Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1680720 > unrolled thread

Re: [PATCH v2] sched/pelt: fix false running accounting

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v2] sched/pelt: fix false running accounting Peter Zijlstra <peterz@infradead.org> - 2017-07-04 09:30 +0200
    Re: [PATCH v2] sched/pelt: fix false running accounting Vincent Guittot <vincent.guittot@linaro.org> - 2017-07-04 09:40 +0200
    Re: [PATCH v2] sched/pelt: fix false running accounting Peter Zijlstra <peterz@infradead.org> - 2017-07-04 10:40 +0200
      Re: [PATCH v2] sched/pelt: fix false running accounting Vincent Guittot <vincent.guittot@linaro.org> - 2017-07-04 11:20 +0200
        Re: [PATCH v2] sched/pelt: fix false running accounting Peter Zijlstra <peterz@infradead.org> - 2017-07-04 11:50 +0200
          Re: [PATCH v2] sched/pelt: fix false running accounting Vincent Guittot <vincent.guittot@linaro.org> - 2017-07-04 12:00 +0200
            Re: [PATCH v2] sched/pelt: fix false running accounting Peter Zijlstra <peterz@infradead.org> - 2017-07-04 13:10 +0200

#1680720 — Re: [PATCH v2] sched/pelt: fix false running accounting

On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote:
> The running state is a subset of runnable state which means that running
> can't be set if runnable (weight) is cleared. There are corner cases
> where the current sched_entity has been already dequeued but cfs_rq->curr
> has not been updated yet and still points to the dequeued sched_entity.
> If ___update_load_avg is called at that time, weight will be 0 and running
> will be set which is not possible.
> 
> This case happens during pick_next_task_fair() when a cfs_rq becomes idles.
> The current sched_entity has been dequeued so se->on_rq is cleared and
> cfs_rq->weight is null. But cfs_rq->curr still points to se (it will be
> cleared when picking the idle thread). Because the cfs_rq becomes idle,
> idle_balance() is called and ends up to call update_blocked_averages()
> with these wrong running and runnable states.
> 
> Add a test in ___update_load_avg to correct the running state in this case.

Cute, however did you find that ?

[toc] | [next] | [standalone]

#1680724

On 4 July 2017 at 09:27, Peter Zijlstra <peterz@infradead.org> wrote:
> On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote:
>> The running state is a subset of runnable state which means that running
>> can't be set if runnable (weight) is cleared. There are corner cases
>> where the current sched_entity has been already dequeued but cfs_rq->curr
>> has not been updated yet and still points to the dequeued sched_entity.
>> If ___update_load_avg is called at that time, weight will be 0 and running
>> will be set which is not possible.
>>
>> This case happens during pick_next_task_fair() when a cfs_rq becomes idles.
>> The current sched_entity has been dequeued so se->on_rq is cleared and
>> cfs_rq->weight is null. But cfs_rq->curr still points to se (it will be
>> cleared when picking the idle thread). Because the cfs_rq becomes idle,
>> idle_balance() is called and ends up to call update_blocked_averages()
>> with these wrong running and runnable states.
>>
>> Add a test in ___update_load_avg to correct the running state in this case.
>
> Cute, however did you find that ?

In fact, while rebasing and running more tests on my patch "update
scale invariance of PELT" that changes how to scale the load and
utilization, I have seen that sometimes the utilization was increasing
but not the load when CPU was going into idle state because the
stolen_idle time was applied as idle time for load but running time
for utilization. This patch has highlighted the problem.

[toc] | [prev] | [next] | [standalone]

#1680760

On Tue, Jul 04, 2017 at 09:27:07AM +0200, Peter Zijlstra wrote:
> On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote:
> > The running state is a subset of runnable state which means that running
> > can't be set if runnable (weight) is cleared. There are corner cases
> > where the current sched_entity has been already dequeued but cfs_rq->curr
> > has not been updated yet and still points to the dequeued sched_entity.
> > If ___update_load_avg is called at that time, weight will be 0 and running
> > will be set which is not possible.
> > 
> > This case happens during pick_next_task_fair() when a cfs_rq becomes idles.
> > The current sched_entity has been dequeued so se->on_rq is cleared and
> > cfs_rq->weight is null. But cfs_rq->curr still points to se (it will be
> > cleared when picking the idle thread). Because the cfs_rq becomes idle,
> > idle_balance() is called and ends up to call update_blocked_averages()
> > with these wrong running and runnable states.
> > 
> > Add a test in ___update_load_avg to correct the running state in this case.
> 
> Cute, however did you find that ?

Hmm,.. could you give a little more detail?

Because if ->on_rq=0, we'll have done dequeue_task() which will have
done update_curr() with ->on_rq, weight and ->running consistently.

Then the above, inconsistent update should not happen, because delta=0.

[toc] | [prev] | [next] | [standalone]

#1680790

On 4 July 2017 at 10:34, Peter Zijlstra <peterz@infradead.org> wrote:
> On Tue, Jul 04, 2017 at 09:27:07AM +0200, Peter Zijlstra wrote:
>> On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote:
>> > The running state is a subset of runnable state which means that running
>> > can't be set if runnable (weight) is cleared. There are corner cases
>> > where the current sched_entity has been already dequeued but cfs_rq->curr
>> > has not been updated yet and still points to the dequeued sched_entity.
>> > If ___update_load_avg is called at that time, weight will be 0 and running
>> > will be set which is not possible.
>> >
>> > This case happens during pick_next_task_fair() when a cfs_rq becomes idles.
>> > The current sched_entity has been dequeued so se->on_rq is cleared and
>> > cfs_rq->weight is null. But cfs_rq->curr still points to se (it will be
>> > cleared when picking the idle thread). Because the cfs_rq becomes idle,
>> > idle_balance() is called and ends up to call update_blocked_averages()
>> > with these wrong running and runnable states.
>> >
>> > Add a test in ___update_load_avg to correct the running state in this case.
>>
>> Cute, however did you find that ?
>
> Hmm,.. could you give a little more detail?
>
> Because if ->on_rq=0, we'll have done dequeue_task() which will have
> done update_curr() with ->on_rq, weight and ->running consistently.
>
> Then the above, inconsistent update should not happen, because delta=0.

In fact, the delta between dequeue_entity_load_avg() and
update_blocked_averages() is not 0 on my platform (hikey) but can be
longer than 60us (at lowest frequency with only 1 task group level)

[toc] | [prev] | [next] | [standalone]

#1680817

On Tue, Jul 04, 2017 at 11:12:34AM +0200, Vincent Guittot wrote:
> On 4 July 2017 at 10:34, Peter Zijlstra <peterz@infradead.org> wrote:
> > On Tue, Jul 04, 2017 at 09:27:07AM +0200, Peter Zijlstra wrote:
> >> On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote:
> >> > The running state is a subset of runnable state which means that running
> >> > can't be set if runnable (weight) is cleared. There are corner cases
> >> > where the current sched_entity has been already dequeued but cfs_rq->curr
> >> > has not been updated yet and still points to the dequeued sched_entity.
> >> > If ___update_load_avg is called at that time, weight will be 0 and running
> >> > will be set which is not possible.
> >> >
> >> > This case happens during pick_next_task_fair() when a cfs_rq becomes idles.
> >> > The current sched_entity has been dequeued so se->on_rq is cleared and
> >> > cfs_rq->weight is null. But cfs_rq->curr still points to se (it will be
> >> > cleared when picking the idle thread). Because the cfs_rq becomes idle,
> >> > idle_balance() is called and ends up to call update_blocked_averages()
> >> > with these wrong running and runnable states.
> >> >
> >> > Add a test in ___update_load_avg to correct the running state in this case.
> >>
> >> Cute, however did you find that ?
> >
> > Hmm,.. could you give a little more detail?
> >
> > Because if ->on_rq=0, we'll have done dequeue_task() which will have
> > done update_curr() with ->on_rq, weight and ->running consistently.
> >
> > Then the above, inconsistent update should not happen, because delta=0.
> 
> In fact, the delta between dequeue_entity_load_avg() and
> update_blocked_averages() is not 0 on my platform (hikey) but can be
> longer than 60us (at lowest frequency with only 1 task group level)

But but but, how can that happen? Should it not all be under the same
rq->lock and thus have only a single update_rq_clock() and thus be at
the same 'instant' ?

[toc] | [prev] | [next] | [standalone]

#1680822

On 4 July 2017 at 11:44, Peter Zijlstra <peterz@infradead.org> wrote:
> On Tue, Jul 04, 2017 at 11:12:34AM +0200, Vincent Guittot wrote:
>> On 4 July 2017 at 10:34, Peter Zijlstra <peterz@infradead.org> wrote:
>> > On Tue, Jul 04, 2017 at 09:27:07AM +0200, Peter Zijlstra wrote:
>> >> On Sat, Jul 01, 2017 at 07:06:13AM +0200, Vincent Guittot wrote:
>> >> > The running state is a subset of runnable state which means that running
>> >> > can't be set if runnable (weight) is cleared. There are corner cases
>> >> > where the current sched_entity has been already dequeued but cfs_rq->curr
>> >> > has not been updated yet and still points to the dequeued sched_entity.
>> >> > If ___update_load_avg is called at that time, weight will be 0 and running
>> >> > will be set which is not possible.
>> >> >
>> >> > This case happens during pick_next_task_fair() when a cfs_rq becomes idles.
>> >> > The current sched_entity has been dequeued so se->on_rq is cleared and
>> >> > cfs_rq->weight is null. But cfs_rq->curr still points to se (it will be
>> >> > cleared when picking the idle thread). Because the cfs_rq becomes idle,
>> >> > idle_balance() is called and ends up to call update_blocked_averages()
>> >> > with these wrong running and runnable states.
>> >> >
>> >> > Add a test in ___update_load_avg to correct the running state in this case.
>> >>
>> >> Cute, however did you find that ?
>> >
>> > Hmm,.. could you give a little more detail?
>> >
>> > Because if ->on_rq=0, we'll have done dequeue_task() which will have
>> > done update_curr() with ->on_rq, weight and ->running consistently.
>> >
>> > Then the above, inconsistent update should not happen, because delta=0.
>>
>> In fact, the delta between dequeue_entity_load_avg() and
>> update_blocked_averages() is not 0 on my platform (hikey) but can be
>> longer than 60us (at lowest frequency with only 1 task group level)
>
> But but but, how can that happen? Should it not all be under the same
> rq->lock and thus have only a single update_rq_clock() and thus be at
> the same 'instant' ?

idle_balance() unlock rq->lock before calling  update_blocked_averages
And update_blocked_averages() starts by calling update_rq_clock()

[toc] | [prev] | [next] | [standalone]

#1680880

On Tue, Jul 04, 2017 at 11:57:12AM +0200, Vincent Guittot wrote:
> On 4 July 2017 at 11:44, Peter Zijlstra <peterz@infradead.org> wrote:

> > But but but, how can that happen? Should it not all be under the same
> > rq->lock and thus have only a single update_rq_clock() and thus be at
> > the same 'instant' ?
> 
> idle_balance() unlock rq->lock before calling  update_blocked_averages
> And update_blocked_averages() starts by calling update_rq_clock()

Ah indeed. Might want to clarify that point.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web