Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20181227102107.GA21156@linaro.org> <CAKfTPtD7O_DU5no36jaQziZDCvy=BMJk+tZbnQk2yVt=pzRfPw@mail.gmail.com>
 <CAMp4zn-bN=DdiKox-xZ++i2n_pgTA7Ms+1hBnAY2AbYv2iqXOA@mail.gmail.com>
 <CAKfTPtAmkHZ5XVHpf6rE8nrSPGzW_uO7a0tvH5JqDSpTzK9D=Q@mail.gmail.com>
 <CAHk-=wja7WN1sDz5_E7hf8c47CmX=yH8xc3h0zmmF8Y_fSR71A@mail.gmail.com>
 <20181228011524.GF2509588@devbig004.ftw2.facebook.com> <CAHk-=wj33F7LKQS3wCYJtb_yCzYhbjPzFqhVS_ZPPNOWTTMHFQ@mail.gmail.com>
 <20181228015352.GG2509588@devbig004.ftw2.facebook.com> <20181228020243.GH2509588@devbig004.ftw2.facebook.com>
 <CAKfTPtA5WBKQ1FMyGqGjn1DdeDbSwT=bTQjD=oW7WMz6iKiS7A@mail.gmail.com> <20181228165451.GJ2509588@devbig004.ftw2.facebook.com>
In-Reply-To: <20181228165451.GJ2509588@devbig004.ftw2.facebook.com>
From:   Vincent Guittot <vincent.guittot@linaro.org>
Date:   Fri, 28 Dec 2018 18:25:37 +0100
Message-ID: <CAKfTPtAMjfnNHu_JdDSi1BSMoXKgGkm2+G_QLsUBDAhq1wFvZQ@mail.gmail.com>
Subject: Re: [PATCH] sched: fix infinity loop in update_blocked_averages
To:     Tejun Heo <tj@kernel.org>
Cc:     Linus Torvalds <torvalds@linux-foundation.org>,
        Sargun Dhillon <sargun@sargun.me>,
        Xie XiuQi <xiexiuqi@huawei.com>,
        Ingo Molnar <mingo@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>, xiezhipeng1@huawei.com,
        huawei.libin@huawei.com,
        linux-kernel <linux-kernel@vger.kernel.org>,
        Dmitry Adamushko <dmitry.adamushko@gmail.com>,
        Rik van Riel <riel@surriel.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, 28 Dec 2018 at 17:54, Tejun Heo <tj@kernel.org> wrote:
>
> Hello,
>
> On Fri, Dec 28, 2018 at 10:30:07AM +0100, Vincent Guittot wrote:
> > > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > > index d1907506318a..88b9118b5191 100644
> > > --- a/kernel/sched/fair.c
> > > +++ b/kernel/sched/fair.c
> > > @@ -7698,7 +7698,8 @@ static void update_blocked_averages(int cpu)
> > >                  * There can be a lot of idle CPU cgroups.  Don't let fully
> > >                  * decayed cfs_rqs linger on the list.
> > >                  */
> > > -               if (cfs_rq_is_decayed(cfs_rq))
> > > +               if (cfs_rq_is_decayed(cfs_rq) &&
> > > +                   rq->tmp_alone_branch == &rq->leaf_cfs_rq_list)
> > >                         list_del_leaf_cfs_rq(cfs_rq);
> >
> > This patch reduces the cases but I don't thinks it's enough because it
> > doesn't cover the case of unregister_fair_sched_group()
> > And we can still break the ordering of the cfs_rq
>
> So, if unregister_fair_sched_group() can corrupt list, the bug is
> there regardless of a9e7f6544b9ce, right?

I don't think so because without a9e7f6544b9ce, the insertion in the
list is done only once and we can't  call unregister_fair_sched_group
while an enqueue is ongoing so tmp_alone_branch always point to
rq->leaf_cfs_rq_list.
a9e7f6544b9ce enables to have tmp_alone_branch not pointing to
rq->leaf_cfs_rq_list

>
> Is there a reason why we're building a dedicated list for avg
> propagation?  AFAICS, it's just doing depth first walk, which can be

we have this list of sched group of the rq to update the load of each
cfs_rq and as a result the load of the task group. This is then used
to compute the share of a task group between CPUs.
This list must be ordered to correctly propagate the updates from leafs to root.

> done without extra space as long as each node has the parent pointer,
> which they do.  Is the dedicated list an optimization?

It prevents to parse and walk all task group struct every time.
Instead, you just have to follow a linked list

Regards,
Vincent
>
> Thanks.
>
> --
> tejun