1 /* $FreeBSD: src/sys/netinet6/ip6_input.c,v 1.11.2.15 2003/01/24 05:11:35 sam Exp $ */
2 /* $DragonFly: src/sys/netinet6/ip6_input.c,v 1.2 2003/06/17 04:28:52 dillon Exp $ */
3 /* $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ */
6 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the name of the project nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * Copyright (c) 1982, 1986, 1988, 1993
36 * The Regents of the University of California. All rights reserved.
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. All advertising materials mentioning features or use of this software
47 * must display the following acknowledgement:
48 * This product includes software developed by the University of
49 * California, Berkeley and its contributors.
50 * 4. Neither the name of the University nor the names of its contributors
51 * may be used to endorse or promote products derived from this software
52 * without specific prior written permission.
54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
66 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
69 #include "opt_ip6fw.h"
71 #include "opt_inet6.h"
72 #include "opt_ipsec.h"
74 #include <sys/param.h>
75 #include <sys/systm.h>
76 #include <sys/malloc.h>
78 #include <sys/domain.h>
79 #include <sys/protosw.h>
80 #include <sys/socket.h>
81 #include <sys/socketvar.h>
82 #include <sys/errno.h>
84 #include <sys/kernel.h>
85 #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>
95 #include <netinet/in.h>
96 #include <netinet/in_systm.h>
98 #include <netinet/ip.h>
99 #include <netinet/ip_icmp.h>
101 #include <netinet/ip6.h>
102 #include <netinet6/in6_var.h>
103 #include <netinet6/ip6_var.h>
104 #include <netinet/in_pcb.h>
105 #include <netinet/icmp6.h>
106 #include <netinet6/in6_ifattach.h>
107 #include <netinet6/nd6.h>
108 #include <netinet6/in6_prefix.h>
111 #include <netinet6/ipsec.h>
113 #include <netinet6/ipsec6.h>
118 #include <netipsec/ipsec.h>
119 #include <netipsec/ipsec6.h>
121 #endif /* FAST_IPSEC */
123 #include <netinet6/ip6_fw.h>
125 #include <netinet6/ip6protosw.h>
127 /* we need it for NLOOP. */
130 #include <net/net_osdep.h>
132 extern int (*fr_checkp) __P((struct ip *, int, struct ifnet *, int, struct mbuf **));
134 extern struct domain inet6domain;
135 extern struct ip6protosw inet6sw[];
137 u_char ip6_protox[IPPROTO_MAX];
138 static int ip6qmaxlen = IFQ_MAXLEN;
139 struct in6_ifaddr *in6_ifaddr;
141 extern struct callout in6_tmpaddrtimer_ch;
143 int ip6_forward_srcrt; /* XXX */
144 int ip6_sourcecheck; /* XXX */
145 int ip6_sourcecheck_interval; /* XXX */
146 const int int6intrq_present = 1;
148 int ip6_ours_check_algorithm;
152 ip6_fw_chk_t *ip6_fw_chk_ptr;
153 ip6_fw_ctl_t *ip6_fw_ctl_ptr;
154 int ip6_fw_enable = 1;
156 struct ip6stat ip6stat;
158 static void ip6_init2 __P((void *));
159 static struct ip6aux *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *));
160 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
162 static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int));
167 * IP6 initialization: fill in IP6 protocol switch table.
168 * All protocols not implemented in kernel go to raw IP6 protocol handler.
173 struct ip6protosw *pr;
178 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
179 panic("sizeof(protosw) != sizeof(ip6protosw)");
181 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
184 for (i = 0; i < IPPROTO_MAX; i++)
185 ip6_protox[i] = pr - inet6sw;
186 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
187 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
188 if (pr->pr_domain->dom_family == PF_INET6 &&
189 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
190 ip6_protox[pr->pr_protocol] = pr - inet6sw;
191 ip6intrq.ifq_maxlen = ip6qmaxlen;
192 register_netisr(NETISR_IPV6, ip6intr);
196 * in many cases, random() here does NOT return random number
197 * as initialization during bootstrap time occur in fixed order.
200 ip6_flow_seq = random() ^ tv.tv_usec;
202 ip6_desync_factor = (random() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR;
211 * to route local address of p2p link to loopback,
212 * assign loopback address first.
214 in6_ifattach(&loif[0], NULL);
217 callout_init(&nd6_timer_ch);
218 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
220 /* router renumbering prefix list maintenance */
221 callout_init(&in6_rr_timer_ch);
222 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
224 /* timer for regeneranation of temporary addresses randomize ID */
225 callout_reset(&in6_tmpaddrtimer_ch,
226 (ip6_temp_preferred_lifetime - ip6_desync_factor -
227 ip6_temp_regen_advance) * hz,
228 in6_tmpaddrtimer, NULL);
232 /* This must be after route_init(), which is now SI_ORDER_THIRD */
233 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
236 * IP6 input interrupt handling. Just pass the packet to ip6_input.
246 IF_DEQUEUE(&ip6intrq, m);
254 extern struct route_in6 ip6_forward_rt;
261 int off = sizeof(struct ip6_hdr), nest;
263 u_int32_t rtalert = ~0;
265 struct ifnet *deliverifp = NULL;
269 * should the inner packet be considered authentic?
270 * see comment in ah4_input().
273 m->m_flags &= ~M_AUTHIPHDR;
274 m->m_flags &= ~M_AUTHIPDGM;
279 * make sure we don't have onion peering information into m_aux.
286 if (m->m_flags & M_EXT) {
288 ip6stat.ip6s_mext2m++;
290 ip6stat.ip6s_mext1++;
292 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
294 if (m->m_flags & M_LOOP) {
295 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
296 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
297 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
299 ip6stat.ip6s_m2m[0]++;
305 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
306 ip6stat.ip6s_total++;
308 #ifndef PULLDOWN_TEST
310 * L2 bridge code and some other code can return mbuf chain
311 * that does not conform to KAME requirement. too bad.
312 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
314 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
317 MGETHDR(n, M_DONTWAIT, MT_HEADER);
320 if (n && n->m_pkthdr.len > MHLEN) {
321 MCLGET(n, M_DONTWAIT);
322 if ((n->m_flags & M_EXT) == 0) {
332 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
333 n->m_len = n->m_pkthdr.len;
337 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
340 if (m->m_len < sizeof(struct ip6_hdr)) {
342 inifp = m->m_pkthdr.rcvif;
343 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
344 ip6stat.ip6s_toosmall++;
345 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
350 ip6 = mtod(m, struct ip6_hdr *);
352 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
353 ip6stat.ip6s_badvers++;
354 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
359 * Check if we want to allow this packet to be processed.
360 * Consider it to be bad if not.
365 if ((*fr_checkp)((struct ip *)ip6, sizeof(*ip6),
366 m->m_pkthdr.rcvif, 0, &m1) != 0)
371 ip6 = mtod(m, struct ip6_hdr *);
374 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
377 * Check with the firewall...
379 if (ip6_fw_enable && ip6_fw_chk_ptr) {
381 /* If ipfw says divert, we have to just drop packet */
382 /* use port as a dummy argument */
383 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
392 * Check against address spoofing/corruption.
394 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
395 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
397 * XXX: "badscope" is not very suitable for a multicast source.
399 ip6stat.ip6s_badscope++;
400 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
403 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
404 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) &&
405 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
406 ip6stat.ip6s_badscope++;
407 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
412 * The following check is not documented in specs. A malicious
413 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
414 * and bypass security checks (act as if it was from 127.0.0.1 by using
415 * IPv6 src ::ffff:127.0.0.1). Be cautious.
417 * This check chokes if we are in an SIIT cloud. As none of BSDs
418 * support IPv4-less kernel compilation, we cannot support SIIT
419 * environment at all. So, it makes more sense for us to reject any
420 * malicious packets for non-SIIT environment, than try to do a
421 * partical support for SIIT environment.
423 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
424 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
425 ip6stat.ip6s_badscope++;
426 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
431 * Reject packets with IPv4 compatible addresses (auto tunnel).
433 * The code forbids auto tunnel relay case in RFC1933 (the check is
434 * stronger than RFC1933). We may want to re-enable it if mech-xx
435 * is revised to forbid relaying case.
437 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
438 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
439 ip6stat.ip6s_badscope++;
440 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
445 /* drop packets if interface ID portion is already filled */
446 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
447 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
448 ip6->ip6_src.s6_addr16[1]) {
449 ip6stat.ip6s_badscope++;
452 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
453 ip6->ip6_dst.s6_addr16[1]) {
454 ip6stat.ip6s_badscope++;
459 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
460 ip6->ip6_src.s6_addr16[1]
461 = htons(m->m_pkthdr.rcvif->if_index);
462 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
463 ip6->ip6_dst.s6_addr16[1]
464 = htons(m->m_pkthdr.rcvif->if_index);
466 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */
468 * We use rt->rt_ifp to determine if the address is ours or not.
469 * If rt_ifp is lo0, the address is ours.
470 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0,
471 * so any address under fe80::%lo0/64 will be mistakenly considered
472 * local. The special case is supplied to handle the case properly
473 * by actually looking at interface addresses
474 * (using in6ifa_ifpwithaddr).
476 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 &&
477 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
478 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) {
479 icmp6_error(m, ICMP6_DST_UNREACH,
480 ICMP6_DST_UNREACH_ADDR, 0);
481 /* m is already freed */
486 deliverifp = m->m_pkthdr.rcvif;
494 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
495 struct in6_multi *in6m = 0;
497 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
499 * See if we belong to the destination multicast group on the
502 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
505 else if (!ip6_mrouter) {
506 ip6stat.ip6s_notmember++;
507 ip6stat.ip6s_cantforward++;
508 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
511 deliverifp = m->m_pkthdr.rcvif;
518 switch (ip6_ours_check_algorithm) {
521 * XXX: I intentionally broke our indentation rule here,
522 * since this switch-case is just for measurement and
523 * therefore should soon be removed.
525 if (ip6_forward_rt.ro_rt != NULL &&
526 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
527 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
528 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
529 ip6stat.ip6s_forward_cachehit++;
531 struct sockaddr_in6 *dst6;
533 if (ip6_forward_rt.ro_rt) {
534 /* route is down or destination is different */
535 ip6stat.ip6s_forward_cachemiss++;
536 RTFREE(ip6_forward_rt.ro_rt);
537 ip6_forward_rt.ro_rt = 0;
540 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
541 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
542 dst6->sin6_len = sizeof(struct sockaddr_in6);
543 dst6->sin6_family = AF_INET6;
544 dst6->sin6_addr = ip6->ip6_dst;
546 ip6_forward_rt.ro_dst.sin6_scope_id =
547 in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst);
550 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
553 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
556 * Accept the packet if the forwarding interface to the destination
557 * according to the routing table is the loopback interface,
558 * unless the associated route has a gateway.
559 * Note that this approach causes to accept a packet if there is a
560 * route to the loopback interface for the destination of the packet.
561 * But we think it's even useful in some situations, e.g. when using
562 * a special daemon which wants to intercept the packet.
564 * XXX: some OSes automatically make a cloned route for the destination
565 * of an outgoing packet. If the outgoing interface of the packet
566 * is a loopback one, the kernel would consider the packet to be
567 * accepted, even if we have no such address assinged on the interface.
568 * We check the cloned flag of the route entry to reject such cases,
569 * assuming that route entries for our own addresses are not made by
570 * cloning (it should be true because in6_addloop explicitly installs
571 * the host route). However, we might have to do an explicit check
572 * while it would be less efficient. Or, should we rather install a
573 * reject route for such a case?
575 if (ip6_forward_rt.ro_rt &&
576 (ip6_forward_rt.ro_rt->rt_flags &
577 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
579 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
582 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
586 * The check below is redundant since the comparison of
587 * the destination and the key of the rtentry has
588 * already done through looking up the routing table.
590 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
591 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
593 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
594 struct in6_ifaddr *ia6 =
595 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
598 * record address information into m_aux.
600 (void)ip6_setdstifaddr(m, ia6);
603 * packets to a tentative, duplicated, or somehow invalid
604 * address must not be accepted.
606 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
607 /* this address is ready */
609 deliverifp = ia6->ia_ifp; /* correct? */
610 /* Count the packet in the ip address stats */
611 ia6->ia_ifa.if_ipackets++;
612 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
615 /* address is not ready, so discard the packet. */
617 "ip6_input: packet to an unready address %s->%s\n",
618 ip6_sprintf(&ip6->ip6_src),
619 ip6_sprintf(&ip6->ip6_dst)));
624 } /* XXX indentation (see above) */
627 * FAITH(Firewall Aided Internet Translator)
630 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
631 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
632 /* XXX do we need more sanity checks? */
634 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
640 * Now there is no reason to process the packet if it's not our own
641 * and we're not a router.
643 if (!ip6_forwarding) {
644 ip6stat.ip6s_cantforward++;
645 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
651 * record address information into m_aux, if we don't have one yet.
652 * note that we are unable to record it, if the address is not listed
653 * as our interface address (e.g. multicast addresses, addresses
654 * within FAITH prefixes and such).
656 if (deliverifp && !ip6_getdstifaddr(m)) {
657 struct in6_ifaddr *ia6;
659 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
661 if (!ip6_setdstifaddr(m, ia6)) {
663 * XXX maybe we should drop the packet here,
664 * as we could not provide enough information
665 * to the upper layers.
672 * Process Hop-by-Hop options header if it's contained.
673 * m may be modified in ip6_hopopts_input().
674 * If a JumboPayload option is included, plen will also be modified.
676 plen = (u_int32_t)ntohs(ip6->ip6_plen);
677 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
680 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
681 #if 0 /*touches NULL pointer*/
682 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
684 return; /* m have already been freed */
688 ip6 = mtod(m, struct ip6_hdr *);
691 * if the payload length field is 0 and the next header field
692 * indicates Hop-by-Hop Options header, then a Jumbo Payload
693 * option MUST be included.
695 if (ip6->ip6_plen == 0 && plen == 0) {
697 * Note that if a valid jumbo payload option is
698 * contained, ip6_hoptops_input() must set a valid
699 * (non-zero) payload length to the variable plen.
701 ip6stat.ip6s_badoptions++;
702 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
703 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
704 icmp6_error(m, ICMP6_PARAM_PROB,
705 ICMP6_PARAMPROB_HEADER,
706 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
709 #ifndef PULLDOWN_TEST
710 /* ip6_hopopts_input() ensures that mbuf is contiguous */
711 hbh = (struct ip6_hbh *)(ip6 + 1);
713 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
714 sizeof(struct ip6_hbh));
716 ip6stat.ip6s_tooshort++;
723 * accept the packet if a router alert option is included
724 * and we act as an IPv6 router.
726 if (rtalert != ~0 && ip6_forwarding)
732 * Check that the amount of data in the buffers
733 * is as at least much as the IPv6 header would have us expect.
734 * Trim mbufs if longer than we expect.
735 * Drop packet if shorter than we expect.
737 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
738 ip6stat.ip6s_tooshort++;
739 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
742 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
743 if (m->m_len == m->m_pkthdr.len) {
744 m->m_len = sizeof(struct ip6_hdr) + plen;
745 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
747 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
751 * Forward if desirable.
753 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
755 * If we are acting as a multicast router, all
756 * incoming multicast packets are passed to the
757 * kernel-level multicast forwarding function.
758 * The packet is returned (relatively) intact; if
759 * ip6_mforward() returns a non-zero value, the packet
760 * must be discarded, else it may be accepted below.
762 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
763 ip6stat.ip6s_cantforward++;
776 ip6 = mtod(m, struct ip6_hdr *);
779 * Malicious party may be able to use IPv4 mapped addr to confuse
780 * tcp/udp stack and bypass security checks (act as if it was from
781 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
783 * For SIIT end node behavior, you may want to disable the check.
784 * However, you will become vulnerable to attacks using IPv4 mapped
787 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
788 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
789 ip6stat.ip6s_badscope++;
790 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
795 * Tell launch routine the next header
797 ip6stat.ip6s_delivered++;
798 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
801 while (nxt != IPPROTO_DONE) {
802 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
803 ip6stat.ip6s_toomanyhdr++;
808 * protection against faulty packet - there should be
809 * more sanity checks in header chain processing.
811 if (m->m_pkthdr.len < off) {
812 ip6stat.ip6s_tooshort++;
813 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
819 * do we need to do it for every header? yeah, other
820 * functions can play with it (like re-allocate and copy).
822 mhist = ip6_addaux(m);
823 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) {
824 hist = mtod(mhist, caddr_t) + mhist->m_len;
825 bcopy(&nxt, hist, sizeof(nxt));
826 mhist->m_len += sizeof(nxt);
828 ip6stat.ip6s_toomanyhdr++;
835 * enforce IPsec policy checking if we are seeing last header.
836 * note that we do not visit this with protocols with pcb layer
837 * code - like udp/tcp/raw ip.
839 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
840 ipsec6_in_reject(m, NULL)) {
841 ipsec6stat.in_polvio++;
846 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
854 * set/grab in6_ifaddr correspond to IPv6 destination address.
855 * XXX backward compatibility wrapper
857 static struct ip6aux *
858 ip6_setdstifaddr(m, ia6)
860 struct in6_ifaddr *ia6;
866 n->ip6a_dstia6 = ia6;
867 return n; /* NULL if failed to set */
878 return n->ip6a_dstia6;
884 * Hop-by-Hop options header processing. If a valid jumbo payload option is
885 * included, the real payload length will be stored in plenp.
888 ip6_hopopts_input(plenp, rtalertp, mp, offp)
890 u_int32_t *rtalertp; /* XXX: should be stored more smart way */
894 struct mbuf *m = *mp;
895 int off = *offp, hbhlen;
899 /* validation of the length of the header */
900 #ifndef PULLDOWN_TEST
901 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
902 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
903 hbhlen = (hbh->ip6h_len + 1) << 3;
905 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
906 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
908 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
909 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
911 ip6stat.ip6s_tooshort++;
914 hbhlen = (hbh->ip6h_len + 1) << 3;
915 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
918 ip6stat.ip6s_tooshort++;
923 hbhlen -= sizeof(struct ip6_hbh);
924 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
926 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
927 hbhlen, rtalertp, plenp) < 0)
936 * Search header for all Hop-by-hop options and process each option.
937 * This function is separate from ip6_hopopts_input() in order to
938 * handle a case where the sending node itself process its hop-by-hop
939 * options header. In such a case, the function is called from ip6_output().
941 * The function assumes that hbh header is located right after the IPv6 header
942 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
943 * opthead + hbhlen is located in continuous memory region.
946 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
955 u_int8_t *opt = opthead;
956 u_int16_t rtalert_val;
958 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
960 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
966 if (hbhlen < IP6OPT_MINLEN) {
967 ip6stat.ip6s_toosmall++;
970 optlen = *(opt + 1) + 2;
973 /* XXX may need check for alignment */
974 if (hbhlen < IP6OPT_RTALERT_LEN) {
975 ip6stat.ip6s_toosmall++;
978 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
980 icmp6_error(m, ICMP6_PARAM_PROB,
981 ICMP6_PARAMPROB_HEADER,
982 erroff + opt + 1 - opthead);
985 optlen = IP6OPT_RTALERT_LEN;
986 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
987 *rtalertp = ntohs(rtalert_val);
990 /* XXX may need check for alignment */
991 if (hbhlen < IP6OPT_JUMBO_LEN) {
992 ip6stat.ip6s_toosmall++;
995 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
997 icmp6_error(m, ICMP6_PARAM_PROB,
998 ICMP6_PARAMPROB_HEADER,
999 erroff + opt + 1 - opthead);
1002 optlen = IP6OPT_JUMBO_LEN;
1005 * IPv6 packets that have non 0 payload length
1006 * must not contain a jumbo payload option.
1008 ip6 = mtod(m, struct ip6_hdr *);
1009 if (ip6->ip6_plen) {
1010 ip6stat.ip6s_badoptions++;
1011 icmp6_error(m, ICMP6_PARAM_PROB,
1012 ICMP6_PARAMPROB_HEADER,
1013 erroff + opt - opthead);
1018 * We may see jumbolen in unaligned location, so
1019 * we'd need to perform bcopy().
1021 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1022 jumboplen = (u_int32_t)htonl(jumboplen);
1026 * if there are multiple jumbo payload options,
1027 * *plenp will be non-zero and the packet will be
1029 * the behavior may need some debate in ipngwg -
1030 * multiple options does not make sense, however,
1031 * there's no explicit mention in specification.
1034 ip6stat.ip6s_badoptions++;
1035 icmp6_error(m, ICMP6_PARAM_PROB,
1036 ICMP6_PARAMPROB_HEADER,
1037 erroff + opt + 2 - opthead);
1043 * jumbo payload length must be larger than 65535.
1045 if (jumboplen <= IPV6_MAXPACKET) {
1046 ip6stat.ip6s_badoptions++;
1047 icmp6_error(m, ICMP6_PARAM_PROB,
1048 ICMP6_PARAMPROB_HEADER,
1049 erroff + opt + 2 - opthead);
1055 default: /* unknown option */
1056 if (hbhlen < IP6OPT_MINLEN) {
1057 ip6stat.ip6s_toosmall++;
1060 optlen = ip6_unknown_opt(opt, m,
1061 erroff + opt - opthead);
1077 * Unknown option processing.
1078 * The third argument `off' is the offset from the IPv6 header to the option,
1079 * which is necessary if the IPv6 header the and option header and IPv6 header
1080 * is not continuous in order to return an ICMPv6 error.
1083 ip6_unknown_opt(optp, m, off)
1088 struct ip6_hdr *ip6;
1090 switch (IP6OPT_TYPE(*optp)) {
1091 case IP6OPT_TYPE_SKIP: /* ignore the option */
1092 return((int)*(optp + 1));
1093 case IP6OPT_TYPE_DISCARD: /* silently discard */
1096 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1097 ip6stat.ip6s_badoptions++;
1098 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1100 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1101 ip6stat.ip6s_badoptions++;
1102 ip6 = mtod(m, struct ip6_hdr *);
1103 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1104 (m->m_flags & (M_BCAST|M_MCAST)))
1107 icmp6_error(m, ICMP6_PARAM_PROB,
1108 ICMP6_PARAMPROB_OPTION, off);
1112 m_freem(m); /* XXX: NOTREACHED */
1117 * Create the "control" list for this pcb.
1118 * The function will not modify mbuf chain at all.
1120 * with KAME mbuf chain restriction:
1121 * The routine will be called from upper layer handlers like tcp6_input().
1122 * Thus the routine assumes that the caller (tcp6_input) have already
1123 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1124 * very first mbuf on the mbuf chain.
1127 ip6_savecontrol(in6p, mp, ip6, m)
1130 struct ip6_hdr *ip6;
1133 struct proc *p = curproc; /* XXX */
1135 int rthdr_exist = 0;
1142 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) {
1146 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1147 SCM_TIMESTAMP, SOL_SOCKET);
1149 mp = &(*mp)->m_next;
1154 /* RFC 2292 sec. 5 */
1155 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1156 struct in6_pktinfo pi6;
1157 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1158 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1159 pi6.ipi6_addr.s6_addr16[1] = 0;
1160 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1161 ? m->m_pkthdr.rcvif->if_index
1163 *mp = sbcreatecontrol((caddr_t) &pi6,
1164 sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1167 mp = &(*mp)->m_next;
1170 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) {
1171 int hlim = ip6->ip6_hlim & 0xff;
1172 *mp = sbcreatecontrol((caddr_t) &hlim,
1173 sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1175 mp = &(*mp)->m_next;
1179 * IPV6_HOPOPTS socket option. We require super-user privilege
1180 * for the option, but it might be too strict, since there might
1181 * be some hop-by-hop options which can be returned to normal user.
1182 * See RFC 2292 section 6.
1184 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1186 * Check if a hop-by-hop options header is contatined in the
1187 * received packet, and if so, store the options as ancillary
1188 * data. Note that a hop-by-hop options header must be
1189 * just after the IPv6 header, which fact is assured through
1190 * the IPv6 input processing.
1192 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1193 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1194 struct ip6_hbh *hbh;
1196 #ifdef PULLDOWN_TEST
1200 #ifndef PULLDOWN_TEST
1201 hbh = (struct ip6_hbh *)(ip6 + 1);
1202 hbhlen = (hbh->ip6h_len + 1) << 3;
1204 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1207 ip6stat.ip6s_tooshort++;
1210 hbh = mtod(ext, struct ip6_hbh *);
1211 hbhlen = (hbh->ip6h_len + 1) << 3;
1212 if (hbhlen != ext->m_len) {
1214 ip6stat.ip6s_tooshort++;
1220 * XXX: We copy whole the header even if a jumbo
1221 * payload option is included, which option is to
1222 * be removed before returning in the RFC 2292.
1223 * Note: this constraint is removed in 2292bis.
1225 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1226 IPV6_HOPOPTS, IPPROTO_IPV6);
1228 mp = &(*mp)->m_next;
1229 #ifdef PULLDOWN_TEST
1235 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1236 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1237 int proto, off, nxt;
1240 * go through the header chain to see if a routing header is
1241 * contained in the packet. We need this information to store
1242 * destination options headers (if any) properly.
1243 * XXX: performance issue. We should record this info when
1244 * processing extension headers in incoming routine.
1247 proto = IPPROTO_IPV6;
1253 newoff = ip6_nexthdr(m, off, proto, &nxt);
1256 if (newoff < off) /* invalid, check for safety */
1258 if ((proto = nxt) == IPPROTO_ROUTING) {
1266 if ((in6p->in6p_flags &
1267 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1268 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1269 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1272 * Search for destination options headers or routing
1273 * header(s) through the header chain, and stores each
1274 * header as ancillary data.
1275 * Note that the order of the headers remains in
1276 * the chain of ancillary data.
1278 while (1) { /* is explicit loop prevention necessary? */
1279 struct ip6_ext *ip6e = NULL;
1281 #ifdef PULLDOWN_TEST
1282 struct mbuf *ext = NULL;
1286 * if it is not an extension header, don't try to
1287 * pull it from the chain.
1290 case IPPROTO_DSTOPTS:
1291 case IPPROTO_ROUTING:
1292 case IPPROTO_HOPOPTS:
1293 case IPPROTO_AH: /* is it possible? */
1299 #ifndef PULLDOWN_TEST
1300 if (off + sizeof(*ip6e) > m->m_len)
1302 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1303 if (nxt == IPPROTO_AH)
1304 elen = (ip6e->ip6e_len + 2) << 2;
1306 elen = (ip6e->ip6e_len + 1) << 3;
1307 if (off + elen > m->m_len)
1310 ext = ip6_pullexthdr(m, off, nxt);
1312 ip6stat.ip6s_tooshort++;
1315 ip6e = mtod(ext, struct ip6_ext *);
1316 if (nxt == IPPROTO_AH)
1317 elen = (ip6e->ip6e_len + 2) << 2;
1319 elen = (ip6e->ip6e_len + 1) << 3;
1320 if (elen != ext->m_len) {
1322 ip6stat.ip6s_tooshort++;
1328 case IPPROTO_DSTOPTS:
1329 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0)
1333 * We also require super-user privilege for
1335 * See the comments on IN6_HOPOPTS.
1340 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1344 mp = &(*mp)->m_next;
1346 case IPPROTO_ROUTING:
1347 if (!in6p->in6p_flags & IN6P_RTHDR)
1350 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1354 mp = &(*mp)->m_next;
1356 case IPPROTO_HOPOPTS:
1357 case IPPROTO_AH: /* is it possible? */
1362 * other cases have been filtered in the above.
1363 * none will visit this case. here we supply
1364 * the code just in case (nxt overwritten or
1367 #ifdef PULLDOWN_TEST
1374 /* proceed with the next header. */
1376 nxt = ip6e->ip6e_nxt;
1378 #ifdef PULLDOWN_TEST
1389 #ifdef PULLDOWN_TEST
1391 * pull single extension header from mbuf chain. returns single mbuf that
1392 * contains the result, or NULL on error.
1394 static struct mbuf *
1395 ip6_pullexthdr(m, off, nxt)
1400 struct ip6_ext ip6e;
1406 case IPPROTO_DSTOPTS:
1407 case IPPROTO_ROUTING:
1408 case IPPROTO_HOPOPTS:
1409 case IPPROTO_AH: /* is it possible? */
1412 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1416 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1417 if (nxt == IPPROTO_AH)
1418 elen = (ip6e.ip6e_len + 2) << 2;
1420 elen = (ip6e.ip6e_len + 1) << 3;
1422 MGET(n, M_DONTWAIT, MT_DATA);
1423 if (n && elen >= MLEN) {
1424 MCLGET(n, M_DONTWAIT);
1425 if ((n->m_flags & M_EXT) == 0) {
1434 if (elen >= M_TRAILINGSPACE(n)) {
1439 m_copydata(m, off, elen, mtod(n, caddr_t));
1446 * Get pointer to the previous header followed by the header
1447 * currently processed.
1448 * XXX: This function supposes that
1449 * M includes all headers,
1450 * the next header field and the header length field of each header
1452 * the sum of each header length equals to OFF.
1453 * Because of these assumptions, this function must be called very
1454 * carefully. Moreover, it will not be used in the near future when
1455 * we develop `neater' mechanism to process extension headers.
1458 ip6_get_prevhdr(m, off)
1462 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1464 if (off == sizeof(struct ip6_hdr))
1465 return(&ip6->ip6_nxt);
1468 struct ip6_ext *ip6e = NULL;
1471 len = sizeof(struct ip6_hdr);
1473 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1476 case IPPROTO_FRAGMENT:
1477 len += sizeof(struct ip6_frag);
1480 len += (ip6e->ip6e_len + 2) << 2;
1483 len += (ip6e->ip6e_len + 1) << 3;
1486 nxt = ip6e->ip6e_nxt;
1489 return(&ip6e->ip6e_nxt);
1496 * get next header offset. m will be retained.
1499 ip6_nexthdr(m, off, proto, nxtp)
1506 struct ip6_ext ip6e;
1511 panic("ip6_nexthdr: m == NULL");
1512 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1517 if (m->m_pkthdr.len < off + sizeof(ip6))
1519 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1521 *nxtp = ip6.ip6_nxt;
1525 case IPPROTO_FRAGMENT:
1527 * terminate parsing if it is not the first fragment,
1528 * it does not make sense to parse through it.
1530 if (m->m_pkthdr.len < off + sizeof(fh))
1532 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1533 if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1536 *nxtp = fh.ip6f_nxt;
1537 off += sizeof(struct ip6_frag);
1541 if (m->m_pkthdr.len < off + sizeof(ip6e))
1543 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1545 *nxtp = ip6e.ip6e_nxt;
1546 off += (ip6e.ip6e_len + 2) << 2;
1549 case IPPROTO_HOPOPTS:
1550 case IPPROTO_ROUTING:
1551 case IPPROTO_DSTOPTS:
1552 if (m->m_pkthdr.len < off + sizeof(ip6e))
1554 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1556 *nxtp = ip6e.ip6e_nxt;
1557 off += (ip6e.ip6e_len + 1) << 3;
1562 case IPPROTO_IPCOMP:
1574 * get offset for the last header in the chain. m will be kept untainted.
1577 ip6_lasthdr(m, off, proto, nxtp)
1591 newoff = ip6_nexthdr(m, off, proto, nxtp);
1594 else if (newoff < off)
1595 return -1; /* invalid */
1596 else if (newoff == off)
1608 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1610 tag = m_tag_get(PACKET_TAG_IPV6_INPUT,
1611 sizeof (struct ip6aux),
1614 m_tag_prepend(m, tag);
1617 bzero(tag+1, sizeof (struct ip6aux));
1618 return tag ? (struct ip6aux*)(tag+1) : NULL;
1625 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1626 return tag ? (struct ip6aux*)(tag+1) : NULL;
1633 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1635 m_tag_delete(m, tag);
1639 * System control for IP6
1642 u_char inet6ctlerrmap[PRC_NCMDS] = {
1644 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1645 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1646 EMSGSIZE, EHOSTUNREACH, 0, 0,