rtmutex: Fix deadlock detector for real

commit 397335f004f41e5fcf7a795e94eb3ab83411a17c upstream.

The current deadlock detection logic does not work reliably due to the
following early exit path:

	/*
	 * Drop out, when the task has no waiters. Note,
	 * top_waiter can be NULL, when we are in the deboosting
	 * mode!
	 */
	if (top_waiter && (!task_has_pi_waiters(task) ||
			   top_waiter != task_top_pi_waiter(task)))
		goto out_unlock_pi;

So this not only exits when the task has no waiters, it also exits
unconditionally when the current waiter is not the top priority waiter
of the task.

So in a nested locking scenario, it might abort the lock chain walk
and therefor miss a potential deadlock.

Simple fix: Continue the chain walk, when deadlock detection is
enabled.

We also avoid the whole enqueue, if we detect the deadlock right away
(A-A). It's an optimization, but also prevents that another waiter who
comes in after the detection and before the task has undone the damage
observes the situation and detects the deadlock and returns
-EDEADLOCK, which is wrong as the other task is not in a deadlock
situation.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Link: http://lkml.kernel.org/r/20140522031949.725272460@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
[bwh: Backported to 3.2: adjust filename]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

diff --git a/kernel/rtmutex.c b/kernel/rtmutex.c
index f9d8482..f672d5e 100644 (file)
--- a/kernel/rtmutex.c
+++ b/kernel/rtmutex.c
@@ -211,9 +211,16 @@ static int rt_mutex_adjust_prio_chain(struct task_struct *task,
         * top_waiter can be NULL, when we are in the deboosting
         * mode!
         */
-       if (top_waiter && (!task_has_pi_waiters(task) ||
-                          top_waiter != task_top_pi_waiter(task)))
-               goto out_unlock_pi;
+       if (top_waiter) {
+               if (!task_has_pi_waiters(task))
+                       goto out_unlock_pi;
+               /*
+                * If deadlock detection is off, we stop here if we
+                * are not the top pi waiter of the task.
+                */
+               if (!detect_deadlock && top_waiter != task_top_pi_waiter(task))
+                       goto out_unlock_pi;
+       }
 
        /*
         * When deadlock detection is off then we check, if further
@@ -229,7 +236,12 @@ static int rt_mutex_adjust_prio_chain(struct task_struct *task,
                goto retry;
        }
 
-       /* Deadlock detection */
+       /*
+        * Deadlock detection. If the lock is the same as the original
+        * lock which caused us to walk the lock chain or if the
+        * current lock is owned by the task which initiated the chain
+        * walk, we detected a deadlock.
+        */
        if (lock == orig_lock || rt_mutex_owner(lock) == top_task) {
                debug_rt_mutex_deadlock(deadlock_detect, orig_waiter, lock);
                raw_spin_unlock(&lock->wait_lock);
@@ -398,6 +410,18 @@ static int task_blocks_on_rt_mutex(struct rt_mutex *lock,
        unsigned long flags;
        int chain_walk = 0, res;
 
+       /*
+        * Early deadlock detection. We really don't want the task to
+        * enqueue on itself just to untangle the mess later. It's not
+        * only an optimization. We drop the locks, so another waiter
+        * can come in before the chain walk detects the deadlock. So
+        * the other will detect the deadlock and return -EDEADLOCK,
+        * which is wrong, as the other waiter is not in a deadlock
+        * situation.
+        */
+       if (detect_deadlock && owner == task)
+               return -EDEADLK;
+
        raw_spin_lock_irqsave(&task->pi_lock, flags);
        __rt_mutex_adjust_prio(task);
        waiter->task = task;