Re: [PATCH 3/4] sched/fair: Do not migrate on wake_affine_weight if weights are equal

[Mel Gorman]

On Mon, Feb 12, 2018 at 06:29:26PM +0100, Peter Zijlstra wrote:
> On Mon, Feb 12, 2018 at 02:58:56PM +0000, Mel Gorman wrote:
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index c1091cb023c4..28c8d9c91955 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -5747,7 +5747,16 @@ wake_affine_weight(struct sched_domain *sd, struct task_struct *p,
> >  		prev_eff_load *= 100 + (sd->imbalance_pct - 100) / 2;
> >  	prev_eff_load *= capacity_of(this_cpu);
> >  
> > -	return this_eff_load <= prev_eff_load ? this_cpu : nr_cpumask_bits;
> > +	/*
> > +	 * If sync, adjust the weight of prev_eff_load such that if
> > +	 * prev_eff == this_eff that select_idle_sibling will consider
> > +	 * stacking the wakee on top of the waker if no other CPU is
> > +	 * idle.
> > +	 */
> > +	if (sync)
> > +		prev_eff_load += 1;
> 
> So where we had <= and would consistently favour pulling the task to the
> waking CPU when all else what equal, you now switch to <, such that when
> things are equal we do not pull.
> 
> That makes sense I suppose.
> 

Yep, with the addenum that when CPU load is equal, it does not
necessarily mean they are equal in terms of memory locality. It might
make more sense to use <= if there were more cases where we stacked
tasks on the same CPU but we avoid that as much as possible for good
reasons.

-- 
Mel Gorman
SUSE Labs