diff options
author | Peter Zijlstra <a.p.zijlstra@chello.nl> | 2011-10-17 05:50:30 -0400 |
---|---|---|
committer | Thomas Gleixner <tglx@linutronix.de> | 2011-10-18 05:36:59 -0400 |
commit | bcd5cff7216f9b2de0a148cc355eac199dc6f1cf (patch) | |
tree | 384674b2b0e16e489f591148982046bf2d25608b /kernel/posix-cpu-timers.c | |
parent | 899e3ee404961a90b828ad527573aaaac39f0ab1 (diff) |
cputimer: Cure lock inversion
There's a lock inversion between the cputimer->lock and rq->lock;
notably the two callchains involved are:
update_rlimit_cpu()
sighand->siglock
set_process_cpu_timer()
cpu_timer_sample_group()
thread_group_cputimer()
cputimer->lock
thread_group_cputime()
task_sched_runtime()
->pi_lock
rq->lock
scheduler_tick()
rq->lock
task_tick_fair()
update_curr()
account_group_exec()
cputimer->lock
Where the first one is enabling a CLOCK_PROCESS_CPUTIME_ID timer, and
the second one is keeping up-to-date.
This problem was introduced by e8abccb7193 ("posix-cpu-timers: Cure
SMP accounting oddities").
Cure the problem by removing the cputimer->lock and rq->lock nesting,
this leaves concurrent enablers doing duplicate work, but the time
wasted should be on the same order otherwise wasted spinning on the
lock and the greater-than assignment filter should ensure we preserve
monotonicity.
Reported-by: Dave Jones <davej@redhat.com>
Reported-by: Simon Kirby <sim@hostway.ca>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: stable@kernel.org
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Link: http://lkml.kernel.org/r/1318928713.21167.4.camel@twins
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Diffstat (limited to 'kernel/posix-cpu-timers.c')
-rw-r--r-- | kernel/posix-cpu-timers.c | 7 |
1 files changed, 4 insertions, 3 deletions
diff --git a/kernel/posix-cpu-timers.c b/kernel/posix-cpu-timers.c index c8008dd58ef2..640ded8f5c48 100644 --- a/kernel/posix-cpu-timers.c +++ b/kernel/posix-cpu-timers.c | |||
@@ -274,9 +274,7 @@ void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times) | |||
274 | struct task_cputime sum; | 274 | struct task_cputime sum; |
275 | unsigned long flags; | 275 | unsigned long flags; |
276 | 276 | ||
277 | spin_lock_irqsave(&cputimer->lock, flags); | ||
278 | if (!cputimer->running) { | 277 | if (!cputimer->running) { |
279 | cputimer->running = 1; | ||
280 | /* | 278 | /* |
281 | * The POSIX timer interface allows for absolute time expiry | 279 | * The POSIX timer interface allows for absolute time expiry |
282 | * values through the TIMER_ABSTIME flag, therefore we have | 280 | * values through the TIMER_ABSTIME flag, therefore we have |
@@ -284,8 +282,11 @@ void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times) | |||
284 | * it. | 282 | * it. |
285 | */ | 283 | */ |
286 | thread_group_cputime(tsk, &sum); | 284 | thread_group_cputime(tsk, &sum); |
285 | spin_lock_irqsave(&cputimer->lock, flags); | ||
286 | cputimer->running = 1; | ||
287 | update_gt_cputime(&cputimer->cputime, &sum); | 287 | update_gt_cputime(&cputimer->cputime, &sum); |
288 | } | 288 | } else |
289 | spin_lock_irqsave(&cputimer->lock, flags); | ||
289 | *times = cputimer->cputime; | 290 | *times = cputimer->cputime; |
290 | spin_unlock_irqrestore(&cputimer->lock, flags); | 291 | spin_unlock_irqrestore(&cputimer->lock, flags); |
291 | } | 292 | } |