Re: [RFC][PATCH v2 5/5] mutex: Give spinners a chance tospin_on_owner if need_resched() triggered while queued

[Paul E. McKenney]

On Fri, Jan 31, 2014 at 09:08:25PM +0100, Peter Zijlstra wrote:
> On Fri, Jan 31, 2014 at 12:01:37PM -0800, Jason Low wrote:
> > On Fri, Jan 31, 2014 at 6:09 AM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> > 
> > > I've downloaded AIM7 from sf.net and I hope I'm running it with 100+
> > > loads but I'm not entirely sure I got this thing right, its not really
> > > making progress with or without patch :/
> > 
> > Ingo's program http://lkml.org/lkml/2006/1/8/50 using the V option may
> > be able to generate similar mutex contention.
> > 
> > Currently still getting soft lockups with the updated version.
> 
> Bugger.. ok clearly I need to think harder still. I'm fairly sure this
> cancelation can work though, just seems tricky to get right :-)

We used to do something similar to avoid passing locks off to tasks that
had been interrupted while spinning, and it was a bit tricky.  But we
had it a bit easier, because we didn't actually have to remove the element
from the queue, just bypass it at lock-grant time.

							Thanx, Paul

> I'll give that proglet from Ingo a go, although that might be Monday ere
> I get to it.
> 

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/