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>
80 * Forward a packet. If some error occurs return the sender
81 * an icmp packet. Note we can't always generate a meaningful
82 * icmp message because icmp doesn't have a large enough repertoire
85 * If not forwarding, just drop the packet. This could be confusing
86 * if ipforwarding was zero but some routing protocol was advancing
87 * us as a gateway to somewhere. However, we must let the routing
88 * protocol deal with that.
92 ip6_forward(struct mbuf *m, int srcrt)
94 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
95 struct sockaddr_in6 *dst = NULL;
96 struct rtentry *rt = NULL;
97 struct route_in6 rin6;
98 int error, type = 0, code = 0;
99 struct mbuf *mcopy = NULL;
100 struct ifnet *origifp; /* maybe unnecessary */
101 u_int32_t inzone, outzone;
102 struct in6_addr src_in6, dst_in6, odst;
104 struct secpolicy *sp = NULL;
109 struct m_tag *fwd_tag;
110 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
114 * Check AH/ESP integrity.
117 * Don't increment ip6s_cantforward because this is the check
118 * before forwarding packet actually.
120 if (ipsec6_in_reject(m, NULL)) {
121 IPSEC6STAT_INC(ips_in_polvio);
128 * Do not forward packets to multicast destination (should be handled
130 * Do not forward packets with unspecified source. It was discussed
131 * in July 2000, on the ipngwg mailing list.
133 if ((m->m_flags & (M_BCAST|M_MCAST)) != 0 ||
134 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
135 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
136 IP6STAT_INC(ip6s_cantforward);
137 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
138 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
139 V_ip6_log_time = time_uptime;
142 "from %s to %s nxt %d received on %s\n",
143 ip6_sprintf(ip6bufs, &ip6->ip6_src),
144 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
146 if_name(m->m_pkthdr.rcvif));
155 if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
156 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
157 icmp6_error(m, ICMP6_TIME_EXCEEDED,
158 ICMP6_TIME_EXCEED_TRANSIT, 0);
161 ip6->ip6_hlim -= IPV6_HLIMDEC;
168 * Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
169 * size of IPv6 + ICMPv6 headers) bytes of the packet in case
170 * we need to generate an ICMP6 message to the src.
171 * Thanks to M_EXT, in most cases copy will not occur.
173 * It is important to save it before IPsec processing as IPsec
174 * processing may modify the mbuf.
176 mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));
179 /* get a security policy for this packet */
180 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND,
181 IP_FORWARDING, &error);
183 IPSEC6STAT_INC(ips_out_inval);
184 IP6STAT_INC(ip6s_cantforward);
187 /* XXX: what icmp ? */
199 switch (sp->policy) {
200 case IPSEC_POLICY_DISCARD:
202 * This packet is just discarded.
204 IPSEC6STAT_INC(ips_out_polvio);
205 IP6STAT_INC(ip6s_cantforward);
209 /* XXX: what icmp ? */
217 case IPSEC_POLICY_BYPASS:
218 case IPSEC_POLICY_NONE:
219 /* no need to do IPsec. */
223 case IPSEC_POLICY_IPSEC:
224 if (sp->req == NULL) {
225 /* XXX should be panic ? */
226 printf("ip6_forward: No IPsec request specified.\n");
227 IP6STAT_INC(ip6s_cantforward);
231 /* XXX: what icmp ? */
242 case IPSEC_POLICY_ENTRUST:
244 /* should be panic ?? */
245 printf("ip6_forward: Invalid policy found. %d\n", sp->policy);
251 struct ipsecrequest *isr = NULL;
254 * when the kernel forwards a packet, it is not proper to apply
255 * IPsec transport mode to the packet is not proper. this check
257 * at present, if there is even a transport mode SA request in the
258 * security policy, the kernel does not apply IPsec to the packet.
259 * this check is not enough because the following case is valid.
260 * ipsec esp/tunnel/xxx-xxx/require esp/transport//require;
262 for (isr = sp->req; isr; isr = isr->next) {
263 if (isr->saidx.mode == IPSEC_MODE_ANY)
265 if (isr->saidx.mode == IPSEC_MODE_TUNNEL)
270 * if there's no need for tunnel mode IPsec, skip.
277 * All the extension headers will become inaccessible
278 * (since they can be encrypted).
279 * Don't panic, we need no more updates to extension headers
280 * on inner IPv6 packet (since they are now encapsulated).
282 * IPv6 [ESP|AH] IPv6 [extension headers] payload
286 * If we need to encapsulate the packet, do it here
287 * ipsec6_proces_packet will send the packet using ip6_output
289 error = ipsec6_process_packet(m, sp->req);
293 if (error == EJUSTRETURN) {
295 * We had a SP with a level of 'use' and no SA. We
296 * will just continue to process the packet without
304 /* mbuf is already reclaimed in ipsec6_process_packet. */
313 printf("ip6_output (ipsec): error code %d\n", error);
316 /* don't show these error codes to the user */
319 IP6STAT_INC(ip6s_cantforward);
322 /* XXX: what icmp ? */
330 * In the FAST IPSec case we have already
331 * re-injected the packet and it has been freed
332 * by the ipsec_done() function. So, just clean
333 * up after ourselves.
342 bzero(&rin6, sizeof(struct route_in6));
343 dst = (struct sockaddr_in6 *)&rin6.ro_dst;
344 dst->sin6_len = sizeof(struct sockaddr_in6);
345 dst->sin6_family = AF_INET6;
346 dst->sin6_addr = ip6->ip6_dst;
348 rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
349 if (rin6.ro_rt != NULL)
350 RT_UNLOCK(rin6.ro_rt);
352 IP6STAT_INC(ip6s_noroute);
353 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
355 icmp6_error(mcopy, ICMP6_DST_UNREACH,
356 ICMP6_DST_UNREACH_NOROUTE, 0);
363 * Source scope check: if a packet can't be delivered to its
364 * destination for the reason that the destination is beyond the scope
365 * of the source address, discard the packet and return an icmp6
366 * destination unreachable error with Code 2 (beyond scope of source
367 * address). We use a local copy of ip6_src, since in6_setscope()
368 * will possibly modify its first argument.
369 * [draft-ietf-ipngwg-icmp-v3-04.txt, Section 3.1]
371 src_in6 = ip6->ip6_src;
372 if (in6_setscope(&src_in6, rt->rt_ifp, &outzone)) {
373 /* XXX: this should not happen */
374 IP6STAT_INC(ip6s_cantforward);
375 IP6STAT_INC(ip6s_badscope);
378 if (in6_setscope(&src_in6, m->m_pkthdr.rcvif, &inzone)) {
379 IP6STAT_INC(ip6s_cantforward);
380 IP6STAT_INC(ip6s_badscope);
383 if (inzone != outzone) {
384 IP6STAT_INC(ip6s_cantforward);
385 IP6STAT_INC(ip6s_badscope);
386 in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
388 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
389 V_ip6_log_time = time_uptime;
392 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
393 ip6_sprintf(ip6bufs, &ip6->ip6_src),
394 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
396 if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
399 icmp6_error(mcopy, ICMP6_DST_UNREACH,
400 ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
405 * Destination scope check: if a packet is going to break the scope
406 * zone of packet's destination address, discard it. This case should
407 * usually be prevented by appropriately-configured routing table, but
408 * we need an explicit check because we may mistakenly forward the
409 * packet to a different zone by (e.g.) a default route.
411 dst_in6 = ip6->ip6_dst;
412 if (in6_setscope(&dst_in6, m->m_pkthdr.rcvif, &inzone) != 0 ||
413 in6_setscope(&dst_in6, rt->rt_ifp, &outzone) != 0 ||
415 IP6STAT_INC(ip6s_cantforward);
416 IP6STAT_INC(ip6s_badscope);
420 if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
421 in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
428 mtu = IN6_LINKMTU(rt->rt_ifp);
431 * When we do IPsec tunnel ingress, we need to play
432 * with the link value (decrement IPsec header size
433 * from mtu value). The code is much simpler than v4
434 * case, as we have the outgoing interface for
435 * encapsulated packet as "rt->rt_ifp".
437 ipsechdrsiz = ipsec_hdrsiz(mcopy, IPSEC_DIR_OUTBOUND,
439 if (ipsechdrsiz < mtu)
442 * if mtu becomes less than minimum MTU,
443 * tell minimum MTU (and I'll need to fragment it).
448 icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
453 if (rt->rt_flags & RTF_GATEWAY)
454 dst = (struct sockaddr_in6 *)rt->rt_gateway;
457 * If we are to forward the packet using the same interface
458 * as one we got the packet from, perhaps we should send a redirect
459 * to sender to shortcut a hop.
460 * Only send redirect if source is sending directly to us,
461 * and if packet was not source routed (or has any options).
462 * Also, don't send redirect if forwarding using a route
463 * modified by a redirect.
465 if (V_ip6_sendredirects && rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
466 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0) {
467 if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) != 0) {
469 * If the incoming interface is equal to the outgoing
470 * one, and the link attached to the interface is
471 * point-to-point, then it will be highly probable
472 * that a routing loop occurs. Thus, we immediately
473 * drop the packet and send an ICMPv6 error message.
475 * type/code is based on suggestion by Rich Draves.
476 * not sure if it is the best pick.
478 icmp6_error(mcopy, ICMP6_DST_UNREACH,
479 ICMP6_DST_UNREACH_ADDR, 0);
486 * Fake scoped addresses. Note that even link-local source or
487 * destinaion can appear, if the originating node just sends the
488 * packet to us (without address resolution for the destination).
489 * Since both icmp6_error and icmp6_redirect_output fill the embedded
490 * link identifiers, we can do this stuff after making a copy for
491 * returning an error.
493 if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) {
495 * See corresponding comments in ip6_output.
496 * XXX: but is it possible that ip6_forward() sends a packet
497 * to a loopback interface? I don't think so, and thus
498 * I bark here. (jinmei@kame.net)
499 * XXX: it is common to route invalid packets to loopback.
500 * also, the codepath will be visited on use of ::1 in
506 if ((rt->rt_flags & (RTF_BLACKHOLE|RTF_REJECT)) == 0)
509 printf("ip6_forward: outgoing interface is loopback. "
510 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
511 ip6_sprintf(ip6bufs, &ip6->ip6_src),
512 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
513 ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
514 if_name(rt->rt_ifp));
517 /* we can just use rcvif in forwarding. */
518 origifp = m->m_pkthdr.rcvif;
521 origifp = rt->rt_ifp;
523 * clear embedded scope identifiers if necessary.
524 * in6_clearscope will touch the addresses only when necessary.
526 in6_clearscope(&ip6->ip6_src);
527 in6_clearscope(&ip6->ip6_dst);
529 /* Jump over all PFIL processing if hooks are not active. */
530 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
534 /* Run through list of hooks for output packets. */
535 error = pfil_run_hooks(&V_inet6_pfil_hook, &m, rt->rt_ifp, PFIL_OUT, NULL);
536 if (error != 0 || m == NULL)
537 goto freecopy; /* consumed by filter */
538 ip6 = mtod(m, struct ip6_hdr *);
540 /* See if destination IP address was changed by packet filter. */
541 if (!IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst)) {
542 m->m_flags |= M_SKIP_FIREWALL;
543 /* If destination is now ourself drop to ip6_input(). */
544 if (in6_localip(&ip6->ip6_dst)) {
545 m->m_flags |= M_FASTFWD_OURS;
546 if (m->m_pkthdr.rcvif == NULL)
547 m->m_pkthdr.rcvif = V_loif;
548 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) {
549 m->m_pkthdr.csum_flags |=
550 CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR;
551 m->m_pkthdr.csum_data = 0xffff;
554 if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6)
555 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
557 error = netisr_queue(NETISR_IPV6, m);
560 goto again; /* Redo the routing table lookup. */
563 /* See if local, if yes, send it to netisr. */
564 if (m->m_flags & M_FASTFWD_OURS) {
565 if (m->m_pkthdr.rcvif == NULL)
566 m->m_pkthdr.rcvif = V_loif;
567 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) {
568 m->m_pkthdr.csum_flags |=
569 CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR;
570 m->m_pkthdr.csum_data = 0xffff;
573 if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6)
574 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
576 error = netisr_queue(NETISR_IPV6, m);
579 /* Or forward to some other address? */
580 if ((m->m_flags & M_IP6_NEXTHOP) &&
581 (fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) {
582 dst = (struct sockaddr_in6 *)&rin6.ro_dst;
583 bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in6));
584 m->m_flags |= M_SKIP_FIREWALL;
585 m->m_flags &= ~M_IP6_NEXTHOP;
586 m_tag_delete(m, fwd_tag);
591 error = nd6_output(rt->rt_ifp, origifp, m, dst, rt);
593 in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
594 IP6STAT_INC(ip6s_cantforward);
596 IP6STAT_INC(ip6s_forward);
597 in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
599 IP6STAT_INC(ip6s_redirectsent);
610 if (type == ND_REDIRECT) {
611 icmp6_redirect_output(mcopy, rt);
617 /* xxx MTU is constant in PPP? */
621 /* Tell source to slow down like source quench in IP? */
624 case ENETUNREACH: /* shouldn't happen, checked above */
629 type = ICMP6_DST_UNREACH;
630 code = ICMP6_DST_UNREACH_ADDR;
633 icmp6_error(mcopy, type, code, 0);