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

[Vincent Guittot]

On Mon, 11 Apr 2022 at 10:25, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>
> On Mon, Apr 11, 2022 at 02:16:56PM +0800, Kuyo Chang 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 zero and
> > 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.
> >
> > After all, I refer the following commit patch to do the similar thing at
> > enqueue_load_avg.
> > sched/pelt: Relax the sync of load_sum with load_avg
>
>   2d02fa8cc21a ("sched/pelt: Relax the sync of load_sum with load_avg")
>
> >
> > After long time testing, the kernel warning was gone and the system runs
> > as well as before.
> >
>
> Could probably do with a Fixes: tag too
>
> > Signed-off-by: kuyo chang <kuyo.chang@xxxxxxxxxxxx>
> > ---
> >  kernel/sched/fair.c | 6 ++++--
> >  1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index d4bd299d67ab..30d8b6dba249 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -3074,8 +3074,10 @@ account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se)
> >  static inline void
> >  enqueue_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *se)
> >  {
> > -     cfs_rq->avg.load_avg += se->avg.load_avg;
> > -     cfs_rq->avg.load_sum += se_weight(se) * se->avg.load_sum;
> > +     add_positive(&cfs_rq->avg.load_avg, se->avg.load_avg);
> > +     add_positive(&cfs_rq->avg.load_sum, se_weight(se) * se->avg.load_sum);
> > +     cfs_rq->avg.load_sum = max_t(u32, cfs_rq->avg.load_sum,
> > +                                       cfs_rq->avg.load_avg * PELT_MIN_DIVIDER);
> >  }
>
> Yes,. that looks right. Looking at this again though, I wonder if we
> should perhaps clean is all up a bit.

adding se->load_avg|sum in cfs_rq should not create a situation where
cfs->load_sum is null but not cfs->load_avg unless se->load_avg|sum
are wrong from the beginning which seems to be the case here

>
> Does something like the below on top make sense?
>
>
> ---
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -3071,23 +3071,37 @@ account_entity_dequeue(struct cfs_rq *cf
>  } while (0)
>
>  #ifdef CONFIG_SMP
> +
> +/*
> + * Because of rounding, se->util_sum might ends up being +1 more than
> + * cfs->util_sum. Although this is not a problem by itself, detaching
> + * a lot of tasks with the rounding problem between 2 updates of
> + * util_avg (~1ms) can make cfs->util_sum becoming null whereas
> + * cfs_util_avg is not.
> + *
> + * Check that util_sum is still above its lower bound for the new
> + * util_avg. Given that period_contrib might have moved since the last
> + * sync, we are only sure that util_sum must be above or equal to
> + *    util_avg * minimum possible divider
> + */
> +#define __update_sa(sa, name, delta_avg, delta_sum) do {       \
> +       add_positive(&(sa)->name##_avg, delta_avg);             \
> +       add_positive(&(sa)->name##_sum, delta_sum);             \
> +       (sa)->name##_sum = max_t(typeof((sa)->name##_sum),      \
> +                              (sa)->name##_sum,                \
> +                              (sa)->name##_avg * PELT_MIN_DIVIDER); \
> +} while (0)
> +
>  static inline void
>  enqueue_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *se)
>  {
> -       add_positive(&cfs_rq->avg.load_avg, se->avg.load_avg);
> -       add_positive(&cfs_rq->avg.load_sum, se_weight(se) * se->avg.load_sum);
> -       cfs_rq->avg.load_sum = max_t(u32, cfs_rq->avg.load_sum,
> -                                         cfs_rq->avg.load_avg * PELT_MIN_DIVIDER);
> +       __update_sa(&cfs_rq->avg, load, se->avg.load_avg, se->avg.load_sum);
>  }
>
>  static inline void
>  dequeue_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *se)
>  {
> -       sub_positive(&cfs_rq->avg.load_avg, se->avg.load_avg);
> -       sub_positive(&cfs_rq->avg.load_sum, se_weight(se) * se->avg.load_sum);
> -       /* See update_cfs_rq_load_avg() */
> -       cfs_rq->avg.load_sum = max_t(u32, cfs_rq->avg.load_sum,
> -                                         cfs_rq->avg.load_avg * PELT_MIN_DIVIDER);
> +       __update_sa(&cfs_rq->avg, load, -se->avg.load_avg, -se->avg.load_sum);
>  }
>  #else
>  static inline void
> @@ -3529,12 +3543,7 @@ update_tg_cfs_util(struct cfs_rq *cfs_rq
>         se->avg.util_sum = new_sum;
>
>         /* Update parent cfs_rq utilization */
> -       add_positive(&cfs_rq->avg.util_avg, delta_avg);
> -       add_positive(&cfs_rq->avg.util_sum, delta_sum);
> -
> -       /* See update_cfs_rq_load_avg() */
> -       cfs_rq->avg.util_sum = max_t(u32, cfs_rq->avg.util_sum,
> -                                         cfs_rq->avg.util_avg * PELT_MIN_DIVIDER);
> +       __update_sa(&cfs_rq->avg, util, delta_avg, delta_sum);
>  }
>
>  static inline void
> @@ -3560,11 +3569,7 @@ update_tg_cfs_runnable(struct cfs_rq *cf
>         se->avg.runnable_sum = new_sum;
>
>         /* Update parent cfs_rq runnable */
> -       add_positive(&cfs_rq->avg.runnable_avg, delta_avg);
> -       add_positive(&cfs_rq->avg.runnable_sum, delta_sum);
> -       /* See update_cfs_rq_load_avg() */
> -       cfs_rq->avg.runnable_sum = max_t(u32, cfs_rq->avg.runnable_sum,
> -                                             cfs_rq->avg.runnable_avg * PELT_MIN_DIVIDER);
> +       __update_sa(&cfs_rq->avg, runnable, delta_avg, delta_sum);
>  }
>
>  static inline void
> @@ -3628,11 +3633,7 @@ update_tg_cfs_load(struct cfs_rq *cfs_rq
>
>         se->avg.load_sum = runnable_sum;
>         se->avg.load_avg = load_avg;
> -       add_positive(&cfs_rq->avg.load_avg, delta_avg);
> -       add_positive(&cfs_rq->avg.load_sum, delta_sum);
> -       /* See update_cfs_rq_load_avg() */
> -       cfs_rq->avg.load_sum = max_t(u32, cfs_rq->avg.load_sum,
> -                                         cfs_rq->avg.load_avg * PELT_MIN_DIVIDER);
> +       __update_sa(&cfs_rq->avg, load, delta_avg, delta_sum);
>  }
>
>  static inline void add_tg_cfs_propagate(struct cfs_rq *cfs_rq, long runnable_sum)
> @@ -3747,33 +3748,13 @@ update_cfs_rq_load_avg(u64 now, struct c
>                 raw_spin_unlock(&cfs_rq->removed.lock);
>
>                 r = removed_load;
> -               sub_positive(&sa->load_avg, r);
> -               sub_positive(&sa->load_sum, r * divider);
> -               /* See sa->util_sum below */
> -               sa->load_sum = max_t(u32, sa->load_sum, sa->load_avg * PELT_MIN_DIVIDER);
> +               __update_sa(sa, load, r, r*divider);
>
>                 r = removed_util;
> -               sub_positive(&sa->util_avg, r);
> -               sub_positive(&sa->util_sum, r * divider);
> -               /*
> -                * Because of rounding, se->util_sum might ends up being +1 more than
> -                * cfs->util_sum. Although this is not a problem by itself, detaching
> -                * a lot of tasks with the rounding problem between 2 updates of
> -                * util_avg (~1ms) can make cfs->util_sum becoming null whereas
> -                * cfs_util_avg is not.
> -                * Check that util_sum is still above its lower bound for the new
> -                * util_avg. Given that period_contrib might have moved since the last
> -                * sync, we are only sure that util_sum must be above or equal to
> -                *    util_avg * minimum possible divider
> -                */
> -               sa->util_sum = max_t(u32, sa->util_sum, sa->util_avg * PELT_MIN_DIVIDER);
> +               __update_sa(sa, util, r, r*divider);
>
>                 r = removed_runnable;
> -               sub_positive(&sa->runnable_avg, r);
> -               sub_positive(&sa->runnable_sum, r * divider);
> -               /* See sa->util_sum above */
> -               sa->runnable_sum = max_t(u32, sa->runnable_sum,
> -                                             sa->runnable_avg * PELT_MIN_DIVIDER);
> +               __update_sa(sa, runnable, r, r*divider);
>
>                 /*
>                  * removed_runnable is the unweighted version of removed_load so we
> @@ -3861,17 +3842,8 @@ static void attach_entity_load_avg(struc
>  static void detach_entity_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *se)
>  {
>         dequeue_load_avg(cfs_rq, se);
> -       sub_positive(&cfs_rq->avg.util_avg, se->avg.util_avg);
> -       sub_positive(&cfs_rq->avg.util_sum, se->avg.util_sum);
> -       /* See update_cfs_rq_load_avg() */
> -       cfs_rq->avg.util_sum = max_t(u32, cfs_rq->avg.util_sum,
> -                                         cfs_rq->avg.util_avg * PELT_MIN_DIVIDER);
> -
> -       sub_positive(&cfs_rq->avg.runnable_avg, se->avg.runnable_avg);
> -       sub_positive(&cfs_rq->avg.runnable_sum, se->avg.runnable_sum);
> -       /* See update_cfs_rq_load_avg() */
> -       cfs_rq->avg.runnable_sum = max_t(u32, cfs_rq->avg.runnable_sum,
> -                                             cfs_rq->avg.runnable_avg * PELT_MIN_DIVIDER);
> +       __update_sa(&cfs_rq->avg, util, se->avg.util_avg, se->avg.util_sum);
> +       __update_sa(&cfs_rq->avg, runnable, se->avg.runnable_avg, se->avg.runnable_sum);
>
>         add_tg_cfs_propagate(cfs_rq, -se->avg.load_sum);
>