2 /* $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei 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 * Copyright (c) 1982, 1986, 1988, 1993
35 * The Regents of the University of California. All rights reserved.
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 * must display the following acknowledgement:
47 * This product includes software developed by the University of
48 * California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 * may be used to endorse or promote products derived from this software
51 * without specific prior written permission.
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
65 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
68 #include "opt_ip6fw.h"
70 #include "opt_inet6.h"
71 #include "opt_ipsec.h"
72 #include "opt_pfil_hooks.h"
74 #include <sys/param.h>
75 #include <sys/systm.h>
76 #include <sys/malloc.h>
79 #include <sys/domain.h>
80 #include <sys/protosw.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/errno.h>
85 #include <sys/kernel.h>
86 #include <sys/syslog.h>
89 #include <net/if_types.h>
90 #include <net/if_dl.h>
91 #include <net/route.h>
92 #include <net/netisr.h>
93 #include <net/intrq.h>
98 #include <netinet/in.h>
99 #include <netinet/in_systm.h>
101 #include <netinet/ip.h>
102 #include <netinet/ip_icmp.h>
104 #include <netinet/ip6.h>
105 #include <netinet6/in6_var.h>
106 #include <netinet6/ip6_var.h>
107 #include <netinet/in_pcb.h>
108 #include <netinet/icmp6.h>
109 #include <netinet6/in6_ifattach.h>
110 #include <netinet6/nd6.h>
111 #include <netinet6/in6_prefix.h>
114 #include <netinet6/ipsec.h>
116 #include <netinet6/ipsec6.h>
120 #include <netinet6/ip6_fw.h>
122 #include <netinet6/ip6protosw.h>
124 #include <net/net_osdep.h>
126 extern struct domain inet6domain;
128 u_char ip6_protox[IPPROTO_MAX];
129 static int ip6qmaxlen = IFQ_MAXLEN;
130 struct in6_ifaddr *in6_ifaddr;
132 extern struct callout in6_tmpaddrtimer_ch;
134 int ip6_forward_srcrt; /* XXX */
135 int ip6_sourcecheck; /* XXX */
136 int ip6_sourcecheck_interval; /* XXX */
138 int ip6_ours_check_algorithm;
142 ip6_fw_chk_t *ip6_fw_chk_ptr;
143 ip6_fw_ctl_t *ip6_fw_ctl_ptr;
144 int ip6_fw_enable = 1;
146 struct ip6stat ip6stat;
148 static void ip6_init2 __P((void *));
149 static struct mbuf *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *));
150 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
152 static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int));
157 * IP6 initialization: fill in IP6 protocol switch table.
158 * All protocols not implemented in kernel go to raw IP6 protocol handler.
163 struct ip6protosw *pr;
168 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
169 panic("sizeof(protosw) != sizeof(ip6protosw)");
171 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
174 for (i = 0; i < IPPROTO_MAX; i++)
175 ip6_protox[i] = pr - inet6sw;
176 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
177 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
178 if (pr->pr_domain->dom_family == PF_INET6 &&
179 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
180 ip6_protox[pr->pr_protocol] = pr - inet6sw;
181 ip6intrq.ifq_maxlen = ip6qmaxlen;
182 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF);
183 ip6intrq_present = 1;
184 register_netisr(NETISR_IPV6, ip6intr);
188 * in many cases, random() here does NOT return random number
189 * as initialization during bootstrap time occur in fixed order.
192 ip6_flow_seq = random() ^ tv.tv_usec;
194 ip6_desync_factor = (random() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR;
203 * to route local address of p2p link to loopback,
204 * assign loopback address first.
206 in6_ifattach(&loif[0], NULL);
209 callout_init(&nd6_timer_ch, 0);
210 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
212 /* router renumbering prefix list maintenance */
213 callout_init(&in6_rr_timer_ch, 0);
214 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
216 /* timer for regeneranation of temporary addresses randomize ID */
217 callout_reset(&in6_tmpaddrtimer_ch,
218 (ip6_temp_preferred_lifetime - ip6_desync_factor -
219 ip6_temp_regen_advance) * hz,
220 in6_tmpaddrtimer, NULL);
224 /* This must be after route_init(), which is now SI_ORDER_THIRD */
225 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
228 * IP6 input interrupt handling. Just pass the packet to ip6_input.
238 IF_DEQUEUE(&ip6intrq, m);
246 extern struct route_in6 ip6_forward_rt;
253 int off = sizeof(struct ip6_hdr), nest;
255 u_int32_t rtalert = ~0;
257 struct ifnet *deliverifp = NULL;
259 struct packet_filter_hook *pfh;
262 #endif /* PFIL_HOOKS */
266 * should the inner packet be considered authentic?
267 * see comment in ah4_input().
270 m->m_flags &= ~M_AUTHIPHDR;
271 m->m_flags &= ~M_AUTHIPDGM;
276 * make sure we don't have onion peering information into m_aux.
283 if (m->m_flags & M_EXT) {
285 ip6stat.ip6s_mext2m++;
287 ip6stat.ip6s_mext1++;
289 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
291 if (m->m_flags & M_LOOP) {
292 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
293 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
294 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
296 ip6stat.ip6s_m2m[0]++;
302 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
303 ip6stat.ip6s_total++;
305 #ifndef PULLDOWN_TEST
307 * L2 bridge code and some other code can return mbuf chain
308 * that does not conform to KAME requirement. too bad.
309 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
311 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
314 MGETHDR(n, M_DONTWAIT, MT_HEADER);
317 if (n && m->m_pkthdr.len > MHLEN) {
318 MCLGET(n, M_DONTWAIT);
319 if ((n->m_flags & M_EXT) == 0) {
329 m_copydata(m, 0, m->m_pkthdr.len, mtod(n, caddr_t));
330 n->m_len = m->m_pkthdr.len;
334 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
337 if (m->m_len < sizeof(struct ip6_hdr)) {
339 inifp = m->m_pkthdr.rcvif;
340 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
341 ip6stat.ip6s_toosmall++;
342 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
347 ip6 = mtod(m, struct ip6_hdr *);
349 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
350 ip6stat.ip6s_badvers++;
351 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
357 * Run through list of hooks for input packets. If there are any
358 * filters which require that additional packets in the flow are
359 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
360 * Note that filters must _never_ set this flag, as another filter
361 * in the list may have previously cleared it.
364 pfh = pfil_hook_get(PFIL_IN, &inet6sw[ip6_protox[IPPROTO_IPV6]].pr_pfh);
365 for (; pfh; pfh = pfh->pfil_link.tqe_next)
366 if (pfh->pfil_func) {
367 rv = pfh->pfil_func(ip6, sizeof(*ip6),
368 m->m_pkthdr.rcvif, 0, &m0);
374 ip6 = mtod(m, struct ip6_hdr *);
376 #endif /* PFIL_HOOKS */
378 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
381 * Check with the firewall...
383 if (ip6_fw_enable && ip6_fw_chk_ptr) {
385 /* If ipfw says divert, we have to just drop packet */
386 /* use port as a dummy argument */
387 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
396 * Check against address spoofing/corruption.
398 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
399 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
401 * XXX: "badscope" is not very suitable for a multicast source.
403 ip6stat.ip6s_badscope++;
404 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
407 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
408 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) &&
409 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
410 ip6stat.ip6s_badscope++;
411 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
416 * The following check is not documented in specs. A malicious
417 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
418 * and bypass security checks (act as if it was from 127.0.0.1 by using
419 * IPv6 src ::ffff:127.0.0.1). Be cautious.
421 * This check chokes if we are in an SIIT cloud. As none of BSDs
422 * support IPv4-less kernel compilation, we cannot support SIIT
423 * environment at all. So, it makes more sense for us to reject any
424 * malicious packets for non-SIIT environment, than try to do a
425 * partical support for SIIT environment.
427 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
428 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
429 ip6stat.ip6s_badscope++;
430 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
435 * Reject packets with IPv4 compatible addresses (auto tunnel).
437 * The code forbids auto tunnel relay case in RFC1933 (the check is
438 * stronger than RFC1933). We may want to re-enable it if mech-xx
439 * is revised to forbid relaying case.
441 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
442 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
443 ip6stat.ip6s_badscope++;
444 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
449 /* drop packets if interface ID portion is already filled */
450 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
451 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
452 ip6->ip6_src.s6_addr16[1]) {
453 ip6stat.ip6s_badscope++;
456 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
457 ip6->ip6_dst.s6_addr16[1]) {
458 ip6stat.ip6s_badscope++;
463 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
464 ip6->ip6_src.s6_addr16[1]
465 = htons(m->m_pkthdr.rcvif->if_index);
466 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
467 ip6->ip6_dst.s6_addr16[1]
468 = htons(m->m_pkthdr.rcvif->if_index);
470 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */
472 * We use rt->rt_ifp to determine if the address is ours or not.
473 * If rt_ifp is lo0, the address is ours.
474 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0,
475 * so any address under fe80::%lo0/64 will be mistakenly considered
476 * local. The special case is supplied to handle the case properly
477 * by actually looking at interface addresses
478 * (using in6ifa_ifpwithaddr).
480 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 &&
481 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
482 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) {
483 icmp6_error(m, ICMP6_DST_UNREACH,
484 ICMP6_DST_UNREACH_ADDR, 0);
485 /* m is already freed */
490 deliverifp = m->m_pkthdr.rcvif;
498 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
499 struct in6_multi *in6m = 0;
501 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
503 * See if we belong to the destination multicast group on the
506 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
509 else if (!ip6_mrouter) {
510 ip6stat.ip6s_notmember++;
511 ip6stat.ip6s_cantforward++;
512 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
515 deliverifp = m->m_pkthdr.rcvif;
522 switch (ip6_ours_check_algorithm) {
525 * XXX: I intentionally broke our indentation rule here,
526 * since this switch-case is just for measurement and
527 * therefore should soon be removed.
529 if (ip6_forward_rt.ro_rt != NULL &&
530 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
531 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
532 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
533 ip6stat.ip6s_forward_cachehit++;
535 struct sockaddr_in6 *dst6;
537 if (ip6_forward_rt.ro_rt) {
538 /* route is down or destination is different */
539 ip6stat.ip6s_forward_cachemiss++;
540 RTFREE(ip6_forward_rt.ro_rt);
541 ip6_forward_rt.ro_rt = 0;
544 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
545 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
546 dst6->sin6_len = sizeof(struct sockaddr_in6);
547 dst6->sin6_family = AF_INET6;
548 dst6->sin6_addr = ip6->ip6_dst;
550 ip6_forward_rt.ro_dst.sin6_scope_id =
551 in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst);
554 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
557 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
560 * Accept the packet if the forwarding interface to the destination
561 * according to the routing table is the loopback interface,
562 * unless the associated route has a gateway.
563 * Note that this approach causes to accept a packet if there is a
564 * route to the loopback interface for the destination of the packet.
565 * But we think it's even useful in some situations, e.g. when using
566 * a special daemon which wants to intercept the packet.
568 * XXX: some OSes automatically make a cloned route for the destination
569 * of an outgoing packet. If the outgoing interface of the packet
570 * is a loopback one, the kernel would consider the packet to be
571 * accepted, even if we have no such address assinged on the interface.
572 * We check the cloned flag of the route entry to reject such cases,
573 * assuming that route entries for our own addresses are not made by
574 * cloning (it should be true because in6_addloop explicitly installs
575 * the host route). However, we might have to do an explicit check
576 * while it would be less efficient. Or, should we rather install a
577 * reject route for such a case?
579 if (ip6_forward_rt.ro_rt &&
580 (ip6_forward_rt.ro_rt->rt_flags &
581 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
583 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
586 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
590 * The check below is redundant since the comparison of
591 * the destination and the key of the rtentry has
592 * already done through looking up the routing table.
594 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
595 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
597 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
598 struct in6_ifaddr *ia6 =
599 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
602 * record address information into m_aux.
604 (void)ip6_setdstifaddr(m, ia6);
607 * packets to a tentative, duplicated, or somehow invalid
608 * address must not be accepted.
610 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
611 /* this address is ready */
613 deliverifp = ia6->ia_ifp; /* correct? */
614 /* Count the packet in the ip address stats */
615 ia6->ia_ifa.if_ipackets++;
616 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
619 /* address is not ready, so discard the packet. */
621 "ip6_input: packet to an unready address %s->%s\n",
622 ip6_sprintf(&ip6->ip6_src),
623 ip6_sprintf(&ip6->ip6_dst)));
628 } /* XXX indentation (see above) */
631 * FAITH(Firewall Aided Internet Translator)
634 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
635 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
636 /* XXX do we need more sanity checks? */
638 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
644 * Now there is no reason to process the packet if it's not our own
645 * and we're not a router.
647 if (!ip6_forwarding) {
648 ip6stat.ip6s_cantforward++;
649 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
655 * record address information into m_aux, if we don't have one yet.
656 * note that we are unable to record it, if the address is not listed
657 * as our interface address (e.g. multicast addresses, addresses
658 * within FAITH prefixes and such).
660 if (deliverifp && !ip6_getdstifaddr(m)) {
661 struct in6_ifaddr *ia6;
663 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
665 if (!ip6_setdstifaddr(m, ia6)) {
667 * XXX maybe we should drop the packet here,
668 * as we could not provide enough information
669 * to the upper layers.
676 * Process Hop-by-Hop options header if it's contained.
677 * m may be modified in ip6_hopopts_input().
678 * If a JumboPayload option is included, plen will also be modified.
680 plen = (u_int32_t)ntohs(ip6->ip6_plen);
681 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
684 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
685 #if 0 /*touches NULL pointer*/
686 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
688 return; /* m have already been freed */
692 ip6 = mtod(m, struct ip6_hdr *);
695 * if the payload length field is 0 and the next header field
696 * indicates Hop-by-Hop Options header, then a Jumbo Payload
697 * option MUST be included.
699 if (ip6->ip6_plen == 0 && plen == 0) {
701 * Note that if a valid jumbo payload option is
702 * contained, ip6_hoptops_input() must set a valid
703 * (non-zero) payload length to the variable plen.
705 ip6stat.ip6s_badoptions++;
706 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
707 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
708 icmp6_error(m, ICMP6_PARAM_PROB,
709 ICMP6_PARAMPROB_HEADER,
710 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
713 #ifndef PULLDOWN_TEST
714 /* ip6_hopopts_input() ensures that mbuf is contiguous */
715 hbh = (struct ip6_hbh *)(ip6 + 1);
717 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
718 sizeof(struct ip6_hbh));
720 ip6stat.ip6s_tooshort++;
727 * accept the packet if a router alert option is included
728 * and we act as an IPv6 router.
730 if (rtalert != ~0 && ip6_forwarding)
736 * Check that the amount of data in the buffers
737 * is as at least much as the IPv6 header would have us expect.
738 * Trim mbufs if longer than we expect.
739 * Drop packet if shorter than we expect.
741 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
742 ip6stat.ip6s_tooshort++;
743 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
746 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
747 if (m->m_len == m->m_pkthdr.len) {
748 m->m_len = sizeof(struct ip6_hdr) + plen;
749 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
751 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
755 * Forward if desirable.
757 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
759 * If we are acting as a multicast router, all
760 * incoming multicast packets are passed to the
761 * kernel-level multicast forwarding function.
762 * The packet is returned (relatively) intact; if
763 * ip6_mforward() returns a non-zero value, the packet
764 * must be discarded, else it may be accepted below.
766 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
767 ip6stat.ip6s_cantforward++;
780 ip6 = mtod(m, struct ip6_hdr *);
783 * Malicious party may be able to use IPv4 mapped addr to confuse
784 * tcp/udp stack and bypass security checks (act as if it was from
785 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
787 * For SIIT end node behavior, you may want to disable the check.
788 * However, you will become vulnerable to attacks using IPv4 mapped
791 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
792 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
793 ip6stat.ip6s_badscope++;
794 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
799 * Tell launch routine the next header
801 ip6stat.ip6s_delivered++;
802 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
805 while (nxt != IPPROTO_DONE) {
806 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
807 ip6stat.ip6s_toomanyhdr++;
812 * protection against faulty packet - there should be
813 * more sanity checks in header chain processing.
815 if (m->m_pkthdr.len < off) {
816 ip6stat.ip6s_tooshort++;
817 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
823 * do we need to do it for every header? yeah, other
824 * functions can play with it (like re-allocate and copy).
826 mhist = ip6_addaux(m);
827 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) {
828 hist = mtod(mhist, caddr_t) + mhist->m_len;
829 bcopy(&nxt, hist, sizeof(nxt));
830 mhist->m_len += sizeof(nxt);
832 ip6stat.ip6s_toomanyhdr++;
839 * enforce IPsec policy checking if we are seeing last header.
840 * note that we do not visit this with protocols with pcb layer
841 * code - like udp/tcp/raw ip.
843 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
844 ipsec6_in_reject(m, NULL)) {
845 ipsec6stat.in_polvio++;
850 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
858 * set/grab in6_ifaddr correspond to IPv6 destination address.
859 * XXX backward compatibility wrapper
862 ip6_setdstifaddr(m, ia6)
864 struct in6_ifaddr *ia6;
870 mtod(n, struct ip6aux *)->ip6a_dstia6 = ia6;
871 return n; /* NULL if failed to set */
882 return mtod(n, struct ip6aux *)->ip6a_dstia6;
888 * Hop-by-Hop options header processing. If a valid jumbo payload option is
889 * included, the real payload length will be stored in plenp.
892 ip6_hopopts_input(plenp, rtalertp, mp, offp)
894 u_int32_t *rtalertp; /* XXX: should be stored more smart way */
898 struct mbuf *m = *mp;
899 int off = *offp, hbhlen;
903 /* validation of the length of the header */
904 #ifndef PULLDOWN_TEST
905 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
906 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
907 hbhlen = (hbh->ip6h_len + 1) << 3;
909 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
910 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
912 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
913 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
915 ip6stat.ip6s_tooshort++;
918 hbhlen = (hbh->ip6h_len + 1) << 3;
919 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
922 ip6stat.ip6s_tooshort++;
927 hbhlen -= sizeof(struct ip6_hbh);
928 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
930 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
931 hbhlen, rtalertp, plenp) < 0)
940 * Search header for all Hop-by-hop options and process each option.
941 * This function is separate from ip6_hopopts_input() in order to
942 * handle a case where the sending node itself process its hop-by-hop
943 * options header. In such a case, the function is called from ip6_output().
945 * The function assumes that hbh header is located right after the IPv6 header
946 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
947 * opthead + hbhlen is located in continuous memory region.
950 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
959 u_int8_t *opt = opthead;
960 u_int16_t rtalert_val;
962 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
964 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
970 if (hbhlen < IP6OPT_MINLEN) {
971 ip6stat.ip6s_toosmall++;
974 optlen = *(opt + 1) + 2;
977 /* XXX may need check for alignment */
978 if (hbhlen < IP6OPT_RTALERT_LEN) {
979 ip6stat.ip6s_toosmall++;
982 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
984 icmp6_error(m, ICMP6_PARAM_PROB,
985 ICMP6_PARAMPROB_HEADER,
986 erroff + opt + 1 - opthead);
989 optlen = IP6OPT_RTALERT_LEN;
990 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
991 *rtalertp = ntohs(rtalert_val);
994 /* XXX may need check for alignment */
995 if (hbhlen < IP6OPT_JUMBO_LEN) {
996 ip6stat.ip6s_toosmall++;
999 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1001 icmp6_error(m, ICMP6_PARAM_PROB,
1002 ICMP6_PARAMPROB_HEADER,
1003 erroff + opt + 1 - opthead);
1006 optlen = IP6OPT_JUMBO_LEN;
1009 * IPv6 packets that have non 0 payload length
1010 * must not contain a jumbo payload option.
1012 ip6 = mtod(m, struct ip6_hdr *);
1013 if (ip6->ip6_plen) {
1014 ip6stat.ip6s_badoptions++;
1015 icmp6_error(m, ICMP6_PARAM_PROB,
1016 ICMP6_PARAMPROB_HEADER,
1017 erroff + opt - opthead);
1022 * We may see jumbolen in unaligned location, so
1023 * we'd need to perform bcopy().
1025 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1026 jumboplen = (u_int32_t)htonl(jumboplen);
1030 * if there are multiple jumbo payload options,
1031 * *plenp will be non-zero and the packet will be
1033 * the behavior may need some debate in ipngwg -
1034 * multiple options does not make sense, however,
1035 * there's no explicit mention in specification.
1038 ip6stat.ip6s_badoptions++;
1039 icmp6_error(m, ICMP6_PARAM_PROB,
1040 ICMP6_PARAMPROB_HEADER,
1041 erroff + opt + 2 - opthead);
1047 * jumbo payload length must be larger than 65535.
1049 if (jumboplen <= IPV6_MAXPACKET) {
1050 ip6stat.ip6s_badoptions++;
1051 icmp6_error(m, ICMP6_PARAM_PROB,
1052 ICMP6_PARAMPROB_HEADER,
1053 erroff + opt + 2 - opthead);
1059 default: /* unknown option */
1060 if (hbhlen < IP6OPT_MINLEN) {
1061 ip6stat.ip6s_toosmall++;
1064 optlen = ip6_unknown_opt(opt, m,
1065 erroff + opt - opthead);
1081 * Unknown option processing.
1082 * The third argument `off' is the offset from the IPv6 header to the option,
1083 * which is necessary if the IPv6 header the and option header and IPv6 header
1084 * is not continuous in order to return an ICMPv6 error.
1087 ip6_unknown_opt(optp, m, off)
1092 struct ip6_hdr *ip6;
1094 switch (IP6OPT_TYPE(*optp)) {
1095 case IP6OPT_TYPE_SKIP: /* ignore the option */
1096 return((int)*(optp + 1));
1097 case IP6OPT_TYPE_DISCARD: /* silently discard */
1100 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1101 ip6stat.ip6s_badoptions++;
1102 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1104 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1105 ip6stat.ip6s_badoptions++;
1106 ip6 = mtod(m, struct ip6_hdr *);
1107 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1108 (m->m_flags & (M_BCAST|M_MCAST)))
1111 icmp6_error(m, ICMP6_PARAM_PROB,
1112 ICMP6_PARAMPROB_OPTION, off);
1116 m_freem(m); /* XXX: NOTREACHED */
1121 * Create the "control" list for this pcb.
1122 * The function will not modify mbuf chain at all.
1124 * with KAME mbuf chain restriction:
1125 * The routine will be called from upper layer handlers like tcp6_input().
1126 * Thus the routine assumes that the caller (tcp6_input) have already
1127 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1128 * very first mbuf on the mbuf chain.
1131 ip6_savecontrol(in6p, mp, ip6, m)
1134 struct ip6_hdr *ip6;
1137 #if __FreeBSD_version >= 500000
1138 struct thread *td = curthread; /* XXX */
1140 struct proc *td = curproc; /* XXX */
1143 int rthdr_exist = 0;
1146 if (td && !suser(td))
1150 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) {
1154 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1155 SCM_TIMESTAMP, SOL_SOCKET);
1157 mp = &(*mp)->m_next;
1162 /* RFC 2292 sec. 5 */
1163 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1164 struct in6_pktinfo pi6;
1165 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1166 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1167 pi6.ipi6_addr.s6_addr16[1] = 0;
1168 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1169 ? m->m_pkthdr.rcvif->if_index
1171 *mp = sbcreatecontrol((caddr_t) &pi6,
1172 sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1175 mp = &(*mp)->m_next;
1178 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) {
1179 int hlim = ip6->ip6_hlim & 0xff;
1180 *mp = sbcreatecontrol((caddr_t) &hlim,
1181 sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1183 mp = &(*mp)->m_next;
1187 * IPV6_HOPOPTS socket option. We require super-user privilege
1188 * for the option, but it might be too strict, since there might
1189 * be some hop-by-hop options which can be returned to normal user.
1190 * See RFC 2292 section 6.
1192 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1194 * Check if a hop-by-hop options header is contatined in the
1195 * received packet, and if so, store the options as ancillary
1196 * data. Note that a hop-by-hop options header must be
1197 * just after the IPv6 header, which fact is assured through
1198 * the IPv6 input processing.
1200 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1201 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1202 struct ip6_hbh *hbh;
1204 #ifdef PULLDOWN_TEST
1208 #ifndef PULLDOWN_TEST
1209 hbh = (struct ip6_hbh *)(ip6 + 1);
1210 hbhlen = (hbh->ip6h_len + 1) << 3;
1212 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1215 ip6stat.ip6s_tooshort++;
1218 hbh = mtod(ext, struct ip6_hbh *);
1219 hbhlen = (hbh->ip6h_len + 1) << 3;
1220 if (hbhlen != ext->m_len) {
1222 ip6stat.ip6s_tooshort++;
1228 * XXX: We copy whole the header even if a jumbo
1229 * payload option is included, which option is to
1230 * be removed before returning in the RFC 2292.
1231 * Note: this constraint is removed in 2292bis.
1233 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1234 IPV6_HOPOPTS, IPPROTO_IPV6);
1236 mp = &(*mp)->m_next;
1237 #ifdef PULLDOWN_TEST
1243 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1244 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1245 int proto, off, nxt;
1248 * go through the header chain to see if a routing header is
1249 * contained in the packet. We need this information to store
1250 * destination options headers (if any) properly.
1251 * XXX: performance issue. We should record this info when
1252 * processing extension headers in incoming routine.
1255 proto = IPPROTO_IPV6;
1261 newoff = ip6_nexthdr(m, off, proto, &nxt);
1264 if (newoff < off) /* invalid, check for safety */
1266 if ((proto = nxt) == IPPROTO_ROUTING) {
1274 if ((in6p->in6p_flags &
1275 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1276 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1277 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1280 * Search for destination options headers or routing
1281 * header(s) through the header chain, and stores each
1282 * header as ancillary data.
1283 * Note that the order of the headers remains in
1284 * the chain of ancillary data.
1286 while (1) { /* is explicit loop prevention necessary? */
1287 struct ip6_ext *ip6e = NULL;
1289 #ifdef PULLDOWN_TEST
1290 struct mbuf *ext = NULL;
1294 * if it is not an extension header, don't try to
1295 * pull it from the chain.
1298 case IPPROTO_DSTOPTS:
1299 case IPPROTO_ROUTING:
1300 case IPPROTO_HOPOPTS:
1301 case IPPROTO_AH: /* is it possible? */
1307 #ifndef PULLDOWN_TEST
1308 if (off + sizeof(*ip6e) > m->m_len)
1310 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1311 if (nxt == IPPROTO_AH)
1312 elen = (ip6e->ip6e_len + 2) << 2;
1314 elen = (ip6e->ip6e_len + 1) << 3;
1315 if (off + elen > m->m_len)
1318 ext = ip6_pullexthdr(m, off, nxt);
1320 ip6stat.ip6s_tooshort++;
1323 ip6e = mtod(ext, struct ip6_ext *);
1324 if (nxt == IPPROTO_AH)
1325 elen = (ip6e->ip6e_len + 2) << 2;
1327 elen = (ip6e->ip6e_len + 1) << 3;
1328 if (elen != ext->m_len) {
1330 ip6stat.ip6s_tooshort++;
1336 case IPPROTO_DSTOPTS:
1337 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0)
1341 * We also require super-user privilege for
1343 * See the comments on IN6_HOPOPTS.
1348 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1352 mp = &(*mp)->m_next;
1354 case IPPROTO_ROUTING:
1355 if (!in6p->in6p_flags & IN6P_RTHDR)
1358 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1362 mp = &(*mp)->m_next;
1364 case IPPROTO_HOPOPTS:
1365 case IPPROTO_AH: /* is it possible? */
1370 * other cases have been filtered in the above.
1371 * none will visit this case. here we supply
1372 * the code just in case (nxt overwritten or
1375 #ifdef PULLDOWN_TEST
1382 /* proceed with the next header. */
1384 nxt = ip6e->ip6e_nxt;
1386 #ifdef PULLDOWN_TEST
1397 #ifdef PULLDOWN_TEST
1399 * pull single extension header from mbuf chain. returns single mbuf that
1400 * contains the result, or NULL on error.
1402 static struct mbuf *
1403 ip6_pullexthdr(m, off, nxt)
1408 struct ip6_ext ip6e;
1414 case IPPROTO_DSTOPTS:
1415 case IPPROTO_ROUTING:
1416 case IPPROTO_HOPOPTS:
1417 case IPPROTO_AH: /* is it possible? */
1420 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1424 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1425 if (nxt == IPPROTO_AH)
1426 elen = (ip6e.ip6e_len + 2) << 2;
1428 elen = (ip6e.ip6e_len + 1) << 3;
1430 MGET(n, M_DONTWAIT, MT_DATA);
1431 if (n && elen >= MLEN) {
1432 MCLGET(n, M_DONTWAIT);
1433 if ((n->m_flags & M_EXT) == 0) {
1442 if (elen >= M_TRAILINGSPACE(n)) {
1447 m_copydata(m, off, elen, mtod(n, caddr_t));
1454 * Get pointer to the previous header followed by the header
1455 * currently processed.
1456 * XXX: This function supposes that
1457 * M includes all headers,
1458 * the next header field and the header length field of each header
1460 * the sum of each header length equals to OFF.
1461 * Because of these assumptions, this function must be called very
1462 * carefully. Moreover, it will not be used in the near future when
1463 * we develop `neater' mechanism to process extension headers.
1466 ip6_get_prevhdr(m, off)
1470 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1472 if (off == sizeof(struct ip6_hdr))
1473 return(&ip6->ip6_nxt);
1476 struct ip6_ext *ip6e = NULL;
1479 len = sizeof(struct ip6_hdr);
1481 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1484 case IPPROTO_FRAGMENT:
1485 len += sizeof(struct ip6_frag);
1488 len += (ip6e->ip6e_len + 2) << 2;
1491 len += (ip6e->ip6e_len + 1) << 3;
1494 nxt = ip6e->ip6e_nxt;
1497 return(&ip6e->ip6e_nxt);
1504 * get next header offset. m will be retained.
1507 ip6_nexthdr(m, off, proto, nxtp)
1514 struct ip6_ext ip6e;
1519 panic("ip6_nexthdr: m == NULL");
1520 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1525 if (m->m_pkthdr.len < off + sizeof(ip6))
1527 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1529 *nxtp = ip6.ip6_nxt;
1533 case IPPROTO_FRAGMENT:
1535 * terminate parsing if it is not the first fragment,
1536 * it does not make sense to parse through it.
1538 if (m->m_pkthdr.len < off + sizeof(fh))
1540 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1541 if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1544 *nxtp = fh.ip6f_nxt;
1545 off += sizeof(struct ip6_frag);
1549 if (m->m_pkthdr.len < off + sizeof(ip6e))
1551 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1553 *nxtp = ip6e.ip6e_nxt;
1554 off += (ip6e.ip6e_len + 2) << 2;
1557 case IPPROTO_HOPOPTS:
1558 case IPPROTO_ROUTING:
1559 case IPPROTO_DSTOPTS:
1560 if (m->m_pkthdr.len < off + sizeof(ip6e))
1562 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1564 *nxtp = ip6e.ip6e_nxt;
1565 off += (ip6e.ip6e_len + 1) << 3;
1570 case IPPROTO_IPCOMP:
1582 * get offset for the last header in the chain. m will be kept untainted.
1585 ip6_lasthdr(m, off, proto, nxtp)
1599 newoff = ip6_nexthdr(m, off, proto, nxtp);
1602 else if (newoff < off)
1603 return -1; /* invalid */
1604 else if (newoff == off)
1619 if (sizeof(struct ip6aux) > MHLEN)
1620 panic("assumption failed on sizeof(ip6aux)");
1622 n = m_aux_find(m, AF_INET6, -1);
1624 if (n->m_len < sizeof(struct ip6aux)) {
1625 printf("conflicting use of ip6aux");
1629 n = m_aux_add(m, AF_INET6, -1);
1630 n->m_len = sizeof(struct ip6aux);
1631 bzero(mtod(n, caddr_t), n->m_len);
1642 n = m_aux_find(m, AF_INET6, -1);
1643 if (n && n->m_len < sizeof(struct ip6aux)) {
1644 printf("conflicting use of ip6aux");
1656 n = m_aux_find(m, AF_INET6, -1);
1662 * System control for IP6
1665 u_char inet6ctlerrmap[PRC_NCMDS] = {
1667 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1668 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1669 EMSGSIZE, EHOSTUNREACH, 0, 0,