Re: [PATCH sched_ext/for-6.12] sched_ext: Handle cases where pick_task_scx() is called without preceding balance_scx()

[Tejun Heo <tj@kernel.org>]

On Thu, Sep 05, 2024 at 03:17:13PM -1000, Tejun Heo wrote:
> Did a bit of testing and it seems like it's mostly coming from delayed
> dequeue handling. pick_next_entity() does this:
> 
> 	struct sched_entity *se = pick_eevdf(cfs_rq);
> 	if (se->sched_delayed) {
> 		dequeue_entities(rq, se, DEQUEUE_SLEEP | DEQUEUE_DELAYED);
> 		SCHED_WARN_ON(se->sched_delayed);
> 		SCHED_WARN_ON(se->on_rq);
> 		return NULL;
> 	}
> 
> rq->cfs.nr_running includes the number of delay dequeued tasks which aren't
> really runnable, so it seems like balance_fair() saying yes and
> pick_next_entity() then hitting a delayed task. Maybe the solution is
> tracking the number of delayed ones and subtracting that from nr_running?
> I'm trying that but can't get the delayed count straight for some reason.

Backported http://lkml.kernel.org/r/Ztpjt5Pz9pJliblL@slm.duckdns.org to
v6.10 and it doesn't trigger (at least not easily) while the warning
triggers immediately on the current tip/sched/core. It does look like the
problem is delayed dequeue.

Thanks.

-- 
tejun