Re: [PATCH -v11][RFC] mutex: implement adaptive spinning

[Nick Piggin <npiggin@suse.de>]

On Thu, Jan 15, 2009 at 08:44:03AM +0100, Peter Zijlstra wrote:
> On Thu, 2009-01-15 at 01:46 +0100, Nick Piggin wrote:
> 
> > Hmm, well this is rather a slow path, I would say. I'd prefer not to
> > modify schedule in this way (if we just get scheduled back on after
> > being switched away, the subsequent call to schedule is going to be
> > cache hot and not do too much work).
> > 
> > preempt_enable_noresched maybe if you really care, would close up the
> > window even more. But is it really worthwhile? We'd want to see numbers
> > (when in doubt, keep it  simpler).
> 
> I initially did the preempt_enable_no_resched() thing and that showed
> some improvement for PREEMPT=y kernels (lost the numbers though).
> 
> When I redid all the patches I tried closing that last hole by doing
> that __schedule() thing, never realizing that schedule() would then get
> extra overhead,.. d'0h.
> 
> I agree that that isn't worth it. I shall revert to
> preempt_enable_no_resched() and try to get some new numbers.

Thanks!