linux-kernel - Re: [RFC PATCH] sched/fair: Prevent from cpufreq not being updated when delayed-task is iowait

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAB8ipk-CcirMttnxDUj_VGfgd3Z9SgeZf+jmLTEV7YBabvVP3w@mail.gmail.com>
Date: Thu, 27 Feb 2025 10:12:21 +0800
From: Xuewen Yan <xuewen.yan94@...il.com>
To: Christian Loehle <christian.loehle@....com>
Cc: Xuewen Yan <xuewen.yan@...soc.com>, peterz@...radead.org, mingo@...hat.com, 
	juri.lelli@...hat.com, vincent.guittot@...aro.org, dietmar.eggemann@....com, 
	Hongyan Xia <Hongyan.Xia2@....com>, Pierre Gondois <pierre.gondois@....com>, 
	Luis Machado <luis.machado@....com>, rostedt@...dmis.org, bsegall@...gle.com, 
	mgorman@...e.de, vschneid@...hat.com, ke.wang@...soc.com, di.shen@...soc.com, 
	linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH] sched/fair: Prevent from cpufreq not being updated
 when delayed-task is iowait

On Wed, Feb 26, 2025 at 8:08 PM Christian Loehle
<christian.loehle@....com> wrote:
>
> On 2/26/25 11:43, Xuewen Yan wrote:
> > Because the sched-delayed task maybe in io-wait state,
> > so we should place the requeue_delayed_entity() after the
> > cpufreq_update_util(), to prevent not boosting iowait cpufreq
> > before return.
> >
> > Signed-off-by: Xuewen Yan <xuewen.yan@...soc.com>
> > ---
> >  kernel/sched/fair.c | 10 +++++-----
> >  1 file changed, 5 insertions(+), 5 deletions(-)
> >
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index 2d6d5582c3e9..040674734128 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -6931,11 +6931,6 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
> >       if (!(p->se.sched_delayed && (task_on_rq_migrating(p) || (flags & ENQUEUE_RESTORE))))
> >               util_est_enqueue(&rq->cfs, p);
> >
> > -     if (flags & ENQUEUE_DELAYED) {
> > -             requeue_delayed_entity(se);
> > -             return;
> > -     }
> > -
> >       /*
> >        * If in_iowait is set, the code below may not trigger any cpufreq
> >        * utilization updates, so do it here explicitly with the IOWAIT flag
> > @@ -6944,6 +6939,11 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
> >       if (p->in_iowait)
> >               cpufreq_update_util(rq, SCHED_CPUFREQ_IOWAIT);
> >
> > +     if (flags & ENQUEUE_DELAYED) {
> > +             requeue_delayed_entity(se);
> > +             return;
> > +     }
> > +
> >       if (task_new && se->sched_delayed)
> >               h_nr_runnable = 0;
> >
>
> I understand that iowait cpufreq update isn't happening now (and that's a bug),
> but if we reorder we may call cpufreq_update_util(rq, SCHED_CPUFREQ_IOWAIT)
> followed by the cpufreq_update_util() in update_load_avg() of
> requeue_delayed_entity()
>         update_load_avg()
>                 cpufreq_update_util()
>
> and the latter will likely be dropped by the governor, so the update
> won't include util of the (re)-enqueuing task, right?

emmm, util_est has already included the util of requeue-task.
However, the cfs_rq->avg.runnable_avg would indeed be slightly different.
If we want to include both iowait and runnable_avg, perhaps we should
add the iowait check to the update_load_avg() function, but this would
make the code more complex.

>
> I'll give it some more thought.