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: <1683167226-7012-1-git-send-email-wangwenliang.1995@bytedance.com>
 <CACGkMEs_4kUzc6iSBWvhZA1+U70Pp0o+WhE0aQnC-5pECW7QXA@mail.gmail.com>
 <20230507093328-mutt-send-email-mst@kernel.org> <2b5cf90a-efa8-52a7-9277-77722622c128@redhat.com>
 <20230508020717-mutt-send-email-mst@kernel.org> <CACGkMEuQdy8xi=eD4v7-UNQ12xOUdnuyQ73vvC6vdGXUfeasug@mail.gmail.com>
 <20230508024433-mutt-send-email-mst@kernel.org>
In-Reply-To: <20230508024433-mutt-send-email-mst@kernel.org>
From:   Jason Wang <jasowang@redhat.com>
Date:   Tue, 9 May 2023 11:14:30 +0800
Message-ID: <CACGkMEuYcR21_k0hyisWzTVHG4+a3Y=ym101Z5P8TSWyNkHWxA@mail.gmail.com>
Subject: Re: [PATCH v4] virtio_net: suppress cpu stall when free_unused_bufs
To:     "Michael S. Tsirkin" <mst@redhat.com>
Cc:     Wenliang Wang <wangwenliang.1995@bytedance.com>,
        davem@davemloft.net, edumazet@google.com, kuba@kernel.org,
        pabeni@redhat.com, zhengqi.arch@bytedance.com,
        willemdebruijn.kernel@gmail.com,
        virtualization@lists.linux-foundation.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, xuanzhuo@linux.alibaba.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Mon, May 8, 2023 at 2:47=E2=80=AFPM Michael S. Tsirkin <mst@redhat.com> =
wrote:
>
> On Mon, May 08, 2023 at 02:13:42PM +0800, Jason Wang wrote:
> > On Mon, May 8, 2023 at 2:08=E2=80=AFPM Michael S. Tsirkin <mst@redhat.c=
om> wrote:
> > >
> > > On Mon, May 08, 2023 at 11:12:03AM +0800, Jason Wang wrote:
> > > >
> > > > =E5=9C=A8 2023/5/7 21:34, Michael S. Tsirkin =E5=86=99=E9=81=93:
> > > > > On Fri, May 05, 2023 at 11:28:25AM +0800, Jason Wang wrote:
> > > > > > On Thu, May 4, 2023 at 10:27=E2=80=AFAM Wenliang Wang
> > > > > > <wangwenliang.1995@bytedance.com> wrote:
> > > > > > > For multi-queue and large ring-size use case, the following e=
rror
> > > > > > > occurred when free_unused_bufs:
> > > > > > > rcu: INFO: rcu_sched self-detected stall on CPU.
> > > > > > >
> > > > > > > Fixes: 986a4f4d452d ("virtio_net: multiqueue support")
> > > > > > > Signed-off-by: Wenliang Wang <wangwenliang.1995@bytedance.com=
>
> > > > > > > ---
> > > > > > > v2:
> > > > > > > -add need_resched check.
> > > > > > > -apply same logic to sq.
> > > > > > > v3:
> > > > > > > -use cond_resched instead.
> > > > > > > v4:
> > > > > > > -add fixes tag
> > > > > > > ---
> > > > > > >   drivers/net/virtio_net.c | 2 ++
> > > > > > >   1 file changed, 2 insertions(+)
> > > > > > >
> > > > > > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_ne=
t.c
> > > > > > > index 8d8038538fc4..a12ae26db0e2 100644
> > > > > > > --- a/drivers/net/virtio_net.c
> > > > > > > +++ b/drivers/net/virtio_net.c
> > > > > > > @@ -3560,12 +3560,14 @@ static void free_unused_bufs(struct v=
irtnet_info *vi)
> > > > > > >                  struct virtqueue *vq =3D vi->sq[i].vq;
> > > > > > >                  while ((buf =3D virtqueue_detach_unused_buf(=
vq)) !=3D NULL)
> > > > > > >                          virtnet_sq_free_unused_buf(vq, buf);
> > > > > > > +               cond_resched();
> > > > > > Does this really address the case when the virtqueue is very la=
rge?
> > > > > >
> > > > > > Thanks
> > > > >
> > > > > it does in that a very large queue is still just 64k in size.
> > > > > we might however have 64k of these queues.
> > > >
> > > >
> > > > Ok, but we have other similar loops especially the refill, I think =
we may
> > > > need cond_resched() there as well.
> > > >
> > > > Thanks
> > > >
> > >
> > > Refill is already per vq isn't it?
> >
> > Not for the refill_work().
> >
> > Thanks
>
> Good point, refill_work probably needs cond_resched, too.
> And I guess virtnet_open?

Yes, let me draft a patch.

Thanks

>
>
> > >
> > >
> > > > >
> > > > > > >          }
> > > > > > >
> > > > > > >          for (i =3D 0; i < vi->max_queue_pairs; i++) {
> > > > > > >                  struct virtqueue *vq =3D vi->rq[i].vq;
> > > > > > >                  while ((buf =3D virtqueue_detach_unused_buf(=
vq)) !=3D NULL)
> > > > > > >                          virtnet_rq_free_unused_buf(vq, buf);
> > > > > > > +               cond_resched();
> > > > > > >          }
> > > > > > >   }
> > > > > > >
> > > > > > > --
> > > > > > > 2.20.1
> > > > > > >
> > >
>