Re: [PATCH 1/1] [PATCH v2]sched/pelt: Refine the enqueue_load_avg calculate method

[Kuyo Chang]

On Thu, 2022-04-14 at 09:37 +0200, Vincent Guittot wrote:
> On Thu, 14 Apr 2022 at 03:59, Kuyo Chang <kuyo.chang@xxxxxxxxxxxx>
> wrote:
> > 
> > From: kuyo chang <kuyo.chang@xxxxxxxxxxxx>
> > 
> > I meet the warning message at cfs_rq_is_decayed at below code.
> > 
> > SCHED_WARN_ON(cfs_rq->avg.load_avg ||
> >                     cfs_rq->avg.util_avg ||
> >                     cfs_rq->avg.runnable_avg)
> > 
> > Following is the calltrace.
> > 
> > Call trace:
> > __update_blocked_fair
> > update_blocked_averages
> > newidle_balance
> > pick_next_task_fair
> > __schedule
> > schedule
> > pipe_read
> > vfs_read
> > ksys_read
> > 
> > After code analyzing and some debug messages, I found it exits a
> > corner
> > case at attach_entity_load_avg which will cause load_sum is null
> > but
> > load_avg is not.
> > Consider se_weight is 88761 according by sched_prio_to_weight
> > table.
> > And assume the get_pelt_divider() is 47742, se->avg.load_avg is 1.
> > By the calculating for se->avg.load_sum as following will become
> > zero
> > as following.
> > se->avg.load_sum =
> >         div_u64(se->avg.load_avg * se->avg.load_sum,
> > se_weight(se));
> > se->avg.load_sum = 1*47742/88761 = 0.
> > 
> > After enqueue_load_avg code as below.
> > cfs_rq->avg.load_avg += se->avg.load_avg;
> > cfs_rq->avg.load_sum += se_weight(se) * se->avg.load_sum;
> > 
> > Then the load_sum for cfs_rq will be 1 while the load_sum for
> > cfs_rq is 0.
> > So it will hit the warning message.
> > 
> > In order to fix the corner case, make sure the se->load_avg|sum is
> > correct
> > before enqueue_load_avg.
> > 
> > After long time testing, the kernel warning was gone and the system
> > runs
> > as well as before.
> 
> This needs a fix tag:
> Fixes: f207934fb79d ("sched/fair: Align PELT windows between cfs_rq
> and its se")

Thanks for your friendly reminder.


> > 
> > Signed-off-by: kuyo chang <kuyo.chang@xxxxxxxxxxxx>
> 
> Reviewed-by: Vincent Guittot <vincent.guittot@xxxxxxxxxx>
> 
> > ---
> > 
> > v1->v2:
> > 
> > (1)Thanks for suggestion from Peter Zijlstra & Vincent Guittot.
> > (2)By suggestion from Vincent Guittot,
> > rework the se->load_sum calculation method for fix the corner case,
> > make sure the se->load_avg|sum is correct before enqueue_load_avg.
> > (3)Rework changlog.
> > 
> >  kernel/sched/fair.c | 8 +++++---
> >  1 file changed, 5 insertions(+), 3 deletions(-)
> > 
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index d4bd299d67ab..159274482c4e 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -3829,10 +3829,12 @@ static void attach_entity_load_avg(struct
> > cfs_rq *cfs_rq, struct sched_entity *s
> > 
> >         se->avg.runnable_sum = se->avg.runnable_avg * divider;
> > 
> > -       se->avg.load_sum = divider;
> > -       if (se_weight(se)) {
> > +       se->avg.load_sum = se->avg.load_avg * divider;
> > +       if (se_weight(se) < se->avg.load_sum) {
> >                 se->avg.load_sum =
> > -                       div_u64(se->avg.load_avg * se-
> > >avg.load_sum, se_weight(se));
> > +                       div_u64(se->avg.load_sum, se_weight(se));
> > +       } else {
> > +               se->avg.load_sum = 1;
> >         }
> > 
> >         enqueue_load_avg(cfs_rq, se);
> > --
> > 2.18.0
> >