netfilter: conntrack: make conntrack userspace helpers work again

Florian Westphal says:

"Problem is that after the helper hook was merged back into the confirm
one, the queueing itself occurs from the confirm hook, i.e. we queue
from the last netfilter callback in the hook-list.

Therefore, on return, the packet bypasses the confirm action and the
connection is never committed to the main conntrack table.

To fix this there are several ways:
1. revert the 'Fixes' commit and have a extra helper hook again.
Works, but has the drawback of adding another indirect call for
everyone.

2. Special case this: split the hooks only when userspace helper
gets added, so queueing occurs at a lower priority again,
and normal enqueue reinject would eventually call the last hook.

3. Extend the existing nf_queue ct update hook to allow a forced
confirmation (plus run the seqadj code).

This goes for 3)."

Fixes: 827318feb69cb ("netfilter: conntrack: remove helper hook again")
Reviewed-by: Florian Westphal <[e-mail]>
Signed-off-by: Pablo Neira Ayuso <[e-mail]>

net/netfilter/nf_conntrack_core.c

···
 	nf_conntrack_get(skb_nfct(nskb));
 }
 
-static int nf_conntrack_update(struct net *net, struct sk_buff *skb)
+static int __nf_conntrack_update(struct net *net, struct sk_buff *skb,
+				 struct nf_conn *ct)
 {
 	struct nf_conntrack_tuple_hash *h;
 	struct nf_conntrack_tuple tuple;
 	enum ip_conntrack_info ctinfo;
 	struct nf_nat_hook *nat_hook;
 	unsigned int status;
-	struct nf_conn *ct;
 	int dataoff;
 	u16 l3num;
 	u8 l4num;
-
-	ct = nf_ct_get(skb, &ctinfo);
-	if (!ct || nf_ct_is_confirmed(ct))
-		return 0;
 
 	l3num = nf_ct_l3num(ct);
 
···
 		return -1;
 
 	return 0;
+}
+
+/* This packet is coming from userspace via nf_queue, complete the packet
+ * processing after the helper invocation in nf_confirm().
+ */
+static int nf_confirm_cthelper(struct sk_buff *skb, struct nf_conn *ct,
+			       enum ip_conntrack_info ctinfo)
+{
+	const struct nf_conntrack_helper *helper;
+	const struct nf_conn_help *help;
+	unsigned int protoff;
+
+	help = nfct_help(ct);
+	if (!help)
+		return 0;
+
+	helper = rcu_dereference(help->helper);
+	if (!(helper->flags & NF_CT_HELPER_F_USERSPACE))
+		return 0;
+
+	switch (nf_ct_l3num(ct)) {
+	case NFPROTO_IPV4:
+		protoff = skb_network_offset(skb) + ip_hdrlen(skb);
+		break;
+#if IS_ENABLED(CONFIG_IPV6)
+	case NFPROTO_IPV6: {
+		__be16 frag_off;
+		u8 pnum;
+
+		pnum = ipv6_hdr(skb)->nexthdr;
+		protoff = ipv6_skip_exthdr(skb, sizeof(struct ipv6hdr), &pnum,
+					   &frag_off);
+		if (protoff < 0 || (frag_off & htons(~0x7)) != 0)
+			return 0;
+		break;
+	}
+#endif
+	default:
+		return 0;
+	}
+
+	if (test_bit(IPS_SEQ_ADJUST_BIT, &ct->status) &&
+	    !nf_is_loopback_packet(skb)) {
+		if (!nf_ct_seq_adjust(skb, ct, ctinfo, protoff)) {
+			NF_CT_STAT_INC_ATOMIC(nf_ct_net(ct), drop);
+			return -1;
+		}
+	}
+
+	/* We've seen it coming out the other side: confirm it */
+	return nf_conntrack_confirm(skb) == NF_DROP ? - 1 : 0;
+}
+
+static int nf_conntrack_update(struct net *net, struct sk_buff *skb)
+{
+	enum ip_conntrack_info ctinfo;
+	struct nf_conn *ct;
+	int err;
+
+	ct = nf_ct_get(skb, &ctinfo);
+	if (!ct)
+		return 0;
+
+	if (!nf_ct_is_confirmed(ct)) {
+		err = __nf_conntrack_update(net, skb, ct);
+		if (err < 0)
+			return err;
+	}
+
+	return nf_confirm_cthelper(skb, ct, ctinfo);
 }
 
 static bool nf_conntrack_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple,