Subject: Re: [BUG] How to get real-time priority using idle priority
From: Mike Galbraith <efault@gmx.de>
To: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Brian Rogers <brian@xyzw.org>, Ingo Molnar <mingo@elte.hu>,
       linux-kernel@vger.kernel.org
In-Reply-To: <1232015423.13856.5.camel@marge.simson.net>
References: <4969D0D7.2060401@xyzw.org>
	 <1231736941.6003.7.camel@marge.simson.net>
	 <1231765433.5789.35.camel@marge.simson.net> <20090112131406.GB670@elte.hu>
	 <496BE8F6.1040308@xyzw.org>  <1232011723.26761.36.camel@marge.simson.net>
	 <1232014456.8870.26.camel@laptop>
	 <1232015423.13856.5.camel@marge.simson.net>
Content-Type: text/plain
Date: Thu, 15 Jan 2009 12:37:08 +0100
Message-Id: <1232019428.5720.8.camel@marge.simson.net>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1515
Lines: 56

On Thu, 2009-01-15 at 11:30 +0100, Mike Galbraith wrote:
> On Thu, 2009-01-15 at 11:14 +0100, Peter Zijlstra wrote:
> >
> > Which leads me to suggest the following
> > 
> > ---
> > diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
> > index 8e1352c..f2d2d94 100644
> > --- a/kernel/sched_fair.c
> > +++ b/kernel/sched_fair.c
> > @@ -283,7 +283,7 @@ static void update_min_vruntime(struct cfs_rq
> > *cfs_rq)
> >  						   struct sched_entity,
> >  						   run_node);
> >  
> > -		if (vruntime == cfs_rq->min_vruntime)
> > +		if (!cfs_rq->curr)
> >  			vruntime = se->vruntime;
> >  		else
> >  			vruntime = min_vruntime(vruntime, se->vruntime);
> 
> Aha.  Yeah, I'll re-test with that instead.

Works a treat.

> > The below can be split into 3 patches:
> > 
> >  - the idle weight change (do we really need that? why?)
> 
> I saw idle tasks slamming extremely far.  I'll verify, less is more.

time advanced in 100ms
weight=2
64765.988352
67012.881408
88501.412352

weight=3
35496.181411
34130.971298
35497.411573

Measured from an RT shell doing..
   while sleep .1; do cat /proc/sched_debug >> /debug; done
..for a pinned chew.  Not necessarily gnats-arse accurate, but good
enough to see the margin of error is pretty high with weight=2.

Your call.

	-Mike

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