContinuity(ZSTD_DCtx *dctx, const void *dst)
-{
-	if (dst != dctx->previousDstEnd) { /* not contiguous */
-		dctx->dictEnd = dctx->previousDstEnd;
-		dctx->vBase = (const char *)dst - ((const char *)(dctx->previousDstEnd) - (const char *)(dctx->base));
-		dctx->base = dst;
-		dctx->previousDstEnd = dst;
-	}
-}
-
-size_t ZSTD_decompressBlock(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	size_t dSize;
-	ZSTD_checkContinuity(dctx, dst);
-	dSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize);
-	dctx->previousDstEnd = (char *)dst + dSize;
-	return dSize;
-}
-
-/** ZSTD_insertBlock() :
-	insert `src` block into `dctx` history. Useful to track uncompressed blocks. */
-size_t ZSTD_insertBlock(ZSTD_DCtx *dctx, const void *blockStart, size_t blockSize)
-{
-	ZSTD_checkContinuity(dctx, blockStart);
-	dctx->previousDstEnd = (const char *)blockStart + blockSize;
-	return blockSize;
-}
-
-size_t ZSTD_generateNxBytes(void *dst, size_t dstCapacity, BYTE byte, size_t length)
-{
-	if (length > dstCapacity)
-		return ERROR(dstSize_tooSmall);
-	memset(dst, byte, length);
-	return length;
-}
-
-/** ZSTD_findFrameCompressedSize() :
- *  compatible with legacy mode
- *  `src` must point to the start of a ZSTD frame, ZSTD legacy frame, or skippable frame
- *  `srcSize` must be at least as large as the frame contained
- *  @return : the compressed size of the frame starting at `src` */
-size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize)
-{
-	if (srcSize >= ZSTD_skippableHeaderSize && (ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
-		return ZSTD_skippableHeaderSize + ZSTD_readLE32((const BYTE *)src + 4);
-	} else {
-		const BYTE *ip = (const BYTE *)src;
-		const BYTE *const ipstart = ip;
-		size_t remainingSize = srcSize;
-		ZSTD_frameParams fParams;
-
-		size_t const headerSize = ZSTD_frameHeaderSize(ip, remainingSize);
-		if (ZSTD_isError(headerSize))
-			return headerSize;
-
-		/* Frame Header */
-		{
-			size_t const ret = ZSTD_getFrameParams(&fParams, ip, remainingSize);
-			if (ZSTD_isError(ret))
-				return ret;
-			if (ret > 0)
-				return ERROR(srcSize_wrong);
-		}
-
-		ip += headerSize;
-		remainingSize -= headerSize;
-
-		/* Loop on each block */
-		while (1) {
-			blockProperties_t blockProperties;
-			size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
-			if (ZSTD_isError(cBlockSize))
-				return cBlockSize;
-
-			if (ZSTD_blockHeaderSize + cBlockSize > remainingSize)
-				return ERROR(srcSize_wrong);
-
-			ip += ZSTD_blockHeaderSize + cBlockSize;
-			remainingSize -= ZSTD_blockHeaderSize + cBlockSize;
-
-			if (blockProperties.lastBlock)
-				break;
-		}
-
-		if (fParams.checksumFlag) { /* Frame content checksum */
-			if (remainingSize < 4)
-				return ERROR(srcSize_wrong);
-			ip += 4;
-			remainingSize -= 4;
-		}
-
-		return ip - ipstart;
-	}
-}
-
-/*! ZSTD_decompressFrame() :
-*   @dctx must be properly initialized */
-static size_t ZSTD_decompressFrame(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void **srcPtr, size_t *srcSizePtr)
-{
-	const BYTE *ip = (const BYTE *)(*srcPtr);
-	BYTE *const ostart = (BYTE * const)dst;
-	BYTE *const oend = ostart + dstCapacity;
-	BYTE *op = ostart;
-	size_t remainingSize = *srcSizePtr;
-
-	/* check */
-	if (remainingSize < ZSTD_frameHeaderSize_min + ZSTD_blockHeaderSize)
-		return ERROR(srcSize_wrong);
-
-	/* Frame Header */
-	{
-		size_t const frameHeaderSize = ZSTD_frameHeaderSize(ip, ZSTD_frameHeaderSize_prefix);
-		if (ZSTD_isError(frameHeaderSize))
-			return frameHeaderSize;
-		if (remainingSize < frameHeaderSize + ZSTD_blockHeaderSize)
-			return ERROR(srcSize_wrong);
-		CHECK_F(ZSTD_decodeFrameHeader(dctx, ip, frameHeaderSize));
-		ip += frameHeaderSize;
-		remainingSize -= frameHeaderSize;
-	}
-
-	/* Loop on each block */
-	while (1) {
-		size_t decodedSize;
-		blockProperties_t blockProperties;
-		size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
-		if (ZSTD_isError(cBlockSize))
-			return cBlockSize;
-
-		ip += ZSTD_blockHeaderSize;
-		remainingSize -= ZSTD_blockHeaderSize;
-		if (cBlockSize > remainingSize)
-			return ERROR(srcSize_wrong);
-
-		switch (blockProperties.blockType) {
-		case bt_compressed: decodedSize = ZSTD_decompressBlock_internal(dctx, op, oend - op, ip, cBlockSize); break;
-		case bt_raw: decodedSize = ZSTD_copyRawBlock(op, oend - op, ip, cBlockSize); break;
-		case bt_rle: decodedSize = ZSTD_generateNxBytes(op, oend - op, *ip, blockProperties.origSize); break;
-		case bt_reserved:
-		default: return ERROR(corruption_detected);
-		}
-
-		if (ZSTD_isError(decodedSize))
-			return decodedSize;
-		if (dctx->fParams.checksumFlag)
-			xxh64_update(&dctx->xxhState, op, decodedSize);
-		op += decodedSize;
-		ip += cBlockSize;
-		remainingSize -= cBlockSize;
-		if (blockProperties.lastBlock)
-			break;
-	}
-
-	if (dctx->fParams.checksumFlag) { /* Frame content checksum verification */
-		U32 const checkCalc = (U32)xxh64_digest(&dctx->xxhState);
-		U32 checkRead;
-		if (remainingSize < 4)
-			return ERROR(checksum_wrong);
-		checkRead = ZSTD_readLE32(ip);
-		if (checkRead != checkCalc)
-			return ERROR(checksum_wrong);
-		ip += 4;
-		remainingSize -= 4;
-	}
-
-	/* Allow caller to get size read */
-	*srcPtr = ip;
-	*srcSizePtr = remainingSize;
-	return op - ostart;
-}
-
-static const void *ZSTD_DDictDictContent(const ZSTD_DDict *ddict);
-static size_t ZSTD_DDictDictSize(const ZSTD_DDict *ddict);
-
-static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize,
-					const ZSTD_DDict *ddict)
-{
-	void *const dststart = dst;
-
-	if (ddict) {
-		if (dict) {
-			/* programmer error, these two cases should be mutually exclusive */
-			return ERROR(GENERIC);
-		}
-
-		dict = ZSTD_DDictDictContent(ddict);
-		dictSize = ZSTD_DDictDictSize(ddict);
-	}
-
-	while (srcSize >= ZSTD_frameHeaderSize_prefix) {
-		U32 magicNumber;
-
-		magicNumber = ZSTD_readLE32(src);
-		if (magicNumber != ZSTD_MAGICNUMBER) {
-			if ((magicNumber & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
-				size_t skippableSize;
-				if (srcSize < ZSTD_skippableHeaderSize)
-					return ERROR(srcSize_wrong);
-				skippableSize = ZSTD_readLE32((const BYTE *)src + 4) + ZSTD_skippableHeaderSize;
-				if (srcSize < skippableSize) {
-					return ERROR(srcSize_wrong);
-				}
-
-				src = (const BYTE *)src + skippableSize;
-				srcSize -= skippableSize;
-				continue;
-			} else {
-				return ERROR(prefix_unknown);
-			}
-		}
-
-		if (ddict) {
-			/* we were called from ZSTD_decompress_usingDDict */
-			ZSTD_refDDict(dctx, ddict);
-		} else {
-			/* this will initialize correctly with no dict if dict == NULL, so
-			 * use this in all cases but ddict */
-			CHECK_F(ZSTD_decompressBegin_usingDict(dctx, dict, dictSize));
-		}
-		ZSTD_checkContinuity(dctx, dst);
-
-		{
-			const size_t res = ZSTD_decompressFrame(dctx, dst, dstCapacity, &src, &srcSize);
-			if (ZSTD_isError(res))
-				return res;
-			/* don't need to bounds check this, ZSTD_decompressFrame will have
-			 * already */
-			dst = (BYTE *)dst + res;
-			dstCapacity -= res;
-		}
-	}
-
-	if (srcSize)
-		return ERROR(srcSize_wrong); /* input not entirely consumed */
-
-	return (BYTE *)dst - (BYTE *)dststart;
-}
-
-size_t ZSTD_decompress_usingDict(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize)
-{
-	return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, dict, dictSize, NULL);
-}
-
-size_t ZSTD_decompressDCtx(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	return ZSTD_decompress_usingDict(dctx, dst, dstCapacity, src, srcSize, NULL, 0);
-}
-
-/*-**************************************
-*   Advanced Streaming Decompression API
-*   Bufferless and synchronous
-****************************************/
-size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx *dctx) { return dctx->expected; }
-
-ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx *dctx)
-{
-	switch (dctx->stage) {
-	default: /* should not happen */
-	case ZSTDds_getFrameHeaderSize:
-	case ZSTDds_decodeFrameHeader: return ZSTDnit_frameHeader;
-	case ZSTDds_decodeBlockHeader: return ZSTDnit_blockHeader;
-	case ZSTDds_decompressBlock: return ZSTDnit_block;
-	case ZSTDds_decompressLastBlock: return ZSTDnit_lastBlock;
-	case ZSTDds_checkChecksum: return ZSTDnit_checksum;
-	case ZSTDds_decodeSkippableHeader:
-	case ZSTDds_skipFrame: return ZSTDnit_skippableFrame;
-	}
-}
-
-int ZSTD_isSkipFrame(ZSTD_DCtx *dctx) { return dctx->stage == ZSTDds_skipFrame; } /* for zbuff */
-
-/** ZSTD_decompressContinue() :
-*   @return : nb of bytes generated into `dst` (necessarily <= `dstCapacity)
-*             or an error code, which can be tested using ZSTD_isError() */
-size_t ZSTD_decompressContinue(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	/* Sanity check */
-	if (srcSize != dctx->expected)
-		return ERROR(srcSize_wrong);
-	if (dstCapacity)
-		ZSTD_checkContinuity(dctx, dst);
-
-	switch (dctx->stage) {
-	case ZSTDds_getFrameHeaderSize:
-		if (srcSize != ZSTD_frameHeaderSize_prefix)
-			return ERROR(srcSize_wrong);					/* impossible */
-		if ((ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */
-			memcpy(dctx->headerBuffer, src, ZSTD_frameHeaderSize_prefix);
-			dctx->expected = ZSTD_skippableHeaderSize - ZSTD_frameHeaderSize_prefix; /* magic number + skippable frame length */
-			dctx->stage = ZSTDds_decodeSkippableHeader;
-			return 0;
-		}
-		dctx->headerSize = ZSTD_frameHeaderSize(src, ZSTD_frameHeaderSize_prefix);
-		if (ZSTD_isError(dctx->headerSize))
-			return dctx->headerSize;
-		memcpy(dctx->headerBuffer, src, ZSTD_frameHeaderSize_prefix);
-		if (dctx->headerSize > ZSTD_frameHeaderSize_prefix) {
-			dctx->expected = dctx->headerSize - ZSTD_frameHeaderSize_prefix;
-			dctx->stage = ZSTDds_decodeFrameHeader;
-			return 0;
-		}
-		dctx->expected = 0; /* not necessary to copy more */
-		fallthrough;
-
-	case ZSTDds_decodeFrameHeader:
-		memcpy(dctx->headerBuffer + ZSTD_frameHeaderSize_prefix, src, dctx->expected);
-		CHECK_F(ZSTD_decodeFrameHeader(dctx, dctx->headerBuffer, dctx->headerSize));
-		dctx->expected = ZSTD_blockHeaderSize;
-		dctx->stage = ZSTDds_decodeBlockHeader;
-		return 0;
-
-	case ZSTDds_decodeBlockHeader: {
-		blockProperties_t bp;
-		size_t const cBlockSize = ZSTD_getcBlockSize(src, ZSTD_blockHeaderSize, &bp);
-		if (ZSTD_isError(cBlockSize))
-			return cBlockSize;
-		dctx->expected = cBlockSize;
-		dctx->bType = bp.blockType;
-		dctx->rleSize = bp.origSize;
-		if (cBlockSize) {
-			dctx->stage = bp.lastBlock ? ZSTDds_decompressLastBlock : ZSTDds_decompressBlock;
-			return 0;
-		}
-		/* empty block */
-		if (bp.lastBlock) {
-			if (dctx->fParams.checksumFlag) {
-				dctx->expected = 4;
-				dctx->stage = ZSTDds_checkChecksum;
-			} else {
-				dctx->expected = 0; /* end of frame */
-				dctx->stage = ZSTDds_getFrameHeaderSize;
-			}
-		} else {
-			dctx->expected = 3; /* go directly to next header */
-			dctx->stage = ZSTDds_decodeBlockHeader;
-		}
-		return 0;
-	}
-	case ZSTDds_decompressLastBlock:
-	case ZSTDds_decompressBlock: {
-		size_t rSize;
-		switch (dctx->bType) {
-		case bt_compressed: rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize); break;
-		case bt_raw: rSize = ZSTD_copyRawBlock(dst, dstCapacity, src, srcSize); break;
-		case bt_rle: rSize = ZSTD_setRleBlock(dst, dstCapacity, src, srcSize, dctx->rleSize); break;
-		case bt_reserved: /* should never happen */
-		default: return ERROR(corruption_detected);
-		}
-		if (ZSTD_isError(rSize))
-			return rSize;
-		if (dctx->fParams.checksumFlag)
-			xxh64_update(&dctx->xxhState, dst, rSize);
-
-		if (dctx->stage == ZSTDds_decompressLastBlock) { /* end of frame */
-			if (dctx->fParams.checksumFlag) {	/* another round for frame checksum */
-				dctx->expected = 4;
-				dctx->stage = ZSTDds_checkChecksum;
-			} else {
-				dctx->expected = 0; /* ends here */
-				dctx->stage = ZSTDds_getFrameHeaderSize;
-			}
-		} else {
-			dctx->stage = ZSTDds_decodeBlockHeader;
-			dctx->expected = ZSTD_blockHeaderSize;
-			dctx->previousDstEnd = (char *)dst + rSize;
-		}
-		return rSize;
-	}
-	case ZSTDds_checkChecksum: {
-		U32 const h32 = (U32)xxh64_digest(&dctx->xxhState);
-		U32 const check32 = ZSTD_readLE32(src); /* srcSize == 4, guaranteed by dctx->expected */
-		if (check32 != h32)
-			return ERROR(checksum_wrong);
-		dctx->expected = 0;
-		dctx->stage = ZSTDds_getFrameHeaderSize;
-		return 0;
-	}
-	case ZSTDds_decodeSkippableHeader: {
-		memcpy(dctx->headerBuffer + ZSTD_frameHeaderSize_prefix, src, dctx->expected);
-		dctx->expected = ZSTD_readLE32(dctx->headerBuffer + 4);
-		dctx->stage = ZSTDds_skipFrame;
-		return 0;
-	}
-	case ZSTDds_skipFrame: {
-		dctx->expected = 0;
-		dctx->stage = ZSTDds_getFrameHeaderSize;
-		return 0;
-	}
-	default:
-		return ERROR(GENERIC); /* impossible */
-	}
-}
-
-static size_t ZSTD_refDictContent(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
-{
-	dctx->dictEnd = dctx->previousDstEnd;
-	dctx->vBase = (const char *)dict - ((const char *)(dctx->previousDstEnd) - (const char *)(dctx->base));
-	dctx->base = dict;
-	dctx->previousDstEnd = (const char *)dict + dictSize;
-	return 0;
-}
-
-/* ZSTD_loadEntropy() :
- * dict : must point at beginning of a valid zstd dictionary
- * @return : size of entropy tables read */
-static size_t ZSTD_loadEntropy(ZSTD_entropyTables_t *entropy, const void *const dict, size_t const dictSize)
-{
-	const BYTE *dictPtr = (const BYTE *)dict;
-	const BYTE *const dictEnd = dictPtr + dictSize;
-
-	if (dictSize <= 8)
-		return ERROR(dictionary_corrupted);
-	dictPtr += 8; /* skip header = magic + dictID */
-
-	{
-		size_t const hSize = HUF_readDTableX4_wksp(entropy->hufTable, dictPtr, dictEnd - dictPtr, entropy->workspace, sizeof(entropy->workspace));
-		if (HUF_isError(hSize))
-			return ERROR(dictionary_corrupted);
-		dictPtr += hSize;
-	}
-
-	{
-		short offcodeNCount[MaxOff + 1];
-		U32 offcodeMaxValue = MaxOff, offcodeLog;
-		size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(offcodeHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (offcodeLog > OffFSELog)
-			return ERROR(dictionary_corrupted);
-		CHECK_E(FSE_buildDTable_wksp(entropy->OFTable, offcodeNCount, offcodeMaxValue, offcodeLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
-		dictPtr += offcodeHeaderSize;
-	}
-
-	{
-		short matchlengthNCount[MaxML + 1];
-		unsigned matchlengthMaxValue = MaxML, matchlengthLog;
-		size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(matchlengthHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (matchlengthLog > MLFSELog)
-			return ERROR(dictionary_corrupted);
-		CHECK_E(FSE_buildDTable_wksp(entropy->MLTable, matchlengthNCount, matchlengthMaxValue, matchlengthLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
-		dictPtr += matchlengthHeaderSize;
-	}
-
-	{
-		short litlengthNCount[MaxLL + 1];
-		unsigned litlengthMaxValue = MaxLL, litlengthLog;
-		size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd - dictPtr);
-		if (FSE_isError(litlengthHeaderSize))
-			return ERROR(dictionary_corrupted);
-		if (litlengthLog > LLFSELog)
-			return ERROR(dictionary_corrupted);
-		CHECK_E(FSE_buildDTable_wksp(entropy->LLTable, litlengthNCount, litlengthMaxValue, litlengthLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
-		dictPtr += litlengthHeaderSize;
-	}
-
-	if (dictPtr + 12 > dictEnd)
-		return ERROR(dictionary_corrupted);
-	{
-		int i;
-		size_t const dictContentSize = (size_t)(dictEnd - (dictPtr + 12));
-		for (i = 0; i < 3; i++) {
-			U32 const rep = ZSTD_readLE32(dictPtr);
-			dictPtr += 4;
-			if (rep == 0 || rep >= dictContentSize)
-				return ERROR(dictionary_corrupted);
-			entropy->rep[i] = rep;
-		}
-	}
-
-	return dictPtr - (const BYTE *)dict;
-}
-
-static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
-{
-	if (dictSize < 8)
-		return ZSTD_refDictContent(dctx, dict, dictSize);
-	{
-		U32 const magic = ZSTD_readLE32(dict);
-		if (magic != ZSTD_DICT_MAGIC) {
-			return ZSTD_refDictContent(dctx, dict, dictSize); /* pure content mode */
-		}
-	}
-	dctx->dictID = ZSTD_readLE32((const char *)dict + 4);
-
-	/* load entropy tables */
-	{
-		size_t const eSize = ZSTD_loadEntropy(&dctx->entropy, dict, dictSize);
-		if (ZSTD_isError(eSize))
-			return ERROR(dictionary_corrupted);
-		dict = (const char *)dict + eSize;
-		dictSize -= eSize;
-	}
-	dctx->litEntropy = dctx->fseEntropy = 1;
-
-	/* reference dictionary content */
-	return ZSTD_refDictContent(dctx, dict, dictSize);
-}
-
-size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
-{
-	CHECK_F(ZSTD_decompressBegin(dctx));
-	if (dict && dictSize)
-		CHECK_E(ZSTD_decompress_insertDictionary(dctx, dict, dictSize), dictionary_corrupted);
-	return 0;
-}
-
-/* ======   ZSTD_DDict   ====== */
-
-struct ZSTD_DDict_s {
-	void *dictBuffer;
-	const void *dictContent;
-	size_t dictSize;
-	ZSTD_entropyTables_t entropy;
-	U32 dictID;
-	U32 entropyPresent;
-	ZSTD_customMem cMem;
-}; /* typedef'd to ZSTD_DDict within "zstd.h" */
-
-size_t ZSTD_DDictWorkspaceBound(void) { return ZSTD_ALIGN(sizeof(ZSTD_stack)) + ZSTD_ALIGN(sizeof(ZSTD_DDict)); }
-
-static const void *ZSTD_DDictDictContent(const ZSTD_DDict *ddict) { return ddict->dictContent; }
-
-static size_t ZSTD_DDictDictSize(const ZSTD_DDict *ddict) { return ddict->dictSize; }
-
-static void ZSTD_refDDict(ZSTD_DCtx *dstDCtx, const ZSTD_DDict *ddict)
-{
-	ZSTD_decompressBegin(dstDCtx); /* init */
-	if (ddict) {		       /* support refDDict on NULL */
-		dstDCtx->dictID = ddict->dictID;
-		dstDCtx->base = ddict->dictContent;
-		dstDCtx->vBase = ddict->dictContent;
-		dstDCtx->dictEnd = (const BYTE *)ddict->dictContent + ddict->dictSize;
-		dstDCtx->previousDstEnd = dstDCtx->dictEnd;
-		if (ddict->entropyPresent) {
-			dstDCtx->litEntropy = 1;
-			dstDCtx->fseEntropy = 1;
-			dstDCtx->LLTptr = ddict->entropy.LLTable;
-			dstDCtx->MLTptr = ddict->entropy.MLTable;
-			dstDCtx->OFTptr = ddict->entropy.OFTable;
-			dstDCtx->HUFptr = ddict->entropy.hufTable;
-			dstDCtx->entropy.rep[0] = ddict->entropy.rep[0];
-			dstDCtx->entropy.rep[1] = ddict->entropy.rep[1];
-			dstDCtx->entropy.rep[2] = ddict->entropy.rep[2];
-		} else {
-			dstDCtx->litEntropy = 0;
-			dstDCtx->fseEntropy = 0;
-		}
-	}
-}
-
-static size_t ZSTD_loadEntropy_inDDict(ZSTD_DDict *ddict)
-{
-	ddict->dictID = 0;
-	ddict->entropyPresent = 0;
-	if (ddict->dictSize < 8)
-		return 0;
-	{
-		U32 const magic = ZSTD_readLE32(ddict->dictContent);
-		if (magic != ZSTD_DICT_MAGIC)
-			return 0; /* pure content mode */
-	}
-	ddict->dictID = ZSTD_readLE32((const char *)ddict->dictContent + 4);
-
-	/* load entropy tables */
-	CHECK_E(ZSTD_loadEntropy(&ddict->entropy, ddict->dictContent, ddict->dictSize), dictionary_corrupted);
-	ddict->entropyPresent = 1;
-	return 0;
-}
-
-static ZSTD_DDict *ZSTD_createDDict_advanced(const void *dict, size_t dictSize, unsigned byReference, ZSTD_customMem customMem)
-{
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-
-	{
-		ZSTD_DDict *const ddict = (ZSTD_DDict *)ZSTD_malloc(sizeof(ZSTD_DDict), customMem);
-		if (!ddict)
-			return NULL;
-		ddict->cMem = customMem;
-
-		if ((byReference) || (!dict) || (!dictSize)) {
-			ddict->dictBuffer = NULL;
-			ddict->dictContent = dict;
-		} else {
-			void *const internalBuffer = ZSTD_malloc(dictSize, customMem);
-			if (!internalBuffer) {
-				ZSTD_freeDDict(ddict);
-				return NULL;
-			}
-			memcpy(internalBuffer, dict, dictSize);
-			ddict->dictBuffer = internalBuffer;
-			ddict->dictContent = internalBuffer;
-		}
-		ddict->dictSize = dictSize;
-		ddict->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001); /* cover both little and big endian */
-		/* parse dictionary content */
-		{
-			size_t const errorCode = ZSTD_loadEntropy_inDDict(ddict);
-			if (ZSTD_isError(errorCode)) {
-				ZSTD_freeDDict(ddict);
-				return NULL;
-			}
-		}
-
-		return ddict;
-	}
-}
-
-/*! ZSTD_initDDict() :
-*   Create a digested dictionary, to start decompression without startup delay.
-*   `dict` content is copied inside DDict.
-*   Consequently, `dict` can be released after `ZSTD_DDict` creation */
-ZSTD_DDict *ZSTD_initDDict(const void *dict, size_t dictSize, void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	return ZSTD_createDDict_advanced(dict, dictSize, 1, stackMem);
-}
-
-size_t ZSTD_freeDDict(ZSTD_DDict *ddict)
-{
-	if (ddict == NULL)
-		return 0; /* support free on NULL */
-	{
-		ZSTD_customMem const cMem = ddict->cMem;
-		ZSTD_free(ddict->dictBuffer, cMem);
-		ZSTD_free(ddict, cMem);
-		return 0;
-	}
-}
-
-/*! ZSTD_getDictID_fromDict() :
- *  Provides the dictID stored within dictionary.
- *  if @return == 0, the dictionary is not conformant with Zstandard specification.
- *  It can still be loaded, but as a content-only dictionary. */
-unsigned ZSTD_getDictID_fromDict(const void *dict, size_t dictSize)
-{
-	if (dictSize < 8)
-		return 0;
-	if (ZSTD_readLE32(dict) != ZSTD_DICT_MAGIC)
-		return 0;
-	return ZSTD_readLE32((const char *)dict + 4);
-}
-
-/*! ZSTD_getDictID_fromDDict() :
- *  Provides the dictID of the dictionary loaded into `ddict`.
- *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
- *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
-unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict *ddict)
-{
-	if (ddict == NULL)
-		return 0;
-	return ZSTD_getDictID_fromDict(ddict->dictContent, ddict->dictSize);
-}
-
-/*! ZSTD_getDictID_fromFrame() :
- *  Provides the dictID required to decompressed the frame stored within `src`.
- *  If @return == 0, the dictID could not be decoded.
- *  This could for one of the following reasons :
- *  - The frame does not require a dictionary to be decoded (most common case).
- *  - The frame was built with dictID intentionally removed. Whatever dictionary is necessary is a hidden information.
- *    Note : this use case also happens when using a non-conformant dictionary.
- *  - `srcSize` is too small, and as a result, the frame header could not be decoded (only possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`).
- *  - This is not a Zstandard frame.
- *  When identifying the exact failure cause, it's possible to used ZSTD_getFrameParams(), which will provide a more precise error code. */
-unsigned ZSTD_getDictID_fromFrame(const void *src, size_t srcSize)
-{
-	ZSTD_frameParams zfp = {0, 0, 0, 0};
-	size_t const hError = ZSTD_getFrameParams(&zfp, src, srcSize);
-	if (ZSTD_isError(hError))
-		return 0;
-	return zfp.dictID;
-}
-
-/*! ZSTD_decompress_usingDDict() :
-*   Decompression using a pre-digested Dictionary
-*   Use dictionary without significant overhead. */
-size_t ZSTD_decompress_usingDDict(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const ZSTD_DDict *ddict)
-{
-	/* pass content and size in case legacy frames are encountered */
-	return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, NULL, 0, ddict);
-}
-
-/*=====================================
-*   Streaming decompression
-*====================================*/
-
-typedef enum { zdss_init, zdss_loadHeader, zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;
-
-/* *** Resource management *** */
-struct ZSTD_DStream_s {
-	ZSTD_DCtx *dctx;
-	ZSTD_DDict *ddictLocal;
-	const ZSTD_DDict *ddict;
-	ZSTD_frameParams fParams;
-	ZSTD_dStreamStage stage;
-	char *inBuff;
-	size_t inBuffSize;
-	size_t inPos;
-	size_t maxWindowSize;
-	char *outBuff;
-	size_t outBuffSize;
-	size_t outStart;
-	size_t outEnd;
-	size_t blockSize;
-	BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX]; /* tmp buffer to store frame header */
-	size_t lhSize;
-	ZSTD_customMem customMem;
-	void *legacyContext;
-	U32 previousLegacyVersion;
-	U32 legacyVersion;
-	U32 hostageByte;
-}; /* typedef'd to ZSTD_DStream within "zstd.h" */
-
-size_t ZSTD_DStreamWorkspaceBound(size_t maxWindowSize)
-{
-	size_t const blockSize = MIN(maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
-	size_t const inBuffSize = blockSize;
-	size_t const outBuffSize = maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
-	return ZSTD_DCtxWorkspaceBound() + ZSTD_ALIGN(sizeof(ZSTD_DStream)) + ZSTD_ALIGN(inBuffSize) + ZSTD_ALIGN(outBuffSize);
-}
-
-static ZSTD_DStream *ZSTD_createDStream_advanced(ZSTD_customMem customMem)
-{
-	ZSTD_DStream *zds;
-
-	if (!customMem.customAlloc || !customMem.customFree)
-		return NULL;
-
-	zds = (ZSTD_DStream *)ZSTD_malloc(sizeof(ZSTD_DStream), customMem);
-	if (zds == NULL)
-		return NULL;
-	memset(zds, 0, sizeof(ZSTD_DStream));
-	memcpy(&zds->customMem, &customMem, sizeof(ZSTD_customMem));
-	zds->dctx = ZSTD_createDCtx_advanced(customMem);
-	if (zds->dctx == NULL) {
-		ZSTD_freeDStream(zds);
-		return NULL;
-	}
-	zds->stage = zdss_init;
-	zds->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT;
-	return zds;
-}
-
-ZSTD_DStream *ZSTD_initDStream(size_t maxWindowSize, void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
-	ZSTD_DStream *zds = ZSTD_createDStream_advanced(stackMem);
-	if (!zds) {
-		return NULL;
-	}
-
-	zds->maxWindowSize = maxWindowSize;
-	zds->stage = zdss_loadHeader;
-	zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
-	ZSTD_freeDDict(zds->ddictLocal);
-	zds->ddictLocal = NULL;
-	zds->ddict = zds->ddictLocal;
-	zds->legacyVersion = 0;
-	zds->hostageByte = 0;
-
-	{
-		size_t const blockSize = MIN(zds->maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
-		size_t const neededOutSize = zds->maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
-
-		zds->inBuff = (char *)ZSTD_malloc(blockSize, zds->customMem);
-		zds->inBuffSize = blockSize;
-		zds->outBuff = (char *)ZSTD_malloc(neededOutSize, zds->customMem);
-		zds->outBuffSize = neededOutSize;
-		if (zds->inBuff == NULL || zds->outBuff == NULL) {
-			ZSTD_freeDStream(zds);
-			return NULL;
-		}
-	}
-	return zds;
-}
-
-ZSTD_DStream *ZSTD_initDStream_usingDDict(size_t maxWindowSize, const ZSTD_DDict *ddict, void *workspace, size_t workspaceSize)
-{
-	ZSTD_DStream *zds = ZSTD_initDStream(maxWindowSize, workspace, workspaceSize);
-	if (zds) {
-		zds->ddict = ddict;
-	}
-	return zds;
-}
-
-size_t ZSTD_freeDStream(ZSTD_DStream *zds)
-{
-	if (zds == NULL)
-		return 0; /* support free on null */
-	{
-		ZSTD_customMem const cMem = zds->customMem;
-		ZSTD_freeDCtx(zds->dctx);
-		zds->dctx = NULL;
-		ZSTD_freeDDict(zds->ddictLocal);
-		zds->ddictLocal = NULL;
-		ZSTD_free(zds->inBuff, cMem);
-		zds->inBuff = NULL;
-		ZSTD_free(zds->outBuff, cMem);
-		zds->outBuff = NULL;
-		ZSTD_free(zds, cMem);
-		return 0;
-	}
-}
-
-/* *** Initialization *** */
-
-size_t ZSTD_DStreamInSize(void) { return ZSTD_BLOCKSIZE_ABSOLUTEMAX + ZSTD_blockHeaderSize; }
-size_t ZSTD_DStreamOutSize(void) { return ZSTD_BLOCKSIZE_ABSOLUTEMAX; }
-
-size_t ZSTD_resetDStream(ZSTD_DStream *zds)
-{
-	zds->stage = zdss_loadHeader;
-	zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
-	zds->legacyVersion = 0;
-	zds->hostageByte = 0;
-	return ZSTD_frameHeaderSize_prefix;
-}
-
-/* *****   Decompression   ***** */
-
-ZSTD_STATIC size_t ZSTD_limitCopy(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
-{
-	size_t const length = MIN(dstCapacity, srcSize);
-	memcpy(dst, src, length);
-	return length;
-}
-
-size_t ZSTD_decompressStream(ZSTD_DStream *zds, ZSTD_outBuffer *output, ZSTD_inBuffer *input)
-{
-	const char *const istart = (const char *)(input->src) + input->pos;
-	const char *const iend = (const char *)(input->src) + input->size;
-	const char *ip = istart;
-	char *const ostart = (char *)(output->dst) + output->pos;
-	char *const oend = (char *)(output->dst) + output->size;
-	char *op = ostart;
-	U32 someMoreWork = 1;
-
-	while (someMoreWork) {
-		switch (zds->stage) {
-		case zdss_init:
-			ZSTD_resetDStream(zds); /* transparent reset on starting decoding a new frame */
-			fallthrough;
-
-		case zdss_loadHeader: {
-			size_t const hSize = ZSTD_getFrameParams(&zds->fParams, zds->headerBuffer, zds->lhSize);
-			if (ZSTD_isError(hSize))
-				return hSize;
-			if (hSize != 0) {				   /* need more input */
-				size_t const toLoad = hSize - zds->lhSize; /* if hSize!=0, hSize > zds->lhSize */
-				if (toLoad > (size_t)(iend - ip)) {	/* not enough input to load full header */
-					memcpy(zds->headerBuffer + zds->lhSize, ip, iend - ip);
-					zds->lhSize += iend - ip;
-					input->pos = input->size;
-					return (MAX(ZSTD_frameHeaderSize_min, hSize) - zds->lhSize) +
-					       ZSTD_blockHeaderSize; /* remaining header bytes + next block header */
-				}
-				memcpy(zds->headerBuffer + zds->lhSize, ip, toLoad);
-				zds->lhSize = hSize;
-				ip += toLoad;
-				break;
-			}
-
-			/* check for single-pass mode opportunity */
-			if (zds->fParams.frameContentSize && zds->fParams.windowSize /* skippable frame if == 0 */
-			    && (U64)(size_t)(oend - op) >= zds->fParams.frameContentSize) {
-				size_t const cSize = ZSTD_findFrameCompressedSize(istart, iend - istart);
-				if (cSize <= (size_t)(iend - istart)) {
-					size_t const decompressedSize = ZSTD_decompress_usingDDict(zds->dctx, op, oend - op, istart, cSize, zds->ddict);
-					if (ZSTD_isError(decompressedSize))
-						return decompressedSize;
-					ip = istart + cSize;
-					op += decompressedSize;
-					zds->dctx->expected = 0;
-					zds->stage = zdss_init;
-					someMoreWork = 0;
-					break;
-				}
-			}
-
-			/* Consume header */
-			ZSTD_refDDict(zds->dctx, zds->ddict);
-			{
-				size_t const h1Size = ZSTD_nextSrcSizeToDecompress(zds->dctx); /* == ZSTD_frameHeaderSize_prefix */
-				CHECK_F(ZSTD_decompressContinue(zds->dctx, NULL, 0, zds->headerBuffer, h1Size));
-				{
-					size_t const h2Size = ZSTD_nextSrcSizeToDecompress(zds->dctx);
-					CHECK_F(ZSTD_decompressContinue(zds->dctx, NULL, 0, zds->headerBuffer + h1Size, h2Size));
-				}
-			}
-
-			zds->fParams.windowSize = MAX(zds->fParams.windowSize, 1U << ZSTD_WINDOWLOG_ABSOLUTEMIN);
-			if (zds->fParams.windowSize > zds->maxWindowSize)
-				return ERROR(frameParameter_windowTooLarge);
-
-			/* Buffers are preallocated, but double check */
-			{
-				size_t const blockSize = MIN(zds->maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
-				size_t const neededOutSize = zds->maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
-				if (zds->inBuffSize < blockSize) {
-					return ERROR(GENERIC);
-				}
-				if (zds->outBuffSize < neededOutSize) {
-					return ERROR(GENERIC);
-				}
-				zds->blockSize = blockSize;
-			}
-			zds->stage = zdss_read;
-		}
-			fallthrough;
-
-		case zdss_read: {
-			size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds->dctx);
-			if (neededInSize == 0) { /* end of frame */
-				zds->stage = zdss_init;
-				someMoreWork = 0;
-				break;
-			}
-			if ((size_t)(iend - ip) >= neededInSize) { /* decode directly from src */
-				const int isSkipFrame = ZSTD_isSkipFrame(zds->dctx);
-				size_t const decodedSize = ZSTD_decompressContinue(zds->dctx, zds->outBuff + zds->outStart,
-										   (isSkipFrame ? 0 : zds->outBuffSize - zds->outStart), ip, neededInSize);
-				if (ZSTD_isError(decodedSize))
-					return decodedSize;
-				ip += neededInSize;
-				if (!decodedSize && !isSkipFrame)
-					break; /* this was just a header */
-				zds->outEnd = zds->outStart + decodedSize;
-				zds->stage = zdss_flush;
-				break;
-			}
-			if (ip == iend) {
-				someMoreWork = 0;
-				break;
-			} /* no more input */
-			zds->stage = zdss_load;
-			/* pass-through */
-		}
-			fallthrough;
-
-		case zdss_load: {
-			size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds->dctx);
-			size_t const toLoad = neededInSize - zds->inPos; /* should always be <= remaining space within inBuff */
-			size_t loadedSize;
-			if (toLoad > zds->inBuffSize - zds->inPos)
-				return ERROR(corruption_detected); /* should never happen */
-			loadedSize = ZSTD_limitCopy(zds->inBuff + zds->inPos, toLoad, ip, iend - ip);
-			ip += loadedSize;
-			zds->inPos += loadedSize;
-			if (loadedSize < toLoad) {
-				someMoreWork = 0;
-				break;
-			} /* not enough input, wait for more */
-
-			/* decode loaded input */
-			{
-				const int isSkipFrame = ZSTD_isSkipFrame(zds->dctx);
-				size_t const decodedSize = ZSTD_decompressContinue(zds->dctx, zds->outBuff + zds->outStart, zds->outBuffSize - zds->outStart,
-										   zds->inBuff, neededInSize);
-				if (ZSTD_isError(decodedSize))
-					return decodedSize;
-				zds->inPos = 0; /* input is consumed */
-				if (!decodedSize && !isSkipFrame) {
-					zds->stage = zdss_read;
-					break;
-				} /* this was just a header */
-				zds->outEnd = zds->outStart + decodedSize;
-				zds->stage = zdss_flush;
-				/* pass-through */
-			}
-		}
-			fallthrough;
-
-		case zdss_flush: {
-			size_t const toFlushSize = zds->outEnd - zds->outStart;
-			size_t const flushedSize = ZSTD_limitCopy(op, oend - op, zds->outBuff + zds->outStart, toFlushSize);
-			op += flushedSize;
-			zds->outStart += flushedSize;
-			if (flushedSize == toFlushSize) { /* flush completed */
-				zds->stage = zdss_read;
-				if (zds->outStart + zds->blockSize > zds->outBuffSize)
-					zds->outStart = zds->outEnd = 0;
-				break;
-			}
-			/* cannot complete flush */
-			someMoreWork = 0;
-			break;
-		}
-		default:
-			return ERROR(GENERIC); /* impossible */
-		}
-	}
-
-	/* result */
-	input->pos += (size_t)(ip - istart);
-	output->pos += (size_t)(op - ostart);
-	{
-		size_t nextSrcSizeHint = ZSTD_nextSrcSizeToDecompress(zds->dctx);
-		if (!nextSrcSizeHint) {			    /* frame fully decoded */
-			if (zds->outEnd == zds->outStart) { /* output fully flushed */
-				if (zds->hostageByte) {
-					if (input->pos >= input->size) {
-						zds->stage = zdss_read;
-						return 1;
-					}	     /* can't release hostage (not present) */
-					input->pos++; /* release hostage */
-				}
-				return 0;
-			}
-			if (!zds->hostageByte) { /* output not fully flushed; keep last byte as hostage; will be released when all output is flushed */
-				input->pos--;    /* note : pos > 0, otherwise, impossible to finish reading last block */
-				zds->hostageByte = 1;
-			}
-			return 1;
-		}
-		nextSrcSizeHint += ZSTD_blockHeaderSize * (ZSTD_nextInputType(zds->dctx) == ZSTDnit_block); /* preload header of next block */
-		if (zds->inPos > nextSrcSizeHint)
-			return ERROR(GENERIC); /* should never happen */
-		nextSrcSizeHint -= zds->inPos; /* already loaded*/
-		return nextSrcSizeHint;
-	}
-}
-
-EXPORT_SYMBOL(ZSTD_DCtxWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initDCtx);
-EXPORT_SYMBOL(ZSTD_decompressDCtx);
-EXPORT_SYMBOL(ZSTD_decompress_usingDict);
-
-EXPORT_SYMBOL(ZSTD_DDictWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initDDict);
-EXPORT_SYMBOL(ZSTD_decompress_usingDDict);
-
-EXPORT_SYMBOL(ZSTD_DStreamWorkspaceBound);
-EXPORT_SYMBOL(ZSTD_initDStream);
-EXPORT_SYMBOL(ZSTD_initDStream_usingDDict);
-EXPORT_SYMBOL(ZSTD_resetDStream);
-EXPORT_SYMBOL(ZSTD_decompressStream);
-EXPORT_SYMBOL(ZSTD_DStreamInSize);
-EXPORT_SYMBOL(ZSTD_DStreamOutSize);
-
-EXPORT_SYMBOL(ZSTD_findFrameCompressedSize);
-EXPORT_SYMBOL(ZSTD_getFrameContentSize);
-EXPORT_SYMBOL(ZSTD_findDecompressedSize);
-
-EXPORT_SYMBOL(ZSTD_isFrame);
-EXPORT_SYMBOL(ZSTD_getDictID_fromDict);
-EXPORT_SYMBOL(ZSTD_getDictID_fromDDict);
-EXPORT_SYMBOL(ZSTD_getDictID_fromFrame);
-
-EXPORT_SYMBOL(ZSTD_getFrameParams);
-EXPORT_SYMBOL(ZSTD_decompressBegin);
-EXPORT_SYMBOL(ZSTD_decompressBegin_usingDict);
-EXPORT_SYMBOL(ZSTD_copyDCtx);
-EXPORT_SYMBOL(ZSTD_nextSrcSizeToDecompress);
-EXPORT_SYMBOL(ZSTD_decompressContinue);
-EXPORT_SYMBOL(ZSTD_nextInputType);
-
-EXPORT_SYMBOL(ZSTD_decompressBlock);
-EXPORT_SYMBOL(ZSTD_insertBlock);
-
-MODULE_LICENSE("Dual BSD/GPL");
-MODULE_DESCRIPTION("Zstd Decompressor");
diff --git a/lib/zstd/decompress/huf_decompress.c b/lib/zstd/decompress/huf_decompress.c
new file mode 100644
index 0000000..5105e59
--- /dev/null
+++ b/lib/zstd/decompress/huf_decompress.c
@@ -0,0 +1,1206 @@
+/* ******************************************************************
+ * huff0 huffman decoder,
+ * part of Finite State Entropy library
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ *
+ *  You can contact the author at :
+ *  - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+****************************************************************** */
+
+/* **************************************************************
+*  Dependencies
+****************************************************************/
+#include "../common/zstd_deps.h"  /* ZSTD_memcpy, ZSTD_memset */
+#include "../common/compiler.h"
+#include "../common/bitstream.h"  /* BIT_* */
+#include "../common/fse.h"        /* to compress headers */
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include "../common/error_private.h"
+
+/* **************************************************************
+*  Macros
+****************************************************************/
+
+/* These two optional macros force the use one way or another of the two
+ * Huffman decompression implementations. You can't force in both directions
+ * at the same time.
+ */
+#if defined(HUF_FORCE_DECOMPRESS_X1) && \
+    defined(HUF_FORCE_DECOMPRESS_X2)
+#error "Cannot force the use of the X1 and X2 decoders at the same time!"
+#endif
+
+
+/* **************************************************************
+*  Error Management
+****************************************************************/
+#define HUF_isError ERR_isError
+
+
+/* **************************************************************
+*  Byte alignment for workSpace management
+****************************************************************/
+#define HUF_ALIGN(x, a)         HUF_ALIGN_MASK((x), (a) - 1)
+#define HUF_ALIGN_MASK(x, mask) (((x) + (mask)) & ~(mask))
+
+
+/* **************************************************************
+*  BMI2 Variant Wrappers
+****************************************************************/
+#if DYNAMIC_BMI2
+
+#define HUF_DGEN(fn)                                                        \
+                                                                            \
+    static size_t fn##_default(                                             \
+                  void* dst,  size_t dstSize,                               \
+            const void* cSrc, size_t cSrcSize,                              \
+            const HUF_DTable* DTable)                                       \
+    {                                                                       \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }                                                                       \
+                                                                            \
+    static TARGET_ATTRIBUTE("bmi2") size_t fn##_bmi2(                       \
+                  void* dst,  size_t dstSize,                               \
+            const void* cSrc, size_t cSrcSize,                              \
+            const HUF_DTable* DTable)                                       \
+    {                                                                       \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }                                                                       \
+                                                                            \
+    static size_t fn(void* dst, size_t dstSize, void const* cSrc,           \
+                     size_t cSrcSize, HUF_DTable const* DTable, int bmi2)   \
+    {                                                                       \
+        if (bmi2) {                                                         \
+            return fn##_bmi2(dst, dstSize, cSrc, cSrcSize, DTable);         \
+        }                                                                   \
+        return fn##_default(dst, dstSize, cSrc, cSrcSize, DTable);          \
+    }
+
+#else
+
+#define HUF_DGEN(fn)                                                        \
+    static size_t fn(void* dst, size_t dstSize, void const* cSrc,           \
+                     size_t cSrcSize, HUF_DTable const* DTable, int bmi2)   \
+    {                                                                       \
+        (void)bmi2;                                                         \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }
+
+#endif
+
+
+/*-***************************/
+/*  generic DTableDesc       */
+/*-***************************/
+typedef struct { BYTE maxTableLog; BYTE tableType; BYTE tableLog; BYTE reserved; } DTableDesc;
+
+static DTableDesc HUF_getDTableDesc(const HUF_DTable* table)
+{
+    DTableDesc dtd;
+    ZSTD_memcpy(&dtd, table, sizeof(dtd));
+    return dtd;
+}
+
+
+#ifndef HUF_FORCE_DECOMPRESS_X2
+
+/*-***************************/
+/*  single-symbol decoding   */
+/*-***************************/
+typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX1;   /* single-symbol decoding */
+
+/*
+ * Packs 4 HUF_DEltX1 structs into a U64. This is used to lay down 4 entries at
+ * a time.
+ */
+static U64 HUF_DEltX1_set4(BYTE symbol, BYTE nbBits) {
+    U64 D4;
+    if (MEM_isLittleEndian()) {
+        D4 = symbol + (nbBits << 8);
+    } else {
+        D4 = (symbol << 8) + nbBits;
+    }
+    D4 *= 0x0001000100010001ULL;
+    return D4;
+}
+
+typedef struct {
+        U32 rankVal[HUF_TABLELOG_ABSOLUTEMAX + 1];
+        U32 rankStart[HUF_TABLELOG_ABSOLUTEMAX + 1];
+        U32 statsWksp[HUF_READ_STATS_WORKSPACE_SIZE_U32];
+        BYTE symbols[HUF_SYMBOLVALUE_MAX + 1];
+        BYTE huffWeight[HUF_SYMBOLVALUE_MAX + 1];
+} HUF_ReadDTableX1_Workspace;
+
+
+size_t HUF_readDTableX1_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize)
+{
+    return HUF_readDTableX1_wksp_bmi2(DTable, src, srcSize, workSpace, wkspSize, /* bmi2 */ 0);
+}
+
+size_t HUF_readDTableX1_wksp_bmi2(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int bmi2)
+{
+    U32 tableLog = 0;
+    U32 nbSymbols = 0;
+    size_t iSize;
+    void* const dtPtr = DTable + 1;
+    HUF_DEltX1* const dt = (HUF_DEltX1*)dtPtr;
+    HUF_ReadDTableX1_Workspace* wksp = (HUF_ReadDTableX1_Workspace*)workSpace;
+
+    DEBUG_STATIC_ASSERT(HUF_DECOMPRESS_WORKSPACE_SIZE >= sizeof(*wksp));
+    if (sizeof(*wksp) > wkspSize) return ERROR(tableLog_tooLarge);
+
+    DEBUG_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable));
+    /* ZSTD_memset(huffWeight, 0, sizeof(huffWeight)); */   /* is not necessary, even though some analyzer complain ... */
+
+    iSize = HUF_readStats_wksp(wksp->huffWeight, HUF_SYMBOLVALUE_MAX + 1, wksp->rankVal, &nbSymbols, &tableLog, src, srcSize, wksp->statsWksp, sizeof(wksp->statsWksp), bmi2);
+    if (HUF_isError(iSize)) return iSize;
+
+    /* Table header */
+    {   DTableDesc dtd = HUF_getDTableDesc(DTable);
+        if (tableLog > (U32)(dtd.maxTableLog+1)) return ERROR(tableLog_tooLarge);   /* DTable too small, Huffman tree cannot fit in */
+        dtd.tableType = 0;
+        dtd.tableLog = (BYTE)tableLog;
+        ZSTD_memcpy(DTable, &dtd, sizeof(dtd));
+    }
+
+    /* Compute symbols and rankStart given rankVal:
+     *
+     * rankVal already contains the number of values of each weight.
+     *
+     * symbols contains the symbols ordered by weight. First are the rankVal[0]
+     * weight 0 symbols, followed by the rankVal[1] weight 1 symbols, and so on.
+     * symbols[0] is filled (but unused) to avoid a branch.
+     *
+     * rankStart contains the offset where each rank belongs in the DTable.
+     * rankStart[0] is not filled because there are no entries in the table for
+     * weight 0.
+     */
+    {
+        int n;
+        int nextRankStart = 0;
+        int const unroll = 4;
+        int const nLimit = (int)nbSymbols - unroll + 1;
+        for (n=0; n<(int)tableLog+1; n++) {
+            U32 const curr = nextRankStart;
+            nextRankStart += wksp->rankVal[n];
+            wksp->rankStart[n] = curr;
+        }
+        for (n=0; n < nLimit; n += unroll) {
+            int u;
+            for (u=0; u < unroll; ++u) {
+                size_t const w = wksp->huffWeight[n+u];
+                wksp->symbols[wksp->rankStart[w]++] = (BYTE)(n+u);
+            }
+        }
+        for (; n < (int)nbSymbols; ++n) {
+            size_t const w = wksp->huffWeight[n];
+            wksp->symbols[wksp->rankStart[w]++] = (BYTE)n;
+        }
+    }
+
+    /* fill DTable
+     * We fill all entries of each weight in order.
+     * That way length is a constant for each iteration of the outter loop.
+     * We can switch based on the length to a different inner loop which is
+     * optimized for that particular case.
+     */
+    {
+        U32 w;
+        int symbol=wksp->rankVal[0];
+        int rankStart=0;
+        for (w=1; w<tableLog+1; ++w) {
+            int const symbolCount = wksp->rankVal[w];
+            int const length = (1 << w) >> 1;
+            int uStart = rankStart;
+            BYTE const nbBits = (BYTE)(tableLog + 1 - w);
+            int s;
+            int u;
+            switch (length) {
+            case 1:
+                for (s=0; s<symbolCount; ++s) {
+                    HUF_DEltX1 D;
+                    D.byte = wksp->symbols[symbol + s];
+                    D.nbBits = nbBits;
+                    dt[uStart] = D;
+                    uStart += 1;
+                }
+                break;
+            case 2:
+                for (s=0; s<symbolCount; ++s) {
+                    HUF_DEltX1 D;
+                    D.byte = wksp->symbols[symbol + s];
+                    D.nbBits = nbBits;
+                    dt[uStart+0] = D;
+                    dt[uStart+1] = D;
+                    uStart += 2;
+                }
+                break;
+            case 4:
+                for (s=0; s<symbolCount; ++s) {
+                    U64 const D4 = HUF_DEltX1_set4(wksp->symbols[symbol + s], nbBits);
+                    MEM_write64(dt + uStart, D4);
+                    uStart += 4;
+                }
+                break;
+            case 8:
+                for (s=0; s<symbolCount; ++s) {
+                    U64 const D4 = HUF_DEltX1_set4(wksp->symbols[symbol + s], nbBits);
+                    MEM_write64(dt + uStart, D4);
+                    MEM_write64(dt + uStart + 4, D4);
+                    uStart += 8;
+                }
+                break;
+            default:
+                for (s=0; s<symbolCount; ++s) {
+                    U64 const D4 = HUF_DEltX1_set4(wksp->symbols[symbol + s], nbBits);
+                    for (u=0; u < length; u += 16) {
+                        MEM_write64(dt + uStart + u + 0, D4);
+                        MEM_write64(dt + uStart + u + 4, D4);
+                        MEM_write64(dt + uStart + u + 8, D4);
+                        MEM_write64(dt + uStart + u + 12, D4);
+                    }
+                    assert(u == length);
+                    uStart += length;
+                }
+                break;
+            }
+            symbol += symbolCount;
+            rankStart += symbolCount * length;
+        }
+    }
+    return iSize;
+}
+
+FORCE_INLINE_TEMPLATE BYTE
+HUF_decodeSymbolX1(BIT_DStream_t* Dstream, const HUF_DEltX1* dt, const U32 dtLog)
+{
+    size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
+    BYTE const c = dt[val].byte;
+    BIT_skipBits(Dstream, dt[val].nbBits);
+    return c;
+}
+
+#define HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr) \
+    *ptr++ = HUF_decodeSymbolX1(DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX1_1(ptr, DStreamPtr)  \
+    if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
+        HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr)
+
+#define HUF_DECODE_SYMBOLX1_2(ptr, DStreamPtr) \
+    if (MEM_64bits()) \
+        HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr)
+
+HINT_INLINE size_t
+HUF_decodeStreamX1(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX1* const dt, const U32 dtLog)
+{
+    BYTE* const pStart = p;
+
+    /* up to 4 symbols at a time */
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-3)) {
+        HUF_DECODE_SYMBOLX1_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX1_1(p, bitDPtr);
+        HUF_DECODE_SYMBOLX1_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX1_0(p, bitDPtr);
+    }
+
+    /* [0-3] symbols remaining */
+    if (MEM_32bits())
+        while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd))
+            HUF_DECODE_SYMBOLX1_0(p, bitDPtr);
+
+    /* no more data to retrieve from bitstream, no need to reload */
+    while (p < pEnd)
+        HUF_DECODE_SYMBOLX1_0(p, bitDPtr);
+
+    return pEnd-pStart;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress1X1_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    BYTE* op = (BYTE*)dst;
+    BYTE* const oend = op + dstSize;
+    const void* dtPtr = DTable + 1;
+    const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr;
+    BIT_DStream_t bitD;
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+    U32 const dtLog = dtd.tableLog;
+
+    CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );
+
+    HUF_decodeStreamX1(op, &bitD, oend, dt, dtLog);
+
+    if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
+
+    return dstSize;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress4X1_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    /* Check */
+    if (cSrcSize < 10) return ERROR(corruption_detected);  /* strict minimum : jump table + 1 byte per stream */
+
+    {   const BYTE* const istart = (const BYTE*) cSrc;
+        BYTE* const ostart = (BYTE*) dst;
+        BYTE* const oend = ostart + dstSize;
+        BYTE* const olimit = oend - 3;
+        const void* const dtPtr = DTable + 1;
+        const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr;
+
+        /* Init */
+        BIT_DStream_t bitD1;
+        BIT_DStream_t bitD2;
+        BIT_DStream_t bitD3;
+        BIT_DStream_t bitD4;
+        size_t const length1 = MEM_readLE16(istart);
+        size_t const length2 = MEM_readLE16(istart+2);
+        size_t const length3 = MEM_readLE16(istart+4);
+        size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
+        const BYTE* const istart1 = istart + 6;  /* jumpTable */
+        const BYTE* const istart2 = istart1 + length1;
+        const BYTE* const istart3 = istart2 + length2;
+        const BYTE* const istart4 = istart3 + length3;
+        const size_t segmentSize = (dstSize+3) / 4;
+        BYTE* const opStart2 = ostart + segmentSize;
+        BYTE* const opStart3 = opStart2 + segmentSize;
+        BYTE* const opStart4 = opStart3 + segmentSize;
+        BYTE* op1 = ostart;
+        BYTE* op2 = opStart2;
+        BYTE* op3 = opStart3;
+        BYTE* op4 = opStart4;
+        DTableDesc const dtd = HUF_getDTableDesc(DTable);
+        U32 const dtLog = dtd.tableLog;
+        U32 endSignal = 1;
+
+        if (length4 > cSrcSize) return ERROR(corruption_detected);   /* overflow */
+        CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
+        CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
+        CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
+        CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );
+
+        /* up to 16 symbols per loop (4 symbols per stream) in 64-bit mode */
+        for ( ; (endSignal) & (op4 < olimit) ; ) {
+            HUF_DECODE_SYMBOLX1_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX1_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX1_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX1_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX1_1(op1, &bitD1);
+            HUF_DECODE_SYMBOLX1_1(op2, &bitD2);
+            HUF_DECODE_SYMBOLX1_1(op3, &bitD3);
+            HUF_DECODE_SYMBOLX1_1(op4, &bitD4);
+            HUF_DECODE_SYMBOLX1_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX1_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX1_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX1_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX1_0(op1, &bitD1);
+            HUF_DECODE_SYMBOLX1_0(op2, &bitD2);
+            HUF_DECODE_SYMBOLX1_0(op3, &bitD3);
+            HUF_DECODE_SYMBOLX1_0(op4, &bitD4);
+            endSignal &= BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished;
+        }
+
+        /* check corruption */
+        /* note : should not be necessary : op# advance in lock step, and we control op4.
+         *        but curiously, binary generated by gcc 7.2 & 7.3 with -mbmi2 runs faster when >=1 test is present */
+        if (op1 > opStart2) return ERROR(corruption_detected);
+        if (op2 > opStart3) return ERROR(corruption_detected);
+        if (op3 > opStart4) return ERROR(corruption_detected);
+        /* note : op4 supposed already verified within main loop */
+
+        /* finish bitStreams one by one */
+        HUF_decodeStreamX1(op1, &bitD1, opStart2, dt, dtLog);
+        HUF_decodeStreamX1(op2, &bitD2, opStart3, dt, dtLog);
+        HUF_decodeStreamX1(op3, &bitD3, opStart4, dt, dtLog);
+        HUF_decodeStreamX1(op4, &bitD4, oend,     dt, dtLog);
+
+        /* check */
+        { U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
+          if (!endCheck) return ERROR(corruption_detected); }
+
+        /* decoded size */
+        return dstSize;
+    }
+}
+
+
+typedef size_t (*HUF_decompress_usingDTable_t)(void *dst, size_t dstSize,
+                                               const void *cSrc,
+                                               size_t cSrcSize,
+                                               const HUF_DTable *DTable);
+
+HUF_DGEN(HUF_decompress1X1_usingDTable_internal)
+HUF_DGEN(HUF_decompress4X1_usingDTable_internal)
+
+
+
+size_t HUF_decompress1X1_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 0) return ERROR(GENERIC);
+    return HUF_decompress1X1_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t const hSize = HUF_readDTableX1_wksp(DCtx, cSrc, cSrcSize, workSpace, wkspSize);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
+}
+
+
+size_t HUF_decompress4X1_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 0) return ERROR(GENERIC);
+    return HUF_decompress4X1_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+static size_t HUF_decompress4X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize, int bmi2)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t const hSize = HUF_readDTableX1_wksp_bmi2(dctx, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress4X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
+}
+
+size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    return HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, 0);
+}
+
+
+#endif /* HUF_FORCE_DECOMPRESS_X2 */
+
+
+#ifndef HUF_FORCE_DECOMPRESS_X1
+
+/* *************************/
+/* double-symbols decoding */
+/* *************************/
+
+typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX2;  /* double-symbols decoding */
+typedef struct { BYTE symbol; BYTE weight; } sortedSymbol_t;
+typedef U32 rankValCol_t[HUF_TABLELOG_MAX + 1];
+typedef rankValCol_t rankVal_t[HUF_TABLELOG_MAX];
+
+
+/* HUF_fillDTableX2Level2() :
+ * `rankValOrigin` must be a table of at least (HUF_TABLELOG_MAX + 1) U32 */
+static void HUF_fillDTableX2Level2(HUF_DEltX2* DTable, U32 sizeLog, const U32 consumed,
+                           const U32* rankValOrigin, const int minWeight,
+                           const sortedSymbol_t* sortedSymbols, const U32 sortedListSize,
+                           U32 nbBitsBaseline, U16 baseSeq, U32* wksp, size_t wkspSize)
+{
+    HUF_DEltX2 DElt;
+    U32* rankVal = wksp;
+
+    assert(wkspSize >= HUF_TABLELOG_MAX + 1);
+    (void)wkspSize;
+    /* get pre-calculated rankVal */
+    ZSTD_memcpy(rankVal, rankValOrigin, sizeof(U32) * (HUF_TABLELOG_MAX + 1));
+
+    /* fill skipped values */
+    if (minWeight>1) {
+        U32 i, skipSize = rankVal[minWeight];
+        MEM_writeLE16(&(DElt.sequence), baseSeq);
+        DElt.nbBits   = (BYTE)(consumed);
+        DElt.length   = 1;
+        for (i = 0; i < skipSize; i++)
+            DTable[i] = DElt;
+    }
+
+    /* fill DTable */
+    {   U32 s; for (s=0; s<sortedListSize; s++) {   /* note : sortedSymbols already skipped */
+            const U32 symbol = sortedSymbols[s].symbol;
+            const U32 weight = sortedSymbols[s].weight;
+            const U32 nbBits = nbBitsBaseline - weight;
+            const U32 length = 1 << (sizeLog-nbBits);
+            const U32 start = rankVal[weight];
+            U32 i = start;
+            const U32 end = start + length;
+
+            MEM_writeLE16(&(DElt.sequence), (U16)(baseSeq + (symbol << 8)));
+            DElt.nbBits = (BYTE)(nbBits + consumed);
+            DElt.length = 2;
+            do { DTable[i++] = DElt; } while (i<end);   /* since length >= 1 */
+
+            rankVal[weight] += length;
+    }   }
+}
+
+
+static void HUF_fillDTableX2(HUF_DEltX2* DTable, const U32 targetLog,
+                           const sortedSymbol_t* sortedList, const U32 sortedListSize,
+                           const U32* rankStart, rankVal_t rankValOrigin, const U32 maxWeight,
+                           const U32 nbBitsBaseline, U32* wksp, size_t wkspSize)
+{
+    U32* rankVal = wksp;
+    const int scaleLog = nbBitsBaseline - targetLog;   /* note : targetLog >= srcLog, hence scaleLog <= 1 */
+    const U32 minBits  = nbBitsBaseline - maxWeight;
+    U32 s;
+
+    assert(wkspSize >= HUF_TABLELOG_MAX + 1);
+    wksp += HUF_TABLELOG_MAX + 1;
+    wkspSize -= HUF_TABLELOG_MAX + 1;
+
+    ZSTD_memcpy(rankVal, rankValOrigin, sizeof(U32) * (HUF_TABLELOG_MAX + 1));
+
+    /* fill DTable */
+    for (s=0; s<sortedListSize; s++) {
+        const U16 symbol = sortedList[s].symbol;
+        const U32 weight = sortedList[s].weight;
+        const U32 nbBits = nbBitsBaseline - weight;
+        const U32 start = rankVal[weight];
+        const U32 length = 1 << (targetLog-nbBits);
+
+        if (targetLog-nbBits >= minBits) {   /* enough room for a second symbol */
+            U32 sortedRank;
+            int minWeight = nbBits + scaleLog;
+            if (minWeight < 1) minWeight = 1;
+            sortedRank = rankStart[minWeight];
+            HUF_fillDTableX2Level2(DTable+start, targetLog-nbBits, nbBits,
+                           rankValOrigin[nbBits], minWeight,
+                           sortedList+sortedRank, sortedListSize-sortedRank,
+                           nbBitsBaseline, symbol, wksp, wkspSize);
+        } else {
+            HUF_DEltX2 DElt;
+            MEM_writeLE16(&(DElt.sequence), symbol);
+            DElt.nbBits = (BYTE)(nbBits);
+            DElt.length = 1;
+            {   U32 const end = start + length;
+                U32 u;
+                for (u = start; u < end; u++) DTable[u] = DElt;
+        }   }
+        rankVal[weight] += length;
+    }
+}
+
+typedef struct {
+    rankValCol_t rankVal[HUF_TABLELOG_MAX];
+    U32 rankStats[HUF_TABLELOG_MAX + 1];
+    U32 rankStart0[HUF_TABLELOG_MAX + 2];
+    sortedSymbol_t sortedSymbol[HUF_SYMBOLVALUE_MAX + 1];
+    BYTE weightList[HUF_SYMBOLVALUE_MAX + 1];
+    U32 calleeWksp[HUF_READ_STATS_WORKSPACE_SIZE_U32];
+} HUF_ReadDTableX2_Workspace;
+
+size_t HUF_readDTableX2_wksp(HUF_DTable* DTable,
+                       const void* src, size_t srcSize,
+                             void* workSpace, size_t wkspSize)
+{
+    U32 tableLog, maxW, sizeOfSort, nbSymbols;
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    U32 const maxTableLog = dtd.maxTableLog;
+    size_t iSize;
+    void* dtPtr = DTable+1;   /* force compiler to avoid strict-aliasing */
+    HUF_DEltX2* const dt = (HUF_DEltX2*)dtPtr;
+    U32 *rankStart;
+
+    HUF_ReadDTableX2_Workspace* const wksp = (HUF_ReadDTableX2_Workspace*)workSpace;
+
+    if (sizeof(*wksp) > wkspSize) return ERROR(GENERIC);
+
+    rankStart = wksp->rankStart0 + 1;
+    ZSTD_memset(wksp->rankStats, 0, sizeof(wksp->rankStats));
+    ZSTD_memset(wksp->rankStart0, 0, sizeof(wksp->rankStart0));
+
+    DEBUG_STATIC_ASSERT(sizeof(HUF_DEltX2) == sizeof(HUF_DTable));   /* if compiler fails here, assertion is wrong */
+    if (maxTableLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
+    /* ZSTD_memset(weightList, 0, sizeof(weightList)); */  /* is not necessary, even though some analyzer complain ... */
+
+    iSize = HUF_readStats_wksp(wksp->weightList, HUF_SYMBOLVALUE_MAX + 1, wksp->rankStats, &nbSymbols, &tableLog, src, srcSize, wksp->calleeWksp, sizeof(wksp->calleeWksp), /* bmi2 */ 0);
+    if (HUF_isError(iSize)) return iSize;
+
+    /* check result */
+    if (tableLog > maxTableLog) return ERROR(tableLog_tooLarge);   /* DTable can't fit code depth */
+
+    /* find maxWeight */
+    for (maxW = tableLog; wksp->rankStats[maxW]==0; maxW--) {}  /* necessarily finds a solution before 0 */
+
+    /* Get start index of each weight */
+    {   U32 w, nextRankStart = 0;
+        for (w=1; w<maxW+1; w++) {
+            U32 curr = nextRankStart;
+            nextRankStart += wksp->rankStats[w];
+            rankStart[w] = curr;
+        }
+        rankStart[0] = nextRankStart;   /* put all 0w symbols at the end of sorted list*/
+        sizeOfSort = nextRankStart;
+    }
+
+    /* sort symbols by weight */
+    {   U32 s;
+        for (s=0; s<nbSymbols; s++) {
+            U32 const w = wksp->weightList[s];
+            U32 const r = rankStart[w]++;
+            wksp->sortedSymbol[r].symbol = (BYTE)s;
+            wksp->sortedSymbol[r].weight = (BYTE)w;
+        }
+        rankStart[0] = 0;   /* forget 0w symbols; this is beginning of weight(1) */
+    }
+
+    /* Build rankVal */
+    {   U32* const rankVal0 = wksp->rankVal[0];
+        {   int const rescale = (maxTableLog-tableLog) - 1;   /* tableLog <= maxTableLog */
+            U32 nextRankVal = 0;
+            U32 w;
+            for (w=1; w<maxW+1; w++) {
+                U32 curr = nextRankVal;
+                nextRankVal += wksp->rankStats[w] << (w+rescale);
+                rankVal0[w] = curr;
+        }   }
+        {   U32 const minBits = tableLog+1 - maxW;
+            U32 consumed;
+            for (consumed = minBits; consumed < maxTableLog - minBits + 1; consumed++) {
+                U32* const rankValPtr = wksp->rankVal[consumed];
+                U32 w;
+                for (w = 1; w < maxW+1; w++) {
+                    rankValPtr[w] = rankVal0[w] >> consumed;
+    }   }   }   }
+
+    HUF_fillDTableX2(dt, maxTableLog,
+                   wksp->sortedSymbol, sizeOfSort,
+                   wksp->rankStart0, wksp->rankVal, maxW,
+                   tableLog+1,
+                   wksp->calleeWksp, sizeof(wksp->calleeWksp) / sizeof(U32));
+
+    dtd.tableLog = (BYTE)maxTableLog;
+    dtd.tableType = 1;
+    ZSTD_memcpy(DTable, &dtd, sizeof(dtd));
+    return iSize;
+}
+
+
+FORCE_INLINE_TEMPLATE U32
+HUF_decodeSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, const U32 dtLog)
+{
+    size_t const val = BIT_lookBitsFast(DStream, dtLog);   /* note : dtLog >= 1 */
+    ZSTD_memcpy(op, dt+val, 2);
+    BIT_skipBits(DStream, dt[val].nbBits);
+    return dt[val].length;
+}
+
+FORCE_INLINE_TEMPLATE U32
+HUF_decodeLastSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, const U32 dtLog)
+{
+    size_t const val = BIT_lookBitsFast(DStream, dtLog);   /* note : dtLog >= 1 */
+    ZSTD_memcpy(op, dt+val, 1);
+    if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits);
+    else {
+        if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) {
+            BIT_skipBits(DStream, dt[val].nbBits);
+            if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8))
+                /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
+                DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8);
+    }   }
+    return 1;
+}
+
+#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \
+    ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \
+    if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
+        ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
+    if (MEM_64bits()) \
+        ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+
+HINT_INLINE size_t
+HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd,
+                const HUF_DEltX2* const dt, const U32 dtLog)
+{
+    BYTE* const pStart = p;
+
+    /* up to 8 symbols at a time */
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-(sizeof(bitDPtr->bitContainer)-1))) {
+        HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
+        HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
+    }
+
+    /* closer to end : up to 2 symbols at a time */
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd-2))
+        HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
+
+    while (p <= pEnd-2)
+        HUF_DECODE_SYMBOLX2_0(p, bitDPtr);   /* no need to reload : reached the end of DStream */
+
+    if (p < pEnd)
+        p += HUF_decodeLastSymbolX2(p, bitDPtr, dt, dtLog);
+
+    return p-pStart;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress1X2_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    BIT_DStream_t bitD;
+
+    /* Init */
+    CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );
+
+    /* decode */
+    {   BYTE* const ostart = (BYTE*) dst;
+        BYTE* const oend = ostart + dstSize;
+        const void* const dtPtr = DTable+1;   /* force compiler to not use strict-aliasing */
+        const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
+        DTableDesc const dtd = HUF_getDTableDesc(DTable);
+        HUF_decodeStreamX2(ostart, &bitD, oend, dt, dtd.tableLog);
+    }
+
+    /* check */
+    if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
+
+    /* decoded size */
+    return dstSize;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress4X2_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    if (cSrcSize < 10) return ERROR(corruption_detected);   /* strict minimum : jump table + 1 byte per stream */
+
+    {   const BYTE* const istart = (const BYTE*) cSrc;
+        BYTE* const ostart = (BYTE*) dst;
+        BYTE* const oend = ostart + dstSize;
+        BYTE* const olimit = oend - (sizeof(size_t)-1);
+        const void* const dtPtr = DTable+1;
+        const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
+
+        /* Init */
+        BIT_DStream_t bitD1;
+        BIT_DStream_t bitD2;
+        BIT_DStream_t bitD3;
+        BIT_DStream_t bitD4;
+        size_t const length1 = MEM_readLE16(istart);
+        size_t const length2 = MEM_readLE16(istart+2);
+        size_t const length3 = MEM_readLE16(istart+4);
+        size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
+        const BYTE* const istart1 = istart + 6;  /* jumpTable */
+        const BYTE* const istart2 = istart1 + length1;
+        const BYTE* const istart3 = istart2 + length2;
+        const BYTE* const istart4 = istart3 + length3;
+        size_t const segmentSize = (dstSize+3) / 4;
+        BYTE* const opStart2 = ostart + segmentSize;
+        BYTE* const opStart3 = opStart2 + segmentSize;
+        BYTE* const opStart4 = opStart3 + segmentSize;
+        BYTE* op1 = ostart;
+        BYTE* op2 = opStart2;
+        BYTE* op3 = opStart3;
+        BYTE* op4 = opStart4;
+        U32 endSignal = 1;
+        DTableDesc const dtd = HUF_getDTableDesc(DTable);
+        U32 const dtLog = dtd.tableLog;
+
+        if (length4 > cSrcSize) return ERROR(corruption_detected);   /* overflow */
+        CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
+        CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
+        CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
+        CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );
+
+        /* 16-32 symbols per loop (4-8 symbols per stream) */
+        for ( ; (endSignal) & (op4 < olimit); ) {
+#if defined(__clang__) && (defined(__x86_64__) || defined(__i386__))
+            HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
+            endSignal &= BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished;
+            HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
+            endSignal &= BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished;
+            endSignal &= BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished;
+#else
+            HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
+            HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
+            HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
+            HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
+            endSignal = (U32)LIKELY((U32)
+                        (BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished)
+                      & (BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished)
+                      & (BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished)
+                      & (BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished));
+#endif
+        }
+
+        /* check corruption */
+        if (op1 > opStart2) return ERROR(corruption_detected);
+        if (op2 > opStart3) return ERROR(corruption_detected);
+        if (op3 > opStart4) return ERROR(corruption_detected);
+        /* note : op4 already verified within main loop */
+
+        /* finish bitStreams one by one */
+        HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog);
+        HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog);
+        HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog);
+        HUF_decodeStreamX2(op4, &bitD4, oend,     dt, dtLog);
+
+        /* check */
+        { U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
+          if (!endCheck) return ERROR(corruption_detected); }
+
+        /* decoded size */
+        return dstSize;
+    }
+}
+
+HUF_DGEN(HUF_decompress1X2_usingDTable_internal)
+HUF_DGEN(HUF_decompress4X2_usingDTable_internal)
+
+size_t HUF_decompress1X2_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 1) return ERROR(GENERIC);
+    return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize,
+                                               workSpace, wkspSize);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
+}
+
+
+size_t HUF_decompress4X2_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 1) return ERROR(GENERIC);
+    return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+static size_t HUF_decompress4X2_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize, int bmi2)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize,
+                                         workSpace, wkspSize);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
+}
+
+size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, /* bmi2 */ 0);
+}
+
+
+#endif /* HUF_FORCE_DECOMPRESS_X1 */
+
+
+/* ***********************************/
+/* Universal decompression selectors */
+/* ***********************************/
+
+size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize,
+                                    const void* cSrc, size_t cSrcSize,
+                                    const HUF_DTable* DTable)
+{
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#else
+    return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
+                           HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#endif
+}
+
+size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize,
+                                    const void* cSrc, size_t cSrcSize,
+                                    const HUF_DTable* DTable)
+{
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#else
+    return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
+                           HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#endif
+}
+
+
+#if !defined(HUF_FORCE_DECOMPRESS_X1) && !defined(HUF_FORCE_DECOMPRESS_X2)
+typedef struct { U32 tableTime; U32 decode256Time; } algo_time_t;
+static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, quad */] =
+{
+    /* single, double, quad */
+    {{0,0}, {1,1}, {2,2}},  /* Q==0 : impossible */
+    {{0,0}, {1,1}, {2,2}},  /* Q==1 : impossible */
+    {{  38,130}, {1313, 74}, {2151, 38}},   /* Q == 2 : 12-18% */
+    {{ 448,128}, {1353, 74}, {2238, 41}},   /* Q == 3 : 18-25% */
+    {{ 556,128}, {1353, 74}, {2238, 47}},   /* Q == 4 : 25-32% */
+    {{ 714,128}, {1418, 74}, {2436, 53}},   /* Q == 5 : 32-38% */
+    {{ 883,128}, {1437, 74}, {2464, 61}},   /* Q == 6 : 38-44% */
+    {{ 897,128}, {1515, 75}, {2622, 68}},   /* Q == 7 : 44-50% */
+    {{ 926,128}, {1613, 75}, {2730, 75}},   /* Q == 8 : 50-56% */
+    {{ 947,128}, {1729, 77}, {3359, 77}},   /* Q == 9 : 56-62% */
+    {{1107,128}, {2083, 81}, {4006, 84}},   /* Q ==10 : 62-69% */
+    {{1177,128}, {2379, 87}, {4785, 88}},   /* Q ==11 : 69-75% */
+    {{1242,128}, {2415, 93}, {5155, 84}},   /* Q ==12 : 75-81% */
+    {{1349,128}, {2644,106}, {5260,106}},   /* Q ==13 : 81-87% */
+    {{1455,128}, {2422,124}, {4174,124}},   /* Q ==14 : 87-93% */
+    {{ 722,128}, {1891,145}, {1936,146}},   /* Q ==15 : 93-99% */
+};
+#endif
+
+/* HUF_selectDecoder() :
+ *  Tells which decoder is likely to decode faster,
+ *  based on a set of pre-computed metrics.
+ * @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 .
+ *  Assumption : 0 < dstSize <= 128 KB */
+U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize)
+{
+    assert(dstSize > 0);
+    assert(dstSize <= 128*1024);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dstSize;
+    (void)cSrcSize;
+    return 0;
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dstSize;
+    (void)cSrcSize;
+    return 1;
+#else
+    /* decoder timing evaluation */
+    {   U32 const Q = (cSrcSize >= dstSize) ? 15 : (U32)(cSrcSize * 16 / dstSize);   /* Q < 16 */
+        U32 const D256 = (U32)(dstSize >> 8);
+        U32 const DTime0 = algoTime[Q][0].tableTime + (algoTime[Q][0].decode256Time * D256);
+        U32 DTime1 = algoTime[Q][1].tableTime + (algoTime[Q][1].decode256Time * D256);
+        DTime1 += DTime1 >> 3;  /* advantage to algorithm using less memory, to reduce cache eviction */
+        return DTime1 < DTime0;
+    }
+#endif
+}
+
+
+size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst,
+                                     size_t dstSize, const void* cSrc,
+                                     size_t cSrcSize, void* workSpace,
+                                     size_t wkspSize)
+{
+    /* validation checks */
+    if (dstSize == 0) return ERROR(dstSize_tooSmall);
+    if (cSrcSize == 0) return ERROR(corruption_detected);
+
+    {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#else
+        return algoNb ? HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                            cSrcSize, workSpace, wkspSize):
+                        HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#endif
+    }
+}
+
+size_t HUF_decompress1X_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                  const void* cSrc, size_t cSrcSize,
+                                  void* workSpace, size_t wkspSize)
+{
+    /* validation checks */
+    if (dstSize == 0) return ERROR(dstSize_tooSmall);
+    if (cSrcSize > dstSize) return ERROR(corruption_detected);   /* invalid */
+    if (cSrcSize == dstSize) { ZSTD_memcpy(dst, cSrc, dstSize); return dstSize; }   /* not compressed */
+    if (cSrcSize == 1) { ZSTD_memset(dst, *(const BYTE*)cSrc, dstSize); return dstSize; }   /* RLE */
+
+    {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize);
+#else
+        return algoNb ? HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize):
+                        HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize);
+#endif
+    }
+}
+
+
+size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
+{
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#else
+    return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
+                           HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#endif
+}
+
+#ifndef HUF_FORCE_DECOMPRESS_X2
+size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2)
+{
+    const BYTE* ip = (const BYTE*) cSrc;
+
+    size_t const hSize = HUF_readDTableX1_wksp_bmi2(dctx, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
+}
+#endif
+
+size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
+{
+    DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#else
+    return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
+                           HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#endif
+}
+
+size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2)
+{
+    /* validation checks */
+    if (dstSize == 0) return ERROR(dstSize_tooSmall);
+    if (cSrcSize == 0) return ERROR(corruption_detected);
+
+    {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#else
+        return algoNb ? HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2) :
+                        HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#endif
+    }
+}
+
diff --git a/lib/zstd/decompress/zstd_ddict.c b/lib/zstd/decompress/zstd_ddict.c
new file mode 100644
index 0000000..dbbc791
--- /dev/null
+++ b/lib/zstd/decompress/zstd_ddict.c
@@ -0,0 +1,241 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* zstd_ddict.c :
+ * concentrates all logic that needs to know the internals of ZSTD_DDict object */
+
+/*-*******************************************************
+*  Dependencies
+*********************************************************/
+#include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
+#include "../common/cpu.h"         /* bmi2 */
+#include "../common/mem.h"         /* low level memory routines */
+#define FSE_STATIC_LINKING_ONLY
+#include "../common/fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include "zstd_decompress_internal.h"
+#include "zstd_ddict.h"
+
+
+
+
+/*-*******************************************************
+*  Types
+*********************************************************/
+struct ZSTD_DDict_s {
+    void* dictBuffer;
+    const void* dictContent;
+    size_t dictSize;
+    ZSTD_entropyDTables_t entropy;
+    U32 dictID;
+    U32 entropyPresent;
+    ZSTD_customMem cMem;
+};  /* typedef'd to ZSTD_DDict within "zstd.h" */
+
+const void* ZSTD_DDict_dictContent(const ZSTD_DDict* ddict)
+{
+    assert(ddict != NULL);
+    return ddict->dictContent;
+}
+
+size_t ZSTD_DDict_dictSize(const ZSTD_DDict* ddict)
+{
+    assert(ddict != NULL);
+    return ddict->dictSize;
+}
+
+void ZSTD_copyDDictParameters(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
+{
+    DEBUGLOG(4, "ZSTD_copyDDictParameters");
+    assert(dctx != NULL);
+    assert(ddict != NULL);
+    dctx->dictID = ddict->dictID;
+    dctx->prefixStart = ddict->dictContent;
+    dctx->virtualStart = ddict->dictContent;
+    dctx->dictEnd = (const BYTE*)ddict->dictContent + ddict->dictSize;
+    dctx->previousDstEnd = dctx->dictEnd;
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    dctx->dictContentBeginForFuzzing = dctx->prefixStart;
+    dctx->dictContentEndForFuzzing = dctx->previousDstEnd;
+#endif
+    if (ddict->entropyPresent) {
+        dctx->litEntropy = 1;
+        dctx->fseEntropy = 1;
+        dctx->LLTptr = ddict->entropy.LLTable;
+        dctx->MLTptr = ddict->entropy.MLTable;
+        dctx->OFTptr = ddict->entropy.OFTable;
+        dctx->HUFptr = ddict->entropy.hufTable;
+        dctx->entropy.rep[0] = ddict->entropy.rep[0];
+        dctx->entropy.rep[1] = ddict->entropy.rep[1];
+        dctx->entropy.rep[2] = ddict->entropy.rep[2];
+    } else {
+        dctx->litEntropy = 0;
+        dctx->fseEntropy = 0;
+    }
+}
+
+
+static size_t
+ZSTD_loadEntropy_intoDDict(ZSTD_DDict* ddict,
+                           ZSTD_dictContentType_e dictContentType)
+{
+    ddict->dictID = 0;
+    ddict->entropyPresent = 0;
+    if (dictContentType == ZSTD_dct_rawContent) return 0;
+
+    if (ddict->dictSize < 8) {
+        if (dictContentType == ZSTD_dct_fullDict)
+            return ERROR(dictionary_corrupted);   /* only accept specified dictionaries */
+        return 0;   /* pure content mode */
+    }
+    {   U32 const magic = MEM_readLE32(ddict->dictContent);
+        if (magic != ZSTD_MAGIC_DICTIONARY) {
+            if (dictContentType == ZSTD_dct_fullDict)
+                return ERROR(dictionary_corrupted);   /* only accept specified dictionaries */
+            return 0;   /* pure content mode */
+        }
+    }
+    ddict->dictID = MEM_readLE32((const char*)ddict->dictContent + ZSTD_FRAMEIDSIZE);
+
+    /* load entropy tables */
+    RETURN_ERROR_IF(ZSTD_isError(ZSTD_loadDEntropy(
+            &ddict->entropy, ddict->dictContent, ddict->dictSize)),
+        dictionary_corrupted, "");
+    ddict->entropyPresent = 1;
+    return 0;
+}
+
+
+static size_t ZSTD_initDDict_internal(ZSTD_DDict* ddict,
+                                      const void* dict, size_t dictSize,
+                                      ZSTD_dictLoadMethod_e dictLoadMethod,
+                                      ZSTD_dictContentType_e dictContentType)
+{
+    if ((dictLoadMethod == ZSTD_dlm_byRef) || (!dict) || (!dictSize)) {
+        ddict->dictBuffer = NULL;
+        ddict->dictContent = dict;
+        if (!dict) dictSize = 0;
+    } else {
+        void* const internalBuffer = ZSTD_customMalloc(dictSize, ddict->cMem);
+        ddict->dictBuffer = internalBuffer;
+        ddict->dictContent = internalBuffer;
+        if (!internalBuffer) return ERROR(memory_allocation);
+        ZSTD_memcpy(internalBuffer, dict, dictSize);
+    }
+    ddict->dictSize = dictSize;
+    ddict->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001);  /* cover both little and big endian */
+
+    /* parse dictionary content */
+    FORWARD_IF_ERROR( ZSTD_loadEntropy_intoDDict(ddict, dictContentType) , "");
+
+    return 0;
+}
+
+ZSTD_DDict* ZSTD_createDDict_advanced(const void* dict, size_t dictSize,
+                                      ZSTD_dictLoadMethod_e dictLoadMethod,
+                                      ZSTD_dictContentType_e dictContentType,
+                                      ZSTD_customMem customMem)
+{
+    if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
+
+    {   ZSTD_DDict* const ddict = (ZSTD_DDict*) ZSTD_customMalloc(sizeof(ZSTD_DDict), customMem);
+        if (ddict == NULL) return NULL;
+        ddict->cMem = customMem;
+        {   size_t const initResult = ZSTD_initDDict_internal(ddict,
+                                            dict, dictSize,
+                                            dictLoadMethod, dictContentType);
+            if (ZSTD_isError(initResult)) {
+                ZSTD_freeDDict(ddict);
+                return NULL;
+        }   }
+        return ddict;
+    }
+}
+
+/*! ZSTD_createDDict() :
+*   Create a digested dictionary, to start decompression without startup delay.
+*   `dict` content is copied inside DDict.
+*   Consequently, `dict` can be released after `ZSTD_DDict` creation */
+ZSTD_DDict* ZSTD_createDDict(const void* dict, size_t dictSize)
+{
+    ZSTD_customMem const allocator = { NULL, NULL, NULL };
+    return ZSTD_createDDict_advanced(dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto, allocator);
+}
+
+/*! ZSTD_createDDict_byReference() :
+ *  Create a digested dictionary, to start decompression without startup delay.
+ *  Dictionary content is simply referenced, it will be accessed during decompression.
+ *  Warning : dictBuffer must outlive DDict (DDict must be freed before dictBuffer) */
+ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize)
+{
+    ZSTD_customMem const allocator = { NULL, NULL, NULL };
+    return ZSTD_createDDict_advanced(dictBuffer, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto, allocator);
+}
+
+
+const ZSTD_DDict* ZSTD_initStaticDDict(
+                                void* sBuffer, size_t sBufferSize,
+                                const void* dict, size_t dictSize,
+                                ZSTD_dictLoadMethod_e dictLoadMethod,
+                                ZSTD_dictContentType_e dictContentType)
+{
+    size_t const neededSpace = sizeof(ZSTD_DDict)
+                             + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
+    ZSTD_DDict* const ddict = (ZSTD_DDict*)sBuffer;
+    assert(sBuffer != NULL);
+    assert(dict != NULL);
+    if ((size_t)sBuffer & 7) return NULL;   /* 8-aligned */
+    if (sBufferSize < neededSpace) return NULL;
+    if (dictLoadMethod == ZSTD_dlm_byCopy) {
+        ZSTD_memcpy(ddict+1, dict, dictSize);  /* local copy */
+        dict = ddict+1;
+    }
+    if (ZSTD_isError( ZSTD_initDDict_internal(ddict,
+                                              dict, dictSize,
+                                              ZSTD_dlm_byRef, dictContentType) ))
+        return NULL;
+    return ddict;
+}
+
+
+size_t ZSTD_freeDDict(ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;   /* support free on NULL */
+    {   ZSTD_customMem const cMem = ddict->cMem;
+        ZSTD_customFree(ddict->dictBuffer, cMem);
+        ZSTD_customFree(ddict, cMem);
+        return 0;
+    }
+}
+
+/*! ZSTD_estimateDDictSize() :
+ *  Estimate amount of memory that will be needed to create a dictionary for decompression.
+ *  Note : dictionary created by reference using ZSTD_dlm_byRef are smaller */
+size_t ZSTD_estimateDDictSize(size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod)
+{
+    return sizeof(ZSTD_DDict) + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
+}
+
+size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;   /* support sizeof on NULL */
+    return sizeof(*ddict) + (ddict->dictBuffer ? ddict->dictSize : 0) ;
+}
+
+/*! ZSTD_getDictID_fromDDict() :
+ *  Provides the dictID of the dictionary loaded into `ddict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;
+    return ZSTD_getDictID_fromDict(ddict->dictContent, ddict->dictSize);
+}
diff --git a/lib/zstd/decompress/zstd_ddict.h b/lib/zstd/decompress/zstd_ddict.h
new file mode 100644
index 0000000..8c1a79d
--- /dev/null
+++ b/lib/zstd/decompress/zstd_ddict.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+#ifndef ZSTD_DDICT_H
+#define ZSTD_DDICT_H
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include "../common/zstd_deps.h"   /* size_t */
+#include <linux/zstd.h>     /* ZSTD_DDict, and several public functions */
+
+
+/*-*******************************************************
+ *  Interface
+ *********************************************************/
+
+/* note: several prototypes are already published in `zstd.h` :
+ * ZSTD_createDDict()
+ * ZSTD_createDDict_byReference()
+ * ZSTD_createDDict_advanced()
+ * ZSTD_freeDDict()
+ * ZSTD_initStaticDDict()
+ * ZSTD_sizeof_DDict()
+ * ZSTD_estimateDDictSize()
+ * ZSTD_getDictID_fromDict()
+ */
+
+const void* ZSTD_DDict_dictContent(const ZSTD_DDict* ddict);
+size_t ZSTD_DDict_dictSize(const ZSTD_DDict* ddict);
+
+void ZSTD_copyDDictParameters(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);
+
+
+
+#endif /* ZSTD_DDICT_H */
diff --git a/lib/zstd/decompress/zstd_decompress.c b/lib/zstd/decompress/zstd_decompress.c
new file mode 100644
index 0000000..b4d81d8
--- /dev/null
+++ b/lib/zstd/decompress/zstd_decompress.c
@@ -0,0 +1,2085 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+/* ***************************************************************
+*  Tuning parameters
+*****************************************************************/
+/*!
+ * HEAPMODE :
+ * Select how default decompression function ZSTD_decompress() allocates its context,
+ * on stack (0), or into heap (1, default; requires malloc()).
+ * Note that functions with explicit context such as ZSTD_decompressDCtx() are unaffected.
+ */
+#ifndef ZSTD_HEAPMODE
+#  define ZSTD_HEAPMODE 1
+#endif
+
+/*!
+*  LEGACY_SUPPORT :
+*  if set to 1+, ZSTD_decompress() can decode older formats (v0.1+)
+*/
+
+/*!
+ *  MAXWINDOWSIZE_DEFAULT :
+ *  maximum window size accepted by DStream __by default__.
+ *  Frames requiring more memory will be rejected.
+ *  It's possible to set a different limit using ZSTD_DCtx_setMaxWindowSize().
+ */
+#ifndef ZSTD_MAXWINDOWSIZE_DEFAULT
+#  define ZSTD_MAXWINDOWSIZE_DEFAULT (((U32)1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT) + 1)
+#endif
+
+/*!
+ *  NO_FORWARD_PROGRESS_MAX :
+ *  maximum allowed nb of calls to ZSTD_decompressStream()
+ *  without any forward progress
+ *  (defined as: no byte read from input, and no byte flushed to output)
+ *  before triggering an error.
+ */
+#ifndef ZSTD_NO_FORWARD_PROGRESS_MAX
+#  define ZSTD_NO_FORWARD_PROGRESS_MAX 16
+#endif
+
+
+/*-*******************************************************
+*  Dependencies
+*********************************************************/
+#include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
+#include "../common/cpu.h"         /* bmi2 */
+#include "../common/mem.h"         /* low level memory routines */
+#define FSE_STATIC_LINKING_ONLY
+#include "../common/fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include <linux/xxhash.h> /* xxh64_reset, xxh64_update, xxh64_digest, XXH64 */
+#include "../common/zstd_internal.h"  /* blockProperties_t */
+#include "zstd_decompress_internal.h"   /* ZSTD_DCtx */
+#include "zstd_ddict.h"  /* ZSTD_DDictDictContent */
+#include "zstd_decompress_block.h"   /* ZSTD_decompressBlock_internal */
+
+
+
+
+/* ***********************************
+ * Multiple DDicts Hashset internals *
+ *************************************/
+
+#define DDICT_HASHSET_MAX_LOAD_FACTOR_COUNT_MULT 4
+#define DDICT_HASHSET_MAX_LOAD_FACTOR_SIZE_MULT 3   /* These two constants represent SIZE_MULT/COUNT_MULT load factor without using a float.
+                                                     * Currently, that means a 0.75 load factor.
+                                                     * So, if count * COUNT_MULT / size * SIZE_MULT != 0, then we've exceeded
+                                                     * the load factor of the ddict hash set.
+                                                     */
+
+#define DDICT_HASHSET_TABLE_BASE_SIZE 64
+#define DDICT_HASHSET_RESIZE_FACTOR 2
+
+/* Hash function to determine starting position of dict insertion within the table
+ * Returns an index between [0, hashSet->ddictPtrTableSize]
+ */
+static size_t ZSTD_DDictHashSet_getIndex(const ZSTD_DDictHashSet* hashSet, U32 dictID) {
+    const U64 hash = xxh64(&dictID, sizeof(U32), 0);
+    /* DDict ptr table size is a multiple of 2, use size - 1 as mask to get index within [0, hashSet->ddictPtrTableSize) */
+    return hash & (hashSet->ddictPtrTableSize - 1);
+}
+
+/* Adds DDict to a hashset without resizing it.
+ * If inserting a DDict with a dictID that already exists in the set, replaces the one in the set.
+ * Returns 0 if successful, or a zstd error code if something went wrong.
+ */
+static size_t ZSTD_DDictHashSet_emplaceDDict(ZSTD_DDictHashSet* hashSet, const ZSTD_DDict* ddict) {
+    const U32 dictID = ZSTD_getDictID_fromDDict(ddict);
+    size_t idx = ZSTD_DDictHashSet_getIndex(hashSet, dictID);
+    const size_t idxRangeMask = hashSet->ddictPtrTableSize - 1;
+    RETURN_ERROR_IF(hashSet->ddictPtrCount == hashSet->ddictPtrTableSize, GENERIC, "Hash set is full!");
+    DEBUGLOG(4, "Hashed index: for dictID: %u is %zu", dictID, idx);
+    while (hashSet->ddictPtrTable[idx] != NULL) {
+        /* Replace existing ddict if inserting ddict with same dictID */
+        if (ZSTD_getDictID_fromDDict(hashSet->ddictPtrTable[idx]) == dictID) {
+            DEBUGLOG(4, "DictID already exists, replacing rather than adding");
+            hashSet->ddictPtrTable[idx] = ddict;
+            return 0;
+        }
+        idx &= idxRangeMask;
+        idx++;
+    }
+    DEBUGLOG(4, "Final idx after probing for dictID %u is: %zu", dictID, idx);
+    hashSet->ddictPtrTable[idx] = ddict;
+    hashSet->ddictPtrCount++;
+    return 0;
+}
+
+/* Expands hash table by factor of DDICT_HASHSET_RESIZE_FACTOR and
+ * rehashes all values, allocates new table, frees old table.
+ * Returns 0 on success, otherwise a zstd error code.
+ */
+static size_t ZSTD_DDictHashSet_expand(ZSTD_DDictHashSet* hashSet, ZSTD_customMem customMem) {
+    size_t newTableSize = hashSet->ddictPtrTableSize * DDICT_HASHSET_RESIZE_FACTOR;
+    const ZSTD_DDict** newTable = (const ZSTD_DDict**)ZSTD_customCalloc(sizeof(ZSTD_DDict*) * newTableSize, customMem);
+    const ZSTD_DDict** oldTable = hashSet->ddictPtrTable;
+    size_t oldTableSize = hashSet->ddictPtrTableSize;
+    size_t i;
+
+    DEBUGLOG(4, "Expanding DDict hash table! Old size: %zu new size: %zu", oldTableSize, newTableSize);
+    RETURN_ERROR_IF(!newTable, memory_allocation, "Expanded hashset allocation failed!");
+    hashSet->ddictPtrTable = newTable;
+    hashSet->ddictPtrTableSize = newTableSize;
+    hashSet->ddictPtrCount = 0;
+    for (i = 0; i < oldTableSize; ++i) {
+        if (oldTable[i] != NULL) {
+            FORWARD_IF_ERROR(ZSTD_DDictHashSet_emplaceDDict(hashSet, oldTable[i]), "");
+        }
+    }
+    ZSTD_customFree((void*)oldTable, customMem);
+    DEBUGLOG(4, "Finished re-hash");
+    return 0;
+}
+
+/* Fetches a DDict with the given dictID
+ * Returns the ZSTD_DDict* with the requested dictID. If it doesn't exist, then returns NULL.
+ */
+static const ZSTD_DDict* ZSTD_DDictHashSet_getDDict(ZSTD_DDictHashSet* hashSet, U32 dictID) {
+    size_t idx = ZSTD_DDictHashSet_getIndex(hashSet, dictID);
+    const size_t idxRangeMask = hashSet->ddictPtrTableSize - 1;
+    DEBUGLOG(4, "Hashed index: for dictID: %u is %zu", dictID, idx);
+    for (;;) {
+        size_t currDictID = ZSTD_getDictID_fromDDict(hashSet->ddictPtrTable[idx]);
+        if (currDictID == dictID || currDictID == 0) {
+            /* currDictID == 0 implies a NULL ddict entry */
+            break;
+        } else {
+            idx &= idxRangeMask;    /* Goes to start of table when we reach the end */
+            idx++;
+        }
+    }
+    DEBUGLOG(4, "Final idx after probing for dictID %u is: %zu", dictID, idx);
+    return hashSet->ddictPtrTable[idx];
+}
+
+/* Allocates space for and returns a ddict hash set
+ * The hash set's ZSTD_DDict* table has all values automatically set to NULL to begin with.
+ * Returns NULL if allocation failed.
+ */
+static ZSTD_DDictHashSet* ZSTD_createDDictHashSet(ZSTD_customMem customMem) {
+    ZSTD_DDictHashSet* ret = (ZSTD_DDictHashSet*)ZSTD_customMalloc(sizeof(ZSTD_DDictHashSet), customMem);
+    DEBUGLOG(4, "Allocating new hash set");
+    if (!ret)
+        return NULL;
+    ret->ddictPtrTable = (const ZSTD_DDict**)ZSTD_customCalloc(DDICT_HASHSET_TABLE_BASE_SIZE * sizeof(ZSTD_DDict*), customMem);
+    if (!ret->ddictPtrTable) {
+        ZSTD_customFree(ret, customMem);
+        return NULL;
+    }
+    ret->ddictPtrTableSize = DDICT_HASHSET_TABLE_BASE_SIZE;
+    ret->ddictPtrCount = 0;
+    return ret;
+}
+
+/* Frees the table of ZSTD_DDict* within a hashset, then frees the hashset itself.
+ * Note: The ZSTD_DDict* within the table are NOT freed.
+ */
+static void ZSTD_freeDDictHashSet(ZSTD_DDictHashSet* hashSet, ZSTD_customMem customMem) {
+    DEBUGLOG(4, "Freeing ddict hash set");
+    if (hashSet && hashSet->ddictPtrTable) {
+        ZSTD_customFree((void*)hashSet->ddictPtrTable, customMem);
+    }
+    if (hashSet) {
+        ZSTD_customFree(hashSet, customMem);
+    }
+}
+
+/* Public function: Adds a DDict into the ZSTD_DDictHashSet, possibly triggering a resize of the hash set.
+ * Returns 0 on success, or a ZSTD error.
+ */
+static size_t ZSTD_DDictHashSet_addDDict(ZSTD_DDictHashSet* hashSet, const ZSTD_DDict* ddict, ZSTD_customMem customMem) {
+    DEBUGLOG(4, "Adding dict ID: %u to hashset with - Count: %zu Tablesize: %zu", ZSTD_getDictID_fromDDict(ddict), hashSet->ddictPtrCount, hashSet->ddictPtrTableSize);
+    if (hashSet->ddictPtrCount * DDICT_HASHSET_MAX_LOAD_FACTOR_COUNT_MULT / hashSet->ddictPtrTableSize * DDICT_HASHSET_MAX_LOAD_FACTOR_SIZE_MULT != 0) {
+        FORWARD_IF_ERROR(ZSTD_DDictHashSet_expand(hashSet, customMem), "");
+    }
+    FORWARD_IF_ERROR(ZSTD_DDictHashSet_emplaceDDict(hashSet, ddict), "");
+    return 0;
+}
+
+/*-*************************************************************
+*   Context management
+***************************************************************/
+size_t ZSTD_sizeof_DCtx (const ZSTD_DCtx* dctx)
+{
+    if (dctx==NULL) return 0;   /* support sizeof NULL */
+    return sizeof(*dctx)
+           + ZSTD_sizeof_DDict(dctx->ddictLocal)
+           + dctx->inBuffSize + dctx->outBuffSize;
+}
+
+size_t ZSTD_estimateDCtxSize(void) { return sizeof(ZSTD_DCtx); }
+
+
+static size_t ZSTD_startingInputLength(ZSTD_format_e format)
+{
+    size_t const startingInputLength = ZSTD_FRAMEHEADERSIZE_PREFIX(format);
+    /* only supports formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless */
+    assert( (format == ZSTD_f_zstd1) || (format == ZSTD_f_zstd1_magicless) );
+    return startingInputLength;
+}
+
+static void ZSTD_DCtx_resetParameters(ZSTD_DCtx* dctx)
+{
+    assert(dctx->streamStage == zdss_init);
+    dctx->format = ZSTD_f_zstd1;
+    dctx->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT;
+    dctx->outBufferMode = ZSTD_bm_buffered;
+    dctx->forceIgnoreChecksum = ZSTD_d_validateChecksum;
+    dctx->refMultipleDDicts = ZSTD_rmd_refSingleDDict;
+}
+
+static void ZSTD_initDCtx_internal(ZSTD_DCtx* dctx)
+{
+    dctx->staticSize  = 0;
+    dctx->ddict       = NULL;
+    dctx->ddictLocal  = NULL;
+    dctx->dictEnd     = NULL;
+    dctx->ddictIsCold = 0;
+    dctx->dictUses = ZSTD_dont_use;
+    dctx->inBuff      = NULL;
+    dctx->inBuffSize  = 0;
+    dctx->outBuffSize = 0;
+    dctx->streamStage = zdss_init;
+    dctx->legacyContext = NULL;
+    dctx->previousLegacyVersion = 0;
+    dctx->noForwardProgress = 0;
+    dctx->oversizedDuration = 0;
+    dctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
+    dctx->ddictSet = NULL;
+    ZSTD_DCtx_resetParameters(dctx);
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    dctx->dictContentEndForFuzzing = NULL;
+#endif
+}
+
+ZSTD_DCtx* ZSTD_initStaticDCtx(void *workspace, size_t workspaceSize)
+{
+    ZSTD_DCtx* const dctx = (ZSTD_DCtx*) workspace;
+
+    if ((size_t)workspace & 7) return NULL;  /* 8-aligned */
+    if (workspaceSize < sizeof(ZSTD_DCtx)) return NULL;  /* minimum size */
+
+    ZSTD_initDCtx_internal(dctx);
+    dctx->staticSize = workspaceSize;
+    dctx->inBuff = (char*)(dctx+1);
+    return dctx;
+}
+
+ZSTD_DCtx* ZSTD_createDCtx_advanced(ZSTD_customMem customMem)
+{
+    if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
+
+    {   ZSTD_DCtx* const dctx = (ZSTD_DCtx*)ZSTD_customMalloc(sizeof(*dctx), customMem);
+        if (!dctx) return NULL;
+        dctx->customMem = customMem;
+        ZSTD_initDCtx_internal(dctx);
+        return dctx;
+    }
+}
+
+ZSTD_DCtx* ZSTD_createDCtx(void)
+{
+    DEBUGLOG(3, "ZSTD_createDCtx");
+    return ZSTD_createDCtx_advanced(ZSTD_defaultCMem);
+}
+
+static void ZSTD_clearDict(ZSTD_DCtx* dctx)
+{
+    ZSTD_freeDDict(dctx->ddictLocal);
+    dctx->ddictLocal = NULL;
+    dctx->ddict = NULL;
+    dctx->dictUses = ZSTD_dont_use;
+}
+
+size_t ZSTD_freeDCtx(ZSTD_DCtx* dctx)
+{
+    if (dctx==NULL) return 0;   /* support free on NULL */
+    RETURN_ERROR_IF(dctx->staticSize, memory_allocation, "not compatible with static DCtx");
+    {   ZSTD_customMem const cMem = dctx->customMem;
+        ZSTD_clearDict(dctx);
+        ZSTD_customFree(dctx->inBuff, cMem);
+        dctx->inBuff = NULL;
+        if (dctx->ddictSet) {
+            ZSTD_freeDDictHashSet(dctx->ddictSet, cMem);
+            dctx->ddictSet = NULL;
+        }
+        ZSTD_customFree(dctx, cMem);
+        return 0;
+    }
+}
+
+/* no longer useful */
+void ZSTD_copyDCtx(ZSTD_DCtx* dstDCtx, const ZSTD_DCtx* srcDCtx)
+{
+    size_t const toCopy = (size_t)((char*)(&dstDCtx->inBuff) - (char*)dstDCtx);
+    ZSTD_memcpy(dstDCtx, srcDCtx, toCopy);  /* no need to copy workspace */
+}
+
+/* Given a dctx with a digested frame params, re-selects the correct ZSTD_DDict based on
+ * the requested dict ID from the frame. If there exists a reference to the correct ZSTD_DDict, then
+ * accordingly sets the ddict to be used to decompress the frame.
+ *
+ * If no DDict is found, then no action is taken, and the ZSTD_DCtx::ddict remains as-is.
+ *
+ * ZSTD_d_refMultipleDDicts must be enabled for this function to be called.
+ */
+static void ZSTD_DCtx_selectFrameDDict(ZSTD_DCtx* dctx) {
+    assert(dctx->refMultipleDDicts && dctx->ddictSet);
+    DEBUGLOG(4, "Adjusting DDict based on requested dict ID from frame");
+    if (dctx->ddict) {
+        const ZSTD_DDict* frameDDict = ZSTD_DDictHashSet_getDDict(dctx->ddictSet, dctx->fParams.dictID);
+        if (frameDDict) {
+            DEBUGLOG(4, "DDict found!");
+            ZSTD_clearDict(dctx);
+            dctx->dictID = dctx->fParams.dictID;
+            dctx->ddict = frameDDict;
+            dctx->dictUses = ZSTD_use_indefinitely;
+        }
+    }
+}
+
+
+/*-*************************************************************
+ *   Frame header decoding
+ ***************************************************************/
+
+/*! ZSTD_isFrame() :
+ *  Tells if the content of `buffer` starts with a valid Frame Identifier.
+ *  Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
+ *  Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
+ *  Note 3 : Skippable Frame Identifiers are considered valid. */
+unsigned ZSTD_isFrame(const void* buffer, size_t size)
+{
+    if (size < ZSTD_FRAMEIDSIZE) return 0;
+    {   U32 const magic = MEM_readLE32(buffer);
+        if (magic == ZSTD_MAGICNUMBER) return 1;
+        if ((magic & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) return 1;
+    }
+    return 0;
+}
+
+/* ZSTD_frameHeaderSize_internal() :
+ *  srcSize must be large enough to reach header size fields.
+ *  note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless.
+ * @return : size of the Frame Header
+ *           or an error code, which can be tested with ZSTD_isError() */
+static size_t ZSTD_frameHeaderSize_internal(const void* src, size_t srcSize, ZSTD_format_e format)
+{
+    size_t const minInputSize = ZSTD_startingInputLength(format);
+    RETURN_ERROR_IF(srcSize < minInputSize, srcSize_wrong, "");
+
+    {   BYTE const fhd = ((const BYTE*)src)[minInputSize-1];
+        U32 const dictID= fhd & 3;
+        U32 const singleSegment = (fhd >> 5) & 1;
+        U32 const fcsId = fhd >> 6;
+        return minInputSize + !singleSegment
+             + ZSTD_did_fieldSize[dictID] + ZSTD_fcs_fieldSize[fcsId]
+             + (singleSegment && !fcsId);
+    }
+}
+
+/* ZSTD_frameHeaderSize() :
+ *  srcSize must be >= ZSTD_frameHeaderSize_prefix.
+ * @return : size of the Frame Header,
+ *           or an error code (if srcSize is too small) */
+size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize)
+{
+    return ZSTD_frameHeaderSize_internal(src, srcSize, ZSTD_f_zstd1);
+}
+
+
+/* ZSTD_getFrameHeader_advanced() :
+ *  decode Frame Header, or require larger `srcSize`.
+ *  note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless
+ * @return : 0, `zfhPtr` is correctly filled,
+ *          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+ *           or an error code, which can be tested using ZSTD_isError() */
+size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format)
+{
+    const BYTE* ip = (const BYTE*)src;
+    size_t const minInputSize = ZSTD_startingInputLength(format);
+
+    ZSTD_memset(zfhPtr, 0, sizeof(*zfhPtr));   /* not strictly necessary, but static analyzer do not understand that zfhPtr is only going to be read only if return value is zero, since they are 2 different signals */
+    if (srcSize < minInputSize) return minInputSize;
+    RETURN_ERROR_IF(src==NULL, GENERIC, "invalid parameter");
+
+    if ( (format != ZSTD_f_zstd1_magicless)
+      && (MEM_readLE32(src) != ZSTD_MAGICNUMBER) ) {
+        if ((MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+            /* skippable frame */
+            if (srcSize < ZSTD_SKIPPABLEHEADERSIZE)
+                return ZSTD_SKIPPABLEHEADERSIZE; /* magic number + frame length */
+            ZSTD_memset(zfhPtr, 0, sizeof(*zfhPtr));
+            zfhPtr->frameContentSize = MEM_readLE32((const char *)src + ZSTD_FRAMEIDSIZE);
+            zfhPtr->frameType = ZSTD_skippableFrame;
+            return 0;
+        }
+        RETURN_ERROR(prefix_unknown, "");
+    }
+
+    /* ensure there is enough `srcSize` to fully read/decode frame header */
+    {   size_t const fhsize = ZSTD_frameHeaderSize_internal(src, srcSize, format);
+        if (srcSize < fhsize) return fhsize;
+        zfhPtr->headerSize = (U32)fhsize;
+    }
+
+    {   BYTE const fhdByte = ip[minInputSize-1];
+        size_t pos = minInputSize;
+        U32 const dictIDSizeCode = fhdByte&3;
+        U32 const checksumFlag = (fhdByte>>2)&1;
+        U32 const singleSegment = (fhdByte>>5)&1;
+        U32 const fcsID = fhdByte>>6;
+        U64 windowSize = 0;
+        U32 dictID = 0;
+        U64 frameContentSize = ZSTD_CONTENTSIZE_UNKNOWN;
+        RETURN_ERROR_IF((fhdByte & 0x08) != 0, frameParameter_unsupported,
+                        "reserved bits, must be zero");
+
+        if (!singleSegment) {
+            BYTE const wlByte = ip[pos++];
+            U32 const windowLog = (wlByte >> 3) + ZSTD_WINDOWLOG_ABSOLUTEMIN;
+            RETURN_ERROR_IF(windowLog > ZSTD_WINDOWLOG_MAX, frameParameter_windowTooLarge, "");
+            windowSize = (1ULL << windowLog);
+            windowSize += (windowSize >> 3) * (wlByte&7);
+        }
+        switch(dictIDSizeCode)
+        {
+            default:
+                assert(0);  /* impossible */
+                ZSTD_FALLTHROUGH;
+            case 0 : break;
+            case 1 : dictID = ip[pos]; pos++; break;
+            case 2 : dictID = MEM_readLE16(ip+pos); pos+=2; break;
+            case 3 : dictID = MEM_readLE32(ip+pos); pos+=4; break;
+        }
+        switch(fcsID)
+        {
+            default:
+                assert(0);  /* impossible */
+                ZSTD_FALLTHROUGH;
+            case 0 : if (singleSegment) frameContentSize = ip[pos]; break;
+            case 1 : frameContentSize = MEM_readLE16(ip+pos)+256; break;
+            case 2 : frameContentSize = MEM_readLE32(ip+pos); break;
+            case 3 : frameContentSize = MEM_readLE64(ip+pos); break;
+        }
+        if (singleSegment) windowSize = frameContentSize;
+
+        zfhPtr->frameType = ZSTD_frame;
+        zfhPtr->frameContentSize = frameContentSize;
+        zfhPtr->windowSize = windowSize;
+        zfhPtr->blockSizeMax = (unsigned) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
+        zfhPtr->dictID = dictID;
+        zfhPtr->checksumFlag = checksumFlag;
+    }
+    return 0;
+}
+
+/* ZSTD_getFrameHeader() :
+ *  decode Frame Header, or require larger `srcSize`.
+ *  note : this function does not consume input, it only reads it.
+ * @return : 0, `zfhPtr` is correctly filled,
+ *          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+ *           or an error code, which can be tested using ZSTD_isError() */
+size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize)
+{
+    return ZSTD_getFrameHeader_advanced(zfhPtr, src, srcSize, ZSTD_f_zstd1);
+}
+
+
+/* ZSTD_getFrameContentSize() :
+ *  compatible with legacy mode
+ * @return : decompressed size of the single frame pointed to be `src` if known, otherwise
+ *         - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
+ *         - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small) */
+unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize)
+{
+    {   ZSTD_frameHeader zfh;
+        if (ZSTD_getFrameHeader(&zfh, src, srcSize) != 0)
+            return ZSTD_CONTENTSIZE_ERROR;
+        if (zfh.frameType == ZSTD_skippableFrame) {
+            return 0;
+        } else {
+            return zfh.frameContentSize;
+    }   }
+}
+
+static size_t readSkippableFrameSize(void const* src, size_t srcSize)
+{
+    size_t const skippableHeaderSize = ZSTD_SKIPPABLEHEADERSIZE;
+    U32 sizeU32;
+
+    RETURN_ERROR_IF(srcSize < ZSTD_SKIPPABLEHEADERSIZE, srcSize_wrong, "");
+
+    sizeU32 = MEM_readLE32((BYTE const*)src + ZSTD_FRAMEIDSIZE);
+    RETURN_ERROR_IF((U32)(sizeU32 + ZSTD_SKIPPABLEHEADERSIZE) < sizeU32,
+                    frameParameter_unsupported, "");
+    {
+        size_t const skippableSize = skippableHeaderSize + sizeU32;
+        RETURN_ERROR_IF(skippableSize > srcSize, srcSize_wrong, "");
+        return skippableSize;
+    }
+}
+
+/* ZSTD_findDecompressedSize() :
+ *  compatible with legacy mode
+ *  `srcSize` must be the exact length of some number of ZSTD compressed and/or
+ *      skippable frames
+ *  @return : decompressed size of the frames contained */
+unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
+{
+    unsigned long long totalDstSize = 0;
+
+    while (srcSize >= ZSTD_startingInputLength(ZSTD_f_zstd1)) {
+        U32 const magicNumber = MEM_readLE32(src);
+
+        if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+            size_t const skippableSize = readSkippableFrameSize(src, srcSize);
+            if (ZSTD_isError(skippableSize)) {
+                return ZSTD_CONTENTSIZE_ERROR;
+            }
+            assert(skippableSize <= srcSize);
+
+            src = (const BYTE *)src + skippableSize;
+            srcSize -= skippableSize;
+            continue;
+        }
+
+        {   unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
+            if (ret >= ZSTD_CONTENTSIZE_ERROR) return ret;
+
+            /* check for overflow */
+            if (totalDstSize + ret < totalDstSize) return ZSTD_CONTENTSIZE_ERROR;
+            totalDstSize += ret;
+        }
+        {   size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
+            if (ZSTD_isError(frameSrcSize)) {
+                return ZSTD_CONTENTSIZE_ERROR;
+            }
+
+            src = (const BYTE *)src + frameSrcSize;
+            srcSize -= frameSrcSize;
+        }
+    }  /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */
+
+    if (srcSize) return ZSTD_CONTENTSIZE_ERROR;
+
+    return totalDstSize;
+}
+
+/* ZSTD_getDecompressedSize() :
+ *  compatible with legacy mode
+ * @return : decompressed size if known, 0 otherwise
+             note : 0 can mean any of the following :
+                   - frame content is empty
+                   - decompressed size field is not present in frame header
+                   - frame header unknown / not supported
+                   - frame header not complete (`srcSize` too small) */
+unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize)
+{
+    unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
+    ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_ERROR < ZSTD_CONTENTSIZE_UNKNOWN);
+    return (ret >= ZSTD_CONTENTSIZE_ERROR) ? 0 : ret;
+}
+
+
+/* ZSTD_decodeFrameHeader() :
+ * `headerSize` must be the size provided by ZSTD_frameHeaderSize().
+ * If multiple DDict references are enabled, also will choose the correct DDict to use.
+ * @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */
+static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx* dctx, const void* src, size_t headerSize)
+{
+    size_t const result = ZSTD_getFrameHeader_advanced(&(dctx->fParams), src, headerSize, dctx->format);
+    if (ZSTD_isError(result)) return result;    /* invalid header */
+    RETURN_ERROR_IF(result>0, srcSize_wrong, "headerSize too small");
+
+    /* Reference DDict requested by frame if dctx references multiple ddicts */
+    if (dctx->refMultipleDDicts == ZSTD_rmd_refMultipleDDicts && dctx->ddictSet) {
+        ZSTD_DCtx_selectFrameDDict(dctx);
+    }
+
+#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    /* Skip the dictID check in fuzzing mode, because it makes the search
+     * harder.
+     */
+    RETURN_ERROR_IF(dctx->fParams.dictID && (dctx->dictID != dctx->fParams.dictID),
+                    dictionary_wrong, "");
+#endif
+    dctx->validateChecksum = (dctx->fParams.checksumFlag && !dctx->forceIgnoreChecksum) ? 1 : 0;
+    if (dctx->validateChecksum) xxh64_reset(&dctx->xxhState, 0);
+    dctx->processedCSize += headerSize;
+    return 0;
+}
+
+static ZSTD_frameSizeInfo ZSTD_errorFrameSizeInfo(size_t ret)
+{
+    ZSTD_frameSizeInfo frameSizeInfo;
+    frameSizeInfo.compressedSize = ret;
+    frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
+    return frameSizeInfo;
+}
+
+static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize)
+{
+    ZSTD_frameSizeInfo frameSizeInfo;
+    ZSTD_memset(&frameSizeInfo, 0, sizeof(ZSTD_frameSizeInfo));
+
+
+    if ((srcSize >= ZSTD_SKIPPABLEHEADERSIZE)
+        && (MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+        frameSizeInfo.compressedSize = readSkippableFrameSize(src, srcSize);
+        assert(ZSTD_isError(frameSizeInfo.compressedSize) ||
+               frameSizeInfo.compressedSize <= srcSize);
+        return frameSizeInfo;
+    } else {
+        const BYTE* ip = (const BYTE*)src;
+        const BYTE* const ipstart = ip;
+        size_t remainingSize = srcSize;
+        size_t nbBlocks = 0;
+        ZSTD_frameHeader zfh;
+
+        /* Extract Frame Header */
+        {   size_t const ret = ZSTD_getFrameHeader(&zfh, src, srcSize);
+            if (ZSTD_isError(ret))
+                return ZSTD_errorFrameSizeInfo(ret);
+            if (ret > 0)
+                return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
+        }
+
+        ip += zfh.headerSize;
+        remainingSize -= zfh.headerSize;
+
+        /* Iterate over each block */
+        while (1) {
+            blockProperties_t blockProperties;
+            size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
+            if (ZSTD_isError(cBlockSize))
+                return ZSTD_errorFrameSizeInfo(cBlockSize);
+
+            if (ZSTD_blockHeaderSize + cBlockSize > remainingSize)
+                return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
+
+            ip += ZSTD_blockHeaderSize + cBlockSize;
+            remainingSize -= ZSTD_blockHeaderSize + cBlockSize;
+            nbBlocks++;
+
+            if (blockProperties.lastBlock) break;
+        }
+
+        /* Final frame content checksum */
+        if (zfh.checksumFlag) {
+            if (remainingSize < 4)
+                return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
+            ip += 4;
+        }
+
+        frameSizeInfo.compressedSize = (size_t)(ip - ipstart);
+        frameSizeInfo.decompressedBound = (zfh.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN)
+                                        ? zfh.frameContentSize
+                                        : nbBlocks * zfh.blockSizeMax;
+        return frameSizeInfo;
+    }
+}
+
+/* ZSTD_findFrameCompressedSize() :
+ *  compatible with legacy mode
+ *  `src` must point to the start of a ZSTD frame, ZSTD legacy frame, or skippable frame
+ *  `srcSize` must be at least as large as the frame contained
+ *  @return : the compressed size of the frame starting at `src` */
+size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize)
+{
+    ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
+    return frameSizeInfo.compressedSize;
+}
+
+/* ZSTD_decompressBound() :
+ *  compatible with legacy mode
+ *  `src` must point to the start of a ZSTD frame or a skippeable frame
+ *  `srcSize` must be at least as large as the frame contained
+ *  @return : the maximum decompressed size of the compressed source
+ */
+unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize)
+{
+    unsigned long long bound = 0;
+    /* Iterate over each frame */
+    while (srcSize > 0) {
+        ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
+        size_t const compressedSize = frameSizeInfo.compressedSize;
+        unsigned long long const decompressedBound = frameSizeInfo.decompressedBound;
+        if (ZSTD_isError(compressedSize) || decompressedBound == ZSTD_CONTENTSIZE_ERROR)
+            return ZSTD_CONTENTSIZE_ERROR;
+        assert(srcSize >= compressedSize);
+        src = (const BYTE*)src + compressedSize;
+        srcSize -= compressedSize;
+        bound += decompressedBound;
+    }
+    return bound;
+}
+
+
+/*-*************************************************************
+ *   Frame decoding
+ ***************************************************************/
+
+/* ZSTD_insertBlock() :
+ *  insert `src` block into `dctx` history. Useful to track uncompressed blocks. */
+size_t ZSTD_insertBlock(ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize)
+{
+    DEBUGLOG(5, "ZSTD_insertBlock: %u bytes", (unsigned)blockSize);
+    ZSTD_checkContinuity(dctx, blockStart, blockSize);
+    dctx->previousDstEnd = (const char*)blockStart + blockSize;
+    return blockSize;
+}
+
+
+static size_t ZSTD_copyRawBlock(void* dst, size_t dstCapacity,
+                          const void* src, size_t srcSize)
+{
+    DEBUGLOG(5, "ZSTD_copyRawBlock");
+    RETURN_ERROR_IF(srcSize > dstCapacity, dstSize_tooSmall, "");
+    if (dst == NULL) {
+        if (srcSize == 0) return 0;
+        RETURN_ERROR(dstBuffer_null, "");
+    }
+    ZSTD_memcpy(dst, src, srcSize);
+    return srcSize;
+}
+
+static size_t ZSTD_setRleBlock(void* dst, size_t dstCapacity,
+                               BYTE b,
+                               size_t regenSize)
+{
+    RETURN_ERROR_IF(regenSize > dstCapacity, dstSize_tooSmall, "");
+    if (dst == NULL) {
+        if (regenSize == 0) return 0;
+        RETURN_ERROR(dstBuffer_null, "");
+    }
+    ZSTD_memset(dst, b, regenSize);
+    return regenSize;
+}
+
+static void ZSTD_DCtx_trace_end(ZSTD_DCtx const* dctx, U64 uncompressedSize, U64 compressedSize, unsigned streaming)
+{
+    (void)dctx;
+    (void)uncompressedSize;
+    (void)compressedSize;
+    (void)streaming;
+}
+
+
+/*! ZSTD_decompressFrame() :
+ * @dctx must be properly initialized
+ *  will update *srcPtr and *srcSizePtr,
+ *  to make *srcPtr progress by one frame. */
+static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
+                                   void* dst, size_t dstCapacity,
+                             const void** srcPtr, size_t *srcSizePtr)
+{
+    const BYTE* const istart = (const BYTE*)(*srcPtr);
+    const BYTE* ip = istart;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = dstCapacity != 0 ? ostart + dstCapacity : ostart;
+    BYTE* op = ostart;
+    size_t remainingSrcSize = *srcSizePtr;
+
+    DEBUGLOG(4, "ZSTD_decompressFrame (srcSize:%i)", (int)*srcSizePtr);
+
+    /* check */
+    RETURN_ERROR_IF(
+        remainingSrcSize < ZSTD_FRAMEHEADERSIZE_MIN(dctx->format)+ZSTD_blockHeaderSize,
+        srcSize_wrong, "");
+
+    /* Frame Header */
+    {   size_t const frameHeaderSize = ZSTD_frameHeaderSize_internal(
+                ip, ZSTD_FRAMEHEADERSIZE_PREFIX(dctx->format), dctx->format);
+        if (ZSTD_isError(frameHeaderSize)) return frameHeaderSize;
+        RETURN_ERROR_IF(remainingSrcSize < frameHeaderSize+ZSTD_blockHeaderSize,
+                        srcSize_wrong, "");
+        FORWARD_IF_ERROR( ZSTD_decodeFrameHeader(dctx, ip, frameHeaderSize) , "");
+        ip += frameHeaderSize; remainingSrcSize -= frameHeaderSize;
+    }
+
+    /* Loop on each block */
+    while (1) {
+        size_t decodedSize;
+        blockProperties_t blockProperties;
+        size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSrcSize, &blockProperties);
+        if (ZSTD_isError(cBlockSize)) return cBlockSize;
+
+        ip += ZSTD_blockHeaderSize;
+        remainingSrcSize -= ZSTD_blockHeaderSize;
+        RETURN_ERROR_IF(cBlockSize > remainingSrcSize, srcSize_wrong, "");
+
+        switch(blockProperties.blockType)
+        {
+        case bt_compressed:
+            decodedSize = ZSTD_decompressBlock_internal(dctx, op, (size_t)(oend-op), ip, cBlockSize, /* frame */ 1);
+            break;
+        case bt_raw :
+            decodedSize = ZSTD_copyRawBlock(op, (size_t)(oend-op), ip, cBlockSize);
+            break;
+        case bt_rle :
+            decodedSize = ZSTD_setRleBlock(op, (size_t)(oend-op), *ip, blockProperties.origSize);
+            break;
+        case bt_reserved :
+        default:
+            RETURN_ERROR(corruption_detected, "invalid block type");
+        }
+
+        if (ZSTD_isError(decodedSize)) return decodedSize;
+        if (dctx->validateChecksum)
+            xxh64_update(&dctx->xxhState, op, decodedSize);
+        if (decodedSize != 0)
+            op += decodedSize;
+        assert(ip != NULL);
+        ip += cBlockSize;
+        remainingSrcSize -= cBlockSize;
+        if (blockProperties.lastBlock) break;
+    }
+
+    if (dctx->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN) {
+        RETURN_ERROR_IF((U64)(op-ostart) != dctx->fParams.frameContentSize,
+                        corruption_detected, "");
+    }
+    if (dctx->fParams.checksumFlag) { /* Frame content checksum verification */
+        RETURN_ERROR_IF(remainingSrcSize<4, checksum_wrong, "");
+        if (!dctx->forceIgnoreChecksum) {
+            U32 const checkCalc = (U32)xxh64_digest(&dctx->xxhState);
+            U32 checkRead;
+            checkRead = MEM_readLE32(ip);
+            RETURN_ERROR_IF(checkRead != checkCalc, checksum_wrong, "");
+        }
+        ip += 4;
+        remainingSrcSize -= 4;
+    }
+    ZSTD_DCtx_trace_end(dctx, (U64)(op-ostart), (U64)(ip-istart), /* streaming */ 0);
+    /* Allow caller to get size read */
+    *srcPtr = ip;
+    *srcSizePtr = remainingSrcSize;
+    return (size_t)(op-ostart);
+}
+
+static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
+                                        void* dst, size_t dstCapacity,
+                                  const void* src, size_t srcSize,
+                                  const void* dict, size_t dictSize,
+                                  const ZSTD_DDict* ddict)
+{
+    void* const dststart = dst;
+    int moreThan1Frame = 0;
+
+    DEBUGLOG(5, "ZSTD_decompressMultiFrame");
+    assert(dict==NULL || ddict==NULL);  /* either dict or ddict set, not both */
+
+    if (ddict) {
+        dict = ZSTD_DDict_dictContent(ddict);
+        dictSize = ZSTD_DDict_dictSize(ddict);
+    }
+
+    while (srcSize >= ZSTD_startingInputLength(dctx->format)) {
+
+
+        {   U32 const magicNumber = MEM_readLE32(src);
+            DEBUGLOG(4, "reading magic number %08X (expecting %08X)",
+                        (unsigned)magicNumber, ZSTD_MAGICNUMBER);
+            if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+                size_t const skippableSize = readSkippableFrameSize(src, srcSize);
+                FORWARD_IF_ERROR(skippableSize, "readSkippableFrameSize failed");
+                assert(skippableSize <= srcSize);
+
+                src = (const BYTE *)src + skippableSize;
+                srcSize -= skippableSize;
+                continue;
+        }   }
+
+        if (ddict) {
+            /* we were called from ZSTD_decompress_usingDDict */
+            FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(dctx, ddict), "");
+        } else {
+            /* this will initialize correctly with no dict if dict == NULL, so
+             * use this in all cases but ddict */
+            FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDict(dctx, dict, dictSize), "");
+        }
+        ZSTD_checkContinuity(dctx, dst, dstCapacity);
+
+        {   const size_t res = ZSTD_decompressFrame(dctx, dst, dstCapacity,
+                                                    &src, &srcSize);
+            RETURN_ERROR_IF(
+                (ZSTD_getErrorCode(res) == ZSTD_error_prefix_unknown)
+             && (moreThan1Frame==1),
+                srcSize_wrong,
+                "At least one frame successfully completed, "
+                "but following bytes are garbage: "
+                "it's more likely to be a srcSize error, "
+                "specifying more input bytes than size of frame(s). "
+                "Note: one could be unlucky, it might be a corruption error instead, "
+                "happening right at the place where we expect zstd magic bytes. "
+                "But this is _much_ less likely than a srcSize field error.");
+            if (ZSTD_isError(res)) return res;
+            assert(res <= dstCapacity);
+            if (res != 0)
+                dst = (BYTE*)dst + res;
+            dstCapacity -= res;
+        }
+        moreThan1Frame = 1;
+    }  /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */
+
+    RETURN_ERROR_IF(srcSize, srcSize_wrong, "input not entirely consumed");
+
+    return (size_t)((BYTE*)dst - (BYTE*)dststart);
+}
+
+size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx,
+                                 void* dst, size_t dstCapacity,
+                           const void* src, size_t srcSize,
+                           const void* dict, size_t dictSize)
+{
+    return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, dict, dictSize, NULL);
+}
+
+
+static ZSTD_DDict const* ZSTD_getDDict(ZSTD_DCtx* dctx)
+{
+    switch (dctx->dictUses) {
+    default:
+        assert(0 /* Impossible */);
+        ZSTD_FALLTHROUGH;
+    case ZSTD_dont_use:
+        ZSTD_clearDict(dctx);
+        return NULL;
+    case ZSTD_use_indefinitely:
+        return dctx->ddict;
+    case ZSTD_use_once:
+        dctx->dictUses = ZSTD_dont_use;
+        return dctx->ddict;
+    }
+}
+
+size_t ZSTD_decompressDCtx(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    return ZSTD_decompress_usingDDict(dctx, dst, dstCapacity, src, srcSize, ZSTD_getDDict(dctx));
+}
+
+
+size_t ZSTD_decompress(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+#if defined(ZSTD_HEAPMODE) && (ZSTD_HEAPMODE>=1)
+    size_t regenSize;
+    ZSTD_DCtx* const dctx = ZSTD_createDCtx();
+    RETURN_ERROR_IF(dctx==NULL, memory_allocation, "NULL pointer!");
+    regenSize = ZSTD_decompressDCtx(dctx, dst, dstCapacity, src, srcSize);
+    ZSTD_freeDCtx(dctx);
+    return regenSize;
+#else   /* stack mode */
+    ZSTD_DCtx dctx;
+    ZSTD_initDCtx_internal(&dctx);
+    return ZSTD_decompressDCtx(&dctx, dst, dstCapacity, src, srcSize);
+#endif
+}
+
+
+/*-**************************************
+*   Advanced Streaming Decompression API
+*   Bufferless and synchronous
+****************************************/
+size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx) { return dctx->expected; }
+
+/*
+ * Similar to ZSTD_nextSrcSizeToDecompress(), but when when a block input can be streamed,
+ * we allow taking a partial block as the input. Currently only raw uncompressed blocks can
+ * be streamed.
+ *
+ * For blocks that can be streamed, this allows us to reduce the latency until we produce
+ * output, and avoid copying the input.
+ *
+ * @param inputSize - The total amount of input that the caller currently has.
+ */
+static size_t ZSTD_nextSrcSizeToDecompressWithInputSize(ZSTD_DCtx* dctx, size_t inputSize) {
+    if (!(dctx->stage == ZSTDds_decompressBlock || dctx->stage == ZSTDds_decompressLastBlock))
+        return dctx->expected;
+    if (dctx->bType != bt_raw)
+        return dctx->expected;
+    return MIN(MAX(inputSize, 1), dctx->expected);
+}
+
+ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx) {
+    switch(dctx->stage)
+    {
+    default:   /* should not happen */
+        assert(0);
+        ZSTD_FALLTHROUGH;
+    case ZSTDds_getFrameHeaderSize:
+        ZSTD_FALLTHROUGH;
+    case ZSTDds_decodeFrameHeader:
+        return ZSTDnit_frameHeader;
+    case ZSTDds_decodeBlockHeader:
+        return ZSTDnit_blockHeader;
+    case ZSTDds_decompressBlock:
+        return ZSTDnit_block;
+    case ZSTDds_decompressLastBlock:
+        return ZSTDnit_lastBlock;
+    case ZSTDds_checkChecksum:
+        return ZSTDnit_checksum;
+    case ZSTDds_decodeSkippableHeader:
+        ZSTD_FALLTHROUGH;
+    case ZSTDds_skipFrame:
+        return ZSTDnit_skippableFrame;
+    }
+}
+
+static int ZSTD_isSkipFrame(ZSTD_DCtx* dctx) { return dctx->stage == ZSTDds_skipFrame; }
+
+/* ZSTD_decompressContinue() :
+ *  srcSize : must be the exact nb of bytes expected (see ZSTD_nextSrcSizeToDecompress())
+ *  @return : nb of bytes generated into `dst` (necessarily <= `dstCapacity)
+ *            or an error code, which can be tested using ZSTD_isError() */
+size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
+{
+    DEBUGLOG(5, "ZSTD_decompressContinue (srcSize:%u)", (unsigned)srcSize);
+    /* Sanity check */
+    RETURN_ERROR_IF(srcSize != ZSTD_nextSrcSizeToDecompressWithInputSize(dctx, srcSize), srcSize_wrong, "not allowed");
+    ZSTD_checkContinuity(dctx, dst, dstCapacity);
+
+    dctx->processedCSize += srcSize;
+
+    switch (dctx->stage)
+    {
+    case ZSTDds_getFrameHeaderSize :
+        assert(src != NULL);
+        if (dctx->format == ZSTD_f_zstd1) {  /* allows header */
+            assert(srcSize >= ZSTD_FRAMEIDSIZE);  /* to read skippable magic number */
+            if ((MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {        /* skippable frame */
+                ZSTD_memcpy(dctx->headerBuffer, src, srcSize);
+                dctx->expected = ZSTD_SKIPPABLEHEADERSIZE - srcSize;  /* remaining to load to get full skippable frame header */
+                dctx->stage = ZSTDds_decodeSkippableHeader;
+                return 0;
+        }   }
+        dctx->headerSize = ZSTD_frameHeaderSize_internal(src, srcSize, dctx->format);
+        if (ZSTD_isError(dctx->headerSize)) return dctx->headerSize;
+        ZSTD_memcpy(dctx->headerBuffer, src, srcSize);
+        dctx->expected = dctx->headerSize - srcSize;
+        dctx->stage = ZSTDds_decodeFrameHeader;
+        return 0;
+
+    case ZSTDds_decodeFrameHeader:
+        assert(src != NULL);
+        ZSTD_memcpy(dctx->headerBuffer + (dctx->headerSize - srcSize), src, srcSize);
+        FORWARD_IF_ERROR(ZSTD_decodeFrameHeader(dctx, dctx->headerBuffer, dctx->headerSize), "");
+        dctx->expected = ZSTD_blockHeaderSize;
+        dctx->stage = ZSTDds_decodeBlockHeader;
+        return 0;
+
+    case ZSTDds_decodeBlockHeader:
+        {   blockProperties_t bp;
+            size_t const cBlockSize = ZSTD_getcBlockSize(src, ZSTD_blockHeaderSize, &bp);
+            if (ZSTD_isError(cBlockSize)) return cBlockSize;
+            RETURN_ERROR_IF(cBlockSize > dctx->fParams.blockSizeMax, corruption_detected, "Block Size Exceeds Maximum");
+            dctx->expected = cBlockSize;
+            dctx->bType = bp.blockType;
+            dctx->rleSize = bp.origSize;
+            if (cBlockSize) {
+                dctx->stage = bp.lastBlock ? ZSTDds_decompressLastBlock : ZSTDds_decompressBlock;
+                return 0;
+            }
+            /* empty block */
+            if (bp.lastBlock) {
+                if (dctx->fParams.checksumFlag) {
+                    dctx->expected = 4;
+                    dctx->stage = ZSTDds_checkChecksum;
+                } else {
+                    dctx->expected = 0; /* end of frame */
+                    dctx->stage = ZSTDds_getFrameHeaderSize;
+                }
+            } else {
+                dctx->expected = ZSTD_blockHeaderSize;  /* jump to next header */
+                dctx->stage = ZSTDds_decodeBlockHeader;
+            }
+            return 0;
+        }
+
+    case ZSTDds_decompressLastBlock:
+    case ZSTDds_decompressBlock:
+        DEBUGLOG(5, "ZSTD_decompressContinue: case ZSTDds_decompressBlock");
+        {   size_t rSize;
+            switch(dctx->bType)
+            {
+            case bt_compressed:
+                DEBUGLOG(5, "ZSTD_decompressContinue: case bt_compressed");
+                rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 1);
+                dctx->expected = 0;  /* Streaming not supported */
+                break;
+            case bt_raw :
+                assert(srcSize <= dctx->expected);
+                rSize = ZSTD_copyRawBlock(dst, dstCapacity, src, srcSize);
+                FORWARD_IF_ERROR(rSize, "ZSTD_copyRawBlock failed");
+                assert(rSize == srcSize);
+                dctx->expected -= rSize;
+                break;
+            case bt_rle :
+                rSize = ZSTD_setRleBlock(dst, dstCapacity, *(const BYTE*)src, dctx->rleSize);
+                dctx->expected = 0;  /* Streaming not supported */
+                break;
+            case bt_reserved :   /* should never happen */
+            default:
+                RETURN_ERROR(corruption_detected, "invalid block type");
+            }
+            FORWARD_IF_ERROR(rSize, "");
+            RETURN_ERROR_IF(rSize > dctx->fParams.blockSizeMax, corruption_detected, "Decompressed Block Size Exceeds Maximum");
+            DEBUGLOG(5, "ZSTD_decompressContinue: decoded size from block : %u", (unsigned)rSize);
+            dctx->decodedSize += rSize;
+            if (dctx->validateChecksum) xxh64_update(&dctx->xxhState, dst, rSize);
+            dctx->previousDstEnd = (char*)dst + rSize;
+
+            /* Stay on the same stage until we are finished streaming the block. */
+            if (dctx->expected > 0) {
+                return rSize;
+            }
+
+            if (dctx->stage == ZSTDds_decompressLastBlock) {   /* end of frame */
+                DEBUGLOG(4, "ZSTD_decompressContinue: decoded size from frame : %u", (unsigned)dctx->decodedSize);
+                RETURN_ERROR_IF(
+                    dctx->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
+                 && dctx->decodedSize != dctx->fParams.frameContentSize,
+                    corruption_detected, "");
+                if (dctx->fParams.checksumFlag) {  /* another round for frame checksum */
+                    dctx->expected = 4;
+                    dctx->stage = ZSTDds_checkChecksum;
+                } else {
+                    ZSTD_DCtx_trace_end(dctx, dctx->decodedSize, dctx->processedCSize, /* streaming */ 1);
+                    dctx->expected = 0;   /* ends here */
+                    dctx->stage = ZSTDds_getFrameHeaderSize;
+                }
+            } else {
+                dctx->stage = ZSTDds_decodeBlockHeader;
+                dctx->expected = ZSTD_blockHeaderSize;
+            }
+            return rSize;
+        }
+
+    case ZSTDds_checkChecksum:
+        assert(srcSize == 4);  /* guaranteed by dctx->expected */
+        {
+            if (dctx->validateChecksum) {
+                U32 const h32 = (U32)xxh64_digest(&dctx->xxhState);
+                U32 const check32 = MEM_readLE32(src);
+                DEBUGLOG(4, "ZSTD_decompressContinue: checksum : calculated %08X :: %08X read", (unsigned)h32, (unsigned)check32);
+                RETURN_ERROR_IF(check32 != h32, checksum_wrong, "");
+            }
+            ZSTD_DCtx_trace_end(dctx, dctx->decodedSize, dctx->processedCSize, /* streaming */ 1);
+            dctx->expected = 0;
+            dctx->stage = ZSTDds_getFrameHeaderSize;
+            return 0;
+        }
+
+    case ZSTDds_decodeSkippableHeader:
+        assert(src != NULL);
+        assert(srcSize <= ZSTD_SKIPPABLEHEADERSIZE);
+        ZSTD_memcpy(dctx->headerBuffer + (ZSTD_SKIPPABLEHEADERSIZE - srcSize), src, srcSize);   /* complete skippable header */
+        dctx->expected = MEM_readLE32(dctx->headerBuffer + ZSTD_FRAMEIDSIZE);   /* note : dctx->expected can grow seriously large, beyond local buffer size */
+        dctx->stage = ZSTDds_skipFrame;
+        return 0;
+
+    case ZSTDds_skipFrame:
+        dctx->expected = 0;
+        dctx->stage = ZSTDds_getFrameHeaderSize;
+        return 0;
+
+    default:
+        assert(0);   /* impossible */
+        RETURN_ERROR(GENERIC, "impossible to reach");   /* some compiler require default to do something */
+    }
+}
+
+
+static size_t ZSTD_refDictContent(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    dctx->dictEnd = dctx->previousDstEnd;
+    dctx->virtualStart = (const char*)dict - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart));
+    dctx->prefixStart = dict;
+    dctx->previousDstEnd = (const char*)dict + dictSize;
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    dctx->dictContentBeginForFuzzing = dctx->prefixStart;
+    dctx->dictContentEndForFuzzing = dctx->previousDstEnd;
+#endif
+    return 0;
+}
+
+/*! ZSTD_loadDEntropy() :
+ *  dict : must point at beginning of a valid zstd dictionary.
+ * @return : size of entropy tables read */
+size_t
+ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
+                  const void* const dict, size_t const dictSize)
+{
+    const BYTE* dictPtr = (const BYTE*)dict;
+    const BYTE* const dictEnd = dictPtr + dictSize;
+
+    RETURN_ERROR_IF(dictSize <= 8, dictionary_corrupted, "dict is too small");
+    assert(MEM_readLE32(dict) == ZSTD_MAGIC_DICTIONARY);   /* dict must be valid */
+    dictPtr += 8;   /* skip header = magic + dictID */
+
+    ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, OFTable) == offsetof(ZSTD_entropyDTables_t, LLTable) + sizeof(entropy->LLTable));
+    ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, MLTable) == offsetof(ZSTD_entropyDTables_t, OFTable) + sizeof(entropy->OFTable));
+    ZSTD_STATIC_ASSERT(sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable) >= HUF_DECOMPRESS_WORKSPACE_SIZE);
+    {   void* const workspace = &entropy->LLTable;   /* use fse tables as temporary workspace; implies fse tables are grouped together */
+        size_t const workspaceSize = sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable);
+#ifdef HUF_FORCE_DECOMPRESS_X1
+        /* in minimal huffman, we always use X1 variants */
+        size_t const hSize = HUF_readDTableX1_wksp(entropy->hufTable,
+                                                dictPtr, dictEnd - dictPtr,
+                                                workspace, workspaceSize);
+#else
+        size_t const hSize = HUF_readDTableX2_wksp(entropy->hufTable,
+                                                dictPtr, (size_t)(dictEnd - dictPtr),
+                                                workspace, workspaceSize);
+#endif
+        RETURN_ERROR_IF(HUF_isError(hSize), dictionary_corrupted, "");
+        dictPtr += hSize;
+    }
+
+    {   short offcodeNCount[MaxOff+1];
+        unsigned offcodeMaxValue = MaxOff, offcodeLog;
+        size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, (size_t)(dictEnd-dictPtr));
+        RETURN_ERROR_IF(FSE_isError(offcodeHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(offcodeMaxValue > MaxOff, dictionary_corrupted, "");
+        RETURN_ERROR_IF(offcodeLog > OffFSELog, dictionary_corrupted, "");
+        ZSTD_buildFSETable( entropy->OFTable,
+                            offcodeNCount, offcodeMaxValue,
+                            OF_base, OF_bits,
+                            offcodeLog,
+                            entropy->workspace, sizeof(entropy->workspace),
+                            /* bmi2 */0);
+        dictPtr += offcodeHeaderSize;
+    }
+
+    {   short matchlengthNCount[MaxML+1];
+        unsigned matchlengthMaxValue = MaxML, matchlengthLog;
+        size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, (size_t)(dictEnd-dictPtr));
+        RETURN_ERROR_IF(FSE_isError(matchlengthHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(matchlengthMaxValue > MaxML, dictionary_corrupted, "");
+        RETURN_ERROR_IF(matchlengthLog > MLFSELog, dictionary_corrupted, "");
+        ZSTD_buildFSETable( entropy->MLTable,
+                            matchlengthNCount, matchlengthMaxValue,
+                            ML_base, ML_bits,
+                            matchlengthLog,
+                            entropy->workspace, sizeof(entropy->workspace),
+                            /* bmi2 */ 0);
+        dictPtr += matchlengthHeaderSize;
+    }
+
+    {   short litlengthNCount[MaxLL+1];
+        unsigned litlengthMaxValue = MaxLL, litlengthLog;
+        size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, (size_t)(dictEnd-dictPtr));
+        RETURN_ERROR_IF(FSE_isError(litlengthHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(litlengthMaxValue > MaxLL, dictionary_corrupted, "");
+        RETURN_ERROR_IF(litlengthLog > LLFSELog, dictionary_corrupted, "");
+        ZSTD_buildFSETable( entropy->LLTable,
+                            litlengthNCount, litlengthMaxValue,
+                            LL_base, LL_bits,
+                            litlengthLog,
+                            entropy->workspace, sizeof(entropy->workspace),
+                            /* bmi2 */ 0);
+        dictPtr += litlengthHeaderSize;
+    }
+
+    RETURN_ERROR_IF(dictPtr+12 > dictEnd, dictionary_corrupted, "");
+    {   int i;
+        size_t const dictContentSize = (size_t)(dictEnd - (dictPtr+12));
+        for (i=0; i<3; i++) {
+            U32 const rep = MEM_readLE32(dictPtr); dictPtr += 4;
+            RETURN_ERROR_IF(rep==0 || rep > dictContentSize,
+                            dictionary_corrupted, "");
+            entropy->rep[i] = rep;
+    }   }
+
+    return (size_t)(dictPtr - (const BYTE*)dict);
+}
+
+static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    if (dictSize < 8) return ZSTD_refDictContent(dctx, dict, dictSize);
+    {   U32 const magic = MEM_readLE32(dict);
+        if (magic != ZSTD_MAGIC_DICTIONARY) {
+            return ZSTD_refDictContent(dctx, dict, dictSize);   /* pure content mode */
+    }   }
+    dctx->dictID = MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE);
+
+    /* load entropy tables */
+    {   size_t const eSize = ZSTD_loadDEntropy(&dctx->entropy, dict, dictSize);
+        RETURN_ERROR_IF(ZSTD_isError(eSize), dictionary_corrupted, "");
+        dict = (const char*)dict + eSize;
+        dictSize -= eSize;
+    }
+    dctx->litEntropy = dctx->fseEntropy = 1;
+
+    /* reference dictionary content */
+    return ZSTD_refDictContent(dctx, dict, dictSize);
+}
+
+size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx)
+{
+    assert(dctx != NULL);
+    dctx->expected = ZSTD_startingInputLength(dctx->format);  /* dctx->format must be properly set */
+    dctx->stage = ZSTDds_getFrameHeaderSize;
+    dctx->processedCSize = 0;
+    dctx->decodedSize = 0;
+    dctx->previousDstEnd = NULL;
+    dctx->prefixStart = NULL;
+    dctx->virtualStart = NULL;
+    dctx->dictEnd = NULL;
+    dctx->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001);  /* cover both little and big endian */
+    dctx->litEntropy = dctx->fseEntropy = 0;
+    dctx->dictID = 0;
+    dctx->bType = bt_reserved;
+    ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.rep) == sizeof(repStartValue));
+    ZSTD_memcpy(dctx->entropy.rep, repStartValue, sizeof(repStartValue));  /* initial repcodes */
+    dctx->LLTptr = dctx->entropy.LLTable;
+    dctx->MLTptr = dctx->entropy.MLTable;
+    dctx->OFTptr = dctx->entropy.OFTable;
+    dctx->HUFptr = dctx->entropy.hufTable;
+    return 0;
+}
+
+size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    FORWARD_IF_ERROR( ZSTD_decompressBegin(dctx) , "");
+    if (dict && dictSize)
+        RETURN_ERROR_IF(
+            ZSTD_isError(ZSTD_decompress_insertDictionary(dctx, dict, dictSize)),
+            dictionary_corrupted, "");
+    return 0;
+}
+
+
+/* ======   ZSTD_DDict   ====== */
+
+size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
+{
+    DEBUGLOG(4, "ZSTD_decompressBegin_usingDDict");
+    assert(dctx != NULL);
+    if (ddict) {
+        const char* const dictStart = (const char*)ZSTD_DDict_dictContent(ddict);
+        size_t const dictSize = ZSTD_DDict_dictSize(ddict);
+        const void* const dictEnd = dictStart + dictSize;
+        dctx->ddictIsCold = (dctx->dictEnd != dictEnd);
+        DEBUGLOG(4, "DDict is %s",
+                    dctx->ddictIsCold ? "~cold~" : "hot!");
+    }
+    FORWARD_IF_ERROR( ZSTD_decompressBegin(dctx) , "");
+    if (ddict) {   /* NULL ddict is equivalent to no dictionary */
+        ZSTD_copyDDictParameters(dctx, ddict);
+    }
+    return 0;
+}
+
+/*! ZSTD_getDictID_fromDict() :
+ *  Provides the dictID stored within dictionary.
+ *  if @return == 0, the dictionary is not conformant with Zstandard specification.
+ *  It can still be loaded, but as a content-only dictionary. */
+unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize)
+{
+    if (dictSize < 8) return 0;
+    if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) return 0;
+    return MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE);
+}
+
+/*! ZSTD_getDictID_fromFrame() :
+ *  Provides the dictID required to decompress frame stored within `src`.
+ *  If @return == 0, the dictID could not be decoded.
+ *  This could for one of the following reasons :
+ *  - The frame does not require a dictionary (most common case).
+ *  - The frame was built with dictID intentionally removed.
+ *    Needed dictionary is a hidden information.
+ *    Note : this use case also happens when using a non-conformant dictionary.
+ *  - `srcSize` is too small, and as a result, frame header could not be decoded.
+ *    Note : possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`.
+ *  - This is not a Zstandard frame.
+ *  When identifying the exact failure cause, it's possible to use
+ *  ZSTD_getFrameHeader(), which will provide a more precise error code. */
+unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize)
+{
+    ZSTD_frameHeader zfp = { 0, 0, 0, ZSTD_frame, 0, 0, 0 };
+    size_t const hError = ZSTD_getFrameHeader(&zfp, src, srcSize);
+    if (ZSTD_isError(hError)) return 0;
+    return zfp.dictID;
+}
+
+
+/*! ZSTD_decompress_usingDDict() :
+*   Decompression using a pre-digested Dictionary
+*   Use dictionary without significant overhead. */
+size_t ZSTD_decompress_usingDDict(ZSTD_DCtx* dctx,
+                                  void* dst, size_t dstCapacity,
+                            const void* src, size_t srcSize,
+                            const ZSTD_DDict* ddict)
+{
+    /* pass content and size in case legacy frames are encountered */
+    return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize,
+                                     NULL, 0,
+                                     ddict);
+}
+
+
+/*=====================================
+*   Streaming decompression
+*====================================*/
+
+ZSTD_DStream* ZSTD_createDStream(void)
+{
+    DEBUGLOG(3, "ZSTD_createDStream");
+    return ZSTD_createDStream_advanced(ZSTD_defaultCMem);
+}
+
+ZSTD_DStream* ZSTD_initStaticDStream(void *workspace, size_t workspaceSize)
+{
+    return ZSTD_initStaticDCtx(workspace, workspaceSize);
+}
+
+ZSTD_DStream* ZSTD_createDStream_advanced(ZSTD_customMem customMem)
+{
+    return ZSTD_createDCtx_advanced(customMem);
+}
+
+size_t ZSTD_freeDStream(ZSTD_DStream* zds)
+{
+    return ZSTD_freeDCtx(zds);
+}
+
+
+/* ***  Initialization  *** */
+
+size_t ZSTD_DStreamInSize(void)  { return ZSTD_BLOCKSIZE_MAX + ZSTD_blockHeaderSize; }
+size_t ZSTD_DStreamOutSize(void) { return ZSTD_BLOCKSIZE_MAX; }
+
+size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx,
+                                   const void* dict, size_t dictSize,
+                                         ZSTD_dictLoadMethod_e dictLoadMethod,
+                                         ZSTD_dictContentType_e dictContentType)
+{
+    RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+    ZSTD_clearDict(dctx);
+    if (dict && dictSize != 0) {
+        dctx->ddictLocal = ZSTD_createDDict_advanced(dict, dictSize, dictLoadMethod, dictContentType, dctx->customMem);
+        RETURN_ERROR_IF(dctx->ddictLocal == NULL, memory_allocation, "NULL pointer!");
+        dctx->ddict = dctx->ddictLocal;
+        dctx->dictUses = ZSTD_use_indefinitely;
+    }
+    return 0;
+}
+
+size_t ZSTD_DCtx_loadDictionary_byReference(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    return ZSTD_DCtx_loadDictionary_advanced(dctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto);
+}
+
+size_t ZSTD_DCtx_loadDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
+{
+    return ZSTD_DCtx_loadDictionary_advanced(dctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto);
+}
+
+size_t ZSTD_DCtx_refPrefix_advanced(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType)
+{
+    FORWARD_IF_ERROR(ZSTD_DCtx_loadDictionary_advanced(dctx, prefix, prefixSize, ZSTD_dlm_byRef, dictContentType), "");
+    dctx->dictUses = ZSTD_use_once;
+    return 0;
+}
+
+size_t ZSTD_DCtx_refPrefix(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize)
+{
+    return ZSTD_DCtx_refPrefix_advanced(dctx, prefix, prefixSize, ZSTD_dct_rawContent);
+}
+
+
+/* ZSTD_initDStream_usingDict() :
+ * return : expected size, aka ZSTD_startingInputLength().
+ * this function cannot fail */
+size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize)
+{
+    DEBUGLOG(4, "ZSTD_initDStream_usingDict");
+    FORWARD_IF_ERROR( ZSTD_DCtx_reset(zds, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_DCtx_loadDictionary(zds, dict, dictSize) , "");
+    return ZSTD_startingInputLength(zds->format);
+}
+
+/* note : this variant can't fail */
+size_t ZSTD_initDStream(ZSTD_DStream* zds)
+{
+    DEBUGLOG(4, "ZSTD_initDStream");
+    return ZSTD_initDStream_usingDDict(zds, NULL);
+}
+
+/* ZSTD_initDStream_usingDDict() :
+ * ddict will just be referenced, and must outlive decompression session
+ * this function cannot fail */
+size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* dctx, const ZSTD_DDict* ddict)
+{
+    FORWARD_IF_ERROR( ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_DCtx_refDDict(dctx, ddict) , "");
+    return ZSTD_startingInputLength(dctx->format);
+}
+
+/* ZSTD_resetDStream() :
+ * return : expected size, aka ZSTD_startingInputLength().
+ * this function cannot fail */
+size_t ZSTD_resetDStream(ZSTD_DStream* dctx)
+{
+    FORWARD_IF_ERROR(ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only), "");
+    return ZSTD_startingInputLength(dctx->format);
+}
+
+
+size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
+{
+    RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+    ZSTD_clearDict(dctx);
+    if (ddict) {
+        dctx->ddict = ddict;
+        dctx->dictUses = ZSTD_use_indefinitely;
+        if (dctx->refMultipleDDicts == ZSTD_rmd_refMultipleDDicts) {
+            if (dctx->ddictSet == NULL) {
+                dctx->ddictSet = ZSTD_createDDictHashSet(dctx->customMem);
+                if (!dctx->ddictSet) {
+                    RETURN_ERROR(memory_allocation, "Failed to allocate memory for hash set!");
+                }
+            }
+            assert(!dctx->staticSize);  /* Impossible: ddictSet cannot have been allocated if static dctx */
+            FORWARD_IF_ERROR(ZSTD_DDictHashSet_addDDict(dctx->ddictSet, ddict, dctx->customMem), "");
+        }
+    }
+    return 0;
+}
+
+/* ZSTD_DCtx_setMaxWindowSize() :
+ * note : no direct equivalence in ZSTD_DCtx_setParameter,
+ * since this version sets windowSize, and the other sets windowLog */
+size_t ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx* dctx, size_t maxWindowSize)
+{
+    ZSTD_bounds const bounds = ZSTD_dParam_getBounds(ZSTD_d_windowLogMax);
+    size_t const min = (size_t)1 << bounds.lowerBound;
+    size_t const max = (size_t)1 << bounds.upperBound;
+    RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+    RETURN_ERROR_IF(maxWindowSize < min, parameter_outOfBound, "");
+    RETURN_ERROR_IF(maxWindowSize > max, parameter_outOfBound, "");
+    dctx->maxWindowSize = maxWindowSize;
+    return 0;
+}
+
+size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format)
+{
+    return ZSTD_DCtx_setParameter(dctx, ZSTD_d_format, (int)format);
+}
+
+ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam)
+{
+    ZSTD_bounds bounds = { 0, 0, 0 };
+    switch(dParam) {
+        case ZSTD_d_windowLogMax:
+            bounds.lowerBound = ZSTD_WINDOWLOG_ABSOLUTEMIN;
+            bounds.upperBound = ZSTD_WINDOWLOG_MAX;
+            return bounds;
+        case ZSTD_d_format:
+            bounds.lowerBound = (int)ZSTD_f_zstd1;
+            bounds.upperBound = (int)ZSTD_f_zstd1_magicless;
+            ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless);
+            return bounds;
+        case ZSTD_d_stableOutBuffer:
+            bounds.lowerBound = (int)ZSTD_bm_buffered;
+            bounds.upperBound = (int)ZSTD_bm_stable;
+            return bounds;
+        case ZSTD_d_forceIgnoreChecksum:
+            bounds.lowerBound = (int)ZSTD_d_validateChecksum;
+            bounds.upperBound = (int)ZSTD_d_ignoreChecksum;
+            return bounds;
+        case ZSTD_d_refMultipleDDicts:
+            bounds.lowerBound = (int)ZSTD_rmd_refSingleDDict;
+            bounds.upperBound = (int)ZSTD_rmd_refMultipleDDicts;
+            return bounds;
+        default:;
+    }
+    bounds.error = ERROR(parameter_unsupported);
+    return bounds;
+}
+
+/* ZSTD_dParam_withinBounds:
+ * @return 1 if value is within dParam bounds,
+ * 0 otherwise */
+static int ZSTD_dParam_withinBounds(ZSTD_dParameter dParam, int value)
+{
+    ZSTD_bounds const bounds = ZSTD_dParam_getBounds(dParam);
+    if (ZSTD_isError(bounds.error)) return 0;
+    if (value < bounds.lowerBound) return 0;
+    if (value > bounds.upperBound) return 0;
+    return 1;
+}
+
+#define CHECK_DBOUNDS(p,v) {                \
+    RETURN_ERROR_IF(!ZSTD_dParam_withinBounds(p, v), parameter_outOfBound, ""); \
+}
+
+size_t ZSTD_DCtx_getParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int* value)
+{
+    switch (param) {
+        case ZSTD_d_windowLogMax:
+            *value = (int)ZSTD_highbit32((U32)dctx->maxWindowSize);
+            return 0;
+        case ZSTD_d_format:
+            *value = (int)dctx->format;
+            return 0;
+        case ZSTD_d_stableOutBuffer:
+            *value = (int)dctx->outBufferMode;
+            return 0;
+        case ZSTD_d_forceIgnoreChecksum:
+            *value = (int)dctx->forceIgnoreChecksum;
+            return 0;
+        case ZSTD_d_refMultipleDDicts:
+            *value = (int)dctx->refMultipleDDicts;
+            return 0;
+        default:;
+    }
+    RETURN_ERROR(parameter_unsupported, "");
+}
+
+size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter dParam, int value)
+{
+    RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+    switch(dParam) {
+        case ZSTD_d_windowLogMax:
+            if (value == 0) value = ZSTD_WINDOWLOG_LIMIT_DEFAULT;
+            CHECK_DBOUNDS(ZSTD_d_windowLogMax, value);
+            dctx->maxWindowSize = ((size_t)1) << value;
+            return 0;
+        case ZSTD_d_format:
+            CHECK_DBOUNDS(ZSTD_d_format, value);
+            dctx->format = (ZSTD_format_e)value;
+            return 0;
+        case ZSTD_d_stableOutBuffer:
+            CHECK_DBOUNDS(ZSTD_d_stableOutBuffer, value);
+            dctx->outBufferMode = (ZSTD_bufferMode_e)value;
+            return 0;
+        case ZSTD_d_forceIgnoreChecksum:
+            CHECK_DBOUNDS(ZSTD_d_forceIgnoreChecksum, value);
+            dctx->forceIgnoreChecksum = (ZSTD_forceIgnoreChecksum_e)value;
+            return 0;
+        case ZSTD_d_refMultipleDDicts:
+            CHECK_DBOUNDS(ZSTD_d_refMultipleDDicts, value);
+            if (dctx->staticSize != 0) {
+                RETURN_ERROR(parameter_unsupported, "Static dctx does not support multiple DDicts!");
+            }
+            dctx->refMultipleDDicts = (ZSTD_refMultipleDDicts_e)value;
+            return 0;
+        default:;
+    }
+    RETURN_ERROR(parameter_unsupported, "");
+}
+
+size_t ZSTD_DCtx_reset(ZSTD_DCtx* dctx, ZSTD_ResetDirective reset)
+{
+    if ( (reset == ZSTD_reset_session_only)
+      || (reset == ZSTD_reset_session_and_parameters) ) {
+        dctx->streamStage = zdss_init;
+        dctx->noForwardProgress = 0;
+    }
+    if ( (reset == ZSTD_reset_parameters)
+      || (reset == ZSTD_reset_session_and_parameters) ) {
+        RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
+        ZSTD_clearDict(dctx);
+        ZSTD_DCtx_resetParameters(dctx);
+    }
+    return 0;
+}
+
+
+size_t ZSTD_sizeof_DStream(const ZSTD_DStream* dctx)
+{
+    return ZSTD_sizeof_DCtx(dctx);
+}
+
+size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize)
+{
+    size_t const blockSize = (size_t) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
+    unsigned long long const neededRBSize = windowSize + blockSize + (WILDCOPY_OVERLENGTH * 2);
+    unsigned long long const neededSize = MIN(frameContentSize, neededRBSize);
+    size_t const minRBSize = (size_t) neededSize;
+    RETURN_ERROR_IF((unsigned long long)minRBSize != neededSize,
+                    frameParameter_windowTooLarge, "");
+    return minRBSize;
+}
+
+size_t ZSTD_estimateDStreamSize(size_t windowSize)
+{
+    size_t const blockSize = MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
+    size_t const inBuffSize = blockSize;  /* no block can be larger */
+    size_t const outBuffSize = ZSTD_decodingBufferSize_min(windowSize, ZSTD_CONTENTSIZE_UNKNOWN);
+    return ZSTD_estimateDCtxSize() + inBuffSize + outBuffSize;
+}
+
+size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, size_t srcSize)
+{
+    U32 const windowSizeMax = 1U << ZSTD_WINDOWLOG_MAX;   /* note : should be user-selectable, but requires an additional parameter (or a dctx) */
+    ZSTD_frameHeader zfh;
+    size_t const err = ZSTD_getFrameHeader(&zfh, src, srcSize);
+    if (ZSTD_isError(err)) return err;
+    RETURN_ERROR_IF(err>0, srcSize_wrong, "");
+    RETURN_ERROR_IF(zfh.windowSize > windowSizeMax,
+                    frameParameter_windowTooLarge, "");
+    return ZSTD_estimateDStreamSize((size_t)zfh.windowSize);
+}
+
+
+/* *****   Decompression   ***** */
+
+static int ZSTD_DCtx_isOverflow(ZSTD_DStream* zds, size_t const neededInBuffSize, size_t const neededOutBuffSize)
+{
+    return (zds->inBuffSize + zds->outBuffSize) >= (neededInBuffSize + neededOutBuffSize) * ZSTD_WORKSPACETOOLARGE_FACTOR;
+}
+
+static void ZSTD_DCtx_updateOversizedDuration(ZSTD_DStream* zds, size_t const neededInBuffSize, size_t const neededOutBuffSize)
+{
+    if (ZSTD_DCtx_isOverflow(zds, neededInBuffSize, neededOutBuffSize))
+        zds->oversizedDuration++;
+    else
+        zds->oversizedDuration = 0;
+}
+
+static int ZSTD_DCtx_isOversizedTooLong(ZSTD_DStream* zds)
+{
+    return zds->oversizedDuration >= ZSTD_WORKSPACETOOLARGE_MAXDURATION;
+}
+
+/* Checks that the output buffer hasn't changed if ZSTD_obm_stable is used. */
+static size_t ZSTD_checkOutBuffer(ZSTD_DStream const* zds, ZSTD_outBuffer const* output)
+{
+    ZSTD_outBuffer const expect = zds->expectedOutBuffer;
+    /* No requirement when ZSTD_obm_stable is not enabled. */
+    if (zds->outBufferMode != ZSTD_bm_stable)
+        return 0;
+    /* Any buffer is allowed in zdss_init, this must be the same for every other call until
+     * the context is reset.
+     */
+    if (zds->streamStage == zdss_init)
+        return 0;
+    /* The buffer must match our expectation exactly. */
+    if (expect.dst == output->dst && expect.pos == output->pos && expect.size == output->size)
+        return 0;
+    RETURN_ERROR(dstBuffer_wrong, "ZSTD_d_stableOutBuffer enabled but output differs!");
+}
+
+/* Calls ZSTD_decompressContinue() with the right parameters for ZSTD_decompressStream()
+ * and updates the stage and the output buffer state. This call is extracted so it can be
+ * used both when reading directly from the ZSTD_inBuffer, and in buffered input mode.
+ * NOTE: You must break after calling this function since the streamStage is modified.
+ */
+static size_t ZSTD_decompressContinueStream(
+            ZSTD_DStream* zds, char** op, char* oend,
+            void const* src, size_t srcSize) {
+    int const isSkipFrame = ZSTD_isSkipFrame(zds);
+    if (zds->outBufferMode == ZSTD_bm_buffered) {
+        size_t const dstSize = isSkipFrame ? 0 : zds->outBuffSize - zds->outStart;
+        size_t const decodedSize = ZSTD_decompressContinue(zds,
+                zds->outBuff + zds->outStart, dstSize, src, srcSize);
+        FORWARD_IF_ERROR(decodedSize, "");
+        if (!decodedSize && !isSkipFrame) {
+            zds->streamStage = zdss_read;
+        } else {
+            zds->outEnd = zds->outStart + decodedSize;
+            zds->streamStage = zdss_flush;
+        }
+    } else {
+        /* Write directly into the output buffer */
+        size_t const dstSize = isSkipFrame ? 0 : (size_t)(oend - *op);
+        size_t const decodedSize = ZSTD_decompressContinue(zds, *op, dstSize, src, srcSize);
+        FORWARD_IF_ERROR(decodedSize, "");
+        *op += decodedSize;
+        /* Flushing is not needed. */
+        zds->streamStage = zdss_read;
+        assert(*op <= oend);
+        assert(zds->outBufferMode == ZSTD_bm_stable);
+    }
+    return 0;
+}
+
+size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input)
+{
+    const char* const src = (const char*)input->src;
+    const char* const istart = input->pos != 0 ? src + input->pos : src;
+    const char* const iend = input->size != 0 ? src + input->size : src;
+    const char* ip = istart;
+    char* const dst = (char*)output->dst;
+    char* const ostart = output->pos != 0 ? dst + output->pos : dst;
+    char* const oend = output->size != 0 ? dst + output->size : dst;
+    char* op = ostart;
+    U32 someMoreWork = 1;
+
+    DEBUGLOG(5, "ZSTD_decompressStream");
+    RETURN_ERROR_IF(
+        input->pos > input->size,
+        srcSize_wrong,
+        "forbidden. in: pos: %u   vs size: %u",
+        (U32)input->pos, (U32)input->size);
+    RETURN_ERROR_IF(
+        output->pos > output->size,
+        dstSize_tooSmall,
+        "forbidden. out: pos: %u   vs size: %u",
+        (U32)output->pos, (U32)output->size);
+    DEBUGLOG(5, "input size : %u", (U32)(input->size - input->pos));
+    FORWARD_IF_ERROR(ZSTD_checkOutBuffer(zds, output), "");
+
+    while (someMoreWork) {
+        switch(zds->streamStage)
+        {
+        case zdss_init :
+            DEBUGLOG(5, "stage zdss_init => transparent reset ");
+            zds->streamStage = zdss_loadHeader;
+            zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
+            zds->legacyVersion = 0;
+            zds->hostageByte = 0;
+            zds->expectedOutBuffer = *output;
+            ZSTD_FALLTHROUGH;
+
+        case zdss_loadHeader :
+            DEBUGLOG(5, "stage zdss_loadHeader (srcSize : %u)", (U32)(iend - ip));
+            {   size_t const hSize = ZSTD_getFrameHeader_advanced(&zds->fParams, zds->headerBuffer, zds->lhSize, zds->format);
+                if (zds->refMultipleDDicts && zds->ddictSet) {
+                    ZSTD_DCtx_selectFrameDDict(zds);
+                }
+                DEBUGLOG(5, "header size : %u", (U32)hSize);
+                if (ZSTD_isError(hSize)) {
+                    return hSize;   /* error */
+                }
+                if (hSize != 0) {   /* need more input */
+                    size_t const toLoad = hSize - zds->lhSize;   /* if hSize!=0, hSize > zds->lhSize */
+                    size_t const remainingInput = (size_t)(iend-ip);
+                    assert(iend >= ip);
+                    if (toLoad > remainingInput) {   /* not enough input to load full header */
+                        if (remainingInput > 0) {
+                            ZSTD_memcpy(zds->headerBuffer + zds->lhSize, ip, remainingInput);
+                            zds->lhSize += remainingInput;
+                        }
+                        input->pos = input->size;
+                        return (MAX((size_t)ZSTD_FRAMEHEADERSIZE_MIN(zds->format), hSize) - zds->lhSize) + ZSTD_blockHeaderSize;   /* remaining header bytes + next block header */
+                    }
+                    assert(ip != NULL);
+                    ZSTD_memcpy(zds->headerBuffer + zds->lhSize, ip, toLoad); zds->lhSize = hSize; ip += toLoad;
+                    break;
+            }   }
+
+            /* check for single-pass mode opportunity */
+            if (zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
+                && zds->fParams.frameType != ZSTD_skippableFrame
+                && (U64)(size_t)(oend-op) >= zds->fParams.frameContentSize) {
+                size_t const cSize = ZSTD_findFrameCompressedSize(istart, (size_t)(iend-istart));
+                if (cSize <= (size_t)(iend-istart)) {
+                    /* shortcut : using single-pass mode */
+                    size_t const decompressedSize = ZSTD_decompress_usingDDict(zds, op, (size_t)(oend-op), istart, cSize, ZSTD_getDDict(zds));
+                    if (ZSTD_isError(decompressedSize)) return decompressedSize;
+                    DEBUGLOG(4, "shortcut to single-pass ZSTD_decompress_usingDDict()")
+                    ip = istart + cSize;
+                    op += decompressedSize;
+                    zds->expected = 0;
+                    zds->streamStage = zdss_init;
+                    someMoreWork = 0;
+                    break;
+            }   }
+
+            /* Check output buffer is large enough for ZSTD_odm_stable. */
+            if (zds->outBufferMode == ZSTD_bm_stable
+                && zds->fParams.frameType != ZSTD_skippableFrame
+                && zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
+                && (U64)(size_t)(oend-op) < zds->fParams.frameContentSize) {
+                RETURN_ERROR(dstSize_tooSmall, "ZSTD_obm_stable passed but ZSTD_outBuffer is too small");
+            }
+
+            /* Consume header (see ZSTDds_decodeFrameHeader) */
+            DEBUGLOG(4, "Consume header");
+            FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(zds, ZSTD_getDDict(zds)), "");
+
+            if ((MEM_readLE32(zds->headerBuffer) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {  /* skippable frame */
+                zds->expected = MEM_readLE32(zds->headerBuffer + ZSTD_FRAMEIDSIZE);
+                zds->stage = ZSTDds_skipFrame;
+            } else {
+                FORWARD_IF_ERROR(ZSTD_decodeFrameHeader(zds, zds->headerBuffer, zds->lhSize), "");
+                zds->expected = ZSTD_blockHeaderSize;
+                zds->stage = ZSTDds_decodeBlockHeader;
+            }
+
+            /* control buffer memory usage */
+            DEBUGLOG(4, "Control max memory usage (%u KB <= max %u KB)",
+                        (U32)(zds->fParams.windowSize >>10),
+                        (U32)(zds->maxWindowSize >> 10) );
+            zds->fParams.windowSize = MAX(zds->fParams.windowSize, 1U << ZSTD_WINDOWLOG_ABSOLUTEMIN);
+            RETURN_ERROR_IF(zds->fParams.windowSize > zds->maxWindowSize,
+                            frameParameter_windowTooLarge, "");
+
+            /* Adapt buffer sizes to frame header instructions */
+            {   size_t const neededInBuffSize = MAX(zds->fParams.blockSizeMax, 4 /* frame checksum */);
+                size_t const neededOutBuffSize = zds->outBufferMode == ZSTD_bm_buffered
+                        ? ZSTD_decodingBufferSize_min(zds->fParams.windowSize, zds->fParams.frameContentSize)
+                        : 0;
+
+                ZSTD_DCtx_updateOversizedDuration(zds, neededInBuffSize, neededOutBuffSize);
+
+                {   int const tooSmall = (zds->inBuffSize < neededInBuffSize) || (zds->outBuffSize < neededOutBuffSize);
+                    int const tooLarge = ZSTD_DCtx_isOversizedTooLong(zds);
+
+                    if (tooSmall || tooLarge) {
+                        size_t const bufferSize = neededInBuffSize + neededOutBuffSize;
+                        DEBUGLOG(4, "inBuff  : from %u to %u",
+                                    (U32)zds->inBuffSize, (U32)neededInBuffSize);
+                        DEBUGLOG(4, "outBuff : from %u to %u",
+                                    (U32)zds->outBuffSize, (U32)neededOutBuffSize);
+                        if (zds->staticSize) {  /* static DCtx */
+                            DEBUGLOG(4, "staticSize : %u", (U32)zds->staticSize);
+                            assert(zds->staticSize >= sizeof(ZSTD_DCtx));  /* controlled at init */
+                            RETURN_ERROR_IF(
+                                bufferSize > zds->staticSize - sizeof(ZSTD_DCtx),
+                                memory_allocation, "");
+                        } else {
+                            ZSTD_customFree(zds->inBuff, zds->customMem);
+                            zds->inBuffSize = 0;
+                            zds->outBuffSize = 0;
+                            zds->inBuff = (char*)ZSTD_customMalloc(bufferSize, zds->customMem);
+                            RETURN_ERROR_IF(zds->inBuff == NULL, memory_allocation, "");
+                        }
+                        zds->inBuffSize = neededInBuffSize;
+                        zds->outBuff = zds->inBuff + zds->inBuffSize;
+                        zds->outBuffSize = neededOutBuffSize;
+            }   }   }
+            zds->streamStage = zdss_read;
+            ZSTD_FALLTHROUGH;
+
+        case zdss_read:
+            DEBUGLOG(5, "stage zdss_read");
+            {   size_t const neededInSize = ZSTD_nextSrcSizeToDecompressWithInputSize(zds, (size_t)(iend - ip));
+                DEBUGLOG(5, "neededInSize = %u", (U32)neededInSize);
+                if (neededInSize==0) {  /* end of frame */
+                    zds->streamStage = zdss_init;
+                    someMoreWork = 0;
+                    break;
+                }
+                if ((size_t)(iend-ip) >= neededInSize) {  /* decode directly from src */
+                    FORWARD_IF_ERROR(ZSTD_decompressContinueStream(zds, &op, oend, ip, neededInSize), "");
+                    ip += neededInSize;
+                    /* Function modifies the stage so we must break */
+                    break;
+            }   }
+            if (ip==iend) { someMoreWork = 0; break; }   /* no more input */
+            zds->streamStage = zdss_load;
+            ZSTD_FALLTHROUGH;
+
+        case zdss_load:
+            {   size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds);
+                size_t const toLoad = neededInSize - zds->inPos;
+                int const isSkipFrame = ZSTD_isSkipFrame(zds);
+                size_t loadedSize;
+                /* At this point we shouldn't be decompressing a block that we can stream. */
+                assert(neededInSize == ZSTD_nextSrcSizeToDecompressWithInputSize(zds, iend - ip));
+                if (isSkipFrame) {
+                    loadedSize = MIN(toLoad, (size_t)(iend-ip));
+                } else {
+                    RETURN_ERROR_IF(toLoad > zds->inBuffSize - zds->inPos,
+                                    corruption_detected,
+                                    "should never happen");
+                    loadedSize = ZSTD_limitCopy(zds->inBuff + zds->inPos, toLoad, ip, (size_t)(iend-ip));
+                }
+                ip += loadedSize;
+                zds->inPos += loadedSize;
+                if (loadedSize < toLoad) { someMoreWork = 0; break; }   /* not enough input, wait for more */
+
+                /* decode loaded input */
+                zds->inPos = 0;   /* input is consumed */
+                FORWARD_IF_ERROR(ZSTD_decompressContinueStream(zds, &op, oend, zds->inBuff, neededInSize), "");
+                /* Function modifies the stage so we must break */
+                break;
+            }
+        case zdss_flush:
+            {   size_t const toFlushSize = zds->outEnd - zds->outStart;
+                size_t const flushedSize = ZSTD_limitCopy(op, (size_t)(oend-op), zds->outBuff + zds->outStart, toFlushSize);
+                op += flushedSize;
+                zds->outStart += flushedSize;
+                if (flushedSize == toFlushSize) {  /* flush completed */
+                    zds->streamStage = zdss_read;
+                    if ( (zds->outBuffSize < zds->fParams.frameContentSize)
+                      && (zds->outStart + zds->fParams.blockSizeMax > zds->outBuffSize) ) {
+                        DEBUGLOG(5, "restart filling outBuff from beginning (left:%i, needed:%u)",
+                                (int)(zds->outBuffSize - zds->outStart),
+                                (U32)zds->fParams.blockSizeMax);
+                        zds->outStart = zds->outEnd = 0;
+                    }
+                    break;
+            }   }
+            /* cannot complete flush */
+            someMoreWork = 0;
+            break;
+
+        default:
+            assert(0);    /* impossible */
+            RETURN_ERROR(GENERIC, "impossible to reach");   /* some compiler require default to do something */
+    }   }
+
+    /* result */
+    input->pos = (size_t)(ip - (const char*)(input->src));
+    output->pos = (size_t)(op - (char*)(output->dst));
+
+    /* Update the expected output buffer for ZSTD_obm_stable. */
+    zds->expectedOutBuffer = *output;
+
+    if ((ip==istart) && (op==ostart)) {  /* no forward progress */
+        zds->noForwardProgress ++;
+        if (zds->noForwardProgress >= ZSTD_NO_FORWARD_PROGRESS_MAX) {
+            RETURN_ERROR_IF(op==oend, dstSize_tooSmall, "");
+            RETURN_ERROR_IF(ip==iend, srcSize_wrong, "");
+            assert(0);
+        }
+    } else {
+        zds->noForwardProgress = 0;
+    }
+    {   size_t nextSrcSizeHint = ZSTD_nextSrcSizeToDecompress(zds);
+        if (!nextSrcSizeHint) {   /* frame fully decoded */
+            if (zds->outEnd == zds->outStart) {  /* output fully flushed */
+                if (zds->hostageByte) {
+                    if (input->pos >= input->size) {
+                        /* can't release hostage (not present) */
+                        zds->streamStage = zdss_read;
+                        return 1;
+                    }
+                    input->pos++;  /* release hostage */
+                }   /* zds->hostageByte */
+                return 0;
+            }  /* zds->outEnd == zds->outStart */
+            if (!zds->hostageByte) { /* output not fully flushed; keep last byte as hostage; will be released when all output is flushed */
+                input->pos--;   /* note : pos > 0, otherwise, impossible to finish reading last block */
+                zds->hostageByte=1;
+            }
+            return 1;
+        }  /* nextSrcSizeHint==0 */
+        nextSrcSizeHint += ZSTD_blockHeaderSize * (ZSTD_nextInputType(zds) == ZSTDnit_block);   /* preload header of next block */
+        assert(zds->inPos <= nextSrcSizeHint);
+        nextSrcSizeHint -= zds->inPos;   /* part already loaded*/
+        return nextSrcSizeHint;
+    }
+}
+
+size_t ZSTD_decompressStream_simpleArgs (
+                            ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity, size_t* dstPos,
+                      const void* src, size_t srcSize, size_t* srcPos)
+{
+    ZSTD_outBuffer output = { dst, dstCapacity, *dstPos };
+    ZSTD_inBuffer  input  = { src, srcSize, *srcPos };
+    /* ZSTD_compress_generic() will check validity of dstPos and srcPos */
+    size_t const cErr = ZSTD_decompressStream(dctx, &output, &input);
+    *dstPos = output.pos;
+    *srcPos = input.pos;
+    return cErr;
+}
diff --git a/lib/zstd/decompress/zstd_decompress_block.c b/lib/zstd/decompress/zstd_decompress_block.c
new file mode 100644
index 0000000..2d101d9
--- /dev/null
+++ b/lib/zstd/decompress/zstd_decompress_block.c
@@ -0,0 +1,1540 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* zstd_decompress_block :
+ * this module takes care of decompressing _compressed_ block */
+
+/*-*******************************************************
+*  Dependencies
+*********************************************************/
+#include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
+#include "../common/compiler.h"    /* prefetch */
+#include "../common/cpu.h"         /* bmi2 */
+#include "../common/mem.h"         /* low level memory routines */
+#define FSE_STATIC_LINKING_ONLY
+#include "../common/fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "../common/huf.h"
+#include "../common/zstd_internal.h"
+#include "zstd_decompress_internal.h"   /* ZSTD_DCtx */
+#include "zstd_ddict.h"  /* ZSTD_DDictDictContent */
+#include "zstd_decompress_block.h"
+
+/*_*******************************************************
+*  Macros
+**********************************************************/
+
+/* These two optional macros force the use one way or another of the two
+ * ZSTD_decompressSequences implementations. You can't force in both directions
+ * at the same time.
+ */
+#if defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+#error "Cannot force the use of the short and the long ZSTD_decompressSequences variants!"
+#endif
+
+
+/*_*******************************************************
+*  Memory operations
+**********************************************************/
+static void ZSTD_copy4(void* dst, const void* src) { ZSTD_memcpy(dst, src, 4); }
+
+
+/*-*************************************************************
+ *   Block decoding
+ ***************************************************************/
+
+/*! ZSTD_getcBlockSize() :
+ *  Provides the size of compressed block from block header `src` */
+size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
+                          blockProperties_t* bpPtr)
+{
+    RETURN_ERROR_IF(srcSize < ZSTD_blockHeaderSize, srcSize_wrong, "");
+
+    {   U32 const cBlockHeader = MEM_readLE24(src);
+        U32 const cSize = cBlockHeader >> 3;
+        bpPtr->lastBlock = cBlockHeader & 1;
+        bpPtr->blockType = (blockType_e)((cBlockHeader >> 1) & 3);
+        bpPtr->origSize = cSize;   /* only useful for RLE */
+        if (bpPtr->blockType == bt_rle) return 1;
+        RETURN_ERROR_IF(bpPtr->blockType == bt_reserved, corruption_detected, "");
+        return cSize;
+    }
+}
+
+
+/* Hidden declaration for fullbench */
+size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
+                          const void* src, size_t srcSize);
+/*! ZSTD_decodeLiteralsBlock() :
+ * @return : nb of bytes read from src (< srcSize )
+ *  note : symbol not declared but exposed for fullbench */
+size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
+                          const void* src, size_t srcSize)   /* note : srcSize < BLOCKSIZE */
+{
+    DEBUGLOG(5, "ZSTD_decodeLiteralsBlock");
+    RETURN_ERROR_IF(srcSize < MIN_CBLOCK_SIZE, corruption_detected, "");
+
+    {   const BYTE* const istart = (const BYTE*) src;
+        symbolEncodingType_e const litEncType = (symbolEncodingType_e)(istart[0] & 3);
+
+        switch(litEncType)
+        {
+        case set_repeat:
+            DEBUGLOG(5, "set_repeat flag : re-using stats from previous compressed literals block");
+            RETURN_ERROR_IF(dctx->litEntropy==0, dictionary_corrupted, "");
+            ZSTD_FALLTHROUGH;
+
+        case set_compressed:
+            RETURN_ERROR_IF(srcSize < 5, corruption_detected, "srcSize >= MIN_CBLOCK_SIZE == 3; here we need up to 5 for case 3");
+            {   size_t lhSize, litSize, litCSize;
+                U32 singleStream=0;
+                U32 const lhlCode = (istart[0] >> 2) & 3;
+                U32 const lhc = MEM_readLE32(istart);
+                size_t hufSuccess;
+                switch(lhlCode)
+                {
+                case 0: case 1: default:   /* note : default is impossible, since lhlCode into [0..3] */
+                    /* 2 - 2 - 10 - 10 */
+                    singleStream = !lhlCode;
+                    lhSize = 3;
+                    litSize  = (lhc >> 4) & 0x3FF;
+                    litCSize = (lhc >> 14) & 0x3FF;
+                    break;
+                case 2:
+                    /* 2 - 2 - 14 - 14 */
+                    lhSize = 4;
+                    litSize  = (lhc >> 4) & 0x3FFF;
+                    litCSize = lhc >> 18;
+                    break;
+                case 3:
+                    /* 2 - 2 - 18 - 18 */
+                    lhSize = 5;
+                    litSize  = (lhc >> 4) & 0x3FFFF;
+                    litCSize = (lhc >> 22) + ((size_t)istart[4] << 10);
+                    break;
+                }
+                RETURN_ERROR_IF(litSize > ZSTD_BLOCKSIZE_MAX, corruption_detected, "");
+                RETURN_ERROR_IF(litCSize + lhSize > srcSize, corruption_detected, "");
+
+                /* prefetch huffman table if cold */
+                if (dctx->ddictIsCold && (litSize > 768 /* heuristic */)) {
+                    PREFETCH_AREA(dctx->HUFptr, sizeof(dctx->entropy.hufTable));
+                }
+
+                if (litEncType==set_repeat) {
+                    if (singleStream) {
+                        hufSuccess = HUF_decompress1X_usingDTable_bmi2(
+                            dctx->litBuffer, litSize, istart+lhSize, litCSize,
+                            dctx->HUFptr, dctx->bmi2);
+                    } else {
+                        hufSuccess = HUF_decompress4X_usingDTable_bmi2(
+                            dctx->litBuffer, litSize, istart+lhSize, litCSize,
+                            dctx->HUFptr, dctx->bmi2);
+                    }
+                } else {
+                    if (singleStream) {
+#if defined(HUF_FORCE_DECOMPRESS_X2)
+                        hufSuccess = HUF_decompress1X_DCtx_wksp(
+                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
+                            istart+lhSize, litCSize, dctx->workspace,
+                            sizeof(dctx->workspace));
+#else
+                        hufSuccess = HUF_decompress1X1_DCtx_wksp_bmi2(
+                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
+                            istart+lhSize, litCSize, dctx->workspace,
+                            sizeof(dctx->workspace), dctx->bmi2);
+#endif
+                    } else {
+                        hufSuccess = HUF_decompress4X_hufOnly_wksp_bmi2(
+                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
+                            istart+lhSize, litCSize, dctx->workspace,
+                            sizeof(dctx->workspace), dctx->bmi2);
+                    }
+                }
+
+                RETURN_ERROR_IF(HUF_isError(hufSuccess), corruption_detected, "");
+
+                dctx->litPtr = dctx->litBuffer;
+                dctx->litSize = litSize;
+                dctx->litEntropy = 1;
+                if (litEncType==set_compressed) dctx->HUFptr = dctx->entropy.hufTable;
+                ZSTD_memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
+                return litCSize + lhSize;
+            }
+
+        case set_basic:
+            {   size_t litSize, lhSize;
+                U32 const lhlCode = ((istart[0]) >> 2) & 3;
+                switch(lhlCode)
+                {
+                case 0: case 2: default:   /* note : default is impossible, since lhlCode into [0..3] */
+                    lhSize = 1;
+                    litSize = istart[0] >> 3;
+                    break;
+                case 1:
+                    lhSize = 2;
+                    litSize = MEM_readLE16(istart) >> 4;
+                    break;
+                case 3:
+                    lhSize = 3;
+                    litSize = MEM_readLE24(istart) >> 4;
+                    break;
+                }
+
+                if (lhSize+litSize+WILDCOPY_OVERLENGTH > srcSize) {  /* risk reading beyond src buffer with wildcopy */
+                    RETURN_ERROR_IF(litSize+lhSize > srcSize, corruption_detected, "");
+                    ZSTD_memcpy(dctx->litBuffer, istart+lhSize, litSize);
+                    dctx->litPtr = dctx->litBuffer;
+                    dctx->litSize = litSize;
+                    ZSTD_memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
+                    return lhSize+litSize;
+                }
+                /* direct reference into compressed stream */
+                dctx->litPtr = istart+lhSize;
+                dctx->litSize = litSize;
+                return lhSize+litSize;
+            }
+
+        case set_rle:
+            {   U32 const lhlCode = ((istart[0]) >> 2) & 3;
+                size_t litSize, lhSize;
+                switch(lhlCode)
+                {
+                case 0: case 2: default:   /* note : default is impossible, since lhlCode into [0..3] */
+                    lhSize = 1;
+                    litSize = istart[0] >> 3;
+                    break;
+                case 1:
+                    lhSize = 2;
+                    litSize = MEM_readLE16(istart) >> 4;
+                    break;
+                case 3:
+                    lhSize = 3;
+                    litSize = MEM_readLE24(istart) >> 4;
+                    RETURN_ERROR_IF(srcSize<4, corruption_detected, "srcSize >= MIN_CBLOCK_SIZE == 3; here we need lhSize+1 = 4");
+                    break;
+                }
+                RETURN_ERROR_IF(litSize > ZSTD_BLOCKSIZE_MAX, corruption_detected, "");
+                ZSTD_memset(dctx->litBuffer, istart[lhSize], litSize + WILDCOPY_OVERLENGTH);
+                dctx->litPtr = dctx->litBuffer;
+                dctx->litSize = litSize;
+                return lhSize+1;
+            }
+        default:
+            RETURN_ERROR(corruption_detected, "impossible");
+        }
+    }
+}
+
+/* Default FSE distribution tables.
+ * These are pre-calculated FSE decoding tables using default distributions as defined in specification :
+ * https://github.com/facebook/zstd/blob/release/doc/zstd_compression_format.md#default-distributions
+ * They were generated programmatically with following method :
+ * - start from default distributions, present in /lib/common/zstd_internal.h
+ * - generate tables normally, using ZSTD_buildFSETable()
+ * - printout the content of tables
+ * - pretify output, report below, test with fuzzer to ensure it's correct */
+
+/* Default FSE distribution table for Literal Lengths */
+static const ZSTD_seqSymbol LL_defaultDTable[(1<<LL_DEFAULTNORMLOG)+1] = {
+     {  1,  1,  1, LL_DEFAULTNORMLOG},  /* header : fastMode, tableLog */
+     /* nextState, nbAddBits, nbBits, baseVal */
+     {  0,  0,  4,    0},  { 16,  0,  4,    0},
+     { 32,  0,  5,    1},  {  0,  0,  5,    3},
+     {  0,  0,  5,    4},  {  0,  0,  5,    6},
+     {  0,  0,  5,    7},  {  0,  0,  5,    9},
+     {  0,  0,  5,   10},  {  0,  0,  5,   12},
+     {  0,  0,  6,   14},  {  0,  1,  5,   16},
+     {  0,  1,  5,   20},  {  0,  1,  5,   22},
+     {  0,  2,  5,   28},  {  0,  3,  5,   32},
+     {  0,  4,  5,   48},  { 32,  6,  5,   64},
+     {  0,  7,  5,  128},  {  0,  8,  6,  256},
+     {  0, 10,  6, 1024},  {  0, 12,  6, 4096},
+     { 32,  0,  4,    0},  {  0,  0,  4,    1},
+     {  0,  0,  5,    2},  { 32,  0,  5,    4},
+     {  0,  0,  5,    5},  { 32,  0,  5,    7},
+     {  0,  0,  5,    8},  { 32,  0,  5,   10},
+     {  0,  0,  5,   11},  {  0,  0,  6,   13},
+     { 32,  1,  5,   16},  {  0,  1,  5,   18},
+     { 32,  1,  5,   22},  {  0,  2,  5,   24},
+     { 32,  3,  5,   32},  {  0,  3,  5,   40},
+     {  0,  6,  4,   64},  { 16,  6,  4,   64},
+     { 32,  7,  5,  128},  {  0,  9,  6,  512},
+     {  0, 11,  6, 2048},  { 48,  0,  4,    0},
+     { 16,  0,  4,    1},  { 32,  0,  5,    2},
+     { 32,  0,  5,    3},  { 32,  0,  5,    5},
+     { 32,  0,  5,    6},  { 32,  0,  5,    8},
+     { 32,  0,  5,    9},  { 32,  0,  5,   11},
+     { 32,  0,  5,   12},  {  0,  0,  6,   15},
+     { 32,  1,  5,   18},  { 32,  1,  5,   20},
+     { 32,  2,  5,   24},  { 32,  2,  5,   28},
+     { 32,  3,  5,   40},  { 32,  4,  5,   48},
+     {  0, 16,  6,65536},  {  0, 15,  6,32768},
+     {  0, 14,  6,16384},  {  0, 13,  6, 8192},
+};   /* LL_defaultDTable */
+
+/* Default FSE distribution table for Offset Codes */
+static const ZSTD_seqSymbol OF_defaultDTable[(1<<OF_DEFAULTNORMLOG)+1] = {
+    {  1,  1,  1, OF_DEFAULTNORMLOG},  /* header : fastMode, tableLog */
+    /* nextState, nbAddBits, nbBits, baseVal */
+    {  0,  0,  5,    0},     {  0,  6,  4,   61},
+    {  0,  9,  5,  509},     {  0, 15,  5,32765},
+    {  0, 21,  5,2097149},   {  0,  3,  5,    5},
+    {  0,  7,  4,  125},     {  0, 12,  5, 4093},
+    {  0, 18,  5,262141},    {  0, 23,  5,8388605},
+    {  0,  5,  5,   29},     {  0,  8,  4,  253},
+    {  0, 14,  5,16381},     {  0, 20,  5,1048573},
+    {  0,  2,  5,    1},     { 16,  7,  4,  125},
+    {  0, 11,  5, 2045},     {  0, 17,  5,131069},
+    {  0, 22,  5,4194301},   {  0,  4,  5,   13},
+    { 16,  8,  4,  253},     {  0, 13,  5, 8189},
+    {  0, 19,  5,524285},    {  0,  1,  5,    1},
+    { 16,  6,  4,   61},     {  0, 10,  5, 1021},
+    {  0, 16,  5,65533},     {  0, 28,  5,268435453},
+    {  0, 27,  5,134217725}, {  0, 26,  5,67108861},
+    {  0, 25,  5,33554429},  {  0, 24,  5,16777213},
+};   /* OF_defaultDTable */
+
+
+/* Default FSE distribution table for Match Lengths */
+static const ZSTD_seqSymbol ML_defaultDTable[(1<<ML_DEFAULTNORMLOG)+1] = {
+    {  1,  1,  1, ML_DEFAULTNORMLOG},  /* header : fastMode, tableLog */
+    /* nextState, nbAddBits, nbBits, baseVal */
+    {  0,  0,  6,    3},  {  0,  0,  4,    4},
+    { 32,  0,  5,    5},  {  0,  0,  5,    6},
+    {  0,  0,  5,    8},  {  0,  0,  5,    9},
+    {  0,  0,  5,   11},  {  0,  0,  6,   13},
+    {  0,  0,  6,   16},  {  0,  0,  6,   19},
+    {  0,  0,  6,   22},  {  0,  0,  6,   25},
+    {  0,  0,  6,   28},  {  0,  0,  6,   31},
+    {  0,  0,  6,   34},  {  0,  1,  6,   37},
+    {  0,  1,  6,   41},  {  0,  2,  6,   47},
+    {  0,  3,  6,   59},  {  0,  4,  6,   83},
+    {  0,  7,  6,  131},  {  0,  9,  6,  515},
+    { 16,  0,  4,    4},  {  0,  0,  4,    5},
+    { 32,  0,  5,    6},  {  0,  0,  5,    7},
+    { 32,  0,  5,    9},  {  0,  0,  5,   10},
+    {  0,  0,  6,   12},  {  0,  0,  6,   15},
+    {  0,  0,  6,   18},  {  0,  0,  6,   21},
+    {  0,  0,  6,   24},  {  0,  0,  6,   27},
+    {  0,  0,  6,   30},  {  0,  0,  6,   33},
+    {  0,  1,  6,   35},  {  0,  1,  6,   39},
+    {  0,  2,  6,   43},  {  0,  3,  6,   51},
+    {  0,  4,  6,   67},  {  0,  5,  6,   99},
+    {  0,  8,  6,  259},  { 32,  0,  4,    4},
+    { 48,  0,  4,    4},  { 16,  0,  4,    5},
+    { 32,  0,  5,    7},  { 32,  0,  5,    8},
+    { 32,  0,  5,   10},  { 32,  0,  5,   11},
+    {  0,  0,  6,   14},  {  0,  0,  6,   17},
+    {  0,  0,  6,   20},  {  0,  0,  6,   23},
+    {  0,  0,  6,   26},  {  0,  0,  6,   29},
+    {  0,  0,  6,   32},  {  0, 16,  6,65539},
+    {  0, 15,  6,32771},  {  0, 14,  6,16387},
+    {  0, 13,  6, 8195},  {  0, 12,  6, 4099},
+    {  0, 11,  6, 2051},  {  0, 10,  6, 1027},
+};   /* ML_defaultDTable */
+
+
+static void ZSTD_buildSeqTable_rle(ZSTD_seqSymbol* dt, U32 baseValue, U32 nbAddBits)
+{
+    void* ptr = dt;
+    ZSTD_seqSymbol_header* const DTableH = (ZSTD_seqSymbol_header*)ptr;
+    ZSTD_seqSymbol* const cell = dt + 1;
+
+    DTableH->tableLog = 0;
+    DTableH->fastMode = 0;
+
+    cell->nbBits = 0;
+    cell->nextState = 0;
+    assert(nbAddBits < 255);
+    cell->nbAdditionalBits = (BYTE)nbAddBits;
+    cell->baseValue = baseValue;
+}
+
+
+/* ZSTD_buildFSETable() :
+ * generate FSE decoding table for one symbol (ll, ml or off)
+ * cannot fail if input is valid =>
+ * all inputs are presumed validated at this stage */
+FORCE_INLINE_TEMPLATE
+void ZSTD_buildFSETable_body(ZSTD_seqSymbol* dt,
+            const short* normalizedCounter, unsigned maxSymbolValue,
+            const U32* baseValue, const U32* nbAdditionalBits,
+            unsigned tableLog, void* wksp, size_t wkspSize)
+{
+    ZSTD_seqSymbol* const tableDecode = dt+1;
+    U32 const maxSV1 = maxSymbolValue + 1;
+    U32 const tableSize = 1 << tableLog;
+
+    U16* symbolNext = (U16*)wksp;
+    BYTE* spread = (BYTE*)(symbolNext + MaxSeq + 1);
+    U32 highThreshold = tableSize - 1;
+
+
+    /* Sanity Checks */
+    assert(maxSymbolValue <= MaxSeq);
+    assert(tableLog <= MaxFSELog);
+    assert(wkspSize >= ZSTD_BUILD_FSE_TABLE_WKSP_SIZE);
+    (void)wkspSize;
+    /* Init, lay down lowprob symbols */
+    {   ZSTD_seqSymbol_header DTableH;
+        DTableH.tableLog = tableLog;
+        DTableH.fastMode = 1;
+        {   S16 const largeLimit= (S16)(1 << (tableLog-1));
+            U32 s;
+            for (s=0; s<maxSV1; s++) {
+                if (normalizedCounter[s]==-1) {
+                    tableDecode[highThreshold--].baseValue = s;
+                    symbolNext[s] = 1;
+                } else {
+                    if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
+                    assert(normalizedCounter[s]>=0);
+                    symbolNext[s] = (U16)normalizedCounter[s];
+        }   }   }
+        ZSTD_memcpy(dt, &DTableH, sizeof(DTableH));
+    }
+
+    /* Spread symbols */
+    assert(tableSize <= 512);
+    /* Specialized symbol spreading for the case when there are
+     * no low probability (-1 count) symbols. When compressing
+     * small blocks we avoid low probability symbols to hit this
+     * case, since header decoding speed matters more.
+     */
+    if (highThreshold == tableSize - 1) {
+        size_t const tableMask = tableSize-1;
+        size_t const step = FSE_TABLESTEP(tableSize);
+        /* First lay down the symbols in order.
+         * We use a uint64_t to lay down 8 bytes at a time. This reduces branch
+         * misses since small blocks generally have small table logs, so nearly
+         * all symbols have counts <= 8. We ensure we have 8 bytes at the end of
+         * our buffer to handle the over-write.
+         */
+        {
+            U64 const add = 0x0101010101010101ull;
+            size_t pos = 0;
+            U64 sv = 0;
+            U32 s;
+            for (s=0; s<maxSV1; ++s, sv += add) {
+                int i;
+                int const n = normalizedCounter[s];
+                MEM_write64(spread + pos, sv);
+                for (i = 8; i < n; i += 8) {
+                    MEM_write64(spread + pos + i, sv);
+                }
+                pos += n;
+            }
+        }
+        /* Now we spread those positions across the table.
+         * The benefit of doing it in two stages is that we avoid the the
+         * variable size inner loop, which caused lots of branch misses.
+         * Now we can run through all the positions without any branch misses.
+         * We unroll the loop twice, since that is what emperically worked best.
+         */
+        {
+            size_t position = 0;
+            size_t s;
+            size_t const unroll = 2;
+            assert(tableSize % unroll == 0); /* FSE_MIN_TABLELOG is 5 */
+            for (s = 0; s < (size_t)tableSize; s += unroll) {
+                size_t u;
+                for (u = 0; u < unroll; ++u) {
+                    size_t const uPosition = (position + (u * step)) & tableMask;
+                    tableDecode[uPosition].baseValue = spread[s + u];
+                }
+                position = (position + (unroll * step)) & tableMask;
+            }
+            assert(position == 0);
+        }
+    } else {
+        U32 const tableMask = tableSize-1;
+        U32 const step = FSE_TABLESTEP(tableSize);
+        U32 s, position = 0;
+        for (s=0; s<maxSV1; s++) {
+            int i;
+            int const n = normalizedCounter[s];
+            for (i=0; i<n; i++) {
+                tableDecode[position].baseValue = s;
+                position = (position + step) & tableMask;
+                while (position > highThreshold) position = (position + step) & tableMask;   /* lowprob area */
+        }   }
+        assert(position == 0); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
+    }
+
+    /* Build Decoding table */
+    {
+        U32 u;
+        for (u=0; u<tableSize; u++) {
+            U32 const symbol = tableDecode[u].baseValue;
+            U32 const nextState = symbolNext[symbol]++;
+            tableDecode[u].nbBits = (BYTE) (tableLog - BIT_highbit32(nextState) );
+            tableDecode[u].nextState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
+            assert(nbAdditionalBits[symbol] < 255);
+            tableDecode[u].nbAdditionalBits = (BYTE)nbAdditionalBits[symbol];
+            tableDecode[u].baseValue = baseValue[symbol];
+        }
+    }
+}
+
+/* Avoids the FORCE_INLINE of the _body() function. */
+static void ZSTD_buildFSETable_body_default(ZSTD_seqSymbol* dt,
+            const short* normalizedCounter, unsigned maxSymbolValue,
+            const U32* baseValue, const U32* nbAdditionalBits,
+            unsigned tableLog, void* wksp, size_t wkspSize)
+{
+    ZSTD_buildFSETable_body(dt, normalizedCounter, maxSymbolValue,
+            baseValue, nbAdditionalBits, tableLog, wksp, wkspSize);
+}
+
+#if DYNAMIC_BMI2
+TARGET_ATTRIBUTE("bmi2") static void ZSTD_buildFSETable_body_bmi2(ZSTD_seqSymbol* dt,
+            const short* normalizedCounter, unsigned maxSymbolValue,
+            const U32* baseValue, const U32* nbAdditionalBits,
+            unsigned tableLog, void* wksp, size_t wkspSize)
+{
+    ZSTD_buildFSETable_body(dt, normalizedCounter, maxSymbolValue,
+            baseValue, nbAdditionalBits, tableLog, wksp, wkspSize);
+}
+#endif
+
+void ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
+            const short* normalizedCounter, unsigned maxSymbolValue,
+            const U32* baseValue, const U32* nbAdditionalBits,
+            unsigned tableLog, void* wksp, size_t wkspSize, int bmi2)
+{
+#if DYNAMIC_BMI2
+    if (bmi2) {
+        ZSTD_buildFSETable_body_bmi2(dt, normalizedCounter, maxSymbolValue,
+                baseValue, nbAdditionalBits, tableLog, wksp, wkspSize);
+        return;
+    }
+#endif
+    (void)bmi2;
+    ZSTD_buildFSETable_body_default(dt, normalizedCounter, maxSymbolValue,
+            baseValue, nbAdditionalBits, tableLog, wksp, wkspSize);
+}
+
+
+/*! ZSTD_buildSeqTable() :
+ * @return : nb bytes read from src,
+ *           or an error code if it fails */
+static size_t ZSTD_buildSeqTable(ZSTD_seqSymbol* DTableSpace, const ZSTD_seqSymbol** DTablePtr,
+                                 symbolEncodingType_e type, unsigned max, U32 maxLog,
+                                 const void* src, size_t srcSize,
+                                 const U32* baseValue, const U32* nbAdditionalBits,
+                                 const ZSTD_seqSymbol* defaultTable, U32 flagRepeatTable,
+                                 int ddictIsCold, int nbSeq, U32* wksp, size_t wkspSize,
+                                 int bmi2)
+{
+    switch(type)
+    {
+    case set_rle :
+        RETURN_ERROR_IF(!srcSize, srcSize_wrong, "");
+        RETURN_ERROR_IF((*(const BYTE*)src) > max, corruption_detected, "");
+        {   U32 const symbol = *(const BYTE*)src;
+            U32 const baseline = baseValue[symbol];
+            U32 const nbBits = nbAdditionalBits[symbol];
+            ZSTD_buildSeqTable_rle(DTableSpace, baseline, nbBits);
+        }
+        *DTablePtr = DTableSpace;
+        return 1;
+    case set_basic :
+        *DTablePtr = defaultTable;
+        return 0;
+    case set_repeat:
+        RETURN_ERROR_IF(!flagRepeatTable, corruption_detected, "");
+        /* prefetch FSE table if used */
+        if (ddictIsCold && (nbSeq > 24 /* heuristic */)) {
+            const void* const pStart = *DTablePtr;
+            size_t const pSize = sizeof(ZSTD_seqSymbol) * (SEQSYMBOL_TABLE_SIZE(maxLog));
+            PREFETCH_AREA(pStart, pSize);
+        }
+        return 0;
+    case set_compressed :
+        {   unsigned tableLog;
+            S16 norm[MaxSeq+1];
+            size_t const headerSize = FSE_readNCount(norm, &max, &tableLog, src, srcSize);
+            RETURN_ERROR_IF(FSE_isError(headerSize), corruption_detected, "");
+            RETURN_ERROR_IF(tableLog > maxLog, corruption_detected, "");
+            ZSTD_buildFSETable(DTableSpace, norm, max, baseValue, nbAdditionalBits, tableLog, wksp, wkspSize, bmi2);
+            *DTablePtr = DTableSpace;
+            return headerSize;
+        }
+    default :
+        assert(0);
+        RETURN_ERROR(GENERIC, "impossible");
+    }
+}
+
+size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
+                             const void* src, size_t srcSize)
+{
+    const BYTE* const istart = (const BYTE*)src;
+    const BYTE* const iend = istart + srcSize;
+    const BYTE* ip = istart;
+    int nbSeq;
+    DEBUGLOG(5, "ZSTD_decodeSeqHeaders");
+
+    /* check */
+    RETURN_ERROR_IF(srcSize < MIN_SEQUENCES_SIZE, srcSize_wrong, "");
+
+    /* SeqHead */
+    nbSeq = *ip++;
+    if (!nbSeq) {
+        *nbSeqPtr=0;
+        RETURN_ERROR_IF(srcSize != 1, srcSize_wrong, "");
+        return 1;
+    }
+    if (nbSeq > 0x7F) {
+        if (nbSeq == 0xFF) {
+            RETURN_ERROR_IF(ip+2 > iend, srcSize_wrong, "");
+            nbSeq = MEM_readLE16(ip) + LONGNBSEQ;
+            ip+=2;
+        } else {
+            RETURN_ERROR_IF(ip >= iend, srcSize_wrong, "");
+            nbSeq = ((nbSeq-0x80)<<8) + *ip++;
+        }
+    }
+    *nbSeqPtr = nbSeq;
+
+    /* FSE table descriptors */
+    RETURN_ERROR_IF(ip+1 > iend, srcSize_wrong, ""); /* minimum possible size: 1 byte for symbol encoding types */
+    {   symbolEncodingType_e const LLtype = (symbolEncodingType_e)(*ip >> 6);
+        symbolEncodingType_e const OFtype = (symbolEncodingType_e)((*ip >> 4) & 3);
+        symbolEncodingType_e const MLtype = (symbolEncodingType_e)((*ip >> 2) & 3);
+        ip++;
+
+        /* Build DTables */
+        {   size_t const llhSize = ZSTD_buildSeqTable(dctx->entropy.LLTable, &dctx->LLTptr,
+                                                      LLtype, MaxLL, LLFSELog,
+                                                      ip, iend-ip,
+                                                      LL_base, LL_bits,
+                                                      LL_defaultDTable, dctx->fseEntropy,
+                                                      dctx->ddictIsCold, nbSeq,
+                                                      dctx->workspace, sizeof(dctx->workspace),
+                                                      dctx->bmi2);
+            RETURN_ERROR_IF(ZSTD_isError(llhSize), corruption_detected, "ZSTD_buildSeqTable failed");
+            ip += llhSize;
+        }
+
+        {   size_t const ofhSize = ZSTD_buildSeqTable(dctx->entropy.OFTable, &dctx->OFTptr,
+                                                      OFtype, MaxOff, OffFSELog,
+                                                      ip, iend-ip,
+                                                      OF_base, OF_bits,
+                                                      OF_defaultDTable, dctx->fseEntropy,
+                                                      dctx->ddictIsCold, nbSeq,
+                                                      dctx->workspace, sizeof(dctx->workspace),
+                                                      dctx->bmi2);
+            RETURN_ERROR_IF(ZSTD_isError(ofhSize), corruption_detected, "ZSTD_buildSeqTable failed");
+            ip += ofhSize;
+        }
+
+        {   size_t const mlhSize = ZSTD_buildSeqTable(dctx->entropy.MLTable, &dctx->MLTptr,
+                                                      MLtype, MaxML, MLFSELog,
+                                                      ip, iend-ip,
+                                                      ML_base, ML_bits,
+                                                      ML_defaultDTable, dctx->fseEntropy,
+                                                      dctx->ddictIsCold, nbSeq,
+                                                      dctx->workspace, sizeof(dctx->workspace),
+                                                      dctx->bmi2);
+            RETURN_ERROR_IF(ZSTD_isError(mlhSize), corruption_detected, "ZSTD_buildSeqTable failed");
+            ip += mlhSize;
+        }
+    }
+
+    return ip-istart;
+}
+
+
+typedef struct {
+    size_t litLength;
+    size_t matchLength;
+    size_t offset;
+    const BYTE* match;
+} seq_t;
+
+typedef struct {
+    size_t state;
+    const ZSTD_seqSymbol* table;
+} ZSTD_fseState;
+
+typedef struct {
+    BIT_DStream_t DStream;
+    ZSTD_fseState stateLL;
+    ZSTD_fseState stateOffb;
+    ZSTD_fseState stateML;
+    size_t prevOffset[ZSTD_REP_NUM];
+    const BYTE* prefixStart;
+    const BYTE* dictEnd;
+    size_t pos;
+} seqState_t;
+
+/*! ZSTD_overlapCopy8() :
+ *  Copies 8 bytes from ip to op and updates op and ip where ip <= op.
+ *  If the offset is < 8 then the offset is spread to at least 8 bytes.
+ *
+ *  Precondition: *ip <= *op
+ *  Postcondition: *op - *op >= 8
+ */
+HINT_INLINE void ZSTD_overlapCopy8(BYTE** op, BYTE const** ip, size_t offset) {
+    assert(*ip <= *op);
+    if (offset < 8) {
+        /* close range match, overlap */
+        static const U32 dec32table[] = { 0, 1, 2, 1, 4, 4, 4, 4 };   /* added */
+        static const int dec64table[] = { 8, 8, 8, 7, 8, 9,10,11 };   /* subtracted */
+        int const sub2 = dec64table[offset];
+        (*op)[0] = (*ip)[0];
+        (*op)[1] = (*ip)[1];
+        (*op)[2] = (*ip)[2];
+        (*op)[3] = (*ip)[3];
+        *ip += dec32table[offset];
+        ZSTD_copy4(*op+4, *ip);
+        *ip -= sub2;
+    } else {
+        ZSTD_copy8(*op, *ip);
+    }
+    *ip += 8;
+    *op += 8;
+    assert(*op - *ip >= 8);
+}
+
+/*! ZSTD_safecopy() :
+ *  Specialized version of memcpy() that is allowed to READ up to WILDCOPY_OVERLENGTH past the input buffer
+ *  and write up to 16 bytes past oend_w (op >= oend_w is allowed).
+ *  This function is only called in the uncommon case where the sequence is near the end of the block. It
+ *  should be fast for a single long sequence, but can be slow for several short sequences.
+ *
+ *  @param ovtype controls the overlap detection
+ *         - ZSTD_no_overlap: The source and destination are guaranteed to be at least WILDCOPY_VECLEN bytes apart.
+ *         - ZSTD_overlap_src_before_dst: The src and dst may overlap and may be any distance apart.
+ *           The src buffer must be before the dst buffer.
+ */
+static void ZSTD_safecopy(BYTE* op, BYTE* const oend_w, BYTE const* ip, ptrdiff_t length, ZSTD_overlap_e ovtype) {
+    ptrdiff_t const diff = op - ip;
+    BYTE* const oend = op + length;
+
+    assert((ovtype == ZSTD_no_overlap && (diff <= -8 || diff >= 8 || op >= oend_w)) ||
+           (ovtype == ZSTD_overlap_src_before_dst && diff >= 0));
+
+    if (length < 8) {
+        /* Handle short lengths. */
+        while (op < oend) *op++ = *ip++;
+        return;
+    }
+    if (ovtype == ZSTD_overlap_src_before_dst) {
+        /* Copy 8 bytes and ensure the offset >= 8 when there can be overlap. */
+        assert(length >= 8);
+        ZSTD_overlapCopy8(&op, &ip, diff);
+        assert(op - ip >= 8);
+        assert(op <= oend);
+    }
+
+    if (oend <= oend_w) {
+        /* No risk of overwrite. */
+        ZSTD_wildcopy(op, ip, length, ovtype);
+        return;
+    }
+    if (op <= oend_w) {
+        /* Wildcopy until we get close to the end. */
+        assert(oend > oend_w);
+        ZSTD_wildcopy(op, ip, oend_w - op, ovtype);
+        ip += oend_w - op;
+        op = oend_w;
+    }
+    /* Handle the leftovers. */
+    while (op < oend) *op++ = *ip++;
+}
+
+/* ZSTD_execSequenceEnd():
+ * This version handles cases that are near the end of the output buffer. It requires
+ * more careful checks to make sure there is no overflow. By separating out these hard
+ * and unlikely cases, we can speed up the common cases.
+ *
+ * NOTE: This function needs to be fast for a single long sequence, but doesn't need
+ * to be optimized for many small sequences, since those fall into ZSTD_execSequence().
+ */
+FORCE_NOINLINE
+size_t ZSTD_execSequenceEnd(BYTE* op,
+                            BYTE* const oend, seq_t sequence,
+                            const BYTE** litPtr, const BYTE* const litLimit,
+                            const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd)
+{
+    BYTE* const oLitEnd = op + sequence.litLength;
+    size_t const sequenceLength = sequence.litLength + sequence.matchLength;
+    const BYTE* const iLitEnd = *litPtr + sequence.litLength;
+    const BYTE* match = oLitEnd - sequence.offset;
+    BYTE* const oend_w = oend - WILDCOPY_OVERLENGTH;
+
+    /* bounds checks : careful of address space overflow in 32-bit mode */
+    RETURN_ERROR_IF(sequenceLength > (size_t)(oend - op), dstSize_tooSmall, "last match must fit within dstBuffer");
+    RETURN_ERROR_IF(sequence.litLength > (size_t)(litLimit - *litPtr), corruption_detected, "try to read beyond literal buffer");
+    assert(op < op + sequenceLength);
+    assert(oLitEnd < op + sequenceLength);
+
+    /* copy literals */
+    ZSTD_safecopy(op, oend_w, *litPtr, sequence.litLength, ZSTD_no_overlap);
+    op = oLitEnd;
+    *litPtr = iLitEnd;
+
+    /* copy Match */
+    if (sequence.offset > (size_t)(oLitEnd - prefixStart)) {
+        /* offset beyond prefix */
+        RETURN_ERROR_IF(sequence.offset > (size_t)(oLitEnd - virtualStart), corruption_detected, "");
+        match = dictEnd - (prefixStart-match);
+        if (match + sequence.matchLength <= dictEnd) {
+            ZSTD_memmove(oLitEnd, match, sequence.matchLength);
+            return sequenceLength;
+        }
+        /* span extDict & currentPrefixSegment */
+        {   size_t const length1 = dictEnd - match;
+            ZSTD_memmove(oLitEnd, match, length1);
+            op = oLitEnd + length1;
+            sequence.matchLength -= length1;
+            match = prefixStart;
+    }   }
+    ZSTD_safecopy(op, oend_w, match, sequence.matchLength, ZSTD_overlap_src_before_dst);
+    return sequenceLength;
+}
+
+HINT_INLINE
+size_t ZSTD_execSequence(BYTE* op,
+                         BYTE* const oend, seq_t sequence,
+                         const BYTE** litPtr, const BYTE* const litLimit,
+                         const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd)
+{
+    BYTE* const oLitEnd = op + sequence.litLength;
+    size_t const sequenceLength = sequence.litLength + sequence.matchLength;
+    BYTE* const oMatchEnd = op + sequenceLength;   /* risk : address space overflow (32-bits) */
+    BYTE* const oend_w = oend - WILDCOPY_OVERLENGTH;   /* risk : address space underflow on oend=NULL */
+    const BYTE* const iLitEnd = *litPtr + sequence.litLength;
+    const BYTE* match = oLitEnd - sequence.offset;
+
+    assert(op != NULL /* Precondition */);
+    assert(oend_w < oend /* No underflow */);
+    /* Handle edge cases in a slow path:
+     *   - Read beyond end of literals
+     *   - Match end is within WILDCOPY_OVERLIMIT of oend
+     *   - 32-bit mode and the match length overflows
+     */
+    if (UNLIKELY(
+            iLitEnd > litLimit ||
+            oMatchEnd > oend_w ||
+            (MEM_32bits() && (size_t)(oend - op) < sequenceLength + WILDCOPY_OVERLENGTH)))
+        return ZSTD_execSequenceEnd(op, oend, sequence, litPtr, litLimit, prefixStart, virtualStart, dictEnd);
+
+    /* Assumptions (everything else goes into ZSTD_execSequenceEnd()) */
+    assert(op <= oLitEnd /* No overflow */);
+    assert(oLitEnd < oMatchEnd /* Non-zero match & no overflow */);
+    assert(oMatchEnd <= oend /* No underflow */);
+    assert(iLitEnd <= litLimit /* Literal length is in bounds */);
+    assert(oLitEnd <= oend_w /* Can wildcopy literals */);
+    assert(oMatchEnd <= oend_w /* Can wildcopy matches */);
+
+    /* Copy Literals:
+     * Split out litLength <= 16 since it is nearly always true. +1.6% on gcc-9.
+     * We likely don't need the full 32-byte wildcopy.
+     */
+    assert(WILDCOPY_OVERLENGTH >= 16);
+    ZSTD_copy16(op, (*litPtr));
+    if (UNLIKELY(sequence.litLength > 16)) {
+        ZSTD_wildcopy(op+16, (*litPtr)+16, sequence.litLength-16, ZSTD_no_overlap);
+    }
+    op = oLitEnd;
+    *litPtr = iLitEnd;   /* update for next sequence */
+
+    /* Copy Match */
+    if (sequence.offset > (size_t)(oLitEnd - prefixStart)) {
+        /* offset beyond prefix -> go into extDict */
+        RETURN_ERROR_IF(UNLIKELY(sequence.offset > (size_t)(oLitEnd - virtualStart)), corruption_detected, "");
+        match = dictEnd + (match - prefixStart);
+        if (match + sequence.matchLength <= dictEnd) {
+            ZSTD_memmove(oLitEnd, match, sequence.matchLength);
+            return sequenceLength;
+        }
+        /* span extDict & currentPrefixSegment */
+        {   size_t const length1 = dictEnd - match;
+            ZSTD_memmove(oLitEnd, match, length1);
+            op = oLitEnd + length1;
+            sequence.matchLength -= length1;
+            match = prefixStart;
+    }   }
+    /* Match within prefix of 1 or more bytes */
+    assert(op <= oMatchEnd);
+    assert(oMatchEnd <= oend_w);
+    assert(match >= prefixStart);
+    assert(sequence.matchLength >= 1);
+
+    /* Nearly all offsets are >= WILDCOPY_VECLEN bytes, which means we can use wildcopy
+     * without overlap checking.
+     */
+    if (LIKELY(sequence.offset >= WILDCOPY_VECLEN)) {
+        /* We bet on a full wildcopy for matches, since we expect matches to be
+         * longer than literals (in general). In silesia, ~10% of matches are longer
+         * than 16 bytes.
+         */
+        ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength, ZSTD_no_overlap);
+        return sequenceLength;
+    }
+    assert(sequence.offset < WILDCOPY_VECLEN);
+
+    /* Copy 8 bytes and spread the offset to be >= 8. */
+    ZSTD_overlapCopy8(&op, &match, sequence.offset);
+
+    /* If the match length is > 8 bytes, then continue with the wildcopy. */
+    if (sequence.matchLength > 8) {
+        assert(op < oMatchEnd);
+        ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength-8, ZSTD_overlap_src_before_dst);
+    }
+    return sequenceLength;
+}
+
+static void
+ZSTD_initFseState(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, const ZSTD_seqSymbol* dt)
+{
+    const void* ptr = dt;
+    const ZSTD_seqSymbol_header* const DTableH = (const ZSTD_seqSymbol_header*)ptr;
+    DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
+    DEBUGLOG(6, "ZSTD_initFseState : val=%u using %u bits",
+                (U32)DStatePtr->state, DTableH->tableLog);
+    BIT_reloadDStream(bitD);
+    DStatePtr->table = dt + 1;
+}
+
+FORCE_INLINE_TEMPLATE void
+ZSTD_updateFseState(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD)
+{
+    ZSTD_seqSymbol const DInfo = DStatePtr->table[DStatePtr->state];
+    U32 const nbBits = DInfo.nbBits;
+    size_t const lowBits = BIT_readBits(bitD, nbBits);
+    DStatePtr->state = DInfo.nextState + lowBits;
+}
+
+FORCE_INLINE_TEMPLATE void
+ZSTD_updateFseStateWithDInfo(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, ZSTD_seqSymbol const DInfo)
+{
+    U32 const nbBits = DInfo.nbBits;
+    size_t const lowBits = BIT_readBits(bitD, nbBits);
+    DStatePtr->state = DInfo.nextState + lowBits;
+}
+
+/* We need to add at most (ZSTD_WINDOWLOG_MAX_32 - 1) bits to read the maximum
+ * offset bits. But we can only read at most (STREAM_ACCUMULATOR_MIN_32 - 1)
+ * bits before reloading. This value is the maximum number of bytes we read
+ * after reloading when we are decoding long offsets.
+ */
+#define LONG_OFFSETS_MAX_EXTRA_BITS_32                       \
+    (ZSTD_WINDOWLOG_MAX_32 > STREAM_ACCUMULATOR_MIN_32       \
+        ? ZSTD_WINDOWLOG_MAX_32 - STREAM_ACCUMULATOR_MIN_32  \
+        : 0)
+
+typedef enum { ZSTD_lo_isRegularOffset, ZSTD_lo_isLongOffset=1 } ZSTD_longOffset_e;
+typedef enum { ZSTD_p_noPrefetch=0, ZSTD_p_prefetch=1 } ZSTD_prefetch_e;
+
+FORCE_INLINE_TEMPLATE seq_t
+ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets, const ZSTD_prefetch_e prefetch)
+{
+    seq_t seq;
+    ZSTD_seqSymbol const llDInfo = seqState->stateLL.table[seqState->stateLL.state];
+    ZSTD_seqSymbol const mlDInfo = seqState->stateML.table[seqState->stateML.state];
+    ZSTD_seqSymbol const ofDInfo = seqState->stateOffb.table[seqState->stateOffb.state];
+    U32 const llBase = llDInfo.baseValue;
+    U32 const mlBase = mlDInfo.baseValue;
+    U32 const ofBase = ofDInfo.baseValue;
+    BYTE const llBits = llDInfo.nbAdditionalBits;
+    BYTE const mlBits = mlDInfo.nbAdditionalBits;
+    BYTE const ofBits = ofDInfo.nbAdditionalBits;
+    BYTE const totalBits = llBits+mlBits+ofBits;
+
+    /* sequence */
+    {   size_t offset;
+        if (ofBits > 1) {
+            ZSTD_STATIC_ASSERT(ZSTD_lo_isLongOffset == 1);
+            ZSTD_STATIC_ASSERT(LONG_OFFSETS_MAX_EXTRA_BITS_32 == 5);
+            assert(ofBits <= MaxOff);
+            if (MEM_32bits() && longOffsets && (ofBits >= STREAM_ACCUMULATOR_MIN_32)) {
+                U32 const extraBits = ofBits - MIN(ofBits, 32 - seqState->DStream.bitsConsumed);
+                offset = ofBase + (BIT_readBitsFast(&seqState->DStream, ofBits - extraBits) << extraBits);
+                BIT_reloadDStream(&seqState->DStream);
+                if (extraBits) offset += BIT_readBitsFast(&seqState->DStream, extraBits);
+                assert(extraBits <= LONG_OFFSETS_MAX_EXTRA_BITS_32);   /* to avoid another reload */
+            } else {
+                offset = ofBase + BIT_readBitsFast(&seqState->DStream, ofBits/*>0*/);   /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
+                if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);
+            }
+            seqState->prevOffset[2] = seqState->prevOffset[1];
+            seqState->prevOffset[1] = seqState->prevOffset[0];
+            seqState->prevOffset[0] = offset;
+        } else {
+            U32 const ll0 = (llBase == 0);
+            if (LIKELY((ofBits == 0))) {
+                if (LIKELY(!ll0))
+                    offset = seqState->prevOffset[0];
+                else {
+                    offset = seqState->prevOffset[1];
+                    seqState->prevOffset[1] = seqState->prevOffset[0];
+                    seqState->prevOffset[0] = offset;
+                }
+            } else {
+                offset = ofBase + ll0 + BIT_readBitsFast(&seqState->DStream, 1);
+                {   size_t temp = (offset==3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
+                    temp += !temp;   /* 0 is not valid; input is corrupted; force offset to 1 */
+                    if (offset != 1) seqState->prevOffset[2] = seqState->prevOffset[1];
+                    seqState->prevOffset[1] = seqState->prevOffset[0];
+                    seqState->prevOffset[0] = offset = temp;
+        }   }   }
+        seq.offset = offset;
+    }
+
+    seq.matchLength = mlBase;
+    if (mlBits > 0)
+        seq.matchLength += BIT_readBitsFast(&seqState->DStream, mlBits/*>0*/);
+
+    if (MEM_32bits() && (mlBits+llBits >= STREAM_ACCUMULATOR_MIN_32-LONG_OFFSETS_MAX_EXTRA_BITS_32))
+        BIT_reloadDStream(&seqState->DStream);
+    if (MEM_64bits() && UNLIKELY(totalBits >= STREAM_ACCUMULATOR_MIN_64-(LLFSELog+MLFSELog+OffFSELog)))
+        BIT_reloadDStream(&seqState->DStream);
+    /* Ensure there are enough bits to read the rest of data in 64-bit mode. */
+    ZSTD_STATIC_ASSERT(16+LLFSELog+MLFSELog+OffFSELog < STREAM_ACCUMULATOR_MIN_64);
+
+    seq.litLength = llBase;
+    if (llBits > 0)
+        seq.litLength += BIT_readBitsFast(&seqState->DStream, llBits/*>0*/);
+
+    if (MEM_32bits())
+        BIT_reloadDStream(&seqState->DStream);
+
+    DEBUGLOG(6, "seq: litL=%u, matchL=%u, offset=%u",
+                (U32)seq.litLength, (U32)seq.matchLength, (U32)seq.offset);
+
+    if (prefetch == ZSTD_p_prefetch) {
+        size_t const pos = seqState->pos + seq.litLength;
+        const BYTE* const matchBase = (seq.offset > pos) ? seqState->dictEnd : seqState->prefixStart;
+        seq.match = matchBase + pos - seq.offset;  /* note : this operation can overflow when seq.offset is really too large, which can only happen when input is corrupted.
+                                                    * No consequence though : no memory access will occur, offset is only used for prefetching */
+        seqState->pos = pos + seq.matchLength;
+    }
+
+    /* ANS state update
+     * gcc-9.0.0 does 2.5% worse with ZSTD_updateFseStateWithDInfo().
+     * clang-9.2.0 does 7% worse with ZSTD_updateFseState().
+     * Naturally it seems like ZSTD_updateFseStateWithDInfo() should be the
+     * better option, so it is the default for other compilers. But, if you
+     * measure that it is worse, please put up a pull request.
+     */
+    {
+#if !defined(__clang__)
+        const int kUseUpdateFseState = 1;
+#else
+        const int kUseUpdateFseState = 0;
+#endif
+        if (kUseUpdateFseState) {
+            ZSTD_updateFseState(&seqState->stateLL, &seqState->DStream);    /* <=  9 bits */
+            ZSTD_updateFseState(&seqState->stateML, &seqState->DStream);    /* <=  9 bits */
+            if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);    /* <= 18 bits */
+            ZSTD_updateFseState(&seqState->stateOffb, &seqState->DStream);  /* <=  8 bits */
+        } else {
+            ZSTD_updateFseStateWithDInfo(&seqState->stateLL, &seqState->DStream, llDInfo);    /* <=  9 bits */
+            ZSTD_updateFseStateWithDInfo(&seqState->stateML, &seqState->DStream, mlDInfo);    /* <=  9 bits */
+            if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);    /* <= 18 bits */
+            ZSTD_updateFseStateWithDInfo(&seqState->stateOffb, &seqState->DStream, ofDInfo);  /* <=  8 bits */
+        }
+    }
+
+    return seq;
+}
+
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+MEM_STATIC int ZSTD_dictionaryIsActive(ZSTD_DCtx const* dctx, BYTE const* prefixStart, BYTE const* oLitEnd)
+{
+    size_t const windowSize = dctx->fParams.windowSize;
+    /* No dictionary used. */
+    if (dctx->dictContentEndForFuzzing == NULL) return 0;
+    /* Dictionary is our prefix. */
+    if (prefixStart == dctx->dictContentBeginForFuzzing) return 1;
+    /* Dictionary is not our ext-dict. */
+    if (dctx->dictEnd != dctx->dictContentEndForFuzzing) return 0;
+    /* Dictionary is not within our window size. */
+    if ((size_t)(oLitEnd - prefixStart) >= windowSize) return 0;
+    /* Dictionary is active. */
+    return 1;
+}
+
+MEM_STATIC void ZSTD_assertValidSequence(
+        ZSTD_DCtx const* dctx,
+        BYTE const* op, BYTE const* oend,
+        seq_t const seq,
+        BYTE const* prefixStart, BYTE const* virtualStart)
+{
+#if DEBUGLEVEL >= 1
+    size_t const windowSize = dctx->fParams.windowSize;
+    size_t const sequenceSize = seq.litLength + seq.matchLength;
+    BYTE const* const oLitEnd = op + seq.litLength;
+    DEBUGLOG(6, "Checking sequence: litL=%u matchL=%u offset=%u",
+            (U32)seq.litLength, (U32)seq.matchLength, (U32)seq.offset);
+    assert(op <= oend);
+    assert((size_t)(oend - op) >= sequenceSize);
+    assert(sequenceSize <= ZSTD_BLOCKSIZE_MAX);
+    if (ZSTD_dictionaryIsActive(dctx, prefixStart, oLitEnd)) {
+        size_t const dictSize = (size_t)((char const*)dctx->dictContentEndForFuzzing - (char const*)dctx->dictContentBeginForFuzzing);
+        /* Offset must be within the dictionary. */
+        assert(seq.offset <= (size_t)(oLitEnd - virtualStart));
+        assert(seq.offset <= windowSize + dictSize);
+    } else {
+        /* Offset must be within our window. */
+        assert(seq.offset <= windowSize);
+    }
+#else
+    (void)dctx, (void)op, (void)oend, (void)seq, (void)prefixStart, (void)virtualStart;
+#endif
+}
+#endif
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
+FORCE_INLINE_TEMPLATE size_t
+DONT_VECTORIZE
+ZSTD_decompressSequences_body( ZSTD_DCtx* dctx,
+                               void* dst, size_t maxDstSize,
+                         const void* seqStart, size_t seqSize, int nbSeq,
+                         const ZSTD_longOffset_e isLongOffset,
+                         const int frame)
+{
+    const BYTE* ip = (const BYTE*)seqStart;
+    const BYTE* const iend = ip + seqSize;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + maxDstSize;
+    BYTE* op = ostart;
+    const BYTE* litPtr = dctx->litPtr;
+    const BYTE* const litEnd = litPtr + dctx->litSize;
+    const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart);
+    const BYTE* const vBase = (const BYTE*) (dctx->virtualStart);
+    const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
+    DEBUGLOG(5, "ZSTD_decompressSequences_body");
+    (void)frame;
+
+    /* Regen sequences */
+    if (nbSeq) {
+        seqState_t seqState;
+        size_t error = 0;
+        dctx->fseEntropy = 1;
+        { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) seqState.prevOffset[i] = dctx->entropy.rep[i]; }
+        RETURN_ERROR_IF(
+            ERR_isError(BIT_initDStream(&seqState.DStream, ip, iend-ip)),
+            corruption_detected, "");
+        ZSTD_initFseState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
+        ZSTD_initFseState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
+        ZSTD_initFseState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
+        assert(dst != NULL);
+
+        ZSTD_STATIC_ASSERT(
+                BIT_DStream_unfinished < BIT_DStream_completed &&
+                BIT_DStream_endOfBuffer < BIT_DStream_completed &&
+                BIT_DStream_completed < BIT_DStream_overflow);
+
+#if defined(__x86_64__)
+        /* Align the decompression loop to 32 + 16 bytes.
+         *
+         * zstd compiled with gcc-9 on an Intel i9-9900k shows 10% decompression
+         * speed swings based on the alignment of the decompression loop. This
+         * performance swing is caused by parts of the decompression loop falling
+         * out of the DSB. The entire decompression loop should fit in the DSB,
+         * when it can't we get much worse performance. You can measure if you've
+         * hit the good case or the bad case with this perf command for some
+         * compressed file test.zst:
+         *
+         *   perf stat -e cycles -e instructions -e idq.all_dsb_cycles_any_uops \
+         *             -e idq.all_mite_cycles_any_uops -- ./zstd -tq test.zst
+         *
+         * If you see most cycles served out of the MITE you've hit the bad case.
+         * If you see most cycles served out of the DSB you've hit the good case.
+         * If it is pretty even then you may be in an okay case.
+         *
+         * I've been able to reproduce this issue on the following CPUs:
+         *   - Kabylake: Macbook Pro (15-inch, 2019) 2.4 GHz Intel Core i9
+         *               Use Instruments->Counters to get DSB/MITE cycles.
+         *               I never got performance swings, but I was able to
+         *               go from the good case of mostly DSB to half of the
+         *               cycles served from MITE.
+         *   - Coffeelake: Intel i9-9900k
+         *
+         * I haven't been able to reproduce the instability or DSB misses on any
+         * of the following CPUS:
+         *   - Haswell
+         *   - Broadwell: Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GH
+         *   - Skylake
+         *
+         * If you are seeing performance stability this script can help test.
+         * It tests on 4 commits in zstd where I saw performance change.
+         *
+         *   https://gist.github.com/terrelln/9889fc06a423fd5ca6e99351564473f4
+         */
+        __asm__(".p2align 5");
+        __asm__("nop");
+        __asm__(".p2align 4");
+#endif
+        for ( ; ; ) {
+            seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_noPrefetch);
+            size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, prefixStart, vBase, dictEnd);
+#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
+            assert(!ZSTD_isError(oneSeqSize));
+            if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequence, prefixStart, vBase);
+#endif
+            DEBUGLOG(6, "regenerated sequence size : %u", (U32)oneSeqSize);
+            BIT_reloadDStream(&(seqState.DStream));
+            op += oneSeqSize;
+            /* gcc and clang both don't like early returns in this loop.
+             * Instead break and check for an error at the end of the loop.
+             */
+            if (UNLIKELY(ZSTD_isError(oneSeqSize))) {
+                error = oneSeqSize;
+                break;
+            }
+            if (UNLIKELY(!--nbSeq)) break;
+        }
+
+        /* check if reached exact end */
+        DEBUGLOG(5, "ZSTD_decompressSequences_body: after decode loop, remaining nbSeq : %i", nbSeq);
+        if (ZSTD_isError(error)) return error;
+        RETURN_ERROR_IF(nbSeq, corruption_detected, "");
+        RETURN_ERROR_IF(BIT_reloadDStream(&seqState.DStream) < BIT_DStream_completed, corruption_detected, "");
+        /* save reps for next block */
+        { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); }
+    }
+
+    /* last literal segment */
+    {   size_t const lastLLSize = litEnd - litPtr;
+        RETURN_ERROR_IF(lastLLSize > (size_t)(oend-op), dstSize_tooSmall, "");
+        if (op != NULL) {
+            ZSTD_memcpy(op, litPtr, lastLLSize);
+            op += lastLLSize;
+        }
+    }
+
+    return op-ostart;
+}
+
+static size_t
+ZSTD_decompressSequences_default(ZSTD_DCtx* dctx,
+                                 void* dst, size_t maxDstSize,
+                           const void* seqStart, size_t seqSize, int nbSeq,
+                           const ZSTD_longOffset_e isLongOffset,
+                           const int frame)
+{
+    return ZSTD_decompressSequences_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
+FORCE_INLINE_TEMPLATE size_t
+ZSTD_decompressSequencesLong_body(
+                               ZSTD_DCtx* dctx,
+                               void* dst, size_t maxDstSize,
+                         const void* seqStart, size_t seqSize, int nbSeq,
+                         const ZSTD_longOffset_e isLongOffset,
+                         const int frame)
+{
+    const BYTE* ip = (const BYTE*)seqStart;
+    const BYTE* const iend = ip + seqSize;
+    BYTE* const ostart = (BYTE*)dst;
+    BYTE* const oend = ostart + maxDstSize;
+    BYTE* op = ostart;
+    const BYTE* litPtr = dctx->litPtr;
+    const BYTE* const litEnd = litPtr + dctx->litSize;
+    const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart);
+    const BYTE* const dictStart = (const BYTE*) (dctx->virtualStart);
+    const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
+    (void)frame;
+
+    /* Regen sequences */
+    if (nbSeq) {
+#define STORED_SEQS 4
+#define STORED_SEQS_MASK (STORED_SEQS-1)
+#define ADVANCED_SEQS 4
+        seq_t sequences[STORED_SEQS];
+        int const seqAdvance = MIN(nbSeq, ADVANCED_SEQS);
+        seqState_t seqState;
+        int seqNb;
+        dctx->fseEntropy = 1;
+        { int i; for (i=0; i<ZSTD_REP_NUM; i++) seqState.prevOffset[i] = dctx->entropy.rep[i]; }
+        seqState.prefixStart = prefixStart;
+        seqState.pos = (size_t)(op-prefixStart);
+        seqState.dictEnd = dictEnd;
+        assert(dst != NULL);
+        assert(iend >= ip);
+        RETURN_ERROR_IF(
+            ERR_isError(BIT_initDStream(&seqState.DStream, ip, iend-ip)),
+            corruption_detected, "");
+        ZSTD_initFseState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
+        ZSTD_initFseState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
+        ZSTD_initFseState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
+
+        /* prepare in advance */
+        for (seqNb=0; (BIT_reloadDStream(&seqState.DStream) <= BIT_DStream_completed) && (seqNb<seqAdvance); seqNb++) {
+            sequences[seqNb] = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_prefetch);
+            PREFETCH_L1(sequences[seqNb].match); PREFETCH_L1(sequences[seqNb].match + sequences[seqNb].matchLength - 1); /* note : it's safe to invoke PREFETCH() on any memory address, including invalid ones */
+        }
+        RETURN_ERROR_IF(seqNb<seqAdvance, corruption_detected, "");
+
+        /* decode and decompress */
+        for ( ; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && (seqNb<nbSeq) ; seqNb++) {
+            seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_prefetch);
+            size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequences[(seqNb-ADVANCED_SEQS) & STORED_SEQS_MASK], &litPtr, litEnd, prefixStart, dictStart, dictEnd);
+#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
+            assert(!ZSTD_isError(oneSeqSize));
+            if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequences[(seqNb-ADVANCED_SEQS) & STORED_SEQS_MASK], prefixStart, dictStart);
+#endif
+            if (ZSTD_isError(oneSeqSize)) return oneSeqSize;
+            PREFETCH_L1(sequence.match); PREFETCH_L1(sequence.match + sequence.matchLength - 1); /* note : it's safe to invoke PREFETCH() on any memory address, including invalid ones */
+            sequences[seqNb & STORED_SEQS_MASK] = sequence;
+            op += oneSeqSize;
+        }
+        RETURN_ERROR_IF(seqNb<nbSeq, corruption_detected, "");
+
+        /* finish queue */
+        seqNb -= seqAdvance;
+        for ( ; seqNb<nbSeq ; seqNb++) {
+            size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequences[seqNb&STORED_SEQS_MASK], &litPtr, litEnd, prefixStart, dictStart, dictEnd);
+#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
+            assert(!ZSTD_isError(oneSeqSize));
+            if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequences[seqNb&STORED_SEQS_MASK], prefixStart, dictStart);
+#endif
+            if (ZSTD_isError(oneSeqSize)) return oneSeqSize;
+            op += oneSeqSize;
+        }
+
+        /* save reps for next block */
+        { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); }
+    }
+
+    /* last literal segment */
+    {   size_t const lastLLSize = litEnd - litPtr;
+        RETURN_ERROR_IF(lastLLSize > (size_t)(oend-op), dstSize_tooSmall, "");
+        if (op != NULL) {
+            ZSTD_memcpy(op, litPtr, lastLLSize);
+            op += lastLLSize;
+        }
+    }
+
+    return op-ostart;
+}
+
+static size_t
+ZSTD_decompressSequencesLong_default(ZSTD_DCtx* dctx,
+                                 void* dst, size_t maxDstSize,
+                           const void* seqStart, size_t seqSize, int nbSeq,
+                           const ZSTD_longOffset_e isLongOffset,
+                           const int frame)
+{
+    return ZSTD_decompressSequencesLong_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
+
+
+
+#if DYNAMIC_BMI2
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
+static TARGET_ATTRIBUTE("bmi2") size_t
+DONT_VECTORIZE
+ZSTD_decompressSequences_bmi2(ZSTD_DCtx* dctx,
+                                 void* dst, size_t maxDstSize,
+                           const void* seqStart, size_t seqSize, int nbSeq,
+                           const ZSTD_longOffset_e isLongOffset,
+                           const int frame)
+{
+    return ZSTD_decompressSequences_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
+static TARGET_ATTRIBUTE("bmi2") size_t
+ZSTD_decompressSequencesLong_bmi2(ZSTD_DCtx* dctx,
+                                 void* dst, size_t maxDstSize,
+                           const void* seqStart, size_t seqSize, int nbSeq,
+                           const ZSTD_longOffset_e isLongOffset,
+                           const int frame)
+{
+    return ZSTD_decompressSequencesLong_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
+
+#endif /* DYNAMIC_BMI2 */
+
+typedef size_t (*ZSTD_decompressSequences_t)(
+                            ZSTD_DCtx* dctx,
+                            void* dst, size_t maxDstSize,
+                            const void* seqStart, size_t seqSize, int nbSeq,
+                            const ZSTD_longOffset_e isLongOffset,
+                            const int frame);
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
+static size_t
+ZSTD_decompressSequences(ZSTD_DCtx* dctx, void* dst, size_t maxDstSize,
+                   const void* seqStart, size_t seqSize, int nbSeq,
+                   const ZSTD_longOffset_e isLongOffset,
+                   const int frame)
+{
+    DEBUGLOG(5, "ZSTD_decompressSequences");
+#if DYNAMIC_BMI2
+    if (dctx->bmi2) {
+        return ZSTD_decompressSequences_bmi2(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+    }
+#endif
+  return ZSTD_decompressSequences_default(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
+
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
+/* ZSTD_decompressSequencesLong() :
+ * decompression function triggered when a minimum share of offsets is considered "long",
+ * aka out of cache.
+ * note : "long" definition seems overloaded here, sometimes meaning "wider than bitstream register", and sometimes meaning "farther than memory cache distance".
+ * This function will try to mitigate main memory latency through the use of prefetching */
+static size_t
+ZSTD_decompressSequencesLong(ZSTD_DCtx* dctx,
+                             void* dst, size_t maxDstSize,
+                             const void* seqStart, size_t seqSize, int nbSeq,
+                             const ZSTD_longOffset_e isLongOffset,
+                             const int frame)
+{
+    DEBUGLOG(5, "ZSTD_decompressSequencesLong");
+#if DYNAMIC_BMI2
+    if (dctx->bmi2) {
+        return ZSTD_decompressSequencesLong_bmi2(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+    }
+#endif
+  return ZSTD_decompressSequencesLong_default(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
+}
+#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
+
+
+
+#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+/* ZSTD_getLongOffsetsShare() :
+ * condition : offTable must be valid
+ * @return : "share" of long offsets (arbitrarily defined as > (1<<23))
+ *           compared to maximum possible of (1<<OffFSELog) */
+static unsigned
+ZSTD_getLongOffsetsShare(const ZSTD_seqSymbol* offTable)
+{
+    const void* ptr = offTable;
+    U32 const tableLog = ((const ZSTD_seqSymbol_header*)ptr)[0].tableLog;
+    const ZSTD_seqSymbol* table = offTable + 1;
+    U32 const max = 1 << tableLog;
+    U32 u, total = 0;
+    DEBUGLOG(5, "ZSTD_getLongOffsetsShare: (tableLog=%u)", tableLog);
+
+    assert(max <= (1 << OffFSELog));  /* max not too large */
+    for (u=0; u<max; u++) {
+        if (table[u].nbAdditionalBits > 22) total += 1;
+    }
+
+    assert(tableLog <= OffFSELog);
+    total <<= (OffFSELog - tableLog);  /* scale to OffFSELog */
+
+    return total;
+}
+#endif
+
+size_t
+ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
+                              void* dst, size_t dstCapacity,
+                        const void* src, size_t srcSize, const int frame)
+{   /* blockType == blockCompressed */
+    const BYTE* ip = (const BYTE*)src;
+    /* isLongOffset must be true if there are long offsets.
+     * Offsets are long if they are larger than 2^STREAM_ACCUMULATOR_MIN.
+     * We don't expect that to be the case in 64-bit mode.
+     * In block mode, window size is not known, so we have to be conservative.
+     * (note: but it could be evaluated from current-lowLimit)
+     */
+    ZSTD_longOffset_e const isLongOffset = (ZSTD_longOffset_e)(MEM_32bits() && (!frame || (dctx->fParams.windowSize > (1ULL << STREAM_ACCUMULATOR_MIN))));
+    DEBUGLOG(5, "ZSTD_decompressBlock_internal (size : %u)", (U32)srcSize);
+
+    RETURN_ERROR_IF(srcSize >= ZSTD_BLOCKSIZE_MAX, srcSize_wrong, "");
+
+    /* Decode literals section */
+    {   size_t const litCSize = ZSTD_decodeLiteralsBlock(dctx, src, srcSize);
+        DEBUGLOG(5, "ZSTD_decodeLiteralsBlock : %u", (U32)litCSize);
+        if (ZSTD_isError(litCSize)) return litCSize;
+        ip += litCSize;
+        srcSize -= litCSize;
+    }
+
+    /* Build Decoding Tables */
+    {
+        /* These macros control at build-time which decompressor implementation
+         * we use. If neither is defined, we do some inspection and dispatch at
+         * runtime.
+         */
+#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+        int usePrefetchDecoder = dctx->ddictIsCold;
+#endif
+        int nbSeq;
+        size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, srcSize);
+        if (ZSTD_isError(seqHSize)) return seqHSize;
+        ip += seqHSize;
+        srcSize -= seqHSize;
+
+        RETURN_ERROR_IF(dst == NULL && nbSeq > 0, dstSize_tooSmall, "NULL not handled");
+
+#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+        if ( !usePrefetchDecoder
+          && (!frame || (dctx->fParams.windowSize > (1<<24)))
+          && (nbSeq>ADVANCED_SEQS) ) {  /* could probably use a larger nbSeq limit */
+            U32 const shareLongOffsets = ZSTD_getLongOffsetsShare(dctx->OFTptr);
+            U32 const minShare = MEM_64bits() ? 7 : 20; /* heuristic values, correspond to 2.73% and 7.81% */
+            usePrefetchDecoder = (shareLongOffsets >= minShare);
+        }
+#endif
+
+        dctx->ddictIsCold = 0;
+
+#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
+    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
+        if (usePrefetchDecoder)
+#endif
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
+            return ZSTD_decompressSequencesLong(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
+#endif
+
+#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
+        /* else */
+        return ZSTD_decompressSequences(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
+#endif
+    }
+}
+
+
+void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst, size_t dstSize)
+{
+    if (dst != dctx->previousDstEnd && dstSize > 0) {   /* not contiguous */
+        dctx->dictEnd = dctx->previousDstEnd;
+        dctx->virtualStart = (const char*)dst - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart));
+        dctx->prefixStart = dst;
+        dctx->previousDstEnd = dst;
+    }
+}
+
+
+size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity,
+                      const void* src, size_t srcSize)
+{
+    size_t dSize;
+    ZSTD_checkContinuity(dctx, dst, dstCapacity);
+    dSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 0);
+    dctx->previousDstEnd = (char*)dst + dSize;
+    return dSize;
+}
diff --git a/lib/zstd/decompress/zstd_decompress_block.h b/lib/zstd/decompress/zstd_decompress_block.h
new file mode 100644
index 0000000..e7f5f66
--- /dev/null
+++ b/lib/zstd/decompress/zstd_decompress_block.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+#ifndef ZSTD_DEC_BLOCK_H
+#define ZSTD_DEC_BLOCK_H
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include "../common/zstd_deps.h"   /* size_t */
+#include <linux/zstd.h>    /* DCtx, and some public functions */
+#include "../common/zstd_internal.h"  /* blockProperties_t, and some public functions */
+#include "zstd_decompress_internal.h"  /* ZSTD_seqSymbol */
+
+
+/* ===   Prototypes   === */
+
+/* note: prototypes already published within `zstd.h` :
+ * ZSTD_decompressBlock()
+ */
+
+/* note: prototypes already published within `zstd_internal.h` :
+ * ZSTD_getcBlockSize()
+ * ZSTD_decodeSeqHeaders()
+ */
+
+
+/* ZSTD_decompressBlock_internal() :
+ * decompress block, starting at `src`,
+ * into destination buffer `dst`.
+ * @return : decompressed block size,
+ *           or an error code (which can be tested using ZSTD_isError())
+ */
+size_t ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
+                               void* dst, size_t dstCapacity,
+                         const void* src, size_t srcSize, const int frame);
+
+/* ZSTD_buildFSETable() :
+ * generate FSE decoding table for one symbol (ll, ml or off)
+ * this function must be called with valid parameters only
+ * (dt is large enough, normalizedCounter distribution total is a power of 2, max is within range, etc.)
+ * in which case it cannot fail.
+ * The workspace must be 4-byte aligned and at least ZSTD_BUILD_FSE_TABLE_WKSP_SIZE bytes, which is
+ * defined in zstd_decompress_internal.h.
+ * Internal use only.
+ */
+void ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
+             const short* normalizedCounter, unsigned maxSymbolValue,
+             const U32* baseValue, const U32* nbAdditionalBits,
+                   unsigned tableLog, void* wksp, size_t wkspSize,
+                   int bmi2);
+
+
+#endif /* ZSTD_DEC_BLOCK_H */
diff --git a/lib/zstd/decompress/zstd_decompress_internal.h b/lib/zstd/decompress/zstd_decompress_internal.h
new file mode 100644
index 0000000..4b9052f
--- /dev/null
+++ b/lib/zstd/decompress/zstd_decompress_internal.h
@@ -0,0 +1,202 @@
+/*
+ * Copyright (c) Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+
+/* zstd_decompress_internal:
+ * objects and definitions shared within lib/decompress modules */
+
+ #ifndef ZSTD_DECOMPRESS_INTERNAL_H
+ #define ZSTD_DECOMPRESS_INTERNAL_H
+
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include "../common/mem.h"             /* BYTE, U16, U32 */
+#include "../common/zstd_internal.h"   /* ZSTD_seqSymbol */
+
+
+
+/*-*******************************************************
+ *  Constants
+ *********************************************************/
+static UNUSED_ATTR const U32 LL_base[MaxLL+1] = {
+                 0,    1,    2,     3,     4,     5,     6,      7,
+                 8,    9,   10,    11,    12,    13,    14,     15,
+                16,   18,   20,    22,    24,    28,    32,     40,
+                48,   64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000,
+                0x2000, 0x4000, 0x8000, 0x10000 };
+
+static UNUSED_ATTR const U32 OF_base[MaxOff+1] = {
+                 0,        1,       1,       5,     0xD,     0x1D,     0x3D,     0x7D,
+                 0xFD,   0x1FD,   0x3FD,   0x7FD,   0xFFD,   0x1FFD,   0x3FFD,   0x7FFD,
+                 0xFFFD, 0x1FFFD, 0x3FFFD, 0x7FFFD, 0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD,
+                 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD, 0x1FFFFFFD, 0x3FFFFFFD, 0x7FFFFFFD };
+
+static UNUSED_ATTR const U32 OF_bits[MaxOff+1] = {
+                     0,  1,  2,  3,  4,  5,  6,  7,
+                     8,  9, 10, 11, 12, 13, 14, 15,
+                    16, 17, 18, 19, 20, 21, 22, 23,
+                    24, 25, 26, 27, 28, 29, 30, 31 };
+
+static UNUSED_ATTR const U32 ML_base[MaxML+1] = {
+                     3,  4,  5,    6,     7,     8,     9,    10,
+                    11, 12, 13,   14,    15,    16,    17,    18,
+                    19, 20, 21,   22,    23,    24,    25,    26,
+                    27, 28, 29,   30,    31,    32,    33,    34,
+                    35, 37, 39,   41,    43,    47,    51,    59,
+                    67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803,
+                    0x1003, 0x2003, 0x4003, 0x8003, 0x10003 };
+
+
+/*-*******************************************************
+ *  Decompression types
+ *********************************************************/
+ typedef struct {
+     U32 fastMode;
+     U32 tableLog;
+ } ZSTD_seqSymbol_header;
+
+ typedef struct {
+     U16  nextState;
+     BYTE nbAdditionalBits;
+     BYTE nbBits;
+     U32  baseValue;
+ } ZSTD_seqSymbol;
+
+ #define SEQSYMBOL_TABLE_SIZE(log)   (1 + (1 << (log)))
+
+#define ZSTD_BUILD_FSE_TABLE_WKSP_SIZE (sizeof(S16) * (MaxSeq + 1) + (1u << MaxFSELog) + sizeof(U64))
+#define ZSTD_BUILD_FSE_TABLE_WKSP_SIZE_U32 ((ZSTD_BUILD_FSE_TABLE_WKSP_SIZE + sizeof(U32) - 1) / sizeof(U32))
+
+typedef struct {
+    ZSTD_seqSymbol LLTable[SEQSYMBOL_TABLE_SIZE(LLFSELog)];    /* Note : Space reserved for FSE Tables */
+    ZSTD_seqSymbol OFTable[SEQSYMBOL_TABLE_SIZE(OffFSELog)];   /* is also used as temporary workspace while building hufTable during DDict creation */
+    ZSTD_seqSymbol MLTable[SEQSYMBOL_TABLE_SIZE(MLFSELog)];    /* and therefore must be at least HUF_DECOMPRESS_WORKSPACE_SIZE large */
+    HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)];  /* can accommodate HUF_decompress4X */
+    U32 rep[ZSTD_REP_NUM];
+    U32 workspace[ZSTD_BUILD_FSE_TABLE_WKSP_SIZE_U32];
+} ZSTD_entropyDTables_t;
+
+typedef enum { ZSTDds_getFrameHeaderSize, ZSTDds_decodeFrameHeader,
+               ZSTDds_decodeBlockHeader, ZSTDds_decompressBlock,
+               ZSTDds_decompressLastBlock, ZSTDds_checkChecksum,
+               ZSTDds_decodeSkippableHeader, ZSTDds_skipFrame } ZSTD_dStage;
+
+typedef enum { zdss_init=0, zdss_loadHeader,
+               zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;
+
+typedef enum {
+    ZSTD_use_indefinitely = -1,  /* Use the dictionary indefinitely */
+    ZSTD_dont_use = 0,           /* Do not use the dictionary (if one exists free it) */
+    ZSTD_use_once = 1            /* Use the dictionary once and set to ZSTD_dont_use */
+} ZSTD_dictUses_e;
+
+/* Hashset for storing references to multiple ZSTD_DDict within ZSTD_DCtx */
+typedef struct {
+    const ZSTD_DDict** ddictPtrTable;
+    size_t ddictPtrTableSize;
+    size_t ddictPtrCount;
+} ZSTD_DDictHashSet;
+
+struct ZSTD_DCtx_s
+{
+    const ZSTD_seqSymbol* LLTptr;
+    const ZSTD_seqSymbol* MLTptr;
+    const ZSTD_seqSymbol* OFTptr;
+    const HUF_DTable* HUFptr;
+    ZSTD_entropyDTables_t entropy;
+    U32 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];   /* space needed when building huffman tables */
+    const void* previousDstEnd;   /* detect continuity */
+    const void* prefixStart;      /* start of current segment */
+    const void* virtualStart;     /* virtual start of previous segment if it was just before current one */
+    const void* dictEnd;          /* end of previous segment */
+    size_t expected;
+    ZSTD_frameHeader fParams;
+    U64 processedCSize;
+    U64 decodedSize;
+    blockType_e bType;            /* used in ZSTD_decompressContinue(), store blockType between block header decoding and block decompression stages */
+    ZSTD_dStage stage;
+    U32 litEntropy;
+    U32 fseEntropy;
+    struct xxh64_state xxhState;
+    size_t headerSize;
+    ZSTD_format_e format;
+    ZSTD_forceIgnoreChecksum_e forceIgnoreChecksum;   /* User specified: if == 1, will ignore checksums in compressed frame. Default == 0 */
+    U32 validateChecksum;         /* if == 1, will validate checksum. Is == 1 if (fParams.checksumFlag == 1) and (forceIgnoreChecksum == 0). */
+    const BYTE* litPtr;
+    ZSTD_customMem customMem;
+    size_t litSize;
+    size_t rleSize;
+    size_t staticSize;
+    int bmi2;                     /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */
+
+    /* dictionary */
+    ZSTD_DDict* ddictLocal;
+    const ZSTD_DDict* ddict;     /* set by ZSTD_initDStream_usingDDict(), or ZSTD_DCtx_refDDict() */
+    U32 dictID;
+    int ddictIsCold;             /* if == 1 : dictionary is "new" for working context, and presumed "cold" (not in cpu cache) */
+    ZSTD_dictUses_e dictUses;
+    ZSTD_DDictHashSet* ddictSet;                    /* Hash set for multiple ddicts */
+    ZSTD_refMultipleDDicts_e refMultipleDDicts;     /* User specified: if == 1, will allow references to multiple DDicts. Default == 0 (disabled) */
+
+    /* streaming */
+    ZSTD_dStreamStage streamStage;
+    char*  inBuff;
+    size_t inBuffSize;
+    size_t inPos;
+    size_t maxWindowSize;
+    char*  outBuff;
+    size_t outBuffSize;
+    size_t outStart;
+    size_t outEnd;
+    size_t lhSize;
+    void* legacyContext;
+    U32 previousLegacyVersion;
+    U32 legacyVersion;
+    U32 hostageByte;
+    int noForwardProgress;
+    ZSTD_bufferMode_e outBufferMode;
+    ZSTD_outBuffer expectedOutBuffer;
+
+    /* workspace */
+    BYTE litBuffer[ZSTD_BLOCKSIZE_MAX + WILDCOPY_OVERLENGTH];
+    BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];
+
+    size_t oversizedDuration;
+
+#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
+    void const* dictContentBeginForFuzzing;
+    void const* dictContentEndForFuzzing;
+#endif
+
+    /* Tracing */
+};  /* typedef'd to ZSTD_DCtx within "zstd.h" */
+
+
+/*-*******************************************************
+ *  Shared internal functions
+ *********************************************************/
+
+/*! ZSTD_loadDEntropy() :
+ *  dict : must point at beginning of a valid zstd dictionary.
+ * @return : size of dictionary header (size of magic number + dict ID + entropy tables) */
+size_t ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
+                   const void* const dict, size_t const dictSize);
+
+/*! ZSTD_checkContinuity() :
+ *  check if next `dst` follows previous position, where decompression ended.
+ *  If yes, do nothing (continue on current segment).
+ *  If not, classify previous segment as "external dictionary", and start a new segment.
+ *  This function cannot fail. */
+void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst, size_t dstSize);
+
+
+#endif /* ZSTD_DECOMPRESS_INTERNAL_H */
diff --git a/lib/zstd/decompress_sources.h b/lib/zstd/decompress_sources.h
new file mode 100644
index 0000000..0fbec50
--- /dev/null
+++ b/lib/zstd/decompress_sources.h
@@ -0,0 +1,28 @@
+/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/*
+ * This file includes every .c file needed for decompression.
+ * It is used by lib/decompress_unzstd.c to include the decompression
+ * source into the translation-unit, so it can be used for kernel
+ * decompression.
+ */
+
+#include "common/debug.c"
+#include "common/entropy_common.c"
+#include "common/error_private.c"
+#include "common/fse_decompress.c"
+#include "common/zstd_common.c"
+#include "decompress/huf_decompress.c"
+#include "decompress/zstd_ddict.c"
+#include "decompress/zstd_decompress.c"
+#include "decompress/zstd_decompress_block.c"
+#include "zstd_decompress_module.c"
diff --git a/lib/zstd/entropy_common.c b/lib/zstd/entropy_common.c
deleted file mode 100644
index 2b0a643..0000000
--- a/lib/zstd/entropy_common.c
+++ /dev/null
@@ -1,243 +0,0 @@
-/*
- * Common functions of New Generation Entropy library
- * Copyright (C) 2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-
-/* *************************************
-*  Dependencies
-***************************************/
-#include "error_private.h" /* ERR_*, ERROR */
-#include "fse.h"
-#include "huf.h"
-#include "mem.h"
-
-/*===   Version   ===*/
-unsigned FSE_versionNumber(void) { return FSE_VERSION_NUMBER; }
-
-/*===   Error Management   ===*/
-unsigned FSE_isError(size_t code) { return ERR_isError(code); }
-
-unsigned HUF_isError(size_t code) { return ERR_isError(code); }
-
-/*-**************************************************************
-*  FSE NCount encoding-decoding
-****************************************************************/
-size_t FSE_readNCount(short *normalizedCounter, unsigned *maxSVPtr, unsigned *tableLogPtr, const void *headerBuffer, size_t hbSize)
-{
-	const BYTE *const istart = (const BYTE *)headerBuffer;
-	const BYTE *const iend = istart + hbSize;
-	const BYTE *ip = istart;
-	int nbBits;
-	int remaining;
-	int threshold;
-	U32 bitStream;
-	int bitCount;
-	unsigned charnum = 0;
-	int previous0 = 0;
-
-	if (hbSize < 4)
-		return ERROR(srcSize_wrong);
-	bitStream = ZSTD_readLE32(ip);
-	nbBits = (bitStream & 0xF) + FSE_MIN_TABLELOG; /* extract tableLog */
-	if (nbBits > FSE_TABLELOG_ABSOLUTE_MAX)
-		return ERROR(tableLog_tooLarge);
-	bitStream >>= 4;
-	bitCount = 4;
-	*tableLogPtr = nbBits;
-	remaining = (1 << nbBits) + 1;
-	threshold = 1 << nbBits;
-	nbBits++;
-
-	while ((remaining > 1) & (charnum <= *maxSVPtr)) {
-		if (previous0) {
-			unsigned n0 = charnum;
-			while ((bitStream & 0xFFFF) == 0xFFFF) {
-				n0 += 24;
-				if (ip < iend - 5) {
-					ip += 2;
-					bitStream = ZSTD_readLE32(ip) >> bitCount;
-				} else {
-					bitStream >>= 16;
-					bitCount += 16;
-				}
-			}
-			while ((bitStream & 3) == 3) {
-				n0 += 3;
-				bitStream >>= 2;
-				bitCount += 2;
-			}
-			n0 += bitStream & 3;
-			bitCount += 2;
-			if (n0 > *maxSVPtr)
-				return ERROR(maxSymbolValue_tooSmall);
-			while (charnum < n0)
-				normalizedCounter[charnum++] = 0;
-			if ((ip <= iend - 7) || (ip + (bitCount >> 3) <= iend - 4)) {
-				ip += bitCount >> 3;
-				bitCount &= 7;
-				bitStream = ZSTD_readLE32(ip) >> bitCount;
-			} else {
-				bitStream >>= 2;
-			}
-		}
-		{
-			int const max = (2 * threshold - 1) - remaining;
-			int count;
-
-			if ((bitStream & (threshold - 1)) < (U32)max) {
-				count = bitStream & (threshold - 1);
-				bitCount += nbBits - 1;
-			} else {
-				count = bitStream & (2 * threshold - 1);
-				if (count >= threshold)
-					count -= max;
-				bitCount += nbBits;
-			}
-
-			count--;				 /* extra accuracy */
-			remaining -= count < 0 ? -count : count; /* -1 means +1 */
-			normalizedCounter[charnum++] = (short)count;
-			previous0 = !count;
-			while (remaining < threshold) {
-				nbBits--;
-				threshold >>= 1;
-			}
-
-			if ((ip <= iend - 7) || (ip + (bitCount >> 3) <= iend - 4)) {
-				ip += bitCount >> 3;
-				bitCount &= 7;
-			} else {
-				bitCount -= (int)(8 * (iend - 4 - ip));
-				ip = iend - 4;
-			}
-			bitStream = ZSTD_readLE32(ip) >> (bitCount & 31);
-		}
-	} /* while ((remaining>1) & (charnum<=*maxSVPtr)) */
-	if (remaining != 1)
-		return ERROR(corruption_detected);
-	if (bitCount > 32)
-		return ERROR(corruption_detected);
-	*maxSVPtr = charnum - 1;
-
-	ip += (bitCount + 7) >> 3;
-	return ip - istart;
-}
-
-/*! HUF_readStats() :
-	Read compact Huffman tree, saved by HUF_writeCTable().
-	`huffWeight` is destination buffer.
-	`rankStats` is assumed to be a table of at least HUF_TABLELOG_MAX U32.
-	@return : size read from `src` , or an error Code .
-	Note : Needed by HUF_readCTable() and HUF_readDTableX?() .
-*/
-size_t HUF_readStats_wksp(BYTE *huffWeight, size_t hwSize, U32 *rankStats, U32 *nbSymbolsPtr, U32 *tableLogPtr, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
-{
-	U32 weightTotal;
-	const BYTE *ip = (const BYTE *)src;
-	size_t iSize;
-	size_t oSize;
-
-	if (!srcSize)
-		return ERROR(srcSize_wrong);
-	iSize = ip[0];
-	/* memset(huffWeight, 0, hwSize);   */ /* is not necessary, even though some analyzer complain ... */
-
-	if (iSize >= 128) { /* special header */
-		oSize = iSize - 127;
-		iSize = ((oSize + 1) / 2);
-		if (iSize + 1 > srcSize)
-			return ERROR(srcSize_wrong);
-		if (oSize >= hwSize)
-			return ERROR(corruption_detected);
-		ip += 1;
-		{
-			U32 n;
-			for (n = 0; n < oSize; n += 2) {
-				huffWeight[n] = ip[n / 2] >> 4;
-				huffWeight[n + 1] = ip[n / 2] & 15;
-			}
-		}
-	} else {						 /* header compressed with FSE (normal case) */
-		if (iSize + 1 > srcSize)
-			return ERROR(srcSize_wrong);
-		oSize = FSE_decompress_wksp(huffWeight, hwSize - 1, ip + 1, iSize, 6, workspace, workspaceSize); /* max (hwSize-1) values decoded, as last one is implied */
-		if (FSE_isError(oSize))
-			return oSize;
-	}
-
-	/* collect weight stats */
-	memset(rankStats, 0, (HUF_TABLELOG_MAX + 1) * sizeof(U32));
-	weightTotal = 0;
-	{
-		U32 n;
-		for (n = 0; n < oSize; n++) {
-			if (huffWeight[n] >= HUF_TABLELOG_MAX)
-				return ERROR(corruption_detected);
-			rankStats[huffWeight[n]]++;
-			weightTotal += (1 << huffWeight[n]) >> 1;
-		}
-	}
-	if (weightTotal == 0)
-		return ERROR(corruption_detected);
-
-	/* get last non-null symbol weight (implied, total must be 2^n) */
-	{
-		U32 const tableLog = BIT_highbit32(weightTotal) + 1;
-		if (tableLog > HUF_TABLELOG_MAX)
-			return ERROR(corruption_detected);
-		*tableLogPtr = tableLog;
-		/* determine last weight */
-		{
-			U32 const total = 1 << tableLog;
-			U32 const rest = total - weightTotal;
-			U32 const verif = 1 << BIT_highbit32(rest);
-			U32 const lastWeight = BIT_highbit32(rest) + 1;
-			if (verif != rest)
-				return ERROR(corruption_detected); /* last value must be a clean power of 2 */
-			huffWeight[oSize] = (BYTE)lastWeight;
-			rankStats[lastWeight]++;
-		}
-	}
-
-	/* check tree construction validity */
-	if ((rankStats[1] < 2) || (rankStats[1] & 1))
-		return ERROR(corruption_detected); /* by construction : at least 2 elts of rank 1, must be even */
-
-	/* results */
-	*nbSymbolsPtr = (U32)(oSize + 1);
-	return iSize + 1;
-}
diff --git a/lib/zstd/error_private.h b/lib/zstd/error_private.h
deleted file mode 100644
index 1a60b31..0000000
--- a/lib/zstd/error_private.h
+++ /dev/null
@@ -1,53 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-/* Note : this module is expected to remain private, do not expose it */
-
-#ifndef ERROR_H_MODULE
-#define ERROR_H_MODULE
-
-/* ****************************************
-*  Dependencies
-******************************************/
-#include <linux/types.h> /* size_t */
-#include <linux/zstd.h>  /* enum list */
-
-/* ****************************************
-*  Compiler-specific
-******************************************/
-#define ERR_STATIC static __attribute__((unused))
-
-/*-****************************************
-*  Customization (error_public.h)
-******************************************/
-typedef ZSTD_ErrorCode ERR_enum;
-#define PREFIX(name) ZSTD_error_##name
-
-/*-****************************************
-*  Error codes handling
-******************************************/
-#define ERROR(name) ((size_t)-PREFIX(name))
-
-ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); }
-
-ERR_STATIC ERR_enum ERR_getErrorCode(size_t code)
-{
-	if (!ERR_isError(code))
-		return (ERR_enum)0;
-	return (ERR_enum)(0 - code);
-}
-
-#endif /* ERROR_H_MODULE */
diff --git a/lib/zstd/fse.h b/lib/zstd/fse.h
deleted file mode 100644
index 7460ab0..0000000
--- a/lib/zstd/fse.h
+++ /dev/null
@@ -1,575 +0,0 @@
-/*
- * FSE : Finite State Entropy codec
- * Public Prototypes declaration
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-#ifndef FSE_H
-#define FSE_H
-
-/*-*****************************************
-*  Dependencies
-******************************************/
-#include <linux/types.h> /* size_t, ptrdiff_t */
-
-/*-*****************************************
-*  FSE_PUBLIC_API : control library symbols visibility
-******************************************/
-#define FSE_PUBLIC_API
-
-/*------   Version   ------*/
-#define FSE_VERSION_MAJOR 0
-#define FSE_VERSION_MINOR 9
-#define FSE_VERSION_RELEASE 0
-
-#define FSE_LIB_VERSION FSE_VERSION_MAJOR.FSE_VERSION_MINOR.FSE_VERSION_RELEASE
-#define FSE_QUOTE(str) #str
-#define FSE_EXPAND_AND_QUOTE(str) FSE_QUOTE(str)
-#define FSE_VERSION_STRING FSE_EXPAND_AND_QUOTE(FSE_LIB_VERSION)
-
-#define FSE_VERSION_NUMBER (FSE_VERSION_MAJOR * 100 * 100 + FSE_VERSION_MINOR * 100 + FSE_VERSION_RELEASE)
-FSE_PUBLIC_API unsigned FSE_versionNumber(void); /**< library version number; to be used when checking dll version */
-
-/*-*****************************************
-*  Tool functions
-******************************************/
-FSE_PUBLIC_API size_t FSE_compressBound(size_t size); /* maximum compressed size */
-
-/* Error Management */
-FSE_PUBLIC_API unsigned FSE_isError(size_t code); /* tells if a return value is an error code */
-
-/*-*****************************************
-*  FSE detailed API
-******************************************/
-/*!
-FSE_compress() does the following:
-1. count symbol occurrence from source[] into table count[]
-2. normalize counters so that sum(count[]) == Power_of_2 (2^tableLog)
-3. save normalized counters to memory buffer using writeNCount()
-4. build encoding table 'CTable' from normalized counters
-5. encode the data stream using encoding table 'CTable'
-
-FSE_decompress() does the following:
-1. read normalized counters with readNCount()
-2. build decoding table 'DTable' from normalized counters
-3. decode the data stream using decoding table 'DTable'
-
-The following API allows targeting specific sub-functions for advanced tasks.
-For example, it's possible to compress several blocks using the same 'CTable',
-or to save and provide normalized distribution using external method.
-*/
-
-/* *** COMPRESSION *** */
-/*! FSE_optimalTableLog():
-	dynamically downsize 'tableLog' when conditions are met.
-	It saves CPU time, by using smaller tables, while preserving or even improving compression ratio.
-	@return : recommended tableLog (necessarily <= 'maxTableLog') */
-FSE_PUBLIC_API unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
-
-/*! FSE_normalizeCount():
-	normalize counts so that sum(count[]) == Power_of_2 (2^tableLog)
-	'normalizedCounter' is a table of short, of minimum size (maxSymbolValue+1).
-	@return : tableLog,
-			  or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_normalizeCount(short *normalizedCounter, unsigned tableLog, const unsigned *count, size_t srcSize, unsigned maxSymbolValue);
-
-/*! FSE_NCountWriteBound():
-	Provides the maximum possible size of an FSE normalized table, given 'maxSymbolValue' and 'tableLog'.
-	Typically useful for allocation purpose. */
-FSE_PUBLIC_API size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog);
-
-/*! FSE_writeNCount():
-	Compactly save 'normalizedCounter' into 'buffer'.
-	@return : size of the compressed table,
-			  or an errorCode, which can be tested using FSE_isError(). */
-FSE_PUBLIC_API size_t FSE_writeNCount(void *buffer, size_t bufferSize, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
-
-/*! Constructor and Destructor of FSE_CTable.
-	Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */
-typedef unsigned FSE_CTable; /* don't allocate that. It's only meant to be more restrictive than void* */
-
-/*! FSE_compress_usingCTable():
-	Compress `src` using `ct` into `dst` which must be already allocated.
-	@return : size of compressed data (<= `dstCapacity`),
-			  or 0 if compressed data could not fit into `dst`,
-			  or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_compress_usingCTable(void *dst, size_t dstCapacity, const void *src, size_t srcSize, const FSE_CTable *ct);
-
-/*!
-Tutorial :
-----------
-The first step is to count all symbols. FSE_count() does this job very fast.
-Result will be saved into 'count', a table of unsigned int, which must be already allocated, and have 'maxSymbolValuePtr[0]+1' cells.
-'src' is a table of bytes of size 'srcSize'. All values within 'src' MUST be <= maxSymbolValuePtr[0]
-maxSymbolValuePtr[0] will be updated, with its real value (necessarily <= original value)
-FSE_count() will return the number of occurrence of the most frequent symbol.
-This can be used to know if there is a single symbol within 'src', and to quickly evaluate its compressibility.
-If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
-
-The next step is to normalize the frequencies.
-FSE_normalizeCount() will ensure that sum of frequencies is == 2 ^'tableLog'.
-It also guarantees a minimum of 1 to any Symbol with frequency >= 1.
-You can use 'tableLog'==0 to mean "use default tableLog value".
-If you are unsure of which tableLog value to use, you can ask FSE_optimalTableLog(),
-which will provide the optimal valid tableLog given sourceSize, maxSymbolValue, and a user-defined maximum (0 means "default").
-
-The result of FSE_normalizeCount() will be saved into a table,
-called 'normalizedCounter', which is a table of signed short.
-'normalizedCounter' must be already allocated, and have at least 'maxSymbolValue+1' cells.
-The return value is tableLog if everything proceeded as expected.
-It is 0 if there is a single symbol within distribution.
-If there is an error (ex: invalid tableLog value), the function will return an ErrorCode (which can be tested using FSE_isError()).
-
-'normalizedCounter' can be saved in a compact manner to a memory area using FSE_writeNCount().
-'buffer' must be already allocated.
-For guaranteed success, buffer size must be at least FSE_headerBound().
-The result of the function is the number of bytes written into 'buffer'.
-If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError(); ex : buffer size too small).
-
-'normalizedCounter' can then be used to create the compression table 'CTable'.
-The space required by 'CTable' must be already allocated, using FSE_createCTable().
-You can then use FSE_buildCTable() to fill 'CTable'.
-If there is an error, both functions will return an ErrorCode (which can be tested using FSE_isError()).
-
-'CTable' can then be used to compress 'src', with FSE_compress_usingCTable().
-Similar to FSE_count(), the convention is that 'src' is assumed to be a table of char of size 'srcSize'
-The function returns the size of compressed data (without header), necessarily <= `dstCapacity`.
-If it returns '0', compressed data could not fit into 'dst'.
-If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
-*/
-
-/* *** DECOMPRESSION *** */
-
-/*! FSE_readNCount():
-	Read compactly saved 'normalizedCounter' from 'rBuffer'.
-	@return : size read from 'rBuffer',
-			  or an errorCode, which can be tested using FSE_isError().
-			  maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */
-FSE_PUBLIC_API size_t FSE_readNCount(short *normalizedCounter, unsigned *maxSymbolValuePtr, unsigned *tableLogPtr, const void *rBuffer, size_t rBuffSize);
-
-/*! Constructor and Destructor of FSE_DTable.
-	Note that its size depends on 'tableLog' */
-typedef unsigned FSE_DTable; /* don't allocate that. It's just a way to be more restrictive than void* */
-
-/*! FSE_buildDTable():
-	Builds 'dt', which must be already allocated, using FSE_createDTable().
-	return : 0, or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_buildDTable_wksp(FSE_DTable *dt, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workspace, size_t workspaceSize);
-
-/*! FSE_decompress_usingDTable():
-	Decompress compressed source `cSrc` of size `cSrcSize` using `dt`
-	into `dst` which must be already allocated.
-	@return : size of regenerated data (necessarily <= `dstCapacity`),
-			  or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_decompress_usingDTable(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt);
-
-/*!
-Tutorial :
-----------
-(Note : these functions only decompress FSE-compressed blocks.
- If block is uncompressed, use memcpy() instead
- If block is a single repeated byte, use memset() instead )
-
-The first step is to obtain the normalized frequencies of symbols.
-This can be performed by FSE_readNCount() if it was saved using FSE_writeNCount().
-'normalizedCounter' must be already allocated, and have at least 'maxSymbolValuePtr[0]+1' cells of signed short.
-In practice, that means it's necessary to know 'maxSymbolValue' beforehand,
-or size the table to handle worst case situations (typically 256).
-FSE_readNCount() will provide 'tableLog' and 'maxSymbolValue'.
-The result of FSE_readNCount() is the number of bytes read from 'rBuffer'.
-Note that 'rBufferSize' must be at least 4 bytes, even if useful information is less than that.
-If there is an error, the function will return an error code, which can be tested using FSE_isError().
-
-The next step is to build the decompression tables 'FSE_DTable' from 'normalizedCounter'.
-This is performed by the function FSE_buildDTable().
-The space required by 'FSE_DTable' must be already allocated using FSE_createDTable().
-If there is an error, the function will return an error code, which can be tested using FSE_isError().
-
-`FSE_DTable` can then be used to decompress `cSrc`, with FSE_decompress_usingDTable().
-`cSrcSize` must be strictly correct, otherwise decompression will fail.
-FSE_decompress_usingDTable() result will tell how many bytes were regenerated (<=`dstCapacity`).
-If there is an error, the function will return an error code, which can be tested using FSE_isError(). (ex: dst buffer too small)
-*/
-
-/* *** Dependency *** */
-#include "bitstream.h"
-
-/* *****************************************
-*  Static allocation
-*******************************************/
-/* FSE buffer bounds */
-#define FSE_NCOUNTBOUND 512
-#define FSE_BLOCKBOUND(size) (size + (size >> 7))
-#define FSE_COMPRESSBOUND(size) (FSE_NCOUNTBOUND + FSE_BLOCKBOUND(size)) /* Macro version, useful for static allocation */
-
-/* It is possible to statically allocate FSE CTable/DTable as a table of FSE_CTable/FSE_DTable using below macros */
-#define FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) (1 + (1 << (maxTableLog - 1)) + ((maxSymbolValue + 1) * 2))
-#define FSE_DTABLE_SIZE_U32(maxTableLog) (1 + (1 << maxTableLog))
-
-/* *****************************************
-*  FSE advanced API
-*******************************************/
-/* FSE_count_wksp() :
- * Same as FSE_count(), but using an externally provided scratch buffer.
- * `workSpace` size must be table of >= `1024` unsigned
- */
-size_t FSE_count_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned *workSpace);
-
-/* FSE_countFast_wksp() :
- * Same as FSE_countFast(), but using an externally provided scratch buffer.
- * `workSpace` must be a table of minimum `1024` unsigned
- */
-size_t FSE_countFast_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *src, size_t srcSize, unsigned *workSpace);
-
-/*! FSE_count_simple
- * Same as FSE_countFast(), but does not use any additional memory (not even on stack).
- * This function is unsafe, and will segfault if any value within `src` is `> *maxSymbolValuePtr` (presuming it's also the size of `count`).
-*/
-size_t FSE_count_simple(unsigned *count, unsigned *maxSymbolValuePtr, const void *src, size_t srcSize);
-
-unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus);
-/**< same as FSE_optimalTableLog(), which used `minus==2` */
-
-size_t FSE_buildCTable_raw(FSE_CTable *ct, unsigned nbBits);
-/**< build a fake FSE_CTable, designed for a flat distribution, where each symbol uses nbBits */
-
-size_t FSE_buildCTable_rle(FSE_CTable *ct, unsigned char symbolValue);
-/**< build a fake FSE_CTable, designed to compress always the same symbolValue */
-
-/* FSE_buildCTable_wksp() :
- * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
- * `wkspSize` must be >= `(1<<tableLog)`.
- */
-size_t FSE_buildCTable_wksp(FSE_CTable *ct, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workSpace, size_t wkspSize);
-
-size_t FSE_buildDTable_raw(FSE_DTable *dt, unsigned nbBits);
-/**< build a fake FSE_DTable, designed to read a flat distribution where each symbol uses nbBits */
-
-size_t FSE_buildDTable_rle(FSE_DTable *dt, unsigned char symbolValue);
-/**< build a fake FSE_DTable, designed to always generate the same symbolValue */
-
-size_t FSE_decompress_wksp(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, unsigned maxLog, void *workspace, size_t workspaceSize);
-/**< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DTABLE_SIZE_U32(maxLog)` */
-
-/* *****************************************
-*  FSE symbol compression API
-*******************************************/
-/*!
-   This API consists of small unitary functions, which highly benefit from being inlined.
-   Hence their body are included in next section.
-*/
-typedef struct {
-	ptrdiff_t value;
-	const void *stateTable;
-	const void *symbolTT;
-	unsigned stateLog;
-} FSE_CState_t;
-
-static void FSE_initCState(FSE_CState_t *CStatePtr, const FSE_CTable *ct);
-
-static void FSE_encodeSymbol(BIT_CStream_t *bitC, FSE_CState_t *CStatePtr, unsigned symbol);
-
-static void FSE_flushCState(BIT_CStream_t *bitC, const FSE_CState_t *CStatePtr);
-
-/**<
-These functions are inner components of FSE_compress_usingCTable().
-They allow the creation of custom streams, mixing multiple tables and bit sources.
-
-A key property to keep in mind is that encoding and decoding are done **in reverse direction**.
-So the first symbol you will encode is the last you will decode, like a LIFO stack.
-
-You will need a few variables to track your CStream. They are :
-
-FSE_CTable    ct;         // Provided by FSE_buildCTable()
-BIT_CStream_t bitStream;  // bitStream tracking structure
-FSE_CState_t  state;      // State tracking structure (can have several)
-
-
-The first thing to do is to init bitStream and state.
-	size_t errorCode = BIT_initCStream(&bitStream, dstBuffer, maxDstSize);
-	FSE_initCState(&state, ct);
-
-Note that BIT_initCStream() can produce an error code, so its result should be tested, using FSE_isError();
-You can then encode your input data, byte after byte.
-FSE_encodeSymbol() outputs a maximum of 'tableLog' bits at a time.
-Remember decoding will be done in reverse direction.
-	FSE_encodeByte(&bitStream, &state, symbol);
-
-At any time, you can also add any bit sequence.
-Note : maximum allowed nbBits is 25, for compatibility with 32-bits decoders
-	BIT_addBits(&bitStream, bitField, nbBits);
-
-The above methods don't commit data to memory, they just store it into local register, for speed.
-Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
-Writing data to memory is a manual operation, performed by the flushBits function.
-	BIT_flushBits(&bitStream);
-
-Your last FSE encoding operation shall be to flush your last state value(s).
-	FSE_flushState(&bitStream, &state);
-
-Finally, you must close the bitStream.
-The function returns the size of CStream in bytes.
-If data couldn't fit into dstBuffer, it will return a 0 ( == not compressible)
-If there is an error, it returns an errorCode (which can be tested using FSE_isError()).
-	size_t size = BIT_closeCStream(&bitStream);
-*/
-
-/* *****************************************
-*  FSE symbol decompression API
-*******************************************/
-typedef struct {
-	size_t state;
-	const void *table; /* precise table may vary, depending on U16 */
-} FSE_DState_t;
-
-static void FSE_initDState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD, const FSE_DTable *dt);
-
-static unsigned char FSE_decodeSymbol(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD);
-
-static unsigned FSE_endOfDState(const FSE_DState_t *DStatePtr);
-
-/**<
-Let's now decompose FSE_decompress_usingDTable() into its unitary components.
-You will decode FSE-encoded symbols from the bitStream,
-and also any other bitFields you put in, **in reverse order**.
-
-You will need a few variables to track your bitStream. They are :
-
-BIT_DStream_t DStream;    // Stream context
-FSE_DState_t  DState;     // State context. Multiple ones are possible
-FSE_DTable*   DTablePtr;  // Decoding table, provided by FSE_buildDTable()
-
-The first thing to do is to init the bitStream.
-	errorCode = BIT_initDStream(&DStream, srcBuffer, srcSize);
-
-You should then retrieve your initial state(s)
-(in reverse flushing order if you have several ones) :
-	errorCode = FSE_initDState(&DState, &DStream, DTablePtr);
-
-You can then decode your data, symbol after symbol.
-For information the maximum number of bits read by FSE_decodeSymbol() is 'tableLog'.
-Keep in mind that symbols are decoded in reverse order, like a LIFO stack (last in, first out).
-	unsigned char symbol = FSE_decodeSymbol(&DState, &DStream);
-
-You can retrieve any bitfield you eventually stored into the bitStream (in reverse order)
-Note : maximum allowed nbBits is 25, for 32-bits compatibility
-	size_t bitField = BIT_readBits(&DStream, nbBits);
-
-All above operations only read from local register (which size depends on size_t).
-Refueling the register from memory is manually performed by the reload method.
-	endSignal = FSE_reloadDStream(&DStream);
-
-BIT_reloadDStream() result tells if there is still some more data to read from DStream.
-BIT_DStream_unfinished : there is still some data left into the DStream.
-BIT_DStream_endOfBuffer : Dstream reached end of buffer. Its container may no longer be completely filled.
-BIT_DStream_completed : Dstream reached its exact end, corresponding in general to decompression completed.
-BIT_DStream_tooFar : Dstream went too far. Decompression result is corrupted.
-
-When reaching end of buffer (BIT_DStream_endOfBuffer), progress slowly, notably if you decode multiple symbols per loop,
-to properly detect the exact end of stream.
-After each decoded symbol, check if DStream is fully consumed using this simple test :
-	BIT_reloadDStream(&DStream) >= BIT_DStream_completed
-
-When it's done, verify decompression is fully completed, by checking both DStream and the relevant states.
-Checking if DStream has reached its end is performed by :
-	BIT_endOfDStream(&DStream);
-Check also the states. There might be some symbols left there, if some high probability ones (>50%) are possible.
-	FSE_endOfDState(&DState);
-*/
-
-/* *****************************************
-*  FSE unsafe API
-*******************************************/
-static unsigned char FSE_decodeSymbolFast(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD);
-/* faster, but works only if nbBits is always >= 1 (otherwise, result will be corrupted) */
-
-/* *****************************************
-*  Implementation of inlined functions
-*******************************************/
-typedef struct {
-	int deltaFindState;
-	U32 deltaNbBits;
-} FSE_symbolCompressionTransform; /* total 8 bytes */
-
-ZSTD_STATIC void FSE_initCState(FSE_CState_t *statePtr, const FSE_CTable *ct)
-{
-	const void *ptr = ct;
-	const U16 *u16ptr = (const U16 *)ptr;
-	const U32 tableLog = ZSTD_read16(ptr);
-	statePtr->value = (ptrdiff_t)1 << tableLog;
-	statePtr->stateTable = u16ptr + 2;
-	statePtr->symbolTT = ((const U32 *)ct + 1 + (tableLog ? (1 << (tableLog - 1)) : 1));
-	statePtr->stateLog = tableLog;
-}
-
-/*! FSE_initCState2() :
-*   Same as FSE_initCState(), but the first symbol to include (which will be the last to be read)
-*   uses the smallest state value possible, saving the cost of this symbol */
-ZSTD_STATIC void FSE_initCState2(FSE_CState_t *statePtr, const FSE_CTable *ct, U32 symbol)
-{
-	FSE_initCState(statePtr, ct);
-	{
-		const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform *)(statePtr->symbolTT))[symbol];
-		const U16 *stateTable = (const U16 *)(statePtr->stateTable);
-		U32 nbBitsOut = (U32)((symbolTT.deltaNbBits + (1 << 15)) >> 16);
-		statePtr->value = (nbBitsOut << 16) - symbolTT.deltaNbBits;
-		statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
-	}
-}
-
-ZSTD_STATIC void FSE_encodeSymbol(BIT_CStream_t *bitC, FSE_CState_t *statePtr, U32 symbol)
-{
-	const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform *)(statePtr->symbolTT))[symbol];
-	const U16 *const stateTable = (const U16 *)(statePtr->stateTable);
-	U32 nbBitsOut = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16);
-	BIT_addBits(bitC, statePtr->value, nbBitsOut);
-	statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
-}
-
-ZSTD_STATIC void FSE_flushCState(BIT_CStream_t *bitC, const FSE_CState_t *statePtr)
-{
-	BIT_addBits(bitC, statePtr->value, statePtr->stateLog);
-	BIT_flushBits(bitC);
-}
-
-/* ======    Decompression    ====== */
-
-typedef struct {
-	U16 tableLog;
-	U16 fastMode;
-} FSE_DTableHeader; /* sizeof U32 */
-
-typedef struct {
-	unsigned short newState;
-	unsigned char symbol;
-	unsigned char nbBits;
-} FSE_decode_t; /* size == U32 */
-
-ZSTD_STATIC void FSE_initDState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD, const FSE_DTable *dt)
-{
-	const void *ptr = dt;
-	const FSE_DTableHeader *const DTableH = (const FSE_DTableHeader *)ptr;
-	DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
-	BIT_reloadDStream(bitD);
-	DStatePtr->table = dt + 1;
-}
-
-ZSTD_STATIC BYTE FSE_peekSymbol(const FSE_DState_t *DStatePtr)
-{
-	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
-	return DInfo.symbol;
-}
-
-ZSTD_STATIC void FSE_updateState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
-{
-	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
-	U32 const nbBits = DInfo.nbBits;
-	size_t const lowBits = BIT_readBits(bitD, nbBits);
-	DStatePtr->state = DInfo.newState + lowBits;
-}
-
-ZSTD_STATIC BYTE FSE_decodeSymbol(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
-{
-	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
-	U32 const nbBits = DInfo.nbBits;
-	BYTE const symbol = DInfo.symbol;
-	size_t const lowBits = BIT_readBits(bitD, nbBits);
-
-	DStatePtr->state = DInfo.newState + lowBits;
-	return symbol;
-}
-
-/*! FSE_decodeSymbolFast() :
-	unsafe, only works if no symbol has a probability > 50% */
-ZSTD_STATIC BYTE FSE_decodeSymbolFast(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
-{
-	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
-	U32 const nbBits = DInfo.nbBits;
-	BYTE const symbol = DInfo.symbol;
-	size_t const lowBits = BIT_readBitsFast(bitD, nbBits);
-
-	DStatePtr->state = DInfo.newState + lowBits;
-	return symbol;
-}
-
-ZSTD_STATIC unsigned FSE_endOfDState(const FSE_DState_t *DStatePtr) { return DStatePtr->state == 0; }
-
-/* **************************************************************
-*  Tuning parameters
-****************************************************************/
-/*!MEMORY_USAGE :
-*  Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; etc.)
-*  Increasing memory usage improves compression ratio
-*  Reduced memory usage can improve speed, due to cache effect
-*  Recommended max value is 14, for 16KB, which nicely fits into Intel x86 L1 cache */
-#ifndef FSE_MAX_MEMORY_USAGE
-#define FSE_MAX_MEMORY_USAGE 14
-#endif
-#ifndef FSE_DEFAULT_MEMORY_USAGE
-#define FSE_DEFAULT_MEMORY_USAGE 13
-#endif
-
-/*!FSE_MAX_SYMBOL_VALUE :
-*  Maximum symbol value authorized.
-*  Required for proper stack allocation */
-#ifndef FSE_MAX_SYMBOL_VALUE
-#define FSE_MAX_SYMBOL_VALUE 255
-#endif
-
-/* **************************************************************
-*  template functions type & suffix
-****************************************************************/
-#define FSE_FUNCTION_TYPE BYTE
-#define FSE_FUNCTION_EXTENSION
-#define FSE_DECODE_TYPE FSE_decode_t
-
-/* ***************************************************************
-*  Constants
-*****************************************************************/
-#define FSE_MAX_TABLELOG (FSE_MAX_MEMORY_USAGE - 2)
-#define FSE_MAX_TABLESIZE (1U << FSE_MAX_TABLELOG)
-#define FSE_MAXTABLESIZE_MASK (FSE_MAX_TABLESIZE - 1)
-#define FSE_DEFAULT_TABLELOG (FSE_DEFAULT_MEMORY_USAGE - 2)
-#define FSE_MIN_TABLELOG 5
-
-#define FSE_TABLELOG_ABSOLUTE_MAX 15
-#if FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX
-#error "FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX is not supported"
-#endif
-
-#define FSE_TABLESTEP(tableSize) ((tableSize >> 1) + (tableSize >> 3) + 3)
-
-#endif /* FSE_H */
diff --git a/lib/zstd/fse_compress.c b/lib/zstd/fse_compress.c
deleted file mode 100644
index ef3d174..0000000
--- a/lib/zstd/fse_compress.c
+++ /dev/null
@@ -1,795 +0,0 @@
-/*
- * FSE : Finite State Entropy encoder
- * Copyright (C) 2013-2015, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-
-/* **************************************************************
-*  Compiler specifics
-****************************************************************/
-#define FORCE_INLINE static __always_inline
-
-/* **************************************************************
-*  Includes
-****************************************************************/
-#include "bitstream.h"
-#include "fse.h"
-#include <linux/compiler.h>
-#include <linux/kernel.h>
-#include <linux/math64.h>
-#include <linux/string.h> /* memcpy, memset */
-
-/* **************************************************************
-*  Error Management
-****************************************************************/
-#define FSE_STATIC_ASSERT(c)                                   \
-	{                                                      \
-		enum { FSE_static_assert = 1 / (int)(!!(c)) }; \
-	} /* use only *after* variable declarations */
-
-/* **************************************************************
-*  Templates
-****************************************************************/
-/*
-  designed to be included
-  for type-specific functions (template emulation in C)
-  Objective is to write these functions only once, for improved maintenance
-*/
-
-/* safety checks */
-#ifndef FSE_FUNCTION_EXTENSION
-#error "FSE_FUNCTION_EXTENSION must be defined"
-#endif
-#ifndef FSE_FUNCTION_TYPE
-#error "FSE_FUNCTION_TYPE must be defined"
-#endif
-
-/* Function names */
-#define FSE_CAT(X, Y) X##Y
-#define FSE_FUNCTION_NAME(X, Y) FSE_CAT(X, Y)
-#define FSE_TYPE_NAME(X, Y) FSE_CAT(X, Y)
-
-/* Function templates */
-
-/* FSE_buildCTable_wksp() :
- * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
- * wkspSize should be sized to handle worst case situation, which is `1<<max_tableLog * sizeof(FSE_FUNCTION_TYPE)`
- * workSpace must also be properly aligned with FSE_FUNCTION_TYPE requirements
- */
-size_t FSE_buildCTable_wksp(FSE_CTable *ct, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workspace, size_t workspaceSize)
-{
-	U32 const tableSize = 1 << tableLog;
-	U32 const tableMask = tableSize - 1;
-	void *const ptr = ct;
-	U16 *const tableU16 = ((U16 *)ptr) + 2;
-	void *const FSCT = ((U32 *)ptr) + 1 /* header */ + (tableLog ? tableSize >> 1 : 1);
-	FSE_symbolCompressionTransform *const symbolTT = (FSE_symbolCompressionTransform *)(FSCT);
-	U32 const step = FSE_TABLESTEP(tableSize);
-	U32 highThreshold = tableSize - 1;
-
-	U32 *cumul;
-	FSE_FUNCTION_TYPE *tableSymbol;
-	size_t spaceUsed32 = 0;
-
-	cumul = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += FSE_MAX_SYMBOL_VALUE + 2;
-	tableSymbol = (FSE_FUNCTION_TYPE *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(sizeof(FSE_FUNCTION_TYPE) * ((size_t)1 << tableLog), sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	/* CTable header */
-	tableU16[-2] = (U16)tableLog;
-	tableU16[-1] = (U16)maxSymbolValue;
-
-	/* For explanations on how to distribute symbol values over the table :
-	*  http://fastcompression.blogspot.fr/2014/02/fse-distributing-symbol-values.html */
-
-	/* symbol start positions */
-	{
-		U32 u;
-		cumul[0] = 0;
-		for (u = 1; u <= maxSymbolValue + 1; u++) {
-			if (normalizedCounter[u - 1] == -1) { /* Low proba symbol */
-				cumul[u] = cumul[u - 1] + 1;
-				tableSymbol[highThreshold--] = (FSE_FUNCTION_TYPE)(u - 1);
-			} else {
-				cumul[u] = cumul[u - 1] + normalizedCounter[u - 1];
-			}
-		}
-		cumul[maxSymbolValue + 1] = tableSize + 1;
-	}
-
-	/* Spread symbols */
-	{
-		U32 position = 0;
-		U32 symbol;
-		for (symbol = 0; symbol <= maxSymbolValue; symbol++) {
-			int nbOccurences;
-			for (nbOccurences = 0; nbOccurences < normalizedCounter[symbol]; nbOccurences++) {
-				tableSymbol[position] = (FSE_FUNCTION_TYPE)symbol;
-				position = (position + step) & tableMask;
-				while (position > highThreshold)
-					position = (position + step) & tableMask; /* Low proba area */
-			}
-		}
-
-		if (position != 0)
-			return ERROR(GENERIC); /* Must have gone through all positions */
-	}
-
-	/* Build table */
-	{
-		U32 u;
-		for (u = 0; u < tableSize; u++) {
-			FSE_FUNCTION_TYPE s = tableSymbol[u];	/* note : static analyzer may not understand tableSymbol is properly initialized */
-			tableU16[cumul[s]++] = (U16)(tableSize + u); /* TableU16 : sorted by symbol order; gives next state value */
-		}
-	}
-
-	/* Build Symbol Transformation Table */
-	{
-		unsigned total = 0;
-		unsigned s;
-		for (s = 0; s <= maxSymbolValue; s++) {
-			switch (normalizedCounter[s]) {
-			case 0: break;
-
-			case -1:
-			case 1:
-				symbolTT[s].deltaNbBits = (tableLog << 16) - (1 << tableLog);
-				symbolTT[s].deltaFindState = total - 1;
-				total++;
-				break;
-			default: {
-				U32 const maxBitsOut = tableLog - BIT_highbit32(normalizedCounter[s] - 1);
-				U32 const minStatePlus = normalizedCounter[s] << maxBitsOut;
-				symbolTT[s].deltaNbBits = (maxBitsOut << 16) - minStatePlus;
-				symbolTT[s].deltaFindState = total - normalizedCounter[s];
-				total += normalizedCounter[s];
-			}
-			}
-		}
-	}
-
-	return 0;
-}
-
-/*-**************************************************************
-*  FSE NCount encoding-decoding
-****************************************************************/
-size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog)
-{
-	size_t const maxHeaderSize = (((maxSymbolValue + 1) * tableLog) >> 3) + 3;
-	return maxSymbolValue ? maxHeaderSize : FSE_NCOUNTBOUND; /* maxSymbolValue==0 ? use default */
-}
-
-static size_t FSE_writeNCount_generic(void *header, size_t headerBufferSize, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog,
-				      unsigned writeIsSafe)
-{
-	BYTE *const ostart = (BYTE *)header;
-	BYTE *out = ostart;
-	BYTE *const oend = ostart + headerBufferSize;
-	int nbBits;
-	const int tableSize = 1 << tableLog;
-	int remaining;
-	int threshold;
-	U32 bitStream;
-	int bitCount;
-	unsigned charnum = 0;
-	int previous0 = 0;
-
-	bitStream = 0;
-	bitCount = 0;
-	/* Table Size */
-	bitStream += (tableLog - FSE_MIN_TABLELOG) << bitCount;
-	bitCount += 4;
-
-	/* Init */
-	remaining = tableSize + 1; /* +1 for extra accuracy */
-	threshold = tableSize;
-	nbBits = tableLog + 1;
-
-	while (remaining > 1) { /* stops at 1 */
-		if (previous0) {
-			unsigned start = charnum;
-			while (!normalizedCounter[charnum])
-				charnum++;
-			while (charnum >= start + 24) {
-				start += 24;
-				bitStream += 0xFFFFU << bitCount;
-				if ((!writeIsSafe) && (out > oend - 2))
-					return ERROR(dstSize_tooSmall); /* Buffer overflow */
-				out[0] = (BYTE)bitStream;
-				out[1] = (BYTE)(bitStream >> 8);
-				out += 2;
-				bitStream >>= 16;
-			}
-			while (charnum >= start + 3) {
-				start += 3;
-				bitStream += 3 << bitCount;
-				bitCount += 2;
-			}
-			bitStream += (charnum - start) << bitCount;
-			bitCount += 2;
-			if (bitCount > 16) {
-				if ((!writeIsSafe) && (out > oend - 2))
-					return ERROR(dstSize_tooSmall); /* Buffer overflow */
-				out[0] = (BYTE)bitStream;
-				out[1] = (BYTE)(bitStream >> 8);
-				out += 2;
-				bitStream >>= 16;
-				bitCount -= 16;
-			}
-		}
-		{
-			int count = normalizedCounter[charnum++];
-			int const max = (2 * threshold - 1) - remaining;
-			remaining -= count < 0 ? -count : count;
-			count++; /* +1 for extra accuracy */
-			if (count >= threshold)
-				count += max; /* [0..max[ [max..threshold[ (...) [threshold+max 2*threshold[ */
-			bitStream += count << bitCount;
-			bitCount += nbBits;
-			bitCount -= (count < max);
-			previous0 = (count == 1);
-			if (remaining < 1)
-				return ERROR(GENERIC);
-			while (remaining < threshold)
-				nbBits--, threshold >>= 1;
-		}
-		if (bitCount > 16) {
-			if ((!writeIsSafe) && (out > oend - 2))
-				return ERROR(dstSize_tooSmall); /* Buffer overflow */
-			out[0] = (BYTE)bitStream;
-			out[1] = (BYTE)(bitStream >> 8);
-			out += 2;
-			bitStream >>= 16;
-			bitCount -= 16;
-		}
-	}
-
-	/* flush remaining bitStream */
-	if ((!writeIsSafe) && (out > oend - 2))
-		return ERROR(dstSize_tooSmall); /* Buffer overflow */
-	out[0] = (BYTE)bitStream;
-	out[1] = (BYTE)(bitStream >> 8);
-	out += (bitCount + 7) / 8;
-
-	if (charnum > maxSymbolValue + 1)
-		return ERROR(GENERIC);
-
-	return (out - ostart);
-}
-
-size_t FSE_writeNCount(void *buffer, size_t bufferSize, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog)
-{
-	if (tableLog > FSE_MAX_TABLELOG)
-		return ERROR(tableLog_tooLarge); /* Unsupported */
-	if (tableLog < FSE_MIN_TABLELOG)
-		return ERROR(GENERIC); /* Unsupported */
-
-	if (bufferSize < FSE_NCountWriteBound(maxSymbolValue, tableLog))
-		return FSE_writeNCount_generic(buffer, bufferSize, normalizedCounter, maxSymbolValue, tableLog, 0);
-
-	return FSE_writeNCount_generic(buffer, bufferSize, normalizedCounter, maxSymbolValue, tableLog, 1);
-}
-
-/*-**************************************************************
-*  Counting histogram
-****************************************************************/
-/*! FSE_count_simple
-	This function counts byte values within `src`, and store the histogram into table `count`.
-	It doesn't use any additional memory.
-	But this function is unsafe : it doesn't check that all values within `src` can fit into `count`.
-	For this reason, prefer using a table `count` with 256 elements.
-	@return : count of most numerous element
-*/
-size_t FSE_count_simple(unsigned *count, unsigned *maxSymbolValuePtr, const void *src, size_t srcSize)
-{
-	const BYTE *ip = (const BYTE *)src;
-	const BYTE *const end = ip + srcSize;
-	unsigned maxSymbolValue = *maxSymbolValuePtr;
-	unsigned max = 0;
-
-	memset(count, 0, (maxSymbolValue + 1) * sizeof(*count));
-	if (srcSize == 0) {
-		*maxSymbolValuePtr = 0;
-		return 0;
-	}
-
-	while (ip < end)
-		count[*ip++]++;
-
-	while (!count[maxSymbolValue])
-		maxSymbolValue--;
-	*maxSymbolValuePtr = maxSymbolValue;
-
-	{
-		U32 s;
-		for (s = 0; s <= maxSymbolValue; s++)
-			if (count[s] > max)
-				max = count[s];
-	}
-
-	return (size_t)max;
-}
-
-/* FSE_count_parallel_wksp() :
- * Same as FSE_count_parallel(), but using an externally provided scratch buffer.
- * `workSpace` size must be a minimum of `1024 * sizeof(unsigned)`` */
-static size_t FSE_count_parallel_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned checkMax,
-				      unsigned *const workSpace)
-{
-	const BYTE *ip = (const BYTE *)source;
-	const BYTE *const iend = ip + sourceSize;
-	unsigned maxSymbolValue = *maxSymbolValuePtr;
-	unsigned max = 0;
-	U32 *const Counting1 = workSpace;
-	U32 *const Counting2 = Counting1 + 256;
-	U32 *const Counting3 = Counting2 + 256;
-	U32 *const Counting4 = Counting3 + 256;
-
-	memset(Counting1, 0, 4 * 256 * sizeof(unsigned));
-
-	/* safety checks */
-	if (!sourceSize) {
-		memset(count, 0, maxSymbolValue + 1);
-		*maxSymbolValuePtr = 0;
-		return 0;
-	}
-	if (!maxSymbolValue)
-		maxSymbolValue = 255; /* 0 == default */
-
-	/* by stripes of 16 bytes */
-	{
-		U32 cached = ZSTD_read32(ip);
-		ip += 4;
-		while (ip < iend - 15) {
-			U32 c = cached;
-			cached = ZSTD_read32(ip);
-			ip += 4;
-			Counting1[(BYTE)c]++;
-			Counting2[(BYTE)(c >> 8)]++;
-			Counting3[(BYTE)(c >> 16)]++;
-			Counting4[c >> 24]++;
-			c = cached;
-			cached = ZSTD_read32(ip);
-			ip += 4;
-			Counting1[(BYTE)c]++;
-			Counting2[(BYTE)(c >> 8)]++;
-			Counting3[(BYTE)(c >> 16)]++;
-			Counting4[c >> 24]++;
-			c = cached;
-			cached = ZSTD_read32(ip);
-			ip += 4;
-			Counting1[(BYTE)c]++;
-			Counting2[(BYTE)(c >> 8)]++;
-			Counting3[(BYTE)(c >> 16)]++;
-			Counting4[c >> 24]++;
-			c = cached;
-			cached = ZSTD_read32(ip);
-			ip += 4;
-			Counting1[(BYTE)c]++;
-			Counting2[(BYTE)(c >> 8)]++;
-			Counting3[(BYTE)(c >> 16)]++;
-			Counting4[c >> 24]++;
-		}
-		ip -= 4;
-	}
-
-	/* finish last symbols */
-	while (ip < iend)
-		Counting1[*ip++]++;
-
-	if (checkMax) { /* verify stats will fit into destination table */
-		U32 s;
-		for (s = 255; s > maxSymbolValue; s--) {
-			Counting1[s] += Counting2[s] + Counting3[s] + Counting4[s];
-			if (Counting1[s])
-				return ERROR(maxSymbolValue_tooSmall);
-		}
-	}
-
-	{
-		U32 s;
-		for (s = 0; s <= maxSymbolValue; s++) {
-			count[s] = Counting1[s] + Counting2[s] + Counting3[s] + Counting4[s];
-			if (count[s] > max)
-				max = count[s];
-		}
-	}
-
-	while (!count[maxSymbolValue])
-		maxSymbolValue--;
-	*maxSymbolValuePtr = maxSymbolValue;
-	return (size_t)max;
-}
-
-/* FSE_countFast_wksp() :
- * Same as FSE_countFast(), but using an externally provided scratch buffer.
- * `workSpace` size must be table of >= `1024` unsigned */
-size_t FSE_countFast_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned *workSpace)
-{
-	if (sourceSize < 1500)
-		return FSE_count_simple(count, maxSymbolValuePtr, source, sourceSize);
-	return FSE_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, 0, workSpace);
-}
-
-/* FSE_count_wksp() :
- * Same as FSE_count(), but using an externally provided scratch buffer.
- * `workSpace` size must be table of >= `1024` unsigned */
-size_t FSE_count_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned *workSpace)
-{
-	if (*maxSymbolValuePtr < 255)
-		return FSE_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, 1, workSpace);
-	*maxSymbolValuePtr = 255;
-	return FSE_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, workSpace);
-}
-
-/*-**************************************************************
-*  FSE Compression Code
-****************************************************************/
-/*! FSE_sizeof_CTable() :
-	FSE_CTable is a variable size structure which contains :
-	`U16 tableLog;`
-	`U16 maxSymbolValue;`
-	`U16 nextStateNumber[1 << tableLog];`                         // This size is variable
-	`FSE_symbolCompressionTransform symbolTT[maxSymbolValue+1];`  // This size is variable
-Allocation is manual (C standard does not support variable-size structures).
-*/
-size_t FSE_sizeof_CTable(unsigned maxSymbolValue, unsigned tableLog)
-{
-	if (tableLog > FSE_MAX_TABLELOG)
-		return ERROR(tableLog_tooLarge);
-	return FSE_CTABLE_SIZE_U32(tableLog, maxSymbolValue) * sizeof(U32);
-}
-
-/* provides the minimum logSize to safely represent a distribution */
-static unsigned FSE_minTableLog(size_t srcSize, unsigned maxSymbolValue)
-{
-	U32 minBitsSrc = BIT_highbit32((U32)(srcSize - 1)) + 1;
-	U32 minBitsSymbols = BIT_highbit32(maxSymbolValue) + 2;
-	U32 minBits = minBitsSrc < minBitsSymbols ? minBitsSrc : minBitsSymbols;
-	return minBits;
-}
-
-unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus)
-{
-	U32 maxBitsSrc = BIT_highbit32((U32)(srcSize - 1)) - minus;
-	U32 tableLog = maxTableLog;
-	U32 minBits = FSE_minTableLog(srcSize, maxSymbolValue);
-	if (tableLog == 0)
-		tableLog = FSE_DEFAULT_TABLELOG;
-	if (maxBitsSrc < tableLog)
-		tableLog = maxBitsSrc; /* Accuracy can be reduced */
-	if (minBits > tableLog)
-		tableLog = minBits; /* Need a minimum to safely represent all symbol values */
-	if (tableLog < FSE_MIN_TABLELOG)
-		tableLog = FSE_MIN_TABLELOG;
-	if (tableLog > FSE_MAX_TABLELOG)
-		tableLog = FSE_MAX_TABLELOG;
-	return tableLog;
-}
-
-unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
-{
-	return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 2);
-}
-
-/* Secondary normalization method.
-   To be used when primary method fails. */
-
-static size_t FSE_normalizeM2(short *norm, U32 tableLog, const unsigned *count, size_t total, U32 maxSymbolValue)
-{
-	short const NOT_YET_ASSIGNED = -2;
-	U32 s;
-	U32 distributed = 0;
-	U32 ToDistribute;
-
-	/* Init */
-	U32 const lowThreshold = (U32)(total >> tableLog);
-	U32 lowOne = (U32)((total * 3) >> (tableLog + 1));
-
-	for (s = 0; s <= maxSymbolValue; s++) {
-		if (count[s] == 0) {
-			norm[s] = 0;
-			continue;
-		}
-		if (count[s] <= lowThreshold) {
-			norm[s] = -1;
-			distributed++;
-			total -= count[s];
-			continue;
-		}
-		if (count[s] <= lowOne) {
-			norm[s] = 1;
-			distributed++;
-			total -= count[s];
-			continue;
-		}
-
-		norm[s] = NOT_YET_ASSIGNED;
-	}
-	ToDistribute = (1 << tableLog) - distributed;
-
-	if ((total / ToDistribute) > lowOne) {
-		/* risk of rounding to zero */
-		lowOne = (U32)((total * 3) / (ToDistribute * 2));
-		for (s = 0; s <= maxSymbolValue; s++) {
-			if ((norm[s] == NOT_YET_ASSIGNED) && (count[s] <= lowOne)) {
-				norm[s] = 1;
-				distributed++;
-				total -= count[s];
-				continue;
-			}
-		}
-		ToDistribute = (1 << tableLog) - distributed;
-	}
-
-	if (distributed == maxSymbolValue + 1) {
-		/* all values are pretty poor;
-		   probably incompressible data (should have already been detected);
-		   find max, then give all remaining points to max */
-		U32 maxV = 0, maxC = 0;
-		for (s = 0; s <= maxSymbolValue; s++)
-			if (count[s] > maxC)
-				maxV = s, maxC = count[s];
-		norm[maxV] += (short)ToDistribute;
-		return 0;
-	}
-
-	if (total == 0) {
-		/* all of the symbols were low enough for the lowOne or lowThreshold */
-		for (s = 0; ToDistribute > 0; s = (s + 1) % (maxSymbolValue + 1))
-			if (norm[s] > 0)
-				ToDistribute--, norm[s]++;
-		return 0;
-	}
-
-	{
-		U64 const vStepLog = 62 - tableLog;
-		U64 const mid = (1ULL << (vStepLog - 1)) - 1;
-		U64 const rStep = div_u64((((U64)1 << vStepLog) * ToDistribute) + mid, (U32)total); /* scale on remaining */
-		U64 tmpTotal = mid;
-		for (s = 0; s <= maxSymbolValue; s++) {
-			if (norm[s] == NOT_YET_ASSIGNED) {
-				U64 const end = tmpTotal + (count[s] * rStep);
-				U32 const sStart = (U32)(tmpTotal >> vStepLog);
-				U32 const sEnd = (U32)(end >> vStepLog);
-				U32 const weight = sEnd - sStart;
-				if (weight < 1)
-					return ERROR(GENERIC);
-				norm[s] = (short)weight;
-				tmpTotal = end;
-			}
-		}
-	}
-
-	return 0;
-}
-
-size_t FSE_normalizeCount(short *normalizedCounter, unsigned tableLog, const unsigned *count, size_t total, unsigned maxSymbolValue)
-{
-	/* Sanity checks */
-	if (tableLog == 0)
-		tableLog = FSE_DEFAULT_TABLELOG;
-	if (tableLog < FSE_MIN_TABLELOG)
-		return ERROR(GENERIC); /* Unsupported size */
-	if (tableLog > FSE_MAX_TABLELOG)
-		return ERROR(tableLog_tooLarge); /* Unsupported size */
-	if (tableLog < FSE_minTableLog(total, maxSymbolValue))
-		return ERROR(GENERIC); /* Too small tableLog, compression potentially impossible */
-
-	{
-		U32 const rtbTable[] = {0, 473195, 504333, 520860, 550000, 700000, 750000, 830000};
-		U64 const scale = 62 - tableLog;
-		U64 const step = div_u64((U64)1 << 62, (U32)total); /* <== here, one division ! */
-		U64 const vStep = 1ULL << (scale - 20);
-		int stillToDistribute = 1 << tableLog;
-		unsigned s;
-		unsigned largest = 0;
-		short largestP = 0;
-		U32 lowThreshold = (U32)(total >> tableLog);
-
-		for (s = 0; s <= maxSymbolValue; s++) {
-			if (count[s] == total)
-				return 0; /* rle special case */
-			if (count[s] == 0) {
-				normalizedCounter[s] = 0;
-				continue;
-			}
-			if (count[s] <= lowThreshold) {
-				normalizedCounter[s] = -1;
-				stillToDistribute--;
-			} else {
-				short proba = (short)((count[s] * step) >> scale);
-				if (proba < 8) {
-					U64 restToBeat = vStep * rtbTable[proba];
-					proba += (count[s] * step) - ((U64)proba << scale) > restToBeat;
-				}
-				if (proba > largestP)
-					largestP = proba, largest = s;
-				normalizedCounter[s] = proba;
-				stillToDistribute -= proba;
-			}
-		}
-		if (-stillToDistribute >= (normalizedCounter[largest] >> 1)) {
-			/* corner case, need another normalization method */
-			size_t const errorCode = FSE_normalizeM2(normalizedCounter, tableLog, count, total, maxSymbolValue);
-			if (FSE_isError(errorCode))
-				return errorCode;
-		} else
-			normalizedCounter[largest] += (short)stillToDistribute;
-	}
-
-	return tableLog;
-}
-
-/* fake FSE_CTable, for raw (uncompressed) input */
-size_t FSE_buildCTable_raw(FSE_CTable *ct, unsigned nbBits)
-{
-	const unsigned tableSize = 1 << nbBits;
-	const unsigned tableMask = tableSize - 1;
-	const unsigned maxSymbolValue = tableMask;
-	void *const ptr = ct;
-	U16 *const tableU16 = ((U16 *)ptr) + 2;
-	void *const FSCT = ((U32 *)ptr) + 1 /* header */ + (tableSize >> 1); /* assumption : tableLog >= 1 */
-	FSE_symbolCompressionTransform *const symbolTT = (FSE_symbolCompressionTransform *)(FSCT);
-	unsigned s;
-
-	/* Sanity checks */
-	if (nbBits < 1)
-		return ERROR(GENERIC); /* min size */
-
-	/* header */
-	tableU16[-2] = (U16)nbBits;
-	tableU16[-1] = (U16)maxSymbolValue;
-
-	/* Build table */
-	for (s = 0; s < tableSize; s++)
-		tableU16[s] = (U16)(tableSize + s);
-
-	/* Build Symbol Transformation Table */
-	{
-		const U32 deltaNbBits = (nbBits << 16) - (1 << nbBits);
-		for (s = 0; s <= maxSymbolValue; s++) {
-			symbolTT[s].deltaNbBits = deltaNbBits;
-			symbolTT[s].deltaFindState = s - 1;
-		}
-	}
-
-	return 0;
-}
-
-/* fake FSE_CTable, for rle input (always same symbol) */
-size_t FSE_buildCTable_rle(FSE_CTable *ct, BYTE symbolValue)
-{
-	void *ptr = ct;
-	U16 *tableU16 = ((U16 *)ptr) + 2;
-	void *FSCTptr = (U32 *)ptr + 2;
-	FSE_symbolCompressionTransform *symbolTT = (FSE_symbolCompressionTransform *)FSCTptr;
-
-	/* header */
-	tableU16[-2] = (U16)0;
-	tableU16[-1] = (U16)symbolValue;
-
-	/* Build table */
-	tableU16[0] = 0;
-	tableU16[1] = 0; /* just in case */
-
-	/* Build Symbol Transformation Table */
-	symbolTT[symbolValue].deltaNbBits = 0;
-	symbolTT[symbolValue].deltaFindState = 0;
-
-	return 0;
-}
-
-static size_t FSE_compress_usingCTable_generic(void *dst, size_t dstSize, const void *src, size_t srcSize, const FSE_CTable *ct, const unsigned fast)
-{
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *ip = iend;
-
-	BIT_CStream_t bitC;
-	FSE_CState_t CState1, CState2;
-
-	/* init */
-	if (srcSize <= 2)
-		return 0;
-	{
-		size_t const initError = BIT_initCStream(&bitC, dst, dstSize);
-		if (FSE_isError(initError))
-			return 0; /* not enough space available to write a bitstream */
-	}
-
-#define FSE_FLUSHBITS(s) (fast ? BIT_flushBitsFast(s) : BIT_flushBits(s))
-
-	if (srcSize & 1) {
-		FSE_initCState2(&CState1, ct, *--ip);
-		FSE_initCState2(&CState2, ct, *--ip);
-		FSE_encodeSymbol(&bitC, &CState1, *--ip);
-		FSE_FLUSHBITS(&bitC);
-	} else {
-		FSE_initCState2(&CState2, ct, *--ip);
-		FSE_initCState2(&CState1, ct, *--ip);
-	}
-
-	/* join to mod 4 */
-	srcSize -= 2;
-	if ((sizeof(bitC.bitContainer) * 8 > FSE_MAX_TABLELOG * 4 + 7) && (srcSize & 2)) { /* test bit 2 */
-		FSE_encodeSymbol(&bitC, &CState2, *--ip);
-		FSE_encodeSymbol(&bitC, &CState1, *--ip);
-		FSE_FLUSHBITS(&bitC);
-	}
-
-	/* 2 or 4 encoding per loop */
-	while (ip > istart) {
-
-		FSE_encodeSymbol(&bitC, &CState2, *--ip);
-
-		if (sizeof(bitC.bitContainer) * 8 < FSE_MAX_TABLELOG * 2 + 7) /* this test must be static */
-			FSE_FLUSHBITS(&bitC);
-
-		FSE_encodeSymbol(&bitC, &CState1, *--ip);
-
-		if (sizeof(bitC.bitContainer) * 8 > FSE_MAX_TABLELOG * 4 + 7) { /* this test must be static */
-			FSE_encodeSymbol(&bitC, &CState2, *--ip);
-			FSE_encodeSymbol(&bitC, &CState1, *--ip);
-		}
-
-		FSE_FLUSHBITS(&bitC);
-	}
-
-	FSE_flushCState(&bitC, &CState2);
-	FSE_flushCState(&bitC, &CState1);
-	return BIT_closeCStream(&bitC);
-}
-
-size_t FSE_compress_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const FSE_CTable *ct)
-{
-	unsigned const fast = (dstSize >= FSE_BLOCKBOUND(srcSize));
-
-	if (fast)
-		return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 1);
-	else
-		return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 0);
-}
-
-size_t FSE_compressBound(size_t size) { return FSE_COMPRESSBOUND(size); }
diff --git a/lib/zstd/fse_decompress.c b/lib/zstd/fse_decompress.c
deleted file mode 100644
index 0b35353..0000000
--- a/lib/zstd/fse_decompress.c
+++ /dev/null
@@ -1,325 +0,0 @@
-/*
- * FSE : Finite State Entropy decoder
- * Copyright (C) 2013-2015, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-
-/* **************************************************************
-*  Compiler specifics
-****************************************************************/
-#define FORCE_INLINE static __always_inline
-
-/* **************************************************************
-*  Includes
-****************************************************************/
-#include "bitstream.h"
-#include "fse.h"
-#include "zstd_internal.h"
-#include <linux/compiler.h>
-#include <linux/kernel.h>
-#include <linux/string.h> /* memcpy, memset */
-
-/* **************************************************************
-*  Error Management
-****************************************************************/
-#define FSE_isError ERR_isError
-#define FSE_STATIC_ASSERT(c)                                   \
-	{                                                      \
-		enum { FSE_static_assert = 1 / (int)(!!(c)) }; \
-	} /* use only *after* variable declarations */
-
-/* **************************************************************
-*  Templates
-****************************************************************/
-/*
-  designed to be included
-  for type-specific functions (template emulation in C)
-  Objective is to write these functions only once, for improved maintenance
-*/
-
-/* safety checks */
-#ifndef FSE_FUNCTION_EXTENSION
-#error "FSE_FUNCTION_EXTENSION must be defined"
-#endif
-#ifndef FSE_FUNCTION_TYPE
-#error "FSE_FUNCTION_TYPE must be defined"
-#endif
-
-/* Function names */
-#define FSE_CAT(X, Y) X##Y
-#define FSE_FUNCTION_NAME(X, Y) FSE_CAT(X, Y)
-#define FSE_TYPE_NAME(X, Y) FSE_CAT(X, Y)
-
-/* Function templates */
-
-size_t FSE_buildDTable_wksp(FSE_DTable *dt, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workspace, size_t workspaceSize)
-{
-	void *const tdPtr = dt + 1; /* because *dt is unsigned, 32-bits aligned on 32-bits */
-	FSE_DECODE_TYPE *const tableDecode = (FSE_DECODE_TYPE *)(tdPtr);
-	U16 *symbolNext = (U16 *)workspace;
-
-	U32 const maxSV1 = maxSymbolValue + 1;
-	U32 const tableSize = 1 << tableLog;
-	U32 highThreshold = tableSize - 1;
-
-	/* Sanity Checks */
-	if (workspaceSize < sizeof(U16) * (FSE_MAX_SYMBOL_VALUE + 1))
-		return ERROR(tableLog_tooLarge);
-	if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE)
-		return ERROR(maxSymbolValue_tooLarge);
-	if (tableLog > FSE_MAX_TABLELOG)
-		return ERROR(tableLog_tooLarge);
-
-	/* Init, lay down lowprob symbols */
-	{
-		FSE_DTableHeader DTableH;
-		DTableH.tableLog = (U16)tableLog;
-		DTableH.fastMode = 1;
-		{
-			S16 const largeLimit = (S16)(1 << (tableLog - 1));
-			U32 s;
-			for (s = 0; s < maxSV1; s++) {
-				if (normalizedCounter[s] == -1) {
-					tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s;
-					symbolNext[s] = 1;
-				} else {
-					if (normalizedCounter[s] >= largeLimit)
-						DTableH.fastMode = 0;
-					symbolNext[s] = normalizedCounter[s];
-				}
-			}
-		}
-		memcpy(dt, &DTableH, sizeof(DTableH));
-	}
-
-	/* Spread symbols */
-	{
-		U32 const tableMask = tableSize - 1;
-		U32 const step = FSE_TABLESTEP(tableSize);
-		U32 s, position = 0;
-		for (s = 0; s < maxSV1; s++) {
-			int i;
-			for (i = 0; i < normalizedCounter[s]; i++) {
-				tableDecode[position].symbol = (FSE_FUNCTION_TYPE)s;
-				position = (position + step) & tableMask;
-				while (position > highThreshold)
-					position = (position + step) & tableMask; /* lowprob area */
-			}
-		}
-		if (position != 0)
-			return ERROR(GENERIC); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
-	}
-
-	/* Build Decoding table */
-	{
-		U32 u;
-		for (u = 0; u < tableSize; u++) {
-			FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
-			U16 nextState = symbolNext[symbol]++;
-			tableDecode[u].nbBits = (BYTE)(tableLog - BIT_highbit32((U32)nextState));
-			tableDecode[u].newState = (U16)((nextState << tableDecode[u].nbBits) - tableSize);
-		}
-	}
-
-	return 0;
-}
-
-/*-*******************************************************
-*  Decompression (Byte symbols)
-*********************************************************/
-size_t FSE_buildDTable_rle(FSE_DTable *dt, BYTE symbolValue)
-{
-	void *ptr = dt;
-	FSE_DTableHeader *const DTableH = (FSE_DTableHeader *)ptr;
-	void *dPtr = dt + 1;
-	FSE_decode_t *const cell = (FSE_decode_t *)dPtr;
-
-	DTableH->tableLog = 0;
-	DTableH->fastMode = 0;
-
-	cell->newState = 0;
-	cell->symbol = symbolValue;
-	cell->nbBits = 0;
-
-	return 0;
-}
-
-size_t FSE_buildDTable_raw(FSE_DTable *dt, unsigned nbBits)
-{
-	void *ptr = dt;
-	FSE_DTableHeader *const DTableH = (FSE_DTableHeader *)ptr;
-	void *dPtr = dt + 1;
-	FSE_decode_t *const dinfo = (FSE_decode_t *)dPtr;
-	const unsigned tableSize = 1 << nbBits;
-	const unsigned tableMask = tableSize - 1;
-	const unsigned maxSV1 = tableMask + 1;
-	unsigned s;
-
-	/* Sanity checks */
-	if (nbBits < 1)
-		return ERROR(GENERIC); /* min size */
-
-	/* Build Decoding Table */
-	DTableH->tableLog = (U16)nbBits;
-	DTableH->fastMode = 1;
-	for (s = 0; s < maxSV1; s++) {
-		dinfo[s].newState = 0;
-		dinfo[s].symbol = (BYTE)s;
-		dinfo[s].nbBits = (BYTE)nbBits;
-	}
-
-	return 0;
-}
-
-FORCE_INLINE size_t FSE_decompress_usingDTable_generic(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt,
-						       const unsigned fast)
-{
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *op = ostart;
-	BYTE *const omax = op + maxDstSize;
-	BYTE *const olimit = omax - 3;
-
-	BIT_DStream_t bitD;
-	FSE_DState_t state1;
-	FSE_DState_t state2;
-
-	/* Init */
-	CHECK_F(BIT_initDStream(&bitD, cSrc, cSrcSize));
-
-	FSE_initDState(&state1, &bitD, dt);
-	FSE_initDState(&state2, &bitD, dt);
-
-#define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD)
-
-	/* 4 symbols per loop */
-	for (; (BIT_reloadDStream(&bitD) == BIT_DStream_unfinished) & (op < olimit); op += 4) {
-		op[0] = FSE_GETSYMBOL(&state1);
-
-		if (FSE_MAX_TABLELOG * 2 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
-			BIT_reloadDStream(&bitD);
-
-		op[1] = FSE_GETSYMBOL(&state2);
-
-		if (FSE_MAX_TABLELOG * 4 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
-		{
-			if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) {
-				op += 2;
-				break;
-			}
-		}
-
-		op[2] = FSE_GETSYMBOL(&state1);
-
-		if (FSE_MAX_TABLELOG * 2 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
-			BIT_reloadDStream(&bitD);
-
-		op[3] = FSE_GETSYMBOL(&state2);
-	}
-
-	/* tail */
-	/* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */
-	while (1) {
-		if (op > (omax - 2))
-			return ERROR(dstSize_tooSmall);
-		*op++ = FSE_GETSYMBOL(&state1);
-		if (BIT_reloadDStream(&bitD) == BIT_DStream_overflow) {
-			*op++ = FSE_GETSYMBOL(&state2);
-			break;
-		}
-
-		if (op > (omax - 2))
-			return ERROR(dstSize_tooSmall);
-		*op++ = FSE_GETSYMBOL(&state2);
-		if (BIT_reloadDStream(&bitD) == BIT_DStream_overflow) {
-			*op++ = FSE_GETSYMBOL(&state1);
-			break;
-		}
-	}
-
-	return op - ostart;
-}
-
-size_t FSE_decompress_usingDTable(void *dst, size_t originalSize, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt)
-{
-	const void *ptr = dt;
-	const FSE_DTableHeader *DTableH = (const FSE_DTableHeader *)ptr;
-	const U32 fastMode = DTableH->fastMode;
-
-	/* select fast mode (static) */
-	if (fastMode)
-		return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1);
-	return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0);
-}
-
-size_t FSE_decompress_wksp(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, unsigned maxLog, void *workspace, size_t workspaceSize)
-{
-	const BYTE *const istart = (const BYTE *)cSrc;
-	const BYTE *ip = istart;
-	unsigned tableLog;
-	unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
-	size_t NCountLength;
-
-	FSE_DTable *dt;
-	short *counting;
-	size_t spaceUsed32 = 0;
-
-	FSE_STATIC_ASSERT(sizeof(FSE_DTable) == sizeof(U32));
-
-	dt = (FSE_DTable *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += FSE_DTABLE_SIZE_U32(maxLog);
-	counting = (short *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(sizeof(short) * (FSE_MAX_SYMBOL_VALUE + 1), sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	/* normal FSE decoding mode */
-	NCountLength = FSE_readNCount(counting, &maxSymbolValue, &tableLog, istart, cSrcSize);
-	if (FSE_isError(NCountLength))
-		return NCountLength;
-	// if (NCountLength >= cSrcSize) return ERROR(srcSize_wrong);   /* too small input size; supposed to be already checked in NCountLength, only remaining
-	// case : NCountLength==cSrcSize */
-	if (tableLog > maxLog)
-		return ERROR(tableLog_tooLarge);
-	ip += NCountLength;
-	cSrcSize -= NCountLength;
-
-	CHECK_F(FSE_buildDTable_wksp(dt, counting, maxSymbolValue, tableLog, workspace, workspaceSize));
-
-	return FSE_decompress_usingDTable(dst, dstCapacity, ip, cSrcSize, dt); /* always return, even if it is an error code */
-}
diff --git a/lib/zstd/huf.h b/lib/zstd/huf.h
deleted file mode 100644
index 923218d..0000000
--- a/lib/zstd/huf.h
+++ /dev/null
@@ -1,212 +0,0 @@
-/*
- * Huffman coder, part of New Generation Entropy library
- * header file
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-#ifndef HUF_H_298734234
-#define HUF_H_298734234
-
-/* *** Dependencies *** */
-#include <linux/types.h> /* size_t */
-
-/* ***   Tool functions *** */
-#define HUF_BLOCKSIZE_MAX (128 * 1024) /**< maximum input size for a single block compressed with HUF_compress */
-size_t HUF_compressBound(size_t size); /**< maximum compressed size (worst case) */
-
-/* Error Management */
-unsigned HUF_isError(size_t code); /**< tells if a return value is an error code */
-
-/* ***   Advanced function   *** */
-
-/** HUF_compress4X_wksp() :
-*   Same as HUF_compress2(), but uses externally allocated `workSpace`, which must be a table of >= 1024 unsigned */
-size_t HUF_compress4X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
-			   size_t wkspSize); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
-
-/* *** Dependencies *** */
-#include "mem.h" /* U32 */
-
-/* *** Constants *** */
-#define HUF_TABLELOG_MAX 12     /* max configured tableLog (for static allocation); can be modified up to HUF_ABSOLUTEMAX_TABLELOG */
-#define HUF_TABLELOG_DEFAULT 11 /* tableLog by default, when not specified */
-#define HUF_SYMBOLVALUE_MAX 255
-
-#define HUF_TABLELOG_ABSOLUTEMAX 15 /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */
-#if (HUF_TABLELOG_MAX > HUF_TABLELOG_ABSOLUTEMAX)
-#error "HUF_TABLELOG_MAX is too large !"
-#endif
-
-/* ****************************************
-*  Static allocation
-******************************************/
-/* HUF buffer bounds */
-#define HUF_CTABLEBOUND 129
-#define HUF_BLOCKBOUND(size) (size + (size >> 8) + 8)			 /* only true if incompressible pre-filtered with fast heuristic */
-#define HUF_COMPRESSBOUND(size) (HUF_CTABLEBOUND + HUF_BLOCKBOUND(size)) /* Macro version, useful for static allocation */
-
-/* static allocation of HUF's Compression Table */
-#define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \
-	U32 name##hb[maxSymbolValue + 1];              \
-	void *name##hv = &(name##hb);                  \
-	HUF_CElt *name = (HUF_CElt *)(name##hv) /* no final ; */
-
-/* static allocation of HUF's DTable */
-typedef U32 HUF_DTable;
-#define HUF_DTABLE_SIZE(maxTableLog) (1 + (1 << (maxTableLog)))
-#define HUF_CREATE_STATIC_DTABLEX2(DTable, maxTableLog) HUF_DTable DTable[HUF_DTABLE_SIZE((maxTableLog)-1)] = {((U32)((maxTableLog)-1) * 0x01000001)}
-#define HUF_CREATE_STATIC_DTABLEX4(DTable, maxTableLog) HUF_DTable DTable[HUF_DTABLE_SIZE(maxTableLog)] = {((U32)(maxTableLog)*0x01000001)}
-
-/* The workspace must have alignment at least 4 and be at least this large */
-#define HUF_COMPRESS_WORKSPACE_SIZE (6 << 10)
-#define HUF_COMPRESS_WORKSPACE_SIZE_U32 (HUF_COMPRESS_WORKSPACE_SIZE / sizeof(U32))
-
-/* The workspace must have alignment at least 4 and be at least this large */
-#define HUF_DECOMPRESS_WORKSPACE_SIZE (3 << 10)
-#define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))
-
-/* ****************************************
-*  Advanced decompression functions
-******************************************/
-size_t HUF_decompress4X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize); /**< decodes RLE and uncompressed */
-size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				size_t workspaceSize);							       /**< considers RLE and uncompressed as errors */
-size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				   size_t workspaceSize); /**< single-symbol decoder */
-size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				   size_t workspaceSize); /**< double-symbols decoder */
-
-/* ****************************************
-*  HUF detailed API
-******************************************/
-/*!
-HUF_compress() does the following:
-1. count symbol occurrence from source[] into table count[] using FSE_count()
-2. (optional) refine tableLog using HUF_optimalTableLog()
-3. build Huffman table from count using HUF_buildCTable()
-4. save Huffman table to memory buffer using HUF_writeCTable_wksp()
-5. encode the data stream using HUF_compress4X_usingCTable()
-
-The following API allows targeting specific sub-functions for advanced tasks.
-For example, it's possible to compress several blocks using the same 'CTable',
-or to save and regenerate 'CTable' using external methods.
-*/
-/* FSE_count() : find it within "fse.h" */
-unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
-typedef struct HUF_CElt_s HUF_CElt; /* incomplete type */
-size_t HUF_writeCTable_wksp(void *dst, size_t maxDstSize, const HUF_CElt *CTable, unsigned maxSymbolValue, unsigned huffLog, void *workspace, size_t workspaceSize);
-size_t HUF_compress4X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable);
-
-typedef enum {
-	HUF_repeat_none,  /**< Cannot use the previous table */
-	HUF_repeat_check, /**< Can use the previous table but it must be checked. Note : The previous table must have been constructed by HUF_compress{1,
-			     4}X_repeat */
-	HUF_repeat_valid  /**< Can use the previous table and it is assumed to be valid */
-} HUF_repeat;
-/** HUF_compress4X_repeat() :
-*   Same as HUF_compress4X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
-*   If it uses hufTable it does not modify hufTable or repeat.
-*   If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
-*   If preferRepeat then the old table will always be used if valid. */
-size_t HUF_compress4X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
-			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat,
-			     int preferRepeat); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
-
-/** HUF_buildCTable_wksp() :
- *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
- *  `workSpace` must be aligned on 4-bytes boundaries, and be at least as large as a table of 1024 unsigned.
- */
-size_t HUF_buildCTable_wksp(HUF_CElt *tree, const U32 *count, U32 maxSymbolValue, U32 maxNbBits, void *workSpace, size_t wkspSize);
-
-/*! HUF_readStats() :
-	Read compact Huffman tree, saved by HUF_writeCTable().
-	`huffWeight` is destination buffer.
-	@return : size read from `src` , or an error Code .
-	Note : Needed by HUF_readCTable() and HUF_readDTableXn() . */
-size_t HUF_readStats_wksp(BYTE *huffWeight, size_t hwSize, U32 *rankStats, U32 *nbSymbolsPtr, U32 *tableLogPtr, const void *src, size_t srcSize,
-			  void *workspace, size_t workspaceSize);
-
-/** HUF_readCTable() :
-*   Loading a CTable saved with HUF_writeCTable() */
-size_t HUF_readCTable_wksp(HUF_CElt *CTable, unsigned maxSymbolValue, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
-
-/*
-HUF_decompress() does the following:
-1. select the decompression algorithm (X2, X4) based on pre-computed heuristics
-2. build Huffman table from save, using HUF_readDTableXn()
-3. decode 1 or 4 segments in parallel using HUF_decompressSXn_usingDTable
-*/
-
-/** HUF_selectDecoder() :
-*   Tells which decoder is likely to decode faster,
-*   based on a set of pre-determined metrics.
-*   @return : 0==HUF_decompress4X2, 1==HUF_decompress4X4 .
-*   Assumption : 0 < cSrcSize < dstSize <= 128 KB */
-U32 HUF_selectDecoder(size_t dstSize, size_t cSrcSize);
-
-size_t HUF_readDTableX2_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
-size_t HUF_readDTableX4_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
-
-size_t HUF_decompress4X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-size_t HUF_decompress4X2_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-size_t HUF_decompress4X4_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-
-/* single stream variants */
-
-size_t HUF_compress1X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
-			   size_t wkspSize); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
-size_t HUF_compress1X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable);
-/** HUF_compress1X_repeat() :
-*   Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
-*   If it uses hufTable it does not modify hufTable or repeat.
-*   If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
-*   If preferRepeat then the old table will always be used if valid. */
-size_t HUF_compress1X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
-			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat,
-			     int preferRepeat); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
-
-size_t HUF_decompress1X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize);
-size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				   size_t workspaceSize); /**< single-symbol decoder */
-size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
-				   size_t workspaceSize); /**< double-symbols decoder */
-
-size_t HUF_decompress1X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize,
-				    const HUF_DTable *DTable); /**< automatic selection of sing or double symbol decoder, based on DTable */
-size_t HUF_decompress1X2_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-size_t HUF_decompress1X4_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
-
-#endif /* HUF_H_298734234 */
diff --git a/lib/zstd/huf_compress.c b/lib/zstd/huf_compress.c
deleted file mode 100644
index fd32838..0000000
--- a/lib/zstd/huf_compress.c
+++ /dev/null
@@ -1,773 +0,0 @@
-/*
- * Huffman encoder, part of New Generation Entropy library
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-
-/* **************************************************************
-*  Includes
-****************************************************************/
-#include "bitstream.h"
-#include "fse.h" /* header compression */
-#include "huf.h"
-#include <linux/kernel.h>
-#include <linux/string.h> /* memcpy, memset */
-
-/* **************************************************************
-*  Error Management
-****************************************************************/
-#define HUF_STATIC_ASSERT(c)                                   \
-	{                                                      \
-		enum { HUF_static_assert = 1 / (int)(!!(c)) }; \
-	} /* use only *after* variable declarations */
-#define CHECK_V_F(e, f)     \
-	size_t const e = f; \
-	if (ERR_isError(e)) \
-	return f
-#define CHECK_F(f)                        \
-	{                                 \
-		CHECK_V_F(_var_err__, f); \
-	}
-
-/* **************************************************************
-*  Utils
-****************************************************************/
-unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
-{
-	return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 1);
-}
-
-/* *******************************************************
-*  HUF : Huffman block compression
-*********************************************************/
-/* HUF_compressWeights() :
- * Same as FSE_compress(), but dedicated to huff0's weights compression.
- * The use case needs much less stack memory.
- * Note : all elements within weightTable are supposed to be <= HUF_TABLELOG_MAX.
- */
-#define MAX_FSE_TABLELOG_FOR_HUFF_HEADER 6
-size_t HUF_compressWeights_wksp(void *dst, size_t dstSize, const void *weightTable, size_t wtSize, void *workspace, size_t workspaceSize)
-{
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *op = ostart;
-	BYTE *const oend = ostart + dstSize;
-
-	U32 maxSymbolValue = HUF_TABLELOG_MAX;
-	U32 tableLog = MAX_FSE_TABLELOG_FOR_HUFF_HEADER;
-
-	FSE_CTable *CTable;
-	U32 *count;
-	S16 *norm;
-	size_t spaceUsed32 = 0;
-
-	HUF_STATIC_ASSERT(sizeof(FSE_CTable) == sizeof(U32));
-
-	CTable = (FSE_CTable *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += FSE_CTABLE_SIZE_U32(MAX_FSE_TABLELOG_FOR_HUFF_HEADER, HUF_TABLELOG_MAX);
-	count = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += HUF_TABLELOG_MAX + 1;
-	norm = (S16 *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(sizeof(S16) * (HUF_TABLELOG_MAX + 1), sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	/* init conditions */
-	if (wtSize <= 1)
-		return 0; /* Not compressible */
-
-	/* Scan input and build symbol stats */
-	{
-		CHECK_V_F(maxCount, FSE_count_simple(count, &maxSymbolValue, weightTable, wtSize));
-		if (maxCount == wtSize)
-			return 1; /* only a single symbol in src : rle */
-		if (maxCount == 1)
-			return 0; /* each symbol present maximum once => not compressible */
-	}
-
-	tableLog = FSE_optimalTableLog(tableLog, wtSize, maxSymbolValue);
-	CHECK_F(FSE_normalizeCount(norm, tableLog, count, wtSize, maxSymbolValue));
-
-	/* Write table description header */
-	{
-		CHECK_V_F(hSize, FSE_writeNCount(op, oend - op, norm, maxSymbolValue, tableLog));
-		op += hSize;
-	}
-
-	/* Compress */
-	CHECK_F(FSE_buildCTable_wksp(CTable, norm, maxSymbolValue, tableLog, workspace, workspaceSize));
-	{
-		CHECK_V_F(cSize, FSE_compress_usingCTable(op, oend - op, weightTable, wtSize, CTable));
-		if (cSize == 0)
-			return 0; /* not enough space for compressed data */
-		op += cSize;
-	}
-
-	return op - ostart;
-}
-
-struct HUF_CElt_s {
-	U16 val;
-	BYTE nbBits;
-}; /* typedef'd to HUF_CElt within "huf.h" */
-
-/*! HUF_writeCTable_wksp() :
-	`CTable` : Huffman tree to save, using huf representation.
-	@return : size of saved CTable */
-size_t HUF_writeCTable_wksp(void *dst, size_t maxDstSize, const HUF_CElt *CTable, U32 maxSymbolValue, U32 huffLog, void *workspace, size_t workspaceSize)
-{
-	BYTE *op = (BYTE *)dst;
-	U32 n;
-
-	BYTE *bitsToWeight;
-	BYTE *huffWeight;
-	size_t spaceUsed32 = 0;
-
-	bitsToWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(HUF_TABLELOG_MAX + 1, sizeof(U32)) >> 2;
-	huffWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX, sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	/* check conditions */
-	if (maxSymbolValue > HUF_SYMBOLVALUE_MAX)
-		return ERROR(maxSymbolValue_tooLarge);
-
-	/* convert to weight */
-	bitsToWeight[0] = 0;
-	for (n = 1; n < huffLog + 1; n++)
-		bitsToWeight[n] = (BYTE)(huffLog + 1 - n);
-	for (n = 0; n < maxSymbolValue; n++)
-		huffWeight[n] = bitsToWeight[CTable[n].nbBits];
-
-	/* attempt weights compression by FSE */
-	{
-		CHECK_V_F(hSize, HUF_compressWeights_wksp(op + 1, maxDstSize - 1, huffWeight, maxSymbolValue, workspace, workspaceSize));
-		if ((hSize > 1) & (hSize < maxSymbolValue / 2)) { /* FSE compressed */
-			op[0] = (BYTE)hSize;
-			return hSize + 1;
-		}
-	}
-
-	/* write raw values as 4-bits (max : 15) */
-	if (maxSymbolValue > (256 - 128))
-		return ERROR(GENERIC); /* should not happen : likely means source cannot be compressed */
-	if (((maxSymbolValue + 1) / 2) + 1 > maxDstSize)
-		return ERROR(dstSize_tooSmall); /* not enough space within dst buffer */
-	op[0] = (BYTE)(128 /*special case*/ + (maxSymbolValue - 1));
-	huffWeight[maxSymbolValue] = 0; /* to be sure it doesn't cause msan issue in final combination */
-	for (n = 0; n < maxSymbolValue; n += 2)
-		op[(n / 2) + 1] = (BYTE)((huffWeight[n] << 4) + huffWeight[n + 1]);
-	return ((maxSymbolValue + 1) / 2) + 1;
-}
-
-size_t HUF_readCTable_wksp(HUF_CElt *CTable, U32 maxSymbolValue, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
-{
-	U32 *rankVal;
-	BYTE *huffWeight;
-	U32 tableLog = 0;
-	U32 nbSymbols = 0;
-	size_t readSize;
-	size_t spaceUsed32 = 0;
-
-	rankVal = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += HUF_TABLELOG_ABSOLUTEMAX + 1;
-	huffWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	/* get symbol weights */
-	readSize = HUF_readStats_wksp(huffWeight, HUF_SYMBOLVALUE_MAX + 1, rankVal, &nbSymbols, &tableLog, src, srcSize, workspace, workspaceSize);
-	if (ERR_isError(readSize))
-		return readSize;
-
-	/* check result */
-	if (tableLog > HUF_TABLELOG_MAX)
-		return ERROR(tableLog_tooLarge);
-	if (nbSymbols > maxSymbolValue + 1)
-		return ERROR(maxSymbolValue_tooSmall);
-
-	/* Prepare base value per rank */
-	{
-		U32 n, nextRankStart = 0;
-		for (n = 1; n <= tableLog; n++) {
-			U32 curr = nextRankStart;
-			nextRankStart += (rankVal[n] << (n - 1));
-			rankVal[n] = curr;
-		}
-	}
-
-	/* fill nbBits */
-	{
-		U32 n;
-		for (n = 0; n < nbSymbols; n++) {
-			const U32 w = huffWeight[n];
-			CTable[n].nbBits = (BYTE)(tableLog + 1 - w);
-		}
-	}
-
-	/* fill val */
-	{
-		U16 nbPerRank[HUF_TABLELOG_MAX + 2] = {0}; /* support w=0=>n=tableLog+1 */
-		U16 valPerRank[HUF_TABLELOG_MAX + 2] = {0};
-		{
-			U32 n;
-			for (n = 0; n < nbSymbols; n++)
-				nbPerRank[CTable[n].nbBits]++;
-		}
-		/* determine stating value per rank */
-		valPerRank[tableLog + 1] = 0; /* for w==0 */
-		{
-			U16 min = 0;
-			U32 n;
-			for (n = tableLog; n > 0; n--) { /* start at n=tablelog <-> w=1 */
-				valPerRank[n] = min;     /* get starting value within each rank */
-				min += nbPerRank[n];
-				min >>= 1;
-			}
-		}
-		/* assign value within rank, symbol order */
-		{
-			U32 n;
-			for (n = 0; n <= maxSymbolValue; n++)
-				CTable[n].val = valPerRank[CTable[n].nbBits]++;
-		}
-	}
-
-	return readSize;
-}
-
-typedef struct nodeElt_s {
-	U32 count;
-	U16 parent;
-	BYTE byte;
-	BYTE nbBits;
-} nodeElt;
-
-static U32 HUF_setMaxHeight(nodeElt *huffNode, U32 lastNonNull, U32 maxNbBits)
-{
-	const U32 largestBits = huffNode[lastNonNull].nbBits;
-	if (largestBits <= maxNbBits)
-		return largestBits; /* early exit : no elt > maxNbBits */
-
-	/* there are several too large elements (at least >= 2) */
-	{
-		int totalCost = 0;
-		const U32 baseCost = 1 << (largestBits - maxNbBits);
-		U32 n = lastNonNull;
-
-		while (huffNode[n].nbBits > maxNbBits) {
-			totalCost += baseCost - (1 << (largestBits - huffNode[n].nbBits));
-			huffNode[n].nbBits = (BYTE)maxNbBits;
-			n--;
-		} /* n stops at huffNode[n].nbBits <= maxNbBits */
-		while (huffNode[n].nbBits == maxNbBits)
-			n--; /* n end at index of smallest symbol using < maxNbBits */
-
-		/* renorm totalCost */
-		totalCost >>= (largestBits - maxNbBits); /* note : totalCost is necessarily a multiple of baseCost */
-
-		/* repay normalized cost */
-		{
-			U32 const noSymbol = 0xF0F0F0F0;
-			U32 rankLast[HUF_TABLELOG_MAX + 2];
-			int pos;
-
-			/* Get pos of last (smallest) symbol per rank */
-			memset(rankLast, 0xF0, sizeof(rankLast));
-			{
-				U32 currNbBits = maxNbBits;
-				for (pos = n; pos >= 0; pos--) {
-					if (huffNode[pos].nbBits >= currNbBits)
-						continue;
-					currNbBits = huffNode[pos].nbBits; /* < maxNbBits */
-					rankLast[maxNbBits - currNbBits] = pos;
-				}
-			}
-
-			while (totalCost > 0) {
-				U32 nBitsToDecrease = BIT_highbit32(totalCost) + 1;
-				for (; nBitsToDecrease > 1; nBitsToDecrease--) {
-					U32 highPos = rankLast[nBitsToDecrease];
-					U32 lowPos = rankLast[nBitsToDecrease - 1];
-					if (highPos == noSymbol)
-						continue;
-					if (lowPos == noSymbol)
-						break;
-					{
-						U32 const highTotal = huffNode[highPos].count;
-						U32 const lowTotal = 2 * huffNode[lowPos].count;
-						if (highTotal <= lowTotal)
-							break;
-					}
-				}
-				/* only triggered when no more rank 1 symbol left => find closest one (note : there is necessarily at least one !) */
-				/* HUF_MAX_TABLELOG test just to please gcc 5+; but it should not be necessary */
-				while ((nBitsToDecrease <= HUF_TABLELOG_MAX) && (rankLast[nBitsToDecrease] == noSymbol))
-					nBitsToDecrease++;
-				totalCost -= 1 << (nBitsToDecrease - 1);
-				if (rankLast[nBitsToDecrease - 1] == noSymbol)
-					rankLast[nBitsToDecrease - 1] = rankLast[nBitsToDecrease]; /* this rank is no longer empty */
-				huffNode[rankLast[nBitsToDecrease]].nbBits++;
-				if (rankLast[nBitsToDecrease] == 0) /* special case, reached largest symbol */
-					rankLast[nBitsToDecrease] = noSymbol;
-				else {
-					rankLast[nBitsToDecrease]--;
-					if (huffNode[rankLast[nBitsToDecrease]].nbBits != maxNbBits - nBitsToDecrease)
-						rankLast[nBitsToDecrease] = noSymbol; /* this rank is now empty */
-				}
-			} /* while (totalCost > 0) */
-
-			while (totalCost < 0) {		       /* Sometimes, cost correction overshoot */
-				if (rankLast[1] == noSymbol) { /* special case : no rank 1 symbol (using maxNbBits-1); let's create one from largest rank 0
-								  (using maxNbBits) */
-					while (huffNode[n].nbBits == maxNbBits)
-						n--;
-					huffNode[n + 1].nbBits--;
-					rankLast[1] = n + 1;
-					totalCost++;
-					continue;
-				}
-				huffNode[rankLast[1] + 1].nbBits--;
-				rankLast[1]++;
-				totalCost++;
-			}
-		}
-	} /* there are several too large elements (at least >= 2) */
-
-	return maxNbBits;
-}
-
-typedef struct {
-	U32 base;
-	U32 curr;
-} rankPos;
-
-static void HUF_sort(nodeElt *huffNode, const U32 *count, U32 maxSymbolValue)
-{
-	rankPos rank[32];
-	U32 n;
-
-	memset(rank, 0, sizeof(rank));
-	for (n = 0; n <= maxSymbolValue; n++) {
-		U32 r = BIT_highbit32(count[n] + 1);
-		rank[r].base++;
-	}
-	for (n = 30; n > 0; n--)
-		rank[n - 1].base += rank[n].base;
-	for (n = 0; n < 32; n++)
-		rank[n].curr = rank[n].base;
-	for (n = 0; n <= maxSymbolValue; n++) {
-		U32 const c = count[n];
-		U32 const r = BIT_highbit32(c + 1) + 1;
-		U32 pos = rank[r].curr++;
-		while ((pos > rank[r].base) && (c > huffNode[pos - 1].count))
-			huffNode[pos] = huffNode[pos - 1], pos--;
-		huffNode[pos].count = c;
-		huffNode[pos].byte = (BYTE)n;
-	}
-}
-
-/** HUF_buildCTable_wksp() :
- *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
- *  `workSpace` must be aligned on 4-bytes boundaries, and be at least as large as a table of 1024 unsigned.
- */
-#define STARTNODE (HUF_SYMBOLVALUE_MAX + 1)
-typedef nodeElt huffNodeTable[2 * HUF_SYMBOLVALUE_MAX + 1 + 1];
-size_t HUF_buildCTable_wksp(HUF_CElt *tree, const U32 *count, U32 maxSymbolValue, U32 maxNbBits, void *workSpace, size_t wkspSize)
-{
-	nodeElt *const huffNode0 = (nodeElt *)workSpace;
-	nodeElt *const huffNode = huffNode0 + 1;
-	U32 n, nonNullRank;
-	int lowS, lowN;
-	U16 nodeNb = STARTNODE;
-	U32 nodeRoot;
-
-	/* safety checks */
-	if (wkspSize < sizeof(huffNodeTable))
-		return ERROR(GENERIC); /* workSpace is not large enough */
-	if (maxNbBits == 0)
-		maxNbBits = HUF_TABLELOG_DEFAULT;
-	if (maxSymbolValue > HUF_SYMBOLVALUE_MAX)
-		return ERROR(GENERIC);
-	memset(huffNode0, 0, sizeof(huffNodeTable));
-
-	/* sort, decreasing order */
-	HUF_sort(huffNode, count, maxSymbolValue);
-
-	/* init for parents */
-	nonNullRank = maxSymbolValue;
-	while (huffNode[nonNullRank].count == 0)
-		nonNullRank--;
-	lowS = nonNullRank;
-	nodeRoot = nodeNb + lowS - 1;
-	lowN = nodeNb;
-	huffNode[nodeNb].count = huffNode[lowS].count + huffNode[lowS - 1].count;
-	huffNode[lowS].parent = huffNode[lowS - 1].parent = nodeNb;
-	nodeNb++;
-	lowS -= 2;
-	for (n = nodeNb; n <= nodeRoot; n++)
-		huffNode[n].count = (U32)(1U << 30);
-	huffNode0[0].count = (U32)(1U << 31); /* fake entry, strong barrier */
-
-	/* create parents */
-	while (nodeNb <= nodeRoot) {
-		U32 n1 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++;
-		U32 n2 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++;
-		huffNode[nodeNb].count = huffNode[n1].count + huffNode[n2].count;
-		huffNode[n1].parent = huffNode[n2].parent = nodeNb;
-		nodeNb++;
-	}
-
-	/* distribute weights (unlimited tree height) */
-	huffNode[nodeRoot].nbBits = 0;
-	for (n = nodeRoot - 1; n >= STARTNODE; n--)
-		huffNode[n].nbBits = huffNode[huffNode[n].parent].nbBits + 1;
-	for (n = 0; n <= nonNullRank; n++)
-		huffNode[n].nbBits = huffNode[huffNode[n].parent].nbBits + 1;
-
-	/* enforce maxTableLog */
-	maxNbBits = HUF_setMaxHeight(huffNode, nonNullRank, maxNbBits);
-
-	/* fill result into tree (val, nbBits) */
-	{
-		U16 nbPerRank[HUF_TABLELOG_MAX + 1] = {0};
-		U16 valPerRank[HUF_TABLELOG_MAX + 1] = {0};
-		if (maxNbBits > HUF_TABLELOG_MAX)
-			return ERROR(GENERIC); /* check fit into table */
-		for (n = 0; n <= nonNullRank; n++)
-			nbPerRank[huffNode[n].nbBits]++;
-		/* determine stating value per rank */
-		{
-			U16 min = 0;
-			for (n = maxNbBits; n > 0; n--) {
-				valPerRank[n] = min; /* get starting value within each rank */
-				min += nbPerRank[n];
-				min >>= 1;
-			}
-		}
-		for (n = 0; n <= maxSymbolValue; n++)
-			tree[huffNode[n].byte].nbBits = huffNode[n].nbBits; /* push nbBits per symbol, symbol order */
-		for (n = 0; n <= maxSymbolValue; n++)
-			tree[n].val = valPerRank[tree[n].nbBits]++; /* assign value within rank, symbol order */
-	}
-
-	return maxNbBits;
-}
-
-static size_t HUF_estimateCompressedSize(HUF_CElt *CTable, const unsigned *count, unsigned maxSymbolValue)
-{
-	size_t nbBits = 0;
-	int s;
-	for (s = 0; s <= (int)maxSymbolValue; ++s) {
-		nbBits += CTable[s].nbBits * count[s];
-	}
-	return nbBits >> 3;
-}
-
-static int HUF_validateCTable(const HUF_CElt *CTable, const unsigned *count, unsigned maxSymbolValue)
-{
-	int bad = 0;
-	int s;
-	for (s = 0; s <= (int)maxSymbolValue; ++s) {
-		bad |= (count[s] != 0) & (CTable[s].nbBits == 0);
-	}
-	return !bad;
-}
-
-static void HUF_encodeSymbol(BIT_CStream_t *bitCPtr, U32 symbol, const HUF_CElt *CTable)
-{
-	BIT_addBitsFast(bitCPtr, CTable[symbol].val, CTable[symbol].nbBits);
-}
-
-size_t HUF_compressBound(size_t size) { return HUF_COMPRESSBOUND(size); }
-
-#define HUF_FLUSHBITS(s)  BIT_flushBits(s)
-
-#define HUF_FLUSHBITS_1(stream)                                            \
-	if (sizeof((stream)->bitContainer) * 8 < HUF_TABLELOG_MAX * 2 + 7) \
-	HUF_FLUSHBITS(stream)
-
-#define HUF_FLUSHBITS_2(stream)                                            \
-	if (sizeof((stream)->bitContainer) * 8 < HUF_TABLELOG_MAX * 4 + 7) \
-	HUF_FLUSHBITS(stream)
-
-size_t HUF_compress1X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable)
-{
-	const BYTE *ip = (const BYTE *)src;
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *const oend = ostart + dstSize;
-	BYTE *op = ostart;
-	size_t n;
-	BIT_CStream_t bitC;
-
-	/* init */
-	if (dstSize < 8)
-		return 0; /* not enough space to compress */
-	{
-		size_t const initErr = BIT_initCStream(&bitC, op, oend - op);
-		if (HUF_isError(initErr))
-			return 0;
-	}
-
-	n = srcSize & ~3; /* join to mod 4 */
-	switch (srcSize & 3) {
-	case 3: HUF_encodeSymbol(&bitC, ip[n + 2], CTable); HUF_FLUSHBITS_2(&bitC);
-		fallthrough;
-	case 2: HUF_encodeSymbol(&bitC, ip[n + 1], CTable); HUF_FLUSHBITS_1(&bitC);
-		fallthrough;
-	case 1: HUF_encodeSymbol(&bitC, ip[n + 0], CTable); HUF_FLUSHBITS(&bitC);
-		fallthrough;
-	case 0:
-	default:;
-	}
-
-	for (; n > 0; n -= 4) { /* note : n&3==0 at this stage */
-		HUF_encodeSymbol(&bitC, ip[n - 1], CTable);
-		HUF_FLUSHBITS_1(&bitC);
-		HUF_encodeSymbol(&bitC, ip[n - 2], CTable);
-		HUF_FLUSHBITS_2(&bitC);
-		HUF_encodeSymbol(&bitC, ip[n - 3], CTable);
-		HUF_FLUSHBITS_1(&bitC);
-		HUF_encodeSymbol(&bitC, ip[n - 4], CTable);
-		HUF_FLUSHBITS(&bitC);
-	}
-
-	return BIT_closeCStream(&bitC);
-}
-
-size_t HUF_compress4X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable)
-{
-	size_t const segmentSize = (srcSize + 3) / 4; /* first 3 segments */
-	const BYTE *ip = (const BYTE *)src;
-	const BYTE *const iend = ip + srcSize;
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *const oend = ostart + dstSize;
-	BYTE *op = ostart;
-
-	if (dstSize < 6 + 1 + 1 + 1 + 8)
-		return 0; /* minimum space to compress successfully */
-	if (srcSize < 12)
-		return 0; /* no saving possible : too small input */
-	op += 6;	  /* jumpTable */
-
-	{
-		CHECK_V_F(cSize, HUF_compress1X_usingCTable(op, oend - op, ip, segmentSize, CTable));
-		if (cSize == 0)
-			return 0;
-		ZSTD_writeLE16(ostart, (U16)cSize);
-		op += cSize;
-	}
-
-	ip += segmentSize;
-	{
-		CHECK_V_F(cSize, HUF_compress1X_usingCTable(op, oend - op, ip, segmentSize, CTable));
-		if (cSize == 0)
-			return 0;
-		ZSTD_writeLE16(ostart + 2, (U16)cSize);
-		op += cSize;
-	}
-
-	ip += segmentSize;
-	{
-		CHECK_V_F(cSize, HUF_compress1X_usingCTable(op, oend - op, ip, segmentSize, CTable));
-		if (cSize == 0)
-			return 0;
-		ZSTD_writeLE16(ostart + 4, (U16)cSize);
-		op += cSize;
-	}
-
-	ip += segmentSize;
-	{
-		CHECK_V_F(cSize, HUF_compress1X_usingCTable(op, oend - op, ip, iend - ip, CTable));
-		if (cSize == 0)
-			return 0;
-		op += cSize;
-	}
-
-	return op - ostart;
-}
-
-static size_t HUF_compressCTable_internal(BYTE *const ostart, BYTE *op, BYTE *const oend, const void *src, size_t srcSize, unsigned singleStream,
-					  const HUF_CElt *CTable)
-{
-	size_t const cSize =
-	    singleStream ? HUF_compress1X_usingCTable(op, oend - op, src, srcSize, CTable) : HUF_compress4X_usingCTable(op, oend - op, src, srcSize, CTable);
-	if (HUF_isError(cSize)) {
-		return cSize;
-	}
-	if (cSize == 0) {
-		return 0;
-	} /* uncompressible */
-	op += cSize;
-	/* check compressibility */
-	if ((size_t)(op - ostart) >= srcSize - 1) {
-		return 0;
-	}
-	return op - ostart;
-}
-
-/* `workSpace` must a table of at least 1024 unsigned */
-static size_t HUF_compress_internal(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog,
-				    unsigned singleStream, void *workSpace, size_t wkspSize, HUF_CElt *oldHufTable, HUF_repeat *repeat, int preferRepeat)
-{
-	BYTE *const ostart = (BYTE *)dst;
-	BYTE *const oend = ostart + dstSize;
-	BYTE *op = ostart;
-
-	U32 *count;
-	size_t const countSize = sizeof(U32) * (HUF_SYMBOLVALUE_MAX + 1);
-	HUF_CElt *CTable;
-	size_t const CTableSize = sizeof(HUF_CElt) * (HUF_SYMBOLVALUE_MAX + 1);
-
-	/* checks & inits */
-	if (wkspSize < sizeof(huffNodeTable) + countSize + CTableSize)
-		return ERROR(GENERIC);
-	if (!srcSize)
-		return 0; /* Uncompressed (note : 1 means rle, so first byte must be correct) */
-	if (!dstSize)
-		return 0; /* cannot fit within dst budget */
-	if (srcSize > HUF_BLOCKSIZE_MAX)
-		return ERROR(srcSize_wrong); /* curr block size limit */
-	if (huffLog > HUF_TABLELOG_MAX)
-		return ERROR(tableLog_tooLarge);
-	if (!maxSymbolValue)
-		maxSymbolValue = HUF_SYMBOLVALUE_MAX;
-	if (!huffLog)
-		huffLog = HUF_TABLELOG_DEFAULT;
-
-	count = (U32 *)workSpace;
-	workSpace = (BYTE *)workSpace + countSize;
-	wkspSize -= countSize;
-	CTable = (HUF_CElt *)workSpace;
-	workSpace = (BYTE *)workSpace + CTableSize;
-	wkspSize -= CTableSize;
-
-	/* Heuristic : If we don't need to check the validity of the old table use the old table for small inputs */
-	if (preferRepeat && repeat && *repeat == HUF_repeat_valid) {
-		return HUF_compressCTable_internal(ostart, op, oend, src, srcSize, singleStream, oldHufTable);
-	}
-
-	/* Scan input and build symbol stats */
-	{
-		CHECK_V_F(largest, FSE_count_wksp(count, &maxSymbolValue, (const BYTE *)src, srcSize, (U32 *)workSpace));
-		if (largest == srcSize) {
-			*ostart = ((const BYTE *)src)[0];
-			return 1;
-		} /* single symbol, rle */
-		if (largest <= (srcSize >> 7) + 1)
-			return 0; /* Fast heuristic : not compressible enough */
-	}
-
-	/* Check validity of previous table */
-	if (repeat && *repeat == HUF_repeat_check && !HUF_validateCTable(oldHufTable, count, maxSymbolValue)) {
-		*repeat = HUF_repeat_none;
-	}
-	/* Heuristic : use existing table for small inputs */
-	if (preferRepeat && repeat && *repeat != HUF_repeat_none) {
-		return HUF_compressCTable_internal(ostart, op, oend, src, srcSize, singleStream, oldHufTable);
-	}
-
-	/* Build Huffman Tree */
-	huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
-	{
-		CHECK_V_F(maxBits, HUF_buildCTable_wksp(CTable, count, maxSymbolValue, huffLog, workSpace, wkspSize));
-		huffLog = (U32)maxBits;
-		/* Zero the unused symbols so we can check it for validity */
-		memset(CTable + maxSymbolValue + 1, 0, CTableSize - (maxSymbolValue + 1) * sizeof(HUF_CElt));
-	}
-
-	/* Write table description header */
-	{
-		CHECK_V_F(hSize, HUF_writeCTable_wksp(op, dstSize, CTable, maxSymbolValue, huffLog, workSpace, wkspSize));
-		/* Check if using the previous table will be beneficial */
-		if (repeat && *repeat != HUF_repeat_none) {
-			size_t const oldSize = HUF_estimateCompressedSize(oldHufTable, count, maxSymbolValue);
-			size_t const newSize = HUF_estimateCompressedSize(CTable, count, maxSymbolValue);
-			if (oldSize <= hSize + newSize || hSize + 12 >= srcSize) {
-				return HUF_compressCTable_internal(ostart, op, oend, src, srcSize, singleStream, oldHufTable);
-			}
-		}
-		/* Use the new table */
-		if (hSize + 12ul >= srcSize) {
-			return 0;
-		}
-		op += hSize;
-		if (repeat) {
-			*repeat = HUF_repeat_none;
-		}
-		if (oldHufTable) {
-			memcpy(oldHufTable, CTable, CTableSize);
-		} /* Save the new table */
-	}
-	return HUF_compressCTable_internal(ostart, op, oend, src, srcSize, singleStream, CTable);
-}
-
-size_t HUF_compress1X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog, void *workSpace,
-			   size_t wkspSize)
-{
-	return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 1 /* single stream */, workSpace, wkspSize, NULL, NULL, 0);
-}
-
-size_t HUF_compress1X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog, void *workSpace,
-			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat, int preferRepeat)
-{
-	return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 1 /* single stream */, workSpace, wkspSize, hufTable, repeat,
-				     preferRepeat);
-}
-
-size_t HUF_compress4X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog, void *workSpace,
-			   size_t wkspSize)
-{
-	return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 0 /* 4 streams */, workSpace, wkspSize, NULL, NULL, 0);
-}
-
-size_t HUF_compress4X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog, void *workSpace,
-			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat, int preferRepeat)
-{
-	return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 0 /* 4 streams */, workSpace, wkspSize, hufTable, repeat,
-				     preferRepeat);
-}
diff --git a/lib/zstd/huf_decompress.c b/lib/zstd/huf_decompress.c
deleted file mode 100644
index 6526482..0000000
--- a/lib/zstd/huf_decompress.c
+++ /dev/null
@@ -1,960 +0,0 @@
-/*
- * Huffman decoder, part of New Generation Entropy library
- * Copyright (C) 2013-2016, Yann Collet.
- *
- * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are
- * met:
- *
- *   * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *   * Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following disclaimer
- * in the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- *
- * You can contact the author at :
- * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
- */
-
-/* **************************************************************
-*  Compiler specifics
-****************************************************************/
-#define FORCE_INLINE static __always_inline
-
-/* **************************************************************
-*  Dependencies
-****************************************************************/
-#include "bitstream.h" /* BIT_* */
-#include "fse.h"       /* header compression */
-#include "huf.h"
-#include <linux/compiler.h>
-#include <linux/kernel.h>
-#include <linux/string.h> /* memcpy, memset */
-
-/* **************************************************************
-*  Error Management
-****************************************************************/
-#define HUF_STATIC_ASSERT(c)                                   \
-	{                                                      \
-		enum { HUF_static_assert = 1 / (int)(!!(c)) }; \
-	} /* use only *after* variable declarations */
-
-/*-***************************/
-/*  generic DTableDesc       */
-/*-***************************/
-
-typedef struct {
-	BYTE maxTableLog;
-	BYTE tableType;
-	BYTE tableLog;
-	BYTE reserved;
-} DTableDesc;
-
-static DTableDesc HUF_getDTableDesc(const HUF_DTable *table)
-{
-	DTableDesc dtd;
-	memcpy(&dtd, table, sizeof(dtd));
-	return dtd;
-}
-
-/*-***************************/
-/*  single-symbol decoding   */
-/*-***************************/
-
-typedef struct {
-	BYTE byte;
-	BYTE nbBits;
-} HUF_DEltX2; /* single-symbol decoding */
-
-size_t HUF_readDTableX2_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
-{
-	U32 tableLog = 0;
-	U32 nbSymbols = 0;
-	size_t iSize;
-	void *const dtPtr = DTable + 1;
-	HUF_DEltX2 *const dt = (HUF_DEltX2 *)dtPtr;
-
-	U32 *rankVal;
-	BYTE *huffWeight;
-	size_t spaceUsed32 = 0;
-
-	rankVal = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += HUF_TABLELOG_ABSOLUTEMAX + 1;
-	huffWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	HUF_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable));
-	/* memset(huffWeight, 0, sizeof(huffWeight)); */ /* is not necessary, even though some analyzer complain ... */
-
-	iSize = HUF_readStats_wksp(huffWeight, HUF_SYMBOLVALUE_MAX + 1, rankVal, &nbSymbols, &tableLog, src, srcSize, workspace, workspaceSize);
-	if (HUF_isError(iSize))
-		return iSize;
-
-	/* Table header */
-	{
-		DTableDesc dtd = HUF_getDTableDesc(DTable);
-		if (tableLog > (U32)(dtd.maxTableLog + 1))
-			return ERROR(tableLog_tooLarge); /* DTable too small, Huffman tree cannot fit in */
-		dtd.tableType = 0;
-		dtd.tableLog = (BYTE)tableLog;
-		memcpy(DTable, &dtd, sizeof(dtd));
-	}
-
-	/* Calculate starting value for each rank */
-	{
-		U32 n, nextRankStart = 0;
-		for (n = 1; n < tableLog + 1; n++) {
-			U32 const curr = nextRankStart;
-			nextRankStart += (rankVal[n] << (n - 1));
-			rankVal[n] = curr;
-		}
-	}
-
-	/* fill DTable */
-	{
-		U32 n;
-		for (n = 0; n < nbSymbols; n++) {
-			U32 const w = huffWeight[n];
-			U32 const length = (1 << w) >> 1;
-			U32 u;
-			HUF_DEltX2 D;
-			D.byte = (BYTE)n;
-			D.nbBits = (BYTE)(tableLog + 1 - w);
-			for (u = rankVal[w]; u < rankVal[w] + length; u++)
-				dt[u] = D;
-			rankVal[w] += length;
-		}
-	}
-
-	return iSize;
-}
-
-static BYTE HUF_decodeSymbolX2(BIT_DStream_t *Dstream, const HUF_DEltX2 *dt, const U32 dtLog)
-{
-	size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
-	BYTE const c = dt[val].byte;
-	BIT_skipBits(Dstream, dt[val].nbBits);
-	return c;
-}
-
-#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) *ptr++ = HUF_decodeSymbolX2(DStreamPtr, dt, dtLog)
-
-#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr)         \
-	if (ZSTD_64bits() || (HUF_TABLELOG_MAX <= 12)) \
-	HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
-
-#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
-	if (ZSTD_64bits())                     \
-	HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
-
-FORCE_INLINE size_t HUF_decodeStreamX2(BYTE *p, BIT_DStream_t *const bitDPtr, BYTE *const pEnd, const HUF_DEltX2 *const dt, const U32 dtLog)
-{
-	BYTE *const pStart = p;
-
-	/* up to 4 symbols at a time */
-	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd - 4)) {
-		HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
-		HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
-		HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
-		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
-	}
-
-	/* closer to the end */
-	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd))
-		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
-
-	/* no more data to retrieve from bitstream, hence no need to reload */
-	while (p < pEnd)
-		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
-
-	return pEnd - pStart;
-}
-
-static size_t HUF_decompress1X2_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	BYTE *op = (BYTE *)dst;
-	BYTE *const oend = op + dstSize;
-	const void *dtPtr = DTable + 1;
-	const HUF_DEltX2 *const dt = (const HUF_DEltX2 *)dtPtr;
-	BIT_DStream_t bitD;
-	DTableDesc const dtd = HUF_getDTableDesc(DTable);
-	U32 const dtLog = dtd.tableLog;
-
-	{
-		size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
-		if (HUF_isError(errorCode))
-			return errorCode;
-	}
-
-	HUF_decodeStreamX2(op, &bitD, oend, dt, dtLog);
-
-	/* check */
-	if (!BIT_endOfDStream(&bitD))
-		return ERROR(corruption_detected);
-
-	return dstSize;
-}
-
-size_t HUF_decompress1X2_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	if (dtd.tableType != 0)
-		return ERROR(GENERIC);
-	return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable *DCtx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	const BYTE *ip = (const BYTE *)cSrc;
-
-	size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize, workspace, workspaceSize);
-	if (HUF_isError(hSize))
-		return hSize;
-	if (hSize >= cSrcSize)
-		return ERROR(srcSize_wrong);
-	ip += hSize;
-	cSrcSize -= hSize;
-
-	return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx);
-}
-
-static size_t HUF_decompress4X2_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	/* Check */
-	if (cSrcSize < 10)
-		return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
-
-	{
-		const BYTE *const istart = (const BYTE *)cSrc;
-		BYTE *const ostart = (BYTE *)dst;
-		BYTE *const oend = ostart + dstSize;
-		const void *const dtPtr = DTable + 1;
-		const HUF_DEltX2 *const dt = (const HUF_DEltX2 *)dtPtr;
-
-		/* Init */
-		BIT_DStream_t bitD1;
-		BIT_DStream_t bitD2;
-		BIT_DStream_t bitD3;
-		BIT_DStream_t bitD4;
-		size_t const length1 = ZSTD_readLE16(istart);
-		size_t const length2 = ZSTD_readLE16(istart + 2);
-		size_t const length3 = ZSTD_readLE16(istart + 4);
-		size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
-		const BYTE *const istart1 = istart + 6; /* jumpTable */
-		const BYTE *const istart2 = istart1 + length1;
-		const BYTE *const istart3 = istart2 + length2;
-		const BYTE *const istart4 = istart3 + length3;
-		const size_t segmentSize = (dstSize + 3) / 4;
-		BYTE *const opStart2 = ostart + segmentSize;
-		BYTE *const opStart3 = opStart2 + segmentSize;
-		BYTE *const opStart4 = opStart3 + segmentSize;
-		BYTE *op1 = ostart;
-		BYTE *op2 = opStart2;
-		BYTE *op3 = opStart3;
-		BYTE *op4 = opStart4;
-		U32 endSignal;
-		DTableDesc const dtd = HUF_getDTableDesc(DTable);
-		U32 const dtLog = dtd.tableLog;
-
-		if (length4 > cSrcSize)
-			return ERROR(corruption_detected); /* overflow */
-		{
-			size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-
-		/* 16-32 symbols per loop (4-8 symbols per stream) */
-		endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-		for (; (endSignal == BIT_DStream_unfinished) && (op4 < (oend - 7));) {
-			HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
-			HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
-			HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
-			HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
-			HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
-			HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
-			HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
-			HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
-			HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
-			HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
-			HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
-			HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
-			HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
-			HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
-			HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
-			HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
-			endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-		}
-
-		/* check corruption */
-		if (op1 > opStart2)
-			return ERROR(corruption_detected);
-		if (op2 > opStart3)
-			return ERROR(corruption_detected);
-		if (op3 > opStart4)
-			return ERROR(corruption_detected);
-		/* note : op4 supposed already verified within main loop */
-
-		/* finish bitStreams one by one */
-		HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog);
-		HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog);
-		HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog);
-		HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog);
-
-		/* check */
-		endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
-		if (!endSignal)
-			return ERROR(corruption_detected);
-
-		/* decoded size */
-		return dstSize;
-	}
-}
-
-size_t HUF_decompress4X2_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	if (dtd.tableType != 0)
-		return ERROR(GENERIC);
-	return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	const BYTE *ip = (const BYTE *)cSrc;
-
-	size_t const hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize, workspace, workspaceSize);
-	if (HUF_isError(hSize))
-		return hSize;
-	if (hSize >= cSrcSize)
-		return ERROR(srcSize_wrong);
-	ip += hSize;
-	cSrcSize -= hSize;
-
-	return HUF_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx);
-}
-
-/* *************************/
-/* double-symbols decoding */
-/* *************************/
-typedef struct {
-	U16 sequence;
-	BYTE nbBits;
-	BYTE length;
-} HUF_DEltX4; /* double-symbols decoding */
-
-typedef struct {
-	BYTE symbol;
-	BYTE weight;
-} sortedSymbol_t;
-
-/* HUF_fillDTableX4Level2() :
- * `rankValOrigin` must be a table of at least (HUF_TABLELOG_MAX + 1) U32 */
-static void HUF_fillDTableX4Level2(HUF_DEltX4 *DTable, U32 sizeLog, const U32 consumed, const U32 *rankValOrigin, const int minWeight,
-				   const sortedSymbol_t *sortedSymbols, const U32 sortedListSize, U32 nbBitsBaseline, U16 baseSeq)
-{
-	HUF_DEltX4 DElt;
-	U32 rankVal[HUF_TABLELOG_MAX + 1];
-
-	/* get pre-calculated rankVal */
-	memcpy(rankVal, rankValOrigin, sizeof(rankVal));
-
-	/* fill skipped values */
-	if (minWeight > 1) {
-		U32 i, skipSize = rankVal[minWeight];
-		ZSTD_writeLE16(&(DElt.sequence), baseSeq);
-		DElt.nbBits = (BYTE)(consumed);
-		DElt.length = 1;
-		for (i = 0; i < skipSize; i++)
-			DTable[i] = DElt;
-	}
-
-	/* fill DTable */
-	{
-		U32 s;
-		for (s = 0; s < sortedListSize; s++) { /* note : sortedSymbols already skipped */
-			const U32 symbol = sortedSymbols[s].symbol;
-			const U32 weight = sortedSymbols[s].weight;
-			const U32 nbBits = nbBitsBaseline - weight;
-			const U32 length = 1 << (sizeLog - nbBits);
-			const U32 start = rankVal[weight];
-			U32 i = start;
-			const U32 end = start + length;
-
-			ZSTD_writeLE16(&(DElt.sequence), (U16)(baseSeq + (symbol << 8)));
-			DElt.nbBits = (BYTE)(nbBits + consumed);
-			DElt.length = 2;
-			do {
-				DTable[i++] = DElt;
-			} while (i < end); /* since length >= 1 */
-
-			rankVal[weight] += length;
-		}
-	}
-}
-
-typedef U32 rankVal_t[HUF_TABLELOG_MAX][HUF_TABLELOG_MAX + 1];
-typedef U32 rankValCol_t[HUF_TABLELOG_MAX + 1];
-
-static void HUF_fillDTableX4(HUF_DEltX4 *DTable, const U32 targetLog, const sortedSymbol_t *sortedList, const U32 sortedListSize, const U32 *rankStart,
-			     rankVal_t rankValOrigin, const U32 maxWeight, const U32 nbBitsBaseline)
-{
-	U32 rankVal[HUF_TABLELOG_MAX + 1];
-	const int scaleLog = nbBitsBaseline - targetLog; /* note : targetLog >= srcLog, hence scaleLog <= 1 */
-	const U32 minBits = nbBitsBaseline - maxWeight;
-	U32 s;
-
-	memcpy(rankVal, rankValOrigin, sizeof(rankVal));
-
-	/* fill DTable */
-	for (s = 0; s < sortedListSize; s++) {
-		const U16 symbol = sortedList[s].symbol;
-		const U32 weight = sortedList[s].weight;
-		const U32 nbBits = nbBitsBaseline - weight;
-		const U32 start = rankVal[weight];
-		const U32 length = 1 << (targetLog - nbBits);
-
-		if (targetLog - nbBits >= minBits) { /* enough room for a second symbol */
-			U32 sortedRank;
-			int minWeight = nbBits + scaleLog;
-			if (minWeight < 1)
-				minWeight = 1;
-			sortedRank = rankStart[minWeight];
-			HUF_fillDTableX4Level2(DTable + start, targetLog - nbBits, nbBits, rankValOrigin[nbBits], minWeight, sortedList + sortedRank,
-					       sortedListSize - sortedRank, nbBitsBaseline, symbol);
-		} else {
-			HUF_DEltX4 DElt;
-			ZSTD_writeLE16(&(DElt.sequence), symbol);
-			DElt.nbBits = (BYTE)(nbBits);
-			DElt.length = 1;
-			{
-				U32 const end = start + length;
-				U32 u;
-				for (u = start; u < end; u++)
-					DTable[u] = DElt;
-			}
-		}
-		rankVal[weight] += length;
-	}
-}
-
-size_t HUF_readDTableX4_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
-{
-	U32 tableLog, maxW, sizeOfSort, nbSymbols;
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	U32 const maxTableLog = dtd.maxTableLog;
-	size_t iSize;
-	void *dtPtr = DTable + 1; /* force compiler to avoid strict-aliasing */
-	HUF_DEltX4 *const dt = (HUF_DEltX4 *)dtPtr;
-	U32 *rankStart;
-
-	rankValCol_t *rankVal;
-	U32 *rankStats;
-	U32 *rankStart0;
-	sortedSymbol_t *sortedSymbol;
-	BYTE *weightList;
-	size_t spaceUsed32 = 0;
-
-	HUF_STATIC_ASSERT((sizeof(rankValCol_t) & 3) == 0);
-
-	rankVal = (rankValCol_t *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += (sizeof(rankValCol_t) * HUF_TABLELOG_MAX) >> 2;
-	rankStats = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += HUF_TABLELOG_MAX + 1;
-	rankStart0 = (U32 *)workspace + spaceUsed32;
-	spaceUsed32 += HUF_TABLELOG_MAX + 2;
-	sortedSymbol = (sortedSymbol_t *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(sizeof(sortedSymbol_t) * (HUF_SYMBOLVALUE_MAX + 1), sizeof(U32)) >> 2;
-	weightList = (BYTE *)((U32 *)workspace + spaceUsed32);
-	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
-
-	if ((spaceUsed32 << 2) > workspaceSize)
-		return ERROR(tableLog_tooLarge);
-	workspace = (U32 *)workspace + spaceUsed32;
-	workspaceSize -= (spaceUsed32 << 2);
-
-	rankStart = rankStart0 + 1;
-	memset(rankStats, 0, sizeof(U32) * (2 * HUF_TABLELOG_MAX + 2 + 1));
-
-	HUF_STATIC_ASSERT(sizeof(HUF_DEltX4) == sizeof(HUF_DTable)); /* if compiler fails here, assertion is wrong */
-	if (maxTableLog > HUF_TABLELOG_MAX)
-		return ERROR(tableLog_tooLarge);
-	/* memset(weightList, 0, sizeof(weightList)); */ /* is not necessary, even though some analyzer complain ... */
-
-	iSize = HUF_readStats_wksp(weightList, HUF_SYMBOLVALUE_MAX + 1, rankStats, &nbSymbols, &tableLog, src, srcSize, workspace, workspaceSize);
-	if (HUF_isError(iSize))
-		return iSize;
-
-	/* check result */
-	if (tableLog > maxTableLog)
-		return ERROR(tableLog_tooLarge); /* DTable can't fit code depth */
-
-	/* find maxWeight */
-	for (maxW = tableLog; rankStats[maxW] == 0; maxW--) {
-	} /* necessarily finds a solution before 0 */
-
-	/* Get start index of each weight */
-	{
-		U32 w, nextRankStart = 0;
-		for (w = 1; w < maxW + 1; w++) {
-			U32 curr = nextRankStart;
-			nextRankStart += rankStats[w];
-			rankStart[w] = curr;
-		}
-		rankStart[0] = nextRankStart; /* put all 0w symbols at the end of sorted list*/
-		sizeOfSort = nextRankStart;
-	}
-
-	/* sort symbols by weight */
-	{
-		U32 s;
-		for (s = 0; s < nbSymbols; s++) {
-			U32 const w = weightList[s];
-			U32 const r = rankStart[w]++;
-			sortedSymbol[r].symbol = (BYTE)s;
-			sortedSymbol[r].weight = (BYTE)w;
-		}
-		rankStart[0] = 0; /* forget 0w symbols; this is beginning of weight(1) */
-	}
-
-	/* Build rankVal */
-	{
-		U32 *const rankVal0 = rankVal[0];
-		{
-			int const rescale = (maxTableLog - tableLog) - 1; /* tableLog <= maxTableLog */
-			U32 nextRankVal = 0;
-			U32 w;
-			for (w = 1; w < maxW + 1; w++) {
-				U32 curr = nextRankVal;
-				nextRankVal += rankStats[w] << (w + rescale);
-				rankVal0[w] = curr;
-			}
-		}
-		{
-			U32 const minBits = tableLog + 1 - maxW;
-			U32 consumed;
-			for (consumed = minBits; consumed < maxTableLog - minBits + 1; consumed++) {
-				U32 *const rankValPtr = rankVal[consumed];
-				U32 w;
-				for (w = 1; w < maxW + 1; w++) {
-					rankValPtr[w] = rankVal0[w] >> consumed;
-				}
-			}
-		}
-	}
-
-	HUF_fillDTableX4(dt, maxTableLog, sortedSymbol, sizeOfSort, rankStart0, rankVal, maxW, tableLog + 1);
-
-	dtd.tableLog = (BYTE)maxTableLog;
-	dtd.tableType = 1;
-	memcpy(DTable, &dtd, sizeof(dtd));
-	return iSize;
-}
-
-static U32 HUF_decodeSymbolX4(void *op, BIT_DStream_t *DStream, const HUF_DEltX4 *dt, const U32 dtLog)
-{
-	size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
-	memcpy(op, dt + val, 2);
-	BIT_skipBits(DStream, dt[val].nbBits);
-	return dt[val].length;
-}
-
-static U32 HUF_decodeLastSymbolX4(void *op, BIT_DStream_t *DStream, const HUF_DEltX4 *dt, const U32 dtLog)
-{
-	size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
-	memcpy(op, dt + val, 1);
-	if (dt[val].length == 1)
-		BIT_skipBits(DStream, dt[val].nbBits);
-	else {
-		if (DStream->bitsConsumed < (sizeof(DStream->bitContainer) * 8)) {
-			BIT_skipBits(DStream, dt[val].nbBits);
-			if (DStream->bitsConsumed > (sizeof(DStream->bitContainer) * 8))
-				/* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
-				DStream->bitsConsumed = (sizeof(DStream->bitContainer) * 8);
-		}
-	}
-	return 1;
-}
-
-#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr)         \
-	if (ZSTD_64bits() || (HUF_TABLELOG_MAX <= 12)) \
-	ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \
-	if (ZSTD_64bits())                     \
-	ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-FORCE_INLINE size_t HUF_decodeStreamX4(BYTE *p, BIT_DStream_t *bitDPtr, BYTE *const pEnd, const HUF_DEltX4 *const dt, const U32 dtLog)
-{
-	BYTE *const pStart = p;
-
-	/* up to 8 symbols at a time */
-	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd - (sizeof(bitDPtr->bitContainer) - 1))) {
-		HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
-		HUF_DECODE_SYMBOLX4_1(p, bitDPtr);
-		HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
-		HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
-	}
-
-	/* closer to end : up to 2 symbols at a time */
-	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd - 2))
-		HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
-
-	while (p <= pEnd - 2)
-		HUF_DECODE_SYMBOLX4_0(p, bitDPtr); /* no need to reload : reached the end of DStream */
-
-	if (p < pEnd)
-		p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog);
-
-	return p - pStart;
-}
-
-static size_t HUF_decompress1X4_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	BIT_DStream_t bitD;
-
-	/* Init */
-	{
-		size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
-		if (HUF_isError(errorCode))
-			return errorCode;
-	}
-
-	/* decode */
-	{
-		BYTE *const ostart = (BYTE *)dst;
-		BYTE *const oend = ostart + dstSize;
-		const void *const dtPtr = DTable + 1; /* force compiler to not use strict-aliasing */
-		const HUF_DEltX4 *const dt = (const HUF_DEltX4 *)dtPtr;
-		DTableDesc const dtd = HUF_getDTableDesc(DTable);
-		HUF_decodeStreamX4(ostart, &bitD, oend, dt, dtd.tableLog);
-	}
-
-	/* check */
-	if (!BIT_endOfDStream(&bitD))
-		return ERROR(corruption_detected);
-
-	/* decoded size */
-	return dstSize;
-}
-
-size_t HUF_decompress1X4_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	if (dtd.tableType != 1)
-		return ERROR(GENERIC);
-	return HUF_decompress1X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable *DCtx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	const BYTE *ip = (const BYTE *)cSrc;
-
-	size_t const hSize = HUF_readDTableX4_wksp(DCtx, cSrc, cSrcSize, workspace, workspaceSize);
-	if (HUF_isError(hSize))
-		return hSize;
-	if (hSize >= cSrcSize)
-		return ERROR(srcSize_wrong);
-	ip += hSize;
-	cSrcSize -= hSize;
-
-	return HUF_decompress1X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx);
-}
-
-static size_t HUF_decompress4X4_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	if (cSrcSize < 10)
-		return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
-
-	{
-		const BYTE *const istart = (const BYTE *)cSrc;
-		BYTE *const ostart = (BYTE *)dst;
-		BYTE *const oend = ostart + dstSize;
-		const void *const dtPtr = DTable + 1;
-		const HUF_DEltX4 *const dt = (const HUF_DEltX4 *)dtPtr;
-
-		/* Init */
-		BIT_DStream_t bitD1;
-		BIT_DStream_t bitD2;
-		BIT_DStream_t bitD3;
-		BIT_DStream_t bitD4;
-		size_t const length1 = ZSTD_readLE16(istart);
-		size_t const length2 = ZSTD_readLE16(istart + 2);
-		size_t const length3 = ZSTD_readLE16(istart + 4);
-		size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
-		const BYTE *const istart1 = istart + 6; /* jumpTable */
-		const BYTE *const istart2 = istart1 + length1;
-		const BYTE *const istart3 = istart2 + length2;
-		const BYTE *const istart4 = istart3 + length3;
-		size_t const segmentSize = (dstSize + 3) / 4;
-		BYTE *const opStart2 = ostart + segmentSize;
-		BYTE *const opStart3 = opStart2 + segmentSize;
-		BYTE *const opStart4 = opStart3 + segmentSize;
-		BYTE *op1 = ostart;
-		BYTE *op2 = opStart2;
-		BYTE *op3 = opStart3;
-		BYTE *op4 = opStart4;
-		U32 endSignal;
-		DTableDesc const dtd = HUF_getDTableDesc(DTable);
-		U32 const dtLog = dtd.tableLog;
-
-		if (length4 > cSrcSize)
-			return ERROR(corruption_detected); /* overflow */
-		{
-			size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-		{
-			size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
-			if (HUF_isError(errorCode))
-				return errorCode;
-		}
-
-		/* 16-32 symbols per loop (4-8 symbols per stream) */
-		endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-		for (; (endSignal == BIT_DStream_unfinished) & (op4 < (oend - (sizeof(bitD4.bitContainer) - 1)));) {
-			HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
-			HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
-			HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
-			HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
-			HUF_DECODE_SYMBOLX4_1(op1, &bitD1);
-			HUF_DECODE_SYMBOLX4_1(op2, &bitD2);
-			HUF_DECODE_SYMBOLX4_1(op3, &bitD3);
-			HUF_DECODE_SYMBOLX4_1(op4, &bitD4);
-			HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
-			HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
-			HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
-			HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
-			HUF_DECODE_SYMBOLX4_0(op1, &bitD1);
-			HUF_DECODE_SYMBOLX4_0(op2, &bitD2);
-			HUF_DECODE_SYMBOLX4_0(op3, &bitD3);
-			HUF_DECODE_SYMBOLX4_0(op4, &bitD4);
-
-			endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-		}
-
-		/* check corruption */
-		if (op1 > opStart2)
-			return ERROR(corruption_detected);
-		if (op2 > opStart3)
-			return ERROR(corruption_detected);
-		if (op3 > opStart4)
-			return ERROR(corruption_detected);
-		/* note : op4 already verified within main loop */
-
-		/* finish bitStreams one by one */
-		HUF_decodeStreamX4(op1, &bitD1, opStart2, dt, dtLog);
-		HUF_decodeStreamX4(op2, &bitD2, opStart3, dt, dtLog);
-		HUF_decodeStreamX4(op3, &bitD3, opStart4, dt, dtLog);
-		HUF_decodeStreamX4(op4, &bitD4, oend, dt, dtLog);
-
-		/* check */
-		{
-			U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
-			if (!endCheck)
-				return ERROR(corruption_detected);
-		}
-
-		/* decoded size */
-		return dstSize;
-	}
-}
-
-size_t HUF_decompress4X4_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc dtd = HUF_getDTableDesc(DTable);
-	if (dtd.tableType != 1)
-		return ERROR(GENERIC);
-	return HUF_decompress4X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	const BYTE *ip = (const BYTE *)cSrc;
-
-	size_t hSize = HUF_readDTableX4_wksp(dctx, cSrc, cSrcSize, workspace, workspaceSize);
-	if (HUF_isError(hSize))
-		return hSize;
-	if (hSize >= cSrcSize)
-		return ERROR(srcSize_wrong);
-	ip += hSize;
-	cSrcSize -= hSize;
-
-	return HUF_decompress4X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx);
-}
-
-/* ********************************/
-/* Generic decompression selector */
-/* ********************************/
-
-size_t HUF_decompress1X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc const dtd = HUF_getDTableDesc(DTable);
-	return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable)
-			     : HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable);
-}
-
-size_t HUF_decompress4X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
-{
-	DTableDesc const dtd = HUF_getDTableDesc(DTable);
-	return dtd.tableType ? HUF_decompress4X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable)
-			     : HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable);
-}
-
-typedef struct {
-	U32 tableTime;
-	U32 decode256Time;
-} algo_time_t;
-static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, quad */] = {
-    /* single, double, quad */
-    {{0, 0}, {1, 1}, {2, 2}},		     /* Q==0 : impossible */
-    {{0, 0}, {1, 1}, {2, 2}},		     /* Q==1 : impossible */
-    {{38, 130}, {1313, 74}, {2151, 38}},     /* Q == 2 : 12-18% */
-    {{448, 128}, {1353, 74}, {2238, 41}},    /* Q == 3 : 18-25% */
-    {{556, 128}, {1353, 74}, {2238, 47}},    /* Q == 4 : 25-32% */
-    {{714, 128}, {1418, 74}, {2436, 53}},    /* Q == 5 : 32-38% */
-    {{883, 128}, {1437, 74}, {2464, 61}},    /* Q == 6 : 38-44% */
-    {{897, 128}, {1515, 75}, {2622, 68}},    /* Q == 7 : 44-50% */
-    {{926, 128}, {1613, 75}, {2730, 75}},    /* Q == 8 : 50-56% */
-    {{947, 128}, {1729, 77}, {3359, 77}},    /* Q == 9 : 56-62% */
-    {{1107, 128}, {2083, 81}, {4006, 84}},   /* Q ==10 : 62-69% */
-    {{1177, 128}, {2379, 87}, {4785, 88}},   /* Q ==11 : 69-75% */
-    {{1242, 128}, {2415, 93}, {5155, 84}},   /* Q ==12 : 75-81% */
-    {{1349, 128}, {2644, 106}, {5260, 106}}, /* Q ==13 : 81-87% */
-    {{1455, 128}, {2422, 124}, {4174, 124}}, /* Q ==14 : 87-93% */
-    {{722, 128}, {1891, 145}, {1936, 146}},  /* Q ==15 : 93-99% */
-};
-
-/** HUF_selectDecoder() :
-*   Tells which decoder is likely to decode faster,
-*   based on a set of pre-determined metrics.
-*   @return : 0==HUF_decompress4X2, 1==HUF_decompress4X4 .
-*   Assumption : 0 < cSrcSize < dstSize <= 128 KB */
-U32 HUF_selectDecoder(size_t dstSize, size_t cSrcSize)
-{
-	/* decoder timing evaluation */
-	U32 const Q = (U32)(cSrcSize * 16 / dstSize); /* Q < 16 since dstSize > cSrcSize */
-	U32 const D256 = (U32)(dstSize >> 8);
-	U32 const DTime0 = algoTime[Q][0].tableTime + (algoTime[Q][0].decode256Time * D256);
-	U32 DTime1 = algoTime[Q][1].tableTime + (algoTime[Q][1].decode256Time * D256);
-	DTime1 += DTime1 >> 3; /* advantage to algorithm using less memory, for cache eviction */
-
-	return DTime1 < DTime0;
-}
-
-typedef size_t (*decompressionAlgo)(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize);
-
-size_t HUF_decompress4X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	/* validation checks */
-	if (dstSize == 0)
-		return ERROR(dstSize_tooSmall);
-	if (cSrcSize > dstSize)
-		return ERROR(corruption_detected); /* invalid */
-	if (cSrcSize == dstSize) {
-		memcpy(dst, cSrc, dstSize);
-		return dstSize;
-	} /* not compressed */
-	if (cSrcSize == 1) {
-		memset(dst, *(const BYTE *)cSrc, dstSize);
-		return dstSize;
-	} /* RLE */
-
-	{
-		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
-		return algoNb ? HUF_decompress4X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
-			      : HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
-	}
-}
-
-size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	/* validation checks */
-	if (dstSize == 0)
-		return ERROR(dstSize_tooSmall);
-	if ((cSrcSize >= dstSize) || (cSrcSize <= 1))
-		return ERROR(corruption_detected); /* invalid */
-
-	{
-		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
-		return algoNb ? HUF_decompress4X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
-			      : HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
-	}
-}
-
-size_t HUF_decompress1X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
-{
-	/* validation checks */
-	if (dstSize == 0)
-		return ERROR(dstSize_tooSmall);
-	if (cSrcSize > dstSize)
-		return ERROR(corruption_detected); /* invalid */
-	if (cSrcSize == dstSize) {
-		memcpy(dst, cSrc, dstSize);
-		return dstSize;
-	} /* not compressed */
-	if (cSrcSize == 1) {
-		memset(dst, *(const BYTE *)cSrc, dstSize);
-		return dstSize;
-	} /* RLE */
-
-	{
-		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
-		return algoNb ? HUF_decompress1X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
-			      : HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
-	}
-}
diff --git a/lib/zstd/mem.h b/lib/zstd/mem.h
deleted file mode 100644
index 93d7a2c..0000000
--- a/lib/zstd/mem.h
+++ /dev/null
@@ -1,151 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-#ifndef MEM_H_MODULE
-#define MEM_H_MODULE
-
-/*-****************************************
-*  Dependencies
-******************************************/
-#include <asm/unaligned.h>
-#include <linux/string.h> /* memcpy */
-#include <linux/types.h>  /* size_t, ptrdiff_t */
-
-/*-****************************************
-*  Compiler specifics
-******************************************/
-#define ZSTD_STATIC static inline
-
-/*-**************************************************************
-*  Basic Types
-*****************************************************************/
-typedef uint8_t BYTE;
-typedef uint16_t U16;
-typedef int16_t S16;
-typedef uint32_t U32;
-typedef int32_t S32;
-typedef uint64_t U64;
-typedef int64_t S64;
-typedef ptrdiff_t iPtrDiff;
-typedef uintptr_t uPtrDiff;
-
-/*-**************************************************************
-*  Memory I/O
-*****************************************************************/
-ZSTD_STATIC unsigned ZSTD_32bits(void) { return sizeof(size_t) == 4; }
-ZSTD_STATIC unsigned ZSTD_64bits(void) { return sizeof(size_t) == 8; }
-
-#if defined(__LITTLE_ENDIAN)
-#define ZSTD_LITTLE_ENDIAN 1
-#else
-#define ZSTD_LITTLE_ENDIAN 0
-#endif
-
-ZSTD_STATIC unsigned ZSTD_isLittleEndian(void) { return ZSTD_LITTLE_ENDIAN; }
-
-ZSTD_STATIC U16 ZSTD_read16(const void *memPtr) { return get_unaligned((const U16 *)memPtr); }
-
-ZSTD_STATIC U32 ZSTD_read32(const void *memPtr) { return get_unaligned((const U32 *)memPtr); }
-
-ZSTD_STATIC U64 ZSTD_read64(const void *memPtr) { return get_unaligned((const U64 *)memPtr); }
-
-ZSTD_STATIC size_t ZSTD_readST(const void *memPtr) { return get_unaligned((const size_t *)memPtr); }
-
-ZSTD_STATIC void ZSTD_write16(void *memPtr, U16 value) { put_unaligned(value, (U16 *)memPtr); }
-
-ZSTD_STATIC void ZSTD_write32(void *memPtr, U32 value) { put_unaligned(value, (U32 *)memPtr); }
-
-ZSTD_STATIC void ZSTD_write64(void *memPtr, U64 value) { put_unaligned(value, (U64 *)memPtr); }
-
-/*=== Little endian r/w ===*/
-
-ZSTD_STATIC U16 ZSTD_readLE16(const void *memPtr) { return get_unaligned_le16(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeLE16(void *memPtr, U16 val) { put_unaligned_le16(val, memPtr); }
-
-ZSTD_STATIC U32 ZSTD_readLE24(const void *memPtr) { return ZSTD_readLE16(memPtr) + (((const BYTE *)memPtr)[2] << 16); }
-
-ZSTD_STATIC void ZSTD_writeLE24(void *memPtr, U32 val)
-{
-	ZSTD_writeLE16(memPtr, (U16)val);
-	((BYTE *)memPtr)[2] = (BYTE)(val >> 16);
-}
-
-ZSTD_STATIC U32 ZSTD_readLE32(const void *memPtr) { return get_unaligned_le32(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeLE32(void *memPtr, U32 val32) { put_unaligned_le32(val32, memPtr); }
-
-ZSTD_STATIC U64 ZSTD_readLE64(const void *memPtr) { return get_unaligned_le64(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeLE64(void *memPtr, U64 val64) { put_unaligned_le64(val64, memPtr); }
-
-ZSTD_STATIC size_t ZSTD_readLEST(const void *memPtr)
-{
-	if (ZSTD_32bits())
-		return (size_t)ZSTD_readLE32(memPtr);
-	else
-		return (size_t)ZSTD_readLE64(memPtr);
-}
-
-ZSTD_STATIC void ZSTD_writeLEST(void *memPtr, size_t val)
-{
-	if (ZSTD_32bits())
-		ZSTD_writeLE32(memPtr, (U32)val);
-	else
-		ZSTD_writeLE64(memPtr, (U64)val);
-}
-
-/*=== Big endian r/w ===*/
-
-ZSTD_STATIC U32 ZSTD_readBE32(const void *memPtr) { return get_unaligned_be32(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeBE32(void *memPtr, U32 val32) { put_unaligned_be32(val32, memPtr); }
-
-ZSTD_STATIC U64 ZSTD_readBE64(const void *memPtr) { return get_unaligned_be64(memPtr); }
-
-ZSTD_STATIC void ZSTD_writeBE64(void *memPtr, U64 val64) { put_unaligned_be64(val64, memPtr); }
-
-ZSTD_STATIC size_t ZSTD_readBEST(const void *memPtr)
-{
-	if (ZSTD_32bits())
-		return (size_t)ZSTD_readBE32(memPtr);
-	else
-		return (size_t)ZSTD_readBE64(memPtr);
-}
-
-ZSTD_STATIC void ZSTD_writeBEST(void *memPtr, size_t val)
-{
-	if (ZSTD_32bits())
-		ZSTD_writeBE32(memPtr, (U32)val);
-	else
-		ZSTD_writeBE64(memPtr, (U64)val);
-}
-
-/* function safe only for comparisons */
-ZSTD_STATIC U32 ZSTD_readMINMATCH(const void *memPtr, U32 length)
-{
-	switch (length) {
-	default:
-	case 4: return ZSTD_read32(memPtr);
-	case 3:
-		if (ZSTD_isLittleEndian())
-			return ZSTD_read32(memPtr) << 8;
-		else
-			return ZSTD_read32(memPtr) >> 8;
-	}
-}
-
-#endif /* MEM_H_MODULE */
diff --git a/lib/zstd/zstd_common.c b/lib/zstd/zstd_common.c
deleted file mode 100644
index a282624..0000000
--- a/lib/zstd/zstd_common.c
+++ /dev/null
@@ -1,75 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-/*-*************************************
-*  Dependencies
-***************************************/
-#include "error_private.h"
-#include "zstd_internal.h" /* declaration of ZSTD_isError, ZSTD_getErrorName, ZSTD_getErrorCode, ZSTD_getErrorString, ZSTD_versionNumber */
-#include <linux/kernel.h>
-
-/*=**************************************************************
-*  Custom allocator
-****************************************************************/
-
-#define stack_push(stack, size)                                 \
-	({                                                      \
-		void *const ptr = ZSTD_PTR_ALIGN((stack)->ptr); \
-		(stack)->ptr = (char *)ptr + (size);            \
-		(stack)->ptr <= (stack)->end ? ptr : NULL;      \
-	})
-
-ZSTD_customMem ZSTD_initStack(void *workspace, size_t workspaceSize)
-{
-	ZSTD_customMem stackMem = {ZSTD_stackAlloc, ZSTD_stackFree, workspace};
-	ZSTD_stack *stack = (ZSTD_stack *)workspace;
-	/* Verify preconditions */
-	if (!workspace || workspaceSize < sizeof(ZSTD_stack) || workspace != ZSTD_PTR_ALIGN(workspace)) {
-		ZSTD_customMem error = {NULL, NULL, NULL};
-		return error;
-	}
-	/* Initialize the stack */
-	stack->ptr = workspace;
-	stack->end = (char *)workspace + workspaceSize;
-	stack_push(stack, sizeof(ZSTD_stack));
-	return stackMem;
-}
-
-void *ZSTD_stackAllocAll(void *opaque, size_t *size)
-{
-	ZSTD_stack *stack = (ZSTD_stack *)opaque;
-	*size = (BYTE const *)stack->end - (BYTE *)ZSTD_PTR_ALIGN(stack->ptr);
-	return stack_push(stack, *size);
-}
-
-void *ZSTD_stackAlloc(void *opaque, size_t size)
-{
-	ZSTD_stack *stack = (ZSTD_stack *)opaque;
-	return stack_push(stack, size);
-}
-void ZSTD_stackFree(void *opaque, void *address)
-{
-	(void)opaque;
-	(void)address;
-}
-
-void *ZSTD_malloc(size_t size, ZSTD_customMem customMem) { return customMem.customAlloc(customMem.opaque, size); }
-
-void ZSTD_free(void *ptr, ZSTD_customMem customMem)
-{
-	if (ptr != NULL)
-		customMem.customFree(customMem.opaque, ptr);
-}
diff --git a/lib/zstd/zstd_compress_module.c b/lib/zstd/zstd_compress_module.c
new file mode 100644
index 0000000..65548a4
--- /dev/null
+++ b/lib/zstd/zstd_compress_module.c
@@ -0,0 +1,160 @@
+// SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/zstd.h>
+
+#include "common/zstd_deps.h"
+#include "common/zstd_internal.h"
+
+#define ZSTD_FORWARD_IF_ERR(ret)            \
+	do {                                \
+		size_t const __ret = (ret); \
+		if (ZSTD_isError(__ret))    \
+			return __ret;       \
+	} while (0)
+
+static size_t zstd_cctx_init(zstd_cctx *cctx, const zstd_parameters *parameters,
+	unsigned long long pledged_src_size)
+{
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_reset(
+		cctx, ZSTD_reset_session_and_parameters));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setPledgedSrcSize(
+		cctx, pledged_src_size));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_windowLog, parameters->cParams.windowLog));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_hashLog, parameters->cParams.hashLog));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_chainLog, parameters->cParams.chainLog));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_searchLog, parameters->cParams.searchLog));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_minMatch, parameters->cParams.minMatch));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_targetLength, parameters->cParams.targetLength));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_strategy, parameters->cParams.strategy));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_contentSizeFlag, parameters->fParams.contentSizeFlag));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_checksumFlag, parameters->fParams.checksumFlag));
+	ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter(
+		cctx, ZSTD_c_dictIDFlag, !parameters->fParams.noDictIDFlag));
+	return 0;
+}
+
+int zstd_min_clevel(void)
+{
+	return ZSTD_minCLevel();
+}
+EXPORT_SYMBOL(zstd_min_clevel);
+
+int zstd_max_clevel(void)
+{
+	return ZSTD_maxCLevel();
+}
+EXPORT_SYMBOL(zstd_max_clevel);
+
+size_t zstd_compress_bound(size_t src_size)
+{
+	return ZSTD_compressBound(src_size);
+}
+EXPORT_SYMBOL(zstd_compress_bound);
+
+zstd_parameters zstd_get_params(int level,
+	unsigned long long estimated_src_size)
+{
+	return ZSTD_getParams(level, estimated_src_size, 0);
+}
+EXPORT_SYMBOL(zstd_get_params);
+
+size_t zstd_cctx_workspace_bound(const zstd_compression_parameters *cparams)
+{
+	return ZSTD_estimateCCtxSize_usingCParams(*cparams);
+}
+EXPORT_SYMBOL(zstd_cctx_workspace_bound);
+
+zstd_cctx *zstd_init_cctx(void *workspace, size_t workspace_size)
+{
+	if (workspace == NULL)
+		return NULL;
+	return ZSTD_initStaticCCtx(workspace, workspace_size);
+}
+EXPORT_SYMBOL(zstd_init_cctx);
+
+size_t zstd_compress_cctx(zstd_cctx *cctx, void *dst, size_t dst_capacity,
+	const void *src, size_t src_size, const zstd_parameters *parameters)
+{
+	ZSTD_FORWARD_IF_ERR(zstd_cctx_init(cctx, parameters, src_size));
+	return ZSTD_compress2(cctx, dst, dst_capacity, src, src_size);
+}
+EXPORT_SYMBOL(zstd_compress_cctx);
+
+size_t zstd_cstream_workspace_bound(const zstd_compression_parameters *cparams)
+{
+	return ZSTD_estimateCStreamSize_usingCParams(*cparams);
+}
+EXPORT_SYMBOL(zstd_cstream_workspace_bound);
+
+zstd_cstream *zstd_init_cstream(const zstd_parameters *parameters,
+	unsigned long long pledged_src_size, void *workspace, size_t workspace_size)
+{
+	zstd_cstream *cstream;
+
+	if (workspace == NULL)
+		return NULL;
+
+	cstream = ZSTD_initStaticCStream(workspace, workspace_size);
+	if (cstream == NULL)
+		return NULL;
+
+	/* 0 means unknown in linux zstd API but means 0 in new zstd API */
+	if (pledged_src_size == 0)
+		pledged_src_size = ZSTD_CONTENTSIZE_UNKNOWN;
+
+	if (ZSTD_isError(zstd_cctx_init(cstream, parameters, pledged_src_size)))
+		return NULL;
+
+	return cstream;
+}
+EXPORT_SYMBOL(zstd_init_cstream);
+
+size_t zstd_reset_cstream(zstd_cstream *cstream,
+	unsigned long long pledged_src_size)
+{
+	return ZSTD_resetCStream(cstream, pledged_src_size);
+}
+EXPORT_SYMBOL(zstd_reset_cstream);
+
+size_t zstd_compress_stream(zstd_cstream *cstream, zstd_out_buffer *output,
+	zstd_in_buffer *input)
+{
+	return ZSTD_compressStream(cstream, output, input);
+}
+EXPORT_SYMBOL(zstd_compress_stream);
+
+size_t zstd_flush_stream(zstd_cstream *cstream, zstd_out_buffer *output)
+{
+	return ZSTD_flushStream(cstream, output);
+}
+EXPORT_SYMBOL(zstd_flush_stream);
+
+size_t zstd_end_stream(zstd_cstream *cstream, zstd_out_buffer *output)
+{
+	return ZSTD_endStream(cstream, output);
+}
+EXPORT_SYMBOL(zstd_end_stream);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("Zstd Compressor");
diff --git a/lib/zstd/zstd_decompress_module.c b/lib/zstd/zstd_decompress_module.c
new file mode 100644
index 0000000..f4ed952
--- /dev/null
+++ b/lib/zstd/zstd_decompress_module.c
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause
+/*
+ * Copyright (c) Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/zstd.h>
+
+#include "common/zstd_deps.h"
+
+/* Common symbols. zstd_compress must depend on zstd_decompress. */
+
+unsigned int zstd_is_error(size_t code)
+{
+	return ZSTD_isError(code);
+}
+EXPORT_SYMBOL(zstd_is_error);
+
+zstd_error_code zstd_get_error_code(size_t code)
+{
+	return ZSTD_getErrorCode(code);
+}
+EXPORT_SYMBOL(zstd_get_error_code);
+
+const char *zstd_get_error_name(size_t code)
+{
+	return ZSTD_getErrorName(code);
+}
+EXPORT_SYMBOL(zstd_get_error_name);
+
+/* Decompression symbols. */
+
+size_t zstd_dctx_workspace_bound(void)
+{
+	return ZSTD_estimateDCtxSize();
+}
+EXPORT_SYMBOL(zstd_dctx_workspace_bound);
+
+zstd_dctx *zstd_init_dctx(void *workspace, size_t workspace_size)
+{
+	if (workspace == NULL)
+		return NULL;
+	return ZSTD_initStaticDCtx(workspace, workspace_size);
+}
+EXPORT_SYMBOL(zstd_init_dctx);
+
+size_t zstd_decompress_dctx(zstd_dctx *dctx, void *dst, size_t dst_capacity,
+	const void *src, size_t src_size)
+{
+	return ZSTD_decompressDCtx(dctx, dst, dst_capacity, src, src_size);
+}
+EXPORT_SYMBOL(zstd_decompress_dctx);
+
+size_t zstd_dstream_workspace_bound(size_t max_window_size)
+{
+	return ZSTD_estimateDStreamSize(max_window_size);
+}
+EXPORT_SYMBOL(zstd_dstream_workspace_bound);
+
+zstd_dstream *zstd_init_dstream(size_t max_window_size, void *workspace,
+	size_t workspace_size)
+{
+	if (workspace == NULL)
+		return NULL;
+	(void)max_window_size;
+	return ZSTD_initStaticDStream(workspace, workspace_size);
+}
+EXPORT_SYMBOL(zstd_init_dstream);
+
+size_t zstd_reset_dstream(zstd_dstream *dstream)
+{
+	return ZSTD_resetDStream(dstream);
+}
+EXPORT_SYMBOL(zstd_reset_dstream);
+
+size_t zstd_decompress_stream(zstd_dstream *dstream, zstd_out_buffer *output,
+	zstd_in_buffer *input)
+{
+	return ZSTD_decompressStream(dstream, output, input);
+}
+EXPORT_SYMBOL(zstd_decompress_stream);
+
+size_t zstd_find_frame_compressed_size(const void *src, size_t src_size)
+{
+	return ZSTD_findFrameCompressedSize(src, src_size);
+}
+EXPORT_SYMBOL(zstd_find_frame_compressed_size);
+
+size_t zstd_get_frame_header(zstd_frame_header *header, const void *src,
+	size_t src_size)
+{
+	return ZSTD_getFrameHeader(header, src, src_size);
+}
+EXPORT_SYMBOL(zstd_get_frame_header);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("Zstd Decompressor");
diff --git a/lib/zstd/zstd_internal.h b/lib/zstd/zstd_internal.h
deleted file mode 100644
index dac7533..0000000
--- a/lib/zstd/zstd_internal.h
+++ /dev/null
@@ -1,273 +0,0 @@
-/**
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-#ifndef ZSTD_CCOMMON_H_MODULE
-#define ZSTD_CCOMMON_H_MODULE
-
-/*-*******************************************************
-*  Compiler specifics
-*********************************************************/
-#define FORCE_INLINE static __always_inline
-#define FORCE_NOINLINE static noinline
-
-/*-*************************************
-*  Dependencies
-***************************************/
-#include "error_private.h"
-#include "mem.h"
-#include <linux/compiler.h>
-#include <linux/kernel.h>
-#include <linux/xxhash.h>
-#include <linux/zstd.h>
-
-/*-*************************************
-*  shared macros
-***************************************/
-#define MIN(a, b) ((a) < (b) ? (a) : (b))
-#define MAX(a, b) ((a) > (b) ? (a) : (b))
-#define CHECK_F(f)                       \
-	{                                \
-		size_t const errcod = f; \
-		if (ERR_isError(errcod)) \
-			return errcod;   \
-	} /* check and Forward error code */
-#define CHECK_E(f, e)                    \
-	{                                \
-		size_t const errcod = f; \
-		if (ERR_isError(errcod)) \
-			return ERROR(e); \
-	} /* check and send Error code */
-#define ZSTD_STATIC_ASSERT(c)                                   \
-	{                                                       \
-		enum { ZSTD_static_assert = 1 / (int)(!!(c)) }; \
-	}
-
-/*-*************************************
-*  Common constants
-***************************************/
-#define ZSTD_OPT_NUM (1 << 12)
-#define ZSTD_DICT_MAGIC 0xEC30A437 /* v0.7+ */
-
-#define ZSTD_REP_NUM 3		      /* number of repcodes */
-#define ZSTD_REP_CHECK (ZSTD_REP_NUM) /* number of repcodes to check by the optimal parser */
-#define ZSTD_REP_MOVE (ZSTD_REP_NUM - 1)
-#define ZSTD_REP_MOVE_OPT (ZSTD_REP_NUM)
-static const U32 repStartValue[ZSTD_REP_NUM] = {1, 4, 8};
-
-#define KB *(1 << 10)
-#define MB *(1 << 20)
-#define GB *(1U << 30)
-
-#define BIT7 128
-#define BIT6 64
-#define BIT5 32
-#define BIT4 16
-#define BIT1 2
-#define BIT0 1
-
-#define ZSTD_WINDOWLOG_ABSOLUTEMIN 10
-static const size_t ZSTD_fcs_fieldSize[4] = {0, 2, 4, 8};
-static const size_t ZSTD_did_fieldSize[4] = {0, 1, 2, 4};
-
-#define ZSTD_BLOCKHEADERSIZE 3 /* C standard doesn't allow `static const` variable to be init using another `static const` variable */
-static const size_t ZSTD_blockHeaderSize = ZSTD_BLOCKHEADERSIZE;
-typedef enum { bt_raw, bt_rle, bt_compressed, bt_reserved } blockType_e;
-
-#define MIN_SEQUENCES_SIZE 1									  /* nbSeq==0 */
-#define MIN_CBLOCK_SIZE (1 /*litCSize*/ + 1 /* RLE or RAW */ + MIN_SEQUENCES_SIZE /* nbSeq==0 */) /* for a non-null block */
-
-#define HufLog 12
-typedef enum { set_basic, set_rle, set_compressed, set_repeat } symbolEncodingType_e;
-
-#define LONGNBSEQ 0x7F00
-
-#define MINMATCH 3
-#define EQUAL_READ32 4
-
-#define Litbits 8
-#define MaxLit ((1 << Litbits) - 1)
-#define MaxML 52
-#define MaxLL 35
-#define MaxOff 28
-#define MaxSeq MAX(MaxLL, MaxML) /* Assumption : MaxOff < MaxLL,MaxML */
-#define MLFSELog 9
-#define LLFSELog 9
-#define OffFSELog 8
-
-static const U32 LL_bits[MaxLL + 1] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
-static const S16 LL_defaultNorm[MaxLL + 1] = {4, 3, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 2, 1, 1, 1, 1, 1, -1, -1, -1, -1};
-#define LL_DEFAULTNORMLOG 6 /* for static allocation */
-static const U32 LL_defaultNormLog = LL_DEFAULTNORMLOG;
-
-static const U32 ML_bits[MaxML + 1] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  0,  0,  0,  0,  0,  0, 0,
-				       0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 4, 4, 5, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
-static const S16 ML_defaultNorm[MaxML + 1] = {1, 4, 3, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  1,  1,  1,  1,  1,  1, 1,
-					      1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, -1, -1, -1, -1, -1, -1, -1};
-#define ML_DEFAULTNORMLOG 6 /* for static allocation */
-static const U32 ML_defaultNormLog = ML_DEFAULTNORMLOG;
-
-static const S16 OF_defaultNorm[MaxOff + 1] = {1, 1, 1, 1, 1, 1, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, -1, -1, -1, -1, -1};
-#define OF_DEFAULTNORMLOG 5 /* for static allocation */
-static const U32 OF_defaultNormLog = OF_DEFAULTNORMLOG;
-
-/*-*******************************************
-*  Shared functions to include for inlining
-*********************************************/
-ZSTD_STATIC void ZSTD_copy8(void *dst, const void *src) {
-	/*
-	 * zstd relies heavily on gcc being able to analyze and inline this
-	 * memcpy() call, since it is called in a tight loop. Preboot mode
-	 * is compiled in freestanding mode, which stops gcc from analyzing
-	 * memcpy(). Use __builtin_memcpy() to tell gcc to analyze this as a
-	 * regular memcpy().
-	 */
-	__builtin_memcpy(dst, src, 8);
-}
-/*! ZSTD_wildcopy() :
-*   custom version of memcpy(), can copy up to 7 bytes too many (8 bytes if length==0) */
-#define WILDCOPY_OVERLENGTH 8
-ZSTD_STATIC void ZSTD_wildcopy(void *dst, const void *src, ptrdiff_t length)
-{
-	const BYTE* ip = (const BYTE*)src;
-	BYTE* op = (BYTE*)dst;
-	BYTE* const oend = op + length;
-#if defined(GCC_VERSION) && GCC_VERSION >= 70000 && GCC_VERSION < 70200
-	/*
-	 * Work around https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81388.
-	 * Avoid the bad case where the loop only runs once by handling the
-	 * special case separately. This doesn't trigger the bug because it
-	 * doesn't involve pointer/integer overflow.
-	 */
-	if (length <= 8)
-		return ZSTD_copy8(dst, src);
-#endif
-	do {
-		ZSTD_copy8(op, ip);
-		op += 8;
-		ip += 8;
-	} while (op < oend);
-}
-
-/*-*******************************************
-*  Private interfaces
-*********************************************/
-typedef struct ZSTD_stats_s ZSTD_stats_t;
-
-typedef struct {
-	U32 off;
-	U32 len;
-} ZSTD_match_t;
-
-typedef struct {
-	U32 price;
-	U32 off;
-	U32 mlen;
-	U32 litlen;
-	U32 rep[ZSTD_REP_NUM];
-} ZSTD_optimal_t;
-
-typedef struct seqDef_s {
-	U32 offset;
-	U16 litLength;
-	U16 matchLength;
-} seqDef;
-
-typedef struct {
-	seqDef *sequencesStart;
-	seqDef *sequences;
-	BYTE *litStart;
-	BYTE *lit;
-	BYTE *llCode;
-	BYTE *mlCode;
-	BYTE *ofCode;
-	U32 longLengthID; /* 0 == no longLength; 1 == Lit.longLength; 2 == Match.longLength; */
-	U32 longLengthPos;
-	/* opt */
-	ZSTD_optimal_t *priceTable;
-	ZSTD_match_t *matchTable;
-	U32 *matchLengthFreq;
-	U32 *litLengthFreq;
-	U32 *litFreq;
-	U32 *offCodeFreq;
-	U32 matchLengthSum;
-	U32 matchSum;
-	U32 litLengthSum;
-	U32 litSum;
-	U32 offCodeSum;
-	U32 log2matchLengthSum;
-	U32 log2matchSum;
-	U32 log2litLengthSum;
-	U32 log2litSum;
-	U32 log2offCodeSum;
-	U32 factor;
-	U32 staticPrices;
-	U32 cachedPrice;
-	U32 cachedLitLength;
-	const BYTE *cachedLiterals;
-} seqStore_t;
-
-const seqStore_t *ZSTD_getSeqStore(const ZSTD_CCtx *ctx);
-void ZSTD_seqToCodes(const seqStore_t *seqStorePtr);
-int ZSTD_isSkipFrame(ZSTD_DCtx *dctx);
-
-/*= Custom memory allocation functions */
-typedef void *(*ZSTD_allocFunction)(void *opaque, size_t size);
-typedef void (*ZSTD_freeFunction)(void *opaque, void *address);
-typedef struct {
-	ZSTD_allocFunction customAlloc;
-	ZSTD_freeFunction customFree;
-	void *opaque;
-} ZSTD_customMem;
-
-void *ZSTD_malloc(size_t size, ZSTD_customMem customMem);
-void ZSTD_free(void *ptr, ZSTD_customMem customMem);
-
-/*====== stack allocation  ======*/
-
-typedef struct {
-	void *ptr;
-	const void *end;
-} ZSTD_stack;
-
-#define ZSTD_ALIGN(x) ALIGN(x, sizeof(size_t))
-#define ZSTD_PTR_ALIGN(p) PTR_ALIGN(p, sizeof(size_t))
-
-ZSTD_customMem ZSTD_initStack(void *workspace, size_t workspaceSize);
-
-void *ZSTD_stackAllocAll(void *opaque, size_t *size);
-void *ZSTD_stackAlloc(void *opaque, size_t size);
-void ZSTD_stackFree(void *opaque, void *address);
-
-/*======  common function  ======*/
-
-ZSTD_STATIC U32 ZSTD_highbit32(U32 val) { return 31 - __builtin_clz(val); }
-
-/* hidden functions */
-
-/* ZSTD_invalidateRepCodes() :
- * ensures next compression will not use repcodes from previous block.
- * Note : only works with regular variant;
- *        do not use with extDict variant ! */
-void ZSTD_invalidateRepCodes(ZSTD_CCtx *cctx);
-
-size_t ZSTD_freeCCtx(ZSTD_CCtx *cctx);
-size_t ZSTD_freeDCtx(ZSTD_DCtx *dctx);
-size_t ZSTD_freeCDict(ZSTD_CDict *cdict);
-size_t ZSTD_freeDDict(ZSTD_DDict *cdict);
-size_t ZSTD_freeCStream(ZSTD_CStream *zcs);
-size_t ZSTD_freeDStream(ZSTD_DStream *zds);
-
-#endif /* ZSTD_CCOMMON_H_MODULE */
diff --git a/lib/zstd/zstd_opt.h b/lib/zstd/zstd_opt.h
deleted file mode 100644
index 55e1b4c..0000000
--- a/lib/zstd/zstd_opt.h
+++ /dev/null
@@ -1,1014 +0,0 @@
-/**
- * Copyright (c) 2016-present, Przemyslaw Skibinski, Yann Collet, Facebook, Inc.
- * All rights reserved.
- *
- * This source code is licensed under the BSD-style license found in the
- * LICENSE file in the root directory of https://github.com/facebook/zstd.
- * An additional grant of patent rights can be found in the PATENTS file in the
- * same directory.
- *
- * This program is free software; you can redistribute it and/or modify it under
- * the terms of the GNU General Public License version 2 as published by the
- * Free Software Foundation. This program is dual-licensed; you may select
- * either version 2 of the GNU General Public License ("GPL") or BSD license
- * ("BSD").
- */
-
-/* Note : this file is intended to be included within zstd_compress.c */
-
-#ifndef ZSTD_OPT_H_91842398743
-#define ZSTD_OPT_H_91842398743
-
-#define ZSTD_LITFREQ_ADD 2
-#define ZSTD_FREQ_DIV 4
-#define ZSTD_MAX_PRICE (1 << 30)
-
-/*-*************************************
-*  Price functions for optimal parser
-***************************************/
-FORCE_INLINE void ZSTD_setLog2Prices(seqStore_t *ssPtr)
-{
-	ssPtr->log2matchLengthSum = ZSTD_highbit32(ssPtr->matchLengthSum + 1);
-	ssPtr->log2litLengthSum = ZSTD_highbit32(ssPtr->litLengthSum + 1);
-	ssPtr->log2litSum = ZSTD_highbit32(ssPtr->litSum + 1);
-	ssPtr->log2offCodeSum = ZSTD_highbit32(ssPtr->offCodeSum + 1);
-	ssPtr->factor = 1 + ((ssPtr->litSum >> 5) / ssPtr->litLengthSum) + ((ssPtr->litSum << 1) / (ssPtr->litSum + ssPtr->matchSum));
-}
-
-ZSTD_STATIC void ZSTD_rescaleFreqs(seqStore_t *ssPtr, const BYTE *src, size_t srcSize)
-{
-	unsigned u;
-
-	ssPtr->cachedLiterals = NULL;
-	ssPtr->cachedPrice = ssPtr->cachedLitLength = 0;
-	ssPtr->staticPrices = 0;
-
-	if (ssPtr->litLengthSum == 0) {
-		if (srcSize <= 1024)
-			ssPtr->staticPrices = 1;
-
-		for (u = 0; u <= MaxLit; u++)
-			ssPtr->litFreq[u] = 0;
-		for (u = 0; u < srcSize; u++)
-			ssPtr->litFreq[src[u]]++;
-
-		ssPtr->litSum = 0;
-		ssPtr->litLengthSum = MaxLL + 1;
-		ssPtr->matchLengthSum = MaxML + 1;
-		ssPtr->offCodeSum = (MaxOff + 1);
-		ssPtr->matchSum = (ZSTD_LITFREQ_ADD << Litbits);
-
-		for (u = 0; u <= MaxLit; u++) {
-			ssPtr->litFreq[u] = 1 + (ssPtr->litFreq[u] >> ZSTD_FREQ_DIV);
-			ssPtr->litSum += ssPtr->litFreq[u];
-		}
-		for (u = 0; u <= MaxLL; u++)
-			ssPtr->litLengthFreq[u] = 1;
-		for (u = 0; u <= MaxML; u++)
-			ssPtr->matchLengthFreq[u] = 1;
-		for (u = 0; u <= MaxOff; u++)
-			ssPtr->offCodeFreq[u] = 1;
-	} else {
-		ssPtr->matchLengthSum = 0;
-		ssPtr->litLengthSum = 0;
-		ssPtr->offCodeSum = 0;
-		ssPtr->matchSum = 0;
-		ssPtr->litSum = 0;
-
-		for (u = 0; u <= MaxLit; u++) {
-			ssPtr->litFreq[u] = 1 + (ssPtr->litFreq[u] >> (ZSTD_FREQ_DIV + 1));
-			ssPtr->litSum += ssPtr->litFreq[u];
-		}
-		for (u = 0; u <= MaxLL; u++) {
-			ssPtr->litLengthFreq[u] = 1 + (ssPtr->litLengthFreq[u] >> (ZSTD_FREQ_DIV + 1));
-			ssPtr->litLengthSum += ssPtr->litLengthFreq[u];
-		}
-		for (u = 0; u <= MaxML; u++) {
-			ssPtr->matchLengthFreq[u] = 1 + (ssPtr->matchLengthFreq[u] >> ZSTD_FREQ_DIV);
-			ssPtr->matchLengthSum += ssPtr->matchLengthFreq[u];
-			ssPtr->matchSum += ssPtr->matchLengthFreq[u] * (u + 3);
-		}
-		ssPtr->matchSum *= ZSTD_LITFREQ_ADD;
-		for (u = 0; u <= MaxOff; u++) {
-			ssPtr->offCodeFreq[u] = 1 + (ssPtr->offCodeFreq[u] >> ZSTD_FREQ_DIV);
-			ssPtr->offCodeSum += ssPtr->offCodeFreq[u];
-		}
-	}
-
-	ZSTD_setLog2Prices(ssPtr);
-}
-
-FORCE_INLINE U32 ZSTD_getLiteralPrice(seqStore_t *ssPtr, U32 litLength, const BYTE *literals)
-{
-	U32 price, u;
-
-	if (ssPtr->staticPrices)
-		return ZSTD_highbit32((U32)litLength + 1) + (litLength * 6);
-
-	if (litLength == 0)
-		return ssPtr->log2litLengthSum - ZSTD_highbit32(ssPtr->litLengthFreq[0] + 1);
-
-	/* literals */
-	if (ssPtr->cachedLiterals == literals) {
-		U32 const additional = litLength - ssPtr->cachedLitLength;
-		const BYTE *literals2 = ssPtr->cachedLiterals + ssPtr->cachedLitLength;
-		price = ssPtr->cachedPrice + additional * ssPtr->log2litSum;
-		for (u = 0; u < additional; u++)
-			price -= ZSTD_highbit32(ssPtr->litFreq[literals2[u]] + 1);
-		ssPtr->cachedPrice = price;
-		ssPtr->cachedLitLength = litLength;
-	} else {
-		price = litLength * ssPtr->log2litSum;
-		for (u = 0; u < litLength; u++)
-			price -= ZSTD_highbit32(ssPtr->litFreq[literals[u]] + 1);
-
-		if (litLength >= 12) {
-			ssPtr->cachedLiterals = literals;
-			ssPtr->cachedPrice = price;
-			ssPtr->cachedLitLength = litLength;
-		}
-	}
-
-	/* literal Length */
-	{
-		const BYTE LL_deltaCode = 19;
-		const BYTE llCode = (litLength > 63) ? (BYTE)ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength];
-		price += LL_bits[llCode] + ssPtr->log2litLengthSum - ZSTD_highbit32(ssPtr->litLengthFreq[llCode] + 1);
-	}
-
-	return price;
-}
-
-FORCE_INLINE U32 ZSTD_getPrice(seqStore_t *seqStorePtr, U32 litLength, const BYTE *literals, U32 offset, U32 matchLength, const int ultra)
-{
-	/* offset */
-	U32 price;
-	BYTE const offCode = (BYTE)ZSTD_highbit32(offset + 1);
-
-	if (seqStorePtr->staticPrices)
-		return ZSTD_getLiteralPrice(seqStorePtr, litLength, literals) + ZSTD_highbit32((U32)matchLength + 1) + 16 + offCode;
-
-	price = offCode + seqStorePtr->log2offCodeSum - ZSTD_highbit32(seqStorePtr->offCodeFreq[offCode] + 1);
-	if (!ultra && offCode >= 20)
-		price += (offCode - 19) * 2;
-
-	/* match Length */
-	{
-		const BYTE ML_deltaCode = 36;
-		const BYTE mlCode = (matchLength > 127) ? (BYTE)ZSTD_highbit32(matchLength) + ML_deltaCode : ML_Code[matchLength];
-		price += ML_bits[mlCode] + seqStorePtr->log2matchLengthSum - ZSTD_highbit32(seqStorePtr->matchLengthFreq[mlCode] + 1);
-	}
-
-	return price + ZSTD_getLiteralPrice(seqStorePtr, litLength, literals) + seqStorePtr->factor;
-}
-
-ZSTD_STATIC void ZSTD_updatePrice(seqStore_t *seqStorePtr, U32 litLength, const BYTE *literals, U32 offset, U32 matchLength)
-{
-	U32 u;
-
-	/* literals */
-	seqStorePtr->litSum += litLength * ZSTD_LITFREQ_ADD;
-	for (u = 0; u < litLength; u++)
-		seqStorePtr->litFreq[literals[u]] += ZSTD_LITFREQ_ADD;
-
-	/* literal Length */
-	{
-		const BYTE LL_deltaCode = 19;
-		const BYTE llCode = (litLength > 63) ? (BYTE)ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength];
-		seqStorePtr->litLengthFreq[llCode]++;
-		seqStorePtr->litLengthSum++;
-	}
-
-	/* match offset */
-	{
-		BYTE const offCode = (BYTE)ZSTD_highbit32(offset + 1);
-		seqStorePtr->offCodeSum++;
-		seqStorePtr->offCodeFreq[offCode]++;
-	}
-
-	/* match Length */
-	{
-		const BYTE ML_deltaCode = 36;
-		const BYTE mlCode = (matchLength > 127) ? (BYTE)ZSTD_highbit32(matchLength) + ML_deltaCode : ML_Code[matchLength];
-		seqStorePtr->matchLengthFreq[mlCode]++;
-		seqStorePtr->matchLengthSum++;
-	}
-
-	ZSTD_setLog2Prices(seqStorePtr);
-}
-
-#define SET_PRICE(pos, mlen_, offset_, litlen_, price_)           \
-	{                                                         \
-		while (last_pos < pos) {                          \
-			opt[last_pos + 1].price = ZSTD_MAX_PRICE; \
-			last_pos++;                               \
-		}                                                 \
-		opt[pos].mlen = mlen_;                            \
-		opt[pos].off = offset_;                           \
-		opt[pos].litlen = litlen_;                        \
-		opt[pos].price = price_;                          \
-	}
-
-/* Update hashTable3 up to ip (excluded)
-   Assumption : always within prefix (i.e. not within extDict) */
-FORCE_INLINE
-U32 ZSTD_insertAndFindFirstIndexHash3(ZSTD_CCtx *zc, const BYTE *ip)
-{
-	U32 *const hashTable3 = zc->hashTable3;
-	U32 const hashLog3 = zc->hashLog3;
-	const BYTE *const base = zc->base;
-	U32 idx = zc->nextToUpdate3;
-	const U32 target = zc->nextToUpdate3 = (U32)(ip - base);
-	const size_t hash3 = ZSTD_hash3Ptr(ip, hashLog3);
-
-	while (idx < target) {
-		hashTable3[ZSTD_hash3Ptr(base + idx, hashLog3)] = idx;
-		idx++;
-	}
-
-	return hashTable3[hash3];
-}
-
-/*-*************************************
-*  Binary Tree search
-***************************************/
-static U32 ZSTD_insertBtAndGetAllMatches(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, U32 nbCompares, const U32 mls, U32 extDict,
-					 ZSTD_match_t *matches, const U32 minMatchLen)
-{
-	const BYTE *const base = zc->base;
-	const U32 curr = (U32)(ip - base);
-	const U32 hashLog = zc->params.cParams.hashLog;
-	const size_t h = ZSTD_hashPtr(ip, hashLog, mls);
-	U32 *const hashTable = zc->hashTable;
-	U32 matchIndex = hashTable[h];
-	U32 *const bt = zc->chainTable;
-	const U32 btLog = zc->params.cParams.chainLog - 1;
-	const U32 btMask = (1U << btLog) - 1;
-	size_t commonLengthSmaller = 0, commonLengthLarger = 0;
-	const BYTE *const dictBase = zc->dictBase;
-	const U32 dictLimit = zc->dictLimit;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-	const BYTE *const prefixStart = base + dictLimit;
-	const U32 btLow = btMask >= curr ? 0 : curr - btMask;
-	const U32 windowLow = zc->lowLimit;
-	U32 *smallerPtr = bt + 2 * (curr & btMask);
-	U32 *largerPtr = bt + 2 * (curr & btMask) + 1;
-	U32 matchEndIdx = curr + 8;
-	U32 dummy32; /* to be nullified at the end */
-	U32 mnum = 0;
-
-	const U32 minMatch = (mls == 3) ? 3 : 4;
-	size_t bestLength = minMatchLen - 1;
-
-	if (minMatch == 3) { /* HC3 match finder */
-		U32 const matchIndex3 = ZSTD_insertAndFindFirstIndexHash3(zc, ip);
-		if (matchIndex3 > windowLow && (curr - matchIndex3 < (1 << 18))) {
-			const BYTE *match;
-			size_t currMl = 0;
-			if ((!extDict) || matchIndex3 >= dictLimit) {
-				match = base + matchIndex3;
-				if (match[bestLength] == ip[bestLength])
-					currMl = ZSTD_count(ip, match, iLimit);
-			} else {
-				match = dictBase + matchIndex3;
-				if (ZSTD_readMINMATCH(match, MINMATCH) ==
-				    ZSTD_readMINMATCH(ip, MINMATCH)) /* assumption : matchIndex3 <= dictLimit-4 (by table construction) */
-					currMl = ZSTD_count_2segments(ip + MINMATCH, match + MINMATCH, iLimit, dictEnd, prefixStart) + MINMATCH;
-			}
-
-			/* save best solution */
-			if (currMl > bestLength) {
-				bestLength = currMl;
-				matches[mnum].off = ZSTD_REP_MOVE_OPT + curr - matchIndex3;
-				matches[mnum].len = (U32)currMl;
-				mnum++;
-				if (currMl > ZSTD_OPT_NUM)
-					goto update;
-				if (ip + currMl == iLimit)
-					goto update; /* best possible, and avoid read overflow*/
-			}
-		}
-	}
-
-	hashTable[h] = curr; /* Update Hash Table */
-
-	while (nbCompares-- && (matchIndex > windowLow)) {
-		U32 *nextPtr = bt + 2 * (matchIndex & btMask);
-		size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
-		const BYTE *match;
-
-		if ((!extDict) || (matchIndex + matchLength >= dictLimit)) {
-			match = base + matchIndex;
-			if (match[matchLength] == ip[matchLength]) {
-				matchLength += ZSTD_count(ip + matchLength + 1, match + matchLength + 1, iLimit) + 1;
-			}
-		} else {
-			match = dictBase + matchIndex;
-			matchLength += ZSTD_count_2segments(ip + matchLength, match + matchLength, iLimit, dictEnd, prefixStart);
-			if (matchIndex + matchLength >= dictLimit)
-				match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
-		}
-
-		if (matchLength > bestLength) {
-			if (matchLength > matchEndIdx - matchIndex)
-				matchEndIdx = matchIndex + (U32)matchLength;
-			bestLength = matchLength;
-			matches[mnum].off = ZSTD_REP_MOVE_OPT + curr - matchIndex;
-			matches[mnum].len = (U32)matchLength;
-			mnum++;
-			if (matchLength > ZSTD_OPT_NUM)
-				break;
-			if (ip + matchLength == iLimit) /* equal : no way to know if inf or sup */
-				break;			/* drop, to guarantee consistency (miss a little bit of compression) */
-		}
-
-		if (match[matchLength] < ip[matchLength]) {
-			/* match is smaller than curr */
-			*smallerPtr = matchIndex;	  /* update smaller idx */
-			commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
-			if (matchIndex <= btLow) {
-				smallerPtr = &dummy32;
-				break;
-			}			  /* beyond tree size, stop the search */
-			smallerPtr = nextPtr + 1; /* new "smaller" => larger of match */
-			matchIndex = nextPtr[1];  /* new matchIndex larger than previous (closer to curr) */
-		} else {
-			/* match is larger than curr */
-			*largerPtr = matchIndex;
-			commonLengthLarger = matchLength;
-			if (matchIndex <= btLow) {
-				largerPtr = &dummy32;
-				break;
-			} /* beyond tree size, stop the search */
-			largerPtr = nextPtr;
-			matchIndex = nextPtr[0];
-		}
-	}
-
-	*smallerPtr = *largerPtr = 0;
-
-update:
-	zc->nextToUpdate = (matchEndIdx > curr + 8) ? matchEndIdx - 8 : curr + 1;
-	return mnum;
-}
-
-/** Tree updater, providing best match */
-static U32 ZSTD_BtGetAllMatches(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, const U32 maxNbAttempts, const U32 mls, ZSTD_match_t *matches,
-				const U32 minMatchLen)
-{
-	if (ip < zc->base + zc->nextToUpdate)
-		return 0; /* skipped area */
-	ZSTD_updateTree(zc, ip, iLimit, maxNbAttempts, mls);
-	return ZSTD_insertBtAndGetAllMatches(zc, ip, iLimit, maxNbAttempts, mls, 0, matches, minMatchLen);
-}
-
-static U32 ZSTD_BtGetAllMatches_selectMLS(ZSTD_CCtx *zc, /* Index table will be updated */
-					  const BYTE *ip, const BYTE *const iHighLimit, const U32 maxNbAttempts, const U32 matchLengthSearch,
-					  ZSTD_match_t *matches, const U32 minMatchLen)
-{
-	switch (matchLengthSearch) {
-	case 3: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 3, matches, minMatchLen);
-	default:
-	case 4: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 4, matches, minMatchLen);
-	case 5: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 5, matches, minMatchLen);
-	case 7:
-	case 6: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 6, matches, minMatchLen);
-	}
-}
-
-/** Tree updater, providing best match */
-static U32 ZSTD_BtGetAllMatches_extDict(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, const U32 maxNbAttempts, const U32 mls,
-					ZSTD_match_t *matches, const U32 minMatchLen)
-{
-	if (ip < zc->base + zc->nextToUpdate)
-		return 0; /* skipped area */
-	ZSTD_updateTree_extDict(zc, ip, iLimit, maxNbAttempts, mls);
-	return ZSTD_insertBtAndGetAllMatches(zc, ip, iLimit, maxNbAttempts, mls, 1, matches, minMatchLen);
-}
-
-static U32 ZSTD_BtGetAllMatches_selectMLS_extDict(ZSTD_CCtx *zc, /* Index table will be updated */
-						  const BYTE *ip, const BYTE *const iHighLimit, const U32 maxNbAttempts, const U32 matchLengthSearch,
-						  ZSTD_match_t *matches, const U32 minMatchLen)
-{
-	switch (matchLengthSearch) {
-	case 3: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 3, matches, minMatchLen);
-	default:
-	case 4: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 4, matches, minMatchLen);
-	case 5: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 5, matches, minMatchLen);
-	case 7:
-	case 6: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 6, matches, minMatchLen);
-	}
-}
-
-/*-*******************************
-*  Optimal parser
-*********************************/
-FORCE_INLINE
-void ZSTD_compressBlock_opt_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const int ultra)
-{
-	seqStore_t *seqStorePtr = &(ctx->seqStore);
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - 8;
-	const BYTE *const base = ctx->base;
-	const BYTE *const prefixStart = base + ctx->dictLimit;
-
-	const U32 maxSearches = 1U << ctx->params.cParams.searchLog;
-	const U32 sufficient_len = ctx->params.cParams.targetLength;
-	const U32 mls = ctx->params.cParams.searchLength;
-	const U32 minMatch = (ctx->params.cParams.searchLength == 3) ? 3 : 4;
-
-	ZSTD_optimal_t *opt = seqStorePtr->priceTable;
-	ZSTD_match_t *matches = seqStorePtr->matchTable;
-	const BYTE *inr;
-	U32 offset, rep[ZSTD_REP_NUM];
-
-	/* init */
-	ctx->nextToUpdate3 = ctx->nextToUpdate;
-	ZSTD_rescaleFreqs(seqStorePtr, (const BYTE *)src, srcSize);
-	ip += (ip == prefixStart);
-	{
-		U32 i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			rep[i] = ctx->rep[i];
-	}
-
-	/* Match Loop */
-	while (ip < ilimit) {
-		U32 cur, match_num, last_pos, litlen, price;
-		U32 u, mlen, best_mlen, best_off, litLength;
-		memset(opt, 0, sizeof(ZSTD_optimal_t));
-		last_pos = 0;
-		litlen = (U32)(ip - anchor);
-
-		/* check repCode */
-		{
-			U32 i, last_i = ZSTD_REP_CHECK + (ip == anchor);
-			for (i = (ip == anchor); i < last_i; i++) {
-				const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : rep[i];
-				if ((repCur > 0) && (repCur < (S32)(ip - prefixStart)) &&
-				    (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(ip - repCur, minMatch))) {
-					mlen = (U32)ZSTD_count(ip + minMatch, ip + minMatch - repCur, iend) + minMatch;
-					if (mlen > sufficient_len || mlen >= ZSTD_OPT_NUM) {
-						best_mlen = mlen;
-						best_off = i;
-						cur = 0;
-						last_pos = 1;
-						goto _storeSequence;
-					}
-					best_off = i - (ip == anchor);
-					do {
-						price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
-						if (mlen > last_pos || price < opt[mlen].price)
-							SET_PRICE(mlen, mlen, i, litlen, price); /* note : macro modifies last_pos */
-						mlen--;
-					} while (mlen >= minMatch);
-				}
-			}
-		}
-
-		match_num = ZSTD_BtGetAllMatches_selectMLS(ctx, ip, iend, maxSearches, mls, matches, minMatch);
-
-		if (!last_pos && !match_num) {
-			ip++;
-			continue;
-		}
-
-		if (match_num && (matches[match_num - 1].len > sufficient_len || matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
-			best_mlen = matches[match_num - 1].len;
-			best_off = matches[match_num - 1].off;
-			cur = 0;
-			last_pos = 1;
-			goto _storeSequence;
-		}
-
-		/* set prices using matches at position = 0 */
-		best_mlen = (last_pos) ? last_pos : minMatch;
-		for (u = 0; u < match_num; u++) {
-			mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
-			best_mlen = matches[u].len;
-			while (mlen <= best_mlen) {
-				price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
-				if (mlen > last_pos || price < opt[mlen].price)
-					SET_PRICE(mlen, mlen, matches[u].off, litlen, price); /* note : macro modifies last_pos */
-				mlen++;
-			}
-		}
-
-		if (last_pos < minMatch) {
-			ip++;
-			continue;
-		}
-
-		/* initialize opt[0] */
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				opt[0].rep[i] = rep[i];
-		}
-		opt[0].mlen = 1;
-		opt[0].litlen = litlen;
-
-		/* check further positions */
-		for (cur = 1; cur <= last_pos; cur++) {
-			inr = ip + cur;
-
-			if (opt[cur - 1].mlen == 1) {
-				litlen = opt[cur - 1].litlen + 1;
-				if (cur > litlen) {
-					price = opt[cur - litlen].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - litlen);
-				} else
-					price = ZSTD_getLiteralPrice(seqStorePtr, litlen, anchor);
-			} else {
-				litlen = 1;
-				price = opt[cur - 1].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - 1);
-			}
-
-			if (cur > last_pos || price <= opt[cur].price)
-				SET_PRICE(cur, 1, 0, litlen, price);
-
-			if (cur == last_pos)
-				break;
-
-			if (inr > ilimit) /* last match must start at a minimum distance of 8 from oend */
-				continue;
-
-			mlen = opt[cur].mlen;
-			if (opt[cur].off > ZSTD_REP_MOVE_OPT) {
-				opt[cur].rep[2] = opt[cur - mlen].rep[1];
-				opt[cur].rep[1] = opt[cur - mlen].rep[0];
-				opt[cur].rep[0] = opt[cur].off - ZSTD_REP_MOVE_OPT;
-			} else {
-				opt[cur].rep[2] = (opt[cur].off > 1) ? opt[cur - mlen].rep[1] : opt[cur - mlen].rep[2];
-				opt[cur].rep[1] = (opt[cur].off > 0) ? opt[cur - mlen].rep[0] : opt[cur - mlen].rep[1];
-				opt[cur].rep[0] =
-				    ((opt[cur].off == ZSTD_REP_MOVE_OPT) && (mlen != 1)) ? (opt[cur - mlen].rep[0] - 1) : (opt[cur - mlen].rep[opt[cur].off]);
-			}
-
-			best_mlen = minMatch;
-			{
-				U32 i, last_i = ZSTD_REP_CHECK + (mlen != 1);
-				for (i = (opt[cur].mlen != 1); i < last_i; i++) { /* check rep */
-					const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (opt[cur].rep[0] - 1) : opt[cur].rep[i];
-					if ((repCur > 0) && (repCur < (S32)(inr - prefixStart)) &&
-					    (ZSTD_readMINMATCH(inr, minMatch) == ZSTD_readMINMATCH(inr - repCur, minMatch))) {
-						mlen = (U32)ZSTD_count(inr + minMatch, inr + minMatch - repCur, iend) + minMatch;
-
-						if (mlen > sufficient_len || cur + mlen >= ZSTD_OPT_NUM) {
-							best_mlen = mlen;
-							best_off = i;
-							last_pos = cur + 1;
-							goto _storeSequence;
-						}
-
-						best_off = i - (opt[cur].mlen != 1);
-						if (mlen > best_mlen)
-							best_mlen = mlen;
-
-						do {
-							if (opt[cur].mlen == 1) {
-								litlen = opt[cur].litlen;
-								if (cur > litlen) {
-									price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, inr - litlen,
-															best_off, mlen - MINMATCH, ultra);
-								} else
-									price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
-							} else {
-								litlen = 0;
-								price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, best_off, mlen - MINMATCH, ultra);
-							}
-
-							if (cur + mlen > last_pos || price <= opt[cur + mlen].price)
-								SET_PRICE(cur + mlen, mlen, i, litlen, price);
-							mlen--;
-						} while (mlen >= minMatch);
-					}
-				}
-			}
-
-			match_num = ZSTD_BtGetAllMatches_selectMLS(ctx, inr, iend, maxSearches, mls, matches, best_mlen);
-
-			if (match_num > 0 && (matches[match_num - 1].len > sufficient_len || cur + matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
-				best_mlen = matches[match_num - 1].len;
-				best_off = matches[match_num - 1].off;
-				last_pos = cur + 1;
-				goto _storeSequence;
-			}
-
-			/* set prices using matches at position = cur */
-			for (u = 0; u < match_num; u++) {
-				mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
-				best_mlen = matches[u].len;
-
-				while (mlen <= best_mlen) {
-					if (opt[cur].mlen == 1) {
-						litlen = opt[cur].litlen;
-						if (cur > litlen)
-							price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, ip + cur - litlen,
-													matches[u].off - 1, mlen - MINMATCH, ultra);
-						else
-							price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
-					} else {
-						litlen = 0;
-						price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, matches[u].off - 1, mlen - MINMATCH, ultra);
-					}
-
-					if (cur + mlen > last_pos || (price < opt[cur + mlen].price))
-						SET_PRICE(cur + mlen, mlen, matches[u].off, litlen, price);
-
-					mlen++;
-				}
-			}
-		}
-
-		best_mlen = opt[last_pos].mlen;
-		best_off = opt[last_pos].off;
-		cur = last_pos - best_mlen;
-
-	/* store sequence */
-_storeSequence: /* cur, last_pos, best_mlen, best_off have to be set */
-		opt[0].mlen = 1;
-
-		while (1) {
-			mlen = opt[cur].mlen;
-			offset = opt[cur].off;
-			opt[cur].mlen = best_mlen;
-			opt[cur].off = best_off;
-			best_mlen = mlen;
-			best_off = offset;
-			if (mlen > cur)
-				break;
-			cur -= mlen;
-		}
-
-		for (u = 0; u <= last_pos;) {
-			u += opt[u].mlen;
-		}
-
-		for (cur = 0; cur < last_pos;) {
-			mlen = opt[cur].mlen;
-			if (mlen == 1) {
-				ip++;
-				cur++;
-				continue;
-			}
-			offset = opt[cur].off;
-			cur += mlen;
-			litLength = (U32)(ip - anchor);
-
-			if (offset > ZSTD_REP_MOVE_OPT) {
-				rep[2] = rep[1];
-				rep[1] = rep[0];
-				rep[0] = offset - ZSTD_REP_MOVE_OPT;
-				offset--;
-			} else {
-				if (offset != 0) {
-					best_off = (offset == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : (rep[offset]);
-					if (offset != 1)
-						rep[2] = rep[1];
-					rep[1] = rep[0];
-					rep[0] = best_off;
-				}
-				if (litLength == 0)
-					offset--;
-			}
-
-			ZSTD_updatePrice(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
-			ZSTD_storeSeq(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
-			anchor = ip = ip + mlen;
-		}
-	} /* for (cur=0; cur < last_pos; ) */
-
-	/* Save reps for next block */
-	{
-		int i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			ctx->repToConfirm[i] = rep[i];
-	}
-
-	/* Last Literals */
-	{
-		size_t const lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-FORCE_INLINE
-void ZSTD_compressBlock_opt_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const int ultra)
-{
-	seqStore_t *seqStorePtr = &(ctx->seqStore);
-	const BYTE *const istart = (const BYTE *)src;
-	const BYTE *ip = istart;
-	const BYTE *anchor = istart;
-	const BYTE *const iend = istart + srcSize;
-	const BYTE *const ilimit = iend - 8;
-	const BYTE *const base = ctx->base;
-	const U32 lowestIndex = ctx->lowLimit;
-	const U32 dictLimit = ctx->dictLimit;
-	const BYTE *const prefixStart = base + dictLimit;
-	const BYTE *const dictBase = ctx->dictBase;
-	const BYTE *const dictEnd = dictBase + dictLimit;
-
-	const U32 maxSearches = 1U << ctx->params.cParams.searchLog;
-	const U32 sufficient_len = ctx->params.cParams.targetLength;
-	const U32 mls = ctx->params.cParams.searchLength;
-	const U32 minMatch = (ctx->params.cParams.searchLength == 3) ? 3 : 4;
-
-	ZSTD_optimal_t *opt = seqStorePtr->priceTable;
-	ZSTD_match_t *matches = seqStorePtr->matchTable;
-	const BYTE *inr;
-
-	/* init */
-	U32 offset, rep[ZSTD_REP_NUM];
-	{
-		U32 i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			rep[i] = ctx->rep[i];
-	}
-
-	ctx->nextToUpdate3 = ctx->nextToUpdate;
-	ZSTD_rescaleFreqs(seqStorePtr, (const BYTE *)src, srcSize);
-	ip += (ip == prefixStart);
-
-	/* Match Loop */
-	while (ip < ilimit) {
-		U32 cur, match_num, last_pos, litlen, price;
-		U32 u, mlen, best_mlen, best_off, litLength;
-		U32 curr = (U32)(ip - base);
-		memset(opt, 0, sizeof(ZSTD_optimal_t));
-		last_pos = 0;
-		opt[0].litlen = (U32)(ip - anchor);
-
-		/* check repCode */
-		{
-			U32 i, last_i = ZSTD_REP_CHECK + (ip == anchor);
-			for (i = (ip == anchor); i < last_i; i++) {
-				const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : rep[i];
-				const U32 repIndex = (U32)(curr - repCur);
-				const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
-				const BYTE *const repMatch = repBase + repIndex;
-				if ((repCur > 0 && repCur <= (S32)curr) &&
-				    (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
-				    && (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch))) {
-					/* repcode detected we should take it */
-					const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
-					mlen = (U32)ZSTD_count_2segments(ip + minMatch, repMatch + minMatch, iend, repEnd, prefixStart) + minMatch;
-
-					if (mlen > sufficient_len || mlen >= ZSTD_OPT_NUM) {
-						best_mlen = mlen;
-						best_off = i;
-						cur = 0;
-						last_pos = 1;
-						goto _storeSequence;
-					}
-
-					best_off = i - (ip == anchor);
-					litlen = opt[0].litlen;
-					do {
-						price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
-						if (mlen > last_pos || price < opt[mlen].price)
-							SET_PRICE(mlen, mlen, i, litlen, price); /* note : macro modifies last_pos */
-						mlen--;
-					} while (mlen >= minMatch);
-				}
-			}
-		}
-
-		match_num = ZSTD_BtGetAllMatches_selectMLS_extDict(ctx, ip, iend, maxSearches, mls, matches, minMatch); /* first search (depth 0) */
-
-		if (!last_pos && !match_num) {
-			ip++;
-			continue;
-		}
-
-		{
-			U32 i;
-			for (i = 0; i < ZSTD_REP_NUM; i++)
-				opt[0].rep[i] = rep[i];
-		}
-		opt[0].mlen = 1;
-
-		if (match_num && (matches[match_num - 1].len > sufficient_len || matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
-			best_mlen = matches[match_num - 1].len;
-			best_off = matches[match_num - 1].off;
-			cur = 0;
-			last_pos = 1;
-			goto _storeSequence;
-		}
-
-		best_mlen = (last_pos) ? last_pos : minMatch;
-
-		/* set prices using matches at position = 0 */
-		for (u = 0; u < match_num; u++) {
-			mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
-			best_mlen = matches[u].len;
-			litlen = opt[0].litlen;
-			while (mlen <= best_mlen) {
-				price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
-				if (mlen > last_pos || price < opt[mlen].price)
-					SET_PRICE(mlen, mlen, matches[u].off, litlen, price);
-				mlen++;
-			}
-		}
-
-		if (last_pos < minMatch) {
-			ip++;
-			continue;
-		}
-
-		/* check further positions */
-		for (cur = 1; cur <= last_pos; cur++) {
-			inr = ip + cur;
-
-			if (opt[cur - 1].mlen == 1) {
-				litlen = opt[cur - 1].litlen + 1;
-				if (cur > litlen) {
-					price = opt[cur - litlen].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - litlen);
-				} else
-					price = ZSTD_getLiteralPrice(seqStorePtr, litlen, anchor);
-			} else {
-				litlen = 1;
-				price = opt[cur - 1].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - 1);
-			}
-
-			if (cur > last_pos || price <= opt[cur].price)
-				SET_PRICE(cur, 1, 0, litlen, price);
-
-			if (cur == last_pos)
-				break;
-
-			if (inr > ilimit) /* last match must start at a minimum distance of 8 from oend */
-				continue;
-
-			mlen = opt[cur].mlen;
-			if (opt[cur].off > ZSTD_REP_MOVE_OPT) {
-				opt[cur].rep[2] = opt[cur - mlen].rep[1];
-				opt[cur].rep[1] = opt[cur - mlen].rep[0];
-				opt[cur].rep[0] = opt[cur].off - ZSTD_REP_MOVE_OPT;
-			} else {
-				opt[cur].rep[2] = (opt[cur].off > 1) ? opt[cur - mlen].rep[1] : opt[cur - mlen].rep[2];
-				opt[cur].rep[1] = (opt[cur].off > 0) ? opt[cur - mlen].rep[0] : opt[cur - mlen].rep[1];
-				opt[cur].rep[0] =
-				    ((opt[cur].off == ZSTD_REP_MOVE_OPT) && (mlen != 1)) ? (opt[cur - mlen].rep[0] - 1) : (opt[cur - mlen].rep[opt[cur].off]);
-			}
-
-			best_mlen = minMatch;
-			{
-				U32 i, last_i = ZSTD_REP_CHECK + (mlen != 1);
-				for (i = (mlen != 1); i < last_i; i++) {
-					const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (opt[cur].rep[0] - 1) : opt[cur].rep[i];
-					const U32 repIndex = (U32)(curr + cur - repCur);
-					const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
-					const BYTE *const repMatch = repBase + repIndex;
-					if ((repCur > 0 && repCur <= (S32)(curr + cur)) &&
-					    (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
-					    && (ZSTD_readMINMATCH(inr, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch))) {
-						/* repcode detected */
-						const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
-						mlen = (U32)ZSTD_count_2segments(inr + minMatch, repMatch + minMatch, iend, repEnd, prefixStart) + minMatch;
-
-						if (mlen > sufficient_len || cur + mlen >= ZSTD_OPT_NUM) {
-							best_mlen = mlen;
-							best_off = i;
-							last_pos = cur + 1;
-							goto _storeSequence;
-						}
-
-						best_off = i - (opt[cur].mlen != 1);
-						if (mlen > best_mlen)
-							best_mlen = mlen;
-
-						do {
-							if (opt[cur].mlen == 1) {
-								litlen = opt[cur].litlen;
-								if (cur > litlen) {
-									price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, inr - litlen,
-															best_off, mlen - MINMATCH, ultra);
-								} else
-									price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
-							} else {
-								litlen = 0;
-								price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, best_off, mlen - MINMATCH, ultra);
-							}
-
-							if (cur + mlen > last_pos || price <= opt[cur + mlen].price)
-								SET_PRICE(cur + mlen, mlen, i, litlen, price);
-							mlen--;
-						} while (mlen >= minMatch);
-					}
-				}
-			}
-
-			match_num = ZSTD_BtGetAllMatches_selectMLS_extDict(ctx, inr, iend, maxSearches, mls, matches, minMatch);
-
-			if (match_num > 0 && (matches[match_num - 1].len > sufficient_len || cur + matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
-				best_mlen = matches[match_num - 1].len;
-				best_off = matches[match_num - 1].off;
-				last_pos = cur + 1;
-				goto _storeSequence;
-			}
-
-			/* set prices using matches at position = cur */
-			for (u = 0; u < match_num; u++) {
-				mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
-				best_mlen = matches[u].len;
-
-				while (mlen <= best_mlen) {
-					if (opt[cur].mlen == 1) {
-						litlen = opt[cur].litlen;
-						if (cur > litlen)
-							price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, ip + cur - litlen,
-													matches[u].off - 1, mlen - MINMATCH, ultra);
-						else
-							price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
-					} else {
-						litlen = 0;
-						price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, matches[u].off - 1, mlen - MINMATCH, ultra);
-					}
-
-					if (cur + mlen > last_pos || (price < opt[cur + mlen].price))
-						SET_PRICE(cur + mlen, mlen, matches[u].off, litlen, price);
-
-					mlen++;
-				}
-			}
-		} /* for (cur = 1; cur <= last_pos; cur++) */
-
-		best_mlen = opt[last_pos].mlen;
-		best_off = opt[last_pos].off;
-		cur = last_pos - best_mlen;
-
-	/* store sequence */
-_storeSequence: /* cur, last_pos, best_mlen, best_off have to be set */
-		opt[0].mlen = 1;
-
-		while (1) {
-			mlen = opt[cur].mlen;
-			offset = opt[cur].off;
-			opt[cur].mlen = best_mlen;
-			opt[cur].off = best_off;
-			best_mlen = mlen;
-			best_off = offset;
-			if (mlen > cur)
-				break;
-			cur -= mlen;
-		}
-
-		for (u = 0; u <= last_pos;) {
-			u += opt[u].mlen;
-		}
-
-		for (cur = 0; cur < last_pos;) {
-			mlen = opt[cur].mlen;
-			if (mlen == 1) {
-				ip++;
-				cur++;
-				continue;
-			}
-			offset = opt[cur].off;
-			cur += mlen;
-			litLength = (U32)(ip - anchor);
-
-			if (offset > ZSTD_REP_MOVE_OPT) {
-				rep[2] = rep[1];
-				rep[1] = rep[0];
-				rep[0] = offset - ZSTD_REP_MOVE_OPT;
-				offset--;
-			} else {
-				if (offset != 0) {
-					best_off = (offset == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : (rep[offset]);
-					if (offset != 1)
-						rep[2] = rep[1];
-					rep[1] = rep[0];
-					rep[0] = best_off;
-				}
-
-				if (litLength == 0)
-					offset--;
-			}
-
-			ZSTD_updatePrice(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
-			ZSTD_storeSeq(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
-			anchor = ip = ip + mlen;
-		}
-	} /* for (cur=0; cur < last_pos; ) */
-
-	/* Save reps for next block */
-	{
-		int i;
-		for (i = 0; i < ZSTD_REP_NUM; i++)
-			ctx->repToConfirm[i] = rep[i];
-	}
-
-	/* Last Literals */
-	{
-		size_t lastLLSize = iend - anchor;
-		memcpy(seqStorePtr->lit, anchor, lastLLSize);
-		seqStorePtr->lit += lastLLSize;
-	}
-}
-
-#endif /* ZSTD_OPT_H_91842398743 */
diff --git a/tools/arch/x86/include/asm/msr-index.h b/tools/arch/x86/include/asm/msr-index.h
index a7c4134..01e2650 100644
--- a/tools/arch/x86/include/asm/msr-index.h
+++ b/tools/arch/x86/include/asm/msr-index.h
@@ -625,6 +625,8 @@
 
 #define MSR_IA32_BNDCFGS_RSVD		0x00000ffc
 
+#define MSR_IA32_XFD			0x000001c4
+#define MSR_IA32_XFD_ERR		0x000001c5
 #define MSR_IA32_XSS			0x00000da0
 
 #define MSR_IA32_APICBASE		0x0000001b
diff --git a/tools/arch/x86/include/uapi/asm/prctl.h b/tools/arch/x86/include/uapi/asm/prctl.h
index 5a6aac9..754a078 100644
--- a/tools/arch/x86/include/uapi/asm/prctl.h
+++ b/tools/arch/x86/include/uapi/asm/prctl.h
@@ -10,6 +10,10 @@
 #define ARCH_GET_CPUID		0x1011
 #define ARCH_SET_CPUID		0x1012
 
+#define ARCH_GET_XCOMP_SUPP	0x1021
+#define ARCH_GET_XCOMP_PERM	0x1022
+#define ARCH_REQ_XCOMP_PERM	0x1023
+
 #define ARCH_MAP_VDSO_X32	0x2001
 #define ARCH_MAP_VDSO_32	0x2002
 #define ARCH_MAP_VDSO_64	0x2003
diff --git a/tools/include/uapi/asm-generic/unistd.h b/tools/include/uapi/asm-generic/unistd.h
index 1c5fb86..4557a8b 100644
--- a/tools/include/uapi/asm-generic/unistd.h
+++ b/tools/include/uapi/asm-generic/unistd.h
@@ -880,8 +880,11 @@ __SYSCALL(__NR_memfd_secret, sys_memfd_secret)
 #define __NR_process_mrelease 448
 __SYSCALL(__NR_process_mrelease, sys_process_mrelease)
 
+#define __NR_futex_waitv 449
+__SYSCALL(__NR_futex_waitv, sys_futex_waitv)
+
 #undef __NR_syscalls
-#define __NR_syscalls 449
+#define __NR_syscalls 450
 
 /*
  * 32 bit systems traditionally used different
diff --git a/tools/include/uapi/drm/i915_drm.h b/tools/include/uapi/drm/i915_drm.h
index bde5860..914ebd9 100644
--- a/tools/include/uapi/drm/i915_drm.h
+++ b/tools/include/uapi/drm/i915_drm.h
@@ -1522,6 +1522,12 @@ struct drm_i915_gem_caching {
 #define I915_TILING_NONE	0
 #define I915_TILING_X		1
 #define I915_TILING_Y		2
+/*
+ * Do not add new tiling types here.  The I915_TILING_* values are for
+ * de-tiling fence registers that no longer exist on modern platforms.  Although
+ * the hardware may support new types of tiling in general (e.g., Tile4), we
+ * do not need to add them to the uapi that is specific to now-defunct ioctls.
+ */
 #define I915_TILING_LAST	I915_TILING_Y
 
 #define I915_BIT_6_SWIZZLE_NONE		0
@@ -1824,6 +1830,7 @@ struct drm_i915_gem_context_param {
  * Extensions:
  *   i915_context_engines_load_balance (I915_CONTEXT_ENGINES_EXT_LOAD_BALANCE)
  *   i915_context_engines_bond (I915_CONTEXT_ENGINES_EXT_BOND)
+ *   i915_context_engines_parallel_submit (I915_CONTEXT_ENGINES_EXT_PARALLEL_SUBMIT)
  */
 #define I915_CONTEXT_PARAM_ENGINES	0xa
 
@@ -1846,6 +1853,55 @@ struct drm_i915_gem_context_param {
  * attempted to use it, never re-use this context param number.
  */
 #define I915_CONTEXT_PARAM_RINGSIZE	0xc
+
+/*
+ * I915_CONTEXT_PARAM_PROTECTED_CONTENT:
+ *
+ * Mark that the context makes use of protected content, which will result
+ * in the context being invalidated when the protected content session is.
+ * Given that the protected content session is killed on suspend, the device
+ * is kept awake for the lifetime of a protected context, so the user should
+ * make sure to dispose of them once done.
+ * This flag can only be set at context creation time and, when set to true,
+ * must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE
+ * to false. This flag can't be set to true in conjunction with setting the
+ * I915_CONTEXT_PARAM_BANNABLE flag to false. Creation example:
+ *
+ * .. code-block:: C
+ *
+ *	struct drm_i915_gem_context_create_ext_setparam p_protected = {
+ *		.base = {
+ *			.name = I915_CONTEXT_CREATE_EXT_SETPARAM,
+ *		},
+ *		.param = {
+ *			.param = I915_CONTEXT_PARAM_PROTECTED_CONTENT,
+ *			.value = 1,
+ *		}
+ *	};
+ *	struct drm_i915_gem_context_create_ext_setparam p_norecover = {
+ *		.base = {
+ *			.name = I915_CONTEXT_CREATE_EXT_SETPARAM,
+ *			.next_extension = to_user_pointer(&p_protected),
+ *		},
+ *		.param = {
+ *			.param = I915_CONTEXT_PARAM_RECOVERABLE,
+ *			.value = 0,
+ *		}
+ *	};
+ *	struct drm_i915_gem_context_create_ext create = {
+ *		.flags = I915_CONTEXT_CREATE_FLAGS_USE_EXTENSIONS,
+ *		.extensions = to_user_pointer(&p_norecover);
+ *	};
+ *
+ *	ctx_id = gem_context_create_ext(drm_fd, &create);
+ *
+ * In addition to the normal failure cases, setting this flag during context
+ * creation can result in the following errors:
+ *
+ * -ENODEV: feature not available
+ * -EPERM: trying to mark a recoverable or not bannable context as protected
+ */
+#define I915_CONTEXT_PARAM_PROTECTED_CONTENT    0xd
 /* Must be kept compact -- no holes and well documented */
 
 	__u64 value;
@@ -2050,6 +2106,135 @@ struct i915_context_engines_bond {
 } __attribute__((packed)) name__
 
 /**
+ * struct i915_context_engines_parallel_submit - Configure engine for
+ * parallel submission.
+ *
+ * Setup a slot in the context engine map to allow multiple BBs to be submitted
+ * in a single execbuf IOCTL. Those BBs will then be scheduled to run on the GPU
+ * in parallel. Multiple hardware contexts are created internally in the i915 to
+ * run these BBs. Once a slot is configured for N BBs only N BBs can be
+ * submitted in each execbuf IOCTL and this is implicit behavior e.g. The user
+ * doesn't tell the execbuf IOCTL there are N BBs, the execbuf IOCTL knows how
+ * many BBs there are based on the slot's configuration. The N BBs are the last
+ * N buffer objects or first N if I915_EXEC_BATCH_FIRST is set.
+ *
+ * The default placement behavior is to create implicit bonds between each
+ * context if each context maps to more than 1 physical engine (e.g. context is
+ * a virtual engine). Also we only allow contexts of same engine class and these
+ * contexts must be in logically contiguous order. Examples of the placement
+ * behavior are described below. Lastly, the default is to not allow BBs to be
+ * preempted mid-batch. Rather insert coordinated preemption points on all
+ * hardware contexts between each set of BBs. Flags could be added in the future
+ * to change both of these default behaviors.
+ *
+ * Returns -EINVAL if hardware context placement configuration is invalid or if
+ * the placement configuration isn't supported on the platform / submission
+ * interface.
+ * Returns -ENODEV if extension isn't supported on the platform / submission
+ * interface.
+ *
+ * .. code-block:: none
+ *
+ *	Examples syntax:
+ *	CS[X] = generic engine of same class, logical instance X
+ *	INVALID = I915_ENGINE_CLASS_INVALID, I915_ENGINE_CLASS_INVALID_NONE
+ *
+ *	Example 1 pseudo code:
+ *	set_engines(INVALID)
+ *	set_parallel(engine_index=0, width=2, num_siblings=1,
+ *		     engines=CS[0],CS[1])
+ *
+ *	Results in the following valid placement:
+ *	CS[0], CS[1]
+ *
+ *	Example 2 pseudo code:
+ *	set_engines(INVALID)
+ *	set_parallel(engine_index=0, width=2, num_siblings=2,
+ *		     engines=CS[0],CS[2],CS[1],CS[3])
+ *
+ *	Results in the following valid placements:
+ *	CS[0], CS[1]
+ *	CS[2], CS[3]
+ *
+ *	This can be thought of as two virtual engines, each containing two
+ *	engines thereby making a 2D array. However, there are bonds tying the
+ *	entries together and placing restrictions on how they can be scheduled.
+ *	Specifically, the scheduler can choose only vertical columns from the 2D
+ *	array. That is, CS[0] is bonded to CS[1] and CS[2] to CS[3]. So if the
+ *	scheduler wants to submit to CS[0], it must also choose CS[1] and vice
+ *	versa. Same for CS[2] requires also using CS[3].
+ *	VE[0] = CS[0], CS[2]
+ *	VE[1] = CS[1], CS[3]
+ *
+ *	Example 3 pseudo code:
+ *	set_engines(INVALID)
+ *	set_parallel(engine_index=0, width=2, num_siblings=2,
+ *		     engines=CS[0],CS[1],CS[1],CS[3])
+ *
+ *	Results in the following valid and invalid placements:
+ *	CS[0], CS[1]
+ *	CS[1], CS[3] - Not logically contiguous, return -EINVAL
+ */
+struct i915_context_engines_parallel_submit {
+	/**
+	 * @base: base user extension.
+	 */
+	struct i915_user_extension base;
+
+	/**
+	 * @engine_index: slot for parallel engine
+	 */
+	__u16 engine_index;
+
+	/**
+	 * @width: number of contexts per parallel engine or in other words the
+	 * number of batches in each submission
+	 */
+	__u16 width;
+
+	/**
+	 * @num_siblings: number of siblings per context or in other words the
+	 * number of possible placements for each submission
+	 */
+	__u16 num_siblings;
+
+	/**
+	 * @mbz16: reserved for future use; must be zero
+	 */
+	__u16 mbz16;
+
+	/**
+	 * @flags: all undefined flags must be zero, currently not defined flags
+	 */
+	__u64 flags;
+
+	/**
+	 * @mbz64: reserved for future use; must be zero
+	 */
+	__u64 mbz64[3];
+
+	/**
+	 * @engines: 2-d array of engine instances to configure parallel engine
+	 *
+	 * length = width (i) * num_siblings (j)
+	 * index = j + i * num_siblings
+	 */
+	struct i915_engine_class_instance engines[0];
+
+} __packed;
+
+#define I915_DEFINE_CONTEXT_ENGINES_PARALLEL_SUBMIT(name__, N__) struct { \
+	struct i915_user_extension base; \
+	__u16 engine_index; \
+	__u16 width; \
+	__u16 num_siblings; \
+	__u16 mbz16; \
+	__u64 flags; \
+	__u64 mbz64[3]; \
+	struct i915_engine_class_instance engines[N__]; \
+} __attribute__((packed)) name__
+
+/**
  * DOC: Context Engine Map uAPI
  *
  * Context engine map is a new way of addressing engines when submitting batch-
@@ -2108,6 +2293,7 @@ struct i915_context_param_engines {
 	__u64 extensions; /* linked chain of extension blocks, 0 terminates */
 #define I915_CONTEXT_ENGINES_EXT_LOAD_BALANCE 0 /* see i915_context_engines_load_balance */
 #define I915_CONTEXT_ENGINES_EXT_BOND 1 /* see i915_context_engines_bond */
+#define I915_CONTEXT_ENGINES_EXT_PARALLEL_SUBMIT 2 /* see i915_context_engines_parallel_submit */
 	struct i915_engine_class_instance engines[0];
 } __attribute__((packed));
 
@@ -2726,14 +2912,20 @@ struct drm_i915_engine_info {
 
 	/** @flags: Engine flags. */
 	__u64 flags;
+#define I915_ENGINE_INFO_HAS_LOGICAL_INSTANCE		(1 << 0)
 
 	/** @capabilities: Capabilities of this engine. */
 	__u64 capabilities;
 #define I915_VIDEO_CLASS_CAPABILITY_HEVC		(1 << 0)
 #define I915_VIDEO_AND_ENHANCE_CLASS_CAPABILITY_SFC	(1 << 1)
 
+	/** @logical_instance: Logical instance of engine */
+	__u16 logical_instance;
+
 	/** @rsvd1: Reserved fields. */
-	__u64 rsvd1[4];
+	__u16 rsvd1[3];
+	/** @rsvd2: Reserved fields. */
+	__u64 rsvd2[3];
 };
 
 /**
@@ -2979,8 +3171,12 @@ struct drm_i915_gem_create_ext {
 	 *
 	 * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see
 	 * struct drm_i915_gem_create_ext_memory_regions.
+	 *
+	 * For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see
+	 * struct drm_i915_gem_create_ext_protected_content.
 	 */
 #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0
+#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1
 	__u64 extensions;
 };
 
@@ -3038,6 +3234,50 @@ struct drm_i915_gem_create_ext_memory_regions {
 	__u64 regions;
 };
 
+/**
+ * struct drm_i915_gem_create_ext_protected_content - The
+ * I915_OBJECT_PARAM_PROTECTED_CONTENT extension.
+ *
+ * If this extension is provided, buffer contents are expected to be protected
+ * by PXP encryption and require decryption for scan out and processing. This
+ * is only possible on platforms that have PXP enabled, on all other scenarios
+ * using this extension will cause the ioctl to fail and return -ENODEV. The
+ * flags parameter is reserved for future expansion and must currently be set
+ * to zero.
+ *
+ * The buffer contents are considered invalid after a PXP session teardown.
+ *
+ * The encryption is guaranteed to be processed correctly only if the object
+ * is submitted with a context created using the
+ * I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks
+ * at submission time on the validity of the objects involved.
+ *
+ * Below is an example on how to create a protected object:
+ *
+ * .. code-block:: C
+ *
+ *      struct drm_i915_gem_create_ext_protected_content protected_ext = {
+ *              .base = { .name = I915_GEM_CREATE_EXT_PROTECTED_CONTENT },
+ *              .flags = 0,
+ *      };
+ *      struct drm_i915_gem_create_ext create_ext = {
+ *              .size = PAGE_SIZE,
+ *              .extensions = (uintptr_t)&protected_ext,
+ *      };
+ *
+ *      int err = ioctl(fd, DRM_IOCTL_I915_GEM_CREATE_EXT, &create_ext);
+ *      if (err) ...
+ */
+struct drm_i915_gem_create_ext_protected_content {
+	/** @base: Extension link. See struct i915_user_extension. */
+	struct i915_user_extension base;
+	/** @flags: reserved for future usage, currently MBZ */
+	__u32 flags;
+};
+
+/* ID of the protected content session managed by i915 when PXP is active */
+#define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xf
+
 #if defined(__cplusplus)
 }
 #endif
diff --git a/tools/include/uapi/linux/prctl.h b/tools/include/uapi/linux/prctl.h
index de45fcd..bb73e9a 100644
--- a/tools/include/uapi/linux/prctl.h
+++ b/tools/include/uapi/linux/prctl.h
@@ -268,5 +268,8 @@ struct prctl_mm_map {
 # define PR_SCHED_CORE_SHARE_TO		2 /* push core_sched cookie to pid */
 # define PR_SCHED_CORE_SHARE_FROM	3 /* pull core_sched cookie to pid */
 # define PR_SCHED_CORE_MAX		4
+# define PR_SCHED_CORE_SCOPE_THREAD		0
+# define PR_SCHED_CORE_SCOPE_THREAD_GROUP	1
+# define PR_SCHED_CORE_SCOPE_PROCESS_GROUP	2
 
 #endif /* _LINUX_PRCTL_H */
diff --git a/tools/include/uapi/sound/asound.h b/tools/include/uapi/sound/asound.h
index 5859ca0..5fbb79e 100644
--- a/tools/include/uapi/sound/asound.h
+++ b/tools/include/uapi/sound/asound.h
@@ -1002,7 +1002,7 @@ typedef int __bitwise snd_ctl_elem_iface_t;
 #define SNDRV_CTL_ELEM_ACCESS_WRITE		(1<<1)
 #define SNDRV_CTL_ELEM_ACCESS_READWRITE		(SNDRV_CTL_ELEM_ACCESS_READ|SNDRV_CTL_ELEM_ACCESS_WRITE)
 #define SNDRV_CTL_ELEM_ACCESS_VOLATILE		(1<<2)	/* control value may be changed without a notification */
-// (1 << 3) is unused.
+/* (1 << 3) is unused. */
 #define SNDRV_CTL_ELEM_ACCESS_TLV_READ		(1<<4)	/* TLV read is possible */
 #define SNDRV_CTL_ELEM_ACCESS_TLV_WRITE		(1<<5)	/* TLV write is possible */
 #define SNDRV_CTL_ELEM_ACCESS_TLV_READWRITE	(SNDRV_CTL_ELEM_ACCESS_TLV_READ|SNDRV_CTL_ELEM_ACCESS_TLV_WRITE)
diff --git a/tools/objtool/check.c b/tools/objtool/check.c
index add3990..2173582 100644
--- a/tools/objtool/check.c
+++ b/tools/objtool/check.c
@@ -3310,6 +3310,9 @@ static bool ignore_unreachable_insn(struct objtool_file *file, struct instructio
 	if (!insn->func)
 		return false;
 
+	if (insn->func->static_call_tramp)
+		return true;
+
 	/*
 	 * CONFIG_UBSAN_TRAP inserts a UD2 when it sees
 	 * __builtin_unreachable().  The BUG() macro has an unreachable() after
diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt
index 2d7df87..3cf7bac 100644
--- a/tools/perf/Documentation/perf-record.txt
+++ b/tools/perf/Documentation/perf-record.txt
@@ -469,7 +469,7 @@
 
 --switch-events::
 Record context switch events i.e. events of type PERF_RECORD_SWITCH or
-PERF_RECORD_SWITCH_CPU_WIDE. In some cases (e.g. Intel PT or CoreSight)
+PERF_RECORD_SWITCH_CPU_WIDE. In some cases (e.g. Intel PT, CoreSight or Arm SPE)
 switch events will be enabled automatically, which can be suppressed by
 by the option --no-switch-events.
 
diff --git a/tools/perf/Makefile.perf b/tools/perf/Makefile.perf
index 0777748..80522bc 100644
--- a/tools/perf/Makefile.perf
+++ b/tools/perf/Makefile.perf
@@ -516,17 +516,17 @@
 $(kvm_ioctl_array): $(kvm_hdr_dir)/kvm.h $(kvm_ioctl_tbl)
 	$(Q)$(SHELL) '$(kvm_ioctl_tbl)' $(kvm_hdr_dir) > $@
 
-socket_ipproto_array := $(beauty_outdir)/socket_ipproto_array.c
-socket_ipproto_tbl := $(srctree)/tools/perf/trace/beauty/socket_ipproto.sh
-
-$(socket_ipproto_array): $(linux_uapi_dir)/in.h $(socket_ipproto_tbl)
-	$(Q)$(SHELL) '$(socket_ipproto_tbl)' $(linux_uapi_dir) > $@
-
-socket_arrays := $(beauty_outdir)/socket_arrays.c
+socket_arrays := $(beauty_outdir)/socket.c
 socket_tbl := $(srctree)/tools/perf/trace/beauty/socket.sh
 
-$(socket_arrays): $(beauty_linux_dir)/socket.h $(socket_tbl)
-	$(Q)$(SHELL) '$(socket_tbl)' $(beauty_linux_dir) > $@
+$(socket_arrays): $(linux_uapi_dir)/in.h $(beauty_linux_dir)/socket.h $(socket_tbl)
+	$(Q)$(SHELL) '$(socket_tbl)' $(linux_uapi_dir) $(beauty_linux_dir) > $@
+
+sockaddr_arrays := $(beauty_outdir)/sockaddr.c
+sockaddr_tbl := $(srctree)/tools/perf/trace/beauty/sockaddr.sh
+
+$(sockaddr_arrays): $(beauty_linux_dir)/socket.h $(sockaddr_tbl)
+	$(Q)$(SHELL) '$(sockaddr_tbl)' $(beauty_linux_dir) > $@
 
 vhost_virtio_ioctl_array := $(beauty_ioctl_outdir)/vhost_virtio_ioctl_array.c
 vhost_virtio_hdr_dir := $(srctree)/tools/include/uapi/linux
@@ -736,8 +736,8 @@
 	$(sndrv_ctl_ioctl_array) \
 	$(kcmp_type_array) \
 	$(kvm_ioctl_array) \
-	$(socket_ipproto_array) \
 	$(socket_arrays) \
+	$(sockaddr_arrays) \
 	$(vhost_virtio_ioctl_array) \
 	$(madvise_behavior_array) \
 	$(mmap_flags_array) \
@@ -1113,8 +1113,8 @@
 		$(OUTPUT)$(sndrv_pcm_ioctl_array) \
 		$(OUTPUT)$(kvm_ioctl_array) \
 		$(OUTPUT)$(kcmp_type_array) \
-		$(OUTPUT)$(socket_ipproto_array) \
 		$(OUTPUT)$(socket_arrays) \
+		$(OUTPUT)$(sockaddr_arrays) \
 		$(OUTPUT)$(vhost_virtio_ioctl_array) \
 		$(OUTPUT)$(perf_ioctl_array) \
 		$(OUTPUT)$(prctl_option_array) \
diff --git a/tools/perf/arch/arm/include/arch-tests.h b/tools/perf/arch/arm/include/arch-tests.h
index c625380..452b3d9 100644
--- a/tools/perf/arch/arm/include/arch-tests.h
+++ b/tools/perf/arch/arm/include/arch-tests.h
@@ -2,6 +2,6 @@
 #ifndef ARCH_TESTS_H
 #define ARCH_TESTS_H
 
-extern struct test arch_tests[];
+extern struct test_suite *arch_tests[];
 
 #endif
diff --git a/tools/perf/arch/arm/tests/arch-tests.c b/tools/perf/arch/arm/tests/arch-tests.c
index 6848101..6956111 100644
--- a/tools/perf/arch/arm/tests/arch-tests.c
+++ b/tools/perf/arch/arm/tests/arch-tests.c
@@ -3,18 +3,10 @@
 #include "tests/tests.h"
 #include "arch-tests.h"
 
-struct test arch_tests[] = {
+struct test_suite *arch_tests[] = {
 #ifdef HAVE_DWARF_UNWIND_SUPPORT
-	{
-		.desc = "DWARF unwind",
-		.func = test__dwarf_unwind,
-	},
+	&suite__dwarf_unwind,
 #endif
-	{
-		.desc = "Vectors page",
-		.func = test__vectors_page,
-	},
-	{
-		.func = NULL,
-	},
+	&suite__vectors_page,
+	NULL,
 };
diff --git a/tools/perf/arch/arm/tests/vectors-page.c b/tools/perf/arch/arm/tests/vectors-page.c
index 7ffdd79..55a8358 100644
--- a/tools/perf/arch/arm/tests/vectors-page.c
+++ b/tools/perf/arch/arm/tests/vectors-page.c
@@ -9,8 +9,7 @@
 
 #define VECTORS__MAP_NAME "[vectors]"
 
-int test__vectors_page(struct test *test __maybe_unused,
-		       int subtest __maybe_unused)
+static int test__vectors_page(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	void *start, *end;
 
@@ -22,3 +21,5 @@ int test__vectors_page(struct test *test __maybe_unused,
 
 	return TEST_OK;
 }
+
+DEFINE_SUITE("Vectors page", vectors_page);
diff --git a/tools/perf/arch/arm64/include/arch-tests.h b/tools/perf/arch/arm64/include/arch-tests.h
index c625380..452b3d9 100644
--- a/tools/perf/arch/arm64/include/arch-tests.h
+++ b/tools/perf/arch/arm64/include/arch-tests.h
@@ -2,6 +2,6 @@
 #ifndef ARCH_TESTS_H
 #define ARCH_TESTS_H
 
-extern struct test arch_tests[];
+extern struct test_suite *arch_tests[];
 
 #endif
diff --git a/tools/perf/arch/arm64/tests/arch-tests.c b/tools/perf/arch/arm64/tests/arch-tests.c
index 5b1543c..ad16b4f 100644
--- a/tools/perf/arch/arm64/tests/arch-tests.c
+++ b/tools/perf/arch/arm64/tests/arch-tests.c
@@ -3,14 +3,9 @@
 #include "tests/tests.h"
 #include "arch-tests.h"
 
-struct test arch_tests[] = {
+struct test_suite *arch_tests[] = {
 #ifdef HAVE_DWARF_UNWIND_SUPPORT
-	{
-		.desc = "DWARF unwind",
-		.func = test__dwarf_unwind,
-	},
+	&suite__dwarf_unwind,
 #endif
-	{
-		.func = NULL,
-	},
+	NULL,
 };
diff --git a/tools/perf/arch/arm64/util/arm-spe.c b/tools/perf/arch/arm64/util/arm-spe.c
index a4420d4..2100d46 100644
--- a/tools/perf/arch/arm64/util/arm-spe.c
+++ b/tools/perf/arch/arm64/util/arm-spe.c
@@ -23,6 +23,7 @@
 #include "../../../util/auxtrace.h"
 #include "../../../util/record.h"
 #include "../../../util/arm-spe.h"
+#include <tools/libc_compat.h> // reallocarray
 
 #define KiB(x) ((x) * 1024)
 #define MiB(x) ((x) * 1024 * 1024)
@@ -31,6 +32,8 @@ struct arm_spe_recording {
 	struct auxtrace_record		itr;
 	struct perf_pmu			*arm_spe_pmu;
 	struct evlist		*evlist;
+	int			wrapped_cnt;
+	bool			*wrapped;
 };
 
 static void arm_spe_set_timestamp(struct auxtrace_record *itr,
@@ -84,6 +87,55 @@ static int arm_spe_info_fill(struct auxtrace_record *itr,
 	return 0;
 }
 
+static void
+arm_spe_snapshot_resolve_auxtrace_defaults(struct record_opts *opts,
+					   bool privileged)
+{
+	/*
+	 * The default snapshot size is the auxtrace mmap size. If neither auxtrace mmap size nor
+	 * snapshot size is specified, then the default is 4MiB for privileged users, 128KiB for
+	 * unprivileged users.
+	 *
+	 * The default auxtrace mmap size is 4MiB/page_size for privileged users, 128KiB for
+	 * unprivileged users. If an unprivileged user does not specify mmap pages, the mmap pages
+	 * will be reduced from the default 512KiB/page_size to 256KiB/page_size, otherwise the
+	 * user is likely to get an error as they exceed their mlock limmit.
+	 */
+
+	/*
+	 * No size were given to '-S' or '-m,', so go with the default
+	 */
+	if (!opts->auxtrace_snapshot_size && !opts->auxtrace_mmap_pages) {
+		if (privileged) {
+			opts->auxtrace_mmap_pages = MiB(4) / page_size;
+		} else {
+			opts->auxtrace_mmap_pages = KiB(128) / page_size;
+			if (opts->mmap_pages == UINT_MAX)
+				opts->mmap_pages = KiB(256) / page_size;
+		}
+	} else if (!opts->auxtrace_mmap_pages && !privileged && opts->mmap_pages == UINT_MAX) {
+		opts->mmap_pages = KiB(256) / page_size;
+	}
+
+	/*
+	 * '-m,xyz' was specified but no snapshot size, so make the snapshot size as big as the
+	 * auxtrace mmap area.
+	 */
+	if (!opts->auxtrace_snapshot_size)
+		opts->auxtrace_snapshot_size = opts->auxtrace_mmap_pages * (size_t)page_size;
+
+	/*
+	 * '-Sxyz' was specified but no auxtrace mmap area, so make the auxtrace mmap area big
+	 * enough to fit the requested snapshot size.
+	 */
+	if (!opts->auxtrace_mmap_pages) {
+		size_t sz = opts->auxtrace_snapshot_size;
+
+		sz = round_up(sz, page_size) / page_size;
+		opts->auxtrace_mmap_pages = roundup_pow_of_two(sz);
+	}
+}
+
 static int arm_spe_recording_options(struct auxtrace_record *itr,
 				     struct evlist *evlist,
 				     struct record_opts *opts)
@@ -115,6 +167,36 @@ static int arm_spe_recording_options(struct auxtrace_record *itr,
 	if (!opts->full_auxtrace)
 		return 0;
 
+	/*
+	 * we are in snapshot mode.
+	 */
+	if (opts->auxtrace_snapshot_mode) {
+		/*
+		 * Command arguments '-Sxyz' and/or '-m,xyz' are missing, so fill those in with
+		 * default values.
+		 */
+		if (!opts->auxtrace_snapshot_size || !opts->auxtrace_mmap_pages)
+			arm_spe_snapshot_resolve_auxtrace_defaults(opts, privileged);
+
+		/*
+		 * Snapshot size can't be bigger than the auxtrace area.
+		 */
+		if (opts->auxtrace_snapshot_size > opts->auxtrace_mmap_pages * (size_t)page_size) {
+			pr_err("Snapshot size %zu must not be greater than AUX area tracing mmap size %zu\n",
+			       opts->auxtrace_snapshot_size,
+			       opts->auxtrace_mmap_pages * (size_t)page_size);
+			return -EINVAL;
+		}
+
+		/*
+		 * Something went wrong somewhere - this shouldn't happen.
+		 */
+		if (!opts->auxtrace_snapshot_size || !opts->auxtrace_mmap_pages) {
+			pr_err("Failed to calculate default snapshot size and/or AUX area tracing mmap pages\n");
+			return -EINVAL;
+		}
+	}
+
 	/* We are in full trace mode but '-m,xyz' wasn't specified */
 	if (!opts->auxtrace_mmap_pages) {
 		if (privileged) {
@@ -138,6 +220,9 @@ static int arm_spe_recording_options(struct auxtrace_record *itr,
 		}
 	}
 
+	if (opts->auxtrace_snapshot_mode)
+		pr_debug2("%sx snapshot size: %zu\n", ARM_SPE_PMU_NAME,
+			  opts->auxtrace_snapshot_size);
 
 	/*
 	 * To obtain the auxtrace buffer file descriptor, the auxtrace event
@@ -166,8 +251,199 @@ static int arm_spe_recording_options(struct auxtrace_record *itr,
 	tracking_evsel->core.attr.sample_period = 1;
 
 	/* In per-cpu case, always need the time of mmap events etc */
-	if (!perf_cpu_map__empty(cpus))
+	if (!perf_cpu_map__empty(cpus)) {
 		evsel__set_sample_bit(tracking_evsel, TIME);
+		evsel__set_sample_bit(tracking_evsel, CPU);
+
+		/* also track task context switch */
+		if (!record_opts__no_switch_events(opts))
+			tracking_evsel->core.attr.context_switch = 1;
+	}
+
+	return 0;
+}
+
+static int arm_spe_parse_snapshot_options(struct auxtrace_record *itr __maybe_unused,
+					 struct record_opts *opts,
+					 const char *str)
+{
+	unsigned long long snapshot_size = 0;
+	char *endptr;
+
+	if (str) {
+		snapshot_size = strtoull(str, &endptr, 0);
+		if (*endptr || snapshot_size > SIZE_MAX)
+			return -1;
+	}
+
+	opts->auxtrace_snapshot_mode = true;
+	opts->auxtrace_snapshot_size = snapshot_size;
+
+	return 0;
+}
+
+static int arm_spe_snapshot_start(struct auxtrace_record *itr)
+{
+	struct arm_spe_recording *ptr =
+			container_of(itr, struct arm_spe_recording, itr);
+	struct evsel *evsel;
+
+	evlist__for_each_entry(ptr->evlist, evsel) {
+		if (evsel->core.attr.type == ptr->arm_spe_pmu->type)
+			return evsel__disable(evsel);
+	}
+	return -EINVAL;
+}
+
+static int arm_spe_snapshot_finish(struct auxtrace_record *itr)
+{
+	struct arm_spe_recording *ptr =
+			container_of(itr, struct arm_spe_recording, itr);
+	struct evsel *evsel;
+
+	evlist__for_each_entry(ptr->evlist, evsel) {
+		if (evsel->core.attr.type == ptr->arm_spe_pmu->type)
+			return evsel__enable(evsel);
+	}
+	return -EINVAL;
+}
+
+static int arm_spe_alloc_wrapped_array(struct arm_spe_recording *ptr, int idx)
+{
+	bool *wrapped;
+	int cnt = ptr->wrapped_cnt, new_cnt, i;
+
+	/*
+	 * No need to allocate, so return early.
+	 */
+	if (idx < cnt)
+		return 0;
+
+	/*
+	 * Make ptr->wrapped as big as idx.
+	 */
+	new_cnt = idx + 1;
+
+	/*
+	 * Free'ed in arm_spe_recording_free().
+	 */
+	wrapped = reallocarray(ptr->wrapped, new_cnt, sizeof(bool));
+	if (!wrapped)
+		return -ENOMEM;
+
+	/*
+	 * init new allocated values.
+	 */
+	for (i = cnt; i < new_cnt; i++)
+		wrapped[i] = false;
+
+	ptr->wrapped_cnt = new_cnt;
+	ptr->wrapped = wrapped;
+
+	return 0;
+}
+
+static bool arm_spe_buffer_has_wrapped(unsigned char *buffer,
+				      size_t buffer_size, u64 head)
+{
+	u64 i, watermark;
+	u64 *buf = (u64 *)buffer;
+	size_t buf_size = buffer_size;
+
+	/*
+	 * Defensively handle the case where head might be continually increasing - if its value is
+	 * equal or greater than the size of the ring buffer, then we can safely determine it has
+	 * wrapped around. Otherwise, continue to detect if head might have wrapped.
+	 */
+	if (head >= buffer_size)
+		return true;
+
+	/*
+	 * We want to look the very last 512 byte (chosen arbitrarily) in the ring buffer.
+	 */
+	watermark = buf_size - 512;
+
+	/*
+	 * The value of head is somewhere within the size of the ring buffer. This can be that there
+	 * hasn't been enough data to fill the ring buffer yet or the trace time was so long that
+	 * head has numerically wrapped around.  To find we need to check if we have data at the
+	 * very end of the ring buffer.  We can reliably do this because mmap'ed pages are zeroed
+	 * out and there is a fresh mapping with every new session.
+	 */
+
+	/*
+	 * head is less than 512 byte from the end of the ring buffer.
+	 */
+	if (head > watermark)
+		watermark = head;
+
+	/*
+	 * Speed things up by using 64 bit transactions (see "u64 *buf" above)
+	 */
+	watermark /= sizeof(u64);
+	buf_size /= sizeof(u64);
+
+	/*
+	 * If we find trace data at the end of the ring buffer, head has been there and has
+	 * numerically wrapped around at least once.
+	 */
+	for (i = watermark; i < buf_size; i++)
+		if (buf[i])
+			return true;
+
+	return false;
+}
+
+static int arm_spe_find_snapshot(struct auxtrace_record *itr, int idx,
+				  struct auxtrace_mmap *mm, unsigned char *data,
+				  u64 *head, u64 *old)
+{
+	int err;
+	bool wrapped;
+	struct arm_spe_recording *ptr =
+			container_of(itr, struct arm_spe_recording, itr);
+
+	/*
+	 * Allocate memory to keep track of wrapping if this is the first
+	 * time we deal with this *mm.
+	 */
+	if (idx >= ptr->wrapped_cnt) {
+		err = arm_spe_alloc_wrapped_array(ptr, idx);
+		if (err)
+			return err;
+	}
+
+	/*
+	 * Check to see if *head has wrapped around.  If it hasn't only the
+	 * amount of data between *head and *old is snapshot'ed to avoid
+	 * bloating the perf.data file with zeros.  But as soon as *head has
+	 * wrapped around the entire size of the AUX ring buffer it taken.
+	 */
+	wrapped = ptr->wrapped[idx];
+	if (!wrapped && arm_spe_buffer_has_wrapped(data, mm->len, *head)) {
+		wrapped = true;
+		ptr->wrapped[idx] = true;
+	}
+
+	pr_debug3("%s: mmap index %d old head %zu new head %zu size %zu\n",
+		  __func__, idx, (size_t)*old, (size_t)*head, mm->len);
+
+	/*
+	 * No wrap has occurred, we can just use *head and *old.
+	 */
+	if (!wrapped)
+		return 0;
+
+	/*
+	 * *head has wrapped around - adjust *head and *old to pickup the
+	 * entire content of the AUX buffer.
+	 */
+	if (*head >= mm->len) {
+		*old = *head - mm->len;
+	} else {
+		*head += mm->len;
+		*old = *head - mm->len;
+	}
 
 	return 0;
 }
@@ -186,6 +462,7 @@ static void arm_spe_recording_free(struct auxtrace_record *itr)
 	struct arm_spe_recording *sper =
 			container_of(itr, struct arm_spe_recording, itr);
 
+	free(sper->wrapped);
 	free(sper);
 }
 
@@ -207,6 +484,10 @@ struct auxtrace_record *arm_spe_recording_init(int *err,
 
 	sper->arm_spe_pmu = arm_spe_pmu;
 	sper->itr.pmu = arm_spe_pmu;
+	sper->itr.snapshot_start = arm_spe_snapshot_start;
+	sper->itr.snapshot_finish = arm_spe_snapshot_finish;
+	sper->itr.find_snapshot = arm_spe_find_snapshot;
+	sper->itr.parse_snapshot_options = arm_spe_parse_snapshot_options;
 	sper->itr.recording_options = arm_spe_recording_options;
 	sper->itr.info_priv_size = arm_spe_info_priv_size;
 	sper->itr.info_fill = arm_spe_info_fill;
diff --git a/tools/perf/arch/powerpc/include/arch-tests.h b/tools/perf/arch/powerpc/include/arch-tests.h
index c625380..452b3d9 100644
--- a/tools/perf/arch/powerpc/include/arch-tests.h
+++ b/tools/perf/arch/powerpc/include/arch-tests.h
@@ -2,6 +2,6 @@
 #ifndef ARCH_TESTS_H
 #define ARCH_TESTS_H
 
-extern struct test arch_tests[];
+extern struct test_suite *arch_tests[];
 
 #endif
diff --git a/tools/perf/arch/powerpc/tests/arch-tests.c b/tools/perf/arch/powerpc/tests/arch-tests.c
index 8c3fbd4..eb98c57 100644
--- a/tools/perf/arch/powerpc/tests/arch-tests.c
+++ b/tools/perf/arch/powerpc/tests/arch-tests.c
@@ -3,14 +3,10 @@
 #include "tests/tests.h"
 #include "arch-tests.h"
 
-struct test arch_tests[] = {
+
+struct test_suite *arch_tests[] = {
 #ifdef HAVE_DWARF_UNWIND_SUPPORT
-	{
-		.desc = "Test dwarf unwind",
-		.func = test__dwarf_unwind,
-	},
+	&suite__dwarf_unwind,
 #endif
-	{
-		.func = NULL,
-	},
+	NULL,
 };
diff --git a/tools/perf/arch/x86/entry/syscalls/syscall_64.tbl b/tools/perf/arch/x86/entry/syscalls/syscall_64.tbl
index 18b5500..fe8f8dd 100644
--- a/tools/perf/arch/x86/entry/syscalls/syscall_64.tbl
+++ b/tools/perf/arch/x86/entry/syscalls/syscall_64.tbl
@@ -370,6 +370,7 @@
 446	common	landlock_restrict_self	sys_landlock_restrict_self
 447	common	memfd_secret		sys_memfd_secret
 448	common	process_mrelease	sys_process_mrelease
+449	common	futex_waitv		sys_futex_waitv
 
 #
 # Due to a historical design error, certain syscalls are numbered differently
diff --git a/tools/perf/arch/x86/include/arch-tests.h b/tools/perf/arch/x86/include/arch-tests.h
index 9599e7a..6a1a1b3 100644
--- a/tools/perf/arch/x86/include/arch-tests.h
+++ b/tools/perf/arch/x86/include/arch-tests.h
@@ -2,15 +2,15 @@
 #ifndef ARCH_TESTS_H
 #define ARCH_TESTS_H
 
-struct test;
+struct test_suite;
 
 /* Tests */
-int test__rdpmc(struct test *test, int subtest);
-int test__insn_x86(struct test *test, int subtest);
-int test__intel_pt_pkt_decoder(struct test *test, int subtest);
-int test__bp_modify(struct test *test, int subtest);
-int test__x86_sample_parsing(struct test *test, int subtest);
+int test__rdpmc(struct test_suite *test, int subtest);
+int test__insn_x86(struct test_suite *test, int subtest);
+int test__intel_pt_pkt_decoder(struct test_suite *test, int subtest);
+int test__bp_modify(struct test_suite *test, int subtest);
+int test__x86_sample_parsing(struct test_suite *test, int subtest);
 
-extern struct test arch_tests[];
+extern struct test_suite *arch_tests[];
 
 #endif
diff --git a/tools/perf/arch/x86/tests/arch-tests.c b/tools/perf/arch/x86/tests/arch-tests.c
index 71aa673..64fb73d 100644
--- a/tools/perf/arch/x86/tests/arch-tests.c
+++ b/tools/perf/arch/x86/tests/arch-tests.c
@@ -3,39 +3,28 @@
 #include "tests/tests.h"
 #include "arch-tests.h"
 
-struct test arch_tests[] = {
-	{
-		.desc = "x86 rdpmc",
-		.func = test__rdpmc,
-	},
-#ifdef HAVE_DWARF_UNWIND_SUPPORT
-	{
-		.desc = "DWARF unwind",
-		.func = test__dwarf_unwind,
-	},
-#endif
+DEFINE_SUITE("x86 rdpmc", rdpmc);
 #ifdef HAVE_AUXTRACE_SUPPORT
-	{
-		.desc = "x86 instruction decoder - new instructions",
-		.func = test__insn_x86,
-	},
-	{
-		.desc = "Intel PT packet decoder",
-		.func = test__intel_pt_pkt_decoder,
-	},
+DEFINE_SUITE("x86 instruction decoder - new instructions", insn_x86);
+DEFINE_SUITE("Intel PT packet decoder", intel_pt_pkt_decoder);
 #endif
 #if defined(__x86_64__)
-	{
-		.desc = "x86 bp modify",
-		.func = test__bp_modify,
-	},
+DEFINE_SUITE("x86 bp modify", bp_modify);
 #endif
-	{
-		.desc = "x86 Sample parsing",
-		.func = test__x86_sample_parsing,
-	},
-	{
-		.func = NULL,
-	},
+DEFINE_SUITE("x86 Sample parsing", x86_sample_parsing);
 
+struct test_suite *arch_tests[] = {
+	&suite__rdpmc,
+#ifdef HAVE_DWARF_UNWIND_SUPPORT
+	&suite__dwarf_unwind,
+#endif
+#ifdef HAVE_AUXTRACE_SUPPORT
+	&suite__insn_x86,
+	&suite__intel_pt_pkt_decoder,
+#endif
+#if defined(__x86_64__)
+	&suite__bp_modify,
+#endif
+	&suite__x86_sample_parsing,
+	NULL,
 };
diff --git a/tools/perf/arch/x86/tests/bp-modify.c b/tools/perf/arch/x86/tests/bp-modify.c
index dffcf9b..0924ccd 100644
--- a/tools/perf/arch/x86/tests/bp-modify.c
+++ b/tools/perf/arch/x86/tests/bp-modify.c
@@ -204,7 +204,7 @@ static int bp_modify2(void)
 	return rip == (unsigned long) bp_1 ? TEST_OK : TEST_FAIL;
 }
 
-int test__bp_modify(struct test *test __maybe_unused,
+int test__bp_modify(struct test_suite *test __maybe_unused,
 		    int subtest __maybe_unused)
 {
 	TEST_ASSERT_VAL("modify test 1 failed\n", !bp_modify1());
diff --git a/tools/perf/arch/x86/tests/insn-x86.c b/tools/perf/arch/x86/tests/insn-x86.c
index 0262b0d..94b490c 100644
--- a/tools/perf/arch/x86/tests/insn-x86.c
+++ b/tools/perf/arch/x86/tests/insn-x86.c
@@ -173,7 +173,7 @@ static int test_data_set(struct test_data *dat_set, int x86_64)
  * verbose (-v) option to see all the instructions and whether or not they
  * decoded successfully.
  */
-int test__insn_x86(struct test *test __maybe_unused, int subtest __maybe_unused)
+int test__insn_x86(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int ret = 0;
 
diff --git a/tools/perf/arch/x86/tests/intel-cqm.c b/tools/perf/arch/x86/tests/intel-cqm.c
index 27dd8cf..cb5b2c6 100644
--- a/tools/perf/arch/x86/tests/intel-cqm.c
+++ b/tools/perf/arch/x86/tests/intel-cqm.c
@@ -37,7 +37,7 @@ static pid_t spawn(void)
  * the last read counter value to avoid triggering a WARN_ON_ONCE() in
  * smp_call_function_many() caused by sending IPIs from NMI context.
  */
-int test__intel_cqm_count_nmi_context(struct test *test __maybe_unused, int subtest __maybe_unused)
+int test__intel_cqm_count_nmi_context(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct evlist *evlist = NULL;
 	struct evsel *evsel = NULL;
diff --git a/tools/perf/arch/x86/tests/intel-pt-pkt-decoder-test.c b/tools/perf/arch/x86/tests/intel-pt-pkt-decoder-test.c
index c933e3d..2fc882a 100644
--- a/tools/perf/arch/x86/tests/intel-pt-pkt-decoder-test.c
+++ b/tools/perf/arch/x86/tests/intel-pt-pkt-decoder-test.c
@@ -289,7 +289,7 @@ static int test_one(struct test_data *d)
  * This test feeds byte sequences to the Intel PT packet decoder and checks the
  * results. Changes to the packet context are also checked.
  */
-int test__intel_pt_pkt_decoder(struct test *test __maybe_unused, int subtest __maybe_unused)
+int test__intel_pt_pkt_decoder(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct test_data *d = data;
 	int ret;
diff --git a/tools/perf/arch/x86/tests/rdpmc.c b/tools/perf/arch/x86/tests/rdpmc.c
index 1ea9166..498413a 100644
--- a/tools/perf/arch/x86/tests/rdpmc.c
+++ b/tools/perf/arch/x86/tests/rdpmc.c
@@ -157,7 +157,7 @@ static int __test__rdpmc(void)
 	return 0;
 }
 
-int test__rdpmc(struct test *test __maybe_unused, int subtest __maybe_unused)
+int test__rdpmc(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int status = 0;
 	int wret = 0;
diff --git a/tools/perf/arch/x86/tests/sample-parsing.c b/tools/perf/arch/x86/tests/sample-parsing.c
index c92db87..bfbd366 100644
--- a/tools/perf/arch/x86/tests/sample-parsing.c
+++ b/tools/perf/arch/x86/tests/sample-parsing.c
@@ -115,7 +115,7 @@ static int do_test(u64 sample_type)
  * For now, the PERF_SAMPLE_WEIGHT_STRUCT is the only X86 specific sample type.
  * The test only checks the PERF_SAMPLE_WEIGHT_STRUCT type.
  */
-int test__x86_sample_parsing(struct test *test __maybe_unused, int subtest __maybe_unused)
+int test__x86_sample_parsing(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	return do_test(PERF_SAMPLE_WEIGHT_STRUCT);
 }
diff --git a/tools/perf/bench/futex-lock-pi.c b/tools/perf/bench/futex-lock-pi.c
index 5d1fe9c..137890f 100644
--- a/tools/perf/bench/futex-lock-pi.c
+++ b/tools/perf/bench/futex-lock-pi.c
@@ -233,6 +233,7 @@ int bench_futex_lock_pi(int argc, const char **argv)
 	print_summary();
 
 	free(worker);
+	perf_cpu_map__put(cpu);
 	return ret;
 err:
 	usage_with_options(bench_futex_lock_pi_usage, options);
diff --git a/tools/perf/bench/futex-requeue.c b/tools/perf/bench/futex-requeue.c
index 97fe31f..f7a5ffe 100644
--- a/tools/perf/bench/futex-requeue.c
+++ b/tools/perf/bench/futex-requeue.c
@@ -294,6 +294,7 @@ int bench_futex_requeue(int argc, const char **argv)
 	print_summary();
 
 	free(worker);
+	perf_cpu_map__put(cpu);
 	return ret;
 err:
 	usage_with_options(bench_futex_requeue_usage, options);
diff --git a/tools/perf/bench/futex-wake-parallel.c b/tools/perf/bench/futex-wake-parallel.c
index e970e6b..0983f40 100644
--- a/tools/perf/bench/futex-wake-parallel.c
+++ b/tools/perf/bench/futex-wake-parallel.c
@@ -329,6 +329,7 @@ int bench_futex_wake_parallel(int argc, const char **argv)
 	print_summary();
 
 	free(blocked_worker);
+	perf_cpu_map__put(cpu);
 	return ret;
 }
 #endif /* HAVE_PTHREAD_BARRIER */
diff --git a/tools/perf/bench/futex-wake.c b/tools/perf/bench/futex-wake.c
index 77f058a..2226a475 100644
--- a/tools/perf/bench/futex-wake.c
+++ b/tools/perf/bench/futex-wake.c
@@ -222,5 +222,6 @@ int bench_futex_wake(int argc, const char **argv)
 	print_summary();
 
 	free(worker);
+	perf_cpu_map__put(cpu);
 	return ret;
 }
diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index 624ea12..0b52e08 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -979,6 +979,8 @@ static struct syscall_fmt syscall_fmts[] = {
 	  .arg = { [2] = { .scnprintf = SCA_GETRANDOM_FLAGS, /* flags */ }, }, },
 	{ .name	    = "getrlimit",
 	  .arg = { [0] = STRARRAY(resource, rlimit_resources), }, },
+	{ .name	    = "getsockopt",
+	  .arg = { [1] = STRARRAY(level, socket_level), }, },
 	{ .name	    = "gettid",	    .errpid = true, },
 	{ .name	    = "ioctl",
 	  .arg = {
@@ -1121,6 +1123,8 @@ static struct syscall_fmt syscall_fmts[] = {
 	  .arg = { [0] = STRARRAY(which, itimers), }, },
 	{ .name	    = "setrlimit",
 	  .arg = { [0] = STRARRAY(resource, rlimit_resources), }, },
+	{ .name	    = "setsockopt",
+	  .arg = { [1] = STRARRAY(level, socket_level), }, },
 	{ .name	    = "socket",
 	  .arg = { [0] = STRARRAY(family, socket_families),
 		   [1] = { .scnprintf = SCA_SK_TYPE, /* type */ },
diff --git a/tools/perf/design.txt b/tools/perf/design.txt
index a42fab3..aa8cfea 100644
--- a/tools/perf/design.txt
+++ b/tools/perf/design.txt
@@ -106,6 +106,9 @@
 	PERF_COUNT_HW_BRANCH_INSTRUCTIONS	= 4,
 	PERF_COUNT_HW_BRANCH_MISSES		= 5,
 	PERF_COUNT_HW_BUS_CYCLES		= 6,
+	PERF_COUNT_HW_STALLED_CYCLES_FRONTEND	= 7,
+	PERF_COUNT_HW_STALLED_CYCLES_BACKEND	= 8,
+	PERF_COUNT_HW_REF_CPU_CYCLES		= 9,
 };
 
 These are standardized types of events that work relatively uniformly
diff --git a/tools/perf/pmu-events/arch/powerpc/power10/metrics.json b/tools/perf/pmu-events/arch/powerpc/power10/metrics.json
new file mode 100644
index 0000000..b57526f
--- /dev/null
+++ b/tools/perf/pmu-events/arch/powerpc/power10/metrics.json
@@ -0,0 +1,676 @@
+[
+    {
+        "BriefDescription": "Percentage of cycles that are run cycles",
+        "MetricExpr": "PM_RUN_CYC / PM_CYC * 100",
+        "MetricGroup": "General",
+        "MetricName": "RUN_CYCLES_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction",
+        "MetricExpr": "PM_CYC / PM_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "CYCLES_PER_INSTRUCTION"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled for any reason",
+        "MetricExpr": "PM_DISP_STALL_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled because there was a flush",
+        "MetricExpr": "PM_DISP_STALL_FLUSH / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_FLUSH_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled because the MMU was handling a translation miss",
+        "MetricExpr": "PM_DISP_STALL_TRANSLATION / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_TRANSLATION_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled waiting to resolve an instruction ERAT miss",
+        "MetricExpr": "PM_DISP_STALL_IERAT_ONLY_MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_IERAT_ONLY_MISS_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled waiting to resolve an instruction TLB miss",
+        "MetricExpr": "PM_DISP_STALL_ITLB_MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_ITLB_MISS_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled due to an icache miss",
+        "MetricExpr": "PM_DISP_STALL_IC_MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_IC_MISS_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled while the instruction was fetched from the local L2",
+        "MetricExpr": "PM_DISP_STALL_IC_L2 / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_IC_L2_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled while the instruction was fetched from the local L3",
+        "MetricExpr": "PM_DISP_STALL_IC_L3 / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_IC_L3_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled while the instruction was fetched from any source beyond the local L3",
+        "MetricExpr": "PM_DISP_STALL_IC_L3MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_IC_L3MISS_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled due to an icache miss after a branch mispredict",
+        "MetricExpr": "PM_DISP_STALL_BR_MPRED_ICMISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_BR_MPRED_ICMISS_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled while instruction was fetched from the local L2 after suffering a branch mispredict",
+        "MetricExpr": "PM_DISP_STALL_BR_MPRED_IC_L2 / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_BR_MPRED_IC_L2_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled while instruction was fetched from the local L3 after suffering a branch mispredict",
+        "MetricExpr": "PM_DISP_STALL_BR_MPRED_IC_L3 / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_BR_MPRED_IC_L3_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled while instruction was fetched from any source beyond the local L3 after suffering a branch mispredict",
+        "MetricExpr": "PM_DISP_STALL_BR_MPRED_IC_L3MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_BR_MPRED_IC_L3MISS_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled due to a branch mispredict",
+        "MetricExpr": "PM_DISP_STALL_BR_MPRED / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_BR_MPRED_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch for any reason",
+        "MetricExpr": "PM_DISP_STALL_HELD_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_HELD_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch because of a synchronizing instruction that requires the ICT to be empty before dispatch",
+        "MetricExpr": "PM_DISP_STALL_HELD_SYNC_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISP_HELD_STALL_SYNC_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch while waiting on the scoreboard",
+        "MetricExpr": "PM_DISP_STALL_HELD_SCOREBOARD_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISP_HELD_STALL_SCOREBOARD_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch due to issue queue full",
+        "MetricExpr": "PM_DISP_STALL_HELD_ISSQ_FULL_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISP_HELD_STALL_ISSQ_FULL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch because the mapper/SRB was full",
+        "MetricExpr": "PM_DISP_STALL_HELD_RENAME_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_HELD_RENAME_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch because the STF mapper/SRB was full",
+        "MetricExpr": "PM_DISP_STALL_HELD_STF_MAPPER_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_HELD_STF_MAPPER_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch because the XVFC mapper/SRB was full",
+        "MetricExpr": "PM_DISP_STALL_HELD_XVFC_MAPPER_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_HELD_XVFC_MAPPER_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch for any other reason",
+        "MetricExpr": "PM_DISP_STALL_HELD_OTHER_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_HELD_OTHER_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction has been dispatched but not issued for any reason",
+        "MetricExpr": "PM_ISSUE_STALL / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "ISSUE_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting to be finished in one of the execution units",
+        "MetricExpr": "PM_EXEC_STALL / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "EXECUTION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction spent executing an NTC instruction that gets flushed some time after dispatch",
+        "MetricExpr": "PM_EXEC_STALL_NTC_FLUSH / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "NTC_FLUSH_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTF instruction finishes at dispatch",
+        "MetricExpr": "PM_EXEC_STALL_FIN_AT_DISP / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "FIN_AT_DISP_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is executing in the branch unit",
+        "MetricExpr": "PM_EXEC_STALL_BRU / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "BRU_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a simple fixed point instruction that is executing in the LSU",
+        "MetricExpr": "PM_EXEC_STALL_SIMPLE_FX / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "SIMPLE_FX_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is executing in the VSU",
+        "MetricExpr": "PM_EXEC_STALL_VSU / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "VSU_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting to be finished in one of the execution units",
+        "MetricExpr": "PM_EXEC_STALL_TRANSLATION / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "TRANSLATION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a load or store that suffered a translation miss",
+        "MetricExpr": "PM_EXEC_STALL_DERAT_ONLY_MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DERAT_ONLY_MISS_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is recovering from a TLB miss",
+        "MetricExpr": "PM_EXEC_STALL_DERAT_DTLB_MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DERAT_DTLB_MISS_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is executing in the LSU",
+        "MetricExpr": "PM_EXEC_STALL_LSU / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "LSU_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a load that is executing in the LSU",
+        "MetricExpr": "PM_EXEC_STALL_LOAD / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "LOAD_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting for a load miss to resolve from either the local L2 or local L3",
+        "MetricExpr": "PM_EXEC_STALL_DMISS_L2L3 / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DMISS_L2L3_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting for a load miss to resolve from either the local L2 or local L3, with an RC dispatch conflict",
+        "MetricExpr": "PM_EXEC_STALL_DMISS_L2L3_CONFLICT / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DMISS_L2L3_CONFLICT_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting for a load miss to resolve from either the local L2 or local L3, without an RC dispatch conflict",
+        "MetricExpr": "PM_EXEC_STALL_DMISS_L2L3_NOCONFLICT / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DMISS_L2L3_NOCONFLICT_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting for a load miss to resolve from a source beyond the local L2 and local L3",
+        "MetricExpr": "PM_EXEC_STALL_DMISS_L3MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DMISS_L3MISS_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting for a load miss to resolve from a neighbor chiplet's L2 or L3 in the same chip",
+        "MetricExpr": "PM_EXEC_STALL_DMISS_L21_L31 / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DMISS_L21_L31_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting for a load miss to resolve from L4, local memory or OpenCAPI chip",
+        "MetricExpr": "PM_EXEC_STALL_DMISS_LMEM / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DMISS_LMEM_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting for a load miss to resolve from a remote chip (cache, L4, memory or OpenCAPI) in the same group",
+        "MetricExpr": "PM_EXEC_STALL_DMISS_OFF_CHIP / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DMISS_OFF_CHIP_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is waiting for a load miss to resolve from a distant chip (cache, L4, memory or OpenCAPI chip)",
+        "MetricExpr": "PM_EXEC_STALL_DMISS_OFF_NODE / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DMISS_OFF_NODE_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is executing a TLBIEL instruction",
+        "MetricExpr": "PM_EXEC_STALL_TLBIEL / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "TLBIEL_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is finishing a load after its data has been reloaded from a data source beyond the local L1, OR when the LSU is processing an L1-hit, OR when the NTF instruction merged with another load in the LMQ",
+        "MetricExpr": "PM_EXEC_STALL_LOAD_FINISH / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "LOAD_FINISH_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a store that is executing in the LSU",
+        "MetricExpr": "PM_EXEC_STALL_STORE / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "STORE_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is in the store unit outside of handling store misses or other special store operations",
+        "MetricExpr": "PM_EXEC_STALL_STORE_PIPE / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "STORE_PIPE_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a store whose cache line was not resident in the L1 and had to wait for allocation of the missing line into the L1",
+        "MetricExpr": "PM_EXEC_STALL_STORE_MISS / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "STORE_MISS_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a TLBIE instruction waiting for a response from the L2",
+        "MetricExpr": "PM_EXEC_STALL_TLBIE / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "TLBIE_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is executing a PTESYNC instruction",
+        "MetricExpr": "PM_EXEC_STALL_PTESYNC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "PTESYNC_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction cannot complete because the thread was blocked",
+        "MetricExpr": "PM_CMPL_STALL / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "COMPLETION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction cannot complete because it was interrupted by ANY exception",
+        "MetricExpr": "PM_CMPL_STALL_EXCEPTION / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "EXCEPTION_COMPLETION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is stuck at finish waiting for the non-speculative finish of either a STCX instruction waiting for its result or a load waiting for non-critical sectors of data and ECC",
+        "MetricExpr": "PM_CMPL_STALL_MEM_ECC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "MEM_ECC_COMPLETION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a STCX instruction waiting for resolution from the nest",
+        "MetricExpr": "PM_CMPL_STALL_STCX / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "STCX_COMPLETION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a LWSYNC instruction waiting to complete",
+        "MetricExpr": "PM_CMPL_STALL_LWSYNC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "LWSYNC_COMPLETION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction is a HWSYNC instruction stuck at finish waiting for a response from the L2",
+        "MetricExpr": "PM_CMPL_STALL_HWSYNC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "HWSYNC_COMPLETION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction required special handling before completion",
+        "MetricExpr": "PM_CMPL_STALL_SPECIAL / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "SPECIAL_COMPLETION_STALL_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when dispatch was stalled because fetch was being held, so there was nothing in the pipeline for this thread",
+        "MetricExpr": "PM_DISP_STALL_FETCH / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_FETCH_CPI"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTC instruction was held at dispatch because of power management",
+        "MetricExpr": "PM_DISP_STALL_HELD_HALT_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "CPI",
+        "MetricName": "DISPATCHED_HELD_HALT_CPI"
+    },
+    {
+        "BriefDescription": "Percentage of flushes per completed instruction",
+        "MetricExpr": "PM_FLUSH / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Others",
+        "MetricName": "FLUSH_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of flushes due to a branch mispredict per completed instruction",
+        "MetricExpr": "PM_FLUSH_MPRED / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Others",
+        "MetricName": "BR_MPRED_FLUSH_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of branch mispredictions per completed instruction",
+        "MetricExpr": "PM_BR_MPRED_CMPL / PM_RUN_INST_CMPL",
+        "MetricGroup": "Others",
+        "MetricName": "BRANCH_MISPREDICTION_RATE"
+    },
+    {
+        "BriefDescription": "Percentage of finished loads that missed in the L1",
+        "MetricExpr": "PM_LD_MISS_L1 / PM_LD_REF_L1 * 100",
+        "MetricGroup": "Others",
+        "MetricName": "L1_LD_MISS_RATIO",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of completed instructions that were loads that missed the L1",
+        "MetricExpr": "PM_LD_MISS_L1 / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Others",
+        "MetricName": "L1_LD_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of completed instructions when the DPTEG required for the load/store instruction in execution was missing from the TLB",
+        "MetricExpr": "PM_DTLB_MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Others",
+        "MetricName": "DTLB_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Average number of completed instructions dispatched per instruction completed",
+        "MetricExpr": "PM_INST_DISP / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "DISPATCH_PER_INST_CMPL"
+    },
+    {
+        "BriefDescription": "Percentage of completed instructions that were a demand load that did not hit in the L1 or L2",
+        "MetricExpr": "PM_DATA_FROM_L2MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "General",
+        "MetricName": "L2_LD_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of completed instructions that were demand fetches that missed the L1 icache",
+        "MetricExpr": "PM_L1_ICACHE_MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Instruction_Misses",
+        "MetricName": "L1_INST_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of completed instructions that were demand fetches that reloaded from beyond the L3 icache",
+        "MetricExpr": "PM_INST_FROM_L3MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "General",
+        "MetricName": "L3_INST_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Average number of completed instructions per cycle",
+        "MetricExpr": "PM_INST_CMPL / PM_CYC",
+        "MetricGroup": "General",
+        "MetricName": "IPC"
+    },
+    {
+        "BriefDescription": "Average number of cycles per completed instruction group",
+        "MetricExpr": "PM_CYC / PM_1PLUS_PPC_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "CYCLES_PER_COMPLETED_INSTRUCTIONS_SET"
+    },
+    {
+        "BriefDescription": "Percentage of cycles when at least 1 instruction dispatched",
+        "MetricExpr": "PM_1PLUS_PPC_DISP / PM_RUN_CYC * 100",
+        "MetricGroup": "General",
+        "MetricName": "CYCLES_ATLEAST_ONE_INST_DISPATCHED",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Average number of finished loads per completed instruction",
+        "MetricExpr": "PM_LD_REF_L1 / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "LOADS_PER_INST"
+    },
+    {
+        "BriefDescription": "Average number of finished stores per completed instruction",
+        "MetricExpr": "PM_ST_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "STORES_PER_INST"
+    },
+    {
+        "BriefDescription": "Percentage of demand loads that reloaded from beyond the L2 per completed instruction",
+        "MetricExpr": "PM_DATA_FROM_L2MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "dL1_Reloads",
+        "MetricName": "DL1_RELOAD_FROM_L2_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of demand loads that reloaded from beyond the L3 per completed instruction",
+        "MetricExpr": "PM_DATA_FROM_L3MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "dL1_Reloads",
+        "MetricName": "DL1_RELOAD_FROM_L3_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of DERAT misses with 4k page size per completed instruction",
+        "MetricExpr": "PM_DERAT_MISS_4K / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_4K_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of DERAT misses with 64k page size per completed instruction",
+        "MetricExpr": "PM_DERAT_MISS_64K / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_64K_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Average number of run cycles per completed instruction",
+        "MetricExpr": "PM_RUN_CYC / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "RUN_CPI"
+    },
+    {
+        "BriefDescription": "Percentage of DERAT misses per completed instruction",
+        "MetricExpr": "PM_DERAT_MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Average number of completed instructions per run cycle",
+        "MetricExpr": "PM_RUN_INST_CMPL / PM_RUN_CYC",
+        "MetricGroup": "General",
+        "MetricName": "RUN_IPC"
+    },
+    {
+        "BriefDescription": "Average number of completed instructions per instruction group",
+        "MetricExpr": "PM_RUN_INST_CMPL / PM_1PLUS_PPC_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "AVERAGE_COMPLETED_INSTRUCTION_SET_SIZE"
+    },
+    {
+        "BriefDescription": "Average number of finished instructions per completed instructions",
+        "MetricExpr": "PM_INST_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "INST_FIN_PER_CMPL"
+    },
+    {
+        "BriefDescription": "Average cycles per completed instruction when the NTF instruction is completing and the finish was overlooked",
+        "MetricExpr": "PM_EXEC_STALL_UNKNOWN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "EXEC_STALL_UNKOWN_CPI"
+    },
+    {
+        "BriefDescription": "Percentage of finished branches that were taken",
+        "MetricExpr": "PM_BR_TAKEN_CMPL / PM_BR_FIN * 100",
+        "MetricGroup": "General",
+        "MetricName": "TAKEN_BRANCHES",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of completed instructions that were a demand load that did not hit in the L1, L2, or the L3",
+        "MetricExpr": "PM_DATA_FROM_L3MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "General",
+        "MetricName": "L3_LD_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Average number of finished branches per completed instruction",
+        "MetricExpr": "PM_BR_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "BRANCHES_PER_INST"
+    },
+    {
+        "BriefDescription": "Average number of instructions finished in the LSU per completed instruction",
+        "MetricExpr": "PM_LSU_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "LSU_PER_INST"
+    },
+    {
+        "BriefDescription": "Average number of instructions finished in the VSU per completed instruction",
+        "MetricExpr": "PM_VSU_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "VSU_PER_INST"
+    },
+    {
+        "BriefDescription": "Average number of TLBIE instructions finished in the LSU per completed instruction",
+        "MetricExpr": "PM_TLBIE_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "TLBIE_PER_INST"
+    },
+    {
+        "BriefDescription": "Average number of STCX instructions finshed per completed instruction",
+        "MetricExpr": "PM_STCX_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "STXC_PER_INST"
+    },
+    {
+        "BriefDescription": "Average number of LARX instructions finshed per completed instruction",
+        "MetricExpr": "PM_LARX_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "LARX_PER_INST"
+    },
+    {
+        "BriefDescription": "Average number of PTESYNC instructions finshed per completed instruction",
+        "MetricExpr": "PM_PTESYNC_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "PTESYNC_PER_INST"
+    },
+    {
+        "BriefDescription": "Average number of simple fixed-point instructions finshed in the store unit per completed instruction",
+        "MetricExpr": "PM_FX_LSU_FIN / PM_RUN_INST_CMPL",
+        "MetricGroup": "General",
+        "MetricName": "FX_PER_INST"
+    },
+    {
+        "BriefDescription": "Percentage of demand load misses that reloaded the L1 cache",
+        "MetricExpr": "PM_LD_DEMAND_MISS_L1 / PM_LD_MISS_L1 * 100",
+        "MetricGroup": "General",
+        "MetricName": "DL1_MISS_RELOADS",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of demand load misses that reloaded from beyond the local L2",
+        "MetricExpr": "PM_DATA_FROM_L2MISS / PM_LD_DEMAND_MISS_L1 * 100",
+        "MetricGroup": "dL1_Reloads",
+        "MetricName": "DL1_RELOAD_FROM_L2_MISS",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of demand load misses that reloaded from beyond the local L3",
+        "MetricExpr": "PM_DATA_FROM_L3MISS / PM_LD_DEMAND_MISS_L1 * 100",
+        "MetricGroup": "dL1_Reloads",
+        "MetricName": "DL1_RELOAD_FROM_L3_MISS",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of cycles stalled due to the NTC instruction waiting for a load miss to resolve from a source beyond the local L2 and local L3",
+        "MetricExpr": "DMISS_L3MISS_STALL_CPI / RUN_CPI * 100",
+        "MetricGroup": "General",
+        "MetricName": "DCACHE_MISS_CPI",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of DERAT misses with 2M page size per completed instruction",
+        "MetricExpr": "PM_DERAT_MISS_2M / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_2M_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of DERAT misses with 16M page size per completed instruction",
+        "MetricExpr": "PM_DERAT_MISS_16M / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_16M_MISS_RATE",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "DERAT miss ratio for 4K page size",
+        "MetricExpr": "PM_DERAT_MISS_4K / PM_DERAT_MISS",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_4K_MISS_RATIO"
+    },
+    {
+        "BriefDescription": "DERAT miss ratio for 2M page size",
+        "MetricExpr": "PM_DERAT_MISS_2M / PM_DERAT_MISS",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_2M_MISS_RATIO"
+    },
+    {
+        "BriefDescription": "DERAT miss ratio for 16M page size",
+        "MetricExpr": "PM_DERAT_MISS_16M / PM_DERAT_MISS",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_16M_MISS_RATIO"
+    },
+    {
+        "BriefDescription": "DERAT miss ratio for 64K page size",
+        "MetricExpr": "PM_DERAT_MISS_64K / PM_DERAT_MISS",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_64K_MISS_RATIO"
+    },
+    {
+        "BriefDescription": "Percentage of DERAT misses that resulted in TLB reloads",
+        "MetricExpr": "PM_DTLB_MISS / PM_DERAT_MISS * 100",
+        "MetricGroup": "Translation",
+        "MetricName": "DERAT_MISS_RELOAD",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of icache misses that were reloaded from beyond the local L3",
+        "MetricExpr": "PM_INST_FROM_L3MISS / PM_L1_ICACHE_MISS * 100",
+        "MetricGroup": "Instruction_Misses",
+        "MetricName": "INST_FROM_L3_MISS",
+        "ScaleUnit": "1%"
+    },
+    {
+        "BriefDescription": "Percentage of icache reloads from the beyond the L3 per completed instruction",
+        "MetricExpr": "PM_INST_FROM_L3MISS / PM_RUN_INST_CMPL * 100",
+        "MetricGroup": "Instruction_Misses",
+        "MetricName": "INST_FROM_L3_MISS_RATE",
+        "ScaleUnit": "1%"
+    }
+]
diff --git a/tools/perf/tests/api-io.c b/tools/perf/tests/api-io.c
index 2ada86a..e91cf2c 100644
--- a/tools/perf/tests/api-io.c
+++ b/tools/perf/tests/api-io.c
@@ -289,8 +289,8 @@ static int test_get_dec(void)
 	return ret;
 }
 
-int test__api_io(struct test *test __maybe_unused,
-		int subtest __maybe_unused)
+static int test__api_io(struct test_suite *test __maybe_unused,
+			int subtest __maybe_unused)
 {
 	int ret = 0;
 
@@ -302,3 +302,5 @@ int test__api_io(struct test *test __maybe_unused,
 		ret = TEST_FAIL;
 	return ret;
 }
+
+DEFINE_SUITE("Test api io", api_io);
diff --git a/tools/perf/tests/attr.c b/tools/perf/tests/attr.c
index 9b40a25..0f73e30 100644
--- a/tools/perf/tests/attr.c
+++ b/tools/perf/tests/attr.c
@@ -178,7 +178,7 @@ static int run_dir(const char *d, const char *perf)
 	return system(cmd) ? TEST_FAIL : TEST_OK;
 }
 
-int test__attr(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__attr(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct stat st;
 	char path_perf[PATH_MAX];
@@ -207,3 +207,5 @@ int test__attr(struct test *test __maybe_unused, int subtest __maybe_unused)
 
 	return TEST_SKIP;
 }
+
+DEFINE_SUITE("Setup struct perf_event_attr", attr);
diff --git a/tools/perf/tests/backward-ring-buffer.c b/tools/perf/tests/backward-ring-buffer.c
index 7447a44..79a980b 100644
--- a/tools/perf/tests/backward-ring-buffer.c
+++ b/tools/perf/tests/backward-ring-buffer.c
@@ -82,7 +82,7 @@ static int do_test(struct evlist *evlist, int mmap_pages,
 }
 
 
-int test__backward_ring_buffer(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__backward_ring_buffer(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int ret = TEST_SKIP, err, sample_count = 0, comm_count = 0;
 	char pid[16], sbuf[STRERR_BUFSIZE];
@@ -167,3 +167,5 @@ int test__backward_ring_buffer(struct test *test __maybe_unused, int subtest __m
 	evlist__delete(evlist);
 	return ret;
 }
+
+DEFINE_SUITE("Read backward ring buffer", backward_ring_buffer);
diff --git a/tools/perf/tests/bitmap.c b/tools/perf/tests/bitmap.c
index 12b805e..3848563 100644
--- a/tools/perf/tests/bitmap.c
+++ b/tools/perf/tests/bitmap.c
@@ -40,7 +40,7 @@ static int test_bitmap(const char *str)
 	return ret;
 }
 
-int test__bitmap_print(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__bitmap_print(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	TEST_ASSERT_VAL("failed to convert map", test_bitmap("1"));
 	TEST_ASSERT_VAL("failed to convert map", test_bitmap("1,5"));
@@ -51,3 +51,5 @@ int test__bitmap_print(struct test *test __maybe_unused, int subtest __maybe_unu
 	TEST_ASSERT_VAL("failed to convert map", test_bitmap("1-10,12-20,22-30,32-40"));
 	return 0;
 }
+
+DEFINE_SUITE("Print bitmap", bitmap_print);
diff --git a/tools/perf/tests/bp_account.c b/tools/perf/tests/bp_account.c
index 489b5060..d1ebb55 100644
--- a/tools/perf/tests/bp_account.c
+++ b/tools/perf/tests/bp_account.c
@@ -19,6 +19,19 @@
 #include "../perf-sys.h"
 #include "cloexec.h"
 
+/*
+ * PowerPC and S390 do not support creation of instruction breakpoints using the
+ * perf_event interface.
+ *
+ * Just disable the test for these architectures until these issues are
+ * resolved.
+ */
+#if defined(__powerpc__) || defined(__s390x__)
+#define BP_ACCOUNT_IS_SUPPORTED 0
+#else
+#define BP_ACCOUNT_IS_SUPPORTED 1
+#endif
+
 static volatile long the_var;
 
 static noinline int test_function(void)
@@ -173,13 +186,18 @@ static int detect_share(int wp_cnt, int bp_cnt)
  *     we create another watchpoint to ensure
  *     the slot accounting is correct
  */
-int test__bp_accounting(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__bp_accounting(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int has_ioctl = detect_ioctl();
 	int wp_cnt = detect_cnt(false);
 	int bp_cnt = detect_cnt(true);
 	int share  = detect_share(wp_cnt, bp_cnt);
 
+	if (!BP_ACCOUNT_IS_SUPPORTED) {
+		pr_debug("Test not supported on this architecture");
+		return TEST_SKIP;
+	}
+
 	pr_debug("watchpoints count %d, breakpoints count %d, has_ioctl %d, share %d\n",
 		 wp_cnt, bp_cnt, has_ioctl, share);
 
@@ -189,18 +207,4 @@ int test__bp_accounting(struct test *test __maybe_unused, int subtest __maybe_un
 	return bp_accounting(wp_cnt, share);
 }
 
-bool test__bp_account_is_supported(void)
-{
-	/*
-	 * PowerPC and S390 do not support creation of instruction
-	 * breakpoints using the perf_event interface.
-	 *
-	 * Just disable the test for these architectures until these
-	 * issues are resolved.
-	 */
-#if defined(__powerpc__) || defined(__s390x__)
-	return false;
-#else
-	return true;
-#endif
-}
+DEFINE_SUITE("Breakpoint accounting", bp_accounting);
diff --git a/tools/perf/tests/bp_signal.c b/tools/perf/tests/bp_signal.c
index ef37353..1f2908f0 100644
--- a/tools/perf/tests/bp_signal.c
+++ b/tools/perf/tests/bp_signal.c
@@ -161,11 +161,16 @@ static long long bp_count(int fd)
 	return count;
 }
 
-int test__bp_signal(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__bp_signal(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct sigaction sa;
 	long long count1, count2, count3;
 
+	if (!BP_SIGNAL_IS_SUPPORTED) {
+		pr_debug("Test not supported on this architecture");
+		return TEST_SKIP;
+	}
+
 	/* setup SIGIO signal handler */
 	memset(&sa, 0, sizeof(struct sigaction));
 	sa.sa_sigaction = (void *) sig_handler;
@@ -285,29 +290,4 @@ int test__bp_signal(struct test *test __maybe_unused, int subtest __maybe_unused
 		TEST_OK : TEST_FAIL;
 }
 
-bool test__bp_signal_is_supported(void)
-{
-	/*
-	 * PowerPC and S390 do not support creation of instruction
-	 * breakpoints using the perf_event interface.
-	 *
-	 * ARM requires explicit rounding down of the instruction
-	 * pointer in Thumb mode, and then requires the single-step
-	 * to be handled explicitly in the overflow handler to avoid
-	 * stepping into the SIGIO handler and getting stuck on the
-	 * breakpointed instruction.
-	 *
-	 * Since arm64 has the same issue with arm for the single-step
-	 * handling, this case also gets stuck on the breakpointed
-	 * instruction.
-	 *
-	 * Just disable the test for these architectures until these
-	 * issues are resolved.
-	 */
-#if defined(__powerpc__) || defined(__s390x__) || defined(__arm__) || \
-    defined(__aarch64__)
-	return false;
-#else
-	return true;
-#endif
-}
+DEFINE_SUITE("Breakpoint overflow signal handler", bp_signal);
diff --git a/tools/perf/tests/bp_signal_overflow.c b/tools/perf/tests/bp_signal_overflow.c
index eb4dbbd..4e897c2 100644
--- a/tools/perf/tests/bp_signal_overflow.c
+++ b/tools/perf/tests/bp_signal_overflow.c
@@ -59,13 +59,18 @@ static long long bp_count(int fd)
 #define EXECUTIONS 10000
 #define THRESHOLD  100
 
-int test__bp_signal_overflow(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__bp_signal_overflow(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct perf_event_attr pe;
 	struct sigaction sa;
 	long long count;
 	int fd, i, fails = 0;
 
+	if (!BP_SIGNAL_IS_SUPPORTED) {
+		pr_debug("Test not supported on this architecture");
+		return TEST_SKIP;
+	}
+
 	/* setup SIGIO signal handler */
 	memset(&sa, 0, sizeof(struct sigaction));
 	sa.sa_sigaction = (void *) sig_handler;
@@ -133,3 +138,5 @@ int test__bp_signal_overflow(struct test *test __maybe_unused, int subtest __may
 
 	return fails ? TEST_FAIL : TEST_OK;
 }
+
+DEFINE_SUITE("Breakpoint overflow sampling", bp_signal_overflow);
diff --git a/tools/perf/tests/bpf.c b/tools/perf/tests/bpf.c
index 2bf146e..329f77f 100644
--- a/tools/perf/tests/bpf.c
+++ b/tools/perf/tests/bpf.c
@@ -62,7 +62,6 @@ static int llseek_loop(void)
 
 static struct {
 	enum test_llvm__testcase prog_id;
-	const char *desc;
 	const char *name;
 	const char *msg_compile_fail;
 	const char *msg_load_fail;
@@ -72,7 +71,6 @@ static struct {
 } bpf_testcase_table[] = {
 	{
 		.prog_id	  = LLVM_TESTCASE_BASE,
-		.desc		  = "Basic BPF filtering",
 		.name		  = "[basic_bpf_test]",
 		.msg_compile_fail = "fix 'perf test LLVM' first",
 		.msg_load_fail	  = "load bpf object failed",
@@ -81,7 +79,6 @@ static struct {
 	},
 	{
 		.prog_id	  = LLVM_TESTCASE_BASE,
-		.desc		  = "BPF pinning",
 		.name		  = "[bpf_pinning]",
 		.msg_compile_fail = "fix kbuild first",
 		.msg_load_fail	  = "check your vmlinux setting?",
@@ -92,7 +89,6 @@ static struct {
 #ifdef HAVE_BPF_PROLOGUE
 	{
 		.prog_id	  = LLVM_TESTCASE_BPF_PROLOGUE,
-		.desc		  = "BPF prologue generation",
 		.name		  = "[bpf_prologue_test]",
 		.msg_compile_fail = "fix kbuild first",
 		.msg_load_fail	  = "check your vmlinux setting?",
@@ -283,18 +279,6 @@ static int __test__bpf(int idx)
 	return ret;
 }
 
-int test__bpf_subtest_get_nr(void)
-{
-	return (int)ARRAY_SIZE(bpf_testcase_table);
-}
-
-const char *test__bpf_subtest_get_desc(int i)
-{
-	if (i < 0 || i >= (int)ARRAY_SIZE(bpf_testcase_table))
-		return NULL;
-	return bpf_testcase_table[i].desc;
-}
-
 static int check_env(void)
 {
 	int err;
@@ -313,7 +297,7 @@ static int check_env(void)
 	}
 
 	err = bpf_load_program(BPF_PROG_TYPE_KPROBE, insns,
-			       sizeof(insns) / sizeof(insns[0]),
+			       ARRAY_SIZE(insns),
 			       license, kver_int, NULL, 0);
 	if (err < 0) {
 		pr_err("Missing basic BPF support, skip this test: %s\n",
@@ -325,7 +309,7 @@ static int check_env(void)
 	return 0;
 }
 
-int test__bpf(struct test *test __maybe_unused, int i)
+static int test__bpf(int i)
 {
 	int err;
 
@@ -343,21 +327,60 @@ int test__bpf(struct test *test __maybe_unused, int i)
 	err = __test__bpf(i);
 	return err;
 }
+#endif
 
+static int test__basic_bpf_test(struct test_suite *test __maybe_unused,
+				int subtest __maybe_unused)
+{
+#ifdef HAVE_LIBBPF_SUPPORT
+	return test__bpf(0);
 #else
-int test__bpf_subtest_get_nr(void)
-{
-	return 0;
-}
-
-const char *test__bpf_subtest_get_desc(int i __maybe_unused)
-{
-	return NULL;
-}
-
-int test__bpf(struct test *test __maybe_unused, int i __maybe_unused)
-{
 	pr_debug("Skip BPF test because BPF support is not compiled\n");
 	return TEST_SKIP;
-}
 #endif
+}
+
+static int test__bpf_pinning(struct test_suite *test __maybe_unused,
+			     int subtest __maybe_unused)
+{
+#ifdef HAVE_LIBBPF_SUPPORT
+	return test__bpf(1);
+#else
+	pr_debug("Skip BPF test because BPF support is not compiled\n");
+	return TEST_SKIP;
+#endif
+}
+
+static int test__bpf_prologue_test(struct test_suite *test __maybe_unused,
+				   int subtest __maybe_unused)
+{
+#if defined(HAVE_LIBBPF_SUPPORT) && defined(HAVE_BPF_PROLOGUE)
+	return test__bpf(2);
+#else
+	pr_debug("Skip BPF test because BPF support is not compiled\n");
+	return TEST_SKIP;
+#endif
+}
+
+
+static struct test_case bpf_tests[] = {
+#ifdef HAVE_LIBBPF_SUPPORT
+	TEST_CASE("Basic BPF filtering", basic_bpf_test),
+	TEST_CASE("BPF pinning", bpf_pinning),
+#ifdef HAVE_BPF_PROLOGUE
+	TEST_CASE("BPF prologue generation", bpf_prologue_test),
+#else
+	TEST_CASE_REASON("BPF prologue generation", bpf_prologue_test, "not compiled in"),
+#endif
+#else
+	TEST_CASE_REASON("Basic BPF filtering", basic_bpf_test, "not compiled in"),
+	TEST_CASE_REASON("BPF pinning", bpf_pinning, "not compiled in"),
+	TEST_CASE_REASON("BPF prologue generation", bpf_prologue_test, "not compiled in"),
+#endif
+	{ .name = NULL, }
+};
+
+struct test_suite suite__bpf = {
+	.desc = "BPF filter",
+	.test_cases = bpf_tests,
+};
diff --git a/tools/perf/tests/builtin-test.c b/tools/perf/tests/builtin-test.c
index da7dc5e..8cb5a1c 100644
--- a/tools/perf/tests/builtin-test.c
+++ b/tools/perf/tests/builtin-test.c
@@ -30,350 +30,134 @@
 
 static bool dont_fork;
 
-struct test __weak arch_tests[] = {
-	{
-		.func = NULL,
-	},
+struct test_suite *__weak arch_tests[] = {
+	NULL,
 };
 
-static struct test generic_tests[] = {
-	{
-		.desc = "vmlinux symtab matches kallsyms",
-		.func = test__vmlinux_matches_kallsyms,
-	},
-	{
-		.desc = "Detect openat syscall event",
-		.func = test__openat_syscall_event,
-	},
-	{
-		.desc = "Detect openat syscall event on all cpus",
-		.func = test__openat_syscall_event_on_all_cpus,
-	},
-	{
-		.desc = "Read samples using the mmap interface",
-		.func = test__basic_mmap,
-	},
-	{
-		.desc = "Test data source output",
-		.func = test__mem,
-	},
-	{
-		.desc = "Parse event definition strings",
-		.func = test__parse_events,
-	},
-	{
-		.desc = "Simple expression parser",
-		.func = test__expr,
-	},
-	{
-		.desc = "PERF_RECORD_* events & perf_sample fields",
-		.func = test__PERF_RECORD,
-	},
-	{
-		.desc = "Parse perf pmu format",
-		.func = test__pmu,
-	},
-	{
-		.desc = "PMU events",
-		.func = test__pmu_events,
-		.subtest = {
-			.skip_if_fail	= false,
-			.get_nr		= test__pmu_events_subtest_get_nr,
-			.get_desc	= test__pmu_events_subtest_get_desc,
-			.skip_reason	= test__pmu_events_subtest_skip_reason,
-		},
-
-	},
-	{
-		.desc = "DSO data read",
-		.func = test__dso_data,
-	},
-	{
-		.desc = "DSO data cache",
-		.func = test__dso_data_cache,
-	},
-	{
-		.desc = "DSO data reopen",
-		.func = test__dso_data_reopen,
-	},
-	{
-		.desc = "Roundtrip evsel->name",
-		.func = test__perf_evsel__roundtrip_name_test,
-	},
-	{
-		.desc = "Parse sched tracepoints fields",
-		.func = test__perf_evsel__tp_sched_test,
-	},
-	{
-		.desc = "syscalls:sys_enter_openat event fields",
-		.func = test__syscall_openat_tp_fields,
-	},
-	{
-		.desc = "Setup struct perf_event_attr",
-		.func = test__attr,
-	},
-	{
-		.desc = "Match and link multiple hists",
-		.func = test__hists_link,
-	},
-	{
-		.desc = "'import perf' in python",
-		.func = test__python_use,
-	},
-	{
-		.desc = "Breakpoint overflow signal handler",
-		.func = test__bp_signal,
-		.is_supported = test__bp_signal_is_supported,
-	},
-	{
-		.desc = "Breakpoint overflow sampling",
-		.func = test__bp_signal_overflow,
-		.is_supported = test__bp_signal_is_supported,
-	},
-	{
-		.desc = "Breakpoint accounting",
-		.func = test__bp_accounting,
-		.is_supported = test__bp_account_is_supported,
-	},
-	{
-		.desc = "Watchpoint",
-		.func = test__wp,
-		.is_supported = test__wp_is_supported,
-		.subtest = {
-			.skip_if_fail	= false,
-			.get_nr		= test__wp_subtest_get_nr,
-			.get_desc	= test__wp_subtest_get_desc,
-			.skip_reason    = test__wp_subtest_skip_reason,
-		},
-	},
-	{
-		.desc = "Number of exit events of a simple workload",
-		.func = test__task_exit,
-	},
-	{
-		.desc = "Software clock events period values",
-		.func = test__sw_clock_freq,
-	},
-	{
-		.desc = "Object code reading",
-		.func = test__code_reading,
-	},
-	{
-		.desc = "Sample parsing",
-		.func = test__sample_parsing,
-	},
-	{
-		.desc = "Use a dummy software event to keep tracking",
-		.func = test__keep_tracking,
-	},
-	{
-		.desc = "Parse with no sample_id_all bit set",
-		.func = test__parse_no_sample_id_all,
-	},
-	{
-		.desc = "Filter hist entries",
-		.func = test__hists_filter,
-	},
-	{
-		.desc = "Lookup mmap thread",
-		.func = test__mmap_thread_lookup,
-	},
-	{
-		.desc = "Share thread maps",
-		.func = test__thread_maps_share,
-	},
-	{
-		.desc = "Sort output of hist entries",
-		.func = test__hists_output,
-	},
-	{
-		.desc = "Cumulate child hist entries",
-		.func = test__hists_cumulate,
-	},
-	{
-		.desc = "Track with sched_switch",
-		.func = test__switch_tracking,
-	},
-	{
-		.desc = "Filter fds with revents mask in a fdarray",
-		.func = test__fdarray__filter,
-	},
-	{
-		.desc = "Add fd to a fdarray, making it autogrow",
-		.func = test__fdarray__add,
-	},
-	{
-		.desc = "kmod_path__parse",
-		.func = test__kmod_path__parse,
-	},
-	{
-		.desc = "Thread map",
-		.func = test__thread_map,
-	},
-	{
-		.desc = "LLVM search and compile",
-		.func = test__llvm,
-		.subtest = {
-			.skip_if_fail	= true,
-			.get_nr		= test__llvm_subtest_get_nr,
-			.get_desc	= test__llvm_subtest_get_desc,
-		},
-	},
-	{
-		.desc = "Session topology",
-		.func = test__session_topology,
-	},
-	{
-		.desc = "BPF filter",
-		.func = test__bpf,
-		.subtest = {
-			.skip_if_fail	= true,
-			.get_nr		= test__bpf_subtest_get_nr,
-			.get_desc	= test__bpf_subtest_get_desc,
-		},
-	},
-	{
-		.desc = "Synthesize thread map",
-		.func = test__thread_map_synthesize,
-	},
-	{
-		.desc = "Remove thread map",
-		.func = test__thread_map_remove,
-	},
-	{
-		.desc = "Synthesize cpu map",
-		.func = test__cpu_map_synthesize,
-	},
-	{
-		.desc = "Synthesize stat config",
-		.func = test__synthesize_stat_config,
-	},
-	{
-		.desc = "Synthesize stat",
-		.func = test__synthesize_stat,
-	},
-	{
-		.desc = "Synthesize stat round",
-		.func = test__synthesize_stat_round,
-	},
-	{
-		.desc = "Synthesize attr update",
-		.func = test__event_update,
-	},
-	{
-		.desc = "Event times",
-		.func = test__event_times,
-	},
-	{
-		.desc = "Read backward ring buffer",
-		.func = test__backward_ring_buffer,
-	},
-	{
-		.desc = "Print cpu map",
-		.func = test__cpu_map_print,
-	},
-	{
-		.desc = "Merge cpu map",
-		.func = test__cpu_map_merge,
-	},
-
-	{
-		.desc = "Probe SDT events",
-		.func = test__sdt_event,
-	},
-	{
-		.desc = "is_printable_array",
-		.func = test__is_printable_array,
-	},
-	{
-		.desc = "Print bitmap",
-		.func = test__bitmap_print,
-	},
-	{
-		.desc = "perf hooks",
-		.func = test__perf_hooks,
-	},
-	{
-		.desc = "builtin clang support",
-		.func = test__clang,
-		.subtest = {
-			.skip_if_fail	= true,
-			.get_nr		= test__clang_subtest_get_nr,
-			.get_desc	= test__clang_subtest_get_desc,
-		}
-	},
-	{
-		.desc = "unit_number__scnprintf",
-		.func = test__unit_number__scnprint,
-	},
-	{
-		.desc = "mem2node",
-		.func = test__mem2node,
-	},
-	{
-		.desc = "time utils",
-		.func = test__time_utils,
-	},
-	{
-		.desc = "Test jit_write_elf",
-		.func = test__jit_write_elf,
-	},
-	{
-		.desc = "Test libpfm4 support",
-		.func = test__pfm,
-		.subtest = {
-			.skip_if_fail	= true,
-			.get_nr		= test__pfm_subtest_get_nr,
-			.get_desc	= test__pfm_subtest_get_desc,
-		}
-	},
-	{
-		.desc = "Test api io",
-		.func = test__api_io,
-	},
-	{
-		.desc = "maps__merge_in",
-		.func = test__maps__merge_in,
-	},
-	{
-		.desc = "Demangle Java",
-		.func = test__demangle_java,
-	},
-	{
-		.desc = "Demangle OCaml",
-		.func = test__demangle_ocaml,
-	},
-	{
-		.desc = "Parse and process metrics",
-		.func = test__parse_metric,
-	},
-	{
-		.desc = "PE file support",
-		.func = test__pe_file_parsing,
-	},
-	{
-		.desc = "Event expansion for cgroups",
-		.func = test__expand_cgroup_events,
-	},
-	{
-		.desc = "Convert perf time to TSC",
-		.func = test__perf_time_to_tsc,
-		.is_supported = test__tsc_is_supported,
-	},
-	{
-		.desc = "dlfilter C API",
-		.func = test__dlfilter,
-	},
-	{
-		.func = NULL,
-	},
+static struct test_suite *generic_tests[] = {
+	&suite__vmlinux_matches_kallsyms,
+	&suite__openat_syscall_event,
+	&suite__openat_syscall_event_on_all_cpus,
+	&suite__basic_mmap,
+	&suite__mem,
+	&suite__parse_events,
+	&suite__expr,
+	&suite__PERF_RECORD,
+	&suite__pmu,
+	&suite__pmu_events,
+	&suite__dso_data,
+	&suite__dso_data_cache,
+	&suite__dso_data_reopen,
+	&suite__perf_evsel__roundtrip_name_test,
+	&suite__perf_evsel__tp_sched_test,
+	&suite__syscall_openat_tp_fields,
+	&suite__attr,
+	&suite__hists_link,
+	&suite__python_use,
+	&suite__bp_signal,
+	&suite__bp_signal_overflow,
+	&suite__bp_accounting,
+	&suite__wp,
+	&suite__task_exit,
+	&suite__sw_clock_freq,
+	&suite__code_reading,
+	&suite__sample_parsing,
+	&suite__keep_tracking,
+	&suite__parse_no_sample_id_all,
+	&suite__hists_filter,
+	&suite__mmap_thread_lookup,
+	&suite__thread_maps_share,
+	&suite__hists_output,
+	&suite__hists_cumulate,
+	&suite__switch_tracking,
+	&suite__fdarray__filter,
+	&suite__fdarray__add,
+	&suite__kmod_path__parse,
+	&suite__thread_map,
+	&suite__llvm,
+	&suite__session_topology,
+	&suite__bpf,
+	&suite__thread_map_synthesize,
+	&suite__thread_map_remove,
+	&suite__cpu_map_synthesize,
+	&suite__synthesize_stat_config,
+	&suite__synthesize_stat,
+	&suite__synthesize_stat_round,
+	&suite__event_update,
+	&suite__event_times,
+	&suite__backward_ring_buffer,
+	&suite__cpu_map_print,
+	&suite__cpu_map_merge,
+	&suite__sdt_event,
+	&suite__is_printable_array,
+	&suite__bitmap_print,
+	&suite__perf_hooks,
+	&suite__clang,
+	&suite__unit_number__scnprint,
+	&suite__mem2node,
+	&suite__time_utils,
+	&suite__jit_write_elf,
+	&suite__pfm,
+	&suite__api_io,
+	&suite__maps__merge_in,
+	&suite__demangle_java,
+	&suite__demangle_ocaml,
+	&suite__parse_metric,
+	&suite__pe_file_parsing,
+	&suite__expand_cgroup_events,
+	&suite__perf_time_to_tsc,
+	&suite__dlfilter,
+	NULL,
 };
 
-static struct test *tests[] = {
+static struct test_suite **tests[] = {
 	generic_tests,
 	arch_tests,
 };
 
+static int num_subtests(const struct test_suite *t)
+{
+	int num;
+
+	if (!t->test_cases)
+		return 0;
+
+	num = 0;
+	while (t->test_cases[num].name)
+		num++;
+
+	return num;
+}
+
+static bool has_subtests(const struct test_suite *t)
+{
+	return num_subtests(t) > 1;
+}
+
+static const char *skip_reason(const struct test_suite *t, int subtest)
+{
+	if (t->test_cases && subtest >= 0)
+		return t->test_cases[subtest].skip_reason;
+
+	return NULL;
+}
+
+static const char *test_description(const struct test_suite *t, int subtest)
+{
+	if (t->test_cases && subtest >= 0)
+		return t->test_cases[subtest].desc;
+
+	return t->desc;
+}
+
+static test_fnptr test_function(const struct test_suite *t, int subtest)
+{
+	if (subtest <= 0)
+		return t->test_cases[0].run_case;
+
+	return t->test_cases[subtest].run_case;
+}
+
 static bool perf_test__matches(const char *desc, int curr, int argc, const char *argv[])
 {
 	int i;
@@ -398,7 +182,7 @@ static bool perf_test__matches(const char *desc, int curr, int argc, const char
 	return false;
 }
 
-static int run_test(struct test *test, int subtest)
+static int run_test(struct test_suite *test, int subtest)
 {
 	int status, err = -1, child = dont_fork ? 0 : fork();
 	char sbuf[STRERR_BUFSIZE];
@@ -430,7 +214,7 @@ static int run_test(struct test *test, int subtest)
 			}
 		}
 
-		err = test->func(test, subtest);
+		err = test_function(test, subtest)(test, subtest);
 		if (!dont_fork)
 			exit(err);
 	}
@@ -450,24 +234,19 @@ static int run_test(struct test *test, int subtest)
 	return err;
 }
 
-#define for_each_test(j, t)	 				\
+#define for_each_test(j, k, t)			\
 	for (j = 0; j < ARRAY_SIZE(tests); j++)	\
-		for (t = &tests[j][0]; t->func; t++)
+		for (k = 0, t = tests[j][k]; tests[j][k]; k++, t = tests[j][k])
 
-static int test_and_print(struct test *t, bool force_skip, int subtest)
+static int test_and_print(struct test_suite *t, int subtest)
 {
 	int err;
 
-	if (!force_skip) {
-		pr_debug("\n--- start ---\n");
-		err = run_test(t, subtest);
-		pr_debug("---- end ----\n");
-	} else {
-		pr_debug("\n--- force skipped ---\n");
-		err = TEST_SKIP;
-	}
+	pr_debug("\n--- start ---\n");
+	err = run_test(t, subtest);
+	pr_debug("---- end ----\n");
 
-	if (!t->subtest.get_nr)
+	if (!has_subtests(t))
 		pr_debug("%s:", t->desc);
 	else
 		pr_debug("%s subtest %d:", t->desc, subtest + 1);
@@ -477,11 +256,10 @@ static int test_and_print(struct test *t, bool force_skip, int subtest)
 		pr_info(" Ok\n");
 		break;
 	case TEST_SKIP: {
-		const char *skip_reason = NULL;
-		if (t->subtest.skip_reason)
-			skip_reason = t->subtest.skip_reason(subtest);
-		if (skip_reason)
-			color_fprintf(stderr, PERF_COLOR_YELLOW, " Skip (%s)\n", skip_reason);
+		const char *reason = skip_reason(t, subtest);
+
+		if (reason)
+			color_fprintf(stderr, PERF_COLOR_YELLOW, " Skip (%s)\n", reason);
 		else
 			color_fprintf(stderr, PERF_COLOR_YELLOW, " Skip\n");
 	}
@@ -580,7 +358,7 @@ struct shell_test {
 	const char *file;
 };
 
-static int shell_test__run(struct test *test, int subdir __maybe_unused)
+static int shell_test__run(struct test_suite *test, int subdir __maybe_unused)
 {
 	int err;
 	char script[PATH_MAX];
@@ -622,24 +400,34 @@ static int run_shell_tests(int argc, const char *argv[], int i, int width,
 	for_each_shell_test(entlist, n_dirs, st.dir, ent) {
 		int curr = i++;
 		char desc[256];
-		struct test test = {
-			.desc = shell_test__description(desc, sizeof(desc), st.dir, ent->d_name),
-			.func = shell_test__run,
+		struct test_case test_cases[] = {
+			{
+				.desc = shell_test__description(desc,
+								sizeof(desc),
+								st.dir,
+								ent->d_name),
+				.run_case = shell_test__run,
+			},
+			{ .name = NULL, }
+		};
+		struct test_suite test_suite = {
+			.desc = test_cases[0].desc,
+			.test_cases = test_cases,
 			.priv = &st,
 		};
 
-		if (!perf_test__matches(test.desc, curr, argc, argv))
+		if (!perf_test__matches(test_suite.desc, curr, argc, argv))
 			continue;
 
 		st.file = ent->d_name;
-		pr_info("%2d: %-*s:", i, width, test.desc);
+		pr_info("%2d: %-*s:", i, width, test_suite.desc);
 
 		if (intlist__find(skiplist, i)) {
 			color_fprintf(stderr, PERF_COLOR_YELLOW, " Skip (user override)\n");
 			continue;
 		}
 
-		test_and_print(&test, false, -1);
+		test_and_print(&test_suite, 0);
 	}
 
 	for (e = 0; e < n_dirs; e++)
@@ -650,33 +438,31 @@ static int run_shell_tests(int argc, const char *argv[], int i, int width,
 
 static int __cmd_test(int argc, const char *argv[], struct intlist *skiplist)
 {
-	struct test *t;
-	unsigned int j;
+	struct test_suite *t;
+	unsigned int j, k;
 	int i = 0;
 	int width = shell_tests__max_desc_width();
 
-	for_each_test(j, t) {
-		int len = strlen(t->desc);
+	for_each_test(j, k, t) {
+		int len = strlen(test_description(t, -1));
 
 		if (width < len)
 			width = len;
 	}
 
-	for_each_test(j, t) {
-		int curr = i++, err;
+	for_each_test(j, k, t) {
+		int curr = i++;
 		int subi;
 
-		if (!perf_test__matches(t->desc, curr, argc, argv)) {
+		if (!perf_test__matches(test_description(t, -1), curr, argc, argv)) {
 			bool skip = true;
 			int subn;
 
-			if (!t->subtest.get_nr)
-				continue;
-
-			subn = t->subtest.get_nr();
+			subn = num_subtests(t);
 
 			for (subi = 0; subi < subn; subi++) {
-				if (perf_test__matches(t->subtest.get_desc(subi), curr, argc, argv))
+				if (perf_test__matches(test_description(t, subi),
+							curr, argc, argv))
 					skip = false;
 			}
 
@@ -684,22 +470,17 @@ static int __cmd_test(int argc, const char *argv[], struct intlist *skiplist)
 				continue;
 		}
 
-		if (t->is_supported && !t->is_supported()) {
-			pr_debug("%2d: %-*s: Disabled\n", i, width, t->desc);
-			continue;
-		}
-
-		pr_info("%2d: %-*s:", i, width, t->desc);
+		pr_info("%2d: %-*s:", i, width, test_description(t, -1));
 
 		if (intlist__find(skiplist, i)) {
 			color_fprintf(stderr, PERF_COLOR_YELLOW, " Skip (user override)\n");
 			continue;
 		}
 
-		if (!t->subtest.get_nr) {
-			test_and_print(t, false, -1);
+		if (!has_subtests(t)) {
+			test_and_print(t, -1);
 		} else {
-			int subn = t->subtest.get_nr();
+			int subn = num_subtests(t);
 			/*
 			 * minus 2 to align with normal testcases.
 			 * For subtest we print additional '.x' in number.
@@ -709,7 +490,6 @@ static int __cmd_test(int argc, const char *argv[], struct intlist *skiplist)
 			 * 35.1: Basic BPF llvm compiling test                          : Ok
 			 */
 			int subw = width > 2 ? width - 2 : width;
-			bool skip = false;
 
 			if (subn <= 0) {
 				color_fprintf(stderr, PERF_COLOR_YELLOW,
@@ -719,21 +499,20 @@ static int __cmd_test(int argc, const char *argv[], struct intlist *skiplist)
 			pr_info("\n");
 
 			for (subi = 0; subi < subn; subi++) {
-				int len = strlen(t->subtest.get_desc(subi));
+				int len = strlen(test_description(t, subi));
 
 				if (subw < len)
 					subw = len;
 			}
 
 			for (subi = 0; subi < subn; subi++) {
-				if (!perf_test__matches(t->subtest.get_desc(subi), curr, argc, argv))
+				if (!perf_test__matches(test_description(t, subi),
+							curr, argc, argv))
 					continue;
 
 				pr_info("%2d.%1d: %-*s:", i, subi + 1, subw,
-					t->subtest.get_desc(subi));
-				err = test_and_print(t, skip, subi);
-				if (err != TEST_OK && t->subtest.skip_if_fail)
-					skip = true;
+					test_description(t, subi));
+				test_and_print(t, subi);
 			}
 		}
 	}
@@ -759,7 +538,7 @@ static int perf_test__list_shell(int argc, const char **argv, int i)
 	for_each_shell_test(entlist, n_dirs, path, ent) {
 		int curr = i++;
 		char bf[256];
-		struct test t = {
+		struct test_suite t = {
 			.desc = shell_test__description(bf, sizeof(bf), path, ent->d_name),
 		};
 
@@ -778,26 +557,25 @@ static int perf_test__list_shell(int argc, const char **argv, int i)
 
 static int perf_test__list(int argc, const char **argv)
 {
-	unsigned int j;
-	struct test *t;
+	unsigned int j, k;
+	struct test_suite *t;
 	int i = 0;
 
-	for_each_test(j, t) {
+	for_each_test(j, k, t) {
 		int curr = i++;
 
-		if (!perf_test__matches(t->desc, curr, argc, argv) ||
-		    (t->is_supported && !t->is_supported()))
+		if (!perf_test__matches(test_description(t, -1), curr, argc, argv))
 			continue;
 
-		pr_info("%2d: %s\n", i, t->desc);
+		pr_info("%2d: %s\n", i, test_description(t, -1));
 
-		if (t->subtest.get_nr) {
-			int subn = t->subtest.get_nr();
+		if (has_subtests(t)) {
+			int subn = num_subtests(t);
 			int subi;
 
 			for (subi = 0; subi < subn; subi++)
 				pr_info("%2d:%1d: %s\n", i, subi + 1,
-					t->subtest.get_desc(subi));
+					test_description(t, subi));
 		}
 	}
 
diff --git a/tools/perf/tests/clang.c b/tools/perf/tests/clang.c
index 2577d3e..a711100 100644
--- a/tools/perf/tests/clang.c
+++ b/tools/perf/tests/clang.c
@@ -3,44 +3,30 @@
 #include "c++/clang-c.h"
 #include <linux/kernel.h>
 
-static struct {
-	int (*func)(void);
-	const char *desc;
-} clang_testcase_table[] = {
-#ifdef HAVE_LIBCLANGLLVM_SUPPORT
-	{
-		.func = test__clang_to_IR,
-		.desc = "builtin clang compile C source to IR",
-	},
-	{
-		.func = test__clang_to_obj,
-		.desc = "builtin clang compile C source to ELF object",
-	},
-#endif
-};
-
-int test__clang_subtest_get_nr(void)
-{
-	return (int)ARRAY_SIZE(clang_testcase_table);
-}
-
-const char *test__clang_subtest_get_desc(int i)
-{
-	if (i < 0 || i >= (int)ARRAY_SIZE(clang_testcase_table))
-		return NULL;
-	return clang_testcase_table[i].desc;
-}
-
 #ifndef HAVE_LIBCLANGLLVM_SUPPORT
-int test__clang(struct test *test __maybe_unused, int i __maybe_unused)
+static int test__clang_to_IR(struct test_suite *test __maybe_unused,
+			     int subtest __maybe_unused)
 {
 	return TEST_SKIP;
 }
-#else
-int test__clang(struct test *test __maybe_unused, int i)
+
+static int test__clang_to_obj(struct test_suite *test __maybe_unused,
+			      int subtest __maybe_unused)
 {
-	if (i < 0 || i >= (int)ARRAY_SIZE(clang_testcase_table))
-		return TEST_FAIL;
-	return clang_testcase_table[i].func();
+	return TEST_SKIP;
 }
 #endif
+
+static struct test_case clang_tests[] = {
+	TEST_CASE_REASON("builtin clang compile C source to IR", clang_to_IR,
+			 "not compiled in"),
+	TEST_CASE_REASON("builtin clang compile C source to ELF object",
+			 clang_to_obj,
+			 "not compiled in"),
+	{ .name = NULL, }
+};
+
+struct test_suite suite__clang = {
+	.desc = "builtin clang support",
+	.test_cases = clang_tests,
+};
diff --git a/tools/perf/tests/code-reading.c b/tools/perf/tests/code-reading.c
index f439bd4..5610767 100644
--- a/tools/perf/tests/code-reading.c
+++ b/tools/perf/tests/code-reading.c
@@ -716,7 +716,7 @@ static int do_test_code_reading(bool try_kcore)
 	return err;
 }
 
-int test__code_reading(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__code_reading(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int ret;
 
@@ -743,3 +743,5 @@ int test__code_reading(struct test *test __maybe_unused, int subtest __maybe_unu
 		return -1;
 	};
 }
+
+DEFINE_SUITE("Object code reading", code_reading);
diff --git a/tools/perf/tests/cpumap.c b/tools/perf/tests/cpumap.c
index 0472b11..89a1550 100644
--- a/tools/perf/tests/cpumap.c
+++ b/tools/perf/tests/cpumap.c
@@ -75,7 +75,7 @@ static int process_event_cpus(struct perf_tool *tool __maybe_unused,
 }
 
 
-int test__cpu_map_synthesize(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__cpu_map_synthesize(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct perf_cpu_map *cpus;
 
@@ -111,7 +111,7 @@ static int cpu_map_print(const char *str)
 	return !strcmp(buf, str);
 }
 
-int test__cpu_map_print(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__cpu_map_print(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	TEST_ASSERT_VAL("failed to convert map", cpu_map_print("1"));
 	TEST_ASSERT_VAL("failed to convert map", cpu_map_print("1,5"));
@@ -123,7 +123,7 @@ int test__cpu_map_print(struct test *test __maybe_unused, int subtest __maybe_un
 	return 0;
 }
 
-int test__cpu_map_merge(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__cpu_map_merge(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct perf_cpu_map *a = perf_cpu_map__new("4,2,1");
 	struct perf_cpu_map *b = perf_cpu_map__new("4,5,7");
@@ -137,3 +137,7 @@ int test__cpu_map_merge(struct test *test __maybe_unused, int subtest __maybe_un
 	perf_cpu_map__put(c);
 	return 0;
 }
+
+DEFINE_SUITE("Synthesize cpu map", cpu_map_synthesize);
+DEFINE_SUITE("Print cpu map", cpu_map_print);
+DEFINE_SUITE("Merge cpu map", cpu_map_merge);
diff --git a/tools/perf/tests/demangle-java-test.c b/tools/perf/tests/demangle-java-test.c
index 8f3b908..44d1be3 100644
--- a/tools/perf/tests/demangle-java-test.c
+++ b/tools/perf/tests/demangle-java-test.c
@@ -7,7 +7,7 @@
 #include "debug.h"
 #include "demangle-java.h"
 
-int test__demangle_java(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__demangle_java(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int ret = TEST_OK;
 	char *buf = NULL;
@@ -40,3 +40,5 @@ int test__demangle_java(struct test *test __maybe_unused, int subtest __maybe_un
 
 	return ret;
 }
+
+DEFINE_SUITE("Demangle Java", demangle_java);
diff --git a/tools/perf/tests/demangle-ocaml-test.c b/tools/perf/tests/demangle-ocaml-test.c
index 0043be8..90a4285 100644
--- a/tools/perf/tests/demangle-ocaml-test.c
+++ b/tools/perf/tests/demangle-ocaml-test.c
@@ -7,7 +7,7 @@
 #include "debug.h"
 #include "demangle-ocaml.h"
 
-int test__demangle_ocaml(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__demangle_ocaml(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int ret = TEST_OK;
 	char *buf = NULL;
@@ -41,3 +41,5 @@ int test__demangle_ocaml(struct test *test __maybe_unused, int subtest __maybe_u
 
 	return ret;
 }
+
+DEFINE_SUITE("Demangle OCaml", demangle_ocaml);
diff --git a/tools/perf/tests/dlfilter-test.c b/tools/perf/tests/dlfilter-test.c
index bc03b5d..84352d5 100644
--- a/tools/perf/tests/dlfilter-test.c
+++ b/tools/perf/tests/dlfilter-test.c
@@ -398,7 +398,7 @@ static void test_data__free(struct test_data *td)
 	}
 }
 
-int test__dlfilter(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__dlfilter(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct test_data td = {.fd = -1};
 	int pid = getpid();
@@ -414,3 +414,5 @@ int test__dlfilter(struct test *test __maybe_unused, int subtest __maybe_unused)
 	test_data__free(&td);
 	return err;
 }
+
+DEFINE_SUITE("dlfilter C API", dlfilter);
diff --git a/tools/perf/tests/dso-data.c b/tools/perf/tests/dso-data.c
index 43e1b01..3419a4a 100644
--- a/tools/perf/tests/dso-data.c
+++ b/tools/perf/tests/dso-data.c
@@ -113,7 +113,7 @@ static int dso__data_fd(struct dso *dso, struct machine *machine)
 	return fd;
 }
 
-int test__dso_data(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__dso_data(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct machine machine;
 	struct dso *dso;
@@ -248,7 +248,7 @@ static int set_fd_limit(int n)
 	return setrlimit(RLIMIT_NOFILE, &rlim);
 }
 
-int test__dso_data_cache(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__dso_data_cache(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct machine machine;
 	long nr_end, nr = open_files_cnt();
@@ -318,7 +318,7 @@ static long new_limit(int count)
 	return ret;
 }
 
-int test__dso_data_reopen(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__dso_data_reopen(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct machine machine;
 	long nr_end, nr = open_files_cnt(), lim = new_limit(3);
@@ -393,3 +393,7 @@ int test__dso_data_reopen(struct test *test __maybe_unused, int subtest __maybe_
 	TEST_ASSERT_VAL("failed leaking files", nr == nr_end);
 	return 0;
 }
+
+DEFINE_SUITE("DSO data read", dso_data);
+DEFINE_SUITE("DSO data cache", dso_data_cache);
+DEFINE_SUITE("DSO data reopen", dso_data_reopen);
diff --git a/tools/perf/tests/dwarf-unwind.c b/tools/perf/tests/dwarf-unwind.c
index c756284..2dab2d2 100644
--- a/tools/perf/tests/dwarf-unwind.c
+++ b/tools/perf/tests/dwarf-unwind.c
@@ -195,7 +195,8 @@ NO_TAIL_CALL_ATTRIBUTE noinline int test_dwarf_unwind__krava_1(struct thread *th
 	return ret;
 }
 
-int test__dwarf_unwind(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__dwarf_unwind(struct test_suite *test __maybe_unused,
+			      int subtest __maybe_unused)
 {
 	struct machine *machine;
 	struct thread *thread;
@@ -237,3 +238,5 @@ int test__dwarf_unwind(struct test *test __maybe_unused, int subtest __maybe_unu
 	machine__delete(machine);
 	return err;
 }
+
+DEFINE_SUITE("Test dwarf unwind", dwarf_unwind);
diff --git a/tools/perf/tests/event-times.c b/tools/perf/tests/event-times.c
index 04ce440..7606eb3 100644
--- a/tools/perf/tests/event-times.c
+++ b/tools/perf/tests/event-times.c
@@ -216,7 +216,7 @@ static int test_times(int (attach)(struct evlist *),
  * and checks that enabled and running times
  * match.
  */
-int test__event_times(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__event_times(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int err, ret = 0;
 
@@ -239,3 +239,5 @@ int test__event_times(struct test *test __maybe_unused, int subtest __maybe_unus
 #undef _T
 	return ret;
 }
+
+DEFINE_SUITE("Event times", event_times);
diff --git a/tools/perf/tests/event_update.c b/tools/perf/tests/event_update.c
index 44a5052..fbb68de 100644
--- a/tools/perf/tests/event_update.c
+++ b/tools/perf/tests/event_update.c
@@ -83,7 +83,7 @@ static int process_event_cpus(struct perf_tool *tool __maybe_unused,
 	return 0;
 }
 
-int test__event_update(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__event_update(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct evsel *evsel;
 	struct event_name tmp;
@@ -123,3 +123,5 @@ int test__event_update(struct test *test __maybe_unused, int subtest __maybe_unu
 	evlist__delete(evlist);
 	return 0;
 }
+
+DEFINE_SUITE("Synthesize attr update", event_update);
diff --git a/tools/perf/tests/evsel-roundtrip-name.c b/tools/perf/tests/evsel-roundtrip-name.c
index 4e09f0a..fdbf176 100644
--- a/tools/perf/tests/evsel-roundtrip-name.c
+++ b/tools/perf/tests/evsel-roundtrip-name.c
@@ -99,7 +99,8 @@ static int __perf_evsel__name_array_test(const char *names[], int nr_names,
 #define perf_evsel__name_array_test(names, distance) \
 	__perf_evsel__name_array_test(names, ARRAY_SIZE(names), distance)
 
-int test__perf_evsel__roundtrip_name_test(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__perf_evsel__roundtrip_name_test(struct test_suite *test __maybe_unused,
+						 int subtest __maybe_unused)
 {
 	int err = 0, ret = 0;
 
@@ -120,3 +121,5 @@ int test__perf_evsel__roundtrip_name_test(struct test *test __maybe_unused, int
 
 	return ret;
 }
+
+DEFINE_SUITE("Roundtrip evsel->name", perf_evsel__roundtrip_name_test);
diff --git a/tools/perf/tests/evsel-tp-sched.c b/tools/perf/tests/evsel-tp-sched.c
index f9e34bd..cf4da3d 100644
--- a/tools/perf/tests/evsel-tp-sched.c
+++ b/tools/perf/tests/evsel-tp-sched.c
@@ -32,7 +32,8 @@ static int evsel__test_field(struct evsel *evsel, const char *name, int size, bo
 	return ret;
 }
 
-int test__perf_evsel__tp_sched_test(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__perf_evsel__tp_sched_test(struct test_suite *test __maybe_unused,
+					   int subtest __maybe_unused)
 {
 	struct evsel *evsel = evsel__newtp("sched", "sched_switch");
 	int ret = 0;
@@ -87,3 +88,5 @@ int test__perf_evsel__tp_sched_test(struct test *test __maybe_unused, int subtes
 	evsel__delete(evsel);
 	return ret;
 }
+
+DEFINE_SUITE("Parse sched tracepoints fields", perf_evsel__tp_sched_test);
diff --git a/tools/perf/tests/expand-cgroup.c b/tools/perf/tests/expand-cgroup.c
index 80cff8a..dfefe5b 100644
--- a/tools/perf/tests/expand-cgroup.c
+++ b/tools/perf/tests/expand-cgroup.c
@@ -221,8 +221,8 @@ static int expand_metric_events(void)
 	return ret;
 }
 
-int test__expand_cgroup_events(struct test *test __maybe_unused,
-			       int subtest __maybe_unused)
+static int test__expand_cgroup_events(struct test_suite *test __maybe_unused,
+				      int subtest __maybe_unused)
 {
 	int ret;
 
@@ -240,3 +240,5 @@ int test__expand_cgroup_events(struct test *test __maybe_unused,
 
 	return ret;
 }
+
+DEFINE_SUITE("Event expansion for cgroups", expand_cgroup_events);
diff --git a/tools/perf/tests/expr.c b/tools/perf/tests/expr.c
index 0777832..c895de4 100644
--- a/tools/perf/tests/expr.c
+++ b/tools/perf/tests/expr.c
@@ -62,11 +62,11 @@ static int test(struct expr_parse_ctx *ctx, const char *e, double val2)
 	return 0;
 }
 
-int test__expr(struct test *t __maybe_unused, int subtest __maybe_unused)
+static int test__expr(struct test_suite *t __maybe_unused, int subtest __maybe_unused)
 {
 	struct expr_id_data *val_ptr;
 	const char *p;
-	double val;
+	double val, num_cpus, num_cores, num_dies, num_packages;
 	int ret;
 	struct expr_parse_ctx *ctx;
 
@@ -134,6 +134,16 @@ int test__expr(struct test *t __maybe_unused, int subtest __maybe_unused)
 	TEST_ASSERT_VAL("find ids", hashmap__find(ctx->ids, "EVENT2,param=3@",
 						    (void **)&val_ptr));
 
+	expr__ctx_clear(ctx);
+	TEST_ASSERT_VAL("find ids",
+			expr__find_ids("dash\\-event1 - dash\\-event2",
+				       NULL, ctx) == 0);
+	TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 2);
+	TEST_ASSERT_VAL("find ids", hashmap__find(ctx->ids, "dash-event1",
+						    (void **)&val_ptr));
+	TEST_ASSERT_VAL("find ids", hashmap__find(ctx->ids, "dash-event2",
+						    (void **)&val_ptr));
+
 	/* Only EVENT1 or EVENT2 need be measured depending on the value of smt_on. */
 	expr__ctx_clear(ctx);
 	TEST_ASSERT_VAL("find ids",
@@ -151,7 +161,31 @@ int test__expr(struct test *t __maybe_unused, int subtest __maybe_unused)
 			NULL, ctx) == 0);
 	TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 0);
 
+	/* Test toplogy constants appear well ordered. */
+	expr__ctx_clear(ctx);
+	TEST_ASSERT_VAL("#num_cpus", expr__parse(&num_cpus, ctx, "#num_cpus") == 0);
+	TEST_ASSERT_VAL("#num_cores", expr__parse(&num_cores, ctx, "#num_cores") == 0);
+	TEST_ASSERT_VAL("#num_cpus >= #num_cores", num_cpus >= num_cores);
+	TEST_ASSERT_VAL("#num_dies", expr__parse(&num_dies, ctx, "#num_dies") == 0);
+	TEST_ASSERT_VAL("#num_cores >= #num_dies", num_cores >= num_dies);
+	TEST_ASSERT_VAL("#num_packages", expr__parse(&num_packages, ctx, "#num_packages") == 0);
+	TEST_ASSERT_VAL("#num_dies >= #num_packages", num_dies >= num_packages);
+
+	/*
+	 * Source count returns the number of events aggregating in a leader
+	 * event including the leader. Check parsing yields an id.
+	 */
+	expr__ctx_clear(ctx);
+	TEST_ASSERT_VAL("source count",
+			expr__find_ids("source_count(EVENT1)",
+			NULL, ctx) == 0);
+	TEST_ASSERT_VAL("source count", hashmap__size(ctx->ids) == 1);
+	TEST_ASSERT_VAL("source count", hashmap__find(ctx->ids, "EVENT1",
+							(void **)&val_ptr));
+
 	expr__ctx_free(ctx);
 
 	return 0;
 }
+
+DEFINE_SUITE("Simple expression parser", expr);
diff --git a/tools/perf/tests/fdarray.c b/tools/perf/tests/fdarray.c
index d9eca8e..40983c3 100644
--- a/tools/perf/tests/fdarray.c
+++ b/tools/perf/tests/fdarray.c
@@ -28,7 +28,7 @@ static int fdarray__fprintf_prefix(struct fdarray *fda, const char *prefix, FILE
 	return printed + fdarray__fprintf(fda, fp);
 }
 
-int test__fdarray__filter(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__fdarray__filter(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int nr_fds, err = TEST_FAIL;
 	struct fdarray *fda = fdarray__new(5, 5);
@@ -89,7 +89,7 @@ int test__fdarray__filter(struct test *test __maybe_unused, int subtest __maybe_
 	return err;
 }
 
-int test__fdarray__add(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__fdarray__add(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int err = TEST_FAIL;
 	struct fdarray *fda = fdarray__new(2, 2);
@@ -158,3 +158,6 @@ int test__fdarray__add(struct test *test __maybe_unused, int subtest __maybe_unu
 out:
 	return err;
 }
+
+DEFINE_SUITE("Filter fds with revents mask in a fdarray", fdarray__filter);
+DEFINE_SUITE("Add fd to a fdarray, making it autogrow", fdarray__add);
diff --git a/tools/perf/tests/genelf.c b/tools/perf/tests/genelf.c
index f797f98..95f3be1 100644
--- a/tools/perf/tests/genelf.c
+++ b/tools/perf/tests/genelf.c
@@ -16,8 +16,8 @@
 
 #define TEMPL "/tmp/perf-test-XXXXXX"
 
-int test__jit_write_elf(struct test *test __maybe_unused,
-			int subtest __maybe_unused)
+static int test__jit_write_elf(struct test_suite *test __maybe_unused,
+			       int subtest __maybe_unused)
 {
 #ifdef HAVE_JITDUMP
 	static unsigned char x86_code[] = {
@@ -49,3 +49,5 @@ int test__jit_write_elf(struct test *test __maybe_unused,
 	return TEST_SKIP;
 #endif
 }
+
+DEFINE_SUITE("Test jit_write_elf", jit_write_elf);
diff --git a/tools/perf/tests/hists_cumulate.c b/tools/perf/tests/hists_cumulate.c
index 890cb1f..17f4fcd 100644
--- a/tools/perf/tests/hists_cumulate.c
+++ b/tools/perf/tests/hists_cumulate.c
@@ -689,7 +689,7 @@ static int test4(struct evsel *evsel, struct machine *machine)
 	return err;
 }
 
-int test__hists_cumulate(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__hists_cumulate(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int err = TEST_FAIL;
 	struct machines machines;
@@ -736,3 +736,5 @@ int test__hists_cumulate(struct test *test __maybe_unused, int subtest __maybe_u
 
 	return err;
 }
+
+DEFINE_SUITE("Cumulate child hist entries", hists_cumulate);
diff --git a/tools/perf/tests/hists_filter.c b/tools/perf/tests/hists_filter.c
index ca6120c..08cbeb9 100644
--- a/tools/perf/tests/hists_filter.c
+++ b/tools/perf/tests/hists_filter.c
@@ -101,7 +101,7 @@ static int add_hist_entries(struct evlist *evlist,
 	return TEST_FAIL;
 }
 
-int test__hists_filter(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__hists_filter(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int err = TEST_FAIL;
 	struct machines machines;
@@ -325,3 +325,5 @@ int test__hists_filter(struct test *test __maybe_unused, int subtest __maybe_unu
 
 	return err;
 }
+
+DEFINE_SUITE("Filter hist entries", hists_filter);
diff --git a/tools/perf/tests/hists_link.c b/tools/perf/tests/hists_link.c
index a024d3f..c575e13 100644
--- a/tools/perf/tests/hists_link.c
+++ b/tools/perf/tests/hists_link.c
@@ -264,7 +264,7 @@ static int validate_link(struct hists *leader, struct hists *other)
 	return __validate_link(leader, 0) || __validate_link(other, 1);
 }
 
-int test__hists_link(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__hists_link(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int err = -1;
 	struct hists *hists, *first_hists;
@@ -339,3 +339,5 @@ int test__hists_link(struct test *test __maybe_unused, int subtest __maybe_unuse
 
 	return err;
 }
+
+DEFINE_SUITE("Match and link multiple hists", hists_link);
diff --git a/tools/perf/tests/hists_output.c b/tools/perf/tests/hists_output.c
index 8973f35..0bde4a7 100644
--- a/tools/perf/tests/hists_output.c
+++ b/tools/perf/tests/hists_output.c
@@ -575,7 +575,7 @@ static int test5(struct evsel *evsel, struct machine *machine)
 	return err;
 }
 
-int test__hists_output(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__hists_output(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int err = TEST_FAIL;
 	struct machines machines;
@@ -623,3 +623,5 @@ int test__hists_output(struct test *test __maybe_unused, int subtest __maybe_unu
 
 	return err;
 }
+
+DEFINE_SUITE("Sort output of hist entries", hists_output);
diff --git a/tools/perf/tests/is_printable_array.c b/tools/perf/tests/is_printable_array.c
index 9c7b3ba..f72de24 100644
--- a/tools/perf/tests/is_printable_array.c
+++ b/tools/perf/tests/is_printable_array.c
@@ -5,7 +5,7 @@
 #include "debug.h"
 #include "print_binary.h"
 
-int test__is_printable_array(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__is_printable_array(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	char buf1[] = { 'k', 'r', 4, 'v', 'a', 0 };
 	char buf2[] = { 'k', 'r', 'a', 'v', 4, 0 };
@@ -36,3 +36,5 @@ int test__is_printable_array(struct test *test __maybe_unused, int subtest __may
 
 	return TEST_OK;
 }
+
+DEFINE_SUITE("is_printable_array", is_printable_array);
diff --git a/tools/perf/tests/keep-tracking.c b/tools/perf/tests/keep-tracking.c
index a0438b0..dd20673 100644
--- a/tools/perf/tests/keep-tracking.c
+++ b/tools/perf/tests/keep-tracking.c
@@ -61,7 +61,7 @@ static int find_comm(struct evlist *evlist, const char *comm)
  * when an event is disabled but a dummy software event is not disabled.  If the
  * test passes %0 is returned, otherwise %-1 is returned.
  */
-int test__keep_tracking(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__keep_tracking(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct record_opts opts = {
 		.mmap_pages	     = UINT_MAX,
@@ -160,3 +160,5 @@ int test__keep_tracking(struct test *test __maybe_unused, int subtest __maybe_un
 
 	return err;
 }
+
+DEFINE_SUITE("Use a dummy software event to keep tracking", keep_tracking);
diff --git a/tools/perf/tests/kmod-path.c b/tools/perf/tests/kmod-path.c
index e483210..dfe1bd5 100644
--- a/tools/perf/tests/kmod-path.c
+++ b/tools/perf/tests/kmod-path.c
@@ -47,7 +47,7 @@ static int test_is_kernel_module(const char *path, int cpumode, bool expect)
 #define M(path, c, e) \
 	TEST_ASSERT_VAL("failed", !test_is_kernel_module(path, c, e))
 
-int test__kmod_path__parse(struct test *t __maybe_unused, int subtest __maybe_unused)
+static int test__kmod_path__parse(struct test_suite *t __maybe_unused, int subtest __maybe_unused)
 {
 	/* path                alloc_name  kmod  comp   name   */
 	T("/xxxx/xxxx/x-x.ko", true      , true, 0    , "[x_x]");
@@ -159,3 +159,5 @@ int test__kmod_path__parse(struct test *t __maybe_unused, int subtest __maybe_un
 
 	return 0;
 }
+
+DEFINE_SUITE("kmod_path__parse", kmod_path__parse);
diff --git a/tools/perf/tests/llvm.c b/tools/perf/tests/llvm.c
index 33e43cc..8ac0a3a 100644
--- a/tools/perf/tests/llvm.c
+++ b/tools/perf/tests/llvm.c
@@ -124,7 +124,7 @@ test_llvm__fetch_bpf_obj(void **p_obj_buf,
 	return ret;
 }
 
-int test__llvm(struct test *test __maybe_unused, int subtest)
+static int test__llvm(int subtest)
 {
 	int ret;
 	void *obj_buf = NULL;
@@ -148,32 +148,72 @@ int test__llvm(struct test *test __maybe_unused, int subtest)
 
 	return ret;
 }
+#endif //HAVE_LIBBPF_SUPPORT
 
-int test__llvm_subtest_get_nr(void)
+static int test__llvm__bpf_base_prog(struct test_suite *test __maybe_unused,
+				     int subtest __maybe_unused)
 {
-	return __LLVM_TESTCASE_MAX;
-}
-
-const char *test__llvm_subtest_get_desc(int subtest)
-{
-	if ((subtest < 0) || (subtest >= __LLVM_TESTCASE_MAX))
-		return NULL;
-
-	return bpf_source_table[subtest].desc;
-}
-#else //HAVE_LIBBPF_SUPPORT
-int test__llvm(struct test *test __maybe_unused, int subtest __maybe_unused)
-{
+#ifdef HAVE_LIBBPF_SUPPORT
+	return test__llvm(LLVM_TESTCASE_BASE);
+#else
+	pr_debug("Skip LLVM test because BPF support is not compiled\n");
 	return TEST_SKIP;
+#endif
 }
 
-int test__llvm_subtest_get_nr(void)
+static int test__llvm__bpf_test_kbuild_prog(struct test_suite *test __maybe_unused,
+					    int subtest __maybe_unused)
 {
-	return 0;
+#ifdef HAVE_LIBBPF_SUPPORT
+	return test__llvm(LLVM_TESTCASE_KBUILD);
+#else
+	pr_debug("Skip LLVM test because BPF support is not compiled\n");
+	return TEST_SKIP;
+#endif
 }
 
-const char *test__llvm_subtest_get_desc(int subtest __maybe_unused)
+static int test__llvm__bpf_test_prologue_prog(struct test_suite *test __maybe_unused,
+					      int subtest __maybe_unused)
 {
-	return NULL;
+#ifdef HAVE_LIBBPF_SUPPORT
+	return test__llvm(LLVM_TESTCASE_BPF_PROLOGUE);
+#else
+	pr_debug("Skip LLVM test because BPF support is not compiled\n");
+	return TEST_SKIP;
+#endif
 }
-#endif // HAVE_LIBBPF_SUPPORT
+
+static int test__llvm__bpf_test_relocation(struct test_suite *test __maybe_unused,
+					   int subtest __maybe_unused)
+{
+#ifdef HAVE_LIBBPF_SUPPORT
+	return test__llvm(LLVM_TESTCASE_BPF_RELOCATION);
+#else
+	pr_debug("Skip LLVM test because BPF support is not compiled\n");
+	return TEST_SKIP;
+#endif
+}
+
+
+static struct test_case llvm_tests[] = {
+#ifdef HAVE_LIBBPF_SUPPORT
+	TEST_CASE("Basic BPF llvm compile", llvm__bpf_base_prog),
+	TEST_CASE("kbuild searching", llvm__bpf_test_kbuild_prog),
+	TEST_CASE("Compile source for BPF prologue generation",
+		  llvm__bpf_test_prologue_prog),
+	TEST_CASE("Compile source for BPF relocation", llvm__bpf_test_relocation),
+#else
+	TEST_CASE_REASON("Basic BPF llvm compile", llvm__bpf_base_prog, "not compiled in"),
+	TEST_CASE_REASON("kbuild searching", llvm__bpf_test_kbuild_prog, "not compiled in"),
+	TEST_CASE_REASON("Compile source for BPF prologue generation",
+			llvm__bpf_test_prologue_prog, "not compiled in"),
+	TEST_CASE_REASON("Compile source for BPF relocation",
+			llvm__bpf_test_relocation, "not compiled in"),
+#endif
+	{ .name = NULL, }
+};
+
+struct test_suite suite__llvm = {
+	.desc = "LLVM search and compile",
+	.test_cases = llvm_tests,
+};
diff --git a/tools/perf/tests/maps.c b/tools/perf/tests/maps.c
index 1ac7291..e308a32 100644
--- a/tools/perf/tests/maps.c
+++ b/tools/perf/tests/maps.c
@@ -33,7 +33,7 @@ static int check_maps(struct map_def *merged, unsigned int size, struct maps *ma
 	return TEST_OK;
 }
 
-int test__maps__merge_in(struct test *t __maybe_unused, int subtest __maybe_unused)
+static int test__maps__merge_in(struct test_suite *t __maybe_unused, int subtest __maybe_unused)
 {
 	struct maps maps;
 	unsigned int i;
@@ -120,3 +120,5 @@ int test__maps__merge_in(struct test *t __maybe_unused, int subtest __maybe_unus
 	maps__exit(&maps);
 	return TEST_OK;
 }
+
+DEFINE_SUITE("maps__merge_in", maps__merge_in);
diff --git a/tools/perf/tests/mem.c b/tools/perf/tests/mem.c
index 673a11a6..56014ec 100644
--- a/tools/perf/tests/mem.c
+++ b/tools/perf/tests/mem.c
@@ -23,7 +23,7 @@ static int check(union perf_mem_data_src data_src,
 	return 0;
 }
 
-int test__mem(struct test *text __maybe_unused, int subtest __maybe_unused)
+static int test__mem(struct test_suite *text __maybe_unused, int subtest __maybe_unused)
 {
 	int ret = 0;
 	union perf_mem_data_src src;
@@ -56,3 +56,5 @@ int test__mem(struct test *text __maybe_unused, int subtest __maybe_unused)
 
 	return ret;
 }
+
+DEFINE_SUITE("Test data source output", mem);
diff --git a/tools/perf/tests/mem2node.c b/tools/perf/tests/mem2node.c
index e4d0d58..b17b863 100644
--- a/tools/perf/tests/mem2node.c
+++ b/tools/perf/tests/mem2node.c
@@ -43,7 +43,7 @@ static unsigned long *get_bitmap(const char *str, int nbits)
 	return bm && map ? bm : NULL;
 }
 
-int test__mem2node(struct test *t __maybe_unused, int subtest __maybe_unused)
+static int test__mem2node(struct test_suite *t __maybe_unused, int subtest __maybe_unused)
 {
 	struct mem2node map;
 	struct memory_node nodes[3];
@@ -77,3 +77,5 @@ int test__mem2node(struct test *t __maybe_unused, int subtest __maybe_unused)
 	mem2node__exit(&map);
 	return 0;
 }
+
+DEFINE_SUITE("mem2node", mem2node);
diff --git a/tools/perf/tests/mmap-basic.c b/tools/perf/tests/mmap-basic.c
index d38757d..90b2fed 100644
--- a/tools/perf/tests/mmap-basic.c
+++ b/tools/perf/tests/mmap-basic.c
@@ -29,7 +29,7 @@
  * Then it checks if the number of syscalls reported as perf events by
  * the kernel corresponds to the number of syscalls made.
  */
-int test__basic_mmap(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__basic_mmap(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int err = -1;
 	union perf_event *event;
@@ -164,3 +164,5 @@ int test__basic_mmap(struct test *test __maybe_unused, int subtest __maybe_unuse
 	perf_thread_map__put(threads);
 	return err;
 }
+
+DEFINE_SUITE("Read samples using the mmap interface", basic_mmap);
diff --git a/tools/perf/tests/mmap-thread-lookup.c b/tools/perf/tests/mmap-thread-lookup.c
index 6f2da7a..a4301fc 100644
--- a/tools/perf/tests/mmap-thread-lookup.c
+++ b/tools/perf/tests/mmap-thread-lookup.c
@@ -224,7 +224,7 @@ static int mmap_events(synth_cb synth)
  *
  * by using all thread objects.
  */
-int test__mmap_thread_lookup(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__mmap_thread_lookup(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	/* perf_event__synthesize_threads synthesize */
 	TEST_ASSERT_VAL("failed with sythesizing all",
@@ -236,3 +236,5 @@ int test__mmap_thread_lookup(struct test *test __maybe_unused, int subtest __may
 
 	return 0;
 }
+
+DEFINE_SUITE("Lookup mmap thread", mmap_thread_lookup);
diff --git a/tools/perf/tests/openat-syscall-all-cpus.c b/tools/perf/tests/openat-syscall-all-cpus.c
index f7dd6c46..cd3dd46 100644
--- a/tools/perf/tests/openat-syscall-all-cpus.c
+++ b/tools/perf/tests/openat-syscall-all-cpus.c
@@ -19,7 +19,8 @@
 #include "stat.h"
 #include "util/counts.h"
 
-int test__openat_syscall_event_on_all_cpus(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__openat_syscall_event_on_all_cpus(struct test_suite *test __maybe_unused,
+						  int subtest __maybe_unused)
 {
 	int err = -1, fd, cpu;
 	struct perf_cpu_map *cpus;
@@ -127,3 +128,5 @@ int test__openat_syscall_event_on_all_cpus(struct test *test __maybe_unused, int
 	perf_thread_map__put(threads);
 	return err;
 }
+
+DEFINE_SUITE("Detect openat syscall event on all cpus", openat_syscall_event_on_all_cpus);
diff --git a/tools/perf/tests/openat-syscall-tp-fields.c b/tools/perf/tests/openat-syscall-tp-fields.c
index 5e4af2f..a7b2800 100644
--- a/tools/perf/tests/openat-syscall-tp-fields.c
+++ b/tools/perf/tests/openat-syscall-tp-fields.c
@@ -22,7 +22,8 @@
 #define AT_FDCWD       -100
 #endif
 
-int test__syscall_openat_tp_fields(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__syscall_openat_tp_fields(struct test_suite *test __maybe_unused,
+					  int subtest __maybe_unused)
 {
 	struct record_opts opts = {
 		.target = {
@@ -142,3 +143,5 @@ int test__syscall_openat_tp_fields(struct test *test __maybe_unused, int subtest
 out:
 	return err;
 }
+
+DEFINE_SUITE("syscalls:sys_enter_openat event fields", syscall_openat_tp_fields);
diff --git a/tools/perf/tests/openat-syscall.c b/tools/perf/tests/openat-syscall.c
index 85a8f0f..7f4c13c4 100644
--- a/tools/perf/tests/openat-syscall.c
+++ b/tools/perf/tests/openat-syscall.c
@@ -13,7 +13,8 @@
 #include "tests.h"
 #include "util/counts.h"
 
-int test__openat_syscall_event(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__openat_syscall_event(struct test_suite *test __maybe_unused,
+				      int subtest __maybe_unused)
 {
 	int err = -1, fd;
 	struct evsel *evsel;
@@ -66,3 +67,5 @@ int test__openat_syscall_event(struct test *test __maybe_unused, int subtest __m
 	perf_thread_map__put(threads);
 	return err;
 }
+
+DEFINE_SUITE("Detect openat syscall event", openat_syscall_event);
diff --git a/tools/perf/tests/parse-events.c b/tools/perf/tests/parse-events.c
index 6af9463..a508f1d 100644
--- a/tools/perf/tests/parse-events.c
+++ b/tools/perf/tests/parse-events.c
@@ -2276,7 +2276,7 @@ static int test_pmu_events_alias(char *event, char *alias)
 	return test_event(&e);
 }
 
-int test__parse_events(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__parse_events(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int ret1, ret2 = 0;
 	char *event, *alias;
@@ -2319,3 +2319,5 @@ do {							\
 
 	return ret2;
 }
+
+DEFINE_SUITE("Parse event definition strings", parse_events);
diff --git a/tools/perf/tests/parse-metric.c b/tools/perf/tests/parse-metric.c
index dfc797e..574b7e4 100644
--- a/tools/perf/tests/parse-metric.c
+++ b/tools/perf/tests/parse-metric.c
@@ -369,7 +369,7 @@ static int test_metric_group(void)
 	return 0;
 }
 
-int test__parse_metric(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__parse_metric(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	TEST_ASSERT_VAL("IPC failed", test_ipc() == 0);
 	TEST_ASSERT_VAL("frontend failed", test_frontend() == 0);
@@ -383,3 +383,5 @@ int test__parse_metric(struct test *test __maybe_unused, int subtest __maybe_unu
 	}
 	return 0;
 }
+
+DEFINE_SUITE("Parse and process metrics", parse_metric);
diff --git a/tools/perf/tests/parse-no-sample-id-all.c b/tools/perf/tests/parse-no-sample-id-all.c
index 4712736..d62e315 100644
--- a/tools/perf/tests/parse-no-sample-id-all.c
+++ b/tools/perf/tests/parse-no-sample-id-all.c
@@ -67,7 +67,8 @@ struct test_attr_event {
  *
  * Return: %0 on success, %-1 if the test fails.
  */
-int test__parse_no_sample_id_all(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__parse_no_sample_id_all(struct test_suite *test __maybe_unused,
+					int subtest __maybe_unused)
 {
 	int err;
 
@@ -103,3 +104,5 @@ int test__parse_no_sample_id_all(struct test *test __maybe_unused, int subtest _
 
 	return 0;
 }
+
+DEFINE_SUITE("Parse with no sample_id_all bit set", parse_no_sample_id_all);
diff --git a/tools/perf/tests/pe-file-parsing.c b/tools/perf/tests/pe-file-parsing.c
index 58b90c4..c09a9fa 100644
--- a/tools/perf/tests/pe-file-parsing.c
+++ b/tools/perf/tests/pe-file-parsing.c
@@ -68,7 +68,7 @@ static int run_dir(const char *d)
 	return TEST_OK;
 }
 
-int test__pe_file_parsing(struct test *test __maybe_unused,
+static int test__pe_file_parsing(struct test_suite *test __maybe_unused,
 			  int subtest __maybe_unused)
 {
 	struct stat st;
@@ -89,10 +89,12 @@ int test__pe_file_parsing(struct test *test __maybe_unused,
 
 #else
 
-int test__pe_file_parsing(struct test *test __maybe_unused,
+static int test__pe_file_parsing(struct test_suite *test __maybe_unused,
 			  int subtest __maybe_unused)
 {
 	return TEST_SKIP;
 }
 
 #endif
+
+DEFINE_SUITE("PE file support", pe_file_parsing);
diff --git a/tools/perf/tests/perf-hooks.c b/tools/perf/tests/perf-hooks.c
index dd865e0..78cdeb8 100644
--- a/tools/perf/tests/perf-hooks.c
+++ b/tools/perf/tests/perf-hooks.c
@@ -26,7 +26,7 @@ static void the_hook(void *_hook_flags)
 	raise(SIGSEGV);
 }
 
-int test__perf_hooks(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__perf_hooks(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int hook_flags = 0;
 
@@ -45,3 +45,5 @@ int test__perf_hooks(struct test *test __maybe_unused, int subtest __maybe_unuse
 		return TEST_FAIL;
 	return TEST_OK;
 }
+
+DEFINE_SUITE("perf hooks", perf_hooks);
diff --git a/tools/perf/tests/perf-record.c b/tools/perf/tests/perf-record.c
index 0df471b..6354465 100644
--- a/tools/perf/tests/perf-record.c
+++ b/tools/perf/tests/perf-record.c
@@ -41,7 +41,7 @@ static int sched__get_first_possible_cpu(pid_t pid, cpu_set_t *maskp)
 	return cpu;
 }
 
-int test__PERF_RECORD(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__PERF_RECORD(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct record_opts opts = {
 		.target = {
@@ -332,3 +332,5 @@ int test__PERF_RECORD(struct test *test __maybe_unused, int subtest __maybe_unus
 out:
 	return (err < 0 || errs > 0) ? -1 : 0;
 }
+
+DEFINE_SUITE("PERF_RECORD_* events & perf_sample fields", PERF_RECORD);
diff --git a/tools/perf/tests/perf-time-to-tsc.c b/tools/perf/tests/perf-time-to-tsc.c
index 7c56bc1..d12d0ad 100644
--- a/tools/perf/tests/perf-time-to-tsc.c
+++ b/tools/perf/tests/perf-time-to-tsc.c
@@ -23,6 +23,16 @@
 #include "pmu.h"
 #include "pmu-hybrid.h"
 
+/*
+ * Except x86_64/i386 and Arm64, other archs don't support TSC in perf.  Just
+ * enable the test for x86_64/i386 and Arm64 archs.
+ */
+#if defined(__x86_64__) || defined(__i386__) || defined(__aarch64__)
+#define TSC_IS_SUPPORTED 1
+#else
+#define TSC_IS_SUPPORTED 0
+#endif
+
 #define CHECK__(x) {				\
 	while ((x) < 0) {			\
 		pr_debug(#x " failed!\n");	\
@@ -45,7 +55,7 @@
  * %0 is returned, otherwise %-1 is returned.  If TSC conversion is not
  * supported then then the test passes but " (not supported)" is printed.
  */
-int test__perf_time_to_tsc(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__perf_time_to_tsc(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct record_opts opts = {
 		.mmap_pages	     = UINT_MAX,
@@ -69,6 +79,11 @@ int test__perf_time_to_tsc(struct test *test __maybe_unused, int subtest __maybe
 	u64 test_time, comm1_time = 0, comm2_time = 0;
 	struct mmap *md;
 
+	if (!TSC_IS_SUPPORTED) {
+		pr_debug("Test not supported on this architecture");
+		return TEST_SKIP;
+	}
+
 	threads = thread_map__new(-1, getpid(), UINT_MAX);
 	CHECK_NOT_NULL__(threads);
 
@@ -185,15 +200,4 @@ int test__perf_time_to_tsc(struct test *test __maybe_unused, int subtest __maybe
 	return err;
 }
 
-bool test__tsc_is_supported(void)
-{
-	/*
-	 * Except x86_64/i386 and Arm64, other archs don't support TSC in perf.
-	 * Just enable the test for x86_64/i386 and Arm64 archs.
-	 */
-#if defined(__x86_64__) || defined(__i386__) || defined(__aarch64__)
-	return true;
-#else
-	return false;
-#endif
-}
+DEFINE_SUITE("Convert perf time to TSC", perf_time_to_tsc);
diff --git a/tools/perf/tests/pfm.c b/tools/perf/tests/pfm.c
index e8fd0da..71b76de 100644
--- a/tools/perf/tests/pfm.c
+++ b/tools/perf/tests/pfm.c
@@ -12,27 +12,6 @@
 #include <linux/kernel.h>
 
 #ifdef HAVE_LIBPFM
-static int test__pfm_events(void);
-static int test__pfm_group(void);
-#endif
-
-static const struct {
-	int (*func)(void);
-	const char *desc;
-} pfm_testcase_table[] = {
-#ifdef HAVE_LIBPFM
-	{
-		.func = test__pfm_events,
-		.desc = "test of individual --pfm-events",
-	},
-	{
-		.func = test__pfm_group,
-		.desc = "test groups of --pfm-events",
-	},
-#endif
-};
-
-#ifdef HAVE_LIBPFM
 static int count_pfm_events(struct perf_evlist *evlist)
 {
 	struct perf_evsel *evsel;
@@ -44,7 +23,8 @@ static int count_pfm_events(struct perf_evlist *evlist)
 	return count;
 }
 
-static int test__pfm_events(void)
+static int test__pfm_events(struct test_suite *test __maybe_unused,
+			    int subtest __maybe_unused)
 {
 	struct evlist *evlist;
 	struct option opt;
@@ -104,7 +84,8 @@ static int test__pfm_events(void)
 	return 0;
 }
 
-static int test__pfm_group(void)
+static int test__pfm_group(struct test_suite *test __maybe_unused,
+			   int subtest __maybe_unused)
 {
 	struct evlist *evlist;
 	struct option opt;
@@ -187,27 +168,27 @@ static int test__pfm_group(void)
 	}
 	return 0;
 }
-#endif
-
-const char *test__pfm_subtest_get_desc(int i)
-{
-	if (i < 0 || i >= (int)ARRAY_SIZE(pfm_testcase_table))
-		return NULL;
-	return pfm_testcase_table[i].desc;
-}
-
-int test__pfm_subtest_get_nr(void)
-{
-	return (int)ARRAY_SIZE(pfm_testcase_table);
-}
-
-int test__pfm(struct test *test __maybe_unused, int i __maybe_unused)
-{
-#ifdef HAVE_LIBPFM
-	if (i < 0 || i >= (int)ARRAY_SIZE(pfm_testcase_table))
-		return TEST_FAIL;
-	return pfm_testcase_table[i].func();
 #else
+static int test__pfm_events(struct test_suite *test __maybe_unused,
+			    int subtest __maybe_unused)
+{
 	return TEST_SKIP;
-#endif
 }
+
+static int test__pfm_group(struct test_suite *test __maybe_unused,
+			   int subtest __maybe_unused)
+{
+	return TEST_SKIP;
+}
+#endif
+
+static struct test_case pfm_tests[] = {
+	TEST_CASE_REASON("test of individual --pfm-events", pfm_events, "not compiled in"),
+	TEST_CASE_REASON("test groups of --pfm-events", pfm_group, "not compiled in"),
+	{ .name = NULL, }
+};
+
+struct test_suite suite__pfm = {
+	.desc = "Test libpfm4 support",
+	.test_cases = pfm_tests,
+};
diff --git a/tools/perf/tests/pmu-events.c b/tools/perf/tests/pmu-events.c
index 9ae894c..df1c9a3 100644
--- a/tools/perf/tests/pmu-events.c
+++ b/tools/perf/tests/pmu-events.c
@@ -418,7 +418,8 @@ static int compare_alias_to_test_event(struct perf_pmu_alias *alias,
 }
 
 /* Verify generated events from pmu-events.c are as expected */
-static int test_pmu_event_table(void)
+static int test__pmu_event_table(struct test_suite *test __maybe_unused,
+				 int subtest __maybe_unused)
 {
 	const struct pmu_event *sys_event_tables = __test_pmu_get_sys_events_table();
 	const struct pmu_events_map *map = __test_pmu_get_events_map();
@@ -705,7 +706,8 @@ static struct perf_pmu_test_pmu test_pmus[] = {
 };
 
 /* Test that aliases generated are as expected */
-static int test_aliases(void)
+static int test__aliases(struct test_suite *test __maybe_unused,
+			int subtest __maybe_unused)
 {
 	struct perf_pmu *pmu = NULL;
 	unsigned long i;
@@ -892,7 +894,8 @@ static int resolve_metric_simple(struct expr_parse_ctx *pctx,
 
 }
 
-static int test_parsing(void)
+static int test__parsing(struct test_suite *test __maybe_unused,
+			 int subtest __maybe_unused)
 {
 	const struct pmu_events_map *cpus_map = pmu_events_map__find();
 	const struct pmu_events_map *map;
@@ -1034,7 +1037,8 @@ static int metric_parse_fake(const char *str)
  * or all defined cpus via the 'fake_pmu'
  * in parse_events.
  */
-static int test_parsing_fake(void)
+static int test__parsing_fake(struct test_suite *test __maybe_unused,
+			      int subtest __maybe_unused)
 {
 	const struct pmu_events_map *map;
 	const struct pmu_event *pe;
@@ -1068,55 +1072,16 @@ static int test_parsing_fake(void)
 	return 0;
 }
 
-static const struct {
-	int (*func)(void);
-	const char *desc;
-} pmu_events_testcase_table[] = {
-	{
-		.func = test_pmu_event_table,
-		.desc = "PMU event table sanity",
-	},
-	{
-		.func = test_aliases,
-		.desc = "PMU event map aliases",
-	},
-	{
-		.func = test_parsing,
-		.desc = "Parsing of PMU event table metrics",
-	},
-	{
-		.func = test_parsing_fake,
-		.desc = "Parsing of PMU event table metrics with fake PMUs",
-	},
+static struct test_case pmu_events_tests[] = {
+	TEST_CASE("PMU event table sanity", pmu_event_table),
+	TEST_CASE("PMU event map aliases", aliases),
+	TEST_CASE_REASON("Parsing of PMU event table metrics", parsing,
+			 "some metrics failed"),
+	TEST_CASE("Parsing of PMU event table metrics with fake PMUs", parsing_fake),
+	{ .name = NULL, }
 };
 
-const char *test__pmu_events_subtest_get_desc(int subtest)
-{
-	if (subtest < 0 ||
-	    subtest >= (int)ARRAY_SIZE(pmu_events_testcase_table))
-		return NULL;
-	return pmu_events_testcase_table[subtest].desc;
-}
-
-const char *test__pmu_events_subtest_skip_reason(int subtest)
-{
-	if (subtest < 0 ||
-	    subtest >= (int)ARRAY_SIZE(pmu_events_testcase_table))
-		return NULL;
-	if (pmu_events_testcase_table[subtest].func != test_parsing)
-		return NULL;
-	return "some metrics failed";
-}
-
-int test__pmu_events_subtest_get_nr(void)
-{
-	return (int)ARRAY_SIZE(pmu_events_testcase_table);
-}
-
-int test__pmu_events(struct test *test __maybe_unused, int subtest)
-{
-	if (subtest < 0 ||
-	    subtest >= (int)ARRAY_SIZE(pmu_events_testcase_table))
-		return TEST_FAIL;
-	return pmu_events_testcase_table[subtest].func();
-}
+struct test_suite suite__pmu_events = {
+	.desc = "PMU events",
+	.test_cases = pmu_events_tests,
+};
diff --git a/tools/perf/tests/pmu.c b/tools/perf/tests/pmu.c
index 714e683..8507bd6 100644
--- a/tools/perf/tests/pmu.c
+++ b/tools/perf/tests/pmu.c
@@ -137,7 +137,7 @@ static struct list_head *test_terms_list(void)
 	return &terms;
 }
 
-int test__pmu(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__pmu(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	char *format = test_format_dir_get();
 	LIST_HEAD(formats);
@@ -177,3 +177,5 @@ int test__pmu(struct test *test __maybe_unused, int subtest __maybe_unused)
 	test_format_dir_put(format);
 	return ret;
 }
+
+DEFINE_SUITE("Parse perf pmu format", pmu);
diff --git a/tools/perf/tests/python-use.c b/tools/perf/tests/python-use.c
index 98c6d47..6b990ee 100644
--- a/tools/perf/tests/python-use.c
+++ b/tools/perf/tests/python-use.c
@@ -9,7 +9,7 @@
 #include "tests.h"
 #include "util/debug.h"
 
-int test__python_use(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__python_use(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	char *cmd;
 	int ret;
@@ -23,3 +23,5 @@ int test__python_use(struct test *test __maybe_unused, int subtest __maybe_unuse
 	free(cmd);
 	return ret;
 }
+
+DEFINE_SUITE("'import perf' in python", python_use);
diff --git a/tools/perf/tests/sample-parsing.c b/tools/perf/tests/sample-parsing.c
index c83a115..b669d22 100644
--- a/tools/perf/tests/sample-parsing.c
+++ b/tools/perf/tests/sample-parsing.c
@@ -368,7 +368,7 @@ static int do_test(u64 sample_type, u64 sample_regs, u64 read_format)
  * checks sample format bits separately and together.  If the test passes %0 is
  * returned, otherwise %-1 is returned.
  */
-int test__sample_parsing(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__sample_parsing(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	const u64 rf[] = {4, 5, 6, 7, 12, 13, 14, 15};
 	u64 sample_type;
@@ -426,3 +426,5 @@ int test__sample_parsing(struct test *test __maybe_unused, int subtest __maybe_u
 
 	return 0;
 }
+
+DEFINE_SUITE("Sample parsing", sample_parsing);
diff --git a/tools/perf/tests/sdt.c b/tools/perf/tests/sdt.c
index ed76c69..9197128 100644
--- a/tools/perf/tests/sdt.c
+++ b/tools/perf/tests/sdt.c
@@ -76,7 +76,7 @@ static int search_cached_probe(const char *target,
 	return ret;
 }
 
-int test__sdt_event(struct test *test __maybe_unused, int subtests __maybe_unused)
+static int test__sdt_event(struct test_suite *test __maybe_unused, int subtests __maybe_unused)
 {
 	int ret = TEST_FAIL;
 	char __tempdir[] = "./test-buildid-XXXXXX";
@@ -114,9 +114,11 @@ int test__sdt_event(struct test *test __maybe_unused, int subtests __maybe_unuse
 	return ret;
 }
 #else
-int test__sdt_event(struct test *test __maybe_unused, int subtests __maybe_unused)
+static int test__sdt_event(struct test_suite *test __maybe_unused, int subtests __maybe_unused)
 {
 	pr_debug("Skip SDT event test because SDT support is not compiled\n");
 	return TEST_SKIP;
 }
 #endif
+
+DEFINE_SUITE("Probe SDT events", sdt_event);
diff --git a/tools/perf/tests/shell/record+zstd_comp_decomp.sh b/tools/perf/tests/shell/record+zstd_comp_decomp.sh
index 8a168cf..49bd875 100755
--- a/tools/perf/tests/shell/record+zstd_comp_decomp.sh
+++ b/tools/perf/tests/shell/record+zstd_comp_decomp.sh
@@ -12,7 +12,7 @@
 
 collect_z_record() {
 	echo "Collecting compressed record file:"
-	[[ "$(uname -m)" != s390x ]] && gflag='-g'
+	[ "$(uname -m)" != s390x ] && gflag='-g'
 	$perf_tool record -o $trace_file $gflag -z -F 5000 -- \
 		dd count=500 if=/dev/urandom of=/dev/null
 }
diff --git a/tools/perf/tests/shell/stat_all_pmu.sh b/tools/perf/tests/shell/stat_all_pmu.sh
index 2de7fd0..b30dba4 100755
--- a/tools/perf/tests/shell/stat_all_pmu.sh
+++ b/tools/perf/tests/shell/stat_all_pmu.sh
@@ -7,11 +7,11 @@
 for p in $(perf list --raw-dump pmu); do
   echo "Testing $p"
   result=$(perf stat -e "$p" true 2>&1)
-  if [[ ! "$result" =~ "$p" ]] && [[ ! "$result" =~ "<not supported>" ]]; then
+  if ! echo "$result" | grep -q "$p" && ! echo "$result" | grep -q "<not supported>" ; then
     # We failed to see the event and it is supported. Possibly the workload was
     # too small so retry with something longer.
     result=$(perf stat -e "$p" perf bench internals synthesize 2>&1)
-    if [[ ! "$result" =~ "$p" ]]; then
+    if ! echo "$result" | grep -q "$p" ; then
       echo "Event '$p' not printed in:"
       echo "$result"
       exit 1
diff --git a/tools/perf/tests/shell/stat_bpf_counters.sh b/tools/perf/tests/shell/stat_bpf_counters.sh
index 2aed20d..13473ae 100755
--- a/tools/perf/tests/shell/stat_bpf_counters.sh
+++ b/tools/perf/tests/shell/stat_bpf_counters.sh
@@ -23,7 +23,7 @@
 
 # skip if --bpf-counters is not supported
 if ! perf stat --bpf-counters true > /dev/null 2>&1; then
-	if [ "$1" == "-v" ]; then
+	if [ "$1" = "-v" ]; then
 		echo "Skipping: --bpf-counters not supported"
 		perf --no-pager stat --bpf-counters true || true
 	fi
diff --git a/tools/perf/tests/shell/test_arm_spe.sh b/tools/perf/tests/shell/test_arm_spe.sh
new file mode 100755
index 0000000..e59044e
--- /dev/null
+++ b/tools/perf/tests/shell/test_arm_spe.sh
@@ -0,0 +1,89 @@
+#!/bin/sh
+# Check Arm SPE trace data recording and synthesized samples
+
+# Uses the 'perf record' to record trace data of Arm SPE events;
+# then verify if any SPE event samples are generated by SPE with
+# 'perf script' and 'perf report' commands.
+
+# SPDX-License-Identifier: GPL-2.0
+# German Gomez <german.gomez@arm.com>, 2021
+
+skip_if_no_arm_spe_event() {
+	perf list | egrep -q 'arm_spe_[0-9]+//' && return 0
+
+	# arm_spe event doesn't exist
+	return 2
+}
+
+skip_if_no_arm_spe_event || exit 2
+
+perfdata=$(mktemp /tmp/__perf_test.perf.data.XXXXX)
+glb_err=0
+
+cleanup_files()
+{
+	rm -f ${perfdata}
+	exit $glb_err
+}
+
+trap cleanup_files exit term int
+
+arm_spe_report() {
+	if [ $2 != 0 ]; then
+		echo "$1: FAIL"
+		glb_err=$2
+	else
+		echo "$1: PASS"
+	fi
+}
+
+perf_script_samples() {
+	echo "Looking at perf.data file for dumping samples:"
+
+	# from arm-spe.c/arm_spe_synth_events()
+	events="(ld1-miss|ld1-access|llc-miss|lld-access|tlb-miss|tlb-access|branch-miss|remote-access|memory)"
+
+	# Below is an example of the samples dumping:
+	#	dd  3048 [002]          1    l1d-access:      ffffaa64999c __GI___libc_write+0x3c (/lib/aarch64-linux-gnu/libc-2.27.so)
+	#	dd  3048 [002]          1    tlb-access:      ffffaa64999c __GI___libc_write+0x3c (/lib/aarch64-linux-gnu/libc-2.27.so)
+	#	dd  3048 [002]          1        memory:      ffffaa64999c __GI___libc_write+0x3c (/lib/aarch64-linux-gnu/libc-2.27.so)
+	perf script -F,-time -i ${perfdata} 2>&1 | \
+		egrep " +$1 +[0-9]+ .* +${events}:(.*:)? +" > /dev/null 2>&1
+}
+
+perf_report_samples() {
+	echo "Looking at perf.data file for reporting samples:"
+
+	# Below is an example of the samples reporting:
+	#   73.04%    73.04%  dd    libc-2.27.so      [.] _dl_addr
+	#    7.71%     7.71%  dd    libc-2.27.so      [.] getenv
+	#    2.59%     2.59%  dd    ld-2.27.so        [.] strcmp
+	perf report --stdio -i ${perfdata} 2>&1 | \
+		egrep " +[0-9]+\.[0-9]+% +[0-9]+\.[0-9]+% +$1 " > /dev/null 2>&1
+}
+
+arm_spe_snapshot_test() {
+	echo "Recording trace with snapshot mode $perfdata"
+	perf record -o ${perfdata} -e arm_spe// -S \
+		-- dd if=/dev/zero of=/dev/null > /dev/null 2>&1 &
+	PERFPID=$!
+
+	# Wait for perf program
+	sleep 1
+
+	# Send signal to snapshot trace data
+	kill -USR2 $PERFPID
+
+	# Stop perf program
+	kill $PERFPID
+	wait $PERFPID
+
+	perf_script_samples dd &&
+	perf_report_samples dd
+
+	err=$?
+	arm_spe_report "SPE snapshot testing" $err
+}
+
+arm_spe_snapshot_test
+exit $glb_err
diff --git a/tools/perf/tests/stat.c b/tools/perf/tests/stat.c
index c191150..2eb096b 100644
--- a/tools/perf/tests/stat.c
+++ b/tools/perf/tests/stat.c
@@ -47,7 +47,8 @@ static int process_stat_config_event(struct perf_tool *tool __maybe_unused,
 	return 0;
 }
 
-int test__synthesize_stat_config(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__synthesize_stat_config(struct test_suite *test __maybe_unused,
+					int subtest __maybe_unused)
 {
 	struct perf_stat_config stat_config = {
 		.aggr_mode	= AGGR_CORE,
@@ -77,7 +78,7 @@ static int process_stat_event(struct perf_tool *tool __maybe_unused,
 	return 0;
 }
 
-int test__synthesize_stat(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__synthesize_stat(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct perf_counts_values count;
 
@@ -103,7 +104,7 @@ static int process_stat_round_event(struct perf_tool *tool __maybe_unused,
 	return 0;
 }
 
-int test__synthesize_stat_round(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__synthesize_stat_round(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	TEST_ASSERT_VAL("failed to synthesize stat_config",
 		!perf_event__synthesize_stat_round(NULL, 0xdeadbeef, PERF_STAT_ROUND_TYPE__INTERVAL,
@@ -111,3 +112,7 @@ int test__synthesize_stat_round(struct test *test __maybe_unused, int subtest __
 
 	return 0;
 }
+
+DEFINE_SUITE("Synthesize stat config", synthesize_stat_config);
+DEFINE_SUITE("Synthesize stat", synthesize_stat);
+DEFINE_SUITE("Synthesize stat round", synthesize_stat_round);
diff --git a/tools/perf/tests/sw-clock.c b/tools/perf/tests/sw-clock.c
index 7498884..9cd6fec 100644
--- a/tools/perf/tests/sw-clock.c
+++ b/tools/perf/tests/sw-clock.c
@@ -133,7 +133,7 @@ static int __test__sw_clock_freq(enum perf_sw_ids clock_id)
 	return err;
 }
 
-int test__sw_clock_freq(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__sw_clock_freq(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int ret;
 
@@ -143,3 +143,5 @@ int test__sw_clock_freq(struct test *test __maybe_unused, int subtest __maybe_un
 
 	return ret;
 }
+
+DEFINE_SUITE("Software clock events period values", sw_clock_freq);
diff --git a/tools/perf/tests/switch-tracking.c b/tools/perf/tests/switch-tracking.c
index 62c0ec2..0c0c232 100644
--- a/tools/perf/tests/switch-tracking.c
+++ b/tools/perf/tests/switch-tracking.c
@@ -321,7 +321,7 @@ static int process_events(struct evlist *evlist,
  * evsel->core.system_wide and evsel->tracking flags (respectively) with other events
  * sometimes enabled or disabled.
  */
-int test__switch_tracking(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__switch_tracking(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	const char *sched_switch = "sched:sched_switch";
 	struct switch_tracking switch_tracking = { .tids = NULL, };
@@ -588,3 +588,5 @@ int test__switch_tracking(struct test *test __maybe_unused, int subtest __maybe_
 	err = -1;
 	goto out;
 }
+
+DEFINE_SUITE("Track with sched_switch", switch_tracking);
diff --git a/tools/perf/tests/task-exit.c b/tools/perf/tests/task-exit.c
index 4c2969d..25f075f 100644
--- a/tools/perf/tests/task-exit.c
+++ b/tools/perf/tests/task-exit.c
@@ -39,7 +39,7 @@ static void workload_exec_failed_signal(int signo __maybe_unused,
  * if the number of exit event reported by the kernel is 1 or not
  * in order to check the kernel returns correct number of event.
  */
-int test__task_exit(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__task_exit(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	int err = -1;
 	union perf_event *event;
@@ -151,3 +151,5 @@ int test__task_exit(struct test *test __maybe_unused, int subtest __maybe_unused
 	evlist__delete(evlist);
 	return err;
 }
+
+DEFINE_SUITE("Number of exit events of a simple workload", task_exit);
diff --git a/tools/perf/tests/tests.h b/tools/perf/tests/tests.h
index fe1306f..8f65098 100644
--- a/tools/perf/tests/tests.h
+++ b/tools/perf/tests/tests.h
@@ -27,112 +27,146 @@ enum {
 	TEST_SKIP = -2,
 };
 
-struct test {
+struct test_suite;
+
+typedef int (*test_fnptr)(struct test_suite *, int);
+
+struct test_case {
+	const char *name;
 	const char *desc;
-	int (*func)(struct test *test, int subtest);
-	struct {
-		bool skip_if_fail;
-		int (*get_nr)(void);
-		const char *(*get_desc)(int subtest);
-		const char *(*skip_reason)(int subtest);
-	} subtest;
-	bool (*is_supported)(void);
+	const char *skip_reason;
+	test_fnptr run_case;
+};
+
+struct test_suite {
+	const char *desc;
+	struct test_case *test_cases;
 	void *priv;
 };
 
-/* Tests */
-int test__vmlinux_matches_kallsyms(struct test *test, int subtest);
-int test__openat_syscall_event(struct test *test, int subtest);
-int test__openat_syscall_event_on_all_cpus(struct test *test, int subtest);
-int test__basic_mmap(struct test *test, int subtest);
-int test__PERF_RECORD(struct test *test, int subtest);
-int test__perf_evsel__roundtrip_name_test(struct test *test, int subtest);
-int test__perf_evsel__tp_sched_test(struct test *test, int subtest);
-int test__syscall_openat_tp_fields(struct test *test, int subtest);
-int test__pmu(struct test *test, int subtest);
-int test__pmu_events(struct test *test, int subtest);
-const char *test__pmu_events_subtest_get_desc(int subtest);
-const char *test__pmu_events_subtest_skip_reason(int subtest);
-int test__pmu_events_subtest_get_nr(void);
-int test__attr(struct test *test, int subtest);
-int test__dso_data(struct test *test, int subtest);
-int test__dso_data_cache(struct test *test, int subtest);
-int test__dso_data_reopen(struct test *test, int subtest);
-int test__parse_events(struct test *test, int subtest);
-int test__hists_link(struct test *test, int subtest);
-int test__python_use(struct test *test, int subtest);
-int test__bp_signal(struct test *test, int subtest);
-int test__bp_signal_overflow(struct test *test, int subtest);
-int test__bp_accounting(struct test *test, int subtest);
-int test__wp(struct test *test, int subtest);
-const char *test__wp_subtest_get_desc(int subtest);
-const char *test__wp_subtest_skip_reason(int subtest);
-int test__wp_subtest_get_nr(void);
-int test__task_exit(struct test *test, int subtest);
-int test__mem(struct test *test, int subtest);
-int test__sw_clock_freq(struct test *test, int subtest);
-int test__code_reading(struct test *test, int subtest);
-int test__sample_parsing(struct test *test, int subtest);
-int test__keep_tracking(struct test *test, int subtest);
-int test__parse_no_sample_id_all(struct test *test, int subtest);
-int test__dwarf_unwind(struct test *test, int subtest);
-int test__expr(struct test *test, int subtest);
-int test__hists_filter(struct test *test, int subtest);
-int test__mmap_thread_lookup(struct test *test, int subtest);
-int test__thread_maps_share(struct test *test, int subtest);
-int test__hists_output(struct test *test, int subtest);
-int test__hists_cumulate(struct test *test, int subtest);
-int test__switch_tracking(struct test *test, int subtest);
-int test__fdarray__filter(struct test *test, int subtest);
-int test__fdarray__add(struct test *test, int subtest);
-int test__kmod_path__parse(struct test *test, int subtest);
-int test__thread_map(struct test *test, int subtest);
-int test__llvm(struct test *test, int subtest);
-const char *test__llvm_subtest_get_desc(int subtest);
-int test__llvm_subtest_get_nr(void);
-int test__bpf(struct test *test, int subtest);
-const char *test__bpf_subtest_get_desc(int subtest);
-int test__bpf_subtest_get_nr(void);
-int test__session_topology(struct test *test, int subtest);
-int test__thread_map_synthesize(struct test *test, int subtest);
-int test__thread_map_remove(struct test *test, int subtest);
-int test__cpu_map_synthesize(struct test *test, int subtest);
-int test__synthesize_stat_config(struct test *test, int subtest);
-int test__synthesize_stat(struct test *test, int subtest);
-int test__synthesize_stat_round(struct test *test, int subtest);
-int test__event_update(struct test *test, int subtest);
-int test__event_times(struct test *test, int subtest);
-int test__backward_ring_buffer(struct test *test, int subtest);
-int test__cpu_map_print(struct test *test, int subtest);
-int test__cpu_map_merge(struct test *test, int subtest);
-int test__sdt_event(struct test *test, int subtest);
-int test__is_printable_array(struct test *test, int subtest);
-int test__bitmap_print(struct test *test, int subtest);
-int test__perf_hooks(struct test *test, int subtest);
-int test__clang(struct test *test, int subtest);
-const char *test__clang_subtest_get_desc(int subtest);
-int test__clang_subtest_get_nr(void);
-int test__unit_number__scnprint(struct test *test, int subtest);
-int test__mem2node(struct test *t, int subtest);
-int test__maps__merge_in(struct test *t, int subtest);
-int test__time_utils(struct test *t, int subtest);
-int test__jit_write_elf(struct test *test, int subtest);
-int test__api_io(struct test *test, int subtest);
-int test__demangle_java(struct test *test, int subtest);
-int test__demangle_ocaml(struct test *test, int subtest);
-int test__pfm(struct test *test, int subtest);
-const char *test__pfm_subtest_get_desc(int subtest);
-int test__pfm_subtest_get_nr(void);
-int test__parse_metric(struct test *test, int subtest);
-int test__pe_file_parsing(struct test *test, int subtest);
-int test__expand_cgroup_events(struct test *test, int subtest);
-int test__perf_time_to_tsc(struct test *test, int subtest);
-int test__dlfilter(struct test *test, int subtest);
+#define DECLARE_SUITE(name) \
+	extern struct test_suite suite__##name;
 
-bool test__bp_signal_is_supported(void);
-bool test__bp_account_is_supported(void);
-bool test__wp_is_supported(void);
-bool test__tsc_is_supported(void);
+#define TEST_CASE(description, _name)			\
+	{						\
+		.name = #_name,				\
+		.desc = description,			\
+		.run_case = test__##_name,		\
+	}
+
+#define TEST_CASE_REASON(description, _name, _reason)	\
+	{						\
+		.name = #_name,				\
+		.desc = description,			\
+		.run_case = test__##_name,		\
+		.skip_reason = _reason,			\
+	}
+
+#define DEFINE_SUITE(description, _name)		\
+	struct test_case tests__##_name[] = {           \
+		TEST_CASE(description, _name),		\
+		{	.name = NULL, }			\
+	};						\
+	struct test_suite suite__##_name = {		\
+		.desc = description,			\
+		.test_cases = tests__##_name,		\
+	}
+
+/* Tests */
+DECLARE_SUITE(vmlinux_matches_kallsyms);
+DECLARE_SUITE(openat_syscall_event);
+DECLARE_SUITE(openat_syscall_event_on_all_cpus);
+DECLARE_SUITE(basic_mmap);
+DECLARE_SUITE(PERF_RECORD);
+DECLARE_SUITE(perf_evsel__roundtrip_name_test);
+DECLARE_SUITE(perf_evsel__tp_sched_test);
+DECLARE_SUITE(syscall_openat_tp_fields);
+DECLARE_SUITE(pmu);
+DECLARE_SUITE(pmu_events);
+DECLARE_SUITE(attr);
+DECLARE_SUITE(dso_data);
+DECLARE_SUITE(dso_data_cache);
+DECLARE_SUITE(dso_data_reopen);
+DECLARE_SUITE(parse_events);
+DECLARE_SUITE(hists_link);
+DECLARE_SUITE(python_use);
+DECLARE_SUITE(bp_signal);
+DECLARE_SUITE(bp_signal_overflow);
+DECLARE_SUITE(bp_accounting);
+DECLARE_SUITE(wp);
+DECLARE_SUITE(task_exit);
+DECLARE_SUITE(mem);
+DECLARE_SUITE(sw_clock_freq);
+DECLARE_SUITE(code_reading);
+DECLARE_SUITE(sample_parsing);
+DECLARE_SUITE(keep_tracking);
+DECLARE_SUITE(parse_no_sample_id_all);
+DECLARE_SUITE(dwarf_unwind);
+DECLARE_SUITE(expr);
+DECLARE_SUITE(hists_filter);
+DECLARE_SUITE(mmap_thread_lookup);
+DECLARE_SUITE(thread_maps_share);
+DECLARE_SUITE(hists_output);
+DECLARE_SUITE(hists_cumulate);
+DECLARE_SUITE(switch_tracking);
+DECLARE_SUITE(fdarray__filter);
+DECLARE_SUITE(fdarray__add);
+DECLARE_SUITE(kmod_path__parse);
+DECLARE_SUITE(thread_map);
+DECLARE_SUITE(llvm);
+DECLARE_SUITE(bpf);
+DECLARE_SUITE(session_topology);
+DECLARE_SUITE(thread_map_synthesize);
+DECLARE_SUITE(thread_map_remove);
+DECLARE_SUITE(cpu_map_synthesize);
+DECLARE_SUITE(synthesize_stat_config);
+DECLARE_SUITE(synthesize_stat);
+DECLARE_SUITE(synthesize_stat_round);
+DECLARE_SUITE(event_update);
+DECLARE_SUITE(event_times);
+DECLARE_SUITE(backward_ring_buffer);
+DECLARE_SUITE(cpu_map_print);
+DECLARE_SUITE(cpu_map_merge);
+DECLARE_SUITE(sdt_event);
+DECLARE_SUITE(is_printable_array);
+DECLARE_SUITE(bitmap_print);
+DECLARE_SUITE(perf_hooks);
+DECLARE_SUITE(clang);
+DECLARE_SUITE(unit_number__scnprint);
+DECLARE_SUITE(mem2node);
+DECLARE_SUITE(maps__merge_in);
+DECLARE_SUITE(time_utils);
+DECLARE_SUITE(jit_write_elf);
+DECLARE_SUITE(api_io);
+DECLARE_SUITE(demangle_java);
+DECLARE_SUITE(demangle_ocaml);
+DECLARE_SUITE(pfm);
+DECLARE_SUITE(parse_metric);
+DECLARE_SUITE(pe_file_parsing);
+DECLARE_SUITE(expand_cgroup_events);
+DECLARE_SUITE(perf_time_to_tsc);
+DECLARE_SUITE(dlfilter);
+
+/*
+ * PowerPC and S390 do not support creation of instruction breakpoints using the
+ * perf_event interface.
+ *
+ * ARM requires explicit rounding down of the instruction pointer in Thumb mode,
+ * and then requires the single-step to be handled explicitly in the overflow
+ * handler to avoid stepping into the SIGIO handler and getting stuck on the
+ * breakpointed instruction.
+ *
+ * Since arm64 has the same issue with arm for the single-step handling, this
+ * case also gets stuck on the breakpointed instruction.
+ *
+ * Just disable the test for these architectures until these issues are
+ * resolved.
+ */
+#if defined(__powerpc__) || defined(__s390x__) || defined(__arm__) || defined(__aarch64__)
+#define BP_SIGNAL_IS_SUPPORTED 0
+#else
+#define BP_SIGNAL_IS_SUPPORTED 1
+#endif
 
 #ifdef HAVE_DWARF_UNWIND_SUPPORT
 struct thread;
@@ -142,7 +176,7 @@ int test__arch_unwind_sample(struct perf_sample *sample,
 #endif
 
 #if defined(__arm__)
-int test__vectors_page(struct test *test, int subtest);
+DECLARE_SUITE(vectors_page);
 #endif
 
 #endif /* TESTS_H */
diff --git a/tools/perf/tests/thread-map.c b/tools/perf/tests/thread-map.c
index d1e208b..e413c13 100644
--- a/tools/perf/tests/thread-map.c
+++ b/tools/perf/tests/thread-map.c
@@ -19,7 +19,7 @@ struct machine;
 #define NAME	(const char *) "perf"
 #define NAMEUL	(unsigned long) NAME
 
-int test__thread_map(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__thread_map(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct perf_thread_map *map;
 
@@ -86,7 +86,7 @@ static int process_event(struct perf_tool *tool __maybe_unused,
 	return 0;
 }
 
-int test__thread_map_synthesize(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__thread_map_synthesize(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct perf_thread_map *threads;
 
@@ -106,7 +106,7 @@ int test__thread_map_synthesize(struct test *test __maybe_unused, int subtest __
 	return 0;
 }
 
-int test__thread_map_remove(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__thread_map_remove(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct perf_thread_map *threads;
 	char *str;
@@ -145,3 +145,7 @@ int test__thread_map_remove(struct test *test __maybe_unused, int subtest __mayb
 	perf_thread_map__put(threads);
 	return 0;
 }
+
+DEFINE_SUITE("Thread map", thread_map);
+DEFINE_SUITE("Synthesize thread map", thread_map_synthesize);
+DEFINE_SUITE("Remove thread map", thread_map_remove);
diff --git a/tools/perf/tests/thread-maps-share.c b/tools/perf/tests/thread-maps-share.c
index 9371484..84edd82 100644
--- a/tools/perf/tests/thread-maps-share.c
+++ b/tools/perf/tests/thread-maps-share.c
@@ -4,7 +4,7 @@
 #include "thread.h"
 #include "debug.h"
 
-int test__thread_maps_share(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__thread_maps_share(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct machines machines;
 	struct machine *machine;
@@ -96,3 +96,5 @@ int test__thread_maps_share(struct test *test __maybe_unused, int subtest __mayb
 	machines__exit(&machines);
 	return 0;
 }
+
+DEFINE_SUITE("Share thread maps", thread_maps_share);
diff --git a/tools/perf/tests/time-utils-test.c b/tools/perf/tests/time-utils-test.c
index fe57ca3..38df103 100644
--- a/tools/perf/tests/time-utils-test.c
+++ b/tools/perf/tests/time-utils-test.c
@@ -131,7 +131,7 @@ static bool test__perf_time__parse_for_ranges(struct test_data *d)
 	return pass;
 }
 
-int test__time_utils(struct test *t __maybe_unused, int subtest __maybe_unused)
+static int test__time_utils(struct test_suite *t __maybe_unused, int subtest __maybe_unused)
 {
 	bool pass = true;
 
@@ -249,3 +249,5 @@ int test__time_utils(struct test *t __maybe_unused, int subtest __maybe_unused)
 
 	return pass ? 0 : TEST_FAIL;
 }
+
+DEFINE_SUITE("time utils", time_utils);
diff --git a/tools/perf/tests/topology.c b/tools/perf/tests/topology.c
index 4574c46..8699861 100644
--- a/tools/perf/tests/topology.c
+++ b/tools/perf/tests/topology.c
@@ -175,7 +175,7 @@ static int check_cpu_topology(char *path, struct perf_cpu_map *map)
 	return 0;
 }
 
-int test__session_topology(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__session_topology(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
 {
 	char path[PATH_MAX];
 	struct perf_cpu_map *map;
@@ -201,3 +201,5 @@ int test__session_topology(struct test *test __maybe_unused, int subtest __maybe
 	unlink(path);
 	return ret;
 }
+
+DEFINE_SUITE("Session topology", session_topology);
diff --git a/tools/perf/tests/unit_number__scnprintf.c b/tools/perf/tests/unit_number__scnprintf.c
index 3721757..88bcada1 100644
--- a/tools/perf/tests/unit_number__scnprintf.c
+++ b/tools/perf/tests/unit_number__scnprintf.c
@@ -7,7 +7,7 @@
 #include "units.h"
 #include "debug.h"
 
-int test__unit_number__scnprint(struct test *t __maybe_unused, int subtest __maybe_unused)
+static int test__unit_number__scnprint(struct test_suite *t __maybe_unused, int subtest __maybe_unused)
 {
 	struct {
 		u64		 n;
@@ -38,3 +38,5 @@ int test__unit_number__scnprint(struct test *t __maybe_unused, int subtest __may
 
 	return TEST_OK;
 }
+
+DEFINE_SUITE("unit_number__scnprintf", unit_number__scnprint);
diff --git a/tools/perf/tests/vmlinux-kallsyms.c b/tools/perf/tests/vmlinux-kallsyms.c
index 4f884aa..e80df13 100644
--- a/tools/perf/tests/vmlinux-kallsyms.c
+++ b/tools/perf/tests/vmlinux-kallsyms.c
@@ -111,7 +111,8 @@ static bool is_ignored_symbol(const char *name, char type)
 	return false;
 }
 
-int test__vmlinux_matches_kallsyms(struct test *test __maybe_unused, int subtest __maybe_unused)
+static int test__vmlinux_matches_kallsyms(struct test_suite *test __maybe_unused,
+					int subtest __maybe_unused)
 {
 	int err = -1;
 	struct rb_node *nd;
@@ -352,3 +353,5 @@ int test__vmlinux_matches_kallsyms(struct test *test __maybe_unused, int subtest
 	machine__exit(&vmlinux);
 	return err;
 }
+
+DEFINE_SUITE("vmlinux symtab matches kallsyms", vmlinux_matches_kallsyms);
diff --git a/tools/perf/tests/wp.c b/tools/perf/tests/wp.c
index 9387fa7..820d942 100644
--- a/tools/perf/tests/wp.c
+++ b/tools/perf/tests/wp.c
@@ -48,6 +48,7 @@ static void get__perf_event_attr(struct perf_event_attr *attr, int wp_type,
 	attr->exclude_hv     = 1;
 }
 
+#ifndef __s390x__
 static int __event(int wp_type, void *wp_addr, unsigned long wp_len)
 {
 	int fd;
@@ -61,9 +62,14 @@ static int __event(int wp_type, void *wp_addr, unsigned long wp_len)
 
 	return fd;
 }
+#endif
 
-static int wp_ro_test(void)
+static int test__wp_ro(struct test_suite *test __maybe_unused,
+		       int subtest __maybe_unused)
 {
+#if defined(__s390x__) || defined(__x86_64__) || defined(__i386__)
+	return TEST_SKIP;
+#else
 	int fd;
 	unsigned long tmp, tmp1 = rand();
 
@@ -79,10 +85,15 @@ static int wp_ro_test(void)
 
 	close(fd);
 	return 0;
+#endif
 }
 
-static int wp_wo_test(void)
+static int test__wp_wo(struct test_suite *test __maybe_unused,
+		       int subtest __maybe_unused)
 {
+#if defined(__s390x__)
+	return TEST_SKIP;
+#else
 	int fd;
 	unsigned long tmp, tmp1 = rand();
 
@@ -98,10 +109,15 @@ static int wp_wo_test(void)
 
 	close(fd);
 	return 0;
+#endif
 }
 
-static int wp_rw_test(void)
+static int test__wp_rw(struct test_suite *test __maybe_unused,
+		       int subtest __maybe_unused)
 {
+#if defined(__s390x__)
+	return TEST_SKIP;
+#else
 	int fd;
 	unsigned long tmp, tmp1 = rand();
 
@@ -118,10 +134,15 @@ static int wp_rw_test(void)
 
 	close(fd);
 	return 0;
+#endif
 }
 
-static int wp_modify_test(void)
+static int test__wp_modify(struct test_suite *test __maybe_unused,
+			   int subtest __maybe_unused)
 {
+#if defined(__s390x__)
+	return TEST_SKIP;
+#else
 	int fd, ret;
 	unsigned long tmp = rand();
 	struct perf_event_attr new_attr;
@@ -163,93 +184,18 @@ static int wp_modify_test(void)
 
 	close(fd);
 	return 0;
-}
-
-static bool wp_ro_supported(void)
-{
-#if defined (__x86_64__) || defined (__i386__)
-	return false;
-#else
-	return true;
 #endif
 }
 
-static const char *wp_ro_skip_msg(void)
-{
-#if defined (__x86_64__) || defined (__i386__)
-	return "missing hardware support";
-#else
-	return NULL;
-#endif
-}
-
-static struct {
-	const char *desc;
-	int (*target_func)(void);
-	bool (*is_supported)(void);
-	const char *(*skip_msg)(void);
-} wp_testcase_table[] = {
-	{
-		.desc = "Read Only Watchpoint",
-		.target_func = &wp_ro_test,
-		.is_supported = &wp_ro_supported,
-		.skip_msg = &wp_ro_skip_msg,
-	},
-	{
-		.desc = "Write Only Watchpoint",
-		.target_func = &wp_wo_test,
-	},
-	{
-		.desc = "Read / Write Watchpoint",
-		.target_func = &wp_rw_test,
-	},
-	{
-		.desc = "Modify Watchpoint",
-		.target_func = &wp_modify_test,
-	},
+static struct test_case wp_tests[] = {
+	TEST_CASE_REASON("Read Only Watchpoint", wp_ro, "missing hardware support"),
+	TEST_CASE_REASON("Write Only Watchpoint", wp_wo, "missing hardware support"),
+	TEST_CASE_REASON("Read / Write Watchpoint", wp_rw, "missing hardware support"),
+	TEST_CASE_REASON("Modify Watchpoint", wp_modify, "missing hardware support"),
+	{ .name = NULL, }
 };
 
-int test__wp_subtest_get_nr(void)
-{
-	return (int)ARRAY_SIZE(wp_testcase_table);
-}
-
-const char *test__wp_subtest_get_desc(int i)
-{
-	if (i < 0 || i >= (int)ARRAY_SIZE(wp_testcase_table))
-		return NULL;
-	return wp_testcase_table[i].desc;
-}
-
-const char *test__wp_subtest_skip_reason(int i)
-{
-	if (i < 0 || i >= (int)ARRAY_SIZE(wp_testcase_table))
-		return NULL;
-	if (!wp_testcase_table[i].skip_msg)
-		return NULL;
-	return wp_testcase_table[i].skip_msg();
-}
-
-int test__wp(struct test *test __maybe_unused, int i)
-{
-	if (i < 0 || i >= (int)ARRAY_SIZE(wp_testcase_table))
-		return TEST_FAIL;
-
-	if (wp_testcase_table[i].is_supported &&
-	    !wp_testcase_table[i].is_supported())
-		return TEST_SKIP;
-
-	return !wp_testcase_table[i].target_func() ? TEST_OK : TEST_FAIL;
-}
-
-/* The s390 so far does not have support for
- * instruction breakpoint using the perf_event_open() system call.
- */
-bool test__wp_is_supported(void)
-{
-#if defined(__s390x__)
-	return false;
-#else
-	return true;
-#endif
-}
+struct test_suite suite__wp = {
+	.desc = "Watchpoint",
+	.test_cases = wp_tests,
+};
diff --git a/tools/perf/trace/beauty/beauty.h b/tools/perf/trace/beauty/beauty.h
index d6dfe68..f527a46 100644
--- a/tools/perf/trace/beauty/beauty.h
+++ b/tools/perf/trace/beauty/beauty.h
@@ -62,6 +62,8 @@ size_t pid__scnprintf_fd(struct trace *trace, pid_t pid, int fd, char *bf, size_
 
 extern struct strarray strarray__socket_families;
 
+extern struct strarray strarray__socket_level;
+
 /**
  * augmented_arg: extra payload for syscall pointer arguments
  
@@ -230,6 +232,9 @@ size_t syscall_arg__scnprintf_sockaddr(char *bf, size_t size, struct syscall_arg
 size_t syscall_arg__scnprintf_socket_protocol(char *bf, size_t size, struct syscall_arg *arg);
 #define SCA_SK_PROTO syscall_arg__scnprintf_socket_protocol
 
+size_t syscall_arg__scnprintf_socket_level(char *bf, size_t size, struct syscall_arg *arg);
+#define SCA_SK_LEVEL syscall_arg__scnprintf_socket_level
+
 size_t syscall_arg__scnprintf_statx_flags(char *bf, size_t size, struct syscall_arg *arg);
 #define SCA_STATX_FLAGS syscall_arg__scnprintf_statx_flags
 
diff --git a/tools/perf/trace/beauty/sockaddr.c b/tools/perf/trace/beauty/sockaddr.c
index cd11063..2e0e867 100644
--- a/tools/perf/trace/beauty/sockaddr.c
+++ b/tools/perf/trace/beauty/sockaddr.c
@@ -7,7 +7,7 @@
 #include <sys/un.h>
 #include <arpa/inet.h>
 
-#include "trace/beauty/generated/socket_arrays.c"
+#include "trace/beauty/generated/sockaddr.c"
 DEFINE_STRARRAY(socket_families, "PF_");
 
 static size_t af_inet__scnprintf(struct sockaddr *sa, char *bf, size_t size)
diff --git a/tools/perf/trace/beauty/sockaddr.sh b/tools/perf/trace/beauty/sockaddr.sh
new file mode 100755
index 0000000..3820e5c
--- /dev/null
+++ b/tools/perf/trace/beauty/sockaddr.sh
@@ -0,0 +1,24 @@
+#!/bin/sh
+# SPDX-License-Identifier: LGPL-2.1
+
+# This one uses a copy from the kernel sources headers that is in a
+# place used just for these tools/perf/beauty/ usage, we shouldn't not
+# put it in tools/include/linux otherwise they would be used in the
+# normal compiler building process and would drag needless stuff from the
+# kernel.
+
+# When what these scripts need is already in tools/include/ then use it,
+# otherwise grab and check the copy from the kernel sources just for these
+# string table building scripts.
+
+[ $# -eq 1 ] && header_dir=$1 || header_dir=tools/perf/trace/beauty/include/linux/
+
+printf "static const char *socket_families[] = {\n"
+# #define AF_LOCAL	1	/* POSIX name for AF_UNIX	*/
+regex='^#define[[:space:]]+AF_(\w+)[[:space:]]+([[:digit:]]+).*'
+
+egrep $regex ${header_dir}/socket.h | \
+	sed -r "s/$regex/\2 \1/g"	| \
+	xargs printf "\t[%s] = \"%s\",\n" | \
+	egrep -v "\"(UNIX|MAX)\""
+printf "};\n"
diff --git a/tools/perf/trace/beauty/socket.c b/tools/perf/trace/beauty/socket.c
index f23a3dda..b0870c7 100644
--- a/tools/perf/trace/beauty/socket.c
+++ b/tools/perf/trace/beauty/socket.c
@@ -9,9 +9,10 @@
 #include <sys/types.h>
 #include <sys/socket.h>
 
+#include "trace/beauty/generated/socket.c"
+
 static size_t socket__scnprintf_ipproto(int protocol, char *bf, size_t size, bool show_prefix)
 {
-#include "trace/beauty/generated/socket_ipproto_array.c"
 	static DEFINE_STRARRAY(socket_ipproto, "IPPROTO_");
 
 	return strarray__scnprintf(&strarray__socket_ipproto, bf, size, "%d", show_prefix, protocol);
@@ -26,3 +27,21 @@ size_t syscall_arg__scnprintf_socket_protocol(char *bf, size_t size, struct sysc
 
 	return syscall_arg__scnprintf_int(bf, size, arg);
 }
+
+static size_t socket__scnprintf_level(int level, char *bf, size_t size, bool show_prefix)
+{
+#if defined(__alpha__) || defined(__hppa__) || defined(__mips__) || defined(__sparc__)
+	const int sol_socket = 0xffff;
+#else
+	const int sol_socket = 1;
+#endif
+	if (level == sol_socket)
+		return scnprintf(bf, size, "%sSOCKET", show_prefix ? "SOL_" : "");
+
+	return strarray__scnprintf(&strarray__socket_level, bf, size, "%d", show_prefix, level);
+}
+
+size_t syscall_arg__scnprintf_socket_level(char *bf, size_t size, struct syscall_arg *arg)
+{
+	return socket__scnprintf_level(arg->val, bf, size, arg->show_string_prefix);
+}
diff --git a/tools/perf/trace/beauty/socket.sh b/tools/perf/trace/beauty/socket.sh
index 3820e5c..76330ac 100755
--- a/tools/perf/trace/beauty/socket.sh
+++ b/tools/perf/trace/beauty/socket.sh
@@ -1,24 +1,28 @@
 #!/bin/sh
 # SPDX-License-Identifier: LGPL-2.1
 
-# This one uses a copy from the kernel sources headers that is in a
-# place used just for these tools/perf/beauty/ usage, we shouldn't not
-# put it in tools/include/linux otherwise they would be used in the
-# normal compiler building process and would drag needless stuff from the
-# kernel.
+if [ $# -gt 0 ] ; then
+	uapi_header_dir=$1
+	beauty_header_dir=$2
+else
+	uapi_header_dir=tools/include/uapi/linux/
+	beauty_header_dir=tools/perf/trace/beauty/include/linux/
+fi
 
-# When what these scripts need is already in tools/include/ then use it,
-# otherwise grab and check the copy from the kernel sources just for these
-# string table building scripts.
+printf "static const char *socket_ipproto[] = {\n"
+ipproto_regex='^[[:space:]]+IPPROTO_(\w+)[[:space:]]+=[[:space:]]+([[:digit:]]+),.*'
 
-[ $# -eq 1 ] && header_dir=$1 || header_dir=tools/perf/trace/beauty/include/linux/
+egrep $ipproto_regex ${uapi_header_dir}/in.h | \
+	sed -r "s/$ipproto_regex/\2 \1/g"	| \
+	sort -n | xargs printf "\t[%s] = \"%s\",\n"
+printf "};\n\n"
 
-printf "static const char *socket_families[] = {\n"
-# #define AF_LOCAL	1	/* POSIX name for AF_UNIX	*/
-regex='^#define[[:space:]]+AF_(\w+)[[:space:]]+([[:digit:]]+).*'
+printf "static const char *socket_level[] = {\n"
+socket_level_regex='^#define[[:space:]]+SOL_(\w+)[[:space:]]+([[:digit:]]+)([[:space:]]+\/.*)?'
 
-egrep $regex ${header_dir}/socket.h | \
-	sed -r "s/$regex/\2 \1/g"	| \
-	xargs printf "\t[%s] = \"%s\",\n" | \
-	egrep -v "\"(UNIX|MAX)\""
-printf "};\n"
+egrep $socket_level_regex ${beauty_header_dir}/socket.h | \
+	sed -r "s/$socket_level_regex/\2 \1/g"	| \
+	sort -n | xargs printf "\t[%s] = \"%s\",\n"
+printf "};\n\n"
+
+printf 'DEFINE_STRARRAY(socket_level, "SOL_");\n'
diff --git a/tools/perf/trace/beauty/socket_ipproto.sh b/tools/perf/trace/beauty/socket_ipproto.sh
deleted file mode 100755
index de0f2f2..0000000
--- a/tools/perf/trace/beauty/socket_ipproto.sh
+++ /dev/null
@@ -1,12 +0,0 @@
-#!/bin/sh
-# SPDX-License-Identifier: LGPL-2.1
-
-[ $# -eq 1 ] && header_dir=$1 || header_dir=tools/include/uapi/linux/
-
-printf "static const char *socket_ipproto[] = {\n"
-regex='^[[:space:]]+IPPROTO_(\w+)[[:space:]]+=[[:space:]]+([[:digit:]]+),.*'
-
-egrep $regex ${header_dir}/in.h | \
-	sed -r "s/$regex/\2 \1/g"	| \
-	sort | xargs printf "\t[%s] = \"%s\",\n"
-printf "};\n"
diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c
index 8511af5..0190068 100644
--- a/tools/perf/util/annotate.c
+++ b/tools/perf/util/annotate.c
@@ -1255,6 +1255,17 @@ int disasm_line__scnprintf(struct disasm_line *dl, char *bf, size_t size, bool r
 	return ins__scnprintf(&dl->ins, bf, size, &dl->ops, max_ins_name);
 }
 
+void annotation__init(struct annotation *notes)
+{
+	pthread_mutex_init(&notes->lock, NULL);
+}
+
+void annotation__exit(struct annotation *notes)
+{
+	annotated_source__delete(notes->src);
+	pthread_mutex_destroy(&notes->lock);
+}
+
 static void annotation_line__add(struct annotation_line *al, struct list_head *head)
 {
 	list_add_tail(&al->node, head);
@@ -3132,7 +3143,7 @@ int symbol__annotate2(struct map_symbol *ms, struct evsel *evsel,
 	notes->nr_events = nr_pcnt;
 
 	annotation__update_column_widths(notes);
-	sym->annotate2 = true;
+	sym->annotate2 = 1;
 
 	return 0;
 
diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h
index 3757416..986f2bb 100644
--- a/tools/perf/util/annotate.h
+++ b/tools/perf/util/annotate.h
@@ -299,6 +299,9 @@ struct annotation {
 	struct annotated_source *src;
 };
 
+void annotation__init(struct annotation *notes);
+void annotation__exit(struct annotation *notes);
+
 static inline int annotation__cycles_width(struct annotation *notes)
 {
 	if (notes->have_cycles && notes->options->show_minmax_cycle)
diff --git a/tools/perf/util/arm-spe-decoder/arm-spe-decoder.c b/tools/perf/util/arm-spe-decoder/arm-spe-decoder.c
index 32fe418..3fc528c 100644
--- a/tools/perf/util/arm-spe-decoder/arm-spe-decoder.c
+++ b/tools/perf/util/arm-spe-decoder/arm-spe-decoder.c
@@ -151,6 +151,7 @@ static int arm_spe_read_record(struct arm_spe_decoder *decoder)
 	u64 payload, ip;
 
 	memset(&decoder->record, 0x0, sizeof(decoder->record));
+	decoder->record.context_id = (u64)-1;
 
 	while (1) {
 		err = arm_spe_get_next_packet(decoder);
@@ -180,6 +181,7 @@ static int arm_spe_read_record(struct arm_spe_decoder *decoder)
 		case ARM_SPE_COUNTER:
 			break;
 		case ARM_SPE_CONTEXT:
+			decoder->record.context_id = payload;
 			break;
 		case ARM_SPE_OP_TYPE:
 			if (idx == SPE_OP_PKT_HDR_CLASS_LD_ST_ATOMIC) {
diff --git a/tools/perf/util/arm-spe-decoder/arm-spe-decoder.h b/tools/perf/util/arm-spe-decoder/arm-spe-decoder.h
index 59bdb73..46a8556 100644
--- a/tools/perf/util/arm-spe-decoder/arm-spe-decoder.h
+++ b/tools/perf/util/arm-spe-decoder/arm-spe-decoder.h
@@ -38,6 +38,7 @@ struct arm_spe_record {
 	u64 timestamp;
 	u64 virt_addr;
 	u64 phys_addr;
+	u64 context_id;
 };
 
 struct arm_spe_insn;
diff --git a/tools/perf/util/arm-spe.c b/tools/perf/util/arm-spe.c
index 58b7069..4748bcf 100644
--- a/tools/perf/util/arm-spe.c
+++ b/tools/perf/util/arm-spe.c
@@ -71,6 +71,7 @@ struct arm_spe {
 	u64				kernel_start;
 
 	unsigned long			num_events;
+	u8				use_ctx_pkt_for_pid;
 };
 
 struct arm_spe_queue {
@@ -100,7 +101,7 @@ static void arm_spe_dump(struct arm_spe *spe __maybe_unused,
 	const char *color = PERF_COLOR_BLUE;
 
 	color_fprintf(stdout, color,
-		      ". ... ARM SPE data: size %zu bytes\n",
+		      ". ... ARM SPE data: size %#zx bytes\n",
 		      len);
 
 	while (len) {
@@ -226,6 +227,44 @@ static inline u8 arm_spe_cpumode(struct arm_spe *spe, u64 ip)
 		PERF_RECORD_MISC_USER;
 }
 
+static void arm_spe_set_pid_tid_cpu(struct arm_spe *spe,
+				    struct auxtrace_queue *queue)
+{
+	struct arm_spe_queue *speq = queue->priv;
+	pid_t tid;
+
+	tid = machine__get_current_tid(spe->machine, speq->cpu);
+	if (tid != -1) {
+		speq->tid = tid;
+		thread__zput(speq->thread);
+	} else
+		speq->tid = queue->tid;
+
+	if ((!speq->thread) && (speq->tid != -1)) {
+		speq->thread = machine__find_thread(spe->machine, -1,
+						    speq->tid);
+	}
+
+	if (speq->thread) {
+		speq->pid = speq->thread->pid_;
+		if (queue->cpu == -1)
+			speq->cpu = speq->thread->cpu;
+	}
+}
+
+static int arm_spe_set_tid(struct arm_spe_queue *speq, pid_t tid)
+{
+	struct arm_spe *spe = speq->spe;
+	int err = machine__set_current_tid(spe->machine, speq->cpu, -1, tid);
+
+	if (err)
+		return err;
+
+	arm_spe_set_pid_tid_cpu(spe, &spe->queues.queue_array[speq->queue_nr]);
+
+	return 0;
+}
+
 static void arm_spe_prep_sample(struct arm_spe *spe,
 				struct arm_spe_queue *speq,
 				union perf_event *event,
@@ -460,6 +499,19 @@ static int arm_spe_run_decoder(struct arm_spe_queue *speq, u64 *timestamp)
 		 * can correlate samples between Arm SPE trace data and other
 		 * perf events with correct time ordering.
 		 */
+
+		/*
+		 * Update pid/tid info.
+		 */
+		record = &speq->decoder->record;
+		if (!spe->timeless_decoding && record->context_id != (u64)-1) {
+			ret = arm_spe_set_tid(speq, record->context_id);
+			if (ret)
+				return ret;
+
+			spe->use_ctx_pkt_for_pid = true;
+		}
+
 		ret = arm_spe_sample(speq);
 		if (ret)
 			return ret;
@@ -586,31 +638,6 @@ static bool arm_spe__is_timeless_decoding(struct arm_spe *spe)
 	return timeless_decoding;
 }
 
-static void arm_spe_set_pid_tid_cpu(struct arm_spe *spe,
-				    struct auxtrace_queue *queue)
-{
-	struct arm_spe_queue *speq = queue->priv;
-	pid_t tid;
-
-	tid = machine__get_current_tid(spe->machine, speq->cpu);
-	if (tid != -1) {
-		speq->tid = tid;
-		thread__zput(speq->thread);
-	} else
-		speq->tid = queue->tid;
-
-	if ((!speq->thread) && (speq->tid != -1)) {
-		speq->thread = machine__find_thread(spe->machine, -1,
-						    speq->tid);
-	}
-
-	if (speq->thread) {
-		speq->pid = speq->thread->pid_;
-		if (queue->cpu == -1)
-			speq->cpu = speq->thread->cpu;
-	}
-}
-
 static int arm_spe_process_queues(struct arm_spe *spe, u64 timestamp)
 {
 	unsigned int queue_nr;
@@ -641,7 +668,12 @@ static int arm_spe_process_queues(struct arm_spe *spe, u64 timestamp)
 			ts = timestamp;
 		}
 
-		arm_spe_set_pid_tid_cpu(spe, queue);
+		/*
+		 * A previous context-switch event has set pid/tid in the machine's context, so
+		 * here we need to update the pid/tid in the thread and SPE queue.
+		 */
+		if (!spe->use_ctx_pkt_for_pid)
+			arm_spe_set_pid_tid_cpu(spe, queue);
 
 		ret = arm_spe_run_decoder(speq, &ts);
 		if (ret < 0) {
@@ -681,6 +713,25 @@ static int arm_spe_process_timeless_queues(struct arm_spe *spe, pid_t tid,
 	return 0;
 }
 
+static int arm_spe_context_switch(struct arm_spe *spe, union perf_event *event,
+				  struct perf_sample *sample)
+{
+	pid_t pid, tid;
+	int cpu;
+
+	if (!(event->header.misc & PERF_RECORD_MISC_SWITCH_OUT))
+		return 0;
+
+	pid = event->context_switch.next_prev_pid;
+	tid = event->context_switch.next_prev_tid;
+	cpu = sample->cpu;
+
+	if (tid == -1)
+		pr_warning("context_switch event has no tid\n");
+
+	return machine__set_current_tid(spe->machine, cpu, pid, tid);
+}
+
 static int arm_spe_process_event(struct perf_session *session,
 				 union perf_event *event,
 				 struct perf_sample *sample,
@@ -718,6 +769,13 @@ static int arm_spe_process_event(struct perf_session *session,
 		}
 	} else if (timestamp) {
 		err = arm_spe_process_queues(spe, timestamp);
+		if (err)
+			return err;
+
+		if (!spe->use_ctx_pkt_for_pid &&
+		    (event->header.type == PERF_RECORD_SWITCH_CPU_WIDE ||
+		    event->header.type == PERF_RECORD_SWITCH))
+			err = arm_spe_context_switch(spe, event, sample);
 	}
 
 	return err;
@@ -783,7 +841,15 @@ static int arm_spe_flush(struct perf_session *session __maybe_unused,
 		return arm_spe_process_timeless_queues(spe, -1,
 				MAX_TIMESTAMP - 1);
 
-	return arm_spe_process_queues(spe, MAX_TIMESTAMP);
+	ret = arm_spe_process_queues(spe, MAX_TIMESTAMP);
+	if (ret)
+		return ret;
+
+	if (!spe->use_ctx_pkt_for_pid)
+		ui__warning("Arm SPE CONTEXT packets not found in the traces.\n"
+			    "Matching of TIDs to SPE events could be inaccurate.\n");
+
+	return 0;
 }
 
 static void arm_spe_free_queue(void *priv)
diff --git a/tools/perf/util/bpf-event.c b/tools/perf/util/bpf-event.c
index 4d3b4cd..a517eaa 100644
--- a/tools/perf/util/bpf-event.c
+++ b/tools/perf/util/bpf-event.c
@@ -33,6 +33,33 @@ struct btf * __weak btf__load_from_kernel_by_id(__u32 id)
        return err ? ERR_PTR(err) : btf;
 }
 
+struct bpf_program * __weak
+bpf_object__next_program(const struct bpf_object *obj, struct bpf_program *prev)
+{
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
+	return bpf_program__next(prev, obj);
+#pragma GCC diagnostic pop
+}
+
+struct bpf_map * __weak
+bpf_object__next_map(const struct bpf_object *obj, const struct bpf_map *prev)
+{
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
+	return bpf_map__next(prev, obj);
+#pragma GCC diagnostic pop
+}
+
+const void * __weak
+btf__raw_data(const struct btf *btf_ro, __u32 *size)
+{
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
+	return btf__get_raw_data(btf_ro, size);
+#pragma GCC diagnostic pop
+}
+
 static int snprintf_hex(char *buf, size_t size, unsigned char *data, size_t len)
 {
 	int ret = 0;
@@ -119,7 +146,11 @@ static int perf_env__fetch_btf(struct perf_env *env,
 	node->data_size = data_size;
 	memcpy(node->data, data, data_size);
 
-	perf_env__insert_btf(env, node);
+	if (!perf_env__insert_btf(env, node)) {
+		/* Insertion failed because of a duplicate. */
+		free(node);
+		return -1;
+	}
 	return 0;
 }
 
diff --git a/tools/perf/util/c++/clang-c.h b/tools/perf/util/c++/clang-c.h
index 2df8a45..d3731a8 100644
--- a/tools/perf/util/c++/clang-c.h
+++ b/tools/perf/util/c++/clang-c.h
@@ -12,8 +12,9 @@ extern "C" {
 extern void perf_clang__init(void);
 extern void perf_clang__cleanup(void);
 
-extern int test__clang_to_IR(void);
-extern int test__clang_to_obj(void);
+struct test_suite;
+extern int test__clang_to_IR(struct test_suite *test, int subtest);
+extern int test__clang_to_obj(struct test_suite *test, int subtest);
 
 extern int perf_clang__compile_bpf(const char *filename,
 				   void **p_obj_buf,
@@ -26,9 +27,6 @@ extern int perf_clang__compile_bpf(const char *filename,
 static inline void perf_clang__init(void) { }
 static inline void perf_clang__cleanup(void) { }
 
-static inline int test__clang_to_IR(void) { return -1; }
-static inline int test__clang_to_obj(void) { return -1;}
-
 static inline int
 perf_clang__compile_bpf(const char *filename __maybe_unused,
 			void **p_obj_buf __maybe_unused,
diff --git a/tools/perf/util/c++/clang-test.cpp b/tools/perf/util/c++/clang-test.cpp
index 21b2360..a4683ca 100644
--- a/tools/perf/util/c++/clang-test.cpp
+++ b/tools/perf/util/c++/clang-test.cpp
@@ -35,7 +35,8 @@
 }
 
 extern "C" {
-int test__clang_to_IR(void)
+int test__clang_to_IR(struct test_suite *test __maybe_unused,
+                      int subtest __maybe_unused)
 {
 	perf_clang_scope _scope;
 
@@ -48,7 +49,8 @@
 	return -1;
 }
 
-int test__clang_to_obj(void)
+int test__clang_to_obj(struct test_suite *test __maybe_unused,
+                       int subtest __maybe_unused)
 {
 	perf_clang_scope _scope;
 
diff --git a/tools/perf/util/cputopo.c b/tools/perf/util/cputopo.c
index ec77e2a..51b429c 100644
--- a/tools/perf/util/cputopo.c
+++ b/tools/perf/util/cputopo.c
@@ -14,14 +14,16 @@
 #include "env.h"
 #include "pmu-hybrid.h"
 
-#define CORE_SIB_FMT \
+#define PACKAGE_CPUS_FMT \
+	"%s/devices/system/cpu/cpu%d/topology/package_cpus_list"
+#define PACKAGE_CPUS_FMT_OLD \
 	"%s/devices/system/cpu/cpu%d/topology/core_siblings_list"
-#define DIE_SIB_FMT \
+#define DIE_CPUS_FMT \
 	"%s/devices/system/cpu/cpu%d/topology/die_cpus_list"
-#define THRD_SIB_FMT \
-	"%s/devices/system/cpu/cpu%d/topology/thread_siblings_list"
-#define THRD_SIB_FMT_NEW \
+#define CORE_CPUS_FMT \
 	"%s/devices/system/cpu/cpu%d/topology/core_cpus_list"
+#define CORE_CPUS_FMT_OLD \
+	"%s/devices/system/cpu/cpu%d/topology/thread_siblings_list"
 #define NODE_ONLINE_FMT \
 	"%s/devices/system/node/online"
 #define NODE_MEMINFO_FMT \
@@ -39,8 +41,12 @@ static int build_cpu_topology(struct cpu_topology *tp, int cpu)
 	u32 i = 0;
 	int ret = -1;
 
-	scnprintf(filename, MAXPATHLEN, CORE_SIB_FMT,
+	scnprintf(filename, MAXPATHLEN, PACKAGE_CPUS_FMT,
 		  sysfs__mountpoint(), cpu);
+	if (access(filename, F_OK) == -1) {
+		scnprintf(filename, MAXPATHLEN, PACKAGE_CPUS_FMT_OLD,
+			sysfs__mountpoint(), cpu);
+	}
 	fp = fopen(filename, "r");
 	if (!fp)
 		goto try_dies;
@@ -54,23 +60,23 @@ static int build_cpu_topology(struct cpu_topology *tp, int cpu)
 	if (p)
 		*p = '\0';
 
-	for (i = 0; i < tp->core_sib; i++) {
-		if (!strcmp(buf, tp->core_siblings[i]))
+	for (i = 0; i < tp->package_cpus_lists; i++) {
+		if (!strcmp(buf, tp->package_cpus_list[i]))
 			break;
 	}
-	if (i == tp->core_sib) {
-		tp->core_siblings[i] = buf;
-		tp->core_sib++;
+	if (i == tp->package_cpus_lists) {
+		tp->package_cpus_list[i] = buf;
+		tp->package_cpus_lists++;
 		buf = NULL;
 		len = 0;
 	}
 	ret = 0;
 
 try_dies:
-	if (!tp->die_siblings)
+	if (!tp->die_cpus_list)
 		goto try_threads;
 
-	scnprintf(filename, MAXPATHLEN, DIE_SIB_FMT,
+	scnprintf(filename, MAXPATHLEN, DIE_CPUS_FMT,
 		  sysfs__mountpoint(), cpu);
 	fp = fopen(filename, "r");
 	if (!fp)
@@ -85,23 +91,23 @@ static int build_cpu_topology(struct cpu_topology *tp, int cpu)
 	if (p)
 		*p = '\0';
 
-	for (i = 0; i < tp->die_sib; i++) {
-		if (!strcmp(buf, tp->die_siblings[i]))
+	for (i = 0; i < tp->die_cpus_lists; i++) {
+		if (!strcmp(buf, tp->die_cpus_list[i]))
 			break;
 	}
-	if (i == tp->die_sib) {
-		tp->die_siblings[i] = buf;
-		tp->die_sib++;
+	if (i == tp->die_cpus_lists) {
+		tp->die_cpus_list[i] = buf;
+		tp->die_cpus_lists++;
 		buf = NULL;
 		len = 0;
 	}
 	ret = 0;
 
 try_threads:
-	scnprintf(filename, MAXPATHLEN, THRD_SIB_FMT_NEW,
+	scnprintf(filename, MAXPATHLEN, CORE_CPUS_FMT,
 		  sysfs__mountpoint(), cpu);
 	if (access(filename, F_OK) == -1) {
-		scnprintf(filename, MAXPATHLEN, THRD_SIB_FMT,
+		scnprintf(filename, MAXPATHLEN, CORE_CPUS_FMT_OLD,
 			  sysfs__mountpoint(), cpu);
 	}
 	fp = fopen(filename, "r");
@@ -115,13 +121,13 @@ static int build_cpu_topology(struct cpu_topology *tp, int cpu)
 	if (p)
 		*p = '\0';
 
-	for (i = 0; i < tp->thread_sib; i++) {
-		if (!strcmp(buf, tp->thread_siblings[i]))
+	for (i = 0; i < tp->core_cpus_lists; i++) {
+		if (!strcmp(buf, tp->core_cpus_list[i]))
 			break;
 	}
-	if (i == tp->thread_sib) {
-		tp->thread_siblings[i] = buf;
-		tp->thread_sib++;
+	if (i == tp->core_cpus_lists) {
+		tp->core_cpus_list[i] = buf;
+		tp->core_cpus_lists++;
 		buf = NULL;
 	}
 	ret = 0;
@@ -139,16 +145,14 @@ void cpu_topology__delete(struct cpu_topology *tp)
 	if (!tp)
 		return;
 
-	for (i = 0 ; i < tp->core_sib; i++)
-		zfree(&tp->core_siblings[i]);
+	for (i = 0 ; i < tp->package_cpus_lists; i++)
+		zfree(&tp->package_cpus_list[i]);
 
-	if (tp->die_sib) {
-		for (i = 0 ; i < tp->die_sib; i++)
-			zfree(&tp->die_siblings[i]);
-	}
+	for (i = 0 ; i < tp->die_cpus_lists; i++)
+		zfree(&tp->die_cpus_list[i]);
 
-	for (i = 0 ; i < tp->thread_sib; i++)
-		zfree(&tp->thread_siblings[i]);
+	for (i = 0 ; i < tp->core_cpus_lists; i++)
+		zfree(&tp->core_cpus_list[i]);
 
 	free(tp);
 }
@@ -164,7 +168,7 @@ static bool has_die_topology(void)
 	if (strncmp(uts.machine, "x86_64", 6))
 		return false;
 
-	scnprintf(filename, MAXPATHLEN, DIE_SIB_FMT,
+	scnprintf(filename, MAXPATHLEN, DIE_CPUS_FMT,
 		  sysfs__mountpoint(), 0);
 	if (access(filename, F_OK) == -1)
 		return false;
@@ -205,13 +209,13 @@ struct cpu_topology *cpu_topology__new(void)
 
 	tp = addr;
 	addr += sizeof(*tp);
-	tp->core_siblings = addr;
+	tp->package_cpus_list = addr;
 	addr += sz;
 	if (has_die) {
-		tp->die_siblings = addr;
+		tp->die_cpus_list = addr;
 		addr += sz;
 	}
-	tp->thread_siblings = addr;
+	tp->core_cpus_list = addr;
 
 	for (i = 0; i < nr; i++) {
 		if (!cpu_map__has(map, i))
diff --git a/tools/perf/util/cputopo.h b/tools/perf/util/cputopo.h
index d9af971..854e18f 100644
--- a/tools/perf/util/cputopo.h
+++ b/tools/perf/util/cputopo.h
@@ -5,12 +5,33 @@
 #include <linux/types.h>
 
 struct cpu_topology {
-	u32	  core_sib;
-	u32	  die_sib;
-	u32	  thread_sib;
-	char	**core_siblings;
-	char	**die_siblings;
-	char	**thread_siblings;
+	/* The number of unique package_cpus_lists below. */
+	u32	  package_cpus_lists;
+	/* The number of unique die_cpu_lists below. */
+	u32	  die_cpus_lists;
+	/* The number of unique core_cpu_lists below. */
+	u32	  core_cpus_lists;
+	/*
+	 * An array of strings where each string is unique and read from
+	 * /sys/devices/system/cpu/cpuX/topology/package_cpus_list. From the ABI
+	 * each of these is a human-readable list of CPUs sharing the same
+	 * physical_package_id. The format is like 0-3, 8-11, 14,17.
+	 */
+	const char **package_cpus_list;
+	/*
+	 * An array of string where each string is unique and from
+	 * /sys/devices/system/cpu/cpuX/topology/die_cpus_list. From the ABI
+	 * each of these is a human-readable list of CPUs within the same die.
+	 * The format is like 0-3, 8-11, 14,17.
+	 */
+	const char **die_cpus_list;
+	/*
+	 * An array of string where each string is unique and from
+	 * /sys/devices/system/cpu/cpuX/topology/core_cpus_list. From the ABI
+	 * each of these is a human-readable list of CPUs within the same
+	 * core. The format is like 0-3, 8-11, 14,17.
+	 */
+	const char **core_cpus_list;
 };
 
 struct numa_topology_node {
diff --git a/tools/perf/util/cs-etm.c b/tools/perf/util/cs-etm.c
index f323adb..4f672f7 100644
--- a/tools/perf/util/cs-etm.c
+++ b/tools/perf/util/cs-etm.c
@@ -537,7 +537,7 @@ static void cs_etm__dump_event(struct cs_etm_queue *etmq,
 
 	fprintf(stdout, "\n");
 	color_fprintf(stdout, color,
-		     ". ... CoreSight %s Trace data: size %zu bytes\n",
+		     ". ... CoreSight %s Trace data: size %#zx bytes\n",
 		     cs_etm_decoder__get_name(etmq->decoder), buffer->size);
 
 	do {
diff --git a/tools/perf/util/env.c b/tools/perf/util/env.c
index 17f1dd0..b990489 100644
--- a/tools/perf/util/env.c
+++ b/tools/perf/util/env.c
@@ -75,12 +75,13 @@ struct bpf_prog_info_node *perf_env__find_bpf_prog_info(struct perf_env *env,
 	return node;
 }
 
-void perf_env__insert_btf(struct perf_env *env, struct btf_node *btf_node)
+bool perf_env__insert_btf(struct perf_env *env, struct btf_node *btf_node)
 {
 	struct rb_node *parent = NULL;
 	__u32 btf_id = btf_node->id;
 	struct btf_node *node;
 	struct rb_node **p;
+	bool ret = true;
 
 	down_write(&env->bpf_progs.lock);
 	p = &env->bpf_progs.btfs.rb_node;
@@ -94,6 +95,7 @@ void perf_env__insert_btf(struct perf_env *env, struct btf_node *btf_node)
 			p = &(*p)->rb_right;
 		} else {
 			pr_debug("duplicated btf %u\n", btf_id);
+			ret = false;
 			goto out;
 		}
 	}
@@ -103,6 +105,7 @@ void perf_env__insert_btf(struct perf_env *env, struct btf_node *btf_node)
 	env->bpf_progs.btfs_cnt++;
 out:
 	up_write(&env->bpf_progs.lock);
+	return ret;
 }
 
 struct btf_node *perf_env__find_btf(struct perf_env *env, __u32 btf_id)
diff --git a/tools/perf/util/env.h b/tools/perf/util/env.h
index 1383876..163e5ec 100644
--- a/tools/perf/util/env.h
+++ b/tools/perf/util/env.h
@@ -167,7 +167,7 @@ void perf_env__insert_bpf_prog_info(struct perf_env *env,
 				    struct bpf_prog_info_node *info_node);
 struct bpf_prog_info_node *perf_env__find_bpf_prog_info(struct perf_env *env,
 							__u32 prog_id);
-void perf_env__insert_btf(struct perf_env *env, struct btf_node *btf_node);
+bool perf_env__insert_btf(struct perf_env *env, struct btf_node *btf_node);
 struct btf_node *perf_env__find_btf(struct perf_env *env, __u32 btf_id);
 
 int perf_env__numa_node(struct perf_env *env, int cpu);
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index ec967fb..a59fb2e 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -3037,3 +3037,15 @@ void evsel__set_leader(struct evsel *evsel, struct evsel *leader)
 {
 	evsel->core.leader = &leader->core;
 }
+
+int evsel__source_count(const struct evsel *evsel)
+{
+	struct evsel *pos;
+	int count = 0;
+
+	evlist__for_each_entry(evsel->evlist, pos) {
+		if (pos->metric_leader == evsel)
+			count++;
+	}
+	return count;
+}
diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index 3ea6871..29d49a8 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -489,6 +489,7 @@ struct evsel *evsel__leader(struct evsel *evsel);
 bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
 bool evsel__is_leader(struct evsel *evsel);
 void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+int evsel__source_count(const struct evsel *evsel);
 
 /*
  * Macro to swap the bit-field postition and size.
diff --git a/tools/perf/util/expr.c b/tools/perf/util/expr.c
index 77c6ad8..1d532b9 100644
--- a/tools/perf/util/expr.c
+++ b/tools/perf/util/expr.c
@@ -5,13 +5,17 @@
 #include <stdlib.h>
 #include <string.h>
 #include "metricgroup.h"
+#include "cpumap.h"
+#include "cputopo.h"
 #include "debug.h"
 #include "expr.h"
 #include "expr-bison.h"
 #include "expr-flex.h"
+#include "smt.h"
 #include <linux/kernel.h>
 #include <linux/zalloc.h>
 #include <ctype.h>
+#include <math.h>
 
 #ifdef PARSER_DEBUG
 extern int expr_debug;
@@ -19,7 +23,10 @@ extern int expr_debug;
 
 struct expr_id_data {
 	union {
-		double val;
+		struct {
+			double val;
+			int source_count;
+		} val;
 		struct {
 			double val;
 			const char *metric_name;
@@ -137,6 +144,13 @@ int expr__add_id(struct expr_parse_ctx *ctx, const char *id)
 /* Caller must make sure id is allocated */
 int expr__add_id_val(struct expr_parse_ctx *ctx, const char *id, double val)
 {
+	return expr__add_id_val_source_count(ctx, id, val, /*source_count=*/1);
+}
+
+/* Caller must make sure id is allocated */
+int expr__add_id_val_source_count(struct expr_parse_ctx *ctx, const char *id,
+				  double val, int source_count)
+{
 	struct expr_id_data *data_ptr = NULL, *old_data = NULL;
 	char *old_key = NULL;
 	int ret;
@@ -144,7 +158,8 @@ int expr__add_id_val(struct expr_parse_ctx *ctx, const char *id, double val)
 	data_ptr = malloc(sizeof(*data_ptr));
 	if (!data_ptr)
 		return -ENOMEM;
-	data_ptr->val = val;
+	data_ptr->val.val = val;
+	data_ptr->val.source_count = source_count;
 	data_ptr->kind = EXPR_ID_DATA__VALUE;
 
 	ret = hashmap__set(ctx->ids, id, data_ptr,
@@ -240,7 +255,7 @@ int expr__resolve_id(struct expr_parse_ctx *ctx, const char *id,
 
 	switch (data->kind) {
 	case EXPR_ID_DATA__VALUE:
-		pr_debug2("lookup(%s): val %f\n", id, data->val);
+		pr_debug2("lookup(%s): val %f\n", id, data->val.val);
 		break;
 	case EXPR_ID_DATA__REF:
 		pr_debug2("lookup(%s): ref metric name %s\n", id,
@@ -251,7 +266,7 @@ int expr__resolve_id(struct expr_parse_ctx *ctx, const char *id,
 			pr_debug("%s failed to count\n", id);
 			return -1;
 		}
-		pr_debug("processing metric: %s EXIT: %f\n", id, data->val);
+		pr_debug("processing metric: %s EXIT: %f\n", id, data->ref.val);
 		break;
 	case EXPR_ID_DATA__REF_VALUE:
 		pr_debug2("lookup(%s): ref val %f metric name %s\n", id,
@@ -366,7 +381,47 @@ int expr__find_ids(const char *expr, const char *one,
 double expr_id_data__value(const struct expr_id_data *data)
 {
 	if (data->kind == EXPR_ID_DATA__VALUE)
-		return data->val;
+		return data->val.val;
 	assert(data->kind == EXPR_ID_DATA__REF_VALUE);
 	return data->ref.val;
 }
+
+double expr_id_data__source_count(const struct expr_id_data *data)
+{
+	assert(data->kind == EXPR_ID_DATA__VALUE);
+	return data->val.source_count;
+}
+
+double expr__get_literal(const char *literal)
+{
+	static struct cpu_topology *topology;
+
+	if (!strcmp("#smt_on", literal))
+		return smt_on() > 0 ? 1.0 : 0.0;
+
+	if (!strcmp("#num_cpus", literal))
+		return cpu__max_present_cpu();
+
+	/*
+	 * Assume that topology strings are consistent, such as CPUs "0-1"
+	 * wouldn't be listed as "0,1", and so after deduplication the number of
+	 * these strings gives an indication of the number of packages, dies,
+	 * etc.
+	 */
+	if (!topology) {
+		topology = cpu_topology__new();
+		if (!topology) {
+			pr_err("Error creating CPU topology");
+			return NAN;
+		}
+	}
+	if (!strcmp("#num_packages", literal))
+		return topology->package_cpus_lists;
+	if (!strcmp("#num_dies", literal))
+		return topology->die_cpus_lists;
+	if (!strcmp("#num_cores", literal))
+		return topology->core_cpus_lists;
+
+	pr_err("Unrecognized literal '%s'", literal);
+	return NAN;
+}
diff --git a/tools/perf/util/expr.h b/tools/perf/util/expr.h
index cf81f91..bd21169 100644
--- a/tools/perf/util/expr.h
+++ b/tools/perf/util/expr.h
@@ -40,6 +40,8 @@ void expr__ctx_free(struct expr_parse_ctx *ctx);
 void expr__del_id(struct expr_parse_ctx *ctx, const char *id);
 int expr__add_id(struct expr_parse_ctx *ctx, const char *id);
 int expr__add_id_val(struct expr_parse_ctx *ctx, const char *id, double val);
+int expr__add_id_val_source_count(struct expr_parse_ctx *ctx, const char *id,
+				double val, int source_count);
 int expr__add_ref(struct expr_parse_ctx *ctx, struct metric_ref *ref);
 int expr__get_id(struct expr_parse_ctx *ctx, const char *id,
 		 struct expr_id_data **data);
@@ -55,5 +57,7 @@ int expr__find_ids(const char *expr, const char *one,
 		   struct expr_parse_ctx *ids);
 
 double expr_id_data__value(const struct expr_id_data *data);
+double expr_id_data__source_count(const struct expr_id_data *data);
+double expr__get_literal(const char *literal);
 
 #endif
diff --git a/tools/perf/util/expr.l b/tools/perf/util/expr.l
index bd20f33..0a13eb2 100644
--- a/tools/perf/util/expr.l
+++ b/tools/perf/util/expr.l
@@ -6,6 +6,7 @@
 #include <linux/compiler.h>
 #include "expr.h"
 #include "expr-bison.h"
+#include <math.h>
 
 char *expr_get_text(yyscan_t yyscanner);
 YYSTYPE *expr_get_lval(yyscan_t yyscanner);
@@ -77,6 +78,17 @@
 	yylval->str = normalize(yylval->str, runtime);
 	return token;
 }
+
+static int literal(yyscan_t scanner)
+{
+	YYSTYPE *yylval = expr_get_lval(scanner);
+
+	yylval->num = expr__get_literal(expr_get_text(scanner));
+	if (isnan(yylval->num))
+		return EXPR_ERROR;
+
+	return LITERAL;
+}
 %}
 
 number		([0-9]+\.?[0-9]*|[0-9]*\.?[0-9]+)
@@ -85,6 +97,7 @@
 spec		\\{sch}
 sym		[0-9a-zA-Z_\.:@?]+
 symbol		({spec}|{sym})+
+literal		#[0-9a-zA-Z_\.\-]+
 
 %%
 	struct expr_scanner_ctx *sctx = expr_get_extra(yyscanner);
@@ -94,7 +107,8 @@
 min		{ return MIN; }
 if		{ return IF; }
 else		{ return ELSE; }
-#smt_on		{ return SMT_ON; }
+source_count	{ return SOURCE_COUNT; }
+{literal}	{ return literal(yyscanner); }
 {number}	{ return value(yyscanner); }
 {symbol}	{ return str(yyscanner, ID, sctx->runtime); }
 "|"		{ return '|'; }
diff --git a/tools/perf/util/expr.y b/tools/perf/util/expr.y
index f969dfa..a30b825 100644
--- a/tools/perf/util/expr.y
+++ b/tools/perf/util/expr.y
@@ -3,8 +3,8 @@
 #define YYDEBUG 1
 #include <assert.h>
 #include <math.h>
+#include <stdlib.h>
 #include "util/debug.h"
-#include "smt.h"
 #define IN_EXPR_Y 1
 #include "expr.h"
 %}
@@ -37,7 +37,7 @@
 	} ids;
 }
 
-%token ID NUMBER MIN MAX IF ELSE SMT_ON D_RATIO EXPR_ERROR
+%token ID NUMBER MIN MAX IF ELSE LITERAL D_RATIO SOURCE_COUNT EXPR_ERROR
 %left MIN MAX IF
 %left '|'
 %left '^'
@@ -46,7 +46,7 @@
 %left '-' '+'
 %left '*' '/' '%'
 %left NEG NOT
-%type <num> NUMBER
+%type <num> NUMBER LITERAL
 %type <str> ID
 %destructor { free ($$); } <str>
 %type <ids> expr if_expr
@@ -83,6 +83,41 @@
 	return result;
 }
 
+static struct ids handle_id(struct expr_parse_ctx *ctx, char *id,
+			    bool compute_ids, bool source_count)
+{
+	struct ids result;
+
+	if (!compute_ids) {
+		/*
+		 * Compute the event's value from ID. If the ID isn't known then
+		 * it isn't used to compute the formula so set to NAN.
+		 */
+		struct expr_id_data *data;
+
+		result.val = NAN;
+		if (expr__resolve_id(ctx, id, &data) == 0) {
+			result.val = source_count
+				? expr_id_data__source_count(data)
+				: expr_id_data__value(data);
+		}
+		result.ids = NULL;
+		free(id);
+	} else {
+		/*
+		 * Set the value to BOTTOM to show that any value is possible
+		 * when the event is computed. Create a set of just the ID.
+		 */
+		result.val = BOTTOM;
+		result.ids = ids__new();
+		if (!result.ids || ids__insert(result.ids, id)) {
+			pr_err("Error creating IDs for '%s'", id);
+			free(id);
+		}
+	}
+	return result;
+}
+
 /*
  * If we're not computing ids or $1 and $3 are constants, compute the new
  * constant value using OP. Its invariant that there are no ids.  If computing
@@ -168,32 +203,8 @@
 	$$.val = $1;
 	$$.ids = NULL;
 }
-| ID
-{
-	if (!compute_ids) {
-		/*
-		 * Compute the event's value from ID. If the ID isn't known then
-		 * it isn't used to compute the formula so set to NAN.
-		 */
-		struct expr_id_data *data;
-
-		$$.val = NAN;
-		if (expr__resolve_id(ctx, $1, &data) == 0)
-			$$.val = expr_id_data__value(data);
-
-		$$.ids = NULL;
-		free($1);
-	} else {
-		/*
-		 * Set the value to BOTTOM to show that any value is possible
-		 * when the event is computed. Create a set of just the ID.
-		 */
-		$$.val = BOTTOM;
-		$$.ids = ids__new();
-		if (!$$.ids || ids__insert($$.ids, $1))
-			YYABORT;
-	}
-}
+| ID				{ $$ = handle_id(ctx, $1, compute_ids, /*source_count=*/false); }
+| SOURCE_COUNT '(' ID ')'	{ $$ = handle_id(ctx, $3, compute_ids, /*source_count=*/true); }
 | expr '|' expr { BINARY_LONG_OP($$, |, $1, $3); }
 | expr '&' expr { BINARY_LONG_OP($$, &, $1, $3); }
 | expr '^' expr { BINARY_LONG_OP($$, ^, $1, $3); }
@@ -280,9 +291,9 @@
 		$$ = union_expr($3, $5);
 	}
 }
-| SMT_ON
+| LITERAL
 {
-	$$.val = smt_on() > 0 ? 1.0 : 0.0;
+	$$.val = $1;
 	$$.ids = NULL;
 }
 ;
diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index 56511db..fda8d14 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -583,21 +583,21 @@ static int write_cpu_topology(struct feat_fd *ff,
 	if (!tp)
 		return -1;
 
-	ret = do_write(ff, &tp->core_sib, sizeof(tp->core_sib));
+	ret = do_write(ff, &tp->package_cpus_lists, sizeof(tp->package_cpus_lists));
 	if (ret < 0)
 		goto done;
 
-	for (i = 0; i < tp->core_sib; i++) {
-		ret = do_write_string(ff, tp->core_siblings[i]);
+	for (i = 0; i < tp->package_cpus_lists; i++) {
+		ret = do_write_string(ff, tp->package_cpus_list[i]);
 		if (ret < 0)
 			goto done;
 	}
-	ret = do_write(ff, &tp->thread_sib, sizeof(tp->thread_sib));
+	ret = do_write(ff, &tp->core_cpus_lists, sizeof(tp->core_cpus_lists));
 	if (ret < 0)
 		goto done;
 
-	for (i = 0; i < tp->thread_sib; i++) {
-		ret = do_write_string(ff, tp->thread_siblings[i]);
+	for (i = 0; i < tp->core_cpus_lists; i++) {
+		ret = do_write_string(ff, tp->core_cpus_list[i]);
 		if (ret < 0)
 			break;
 	}
@@ -617,15 +617,15 @@ static int write_cpu_topology(struct feat_fd *ff,
 			return ret;
 	}
 
-	if (!tp->die_sib)
+	if (!tp->die_cpus_lists)
 		goto done;
 
-	ret = do_write(ff, &tp->die_sib, sizeof(tp->die_sib));
+	ret = do_write(ff, &tp->die_cpus_lists, sizeof(tp->die_cpus_lists));
 	if (ret < 0)
 		goto done;
 
-	for (i = 0; i < tp->die_sib; i++) {
-		ret = do_write_string(ff, tp->die_siblings[i]);
+	for (i = 0; i < tp->die_cpus_lists; i++) {
+		ret = do_write_string(ff, tp->die_cpus_list[i]);
 		if (ret < 0)
 			goto done;
 	}
diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c
index e4fb02b..5c7308e 100644
--- a/tools/perf/util/stat-shadow.c
+++ b/tools/perf/util/stat-shadow.c
@@ -829,10 +829,12 @@ static int prepare_metric(struct evsel **metric_events,
 		struct saved_value *v;
 		struct stats *stats;
 		u64 metric_total = 0;
+		int source_count;
 
 		if (!strcmp(metric_events[i]->name, "duration_time")) {
 			stats = &walltime_nsecs_stats;
 			scale = 1e-9;
+			source_count = 1;
 		} else {
 			v = saved_value_lookup(metric_events[i], cpu, false,
 					       STAT_NONE, 0, st,
@@ -841,6 +843,7 @@ static int prepare_metric(struct evsel **metric_events,
 				break;
 			stats = &v->stats;
 			scale = 1.0;
+			source_count = evsel__source_count(metric_events[i]);
 
 			if (v->metric_other)
 				metric_total = v->metric_total;
@@ -849,7 +852,9 @@ static int prepare_metric(struct evsel **metric_events,
 		if (!n)
 			return -ENOMEM;
 
-		expr__add_id_val(pctx, n, metric_total ? : avg_stats(stats) * scale);
+		expr__add_id_val_source_count(pctx, n,
+					metric_total ? : avg_stats(stats) * scale,
+					source_count);
 	}
 
 	for (j = 0; metric_refs && metric_refs[j].metric_name; j++) {
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index aa1b7c1..b2ed314 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -274,7 +274,7 @@ struct symbol *symbol__new(u64 start, u64 len, u8 binding, u8 type, const char *
 	if (symbol_conf.priv_size) {
 		if (symbol_conf.init_annotation) {
 			struct annotation *notes = (void *)sym;
-			pthread_mutex_init(&notes->lock, NULL);
+			annotation__init(notes);
 		}
 		sym = ((void *)sym) + symbol_conf.priv_size;
 	}
@@ -294,6 +294,13 @@ struct symbol *symbol__new(u64 start, u64 len, u8 binding, u8 type, const char *
 
 void symbol__delete(struct symbol *sym)
 {
+	if (symbol_conf.priv_size) {
+		if (symbol_conf.init_annotation) {
+			struct annotation *notes = symbol__annotation(sym);
+
+			annotation__exit(notes);
+		}
+	}
 	free(((void *)sym) - symbol_conf.priv_size);
 }
 
diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h
index 1661966..fbf866d 100644
--- a/tools/perf/util/symbol.h
+++ b/tools/perf/util/symbol.h
@@ -40,22 +40,33 @@ Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep,
 			     GElf_Shdr *shp, const char *name, size_t *idx);
 #endif
 
-/** struct symbol - symtab entry
- *
- * @ignore - resolvable but tools ignore it (e.g. idle routines)
+/**
+ * A symtab entry. When allocated this may be preceded by an annotation (see
+ * symbol__annotation), a browser_index (see symbol__browser_index) and rb_node
+ * to sort by name (see struct symbol_name_rb_node).
  */
 struct symbol {
 	struct rb_node	rb_node;
+	/** Range of symbol [start, end). */
 	u64		start;
 	u64		end;
+	/** Length of the string name. */
 	u16		namelen;
+	/** ELF symbol type as defined for st_info. E.g STT_OBJECT or STT_FUNC. */
 	u8		type:4;
+	/** ELF binding type as defined for st_info. E.g. STB_WEAK or STB_GLOBAL. */
 	u8		binding:4;
+	/** Set true for kernel symbols of idle routines. */
 	u8		idle:1;
+	/** Resolvable but tools ignore it (e.g. idle routines). */
 	u8		ignore:1;
+	/** Symbol for an inlined function. */
 	u8		inlined:1;
+	/** Has symbol__annotate2 been performed. */
+	u8		annotate2:1;
+	/** Architecture specific. Unused except on PPC where it holds st_other. */
 	u8		arch_sym;
-	bool		annotate2;
+	/** The name of length namelen associated with the symbol. */
 	char		name[];
 };