Re: [PATCH v2] sched: Consolidate cpufreq updates

[Qais Yousef]

On 05/07/24 14:53, Vincent Guittot wrote:

> No I mean going out of idle. On an idle cpu, nothing happens at CFS
> task wakeup and we have to wait for the next tick to apply the new
> freq. This happens for both short task with uclamp min or long
> running/sleeping task (i.e. with high util_est)

I think I found the problem, thanks for catching it!

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index cbe79c8ac2ed..dea9383a906e 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -4714,7 +4714,7 @@ static inline void update_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *s
        if (se->avg.last_update_time && !(flags & SKIP_AGE_LOAD))
                __update_load_avg_se(now, cfs_rq, se);

-       cfs_rq->decayed  = update_cfs_rq_load_avg(now, cfs_rq);
+       cfs_rq->decayed |= update_cfs_rq_load_avg(now, cfs_rq);
        cfs_rq->decayed |= propagate_entity_load_avg(se);

        if (!se->avg.last_update_time && (flags & DO_ATTACH)) {