| Message ID | 52FA3AAC.2050003@huawei.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, Feb 11, 2014 at 10:58:52PM +0800, Qin Chuanyu wrote: > udp packet use 2 buffers at least, one for vnet_hdr and > one for skb->data. Not really, we use 1 buffer now with vnet_hdr inline with data. > we could change the threshold from 2 to 3, so the udp packet > which data buff only using single desc will gain from this. > the guest would avoid from allocating memory dynamically. > the host would avoid from translating indirect desc. > > Signed-off-by: Chuanyu Qin <qinchuanyu@huawei.com> Optimization patch without any performance data? Such a change would need much more testing than that: would have to try various workloads with -net and -blk at least. > --- > drivers/virtio/virtio_ring.c | 2 +- > 1 files changed, 1 insertions(+), 1 deletions(-) > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c > index 28b5338..88d008f 100644 > --- a/drivers/virtio/virtio_ring.c > +++ b/drivers/virtio/virtio_ring.c > @@ -220,7 +220,7 @@ static inline int virtqueue_add(struct virtqueue *_vq, > > /* If the host supports indirect descriptor tables, and we have multiple > * buffers, then go indirect. FIXME: tune this threshold */ > - if (vq->indirect && total_sg > 1 && vq->vq.num_free) { > + if (vq->indirect && total_sg > 2 && vq->vq.num_free) { > head = vring_add_indirect(vq, sgs, next, total_sg, total_out, > total_in, > out_sgs, in_sgs, gfp); > -- > 1.7.3.1.msysgit.0 -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2014/2/11 23:43, Michael S. Tsirkin wrote: > On Tue, Feb 11, 2014 at 10:58:52PM +0800, Qin Chuanyu wrote: >> udp packet use 2 buffers at least, one for vnet_hdr and >> one for skb->data. > > Not really, we use 1 buffer now with vnet_hdr inline with data. > I would use it, Thanks. -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Il 11/02/2014 16:43, Michael S. Tsirkin ha scritto: >> > Signed-off-by: Chuanyu Qin <qinchuanyu@huawei.com> > Optimization patch without any performance data? > Such a change would need much more testing than that: > would have to try various workloads with -net and -blk at least. > -blk and -scsi never use 2 descriptors except for flushes (which go all the way to the platters or flash, and are not as frequent as other I/O). So the patch is in all likelihood a no-op there. Of course, virtio-net does need data. Paolo -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Qin Chuanyu <qinchuanyu@huawei.com> writes: > udp packet use 2 buffers at least, one for vnet_hdr and > one for skb->data. > we could change the threshold from 2 to 3, so the udp packet > which data buff only using single desc will gain from this. > the guest would avoid from allocating memory dynamically. > the host would avoid from translating indirect desc. > > Signed-off-by: Chuanyu Qin <qinchuanyu@huawei.com> We should really adapt the threshold depending on how full the ring is. If it's gotten cramped recently, prefer indirect. See linux/average.h. Such a change would have to be benchmarked carefully though! Cheers, Rusty. -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2014/2/11 23:43, Michael S. Tsirkin wrote: > On Tue, Feb 11, 2014 at 10:58:52PM +0800, Qin Chuanyu wrote: >> udp packet use 2 buffers at least, one for vnet_hdr and >> one for skb->data. > > Not really, we use 1 buffer now with vnet_hdr inline with data. > I have found that there are related patch in Qemu, Is there patch suit for vhost_net right now, Or should I implement it myself ? >> we could change the threshold from 2 to 3, so the udp packet >> which data buff only using single desc will gain from this. >> the guest would avoid from allocating memory dynamically. >> the host would avoid from translating indirect desc. >> >> Signed-off-by: Chuanyu Qin <qinchuanyu@huawei.com> > > Optimization patch without any performance data? > Such a change would need much more testing than that: > would have to try various workloads with -net and -blk at least. > >> --- >> drivers/virtio/virtio_ring.c | 2 +- >> 1 files changed, 1 insertions(+), 1 deletions(-) >> >> diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c >> index 28b5338..88d008f 100644 >> --- a/drivers/virtio/virtio_ring.c >> +++ b/drivers/virtio/virtio_ring.c >> @@ -220,7 +220,7 @@ static inline int virtqueue_add(struct virtqueue *_vq, >> >> /* If the host supports indirect descriptor tables, and we have multiple >> * buffers, then go indirect. FIXME: tune this threshold */ >> - if (vq->indirect && total_sg > 1 && vq->vq.num_free) { >> + if (vq->indirect && total_sg > 2 && vq->vq.num_free) { >> head = vring_add_indirect(vq, sgs, next, total_sg, total_out, >> total_in, >> out_sgs, in_sgs, gfp); >> -- >> 1.7.3.1.msysgit.0 > > -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index 28b5338..88d008f 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -220,7 +220,7 @@ static inline int virtqueue_add(struct virtqueue *_vq, /* If the host supports indirect descriptor tables, and we have multiple * buffers, then go indirect. FIXME: tune this threshold */ - if (vq->indirect && total_sg > 1 && vq->vq.num_free) { + if (vq->indirect && total_sg > 2 && vq->vq.num_free) { head = vring_add_indirect(vq, sgs, next, total_sg, total_out, total_in, out_sgs, in_sgs, gfp);
udp packet use 2 buffers at least, one for vnet_hdr and one for skb->data. we could change the threshold from 2 to 3, so the udp packet which data buff only using single desc will gain from this. the guest would avoid from allocating memory dynamically. the host would avoid from translating indirect desc. Signed-off-by: Chuanyu Qin <qinchuanyu@huawei.com> --- drivers/virtio/virtio_ring.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)