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 <netinet6/ip6_ipsec.h>
75 #include <netipsec/ipsec.h>
76 #include <netipsec/ipsec6.h>
77 #include <netipsec/key.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;
110 struct m_tag *fwd_tag;
111 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
114 * Do not forward packets to multicast destination (should be handled
116 * Do not forward packets with unspecified source. It was discussed
117 * in July 2000, on the ipngwg mailing list.
119 if ((m->m_flags & (M_BCAST|M_MCAST)) != 0 ||
120 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
121 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
122 IP6STAT_INC(ip6s_cantforward);
123 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
124 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
125 V_ip6_log_time = time_uptime;
128 "from %s to %s nxt %d received on %s\n",
129 ip6_sprintf(ip6bufs, &ip6->ip6_src),
130 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
132 if_name(m->m_pkthdr.rcvif));
139 * Check if this packet has an active SA and needs to be dropped
140 * instead of forwarded.
142 if (ip6_ipsec_fwd(m) != 0) {
143 IP6STAT_INC(ip6s_cantforward);
152 if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
153 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
154 icmp6_error(m, ICMP6_TIME_EXCEEDED,
155 ICMP6_TIME_EXCEED_TRANSIT, 0);
158 ip6->ip6_hlim -= IPV6_HLIMDEC;
165 * Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
166 * size of IPv6 + ICMPv6 headers) bytes of the packet in case
167 * we need to generate an ICMP6 message to the src.
168 * Thanks to M_EXT, in most cases copy will not occur.
170 * It is important to save it before IPsec processing as IPsec
171 * processing may modify the mbuf.
173 mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));
176 /* get a security policy for this packet */
177 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, &error);
179 IPSEC6STAT_INC(ips_out_inval);
180 IP6STAT_INC(ip6s_cantforward);
183 /* XXX: what icmp ? */
195 switch (sp->policy) {
196 case IPSEC_POLICY_DISCARD:
198 * This packet is just discarded.
200 IPSEC6STAT_INC(ips_out_polvio);
201 IP6STAT_INC(ip6s_cantforward);
205 /* XXX: what icmp ? */
213 case IPSEC_POLICY_BYPASS:
214 case IPSEC_POLICY_NONE:
215 /* no need to do IPsec. */
219 case IPSEC_POLICY_IPSEC:
220 if (sp->req == NULL) {
221 /* XXX should be panic ? */
222 printf("ip6_forward: No IPsec request specified.\n");
223 IP6STAT_INC(ip6s_cantforward);
227 /* XXX: what icmp ? */
238 case IPSEC_POLICY_ENTRUST:
240 /* should be panic ?? */
241 printf("ip6_forward: Invalid policy found. %d\n", sp->policy);
247 struct ipsecrequest *isr = NULL;
250 * when the kernel forwards a packet, it is not proper to apply
251 * IPsec transport mode to the packet. This check avoid from this.
252 * at present, if there is even a transport mode SA request in the
253 * security policy, the kernel does not apply IPsec to the packet.
254 * this check is not enough because the following case is valid.
255 * ipsec esp/tunnel/xxx-xxx/require esp/transport//require;
257 for (isr = sp->req; isr; isr = isr->next) {
258 if (isr->saidx.mode == IPSEC_MODE_ANY)
260 if (isr->saidx.mode == IPSEC_MODE_TUNNEL)
265 * if there's no need for tunnel mode IPsec, skip.
272 * All the extension headers will become inaccessible
273 * (since they can be encrypted).
274 * Don't panic, we need no more updates to extension headers
275 * on inner IPv6 packet (since they are now encapsulated).
277 * IPv6 [ESP|AH] IPv6 [extension headers] payload
281 * If we need to encapsulate the packet, do it here
282 * ipsec6_proces_packet will send the packet using ip6_output
284 error = ipsec6_process_packet(m, sp->req);
285 /* Release SP if an error occured */
288 if (error == EJUSTRETURN) {
290 * We had a SP with a level of 'use' and no SA. We
291 * will just continue to process the packet without
299 /* mbuf is already reclaimed in ipsec6_process_packet. */
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 ? */
325 * In the FAST IPSec case we have already
326 * re-injected the packet and it has been freed
327 * by the ipsec_done() function. So, just clean
328 * up after ourselves.
337 bzero(&rin6, sizeof(struct route_in6));
338 dst = (struct sockaddr_in6 *)&rin6.ro_dst;
339 dst->sin6_len = sizeof(struct sockaddr_in6);
340 dst->sin6_family = AF_INET6;
341 dst->sin6_addr = ip6->ip6_dst;
343 rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
344 if (rin6.ro_rt != NULL)
345 RT_UNLOCK(rin6.ro_rt);
347 IP6STAT_INC(ip6s_noroute);
348 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
350 icmp6_error(mcopy, ICMP6_DST_UNREACH,
351 ICMP6_DST_UNREACH_NOROUTE, 0);
358 * Source scope check: if a packet can't be delivered to its
359 * destination for the reason that the destination is beyond the scope
360 * of the source address, discard the packet and return an icmp6
361 * destination unreachable error with Code 2 (beyond scope of source
362 * address). We use a local copy of ip6_src, since in6_setscope()
363 * will possibly modify its first argument.
364 * [draft-ietf-ipngwg-icmp-v3-04.txt, Section 3.1]
366 src_in6 = ip6->ip6_src;
367 if (in6_setscope(&src_in6, rt->rt_ifp, &outzone)) {
368 /* XXX: this should not happen */
369 IP6STAT_INC(ip6s_cantforward);
370 IP6STAT_INC(ip6s_badscope);
373 if (in6_setscope(&src_in6, m->m_pkthdr.rcvif, &inzone)) {
374 IP6STAT_INC(ip6s_cantforward);
375 IP6STAT_INC(ip6s_badscope);
378 if (inzone != outzone) {
379 IP6STAT_INC(ip6s_cantforward);
380 IP6STAT_INC(ip6s_badscope);
381 in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
383 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
384 V_ip6_log_time = time_uptime;
387 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
388 ip6_sprintf(ip6bufs, &ip6->ip6_src),
389 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
391 if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
394 icmp6_error(mcopy, ICMP6_DST_UNREACH,
395 ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
400 * Destination scope check: if a packet is going to break the scope
401 * zone of packet's destination address, discard it. This case should
402 * usually be prevented by appropriately-configured routing table, but
403 * we need an explicit check because we may mistakenly forward the
404 * packet to a different zone by (e.g.) a default route.
406 dst_in6 = ip6->ip6_dst;
407 if (in6_setscope(&dst_in6, m->m_pkthdr.rcvif, &inzone) != 0 ||
408 in6_setscope(&dst_in6, rt->rt_ifp, &outzone) != 0 ||
410 IP6STAT_INC(ip6s_cantforward);
411 IP6STAT_INC(ip6s_badscope);
415 if (rt->rt_flags & RTF_GATEWAY)
416 dst = (struct sockaddr_in6 *)rt->rt_gateway;
419 * If we are to forward the packet using the same interface
420 * as one we got the packet from, perhaps we should send a redirect
421 * to sender to shortcut a hop.
422 * Only send redirect if source is sending directly to us,
423 * and if packet was not source routed (or has any options).
424 * Also, don't send redirect if forwarding using a route
425 * modified by a redirect.
427 if (V_ip6_sendredirects && rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
428 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0) {
429 if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) != 0) {
431 * If the incoming interface is equal to the outgoing
432 * one, and the link attached to the interface is
433 * point-to-point, then it will be highly probable
434 * that a routing loop occurs. Thus, we immediately
435 * drop the packet and send an ICMPv6 error message.
437 * type/code is based on suggestion by Rich Draves.
438 * not sure if it is the best pick.
440 icmp6_error(mcopy, ICMP6_DST_UNREACH,
441 ICMP6_DST_UNREACH_ADDR, 0);
448 * Fake scoped addresses. Note that even link-local source or
449 * destinaion can appear, if the originating node just sends the
450 * packet to us (without address resolution for the destination).
451 * Since both icmp6_error and icmp6_redirect_output fill the embedded
452 * link identifiers, we can do this stuff after making a copy for
453 * returning an error.
455 if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) {
457 * See corresponding comments in ip6_output.
458 * XXX: but is it possible that ip6_forward() sends a packet
459 * to a loopback interface? I don't think so, and thus
460 * I bark here. (jinmei@kame.net)
461 * XXX: it is common to route invalid packets to loopback.
462 * also, the codepath will be visited on use of ::1 in
468 if ((rt->rt_flags & (RTF_BLACKHOLE|RTF_REJECT)) == 0)
471 printf("ip6_forward: outgoing interface is loopback. "
472 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
473 ip6_sprintf(ip6bufs, &ip6->ip6_src),
474 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
475 ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
476 if_name(rt->rt_ifp));
479 /* we can just use rcvif in forwarding. */
480 origifp = m->m_pkthdr.rcvif;
483 origifp = rt->rt_ifp;
485 * clear embedded scope identifiers if necessary.
486 * in6_clearscope will touch the addresses only when necessary.
488 in6_clearscope(&ip6->ip6_src);
489 in6_clearscope(&ip6->ip6_dst);
491 /* Jump over all PFIL processing if hooks are not active. */
492 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
496 /* Run through list of hooks for output packets. */
497 error = pfil_run_hooks(&V_inet6_pfil_hook, &m, rt->rt_ifp, PFIL_OUT, NULL);
498 if (error != 0 || m == NULL)
499 goto freecopy; /* consumed by filter */
500 ip6 = mtod(m, struct ip6_hdr *);
502 /* See if destination IP address was changed by packet filter. */
503 if (!IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst)) {
504 m->m_flags |= M_SKIP_FIREWALL;
505 /* If destination is now ourself drop to ip6_input(). */
506 if (in6_localip(&ip6->ip6_dst))
507 m->m_flags |= M_FASTFWD_OURS;
509 goto again; /* Redo the routing table lookup. */
512 /* See if local, if yes, send it to netisr. */
513 if (m->m_flags & M_FASTFWD_OURS) {
514 if (m->m_pkthdr.rcvif == NULL)
515 m->m_pkthdr.rcvif = V_loif;
516 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) {
517 m->m_pkthdr.csum_flags |=
518 CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR;
519 m->m_pkthdr.csum_data = 0xffff;
522 if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6)
523 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
525 error = netisr_queue(NETISR_IPV6, m);
528 /* Or forward to some other address? */
529 if ((m->m_flags & M_IP6_NEXTHOP) &&
530 (fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) {
531 dst = (struct sockaddr_in6 *)&rin6.ro_dst;
532 bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in6));
533 m->m_flags |= M_SKIP_FIREWALL;
534 m->m_flags &= ~M_IP6_NEXTHOP;
535 m_tag_delete(m, fwd_tag);
540 /* See if the size was changed by the packet filter. */
541 if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
542 in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
549 mtu = IN6_LINKMTU(rt->rt_ifp);
552 * When we do IPsec tunnel ingress, we need to play
553 * with the link value (decrement IPsec header size
554 * from mtu value). The code is much simpler than v4
555 * case, as we have the outgoing interface for
556 * encapsulated packet as "rt->rt_ifp".
558 ipsechdrsiz = ipsec_hdrsiz(mcopy, IPSEC_DIR_OUTBOUND,
560 if (ipsechdrsiz < mtu)
563 * if mtu becomes less than minimum MTU,
564 * tell minimum MTU (and I'll need to fragment it).
569 icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
574 error = nd6_output(rt->rt_ifp, origifp, m, dst, rt);
576 in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
577 IP6STAT_INC(ip6s_cantforward);
579 IP6STAT_INC(ip6s_forward);
580 in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
582 IP6STAT_INC(ip6s_redirectsent);
593 if (type == ND_REDIRECT) {
594 icmp6_redirect_output(mcopy, rt);
600 /* xxx MTU is constant in PPP? */
604 /* Tell source to slow down like source quench in IP? */
607 case ENETUNREACH: /* shouldn't happen, checked above */
612 type = ICMP6_DST_UNREACH;
613 code = ICMP6_DST_UNREACH_ADDR;
616 icmp6_error(mcopy, type, code, 0);