vrf: sit mtu should not be updated when vrf netdev is the link
authorStephen Suryaputra <ssuryaextr@gmail.com>
Mon, 6 May 2019 19:00:01 +0000 (15:00 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 16 May 2019 17:41:30 +0000 (19:41 +0200)
[ Upstream commit ff6ab32bd4e073976e4d8797b4d514a172cfe6cb ]

VRF netdev mtu isn't typically set and have an mtu of 65536. When the
link of a tunnel is set, the tunnel mtu is changed from 1480 to the link
mtu minus tunnel header. In the case of VRF netdev is the link, then the
tunnel mtu becomes 65516. So, fix it by not setting the tunnel mtu in
this case.

Signed-off-by: Stephen Suryaputra <ssuryaextr@gmail.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net/ipv6/sit.c

index 8f6cf8e..41b3fe8 100644 (file)
@@ -1084,7 +1084,7 @@ static void ipip6_tunnel_bind_dev(struct net_device *dev)
        if (!tdev && tunnel->parms.link)
                tdev = __dev_get_by_index(tunnel->net, tunnel->parms.link);
 
-       if (tdev) {
+       if (tdev && !netif_is_l3_master(tdev)) {
                int t_hlen = tunnel->hlen + sizeof(struct iphdr);
 
                dev->hard_header_len = tdev->hard_header_len + sizeof(struct iphdr);