2 * Copyright (c) 2008 Paolo Pisati
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD: user/luigi/ipfw3-head/sys/netinet/ipfw/ip_fw_nat.c 200975 2009-12-25 01:15:39Z luigi $");
30 #include <sys/param.h>
31 #include <sys/systm.h>
32 #include <sys/eventhandler.h>
33 #include <sys/malloc.h>
34 #include <sys/kernel.h>
36 #include <sys/module.h>
37 #include <sys/rwlock.h>
39 #define IPFW_INTERNAL /* Access to protected data structures in ip_fw.h. */
41 #include <netinet/libalias/alias.h>
42 #include <netinet/libalias/alias_local.h>
45 #include <netinet/in.h>
46 #include <netinet/ip.h>
47 #include <netinet/ip_var.h>
48 #include <netinet/ip_fw.h>
49 #include <netinet/ipfw/ip_fw_private.h>
50 #include <netinet/tcp.h>
51 #include <netinet/udp.h>
53 #include <machine/in_cksum.h> /* XXX for in_cksum */
55 static VNET_DEFINE(eventhandler_tag, ifaddr_event_tag);
56 #define V_ifaddr_event_tag VNET(ifaddr_event_tag)
59 ifaddr_change(void *arg, struct ifnet *ifp)
63 struct ip_fw_chain *chain;
66 chain = &V_layer3_chain;
68 /* Check every nat entry... */
69 LIST_FOREACH(ptr, &chain->nat, _next) {
70 /* ...using nic 'ifp->if_xname' as dynamic alias address. */
71 if (strncmp(ptr->if_name, ifp->if_xname, IF_NAMESIZE) != 0)
74 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
75 if (ifa->ifa_addr == NULL)
77 if (ifa->ifa_addr->sa_family != AF_INET)
79 ptr->ip = ((struct sockaddr_in *)
80 (ifa->ifa_addr))->sin_addr;
81 LibAliasSetAddress(ptr->lib, ptr->ip);
89 * delete the pointers for nat entry ix, or all of them if ix < 0
92 flush_nat_ptrs(struct ip_fw_chain *chain, const int ix)
97 IPFW_WLOCK_ASSERT(chain);
98 for (i = 0; i < chain->n_rules; i++) {
99 cmd = (ipfw_insn_nat *)ACTION_PTR(chain->map[i]);
100 /* XXX skip log and the like ? */
101 if (cmd->o.opcode == O_NAT && cmd->nat != NULL &&
102 (ix < 0 || cmd->nat->id == ix))
108 del_redir_spool_cfg(struct cfg_nat *n, struct redir_chain *head)
110 struct cfg_redir *r, *tmp_r;
111 struct cfg_spool *s, *tmp_s;
114 LIST_FOREACH_SAFE(r, head, _next, tmp_r) {
115 num = 1; /* Number of alias_link to delete. */
122 /* Delete all libalias redirect entry. */
123 for (i = 0; i < num; i++)
124 LibAliasRedirectDelete(n->lib, r->alink[i]);
125 /* Del spool cfg if any. */
126 LIST_FOREACH_SAFE(s, &r->spool_chain, _next, tmp_s) {
127 LIST_REMOVE(s, _next);
130 free(r->alink, M_IPFW);
131 LIST_REMOVE(r, _next);
135 printf("unknown redirect mode: %u\n", r->mode);
136 /* XXX - panic?!?!? */
143 add_redir_spool_cfg(char *buf, struct cfg_nat *ptr)
145 struct cfg_redir *r, *ser_r;
146 struct cfg_spool *s, *ser_s;
149 for (cnt = 0, off = 0; cnt < ptr->redir_cnt; cnt++) {
150 ser_r = (struct cfg_redir *)&buf[off];
151 r = malloc(SOF_REDIR, M_IPFW, M_WAITOK | M_ZERO);
152 memcpy(r, ser_r, SOF_REDIR);
153 LIST_INIT(&r->spool_chain);
155 r->alink = malloc(sizeof(struct alias_link *) * r->pport_cnt,
156 M_IPFW, M_WAITOK | M_ZERO);
159 r->alink[0] = LibAliasRedirectAddr(ptr->lib, r->laddr,
163 for (i = 0 ; i < r->pport_cnt; i++) {
164 /* If remotePort is all ports, set it to 0. */
165 u_short remotePortCopy = r->rport + i;
166 if (r->rport_cnt == 1 && r->rport == 0)
168 r->alink[i] = LibAliasRedirectPort(ptr->lib,
169 r->laddr, htons(r->lport + i), r->raddr,
170 htons(remotePortCopy), r->paddr,
171 htons(r->pport + i), r->proto);
172 if (r->alink[i] == NULL) {
179 r->alink[0] = LibAliasRedirectProto(ptr->lib ,r->laddr,
180 r->raddr, r->paddr, r->proto);
183 printf("unknown redirect mode: %u\n", r->mode);
186 /* XXX perhaps return an error instead of panic ? */
187 if (r->alink[0] == NULL)
188 panic("LibAliasRedirect* returned NULL");
189 /* LSNAT handling. */
190 for (i = 0; i < r->spool_cnt; i++) {
191 ser_s = (struct cfg_spool *)&buf[off];
192 s = malloc(SOF_REDIR, M_IPFW, M_WAITOK | M_ZERO);
193 memcpy(s, ser_s, SOF_SPOOL);
194 LibAliasAddServer(ptr->lib, r->alink[0],
195 s->addr, htons(s->port));
197 /* Hook spool entry. */
198 LIST_INSERT_HEAD(&r->spool_chain, s, _next);
200 /* And finally hook this redir entry. */
201 LIST_INSERT_HEAD(&ptr->redir_chain, r, _next);
207 ipfw_nat(struct ip_fw_args *args, struct cfg_nat *t, struct mbuf *m)
211 /* XXX - libalias duct tape */
217 mcl = m_megapullup(m, m->m_pkthdr.len);
222 ip = mtod(mcl, struct ip *);
225 * XXX - Libalias checksum offload 'duct tape':
227 * locally generated packets have only pseudo-header checksum
228 * calculated and libalias will break it[1], so mark them for
229 * later fix. Moreover there are cases when libalias modifies
230 * tcp packet data[2], mark them for later fix too.
232 * [1] libalias was never meant to run in kernel, so it does
233 * not have any knowledge about checksum offloading, and
234 * expects a packet with a full internet checksum.
235 * Unfortunately, packets generated locally will have just the
236 * pseudo header calculated, and when libalias tries to adjust
237 * the checksum it will actually compute a wrong value.
239 * [2] when libalias modifies tcp's data content, full TCP
240 * checksum has to be recomputed: the problem is that
241 * libalias does not have any idea about checksum offloading.
242 * To work around this, we do not do checksumming in LibAlias,
243 * but only mark the packets in th_x2 field. If we receive a
244 * marked packet, we calculate correct checksum for it
245 * aware of offloading. Why such a terrible hack instead of
246 * recalculating checksum for each packet?
247 * Because the previous checksum was not checked!
248 * Recalculating checksums for EVERY packet will hide ALL
249 * transmission errors. Yes, marked packets still suffer from
250 * this problem. But, sigh, natd(8) has this problem, too.
252 * TODO: -make libalias mbuf aware (so
253 * it can handle delayed checksum and tso)
256 if (mcl->m_pkthdr.rcvif == NULL &&
257 mcl->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
260 c = mtod(mcl, char *);
261 if (args->oif == NULL)
262 retval = LibAliasIn(t->lib, c,
263 mcl->m_len + M_TRAILINGSPACE(mcl));
265 retval = LibAliasOut(t->lib, c,
266 mcl->m_len + M_TRAILINGSPACE(mcl));
267 if (retval == PKT_ALIAS_RESPOND) {
268 m->m_flags |= M_SKIP_FIREWALL;
269 retval = PKT_ALIAS_OK;
271 if (retval != PKT_ALIAS_OK &&
272 retval != PKT_ALIAS_FOUND_HEADER_FRAGMENT) {
273 /* XXX - should i add some logging? */
278 mcl->m_pkthdr.len = mcl->m_len = ntohs(ip->ip_len);
281 * XXX - libalias checksum offload
282 * 'duct tape' (see above)
285 if ((ip->ip_off & htons(IP_OFFMASK)) == 0 &&
286 ip->ip_p == IPPROTO_TCP) {
289 th = (struct tcphdr *)(ip + 1);
299 ip->ip_len = ntohs(ip->ip_len);
300 cksum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr,
301 htons(ip->ip_p + ip->ip_len - (ip->ip_hl << 2)));
305 th = (struct tcphdr *)(ip + 1);
307 * Maybe it was set in
312 mcl->m_pkthdr.csum_data =
313 offsetof(struct tcphdr, th_sum);
316 uh = (struct udphdr *)(ip + 1);
318 mcl->m_pkthdr.csum_data =
319 offsetof(struct udphdr, uh_sum);
322 /* No hw checksum offloading: do it ourselves */
323 if ((mcl->m_pkthdr.csum_flags & CSUM_DELAY_DATA) == 0) {
324 in_delayed_cksum(mcl);
325 mcl->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
327 ip->ip_len = htons(ip->ip_len);
333 static struct cfg_nat *
334 lookup_nat(struct nat_list *l, int nat_id)
338 LIST_FOREACH(res, l, _next) {
339 if (res->id == nat_id)
346 ipfw_nat_cfg(struct sockopt *sopt)
348 struct cfg_nat *ptr, *ser_n;
350 struct ip_fw_chain *chain = &V_layer3_chain;
352 buf = malloc(NAT_BUF_LEN, M_IPFW, M_WAITOK | M_ZERO);
353 sooptcopyin(sopt, buf, NAT_BUF_LEN, sizeof(struct cfg_nat));
354 ser_n = (struct cfg_nat *)buf;
356 /* check valid parameter ser_n->id > 0 ? */
358 * Find/create nat rule.
361 ptr = lookup_nat(&chain->nat, ser_n->id);
363 /* New rule: allocate and init new instance. */
364 ptr = malloc(sizeof(struct cfg_nat),
365 M_IPFW, M_NOWAIT | M_ZERO);
371 ptr->lib = LibAliasInit(NULL);
372 if (ptr->lib == NULL) {
378 LIST_INIT(&ptr->redir_chain);
380 /* Entry already present: temporarly unhook it. */
381 LIST_REMOVE(ptr, _next);
382 flush_nat_ptrs(chain, ser_n->id);
387 * Basic nat configuration.
391 * XXX - what if this rule doesn't nat any ip and just
393 * do we set aliasaddress to 0.0.0.0?
396 ptr->redir_cnt = ser_n->redir_cnt;
397 ptr->mode = ser_n->mode;
398 LibAliasSetMode(ptr->lib, ser_n->mode, ser_n->mode);
399 LibAliasSetAddress(ptr->lib, ptr->ip);
400 memcpy(ptr->if_name, ser_n->if_name, IF_NAMESIZE);
403 * Redir and LSNAT configuration.
405 /* Delete old cfgs. */
406 del_redir_spool_cfg(ptr, &ptr->redir_chain);
407 /* Add new entries. */
408 add_redir_spool_cfg(&buf[(sizeof(struct cfg_nat))], ptr);
411 LIST_INSERT_HEAD(&chain->nat, ptr, _next);
417 ipfw_nat_del(struct sockopt *sopt)
420 struct ip_fw_chain *chain = &V_layer3_chain;
423 sooptcopyin(sopt, &i, sizeof i, sizeof i);
426 ptr = lookup_nat(&chain->nat, i);
431 LIST_REMOVE(ptr, _next);
432 flush_nat_ptrs(chain, i);
434 del_redir_spool_cfg(ptr, &ptr->redir_chain);
435 LibAliasUninit(ptr->lib);
441 ipfw_nat_get_cfg(struct sockopt *sopt)
448 struct ip_fw_chain *chain;
451 chain = &V_layer3_chain;
453 off = sizeof(nat_cnt);
455 data = malloc(NAT_BUF_LEN, M_IPFW, M_WAITOK | M_ZERO);
457 /* Serialize all the data. */
458 LIST_FOREACH(n, &chain->nat, _next) {
460 if (off + SOF_NAT >= NAT_BUF_LEN)
462 bcopy(n, &data[off], SOF_NAT);
464 LIST_FOREACH(r, &n->redir_chain, _next) {
465 if (off + SOF_REDIR >= NAT_BUF_LEN)
467 bcopy(r, &data[off], SOF_REDIR);
469 LIST_FOREACH(s, &r->spool_chain, _next) {
470 if (off + SOF_SPOOL >= NAT_BUF_LEN)
472 bcopy(s, &data[off], SOF_SPOOL);
477 err = 0; /* all good */
481 bcopy(&nat_cnt, data, sizeof(nat_cnt));
482 sooptcopyout(sopt, data, NAT_BUF_LEN);
484 printf("serialized data buffer not big enough:"
485 "please increase NAT_BUF_LEN\n");
492 ipfw_nat_get_log(struct sockopt *sopt)
497 struct ip_fw_chain *chain;
499 chain = &V_layer3_chain;
502 /* one pass to count, one to copy the data */
504 LIST_FOREACH(ptr, &chain->nat, _next) {
505 if (ptr->lib->logDesc == NULL)
509 size = i * (LIBALIAS_BUF_SIZE + sizeof(int));
510 data = malloc(size, M_IPFW, M_NOWAIT | M_ZERO);
516 LIST_FOREACH(ptr, &chain->nat, _next) {
517 if (ptr->lib->logDesc == NULL)
519 bcopy(&ptr->id, &data[i], sizeof(int));
521 bcopy(ptr->lib->logDesc, &data[i], LIBALIAS_BUF_SIZE);
522 i += LIBALIAS_BUF_SIZE;
525 sooptcopyout(sopt, data, size);
534 IPFW_WLOCK(&V_layer3_chain);
535 /* init ipfw hooks */
536 ipfw_nat_ptr = ipfw_nat;
537 lookup_nat_ptr = lookup_nat;
538 ipfw_nat_cfg_ptr = ipfw_nat_cfg;
539 ipfw_nat_del_ptr = ipfw_nat_del;
540 ipfw_nat_get_cfg_ptr = ipfw_nat_get_cfg;
541 ipfw_nat_get_log_ptr = ipfw_nat_get_log;
542 IPFW_WUNLOCK(&V_layer3_chain);
543 V_ifaddr_event_tag = EVENTHANDLER_REGISTER(
544 ifaddr_event, ifaddr_change,
545 NULL, EVENTHANDLER_PRI_ANY);
549 ipfw_nat_destroy(void)
551 struct cfg_nat *ptr, *ptr_temp;
552 struct ip_fw_chain *chain;
554 chain = &V_layer3_chain;
556 LIST_FOREACH_SAFE(ptr, &chain->nat, _next, ptr_temp) {
557 LIST_REMOVE(ptr, _next);
558 del_redir_spool_cfg(ptr, &ptr->redir_chain);
559 LibAliasUninit(ptr->lib);
562 EVENTHANDLER_DEREGISTER(ifaddr_event, V_ifaddr_event_tag);
563 flush_nat_ptrs(chain, -1 /* flush all */);
564 /* deregister ipfw_nat */
566 lookup_nat_ptr = NULL;
567 ipfw_nat_cfg_ptr = NULL;
568 ipfw_nat_del_ptr = NULL;
569 ipfw_nat_get_cfg_ptr = NULL;
570 ipfw_nat_get_log_ptr = NULL;
575 ipfw_nat_modevent(module_t mod, int type, void *unused)
595 static moduledata_t ipfw_nat_mod = {
601 DECLARE_MODULE(ipfw_nat, ipfw_nat_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY);
602 MODULE_DEPEND(ipfw_nat, libalias, 1, 1, 1);
603 MODULE_DEPEND(ipfw_nat, ipfw, 2, 2, 2);
604 MODULE_VERSION(ipfw_nat, 1);