netfilter: nf_flow_table: move ip header check out of nf_flow_exceeds_mtu
authorFelix Fietkau <nbd@nbd.name>
Mon, 26 Feb 2018 09:15:14 +0000 (10:15 +0100)
committerPablo Neira Ayuso <pablo@netfilter.org>
Tue, 24 Apr 2018 08:28:15 +0000 (10:28 +0200)
Allows the function to be shared with the IPv6 hook code

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
net/netfilter/nf_flow_table_ip.c

index 034fda9..103263e 100644 (file)
@@ -182,9 +182,6 @@ static bool nf_flow_exceeds_mtu(const struct sk_buff *skb, unsigned int mtu)
        if (skb->len <= mtu)
                return false;
 
-       if ((ip_hdr(skb)->frag_off & htons(IP_DF)) == 0)
-               return false;
-
        if (skb_is_gso(skb) && skb_gso_validate_network_len(skb, mtu))
                return false;
 
@@ -223,7 +220,8 @@ nf_flow_offload_ip_hook(void *priv, struct sk_buff *skb,
        flow = container_of(tuplehash, struct flow_offload, tuplehash[dir]);
        rt = (const struct rtable *)flow->tuplehash[dir].tuple.dst_cache;
 
-       if (unlikely(nf_flow_exceeds_mtu(skb, flow->tuplehash[dir].tuple.mtu)))
+       if (unlikely(nf_flow_exceeds_mtu(skb, flow->tuplehash[dir].tuple.mtu)) &&
+           (ip_hdr(skb)->frag_off & htons(IP_DF)) != 0)
                return NF_ACCEPT;
 
        if (skb_try_make_writable(skb, sizeof(*iph)))