netlink: reset network header before passing to taps
authorDaniel Borkmann <dborkman@redhat.com>
Thu, 7 Aug 2014 20:22:47 +0000 (22:22 +0200)
committerDavid S. Miller <davem@davemloft.net>
Thu, 7 Aug 2014 23:02:58 +0000 (16:02 -0700)
netlink doesn't set any network header offset thus when the skb is
being passed to tap devices via dev_queue_xmit_nit(), it emits klog
false positives due to it being unset like:

  ...
  [  124.990397] protocol 0000 is buggy, dev nlmon0
  [  124.990411] protocol 0000 is buggy, dev nlmon0
  ...

So just reset the network header before passing to the device; for
packet sockets that just means nothing will change - mac and net
offset hold the same value just as before.

Reported-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/netlink/af_netlink.c

index a324b4b34c909094f1b1aac14e986e3777bccd5d..2e152e5f218660e94ce2bed0d45925d9676e9bec 100644 (file)
@@ -213,7 +213,7 @@ static int __netlink_deliver_tap_skb(struct sk_buff *skb,
                nskb->protocol = htons((u16) sk->sk_protocol);
                nskb->pkt_type = netlink_is_kernel(sk) ?
                                 PACKET_KERNEL : PACKET_USER;
-
+               skb_reset_network_header(nskb);
                ret = dev_queue_xmit(nskb);
                if (unlikely(ret > 0))
                        ret = net_xmit_errno(ret);