[PATCH 1/4] locking/rtmutex: Implement lockless top-waiter wakeup

[Davidlohr Bueso]

Mark the task for later wakeup after the wait_lock has been released.
This way, once the next task is awoken, it will have a better chance
to of finding the wait_lock free when continuing executing in
__rt_mutex_slowlock() when trying to acquire the rtmutex, calling
try_to_take_rt_mutex(). Upon contended scenarios, other tasks attempting
take the lock may acquire it first, right after the wait_lock is released,
but (a) this can also occur with the current code, as it relies on the
spinlock fairness, and (b) we are dealing with the top-waiter anyway,
so it will always take the lock next.

Signed-off-by: Davidlohr Bueso <dbueso@xxxxxxx>
---
 kernel/locking/rtmutex.c | 21 ++++++++++-----------
 1 file changed, 10 insertions(+), 11 deletions(-)

diff --git a/kernel/locking/rtmutex.c b/kernel/locking/rtmutex.c
index 36573e9..74188d8 100644
--- a/kernel/locking/rtmutex.c
+++ b/kernel/locking/rtmutex.c
@@ -955,14 +955,13 @@ static int task_blocks_on_rt_mutex(struct rt_mutex *lock,
 }
 
 /*
- * Wake up the next waiter on the lock.
- *
  * Remove the top waiter from the current tasks pi waiter list and
- * wake it up.
+ * queue it up.
  *
  * Called with lock->wait_lock held.
  */
-static void wakeup_next_waiter(struct rt_mutex *lock)
+static void mark_wakeup_next_waiter(struct wake_q_head *wake_q,
+				    struct rt_mutex *lock)
 {
 	struct rt_mutex_waiter *waiter;
 	unsigned long flags;
@@ -991,12 +990,7 @@ static void wakeup_next_waiter(struct rt_mutex *lock)
 
 	raw_spin_unlock_irqrestore(&current->pi_lock, flags);
 
-	/*
-	 * It's safe to dereference waiter as it cannot go away as
-	 * long as we hold lock->wait_lock. The waiter task needs to
-	 * acquire it in order to dequeue the waiter.
-	 */
-	wake_up_process(waiter->task);
+	wake_q_add(wake_q, waiter->task);
 }
 
 /*
@@ -1255,6 +1249,8 @@ static inline int rt_mutex_slowtrylock(struct rt_mutex *lock)
 static void __sched
 rt_mutex_slowunlock(struct rt_mutex *lock)
 {
+	WAKE_Q(wake_q);
+
 	raw_spin_lock(&lock->wait_lock);
 
 	debug_rt_mutex_unlock(lock);
@@ -1303,10 +1299,13 @@ rt_mutex_slowunlock(struct rt_mutex *lock)
 	/*
 	 * The wakeup next waiter path does not suffer from the above
 	 * race. See the comments there.
+	 *
+	 * Queue the next waiter for wakeup once we release the wait_lock.
 	 */
-	wakeup_next_waiter(lock);
+	mark_wakeup_next_waiter(&wake_q, lock);
 
 	raw_spin_unlock(&lock->wait_lock);
+	wake_up_q(&wake_q);
 
 	/* Undo pi boosting if necessary: */
 	rt_mutex_adjust_prio(current);
-- 
2.1.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/