futex: Avoid freeing an active timer
authorThomas Gleixner <tglx@linutronix.de>
Mon, 10 Apr 2017 16:03:36 +0000 (18:03 +0200)
committerIngo Molnar <mingo@kernel.org>
Fri, 14 Apr 2017 08:29:53 +0000 (10:29 +0200)
Alexander reported a hrtimer debug_object splat:

  ODEBUG: free active (active state 0) object type: hrtimer hint: hrtimer_wakeup (kernel/time/hrtimer.c:1423)

  debug_object_free (lib/debugobjects.c:603)
  destroy_hrtimer_on_stack (kernel/time/hrtimer.c:427)
  futex_lock_pi (kernel/futex.c:2740)
  do_futex (kernel/futex.c:3399)
  SyS_futex (kernel/futex.c:3447 kernel/futex.c:3415)
  do_syscall_64 (arch/x86/entry/common.c:284)
  entry_SYSCALL64_slow_path (arch/x86/entry/entry_64.S:249)

Which was caused by commit:

  cfafcd117da0 ("futex: Rework futex_lock_pi() to use rt_mutex_*_proxy_lock()")

... losing the hrtimer_cancel() in the shuffle. Where previously the
hrtimer_cancel() was done by rt_mutex_slowlock() we now need to do it
manually.

Reported-by: Alexander Levin <alexander.levin@verizon.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Fixes: cfafcd117da0 ("futex: Rework futex_lock_pi() to use rt_mutex_*_proxy_lock()")
Link: http://lkml.kernel.org/r/alpine.DEB.2.20.1704101802370.2906@nanos
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/futex.c

index c3eebcd..7ac1676 100644 (file)
@@ -2736,8 +2736,10 @@ out_unlock_put_key:
 out_put_key:
        put_futex_key(&q.key);
 out:
-       if (to)
+       if (to) {
+               hrtimer_cancel(&to->timer);
                destroy_hrtimer_on_stack(&to->timer);
+       }
        return ret != -EINTR ? ret : -ERESTARTNOINTR;
 
 uaddr_faulted: