aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorLawrence Brakmo <brakmo@fb.com>2017-06-30 23:02:42 -0400
committerDavid S. Miller <davem@davemloft.net>2017-07-01 19:15:13 -0400
commit8550f328f45db6d37981eb2041bc465810245c03 (patch)
treee3594097d7b3c9dc3176c8f9084431cf24d79cf6
parentae16189efb0fe2bdc5e702bd6221ed6d0ff5babd (diff)
bpf: Support for per connection SYN/SYN-ACK RTOs
This patch adds support for setting a per connection SYN and SYN_ACK RTOs from within a BPF_SOCK_OPS program. For example, to set small RTOs when it is known both hosts are within a datacenter. Signed-off-by: Lawrence Brakmo <brakmo@fb.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-rw-r--r--include/net/tcp.h11
-rw-r--r--include/uapi/linux/bpf.h3
-rw-r--r--net/ipv4/tcp_input.c3
-rw-r--r--net/ipv4/tcp_output.c2
4 files changed, 17 insertions, 2 deletions
diff --git a/include/net/tcp.h b/include/net/tcp.h
index e58500825006..564af2dee236 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -2057,4 +2057,15 @@ static inline int tcp_call_bpf(struct sock *sk, int op)
2057} 2057}
2058#endif 2058#endif
2059 2059
2060static inline u32 tcp_timeout_init(struct sock *sk)
2061{
2062 int timeout;
2063
2064 timeout = tcp_call_bpf(sk, BPF_SOCK_OPS_TIMEOUT_INIT);
2065
2066 if (timeout <= 0)
2067 timeout = TCP_TIMEOUT_INIT;
2068 return timeout;
2069}
2070
2060#endif /* _TCP_H */ 2071#endif /* _TCP_H */
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 01cd485ccd4f..00702b294447 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -748,6 +748,9 @@ struct bpf_sock_ops {
748 */ 748 */
749enum { 749enum {
750 BPF_SOCK_OPS_VOID, 750 BPF_SOCK_OPS_VOID,
751 BPF_SOCK_OPS_TIMEOUT_INIT, /* Should return SYN-RTO value to use or
752 * -1 if default value should be used
753 */
751}; 754};
752 755
753#endif /* _UAPI__LINUX_BPF_H__ */ 756#endif /* _UAPI__LINUX_BPF_H__ */
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 2ab7e2fa9bb9..bcc96654cd7e 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -6406,7 +6406,8 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
6406 } else { 6406 } else {
6407 tcp_rsk(req)->tfo_listener = false; 6407 tcp_rsk(req)->tfo_listener = false;
6408 if (!want_cookie) 6408 if (!want_cookie)
6409 inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT); 6409 inet_csk_reqsk_queue_hash_add(sk, req,
6410 tcp_timeout_init((struct sock *)req));
6410 af_ops->send_synack(sk, dst, &fl, req, &foc, 6411 af_ops->send_synack(sk, dst, &fl, req, &foc,
6411 !want_cookie ? TCP_SYNACK_NORMAL : 6412 !want_cookie ? TCP_SYNACK_NORMAL :
6412 TCP_SYNACK_COOKIE); 6413 TCP_SYNACK_COOKIE);
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 1d79137f3795..47fe0759a877 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -3326,7 +3326,7 @@ static void tcp_connect_init(struct sock *sk)
3326 tp->rcv_wup = tp->rcv_nxt; 3326 tp->rcv_wup = tp->rcv_nxt;
3327 tp->copied_seq = tp->rcv_nxt; 3327 tp->copied_seq = tp->rcv_nxt;
3328 3328
3329 inet_csk(sk)->icsk_rto = TCP_TIMEOUT_INIT; 3329 inet_csk(sk)->icsk_rto = tcp_timeout_init(sk);
3330 inet_csk(sk)->icsk_retransmits = 0; 3330 inet_csk(sk)->icsk_retransmits = 0;
3331 tcp_clear_retrans(tp); 3331 tcp_clear_retrans(tp);
3332} 3332}