Michael S. Tsirkin
2017-Jan-09 23:10 UTC
[PATCH V4 net-next 1/3] vhost: better detection of available buffers
On Mon, Jan 09, 2017 at 10:59:16AM +0800, Jason Wang wrote:> > > On 2017?01?07? 03:55, Michael S. Tsirkin wrote: > > On Fri, Jan 06, 2017 at 10:13:15AM +0800, Jason Wang wrote: > > > This patch tries to do several tweaks on vhost_vq_avail_empty() for a > > > better performance: > > > > > > - check cached avail index first which could avoid userspace memory access. > > > - using unlikely() for the failure of userspace access > > > - check vq->last_avail_idx instead of cached avail index as the last > > > step. > > > > > > This patch is need for batching supports which needs to peek whether > > > or not there's still available buffers in the ring. > > > > > > Reviewed-by: Stefan Hajnoczi <stefanha at redhat.com> > > > Signed-off-by: Jason Wang <jasowang at redhat.com> > > > --- > > > drivers/vhost/vhost.c | 8 ++++++-- > > > 1 file changed, 6 insertions(+), 2 deletions(-) > > > > > > diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c > > > index d643260..9f11838 100644 > > > --- a/drivers/vhost/vhost.c > > > +++ b/drivers/vhost/vhost.c > > > @@ -2241,11 +2241,15 @@ bool vhost_vq_avail_empty(struct vhost_dev *dev, struct vhost_virtqueue *vq) > > > __virtio16 avail_idx; > > > int r; > > > + if (vq->avail_idx != vq->last_avail_idx) > > > + return false; > > > + > > > r = vhost_get_user(vq, avail_idx, &vq->avail->idx); > > > - if (r) > > > + if (unlikely(r)) > > > return false; > > > + vq->avail_idx = vhost16_to_cpu(vq, avail_idx); > > > - return vhost16_to_cpu(vq, avail_idx) == vq->avail_idx; > > > + return vq->avail_idx == vq->last_avail_idx; > > > } > > > EXPORT_SYMBOL_GPL(vhost_vq_avail_empty); > > So again, this did not address the issue I pointed out in v1: > > if we have 1 buffer in RX queue and > > that is not enough to store the whole packet, > > vhost_vq_avail_empty returns false, then we re-read > > the descriptors again and again. > > > > You have saved a single index access but not the more expensive > > descriptor access. > > Looks not, if I understand the code correctly, in this case, get_rx_bufs() > will return zero, and we will try to enable rx kick and exit the loop. > > ThanksI mean this: while (vhost_can_busy_poll(vq->dev, endtime) && vhost_vq_avail_empty(vq->dev, vq)) cpu_relax(); preempt_enable(); r = vhost_get_vq_desc(vq, vq->iov, ARRAY_SIZE(vq->iov), out_num, in_num, NULL, NULL); vhost_vq_avail_empty returns false so we break out of the loop and call vhost_get_vq_desc. -- MST
Jason Wang
2017-Jan-10 02:22 UTC
[PATCH V4 net-next 1/3] vhost: better detection of available buffers
On 2017?01?10? 07:10, Michael S. Tsirkin wrote:> On Mon, Jan 09, 2017 at 10:59:16AM +0800, Jason Wang wrote: >> >> On 2017?01?07? 03:55, Michael S. Tsirkin wrote: >>> On Fri, Jan 06, 2017 at 10:13:15AM +0800, Jason Wang wrote: >>>> This patch tries to do several tweaks on vhost_vq_avail_empty() for a >>>> better performance: >>>> >>>> - check cached avail index first which could avoid userspace memory access. >>>> - using unlikely() for the failure of userspace access >>>> - check vq->last_avail_idx instead of cached avail index as the last >>>> step. >>>> >>>> This patch is need for batching supports which needs to peek whether >>>> or not there's still available buffers in the ring. >>>> >>>> Reviewed-by: Stefan Hajnoczi <stefanha at redhat.com> >>>> Signed-off-by: Jason Wang <jasowang at redhat.com> >>>> --- >>>> drivers/vhost/vhost.c | 8 ++++++-- >>>> 1 file changed, 6 insertions(+), 2 deletions(-) >>>> >>>> diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c >>>> index d643260..9f11838 100644 >>>> --- a/drivers/vhost/vhost.c >>>> +++ b/drivers/vhost/vhost.c >>>> @@ -2241,11 +2241,15 @@ bool vhost_vq_avail_empty(struct vhost_dev *dev, struct vhost_virtqueue *vq) >>>> __virtio16 avail_idx; >>>> int r; >>>> + if (vq->avail_idx != vq->last_avail_idx) >>>> + return false; >>>> + >>>> r = vhost_get_user(vq, avail_idx, &vq->avail->idx); >>>> - if (r) >>>> + if (unlikely(r)) >>>> return false; >>>> + vq->avail_idx = vhost16_to_cpu(vq, avail_idx); >>>> - return vhost16_to_cpu(vq, avail_idx) == vq->avail_idx; >>>> + return vq->avail_idx == vq->last_avail_idx; >>>> } >>>> EXPORT_SYMBOL_GPL(vhost_vq_avail_empty); >>> So again, this did not address the issue I pointed out in v1: >>> if we have 1 buffer in RX queue and >>> that is not enough to store the whole packet, >>> vhost_vq_avail_empty returns false, then we re-read >>> the descriptors again and again. >>> >>> You have saved a single index access but not the more expensive >>> descriptor access. >> Looks not, if I understand the code correctly, in this case, get_rx_bufs() >> will return zero, and we will try to enable rx kick and exit the loop. >> >> Thanks > I mean this: > > while (vhost_can_busy_poll(vq->dev, endtime) && > vhost_vq_avail_empty(vq->dev, vq)) > cpu_relax(); > preempt_enable(); > r = vhost_get_vq_desc(vq, vq->iov, ARRAY_SIZE(vq->iov), > out_num, in_num, NULL, NULL); > > > vhost_vq_avail_empty returns false so we break out of the loop > and call vhost_get_vq_desc. > >But this is the code for polling tx vq not rx I think? Thanks
Michael S. Tsirkin
2017-Jan-10 02:57 UTC
[PATCH V4 net-next 1/3] vhost: better detection of available buffers
On Tue, Jan 10, 2017 at 10:22:42AM +0800, Jason Wang wrote:> > > On 2017?01?10? 07:10, Michael S. Tsirkin wrote: > > On Mon, Jan 09, 2017 at 10:59:16AM +0800, Jason Wang wrote: > > > > > > On 2017?01?07? 03:55, Michael S. Tsirkin wrote: > > > > On Fri, Jan 06, 2017 at 10:13:15AM +0800, Jason Wang wrote: > > > > > This patch tries to do several tweaks on vhost_vq_avail_empty() for a > > > > > better performance: > > > > > > > > > > - check cached avail index first which could avoid userspace memory access. > > > > > - using unlikely() for the failure of userspace access > > > > > - check vq->last_avail_idx instead of cached avail index as the last > > > > > step. > > > > > > > > > > This patch is need for batching supports which needs to peek whether > > > > > or not there's still available buffers in the ring. > > > > > > > > > > Reviewed-by: Stefan Hajnoczi <stefanha at redhat.com> > > > > > Signed-off-by: Jason Wang <jasowang at redhat.com> > > > > > --- > > > > > drivers/vhost/vhost.c | 8 ++++++-- > > > > > 1 file changed, 6 insertions(+), 2 deletions(-) > > > > > > > > > > diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c > > > > > index d643260..9f11838 100644 > > > > > --- a/drivers/vhost/vhost.c > > > > > +++ b/drivers/vhost/vhost.c > > > > > @@ -2241,11 +2241,15 @@ bool vhost_vq_avail_empty(struct vhost_dev *dev, struct vhost_virtqueue *vq) > > > > > __virtio16 avail_idx; > > > > > int r; > > > > > + if (vq->avail_idx != vq->last_avail_idx) > > > > > + return false; > > > > > + > > > > > r = vhost_get_user(vq, avail_idx, &vq->avail->idx); > > > > > - if (r) > > > > > + if (unlikely(r)) > > > > > return false; > > > > > + vq->avail_idx = vhost16_to_cpu(vq, avail_idx); > > > > > - return vhost16_to_cpu(vq, avail_idx) == vq->avail_idx; > > > > > + return vq->avail_idx == vq->last_avail_idx; > > > > > } > > > > > EXPORT_SYMBOL_GPL(vhost_vq_avail_empty); > > > > So again, this did not address the issue I pointed out in v1: > > > > if we have 1 buffer in RX queue and > > > > that is not enough to store the whole packet, > > > > vhost_vq_avail_empty returns false, then we re-read > > > > the descriptors again and again. > > > > > > > > You have saved a single index access but not the more expensive > > > > descriptor access. > > > Looks not, if I understand the code correctly, in this case, get_rx_bufs() > > > will return zero, and we will try to enable rx kick and exit the loop. > > > > > > Thanks > > I mean this: > > > > while (vhost_can_busy_poll(vq->dev, endtime) && > > vhost_vq_avail_empty(vq->dev, vq)) > > cpu_relax(); > > preempt_enable(); > > r = vhost_get_vq_desc(vq, vq->iov, ARRAY_SIZE(vq->iov), > > out_num, in_num, NULL, NULL); > > > > > > vhost_vq_avail_empty returns false so we break out of the loop > > and call vhost_get_vq_desc. > > > > > > But this is the code for polling tx vq not rx I think? > > ThanksOh, right. I'll re-read this. -- MST
Reasonably Related Threads
- [PATCH V4 net-next 1/3] vhost: better detection of available buffers
- [PATCH V4 net-next 1/3] vhost: better detection of available buffers
- [PATCH V4 net-next 1/3] vhost: better detection of available buffers
- [PATCH V4 net-next 1/3] vhost: better detection of available buffers
- [PATCH V4 net-next 1/3] vhost: better detection of available buffers