2 /* $KAME: ip6_forward.c,v 1.69 2001/05/17 03:48:30 itojun Exp $ */
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 #include "opt_inet6.h"
35 #include "opt_ipsec.h"
36 #include "opt_ipstealth.h"
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/malloc.h>
42 #include <sys/domain.h>
43 #include <sys/protosw.h>
44 #include <sys/socket.h>
45 #include <sys/errno.h>
47 #include <sys/kernel.h>
48 #include <sys/syslog.h>
51 #include <net/route.h>
54 #include <netinet/in.h>
55 #include <netinet/in_var.h>
56 #include <netinet/in_systm.h>
57 #include <netinet/ip.h>
58 #include <netinet/ip_var.h>
59 #include <netinet6/in6_var.h>
60 #include <netinet/ip6.h>
61 #include <netinet6/ip6_var.h>
62 #include <netinet6/scope6_var.h>
63 #include <netinet/icmp6.h>
64 #include <netinet6/nd6.h>
66 #include <netinet/in_pcb.h>
69 #include <netinet6/ipsec.h>
71 #include <netinet6/ipsec6.h>
73 #include <netkey/key.h>
77 #include <netipsec/ipsec.h>
78 #include <netipsec/ipsec6.h>
79 #include <netipsec/key.h>
81 #endif /* FAST_IPSEC */
83 #include <net/net_osdep.h>
85 #include <netinet6/ip6protosw.h>
87 struct route_in6 ip6_forward_rt;
90 * Forward a packet. If some error occurs return the sender
91 * an icmp packet. Note we can't always generate a meaningful
92 * icmp message because icmp doesn't have a large enough repertoire
95 * If not forwarding, just drop the packet. This could be confusing
96 * if ipforwarding was zero but some routing protocol was advancing
97 * us as a gateway to somewhere. However, we must let the routing
98 * protocol deal with that.
103 ip6_forward(m, srcrt)
107 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
108 struct sockaddr_in6 *dst = NULL;
109 struct rtentry *rt = NULL;
110 int error, type = 0, code = 0;
111 struct mbuf *mcopy = NULL;
112 struct ifnet *origifp; /* maybe unnecessary */
113 u_int32_t inzone, outzone;
114 struct in6_addr src_in6, dst_in6;
116 struct secpolicy *sp = NULL;
120 GIANT_REQUIRED; /* XXX bz: ip6_forward_rt */
124 * Check AH/ESP integrity.
127 * Don't increment ip6s_cantforward because this is the check
128 * before forwarding packet actually.
130 if (ipsec6_in_reject(m, NULL)) {
131 #if !defined(FAST_IPSEC)
132 ipsec6stat.in_polvio++;
140 * Do not forward packets to multicast destination (should be handled
142 * Do not forward packets with unspecified source. It was discussed
143 * in July 2000, on the ipngwg mailing list.
145 if ((m->m_flags & (M_BCAST|M_MCAST)) != 0 ||
146 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
147 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
148 ip6stat.ip6s_cantforward++;
149 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
150 if (ip6_log_time + ip6_log_interval < time_second) {
151 ip6_log_time = time_second;
154 "from %s to %s nxt %d received on %s\n",
155 ip6_sprintf(&ip6->ip6_src),
156 ip6_sprintf(&ip6->ip6_dst),
158 if_name(m->m_pkthdr.rcvif));
167 if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
168 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
169 icmp6_error(m, ICMP6_TIME_EXCEEDED,
170 ICMP6_TIME_EXCEED_TRANSIT, 0);
173 ip6->ip6_hlim -= IPV6_HLIMDEC;
180 * Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
181 * size of IPv6 + ICMPv6 headers) bytes of the packet in case
182 * we need to generate an ICMP6 message to the src.
183 * Thanks to M_EXT, in most cases copy will not occur.
185 * It is important to save it before IPsec processing as IPsec
186 * processing may modify the mbuf.
188 mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));
191 /* get a security policy for this packet */
192 sp = ipsec6_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND,
193 IP_FORWARDING, &error);
195 ipsec6stat.out_inval++;
196 ip6stat.ip6s_cantforward++;
199 /* XXX: what icmp ? */
211 switch (sp->policy) {
212 case IPSEC_POLICY_DISCARD:
214 * This packet is just discarded.
216 ipsec6stat.out_polvio++;
217 ip6stat.ip6s_cantforward++;
221 /* XXX: what icmp ? */
229 case IPSEC_POLICY_BYPASS:
230 case IPSEC_POLICY_NONE:
231 /* no need to do IPsec. */
235 case IPSEC_POLICY_IPSEC:
236 if (sp->req == NULL) {
237 /* XXX should be panic ? */
238 printf("ip6_forward: No IPsec request specified.\n");
239 ip6stat.ip6s_cantforward++;
243 /* XXX: what icmp ? */
254 case IPSEC_POLICY_ENTRUST:
256 /* should be panic ?? */
257 printf("ip6_forward: Invalid policy found. %d\n", sp->policy);
263 struct ipsecrequest *isr = NULL;
264 struct ipsec_output_state state;
267 * when the kernel forwards a packet, it is not proper to apply
268 * IPsec transport mode to the packet is not proper. this check
270 * at present, if there is even a transport mode SA request in the
271 * security policy, the kernel does not apply IPsec to the packet.
272 * this check is not enough because the following case is valid.
273 * ipsec esp/tunnel/xxx-xxx/require esp/transport//require;
275 for (isr = sp->req; isr; isr = isr->next) {
276 if (isr->saidx.mode == IPSEC_MODE_ANY)
278 if (isr->saidx.mode == IPSEC_MODE_TUNNEL)
283 * if there's no need for tunnel mode IPsec, skip.
290 * All the extension headers will become inaccessible
291 * (since they can be encrypted).
292 * Don't panic, we need no more updates to extension headers
293 * on inner IPv6 packet (since they are now encapsulated).
295 * IPv6 [ESP|AH] IPv6 [extension headers] payload
297 bzero(&state, sizeof(state));
299 state.ro = NULL; /* update at ipsec6_output_tunnel() */
300 state.dst = NULL; /* update at ipsec6_output_tunnel() */
302 error = ipsec6_output_tunnel(&state, sp, 0);
308 /* mbuf is already reclaimed in ipsec6_output_tunnel. */
317 printf("ip6_output (ipsec): error code %d\n", error);
320 /* don't show these error codes to the user */
323 ip6stat.ip6s_cantforward++;
326 /* XXX: what icmp ? */
335 if (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)
359 dst = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
361 /* ip6_forward_rt.ro_dst.sin6_addr is equal to ip6->ip6_dst */
362 if (ip6_forward_rt.ro_rt == 0 ||
363 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) == 0) {
364 if (ip6_forward_rt.ro_rt) {
365 RTFREE(ip6_forward_rt.ro_rt);
366 ip6_forward_rt.ro_rt = 0;
369 /* this probably fails but give it a try again */
370 rtalloc((struct route *)&ip6_forward_rt);
373 if (ip6_forward_rt.ro_rt == 0) {
374 ip6stat.ip6s_noroute++;
375 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
377 icmp6_error(mcopy, ICMP6_DST_UNREACH,
378 ICMP6_DST_UNREACH_NOROUTE, 0);
383 } else if ((rt = ip6_forward_rt.ro_rt) == 0 ||
384 !IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &dst->sin6_addr)) {
385 if (ip6_forward_rt.ro_rt) {
386 RTFREE(ip6_forward_rt.ro_rt);
387 ip6_forward_rt.ro_rt = 0;
389 bzero(dst, sizeof(*dst));
390 dst->sin6_len = sizeof(struct sockaddr_in6);
391 dst->sin6_family = AF_INET6;
392 dst->sin6_addr = ip6->ip6_dst;
394 rtalloc((struct route *)&ip6_forward_rt);
395 if (ip6_forward_rt.ro_rt == 0) {
396 ip6stat.ip6s_noroute++;
397 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
399 icmp6_error(mcopy, ICMP6_DST_UNREACH,
400 ICMP6_DST_UNREACH_NOROUTE, 0);
406 rt = ip6_forward_rt.ro_rt;
412 * Source scope check: if a packet can't be delivered to its
413 * destination for the reason that the destination is beyond the scope
414 * of the source address, discard the packet and return an icmp6
415 * destination unreachable error with Code 2 (beyond scope of source
416 * address). We use a local copy of ip6_src, since in6_setscope()
417 * will possibly modify its first argument.
418 * [draft-ietf-ipngwg-icmp-v3-04.txt, Section 3.1]
420 src_in6 = ip6->ip6_src;
421 if (in6_setscope(&src_in6, rt->rt_ifp, &outzone)) {
422 /* XXX: this should not happen */
423 ip6stat.ip6s_cantforward++;
424 ip6stat.ip6s_badscope++;
428 if (in6_setscope(&src_in6, m->m_pkthdr.rcvif, &inzone)) {
429 ip6stat.ip6s_cantforward++;
430 ip6stat.ip6s_badscope++;
434 if (inzone != outzone
439 ip6stat.ip6s_cantforward++;
440 ip6stat.ip6s_badscope++;
441 in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
443 if (ip6_log_time + ip6_log_interval < time_second) {
444 ip6_log_time = time_second;
447 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
448 ip6_sprintf(&ip6->ip6_src),
449 ip6_sprintf(&ip6->ip6_dst),
451 if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
454 icmp6_error(mcopy, ICMP6_DST_UNREACH,
455 ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
461 * Destination scope check: if a packet is going to break the scope
462 * zone of packet's destination address, discard it. This case should
463 * usually be prevented by appropriately-configured routing table, but
464 * we need an explicit check because we may mistakenly forward the
465 * packet to a different zone by (e.g.) a default route.
467 dst_in6 = ip6->ip6_dst;
468 if (in6_setscope(&dst_in6, m->m_pkthdr.rcvif, &inzone) != 0 ||
469 in6_setscope(&dst_in6, rt->rt_ifp, &outzone) != 0 ||
471 ip6stat.ip6s_cantforward++;
472 ip6stat.ip6s_badscope++;
477 if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
478 in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
482 struct secpolicy *sp;
487 mtu = IN6_LINKMTU(rt->rt_ifp);
490 * When we do IPsec tunnel ingress, we need to play
491 * with the link value (decrement IPsec header size
492 * from mtu value). The code is much simpler than v4
493 * case, as we have the outgoing interface for
494 * encapsulated packet as "rt->rt_ifp".
496 sp = ipsec6_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND,
497 IP_FORWARDING, &ipsecerror);
499 ipsechdrsiz = ipsec6_hdrsiz(mcopy,
500 IPSEC_DIR_OUTBOUND, NULL);
501 if (ipsechdrsiz < mtu)
506 * if mtu becomes less than minimum MTU,
507 * tell minimum MTU (and I'll need to fragment it).
512 icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
518 if (rt->rt_flags & RTF_GATEWAY)
519 dst = (struct sockaddr_in6 *)rt->rt_gateway;
522 * If we are to forward the packet using the same interface
523 * as one we got the packet from, perhaps we should send a redirect
524 * to sender to shortcut a hop.
525 * Only send redirect if source is sending directly to us,
526 * and if packet was not source routed (or has any options).
527 * Also, don't send redirect if forwarding using a route
528 * modified by a redirect.
530 if (rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
534 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0) {
535 if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) != 0) {
537 * If the incoming interface is equal to the outgoing
538 * one, and the link attached to the interface is
539 * point-to-point, then it will be highly probable
540 * that a routing loop occurs. Thus, we immediately
541 * drop the packet and send an ICMPv6 error message.
543 * type/code is based on suggestion by Rich Draves.
544 * not sure if it is the best pick.
546 icmp6_error(mcopy, ICMP6_DST_UNREACH,
547 ICMP6_DST_UNREACH_ADDR, 0);
555 * Fake scoped addresses. Note that even link-local source or
556 * destinaion can appear, if the originating node just sends the
557 * packet to us (without address resolution for the destination).
558 * Since both icmp6_error and icmp6_redirect_output fill the embedded
559 * link identifiers, we can do this stuff after making a copy for
560 * returning an error.
562 if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) {
564 * See corresponding comments in ip6_output.
565 * XXX: but is it possible that ip6_forward() sends a packet
566 * to a loopback interface? I don't think so, and thus
567 * I bark here. (jinmei@kame.net)
568 * XXX: it is common to route invalid packets to loopback.
569 * also, the codepath will be visited on use of ::1 in
575 if ((rt->rt_flags & (RTF_BLACKHOLE|RTF_REJECT)) == 0)
578 printf("ip6_forward: outgoing interface is loopback. "
579 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
580 ip6_sprintf(&ip6->ip6_src),
581 ip6_sprintf(&ip6->ip6_dst),
582 ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
583 if_name(rt->rt_ifp));
586 /* we can just use rcvif in forwarding. */
587 origifp = m->m_pkthdr.rcvif;
590 origifp = rt->rt_ifp;
592 * clear embedded scope identifiers if necessary.
593 * in6_clearscope will touch the addresses only when necessary.
595 in6_clearscope(&ip6->ip6_src);
596 in6_clearscope(&ip6->ip6_dst);
598 /* Jump over all PFIL processing if hooks are not active. */
599 if (!PFIL_HOOKED(&inet6_pfil_hook))
602 /* Run through list of hooks for output packets. */
603 error = pfil_run_hooks(&inet6_pfil_hook, &m, rt->rt_ifp, PFIL_OUT, NULL);
608 ip6 = mtod(m, struct ip6_hdr *);
611 error = nd6_output(rt->rt_ifp, origifp, m, dst, rt);
613 in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
614 ip6stat.ip6s_cantforward++;
616 ip6stat.ip6s_forward++;
617 in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
619 ip6stat.ip6s_redirectsent++;
631 if (type == ND_REDIRECT) {
632 icmp6_redirect_output(mcopy, rt);
638 /* xxx MTU is constant in PPP? */
642 /* Tell source to slow down like source quench in IP? */
645 case ENETUNREACH: /* shouldn't happen, checked above */
650 type = ICMP6_DST_UNREACH;
651 code = ICMP6_DST_UNREACH_ADDR;
654 icmp6_error(mcopy, type, code, 0);