It is a bit annoying that do_exit() takes ->pi_lock to set PF_EXITING.  All
we need is to synchronize with lookup_pi_state() which saw this task
without PF_EXITING under ->pi_lock.
Change do_exit() to use spin_unlock_wait().
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
                schedule();
        }
 
+       tsk->flags |= PF_EXITING;
        /*
         * tsk->flags are checked in the futex code to protect against
         * an exiting task cleaning up the robust pi futexes.
         */
-       spin_lock_irq(&tsk->pi_lock);
-       tsk->flags |= PF_EXITING;
-       spin_unlock_irq(&tsk->pi_lock);
+       smp_mb();
+       spin_unlock_wait(&tsk->pi_lock);
 
        if (unlikely(in_atomic()))
                printk(KERN_INFO "note: %s[%d] exited with preempt_count %d\n",