On 06/15/2016 03:08 PM, Waiman Long wrote:

On 06/15/2016 01:12 PM, Peter Zijlstra wrote:

quoted

On Wed, Jun 15, 2016 at 09:56:59AM -0700, Davidlohr Bueso wrote:

quoted

On Tue, 14 Jun 2016, Waiman Long wrote:

quoted

+++ b/kernel/locking/osq_lock.c

@@ -115,7 +115,7 @@ bool osq_lock(struct optimistic_spin_queue *lock)
     * cmpxchg in an attempt to undo our queueing.
     */

-    while (!READ_ONCE(node->locked)) {
+    while (!smp_load_acquire(&node->locked)) {

Hmm this being a polling path, that barrier can get pretty expensive 
and
last I checked it was unnecessary:

I think he'll go rely on it later on.

In any case, its fairly simple to cure, just add
smp_acquire__after_ctrl_dep() at the end. If we bail because
need_resched() we don't need the acquire I think.

Yes, I only need the acquire barrier when the locking is successful. 
Thanks for the suggestion. I will make the change accordingly.

BTW, when will the smp_acquire__after_ctrl_dep() patch goes into the tip 
tree? My patch will have a dependency on that when I make the change.

Cheers,
Longman