Re: [PATCH V3] sched/deadline: Update rq_clock of later_rq when pushing a task

[Juri Lelli]

On 20/07/18 17:36, Daniel Bristot de Oliveira wrote:
> On 07/20/2018 02:53 PM, Juri Lelli wrote:
> > On 20/07/18 14:48, Peter Zijlstra wrote:
> >> On Fri, Jul 20, 2018 at 02:46:15PM +0200, Peter Zijlstra wrote:
> >>> On Fri, Jul 20, 2018 at 11:16:30AM +0200, Daniel Bristot de Oliveira wrote:
> >>>> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> >>>> index fbfc3f1d368a..8b50eea4b607 100644
> >>>> --- a/kernel/sched/deadline.c
> >>>> +++ b/kernel/sched/deadline.c
> >>>> @@ -2090,8 +2090,14 @@ static int push_dl_task(struct rq *rq)
> >>>>  	sub_rq_bw(&next_task->dl, &rq->dl);
> >>>>  	set_task_cpu(next_task, later_rq->cpu);
> >>>>  	add_rq_bw(&next_task->dl, &later_rq->dl);
> >>>> +
> >>>> +	/*
> >>>> +	 * Update the later_rq clock here, because the clock is used
> >>>> +	 * by the cpufreq_update_util() inside __add_running_bw().
> >>>> +	 */
> >>>> +	update_rq_clock(later_rq);
> >>>>  	add_running_bw(&next_task->dl, &later_rq->dl);
> >>>> -	activate_task(later_rq, next_task, 0);
> >>>> +	activate_task(later_rq, next_task, ENQUEUE_NOCLOCK);
> >>>>  	ret = 1;
> >>>>  
> >>>>  	resched_curr(later_rq);
> >>>
> >>> Why isn't push_rt_task() affected by the very same issue?
> >>
> >> Aah, I see, its the add_running_bw() think; for which RT doesn't have a
> >> counter-part.
> > 
> > Right, but doesn't enqueue_top_rt_rq end-up being called by activate_
> > task on lowest_rq? Mmm.
> 
> 
> AFAICS we have:
> 
> push_rt_task() {
> 	activate_task() {
> 		enqueue_task(,,(flags=0)) {
> 		        if (!(flags & ENQUEUE_NOCLOCK))
> 	        	        update_rq_clock(rq);
> 			enqueue_task_rt() {
> 				enqueue_rt_entity() {
> 					enqueue_top_rt_rq();
> 			}
> 		}
> 	}
> }
> 
> So we will have the clock updated already...
> 
> Am I missing something?

Ah, indeed.