netfilter: don't setup nat info for confirmed ct
authorLiping Zhang <zlpnobody@gmail.com>
Sat, 6 May 2017 12:28:02 +0000 (20:28 +0800)
committerPablo Neira Ayuso <pablo@netfilter.org>
Mon, 15 May 2017 10:42:28 +0000 (12:42 +0200)
We cannot setup nat info if the ct has been confirmed already, else,
different cpu may race to handle the same ct. In extreme situation,
we may hit the "BUG_ON(nf_nat_initialized(ct, maniptype))" in the
nf_nat_setup_info.

Also running the following commands will easily hit NF_CT_ASSERT in
nf_conntrack_alter_reply:
  # nft flush ruleset
  # ping -c 2 -W 1 1.1.1.111 &
  # nft add table t
  # nft add chain t c {type nat hook postrouting priority 0 \;}
  # nft add rule t c snat to 4.5.6.7
  WARNING: CPU: 1 PID: 10065 at net/netfilter/nf_conntrack_core.c:1472
  nf_conntrack_alter_reply+0x9a/0x1a0 [nf_conntrack]
  [...]
  Call Trace:
   nf_nat_setup_info+0xad/0x840 [nf_nat]
   ? deactivate_slab+0x65d/0x6c0
   nft_nat_eval+0xcd/0x100 [nft_nat]
   nft_do_chain+0xff/0x5d0 [nf_tables]
   ? mark_held_locks+0x6f/0xa0
   ? __local_bh_enable_ip+0x70/0xa0
   ? trace_hardirqs_on_caller+0x11f/0x190
   ? ipt_do_table+0x310/0x610
   ? trace_hardirqs_on+0xd/0x10
   ? __local_bh_enable_ip+0x70/0xa0
   ? ipt_do_table+0x32b/0x610
   ? __lock_acquire+0x2ac/0x1580
   ? ipt_do_table+0x32b/0x610
   nft_nat_do_chain+0x65/0x80 [nft_chain_nat_ipv4]
   nf_nat_ipv4_fn+0x1ae/0x240 [nf_nat_ipv4]
   nf_nat_ipv4_out+0x4a/0xf0 [nf_nat_ipv4]
   nft_nat_ipv4_out+0x15/0x20 [nft_chain_nat_ipv4]
   nf_hook_slow+0x2c/0xf0
   ip_output+0x154/0x270

So for the confirmed ct, just ignore it and return NF_ACCEPT.

Fixes: 9a08ecfe74d7 ("netfilter: don't attach a nat extension by default")
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Acked-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
net/netfilter/nf_nat_core.c

index b48d6b5aae8a87d4ea69cae0e025739ebe3f1658..ef0be325a0c6368bfe29ecda39db37dcb178a6d2 100644 (file)
@@ -409,6 +409,10 @@ nf_nat_setup_info(struct nf_conn *ct,
 {
        struct nf_conntrack_tuple curr_tuple, new_tuple;
 
+       /* Can't setup nat info for confirmed ct. */
+       if (nf_ct_is_confirmed(ct))
+               return NF_ACCEPT;
+
        NF_CT_ASSERT(maniptype == NF_NAT_MANIP_SRC ||
                     maniptype == NF_NAT_MANIP_DST);
        BUG_ON(nf_nat_initialized(ct, maniptype));