tcp: add tcp_inet6_sk() helper
Subsystems (3)
PATCH 1/1 tcp: add tcp_inet6_sk() helper
TCP ipv6 fast path dereferences a pointer to get to the inet6 part of a tcp socket, but given the fixed memory placement, we can do better and avoid a possible cache line miss.
This also reduces register pressure, since we let the compiler know about this memory placement.
Signed-off-by Eric Dumazet <[email protected]>
--- net/ipv6/tcp_ipv6.c | 44 ++++++++++++++++++++++++++------------------ 1 file changed, 26 insertions(+), 18 deletions(-)
net/ipv6/tcp_ipv6.c +26 -18
@@ -90,6 +90,17 @@ static struct tcp_md5sig_key *tcp_v6_md5_do_lookup(const struct sock *sk, }#endif +/* Helper returning the inet6 address from a given tcp socket. +* It can be used in TCP stack instead of inet6_sk(sk). +* This avoids a dereference and allow compiler optimizations. +*/ +static struct ipv6_pinfo *tcp_inet6_sk (const struct sock *sk) +{ + struct tcp6_sock *tcp6 = container_of(tcp_sk(sk), struct tcp6_sock, tcp); + +return &tcp6->inet6; +} +static void inet6_sk_rx_dst_set (struct sock *sk,const struct sk_buff *skb) {struct dst_entry *dst =skb_dst (skb);
@@ -99,7 +110,7 @@ static void inet6_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb) sk->sk_rx_dst = dst;inet_sk (sk)->rx_dst_ifindex = skb->skb_iif; -inet6_sk (sk)->rx_dst_cookie =rt6_get_cookie (rt); +tcp_inet6_sk (sk)->rx_dst_cookie =rt6_get_cookie (rt); } }
@@ -138,7 +149,7 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,struct sockaddr_in6 *usin = (struct sockaddr_in6 *) uaddr;struct inet_sock *inet =inet_sk (sk);struct inet_connection_sock *icsk =inet_csk (sk); -struct ipv6_pinfo *np =inet6_sk (sk); +struct ipv6_pinfo *np =tcp_inet6_sk (sk);struct tcp_sock *tp =tcp_sk (sk);struct in6_addr *saddr =NULL , *final_p, final;struct ipv6_txoptions *opt;
@@ -390,7 +401,7 @@ static int tcp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,if (sk->sk_state ==TCP_CLOSE ) goto out; -if (ipv6_hdr (skb)->hop_limit <inet6_sk (sk)->min_hopcount) { +if (ipv6_hdr (skb)->hop_limit <tcp_inet6_sk (sk)->min_hopcount) {__NET_INC_STATS (net,LINUX_MIB_TCPMINTTLDROP ); goto out; }
@@ -405,7 +416,7 @@ static int tcp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt, goto out; } - np =inet6_sk (sk); + np =tcp_inet6_sk (sk);if (type ==NDISC_REDIRECT ) {if (!sock_owned_by_user (sk)) {
@@ -478,7 +489,7 @@ static int tcp_v6_send_synack(const struct sock *sk, struct dst_entry *dst, enum tcp_synack_type synack_type) {struct inet_request_sock *ireq =inet_rsk (req); -struct ipv6_pinfo *np =inet6_sk (sk); +struct ipv6_pinfo *np =tcp_inet6_sk (sk);struct ipv6_txoptions *opt;struct flowi6 *fl6 = &fl->u.ip6;struct sk_buff *skb;
@@ -737,7 +748,7 @@ static void tcp_v6_init_req(struct request_sock *req, {bool l3_slave =ipv6_l3mdev_skb (TCP_SKB_CB (skb)->header.h6.flags);struct inet_request_sock *ireq =inet_rsk (req); -const struct ipv6_pinfo *np =inet6_sk (sk_listener); +const struct ipv6_pinfo *np =tcp_inet6_sk (sk_listener); ireq->ir_v6_rmt_addr =ipv6_hdr (skb)->saddr; ireq->ir_v6_loc_addr =ipv6_hdr (skb)->daddr;
@@ -1066,9 +1077,8 @@ static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff * {struct inet_request_sock *ireq;struct ipv6_pinfo *newnp; -const struct ipv6_pinfo *np =inet6_sk (sk); +const struct ipv6_pinfo *np =tcp_inet6_sk (sk);struct ipv6_txoptions *opt; -struct tcp6_sock *newtcp6sk;struct inet_sock *newinet;struct tcp_sock *newtp;struct sock *newsk;
@@ -1088,11 +1098,10 @@ static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *if (!newsk)return NULL ; - newtcp6sk = (struct tcp6_sock *)newsk; -inet_sk (newsk)->pinet6 = &newtcp6sk->inet6; +inet_sk (newsk)->pinet6 =tcp_inet6_sk (newsk); newinet =inet_sk (newsk); - newnp =inet6_sk (newsk); + newnp =tcp_inet6_sk (newsk); newtp =tcp_sk (newsk);memcpy (newnp, np,sizeof (struct ipv6_pinfo ));
@@ -1156,12 +1165,11 @@ static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *ip6_dst_store (newsk, dst,NULL ,NULL );inet6_sk_rx_dst_set (newsk, skb); - newtcp6sk = (struct tcp6_sock *)newsk; -inet_sk (newsk)->pinet6 = &newtcp6sk->inet6; +inet_sk (newsk)->pinet6 =tcp_inet6_sk (newsk); newtp =tcp_sk (newsk); newinet =inet_sk (newsk); - newnp =inet6_sk (newsk); + newnp =tcp_inet6_sk (newsk);memcpy (newnp, np,sizeof (struct ipv6_pinfo ));
@@ -1276,9 +1284,9 @@ static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff **/ static int tcp_v6_do_rcv (struct sock *sk,struct sk_buff *skb) { -struct ipv6_pinfo *np =inet6_sk (sk); -struct tcp_sock *tp; +struct ipv6_pinfo *np =tcp_inet6_sk (sk);struct sk_buff *opt_skb =NULL ; +struct tcp_sock *tp;/* Imagine: socket is IPv6. IPv4 packet arrives, goes to IPv4 receive handler and backlogged.
@@ -1524,7 +1532,7 @@ static int tcp_v6_rcv(struct sk_buff *skb)return 0 ; } } -if (hdr->hop_limit <inet6_sk (sk)->min_hopcount) { +if (hdr->hop_limit <tcp_inet6_sk (sk)->min_hopcount) {__NET_INC_STATS (net,LINUX_MIB_TCPMINTTLDROP ); goto discard_and_relse; }
@@ -1669,7 +1677,7 @@ static void tcp_v6_early_demux(struct sk_buff *skb)struct dst_entry *dst =READ_ONCE (sk->sk_rx_dst);if (dst) - dst =dst_check (dst,inet6_sk (sk)->rx_dst_cookie); + dst =dst_check (dst,tcp_inet6_sk (sk)->rx_dst_cookie);if (dst &&inet_sk (sk)->rx_dst_ifindex == skb->skb_iif)skb_dst_set_noref (skb, dst);
Signature · 1 line
2.21.0.225.g810b269d1ac-goog
From: Eric Dumazet <[email protected]>
Date: Tue, 19 Mar 2019 07:01:08 -0700
Quoted 8 lines
TCP ipv6 fast path dereferences a pointer to get to the inet6 part of a tcp socket, but given the fixed memory placement, we can do better and avoid a possible cache line miss.
This also reduces register pressure, since we let the compiler know about this memory placement.
Signed-off-by Eric Dumazet <[email protected]>
Applied.