* [PATCH -next] virtio_net: Avoid loop in virtnet_poll @ 2020-08-02 3:41 Mao Wenan 2020-08-02 4:40 ` Michael S. Tsirkin 0 siblings, 1 reply; 17+ messages in thread From: Mao Wenan @ 2020-08-02 3:41 UTC (permalink / raw) To: mst, jasowang, davem, kuba; +Cc: netdev, Mao Wenan The loop may exist if vq->broken is true, virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split will return NULL, so virtnet_poll will reschedule napi to receive packet, it will lead cpu usage(si) to 100%. call trace as below: virtnet_poll virtnet_receive virtqueue_get_buf_ctx virtqueue_get_buf_ctx_packed virtqueue_get_buf_ctx_split virtqueue_napi_complete virtqueue_napi_schedule //it will reschedule napi Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> --- drivers/net/virtio_net.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index ba38765..a058da1 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -327,7 +327,8 @@ static void virtqueue_napi_complete(struct napi_struct *napi, opaque = virtqueue_enable_cb_prepare(vq); if (napi_complete_done(napi, processed)) { - if (unlikely(virtqueue_poll(vq, opaque))) + if (unlikely(virtqueue_poll(vq, opaque)) && + unlikely(!virtqueue_is_broken(vq))) virtqueue_napi_schedule(napi, vq); } else { virtqueue_disable_cb(vq); -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: [PATCH -next] virtio_net: Avoid loop in virtnet_poll 2020-08-02 3:41 [PATCH -next] virtio_net: Avoid loop in virtnet_poll Mao Wenan @ 2020-08-02 4:40 ` Michael S. Tsirkin 2020-08-02 5:56 ` Mao Wenan 0 siblings, 1 reply; 17+ messages in thread From: Michael S. Tsirkin @ 2020-08-02 4:40 UTC (permalink / raw) To: Mao Wenan; +Cc: jasowang, davem, kuba, netdev On Sun, Aug 02, 2020 at 11:41:23AM +0800, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_napi_schedule //it will reschedule napi > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> I think it's more a bug in virtqueue_poll : virtqueue_get_buf reports NULL on broken, so virtqueue_poll should report false for consistency. > --- > drivers/net/virtio_net.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c > index ba38765..a058da1 100644 > --- a/drivers/net/virtio_net.c > +++ b/drivers/net/virtio_net.c > @@ -327,7 +327,8 @@ static void virtqueue_napi_complete(struct napi_struct *napi, > > opaque = virtqueue_enable_cb_prepare(vq); > if (napi_complete_done(napi, processed)) { > - if (unlikely(virtqueue_poll(vq, opaque))) > + if (unlikely(virtqueue_poll(vq, opaque)) && > + unlikely(!virtqueue_is_broken(vq))) > virtqueue_napi_schedule(napi, vq); > } else { > virtqueue_disable_cb(vq); > -- > 1.8.3.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll 2020-08-02 4:40 ` Michael S. Tsirkin @ 2020-08-02 5:56 ` Mao Wenan 0 siblings, 0 replies; 17+ messages in thread From: Mao Wenan @ 2020-08-02 5:56 UTC (permalink / raw) To: mst, jasowang; +Cc: virtualization, linux-kernel, kernel-janitors, Mao Wenan The loop may exist if vq->broken is true, virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split will return NULL, so virtnet_poll will reschedule napi to receive packet, it will lead cpu usage(si) to 100%. call trace as below: virtnet_poll virtnet_receive virtqueue_get_buf_ctx virtqueue_get_buf_ctx_packed virtqueue_get_buf_ctx_split virtqueue_napi_complete virtqueue_poll //return true virtqueue_napi_schedule //it will reschedule napi To fix this, return false if vq is broken in virtqueue_poll. Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> --- v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> drivers/virtio/virtio_ring.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index 58b96ba..4f7c73e 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) { struct vring_virtqueue *vq = to_vvq(_vq); + if (unlikely(vq->broken)) + return false; + virtio_mb(vq->weak_barriers); return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : virtqueue_poll_split(_vq, last_used_idx); -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll @ 2020-08-02 5:56 ` Mao Wenan 0 siblings, 0 replies; 17+ messages in thread From: Mao Wenan @ 2020-08-02 5:56 UTC (permalink / raw) To: mst, jasowang; +Cc: virtualization, linux-kernel, kernel-janitors, Mao Wenan The loop may exist if vq->broken is true, virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split will return NULL, so virtnet_poll will reschedule napi to receive packet, it will lead cpu usage(si) to 100%. call trace as below: virtnet_poll virtnet_receive virtqueue_get_buf_ctx virtqueue_get_buf_ctx_packed virtqueue_get_buf_ctx_split virtqueue_napi_complete virtqueue_poll //return true virtqueue_napi_schedule //it will reschedule napi To fix this, return false if vq is broken in virtqueue_poll. Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> --- v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> drivers/virtio/virtio_ring.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index 58b96ba..4f7c73e 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) { struct vring_virtqueue *vq = to_vvq(_vq); + if (unlikely(vq->broken)) + return false; + virtio_mb(vq->weak_barriers); return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : virtqueue_poll_split(_vq, last_used_idx); -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll 2020-08-02 5:56 ` Mao Wenan (?) @ 2020-08-02 6:25 ` Michael S. Tsirkin -1 siblings, 0 replies; 17+ messages in thread From: Michael S. Tsirkin @ 2020-08-02 6:25 UTC (permalink / raw) To: Mao Wenan; +Cc: jasowang, virtualization, linux-kernel, kernel-janitors On Sun, Aug 02, 2020 at 01:56:33PM +0800, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > To fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> Looks good: Acked-by: Michael S. Tsirkin <mst@redhat.com> > --- > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); > -- > 1.8.3.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll @ 2020-08-02 6:25 ` Michael S. Tsirkin 0 siblings, 0 replies; 17+ messages in thread From: Michael S. Tsirkin @ 2020-08-02 6:25 UTC (permalink / raw) To: Mao Wenan; +Cc: kernel-janitors, linux-kernel, virtualization On Sun, Aug 02, 2020 at 01:56:33PM +0800, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > To fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> Looks good: Acked-by: Michael S. Tsirkin <mst@redhat.com> > --- > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); > -- > 1.8.3.1 _______________________________________________ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll @ 2020-08-02 6:25 ` Michael S. Tsirkin 0 siblings, 0 replies; 17+ messages in thread From: Michael S. Tsirkin @ 2020-08-02 6:25 UTC (permalink / raw) To: Mao Wenan; +Cc: jasowang, virtualization, linux-kernel, kernel-janitors On Sun, Aug 02, 2020 at 01:56:33PM +0800, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > To fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> Looks good: Acked-by: Michael S. Tsirkin <mst@redhat.com> > --- > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); > -- > 1.8.3.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll 2020-08-02 5:56 ` Mao Wenan (?) @ 2020-08-02 6:26 ` Michael S. Tsirkin -1 siblings, 0 replies; 17+ messages in thread From: Michael S. Tsirkin @ 2020-08-02 6:26 UTC (permalink / raw) To: Mao Wenan; +Cc: jasowang, virtualization, linux-kernel, kernel-janitors Just noticed the subject is wrong: this is no longer a virtio_net patch. On Sun, Aug 02, 2020 at 01:56:33PM +0800, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > To fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> > --- > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); > -- > 1.8.3.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll @ 2020-08-02 6:26 ` Michael S. Tsirkin 0 siblings, 0 replies; 17+ messages in thread From: Michael S. Tsirkin @ 2020-08-02 6:26 UTC (permalink / raw) To: Mao Wenan; +Cc: kernel-janitors, linux-kernel, virtualization Just noticed the subject is wrong: this is no longer a virtio_net patch. On Sun, Aug 02, 2020 at 01:56:33PM +0800, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > To fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> > --- > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); > -- > 1.8.3.1 _______________________________________________ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll @ 2020-08-02 6:26 ` Michael S. Tsirkin 0 siblings, 0 replies; 17+ messages in thread From: Michael S. Tsirkin @ 2020-08-02 6:26 UTC (permalink / raw) To: Mao Wenan; +Cc: jasowang, virtualization, linux-kernel, kernel-janitors Just noticed the subject is wrong: this is no longer a virtio_net patch. On Sun, Aug 02, 2020 at 01:56:33PM +0800, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > To fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> > --- > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); > -- > 1.8.3.1 ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll 2020-08-02 6:26 ` Michael S. Tsirkin @ 2020-08-02 7:31 ` maowenan -1 siblings, 0 replies; 17+ messages in thread From: maowenan @ 2020-08-02 7:31 UTC (permalink / raw) To: Michael S. Tsirkin Cc: jasowang, virtualization, linux-kernel, kernel-janitors On 8/2/20 2:26 PM, Michael S. Tsirkin wrote: > > Just noticed the subject is wrong: this is no longer > a virtio_net patch. thanks, I will change the subject and send v3. ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v2] virtio_net: Avoid loop in virtnet_poll @ 2020-08-02 7:31 ` maowenan 0 siblings, 0 replies; 17+ messages in thread From: maowenan @ 2020-08-02 7:31 UTC (permalink / raw) To: Michael S. Tsirkin Cc: jasowang, virtualization, linux-kernel, kernel-janitors On 8/2/20 2:26 PM, Michael S. Tsirkin wrote: > > Just noticed the subject is wrong: this is no longer > a virtio_net patch. thanks, I will change the subject and send v3. ^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH -next v3] virtio_ring: Avoid loop when vq is broken in virtqueue_poll 2020-08-02 6:26 ` Michael S. Tsirkin @ 2020-08-02 7:44 ` Mao Wenan -1 siblings, 0 replies; 17+ messages in thread From: Mao Wenan @ 2020-08-02 7:44 UTC (permalink / raw) To: mst, jasowang; +Cc: virtualization, linux-kernel, kernel-janitors, Mao Wenan The loop may exist if vq->broken is true, virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split will return NULL, so virtnet_poll will reschedule napi to receive packet, it will lead cpu usage(si) to 100%. call trace as below: virtnet_poll virtnet_receive virtqueue_get_buf_ctx virtqueue_get_buf_ctx_packed virtqueue_get_buf_ctx_split virtqueue_napi_complete virtqueue_poll //return true virtqueue_napi_schedule //it will reschedule napi to fix this, return false if vq is broken in virtqueue_poll. Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> --- v2->v3: change subject, original is : "virtio_net: Avoid loop in virtnet_poll" v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> drivers/virtio/virtio_ring.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index 58b96ba..4f7c73e 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) { struct vring_virtqueue *vq = to_vvq(_vq); + if (unlikely(vq->broken)) + return false; + virtio_mb(vq->weak_barriers); return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : virtqueue_poll_split(_vq, last_used_idx); -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH -next v3] virtio_ring: Avoid loop when vq is broken in virtqueue_poll @ 2020-08-02 7:44 ` Mao Wenan 0 siblings, 0 replies; 17+ messages in thread From: Mao Wenan @ 2020-08-02 7:44 UTC (permalink / raw) To: mst, jasowang; +Cc: virtualization, linux-kernel, kernel-janitors, Mao Wenan The loop may exist if vq->broken is true, virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split will return NULL, so virtnet_poll will reschedule napi to receive packet, it will lead cpu usage(si) to 100%. call trace as below: virtnet_poll virtnet_receive virtqueue_get_buf_ctx virtqueue_get_buf_ctx_packed virtqueue_get_buf_ctx_split virtqueue_napi_complete virtqueue_poll //return true virtqueue_napi_schedule //it will reschedule napi to fix this, return false if vq is broken in virtqueue_poll. Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> --- v2->v3: change subject, original is : "virtio_net: Avoid loop in virtnet_poll" v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> drivers/virtio/virtio_ring.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index 58b96ba..4f7c73e 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) { struct vring_virtqueue *vq = to_vvq(_vq); + if (unlikely(vq->broken)) + return false; + virtio_mb(vq->weak_barriers); return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : virtqueue_poll_split(_vq, last_used_idx); -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: [PATCH -next v3] virtio_ring: Avoid loop when vq is broken in virtqueue_poll 2020-08-02 7:44 ` Mao Wenan (?) @ 2020-08-04 2:42 ` Jason Wang -1 siblings, 0 replies; 17+ messages in thread From: Jason Wang @ 2020-08-04 2:42 UTC (permalink / raw) To: Mao Wenan, mst; +Cc: virtualization, linux-kernel, kernel-janitors On 2020/8/2 下午3:44, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > to fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> > Acked-by: Michael S. Tsirkin <mst@redhat.com> > --- > v2->v3: change subject, original is : "virtio_net: Avoid loop in virtnet_poll" > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); Acked-by: Jason Wang <jasowang@redhat.com> ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v3] virtio_ring: Avoid loop when vq is broken in virtqueue_poll @ 2020-08-04 2:42 ` Jason Wang 0 siblings, 0 replies; 17+ messages in thread From: Jason Wang @ 2020-08-04 2:42 UTC (permalink / raw) To: Mao Wenan, mst; +Cc: kernel-janitors, linux-kernel, virtualization On 2020/8/2 下午3:44, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > to fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> > Acked-by: Michael S. Tsirkin <mst@redhat.com> > --- > v2->v3: change subject, original is : "virtio_net: Avoid loop in virtnet_poll" > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); Acked-by: Jason Wang <jasowang@redhat.com> _______________________________________________ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH -next v3] virtio_ring: Avoid loop when vq is broken in virtqueue_poll @ 2020-08-04 2:42 ` Jason Wang 0 siblings, 0 replies; 17+ messages in thread From: Jason Wang @ 2020-08-04 2:42 UTC (permalink / raw) To: Mao Wenan, mst; +Cc: virtualization, linux-kernel, kernel-janitors On 2020/8/2 下午3:44, Mao Wenan wrote: > The loop may exist if vq->broken is true, > virtqueue_get_buf_ctx_packed or virtqueue_get_buf_ctx_split > will return NULL, so virtnet_poll will reschedule napi to > receive packet, it will lead cpu usage(si) to 100%. > > call trace as below: > virtnet_poll > virtnet_receive > virtqueue_get_buf_ctx > virtqueue_get_buf_ctx_packed > virtqueue_get_buf_ctx_split > virtqueue_napi_complete > virtqueue_poll //return true > virtqueue_napi_schedule //it will reschedule napi > > to fix this, return false if vq is broken in virtqueue_poll. > > Signed-off-by: Mao Wenan <wenan.mao@linux.alibaba.com> > Acked-by: Michael S. Tsirkin <mst@redhat.com> > --- > v2->v3: change subject, original is : "virtio_net: Avoid loop in virtnet_poll" > v1->v2: fix it in virtqueue_poll suggested by Michael S. Tsirkin <mst@redhat.com> > drivers/virtio/virtio_ring.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 58b96ba..4f7c73e 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -1960,6 +1960,9 @@ bool virtqueue_poll(struct virtqueue *_vq, unsigned last_used_idx) > { > struct vring_virtqueue *vq = to_vvq(_vq); > > + if (unlikely(vq->broken)) > + return false; > + > virtio_mb(vq->weak_barriers); > return vq->packed_ring ? virtqueue_poll_packed(_vq, last_used_idx) : > virtqueue_poll_split(_vq, last_used_idx); Acked-by: Jason Wang <jasowang@redhat.com> ^ permalink raw reply [flat|nested] 17+ messages in thread
end of thread, other threads:[~2020-08-04 2:43 UTC | newest] Thread overview: 17+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2020-08-02 3:41 [PATCH -next] virtio_net: Avoid loop in virtnet_poll Mao Wenan 2020-08-02 4:40 ` Michael S. Tsirkin 2020-08-02 5:56 ` [PATCH -next v2] " Mao Wenan 2020-08-02 5:56 ` Mao Wenan 2020-08-02 6:25 ` Michael S. Tsirkin 2020-08-02 6:25 ` Michael S. Tsirkin 2020-08-02 6:25 ` Michael S. Tsirkin 2020-08-02 6:26 ` Michael S. Tsirkin 2020-08-02 6:26 ` Michael S. Tsirkin 2020-08-02 6:26 ` Michael S. Tsirkin 2020-08-02 7:31 ` maowenan 2020-08-02 7:31 ` maowenan 2020-08-02 7:44 ` [PATCH -next v3] virtio_ring: Avoid loop when vq is broken in virtqueue_poll Mao Wenan 2020-08-02 7:44 ` Mao Wenan 2020-08-04 2:42 ` Jason Wang 2020-08-04 2:42 ` Jason Wang 2020-08-04 2:42 ` Jason Wang
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.