Date:   Wed, 13 May 2020 11:22:18 -0700
From:   bsegall@...gle.com
To:     Vincent Guittot <vincent.guittot@...aro.org>
Cc:     Ben Segall <bsegall@...gle.com>, Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Juri Lelli <juri.lelli@...hat.com>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Mel Gorman <mgorman@...e.de>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        Phil Auld <pauld@...hat.com>, Tao Zhou <ouwen210@...mail.com>
Subject: Re: [PATCH] sched/fair: fix unthrottle_cfs_rq for leaf_cfs_rq list

Vincent Guittot <vincent.guittot@...aro.org> writes:

> On Tue, 12 May 2020 at 20:59, <bsegall@...gle.com> wrote:
>>
>> Vincent Guittot <vincent.guittot@...aro.org> writes:
>>
>> > Although not exactly identical, unthrottle_cfs_rq() and enqueue_task_fair()
>> > are quite close and follow the same sequence for enqueuing an entity in the
>> > cfs hierarchy. Modify unthrottle_cfs_rq() to use the same pattern as
>> > enqueue_task_fair(). This fixes a problem already faced with the latter and
>> > add an optimization in the last for_each_sched_entity loop.
>> >
>> > Fixes: fe61468b2cb (sched/fair: Fix enqueue_task_fair warning)
>> > Reported-by Tao Zhou <zohooouoto@...o.com.cn>
>> > Signed-off-by: Vincent Guittot <vincent.guittot@...aro.org>
>> > ---
>> >
>> > This path applies on top of 20200507203612.GF19331@...ien.usersys.redhat.com
>> > and fixes similar problem for unthrottle_cfs_rq()
>> >
>> >  kernel/sched/fair.c | 37 ++++++++++++++++++++++++++++---------
>> >  1 file changed, 28 insertions(+), 9 deletions(-)
>> >
>> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
>> > index e2450c2e0747..4b73518aa25c 100644
>> > --- a/kernel/sched/fair.c
>> > +++ b/kernel/sched/fair.c
>> > @@ -4803,26 +4803,44 @@ void unthrottle_cfs_rq(struct cfs_rq *cfs_rq)
>> >       idle_task_delta = cfs_rq->idle_h_nr_running;
>> >       for_each_sched_entity(se) {
>> >               if (se->on_rq)
>> > -                     enqueue = 0;
>> > +                     break;
>> > +             cfs_rq = cfs_rq_of(se);
>> > +             enqueue_entity(cfs_rq, se, ENQUEUE_WAKEUP);
>> >
>> > +             cfs_rq->h_nr_running += task_delta;
>> > +             cfs_rq->idle_h_nr_running += idle_task_delta;
>> > +
>> > +             /* end evaluation on encountering a throttled cfs_rq */
>> > +             if (cfs_rq_throttled(cfs_rq))
>> > +                     goto unthrottle_throttle;
>> > +     }
>> > +
>> > +     for_each_sched_entity(se) {
>> >               cfs_rq = cfs_rq_of(se);
>> > -             if (enqueue) {
>> > -                     enqueue_entity(cfs_rq, se, ENQUEUE_WAKEUP);
>> > -             } else {
>> > -                     update_load_avg(cfs_rq, se, 0);
>> > -                     se_update_runnable(se);
>> > -             }
>> > +
>> > +             update_load_avg(cfs_rq, se, UPDATE_TG);
>> > +             se_update_runnable(se);
>> >
>> >               cfs_rq->h_nr_running += task_delta;
>> >               cfs_rq->idle_h_nr_running += idle_task_delta;
>> >
>> > +
>> > +             /* end evaluation on encountering a throttled cfs_rq */
>> >               if (cfs_rq_throttled(cfs_rq))
>> > -                     break;
>> > +                     goto unthrottle_throttle;
>> > +
>> > +             /*
>> > +              * One parent has been throttled and cfs_rq removed from the
>> > +              * list. Add it back to not break the leaf list.
>> > +              */
>> > +             if (throttled_hierarchy(cfs_rq))
>> > +                     list_add_leaf_cfs_rq(cfs_rq);
>> >       }
>> >
>> >       if (!se)
>>
>> The if is no longer necessary, unlike in enqueue, where the skip goto
>
> Yes. Good point
>
>> goes to this if statement rather than past (but enqueue could be changed
>> to match this). Also in general if we are making these loops absolutely
>
> There is a patch on mailing that skip the if statement. I'm going to
> update it to remove the if
>
>> identical we should probably pull them out to a common function (ideally
>> including the goto target and following loop as well).
>
> I tried that but was not convinced by the result which generated a lot
> of arguments. I didn't want to delay the fix for such cleanup but I
> will have a closer look after. Also the same kind identical sequence
> and clean up can be done with dequeue_task_fair and throtthle_cfs_rq.
> But Those don't have the problem we are fixing here
>
>>
>> >               add_nr_running(rq, task_delta);
>> >
>> > +unthrottle_throttle:
>> >       /*
>> >        * The cfs_rq_throttled() breaks in the above iteration can result in
>> >        * incomplete leaf list maintenance, resulting in triggering the
>> > @@ -4831,7 +4849,8 @@ void unthrottle_cfs_rq(struct cfs_rq *cfs_rq)
>> >       for_each_sched_entity(se) {
>> >               cfs_rq = cfs_rq_of(se);
>> >
>> > -             list_add_leaf_cfs_rq(cfs_rq);
>> > +             if (list_add_leaf_cfs_rq(cfs_rq))
>> > +                     break;
>>
>> Do we also need to handle the case of tg_unthrottle_up followed by early exit
>> from unthrottle_cfs_rq? I do not have enough of an idea what
>> list_add_leaf_cfs_rq is doing to say.
>
> If you are speaking about the 'if (!cfs_rq->load.weight) return;"
> after walk_tg_tree_from(). I also thought it was needed but after more
> analyses, I concluded that if cfs_rq->load.weight == 0 , no child has
> been added in the leaf_cfs_rq_list in such case

Hmm, yes, if load.weight is 0 it should not have done anything there.

>
>
>>
>> >       }
>> >
>> >       assert_list_leaf_cfs_rq(rq);

Powered by blists - more mailing lists