perf: Fix a race condition in perf_remove_from_context()
authorCong Wang <cwang@twopensource.com>
Tue, 2 Sep 2014 22:27:20 +0000 (15:27 -0700)
committerBen Hutchings <ben@decadent.org.uk>
Wed, 5 Nov 2014 20:27:41 +0000 (20:27 +0000)
commit189613e5ef510bc6f58feeb023b57d71c1970a4c
tree64858770e2b9658686654de978eaf553d996b882
parent126531ff8a7a8bec5517b6baaada5b2e7560fc89
perf: Fix a race condition in perf_remove_from_context()

commit 3577af70a2ce4853d58e57d832e687d739281479 upstream.

We saw a kernel soft lockup in perf_remove_from_context(),
it looks like the `perf` process, when exiting, could not go
out of the retry loop. Meanwhile, the target process was forking
a child. So either the target process should execute the smp
function call to deactive the event (if it was running) or it should
do a context switch which deactives the event.

It seems we optimize out a context switch in perf_event_context_sched_out(),
and what's more important, we still test an obsolete task pointer when
retrying, so no one actually would deactive that event in this situation.
Fix it directly by reloading the task pointer in perf_remove_from_context().

This should cure the above soft lockup.

Signed-off-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1409696840-843-1-git-send-email-xiyou.wangcong@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
kernel/events/core.c