netfilter: nf_nat: add nat type hooks to nat core

Currently the packet rewrite and instantiation of nat NULL bindings
happens from the protocol specific nat backend.

Invocation occurs either via ip(6)table_nat or the nf_tables nat chain type.

Invocation looks like this (simplified):
NF_HOOK()
   |
   `---iptable_nat
	 |
	 `---> nf_nat_l3proto_ipv4 -> nf_nat_packet
	               |
          new packet? pass skb though iptables nat chain
                       |
		       `---> iptable_nat: ipt_do_table

In nft case, this looks the same (nft_chain_nat_ipv4 instead of
iptable_nat).

This is a problem for two reasons:
1. Can't use iptables nat and nf_tables nat at the same time,
   as the first user adds a nat binding (nf_nat_l3proto_ipv4 adds a
   NULL binding if do_table() did not find a matching nat rule so we
   can detect post-nat tuple collisions).
2. If you use e.g. nft_masq, snat, redir, etc. uses must also register
   an empty base chain so that the nat core gets called fro NF_HOOK()
   to do the reverse translation, which is neither obvious nor user
   friendly.

After this change, the base hook gets registered not from iptable_nat or
nftables nat hooks, but from the l3 nat core.

iptables/nft nat base hooks get registered with the nat core instead:

NF_HOOK()
   |
   `---> nf_nat_l3proto_ipv4 -> nf_nat_packet
		|
         new packet? pass skb through iptables/nftables nat chains
                |
		+-> iptables_nat: ipt_do_table
	        +-> nft nat chain x
	        `-> nft nat chain y

The nat core deals with null bindings and reverse translation.
When no mapping exists, it calls the registered nat lookup hooks until
one creates a new mapping.
If both iptables and nftables nat hooks exist, the first matching
one is used (i.e., higher priority wins).

Also, nft users do not need to create empty nat hooks anymore,
nat core always registers the base hooks that take care of reverse/reply
translation.

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

3 files changed, 106 insertions, 112 deletions

diff --git a/net/ipv4/netfilter/iptable_nat.c b/net/ipv4/netfilter/iptable_nat.c
index 529d89ec31e8..a317445448bf 100644
--- a/net/ipv4/netfilter/iptable_nat.c
+++ b/net/ipv4/netfilter/iptable_nat.c
@@ -38,69 +38,58 @@ static unsigned int iptable_nat_do_chain(void *priv,
 	return ipt_do_table(skb, state, state->net->ipv4.nat_table);
 }
 
-static unsigned int iptable_nat_ipv4_fn(void *priv,
-					struct sk_buff *skb,
-					const struct nf_hook_state *state)
-{
-	return nf_nat_ipv4_fn(priv, skb, state, iptable_nat_do_chain);
-}
-
-static unsigned int iptable_nat_ipv4_in(void *priv,
-					struct sk_buff *skb,
-					const struct nf_hook_state *state)
-{
-	return nf_nat_ipv4_in(priv, skb, state, iptable_nat_do_chain);
-}
-
-static unsigned int iptable_nat_ipv4_out(void *priv,
-					 struct sk_buff *skb,
-					 const struct nf_hook_state *state)
-{
-	return nf_nat_ipv4_out(priv, skb, state, iptable_nat_do_chain);
-}
-
-static unsigned int iptable_nat_ipv4_local_fn(void *priv,
-					      struct sk_buff *skb,
-					      const struct nf_hook_state *state)
-{
-	return nf_nat_ipv4_local_fn(priv, skb, state, iptable_nat_do_chain);
-}
-
 static const struct nf_hook_ops nf_nat_ipv4_ops[] = {
-	/* Before packet filtering, change destination */
 	{
-		.hook		= iptable_nat_ipv4_in,
+		.hook		= iptable_nat_do_chain,
 		.pf		= NFPROTO_IPV4,
-		.nat_hook	= true,
 		.hooknum	= NF_INET_PRE_ROUTING,
 		.priority	= NF_IP_PRI_NAT_DST,
 	},
-	/* After packet filtering, change source */
 	{
-		.hook		= iptable_nat_ipv4_out,
+		.hook		= iptable_nat_do_chain,
 		.pf		= NFPROTO_IPV4,
-		.nat_hook	= true,
 		.hooknum	= NF_INET_POST_ROUTING,
 		.priority	= NF_IP_PRI_NAT_SRC,
 	},
-	/* Before packet filtering, change destination */
 	{
-		.hook		= iptable_nat_ipv4_local_fn,
+		.hook		= iptable_nat_do_chain,
 		.pf		= NFPROTO_IPV4,
-		.nat_hook	= true,
 		.hooknum	= NF_INET_LOCAL_OUT,
 		.priority	= NF_IP_PRI_NAT_DST,
 	},
-	/* After packet filtering, change source */
 	{
-		.hook		= iptable_nat_ipv4_fn,
+		.hook		= iptable_nat_do_chain,
 		.pf		= NFPROTO_IPV4,
-		.nat_hook	= true,
 		.hooknum	= NF_INET_LOCAL_IN,
 		.priority	= NF_IP_PRI_NAT_SRC,
 	},
 };
 
+static int ipt_nat_register_lookups(struct net *net)
+{
+	int i, ret;
+
+	for (i = 0; i < ARRAY_SIZE(nf_nat_ipv4_ops); i++) {
+		ret = nf_nat_l3proto_ipv4_register_fn(net, &nf_nat_ipv4_ops[i]);
+		if (ret) {
+			while (i)
+				nf_nat_l3proto_ipv4_unregister_fn(net, &nf_nat_ipv4_ops[--i]);
+
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
+static void ipt_nat_unregister_lookups(struct net *net)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(nf_nat_ipv4_ops); i++)
+		nf_nat_l3proto_ipv4_unregister_fn(net, &nf_nat_ipv4_ops[i]);
+}
+
 static int __net_init iptable_nat_table_init(struct net *net)
 {
 	struct ipt_replace *repl;
@@ -113,7 +102,18 @@ static int __net_init iptable_nat_table_init(struct net *net)
 	if (repl == NULL)
 		return -ENOMEM;
 	ret = ipt_register_table(net, &nf_nat_ipv4_table, repl,
-				 nf_nat_ipv4_ops, &net->ipv4.nat_table);
+				 NULL, &net->ipv4.nat_table);
+	if (ret < 0) {
+		kfree(repl);
+		return ret;
+	}
+
+	ret = ipt_nat_register_lookups(net);
+	if (ret < 0) {
+		ipt_unregister_table(net, net->ipv4.nat_table, NULL);
+		net->ipv4.nat_table = NULL;
+	}
+
 	kfree(repl);
 	return ret;
 }
@@ -122,7 +122,8 @@ static void __net_exit iptable_nat_net_exit(struct net *net)
 {
 	if (!net->ipv4.nat_table)
 		return;
-	ipt_unregister_table(net, net->ipv4.nat_table, nf_nat_ipv4_ops);
+	ipt_nat_unregister_lookups(net);
+	ipt_unregister_table(net, net->ipv4.nat_table, NULL);
 	net->ipv4.nat_table = NULL;
 }
 
diff --git a/net/ipv4/netfilter/nf_nat_l3proto_ipv4.c b/net/ipv4/netfilter/nf_nat_l3proto_ipv4.c
index 29b5aceac66d..6115bf1ff6f0 100644
--- a/net/ipv4/netfilter/nf_nat_l3proto_ipv4.c
+++ b/net/ipv4/netfilter/nf_nat_l3proto_ipv4.c
@@ -241,12 +241,9 @@ int nf_nat_icmp_reply_translation(struct sk_buff *skb,
 }
 EXPORT_SYMBOL_GPL(nf_nat_icmp_reply_translation);
 
-unsigned int
+static unsigned int
 nf_nat_ipv4_fn(void *priv, struct sk_buff *skb,
-	       const struct nf_hook_state *state,
-	       unsigned int (*do_chain)(void *priv,
-					struct sk_buff *skb,
-					const struct nf_hook_state *state))
+	       const struct nf_hook_state *state)
 {
 	struct nf_conn *ct;
 	enum ip_conntrack_info ctinfo;
@@ -265,35 +262,28 @@ nf_nat_ipv4_fn(void *priv, struct sk_buff *skb,
 		}
 	}
 
-	return nf_nat_inet_fn(priv, skb, state, do_chain);
+	return nf_nat_inet_fn(priv, skb, state);
 }
 EXPORT_SYMBOL_GPL(nf_nat_ipv4_fn);
 
-unsigned int
+static unsigned int
 nf_nat_ipv4_in(void *priv, struct sk_buff *skb,
-	       const struct nf_hook_state *state,
-	       unsigned int (*do_chain)(void *priv,
-					 struct sk_buff *skb,
-					 const struct nf_hook_state *state))
+	       const struct nf_hook_state *state)
 {
 	unsigned int ret;
 	__be32 daddr = ip_hdr(skb)->daddr;
 
-	ret = nf_nat_ipv4_fn(priv, skb, state, do_chain);
+	ret = nf_nat_ipv4_fn(priv, skb, state);
 	if (ret != NF_DROP && ret != NF_STOLEN &&
 	    daddr != ip_hdr(skb)->daddr)
 		skb_dst_drop(skb);
 
 	return ret;
 }
-EXPORT_SYMBOL_GPL(nf_nat_ipv4_in);
 
-unsigned int
+static unsigned int
 nf_nat_ipv4_out(void *priv, struct sk_buff *skb,
-		const struct nf_hook_state *state,
-		unsigned int (*do_chain)(void *priv,
-					  struct sk_buff *skb,
-					  const struct nf_hook_state *state))
+		const struct nf_hook_state *state)
 {
 #ifdef CONFIG_XFRM
 	const struct nf_conn *ct;
@@ -302,7 +292,7 @@ nf_nat_ipv4_out(void *priv, struct sk_buff *skb,
 #endif
 	unsigned int ret;
 
-	ret = nf_nat_ipv4_fn(priv, skb, state, do_chain);
+	ret = nf_nat_ipv4_fn(priv, skb, state);
 #ifdef CONFIG_XFRM
 	if (ret != NF_DROP && ret != NF_STOLEN &&
 	    !(IPCB(skb)->flags & IPSKB_XFRM_TRANSFORMED) &&
@@ -322,21 +312,17 @@ nf_nat_ipv4_out(void *priv, struct sk_buff *skb,
 #endif
 	return ret;
 }
-EXPORT_SYMBOL_GPL(nf_nat_ipv4_out);
 
-unsigned int
+static unsigned int
 nf_nat_ipv4_local_fn(void *priv, struct sk_buff *skb,
-		     const struct nf_hook_state *state,
-		     unsigned int (*do_chain)(void *priv,
-					       struct sk_buff *skb,
-					       const struct nf_hook_state *state))
+		     const struct nf_hook_state *state)
 {
 	const struct nf_conn *ct;
 	enum ip_conntrack_info ctinfo;
 	unsigned int ret;
 	int err;
 
-	ret = nf_nat_ipv4_fn(priv, skb, state, do_chain);
+	ret = nf_nat_ipv4_fn(priv, skb, state);
 	if (ret != NF_DROP && ret != NF_STOLEN &&
 	    (ct = nf_ct_get(skb, &ctinfo)) != NULL) {
 		enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
@@ -360,7 +346,49 @@ nf_nat_ipv4_local_fn(void *priv, struct sk_buff *skb,
 	}
 	return ret;
 }
-EXPORT_SYMBOL_GPL(nf_nat_ipv4_local_fn);
+
+static const struct nf_hook_ops nf_nat_ipv4_ops[] = {
+	/* Before packet filtering, change destination */
+	{
+		.hook		= nf_nat_ipv4_in,
+		.pf		= NFPROTO_IPV4,
+		.hooknum	= NF_INET_PRE_ROUTING,
+		.priority	= NF_IP_PRI_NAT_DST,
+	},
+	/* After packet filtering, change source */
+	{
+		.hook		= nf_nat_ipv4_out,
+		.pf		= NFPROTO_IPV4,
+		.hooknum	= NF_INET_POST_ROUTING,
+		.priority	= NF_IP_PRI_NAT_SRC,
+	},
+	/* Before packet filtering, change destination */
+	{
+		.hook		= nf_nat_ipv4_local_fn,
+		.pf		= NFPROTO_IPV4,
+		.hooknum	= NF_INET_LOCAL_OUT,
+		.priority	= NF_IP_PRI_NAT_DST,
+	},
+	/* After packet filtering, change source */
+	{
+		.hook		= nf_nat_ipv4_fn,
+		.pf		= NFPROTO_IPV4,
+		.hooknum	= NF_INET_LOCAL_IN,
+		.priority	= NF_IP_PRI_NAT_SRC,
+	},
+};
+
+int nf_nat_l3proto_ipv4_register_fn(struct net *net, const struct nf_hook_ops *ops)
+{
+	return nf_nat_register_fn(net, ops, nf_nat_ipv4_ops, ARRAY_SIZE(nf_nat_ipv4_ops));
+}
+EXPORT_SYMBOL_GPL(nf_nat_l3proto_ipv4_register_fn);
+
+void nf_nat_l3proto_ipv4_unregister_fn(struct net *net, const struct nf_hook_ops *ops)
+{
+	nf_nat_unregister_fn(net, ops, ARRAY_SIZE(nf_nat_ipv4_ops));
+}
+EXPORT_SYMBOL_GPL(nf_nat_l3proto_ipv4_unregister_fn);
 
 static int __init nf_nat_l3proto_ipv4_init(void)
 {
diff --git a/net/ipv4/netfilter/nft_chain_nat_ipv4.c b/net/ipv4/netfilter/nft_chain_nat_ipv4.c
index bbcb624b6b81..a3c4ea303e3e 100644
--- a/net/ipv4/netfilter/nft_chain_nat_ipv4.c
+++ b/net/ipv4/netfilter/nft_chain_nat_ipv4.c
@@ -27,8 +27,8 @@
 #include <net/ip.h>
 
 static unsigned int nft_nat_do_chain(void *priv,
-				      struct sk_buff *skb,
-				      const struct nf_hook_state *state)
+				     struct sk_buff *skb,
+				     const struct nf_hook_state *state)
 {
 	struct nft_pktinfo pkt;
 
@@ -38,49 +38,14 @@ static unsigned int nft_nat_do_chain(void *priv,
 	return nft_do_chain(&pkt, priv);
 }
 
-static unsigned int nft_nat_ipv4_fn(void *priv,
-				    struct sk_buff *skb,
-				    const struct nf_hook_state *state)
-{
-	return nf_nat_ipv4_fn(priv, skb, state, nft_nat_do_chain);
-}
-
-static unsigned int nft_nat_ipv4_in(void *priv,
-				    struct sk_buff *skb,
-				    const struct nf_hook_state *state)
-{
-	return nf_nat_ipv4_in(priv, skb, state, nft_nat_do_chain);
-}
-
-static unsigned int nft_nat_ipv4_out(void *priv,
-				     struct sk_buff *skb,
-				     const struct nf_hook_state *state)
-{
-	return nf_nat_ipv4_out(priv, skb, state, nft_nat_do_chain);
-}
-
-static unsigned int nft_nat_ipv4_local_fn(void *priv,
-					  struct sk_buff *skb,
-					  const struct nf_hook_state *state)
-{
-	return nf_nat_ipv4_local_fn(priv, skb, state, nft_nat_do_chain);
-}
-
 static int nft_nat_ipv4_reg(struct net *net, const struct nf_hook_ops *ops)
 {
-	int ret = nf_register_net_hook(net, ops);
-	if (ret == 0) {
-		ret = nf_ct_netns_get(net, NFPROTO_IPV4);
-		if (ret)
-			 nf_unregister_net_hook(net, ops);
-	}
-	return ret;
+	return nf_nat_l3proto_ipv4_register_fn(net, ops);
 }
 
 static void nft_nat_ipv4_unreg(struct net *net, const struct nf_hook_ops *ops)
 {
-	nf_unregister_net_hook(net, ops);
-	nf_ct_netns_put(net, NFPROTO_IPV4);
+	nf_nat_l3proto_ipv4_unregister_fn(net, ops);
 }
 
 static const struct nft_chain_type nft_chain_nat_ipv4 = {
@@ -93,10 +58,10 @@ static const struct nft_chain_type nft_chain_nat_ipv4 = {
 			  (1 << NF_INET_LOCAL_OUT) |
 			  (1 << NF_INET_LOCAL_IN),
 	.hooks		= {
-		[NF_INET_PRE_ROUTING]	= nft_nat_ipv4_in,
-		[NF_INET_POST_ROUTING]	= nft_nat_ipv4_out,
-		[NF_INET_LOCAL_OUT]	= nft_nat_ipv4_local_fn,
-		[NF_INET_LOCAL_IN]	= nft_nat_ipv4_fn,
+		[NF_INET_PRE_ROUTING]	= nft_nat_do_chain,
+		[NF_INET_POST_ROUTING]	= nft_nat_do_chain,
+		[NF_INET_LOCAL_OUT]	= nft_nat_do_chain,
+		[NF_INET_LOCAL_IN]	= nft_nat_do_chain,
 	},
 	.ops_register = nft_nat_ipv4_reg,
 	.ops_unregister = nft_nat_ipv4_unreg,