ipv4: Should use consistent conditional judgement for ip fragment in __ip_append_data...

commit 0a28cfd51e17f4f0a056bcf66bfbe492c3b99f38 upstream.

There is an inconsistent conditional judgement in __ip_append_data and
ip_finish_output functions, the variable length in __ip_append_data just
include the length of application's payload and udp header, don't include
the length of ip header, but in ip_finish_output use
(skb->len > ip_skb_dst_mtu(skb)) as judgement, and skb->len include the
length of ip header.

That causes some particular application's udp payload whose length is
between (MTU - IP Header) and MTU were fragmented by ip_fragment even
though the rst->dev support UFO feature.

Add the length of ip header to length in __ip_append_data to keep
consistent conditional judgement as ip_finish_output for ip fragment.

Signed-off-by: Zheng Li <james.z.li@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index f3403a3ce290cca1e87ce1e1515e303d85676f64..0efa401c39f4d6c027f72ee1532106218ced95a3 100644 (file)
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -923,7 +923,7 @@ static int __ip_append_data(struct sock *sk,
 
        cork->length += length;
        if ((skb && skb_is_gso(skb)) ||
-           ((length > mtu) &&
+           (((length + fragheaderlen) > mtu) &&
            (skb_queue_len(queue) <= 1) &&
            (sk->sk_protocol == IPPROTO_UDP) &&
            (rt->dst.dev->features & NETIF_F_UFO) && !rt->dst.header_len &&