netfilter: nf_conntrack: IPS_UNTRACKED bit
[deliverable/linux.git] / net / ipv4 / netfilter / nf_nat_standalone.c
CommitLineData
5b1158e9
JK
1/* (C) 1999-2001 Paul `Rusty' Russell
2 * (C) 2002-2006 Netfilter Core Team <coreteam@netfilter.org>
3 *
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License version 2 as
6 * published by the Free Software Foundation.
7 */
8#include <linux/types.h>
9#include <linux/icmp.h>
5a0e3ad6 10#include <linux/gfp.h>
5b1158e9
JK
11#include <linux/ip.h>
12#include <linux/netfilter.h>
13#include <linux/netfilter_ipv4.h>
14#include <linux/module.h>
15#include <linux/skbuff.h>
16#include <linux/proc_fs.h>
17#include <net/ip.h>
18#include <net/checksum.h>
19#include <linux/spinlock.h>
20
21#include <net/netfilter/nf_conntrack.h>
22#include <net/netfilter/nf_conntrack_core.h>
2d59e5ca 23#include <net/netfilter/nf_conntrack_extend.h>
5b1158e9
JK
24#include <net/netfilter/nf_nat.h>
25#include <net/netfilter/nf_nat_rule.h>
26#include <net/netfilter/nf_nat_protocol.h>
27#include <net/netfilter/nf_nat_core.h>
28#include <net/netfilter/nf_nat_helper.h>
29#include <linux/netfilter_ipv4/ip_tables.h>
30
5b1158e9
JK
31#ifdef CONFIG_XFRM
32static void nat_decode_session(struct sk_buff *skb, struct flowi *fl)
33{
72b72949
JE
34 const struct nf_conn *ct;
35 const struct nf_conntrack_tuple *t;
5b1158e9
JK
36 enum ip_conntrack_info ctinfo;
37 enum ip_conntrack_dir dir;
38 unsigned long statusbit;
39
40 ct = nf_ct_get(skb, &ctinfo);
41 if (ct == NULL)
42 return;
43 dir = CTINFO2DIR(ctinfo);
44 t = &ct->tuplehash[dir].tuple;
45
46 if (dir == IP_CT_DIR_ORIGINAL)
47 statusbit = IPS_DST_NAT;
48 else
49 statusbit = IPS_SRC_NAT;
50
51 if (ct->status & statusbit) {
52 fl->fl4_dst = t->dst.u3.ip;
53 if (t->dst.protonum == IPPROTO_TCP ||
6185f870 54 t->dst.protonum == IPPROTO_UDP ||
4910a087 55 t->dst.protonum == IPPROTO_UDPLITE ||
9d908a69
PM
56 t->dst.protonum == IPPROTO_DCCP ||
57 t->dst.protonum == IPPROTO_SCTP)
5b1158e9
JK
58 fl->fl_ip_dport = t->dst.u.tcp.port;
59 }
60
61 statusbit ^= IPS_NAT_MASK;
62
63 if (ct->status & statusbit) {
64 fl->fl4_src = t->src.u3.ip;
65 if (t->dst.protonum == IPPROTO_TCP ||
6185f870 66 t->dst.protonum == IPPROTO_UDP ||
4910a087 67 t->dst.protonum == IPPROTO_UDPLITE ||
9d908a69
PM
68 t->dst.protonum == IPPROTO_DCCP ||
69 t->dst.protonum == IPPROTO_SCTP)
5b1158e9
JK
70 fl->fl_ip_sport = t->src.u.tcp.port;
71 }
72}
73#endif
74
75static unsigned int
76nf_nat_fn(unsigned int hooknum,
3db05fea 77 struct sk_buff *skb,
5b1158e9
JK
78 const struct net_device *in,
79 const struct net_device *out,
80 int (*okfn)(struct sk_buff *))
81{
82 struct nf_conn *ct;
83 enum ip_conntrack_info ctinfo;
84 struct nf_conn_nat *nat;
5b1158e9
JK
85 /* maniptype == SRC for postrouting. */
86 enum nf_nat_manip_type maniptype = HOOK2MANIP(hooknum);
87
88 /* We never see fragments: conntrack defrags on pre-routing
89 and local-out, and nf_nat_out protects post-routing. */
3db05fea 90 NF_CT_ASSERT(!(ip_hdr(skb)->frag_off & htons(IP_MF | IP_OFFSET)));
5b1158e9 91
3db05fea 92 ct = nf_ct_get(skb, &ctinfo);
5b1158e9
JK
93 /* Can't track? It's not due to stress, or conntrack would
94 have dropped it. Hence it's the user's responsibilty to
95 packet filter it out, or implement conntrack/NAT for that
96 protocol. 8) --RR */
42cf800c 97 if (!ct)
5b1158e9 98 return NF_ACCEPT;
5b1158e9
JK
99
100 /* Don't try to NAT if this packet is not conntracked */
5bfddbd4 101 if (nf_ct_is_untracked(ct))
5b1158e9
JK
102 return NF_ACCEPT;
103
104 nat = nfct_nat(ct);
2d59e5ca 105 if (!nat) {
8c87238b
PM
106 /* NAT module was loaded late. */
107 if (nf_ct_is_confirmed(ct))
108 return NF_ACCEPT;
2d59e5ca
YK
109 nat = nf_ct_ext_add(ct, NF_CT_EXT_NAT, GFP_ATOMIC);
110 if (nat == NULL) {
0d53778e 111 pr_debug("failed to add NAT extension\n");
2d59e5ca
YK
112 return NF_ACCEPT;
113 }
114 }
5b1158e9
JK
115
116 switch (ctinfo) {
117 case IP_CT_RELATED:
118 case IP_CT_RELATED+IP_CT_IS_REPLY:
3db05fea 119 if (ip_hdr(skb)->protocol == IPPROTO_ICMP) {
5b1158e9 120 if (!nf_nat_icmp_reply_translation(ct, ctinfo,
3db05fea 121 hooknum, skb))
5b1158e9
JK
122 return NF_DROP;
123 else
124 return NF_ACCEPT;
125 }
126 /* Fall thru... (Only ICMPs can be IP_CT_IS_REPLY) */
127 case IP_CT_NEW:
5b1158e9
JK
128
129 /* Seen it before? This can happen for loopback, retrans,
130 or local packets.. */
131 if (!nf_nat_initialized(ct, maniptype)) {
132 unsigned int ret;
133
8c87238b 134 if (hooknum == NF_INET_LOCAL_IN)
5b1158e9 135 /* LOCAL_IN hook doesn't have a chain! */
ba4c7cba 136 ret = alloc_null_binding(ct, hooknum);
5b1158e9 137 else
3db05fea 138 ret = nf_nat_rule_find(skb, hooknum, in, out,
ba4c7cba 139 ct);
5b1158e9 140
22068311 141 if (ret != NF_ACCEPT)
5b1158e9 142 return ret;
5b1158e9 143 } else
0d53778e
PM
144 pr_debug("Already setup manip %s for ct %p\n",
145 maniptype == IP_NAT_MANIP_SRC ? "SRC" : "DST",
146 ct);
5b1158e9
JK
147 break;
148
149 default:
150 /* ESTABLISHED */
151 NF_CT_ASSERT(ctinfo == IP_CT_ESTABLISHED ||
152 ctinfo == (IP_CT_ESTABLISHED+IP_CT_IS_REPLY));
5b1158e9
JK
153 }
154
3db05fea 155 return nf_nat_packet(ct, ctinfo, hooknum, skb);
5b1158e9
JK
156}
157
158static unsigned int
159nf_nat_in(unsigned int hooknum,
3db05fea 160 struct sk_buff *skb,
e905a9ed
YH
161 const struct net_device *in,
162 const struct net_device *out,
163 int (*okfn)(struct sk_buff *))
5b1158e9
JK
164{
165 unsigned int ret;
3db05fea 166 __be32 daddr = ip_hdr(skb)->daddr;
5b1158e9 167
3db05fea 168 ret = nf_nat_fn(hooknum, skb, in, out, okfn);
5b1158e9 169 if (ret != NF_DROP && ret != NF_STOLEN &&
adf30907
ED
170 daddr != ip_hdr(skb)->daddr)
171 skb_dst_drop(skb);
172
5b1158e9
JK
173 return ret;
174}
175
176static unsigned int
177nf_nat_out(unsigned int hooknum,
3db05fea 178 struct sk_buff *skb,
5b1158e9
JK
179 const struct net_device *in,
180 const struct net_device *out,
181 int (*okfn)(struct sk_buff *))
182{
183#ifdef CONFIG_XFRM
72b72949 184 const struct nf_conn *ct;
5b1158e9
JK
185 enum ip_conntrack_info ctinfo;
186#endif
187 unsigned int ret;
188
189 /* root is playing with raw sockets. */
3db05fea
HX
190 if (skb->len < sizeof(struct iphdr) ||
191 ip_hdrlen(skb) < sizeof(struct iphdr))
5b1158e9
JK
192 return NF_ACCEPT;
193
3db05fea 194 ret = nf_nat_fn(hooknum, skb, in, out, okfn);
5b1158e9
JK
195#ifdef CONFIG_XFRM
196 if (ret != NF_DROP && ret != NF_STOLEN &&
3db05fea 197 (ct = nf_ct_get(skb, &ctinfo)) != NULL) {
5b1158e9
JK
198 enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
199
3666ed1c
JP
200 if ((ct->tuplehash[dir].tuple.src.u3.ip !=
201 ct->tuplehash[!dir].tuple.dst.u3.ip) ||
202 (ct->tuplehash[dir].tuple.src.u.all !=
203 ct->tuplehash[!dir].tuple.dst.u.all)
204 )
3db05fea 205 return ip_xfrm_me_harder(skb) == 0 ? ret : NF_DROP;
5b1158e9
JK
206 }
207#endif
208 return ret;
209}
210
211static unsigned int
212nf_nat_local_fn(unsigned int hooknum,
3db05fea 213 struct sk_buff *skb,
5b1158e9
JK
214 const struct net_device *in,
215 const struct net_device *out,
216 int (*okfn)(struct sk_buff *))
217{
72b72949 218 const struct nf_conn *ct;
5b1158e9
JK
219 enum ip_conntrack_info ctinfo;
220 unsigned int ret;
221
222 /* root is playing with raw sockets. */
3db05fea
HX
223 if (skb->len < sizeof(struct iphdr) ||
224 ip_hdrlen(skb) < sizeof(struct iphdr))
5b1158e9
JK
225 return NF_ACCEPT;
226
3db05fea 227 ret = nf_nat_fn(hooknum, skb, in, out, okfn);
5b1158e9 228 if (ret != NF_DROP && ret != NF_STOLEN &&
3db05fea 229 (ct = nf_ct_get(skb, &ctinfo)) != NULL) {
5b1158e9
JK
230 enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
231
232 if (ct->tuplehash[dir].tuple.dst.u3.ip !=
848c29fd 233 ct->tuplehash[!dir].tuple.src.u3.ip) {
3db05fea 234 if (ip_route_me_harder(skb, RTN_UNSPEC))
5b1158e9 235 ret = NF_DROP;
848c29fd
PM
236 }
237#ifdef CONFIG_XFRM
238 else if (ct->tuplehash[dir].tuple.dst.u.all !=
239 ct->tuplehash[!dir].tuple.src.u.all)
3db05fea 240 if (ip_xfrm_me_harder(skb))
848c29fd
PM
241 ret = NF_DROP;
242#endif
5b1158e9
JK
243 }
244 return ret;
245}
246
5b1158e9
JK
247/* We must be after connection tracking and before packet filtering. */
248
1999414a 249static struct nf_hook_ops nf_nat_ops[] __read_mostly = {
5b1158e9
JK
250 /* Before packet filtering, change destination */
251 {
252 .hook = nf_nat_in,
253 .owner = THIS_MODULE,
24c232d8 254 .pf = NFPROTO_IPV4,
6e23ae2a 255 .hooknum = NF_INET_PRE_ROUTING,
5b1158e9
JK
256 .priority = NF_IP_PRI_NAT_DST,
257 },
258 /* After packet filtering, change source */
259 {
260 .hook = nf_nat_out,
261 .owner = THIS_MODULE,
24c232d8 262 .pf = NFPROTO_IPV4,
6e23ae2a 263 .hooknum = NF_INET_POST_ROUTING,
5b1158e9
JK
264 .priority = NF_IP_PRI_NAT_SRC,
265 },
5b1158e9
JK
266 /* Before packet filtering, change destination */
267 {
268 .hook = nf_nat_local_fn,
269 .owner = THIS_MODULE,
24c232d8 270 .pf = NFPROTO_IPV4,
6e23ae2a 271 .hooknum = NF_INET_LOCAL_OUT,
5b1158e9
JK
272 .priority = NF_IP_PRI_NAT_DST,
273 },
274 /* After packet filtering, change source */
275 {
276 .hook = nf_nat_fn,
277 .owner = THIS_MODULE,
24c232d8 278 .pf = NFPROTO_IPV4,
6e23ae2a 279 .hooknum = NF_INET_LOCAL_IN,
5b1158e9
JK
280 .priority = NF_IP_PRI_NAT_SRC,
281 },
5b1158e9
JK
282};
283
284static int __init nf_nat_standalone_init(void)
285{
2d59e5ca 286 int ret = 0;
5b1158e9 287
591e6206 288 need_ipv4_conntrack();
5b1158e9 289
5b1158e9
JK
290#ifdef CONFIG_XFRM
291 BUG_ON(ip_nat_decode_session != NULL);
051578cc 292 rcu_assign_pointer(ip_nat_decode_session, nat_decode_session);
5b1158e9
JK
293#endif
294 ret = nf_nat_rule_init();
295 if (ret < 0) {
654d0fbd 296 pr_err("nf_nat_init: can't setup rules.\n");
5b1158e9
JK
297 goto cleanup_decode_session;
298 }
299 ret = nf_register_hooks(nf_nat_ops, ARRAY_SIZE(nf_nat_ops));
300 if (ret < 0) {
654d0fbd 301 pr_err("nf_nat_init: can't register hooks.\n");
5b1158e9
JK
302 goto cleanup_rule_init;
303 }
5b1158e9
JK
304 return ret;
305
306 cleanup_rule_init:
307 nf_nat_rule_cleanup();
308 cleanup_decode_session:
309#ifdef CONFIG_XFRM
051578cc 310 rcu_assign_pointer(ip_nat_decode_session, NULL);
5b1158e9
JK
311 synchronize_net();
312#endif
5b1158e9
JK
313 return ret;
314}
315
316static void __exit nf_nat_standalone_fini(void)
317{
318 nf_unregister_hooks(nf_nat_ops, ARRAY_SIZE(nf_nat_ops));
319 nf_nat_rule_cleanup();
5b1158e9 320#ifdef CONFIG_XFRM
051578cc 321 rcu_assign_pointer(ip_nat_decode_session, NULL);
5b1158e9
JK
322 synchronize_net();
323#endif
324 /* Conntrack caches are unregistered in nf_conntrack_cleanup */
325}
326
327module_init(nf_nat_standalone_init);
328module_exit(nf_nat_standalone_fini);
329
330MODULE_LICENSE("GPL");
331MODULE_ALIAS("ip_nat");
This page took 0.408716 seconds and 5 git commands to generate.