diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index d096b04..9d24870 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -1703,10 +1703,8 @@
 
 		perf_sample_data_init(&data, 0, event->hw.last_period);
 
-		if (has_branch_stack(event)) {
-			data.br_stack = &cpuc->lbr_stack;
-			data.sample_flags |= PERF_SAMPLE_BRANCH_STACK;
-		}
+		if (has_branch_stack(event))
+			perf_sample_save_brstack(&data, event, &cpuc->lbr_stack);
 
 		if (perf_event_overflow(event, &data, regs))
 			x86_pmu_stop(event, 0);
diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c
index a2e566e..df88576 100644
--- a/arch/x86/events/intel/ds.c
+++ b/arch/x86/events/intel/ds.c
@@ -1229,12 +1229,14 @@
 		  struct perf_event *event, bool add)
 {
 	struct pmu *pmu = event->pmu;
+
 	/*
 	 * Make sure we get updated with the first PEBS
 	 * event. It will trigger also during removal, but
 	 * that does not hurt:
 	 */
-	bool update = cpuc->n_pebs == 1;
+	if (cpuc->n_pebs == 1)
+		cpuc->pebs_data_cfg = PEBS_UPDATE_DS_SW;
 
 	if (needed_cb != pebs_needs_sched_cb(cpuc)) {
 		if (!needed_cb)
@@ -1242,7 +1244,7 @@
 		else
 			perf_sched_cb_dec(pmu);
 
-		update = true;
+		cpuc->pebs_data_cfg |= PEBS_UPDATE_DS_SW;
 	}
 
 	/*
@@ -1252,24 +1254,13 @@
 	if (x86_pmu.intel_cap.pebs_baseline && add) {
 		u64 pebs_data_cfg;
 
-		/* Clear pebs_data_cfg and pebs_record_size for first PEBS. */
-		if (cpuc->n_pebs == 1) {
-			cpuc->pebs_data_cfg = 0;
-			cpuc->pebs_record_size = sizeof(struct pebs_basic);
-		}
-
 		pebs_data_cfg = pebs_update_adaptive_cfg(event);
-
-		/* Update pebs_record_size if new event requires more data. */
-		if (pebs_data_cfg & ~cpuc->pebs_data_cfg) {
-			cpuc->pebs_data_cfg |= pebs_data_cfg;
-			adaptive_pebs_record_size_update();
-			update = true;
-		}
+		/*
+		 * Be sure to update the thresholds when we change the record.
+		 */
+		if (pebs_data_cfg & ~cpuc->pebs_data_cfg)
+			cpuc->pebs_data_cfg |= pebs_data_cfg | PEBS_UPDATE_DS_SW;
 	}
-
-	if (update)
-		pebs_update_threshold(cpuc);
 }
 
 void intel_pmu_pebs_add(struct perf_event *event)
@@ -1326,9 +1317,17 @@
 	wrmsrl(base + idx, value);
 }
 
+static inline void intel_pmu_drain_large_pebs(struct cpu_hw_events *cpuc)
+{
+	if (cpuc->n_pebs == cpuc->n_large_pebs &&
+	    cpuc->n_pebs != cpuc->n_pebs_via_pt)
+		intel_pmu_drain_pebs_buffer();
+}
+
 void intel_pmu_pebs_enable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
+	u64 pebs_data_cfg = cpuc->pebs_data_cfg & ~PEBS_UPDATE_DS_SW;
 	struct hw_perf_event *hwc = &event->hw;
 	struct debug_store *ds = cpuc->ds;
 	unsigned int idx = hwc->idx;
@@ -1344,11 +1343,22 @@
 
 	if (x86_pmu.intel_cap.pebs_baseline) {
 		hwc->config |= ICL_EVENTSEL_ADAPTIVE;
-		if (cpuc->pebs_data_cfg != cpuc->active_pebs_data_cfg) {
-			wrmsrl(MSR_PEBS_DATA_CFG, cpuc->pebs_data_cfg);
-			cpuc->active_pebs_data_cfg = cpuc->pebs_data_cfg;
+		if (pebs_data_cfg != cpuc->active_pebs_data_cfg) {
+			/*
+			 * drain_pebs() assumes uniform record size;
+			 * hence we need to drain when changing said
+			 * size.
+			 */
+			intel_pmu_drain_large_pebs(cpuc);
+			adaptive_pebs_record_size_update();
+			wrmsrl(MSR_PEBS_DATA_CFG, pebs_data_cfg);
+			cpuc->active_pebs_data_cfg = pebs_data_cfg;
 		}
 	}
+	if (cpuc->pebs_data_cfg & PEBS_UPDATE_DS_SW) {
+		cpuc->pebs_data_cfg = pebs_data_cfg;
+		pebs_update_threshold(cpuc);
+	}
 
 	if (idx >= INTEL_PMC_IDX_FIXED) {
 		if (x86_pmu.intel_cap.pebs_format < 5)
@@ -1391,9 +1401,7 @@
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
 
-	if (cpuc->n_pebs == cpuc->n_large_pebs &&
-	    cpuc->n_pebs != cpuc->n_pebs_via_pt)
-		intel_pmu_drain_pebs_buffer();
+	intel_pmu_drain_large_pebs(cpuc);
 
 	cpuc->pebs_enabled &= ~(1ULL << hwc->idx);
 
diff --git a/arch/x86/include/asm/perf_event.h b/arch/x86/include/asm/perf_event.h
index 8fc15ed..abf09882 100644
--- a/arch/x86/include/asm/perf_event.h
+++ b/arch/x86/include/asm/perf_event.h
@@ -121,6 +121,9 @@
 #define PEBS_DATACFG_LBRS	BIT_ULL(3)
 #define PEBS_DATACFG_LBR_SHIFT	24
 
+/* Steal the highest bit of pebs_data_cfg for SW usage */
+#define PEBS_UPDATE_DS_SW	BIT_ULL(63)
+
 /*
  * Intel "Architectural Performance Monitoring" CPUID
  * detection/enumeration details:
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 68baa81..db016e4 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -10150,8 +10150,20 @@
 	perf_trace_buf_update(record, event_type);
 
 	hlist_for_each_entry_rcu(event, head, hlist_entry) {
-		if (perf_tp_event_match(event, &data, regs))
+		if (perf_tp_event_match(event, &data, regs)) {
 			perf_swevent_event(event, count, &data, regs);
+
+			/*
+			 * Here use the same on-stack perf_sample_data,
+			 * some members in data are event-specific and
+			 * need to be re-computed for different sweveents.
+			 * Re-initialize data->sample_flags safely to avoid
+			 * the problem that next event skips preparing data
+			 * because data->sample_flags is set.
+			 */
+			perf_sample_data_init(&data, 0, 0);
+			perf_sample_save_raw_data(&data, &raw);
+		}
 	}
 
 	/*