blob: 2b03a83f7a22c6ee032d4cd3219ad5fb8f5356d1 [file] [log] [blame]
From 3cd7ca15b8c31063269d6c1d05c912870a5b55de Mon Sep 17 00:00:00 2001
From: Chuck Lever <>
Date: Wed, 9 Oct 2019 13:07:38 -0400
Subject: [PATCH] xprtrdma: Close window between waking RPC senders and posting
commit 2ae50ad68cd79224198b525f7bd645c9da98b6ff upstream.
A recent clean up attempted to separate Receive handling and RPC
Reply processing, in the name of clean layering.
Unfortunately, we can't do this because the Receive Queue has to be
refilled _after_ the most recent credit update from the responder
is parsed from the transport header, but _before_ we wake up the
next RPC sender. That is right in the middle of
Usually this isn't a problem because current responder
implementations don't vary their credit grant. The one exception is
when a connection is established: the grant goes from one to a much
larger number on the first Receive. The requester MUST post enough
Receives right then so that any outstanding requests can be sent
without risking RNR and connection loss.
Fixes: 6ceea36890a0 ("xprtrdma: Refactor Receive accounting")
Signed-off-by: Chuck Lever <>
Signed-off-by: Anna Schumaker <>
Signed-off-by: Paul Gortmaker <>
diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c
index 85115a2e2639..6d70cd45ef0b 100644
--- a/net/sunrpc/xprtrdma/rpc_rdma.c
+++ b/net/sunrpc/xprtrdma/rpc_rdma.c
@@ -1365,6 +1365,7 @@ void rpcrdma_reply_handler(struct rpcrdma_rep *rep)
xprt->cwnd = credits << RPC_CWNDSHIFT;
+ rpcrdma_post_recvs(r_xprt, false);
req = rpcr_to_rdmar(rqst);
if (req->rl_reply) {
diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c
index 3f67c395845d..c07462946ecb 100644
--- a/net/sunrpc/xprtrdma/verbs.c
+++ b/net/sunrpc/xprtrdma/verbs.c
@@ -81,7 +81,6 @@ rpcrdma_regbuf_alloc(size_t size, enum dma_data_direction direction,
gfp_t flags);
static void rpcrdma_regbuf_dma_unmap(struct rpcrdma_regbuf *rb);
static void rpcrdma_regbuf_free(struct rpcrdma_regbuf *rb);
-static void rpcrdma_post_recvs(struct rpcrdma_xprt *r_xprt, bool temp);
/* Wait for outstanding transport work to finish. ib_drain_qp
* handles the drains in the wrong order for us, so open code
@@ -169,7 +168,6 @@ rpcrdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
wc->byte_len, DMA_FROM_DEVICE);
- rpcrdma_post_recvs(r_xprt, false);
@@ -1502,8 +1500,13 @@ rpcrdma_ep_post(struct rpcrdma_ia *ia,
return 0;
-static void
-rpcrdma_post_recvs(struct rpcrdma_xprt *r_xprt, bool temp)
+ * rpcrdma_post_recvs - Refill the Receive Queue
+ * @r_xprt: controlling transport instance
+ * @temp: mark Receive buffers to be deleted after use
+ *
+ */
+void rpcrdma_post_recvs(struct rpcrdma_xprt *r_xprt, bool temp)
struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
struct rpcrdma_ep *ep = &r_xprt->rx_ep;
diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
index d1e0749bcbc4..6554f89ebaa1 100644
--- a/net/sunrpc/xprtrdma/xprt_rdma.h
+++ b/net/sunrpc/xprtrdma/xprt_rdma.h
@@ -496,6 +496,7 @@ void rpcrdma_ep_disconnect(struct rpcrdma_ep *, struct rpcrdma_ia *);
int rpcrdma_ep_post(struct rpcrdma_ia *, struct rpcrdma_ep *,
struct rpcrdma_req *);
+void rpcrdma_post_recvs(struct rpcrdma_xprt *r_xprt, bool temp);
* Buffer calls - xprtrdma/verbs.c