Re: [RFC][PATCH 2/3] sched: On-demand cfs_rq list

[Peter Zijlstra]

On Fri, 2010-09-03 at 04:33 +0100, Paul Turner wrote:
> > @@ -668,6 +693,11 @@ static void update_cfs_load(struct cfs_r
> >                cfs_rq->load_period /= 2;
> >                cfs_rq->load_avg /= 2;
> >        }
> > +
> > +       if (lb && !cfs_rq->nr_running) {
> > +               if (cfs_rq->load_period < (period / 8))
> > +                       list_del_leaf_cfs_rq(cfs_rq);
> > +       }
> >  }
> >
> 
> In the case of zero load, load_avg above will decay however
> load_period will remain in the range  [period/2, period], any entity
> that has passed period/8 time will remain on the leaf rq list.

Right, already changed that to cfs_rq->load_avg < (period / 8) after you
pointed that out on irc..

> The lb condition could also be relaxed in the dequeue_entity case.
> 
Possibly yeah, we'll have to see if that makes much difference.
--
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/