Skip to contentSkip to conversation

perf: Fix hotplug splat

tip:perf/core tip-bot for Peter Zijlstra 1 participant 1 message Feb 27, 2014

Subsystems (2)
Stable branch perf events
lore
Share & tools
lore

tip-bot for Peter Zijlstra

author merge bot
#0 lore raw
Commit-ID:  e3703f8cdfcf39c25c4338c3ad8e68891cca3731
Gitweb:     http://git.kernel.org/tip/e3703f8cdfcf39c25c4338c3ad8e68891cca3731
Author:     Peter Zijlstra <[email protected]>
AuthorDate: Mon, 24 Feb 2014 12:06:12 +0100
Committer:  Ingo Molnar <[email protected]>
CommitDate: Thu, 27 Feb 2014 12:38:03 +0100

perf: Fix hotplug splat

Drew Richardson reported that he could make the kernel go *boom* when hotplugging while having perf events active.

It turned out that when you have a group event, the code in __perf_event_exit_context() fails to remove the group siblings from the context.

We then proceed with destroying and freeing the event, and when you re-plug the CPU and try and add another event to that CPU, things go *boom* because you've still got dead entries there.

Reported-by Drew Richardson <[email protected]>
Signed-off-by Peter Zijlstra <[email protected]>
Cc Will Deacon <[email protected]>
Cc <[email protected]>
Link http://lkml.kernel.org/n/[email protected] open here
Signed-off-by Ingo Molnar <[email protected]>
---
 kernel/events/core.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
kernel/events/core.c +6 -6
index 56003c6..fa0b2d4 100644
@@ -7856,14 +7856,14 @@ static void perf_pmu_rotate_stop(struct pmu *pmu)
 static void __perf_event_exit_context(void *__info)
 {
 	struct perf_event_context *ctx = __info;
-	struct perf_event *event, *tmp;
+	struct perf_event *event;
 
 	perf_pmu_rotate_stop(ctx->pmu);
 
-	list_for_each_entry_safe(event, tmp, &ctx->pinned_groups, group_entry)
-		__perf_remove_from_context(event);
-	list_for_each_entry_safe(event, tmp, &ctx->flexible_groups, group_entry)
+	rcu_read_lock();
+	list_for_each_entry_rcu(event, &ctx->event_list, event_entry)
 		__perf_remove_from_context(event);
+	rcu_read_unlock();
 }
 
 static void perf_event_exit_cpu_context(int cpu)

@@ -7887,11 +7887,11 @@ static void perf_event_exit_cpu(int cpu)
 {
 	struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu);
 
+	perf_event_exit_cpu_context(cpu);
+
 	mutex_lock(&swhash->hlist_mutex);
 	swevent_hlist_release(swhash);
 	mutex_unlock(&swhash->hlist_mutex);
-
-	perf_event_exit_cpu_context(cpu);
 }
 #else
 static inline void perf_event_exit_cpu(int cpu) { }