Re: [PATCH] tcp: RST: binding oif to iif for tcp v4

On 02/03/2012 01:31 PM, Eric Dumazet wrote:

Le vendredi 03 février 2012 à 12:16 -0800, Shawn Lu a écrit :

quoted

Binding RST packet outgoing interface to incomming interface
for tcp v4. This has few benefits:
1. tcp_v6_send_reset already did that.

I dont think so. ipv6 makes no special provision for RST.

quoted

2. This helps tcp connect with SO_BINDTODEVICE set. When connection
is lost, we still able to sending out RST using same interface.

I dont understand this.

I have been using a similar patch for years now.  When using routing
rules that depend on binding to an interface, the RST isn't always bound
properly without this fix.

My code looks like this:

@@ -650,6 +650,7 @@ 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;
+	arg.bound_dev_if = skb_rtable(skb)->rt_iif;

  	net = dev_net(skb_dst(skb)->dev);
  	ip_send_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,

It was originally written by Patrick McHardy, but I've hacked it
a bit over the years to keep it compiling.  It seems to work fine.

Thanks,
Ben


-- 
Ben Greear [off-list ref]
Candela Technologies Inc  http://www.candelatech.com