Re: [PATCH -v4 8/8] locking/mutex: Enable optimistic spinning of woken waiter

From: Peter Zijlstra
Date: Tue Oct 18 2016 - 08:20:25 EST


On Mon, Oct 17, 2016 at 07:21:28PM -0400, Waiman Long wrote:
> On 10/07/2016 10:52 AM, Peter Zijlstra wrote:
> >@@ -600,7 +630,7 @@ __mutex_lock_common(struct mutex *lock,
> > mutex_acquire_nest(&lock->dep_map, subclass, 0, nest_lock, ip);
> >
> > if (__mutex_trylock(lock, false) ||
> >- mutex_optimistic_spin(lock, ww_ctx, use_ww_ctx)) {
> >+ mutex_optimistic_spin(lock, ww_ctx, use_ww_ctx, false)) {
> > /* got the lock, yay! */
> > lock_acquired(&lock->dep_map, ip);
> > if (use_ww_ctx)
> >@@ -669,7 +699,8 @@ __mutex_lock_common(struct mutex *lock,
> > * state back to RUNNING and fall through the next schedule(),
> > * or we must see its unlock and acquire.
> > */
> >- if (__mutex_trylock(lock, first))
> >+ if ((first&& mutex_optimistic_spin(lock, ww_ctx, use_ww_ctx, true)) ||
> >+ __mutex_trylock(lock, first))
>
> Do we need a __mutex_trylock() here? mutex_optimistic_spin() will do the
> trylock and we have one at the top of the loop.

Yes, mutex_optimistic_spin() can be a no-op.