tcp_v4_send_reset: binding oif to iif in no sock case
authorShawn Lu <shawn.lu@ericsson.com>
Sat, 4 Feb 2012 12:38:09 +0000 (12:38 +0000)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 1 Mar 2012 00:31:01 +0000 (16:31 -0800)
[ Upstream commit e2446eaab5585555a38ea0df4e01ff313dbb4ac9 ]

Binding RST packet outgoing interface to incoming interface
for tcp v4 when there is no socket associate with it.
when sk is not NULL, using sk->sk_bound_dev_if instead.
(suggested by Eric Dumazet).

This has few benefits:
1. tcp_v6_send_reset already did that.
2. This helps tcp connect with SO_BINDTODEVICE set. When
connection is lost, we still able to sending out RST using
same interface.
3. we are sending reply, it is most likely to be succeed
if iif is used

Signed-off-by: Shawn Lu <shawn.lu@ericsson.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net/ipv4/tcp_ipv4.c

index c89e354..eb90aa8 100644 (file)
@@ -650,6 +650,11 @@ static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
                                      arg.iov[0].iov_len, IPPROTO_TCP, 0);
        arg.csumoffset = offsetof(struct tcphdr, check) / 2;
        arg.flags = (sk && inet_sk(sk)->transparent) ? IP_REPLY_ARG_NOSRCCHECK : 0;
+       /* When socket is gone, all binding information is lost.
+        * routing might fail in this case. using iif for oif to
+        * make sure we can deliver it
+        */
+       arg.bound_dev_if = sk ? sk->sk_bound_dev_if : inet_iif(skb);
 
        net = dev_net(skb_dst(skb)->dev);
        arg.tos = ip_hdr(skb)->tos;