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: <20200120145635.GA30904@blackbody.suse.cz> <20200124114017.8363-1-mkoutny@suse.com>
 <20200124114017.8363-2-mkoutny@suse.com> <CAJuCfpGjC=YwY=oNnYFNDp2nCuR9YhSU95=xbbeoDEheemte+Q@mail.gmail.com>
In-Reply-To: <CAJuCfpGjC=YwY=oNnYFNDp2nCuR9YhSU95=xbbeoDEheemte+Q@mail.gmail.com>
From:   Suren Baghdasaryan <surenb@google.com>
Date:   Wed, 5 Feb 2020 09:27:45 -0800
Message-ID: <CAJuCfpEq9c6cdEdqZOq6KDmxSHLNzby=46psRYpGxYhDT31yew@mail.gmail.com>
Subject: Re: [PATCH v2 1/3] cgroup: Iterate tasks that did not finish do_exit()
To:     =?UTF-8?Q?Michal_Koutn=C3=BD?= <mkoutny@suse.com>
Cc:     cgroups mailinglist <cgroups@vger.kernel.org>,
        alex.shi@linux.alibaba.com, Roman Gushchin <guro@fb.com>,
        Johannes Weiner <hannes@cmpxchg.org>,
        kernel-team <kernel-team@android.com>,
        JeiFeng Lee <linger.lee@mediatek.com>,
        linux-arm-kernel@lists.infradead.org,
        LKML <linux-kernel@vger.kernel.org>,
        linux-kselftest@vger.kernel.org,
        linux-mediatek@lists.infradead.org, Li Zefan <lizefan@huawei.com>,
        matthias.bgg@gmail.com, shuah@kernel.org,
        Tejun Heo <tj@kernel.org>, Tom Cherry <tomcherry@google.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Jan 24, 2020 at 2:56 PM Suren Baghdasaryan <surenb@google.com> wrot=
e:
>
> On Fri, Jan 24, 2020 at 3:40 AM Michal Koutn=C3=BD <mkoutny@suse.com> wro=
te:
> >
> > PF_EXITING is set earlier than actual removal from css_set when a task
> > is exitting. This can confuse cgroup.procs readers who see no PF_EXITIN=
G
> > tasks, however, rmdir is checking against css_set membership so it can
> > transitionally fail with EBUSY.
> >
> > Fix this by listing tasks that weren't unlinked from css_set active
> > lists.
> > It may happen that other users of the task iterator (without
> > CSS_TASK_ITER_PROCS) spot a PF_EXITING task before cgroup_exit(). This
> > is equal to the state before commit c03cd7738a83 ("cgroup: Include dyin=
g
> > leaders with live threads in PROCS iterations") but it may be reviewed
> > later.
> >
> > Reported-by: Suren Baghdasaryan <surenb@google.com>
> > Fixes: c03cd7738a83 ("cgroup: Include dying leaders with live threads i=
n PROCS iterations")
> > Signed-off-by: Michal Koutn=C3=BD <mkoutny@suse.com>
> > ---
> >  include/linux/cgroup.h |  1 +
> >  kernel/cgroup/cgroup.c | 23 ++++++++++++++++-------
> >  2 files changed, 17 insertions(+), 7 deletions(-)
> >
> > diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
> > index d7ddebd0cdec..e75d2191226b 100644
> > --- a/include/linux/cgroup.h
> > +++ b/include/linux/cgroup.h
> > @@ -62,6 +62,7 @@ struct css_task_iter {
> >         struct list_head                *mg_tasks_head;
> >         struct list_head                *dying_tasks_head;
> >
> > +       struct list_head                *cur_tasks_head;
> >         struct css_set                  *cur_cset;
> >         struct css_set                  *cur_dcset;
> >         struct task_struct              *cur_task;
> > diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
> > index 735af8f15f95..a6e3619e013b 100644
> > --- a/kernel/cgroup/cgroup.c
> > +++ b/kernel/cgroup/cgroup.c
> > @@ -4404,12 +4404,16 @@ static void css_task_iter_advance_css_set(struc=
t css_task_iter *it)
> >                 }
> >         } while (!css_set_populated(cset) && list_empty(&cset->dying_ta=
sks));
> >
> > -       if (!list_empty(&cset->tasks))
> > +       if (!list_empty(&cset->tasks)) {
> >                 it->task_pos =3D cset->tasks.next;
> > -       else if (!list_empty(&cset->mg_tasks))
> > +               it->cur_tasks_head =3D &cset->tasks;
> > +       } else if (!list_empty(&cset->mg_tasks)) {
> >                 it->task_pos =3D cset->mg_tasks.next;
> > -       else
> > +               it->cur_tasks_head =3D &cset->mg_tasks;
> > +       } else {
> >                 it->task_pos =3D cset->dying_tasks.next;
> > +               it->cur_tasks_head =3D &cset->dying_tasks;
> > +       }
> >
> >         it->tasks_head =3D &cset->tasks;
> >         it->mg_tasks_head =3D &cset->mg_tasks;
> > @@ -4467,10 +4471,14 @@ static void css_task_iter_advance(struct css_ta=
sk_iter *it)
> >                 else
> >                         it->task_pos =3D it->task_pos->next;
> >
> > -               if (it->task_pos =3D=3D it->tasks_head)
> > +               if (it->task_pos =3D=3D it->tasks_head) {
> >                         it->task_pos =3D it->mg_tasks_head->next;
> > -               if (it->task_pos =3D=3D it->mg_tasks_head)
> > +                       it->cur_tasks_head =3D it->mg_tasks_head;
> > +               }
> > +               if (it->task_pos =3D=3D it->mg_tasks_head) {
> >                         it->task_pos =3D it->dying_tasks_head->next;
> > +                       it->cur_tasks_head =3D it->dying_tasks_head;
> > +               }
> >                 if (it->task_pos =3D=3D it->dying_tasks_head)
> >                         css_task_iter_advance_css_set(it);
> >         } else {
> > @@ -4489,11 +4497,12 @@ static void css_task_iter_advance(struct css_ta=
sk_iter *it)
> >                         goto repeat;
> >
> >                 /* and dying leaders w/o live member threads */
> > -               if (!atomic_read(&task->signal->live))
> > +               if (it->cur_tasks_head =3D=3D it->dying_tasks_head &&
> > +                   !atomic_read(&task->signal->live))
> >                         goto repeat;
> >         } else {
> >                 /* skip all dying ones */
> > -               if (task->flags & PF_EXITING)
> > +               if (it->cur_tasks_head =3D=3D it->dying_tasks_head)
> >                         goto repeat;
> >         }
> >  }
> > --
> > 2.24.1
> >
>
> Tested-by: Suren Baghdasaryan <surenb@google.com>
>
> Thanks!

Hi Folks,
If this new version looks good could we get an Ack please? I need to
start backporting this fix to Android and would like a confirmation
before doing so.
Thanks!