Re: [PATCH] mutex: have non-spinning mutexes on s390 by default

[Peter Zijlstra <peterz@infradead.org>]

On Thu, 2009-04-09 at 10:50 -0700, Jeremy Fitzhardinge wrote:
> Peter Zijlstra wrote:
> > On Thu, 2009-04-09 at 18:53 +0200, Peter Zijlstra wrote:
> >
> >   
> >> I was looking at how an monitor-wait could be used here, but that
> >> appears non-trivial, there's two variables we're watching, lock->owner
> >> and rq->curr, either could change.
> >>
> >> Reducing that to 1 seems an interesting problem :-)
> >>     
> >
> > How about something like this?
> >
> > Does it make sense to implement an monitor-wait spinlock for the virt
> > case as well? Avi, Jeremy?
> >   
> 
> Last time I tried to put mwait in a spinlock, Arjan (or HPA?) said that 
> mwait takes approx a week and a half to wake up, and that it was really 
> only useful for idle power savings.

Yeah, sad that.

> Has that changed?

Nothing much, I was thinking perhaps it would make sense for the virt
case, but if its not properly virtualized then its pretty useless
indeed.

monitor-wait is basically a hardware/hv futex like thing, so I thought
it might help -- spinning in a guest is pretty painful.