Message ID | 20240612170851.1004604-1-jiri@resnulli.us (mailing list archive) |
---|---|
State | Superseded |
Delegated to: | Netdev Maintainers |
Headers | show |
Series | [net-next,v2] virtio_net: add support for Byte Queue Limits | expand |
On Wed, Jun 12, 2024 at 07:08:51PM +0200, Jiri Pirko wrote: > From: Jiri Pirko <jiri@nvidia.com> > > Add support for Byte Queue Limits (BQL). > > Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. > Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is > running in background. Netperf TCP_RR results: > > NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 > NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 > NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 > NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 > NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 > NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 > BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 > BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 > BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 > BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 > BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 > BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 > > Signed-off-by: Jiri Pirko <jiri@nvidia.com> I see you now support both napi and non-napi. Thanks a lot Jiri! Just coming out of a national holiday here, as usual with a backlog - pls allow until Monday to review. Thanks! > --- > v1->v2: > - moved netdev_tx_completed_queue() call into __free_old_xmit(), > propagate use_napi flag to __free_old_xmit() and only call > netdev_tx_completed_queue() in case it is true > - added forgotten call to netdev_tx_reset_queue() > - fixed stats for xdp packets > - fixed bql accounting when __free_old_xmit() is called from xdp path > - handle the !use_napi case in start_xmit() kick section > --- > drivers/net/virtio_net.c | 50 +++++++++++++++++++++++++--------------- > 1 file changed, 32 insertions(+), 18 deletions(-) > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c > index 61a57d134544..5863c663ccab 100644 > --- a/drivers/net/virtio_net.c > +++ b/drivers/net/virtio_net.c > @@ -84,7 +84,9 @@ struct virtnet_stat_desc { > > struct virtnet_sq_free_stats { > u64 packets; > + u64 xdp_packets; > u64 bytes; > + u64 xdp_bytes; > }; > > struct virtnet_sq_stats { > @@ -506,29 +508,33 @@ static struct xdp_frame *ptr_to_xdp(void *ptr) > return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG); > } > > -static void __free_old_xmit(struct send_queue *sq, bool in_napi, > +static void __free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, > + bool in_napi, bool use_napi, > struct virtnet_sq_free_stats *stats) > { > unsigned int len; > void *ptr; > > while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) { > - ++stats->packets; > - > if (!is_xdp_frame(ptr)) { > struct sk_buff *skb = ptr; > > pr_debug("Sent skb %p\n", skb); > > + stats->packets++; > stats->bytes += skb->len; > napi_consume_skb(skb, in_napi); > } else { > struct xdp_frame *frame = ptr_to_xdp(ptr); > > - stats->bytes += xdp_get_frame_len(frame); > + stats->xdp_packets++; > + stats->xdp_bytes += xdp_get_frame_len(frame); > xdp_return_frame(frame); > } > } > + if (use_napi) > + netdev_tx_completed_queue(txq, stats->packets, stats->bytes); > + > } > > /* Converting between virtqueue no. and kernel tx/rx queue no. > @@ -955,21 +961,22 @@ static void virtnet_rq_unmap_free_buf(struct virtqueue *vq, void *buf) > virtnet_rq_free_buf(vi, rq, buf); > } > > -static void free_old_xmit(struct send_queue *sq, bool in_napi) > +static void free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, > + bool in_napi, bool use_napi) > { > struct virtnet_sq_free_stats stats = {0}; > > - __free_old_xmit(sq, in_napi, &stats); > + __free_old_xmit(sq, txq, in_napi, use_napi, &stats); > > /* Avoid overhead when no packets have been processed > * happens when called speculatively from start_xmit. > */ > - if (!stats.packets) > + if (!stats.packets && !stats.xdp_packets) > return; > > u64_stats_update_begin(&sq->stats.syncp); > - u64_stats_add(&sq->stats.bytes, stats.bytes); > - u64_stats_add(&sq->stats.packets, stats.packets); > + u64_stats_add(&sq->stats.bytes, stats.bytes + stats.xdp_bytes); > + u64_stats_add(&sq->stats.packets, stats.packets + stats.xdp_packets); > u64_stats_update_end(&sq->stats.syncp); > } > > @@ -1003,7 +1010,9 @@ static void check_sq_full_and_disable(struct virtnet_info *vi, > * early means 16 slots are typically wasted. > */ > if (sq->vq->num_free < 2+MAX_SKB_FRAGS) { > - netif_stop_subqueue(dev, qnum); > + struct netdev_queue *txq = netdev_get_tx_queue(dev, qnum); > + > + netif_tx_stop_queue(txq); > u64_stats_update_begin(&sq->stats.syncp); > u64_stats_inc(&sq->stats.stop); > u64_stats_update_end(&sq->stats.syncp); > @@ -1012,7 +1021,7 @@ static void check_sq_full_and_disable(struct virtnet_info *vi, > virtqueue_napi_schedule(&sq->napi, sq->vq); > } else if (unlikely(!virtqueue_enable_cb_delayed(sq->vq))) { > /* More just got used, free them then recheck. */ > - free_old_xmit(sq, false); > + free_old_xmit(sq, txq, false, use_napi); > if (sq->vq->num_free >= 2+MAX_SKB_FRAGS) { > netif_start_subqueue(dev, qnum); > u64_stats_update_begin(&sq->stats.syncp); > @@ -1138,7 +1147,8 @@ static int virtnet_xdp_xmit(struct net_device *dev, > } > > /* Free up any pending old buffers before queueing new ones. */ > - __free_old_xmit(sq, false, &stats); > + __free_old_xmit(sq, netdev_get_tx_queue(dev, sq - vi->sq), > + false, sq->napi.weight, &stats); > > for (i = 0; i < n; i++) { > struct xdp_frame *xdpf = frames[i]; > @@ -2313,7 +2323,7 @@ static void virtnet_poll_cleantx(struct receive_queue *rq) > > do { > virtqueue_disable_cb(sq->vq); > - free_old_xmit(sq, true); > + free_old_xmit(sq, txq, true, true); > } while (unlikely(!virtqueue_enable_cb_delayed(sq->vq))); > > if (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) { > @@ -2412,6 +2422,7 @@ static int virtnet_enable_queue_pair(struct virtnet_info *vi, int qp_index) > goto err_xdp_reg_mem_model; > > virtnet_napi_enable(vi->rq[qp_index].vq, &vi->rq[qp_index].napi); > + netdev_tx_reset_queue(netdev_get_tx_queue(vi->dev, qp_index)); > virtnet_napi_tx_enable(vi, vi->sq[qp_index].vq, &vi->sq[qp_index].napi); > > return 0; > @@ -2471,7 +2482,7 @@ static int virtnet_poll_tx(struct napi_struct *napi, int budget) > txq = netdev_get_tx_queue(vi->dev, index); > __netif_tx_lock(txq, raw_smp_processor_id()); > virtqueue_disable_cb(sq->vq); > - free_old_xmit(sq, true); > + free_old_xmit(sq, txq, true, true); > > if (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) { > if (netif_tx_queue_stopped(txq)) { > @@ -2559,17 +2570,18 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev) > struct send_queue *sq = &vi->sq[qnum]; > int err; > struct netdev_queue *txq = netdev_get_tx_queue(dev, qnum); > - bool kick = !netdev_xmit_more(); > + bool xmit_more = netdev_xmit_more(); > bool use_napi = sq->napi.weight; > + bool kick; > > /* Free up any pending old buffers before queueing new ones. */ > do { > if (use_napi) > virtqueue_disable_cb(sq->vq); > > - free_old_xmit(sq, false); > + free_old_xmit(sq, txq, false, use_napi); > > - } while (use_napi && kick && > + } while (use_napi && !xmit_more && > unlikely(!virtqueue_enable_cb_delayed(sq->vq))); > > /* timestamp packet in software */ > @@ -2598,7 +2610,9 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev) > > check_sq_full_and_disable(vi, dev, sq); > > - if (kick || netif_xmit_stopped(txq)) { > + kick = use_napi ? __netdev_tx_sent_queue(txq, skb->len, xmit_more) : > + !xmit_more && netif_xmit_stopped(txq); > + if (kick) { > if (virtqueue_kick_prepare(sq->vq) && virtqueue_notify(sq->vq)) { > u64_stats_update_begin(&sq->stats.syncp); > u64_stats_inc(&sq->stats.kicks); > -- > 2.45.1
Hello Jiri, On Thu, Jun 13, 2024 at 1:08 AM Jiri Pirko <jiri@resnulli.us> wrote: > > From: Jiri Pirko <jiri@nvidia.com> > > Add support for Byte Queue Limits (BQL). > > Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. > Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is > running in background. Netperf TCP_RR results: > > NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 > NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 > NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 > NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 > NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 > NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 > BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 > BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 > BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 > BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 > BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 > BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 I cannot get such a huge improvement when I was doing multiple tests between two VMs. I'm pretty sure the BQL feature is working, but the numbers look the same with/without BQL. VM 1 (client): 16 cpus, x86_64, 4 queues, the latest net-next kernel with/without this patch, pfifo_fast, napi_tx=true, napi_weight=128 VM 2 (server): 16 cpus, aarch64, 4 queues, the latest net-next kernel without this patch, pfifo_fast What the 'ping' command shows to me between two VMs is : rtt min/avg/max/mdev = 0.233/0.257/0.300/0.024 ms I started 50 netperfs to communicate the other side with the following command: #!/bin/bash for i in $(seq 5000 5050); do netperf -p $i -H [ip addr] -l 60 -t TCP_RR -- -r 64,64 > /dev/null 2>&1 & done The results are around 30423.62 txkB/s. If I remove '-r 64 64', they are still the same/similar. Am I missing something? Thanks, Jason
On Thu, Jun 13, 2024 at 1:09 AM Jiri Pirko <jiri@resnulli.us> wrote: > > From: Jiri Pirko <jiri@nvidia.com> > > Add support for Byte Queue Limits (BQL). > > Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. > Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is > running in background. Netperf TCP_RR results: > > NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 > NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 > NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 > NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 > NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 > NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 > BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 > BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 > BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 > BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 > BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 > BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 > > Signed-off-by: Jiri Pirko <jiri@nvidia.com> > --- > v1->v2: > - moved netdev_tx_completed_queue() call into __free_old_xmit(), > propagate use_napi flag to __free_old_xmit() and only call > netdev_tx_completed_queue() in case it is true > - added forgotten call to netdev_tx_reset_queue() > - fixed stats for xdp packets > - fixed bql accounting when __free_old_xmit() is called from xdp path > - handle the !use_napi case in start_xmit() kick section > --- > drivers/net/virtio_net.c | 50 +++++++++++++++++++++++++--------------- > 1 file changed, 32 insertions(+), 18 deletions(-) > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c > index 61a57d134544..5863c663ccab 100644 > --- a/drivers/net/virtio_net.c > +++ b/drivers/net/virtio_net.c > @@ -84,7 +84,9 @@ struct virtnet_stat_desc { > > struct virtnet_sq_free_stats { > u64 packets; > + u64 xdp_packets; > u64 bytes; > + u64 xdp_bytes; > }; > > struct virtnet_sq_stats { > @@ -506,29 +508,33 @@ static struct xdp_frame *ptr_to_xdp(void *ptr) > return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG); > } > > -static void __free_old_xmit(struct send_queue *sq, bool in_napi, > +static void __free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, > + bool in_napi, bool use_napi, > struct virtnet_sq_free_stats *stats) > { > unsigned int len; > void *ptr; > > while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) { > - ++stats->packets; > - > if (!is_xdp_frame(ptr)) { > struct sk_buff *skb = ptr; > > pr_debug("Sent skb %p\n", skb); > > + stats->packets++; > stats->bytes += skb->len; > napi_consume_skb(skb, in_napi); > } else { > struct xdp_frame *frame = ptr_to_xdp(ptr); > > - stats->bytes += xdp_get_frame_len(frame); > + stats->xdp_packets++; > + stats->xdp_bytes += xdp_get_frame_len(frame); > xdp_return_frame(frame); > } > } > + if (use_napi) > + netdev_tx_completed_queue(txq, stats->packets, stats->bytes); > + > } I wonder if this works correctly, for example NAPI could be enabled after queued but before sent. So __netdev_tx_sent_queue() is not called before. Thanks
Fri, Jun 14, 2024 at 11:54:04AM CEST, kerneljasonxing@gmail.com wrote: >Hello Jiri, > >On Thu, Jun 13, 2024 at 1:08 AM Jiri Pirko <jiri@resnulli.us> wrote: >> >> From: Jiri Pirko <jiri@nvidia.com> >> >> Add support for Byte Queue Limits (BQL). >> >> Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. >> Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is >> running in background. Netperf TCP_RR results: >> >> NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 >> NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 >> NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 >> NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 >> NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 >> NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 >> BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 >> BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 >> BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 >> BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 >> BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 >> BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 > >I cannot get such a huge improvement when I was doing multiple tests >between two VMs. I'm pretty sure the BQL feature is working, but the >numbers look the same with/without BQL. > >VM 1 (client): >16 cpus, x86_64, 4 queues, the latest net-next kernel with/without >this patch, pfifo_fast, napi_tx=true, napi_weight=128 > >VM 2 (server): >16 cpus, aarch64, 4 queues, the latest net-next kernel without this >patch, pfifo_fast > >What the 'ping' command shows to me between two VMs is : rtt >min/avg/max/mdev = 0.233/0.257/0.300/0.024 ms > >I started 50 netperfs to communicate the other side with the following command: >#!/bin/bash > >for i in $(seq 5000 5050); >do >netperf -p $i -H [ip addr] -l 60 -t TCP_RR -- -r 64,64 > /dev/null 2>&1 & >done > >The results are around 30423.62 txkB/s. If I remove '-r 64 64', they >are still the same/similar. You have to stress the line by parallel TCP_STREAM instances (50 in my case). For consistent results, use -p portnum,locport to specify the local port. Then run TCP_RR, also use -p portnum,locport to specify the local port. Also, double check the CONFIG_BQL=y is set. > >Am I missing something? > >Thanks, >Jason
Mon, Jun 17, 2024 at 04:34:26AM CEST, jasowang@redhat.com wrote: >On Thu, Jun 13, 2024 at 1:09 AM Jiri Pirko <jiri@resnulli.us> wrote: >> >> From: Jiri Pirko <jiri@nvidia.com> >> >> Add support for Byte Queue Limits (BQL). >> >> Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. >> Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is >> running in background. Netperf TCP_RR results: >> >> NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 >> NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 >> NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 >> NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 >> NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 >> NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 >> BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 >> BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 >> BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 >> BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 >> BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 >> BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 >> >> Signed-off-by: Jiri Pirko <jiri@nvidia.com> >> --- >> v1->v2: >> - moved netdev_tx_completed_queue() call into __free_old_xmit(), >> propagate use_napi flag to __free_old_xmit() and only call >> netdev_tx_completed_queue() in case it is true >> - added forgotten call to netdev_tx_reset_queue() >> - fixed stats for xdp packets >> - fixed bql accounting when __free_old_xmit() is called from xdp path >> - handle the !use_napi case in start_xmit() kick section >> --- >> drivers/net/virtio_net.c | 50 +++++++++++++++++++++++++--------------- >> 1 file changed, 32 insertions(+), 18 deletions(-) >> >> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c >> index 61a57d134544..5863c663ccab 100644 >> --- a/drivers/net/virtio_net.c >> +++ b/drivers/net/virtio_net.c >> @@ -84,7 +84,9 @@ struct virtnet_stat_desc { >> >> struct virtnet_sq_free_stats { >> u64 packets; >> + u64 xdp_packets; >> u64 bytes; >> + u64 xdp_bytes; >> }; >> >> struct virtnet_sq_stats { >> @@ -506,29 +508,33 @@ static struct xdp_frame *ptr_to_xdp(void *ptr) >> return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG); >> } >> >> -static void __free_old_xmit(struct send_queue *sq, bool in_napi, >> +static void __free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, >> + bool in_napi, bool use_napi, >> struct virtnet_sq_free_stats *stats) >> { >> unsigned int len; >> void *ptr; >> >> while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) { >> - ++stats->packets; >> - >> if (!is_xdp_frame(ptr)) { >> struct sk_buff *skb = ptr; >> >> pr_debug("Sent skb %p\n", skb); >> >> + stats->packets++; >> stats->bytes += skb->len; >> napi_consume_skb(skb, in_napi); >> } else { >> struct xdp_frame *frame = ptr_to_xdp(ptr); >> >> - stats->bytes += xdp_get_frame_len(frame); >> + stats->xdp_packets++; >> + stats->xdp_bytes += xdp_get_frame_len(frame); >> xdp_return_frame(frame); >> } >> } >> + if (use_napi) >> + netdev_tx_completed_queue(txq, stats->packets, stats->bytes); >> + >> } > >I wonder if this works correctly, for example NAPI could be enabled >after queued but before sent. So __netdev_tx_sent_queue() is not >called before. How is that possible? Napi weight can't change when link is up. Or am I missing something? > >Thanks >
On Mon, Jun 17, 2024 at 5:15 PM Jiri Pirko <jiri@resnulli.us> wrote: > > Fri, Jun 14, 2024 at 11:54:04AM CEST, kerneljasonxing@gmail.com wrote: > >Hello Jiri, > > > >On Thu, Jun 13, 2024 at 1:08 AM Jiri Pirko <jiri@resnulli.us> wrote: > >> > >> From: Jiri Pirko <jiri@nvidia.com> > >> > >> Add support for Byte Queue Limits (BQL). > >> > >> Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. > >> Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is > >> running in background. Netperf TCP_RR results: > >> > >> NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 > >> NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 > >> NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 > >> NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 > >> NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 > >> NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 > >> BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 > >> BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 > >> BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 > >> BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 > >> BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 > >> BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 > > > >I cannot get such a huge improvement when I was doing multiple tests > >between two VMs. I'm pretty sure the BQL feature is working, but the > >numbers look the same with/without BQL. > > > >VM 1 (client): > >16 cpus, x86_64, 4 queues, the latest net-next kernel with/without > >this patch, pfifo_fast, napi_tx=true, napi_weight=128 > > > >VM 2 (server): > >16 cpus, aarch64, 4 queues, the latest net-next kernel without this > >patch, pfifo_fast > > > >What the 'ping' command shows to me between two VMs is : rtt > >min/avg/max/mdev = 0.233/0.257/0.300/0.024 ms > > > >I started 50 netperfs to communicate the other side with the following command: > >#!/bin/bash > > > >for i in $(seq 5000 5050); > >do > >netperf -p $i -H [ip addr] -l 60 -t TCP_RR -- -r 64,64 > /dev/null 2>&1 & > >done > > > >The results are around 30423.62 txkB/s. If I remove '-r 64 64', they > >are still the same/similar. > > You have to stress the line by parallel TCP_STREAM instances (50 in my > case). For consistent results, use -p portnum,locport to specify the > local port. Thanks. Even though the results of TCP_RR mode vary sometimes, I can see a big improvement in the total value of those results under such circumstances. With BQL, the throughput is 2159.17 Without BQL, it's 1099.33 Please feel free to add the tag: Tested-by: Jason Xing <kerneljasonxing@gmail.com> Thanks, Jason
On Wed, Jun 12, 2024 at 07:08:51PM +0200, Jiri Pirko wrote: > From: Jiri Pirko <jiri@nvidia.com> > > Add support for Byte Queue Limits (BQL). > > Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. > Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is > running in background. Netperf TCP_RR results: > > NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 > NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 > NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 > NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 > NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 > NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 > BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 > BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 > BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 > BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 > BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 > BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 > > Signed-off-by: Jiri Pirko <jiri@nvidia.com> > --- > v1->v2: > - moved netdev_tx_completed_queue() call into __free_old_xmit(), > propagate use_napi flag to __free_old_xmit() and only call > netdev_tx_completed_queue() in case it is true > - added forgotten call to netdev_tx_reset_queue() > - fixed stats for xdp packets > - fixed bql accounting when __free_old_xmit() is called from xdp path > - handle the !use_napi case in start_xmit() kick section > --- > drivers/net/virtio_net.c | 50 +++++++++++++++++++++++++--------------- > 1 file changed, 32 insertions(+), 18 deletions(-) > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c > index 61a57d134544..5863c663ccab 100644 > --- a/drivers/net/virtio_net.c > +++ b/drivers/net/virtio_net.c > @@ -84,7 +84,9 @@ struct virtnet_stat_desc { > > struct virtnet_sq_free_stats { > u64 packets; > + u64 xdp_packets; > u64 bytes; > + u64 xdp_bytes; > }; > > struct virtnet_sq_stats { > @@ -506,29 +508,33 @@ static struct xdp_frame *ptr_to_xdp(void *ptr) > return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG); > } > > -static void __free_old_xmit(struct send_queue *sq, bool in_napi, > +static void __free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, > + bool in_napi, bool use_napi, > struct virtnet_sq_free_stats *stats) > { > unsigned int len; > void *ptr; > > while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) { > - ++stats->packets; > - > if (!is_xdp_frame(ptr)) { > struct sk_buff *skb = ptr; > > pr_debug("Sent skb %p\n", skb); > > + stats->packets++; > stats->bytes += skb->len; > napi_consume_skb(skb, in_napi); > } else { > struct xdp_frame *frame = ptr_to_xdp(ptr); > > - stats->bytes += xdp_get_frame_len(frame); > + stats->xdp_packets++; > + stats->xdp_bytes += xdp_get_frame_len(frame); > xdp_return_frame(frame); > } > } > + if (use_napi) > + netdev_tx_completed_queue(txq, stats->packets, stats->bytes); > + > } > > /* Converting between virtqueue no. and kernel tx/rx queue no. > @@ -955,21 +961,22 @@ static void virtnet_rq_unmap_free_buf(struct virtqueue *vq, void *buf) > virtnet_rq_free_buf(vi, rq, buf); > } > > -static void free_old_xmit(struct send_queue *sq, bool in_napi) > +static void free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, > + bool in_napi, bool use_napi) > { > struct virtnet_sq_free_stats stats = {0}; > > - __free_old_xmit(sq, in_napi, &stats); > + __free_old_xmit(sq, txq, in_napi, use_napi, &stats); > > /* Avoid overhead when no packets have been processed > * happens when called speculatively from start_xmit. > */ > - if (!stats.packets) > + if (!stats.packets && !stats.xdp_packets) > return; > > u64_stats_update_begin(&sq->stats.syncp); > - u64_stats_add(&sq->stats.bytes, stats.bytes); > - u64_stats_add(&sq->stats.packets, stats.packets); > + u64_stats_add(&sq->stats.bytes, stats.bytes + stats.xdp_bytes); > + u64_stats_add(&sq->stats.packets, stats.packets + stats.xdp_packets); > u64_stats_update_end(&sq->stats.syncp); > } > > @@ -1003,7 +1010,9 @@ static void check_sq_full_and_disable(struct virtnet_info *vi, > * early means 16 slots are typically wasted. > */ > if (sq->vq->num_free < 2+MAX_SKB_FRAGS) { > - netif_stop_subqueue(dev, qnum); > + struct netdev_queue *txq = netdev_get_tx_queue(dev, qnum); > + > + netif_tx_stop_queue(txq); > u64_stats_update_begin(&sq->stats.syncp); > u64_stats_inc(&sq->stats.stop); > u64_stats_update_end(&sq->stats.syncp); > @@ -1012,7 +1021,7 @@ static void check_sq_full_and_disable(struct virtnet_info *vi, > virtqueue_napi_schedule(&sq->napi, sq->vq); > } else if (unlikely(!virtqueue_enable_cb_delayed(sq->vq))) { > /* More just got used, free them then recheck. */ > - free_old_xmit(sq, false); > + free_old_xmit(sq, txq, false, use_napi); > if (sq->vq->num_free >= 2+MAX_SKB_FRAGS) { > netif_start_subqueue(dev, qnum); > u64_stats_update_begin(&sq->stats.syncp); > @@ -1138,7 +1147,8 @@ static int virtnet_xdp_xmit(struct net_device *dev, > } > > /* Free up any pending old buffers before queueing new ones. */ > - __free_old_xmit(sq, false, &stats); > + __free_old_xmit(sq, netdev_get_tx_queue(dev, sq - vi->sq), > + false, sq->napi.weight, &stats); So what happens if sq->napi.weight changes while packet is in the queue? What prevents that? > > for (i = 0; i < n; i++) { > struct xdp_frame *xdpf = frames[i]; > @@ -2313,7 +2323,7 @@ static void virtnet_poll_cleantx(struct receive_queue *rq) > > do { > virtqueue_disable_cb(sq->vq); > - free_old_xmit(sq, true); > + free_old_xmit(sq, txq, true, true); > } while (unlikely(!virtqueue_enable_cb_delayed(sq->vq))); > > if (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) { > @@ -2412,6 +2422,7 @@ static int virtnet_enable_queue_pair(struct virtnet_info *vi, int qp_index) > goto err_xdp_reg_mem_model; > > virtnet_napi_enable(vi->rq[qp_index].vq, &vi->rq[qp_index].napi); > + netdev_tx_reset_queue(netdev_get_tx_queue(vi->dev, qp_index)); > virtnet_napi_tx_enable(vi, vi->sq[qp_index].vq, &vi->sq[qp_index].napi); > > return 0; > @@ -2471,7 +2482,7 @@ static int virtnet_poll_tx(struct napi_struct *napi, int budget) > txq = netdev_get_tx_queue(vi->dev, index); > __netif_tx_lock(txq, raw_smp_processor_id()); > virtqueue_disable_cb(sq->vq); > - free_old_xmit(sq, true); > + free_old_xmit(sq, txq, true, true); > > if (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) { > if (netif_tx_queue_stopped(txq)) { > @@ -2559,17 +2570,18 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev) > struct send_queue *sq = &vi->sq[qnum]; > int err; > struct netdev_queue *txq = netdev_get_tx_queue(dev, qnum); > - bool kick = !netdev_xmit_more(); > + bool xmit_more = netdev_xmit_more(); > bool use_napi = sq->napi.weight; > + bool kick; > > /* Free up any pending old buffers before queueing new ones. */ > do { > if (use_napi) > virtqueue_disable_cb(sq->vq); > > - free_old_xmit(sq, false); > + free_old_xmit(sq, txq, false, use_napi); > > - } while (use_napi && kick && > + } while (use_napi && !xmit_more && > unlikely(!virtqueue_enable_cb_delayed(sq->vq))); > > /* timestamp packet in software */ > @@ -2598,7 +2610,9 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev) > > check_sq_full_and_disable(vi, dev, sq); > > - if (kick || netif_xmit_stopped(txq)) { > + kick = use_napi ? __netdev_tx_sent_queue(txq, skb->len, xmit_more) : > + !xmit_more && netif_xmit_stopped(txq); > + if (kick) { > if (virtqueue_kick_prepare(sq->vq) && virtqueue_notify(sq->vq)) { > u64_stats_update_begin(&sq->stats.syncp); > u64_stats_inc(&sq->stats.kicks); > -- > 2.45.1
On Mon, Jun 17, 2024 at 5:18 PM Jiri Pirko <jiri@resnulli.us> wrote: > > Mon, Jun 17, 2024 at 04:34:26AM CEST, jasowang@redhat.com wrote: > >On Thu, Jun 13, 2024 at 1:09 AM Jiri Pirko <jiri@resnulli.us> wrote: > >> > >> From: Jiri Pirko <jiri@nvidia.com> > >> > >> Add support for Byte Queue Limits (BQL). > >> > >> Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. > >> Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is > >> running in background. Netperf TCP_RR results: > >> > >> NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 > >> NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 > >> NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 > >> NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 > >> NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 > >> NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 > >> BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 > >> BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 > >> BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 > >> BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 > >> BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 > >> BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 > >> > >> Signed-off-by: Jiri Pirko <jiri@nvidia.com> > >> --- > >> v1->v2: > >> - moved netdev_tx_completed_queue() call into __free_old_xmit(), > >> propagate use_napi flag to __free_old_xmit() and only call > >> netdev_tx_completed_queue() in case it is true > >> - added forgotten call to netdev_tx_reset_queue() > >> - fixed stats for xdp packets > >> - fixed bql accounting when __free_old_xmit() is called from xdp path > >> - handle the !use_napi case in start_xmit() kick section > >> --- > >> drivers/net/virtio_net.c | 50 +++++++++++++++++++++++++--------------- > >> 1 file changed, 32 insertions(+), 18 deletions(-) > >> > >> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c > >> index 61a57d134544..5863c663ccab 100644 > >> --- a/drivers/net/virtio_net.c > >> +++ b/drivers/net/virtio_net.c > >> @@ -84,7 +84,9 @@ struct virtnet_stat_desc { > >> > >> struct virtnet_sq_free_stats { > >> u64 packets; > >> + u64 xdp_packets; > >> u64 bytes; > >> + u64 xdp_bytes; > >> }; > >> > >> struct virtnet_sq_stats { > >> @@ -506,29 +508,33 @@ static struct xdp_frame *ptr_to_xdp(void *ptr) > >> return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG); > >> } > >> > >> -static void __free_old_xmit(struct send_queue *sq, bool in_napi, > >> +static void __free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, > >> + bool in_napi, bool use_napi, > >> struct virtnet_sq_free_stats *stats) > >> { > >> unsigned int len; > >> void *ptr; > >> > >> while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) { > >> - ++stats->packets; > >> - > >> if (!is_xdp_frame(ptr)) { > >> struct sk_buff *skb = ptr; > >> > >> pr_debug("Sent skb %p\n", skb); > >> > >> + stats->packets++; > >> stats->bytes += skb->len; > >> napi_consume_skb(skb, in_napi); > >> } else { > >> struct xdp_frame *frame = ptr_to_xdp(ptr); > >> > >> - stats->bytes += xdp_get_frame_len(frame); > >> + stats->xdp_packets++; > >> + stats->xdp_bytes += xdp_get_frame_len(frame); > >> xdp_return_frame(frame); > >> } > >> } > >> + if (use_napi) > >> + netdev_tx_completed_queue(txq, stats->packets, stats->bytes); > >> + > >> } > > > >I wonder if this works correctly, for example NAPI could be enabled > >after queued but before sent. So __netdev_tx_sent_queue() is not > >called before. > > How is that possible? Napi weight can't change when link is up. Or am I > missing something? Something like this: 1) packet were queued 2) if down 3) enable NAPI 4) if up 5) packet were sent ? Thanks > > > > >Thanks > > >
Mon, Jun 17, 2024 at 06:14:11PM CEST, mst@redhat.com wrote: >On Wed, Jun 12, 2024 at 07:08:51PM +0200, Jiri Pirko wrote: >> From: Jiri Pirko <jiri@nvidia.com> >> >> Add support for Byte Queue Limits (BQL). >> >> Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. >> Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is >> running in background. Netperf TCP_RR results: >> >> NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 >> NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 >> NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 >> NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 >> NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 >> NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 >> BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 >> BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 >> BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 >> BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 >> BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 >> BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 >> >> Signed-off-by: Jiri Pirko <jiri@nvidia.com> >> --- >> v1->v2: >> - moved netdev_tx_completed_queue() call into __free_old_xmit(), >> propagate use_napi flag to __free_old_xmit() and only call >> netdev_tx_completed_queue() in case it is true >> - added forgotten call to netdev_tx_reset_queue() >> - fixed stats for xdp packets >> - fixed bql accounting when __free_old_xmit() is called from xdp path >> - handle the !use_napi case in start_xmit() kick section >> --- >> drivers/net/virtio_net.c | 50 +++++++++++++++++++++++++--------------- >> 1 file changed, 32 insertions(+), 18 deletions(-) >> >> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c >> index 61a57d134544..5863c663ccab 100644 >> --- a/drivers/net/virtio_net.c >> +++ b/drivers/net/virtio_net.c >> @@ -84,7 +84,9 @@ struct virtnet_stat_desc { >> >> struct virtnet_sq_free_stats { >> u64 packets; >> + u64 xdp_packets; >> u64 bytes; >> + u64 xdp_bytes; >> }; >> >> struct virtnet_sq_stats { >> @@ -506,29 +508,33 @@ static struct xdp_frame *ptr_to_xdp(void *ptr) >> return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG); >> } >> >> -static void __free_old_xmit(struct send_queue *sq, bool in_napi, >> +static void __free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, >> + bool in_napi, bool use_napi, >> struct virtnet_sq_free_stats *stats) >> { >> unsigned int len; >> void *ptr; >> >> while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) { >> - ++stats->packets; >> - >> if (!is_xdp_frame(ptr)) { >> struct sk_buff *skb = ptr; >> >> pr_debug("Sent skb %p\n", skb); >> >> + stats->packets++; >> stats->bytes += skb->len; >> napi_consume_skb(skb, in_napi); >> } else { >> struct xdp_frame *frame = ptr_to_xdp(ptr); >> >> - stats->bytes += xdp_get_frame_len(frame); >> + stats->xdp_packets++; >> + stats->xdp_bytes += xdp_get_frame_len(frame); >> xdp_return_frame(frame); >> } >> } >> + if (use_napi) >> + netdev_tx_completed_queue(txq, stats->packets, stats->bytes); >> + >> } >> >> /* Converting between virtqueue no. and kernel tx/rx queue no. >> @@ -955,21 +961,22 @@ static void virtnet_rq_unmap_free_buf(struct virtqueue *vq, void *buf) >> virtnet_rq_free_buf(vi, rq, buf); >> } >> >> -static void free_old_xmit(struct send_queue *sq, bool in_napi) >> +static void free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, >> + bool in_napi, bool use_napi) >> { >> struct virtnet_sq_free_stats stats = {0}; >> >> - __free_old_xmit(sq, in_napi, &stats); >> + __free_old_xmit(sq, txq, in_napi, use_napi, &stats); >> >> /* Avoid overhead when no packets have been processed >> * happens when called speculatively from start_xmit. >> */ >> - if (!stats.packets) >> + if (!stats.packets && !stats.xdp_packets) >> return; >> >> u64_stats_update_begin(&sq->stats.syncp); >> - u64_stats_add(&sq->stats.bytes, stats.bytes); >> - u64_stats_add(&sq->stats.packets, stats.packets); >> + u64_stats_add(&sq->stats.bytes, stats.bytes + stats.xdp_bytes); >> + u64_stats_add(&sq->stats.packets, stats.packets + stats.xdp_packets); >> u64_stats_update_end(&sq->stats.syncp); >> } >> >> @@ -1003,7 +1010,9 @@ static void check_sq_full_and_disable(struct virtnet_info *vi, >> * early means 16 slots are typically wasted. >> */ >> if (sq->vq->num_free < 2+MAX_SKB_FRAGS) { >> - netif_stop_subqueue(dev, qnum); >> + struct netdev_queue *txq = netdev_get_tx_queue(dev, qnum); >> + >> + netif_tx_stop_queue(txq); >> u64_stats_update_begin(&sq->stats.syncp); >> u64_stats_inc(&sq->stats.stop); >> u64_stats_update_end(&sq->stats.syncp); >> @@ -1012,7 +1021,7 @@ static void check_sq_full_and_disable(struct virtnet_info *vi, >> virtqueue_napi_schedule(&sq->napi, sq->vq); >> } else if (unlikely(!virtqueue_enable_cb_delayed(sq->vq))) { >> /* More just got used, free them then recheck. */ >> - free_old_xmit(sq, false); >> + free_old_xmit(sq, txq, false, use_napi); >> if (sq->vq->num_free >= 2+MAX_SKB_FRAGS) { >> netif_start_subqueue(dev, qnum); >> u64_stats_update_begin(&sq->stats.syncp); >> @@ -1138,7 +1147,8 @@ static int virtnet_xdp_xmit(struct net_device *dev, >> } >> >> /* Free up any pending old buffers before queueing new ones. */ >> - __free_old_xmit(sq, false, &stats); >> + __free_old_xmit(sq, netdev_get_tx_queue(dev, sq - vi->sq), >> + false, sq->napi.weight, &stats); > > >So what happens if sq->napi.weight changes while packet is in the queue? >What prevents that? Sigh. You are correct. That is what Jason points out as well. > > > >> >> for (i = 0; i < n; i++) { >> struct xdp_frame *xdpf = frames[i]; >> @@ -2313,7 +2323,7 @@ static void virtnet_poll_cleantx(struct receive_queue *rq) >> >> do { >> virtqueue_disable_cb(sq->vq); >> - free_old_xmit(sq, true); >> + free_old_xmit(sq, txq, true, true); >> } while (unlikely(!virtqueue_enable_cb_delayed(sq->vq))); >> >> if (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) { >> @@ -2412,6 +2422,7 @@ static int virtnet_enable_queue_pair(struct virtnet_info *vi, int qp_index) >> goto err_xdp_reg_mem_model; >> >> virtnet_napi_enable(vi->rq[qp_index].vq, &vi->rq[qp_index].napi); >> + netdev_tx_reset_queue(netdev_get_tx_queue(vi->dev, qp_index)); >> virtnet_napi_tx_enable(vi, vi->sq[qp_index].vq, &vi->sq[qp_index].napi); >> >> return 0; >> @@ -2471,7 +2482,7 @@ static int virtnet_poll_tx(struct napi_struct *napi, int budget) >> txq = netdev_get_tx_queue(vi->dev, index); >> __netif_tx_lock(txq, raw_smp_processor_id()); >> virtqueue_disable_cb(sq->vq); >> - free_old_xmit(sq, true); >> + free_old_xmit(sq, txq, true, true); >> >> if (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) { >> if (netif_tx_queue_stopped(txq)) { >> @@ -2559,17 +2570,18 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev) >> struct send_queue *sq = &vi->sq[qnum]; >> int err; >> struct netdev_queue *txq = netdev_get_tx_queue(dev, qnum); >> - bool kick = !netdev_xmit_more(); >> + bool xmit_more = netdev_xmit_more(); >> bool use_napi = sq->napi.weight; >> + bool kick; >> >> /* Free up any pending old buffers before queueing new ones. */ >> do { >> if (use_napi) >> virtqueue_disable_cb(sq->vq); >> >> - free_old_xmit(sq, false); >> + free_old_xmit(sq, txq, false, use_napi); >> >> - } while (use_napi && kick && >> + } while (use_napi && !xmit_more && >> unlikely(!virtqueue_enable_cb_delayed(sq->vq))); >> >> /* timestamp packet in software */ >> @@ -2598,7 +2610,9 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev) >> >> check_sq_full_and_disable(vi, dev, sq); >> >> - if (kick || netif_xmit_stopped(txq)) { >> + kick = use_napi ? __netdev_tx_sent_queue(txq, skb->len, xmit_more) : >> + !xmit_more && netif_xmit_stopped(txq); >> + if (kick) { >> if (virtqueue_kick_prepare(sq->vq) && virtqueue_notify(sq->vq)) { >> u64_stats_update_begin(&sq->stats.syncp); >> u64_stats_inc(&sq->stats.kicks); >> -- >> 2.45.1 >
Tue, Jun 18, 2024 at 02:53:42AM CEST, jasowang@redhat.com wrote: >On Mon, Jun 17, 2024 at 5:18 PM Jiri Pirko <jiri@resnulli.us> wrote: >> >> Mon, Jun 17, 2024 at 04:34:26AM CEST, jasowang@redhat.com wrote: >> >On Thu, Jun 13, 2024 at 1:09 AM Jiri Pirko <jiri@resnulli.us> wrote: >> >> >> >> From: Jiri Pirko <jiri@nvidia.com> >> >> >> >> Add support for Byte Queue Limits (BQL). >> >> >> >> Tested on qemu emulated virtio_net device with 1, 2 and 4 queues. >> >> Tested with fq_codel and pfifo_fast. Super netperf with 50 threads is >> >> running in background. Netperf TCP_RR results: >> >> >> >> NOBQL FQC 1q: 159.56 159.33 158.50 154.31 agv: 157.925 >> >> NOBQL FQC 2q: 184.64 184.96 174.73 174.15 agv: 179.62 >> >> NOBQL FQC 4q: 994.46 441.96 416.50 499.56 agv: 588.12 >> >> NOBQL PFF 1q: 148.68 148.92 145.95 149.48 agv: 148.2575 >> >> NOBQL PFF 2q: 171.86 171.20 170.42 169.42 agv: 170.725 >> >> NOBQL PFF 4q: 1505.23 1137.23 2488.70 3507.99 agv: 2159.7875 >> >> BQL FQC 1q: 1332.80 1297.97 1351.41 1147.57 agv: 1282.4375 >> >> BQL FQC 2q: 768.30 817.72 864.43 974.40 agv: 856.2125 >> >> BQL FQC 4q: 945.66 942.68 878.51 822.82 agv: 897.4175 >> >> BQL PFF 1q: 149.69 151.49 149.40 147.47 agv: 149.5125 >> >> BQL PFF 2q: 2059.32 798.74 1844.12 381.80 agv: 1270.995 >> >> BQL PFF 4q: 1871.98 4420.02 4916.59 13268.16 agv: 6119.1875 >> >> >> >> Signed-off-by: Jiri Pirko <jiri@nvidia.com> >> >> --- >> >> v1->v2: >> >> - moved netdev_tx_completed_queue() call into __free_old_xmit(), >> >> propagate use_napi flag to __free_old_xmit() and only call >> >> netdev_tx_completed_queue() in case it is true >> >> - added forgotten call to netdev_tx_reset_queue() >> >> - fixed stats for xdp packets >> >> - fixed bql accounting when __free_old_xmit() is called from xdp path >> >> - handle the !use_napi case in start_xmit() kick section >> >> --- >> >> drivers/net/virtio_net.c | 50 +++++++++++++++++++++++++--------------- >> >> 1 file changed, 32 insertions(+), 18 deletions(-) >> >> >> >> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c >> >> index 61a57d134544..5863c663ccab 100644 >> >> --- a/drivers/net/virtio_net.c >> >> +++ b/drivers/net/virtio_net.c >> >> @@ -84,7 +84,9 @@ struct virtnet_stat_desc { >> >> >> >> struct virtnet_sq_free_stats { >> >> u64 packets; >> >> + u64 xdp_packets; >> >> u64 bytes; >> >> + u64 xdp_bytes; >> >> }; >> >> >> >> struct virtnet_sq_stats { >> >> @@ -506,29 +508,33 @@ static struct xdp_frame *ptr_to_xdp(void *ptr) >> >> return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG); >> >> } >> >> >> >> -static void __free_old_xmit(struct send_queue *sq, bool in_napi, >> >> +static void __free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, >> >> + bool in_napi, bool use_napi, >> >> struct virtnet_sq_free_stats *stats) >> >> { >> >> unsigned int len; >> >> void *ptr; >> >> >> >> while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) { >> >> - ++stats->packets; >> >> - >> >> if (!is_xdp_frame(ptr)) { >> >> struct sk_buff *skb = ptr; >> >> >> >> pr_debug("Sent skb %p\n", skb); >> >> >> >> + stats->packets++; >> >> stats->bytes += skb->len; >> >> napi_consume_skb(skb, in_napi); >> >> } else { >> >> struct xdp_frame *frame = ptr_to_xdp(ptr); >> >> >> >> - stats->bytes += xdp_get_frame_len(frame); >> >> + stats->xdp_packets++; >> >> + stats->xdp_bytes += xdp_get_frame_len(frame); >> >> xdp_return_frame(frame); >> >> } >> >> } >> >> + if (use_napi) >> >> + netdev_tx_completed_queue(txq, stats->packets, stats->bytes); >> >> + >> >> } >> > >> >I wonder if this works correctly, for example NAPI could be enabled >> >after queued but before sent. So __netdev_tx_sent_queue() is not >> >called before. >> >> How is that possible? Napi weight can't change when link is up. Or am I >> missing something? > >Something like this: > >1) packet were queued >2) if down >3) enable NAPI >4) if up >5) packet were sent Gotcha, will try to fix. Thanks! > >? > >Thanks > >> >> > >> >Thanks >> > >> >
diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index 61a57d134544..5863c663ccab 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -84,7 +84,9 @@ struct virtnet_stat_desc { struct virtnet_sq_free_stats { u64 packets; + u64 xdp_packets; u64 bytes; + u64 xdp_bytes; }; struct virtnet_sq_stats { @@ -506,29 +508,33 @@ static struct xdp_frame *ptr_to_xdp(void *ptr) return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG); } -static void __free_old_xmit(struct send_queue *sq, bool in_napi, +static void __free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, + bool in_napi, bool use_napi, struct virtnet_sq_free_stats *stats) { unsigned int len; void *ptr; while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) { - ++stats->packets; - if (!is_xdp_frame(ptr)) { struct sk_buff *skb = ptr; pr_debug("Sent skb %p\n", skb); + stats->packets++; stats->bytes += skb->len; napi_consume_skb(skb, in_napi); } else { struct xdp_frame *frame = ptr_to_xdp(ptr); - stats->bytes += xdp_get_frame_len(frame); + stats->xdp_packets++; + stats->xdp_bytes += xdp_get_frame_len(frame); xdp_return_frame(frame); } } + if (use_napi) + netdev_tx_completed_queue(txq, stats->packets, stats->bytes); + } /* Converting between virtqueue no. and kernel tx/rx queue no. @@ -955,21 +961,22 @@ static void virtnet_rq_unmap_free_buf(struct virtqueue *vq, void *buf) virtnet_rq_free_buf(vi, rq, buf); } -static void free_old_xmit(struct send_queue *sq, bool in_napi) +static void free_old_xmit(struct send_queue *sq, struct netdev_queue *txq, + bool in_napi, bool use_napi) { struct virtnet_sq_free_stats stats = {0}; - __free_old_xmit(sq, in_napi, &stats); + __free_old_xmit(sq, txq, in_napi, use_napi, &stats); /* Avoid overhead when no packets have been processed * happens when called speculatively from start_xmit. */ - if (!stats.packets) + if (!stats.packets && !stats.xdp_packets) return; u64_stats_update_begin(&sq->stats.syncp); - u64_stats_add(&sq->stats.bytes, stats.bytes); - u64_stats_add(&sq->stats.packets, stats.packets); + u64_stats_add(&sq->stats.bytes, stats.bytes + stats.xdp_bytes); + u64_stats_add(&sq->stats.packets, stats.packets + stats.xdp_packets); u64_stats_update_end(&sq->stats.syncp); } @@ -1003,7 +1010,9 @@ static void check_sq_full_and_disable(struct virtnet_info *vi, * early means 16 slots are typically wasted. */ if (sq->vq->num_free < 2+MAX_SKB_FRAGS) { - netif_stop_subqueue(dev, qnum); + struct netdev_queue *txq = netdev_get_tx_queue(dev, qnum); + + netif_tx_stop_queue(txq); u64_stats_update_begin(&sq->stats.syncp); u64_stats_inc(&sq->stats.stop); u64_stats_update_end(&sq->stats.syncp); @@ -1012,7 +1021,7 @@ static void check_sq_full_and_disable(struct virtnet_info *vi, virtqueue_napi_schedule(&sq->napi, sq->vq); } else if (unlikely(!virtqueue_enable_cb_delayed(sq->vq))) { /* More just got used, free them then recheck. */ - free_old_xmit(sq, false); + free_old_xmit(sq, txq, false, use_napi); if (sq->vq->num_free >= 2+MAX_SKB_FRAGS) { netif_start_subqueue(dev, qnum); u64_stats_update_begin(&sq->stats.syncp); @@ -1138,7 +1147,8 @@ static int virtnet_xdp_xmit(struct net_device *dev, } /* Free up any pending old buffers before queueing new ones. */ - __free_old_xmit(sq, false, &stats); + __free_old_xmit(sq, netdev_get_tx_queue(dev, sq - vi->sq), + false, sq->napi.weight, &stats); for (i = 0; i < n; i++) { struct xdp_frame *xdpf = frames[i]; @@ -2313,7 +2323,7 @@ static void virtnet_poll_cleantx(struct receive_queue *rq) do { virtqueue_disable_cb(sq->vq); - free_old_xmit(sq, true); + free_old_xmit(sq, txq, true, true); } while (unlikely(!virtqueue_enable_cb_delayed(sq->vq))); if (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) { @@ -2412,6 +2422,7 @@ static int virtnet_enable_queue_pair(struct virtnet_info *vi, int qp_index) goto err_xdp_reg_mem_model; virtnet_napi_enable(vi->rq[qp_index].vq, &vi->rq[qp_index].napi); + netdev_tx_reset_queue(netdev_get_tx_queue(vi->dev, qp_index)); virtnet_napi_tx_enable(vi, vi->sq[qp_index].vq, &vi->sq[qp_index].napi); return 0; @@ -2471,7 +2482,7 @@ static int virtnet_poll_tx(struct napi_struct *napi, int budget) txq = netdev_get_tx_queue(vi->dev, index); __netif_tx_lock(txq, raw_smp_processor_id()); virtqueue_disable_cb(sq->vq); - free_old_xmit(sq, true); + free_old_xmit(sq, txq, true, true); if (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) { if (netif_tx_queue_stopped(txq)) { @@ -2559,17 +2570,18 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev) struct send_queue *sq = &vi->sq[qnum]; int err; struct netdev_queue *txq = netdev_get_tx_queue(dev, qnum); - bool kick = !netdev_xmit_more(); + bool xmit_more = netdev_xmit_more(); bool use_napi = sq->napi.weight; + bool kick; /* Free up any pending old buffers before queueing new ones. */ do { if (use_napi) virtqueue_disable_cb(sq->vq); - free_old_xmit(sq, false); + free_old_xmit(sq, txq, false, use_napi); - } while (use_napi && kick && + } while (use_napi && !xmit_more && unlikely(!virtqueue_enable_cb_delayed(sq->vq))); /* timestamp packet in software */ @@ -2598,7 +2610,9 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev) check_sq_full_and_disable(vi, dev, sq); - if (kick || netif_xmit_stopped(txq)) { + kick = use_napi ? __netdev_tx_sent_queue(txq, skb->len, xmit_more) : + !xmit_more && netif_xmit_stopped(txq); + if (kick) { if (virtqueue_kick_prepare(sq->vq) && virtqueue_notify(sq->vq)) { u64_stats_update_begin(&sq->stats.syncp); u64_stats_inc(&sq->stats.kicks);