netfilter: conntrack: handle tcp challenge acks during connection reuse

When a connection is re-used, following can happen:
[ connection starts to close, fin sent in either direction ]
 > syn   # initator quickly reuses connection
 < ack   # peer sends a challenge ack
 > rst   # rst, sequence number == ack_seq of previous challenge ack
 > syn   # this syn is expected to pass

Problem is that the rst will fail window validation, so it gets
tagged as invalid.

If ruleset drops such packets, we get repeated syn-retransmits until
initator gives up or peer starts responding with syn/ack.

Before the commit indicated in the "Fixes" tag below this used to work:

The challenge-ack made conntrack re-init state based on the challenge
ack itself, so the following rst would pass window validation.

Add challenge-ack support: If we get ack for syn, record the ack_seq,
and then check if the rst sequence number matches the last ack number
seen in reverse direction.

Fixes: c7aab4f170 ("netfilter: nf_conntrack_tcp: re-init for syn packets only")
Reported-by: Michal Tesar <mtesar@redhat.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

net/netfilter/nf_conntrack_proto_tcp.c

@@ -1068,6 +1068,13 @@ int nf_conntrack_tcp_packet(struct nf_conn *ct,
 				ct->proto.tcp.last_flags |=
 					IP_CT_EXP_CHALLENGE_ACK;
 		}
+
+		/* possible challenge ack reply to syn */
+		if (old_state == TCP_CONNTRACK_SYN_SENT &&
+		    index == TCP_ACK_SET &&
+		    dir == IP_CT_DIR_REPLY)
+			ct->proto.tcp.last_ack = ntohl(th->ack_seq);
+
 		spin_unlock_bh(&ct->lock);
 		nf_ct_l4proto_log_invalid(skb, ct, state,
 					  "packet (index %d) in dir %d ignored, state %s",
@@ -1193,6 +1200,14 @@ int nf_conntrack_tcp_packet(struct nf_conn *ct,
 			 * segments we ignored. */
 			goto in_window;
 		}
+
+		/* Reset in response to a challenge-ack we let through earlier */
+		if (old_state == TCP_CONNTRACK_SYN_SENT &&
+		    ct->proto.tcp.last_index == TCP_ACK_SET &&
+		    ct->proto.tcp.last_dir == IP_CT_DIR_REPLY &&
+		    ntohl(th->seq) == ct->proto.tcp.last_ack)
+			goto in_window;
+
 		break;
 	default:
 		/* Keep compilers happy. */