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: <CAJuCfpG=bADguJWdxV4ZhTze9fmKP2bhsbf0xzbd06Fhr4_U5w@mail.gmail.com>
 <20221012062034.486-1-hdanton@sina.com>
In-Reply-To: <20221012062034.486-1-hdanton@sina.com>
From:   Suren Baghdasaryan <surenb@google.com>
Date:   Wed, 12 Oct 2022 08:40:17 -0700
Message-ID: <CAJuCfpFsx6C1tRnAXVPz3Nn=odqDuQwgrEr5qVWWYh7t6Tpukw@mail.gmail.com>
Subject: Re: PSI idle-shutoff
To:     Hillf Danton <hdanton@sina.com>
Cc:     Pavan Kondeti <quic_pkondeti@quicinc.com>,
        Johannes Weiner <hannes@cmpxchg.org>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, quic_charante@quicinc.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Oct 11, 2022 at 11:20 PM Hillf Danton <hdanton@sina.com> wrote:
>
> On 11 Oct 2022 10:11:58 -0700 Suren Baghdasaryan <surenb@google.com>
> >On Tue, Oct 11, 2022 at 4:38 AM Hillf Danton <hdanton@sina.com> wrote:
> >>
> >> Given activities on remote CPUs, can you specify what prevents psi_avgs_work
> >> from being scheduled on remote CPUs if for example the local CPU has been
> >> idle for a second?
> >
> > I'm not a scheduler expert but I can imagine some work that finished
> > running on a big core A and generated some activity since the last
> > time psi_avgs_work executed.  With no other activity the next
> > psi_avgs_work could be scheduled on a small core B to conserve power.
>
> Given core A and B, nothing prevents.
>
> > There might be other cases involving cpuset limitation changes or cpu
> > offlining but I didn't think too hard about these. The bottom line, I
> > don't think we should be designing mechanisms which rely on
> > assumptions about how tasks will be scheduled. Even if these
>
> The tasks here makes me guess that we are on different pages - scheduling
> work has little to do with how tasks are scheduled, and is no more than
> queuing work on the system_wq in the case of psi_avgs_work,

I must have misunderstood your question then. My original concern was
that in the above example your suggested patch would not reschedule
psi_avgs_work to aggregate the activity recorded from core A. Easily
fixable but looks like a simpler approach is possible.

>
> > assumptions are correct today they might change in the future and
> > things will break in unexpected places.
>
> with nothing assumed.
>