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/netisr.h>
55 #include <net/route.h>
58 #include <netinet/in.h>
59 #include <netinet/in_var.h>
60 #include <netinet/in_systm.h>
61 #include <netinet/ip.h>
62 #include <netinet/ip_var.h>
63 #include <netinet6/in6_var.h>
64 #include <netinet/ip6.h>
65 #include <netinet6/ip6_var.h>
66 #include <netinet6/scope6_var.h>
67 #include <netinet/icmp6.h>
68 #include <netinet6/nd6.h>
70 #include <netinet/in_pcb.h>
73 #include <netipsec/ipsec.h>
74 #include <netipsec/ipsec6.h>
75 #include <netipsec/key.h>
78 #include <netinet6/ip6protosw.h>
81 * Forward a packet. If some error occurs return the sender
82 * an icmp packet. Note we can't always generate a meaningful
83 * icmp message because icmp doesn't have a large enough repertoire
86 * If not forwarding, just drop the packet. This could be confusing
87 * if ipforwarding was zero but some routing protocol was advancing
88 * us as a gateway to somewhere. However, we must let the routing
89 * protocol deal with that.
93 ip6_forward(struct mbuf *m, int srcrt)
95 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
96 struct sockaddr_in6 *dst = NULL;
97 struct rtentry *rt = NULL;
98 struct route_in6 rin6;
99 int error, type = 0, code = 0;
100 struct mbuf *mcopy = NULL;
101 struct ifnet *origifp; /* maybe unnecessary */
102 u_int32_t inzone, outzone;
103 struct in6_addr src_in6, dst_in6, odst;
105 struct secpolicy *sp = NULL;
111 struct m_tag *fwd_tag;
112 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
116 * Check AH/ESP integrity.
119 * Don't increment ip6s_cantforward because this is the check
120 * before forwarding packet actually.
122 if (ipsec6_in_reject(m, NULL)) {
123 IPSEC6STAT_INC(ips_in_polvio);
130 * Do not forward packets to multicast destination (should be handled
132 * Do not forward packets with unspecified source. It was discussed
133 * in July 2000, on the ipngwg mailing list.
135 if ((m->m_flags & (M_BCAST|M_MCAST)) != 0 ||
136 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
137 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
138 IP6STAT_INC(ip6s_cantforward);
139 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
140 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
141 V_ip6_log_time = time_uptime;
144 "from %s to %s nxt %d received on %s\n",
145 ip6_sprintf(ip6bufs, &ip6->ip6_src),
146 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
148 if_name(m->m_pkthdr.rcvif));
157 if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
158 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
159 icmp6_error(m, ICMP6_TIME_EXCEEDED,
160 ICMP6_TIME_EXCEED_TRANSIT, 0);
163 ip6->ip6_hlim -= IPV6_HLIMDEC;
170 * Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
171 * size of IPv6 + ICMPv6 headers) bytes of the packet in case
172 * we need to generate an ICMP6 message to the src.
173 * Thanks to M_EXT, in most cases copy will not occur.
175 * It is important to save it before IPsec processing as IPsec
176 * processing may modify the mbuf.
178 mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));
181 /* get a security policy for this packet */
182 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND,
183 IP_FORWARDING, &error);
185 IPSEC6STAT_INC(ips_out_inval);
186 IP6STAT_INC(ip6s_cantforward);
189 /* XXX: what icmp ? */
201 switch (sp->policy) {
202 case IPSEC_POLICY_DISCARD:
204 * This packet is just discarded.
206 IPSEC6STAT_INC(ips_out_polvio);
207 IP6STAT_INC(ip6s_cantforward);
211 /* XXX: what icmp ? */
219 case IPSEC_POLICY_BYPASS:
220 case IPSEC_POLICY_NONE:
221 /* no need to do IPsec. */
225 case IPSEC_POLICY_IPSEC:
226 if (sp->req == NULL) {
227 /* XXX should be panic ? */
228 printf("ip6_forward: No IPsec request specified.\n");
229 IP6STAT_INC(ip6s_cantforward);
233 /* XXX: what icmp ? */
244 case IPSEC_POLICY_ENTRUST:
246 /* should be panic ?? */
247 printf("ip6_forward: Invalid policy found. %d\n", sp->policy);
253 struct ipsecrequest *isr = NULL;
256 * when the kernel forwards a packet, it is not proper to apply
257 * IPsec transport mode to the packet is not proper. this check
259 * at present, if there is even a transport mode SA request in the
260 * security policy, the kernel does not apply IPsec to the packet.
261 * this check is not enough because the following case is valid.
262 * ipsec esp/tunnel/xxx-xxx/require esp/transport//require;
264 for (isr = sp->req; isr; isr = isr->next) {
265 if (isr->saidx.mode == IPSEC_MODE_ANY)
267 if (isr->saidx.mode == IPSEC_MODE_TUNNEL)
272 * if there's no need for tunnel mode IPsec, skip.
279 * All the extension headers will become inaccessible
280 * (since they can be encrypted).
281 * Don't panic, we need no more updates to extension headers
282 * on inner IPv6 packet (since they are now encapsulated).
284 * IPv6 [ESP|AH] IPv6 [extension headers] payload
288 * If we need to encapsulate the packet, do it here
289 * ipsec6_proces_packet will send the packet using ip6_output
291 error = ipsec6_process_packet(m, sp->req);
295 if (error == EJUSTRETURN) {
297 * We had a SP with a level of 'use' and no SA. We
298 * will just continue to process the packet without
306 /* mbuf is already reclaimed in ipsec6_process_packet. */
315 printf("ip6_output (ipsec): error code %d\n", error);
318 /* don't show these error codes to the user */
321 IP6STAT_INC(ip6s_cantforward);
324 /* XXX: what icmp ? */
332 * In the FAST IPSec case we have already
333 * re-injected the packet and it has been freed
334 * by the ipsec_done() function. So, just clean
335 * up after ourselves.
344 bzero(&rin6, sizeof(struct route_in6));
345 dst = (struct sockaddr_in6 *)&rin6.ro_dst;
346 dst->sin6_len = sizeof(struct sockaddr_in6);
347 dst->sin6_family = AF_INET6;
348 dst->sin6_addr = ip6->ip6_dst;
350 rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
351 if (rin6.ro_rt != NULL)
352 RT_UNLOCK(rin6.ro_rt);
354 IP6STAT_INC(ip6s_noroute);
355 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
357 icmp6_error(mcopy, ICMP6_DST_UNREACH,
358 ICMP6_DST_UNREACH_NOROUTE, 0);
365 * Source scope check: if a packet can't be delivered to its
366 * destination for the reason that the destination is beyond the scope
367 * of the source address, discard the packet and return an icmp6
368 * destination unreachable error with Code 2 (beyond scope of source
369 * address). We use a local copy of ip6_src, since in6_setscope()
370 * will possibly modify its first argument.
371 * [draft-ietf-ipngwg-icmp-v3-04.txt, Section 3.1]
373 src_in6 = ip6->ip6_src;
374 if (in6_setscope(&src_in6, rt->rt_ifp, &outzone)) {
375 /* XXX: this should not happen */
376 IP6STAT_INC(ip6s_cantforward);
377 IP6STAT_INC(ip6s_badscope);
380 if (in6_setscope(&src_in6, m->m_pkthdr.rcvif, &inzone)) {
381 IP6STAT_INC(ip6s_cantforward);
382 IP6STAT_INC(ip6s_badscope);
385 if (inzone != outzone
390 IP6STAT_INC(ip6s_cantforward);
391 IP6STAT_INC(ip6s_badscope);
392 in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
394 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
395 V_ip6_log_time = time_uptime;
398 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
399 ip6_sprintf(ip6bufs, &ip6->ip6_src),
400 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
402 if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
405 icmp6_error(mcopy, ICMP6_DST_UNREACH,
406 ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
411 * Destination scope check: if a packet is going to break the scope
412 * zone of packet's destination address, discard it. This case should
413 * usually be prevented by appropriately-configured routing table, but
414 * we need an explicit check because we may mistakenly forward the
415 * packet to a different zone by (e.g.) a default route.
417 dst_in6 = ip6->ip6_dst;
418 if (in6_setscope(&dst_in6, m->m_pkthdr.rcvif, &inzone) != 0 ||
419 in6_setscope(&dst_in6, rt->rt_ifp, &outzone) != 0 ||
421 IP6STAT_INC(ip6s_cantforward);
422 IP6STAT_INC(ip6s_badscope);
426 if (rt->rt_flags & RTF_GATEWAY)
427 dst = (struct sockaddr_in6 *)rt->rt_gateway;
430 * If we are to forward the packet using the same interface
431 * as one we got the packet from, perhaps we should send a redirect
432 * to sender to shortcut a hop.
433 * Only send redirect if source is sending directly to us,
434 * and if packet was not source routed (or has any options).
435 * Also, don't send redirect if forwarding using a route
436 * modified by a redirect.
438 if (V_ip6_sendredirects && rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
442 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0) {
443 if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) != 0) {
445 * If the incoming interface is equal to the outgoing
446 * one, and the link attached to the interface is
447 * point-to-point, then it will be highly probable
448 * that a routing loop occurs. Thus, we immediately
449 * drop the packet and send an ICMPv6 error message.
451 * type/code is based on suggestion by Rich Draves.
452 * not sure if it is the best pick.
454 icmp6_error(mcopy, ICMP6_DST_UNREACH,
455 ICMP6_DST_UNREACH_ADDR, 0);
462 * Fake scoped addresses. Note that even link-local source or
463 * destinaion can appear, if the originating node just sends the
464 * packet to us (without address resolution for the destination).
465 * Since both icmp6_error and icmp6_redirect_output fill the embedded
466 * link identifiers, we can do this stuff after making a copy for
467 * returning an error.
469 if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) {
471 * See corresponding comments in ip6_output.
472 * XXX: but is it possible that ip6_forward() sends a packet
473 * to a loopback interface? I don't think so, and thus
474 * I bark here. (jinmei@kame.net)
475 * XXX: it is common to route invalid packets to loopback.
476 * also, the codepath will be visited on use of ::1 in
482 if ((rt->rt_flags & (RTF_BLACKHOLE|RTF_REJECT)) == 0)
485 printf("ip6_forward: outgoing interface is loopback. "
486 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
487 ip6_sprintf(ip6bufs, &ip6->ip6_src),
488 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
489 ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
490 if_name(rt->rt_ifp));
493 /* we can just use rcvif in forwarding. */
494 origifp = m->m_pkthdr.rcvif;
497 origifp = rt->rt_ifp;
499 * clear embedded scope identifiers if necessary.
500 * in6_clearscope will touch the addresses only when necessary.
502 in6_clearscope(&ip6->ip6_src);
503 in6_clearscope(&ip6->ip6_dst);
505 /* Jump over all PFIL processing if hooks are not active. */
506 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
510 /* Run through list of hooks for output packets. */
511 error = pfil_run_hooks(&V_inet6_pfil_hook, &m, rt->rt_ifp, PFIL_OUT, NULL);
512 if (error != 0 || m == NULL)
513 goto freecopy; /* consumed by filter */
514 ip6 = mtod(m, struct ip6_hdr *);
516 /* See if destination IP address was changed by packet filter. */
517 if (!IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst)) {
518 m->m_flags |= M_SKIP_FIREWALL;
519 /* If destination is now ourself drop to ip6_input(). */
520 if (in6_localip(&ip6->ip6_dst))
521 m->m_flags |= M_FASTFWD_OURS;
523 goto again; /* Redo the routing table lookup. */
526 /* See if local, if yes, send it to netisr. */
527 if (m->m_flags & M_FASTFWD_OURS) {
528 if (m->m_pkthdr.rcvif == NULL)
529 m->m_pkthdr.rcvif = V_loif;
530 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) {
531 m->m_pkthdr.csum_flags |=
532 CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR;
533 m->m_pkthdr.csum_data = 0xffff;
536 if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6)
537 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
539 error = netisr_queue(NETISR_IPV6, m);
542 /* Or forward to some other address? */
543 if ((m->m_flags & M_IP6_NEXTHOP) &&
544 (fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) {
545 dst = (struct sockaddr_in6 *)&rin6.ro_dst;
546 bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in6));
547 m->m_flags |= M_SKIP_FIREWALL;
548 m->m_flags &= ~M_IP6_NEXTHOP;
549 m_tag_delete(m, fwd_tag);
554 /* See if the size was changed by the packet filter. */
555 if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
556 in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
560 struct secpolicy *sp;
565 mtu = IN6_LINKMTU(rt->rt_ifp);
568 * When we do IPsec tunnel ingress, we need to play
569 * with the link value (decrement IPsec header size
570 * from mtu value). The code is much simpler than v4
571 * case, as we have the outgoing interface for
572 * encapsulated packet as "rt->rt_ifp".
574 sp = ipsec_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND,
575 IP_FORWARDING, &ipsecerror);
577 ipsechdrsiz = ipsec_hdrsiz(mcopy,
578 IPSEC_DIR_OUTBOUND, NULL);
579 if (ipsechdrsiz < mtu)
584 * if mtu becomes less than minimum MTU,
585 * tell minimum MTU (and I'll need to fragment it).
590 icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
595 error = nd6_output(rt->rt_ifp, origifp, m, dst, rt);
597 in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
598 IP6STAT_INC(ip6s_cantforward);
600 IP6STAT_INC(ip6s_forward);
601 in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
603 IP6STAT_INC(ip6s_redirectsent);
614 if (type == ND_REDIRECT) {
615 icmp6_redirect_output(mcopy, rt);
621 /* xxx MTU is constant in PPP? */
625 /* Tell source to slow down like source quench in IP? */
628 case ENETUNREACH: /* shouldn't happen, checked above */
633 type = ICMP6_DST_UNREACH;
634 code = ICMP6_DST_UNREACH_ADDR;
637 icmp6_error(mcopy, type, code, 0);