net: ipv6: fix TCP early demux

IPv6 needs a cookie in dst_check() call. We need to add rx_dst_cookie and provide a family independent sk_rx_dst_set(sk, skb) method to properly support IPv6 TCP early demux.

Eric Dumazet committed 14 years ago

Commit 5d299f31 parent b5497ee7 files · +41 −16Browse files

Unified · Split

7 files

include/linux/ipv6.h +1 −0
include/net/inet_connection_sock.h +1 −0
include/net/inet_sock.h +0 −9
net/ipv4/tcp_input.c +3 −1
net/ipv4/tcp_ipv4.c +10 −3
net/ipv4/tcp_minisocks.c +1 −1
net/ipv6/tcp_ipv6.c +25 −2

include/linux/ipv6.h+1−0

@@ -369,6 +369,7 @@ struct ipv6_pinfo { · 368 unmodified lines
369	__u8 rcv_tclass;	369	__u8 rcv_tclass;
370		370
371	__u32 dst_cookie;	371	__u32 dst_cookie;
		372	__u32 rx_dst_cookie;
372		373
373	struct ipv6_mc_socklist __rcu *ipv6_mc_list;	374	struct ipv6_mc_socklist __rcu *ipv6_mc_list;
374	struct ipv6_ac_socklist *ipv6_ac_list;	375	struct ipv6_ac_socklist *ipv6_ac_list;

include/net/inet_connection_sock.h+1−0

@@ -39,6 +39,7 @@ struct inet_connection_sock_af_ops { · 38 unmodified lines
39	int (queue_xmit)(struct sk_buff skb, struct flowi *fl);	39	int (queue_xmit)(struct sk_buff skb, struct flowi *fl);
40	void (send_check)(struct sock sk, struct sk_buff *skb);	40	void (send_check)(struct sock sk, struct sk_buff *skb);
41	int (rebuild_header)(struct sock sk);	41	int (rebuild_header)(struct sock sk);
		42	void (sk_rx_dst_set)(struct sock sk, const struct sk_buff *skb);
42	int (conn_request)(struct sock sk, struct sk_buff *skb);	43	int (conn_request)(struct sock sk, struct sk_buff *skb);
43	struct sock (syn_recv_sock)(struct sock sk, struct sk_buff skb,	44	struct sock (syn_recv_sock)(struct sock sk, struct sk_buff skb,
44	struct request_sock *req,	45	struct request_sock *req,

include/net/inet_sock.h+0−9

@@ -249,13 +249,4 @@ static inline __u8 inet_sk_flowi_flags(const struct sock *sk) · 248 unmodified lines
249	return flags;	249	return flags;
250	}	250	}
251		251
252	static inline void inet_sk_rx_dst_set(struct sock sk, const struct sk_buff skb)
253	{
254	struct dst_entry *dst = skb_dst(skb);
255
256	dst_hold(dst);
257	sk->sk_rx_dst = dst;
258	inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
259	}
260
261	#endif /* _INET_SOCK_H */	252	#endif /* _INET_SOCK_H */

net/ipv4/tcp_input.c+3−1

 {
 	struct tcp_sock *tp = tcp_sk(sk);
+	if (unlikely(sk->sk_rx_dst == NULL))
+		inet_csk(sk)->icsk_af_ops->sk_rx_dst_set(sk, skb);
 	/*
 	 *	Header prediction.
 	 *	The code loosely follows the one in the famous
 	tcp_set_state(sk, TCP_ESTABLISHED);
 	if (skb != NULL) {
-		inet_sk_rx_dst_set(sk, skb);
+		icsk->icsk_af_ops->sk_rx_dst_set(sk, skb);
 		security_inet_conn_established(sk, skb);
 	}

net/ipv4/tcp_ipv4.c+10−3

 				sk->sk_rx_dst = NULL;
 			}
 		}
-		if (unlikely(sk->sk_rx_dst == NULL))
-			inet_sk_rx_dst_set(sk, skb);
 		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
 			rsk = sk;
 			goto reset;
 	.twsk_destructor= tcp_twsk_destructor,
 };
+static void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
+{
+	struct dst_entry *dst = skb_dst(skb);
+	dst_hold(dst);
+	sk->sk_rx_dst = dst;
+	inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
+}
 const struct inet_connection_sock_af_ops ipv4_specific = {
 	.queue_xmit	   = ip_queue_xmit,
 	.send_check	   = tcp_v4_send_check,
 	.rebuild_header	   = inet_sk_rebuild_header,
+	.sk_rx_dst_set	   = inet_sk_rx_dst_set,
 	.conn_request	   = tcp_v4_conn_request,
 	.syn_recv_sock	   = tcp_v4_syn_recv_sock,
 	.net_header_len	   = sizeof(struct iphdr),

net/ipv4/tcp_minisocks.c+1−1

@@ -387,7 +387,7 @@ struct sock tcp_create_openreq_child(struct sock sk, struct request_sock *req, · 386 unmodified lines
387	struct tcp_sock *oldtp = tcp_sk(sk);	387	struct tcp_sock *oldtp = tcp_sk(sk);
388	struct tcp_cookie_values *oldcvp = oldtp->cookie_values;	388	struct tcp_cookie_values *oldcvp = oldtp->cookie_values;
389		389
390	inet_sk_rx_dst_set(newsk, skb);	390	newicsk->icsk_af_ops->sk_rx_dst_set(newsk, skb);
391		391
392	/* TCP Cookie Transactions require space for the cookie pair,	392	/* TCP Cookie Transactions require space for the cookie pair,
393	* as it differs for each connection. There is no need to	393	* as it differs for each connection. There is no need to

net/ipv6/tcp_ipv6.c+25−2

 		opt_skb = skb_clone(skb, sk_gfp_atomic(sk, GFP_ATOMIC));
 	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
+		struct dst_entry *dst = sk->sk_rx_dst;
 		sock_rps_save_rxhash(sk, skb);
+		if (dst) {
+			if (inet_sk(sk)->rx_dst_ifindex != skb->skb_iif ||
+			    dst->ops->check(dst, np->rx_dst_cookie) == NULL) {
+				dst_release(dst);
+				sk->sk_rx_dst = NULL;
+			}
+		}
 		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len))
 			goto reset;
 		if (opt_skb)
 			struct dst_entry *dst = sk->sk_rx_dst;
 			struct inet_sock *icsk = inet_sk(sk);
 			if (dst)
-				dst = dst_check(dst, 0);
+				dst = dst_check(dst, inet6_sk(sk)->rx_dst_cookie);
 			if (dst &&
-			    icsk->rx_dst_ifindex == inet6_iif(skb))
+			    icsk->rx_dst_ifindex == skb->skb_iif)
 				skb_dst_set_noref(skb, dst);
 		}
 	}
 	.twsk_destructor= tcp_twsk_destructor,
 };
+static void inet6_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
+{
+	struct dst_entry *dst = skb_dst(skb);
+	const struct rt6_info *rt = (const struct rt6_info *)dst;
+	dst_hold(dst);
+	sk->sk_rx_dst = dst;
+	inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
+	if (rt->rt6i_node)
+		inet6_sk(sk)->rx_dst_cookie = rt->rt6i_node->fn_sernum;
+}
 static const struct inet_connection_sock_af_ops ipv6_specific = {
 	.queue_xmit	   = inet6_csk_xmit,
 	.send_check	   = tcp_v6_send_check,
 	.rebuild_header	   = inet6_sk_rebuild_header,
+	.sk_rx_dst_set	   = inet6_sk_rx_dst_set,
 	.conn_request	   = tcp_v6_conn_request,
 	.syn_recv_sock	   = tcp_v6_syn_recv_sock,
 	.net_header_len	   = sizeof(struct ipv6hdr),