[PATCH] futex: PI state locking fix
authorIngo Molnar <mingo@elte.hu>
Fri, 16 Mar 2007 21:38:31 +0000 (13:38 -0800)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Sat, 17 Mar 2007 02:25:06 +0000 (19:25 -0700)
Testing of -rt by IBM uncovered a locking bug in wake_futex_pi(): the PI
state needs to be locked before we access it.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Chuck Ebbert <cebbert@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/futex.c

index e749e7df14b1a598c201f61b4d9b5bf8fb7c3d97..5a270b5e3f95de173e44194cd7e5b3d511ec6ff8 100644 (file)
@@ -565,6 +565,7 @@ static int wake_futex_pi(u32 __user *uaddr, u32 uval, struct futex_q *this)
        if (!pi_state)
                return -EINVAL;
 
+       spin_lock(&pi_state->pi_mutex.wait_lock);
        new_owner = rt_mutex_next_owner(&pi_state->pi_mutex);
 
        /*
@@ -604,6 +605,7 @@ static int wake_futex_pi(u32 __user *uaddr, u32 uval, struct futex_q *this)
        pi_state->owner = new_owner;
        spin_unlock_irq(&new_owner->pi_lock);
 
+       spin_unlock(&pi_state->pi_mutex.wait_lock);
        rt_mutex_unlock(&pi_state->pi_mutex);
 
        return 0;