From 13bfff25c081f4e060af761c4082b5a96f756810 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 7 Dec 2016 08:29:10 -0800 Subject: [PATCH] net: rfs: add a jump label RFS is not commonly used, so add a jump label to avoid some conditionals in fast path. Signed-off-by: Eric Dumazet Cc: Paolo Abeni Signed-off-by: David S. Miller --- include/linux/netdevice.h | 1 + include/net/sock.h | 25 ++++++++++++++----------- net/core/dev.c | 2 ++ net/core/sysctl_net_core.c | 5 ++++- 4 files changed, 21 insertions(+), 12 deletions(-) diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index 1ff5ea6e1221..994f7423a74b 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -192,6 +192,7 @@ struct net_device_stats { #ifdef CONFIG_RPS #include extern struct static_key rps_needed; +extern struct static_key rfs_needed; #endif struct neighbour; diff --git a/include/net/sock.h b/include/net/sock.h index 1749e38d0301..2729e77950b7 100644 --- a/include/net/sock.h +++ b/include/net/sock.h @@ -913,17 +913,20 @@ static inline void sock_rps_record_flow_hash(__u32 hash) static inline void sock_rps_record_flow(const struct sock *sk) { #ifdef CONFIG_RPS - /* Reading sk->sk_rxhash might incur an expensive cache line miss. - * - * TCP_ESTABLISHED does cover almost all states where RFS - * might be useful, and is cheaper [1] than testing : - * IPv4: inet_sk(sk)->inet_daddr - * IPv6: ipv6_addr_any(&sk->sk_v6_daddr) - * OR an additional socket flag - * [1] : sk_state and sk_prot are in the same cache line. - */ - if (sk->sk_state == TCP_ESTABLISHED) - sock_rps_record_flow_hash(sk->sk_rxhash); + if (static_key_false(&rfs_needed)) { + /* Reading sk->sk_rxhash might incur an expensive cache line + * miss. + * + * TCP_ESTABLISHED does cover almost all states where RFS + * might be useful, and is cheaper [1] than testing : + * IPv4: inet_sk(sk)->inet_daddr + * IPv6: ipv6_addr_any(&sk->sk_v6_daddr) + * OR an additional socket flag + * [1] : sk_state and sk_prot are in the same cache line. + */ + if (sk->sk_state == TCP_ESTABLISHED) + sock_rps_record_flow_hash(sk->sk_rxhash); + } #endif } diff --git a/net/core/dev.c b/net/core/dev.c index bffb5253e778..1d33ce03365f 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -3447,6 +3447,8 @@ EXPORT_SYMBOL(rps_cpu_mask); struct static_key rps_needed __read_mostly; EXPORT_SYMBOL(rps_needed); +struct static_key rfs_needed __read_mostly; +EXPORT_SYMBOL(rfs_needed); static struct rps_dev_flow * set_rps_cpu(struct net_device *dev, struct sk_buff *skb, diff --git a/net/core/sysctl_net_core.c b/net/core/sysctl_net_core.c index 0df2aa652530..2a46e4009f62 100644 --- a/net/core/sysctl_net_core.c +++ b/net/core/sysctl_net_core.c @@ -79,10 +79,13 @@ static int rps_sock_flow_sysctl(struct ctl_table *table, int write, if (sock_table != orig_sock_table) { rcu_assign_pointer(rps_sock_flow_table, sock_table); - if (sock_table) + if (sock_table) { static_key_slow_inc(&rps_needed); + static_key_slow_inc(&rfs_needed); + } if (orig_sock_table) { static_key_slow_dec(&rps_needed); + static_key_slow_dec(&rfs_needed); synchronize_rcu(); vfree(orig_sock_table); } -- 2.20.1