Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.33 as permitted sender) client-ip=23.128.96.33;
MIME-Version: 1.0
References: <20231120073821.1304-1-xuewen.yan@unisoc.com> <7c223b0c-d39a-4d53-8f00-7fbec6b49b6e@bytedance.com>
In-Reply-To: <7c223b0c-d39a-4d53-8f00-7fbec6b49b6e@bytedance.com>
From:   Xuewen Yan <xuewen.yan94@gmail.com>
Date:   Tue, 12 Dec 2023 10:51:05 +0800
Message-ID: <CAB8ipk9WBNE8Yk9T9SXovnpyUtGu=ps8_i=3hbb4PS9PXovR+w@mail.gmail.com>
Subject: Re: [PATCH] sched/eevdf: Avoid NULL in pick_eevdf
To:     Abel Wu <wuyun.abel@bytedance.com>
Cc:     Xuewen Yan <xuewen.yan@unisoc.com>, mingo@redhat.com,
        peterz@infradead.org, juri.lelli@redhat.com,
        vincent.guittot@linaro.org, dietmar.eggemann@arm.com,
        rostedt@goodmis.org, bsegall@google.com, mgorman@suse.de,
        bristot@redhat.com, vschneid@redhat.com,
        linux-kernel@vger.kernel.org, ke.wang@unisoc.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hi Abel

On Mon, Nov 20, 2023 at 4:49=E2=80=AFPM Abel Wu <wuyun.abel@bytedance.com> =
wrote:
>
> Hi Xuewen, the pick part has been re-worked, would you please re-test
> with the newest branch?
>
> Thanks,
>         Abel
>
> https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git/?h=3Dsched%2F=
core

These patches would be merged into 6.6? If not, the pr_err also has a deadl=
ock.
And should be changed to printk_deferred.


954 static struct sched_entity *pick_eevdf(struct cfs_rq *cfs_rq)
  955 {
  956         struct sched_entity *se =3D __pick_eevdf(cfs_rq);
  957
  958         if (!se) {
  959                 struct sched_entity *left =3D __pick_first_entity(cfs=
_rq);
  960                 if (left) {
  961                         pr_err("EEVDF scheduling fail, picking
leftmost\n"); <<<<
  962                         return left;
  963                 }
  964         }
  965
  966         return se;
  967 }


>
> On 11/20/23 3:38 PM, Xuewen Yan Wrote:
> > Now in pick_eevdf function, add the pick_first_entity to prevent
> > picking null when using eevdf, however, the leftmost may be null.
> > As a result, it would cause oops because the se is NULL.
> >
> > Fix this by compare the curr and left, if the left is null, set
> > the se be curr.
> >
> > Fixes: 147f3efaa241 ("sched/fair: Implement an EEVDF-like scheduling po=
licy")
> > Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
> > ---
> >   kernel/sched/fair.c | 18 +++++++++++++++++-
> >   1 file changed, 17 insertions(+), 1 deletion(-)
> >
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index d7a3c63a2171..10916f6778ac 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -951,12 +951,28 @@ static struct sched_entity *__pick_eevdf(struct c=
fs_rq *cfs_rq)
> >       return NULL;
> >   }
> >
> > +/* Just simply choose the se with the smallest vruntime */
> > +static struct sched_entity *__pick_cfs(struct cfs_rq *cfs_rq)
> > +{
> > +     struct sched_entity *curr =3D cfs_rq->curr;
> > +     struct sched_entity *left =3D __pick_first_entity(cfs_rq);
> > +
> > +     /*
> > +      * If curr is set we have to see if its left of the leftmost enti=
ty
> > +      * still in the tree, provided there was anything in the tree at =
all.
> > +      */
> > +     if (!left || (curr && entity_before(curr, left)))
> > +             left =3D curr;
> > +
> > +     return left;
> > +}
> > +
> >   static struct sched_entity *pick_eevdf(struct cfs_rq *cfs_rq)
> >   {
> >       struct sched_entity *se =3D __pick_eevdf(cfs_rq);
> >
> >       if (!se) {
> > -             struct sched_entity *left =3D __pick_first_entity(cfs_rq)=
;
> > +             struct sched_entity *left =3D __pick_cfs(cfs_rq);
> >               if (left) {
> >                       pr_err("EEVDF scheduling fail, picking leftmost\n=
");
> >                       return left;