From 11ef7a8996d5d433c9cd75d80651297eccbf6d42 Mon Sep 17 00:00:00 2001 From: Tom Herbert Date: Mon, 30 Jun 2014 09:50:40 -0700 Subject: net: Performance fix for process_backlog In process_backlog the input_pkt_queue is only checked once for new packets and quota is artificially reduced to reflect precisely the number of packets on the input_pkt_queue so that the loop exits appropriately. This patches changes the behavior to be more straightforward and less convoluted. Packets are processed until either the quota is met or there are no more packets to process. This patch seems to provide a small, but noticeable performance improvement. The performance improvement is a result of staying in the process_backlog loop longer which can reduce number of IPI's. Performance data using super_netperf TCP_RR with 200 flows: Before fix: 88.06% CPU utilization 125/190/309 90/95/99% latencies 1.46808e+06 tps 1145382 intrs.sec. With fix: 87.73% CPU utilization 122/183/296 90/95/99% latencies 1.4921e+06 tps 1021674.30 intrs./sec. Signed-off-by: Tom Herbert Acked-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/dev.c | 21 ++++++++++----------- 1 file changed, 10 insertions(+), 11 deletions(-) (limited to 'net/core') diff --git a/net/core/dev.c b/net/core/dev.c index 30eedf677913..77c19c7bb490 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -4227,9 +4227,8 @@ static int process_backlog(struct napi_struct *napi, int quota) #endif napi->weight = weight_p; local_irq_disable(); - while (work < quota) { + while (1) { struct sk_buff *skb; - unsigned int qlen; while ((skb = __skb_dequeue(&sd->process_queue))) { local_irq_enable(); @@ -4243,24 +4242,24 @@ static int process_backlog(struct napi_struct *napi, int quota) } rps_lock(sd); - qlen = skb_queue_len(&sd->input_pkt_queue); - if (qlen) - skb_queue_splice_tail_init(&sd->input_pkt_queue, - &sd->process_queue); - - if (qlen < quota - work) { + if (skb_queue_empty(&sd->input_pkt_queue)) { /* * Inline a custom version of __napi_complete(). * only current cpu owns and manipulates this napi, - * and NAPI_STATE_SCHED is the only possible flag set on backlog. - * we can use a plain write instead of clear_bit(), + * and NAPI_STATE_SCHED is the only possible flag set + * on backlog. + * We can use a plain write instead of clear_bit(), * and we dont need an smp_mb() memory barrier. */ list_del(&napi->poll_list); napi->state = 0; + rps_unlock(sd); - quota = work + qlen; + break; } + + skb_queue_splice_tail_init(&sd->input_pkt_queue, + &sd->process_queue); rps_unlock(sd); } local_irq_enable(); -- cgit v1.2.3 From 54951194656e4853e441266fd095f880bc0398f3 Mon Sep 17 00:00:00 2001 From: Loic Prylli Date: Tue, 1 Jul 2014 21:39:43 -0700 Subject: net: Fix NETDEV_CHANGE notifier usage causing spurious arp flush A bug was introduced in NETDEV_CHANGE notifier sequence causing the arp table to be sometimes spuriously cleared (including manual arp entries marked permanent), upon network link carrier changes. The changed argument for the notifier was applied only to a single caller of NETDEV_CHANGE, missing among others netdev_state_change(). So upon net_carrier events induced by the network, which are triggering a call to netdev_state_change(), arp_netdev_event() would decide whether to clear or not arp cache based on random/junk stack values (a kind of read buffer overflow). Fixes: be9efd365328 ("net: pass changed flags along with NETDEV_CHANGE event") Fixes: 6c8b4e3ff81b ("arp: flush arp cache on IFF_NOARP change") Signed-off-by: Loic Prylli Signed-off-by: David S. Miller --- net/core/dev.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'net/core') diff --git a/net/core/dev.c b/net/core/dev.c index 77c19c7bb490..7990984ca364 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -148,6 +148,9 @@ struct list_head ptype_all __read_mostly; /* Taps */ static struct list_head offload_base __read_mostly; static int netif_rx_internal(struct sk_buff *skb); +static int call_netdevice_notifiers_info(unsigned long val, + struct net_device *dev, + struct netdev_notifier_info *info); /* * The @dev_base_head list is protected by @dev_base_lock and the rtnl @@ -1207,7 +1210,11 @@ EXPORT_SYMBOL(netdev_features_change); void netdev_state_change(struct net_device *dev) { if (dev->flags & IFF_UP) { - call_netdevice_notifiers(NETDEV_CHANGE, dev); + struct netdev_notifier_change_info change_info; + + change_info.flags_changed = 0; + call_netdevice_notifiers_info(NETDEV_CHANGE, dev, + &change_info.info); rtmsg_ifinfo(RTM_NEWLINK, dev, 0, GFP_KERNEL); } } -- cgit v1.2.3 From 9ecf07a1d8f70f72ec99a0f102c8aa24609d84f4 Mon Sep 17 00:00:00 2001 From: Mathias Krause Date: Sat, 12 Jul 2014 22:36:44 +0200 Subject: neigh: sysctl - simplify address calculation of gc_* variables The code in neigh_sysctl_register() relies on a specific layout of struct neigh_table, namely that the 'gc_*' variables are directly following the 'parms' member in a specific order. The code, though, expresses this in the most ugly way. Get rid of the ugly casts and use the 'tbl' pointer to get a handle to the table. This way we can refer to the 'gc_*' variables directly. Similarly seen in the grsecurity patch, written by Brad Spengler. Signed-off-by: Mathias Krause Cc: Brad Spengler Signed-off-by: David S. Miller --- include/net/neighbour.h | 1 - net/core/neighbour.c | 9 +++++---- 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'net/core') diff --git a/include/net/neighbour.h b/include/net/neighbour.h index 7277caf3743d..47f425464f84 100644 --- a/include/net/neighbour.h +++ b/include/net/neighbour.h @@ -203,7 +203,6 @@ struct neigh_table { void (*proxy_redo)(struct sk_buff *skb); char *id; struct neigh_parms parms; - /* HACK. gc_* should follow parms without a gap! */ int gc_interval; int gc_thresh1; int gc_thresh2; diff --git a/net/core/neighbour.c b/net/core/neighbour.c index 32d872eec7f5..559890b0f0a2 100644 --- a/net/core/neighbour.c +++ b/net/core/neighbour.c @@ -3059,11 +3059,12 @@ int neigh_sysctl_register(struct net_device *dev, struct neigh_parms *p, memset(&t->neigh_vars[NEIGH_VAR_GC_INTERVAL], 0, sizeof(t->neigh_vars[NEIGH_VAR_GC_INTERVAL])); } else { + struct neigh_table *tbl = p->tbl; dev_name_source = "default"; - t->neigh_vars[NEIGH_VAR_GC_INTERVAL].data = (int *)(p + 1); - t->neigh_vars[NEIGH_VAR_GC_THRESH1].data = (int *)(p + 1) + 1; - t->neigh_vars[NEIGH_VAR_GC_THRESH2].data = (int *)(p + 1) + 2; - t->neigh_vars[NEIGH_VAR_GC_THRESH3].data = (int *)(p + 1) + 3; + t->neigh_vars[NEIGH_VAR_GC_INTERVAL].data = &tbl->gc_interval; + t->neigh_vars[NEIGH_VAR_GC_THRESH1].data = &tbl->gc_thresh1; + t->neigh_vars[NEIGH_VAR_GC_THRESH2].data = &tbl->gc_thresh2; + t->neigh_vars[NEIGH_VAR_GC_THRESH3].data = &tbl->gc_thresh3; } if (handler) { -- cgit v1.2.3 From c3caf1192f904de2f1381211f564537235d50de3 Mon Sep 17 00:00:00 2001 From: Jerry Chu Date: Mon, 14 Jul 2014 15:54:46 -0700 Subject: net-gre-gro: Fix a bug that breaks the forwarding path Fixed a bug that was introduced by my GRE-GRO patch (bf5a755f5e9186406bbf50f4087100af5bd68e40 net-gre-gro: Add GRE support to the GRO stack) that breaks the forwarding path because various GSO related fields were not set. The bug will cause on the egress path either the GSO code to fail, or a GRE-TSO capable (NETIF_F_GSO_GRE) NICs to choke. The following fix has been tested for both cases. Signed-off-by: H.K. Jerry Chu Signed-off-by: David S. Miller --- net/core/dev.c | 2 ++ net/ipv4/af_inet.c | 3 +++ net/ipv4/gre_offload.c | 3 +++ net/ipv4/tcp_offload.c | 2 +- net/ipv6/tcpv6_offload.c | 2 +- 5 files changed, 10 insertions(+), 2 deletions(-) (limited to 'net/core') diff --git a/net/core/dev.c b/net/core/dev.c index 7990984ca364..367a586d0c8a 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -4096,6 +4096,8 @@ static void napi_reuse_skb(struct napi_struct *napi, struct sk_buff *skb) skb->vlan_tci = 0; skb->dev = napi->dev; skb->skb_iif = 0; + skb->encapsulation = 0; + skb_shinfo(skb)->gso_type = 0; skb->truesize = SKB_TRUESIZE(skb_end_offset(skb)); napi->skb = skb; diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c index d5e6836cf772..d156b3c5f363 100644 --- a/net/ipv4/af_inet.c +++ b/net/ipv4/af_inet.c @@ -1429,6 +1429,9 @@ static int inet_gro_complete(struct sk_buff *skb, int nhoff) int proto = iph->protocol; int err = -ENOSYS; + if (skb->encapsulation) + skb_set_inner_network_header(skb, nhoff); + csum_replace2(&iph->check, iph->tot_len, newlen); iph->tot_len = newlen; diff --git a/net/ipv4/gre_offload.c b/net/ipv4/gre_offload.c index eb92deb12666..f0bdd47bbbcb 100644 --- a/net/ipv4/gre_offload.c +++ b/net/ipv4/gre_offload.c @@ -263,6 +263,9 @@ static int gre_gro_complete(struct sk_buff *skb, int nhoff) int err = -ENOENT; __be16 type; + skb->encapsulation = 1; + skb_shinfo(skb)->gso_type = SKB_GSO_GRE; + type = greh->protocol; if (greh->flags & GRE_KEY) grehlen += GRE_HEADER_SECTION; diff --git a/net/ipv4/tcp_offload.c b/net/ipv4/tcp_offload.c index 4e86c59ec7f7..55046ecd083e 100644 --- a/net/ipv4/tcp_offload.c +++ b/net/ipv4/tcp_offload.c @@ -309,7 +309,7 @@ static int tcp4_gro_complete(struct sk_buff *skb, int thoff) th->check = ~tcp_v4_check(skb->len - thoff, iph->saddr, iph->daddr, 0); - skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4; + skb_shinfo(skb)->gso_type |= SKB_GSO_TCPV4; return tcp_gro_complete(skb); } diff --git a/net/ipv6/tcpv6_offload.c b/net/ipv6/tcpv6_offload.c index 8517d3cd1aed..01b0ff9a0c2c 100644 --- a/net/ipv6/tcpv6_offload.c +++ b/net/ipv6/tcpv6_offload.c @@ -73,7 +73,7 @@ static int tcp6_gro_complete(struct sk_buff *skb, int thoff) th->check = ~tcp_v6_check(skb->len - thoff, &iph->saddr, &iph->daddr, 0); - skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6; + skb_shinfo(skb)->gso_type |= SKB_GSO_TCPV6; return tcp_gro_complete(skb); } -- cgit v1.2.3