Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210617212654.1529125-1-surenb@google.com> <YNHr81D/fPD2Q8kM@hirez.programming.kicks-ass.net>
 <YNH8x9HRyfvF53Pl@cmpxchg.org>
In-Reply-To: <YNH8x9HRyfvF53Pl@cmpxchg.org>
From:   Suren Baghdasaryan <surenb@google.com>
Date:   Tue, 22 Jun 2021 08:48:52 -0700
Message-ID: <CAJuCfpEL+Sp8e7z=JqLNuu5UKZCmCnx0nPugxXnFYd0w=7qbBQ@mail.gmail.com>
Subject: Re: [PATCH 1/1] psi: stop relying on timer_pending for poll_work rescheduling
To:     Johannes Weiner <hannes@cmpxchg.org>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>,
        Juri Lelli <juri.lelli@redhat.com>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        Dietmar Eggemann <dietmar.eggemann@arm.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Benjamin Segall <bsegall@google.com>,
        Mel Gorman <mgorman@suse.de>,
        Daniel Bristot de Oliveira <bristot@redhat.com>,
        matthias.bgg@gmail.com, Minchan Kim <minchan@google.com>,
        Tim Murray <timmurray@google.com>,
        YT Chang <yt.chang@mediatek.com>,
        =?UTF-8?B?V2VuanUgWHUgKOiuuOaWh+S4vik=?= <wenju.xu@mediatek.com>,
        =?UTF-8?B?Sm9uYXRoYW4gSk1DaGVuICjpmbPlrrbmmI4p?= 
        <jonathan.jmchen@mediatek.com>,
        LKML <linux-kernel@vger.kernel.org>,
        linux-arm-kernel@lists.infradead.org,
        linux-mediatek@lists.infradead.org,
        kernel-team <kernel-team@android.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Jun 22, 2021 at 8:08 AM Johannes Weiner <hannes@cmpxchg.org> wrote:
>
> On Tue, Jun 22, 2021 at 03:56:03PM +0200, Peter Zijlstra wrote:
> > On Thu, Jun 17, 2021 at 02:26:54PM -0700, Suren Baghdasaryan wrote:
> > > Fixes: 461daba06bdc ("psi: eliminate kthread_worker from psi trigger scheduling mechanism")
> > > Reported-by: Kathleen Chang <yt.chang@mediatek.com>
> > > Reported-by: Wenju Xu <wenju.xu@mediatek.com>
> > > Reported-by: Jonathan Chen <jonathan.jmchen@mediatek.com>
> > > Signed-off-by: Suren Baghdasaryan <surenb@google.com>
> >
> > Johannes?
>
> Looks generally good to me, I'd just want to get to the bottom of the
> memory ordering before acking...
>
> > > -/* Schedule polling if it's not already scheduled. */
> > > -static void psi_schedule_poll_work(struct psi_group *group, unsigned long delay)
> > > +/* Schedule polling if it's not already scheduled or forced. */
> > > +static void psi_schedule_poll_work(struct psi_group *group, unsigned long delay,
> > > +                              bool force)
> > >  {
> > >     struct task_struct *task;
> > >
> > > -   /*
> > > -    * Do not reschedule if already scheduled.
> > > -    * Possible race with a timer scheduled after this check but before
> > > -    * mod_timer below can be tolerated because group->polling_next_update
> > > -    * will keep updates on schedule.
> > > -    */
> > > -   if (timer_pending(&group->poll_timer))
> > > +   /* cmpxchg should be called even when !force to set poll_scheduled */
> > > +   if (atomic_cmpxchg(&group->poll_scheduled, 0, 1) != 0 && !force)

Will change to Peter's suggested "if
(atomic_cmpxchg(&group->poll_scheduled, 0, 1) && !force)" once the
ordering question is finalized.

> >
> > Do you care about memory ordering here? Afaict the whole thing is
> > supposed to be ordered by ->trigger_lock, so you don't.
>
> It's not always held when we get here.
>
> The worker holds it when it reschedules itself, to serialize against
> userspace destroying the trigger itself. But the scheduler doesn't
> hold it when it kicks the worker on an actionable task change.
>
> That said, I think the ordering we care about there is that when the
> scheduler side sees the worker still queued, the worker must see the
> scheduler's updates to the percpu states and process them correctly.
> But that should be ensured already by the ordering implied by the
> seqcount sections around both the writer and the reader side.

Thanks Johannes! I have nothing to add here really.

>
> Is there another possible race that I'm missing?