2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
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.
13 * 3. Neither the name of the project nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * $KAME: ip6_forward.c,v 1.69 2001/05/17 03:48:30 itojun Exp $
32 #include <sys/cdefs.h>
33 __FBSDID("$FreeBSD$");
36 #include "opt_inet6.h"
38 #include "opt_ipsec.h"
39 #include "opt_ipstealth.h"
41 #include <sys/param.h>
42 #include <sys/systm.h>
43 #include <sys/malloc.h>
45 #include <sys/domain.h>
46 #include <sys/protosw.h>
47 #include <sys/socket.h>
48 #include <sys/errno.h>
50 #include <sys/kernel.h>
51 #include <sys/syslog.h>
54 #include <net/if_var.h>
55 #include <net/netisr.h>
56 #include <net/route.h>
59 #include <netinet/in.h>
60 #include <netinet/in_var.h>
61 #include <netinet/in_systm.h>
62 #include <netinet/ip.h>
63 #include <netinet/ip_var.h>
64 #include <netinet6/in6_var.h>
65 #include <netinet/ip6.h>
66 #include <netinet6/ip6_var.h>
67 #include <netinet6/scope6_var.h>
68 #include <netinet/icmp6.h>
69 #include <netinet6/nd6.h>
71 #include <netinet/in_pcb.h>
74 #include <netipsec/ipsec.h>
75 #include <netipsec/ipsec6.h>
76 #include <netipsec/key.h>
79 #include <netinet6/ip6protosw.h>
82 * Forward a packet. If some error occurs return the sender
83 * an icmp packet. Note we can't always generate a meaningful
84 * icmp message because icmp doesn't have a large enough repertoire
87 * If not forwarding, just drop the packet. This could be confusing
88 * if ipforwarding was zero but some routing protocol was advancing
89 * us as a gateway to somewhere. However, we must let the routing
90 * protocol deal with that.
94 ip6_forward(struct mbuf *m, int srcrt)
96 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
97 struct sockaddr_in6 *dst = NULL;
98 struct rtentry *rt = NULL;
99 struct route_in6 rin6;
100 int error, type = 0, code = 0;
101 struct mbuf *mcopy = NULL;
102 struct ifnet *origifp; /* maybe unnecessary */
103 u_int32_t inzone, outzone;
104 struct in6_addr src_in6, dst_in6, odst;
106 struct secpolicy *sp = NULL;
112 struct m_tag *fwd_tag;
113 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
117 * Check AH/ESP integrity.
120 * Don't increment ip6s_cantforward because this is the check
121 * before forwarding packet actually.
123 if (ipsec6_in_reject(m, NULL)) {
124 IPSEC6STAT_INC(ips_in_polvio);
131 * Do not forward packets to multicast destination (should be handled
133 * Do not forward packets with unspecified source. It was discussed
134 * in July 2000, on the ipngwg mailing list.
136 if ((m->m_flags & (M_BCAST|M_MCAST)) != 0 ||
137 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
138 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
139 IP6STAT_INC(ip6s_cantforward);
140 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
141 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
142 V_ip6_log_time = time_uptime;
145 "from %s to %s nxt %d received on %s\n",
146 ip6_sprintf(ip6bufs, &ip6->ip6_src),
147 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
149 if_name(m->m_pkthdr.rcvif));
158 if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
159 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
160 icmp6_error(m, ICMP6_TIME_EXCEEDED,
161 ICMP6_TIME_EXCEED_TRANSIT, 0);
164 ip6->ip6_hlim -= IPV6_HLIMDEC;
171 * Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
172 * size of IPv6 + ICMPv6 headers) bytes of the packet in case
173 * we need to generate an ICMP6 message to the src.
174 * Thanks to M_EXT, in most cases copy will not occur.
176 * It is important to save it before IPsec processing as IPsec
177 * processing may modify the mbuf.
179 mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));
182 /* get a security policy for this packet */
183 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND,
184 IP_FORWARDING, &error);
186 IPSEC6STAT_INC(ips_out_inval);
187 IP6STAT_INC(ip6s_cantforward);
190 /* XXX: what icmp ? */
202 switch (sp->policy) {
203 case IPSEC_POLICY_DISCARD:
205 * This packet is just discarded.
207 IPSEC6STAT_INC(ips_out_polvio);
208 IP6STAT_INC(ip6s_cantforward);
212 /* XXX: what icmp ? */
220 case IPSEC_POLICY_BYPASS:
221 case IPSEC_POLICY_NONE:
222 /* no need to do IPsec. */
226 case IPSEC_POLICY_IPSEC:
227 if (sp->req == NULL) {
228 /* XXX should be panic ? */
229 printf("ip6_forward: No IPsec request specified.\n");
230 IP6STAT_INC(ip6s_cantforward);
234 /* XXX: what icmp ? */
245 case IPSEC_POLICY_ENTRUST:
247 /* should be panic ?? */
248 printf("ip6_forward: Invalid policy found. %d\n", sp->policy);
254 struct ipsecrequest *isr = NULL;
255 struct ipsec_output_state state;
258 * when the kernel forwards a packet, it is not proper to apply
259 * IPsec transport mode to the packet is not proper. this check
261 * at present, if there is even a transport mode SA request in the
262 * security policy, the kernel does not apply IPsec to the packet.
263 * this check is not enough because the following case is valid.
264 * ipsec esp/tunnel/xxx-xxx/require esp/transport//require;
266 for (isr = sp->req; isr; isr = isr->next) {
267 if (isr->saidx.mode == IPSEC_MODE_ANY)
269 if (isr->saidx.mode == IPSEC_MODE_TUNNEL)
274 * if there's no need for tunnel mode IPsec, skip.
281 * All the extension headers will become inaccessible
282 * (since they can be encrypted).
283 * Don't panic, we need no more updates to extension headers
284 * on inner IPv6 packet (since they are now encapsulated).
286 * IPv6 [ESP|AH] IPv6 [extension headers] payload
288 bzero(&state, sizeof(state));
290 state.ro = NULL; /* update at ipsec6_output_tunnel() */
291 state.dst = NULL; /* update at ipsec6_output_tunnel() */
293 error = ipsec6_output_tunnel(&state, sp, 0);
299 /* mbuf is already reclaimed in ipsec6_output_tunnel. */
308 printf("ip6_output (ipsec): error code %d\n", error);
311 /* don't show these error codes to the user */
314 IP6STAT_INC(ip6s_cantforward);
317 /* XXX: what icmp ? */
326 * In the FAST IPSec case we have already
327 * re-injected the packet and it has been freed
328 * by the ipsec_done() function. So, just clean
329 * up after ourselves.
335 if ((m != NULL) && (ip6 != mtod(m, struct ip6_hdr *)) ){
337 * now tunnel mode headers are added. we are originating
338 * packet instead of forwarding the packet.
340 ip6_output(m, NULL, NULL, IPV6_FORWARDING/*XXX*/, NULL, NULL,
346 dst = (struct sockaddr_in6 *)state.dst;
347 rt = state.ro ? state.ro->ro_rt : NULL;
348 if (dst != NULL && rt != NULL)
356 bzero(&rin6, sizeof(struct route_in6));
357 dst = (struct sockaddr_in6 *)&rin6.ro_dst;
358 dst->sin6_len = sizeof(struct sockaddr_in6);
359 dst->sin6_family = AF_INET6;
360 dst->sin6_addr = ip6->ip6_dst;
362 rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
363 if (rin6.ro_rt != NULL)
364 RT_UNLOCK(rin6.ro_rt);
366 IP6STAT_INC(ip6s_noroute);
367 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
369 icmp6_error(mcopy, ICMP6_DST_UNREACH,
370 ICMP6_DST_UNREACH_NOROUTE, 0);
380 * Source scope check: if a packet can't be delivered to its
381 * destination for the reason that the destination is beyond the scope
382 * of the source address, discard the packet and return an icmp6
383 * destination unreachable error with Code 2 (beyond scope of source
384 * address). We use a local copy of ip6_src, since in6_setscope()
385 * will possibly modify its first argument.
386 * [draft-ietf-ipngwg-icmp-v3-04.txt, Section 3.1]
388 src_in6 = ip6->ip6_src;
389 if (in6_setscope(&src_in6, rt->rt_ifp, &outzone)) {
390 /* XXX: this should not happen */
391 IP6STAT_INC(ip6s_cantforward);
392 IP6STAT_INC(ip6s_badscope);
395 if (in6_setscope(&src_in6, m->m_pkthdr.rcvif, &inzone)) {
396 IP6STAT_INC(ip6s_cantforward);
397 IP6STAT_INC(ip6s_badscope);
400 if (inzone != outzone
405 IP6STAT_INC(ip6s_cantforward);
406 IP6STAT_INC(ip6s_badscope);
407 in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
409 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
410 V_ip6_log_time = time_uptime;
413 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
414 ip6_sprintf(ip6bufs, &ip6->ip6_src),
415 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
417 if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
420 icmp6_error(mcopy, ICMP6_DST_UNREACH,
421 ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
426 * Destination scope check: if a packet is going to break the scope
427 * zone of packet's destination address, discard it. This case should
428 * usually be prevented by appropriately-configured routing table, but
429 * we need an explicit check because we may mistakenly forward the
430 * packet to a different zone by (e.g.) a default route.
432 dst_in6 = ip6->ip6_dst;
433 if (in6_setscope(&dst_in6, m->m_pkthdr.rcvif, &inzone) != 0 ||
434 in6_setscope(&dst_in6, rt->rt_ifp, &outzone) != 0 ||
436 IP6STAT_INC(ip6s_cantforward);
437 IP6STAT_INC(ip6s_badscope);
441 if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
442 in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
446 struct secpolicy *sp;
451 mtu = IN6_LINKMTU(rt->rt_ifp);
454 * When we do IPsec tunnel ingress, we need to play
455 * with the link value (decrement IPsec header size
456 * from mtu value). The code is much simpler than v4
457 * case, as we have the outgoing interface for
458 * encapsulated packet as "rt->rt_ifp".
460 sp = ipsec_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND,
461 IP_FORWARDING, &ipsecerror);
463 ipsechdrsiz = ipsec_hdrsiz(mcopy,
464 IPSEC_DIR_OUTBOUND, NULL);
465 if (ipsechdrsiz < mtu)
470 * if mtu becomes less than minimum MTU,
471 * tell minimum MTU (and I'll need to fragment it).
476 icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
481 if (rt->rt_flags & RTF_GATEWAY)
482 dst = (struct sockaddr_in6 *)rt->rt_gateway;
485 * If we are to forward the packet using the same interface
486 * as one we got the packet from, perhaps we should send a redirect
487 * to sender to shortcut a hop.
488 * Only send redirect if source is sending directly to us,
489 * and if packet was not source routed (or has any options).
490 * Also, don't send redirect if forwarding using a route
491 * modified by a redirect.
493 if (V_ip6_sendredirects && rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
497 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0) {
498 if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) != 0) {
500 * If the incoming interface is equal to the outgoing
501 * one, and the link attached to the interface is
502 * point-to-point, then it will be highly probable
503 * that a routing loop occurs. Thus, we immediately
504 * drop the packet and send an ICMPv6 error message.
506 * type/code is based on suggestion by Rich Draves.
507 * not sure if it is the best pick.
509 icmp6_error(mcopy, ICMP6_DST_UNREACH,
510 ICMP6_DST_UNREACH_ADDR, 0);
517 * Fake scoped addresses. Note that even link-local source or
518 * destinaion can appear, if the originating node just sends the
519 * packet to us (without address resolution for the destination).
520 * Since both icmp6_error and icmp6_redirect_output fill the embedded
521 * link identifiers, we can do this stuff after making a copy for
522 * returning an error.
524 if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) {
526 * See corresponding comments in ip6_output.
527 * XXX: but is it possible that ip6_forward() sends a packet
528 * to a loopback interface? I don't think so, and thus
529 * I bark here. (jinmei@kame.net)
530 * XXX: it is common to route invalid packets to loopback.
531 * also, the codepath will be visited on use of ::1 in
537 if ((rt->rt_flags & (RTF_BLACKHOLE|RTF_REJECT)) == 0)
540 printf("ip6_forward: outgoing interface is loopback. "
541 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
542 ip6_sprintf(ip6bufs, &ip6->ip6_src),
543 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
544 ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
545 if_name(rt->rt_ifp));
548 /* we can just use rcvif in forwarding. */
549 origifp = m->m_pkthdr.rcvif;
552 origifp = rt->rt_ifp;
554 * clear embedded scope identifiers if necessary.
555 * in6_clearscope will touch the addresses only when necessary.
557 in6_clearscope(&ip6->ip6_src);
558 in6_clearscope(&ip6->ip6_dst);
560 /* Jump over all PFIL processing if hooks are not active. */
561 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
565 /* Run through list of hooks for output packets. */
566 error = pfil_run_hooks(&V_inet6_pfil_hook, &m, rt->rt_ifp, PFIL_OUT, NULL);
567 if (error != 0 || m == NULL)
568 goto freecopy; /* consumed by filter */
569 ip6 = mtod(m, struct ip6_hdr *);
571 /* See if destination IP address was changed by packet filter. */
572 if (!IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst)) {
573 m->m_flags |= M_SKIP_FIREWALL;
574 /* If destination is now ourself drop to ip6_input(). */
575 if (in6_localip(&ip6->ip6_dst)) {
576 m->m_flags |= M_FASTFWD_OURS;
577 if (m->m_pkthdr.rcvif == NULL)
578 m->m_pkthdr.rcvif = V_loif;
579 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) {
580 m->m_pkthdr.csum_flags |=
581 CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR;
582 m->m_pkthdr.csum_data = 0xffff;
585 if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6)
586 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
588 error = netisr_queue(NETISR_IPV6, m);
591 goto again; /* Redo the routing table lookup. */
594 /* See if local, if yes, send it to netisr. */
595 if (m->m_flags & M_FASTFWD_OURS) {
596 if (m->m_pkthdr.rcvif == NULL)
597 m->m_pkthdr.rcvif = V_loif;
598 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) {
599 m->m_pkthdr.csum_flags |=
600 CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR;
601 m->m_pkthdr.csum_data = 0xffff;
604 if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6)
605 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
607 error = netisr_queue(NETISR_IPV6, m);
610 /* Or forward to some other address? */
611 if ((m->m_flags & M_IP6_NEXTHOP) &&
612 (fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) {
613 dst = (struct sockaddr_in6 *)&rin6.ro_dst;
614 bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in6));
615 m->m_flags |= M_SKIP_FIREWALL;
616 m->m_flags &= ~M_IP6_NEXTHOP;
617 m_tag_delete(m, fwd_tag);
622 error = nd6_output(rt->rt_ifp, origifp, m, dst, rt);
624 in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
625 IP6STAT_INC(ip6s_cantforward);
627 IP6STAT_INC(ip6s_forward);
628 in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
630 IP6STAT_INC(ip6s_redirectsent);
641 if (type == ND_REDIRECT) {
642 icmp6_redirect_output(mcopy, rt);
648 /* xxx MTU is constant in PPP? */
652 /* Tell source to slow down like source quench in IP? */
655 case ENETUNREACH: /* shouldn't happen, checked above */
660 type = ICMP6_DST_UNREACH;
661 code = ICMP6_DST_UNREACH_ADDR;
664 icmp6_error(mcopy, type, code, 0);