Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230518013414.3053254-1-joshdon@google.com> <20230518013414.3053254-2-joshdon@google.com>
 <ZG0z1enh3TNbsxQh@slm.duckdns.org>
In-Reply-To: <ZG0z1enh3TNbsxQh@slm.duckdns.org>
From:   Josh Don <joshdon@google.com>
Date:   Wed, 24 May 2023 17:06:08 -0700
Message-ID: <CABk29Nvh6jxv3cWampfjiz57=vvQVX8ThtaJPXF=vc3aqme6YQ@mail.gmail.com>
Subject: Re: [PATCH 2/2] sched: add throttled time stat for throttled children
To:     Tejun Heo <tj@kernel.org>
Cc:     Ingo Molnar <mingo@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Juri Lelli <juri.lelli@redhat.com>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        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>,
        linux-kernel@vger.kernel.org,
        Xiangling Kong <xiangling@google.com>,
        Johannes Weiner <hannes@cmpxchg.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hey Tejun,

On Tue, May 23, 2023 at 2:44=E2=80=AFPM Tejun Heo <tj@kernel.org> wrote:
>
> Hello,
>
> (cc'ing Johannes)
>
> On Wed, May 17, 2023 at 06:34:14PM -0700, Josh Don wrote:
> > We currently export the total throttled time for cgroups that are given
> > a bandwidth limit. This patch extends this accounting to also account
> > the total time that each children cgroup has been throttled.
> >
> > This is useful to understand the degree to which children have been
> > affected by the throttling control. Children which are not runnable
> > during the entire throttled period, for example, will not show any
> > self-throttling time during this period.
> ...
> > @@ -11204,20 +11217,24 @@ static int cpu_extra_stat_show(struct seq_fil=
e *sf,
> >       {
> >               struct task_group *tg =3D css_tg(css);
> >               struct cfs_bandwidth *cfs_b =3D &tg->cfs_bandwidth;
> > -             u64 throttled_usec, burst_usec;
> > +             u64 throttled_usec, burst_usec, throttled_self_usec;
> >
> >               throttled_usec =3D cfs_b->throttled_time;
> >               do_div(throttled_usec, NSEC_PER_USEC);
> > +             throttled_self_usec =3D throttled_time_self(tg);
> > +             do_div(throttled_self_usec, NSEC_PER_USEC);
> >               burst_usec =3D cfs_b->burst_time;
> >               do_div(burst_usec, NSEC_PER_USEC);
> >
> >               seq_printf(sf, "nr_periods %d\n"
> >                          "nr_throttled %d\n"
> >                          "throttled_usec %llu\n"
> > +                        "throttled_self_usec %llu\n"
>
> This is fine in principle but I think it'd be better to keep it consisten=
t
> with how non-hierarchical events are in memory.events.local. ie. Can we
> please add cpu.stat.local instead of adding the _self key to cpu.stat?

It seemed unfortunate to split it out into a separate interface, since
there wouldn't be much to put there, but I don't have a strong
objection to this. I'll go ahead and make that change for v2.

Best,
Josh