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_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $
33 * Copyright (c) 1982, 1986, 1988, 1993
34 * The Regents of the University of California. All rights reserved.
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 4. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
63 #include <sys/cdefs.h>
64 __FBSDID("$FreeBSD$");
67 #include "opt_inet6.h"
68 #include "opt_ipsec.h"
70 #include <sys/param.h>
71 #include <sys/systm.h>
72 #include <sys/malloc.h>
75 #include <sys/domain.h>
76 #include <sys/protosw.h>
77 #include <sys/socket.h>
78 #include <sys/socketvar.h>
79 #include <sys/errno.h>
81 #include <sys/kernel.h>
82 #include <sys/syslog.h>
85 #include <net/if_types.h>
86 #include <net/if_dl.h>
87 #include <net/route.h>
88 #include <net/netisr.h>
91 #include <netinet/in.h>
92 #include <netinet/in_systm.h>
94 #include <netinet/ip.h>
95 #include <netinet/ip_icmp.h>
97 #include <netinet/ip6.h>
98 #include <netinet6/in6_var.h>
99 #include <netinet6/ip6_var.h>
100 #include <netinet/in_pcb.h>
101 #include <netinet/icmp6.h>
102 #include <netinet6/scope6_var.h>
103 #include <netinet6/in6_ifattach.h>
104 #include <netinet6/nd6.h>
107 #include <netipsec/ipsec.h>
108 #include <netinet6/ip6_ipsec.h>
109 #include <netipsec/ipsec6.h>
112 #include <netinet6/ip6protosw.h>
114 extern struct domain inet6domain;
116 u_char ip6_protox[IPPROTO_MAX];
117 static struct ifqueue ip6intrq;
118 static int ip6qmaxlen = IFQ_MAXLEN;
119 struct in6_ifaddr *in6_ifaddr;
121 extern struct callout in6_tmpaddrtimer_ch;
123 int ip6_forward_srcrt; /* XXX */
124 int ip6_sourcecheck; /* XXX */
125 int ip6_sourcecheck_interval; /* XXX */
127 int ip6_ours_check_algorithm;
129 struct pfil_head inet6_pfil_hook;
131 struct ip6stat ip6stat;
133 static void ip6_init2(void *);
134 static struct ip6aux *ip6_setdstifaddr(struct mbuf *, struct in6_ifaddr *);
135 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
137 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
141 * IP6 initialization: fill in IP6 protocol switch table.
142 * All protocols not implemented in kernel go to raw IP6 protocol handler.
147 struct ip6protosw *pr;
151 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
152 panic("sizeof(protosw) != sizeof(ip6protosw)");
154 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
158 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
159 for (i = 0; i < IPPROTO_MAX; i++)
160 ip6_protox[i] = pr - inet6sw;
162 * Cycle through IP protocols and put them into the appropriate place
165 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
166 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
167 if (pr->pr_domain->dom_family == PF_INET6 &&
168 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) {
169 /* Be careful to only index valid IP protocols. */
170 if (pr->pr_protocol < IPPROTO_MAX)
171 ip6_protox[pr->pr_protocol] = pr - inet6sw;
174 /* Initialize packet filter hooks. */
175 inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
176 inet6_pfil_hook.ph_af = AF_INET6;
177 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0)
178 printf("%s: WARNING: unable to register pfil hook, "
179 "error %d\n", __func__, i);
181 ip6intrq.ifq_maxlen = ip6qmaxlen;
182 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF);
183 netisr_register(NETISR_IPV6, ip6_input, &ip6intrq, 0);
185 addrsel_policy_init();
188 ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
192 ip6_init2(void *dummy)
196 callout_init(&nd6_timer_ch, 0);
197 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
199 /* timer for regeneranation of temporary addresses randomize ID */
200 callout_init(&in6_tmpaddrtimer_ch, 0);
201 callout_reset(&in6_tmpaddrtimer_ch,
202 (ip6_temp_preferred_lifetime - ip6_desync_factor -
203 ip6_temp_regen_advance) * hz,
204 in6_tmpaddrtimer, NULL);
208 /* This must be after route_init(), which is now SI_ORDER_THIRD */
209 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
211 extern struct route_in6 ip6_forward_rt;
214 ip6_input(struct mbuf *m)
217 int off = sizeof(struct ip6_hdr), nest;
219 u_int32_t rtalert = ~0;
221 struct ifnet *deliverifp = NULL;
222 struct in6_addr odst;
225 GIANT_REQUIRED; /* XXX for now */
229 * should the inner packet be considered authentic?
230 * see comment in ah4_input().
231 * NB: m cannot be NULL when passed to the input routine
234 m->m_flags &= ~M_AUTHIPHDR;
235 m->m_flags &= ~M_AUTHIPDGM;
240 * make sure we don't have onion peering information into m_tag.
247 if (m->m_flags & M_EXT) {
249 ip6stat.ip6s_mext2m++;
251 ip6stat.ip6s_mext1++;
253 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
255 if (m->m_flags & M_LOOP) {
256 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
257 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
258 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
260 ip6stat.ip6s_m2m[0]++;
266 /* drop the packet if IPv6 operation is disabled on the IF */
267 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
272 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
273 ip6stat.ip6s_total++;
275 #ifndef PULLDOWN_TEST
277 * L2 bridge code and some other code can return mbuf chain
278 * that does not conform to KAME requirement. too bad.
279 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
281 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
284 MGETHDR(n, M_DONTWAIT, MT_HEADER);
287 if (n && n->m_pkthdr.len > MHLEN) {
288 MCLGET(n, M_DONTWAIT);
289 if ((n->m_flags & M_EXT) == 0) {
296 return; /* ENOBUFS */
299 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
300 n->m_len = n->m_pkthdr.len;
304 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
307 if (m->m_len < sizeof(struct ip6_hdr)) {
309 inifp = m->m_pkthdr.rcvif;
310 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
311 ip6stat.ip6s_toosmall++;
312 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
317 ip6 = mtod(m, struct ip6_hdr *);
319 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
320 ip6stat.ip6s_badvers++;
321 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
325 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
328 * Check against address spoofing/corruption.
330 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
331 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
333 * XXX: "badscope" is not very suitable for a multicast source.
335 ip6stat.ip6s_badscope++;
336 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
339 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
340 !(m->m_flags & M_LOOP)) {
342 * In this case, the packet should come from the loopback
343 * interface. However, we cannot just check the if_flags,
344 * because ip6_mloopback() passes the "actual" interface
345 * as the outgoing/incoming interface.
347 ip6stat.ip6s_badscope++;
348 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
353 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
354 /* packet is dropped by traffic conditioner */
359 * The following check is not documented in specs. A malicious
360 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
361 * and bypass security checks (act as if it was from 127.0.0.1 by using
362 * IPv6 src ::ffff:127.0.0.1). Be cautious.
364 * This check chokes if we are in an SIIT cloud. As none of BSDs
365 * support IPv4-less kernel compilation, we cannot support SIIT
366 * environment at all. So, it makes more sense for us to reject any
367 * malicious packets for non-SIIT environment, than try to do a
368 * partial support for SIIT environment.
370 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
371 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
372 ip6stat.ip6s_badscope++;
373 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
378 * Reject packets with IPv4 compatible addresses (auto tunnel).
380 * The code forbids auto tunnel relay case in RFC1933 (the check is
381 * stronger than RFC1933). We may want to re-enable it if mech-xx
382 * is revised to forbid relaying case.
384 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
385 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
386 ip6stat.ip6s_badscope++;
387 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
393 * Run through list of hooks for input packets.
395 * NB: Beware of the destination address changing
396 * (e.g. by NAT rewriting). When this happens,
397 * tell ip6_forward to do the right thing.
401 /* Jump over all PFIL processing if hooks are not active. */
402 if (!PFIL_HOOKED(&inet6_pfil_hook))
405 if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN, NULL))
407 if (m == NULL) /* consumed by filter */
409 ip6 = mtod(m, struct ip6_hdr *);
410 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
414 * Disambiguate address scope zones (if there is ambiguity).
415 * We first make sure that the original source or destination address
416 * is not in our internal form for scoped addresses. Such addresses
417 * are not necessarily invalid spec-wise, but we cannot accept them due
418 * to the usage conflict.
419 * in6_setscope() then also checks and rejects the cases where src or
420 * dst are the loopback address and the receiving interface
423 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
424 ip6stat.ip6s_badscope++; /* XXX */
427 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
428 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
429 ip6stat.ip6s_badscope++;
436 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
437 struct in6_multi *in6m = 0;
439 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
441 * See if we belong to the destination multicast group on the
444 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
447 else if (!ip6_mrouter) {
448 ip6stat.ip6s_notmember++;
449 ip6stat.ip6s_cantforward++;
450 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
453 deliverifp = m->m_pkthdr.rcvif;
460 if (ip6_forward_rt.ro_rt != NULL &&
461 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
462 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
463 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
464 ip6stat.ip6s_forward_cachehit++;
466 struct sockaddr_in6 *dst6;
468 if (ip6_forward_rt.ro_rt) {
469 /* route is down or destination is different */
470 ip6stat.ip6s_forward_cachemiss++;
471 RTFREE(ip6_forward_rt.ro_rt);
472 ip6_forward_rt.ro_rt = 0;
475 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
476 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
477 dst6->sin6_len = sizeof(struct sockaddr_in6);
478 dst6->sin6_family = AF_INET6;
479 dst6->sin6_addr = ip6->ip6_dst;
481 rtalloc((struct route *)&ip6_forward_rt);
484 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
487 * Accept the packet if the forwarding interface to the destination
488 * according to the routing table is the loopback interface,
489 * unless the associated route has a gateway.
490 * Note that this approach causes to accept a packet if there is a
491 * route to the loopback interface for the destination of the packet.
492 * But we think it's even useful in some situations, e.g. when using
493 * a special daemon which wants to intercept the packet.
495 * XXX: some OSes automatically make a cloned route for the destination
496 * of an outgoing packet. If the outgoing interface of the packet
497 * is a loopback one, the kernel would consider the packet to be
498 * accepted, even if we have no such address assinged on the interface.
499 * We check the cloned flag of the route entry to reject such cases,
500 * assuming that route entries for our own addresses are not made by
501 * cloning (it should be true because in6_addloop explicitly installs
502 * the host route). However, we might have to do an explicit check
503 * while it would be less efficient. Or, should we rather install a
504 * reject route for such a case?
506 if (ip6_forward_rt.ro_rt &&
507 (ip6_forward_rt.ro_rt->rt_flags &
508 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
510 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
513 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
517 * The check below is redundant since the comparison of
518 * the destination and the key of the rtentry has
519 * already done through looking up the routing table.
521 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
522 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
524 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
525 struct in6_ifaddr *ia6 =
526 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
529 * record address information into m_tag.
531 (void)ip6_setdstifaddr(m, ia6);
534 * packets to a tentative, duplicated, or somehow invalid
535 * address must not be accepted.
537 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
538 /* this address is ready */
540 deliverifp = ia6->ia_ifp; /* correct? */
541 /* Count the packet in the ip address stats */
542 ia6->ia_ifa.if_ipackets++;
543 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
546 char ip6bufs[INET6_ADDRSTRLEN];
547 char ip6bufd[INET6_ADDRSTRLEN];
548 /* address is not ready, so discard the packet. */
550 "ip6_input: packet to an unready address %s->%s\n",
551 ip6_sprintf(ip6bufs, &ip6->ip6_src),
552 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
559 * FAITH (Firewall Aided Internet Translator)
562 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
563 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
564 /* XXX do we need more sanity checks? */
566 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
572 * Now there is no reason to process the packet if it's not our own
573 * and we're not a router.
575 if (!ip6_forwarding) {
576 ip6stat.ip6s_cantforward++;
577 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
583 * record address information into m_tag, if we don't have one yet.
584 * note that we are unable to record it, if the address is not listed
585 * as our interface address (e.g. multicast addresses, addresses
586 * within FAITH prefixes and such).
588 if (deliverifp && !ip6_getdstifaddr(m)) {
589 struct in6_ifaddr *ia6;
591 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
593 if (!ip6_setdstifaddr(m, ia6)) {
595 * XXX maybe we should drop the packet here,
596 * as we could not provide enough information
597 * to the upper layers.
604 * Process Hop-by-Hop options header if it's contained.
605 * m may be modified in ip6_hopopts_input().
606 * If a JumboPayload option is included, plen will also be modified.
608 plen = (u_int32_t)ntohs(ip6->ip6_plen);
609 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
612 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
613 #if 0 /*touches NULL pointer*/
614 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
616 return; /* m have already been freed */
620 ip6 = mtod(m, struct ip6_hdr *);
623 * if the payload length field is 0 and the next header field
624 * indicates Hop-by-Hop Options header, then a Jumbo Payload
625 * option MUST be included.
627 if (ip6->ip6_plen == 0 && plen == 0) {
629 * Note that if a valid jumbo payload option is
630 * contained, ip6_hopopts_input() must set a valid
631 * (non-zero) payload length to the variable plen.
633 ip6stat.ip6s_badoptions++;
634 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
635 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
636 icmp6_error(m, ICMP6_PARAM_PROB,
637 ICMP6_PARAMPROB_HEADER,
638 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
641 #ifndef PULLDOWN_TEST
642 /* ip6_hopopts_input() ensures that mbuf is contiguous */
643 hbh = (struct ip6_hbh *)(ip6 + 1);
645 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
646 sizeof(struct ip6_hbh));
648 ip6stat.ip6s_tooshort++;
655 * If we are acting as a router and the packet contains a
656 * router alert option, see if we know the option value.
657 * Currently, we only support the option value for MLD, in which
658 * case we should pass the packet to the multicast routing
661 if (rtalert != ~0 && ip6_forwarding) {
663 case IP6OPT_RTALERT_MLD:
668 * RFC2711 requires unrecognized values must be
678 * Check that the amount of data in the buffers
679 * is as at least much as the IPv6 header would have us expect.
680 * Trim mbufs if longer than we expect.
681 * Drop packet if shorter than we expect.
683 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
684 ip6stat.ip6s_tooshort++;
685 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
688 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
689 if (m->m_len == m->m_pkthdr.len) {
690 m->m_len = sizeof(struct ip6_hdr) + plen;
691 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
693 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
697 * Forward if desirable.
699 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
701 * If we are acting as a multicast router, all
702 * incoming multicast packets are passed to the
703 * kernel-level multicast forwarding function.
704 * The packet is returned (relatively) intact; if
705 * ip6_mforward() returns a non-zero value, the packet
706 * must be discarded, else it may be accepted below.
708 if (ip6_mrouter && ip6_mforward &&
709 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
710 ip6stat.ip6s_cantforward++;
719 ip6_forward(m, srcrt);
723 ip6 = mtod(m, struct ip6_hdr *);
726 * Malicious party may be able to use IPv4 mapped addr to confuse
727 * tcp/udp stack and bypass security checks (act as if it was from
728 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
730 * For SIIT end node behavior, you may want to disable the check.
731 * However, you will become vulnerable to attacks using IPv4 mapped
734 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
735 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
736 ip6stat.ip6s_badscope++;
737 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
742 * Tell launch routine the next header
744 ip6stat.ip6s_delivered++;
745 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
748 while (nxt != IPPROTO_DONE) {
749 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
750 ip6stat.ip6s_toomanyhdr++;
755 * protection against faulty packet - there should be
756 * more sanity checks in header chain processing.
758 if (m->m_pkthdr.len < off) {
759 ip6stat.ip6s_tooshort++;
760 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
766 * enforce IPsec policy checking if we are seeing last header.
767 * note that we do not visit this with protocols with pcb layer
768 * code - like udp/tcp/raw ip.
770 if (ip6_ipsec_input(m, nxt))
773 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
781 * set/grab in6_ifaddr correspond to IPv6 destination address.
782 * XXX backward compatibility wrapper
784 static struct ip6aux *
785 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6)
789 ip6a = ip6_addaux(m);
791 ip6a->ip6a_dstia6 = ia6;
792 return ip6a; /* NULL if failed to set */
796 ip6_getdstifaddr(struct mbuf *m)
800 ip6a = ip6_findaux(m);
802 return ip6a->ip6a_dstia6;
808 * Hop-by-Hop options header processing. If a valid jumbo payload option is
809 * included, the real payload length will be stored in plenp.
811 * rtalertp - XXX: should be stored more smart way
814 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
815 struct mbuf **mp, int *offp)
817 struct mbuf *m = *mp;
818 int off = *offp, hbhlen;
822 /* validation of the length of the header */
823 #ifndef PULLDOWN_TEST
824 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
825 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
826 hbhlen = (hbh->ip6h_len + 1) << 3;
828 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
829 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
831 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
832 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
834 ip6stat.ip6s_tooshort++;
837 hbhlen = (hbh->ip6h_len + 1) << 3;
838 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
841 ip6stat.ip6s_tooshort++;
846 hbhlen -= sizeof(struct ip6_hbh);
847 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
849 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
850 hbhlen, rtalertp, plenp) < 0)
859 * Search header for all Hop-by-hop options and process each option.
860 * This function is separate from ip6_hopopts_input() in order to
861 * handle a case where the sending node itself process its hop-by-hop
862 * options header. In such a case, the function is called from ip6_output().
864 * The function assumes that hbh header is located right after the IPv6 header
865 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
866 * opthead + hbhlen is located in continuous memory region.
869 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
870 u_int32_t *rtalertp, u_int32_t *plenp)
874 u_int8_t *opt = opthead;
875 u_int16_t rtalert_val;
877 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
879 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
885 if (hbhlen < IP6OPT_MINLEN) {
886 ip6stat.ip6s_toosmall++;
889 optlen = *(opt + 1) + 2;
891 case IP6OPT_ROUTER_ALERT:
892 /* XXX may need check for alignment */
893 if (hbhlen < IP6OPT_RTALERT_LEN) {
894 ip6stat.ip6s_toosmall++;
897 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
899 icmp6_error(m, ICMP6_PARAM_PROB,
900 ICMP6_PARAMPROB_HEADER,
901 erroff + opt + 1 - opthead);
904 optlen = IP6OPT_RTALERT_LEN;
905 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
906 *rtalertp = ntohs(rtalert_val);
909 /* XXX may need check for alignment */
910 if (hbhlen < IP6OPT_JUMBO_LEN) {
911 ip6stat.ip6s_toosmall++;
914 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
916 icmp6_error(m, ICMP6_PARAM_PROB,
917 ICMP6_PARAMPROB_HEADER,
918 erroff + opt + 1 - opthead);
921 optlen = IP6OPT_JUMBO_LEN;
924 * IPv6 packets that have non 0 payload length
925 * must not contain a jumbo payload option.
927 ip6 = mtod(m, struct ip6_hdr *);
929 ip6stat.ip6s_badoptions++;
930 icmp6_error(m, ICMP6_PARAM_PROB,
931 ICMP6_PARAMPROB_HEADER,
932 erroff + opt - opthead);
937 * We may see jumbolen in unaligned location, so
938 * we'd need to perform bcopy().
940 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
941 jumboplen = (u_int32_t)htonl(jumboplen);
945 * if there are multiple jumbo payload options,
946 * *plenp will be non-zero and the packet will be
948 * the behavior may need some debate in ipngwg -
949 * multiple options does not make sense, however,
950 * there's no explicit mention in specification.
953 ip6stat.ip6s_badoptions++;
954 icmp6_error(m, ICMP6_PARAM_PROB,
955 ICMP6_PARAMPROB_HEADER,
956 erroff + opt + 2 - opthead);
962 * jumbo payload length must be larger than 65535.
964 if (jumboplen <= IPV6_MAXPACKET) {
965 ip6stat.ip6s_badoptions++;
966 icmp6_error(m, ICMP6_PARAM_PROB,
967 ICMP6_PARAMPROB_HEADER,
968 erroff + opt + 2 - opthead);
974 default: /* unknown option */
975 if (hbhlen < IP6OPT_MINLEN) {
976 ip6stat.ip6s_toosmall++;
979 optlen = ip6_unknown_opt(opt, m,
980 erroff + opt - opthead);
996 * Unknown option processing.
997 * The third argument `off' is the offset from the IPv6 header to the option,
998 * which is necessary if the IPv6 header the and option header and IPv6 header
999 * is not continuous in order to return an ICMPv6 error.
1002 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1004 struct ip6_hdr *ip6;
1006 switch (IP6OPT_TYPE(*optp)) {
1007 case IP6OPT_TYPE_SKIP: /* ignore the option */
1008 return ((int)*(optp + 1));
1009 case IP6OPT_TYPE_DISCARD: /* silently discard */
1012 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1013 ip6stat.ip6s_badoptions++;
1014 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1016 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1017 ip6stat.ip6s_badoptions++;
1018 ip6 = mtod(m, struct ip6_hdr *);
1019 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1020 (m->m_flags & (M_BCAST|M_MCAST)))
1023 icmp6_error(m, ICMP6_PARAM_PROB,
1024 ICMP6_PARAMPROB_OPTION, off);
1028 m_freem(m); /* XXX: NOTREACHED */
1033 * Create the "control" list for this pcb.
1034 * These functions will not modify mbuf chain at all.
1036 * With KAME mbuf chain restriction:
1037 * The routine will be called from upper layer handlers like tcp6_input().
1038 * Thus the routine assumes that the caller (tcp6_input) have already
1039 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1040 * very first mbuf on the mbuf chain.
1042 * ip6_savecontrol_v4 will handle those options that are possible to be
1043 * set on a v4-mapped socket.
1044 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1045 * options and handle the v6-only ones itself.
1048 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1051 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1054 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1058 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1059 SCM_TIMESTAMP, SOL_SOCKET);
1061 mp = &(*mp)->m_next;
1065 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1071 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1072 /* RFC 2292 sec. 5 */
1073 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1074 struct in6_pktinfo pi6;
1076 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1077 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1079 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1081 *mp = sbcreatecontrol((caddr_t) &pi6,
1082 sizeof(struct in6_pktinfo),
1083 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1085 mp = &(*mp)->m_next;
1088 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1089 int hlim = ip6->ip6_hlim & 0xff;
1091 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1092 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1095 mp = &(*mp)->m_next;
1104 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1106 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1109 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1113 if ((in6p->inp_flags & IN6P_TCLASS) != 0) {
1117 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1120 tclass = flowinfo & 0xff;
1121 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass),
1122 IPV6_TCLASS, IPPROTO_IPV6);
1124 mp = &(*mp)->m_next;
1128 * IPV6_HOPOPTS socket option. Recall that we required super-user
1129 * privilege for the option (see ip6_ctloutput), but it might be too
1130 * strict, since there might be some hop-by-hop options which can be
1131 * returned to normal user.
1132 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1134 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1136 * Check if a hop-by-hop options header is contatined in the
1137 * received packet, and if so, store the options as ancillary
1138 * data. Note that a hop-by-hop options header must be
1139 * just after the IPv6 header, which is assured through the
1140 * IPv6 input processing.
1142 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1143 struct ip6_hbh *hbh;
1145 #ifdef PULLDOWN_TEST
1149 #ifndef PULLDOWN_TEST
1150 hbh = (struct ip6_hbh *)(ip6 + 1);
1151 hbhlen = (hbh->ip6h_len + 1) << 3;
1153 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1156 ip6stat.ip6s_tooshort++;
1159 hbh = mtod(ext, struct ip6_hbh *);
1160 hbhlen = (hbh->ip6h_len + 1) << 3;
1161 if (hbhlen != ext->m_len) {
1163 ip6stat.ip6s_tooshort++;
1169 * XXX: We copy the whole header even if a
1170 * jumbo payload option is included, the option which
1171 * is to be removed before returning according to
1173 * Note: this constraint is removed in RFC3542
1175 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1176 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1179 mp = &(*mp)->m_next;
1180 #ifdef PULLDOWN_TEST
1186 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1187 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1190 * Search for destination options headers or routing
1191 * header(s) through the header chain, and stores each
1192 * header as ancillary data.
1193 * Note that the order of the headers remains in
1194 * the chain of ancillary data.
1196 while (1) { /* is explicit loop prevention necessary? */
1197 struct ip6_ext *ip6e = NULL;
1199 #ifdef PULLDOWN_TEST
1200 struct mbuf *ext = NULL;
1204 * if it is not an extension header, don't try to
1205 * pull it from the chain.
1208 case IPPROTO_DSTOPTS:
1209 case IPPROTO_ROUTING:
1210 case IPPROTO_HOPOPTS:
1211 case IPPROTO_AH: /* is it possible? */
1217 #ifndef PULLDOWN_TEST
1218 if (off + sizeof(*ip6e) > m->m_len)
1220 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1221 if (nxt == IPPROTO_AH)
1222 elen = (ip6e->ip6e_len + 2) << 2;
1224 elen = (ip6e->ip6e_len + 1) << 3;
1225 if (off + elen > m->m_len)
1228 ext = ip6_pullexthdr(m, off, nxt);
1230 ip6stat.ip6s_tooshort++;
1233 ip6e = mtod(ext, struct ip6_ext *);
1234 if (nxt == IPPROTO_AH)
1235 elen = (ip6e->ip6e_len + 2) << 2;
1237 elen = (ip6e->ip6e_len + 1) << 3;
1238 if (elen != ext->m_len) {
1240 ip6stat.ip6s_tooshort++;
1246 case IPPROTO_DSTOPTS:
1247 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1250 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1252 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1255 mp = &(*mp)->m_next;
1257 case IPPROTO_ROUTING:
1258 if (!in6p->inp_flags & IN6P_RTHDR)
1261 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1262 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1265 mp = &(*mp)->m_next;
1267 case IPPROTO_HOPOPTS:
1268 case IPPROTO_AH: /* is it possible? */
1273 * other cases have been filtered in the above.
1274 * none will visit this case. here we supply
1275 * the code just in case (nxt overwritten or
1278 #ifdef PULLDOWN_TEST
1285 /* proceed with the next header. */
1287 nxt = ip6e->ip6e_nxt;
1289 #ifdef PULLDOWN_TEST
1301 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1305 struct ip6_mtuinfo mtuctl;
1307 so = in6p->inp_socket;
1313 if (so == NULL) /* I believe this is impossible */
1314 panic("ip6_notify_pmtu: socket is NULL");
1317 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1318 mtuctl.ip6m_mtu = *mtu;
1319 mtuctl.ip6m_addr = *dst;
1320 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1323 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1324 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1327 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1330 /* XXX: should count statistics */
1337 #ifdef PULLDOWN_TEST
1339 * pull single extension header from mbuf chain. returns single mbuf that
1340 * contains the result, or NULL on error.
1342 static struct mbuf *
1343 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1345 struct ip6_ext ip6e;
1351 case IPPROTO_DSTOPTS:
1352 case IPPROTO_ROUTING:
1353 case IPPROTO_HOPOPTS:
1354 case IPPROTO_AH: /* is it possible? */
1357 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1361 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1362 if (nxt == IPPROTO_AH)
1363 elen = (ip6e.ip6e_len + 2) << 2;
1365 elen = (ip6e.ip6e_len + 1) << 3;
1367 MGET(n, M_DONTWAIT, MT_DATA);
1368 if (n && elen >= MLEN) {
1369 MCLGET(n, M_DONTWAIT);
1370 if ((n->m_flags & M_EXT) == 0) {
1379 if (elen >= M_TRAILINGSPACE(n)) {
1384 m_copydata(m, off, elen, mtod(n, caddr_t));
1391 * Get pointer to the previous header followed by the header
1392 * currently processed.
1393 * XXX: This function supposes that
1394 * M includes all headers,
1395 * the next header field and the header length field of each header
1397 * the sum of each header length equals to OFF.
1398 * Because of these assumptions, this function must be called very
1399 * carefully. Moreover, it will not be used in the near future when
1400 * we develop `neater' mechanism to process extension headers.
1403 ip6_get_prevhdr(struct mbuf *m, int off)
1405 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1407 if (off == sizeof(struct ip6_hdr))
1408 return (&ip6->ip6_nxt);
1411 struct ip6_ext *ip6e = NULL;
1414 len = sizeof(struct ip6_hdr);
1416 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1419 case IPPROTO_FRAGMENT:
1420 len += sizeof(struct ip6_frag);
1423 len += (ip6e->ip6e_len + 2) << 2;
1426 len += (ip6e->ip6e_len + 1) << 3;
1429 nxt = ip6e->ip6e_nxt;
1432 return (&ip6e->ip6e_nxt);
1439 * get next header offset. m will be retained.
1442 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1445 struct ip6_ext ip6e;
1450 panic("ip6_nexthdr: m == NULL");
1451 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1456 if (m->m_pkthdr.len < off + sizeof(ip6))
1458 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1460 *nxtp = ip6.ip6_nxt;
1464 case IPPROTO_FRAGMENT:
1466 * terminate parsing if it is not the first fragment,
1467 * it does not make sense to parse through it.
1469 if (m->m_pkthdr.len < off + sizeof(fh))
1471 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1472 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1473 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1476 *nxtp = fh.ip6f_nxt;
1477 off += sizeof(struct ip6_frag);
1481 if (m->m_pkthdr.len < off + sizeof(ip6e))
1483 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1485 *nxtp = ip6e.ip6e_nxt;
1486 off += (ip6e.ip6e_len + 2) << 2;
1489 case IPPROTO_HOPOPTS:
1490 case IPPROTO_ROUTING:
1491 case IPPROTO_DSTOPTS:
1492 if (m->m_pkthdr.len < off + sizeof(ip6e))
1494 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1496 *nxtp = ip6e.ip6e_nxt;
1497 off += (ip6e.ip6e_len + 1) << 3;
1502 case IPPROTO_IPCOMP:
1514 * get offset for the last header in the chain. m will be kept untainted.
1517 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1527 newoff = ip6_nexthdr(m, off, proto, nxtp);
1530 else if (newoff < off)
1531 return -1; /* invalid */
1532 else if (newoff == off)
1541 ip6_addaux(struct mbuf *m)
1545 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1547 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1550 m_tag_prepend(m, mtag);
1551 bzero(mtag + 1, sizeof(struct ip6aux));
1554 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1558 ip6_findaux(struct mbuf *m)
1562 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1563 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1567 ip6_delaux(struct mbuf *m)
1571 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1573 m_tag_delete(m, mtag);
1577 * System control for IP6
1580 u_char inet6ctlerrmap[PRC_NCMDS] = {
1582 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1583 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1584 EMSGSIZE, EHOSTUNREACH, 0, 0,