search for: rx_batch

...G_MORE as a hint for under layer socket. The backend (e.g tap) can then batch the packets temporarily in a list and submit it all once the number of bacthed exceeds a limitation. Tests shows obvious improvement on guest pktgen over over mlx4(noqueue) on host: Mpps -+% rx_batched=0 0.90 +0% rx_batched=4 0.97 +7.8% rx_batched=8 0.97 +7.8% rx_batched=16 0.98 +8.9% rx_batched=32 1.03 +14.4% rx_batched=48 1.09 +21.1% rx_batched=64 1.02 +13.3% Changes from V2: - remove uselss queue limitation check (and we don't dr...

...G_MORE as a hint for under layer socket. The backend (e.g tap) can then batch the packets temporarily in a list and submit it all once the number of bacthed exceeds a limitation. Tests shows obvious improvement on guest pktgen over over mlx4(noqueue) on host: Mpps -+% rx_batched=0 0.90 +0% rx_batched=4 0.97 +7.8% rx_batched=8 0.97 +7.8% rx_batched=16 0.98 +8.9% rx_batched=32 1.03 +14.4% rx_batched=48 1.09 +21.1% rx_batched=64 1.02 +13.3% Changes from V2: - remove uselss queue limitation check (and we don't dr...

...k stack. This is done through accepting MSG_MORE as a hint from sendmsg() caller, if it was set, batch the packet temporarily in a linked list and submit them all once MSG_MORE were cleared. Tests were done by pktgen (burst=128) in guest over mlx4(noqueue) on host: Mpps -+% rx_batched=0 0.90 +0% rx_batched=4 0.97 +7.8% rx_batched=8 0.97 +7.8% rx_batched=16 0.98 +8.9% rx_batched=32 1.03 +14.4% rx_batched=48 1.09 +21.1% rx_batched=64 1.02 +13.3% The maximum number of batched packets were specified through a module parameter. Signed-off-by: Jas...

...G_MORE as a hint for under layer socket. The backend (e.g tap) can then batch the packets temporarily in a list and submit it all once the number of bacthed exceeds a limitation. Tests shows obvious improvement on guest pktgen over over mlx4(noqueue) on host: Mpps -+% rx_batched=0 0.90 +0% rx_batched=4 0.97 +7.8% rx_batched=8 0.97 +7.8% rx_batched=16 0.98 +8.9% rx_batched=32 1.03 +14.4% rx_batched=48 1.09 +21.1% rx_batched=64 1.02 +13.3% Changes from V1: - drop NAPI handler since we don't use NAPI now - fix t...

...G_MORE as a hint for under layer socket. The backend (e.g tap) can then batch the packets temporarily in a list and submit it all once the number of bacthed exceeds a limitation. Tests shows obvious improvement on guest pktgen over over mlx4(noqueue) on host: Mpps -+% rx_batched=0 0.90 +0% rx_batched=4 0.97 +7.8% rx_batched=8 0.97 +7.8% rx_batched=16 0.98 +8.9% rx_batched=32 1.03 +14.4% rx_batched=48 1.09 +21.1% rx_batched=64 1.02 +13.3% Changes from V1: - drop NAPI handler since we don't use NAPI now - fix t...

...k stack. This is done through accepting MSG_MORE as a hint from sendmsg() caller, if it was set, batch the packet temporarily in a linked list and submit them all once MSG_MORE were cleared. Tests were done by pktgen (burst=128) in guest over mlx4(noqueue) on host: Mpps -+% rx_batched=0 0.90 +0% rx_batched=4 0.97 +7.8% rx_batched=8 0.97 +7.8% rx_batched=16 0.98 +8.9% rx_batched=32 1.03 +14.4% rx_batched=48 1.09 +21.1% rx_batched=64 1.02 +13.3% The maximum number of batched packets were specified through a module parameter. Signed-off-by: Jas...

...+++----- 1 file changed, 70 insertions(+), 6 deletions(-) diff --git a/drivers/net/tun.c b/drivers/net/tun.c index 8c1d3bd..13890ac 100644 --- a/drivers/net/tun.c +++ b/drivers/net/tun.c @@ -218,6 +218,7 @@ struct tun_struct { struct list_head disabled; void *security; u32 flow_count; + u32 rx_batched; struct tun_pcpu_stats __percpu *pcpu_stats; }; @@ -522,6 +523,7 @@ static void tun_queue_purge(struct tun_file *tfile) while ((skb = skb_array_consume(&tfile->tx_array)) != NULL) kfree_skb(skb); + skb_queue_purge(&tfile->sk.sk_write_queue); skb_queue_purge(&tfile-...

...tions(-) > > diff --git a/drivers/net/tun.c b/drivers/net/tun.c > index cd8e02c..6c93926 100644 > --- a/drivers/net/tun.c > +++ b/drivers/net/tun.c > @@ -218,6 +218,7 @@ struct tun_struct { > struct list_head disabled; > void *security; > u32 flow_count; > + u32 rx_batched; > struct tun_pcpu_stats __percpu *pcpu_stats; > }; > > @@ -522,6 +523,7 @@ static void tun_queue_purge(struct tun_file *tfile) > while ((skb = skb_array_consume(&tfile->tx_array)) != NULL) > kfree_skb(skb); > > + skb_queue_purge(&tfile->sk.sk_wri...

...tions(-) > > diff --git a/drivers/net/tun.c b/drivers/net/tun.c > index cd8e02c..6c93926 100644 > --- a/drivers/net/tun.c > +++ b/drivers/net/tun.c > @@ -218,6 +218,7 @@ struct tun_struct { > struct list_head disabled; > void *security; > u32 flow_count; > + u32 rx_batched; > struct tun_pcpu_stats __percpu *pcpu_stats; > }; > > @@ -522,6 +523,7 @@ static void tun_queue_purge(struct tun_file *tfile) > while ((skb = skb_array_consume(&tfile->tx_array)) != NULL) > kfree_skb(skb); > > + skb_queue_purge(&tfile->sk.sk_wri...

...+++----- 1 file changed, 70 insertions(+), 6 deletions(-) diff --git a/drivers/net/tun.c b/drivers/net/tun.c index cd8e02c..6c93926 100644 --- a/drivers/net/tun.c +++ b/drivers/net/tun.c @@ -218,6 +218,7 @@ struct tun_struct { struct list_head disabled; void *security; u32 flow_count; + u32 rx_batched; struct tun_pcpu_stats __percpu *pcpu_stats; }; @@ -522,6 +523,7 @@ static void tun_queue_purge(struct tun_file *tfile) while ((skb = skb_array_consume(&tfile->tx_array)) != NULL) kfree_skb(skb); + skb_queue_purge(&tfile->sk.sk_write_queue); skb_queue_purge(&tfile-...

Hi: This series tries to implement tx batching support for vhost. This was done by using MSG_MORE as a hint for under layer socket. The backend (e.g tap) can then batch the packets temporarily in a list and submit it all once the number of bacthed exceeds a limitation. Tests shows obvious improvement on guest pktgen over over mlx4(noqueue) on host: Mpps -+%

Hi: This series tries to implement tx batching support for vhost. This was done by using MSG_MORE as a hint for under layer socket. The backend (e.g tap) can then batch the packets temporarily in a list and submit it all once the number of bacthed exceeds a limitation. Tests shows obvious improvement on guest pktgen over over mlx4(noqueue) on host: Mpps -+%

Hi: This series tries to implement tx batching support for vhost. This was done by using MSG_MORE as a hint for under layer socket. The backend (e.g tap) can then batch the packets temporarily in a list and submit it all once the number of bacthed exceeds a limitation. Tests shows obvious improvement on guest pktgen over over mlx4(noqueue) on host: Mpps -+%

Hi: This series tries to implement tx batching support for vhost. This was done by using MSG_MORE as a hint for under layer socket. The backend (e.g tap) can then batch the packets temporarily in a list and submit it all once the number of bacthed exceeds a limitation. Tests shows obvious improvement on guest pktgen over over mlx4(noqueue) on host: Mpps -+%

...+0800 Jason Wang <jasowang at redhat.com> wrote: > diff --git a/drivers/net/tun.c b/drivers/net/tun.c > index cd8e02c..a268ed9 100644 > --- a/drivers/net/tun.c > +++ b/drivers/net/tun.c > @@ -75,6 +75,10 @@ > > #include <linux/uaccess.h> > > +static int rx_batched; > +module_param(rx_batched, int, 0444); > +MODULE_PARM_DESC(rx_batched, "Number of packets batched in rx"); > + > /* Uncomment to enable debugging */ I like the concept or rx batching. But controlling it via a module parameter is one of the worst API choices. Ethtool wou...

...+0800 Jason Wang <jasowang at redhat.com> wrote: > diff --git a/drivers/net/tun.c b/drivers/net/tun.c > index cd8e02c..a268ed9 100644 > --- a/drivers/net/tun.c > +++ b/drivers/net/tun.c > @@ -75,6 +75,10 @@ > > #include <linux/uaccess.h> > > +static int rx_batched; > +module_param(rx_batched, int, 0444); > +MODULE_PARM_DESC(rx_batched, "Number of packets batched in rx"); > + > /* Uncomment to enable debugging */ I like the concept or rx batching. But controlling it via a module parameter is one of the worst API choices. Ethtool wou...

From: Jason Wang <jasowang at redhat.com> Date: Wed, 28 Dec 2016 16:09:31 +0800 > + spin_lock(&queue->lock); > + qlen = skb_queue_len(queue); > + if (qlen > rx_batched) > + goto drop; > + __skb_queue_tail(queue, skb); > + if (!more || qlen + 1 > rx_batched) { > + __skb_queue_head_init(&process_queue); > + skb_queue_splice_tail_init(queue, &process_queue); > + rcv = true; > + } > + spin_unlock(&queue->lock); Since y...

...15 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile, > skb_probe_transport_header(skb, 0); > > rxhash = skb_get_hash(skb); > + > #ifndef CONFIG_4KSTACKS > - local_bh_disable(); > - netif_receive_skb(skb); > - local_bh_enable(); > + if (!rx_batched) { > + local_bh_disable(); > + netif_receive_skb(skb); > + local_bh_enable(); > + } else { > + tun_rx_batched(tfile, skb, more); > + } > #else > netif_rx_ni(skb); > #endif If rx_batched has been set, and we are talking to clients not using this new MSG_MORE fac...

From: Jason Wang <jasowang at redhat.com> Date: Wed, 28 Dec 2016 16:09:31 +0800 > + spin_lock(&queue->lock); > + qlen = skb_queue_len(queue); > + if (qlen > rx_batched) > + goto drop; > + __skb_queue_tail(queue, skb); > + if (!more || qlen + 1 > rx_batched) { > + __skb_queue_head_init(&process_queue); > + skb_queue_splice_tail_init(queue, &process_queue); > + rcv = true; > + } > + spin_unlock(&queue->lock); Since y...

...15 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile, > skb_probe_transport_header(skb, 0); > > rxhash = skb_get_hash(skb); > + > #ifndef CONFIG_4KSTACKS > - local_bh_disable(); > - netif_receive_skb(skb); > - local_bh_enable(); > + if (!rx_batched) { > + local_bh_disable(); > + netif_receive_skb(skb); > + local_bh_enable(); > + } else { > + tun_rx_batched(tfile, skb, more); > + } > #else > netif_rx_ni(skb); > #endif If rx_batched has been set, and we are talking to clients not using this new MSG_MORE fac...