Messages in this thread | | | Date | Tue, 28 Feb 2017 09:58:21 +0000 | From | Chris Wilson <> | Subject | Re: [PATCH] locking/ww_mutex: Replace cpu_relax with cond_resched for tests |
| |
On Tue, Feb 28, 2017 at 10:48:53AM +0100, Peter Zijlstra wrote: > On Tue, Feb 28, 2017 at 09:40:11AM +0000, Chris Wilson wrote: > > When busy-spinning on a ww_mutex_trylock(), we depend upon the other > > thread advancing and releasing the lock. This can not happen on a single > > CPU unless we relinquish it. > > Thanks, do you have any clue about those other ww-mutex test fails? So > far I've not had much luck poking at them.
Not yet, they didn't seem quite so self-explanatory as this trace :( -Chris
-- Chris Wilson, Intel Open Source Technology Centre
| |