2 * (C) 1999-2001 Paul `Rusty' Russell
3 * (C) 2002-2006 Netfilter Core Team <coreteam@netfilter.org>
4 * (C) 2011 Patrick McHardy <kaber@trash.net>
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 as
8 * published by the Free Software Foundation.
11 #include <linux/module.h>
12 #include <linux/types.h>
13 #include <linux/timer.h>
14 #include <linux/skbuff.h>
15 #include <linux/gfp.h>
17 #include <linux/jhash.h>
18 #include <linux/rtnetlink.h>
20 #include <net/netfilter/nf_conntrack.h>
21 #include <net/netfilter/nf_conntrack_core.h>
22 #include <net/netfilter/nf_nat.h>
23 #include <net/netfilter/nf_nat_l3proto.h>
24 #include <net/netfilter/nf_nat_l4proto.h>
25 #include <net/netfilter/nf_nat_core.h>
26 #include <net/netfilter/nf_nat_helper.h>
27 #include <net/netfilter/nf_conntrack_helper.h>
28 #include <net/netfilter/nf_conntrack_seqadj.h>
29 #include <net/netfilter/nf_conntrack_l3proto.h>
30 #include <net/netfilter/nf_conntrack_zones.h>
31 #include <linux/netfilter/nf_nat.h>
33 static DEFINE_SPINLOCK(nf_nat_lock);
35 static DEFINE_MUTEX(nf_nat_proto_mutex);
36 static const struct nf_nat_l3proto __rcu *nf_nat_l3protos[NFPROTO_NUMPROTO]
38 static const struct nf_nat_l4proto __rcu **nf_nat_l4protos[NFPROTO_NUMPROTO]
41 static struct hlist_head *nf_nat_bysource __read_mostly;
42 static unsigned int nf_nat_htable_size __read_mostly;
43 static unsigned int nf_nat_hash_rnd __read_mostly;
45 inline const struct nf_nat_l3proto *
46 __nf_nat_l3proto_find(u8 family)
48 return rcu_dereference(nf_nat_l3protos[family]);
51 inline const struct nf_nat_l4proto *
52 __nf_nat_l4proto_find(u8 family, u8 protonum)
54 return rcu_dereference(nf_nat_l4protos[family][protonum]);
56 EXPORT_SYMBOL_GPL(__nf_nat_l4proto_find);
59 static void __nf_nat_decode_session(struct sk_buff *skb, struct flowi *fl)
61 const struct nf_nat_l3proto *l3proto;
62 const struct nf_conn *ct;
63 enum ip_conntrack_info ctinfo;
64 enum ip_conntrack_dir dir;
65 unsigned long statusbit;
68 ct = nf_ct_get(skb, &ctinfo);
72 family = ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.src.l3num;
74 l3proto = __nf_nat_l3proto_find(family);
78 dir = CTINFO2DIR(ctinfo);
79 if (dir == IP_CT_DIR_ORIGINAL)
80 statusbit = IPS_DST_NAT;
82 statusbit = IPS_SRC_NAT;
84 l3proto->decode_session(skb, ct, dir, statusbit, fl);
89 int nf_xfrm_me_harder(struct net *net, struct sk_buff *skb, unsigned int family)
93 struct dst_entry *dst;
96 err = xfrm_decode_session(skb, &fl, family);
102 dst = ((struct xfrm_dst *)dst)->route;
105 dst = xfrm_lookup(net, dst, &fl, skb->sk, 0);
110 skb_dst_set(skb, dst);
112 /* Change in oif may mean change in hh_len. */
113 hh_len = skb_dst(skb)->dev->hard_header_len;
114 if (skb_headroom(skb) < hh_len &&
115 pskb_expand_head(skb, hh_len - skb_headroom(skb), 0, GFP_ATOMIC))
119 EXPORT_SYMBOL(nf_xfrm_me_harder);
120 #endif /* CONFIG_XFRM */
122 /* We keep an extra hash for each conntrack, for fast searching. */
123 static inline unsigned int
124 hash_by_src(const struct net *n, const struct nf_conntrack_tuple *tuple)
128 get_random_once(&nf_nat_hash_rnd, sizeof(nf_nat_hash_rnd));
130 /* Original src, to ensure we map it consistently if poss. */
131 hash = jhash2((u32 *)&tuple->src, sizeof(tuple->src) / sizeof(u32),
132 tuple->dst.protonum ^ nf_nat_hash_rnd ^ net_hash_mix(n));
134 return reciprocal_scale(hash, nf_nat_htable_size);
137 /* Is this tuple already taken? (not by us) */
139 nf_nat_used_tuple(const struct nf_conntrack_tuple *tuple,
140 const struct nf_conn *ignored_conntrack)
142 /* Conntrack tracking doesn't keep track of outgoing tuples; only
143 * incoming ones. NAT means they don't have a fixed mapping,
144 * so we invert the tuple and look for the incoming reply.
146 * We could keep a separate hash if this proves too slow.
148 struct nf_conntrack_tuple reply;
150 nf_ct_invert_tuplepr(&reply, tuple);
151 return nf_conntrack_tuple_taken(&reply, ignored_conntrack);
153 EXPORT_SYMBOL(nf_nat_used_tuple);
155 /* If we source map this tuple so reply looks like reply_tuple, will
156 * that meet the constraints of range.
158 static int in_range(const struct nf_nat_l3proto *l3proto,
159 const struct nf_nat_l4proto *l4proto,
160 const struct nf_conntrack_tuple *tuple,
161 const struct nf_nat_range *range)
163 /* If we are supposed to map IPs, then we must be in the
164 * range specified, otherwise let this drag us onto a new src IP.
166 if (range->flags & NF_NAT_RANGE_MAP_IPS &&
167 !l3proto->in_range(tuple, range))
170 if (!(range->flags & NF_NAT_RANGE_PROTO_SPECIFIED) ||
171 l4proto->in_range(tuple, NF_NAT_MANIP_SRC,
172 &range->min_proto, &range->max_proto))
179 same_src(const struct nf_conn *ct,
180 const struct nf_conntrack_tuple *tuple)
182 const struct nf_conntrack_tuple *t;
184 t = &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple;
185 return (t->dst.protonum == tuple->dst.protonum &&
186 nf_inet_addr_cmp(&t->src.u3, &tuple->src.u3) &&
187 t->src.u.all == tuple->src.u.all);
190 /* Only called for SRC manip */
192 find_appropriate_src(struct net *net,
193 const struct nf_conntrack_zone *zone,
194 const struct nf_nat_l3proto *l3proto,
195 const struct nf_nat_l4proto *l4proto,
196 const struct nf_conntrack_tuple *tuple,
197 struct nf_conntrack_tuple *result,
198 const struct nf_nat_range *range)
200 unsigned int h = hash_by_src(net, tuple);
201 const struct nf_conn_nat *nat;
202 const struct nf_conn *ct;
204 hlist_for_each_entry_rcu(nat, &nf_nat_bysource[h], bysource) {
206 if (same_src(ct, tuple) &&
207 net_eq(net, nf_ct_net(ct)) &&
208 nf_ct_zone_equal(ct, zone, IP_CT_DIR_ORIGINAL)) {
209 /* Copy source part from reply tuple. */
210 nf_ct_invert_tuplepr(result,
211 &ct->tuplehash[IP_CT_DIR_REPLY].tuple);
212 result->dst = tuple->dst;
214 if (in_range(l3proto, l4proto, result, range))
221 /* For [FUTURE] fragmentation handling, we want the least-used
222 * src-ip/dst-ip/proto triple. Fairness doesn't come into it. Thus
223 * if the range specifies 1.2.3.4 ports 10000-10005 and 1.2.3.5 ports
224 * 1-65535, we don't do pro-rata allocation based on ports; we choose
225 * the ip with the lowest src-ip/dst-ip/proto usage.
228 find_best_ips_proto(const struct nf_conntrack_zone *zone,
229 struct nf_conntrack_tuple *tuple,
230 const struct nf_nat_range *range,
231 const struct nf_conn *ct,
232 enum nf_nat_manip_type maniptype)
234 union nf_inet_addr *var_ipp;
237 u32 minip, maxip, j, dist;
240 /* No IP mapping? Do nothing. */
241 if (!(range->flags & NF_NAT_RANGE_MAP_IPS))
244 if (maniptype == NF_NAT_MANIP_SRC)
245 var_ipp = &tuple->src.u3;
247 var_ipp = &tuple->dst.u3;
249 /* Fast path: only one choice. */
250 if (nf_inet_addr_cmp(&range->min_addr, &range->max_addr)) {
251 *var_ipp = range->min_addr;
255 if (nf_ct_l3num(ct) == NFPROTO_IPV4)
256 max = sizeof(var_ipp->ip) / sizeof(u32) - 1;
258 max = sizeof(var_ipp->ip6) / sizeof(u32) - 1;
260 /* Hashing source and destination IPs gives a fairly even
261 * spread in practice (if there are a small number of IPs
262 * involved, there usually aren't that many connections
263 * anyway). The consistency means that servers see the same
264 * client coming from the same IP (some Internet Banking sites
265 * like this), even across reboots.
267 j = jhash2((u32 *)&tuple->src.u3, sizeof(tuple->src.u3) / sizeof(u32),
268 range->flags & NF_NAT_RANGE_PERSISTENT ?
269 0 : (__force u32)tuple->dst.u3.all[max] ^ zone->id);
272 for (i = 0; i <= max; i++) {
273 /* If first bytes of the address are at the maximum, use the
274 * distance. Otherwise use the full range.
277 minip = ntohl((__force __be32)range->min_addr.all[i]);
278 maxip = ntohl((__force __be32)range->max_addr.all[i]);
279 dist = maxip - minip + 1;
285 var_ipp->all[i] = (__force __u32)
286 htonl(minip + reciprocal_scale(j, dist));
287 if (var_ipp->all[i] != range->max_addr.all[i])
290 if (!(range->flags & NF_NAT_RANGE_PERSISTENT))
291 j ^= (__force u32)tuple->dst.u3.all[i];
295 /* Manipulate the tuple into the range given. For NF_INET_POST_ROUTING,
296 * we change the source to map into the range. For NF_INET_PRE_ROUTING
297 * and NF_INET_LOCAL_OUT, we change the destination to map into the
298 * range. It might not be possible to get a unique tuple, but we try.
299 * At worst (or if we race), we will end up with a final duplicate in
300 * __ip_conntrack_confirm and drop the packet. */
302 get_unique_tuple(struct nf_conntrack_tuple *tuple,
303 const struct nf_conntrack_tuple *orig_tuple,
304 const struct nf_nat_range *range,
306 enum nf_nat_manip_type maniptype)
308 const struct nf_conntrack_zone *zone;
309 const struct nf_nat_l3proto *l3proto;
310 const struct nf_nat_l4proto *l4proto;
311 struct net *net = nf_ct_net(ct);
313 zone = nf_ct_zone(ct);
316 l3proto = __nf_nat_l3proto_find(orig_tuple->src.l3num);
317 l4proto = __nf_nat_l4proto_find(orig_tuple->src.l3num,
318 orig_tuple->dst.protonum);
320 /* 1) If this srcip/proto/src-proto-part is currently mapped,
321 * and that same mapping gives a unique tuple within the given
324 * This is only required for source (ie. NAT/masq) mappings.
325 * So far, we don't do local source mappings, so multiple
326 * manips not an issue.
328 if (maniptype == NF_NAT_MANIP_SRC &&
329 !(range->flags & NF_NAT_RANGE_PROTO_RANDOM_ALL)) {
330 /* try the original tuple first */
331 if (in_range(l3proto, l4proto, orig_tuple, range)) {
332 if (!nf_nat_used_tuple(orig_tuple, ct)) {
333 *tuple = *orig_tuple;
336 } else if (find_appropriate_src(net, zone, l3proto, l4proto,
337 orig_tuple, tuple, range)) {
338 pr_debug("get_unique_tuple: Found current src map\n");
339 if (!nf_nat_used_tuple(tuple, ct))
344 /* 2) Select the least-used IP/proto combination in the given range */
345 *tuple = *orig_tuple;
346 find_best_ips_proto(zone, tuple, range, ct, maniptype);
348 /* 3) The per-protocol part of the manip is made to map into
349 * the range to make a unique tuple.
352 /* Only bother mapping if it's not already in range and unique */
353 if (!(range->flags & NF_NAT_RANGE_PROTO_RANDOM_ALL)) {
354 if (range->flags & NF_NAT_RANGE_PROTO_SPECIFIED) {
355 if (l4proto->in_range(tuple, maniptype,
357 &range->max_proto) &&
358 (range->min_proto.all == range->max_proto.all ||
359 !nf_nat_used_tuple(tuple, ct)))
361 } else if (!nf_nat_used_tuple(tuple, ct)) {
366 /* Last change: get protocol to try to obtain unique tuple. */
367 l4proto->unique_tuple(l3proto, tuple, range, maniptype, ct);
372 struct nf_conn_nat *nf_ct_nat_ext_add(struct nf_conn *ct)
374 struct nf_conn_nat *nat = nfct_nat(ct);
378 if (!nf_ct_is_confirmed(ct))
379 nat = nf_ct_ext_add(ct, NF_CT_EXT_NAT, GFP_ATOMIC);
383 EXPORT_SYMBOL_GPL(nf_ct_nat_ext_add);
386 nf_nat_setup_info(struct nf_conn *ct,
387 const struct nf_nat_range *range,
388 enum nf_nat_manip_type maniptype)
390 struct net *net = nf_ct_net(ct);
391 struct nf_conntrack_tuple curr_tuple, new_tuple;
392 struct nf_conn_nat *nat;
394 /* nat helper or nfctnetlink also setup binding */
395 nat = nf_ct_nat_ext_add(ct);
399 NF_CT_ASSERT(maniptype == NF_NAT_MANIP_SRC ||
400 maniptype == NF_NAT_MANIP_DST);
401 BUG_ON(nf_nat_initialized(ct, maniptype));
403 /* What we've got will look like inverse of reply. Normally
404 * this is what is in the conntrack, except for prior
405 * manipulations (future optimization: if num_manips == 0,
406 * orig_tp = ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple)
408 nf_ct_invert_tuplepr(&curr_tuple,
409 &ct->tuplehash[IP_CT_DIR_REPLY].tuple);
411 get_unique_tuple(&new_tuple, &curr_tuple, range, ct, maniptype);
413 if (!nf_ct_tuple_equal(&new_tuple, &curr_tuple)) {
414 struct nf_conntrack_tuple reply;
416 /* Alter conntrack table so will recognize replies. */
417 nf_ct_invert_tuplepr(&reply, &new_tuple);
418 nf_conntrack_alter_reply(ct, &reply);
420 /* Non-atomic: we own this at the moment. */
421 if (maniptype == NF_NAT_MANIP_SRC)
422 ct->status |= IPS_SRC_NAT;
424 ct->status |= IPS_DST_NAT;
427 nfct_seqadj_ext_add(ct);
430 if (maniptype == NF_NAT_MANIP_SRC) {
431 unsigned int srchash;
433 srchash = hash_by_src(net,
434 &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple);
435 spin_lock_bh(&nf_nat_lock);
436 /* nf_conntrack_alter_reply might re-allocate extension aera */
439 hlist_add_head_rcu(&nat->bysource,
440 &nf_nat_bysource[srchash]);
441 spin_unlock_bh(&nf_nat_lock);
445 if (maniptype == NF_NAT_MANIP_DST)
446 ct->status |= IPS_DST_NAT_DONE;
448 ct->status |= IPS_SRC_NAT_DONE;
452 EXPORT_SYMBOL(nf_nat_setup_info);
455 __nf_nat_alloc_null_binding(struct nf_conn *ct, enum nf_nat_manip_type manip)
457 /* Force range to this IP; let proto decide mapping for
458 * per-proto parts (hence not IP_NAT_RANGE_PROTO_SPECIFIED).
459 * Use reply in case it's already been mangled (eg local packet).
461 union nf_inet_addr ip =
462 (manip == NF_NAT_MANIP_SRC ?
463 ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.u3 :
464 ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.u3);
465 struct nf_nat_range range = {
466 .flags = NF_NAT_RANGE_MAP_IPS,
470 return nf_nat_setup_info(ct, &range, manip);
474 nf_nat_alloc_null_binding(struct nf_conn *ct, unsigned int hooknum)
476 return __nf_nat_alloc_null_binding(ct, HOOK2MANIP(hooknum));
478 EXPORT_SYMBOL_GPL(nf_nat_alloc_null_binding);
480 /* Do packet manipulations according to nf_nat_setup_info. */
481 unsigned int nf_nat_packet(struct nf_conn *ct,
482 enum ip_conntrack_info ctinfo,
483 unsigned int hooknum,
486 const struct nf_nat_l3proto *l3proto;
487 const struct nf_nat_l4proto *l4proto;
488 enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
489 unsigned long statusbit;
490 enum nf_nat_manip_type mtype = HOOK2MANIP(hooknum);
492 if (mtype == NF_NAT_MANIP_SRC)
493 statusbit = IPS_SRC_NAT;
495 statusbit = IPS_DST_NAT;
497 /* Invert if this is reply dir. */
498 if (dir == IP_CT_DIR_REPLY)
499 statusbit ^= IPS_NAT_MASK;
501 /* Non-atomic: these bits don't change. */
502 if (ct->status & statusbit) {
503 struct nf_conntrack_tuple target;
505 /* We are aiming to look like inverse of other direction. */
506 nf_ct_invert_tuplepr(&target, &ct->tuplehash[!dir].tuple);
508 l3proto = __nf_nat_l3proto_find(target.src.l3num);
509 l4proto = __nf_nat_l4proto_find(target.src.l3num,
510 target.dst.protonum);
511 if (!l3proto->manip_pkt(skb, 0, l4proto, &target, mtype))
516 EXPORT_SYMBOL_GPL(nf_nat_packet);
518 struct nf_nat_proto_clean {
523 /* kill conntracks with affected NAT section */
524 static int nf_nat_proto_remove(struct nf_conn *i, void *data)
526 const struct nf_nat_proto_clean *clean = data;
527 struct nf_conn_nat *nat = nfct_nat(i);
532 if ((clean->l3proto && nf_ct_l3num(i) != clean->l3proto) ||
533 (clean->l4proto && nf_ct_protonum(i) != clean->l4proto))
536 return i->status & IPS_NAT_MASK ? 1 : 0;
539 static int nf_nat_proto_clean(struct nf_conn *ct, void *data)
541 struct nf_conn_nat *nat = nfct_nat(ct);
543 if (nf_nat_proto_remove(ct, data))
546 if (!nat || !nat->ct)
549 /* This netns is being destroyed, and conntrack has nat null binding.
550 * Remove it from bysource hash, as the table will be freed soon.
552 * Else, when the conntrack is destoyed, nf_nat_cleanup_conntrack()
553 * will delete entry from already-freed table.
555 if (!del_timer(&ct->timeout))
558 spin_lock_bh(&nf_nat_lock);
559 hlist_del_rcu(&nat->bysource);
560 ct->status &= ~IPS_NAT_DONE_MASK;
562 spin_unlock_bh(&nf_nat_lock);
564 add_timer(&ct->timeout);
566 /* don't delete conntrack. Although that would make things a lot
567 * simpler, we'd end up flushing all conntracks on nat rmmod.
572 static void nf_nat_l4proto_clean(u8 l3proto, u8 l4proto)
574 struct nf_nat_proto_clean clean = {
582 nf_ct_iterate_cleanup(net, nf_nat_proto_remove, &clean, 0, 0);
586 static void nf_nat_l3proto_clean(u8 l3proto)
588 struct nf_nat_proto_clean clean = {
596 nf_ct_iterate_cleanup(net, nf_nat_proto_remove, &clean, 0, 0);
600 /* Protocol registration. */
601 int nf_nat_l4proto_register(u8 l3proto, const struct nf_nat_l4proto *l4proto)
603 const struct nf_nat_l4proto **l4protos;
607 mutex_lock(&nf_nat_proto_mutex);
608 if (nf_nat_l4protos[l3proto] == NULL) {
609 l4protos = kmalloc(IPPROTO_MAX * sizeof(struct nf_nat_l4proto *),
611 if (l4protos == NULL) {
616 for (i = 0; i < IPPROTO_MAX; i++)
617 RCU_INIT_POINTER(l4protos[i], &nf_nat_l4proto_unknown);
619 /* Before making proto_array visible to lockless readers,
620 * we must make sure its content is committed to memory.
624 nf_nat_l4protos[l3proto] = l4protos;
627 if (rcu_dereference_protected(
628 nf_nat_l4protos[l3proto][l4proto->l4proto],
629 lockdep_is_held(&nf_nat_proto_mutex)
630 ) != &nf_nat_l4proto_unknown) {
634 RCU_INIT_POINTER(nf_nat_l4protos[l3proto][l4proto->l4proto], l4proto);
636 mutex_unlock(&nf_nat_proto_mutex);
639 EXPORT_SYMBOL_GPL(nf_nat_l4proto_register);
641 /* No one stores the protocol anywhere; simply delete it. */
642 void nf_nat_l4proto_unregister(u8 l3proto, const struct nf_nat_l4proto *l4proto)
644 mutex_lock(&nf_nat_proto_mutex);
645 RCU_INIT_POINTER(nf_nat_l4protos[l3proto][l4proto->l4proto],
646 &nf_nat_l4proto_unknown);
647 mutex_unlock(&nf_nat_proto_mutex);
650 nf_nat_l4proto_clean(l3proto, l4proto->l4proto);
652 EXPORT_SYMBOL_GPL(nf_nat_l4proto_unregister);
654 int nf_nat_l3proto_register(const struct nf_nat_l3proto *l3proto)
658 err = nf_ct_l3proto_try_module_get(l3proto->l3proto);
662 mutex_lock(&nf_nat_proto_mutex);
663 RCU_INIT_POINTER(nf_nat_l4protos[l3proto->l3proto][IPPROTO_TCP],
664 &nf_nat_l4proto_tcp);
665 RCU_INIT_POINTER(nf_nat_l4protos[l3proto->l3proto][IPPROTO_UDP],
666 &nf_nat_l4proto_udp);
667 mutex_unlock(&nf_nat_proto_mutex);
669 RCU_INIT_POINTER(nf_nat_l3protos[l3proto->l3proto], l3proto);
672 EXPORT_SYMBOL_GPL(nf_nat_l3proto_register);
674 void nf_nat_l3proto_unregister(const struct nf_nat_l3proto *l3proto)
676 mutex_lock(&nf_nat_proto_mutex);
677 RCU_INIT_POINTER(nf_nat_l3protos[l3proto->l3proto], NULL);
678 mutex_unlock(&nf_nat_proto_mutex);
681 nf_nat_l3proto_clean(l3proto->l3proto);
682 nf_ct_l3proto_module_put(l3proto->l3proto);
684 EXPORT_SYMBOL_GPL(nf_nat_l3proto_unregister);
686 /* No one using conntrack by the time this called. */
687 static void nf_nat_cleanup_conntrack(struct nf_conn *ct)
689 struct nf_conn_nat *nat = nf_ct_ext_find(ct, NF_CT_EXT_NAT);
691 if (nat == NULL || nat->ct == NULL)
694 NF_CT_ASSERT(nat->ct->status & IPS_SRC_NAT_DONE);
696 spin_lock_bh(&nf_nat_lock);
697 hlist_del_rcu(&nat->bysource);
698 spin_unlock_bh(&nf_nat_lock);
701 static void nf_nat_move_storage(void *new, void *old)
703 struct nf_conn_nat *new_nat = new;
704 struct nf_conn_nat *old_nat = old;
705 struct nf_conn *ct = old_nat->ct;
707 if (!ct || !(ct->status & IPS_SRC_NAT_DONE))
710 spin_lock_bh(&nf_nat_lock);
711 hlist_replace_rcu(&old_nat->bysource, &new_nat->bysource);
712 spin_unlock_bh(&nf_nat_lock);
715 static struct nf_ct_ext_type nat_extend __read_mostly = {
716 .len = sizeof(struct nf_conn_nat),
717 .align = __alignof__(struct nf_conn_nat),
718 .destroy = nf_nat_cleanup_conntrack,
719 .move = nf_nat_move_storage,
721 .flags = NF_CT_EXT_F_PREALLOC,
724 #if IS_ENABLED(CONFIG_NF_CT_NETLINK)
726 #include <linux/netfilter/nfnetlink.h>
727 #include <linux/netfilter/nfnetlink_conntrack.h>
729 static const struct nla_policy protonat_nla_policy[CTA_PROTONAT_MAX+1] = {
730 [CTA_PROTONAT_PORT_MIN] = { .type = NLA_U16 },
731 [CTA_PROTONAT_PORT_MAX] = { .type = NLA_U16 },
734 static int nfnetlink_parse_nat_proto(struct nlattr *attr,
735 const struct nf_conn *ct,
736 struct nf_nat_range *range)
738 struct nlattr *tb[CTA_PROTONAT_MAX+1];
739 const struct nf_nat_l4proto *l4proto;
742 err = nla_parse_nested(tb, CTA_PROTONAT_MAX, attr, protonat_nla_policy);
746 l4proto = __nf_nat_l4proto_find(nf_ct_l3num(ct), nf_ct_protonum(ct));
747 if (l4proto->nlattr_to_range)
748 err = l4proto->nlattr_to_range(tb, range);
753 static const struct nla_policy nat_nla_policy[CTA_NAT_MAX+1] = {
754 [CTA_NAT_V4_MINIP] = { .type = NLA_U32 },
755 [CTA_NAT_V4_MAXIP] = { .type = NLA_U32 },
756 [CTA_NAT_V6_MINIP] = { .len = sizeof(struct in6_addr) },
757 [CTA_NAT_V6_MAXIP] = { .len = sizeof(struct in6_addr) },
758 [CTA_NAT_PROTO] = { .type = NLA_NESTED },
762 nfnetlink_parse_nat(const struct nlattr *nat,
763 const struct nf_conn *ct, struct nf_nat_range *range,
764 const struct nf_nat_l3proto *l3proto)
766 struct nlattr *tb[CTA_NAT_MAX+1];
769 memset(range, 0, sizeof(*range));
771 err = nla_parse_nested(tb, CTA_NAT_MAX, nat, nat_nla_policy);
775 err = l3proto->nlattr_to_range(tb, range);
779 if (!tb[CTA_NAT_PROTO])
782 return nfnetlink_parse_nat_proto(tb[CTA_NAT_PROTO], ct, range);
785 /* This function is called under rcu_read_lock() */
787 nfnetlink_parse_nat_setup(struct nf_conn *ct,
788 enum nf_nat_manip_type manip,
789 const struct nlattr *attr)
791 struct nf_nat_range range;
792 const struct nf_nat_l3proto *l3proto;
795 /* Should not happen, restricted to creating new conntracks
798 if (WARN_ON_ONCE(nf_nat_initialized(ct, manip)))
801 /* Make sure that L3 NAT is there by when we call nf_nat_setup_info to
802 * attach the null binding, otherwise this may oops.
804 l3proto = __nf_nat_l3proto_find(nf_ct_l3num(ct));
808 /* No NAT information has been passed, allocate the null-binding */
810 return __nf_nat_alloc_null_binding(ct, manip);
812 err = nfnetlink_parse_nat(attr, ct, &range, l3proto);
816 return nf_nat_setup_info(ct, &range, manip);
820 nfnetlink_parse_nat_setup(struct nf_conn *ct,
821 enum nf_nat_manip_type manip,
822 const struct nlattr *attr)
828 static void __net_exit nf_nat_net_exit(struct net *net)
830 struct nf_nat_proto_clean clean = {};
832 nf_ct_iterate_cleanup(net, nf_nat_proto_clean, &clean, 0, 0);
835 static struct pernet_operations nf_nat_net_ops = {
836 .exit = nf_nat_net_exit,
839 static struct nf_ct_helper_expectfn follow_master_nat = {
840 .name = "nat-follow-master",
841 .expectfn = nf_nat_follow_master,
844 static int __init nf_nat_init(void)
848 /* Leave them the same for the moment. */
849 nf_nat_htable_size = nf_conntrack_htable_size;
851 nf_nat_bysource = nf_ct_alloc_hashtable(&nf_nat_htable_size, 0);
852 if (!nf_nat_bysource)
855 ret = nf_ct_extend_register(&nat_extend);
857 nf_ct_free_hashtable(nf_nat_bysource, nf_nat_htable_size);
858 printk(KERN_ERR "nf_nat_core: Unable to register extension\n");
862 ret = register_pernet_subsys(&nf_nat_net_ops);
866 nf_ct_helper_expectfn_register(&follow_master_nat);
868 /* Initialize fake conntrack so that NAT will skip it */
869 nf_ct_untracked_status_or(IPS_NAT_DONE_MASK);
871 BUG_ON(nfnetlink_parse_nat_setup_hook != NULL);
872 RCU_INIT_POINTER(nfnetlink_parse_nat_setup_hook,
873 nfnetlink_parse_nat_setup);
875 BUG_ON(nf_nat_decode_session_hook != NULL);
876 RCU_INIT_POINTER(nf_nat_decode_session_hook, __nf_nat_decode_session);
881 nf_ct_free_hashtable(nf_nat_bysource, nf_nat_htable_size);
882 nf_ct_extend_unregister(&nat_extend);
886 static void __exit nf_nat_cleanup(void)
890 unregister_pernet_subsys(&nf_nat_net_ops);
891 nf_ct_extend_unregister(&nat_extend);
892 nf_ct_helper_expectfn_unregister(&follow_master_nat);
893 RCU_INIT_POINTER(nfnetlink_parse_nat_setup_hook, NULL);
895 RCU_INIT_POINTER(nf_nat_decode_session_hook, NULL);
897 for (i = 0; i < NFPROTO_NUMPROTO; i++)
898 kfree(nf_nat_l4protos[i]);
900 nf_ct_free_hashtable(nf_nat_bysource, nf_nat_htable_size);
903 MODULE_LICENSE("GPL");
905 module_init(nf_nat_init);
906 module_exit(nf_nat_cleanup);