perf/core: Set cgroup in CPU contexts for new cgroup events

There's a perf stat bug easy to observer on a machine with only one cgroup:

  $ perf stat -e cycles -I 1000 -C 0 -G /
  #          time             counts unit events
      1.000161699      <not counted>      cycles                    /
      2.000355591      <not counted>      cycles                    /
      3.000565154      <not counted>      cycles                    /
      4.000951350      <not counted>      cycles                    /

We'd expect some output there.

The underlying problem is that there is an optimization in
perf_cgroup_sched_{in,out}() that skips the switch of cgroup events
if the old and new cgroups in a task switch are the same.

This optimization interacts with the current code in two ways
that cause a CPU context's cgroup (cpuctx->cgrp) to be NULL even if a
cgroup event matches the current task. These are:

  1. On creation of the first cgroup event in a CPU: In current code,
  cpuctx->cpu is only set in perf_cgroup_sched_in, but due to the
  aforesaid optimization, perf_cgroup_sched_in will run until the next
  cgroup switches in that CPU. This may happen late or never happen,
  depending on system's number of cgroups, CPU load, etc.

  2. On deletion of the last cgroup event in a cpuctx: In list_del_event,
  cpuctx->cgrp is set NULL. Any new cgroup event will not be sched in
  because cpuctx->cgrp == NULL until a cgroup switch occurs and
  perf_cgroup_sched_in is executed (updating cpuctx->cgrp).

This patch fixes both problems by setting cpuctx->cgrp in list_add_event,
mirroring what list_del_event does when removing a cgroup event from CPU
context, as introduced in:

  commit 68cacd29167b ("perf_events: Fix stale ->cgrp pointer in update_cgrp_time_from_cpuctx()")

With this patch, cpuctx->cgrp is always set/clear when installing/removing
the first/last cgroup event in/from the CPU context. With cpuctx->cgrp
correctly set, event_filter_match works as intended when events are
sched in/out.

After the fix, the output is as expected:

  $ perf stat -e cycles -I 1000 -a -G /
  #         time             counts unit events
     1.004699159          627342882      cycles                    /
     2.007397156          615272690      cycles                    /
     3.010019057          616726074      cycles                    /

Signed-off-by: David Carrillo-Cisneros <davidcc@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul Turner <pjt@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vegard Nossum <vegard.nossum@gmail.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Link: http://lkml.kernel.org/r/1470124092-113192-1-git-send-email-davidcc@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

1 files changed, 36 insertions, 18 deletions

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 87d02b8cb87e..1903b8f3a705 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -843,6 +843,32 @@ perf_cgroup_mark_enabled(struct perf_event *event,
 		}
 	}
 }
+
+/*
+ * Update cpuctx->cgrp so that it is set when first cgroup event is added and
+ * cleared when last cgroup event is removed.
+ */
+static inline void
+list_update_cgroup_event(struct perf_event *event,
+			 struct perf_event_context *ctx, bool add)
+{
+	struct perf_cpu_context *cpuctx;
+
+	if (!is_cgroup_event(event))
+		return;
+
+	if (add && ctx->nr_cgroups++)
+		return;
+	else if (!add && --ctx->nr_cgroups)
+		return;
+	/*
+	 * Because cgroup events are always per-cpu events,
+	 * this will always be called from the right CPU.
+	 */
+	cpuctx = __get_cpu_context(ctx);
+	cpuctx->cgrp = add ? event->cgrp : NULL;
+}
+
 #else /* !CONFIG_CGROUP_PERF */
 
 static inline bool
@@ -920,6 +946,13 @@ perf_cgroup_mark_enabled(struct perf_event *event,
 			 struct perf_event_context *ctx)
 {
 }
+
+static inline void
+list_update_cgroup_event(struct perf_event *event,
+			 struct perf_event_context *ctx, bool add)
+{
+}
+
 #endif
 
 /*
@@ -1392,6 +1425,7 @@ ctx_group_list(struct perf_event *event, struct perf_event_context *ctx)
 static void
 list_add_event(struct perf_event *event, struct perf_event_context *ctx)
 {
+
 	lockdep_assert_held(&ctx->lock);
 
 	WARN_ON_ONCE(event->attach_state & PERF_ATTACH_CONTEXT);
@@ -1412,8 +1446,7 @@ list_add_event(struct perf_event *event, struct perf_event_context *ctx)
 		list_add_tail(&event->group_entry, list);
 	}
 
-	if (is_cgroup_event(event))
-		ctx->nr_cgroups++;
+	list_update_cgroup_event(event, ctx, true);
 
 	list_add_rcu(&event->event_entry, &ctx->event_list);
 	ctx->nr_events++;
@@ -1581,8 +1614,6 @@ static void perf_group_attach(struct perf_event *event)
 static void
 list_del_event(struct perf_event *event, struct perf_event_context *ctx)
 {
-	struct perf_cpu_context *cpuctx;
-
 	WARN_ON_ONCE(event->ctx != ctx);
 	lockdep_assert_held(&ctx->lock);
 
@@ -1594,20 +1625,7 @@ list_del_event(struct perf_event *event, struct perf_event_context *ctx)
 
 	event->attach_state &= ~PERF_ATTACH_CONTEXT;
 
-	if (is_cgroup_event(event)) {
-		ctx->nr_cgroups--;
-		/*
-		 * Because cgroup events are always per-cpu events, this will
-		 * always be called from the right CPU.
-		 */
-		cpuctx = __get_cpu_context(ctx);
-		/*
-		 * If there are no more cgroup events then clear cgrp to avoid
-		 * stale pointer in update_cgrp_time_from_cpuctx().
-		 */
-		if (!ctx->nr_cgroups)
-			cpuctx->cgrp = NULL;
-	}
+	list_update_cgroup_event(event, ctx, false);
 
 	ctx->nr_events--;
 	if (event->attr.inherit_stat)