Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <cover.1648228023.git.tim.c.chen@linux.intel.com>
 <f59f1dfeec4692e7a373cc810168912a2d2f8f3c.1648228023.git.tim.c.chen@linux.intel.com>
 <CAKfTPtBDA1uZ1xvo_uhqnNVg69DPXXJBeo0+aQdVJiosB_qgrw@mail.gmail.com> <20220331114709.GS8939@worktop.programming.kicks-ass.net>
In-Reply-To: <20220331114709.GS8939@worktop.programming.kicks-ass.net>
From:   Vincent Guittot <vincent.guittot@linaro.org>
Date:   Thu, 31 Mar 2022 14:15:38 +0200
Message-ID: <CAKfTPtD-gezaAu5kn_vRjWdhHdu_NQLufpTFbSAKQtxPKPO_ug@mail.gmail.com>
Subject: Re: [PATCH 2/2] sched/fair: Simple runqueue order on migrate
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     Tim Chen <tim.c.chen@linux.intel.com>, Ingo Molnar <mingo@elte.hu>,
        Juri Lelli <juri.lelli@redhat.com>,
        Yu Chen <yu.c.chen@intel.com>,
        Walter Mack <walter.mack@intel.com>,
        Mel Gorman <mgorman@suse.de>, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, 31 Mar 2022 at 13:47, Peter Zijlstra <peterz@infradead.org> wrote:
>
>
> It's been 3 months since I wrote these patches, so memory is vague at
> best :/
>
> On Tue, Mar 29, 2022 at 11:03:44AM +0200, Vincent Guittot wrote:
>
> > > +static void place_entity_migrate(struct cfs_rq *cfs_rq, struct sched_entity *se)
> > > +{
> > > +       if (!sched_feat(PLACE_MIGRATE))
> > > +               return;
> > > +
> > > +       if (cfs_rq->nr_running < se->migrated) {
> > > +               /*
> > > +                * Migrated to a shorter runqueue, go first because
> > > +                * we were under-served on the old runqueue.
> > > +                */
> > > +               se->vruntime = cfs_rq->min_vruntime;
> > > +               return;
> > > +       }
> > > +
> > > +       /*
> > > +        * Migrated to a longer runqueue, go last because
> > > +        * we got over-served on the old runqueue.
> > > +        */
> > > +       se->vruntime = cfs_rq->min_vruntime + sched_vslice(cfs_rq, se);
> > > +}
> > > +
> > >  static void check_enqueue_throttle(struct cfs_rq *cfs_rq);
> > >
> > >  static inline bool cfs_bandwidth_used(void);
> > > @@ -4296,6 +4317,8 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
> > >
> > >         if (flags & ENQUEUE_WAKEUP)
> > >                 place_entity(cfs_rq, se, 0);
> > > +       else if (se->migrated)
> > > +               place_entity_migrate(cfs_rq, se);
> > >
> > >         check_schedstat_required();
> > >         update_stats_enqueue_fair(cfs_rq, se, flags);
>
> > > @@ -6973,14 +6997,15 @@ static void migrate_task_rq_fair(struct task_struct *p, int new_cpu)
> > >                  * wakee task is less decayed, but giving the wakee more load
> > >                  * sounds not bad.
> > >                  */
> > > -               remove_entity_load_avg(&p->se);
> > > +               remove_entity_load_avg(se);
> > >         }
> > >
> > >         /* Tell new CPU we are migrated */
> > > -       p->se.avg.last_update_time = 0;
> > > +       se->avg.last_update_time = 0;
> > >
> > >         /* We have migrated, no longer consider this task hot */
> > > -       p->se.migrated = 1;
> > > +       for_each_sched_entity(se)
> > > +               se->migrated = READ_ONCE(cfs_rq_of(se)->nr_running) + !se->on_rq;
> >
> > Why do we need to loop on se ? Isn't p->se enough ?
>
> Yeah; I really don't recall why I did that. And looking at it now, it
> doesn't really make much sense. I suppose it will trigger
> place_entity_migrate() for the group entries, but on the old CPU.

Yes, that is the reason for my question. task is still connected to
prev cfs so you trigger place_entity_migrate() on the sched_entity of
previous hierarchy. Also this will only happen during enqueue if the
se has not run in the meantime. So it 's probably to be a nop most of
the time as the parent cfs is already enqueued when we migrate a
runnable task and in the other case it will break sched_group vruntime
and fairness

>
>