]> git.baikalelectronics.ru Git - kernel.git/commitdiff
netfilter: nat: fix spurious connection timeouts
authorFlorian Westphal <fw@strlen.de>
Fri, 8 Feb 2019 15:39:52 +0000 (16:39 +0100)
committerPablo Neira Ayuso <pablo@netfilter.org>
Mon, 11 Feb 2019 16:43:17 +0000 (17:43 +0100)
Sander Eikelenboom bisected a NAT related regression down
to the l4proto->manip_pkt indirection removal.

I forgot that ICMP(v6) errors (e.g. PKTTOOBIG) can be set as related
to the existing conntrack entry.

Therefore, when passing the skb to nf_nat_ipv4/6_manip_pkt(), that
ended up calling the wrong l4 manip function, as tuple->dst.protonum
is the original flows l4 protocol (TCP, UDP, etc).

Set the dst protocol field to ICMP(v6), we already have a private copy
of the tuple due to the inversion of src/dst.

Reported-by: Sander Eikelenboom <linux@eikelenboom.it>
Tested-by: Sander Eikelenboom <linux@eikelenboom.it>
Fixes: 7204bc8f0c7fc ("netfilter: nat: remove l4proto->manip_pkt")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
net/ipv4/netfilter/nf_nat_l3proto_ipv4.c
net/ipv6/netfilter/nf_nat_l3proto_ipv6.c

index 2687db015b6f1b3a51caecb26a38ca11c2384938..fa2ba7c500e41139ca03641cf33ec4e2a16764a9 100644 (file)
@@ -215,6 +215,7 @@ int nf_nat_icmp_reply_translation(struct sk_buff *skb,
 
        /* Change outer to look like the reply to an incoming packet */
        nf_ct_invert_tuplepr(&target, &ct->tuplehash[!dir].tuple);
+       target.dst.protonum = IPPROTO_ICMP;
        if (!nf_nat_ipv4_manip_pkt(skb, 0, &target, manip))
                return 0;
 
index 23022447eb497ec279c6585da86dbff09f8d5973..7a41ee3c11b4d076c91c57095472b6a48b4112ba 100644 (file)
@@ -226,6 +226,7 @@ int nf_nat_icmpv6_reply_translation(struct sk_buff *skb,
        }
 
        nf_ct_invert_tuplepr(&target, &ct->tuplehash[!dir].tuple);
+       target.dst.protonum = IPPROTO_ICMPV6;
        if (!nf_nat_ipv6_manip_pkt(skb, 0, &target, manip))
                return 0;