perf/core: Fix __perf_read_group_add() locking

Event timestamps are serialized using ctx->lock, make sure to hold it
over reading all values.

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>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

kernel/events/core.c

@@ -4453,6 +4453,8 @@ static int __perf_read_group_add(struct perf_event *leader,
 	if (ret)
 		return ret;
 
+	raw_spin_lock_irqsave(&ctx->lock, flags);
+
 	/*
 	 * Since we co-schedule groups, {enabled,running} times of siblings
 	 * will be identical to those of the leader, so we only publish one
@@ -4475,8 +4477,6 @@ static int __perf_read_group_add(struct perf_event *leader,
 	if (read_format & PERF_FORMAT_ID)
 		values[n++] = primary_event_id(leader);
 
-	raw_spin_lock_irqsave(&ctx->lock, flags);
-
 	list_for_each_entry(sub, &leader->sibling_list, group_entry) {
 		values[n++] += perf_event_count(sub);
 		if (read_format & PERF_FORMAT_ID)