releases/3.19.2/locking-rtmutex-avoid-a-null-pointer-dereference-on-deadlock.patch

   1 From 8d1e5a1a1ccf5ae9d8a5a0ee7960202ccb0c5429 Mon Sep 17 00:00:00 2001
   2 From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
   3 Date: Tue, 17 Feb 2015 16:43:43 +0100
   4 Subject: locking/rtmutex: Avoid a NULL pointer dereference on deadlock
   5
   6 From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
   7
   8 commit 8d1e5a1a1ccf5ae9d8a5a0ee7960202ccb0c5429 upstream.
   9
  10 With task_blocks_on_rt_mutex() returning early -EDEADLK we never
  11 add the waiter to the waitqueue. Later, we try to remove it via
  12 remove_waiter() and go boom in rt_mutex_top_waiter() because
  13 rb_entry() gives a NULL pointer.
  14
  15 ( Tested on v3.18-RT where rtmutex is used for regular mutex and I
  16   tried to get one twice in a row. )
  17
  18 Not sure when this started but I guess 397335f004f4 ("rtmutex: Fix
  19 deadlock detector for real") or commit 3d5c9340d194 ("rtmutex:
  20 Handle deadlock detection smarter").
  21
  22 Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
  23 Acked-by: Peter Zijlstra <peterz@infradead.org>
  24 Cc: Thomas Gleixner <tglx@linutronix.de>
  25 Link: http://lkml.kernel.org/r/1424187823-19600-1-git-send-email-bigeasy@linutronix.de
  26 Signed-off-by: Ingo Molnar <mingo@kernel.org>
  27 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
  28
  29 ---
  30  kernel/locking/rtmutex.c |    3 ++-
  31  1 file changed, 2 insertions(+), 1 deletion(-)
  32
  33 --- a/kernel/locking/rtmutex.c
  34 +++ b/kernel/locking/rtmutex.c
  35 @@ -1193,7 +1193,8 @@ rt_mutex_slowlock(struct rt_mutex *lock,
  36         set_current_state(TASK_RUNNING);
  37
  38         if (unlikely(ret)) {
  39 -               remove_waiter(lock, &waiter);
  40 +               if (rt_mutex_has_waiters(lock))
  41 +                       remove_waiter(lock, &waiter);
  42                 rt_mutex_handle_deadlock(ret, chwalk, &waiter);
  43         }
  44