sched: Fix hrtimer_cancel()/rq->lock deadlock
authorBen Segall <bsegall@google.com>
Wed, 16 Oct 2013 18:16:22 +0000 (11:16 -0700)
committerIngo Molnar <mingo@kernel.org>
Tue, 29 Oct 2013 11:02:21 +0000 (12:02 +0100)
commit927b54fccbf04207ec92f669dce6806848cbec7d
tree4ee5a37f4796df9883c6beb9b7a91ca8c8534049
parentdb06e78cc13d70f10877e0557becc88ab3ad2be8
sched: Fix hrtimer_cancel()/rq->lock deadlock

__start_cfs_bandwidth calls hrtimer_cancel while holding rq->lock,
waiting for the hrtimer to finish. However, if sched_cfs_period_timer
runs for another loop iteration, the hrtimer can attempt to take
rq->lock, resulting in deadlock.

Fix this by ensuring that cfs_b->timer_active is cleared only if the
_latest_ call to do_sched_cfs_period_timer is returning as idle. Then
__start_cfs_bandwidth can just call hrtimer_try_to_cancel and wait for
that to succeed or timer_active == 1.

Signed-off-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Cc: pjt@google.com
Link: http://lkml.kernel.org/r/20131016181622.22647.16643.stgit@sword-of-the-dawn.mtv.corp.google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/sched/fair.c