bridge: correct IPv6 checksum after pull
stephen hemminger [Tue, 15 Nov 2011 08:09:14 +0000 (08:09 +0000)]
commit fa2da8cdae1dd64f78fc915ca1d1a4a93c71e7cb upstream.

Bridge multicast snooping of ICMPv6 would incorrectly report a checksum problem
when used with Ethernet devices like sky2 that use CHECKSUM_COMPLETE.
When bytes are removed from skb, the computed checksum needs to be adjusted.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Tested-by: Martin Volf <martin.volf.42@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

net/bridge/br_multicast.c

index 995cbe0..e79ff75 100644 (file)
@@ -1501,6 +1501,8 @@ static int br_multicast_ipv6_rcv(struct net_bridge *br,
 
        __skb_pull(skb2, offset);
        skb_reset_transport_header(skb2);
+       skb_postpull_rcsum(skb2, skb_network_header(skb2),
+                          skb_network_header_len(skb2));
 
        icmp6_type = icmp6_hdr(skb2)->icmp6_type;