perf/core: Explain perf_sched_mutex

To clarify why atomic_inc_return(&perf_sched_events) is not sufficient and
a mutex is needed to order static branch enabling vs the atomic counter
increment, this adds a comment with a short explanation.

Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20170829140103.6563-1-alexander.shishkin@linux.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

1 files changed, 5 insertions, 0 deletions

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 6bc21e202ae4..5ee62714f9a6 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -9394,6 +9394,11 @@ static void account_event(struct perf_event *event)
                 inc = true;
 
         if (inc) {
+                /*
+                 * We need the mutex here because static_branch_enable()
+                 * must complete *before* the perf_sched_count increment
+                 * becomes visible.
+                 */
                 if (atomic_inc_not_zero(&perf_sched_count))
                         goto enabled;