rt-mutex: fix stale return value
authorThomas Gleixner <tglx@linutronix.de>
Fri, 8 Jun 2007 20:46:57 +0000 (13:46 -0700)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Sat, 9 Jun 2007 00:23:34 +0000 (17:23 -0700)
Alexey Kuznetsov found some problems in the pi-futex code.

The major problem is a stale return value in rt_mutex_slowlock():

When the pi chain walk returns -EDEADLK, but the waiter was woken up during
the phases where the locks were dropped, the rtmutex could be acquired, but
due to the stale return value -EDEADLK returned to the caller.

Reset the return value in the retry path.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Cc: Ulrich Drepper <drepper@redhat.com>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/rtmutex.c

index 12879f6c1ec3b9a53069690f0205082ea581565c..a273183c37a037307b1f1915171fffcce1ff413f 100644 (file)
@@ -636,9 +636,16 @@ rt_mutex_slowlock(struct rt_mutex *lock, int state,
                         * all over without going into schedule to try
                         * to get the lock now:
                         */
-                       if (unlikely(!waiter.task))
+                       if (unlikely(!waiter.task)) {
+                               /*
+                                * Reset the return value. We might
+                                * have returned with -EDEADLK and the
+                                * owner released the lock while we
+                                * were walking the pi chain.
+                                */
+                               ret = 0;
                                continue;
-
+                       }
                        if (unlikely(ret))
                                break;
                }