Re: [PATCH] sched/fair: Prevent cfs_rq from being unthrottled with zero runtime_remaining

[Aaron Lu <ziqianlu@bytedance.com>]

On Tue, Sep 30, 2025 at 07:07:17PM +0800, Aaron Lu wrote:
> On Tue, Sep 30, 2025 at 02:28:16PM +0530, K Prateek Nayak wrote:
> > Hello Aaron,
> > 
> > On 9/30/2025 1:26 PM, Aaron Lu wrote:
> > > On Mon, Sep 29, 2025 at 03:04:03PM +0530, K Prateek Nayak wrote:
> > > ... ...
> > >> Can we instead do a check_enqueue_throttle() in enqueue_throttled_task()
> > >> if we find cfs_rq->throttled_limbo_list to be empty?
> > >>
> > >> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > >> index 18a30ae35441..fd2d4dad9c27 100644
> > >> --- a/kernel/sched/fair.c
> > >> +++ b/kernel/sched/fair.c
> > >> @@ -5872,6 +5872,8 @@ static bool enqueue_throttled_task(struct task_struct *p)
> > >>  	 */
> > >>  	if (throttled_hierarchy(cfs_rq) &&
> > >>  	    !task_current_donor(rq_of(cfs_rq), p)) {
> > >                 /*
> > >                  * Make sure to throttle this cfs_rq or it can be unthrottled
> > >                  * with no runtime_remaining and gets throttled again on its
> > >                  * unthrottle path.
> > >                  */
> > >> +		if (list_empty(&cfs_rq->throttled_limbo_list))
> > >> +			check_enqueue_throttle(cfs_rq);
> > > 
> > > BTW, do you think a comment is needed? Something like the above, not
> > > sure if it's too redundant though, feel free to let me know your
> > > thoughts, thanks.
> > 
> > Now that I'm looking at it again, I think we should actually do a:
> > 
> >     for_each_entity(se)
> >         check_enqueue_throttle(cfs_rq_of(se));
> > 
> > The reason being, we can have:
> > 
> >     root -> A (throttled) -> B -> C
> > 
> > Consider B has runtime_remaining = 0, and subsequently a throttled task
> > is queued onto C. Ideally, we should start the B/W timer for B at that
> > point but we bail out after queuing it on C. Thoughts?
> 
> Yes agree the B/W timer should also be considered.

On another thought, do we really need care about B/W timer for B?

I mean, when C is unthrottled and gets enqueued on B,
check_enqueue_throttle() will do the right thing for B so I don't
think we need to do this hierarchy check_enqueue_throttle() here.

I think the only difference with your suggestion and my patch is, with
your suggestion, it's possible for a runtime_enabled cfs_rq to reach
tg_unthrottle_up() with runtime_remaining equals to 0 but since it
doesn't have any tasks in its limbo list, it will not do any enqueue so
won't possibly trigger throttle there, so it's still fine. i.e. I think
your original suggestion works.