From: "Gautham R. Shenoy" <gautham.shenoy@amd.com>
To: Chen Yu <yu.c.chen@intel.com>
Cc: Aaron Lu <aaron.lu@intel.com>,
Peter Zijlstra <peterz@infradead.org>,
Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
Ingo Molnar <mingo@redhat.com>,
Vincent Guittot <vincent.guittot@linaro.org>,
Juri Lelli <juri.lelli@redhat.com>,
Tim Chen <tim.c.chen@intel.com>,
Dietmar Eggemann <dietmar.eggemann@arm.com>,
Steven Rostedt <rostedt@goodmis.org>,
Ben Segall <bsegall@google.com>, Mel Gorman <mgorman@suse.de>,
Daniel Bristot de Oliveira <bristot@redhat.com>,
Valentin Schneider <vschneid@redhat.com>,
K Prateek Nayak <kprateek.nayak@amd.com>,
linux-kernel@vger.kernel.org
Subject: Re: [RFC PATCH 2/2] sched/fair: skip the cache hot CPU in select_idle_cpu()
Date: Fri, 15 Sep 2023 20:48:14 +0530 [thread overview]
Message-ID: <ZQR1tovjkTZ0VcP6@BLR-5CG11610CF.amd.com> (raw)
In-Reply-To: <ZQL39qpKQncZWs94@chenyu5-mobl2.ccr.corp.intel.com>
Hello Chen Yu,
On Thu, Sep 14, 2023 at 08:09:26PM +0800, Chen Yu wrote:
[..snip..]
> >
> > So despite "reserving" the CPU for p1, which is likely to have its
> > data still hot in the case, we would have scheduled p1', thus
> > defeating the whole purpose of reservation.
> >
>
> I see. So you mean, although we reserve the CPU for the wakee,
> the wakee might not choose its previous CPU, which is against our
> goal.
Yes, but only because some other task could have run on the previous
CPU. That other task could be something that was woken up on that CPU
due to:
1) wake-affine choosing that CPU
2) newidle-balance pulling the other task on that CPU
3) !wake-affine && that CPU was also the other task's previous CPU
It could also be due to this wakee task being woken up on the waker
CPU due to wake-affine.
>
> The reason to prevent the wakee choosing its previous CPU could be:
> 1. wake_affine() choose the waker's CPU rather the wakee's previous CPU, or
> 2. the wakee's CPU has already been taken by someone else, via newidle_balance().
>
> For 1, I think Prateek has expressed the concern. One mitigation method could be
> that, we give penalty to that wakee, if it decides not to choose its previous CPU:
We would be penalizing the task for something that the scheduler
decides :-)
As you point out below, in the presence of the WF_SYNC flag,
wake_affine_idle() prefer the waker CPU over the previous CPU when
they are on different LLCs and when the waker is the only task.
This strategy makes sense for two reasons:
1) The wakee may be consuming the data produced by the waker.
2) Since the wakeup will happen on the local CPU, there is no risk of
task-stacking, exactly what your SIS_CURRENT patchset was
attempting.
But this strategy would also result in increased task-migration. Which
both Mattieu and you have found is not so beneficial for workloads
such as hackbench. Is it only because task's data is still hot in the
previous CPU's cache ? Or is there more to it ?
It would be good to confirm if this is why lower migration is better
for these kinds of workloads.
>
> "
> new_cpu = select_idle_sibling(p, prev_cpu, new_cpu);
> if (new_cpu != prev_cpu)
> p->burst_sleep_avg >>= 1;
> So the duration of reservation could be shrinked.
> "
>
> For 2, maybe inhit the newidle balance, something in my mind:
>
>
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -12022,6 +12022,7 @@ static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> u64 t0, t1, curr_cost = 0;
> struct sched_domain *sd;
> int pulled_task = 0;
> + bool cache_hot = false;
>
> update_misfit_status(NULL, this_rq);
>
> @@ -12055,8 +12056,19 @@ static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> rcu_read_lock();
> sd = rcu_dereference_check_sched_domain(this_rq->sd);
>
> + if (sched_feat(SIS_CACHE)) {
> + s64 delta = this_rq->cache_hot_timeout - sched_clock_cpu(this_cpu);
> +
> + /*
> + * If a short time later, a short sleeping task will be woken up
> + * on this idle CPU, do not launch the newidle balance.
> + */
> + if (delta > 0 && delta < this_rq->max_idle_balance_cost)
> + cache_hot = true;
> + }
> +
> if (!READ_ONCE(this_rq->rd->overload) ||
> - (sd && this_rq->avg_idle < sd->max_newidle_lb_cost)) {
> + (sd && this_rq->avg_idle < sd->max_newidle_lb_cost) || cache_hot) {
>
> if (sd)
> update_next_balance(sd, &next_balance);
If the benefit that the workload obtains is really due to the data
being hot near its previous CPU, then this seems a sensible thing to
do.
It would be good to confirm this. Let me get some IBS data for
hackbench which is the workload which likes a sticky wakeup.
--
Thanks and Regards
gautham.
>
>
> > To be honest, this isn't so bad, because we have been able to avoid a
> > migration in this case.
> >
> > >
> > > Or do you mean, in select_idle_cpu(), we will re-check p1's previous
> > > CPU but it is skipped due to cache-hot?
> >
> > I had originally thought about this, but then as you pointed out we
> > have an opportunity to pick the previous cpu in the early checks
> > inside select_idle_sibling().
> >
> > >
> > > > Have you considered recording p1's identity in the
> > > > rq->cache_hot_sleeper so that in select_task_rq_fair(), we can simply
> > > > return the previous CPU if it is cache hot and the wakee is
> > > > rq->cache_hot_sleeper, thus avoiding the whole select_idle_sibling
> > > > scan.
> > > >
> > >
> > > Yes this seems to be donable, and one problem would be, if there are
> > > more than 2 dequeued tasks prefer the same (previous) CPU, which task
> > > should be the rq->cache_hot_sleeper. And per Mathieu's feedback[1], we
> > > want to deal with multiple dequeued tasks. If we record all of them,
> > > it might be costly.
> >
> > If there are multiple dequeued tasks, then it doesn't make sense to
> > record the identity of the tasks. However, we need the bail out to be
> > much earlier, in select_task_rq_fair(), perhaps even before the
> > want_affine() checks.
> >
> > After all, if the previous CPU is idle, and its cache_hot_timeout
> > hasn't expired, and if the wakee's sleep duration is less than the
> > cache_hot_timeout, why don't we just pick it here and be done with it?
> >
>
> Yes we can return the previous CPU earlier, one concern is that, should
> we honor WF_SYNC flag or not, because in wake_affine_idle(), WF_SYNC
> seems to have a higher priority than available_idle_cpu(prev_cpu). Say,
> if the current CPU has 1 running task, and the previous CPU is idle,
> wake_affine_idle() still prefers the current CPU.
>
> thanks,
> Chenyu
next prev parent reply other threads:[~2023-09-15 15:18 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-09-11 2:49 [RFC PATCH 0/2] Makes it easier for the wakee to choose previous CPU Chen Yu
2023-09-11 2:49 ` [RFC PATCH 1/2] sched/fair: Record the average sleep time of a task Chen Yu
2023-09-11 2:50 ` [RFC PATCH 2/2] sched/fair: skip the cache hot CPU in select_idle_cpu() Chen Yu
2023-09-11 7:26 ` Aaron Lu
2023-09-11 8:40 ` Chen Yu
2023-09-13 6:22 ` Gautham R. Shenoy
2023-09-13 7:25 ` Chen Yu
2023-09-14 7:06 ` Gautham R. Shenoy
2023-09-14 12:09 ` Chen Yu
2023-09-15 15:18 ` Gautham R. Shenoy [this message]
2023-09-19 9:01 ` Chen Yu
2023-09-11 8:29 ` K Prateek Nayak
2023-09-11 10:19 ` Chen Yu
2023-09-12 3:05 ` K Prateek Nayak
2023-09-12 12:32 ` Chen Yu
2023-09-12 14:26 ` K Prateek Nayak
2023-09-13 2:57 ` Chen Yu
2023-09-14 4:13 ` K Prateek Nayak
2023-09-14 11:01 ` Chen Yu
2023-09-15 3:21 ` K Prateek Nayak
2023-09-12 9:39 ` Mike Galbraith
2023-09-12 14:51 ` Chen Yu
2023-09-12 6:32 ` Mike Galbraith
2023-09-11 15:26 ` Mathieu Desnoyers
2023-09-11 15:43 ` Mathieu Desnoyers
2023-09-12 11:53 ` Chen Yu
2023-09-12 14:06 ` Mathieu Desnoyers
2023-09-12 14:14 ` Chen Yu
2023-09-12 15:18 ` Mathieu Desnoyers
2023-09-13 3:02 ` Chen Yu
2023-09-20 12:34 ` Chen Yu
2023-09-14 5:30 ` K Prateek Nayak
2023-09-14 10:43 ` Chen Yu
2023-09-15 3:37 ` K Prateek Nayak
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ZQR1tovjkTZ0VcP6@BLR-5CG11610CF.amd.com \
--to=gautham.shenoy@amd.com \
--cc=aaron.lu@intel.com \
--cc=bristot@redhat.com \
--cc=bsegall@google.com \
--cc=dietmar.eggemann@arm.com \
--cc=juri.lelli@redhat.com \
--cc=kprateek.nayak@amd.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mathieu.desnoyers@efficios.com \
--cc=mgorman@suse.de \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=tim.c.chen@intel.com \
--cc=vincent.guittot@linaro.org \
--cc=vschneid@redhat.com \
--cc=yu.c.chen@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®