sched/deadline: Fix races between rt_mutex_setprio() and dl_task_timer()
authorJuri Lelli <juri.lelli@arm.com>
Fri, 24 Oct 2014 09:16:38 +0000 (10:16 +0100)
committerIngo Molnar <mingo@kernel.org>
Tue, 28 Oct 2014 09:46:01 +0000 (10:46 +0100)
commitaee38ea95419c818dfdde52b115aeffe9cbb259b
tree438cee90383ef1e33e33372f3a70b77a524a2e9b
parent64be6f1f5f710f5995d41caf8a1767fe6d2b5a87
sched/deadline: Fix races between rt_mutex_setprio() and dl_task_timer()

dl_task_timer() is racy against several paths. Daniel noticed that
the replenishment timer may experience a race condition against an
enqueue_dl_entity() called from rt_mutex_setprio(). With his own
words:

 rt_mutex_setprio() resets p->dl.dl_throttled. So the pattern is:
 start_dl_timer() throttled = 1, rt_mutex_setprio() throlled = 0,
 sched_switch() -> enqueue_task(), dl_task_timer-> enqueue_task()
 throttled is 0

=> BUG_ON(on_dl_rq(dl_se)) fires as the scheduling entity is already
enqueued on the -deadline runqueue.

As we do for the other races, we just bail out in the replenishment
timer code.

Reported-by: Daniel Wagner <daniel.wagner@bmw-carit.de>
Tested-by: Daniel Wagner <daniel.wagner@bmw-carit.de>
Signed-off-by: Juri Lelli <juri.lelli@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: vincent@legout.info
Cc: Dario Faggioli <raistlin@linux.it>
Cc: Michael Trimarchi <michael@amarulasolutions.com>
Cc: Fabio Checconi <fchecconi@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1414142198-18552-5-git-send-email-juri.lelli@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/sched/deadline.c