2 /* $KAME: ip6_input.c,v 1.194 2001/05/27 13:28:35 itojun Exp $ */
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * 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;
127 extern struct ip6protosw inet6sw[];
129 u_char ip6_protox[IPPROTO_MAX];
130 static int ip6qmaxlen = IFQ_MAXLEN;
131 struct in6_ifaddr *in6_ifaddr;
133 extern struct callout in6_tmpaddrtimer_ch;
135 int ip6_forward_srcrt; /* XXX */
136 int ip6_sourcecheck; /* XXX */
137 int ip6_sourcecheck_interval; /* XXX */
139 int ip6_ours_check_algorithm;
143 ip6_fw_chk_t *ip6_fw_chk_ptr;
144 ip6_fw_ctl_t *ip6_fw_ctl_ptr;
145 int ip6_fw_enable = 1;
147 struct ip6stat ip6stat;
149 static void ip6_init2 __P((void *));
150 static struct mbuf *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *));
152 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
154 static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int));
159 * IP6 initialization: fill in IP6 protocol switch table.
160 * All protocols not implemented in kernel go to raw IP6 protocol handler.
165 struct ip6protosw *pr;
170 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
171 panic("sizeof(protosw) != sizeof(ip6protosw)");
173 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
176 for (i = 0; i < IPPROTO_MAX; i++)
177 ip6_protox[i] = pr - inet6sw;
178 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
179 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
180 if (pr->pr_domain->dom_family == PF_INET6 &&
181 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
182 ip6_protox[pr->pr_protocol] = pr - inet6sw;
183 ip6intrq.ifq_maxlen = ip6qmaxlen;
184 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", MTX_DEF);
185 ip6intrq_present = 1;
186 register_netisr(NETISR_IPV6, ip6intr);
190 * in many cases, random() here does NOT return random number
191 * as initialization during bootstrap time occur in fixed order.
194 ip6_flow_seq = random() ^ tv.tv_usec;
196 ip6_desync_factor = (random() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR;
205 * to route local address of p2p link to loopback,
206 * assign loopback address first.
208 in6_ifattach(&loif[0], NULL);
211 callout_init(&nd6_timer_ch, 0);
212 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
214 /* router renumbering prefix list maintenance */
215 callout_init(&in6_rr_timer_ch, 0);
216 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
218 /* timer for regeneranation of temporary addresses randomize ID */
219 callout_reset(&in6_tmpaddrtimer_ch,
220 (ip6_temp_preferred_lifetime - ip6_desync_factor -
221 ip6_temp_regen_advance) * hz,
222 in6_tmpaddrtimer, NULL);
226 /* This must be after route_init(), which is now SI_ORDER_THIRD */
227 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
230 * IP6 input interrupt handling. Just pass the packet to ip6_input.
240 IF_DEQUEUE(&ip6intrq, m);
248 extern struct route_in6 ip6_forward_rt;
255 int off = sizeof(struct ip6_hdr), nest;
257 u_int32_t rtalert = ~0;
259 struct ifnet *deliverifp = NULL;
261 struct packet_filter_hook *pfh;
264 #endif /* PFIL_HOOKS */
268 * should the inner packet be considered authentic?
269 * see comment in ah4_input().
272 m->m_flags &= ~M_AUTHIPHDR;
273 m->m_flags &= ~M_AUTHIPDGM;
278 * make sure we don't have onion peering information into m_aux.
283 * mbuf statistics by kazu
285 if (m->m_flags & M_EXT) {
287 ip6stat.ip6s_mext2m++;
289 ip6stat.ip6s_mext1++;
291 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
293 if (m->m_flags & M_LOOP) {
294 ip6stat.ip6s_m2m[loif[0].if_index]++; /*XXX*/
295 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
296 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
298 ip6stat.ip6s_m2m[0]++;
304 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
305 ip6stat.ip6s_total++;
307 #ifndef PULLDOWN_TEST
309 * L2 bridge code and some other code can return mbuf chain
310 * that does not conform to KAME requirement. too bad.
311 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
313 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
316 MGETHDR(n, M_DONTWAIT, MT_HEADER);
319 if (n && m->m_pkthdr.len > MHLEN) {
320 MCLGET(n, M_DONTWAIT);
321 if ((n->m_flags & M_EXT) == 0) {
331 m_copydata(m, 0, m->m_pkthdr.len, mtod(n, caddr_t));
332 n->m_len = m->m_pkthdr.len;
336 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
339 if (m->m_len < sizeof(struct ip6_hdr)) {
341 inifp = m->m_pkthdr.rcvif;
342 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
343 ip6stat.ip6s_toosmall++;
344 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
349 ip6 = mtod(m, struct ip6_hdr *);
351 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
352 ip6stat.ip6s_badvers++;
353 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
359 * Run through list of hooks for input packets. If there are any
360 * filters which require that additional packets in the flow are
361 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
362 * Note that filters must _never_ set this flag, as another filter
363 * in the list may have previously cleared it.
366 pfh = pfil_hook_get(PFIL_IN, &inet6sw[ip6_protox[IPPROTO_IPV6]].pr_pfh);
367 for (; pfh; pfh = pfh->pfil_link.tqe_next)
368 if (pfh->pfil_func) {
369 rv = pfh->pfil_func(ip6, sizeof(*ip6),
370 m->m_pkthdr.rcvif, 0, &m0);
376 ip6 = mtod(m, struct ip6_hdr *);
378 #endif /* PFIL_HOOKS */
380 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
383 * Check with the firewall...
385 if (ip6_fw_enable && ip6_fw_chk_ptr) {
387 /* If ipfw says divert, we have to just drop packet */
388 /* use port as a dummy argument */
389 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
398 * Check against address spoofing/corruption.
400 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
401 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
403 * XXX: "badscope" is not very suitable for a multicast source.
405 ip6stat.ip6s_badscope++;
406 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
409 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
410 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) &&
411 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
412 ip6stat.ip6s_badscope++;
413 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
417 * The following check is not documented in specs. A malicious
418 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
419 * and bypass security checks (act as if it was from 127.0.0.1 by using
420 * IPv6 src ::ffff:127.0.0.1). Be cautious.
422 * This check chokes if we are in an SIIT cloud. As none of BSDs
423 * support IPv4-less kernel compilation, we cannot support SIIT
424 * environment at all. So, it makes more sense for us to reject any
425 * malicious packets for non-SIIT environment, than try to do a
426 * partical support for SIIT environment.
428 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
429 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
430 ip6stat.ip6s_badscope++;
431 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
436 * Reject packets with IPv4 compatible addresses (auto tunnel).
438 * The code forbids auto tunnel relay case in RFC1933 (the check is
439 * stronger than RFC1933). We may want to re-enable it if mech-xx
440 * is revised to forbid relaying case.
442 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
443 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
444 ip6stat.ip6s_badscope++;
445 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
450 /* drop packets if interface ID portion is already filled */
451 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
452 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
453 ip6->ip6_src.s6_addr16[1]) {
454 ip6stat.ip6s_badscope++;
457 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
458 ip6->ip6_dst.s6_addr16[1]) {
459 ip6stat.ip6s_badscope++;
464 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
465 ip6->ip6_src.s6_addr16[1]
466 = htons(m->m_pkthdr.rcvif->if_index);
467 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
468 ip6->ip6_dst.s6_addr16[1]
469 = htons(m->m_pkthdr.rcvif->if_index);
471 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */
473 * We use rt->rt_ifp to determine if the address is ours or not.
474 * If rt_ifp is lo0, the address is ours.
475 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0,
476 * so any address under fe80::%lo0/64 will be mistakenly considered
477 * local. The special case is supplied to handle the case properly
478 * by actually looking at interface addresses
479 * (using in6ifa_ifpwithaddr).
481 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 &&
482 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
483 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) {
484 icmp6_error(m, ICMP6_DST_UNREACH,
485 ICMP6_DST_UNREACH_ADDR, 0);
486 /* m is already freed */
491 deliverifp = m->m_pkthdr.rcvif;
499 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
500 struct in6_multi *in6m = 0;
502 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
504 * See if we belong to the destination multicast group on the
507 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
510 else if (!ip6_mrouter) {
511 ip6stat.ip6s_notmember++;
512 ip6stat.ip6s_cantforward++;
513 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
516 deliverifp = m->m_pkthdr.rcvif;
523 switch (ip6_ours_check_algorithm) {
526 * XXX: I intentionally broke our indentation rule here,
527 * since this switch-case is just for measurement and
528 * therefore should soon be removed.
530 if (ip6_forward_rt.ro_rt != NULL &&
531 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
532 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
533 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
534 ip6stat.ip6s_forward_cachehit++;
536 struct sockaddr_in6 *dst6;
538 if (ip6_forward_rt.ro_rt) {
539 /* route is down or destination is different */
540 ip6stat.ip6s_forward_cachemiss++;
541 RTFREE(ip6_forward_rt.ro_rt);
542 ip6_forward_rt.ro_rt = 0;
545 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
546 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
547 dst6->sin6_len = sizeof(struct sockaddr_in6);
548 dst6->sin6_family = AF_INET6;
549 dst6->sin6_addr = ip6->ip6_dst;
551 ip6_forward_rt.ro_dst.sin6_scope_id =
552 in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst);
555 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
558 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
561 * Accept the packet if the forwarding interface to the destination
562 * according to the routing table is the loopback interface,
563 * unless the associated route has a gateway.
564 * Note that this approach causes to accept a packet if there is a
565 * route to the loopback interface for the destination of the packet.
566 * But we think it's even useful in some situations, e.g. when using
567 * a special daemon which wants to intercept the packet.
569 * XXX: some OSes automatically make a cloned route for the destination
570 * of an outgoing packet. If the outgoing interface of the packet
571 * is a loopback one, the kernel would consider the packet to be
572 * accepted, even if we have no such address assinged on the interface.
573 * We check the cloned flag of the route entry to reject such cases,
574 * assuming that route entries for our own addresses are not made by
575 * cloning (it should be true because in6_addloop explicitly installs
576 * the host route). However, we might have to do an explicit check
577 * while it would be less efficient. Or, should we rather install a
578 * reject route for such a case?
580 if (ip6_forward_rt.ro_rt &&
581 (ip6_forward_rt.ro_rt->rt_flags &
582 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
584 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
587 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
591 * The check below is redundant since the comparison of
592 * the destination and the key of the rtentry has
593 * already done through looking up the routing table.
595 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
596 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
598 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
599 struct in6_ifaddr *ia6 =
600 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
603 * record address information into m_aux.
605 (void)ip6_setdstifaddr(m, ia6);
608 * packets to a tentative, duplicated, or somehow invalid
609 * address must not be accepted.
611 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
612 /* this address is ready */
614 deliverifp = ia6->ia_ifp; /* correct? */
615 /* Count the packet in the ip address stats */
616 ia6->ia_ifa.if_ipackets++;
617 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
620 /* address is not ready, so discard the packet. */
622 "ip6_input: packet to an unready address %s->%s\n",
623 ip6_sprintf(&ip6->ip6_src),
624 ip6_sprintf(&ip6->ip6_dst)));
629 } /* XXX indentation (see above) */
632 * FAITH(Firewall Aided Internet Translator)
635 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
636 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
637 /* XXX do we need more sanity checks? */
639 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /*faith*/
645 * Now there is no reason to process the packet if it's not our own
646 * and we're not a router.
648 if (!ip6_forwarding) {
649 ip6stat.ip6s_cantforward++;
650 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
656 * record address information into m_aux, if we don't have one yet.
657 * note that we are unable to record it, if the address is not listed
658 * as our interface address (e.g. multicast addresses, addresses
659 * within FAITH prefixes and such).
661 if (deliverifp && !ip6_getdstifaddr(m)) {
662 struct in6_ifaddr *ia6;
664 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
666 if (!ip6_setdstifaddr(m, ia6)) {
668 * XXX maybe we should drop the packet here,
669 * as we could not provide enough information
670 * to the upper layers.
677 * Process Hop-by-Hop options header if it's contained.
678 * m may be modified in ip6_hopopts_input().
679 * If a JumboPayload option is included, plen will also be modified.
681 plen = (u_int32_t)ntohs(ip6->ip6_plen);
682 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
685 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
686 #if 0 /*touches NULL pointer*/
687 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
689 return; /* m have already been freed */
693 ip6 = mtod(m, struct ip6_hdr *);
696 * if the payload length field is 0 and the next header field
697 * indicates Hop-by-Hop Options header, then a Jumbo Payload
698 * option MUST be included.
700 if (ip6->ip6_plen == 0 && plen == 0) {
702 * Note that if a valid jumbo payload option is
703 * contained, ip6_hoptops_input() must set a valid
704 * (non-zero) payload length to the variable plen.
706 ip6stat.ip6s_badoptions++;
707 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
708 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
709 icmp6_error(m, ICMP6_PARAM_PROB,
710 ICMP6_PARAMPROB_HEADER,
711 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
714 #ifndef PULLDOWN_TEST
715 /* ip6_hopopts_input() ensures that mbuf is contiguous */
716 hbh = (struct ip6_hbh *)(ip6 + 1);
718 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
719 sizeof(struct ip6_hbh));
721 ip6stat.ip6s_tooshort++;
728 * accept the packet if a router alert option is included
729 * and we act as an IPv6 router.
731 if (rtalert != ~0 && ip6_forwarding)
737 * Check that the amount of data in the buffers
738 * is as at least much as the IPv6 header would have us expect.
739 * Trim mbufs if longer than we expect.
740 * Drop packet if shorter than we expect.
742 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
743 ip6stat.ip6s_tooshort++;
744 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
747 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
748 if (m->m_len == m->m_pkthdr.len) {
749 m->m_len = sizeof(struct ip6_hdr) + plen;
750 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
752 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
756 * Forward if desirable.
758 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
760 * If we are acting as a multicast router, all
761 * incoming multicast packets are passed to the
762 * kernel-level multicast forwarding function.
763 * The packet is returned (relatively) intact; if
764 * ip6_mforward() returns a non-zero value, the packet
765 * must be discarded, else it may be accepted below.
767 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
768 ip6stat.ip6s_cantforward++;
781 ip6 = mtod(m, struct ip6_hdr *);
784 * Malicious party may be able to use IPv4 mapped addr to confuse
785 * tcp/udp stack and bypass security checks (act as if it was from
786 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
788 * For SIIT end node behavior, you may want to disable the check.
789 * However, you will become vulnerable to attacks using IPv4 mapped
792 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
793 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
794 ip6stat.ip6s_badscope++;
795 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
800 * Tell launch routine the next header
802 ip6stat.ip6s_delivered++;
803 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
806 while (nxt != IPPROTO_DONE) {
807 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
808 ip6stat.ip6s_toomanyhdr++;
813 * protection against faulty packet - there should be
814 * more sanity checks in header chain processing.
816 if (m->m_pkthdr.len < off) {
817 ip6stat.ip6s_tooshort++;
818 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
824 * do we need to do it for every header? yeah, other
825 * functions can play with it (like re-allocate and copy).
827 mhist = ip6_addaux(m);
828 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) {
829 hist = mtod(mhist, caddr_t) + mhist->m_len;
830 bcopy(&nxt, hist, sizeof(nxt));
831 mhist->m_len += sizeof(nxt);
833 ip6stat.ip6s_toomanyhdr++;
840 * enforce IPsec policy checking if we are seeing last header.
841 * note that we do not visit this with protocols with pcb layer
842 * code - like udp/tcp/raw ip.
844 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
845 ipsec6_in_reject(m, NULL)) {
846 ipsec6stat.in_polvio++;
851 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
859 * set/grab in6_ifaddr correspond to IPv6 destination address.
860 * XXX backward compatibility wrapper
863 ip6_setdstifaddr(m, ia6)
865 struct in6_ifaddr *ia6;
871 mtod(n, struct ip6aux *)->ip6a_dstia6 = ia6;
872 return n; /* NULL if failed to set */
883 return mtod(n, struct ip6aux *)->ip6a_dstia6;
889 * Hop-by-Hop options header processing. If a valid jumbo payload option is
890 * included, the real payload length will be stored in plenp.
893 ip6_hopopts_input(plenp, rtalertp, mp, offp)
895 u_int32_t *rtalertp; /* XXX: should be stored more smart way */
899 struct mbuf *m = *mp;
900 int off = *offp, hbhlen;
904 /* validation of the length of the header */
905 #ifndef PULLDOWN_TEST
906 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
907 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
908 hbhlen = (hbh->ip6h_len + 1) << 3;
910 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
911 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
913 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
914 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
916 ip6stat.ip6s_tooshort++;
919 hbhlen = (hbh->ip6h_len + 1) << 3;
920 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
923 ip6stat.ip6s_tooshort++;
928 hbhlen -= sizeof(struct ip6_hbh);
929 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
931 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
932 hbhlen, rtalertp, plenp) < 0)
941 * Search header for all Hop-by-hop options and process each option.
942 * This function is separate from ip6_hopopts_input() in order to
943 * handle a case where the sending node itself process its hop-by-hop
944 * options header. In such a case, the function is called from ip6_output().
946 * The function assumes that hbh header is located right after the IPv6 header
947 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
948 * opthead + hbhlen is located in continuous memory region.
951 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
960 u_int8_t *opt = opthead;
961 u_int16_t rtalert_val;
963 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
965 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
971 if (hbhlen < IP6OPT_MINLEN) {
972 ip6stat.ip6s_toosmall++;
975 optlen = *(opt + 1) + 2;
978 /* XXX may need check for alignment */
979 if (hbhlen < IP6OPT_RTALERT_LEN) {
980 ip6stat.ip6s_toosmall++;
983 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
985 icmp6_error(m, ICMP6_PARAM_PROB,
986 ICMP6_PARAMPROB_HEADER,
987 erroff + opt + 1 - opthead);
990 optlen = IP6OPT_RTALERT_LEN;
991 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
992 *rtalertp = ntohs(rtalert_val);
995 /* XXX may need check for alignment */
996 if (hbhlen < IP6OPT_JUMBO_LEN) {
997 ip6stat.ip6s_toosmall++;
1000 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1002 icmp6_error(m, ICMP6_PARAM_PROB,
1003 ICMP6_PARAMPROB_HEADER,
1004 erroff + opt + 1 - opthead);
1007 optlen = IP6OPT_JUMBO_LEN;
1010 * IPv6 packets that have non 0 payload length
1011 * must not contain a jumbo payload option.
1013 ip6 = mtod(m, struct ip6_hdr *);
1014 if (ip6->ip6_plen) {
1015 ip6stat.ip6s_badoptions++;
1016 icmp6_error(m, ICMP6_PARAM_PROB,
1017 ICMP6_PARAMPROB_HEADER,
1018 erroff + opt - opthead);
1023 * We may see jumbolen in unaligned location, so
1024 * we'd need to perform bcopy().
1026 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1027 jumboplen = (u_int32_t)htonl(jumboplen);
1031 * if there are multiple jumbo payload options,
1032 * *plenp will be non-zero and the packet will be
1034 * the behavior may need some debate in ipngwg -
1035 * multiple options does not make sense, however,
1036 * there's no explicit mention in specification.
1039 ip6stat.ip6s_badoptions++;
1040 icmp6_error(m, ICMP6_PARAM_PROB,
1041 ICMP6_PARAMPROB_HEADER,
1042 erroff + opt + 2 - opthead);
1048 * jumbo payload length must be larger than 65535.
1050 if (jumboplen <= IPV6_MAXPACKET) {
1051 ip6stat.ip6s_badoptions++;
1052 icmp6_error(m, ICMP6_PARAM_PROB,
1053 ICMP6_PARAMPROB_HEADER,
1054 erroff + opt + 2 - opthead);
1060 default: /* unknown option */
1061 if (hbhlen < IP6OPT_MINLEN) {
1062 ip6stat.ip6s_toosmall++;
1065 optlen = ip6_unknown_opt(opt, m,
1066 erroff + opt - opthead);
1082 * Unknown option processing.
1083 * The third argument `off' is the offset from the IPv6 header to the option,
1084 * which is necessary if the IPv6 header the and option header and IPv6 header
1085 * is not continuous in order to return an ICMPv6 error.
1088 ip6_unknown_opt(optp, m, off)
1093 struct ip6_hdr *ip6;
1095 switch (IP6OPT_TYPE(*optp)) {
1096 case IP6OPT_TYPE_SKIP: /* ignore the option */
1097 return((int)*(optp + 1));
1098 case IP6OPT_TYPE_DISCARD: /* silently discard */
1101 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1102 ip6stat.ip6s_badoptions++;
1103 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1105 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1106 ip6stat.ip6s_badoptions++;
1107 ip6 = mtod(m, struct ip6_hdr *);
1108 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1109 (m->m_flags & (M_BCAST|M_MCAST)))
1112 icmp6_error(m, ICMP6_PARAM_PROB,
1113 ICMP6_PARAMPROB_OPTION, off);
1117 m_freem(m); /* XXX: NOTREACHED */
1122 * Create the "control" list for this pcb.
1123 * The function will not modify mbuf chain at all.
1125 * with KAME mbuf chain restriction:
1126 * The routine will be called from upper layer handlers like tcp6_input().
1127 * Thus the routine assumes that the caller (tcp6_input) have already
1128 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1129 * very first mbuf on the mbuf chain.
1132 ip6_savecontrol(in6p, mp, ip6, m)
1135 struct ip6_hdr *ip6;
1138 struct proc *p = curproc; /* XXX */
1140 int rthdr_exist = 0;
1147 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) {
1151 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1152 SCM_TIMESTAMP, SOL_SOCKET);
1154 mp = &(*mp)->m_next;
1159 /* RFC 2292 sec. 5 */
1160 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1161 struct in6_pktinfo pi6;
1162 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1163 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1164 pi6.ipi6_addr.s6_addr16[1] = 0;
1165 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1166 ? m->m_pkthdr.rcvif->if_index
1168 *mp = sbcreatecontrol((caddr_t) &pi6,
1169 sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1172 mp = &(*mp)->m_next;
1175 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) {
1176 int hlim = ip6->ip6_hlim & 0xff;
1177 *mp = sbcreatecontrol((caddr_t) &hlim,
1178 sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1180 mp = &(*mp)->m_next;
1184 * IPV6_HOPOPTS socket option. We require super-user privilege
1185 * for the option, but it might be too strict, since there might
1186 * be some hop-by-hop options which can be returned to normal user.
1187 * See RFC 2292 section 6.
1189 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1191 * Check if a hop-by-hop options header is contatined in the
1192 * received packet, and if so, store the options as ancillary
1193 * data. Note that a hop-by-hop options header must be
1194 * just after the IPv6 header, which fact is assured through
1195 * the IPv6 input processing.
1197 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1198 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1199 struct ip6_hbh *hbh;
1201 #ifdef PULLDOWN_TEST
1205 #ifndef PULLDOWN_TEST
1206 hbh = (struct ip6_hbh *)(ip6 + 1);
1207 hbhlen = (hbh->ip6h_len + 1) << 3;
1209 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1212 ip6stat.ip6s_tooshort++;
1215 hbh = mtod(ext, struct ip6_hbh *);
1216 hbhlen = (hbh->ip6h_len + 1) << 3;
1217 if (hbhlen != ext->m_len) {
1219 ip6stat.ip6s_tooshort++;
1225 * XXX: We copy whole the header even if a jumbo
1226 * payload option is included, which option is to
1227 * be removed before returning in the RFC 2292.
1228 * Note: this constraint is removed in 2292bis.
1230 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1231 IPV6_HOPOPTS, IPPROTO_IPV6);
1233 mp = &(*mp)->m_next;
1234 #ifdef PULLDOWN_TEST
1240 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1241 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1242 int proto, off, nxt;
1245 * go through the header chain to see if a routing header is
1246 * contained in the packet. We need this information to store
1247 * destination options headers (if any) properly.
1248 * XXX: performance issue. We should record this info when
1249 * processing extension headers in incoming routine.
1252 proto = IPPROTO_IPV6;
1258 newoff = ip6_nexthdr(m, off, proto, &nxt);
1261 if (newoff < off) /* invalid, check for safety */
1263 if ((proto = nxt) == IPPROTO_ROUTING) {
1271 if ((in6p->in6p_flags &
1272 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1273 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1274 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1277 * Search for destination options headers or routing
1278 * header(s) through the header chain, and stores each
1279 * header as ancillary data.
1280 * Note that the order of the headers remains in
1281 * the chain of ancillary data.
1283 while (1) { /* is explicit loop prevention necessary? */
1284 struct ip6_ext *ip6e = NULL;
1286 #ifdef PULLDOWN_TEST
1287 struct mbuf *ext = NULL;
1291 * if it is not an extension header, don't try to
1292 * pull it from the chain.
1295 case IPPROTO_DSTOPTS:
1296 case IPPROTO_ROUTING:
1297 case IPPROTO_HOPOPTS:
1298 case IPPROTO_AH: /* is it possible? */
1304 #ifndef PULLDOWN_TEST
1305 if (off + sizeof(*ip6e) > m->m_len)
1307 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1308 if (nxt == IPPROTO_AH)
1309 elen = (ip6e->ip6e_len + 2) << 2;
1311 elen = (ip6e->ip6e_len + 1) << 3;
1312 if (off + elen > m->m_len)
1315 ext = ip6_pullexthdr(m, off, nxt);
1317 ip6stat.ip6s_tooshort++;
1320 ip6e = mtod(ext, struct ip6_ext *);
1321 if (nxt == IPPROTO_AH)
1322 elen = (ip6e->ip6e_len + 2) << 2;
1324 elen = (ip6e->ip6e_len + 1) << 3;
1325 if (elen != ext->m_len) {
1327 ip6stat.ip6s_tooshort++;
1333 case IPPROTO_DSTOPTS:
1334 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0)
1338 * We also require super-user privilege for
1340 * See the comments on IN6_HOPOPTS.
1345 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1349 mp = &(*mp)->m_next;
1351 case IPPROTO_ROUTING:
1352 if (!in6p->in6p_flags & IN6P_RTHDR)
1355 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1359 mp = &(*mp)->m_next;
1361 case IPPROTO_HOPOPTS:
1362 case IPPROTO_AH: /* is it possible? */
1367 * other cases have been filtered in the above.
1368 * none will visit this case. here we supply
1369 * the code just in case (nxt overwritten or
1372 #ifdef PULLDOWN_TEST
1379 /* proceed with the next header. */
1381 nxt = ip6e->ip6e_nxt;
1383 #ifdef PULLDOWN_TEST
1394 #ifdef PULLDOWN_TEST
1396 * pull single extension header from mbuf chain. returns single mbuf that
1397 * contains the result, or NULL on error.
1399 static struct mbuf *
1400 ip6_pullexthdr(m, off, nxt)
1405 struct ip6_ext ip6e;
1411 case IPPROTO_DSTOPTS:
1412 case IPPROTO_ROUTING:
1413 case IPPROTO_HOPOPTS:
1414 case IPPROTO_AH: /* is it possible? */
1417 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1421 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1422 if (nxt == IPPROTO_AH)
1423 elen = (ip6e.ip6e_len + 2) << 2;
1425 elen = (ip6e.ip6e_len + 1) << 3;
1427 MGET(n, M_DONTWAIT, MT_DATA);
1428 if (n && elen >= MLEN) {
1429 MCLGET(n, M_DONTWAIT);
1430 if ((n->m_flags & M_EXT) == 0) {
1439 if (elen >= M_TRAILINGSPACE(n)) {
1444 m_copydata(m, off, elen, mtod(n, caddr_t));
1451 * Get pointer to the previous header followed by the header
1452 * currently processed.
1453 * XXX: This function supposes that
1454 * M includes all headers,
1455 * the next header field and the header length field of each header
1457 * the sum of each header length equals to OFF.
1458 * Because of these assumptions, this function must be called very
1459 * carefully. Moreover, it will not be used in the near future when
1460 * we develop `neater' mechanism to process extension headers.
1463 ip6_get_prevhdr(m, off)
1467 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1469 if (off == sizeof(struct ip6_hdr))
1470 return(&ip6->ip6_nxt);
1473 struct ip6_ext *ip6e = NULL;
1476 len = sizeof(struct ip6_hdr);
1478 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1481 case IPPROTO_FRAGMENT:
1482 len += sizeof(struct ip6_frag);
1485 len += (ip6e->ip6e_len + 2) << 2;
1488 len += (ip6e->ip6e_len + 1) << 3;
1491 nxt = ip6e->ip6e_nxt;
1494 return(&ip6e->ip6e_nxt);
1501 * get next header offset. m will be retained.
1504 ip6_nexthdr(m, off, proto, nxtp)
1511 struct ip6_ext ip6e;
1516 panic("ip6_nexthdr: m == NULL");
1517 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1522 if (m->m_pkthdr.len < off + sizeof(ip6))
1524 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1526 *nxtp = ip6.ip6_nxt;
1530 case IPPROTO_FRAGMENT:
1532 * terminate parsing if it is not the first fragment,
1533 * it does not make sense to parse through it.
1535 if (m->m_pkthdr.len < off + sizeof(fh))
1537 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1538 if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1541 *nxtp = fh.ip6f_nxt;
1542 off += sizeof(struct ip6_frag);
1546 if (m->m_pkthdr.len < off + sizeof(ip6e))
1548 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1550 *nxtp = ip6e.ip6e_nxt;
1551 off += (ip6e.ip6e_len + 2) << 2;
1554 case IPPROTO_HOPOPTS:
1555 case IPPROTO_ROUTING:
1556 case IPPROTO_DSTOPTS:
1557 if (m->m_pkthdr.len < off + sizeof(ip6e))
1559 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1561 *nxtp = ip6e.ip6e_nxt;
1562 off += (ip6e.ip6e_len + 1) << 3;
1567 case IPPROTO_IPCOMP:
1579 * get offset for the last header in the chain. m will be kept untainted.
1582 ip6_lasthdr(m, off, proto, nxtp)
1596 newoff = ip6_nexthdr(m, off, proto, nxtp);
1599 else if (newoff < off)
1600 return -1; /* invalid */
1601 else if (newoff == off)
1616 if (sizeof(struct ip6aux) > MHLEN)
1617 panic("assumption failed on sizeof(ip6aux)");
1619 n = m_aux_find(m, AF_INET6, -1);
1621 if (n->m_len < sizeof(struct ip6aux)) {
1622 printf("conflicting use of ip6aux");
1626 n = m_aux_add(m, AF_INET6, -1);
1627 n->m_len = sizeof(struct ip6aux);
1628 bzero(mtod(n, caddr_t), n->m_len);
1639 n = m_aux_find(m, AF_INET6, -1);
1640 if (n && n->m_len < sizeof(struct ip6aux)) {
1641 printf("conflicting use of ip6aux");
1653 n = m_aux_find(m, AF_INET6, -1);
1659 * System control for IP6
1662 u_char inet6ctlerrmap[PRC_NCMDS] = {
1664 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1665 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1666 EMSGSIZE, EHOSTUNREACH, 0, 0,