Re: [RFC][PATCH 08/10] sched/fair: Implement delayed dequeue

[Peter Zijlstra]

On Thu, Apr 18, 2024 at 06:24:59PM +0200, Mike Galbraith wrote:
> Greetings,
> 
> I tossed a couple rocks at it today, and seem to have hit the little
> bugger.  The root cause seems to be doing the delay dequeue business on
> exiting tasks.  Hunk #1 of hacklet below seems to quell the explosions.
> 

Neat, Thanks! I was just about ready to go look at this again.

> 
> ---
>  kernel/sched/fair.c |    5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -5374,6 +5374,7 @@ dequeue_entity(struct cfs_rq *cfs_rq, st
>  		update_curr(cfs_rq);
> 
>  		if (sched_feat(DELAY_DEQUEUE) && sleep &&
> +		    !(entity_is_task(se) && (task_of(se)->flags & PF_EXITING)) &&
>  		    !entity_eligible(cfs_rq, se)) {
>  			if (cfs_rq->next == se)
>  				cfs_rq->next = NULL;
> @@ -5495,14 +5496,14 @@ pick_next_entity(struct rq *rq, struct c
>  	}
> 
>  	struct sched_entity *se = pick_eevdf(cfs_rq);
> -	if (se->sched_delayed) {
> +	while (se && se->sched_delayed) {
>  		dequeue_entities(rq, se, DEQUEUE_SLEEP | DEQUEUE_DELAYED);
>  		SCHED_WARN_ON(se->sched_delayed);
>  		SCHED_WARN_ON(se->on_rq);
>  		if (sched_feat(DELAY_ZERO) && se->vlag > 0)
>  			se->vlag = 0;
> 
> -		return NULL;
> +		se = pick_eevdf(cfs_rq);
>  	}
>  	return se;
>  }
>