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 ip_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;
227 * should the inner packet be considered authentic?
228 * see comment in ah4_input().
229 * NB: m cannot be NULL when passed to the input routine
232 m->m_flags &= ~M_AUTHIPHDR;
233 m->m_flags &= ~M_AUTHIPDGM;
238 * make sure we don't have onion peering information into m_tag.
245 if (m->m_flags & M_EXT) {
247 ip6stat.ip6s_mext2m++;
249 ip6stat.ip6s_mext1++;
251 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
253 if (m->m_flags & M_LOOP) {
254 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
255 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
256 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
258 ip6stat.ip6s_m2m[0]++;
264 /* drop the packet if IPv6 operation is disabled on the IF */
265 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
270 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
271 ip6stat.ip6s_total++;
273 #ifndef PULLDOWN_TEST
275 * L2 bridge code and some other code can return mbuf chain
276 * that does not conform to KAME requirement. too bad.
277 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
279 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
282 MGETHDR(n, M_DONTWAIT, MT_HEADER);
285 if (n && n->m_pkthdr.len > MHLEN) {
286 MCLGET(n, M_DONTWAIT);
287 if ((n->m_flags & M_EXT) == 0) {
294 return; /* ENOBUFS */
297 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
298 n->m_len = n->m_pkthdr.len;
302 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
305 if (m->m_len < sizeof(struct ip6_hdr)) {
307 inifp = m->m_pkthdr.rcvif;
308 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
309 ip6stat.ip6s_toosmall++;
310 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
315 ip6 = mtod(m, struct ip6_hdr *);
317 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
318 ip6stat.ip6s_badvers++;
319 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
323 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
326 * Check against address spoofing/corruption.
328 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
329 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
331 * XXX: "badscope" is not very suitable for a multicast source.
333 ip6stat.ip6s_badscope++;
334 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
337 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
338 !(m->m_flags & M_LOOP)) {
340 * In this case, the packet should come from the loopback
341 * interface. However, we cannot just check the if_flags,
342 * because ip6_mloopback() passes the "actual" interface
343 * as the outgoing/incoming interface.
345 ip6stat.ip6s_badscope++;
346 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
351 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
352 /* packet is dropped by traffic conditioner */
357 * The following check is not documented in specs. A malicious
358 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
359 * and bypass security checks (act as if it was from 127.0.0.1 by using
360 * IPv6 src ::ffff:127.0.0.1). Be cautious.
362 * This check chokes if we are in an SIIT cloud. As none of BSDs
363 * support IPv4-less kernel compilation, we cannot support SIIT
364 * environment at all. So, it makes more sense for us to reject any
365 * malicious packets for non-SIIT environment, than try to do a
366 * partial support for SIIT environment.
368 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
369 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
370 ip6stat.ip6s_badscope++;
371 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
376 * Reject packets with IPv4 compatible addresses (auto tunnel).
378 * The code forbids auto tunnel relay case in RFC1933 (the check is
379 * stronger than RFC1933). We may want to re-enable it if mech-xx
380 * is revised to forbid relaying case.
382 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
383 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
384 ip6stat.ip6s_badscope++;
385 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
391 * Run through list of hooks for input packets.
393 * NB: Beware of the destination address changing
394 * (e.g. by NAT rewriting). When this happens,
395 * tell ip6_forward to do the right thing.
399 /* Jump over all PFIL processing if hooks are not active. */
400 if (!PFIL_HOOKED(&inet6_pfil_hook))
403 if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN, NULL))
405 if (m == NULL) /* consumed by filter */
407 ip6 = mtod(m, struct ip6_hdr *);
408 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
412 * Disambiguate address scope zones (if there is ambiguity).
413 * We first make sure that the original source or destination address
414 * is not in our internal form for scoped addresses. Such addresses
415 * are not necessarily invalid spec-wise, but we cannot accept them due
416 * to the usage conflict.
417 * in6_setscope() then also checks and rejects the cases where src or
418 * dst are the loopback address and the receiving interface
421 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
422 ip6stat.ip6s_badscope++; /* XXX */
425 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
426 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
427 ip6stat.ip6s_badscope++;
434 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
435 struct in6_multi *in6m = 0;
437 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
439 * See if we belong to the destination multicast group on the
442 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
445 else if (!ip6_mrouter) {
446 ip6stat.ip6s_notmember++;
447 ip6stat.ip6s_cantforward++;
448 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
451 deliverifp = m->m_pkthdr.rcvif;
458 if (ip6_forward_rt.ro_rt != NULL &&
459 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
460 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
461 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
462 ip6stat.ip6s_forward_cachehit++;
464 struct sockaddr_in6 *dst6;
466 if (ip6_forward_rt.ro_rt) {
467 /* route is down or destination is different */
468 ip6stat.ip6s_forward_cachemiss++;
469 RTFREE(ip6_forward_rt.ro_rt);
470 ip6_forward_rt.ro_rt = 0;
473 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
474 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
475 dst6->sin6_len = sizeof(struct sockaddr_in6);
476 dst6->sin6_family = AF_INET6;
477 dst6->sin6_addr = ip6->ip6_dst;
479 rtalloc((struct route *)&ip6_forward_rt);
482 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
485 * Accept the packet if the forwarding interface to the destination
486 * according to the routing table is the loopback interface,
487 * unless the associated route has a gateway.
488 * Note that this approach causes to accept a packet if there is a
489 * route to the loopback interface for the destination of the packet.
490 * But we think it's even useful in some situations, e.g. when using
491 * a special daemon which wants to intercept the packet.
493 * XXX: some OSes automatically make a cloned route for the destination
494 * of an outgoing packet. If the outgoing interface of the packet
495 * is a loopback one, the kernel would consider the packet to be
496 * accepted, even if we have no such address assinged on the interface.
497 * We check the cloned flag of the route entry to reject such cases,
498 * assuming that route entries for our own addresses are not made by
499 * cloning (it should be true because in6_addloop explicitly installs
500 * the host route). However, we might have to do an explicit check
501 * while it would be less efficient. Or, should we rather install a
502 * reject route for such a case?
504 if (ip6_forward_rt.ro_rt &&
505 (ip6_forward_rt.ro_rt->rt_flags &
506 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
508 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
511 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
515 * The check below is redundant since the comparison of
516 * the destination and the key of the rtentry has
517 * already done through looking up the routing table.
519 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
520 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
522 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
523 struct in6_ifaddr *ia6 =
524 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
527 * record address information into m_tag.
529 (void)ip6_setdstifaddr(m, ia6);
532 * packets to a tentative, duplicated, or somehow invalid
533 * address must not be accepted.
535 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
536 /* this address is ready */
538 deliverifp = ia6->ia_ifp; /* correct? */
539 /* Count the packet in the ip address stats */
540 ia6->ia_ifa.if_ipackets++;
541 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
544 char ip6bufs[INET6_ADDRSTRLEN];
545 char ip6bufd[INET6_ADDRSTRLEN];
546 /* address is not ready, so discard the packet. */
548 "ip6_input: packet to an unready address %s->%s\n",
549 ip6_sprintf(ip6bufs, &ip6->ip6_src),
550 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
557 * FAITH (Firewall Aided Internet Translator)
560 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
561 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
562 /* XXX do we need more sanity checks? */
564 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
570 * Now there is no reason to process the packet if it's not our own
571 * and we're not a router.
573 if (!ip6_forwarding) {
574 ip6stat.ip6s_cantforward++;
575 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
581 * record address information into m_tag, if we don't have one yet.
582 * note that we are unable to record it, if the address is not listed
583 * as our interface address (e.g. multicast addresses, addresses
584 * within FAITH prefixes and such).
586 if (deliverifp && !ip6_getdstifaddr(m)) {
587 struct in6_ifaddr *ia6;
589 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
591 if (!ip6_setdstifaddr(m, ia6)) {
593 * XXX maybe we should drop the packet here,
594 * as we could not provide enough information
595 * to the upper layers.
602 * Process Hop-by-Hop options header if it's contained.
603 * m may be modified in ip6_hopopts_input().
604 * If a JumboPayload option is included, plen will also be modified.
606 plen = (u_int32_t)ntohs(ip6->ip6_plen);
607 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
610 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
611 #if 0 /*touches NULL pointer*/
612 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
614 return; /* m have already been freed */
618 ip6 = mtod(m, struct ip6_hdr *);
621 * if the payload length field is 0 and the next header field
622 * indicates Hop-by-Hop Options header, then a Jumbo Payload
623 * option MUST be included.
625 if (ip6->ip6_plen == 0 && plen == 0) {
627 * Note that if a valid jumbo payload option is
628 * contained, ip6_hopopts_input() must set a valid
629 * (non-zero) payload length to the variable plen.
631 ip6stat.ip6s_badoptions++;
632 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
633 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
634 icmp6_error(m, ICMP6_PARAM_PROB,
635 ICMP6_PARAMPROB_HEADER,
636 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
639 #ifndef PULLDOWN_TEST
640 /* ip6_hopopts_input() ensures that mbuf is contiguous */
641 hbh = (struct ip6_hbh *)(ip6 + 1);
643 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
644 sizeof(struct ip6_hbh));
646 ip6stat.ip6s_tooshort++;
653 * If we are acting as a router and the packet contains a
654 * router alert option, see if we know the option value.
655 * Currently, we only support the option value for MLD, in which
656 * case we should pass the packet to the multicast routing
659 if (rtalert != ~0 && ip6_forwarding) {
661 case IP6OPT_RTALERT_MLD:
666 * RFC2711 requires unrecognized values must be
676 * Check that the amount of data in the buffers
677 * is as at least much as the IPv6 header would have us expect.
678 * Trim mbufs if longer than we expect.
679 * Drop packet if shorter than we expect.
681 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
682 ip6stat.ip6s_tooshort++;
683 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
686 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
687 if (m->m_len == m->m_pkthdr.len) {
688 m->m_len = sizeof(struct ip6_hdr) + plen;
689 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
691 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
695 * Forward if desirable.
697 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
699 * If we are acting as a multicast router, all
700 * incoming multicast packets are passed to the
701 * kernel-level multicast forwarding function.
702 * The packet is returned (relatively) intact; if
703 * ip6_mforward() returns a non-zero value, the packet
704 * must be discarded, else it may be accepted below.
706 if (ip6_mrouter && ip6_mforward &&
707 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
708 ip6stat.ip6s_cantforward++;
717 ip6_forward(m, srcrt);
721 ip6 = mtod(m, struct ip6_hdr *);
724 * Malicious party may be able to use IPv4 mapped addr to confuse
725 * tcp/udp stack and bypass security checks (act as if it was from
726 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
728 * For SIIT end node behavior, you may want to disable the check.
729 * However, you will become vulnerable to attacks using IPv4 mapped
732 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
733 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
734 ip6stat.ip6s_badscope++;
735 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
740 * Tell launch routine the next header
742 ip6stat.ip6s_delivered++;
743 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
746 while (nxt != IPPROTO_DONE) {
747 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
748 ip6stat.ip6s_toomanyhdr++;
753 * protection against faulty packet - there should be
754 * more sanity checks in header chain processing.
756 if (m->m_pkthdr.len < off) {
757 ip6stat.ip6s_tooshort++;
758 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
764 * enforce IPsec policy checking if we are seeing last header.
765 * note that we do not visit this with protocols with pcb layer
766 * code - like udp/tcp/raw ip.
768 if (ip6_ipsec_input(m, nxt))
771 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
779 * set/grab in6_ifaddr correspond to IPv6 destination address.
780 * XXX backward compatibility wrapper
782 static struct ip6aux *
783 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6)
787 ip6a = ip6_addaux(m);
789 ip6a->ip6a_dstia6 = ia6;
790 return ip6a; /* NULL if failed to set */
794 ip6_getdstifaddr(struct mbuf *m)
798 ip6a = ip6_findaux(m);
800 return ip6a->ip6a_dstia6;
806 * Hop-by-Hop options header processing. If a valid jumbo payload option is
807 * included, the real payload length will be stored in plenp.
809 * rtalertp - XXX: should be stored more smart way
812 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
813 struct mbuf **mp, int *offp)
815 struct mbuf *m = *mp;
816 int off = *offp, hbhlen;
820 /* validation of the length of the header */
821 #ifndef PULLDOWN_TEST
822 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
823 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
824 hbhlen = (hbh->ip6h_len + 1) << 3;
826 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
827 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
829 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
830 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
832 ip6stat.ip6s_tooshort++;
835 hbhlen = (hbh->ip6h_len + 1) << 3;
836 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
839 ip6stat.ip6s_tooshort++;
844 hbhlen -= sizeof(struct ip6_hbh);
845 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
847 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
848 hbhlen, rtalertp, plenp) < 0)
857 * Search header for all Hop-by-hop options and process each option.
858 * This function is separate from ip6_hopopts_input() in order to
859 * handle a case where the sending node itself process its hop-by-hop
860 * options header. In such a case, the function is called from ip6_output().
862 * The function assumes that hbh header is located right after the IPv6 header
863 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
864 * opthead + hbhlen is located in continuous memory region.
867 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
868 u_int32_t *rtalertp, u_int32_t *plenp)
872 u_int8_t *opt = opthead;
873 u_int16_t rtalert_val;
875 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
877 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
883 if (hbhlen < IP6OPT_MINLEN) {
884 ip6stat.ip6s_toosmall++;
887 optlen = *(opt + 1) + 2;
889 case IP6OPT_ROUTER_ALERT:
890 /* XXX may need check for alignment */
891 if (hbhlen < IP6OPT_RTALERT_LEN) {
892 ip6stat.ip6s_toosmall++;
895 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
897 icmp6_error(m, ICMP6_PARAM_PROB,
898 ICMP6_PARAMPROB_HEADER,
899 erroff + opt + 1 - opthead);
902 optlen = IP6OPT_RTALERT_LEN;
903 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
904 *rtalertp = ntohs(rtalert_val);
907 /* XXX may need check for alignment */
908 if (hbhlen < IP6OPT_JUMBO_LEN) {
909 ip6stat.ip6s_toosmall++;
912 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
914 icmp6_error(m, ICMP6_PARAM_PROB,
915 ICMP6_PARAMPROB_HEADER,
916 erroff + opt + 1 - opthead);
919 optlen = IP6OPT_JUMBO_LEN;
922 * IPv6 packets that have non 0 payload length
923 * must not contain a jumbo payload option.
925 ip6 = mtod(m, struct ip6_hdr *);
927 ip6stat.ip6s_badoptions++;
928 icmp6_error(m, ICMP6_PARAM_PROB,
929 ICMP6_PARAMPROB_HEADER,
930 erroff + opt - opthead);
935 * We may see jumbolen in unaligned location, so
936 * we'd need to perform bcopy().
938 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
939 jumboplen = (u_int32_t)htonl(jumboplen);
943 * if there are multiple jumbo payload options,
944 * *plenp will be non-zero and the packet will be
946 * the behavior may need some debate in ipngwg -
947 * multiple options does not make sense, however,
948 * there's no explicit mention in specification.
951 ip6stat.ip6s_badoptions++;
952 icmp6_error(m, ICMP6_PARAM_PROB,
953 ICMP6_PARAMPROB_HEADER,
954 erroff + opt + 2 - opthead);
960 * jumbo payload length must be larger than 65535.
962 if (jumboplen <= IPV6_MAXPACKET) {
963 ip6stat.ip6s_badoptions++;
964 icmp6_error(m, ICMP6_PARAM_PROB,
965 ICMP6_PARAMPROB_HEADER,
966 erroff + opt + 2 - opthead);
972 default: /* unknown option */
973 if (hbhlen < IP6OPT_MINLEN) {
974 ip6stat.ip6s_toosmall++;
977 optlen = ip6_unknown_opt(opt, m,
978 erroff + opt - opthead);
994 * Unknown option processing.
995 * The third argument `off' is the offset from the IPv6 header to the option,
996 * which is necessary if the IPv6 header the and option header and IPv6 header
997 * is not continuous in order to return an ICMPv6 error.
1000 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1002 struct ip6_hdr *ip6;
1004 switch (IP6OPT_TYPE(*optp)) {
1005 case IP6OPT_TYPE_SKIP: /* ignore the option */
1006 return ((int)*(optp + 1));
1007 case IP6OPT_TYPE_DISCARD: /* silently discard */
1010 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1011 ip6stat.ip6s_badoptions++;
1012 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1014 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1015 ip6stat.ip6s_badoptions++;
1016 ip6 = mtod(m, struct ip6_hdr *);
1017 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1018 (m->m_flags & (M_BCAST|M_MCAST)))
1021 icmp6_error(m, ICMP6_PARAM_PROB,
1022 ICMP6_PARAMPROB_OPTION, off);
1026 m_freem(m); /* XXX: NOTREACHED */
1031 * Create the "control" list for this pcb.
1032 * These functions will not modify mbuf chain at all.
1034 * With KAME mbuf chain restriction:
1035 * The routine will be called from upper layer handlers like tcp6_input().
1036 * Thus the routine assumes that the caller (tcp6_input) have already
1037 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1038 * very first mbuf on the mbuf chain.
1040 * ip6_savecontrol_v4 will handle those options that are possible to be
1041 * set on a v4-mapped socket.
1042 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1043 * options and handle the v6-only ones itself.
1046 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp)
1048 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1051 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1055 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1056 SCM_TIMESTAMP, SOL_SOCKET);
1058 mp = &(*mp)->m_next;
1062 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1065 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1066 /* RFC 2292 sec. 5 */
1067 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1068 struct in6_pktinfo pi6;
1070 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1071 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1073 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1075 *mp = sbcreatecontrol((caddr_t) &pi6,
1076 sizeof(struct in6_pktinfo),
1077 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1079 mp = &(*mp)->m_next;
1082 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1083 int hlim = ip6->ip6_hlim & 0xff;
1085 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1086 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1089 mp = &(*mp)->m_next;
1096 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1098 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1100 if (ip6_savecontrol_v4(in6p, m, mp) != 0)
1103 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1107 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1110 tclass = flowinfo & 0xff;
1111 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass),
1112 IPV6_TCLASS, IPPROTO_IPV6);
1114 mp = &(*mp)->m_next;
1118 * IPV6_HOPOPTS socket option. Recall that we required super-user
1119 * privilege for the option (see ip6_ctloutput), but it might be too
1120 * strict, since there might be some hop-by-hop options which can be
1121 * returned to normal user.
1122 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1124 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1126 * Check if a hop-by-hop options header is contatined in the
1127 * received packet, and if so, store the options as ancillary
1128 * data. Note that a hop-by-hop options header must be
1129 * just after the IPv6 header, which is assured through the
1130 * IPv6 input processing.
1132 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1133 struct ip6_hbh *hbh;
1135 #ifdef PULLDOWN_TEST
1139 #ifndef PULLDOWN_TEST
1140 hbh = (struct ip6_hbh *)(ip6 + 1);
1141 hbhlen = (hbh->ip6h_len + 1) << 3;
1143 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1146 ip6stat.ip6s_tooshort++;
1149 hbh = mtod(ext, struct ip6_hbh *);
1150 hbhlen = (hbh->ip6h_len + 1) << 3;
1151 if (hbhlen != ext->m_len) {
1153 ip6stat.ip6s_tooshort++;
1159 * XXX: We copy the whole header even if a
1160 * jumbo payload option is included, the option which
1161 * is to be removed before returning according to
1163 * Note: this constraint is removed in RFC3542
1165 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1166 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1169 mp = &(*mp)->m_next;
1170 #ifdef PULLDOWN_TEST
1176 if ((in6p->in6p_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1177 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1180 * Search for destination options headers or routing
1181 * header(s) through the header chain, and stores each
1182 * header as ancillary data.
1183 * Note that the order of the headers remains in
1184 * the chain of ancillary data.
1186 while (1) { /* is explicit loop prevention necessary? */
1187 struct ip6_ext *ip6e = NULL;
1189 #ifdef PULLDOWN_TEST
1190 struct mbuf *ext = NULL;
1194 * if it is not an extension header, don't try to
1195 * pull it from the chain.
1198 case IPPROTO_DSTOPTS:
1199 case IPPROTO_ROUTING:
1200 case IPPROTO_HOPOPTS:
1201 case IPPROTO_AH: /* is it possible? */
1207 #ifndef PULLDOWN_TEST
1208 if (off + sizeof(*ip6e) > m->m_len)
1210 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1211 if (nxt == IPPROTO_AH)
1212 elen = (ip6e->ip6e_len + 2) << 2;
1214 elen = (ip6e->ip6e_len + 1) << 3;
1215 if (off + elen > m->m_len)
1218 ext = ip6_pullexthdr(m, off, nxt);
1220 ip6stat.ip6s_tooshort++;
1223 ip6e = mtod(ext, struct ip6_ext *);
1224 if (nxt == IPPROTO_AH)
1225 elen = (ip6e->ip6e_len + 2) << 2;
1227 elen = (ip6e->ip6e_len + 1) << 3;
1228 if (elen != ext->m_len) {
1230 ip6stat.ip6s_tooshort++;
1236 case IPPROTO_DSTOPTS:
1237 if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1240 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1242 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1245 mp = &(*mp)->m_next;
1247 case IPPROTO_ROUTING:
1248 if (!in6p->in6p_flags & IN6P_RTHDR)
1251 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1252 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1255 mp = &(*mp)->m_next;
1257 case IPPROTO_HOPOPTS:
1258 case IPPROTO_AH: /* is it possible? */
1263 * other cases have been filtered in the above.
1264 * none will visit this case. here we supply
1265 * the code just in case (nxt overwritten or
1268 #ifdef PULLDOWN_TEST
1275 /* proceed with the next header. */
1277 nxt = ip6e->ip6e_nxt;
1279 #ifdef PULLDOWN_TEST
1291 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1295 struct ip6_mtuinfo mtuctl;
1297 so = in6p->inp_socket;
1303 if (so == NULL) /* I believe this is impossible */
1304 panic("ip6_notify_pmtu: socket is NULL");
1307 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1308 mtuctl.ip6m_mtu = *mtu;
1309 mtuctl.ip6m_addr = *dst;
1310 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1313 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1314 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1317 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1320 /* XXX: should count statistics */
1327 #ifdef PULLDOWN_TEST
1329 * pull single extension header from mbuf chain. returns single mbuf that
1330 * contains the result, or NULL on error.
1332 static struct mbuf *
1333 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1335 struct ip6_ext ip6e;
1341 case IPPROTO_DSTOPTS:
1342 case IPPROTO_ROUTING:
1343 case IPPROTO_HOPOPTS:
1344 case IPPROTO_AH: /* is it possible? */
1347 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1351 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1352 if (nxt == IPPROTO_AH)
1353 elen = (ip6e.ip6e_len + 2) << 2;
1355 elen = (ip6e.ip6e_len + 1) << 3;
1357 MGET(n, M_DONTWAIT, MT_DATA);
1358 if (n && elen >= MLEN) {
1359 MCLGET(n, M_DONTWAIT);
1360 if ((n->m_flags & M_EXT) == 0) {
1369 if (elen >= M_TRAILINGSPACE(n)) {
1374 m_copydata(m, off, elen, mtod(n, caddr_t));
1381 * Get pointer to the previous header followed by the header
1382 * currently processed.
1383 * XXX: This function supposes that
1384 * M includes all headers,
1385 * the next header field and the header length field of each header
1387 * the sum of each header length equals to OFF.
1388 * Because of these assumptions, this function must be called very
1389 * carefully. Moreover, it will not be used in the near future when
1390 * we develop `neater' mechanism to process extension headers.
1393 ip6_get_prevhdr(struct mbuf *m, int off)
1395 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1397 if (off == sizeof(struct ip6_hdr))
1398 return (&ip6->ip6_nxt);
1401 struct ip6_ext *ip6e = NULL;
1404 len = sizeof(struct ip6_hdr);
1406 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1409 case IPPROTO_FRAGMENT:
1410 len += sizeof(struct ip6_frag);
1413 len += (ip6e->ip6e_len + 2) << 2;
1416 len += (ip6e->ip6e_len + 1) << 3;
1419 nxt = ip6e->ip6e_nxt;
1422 return (&ip6e->ip6e_nxt);
1429 * get next header offset. m will be retained.
1432 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1435 struct ip6_ext ip6e;
1440 panic("ip6_nexthdr: m == NULL");
1441 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1446 if (m->m_pkthdr.len < off + sizeof(ip6))
1448 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1450 *nxtp = ip6.ip6_nxt;
1454 case IPPROTO_FRAGMENT:
1456 * terminate parsing if it is not the first fragment,
1457 * it does not make sense to parse through it.
1459 if (m->m_pkthdr.len < off + sizeof(fh))
1461 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1462 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1463 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1466 *nxtp = fh.ip6f_nxt;
1467 off += sizeof(struct ip6_frag);
1471 if (m->m_pkthdr.len < off + sizeof(ip6e))
1473 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1475 *nxtp = ip6e.ip6e_nxt;
1476 off += (ip6e.ip6e_len + 2) << 2;
1479 case IPPROTO_HOPOPTS:
1480 case IPPROTO_ROUTING:
1481 case IPPROTO_DSTOPTS:
1482 if (m->m_pkthdr.len < off + sizeof(ip6e))
1484 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1486 *nxtp = ip6e.ip6e_nxt;
1487 off += (ip6e.ip6e_len + 1) << 3;
1492 case IPPROTO_IPCOMP:
1504 * get offset for the last header in the chain. m will be kept untainted.
1507 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1517 newoff = ip6_nexthdr(m, off, proto, nxtp);
1520 else if (newoff < off)
1521 return -1; /* invalid */
1522 else if (newoff == off)
1531 ip6_addaux(struct mbuf *m)
1535 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1537 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1540 m_tag_prepend(m, mtag);
1541 bzero(mtag + 1, sizeof(struct ip6aux));
1544 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1548 ip6_findaux(struct mbuf *m)
1552 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1553 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1557 ip6_delaux(struct mbuf *m)
1561 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1563 m_tag_delete(m, mtag);
1567 * System control for IP6
1570 u_char inet6ctlerrmap[PRC_NCMDS] = {
1572 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1573 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1574 EMSGSIZE, EHOSTUNREACH, 0, 0,