diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 245bafd..c816075 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -944,8 +944,10 @@
 
 extern int perf_num_counters(void);
 extern const char *perf_pmu_name(void);
-extern void __perf_event_task_sched_in(struct task_struct *task);
-extern void __perf_event_task_sched_out(struct task_struct *task, struct task_struct *next);
+extern void __perf_event_task_sched_in(struct task_struct *prev,
+				       struct task_struct *task);
+extern void __perf_event_task_sched_out(struct task_struct *prev,
+					struct task_struct *next);
 extern int perf_event_init_task(struct task_struct *child);
 extern void perf_event_exit_task(struct task_struct *child);
 extern void perf_event_free_task(struct task_struct *task);
@@ -1059,17 +1061,20 @@
 
 extern struct jump_label_key perf_sched_events;
 
-static inline void perf_event_task_sched_in(struct task_struct *task)
+static inline void perf_event_task_sched_in(struct task_struct *prev,
+					    struct task_struct *task)
 {
 	if (static_branch(&perf_sched_events))
-		__perf_event_task_sched_in(task);
+		__perf_event_task_sched_in(prev, task);
 }
 
-static inline void perf_event_task_sched_out(struct task_struct *task, struct task_struct *next)
+static inline void perf_event_task_sched_out(struct task_struct *prev,
+					     struct task_struct *next)
 {
 	perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, NULL, 0);
 
-	__perf_event_task_sched_out(task, next);
+	if (static_branch(&perf_sched_events))
+		__perf_event_task_sched_out(prev, next);
 }
 
 extern void perf_event_mmap(struct vm_area_struct *vma);
@@ -1139,10 +1144,11 @@
 extern void perf_event_task_tick(void);
 #else
 static inline void
-perf_event_task_sched_in(struct task_struct *task)			{ }
+perf_event_task_sched_in(struct task_struct *prev,
+			 struct task_struct *task)			{ }
 static inline void
-perf_event_task_sched_out(struct task_struct *task,
-			    struct task_struct *next)			{ }
+perf_event_task_sched_out(struct task_struct *prev,
+			  struct task_struct *next)			{ }
 static inline int perf_event_init_task(struct task_struct *child)	{ return 0; }
 static inline void perf_event_exit_task(struct task_struct *child)	{ }
 static inline void perf_event_free_task(struct task_struct *task)	{ }
diff --git a/kernel/events/core.c b/kernel/events/core.c
index b8785e2..45847fb 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -399,14 +399,54 @@
 	local_irq_restore(flags);
 }
 
-static inline void perf_cgroup_sched_out(struct task_struct *task)
+static inline void perf_cgroup_sched_out(struct task_struct *task,
+					 struct task_struct *next)
 {
-	perf_cgroup_switch(task, PERF_CGROUP_SWOUT);
+	struct perf_cgroup *cgrp1;
+	struct perf_cgroup *cgrp2 = NULL;
+
+	/*
+	 * we come here when we know perf_cgroup_events > 0
+	 */
+	cgrp1 = perf_cgroup_from_task(task);
+
+	/*
+	 * next is NULL when called from perf_event_enable_on_exec()
+	 * that will systematically cause a cgroup_switch()
+	 */
+	if (next)
+		cgrp2 = perf_cgroup_from_task(next);
+
+	/*
+	 * only schedule out current cgroup events if we know
+	 * that we are switching to a different cgroup. Otherwise,
+	 * do no touch the cgroup events.
+	 */
+	if (cgrp1 != cgrp2)
+		perf_cgroup_switch(task, PERF_CGROUP_SWOUT);
 }
 
-static inline void perf_cgroup_sched_in(struct task_struct *task)
+static inline void perf_cgroup_sched_in(struct task_struct *prev,
+					struct task_struct *task)
 {
-	perf_cgroup_switch(task, PERF_CGROUP_SWIN);
+	struct perf_cgroup *cgrp1;
+	struct perf_cgroup *cgrp2 = NULL;
+
+	/*
+	 * we come here when we know perf_cgroup_events > 0
+	 */
+	cgrp1 = perf_cgroup_from_task(task);
+
+	/* prev can never be NULL */
+	cgrp2 = perf_cgroup_from_task(prev);
+
+	/*
+	 * only need to schedule in cgroup events if we are changing
+	 * cgroup during ctxsw. Cgroup events were not scheduled
+	 * out of ctxsw out if that was not the case.
+	 */
+	if (cgrp1 != cgrp2)
+		perf_cgroup_switch(task, PERF_CGROUP_SWIN);
 }
 
 static inline int perf_cgroup_connect(int fd, struct perf_event *event,
@@ -518,11 +558,13 @@
 {
 }
 
-static inline void perf_cgroup_sched_out(struct task_struct *task)
+static inline void perf_cgroup_sched_out(struct task_struct *task,
+					 struct task_struct *next)
 {
 }
 
-static inline void perf_cgroup_sched_in(struct task_struct *task)
+static inline void perf_cgroup_sched_in(struct task_struct *prev,
+					struct task_struct *task)
 {
 }
 
@@ -1988,7 +2030,7 @@
 	 * cgroup event are system-wide mode only
 	 */
 	if (atomic_read(&__get_cpu_var(perf_cgroup_events)))
-		perf_cgroup_sched_out(task);
+		perf_cgroup_sched_out(task, next);
 }
 
 static void task_ctx_sched_out(struct perf_event_context *ctx)
@@ -2153,7 +2195,8 @@
  * accessing the event control register. If a NMI hits, then it will
  * keep the event running.
  */
-void __perf_event_task_sched_in(struct task_struct *task)
+void __perf_event_task_sched_in(struct task_struct *prev,
+				struct task_struct *task)
 {
 	struct perf_event_context *ctx;
 	int ctxn;
@@ -2171,7 +2214,7 @@
 	 * cgroup event are system-wide mode only
 	 */
 	if (atomic_read(&__get_cpu_var(perf_cgroup_events)))
-		perf_cgroup_sched_in(task);
+		perf_cgroup_sched_in(prev, task);
 }
 
 static u64 perf_calculate_period(struct perf_event *event, u64 nsec, u64 count)
@@ -2427,7 +2470,7 @@
 	 * ctxswin cgroup events which are already scheduled
 	 * in.
 	 */
-	perf_cgroup_sched_out(current);
+	perf_cgroup_sched_out(current, NULL);
 
 	raw_spin_lock(&ctx->lock);
 	task_ctx_sched_out(ctx);
diff --git a/kernel/sched.c b/kernel/sched.c
index ccacdbd..0408cdc 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -3065,7 +3065,7 @@
 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
 	local_irq_disable();
 #endif /* __ARCH_WANT_INTERRUPTS_ON_CTXSW */
-	perf_event_task_sched_in(current);
+	perf_event_task_sched_in(prev, current);
 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
 	local_irq_enable();
 #endif /* __ARCH_WANT_INTERRUPTS_ON_CTXSW */