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 1.10 2003/11/08 07:57:51 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>
94 #include <netinet/in.h>
95 #include <netinet/in_systm.h>
97 #include <netinet/ip.h>
98 #include <netinet/ip_icmp.h>
100 #include <netinet/ip6.h>
101 #include <netinet6/in6_var.h>
102 #include <netinet6/ip6_var.h>
103 #include <netinet/in_pcb.h>
104 #include <netinet/icmp6.h>
105 #include <netinet6/in6_ifattach.h>
106 #include <netinet6/nd6.h>
107 #include <netinet6/in6_prefix.h>
110 #include <netinet6/ipsec.h>
112 #include <netinet6/ipsec6.h>
120 #endif /* FAST_IPSEC */
122 #include <net/ip6fw/ip6_fw.h>
124 #include <netinet6/ip6protosw.h>
126 /* we need it for NLOOP. */
127 #include "use_loop.h"
129 #include <net/net_osdep.h>
131 extern int (*fr_checkp) (struct ip *, int, struct ifnet *, int, struct mbuf **);
133 extern struct domain inet6domain;
134 extern struct ip6protosw inet6sw[];
136 u_char ip6_protox[IPPROTO_MAX];
137 struct in6_ifaddr *in6_ifaddr;
139 extern struct callout in6_tmpaddrtimer_ch;
141 int ip6_forward_srcrt; /* XXX */
142 int ip6_sourcecheck; /* XXX */
143 int ip6_sourcecheck_interval; /* XXX */
145 int ip6_ours_check_algorithm;
148 ip6_fw_chk_t *ip6_fw_chk_ptr;
149 ip6_fw_ctl_t *ip6_fw_ctl_ptr;
150 int ip6_fw_enable = 1;
152 struct ip6stat ip6stat;
154 static void ip6_init2 (void *);
155 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *);
156 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *);
158 static struct mbuf *ip6_pullexthdr (struct mbuf *, size_t, int);
162 * IP6 initialization: fill in IP6 protocol switch table.
163 * All protocols not implemented in kernel go to raw IP6 protocol handler.
168 struct ip6protosw *pr;
173 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
174 panic("sizeof(protosw) != sizeof(ip6protosw)");
176 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
179 for (i = 0; i < IPPROTO_MAX; i++)
180 ip6_protox[i] = pr - inet6sw;
181 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
182 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
183 if (pr->pr_domain->dom_family == PF_INET6 &&
184 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
185 ip6_protox[pr->pr_protocol] = pr - inet6sw;
186 netisr_register(NETISR_IPV6, cpu0_portfn, ip6_input);
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);
212 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
214 /* router renumbering prefix list maintenance */
215 callout_init(&in6_rr_timer_ch);
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);
229 extern struct route_in6 ip6_forward_rt;
236 int off = sizeof(struct ip6_hdr), nest;
238 u_int32_t rtalert = ~0;
240 struct ifnet *deliverifp = NULL;
244 * should the inner packet be considered authentic?
245 * see comment in ah4_input().
248 m->m_flags &= ~M_AUTHIPHDR;
249 m->m_flags &= ~M_AUTHIPDGM;
254 * make sure we don't have onion peering information into m_aux.
261 if (m->m_flags & M_EXT) {
263 ip6stat.ip6s_mext2m++;
265 ip6stat.ip6s_mext1++;
267 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
269 if (m->m_flags & M_LOOP) {
270 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
271 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
272 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
274 ip6stat.ip6s_m2m[0]++;
280 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
281 ip6stat.ip6s_total++;
283 #ifndef PULLDOWN_TEST
285 * L2 bridge code and some other code can return mbuf chain
286 * that does not conform to KAME requirement. too bad.
287 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
289 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
292 MGETHDR(n, M_DONTWAIT, MT_HEADER);
295 if (n && n->m_pkthdr.len > MHLEN) {
296 MCLGET(n, M_DONTWAIT);
297 if ((n->m_flags & M_EXT) == 0) {
307 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
308 n->m_len = n->m_pkthdr.len;
312 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
315 if (m->m_len < sizeof(struct ip6_hdr)) {
317 inifp = m->m_pkthdr.rcvif;
318 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
319 ip6stat.ip6s_toosmall++;
320 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
325 ip6 = mtod(m, struct ip6_hdr *);
327 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
328 ip6stat.ip6s_badvers++;
329 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
334 * Check if we want to allow this packet to be processed.
335 * Consider it to be bad if not.
340 if ((*fr_checkp)((struct ip *)ip6, sizeof(*ip6),
341 m->m_pkthdr.rcvif, 0, &m1) != 0)
346 ip6 = mtod(m, struct ip6_hdr *);
349 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
352 * Check with the firewall...
354 if (ip6_fw_enable && ip6_fw_chk_ptr) {
356 /* If ipfw says divert, we have to just drop packet */
357 /* use port as a dummy argument */
358 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
367 * Check against address spoofing/corruption.
369 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
370 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
372 * XXX: "badscope" is not very suitable for a multicast source.
374 ip6stat.ip6s_badscope++;
375 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
378 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
379 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) &&
380 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
381 ip6stat.ip6s_badscope++;
382 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
387 * The following check is not documented in specs. A malicious
388 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
389 * and bypass security checks (act as if it was from 127.0.0.1 by using
390 * IPv6 src ::ffff:127.0.0.1). Be cautious.
392 * This check chokes if we are in an SIIT cloud. As none of BSDs
393 * support IPv4-less kernel compilation, we cannot support SIIT
394 * environment at all. So, it makes more sense for us to reject any
395 * malicious packets for non-SIIT environment, than try to do a
396 * partical support for SIIT environment.
398 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
399 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
400 ip6stat.ip6s_badscope++;
401 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
406 * Reject packets with IPv4 compatible addresses (auto tunnel).
408 * The code forbids auto tunnel relay case in RFC1933 (the check is
409 * stronger than RFC1933). We may want to re-enable it if mech-xx
410 * is revised to forbid relaying case.
412 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
413 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
414 ip6stat.ip6s_badscope++;
415 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
420 /* drop packets if interface ID portion is already filled */
421 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
422 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
423 ip6->ip6_src.s6_addr16[1]) {
424 ip6stat.ip6s_badscope++;
427 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
428 ip6->ip6_dst.s6_addr16[1]) {
429 ip6stat.ip6s_badscope++;
434 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
435 ip6->ip6_src.s6_addr16[1]
436 = htons(m->m_pkthdr.rcvif->if_index);
437 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
438 ip6->ip6_dst.s6_addr16[1]
439 = htons(m->m_pkthdr.rcvif->if_index);
441 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */
443 * We use rt->rt_ifp to determine if the address is ours or not.
444 * If rt_ifp is lo0, the address is ours.
445 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0,
446 * so any address under fe80::%lo0/64 will be mistakenly considered
447 * local. The special case is supplied to handle the case properly
448 * by actually looking at interface addresses
449 * (using in6ifa_ifpwithaddr).
451 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 &&
452 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
453 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) {
454 icmp6_error(m, ICMP6_DST_UNREACH,
455 ICMP6_DST_UNREACH_ADDR, 0);
456 /* m is already freed */
461 deliverifp = m->m_pkthdr.rcvif;
469 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
470 struct in6_multi *in6m = 0;
472 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
474 * See if we belong to the destination multicast group on the
477 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
480 else if (!ip6_mrouter) {
481 ip6stat.ip6s_notmember++;
482 ip6stat.ip6s_cantforward++;
483 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
486 deliverifp = m->m_pkthdr.rcvif;
493 switch (ip6_ours_check_algorithm) {
496 * XXX: I intentionally broke our indentation rule here,
497 * since this switch-case is just for measurement and
498 * therefore should soon be removed.
500 if (ip6_forward_rt.ro_rt != NULL &&
501 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
502 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
503 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
504 ip6stat.ip6s_forward_cachehit++;
506 struct sockaddr_in6 *dst6;
508 if (ip6_forward_rt.ro_rt) {
509 /* route is down or destination is different */
510 ip6stat.ip6s_forward_cachemiss++;
511 RTFREE(ip6_forward_rt.ro_rt);
512 ip6_forward_rt.ro_rt = 0;
515 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
516 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
517 dst6->sin6_len = sizeof(struct sockaddr_in6);
518 dst6->sin6_family = AF_INET6;
519 dst6->sin6_addr = ip6->ip6_dst;
521 ip6_forward_rt.ro_dst.sin6_scope_id =
522 in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst);
525 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
528 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
531 * Accept the packet if the forwarding interface to the destination
532 * according to the routing table is the loopback interface,
533 * unless the associated route has a gateway.
534 * Note that this approach causes to accept a packet if there is a
535 * route to the loopback interface for the destination of the packet.
536 * But we think it's even useful in some situations, e.g. when using
537 * a special daemon which wants to intercept the packet.
539 * XXX: some OSes automatically make a cloned route for the destination
540 * of an outgoing packet. If the outgoing interface of the packet
541 * is a loopback one, the kernel would consider the packet to be
542 * accepted, even if we have no such address assinged on the interface.
543 * We check the cloned flag of the route entry to reject such cases,
544 * assuming that route entries for our own addresses are not made by
545 * cloning (it should be true because in6_addloop explicitly installs
546 * the host route). However, we might have to do an explicit check
547 * while it would be less efficient. Or, should we rather install a
548 * reject route for such a case?
550 if (ip6_forward_rt.ro_rt &&
551 (ip6_forward_rt.ro_rt->rt_flags &
552 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
554 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
557 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
561 * The check below is redundant since the comparison of
562 * the destination and the key of the rtentry has
563 * already done through looking up the routing table.
565 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
566 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
568 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
569 struct in6_ifaddr *ia6 =
570 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
573 * record address information into m_aux.
575 (void)ip6_setdstifaddr(m, ia6);
578 * packets to a tentative, duplicated, or somehow invalid
579 * address must not be accepted.
581 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
582 /* this address is ready */
584 deliverifp = ia6->ia_ifp; /* correct? */
585 /* Count the packet in the ip address stats */
586 ia6->ia_ifa.if_ipackets++;
587 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
590 /* address is not ready, so discard the packet. */
592 "ip6_input: packet to an unready address %s->%s\n",
593 ip6_sprintf(&ip6->ip6_src),
594 ip6_sprintf(&ip6->ip6_dst)));
599 } /* XXX indentation (see above) */
602 * FAITH(Firewall Aided Internet Translator)
605 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
606 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
607 /* XXX do we need more sanity checks? */
609 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
615 * Now there is no reason to process the packet if it's not our own
616 * and we're not a router.
618 if (!ip6_forwarding) {
619 ip6stat.ip6s_cantforward++;
620 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
626 * record address information into m_aux, if we don't have one yet.
627 * note that we are unable to record it, if the address is not listed
628 * as our interface address (e.g. multicast addresses, addresses
629 * within FAITH prefixes and such).
631 if (deliverifp && !ip6_getdstifaddr(m)) {
632 struct in6_ifaddr *ia6;
634 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
636 if (!ip6_setdstifaddr(m, ia6)) {
638 * XXX maybe we should drop the packet here,
639 * as we could not provide enough information
640 * to the upper layers.
647 * Process Hop-by-Hop options header if it's contained.
648 * m may be modified in ip6_hopopts_input().
649 * If a JumboPayload option is included, plen will also be modified.
651 plen = (u_int32_t)ntohs(ip6->ip6_plen);
652 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
655 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
656 #if 0 /*touches NULL pointer*/
657 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
659 return; /* m have already been freed */
663 ip6 = mtod(m, struct ip6_hdr *);
666 * if the payload length field is 0 and the next header field
667 * indicates Hop-by-Hop Options header, then a Jumbo Payload
668 * option MUST be included.
670 if (ip6->ip6_plen == 0 && plen == 0) {
672 * Note that if a valid jumbo payload option is
673 * contained, ip6_hoptops_input() must set a valid
674 * (non-zero) payload length to the variable plen.
676 ip6stat.ip6s_badoptions++;
677 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
678 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
679 icmp6_error(m, ICMP6_PARAM_PROB,
680 ICMP6_PARAMPROB_HEADER,
681 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
684 #ifndef PULLDOWN_TEST
685 /* ip6_hopopts_input() ensures that mbuf is contiguous */
686 hbh = (struct ip6_hbh *)(ip6 + 1);
688 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
689 sizeof(struct ip6_hbh));
691 ip6stat.ip6s_tooshort++;
698 * accept the packet if a router alert option is included
699 * and we act as an IPv6 router.
701 if (rtalert != ~0 && ip6_forwarding)
707 * Check that the amount of data in the buffers
708 * is as at least much as the IPv6 header would have us expect.
709 * Trim mbufs if longer than we expect.
710 * Drop packet if shorter than we expect.
712 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
713 ip6stat.ip6s_tooshort++;
714 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
717 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
718 if (m->m_len == m->m_pkthdr.len) {
719 m->m_len = sizeof(struct ip6_hdr) + plen;
720 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
722 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
726 * Forward if desirable.
728 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
730 * If we are acting as a multicast router, all
731 * incoming multicast packets are passed to the
732 * kernel-level multicast forwarding function.
733 * The packet is returned (relatively) intact; if
734 * ip6_mforward() returns a non-zero value, the packet
735 * must be discarded, else it may be accepted below.
737 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
738 ip6stat.ip6s_cantforward++;
751 ip6 = mtod(m, struct ip6_hdr *);
754 * Malicious party may be able to use IPv4 mapped addr to confuse
755 * tcp/udp stack and bypass security checks (act as if it was from
756 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
758 * For SIIT end node behavior, you may want to disable the check.
759 * However, you will become vulnerable to attacks using IPv4 mapped
762 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
763 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
764 ip6stat.ip6s_badscope++;
765 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
770 * Tell launch routine the next header
772 ip6stat.ip6s_delivered++;
773 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
776 while (nxt != IPPROTO_DONE) {
777 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
778 ip6stat.ip6s_toomanyhdr++;
783 * protection against faulty packet - there should be
784 * more sanity checks in header chain processing.
786 if (m->m_pkthdr.len < off) {
787 ip6stat.ip6s_tooshort++;
788 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
794 * do we need to do it for every header? yeah, other
795 * functions can play with it (like re-allocate and copy).
797 mhist = ip6_addaux(m);
798 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) {
799 hist = mtod(mhist, caddr_t) + mhist->m_len;
800 bcopy(&nxt, hist, sizeof(nxt));
801 mhist->m_len += sizeof(nxt);
803 ip6stat.ip6s_toomanyhdr++;
810 * enforce IPsec policy checking if we are seeing last header.
811 * note that we do not visit this with protocols with pcb layer
812 * code - like udp/tcp/raw ip.
814 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
815 ipsec6_in_reject(m, NULL)) {
816 ipsec6stat.in_polvio++;
821 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
829 * set/grab in6_ifaddr correspond to IPv6 destination address.
830 * XXX backward compatibility wrapper
832 static struct ip6aux *
833 ip6_setdstifaddr(m, ia6)
835 struct in6_ifaddr *ia6;
841 n->ip6a_dstia6 = ia6;
842 return n; /* NULL if failed to set */
853 return n->ip6a_dstia6;
859 * Hop-by-Hop options header processing. If a valid jumbo payload option is
860 * included, the real payload length will be stored in plenp.
863 ip6_hopopts_input(plenp, rtalertp, mp, offp)
865 u_int32_t *rtalertp; /* XXX: should be stored more smart way */
869 struct mbuf *m = *mp;
870 int off = *offp, hbhlen;
874 /* validation of the length of the header */
875 #ifndef PULLDOWN_TEST
876 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
877 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
878 hbhlen = (hbh->ip6h_len + 1) << 3;
880 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
881 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
883 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
884 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
886 ip6stat.ip6s_tooshort++;
889 hbhlen = (hbh->ip6h_len + 1) << 3;
890 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
893 ip6stat.ip6s_tooshort++;
898 hbhlen -= sizeof(struct ip6_hbh);
899 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
901 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
902 hbhlen, rtalertp, plenp) < 0)
911 * Search header for all Hop-by-hop options and process each option.
912 * This function is separate from ip6_hopopts_input() in order to
913 * handle a case where the sending node itself process its hop-by-hop
914 * options header. In such a case, the function is called from ip6_output().
916 * The function assumes that hbh header is located right after the IPv6 header
917 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
918 * opthead + hbhlen is located in continuous memory region.
921 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
930 u_int8_t *opt = opthead;
931 u_int16_t rtalert_val;
933 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
935 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
941 if (hbhlen < IP6OPT_MINLEN) {
942 ip6stat.ip6s_toosmall++;
945 optlen = *(opt + 1) + 2;
948 /* XXX may need check for alignment */
949 if (hbhlen < IP6OPT_RTALERT_LEN) {
950 ip6stat.ip6s_toosmall++;
953 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
955 icmp6_error(m, ICMP6_PARAM_PROB,
956 ICMP6_PARAMPROB_HEADER,
957 erroff + opt + 1 - opthead);
960 optlen = IP6OPT_RTALERT_LEN;
961 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
962 *rtalertp = ntohs(rtalert_val);
965 /* XXX may need check for alignment */
966 if (hbhlen < IP6OPT_JUMBO_LEN) {
967 ip6stat.ip6s_toosmall++;
970 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
972 icmp6_error(m, ICMP6_PARAM_PROB,
973 ICMP6_PARAMPROB_HEADER,
974 erroff + opt + 1 - opthead);
977 optlen = IP6OPT_JUMBO_LEN;
980 * IPv6 packets that have non 0 payload length
981 * must not contain a jumbo payload option.
983 ip6 = mtod(m, struct ip6_hdr *);
985 ip6stat.ip6s_badoptions++;
986 icmp6_error(m, ICMP6_PARAM_PROB,
987 ICMP6_PARAMPROB_HEADER,
988 erroff + opt - opthead);
993 * We may see jumbolen in unaligned location, so
994 * we'd need to perform bcopy().
996 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
997 jumboplen = (u_int32_t)htonl(jumboplen);
1001 * if there are multiple jumbo payload options,
1002 * *plenp will be non-zero and the packet will be
1004 * the behavior may need some debate in ipngwg -
1005 * multiple options does not make sense, however,
1006 * there's no explicit mention in specification.
1009 ip6stat.ip6s_badoptions++;
1010 icmp6_error(m, ICMP6_PARAM_PROB,
1011 ICMP6_PARAMPROB_HEADER,
1012 erroff + opt + 2 - opthead);
1018 * jumbo payload length must be larger than 65535.
1020 if (jumboplen <= IPV6_MAXPACKET) {
1021 ip6stat.ip6s_badoptions++;
1022 icmp6_error(m, ICMP6_PARAM_PROB,
1023 ICMP6_PARAMPROB_HEADER,
1024 erroff + opt + 2 - opthead);
1030 default: /* unknown option */
1031 if (hbhlen < IP6OPT_MINLEN) {
1032 ip6stat.ip6s_toosmall++;
1035 optlen = ip6_unknown_opt(opt, m,
1036 erroff + opt - opthead);
1052 * Unknown option processing.
1053 * The third argument `off' is the offset from the IPv6 header to the option,
1054 * which is necessary if the IPv6 header the and option header and IPv6 header
1055 * is not continuous in order to return an ICMPv6 error.
1058 ip6_unknown_opt(optp, m, off)
1063 struct ip6_hdr *ip6;
1065 switch (IP6OPT_TYPE(*optp)) {
1066 case IP6OPT_TYPE_SKIP: /* ignore the option */
1067 return((int)*(optp + 1));
1068 case IP6OPT_TYPE_DISCARD: /* silently discard */
1071 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1072 ip6stat.ip6s_badoptions++;
1073 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1075 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1076 ip6stat.ip6s_badoptions++;
1077 ip6 = mtod(m, struct ip6_hdr *);
1078 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1079 (m->m_flags & (M_BCAST|M_MCAST)))
1082 icmp6_error(m, ICMP6_PARAM_PROB,
1083 ICMP6_PARAMPROB_OPTION, off);
1087 m_freem(m); /* XXX: NOTREACHED */
1092 * Create the "control" list for this pcb.
1093 * The function will not modify mbuf chain at all.
1095 * with KAME mbuf chain restriction:
1096 * The routine will be called from upper layer handlers like tcp6_input().
1097 * Thus the routine assumes that the caller (tcp6_input) have already
1098 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1099 * very first mbuf on the mbuf chain.
1102 ip6_savecontrol(in6p, mp, ip6, m)
1105 struct ip6_hdr *ip6;
1108 struct thread *td = curthread; /* XXX */
1110 int rthdr_exist = 0;
1117 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) {
1121 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1122 SCM_TIMESTAMP, SOL_SOCKET);
1124 mp = &(*mp)->m_next;
1129 /* RFC 2292 sec. 5 */
1130 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1131 struct in6_pktinfo pi6;
1132 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1133 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1134 pi6.ipi6_addr.s6_addr16[1] = 0;
1135 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1136 ? m->m_pkthdr.rcvif->if_index
1138 *mp = sbcreatecontrol((caddr_t) &pi6,
1139 sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1142 mp = &(*mp)->m_next;
1145 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) {
1146 int hlim = ip6->ip6_hlim & 0xff;
1147 *mp = sbcreatecontrol((caddr_t) &hlim,
1148 sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1150 mp = &(*mp)->m_next;
1154 * IPV6_HOPOPTS socket option. We require super-user privilege
1155 * for the option, but it might be too strict, since there might
1156 * be some hop-by-hop options which can be returned to normal user.
1157 * See RFC 2292 section 6.
1159 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1161 * Check if a hop-by-hop options header is contatined in the
1162 * received packet, and if so, store the options as ancillary
1163 * data. Note that a hop-by-hop options header must be
1164 * just after the IPv6 header, which fact is assured through
1165 * the IPv6 input processing.
1167 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1168 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1169 struct ip6_hbh *hbh;
1171 #ifdef PULLDOWN_TEST
1175 #ifndef PULLDOWN_TEST
1176 hbh = (struct ip6_hbh *)(ip6 + 1);
1177 hbhlen = (hbh->ip6h_len + 1) << 3;
1179 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1182 ip6stat.ip6s_tooshort++;
1185 hbh = mtod(ext, struct ip6_hbh *);
1186 hbhlen = (hbh->ip6h_len + 1) << 3;
1187 if (hbhlen != ext->m_len) {
1189 ip6stat.ip6s_tooshort++;
1195 * XXX: We copy whole the header even if a jumbo
1196 * payload option is included, which option is to
1197 * be removed before returning in the RFC 2292.
1198 * Note: this constraint is removed in 2292bis.
1200 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1201 IPV6_HOPOPTS, IPPROTO_IPV6);
1203 mp = &(*mp)->m_next;
1204 #ifdef PULLDOWN_TEST
1210 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1211 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1212 int proto, off, nxt;
1215 * go through the header chain to see if a routing header is
1216 * contained in the packet. We need this information to store
1217 * destination options headers (if any) properly.
1218 * XXX: performance issue. We should record this info when
1219 * processing extension headers in incoming routine.
1222 proto = IPPROTO_IPV6;
1228 newoff = ip6_nexthdr(m, off, proto, &nxt);
1231 if (newoff < off) /* invalid, check for safety */
1233 if ((proto = nxt) == IPPROTO_ROUTING) {
1241 if ((in6p->in6p_flags &
1242 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1243 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1244 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1247 * Search for destination options headers or routing
1248 * header(s) through the header chain, and stores each
1249 * header as ancillary data.
1250 * Note that the order of the headers remains in
1251 * the chain of ancillary data.
1253 while (1) { /* is explicit loop prevention necessary? */
1254 struct ip6_ext *ip6e = NULL;
1256 #ifdef PULLDOWN_TEST
1257 struct mbuf *ext = NULL;
1261 * if it is not an extension header, don't try to
1262 * pull it from the chain.
1265 case IPPROTO_DSTOPTS:
1266 case IPPROTO_ROUTING:
1267 case IPPROTO_HOPOPTS:
1268 case IPPROTO_AH: /* is it possible? */
1274 #ifndef PULLDOWN_TEST
1275 if (off + sizeof(*ip6e) > m->m_len)
1277 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1278 if (nxt == IPPROTO_AH)
1279 elen = (ip6e->ip6e_len + 2) << 2;
1281 elen = (ip6e->ip6e_len + 1) << 3;
1282 if (off + elen > m->m_len)
1285 ext = ip6_pullexthdr(m, off, nxt);
1287 ip6stat.ip6s_tooshort++;
1290 ip6e = mtod(ext, struct ip6_ext *);
1291 if (nxt == IPPROTO_AH)
1292 elen = (ip6e->ip6e_len + 2) << 2;
1294 elen = (ip6e->ip6e_len + 1) << 3;
1295 if (elen != ext->m_len) {
1297 ip6stat.ip6s_tooshort++;
1303 case IPPROTO_DSTOPTS:
1304 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0)
1308 * We also require super-user privilege for
1310 * See the comments on IN6_HOPOPTS.
1315 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1319 mp = &(*mp)->m_next;
1321 case IPPROTO_ROUTING:
1322 if (!in6p->in6p_flags & IN6P_RTHDR)
1325 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1329 mp = &(*mp)->m_next;
1331 case IPPROTO_HOPOPTS:
1332 case IPPROTO_AH: /* is it possible? */
1337 * other cases have been filtered in the above.
1338 * none will visit this case. here we supply
1339 * the code just in case (nxt overwritten or
1342 #ifdef PULLDOWN_TEST
1349 /* proceed with the next header. */
1351 nxt = ip6e->ip6e_nxt;
1353 #ifdef PULLDOWN_TEST
1364 #ifdef PULLDOWN_TEST
1366 * pull single extension header from mbuf chain. returns single mbuf that
1367 * contains the result, or NULL on error.
1369 static struct mbuf *
1370 ip6_pullexthdr(m, off, nxt)
1375 struct ip6_ext ip6e;
1381 case IPPROTO_DSTOPTS:
1382 case IPPROTO_ROUTING:
1383 case IPPROTO_HOPOPTS:
1384 case IPPROTO_AH: /* is it possible? */
1387 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1391 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1392 if (nxt == IPPROTO_AH)
1393 elen = (ip6e.ip6e_len + 2) << 2;
1395 elen = (ip6e.ip6e_len + 1) << 3;
1397 MGET(n, M_DONTWAIT, MT_DATA);
1398 if (n && elen >= MLEN) {
1399 MCLGET(n, M_DONTWAIT);
1400 if ((n->m_flags & M_EXT) == 0) {
1409 if (elen >= M_TRAILINGSPACE(n)) {
1414 m_copydata(m, off, elen, mtod(n, caddr_t));
1421 * Get pointer to the previous header followed by the header
1422 * currently processed.
1423 * XXX: This function supposes that
1424 * M includes all headers,
1425 * the next header field and the header length field of each header
1427 * the sum of each header length equals to OFF.
1428 * Because of these assumptions, this function must be called very
1429 * carefully. Moreover, it will not be used in the near future when
1430 * we develop `neater' mechanism to process extension headers.
1433 ip6_get_prevhdr(m, off)
1437 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1439 if (off == sizeof(struct ip6_hdr))
1440 return(&ip6->ip6_nxt);
1443 struct ip6_ext *ip6e = NULL;
1446 len = sizeof(struct ip6_hdr);
1448 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1451 case IPPROTO_FRAGMENT:
1452 len += sizeof(struct ip6_frag);
1455 len += (ip6e->ip6e_len + 2) << 2;
1458 len += (ip6e->ip6e_len + 1) << 3;
1461 nxt = ip6e->ip6e_nxt;
1464 return(&ip6e->ip6e_nxt);
1471 * get next header offset. m will be retained.
1474 ip6_nexthdr(m, off, proto, nxtp)
1481 struct ip6_ext ip6e;
1486 panic("ip6_nexthdr: m == NULL");
1487 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1492 if (m->m_pkthdr.len < off + sizeof(ip6))
1494 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1496 *nxtp = ip6.ip6_nxt;
1500 case IPPROTO_FRAGMENT:
1502 * terminate parsing if it is not the first fragment,
1503 * it does not make sense to parse through it.
1505 if (m->m_pkthdr.len < off + sizeof(fh))
1507 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1508 if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1511 *nxtp = fh.ip6f_nxt;
1512 off += sizeof(struct ip6_frag);
1516 if (m->m_pkthdr.len < off + sizeof(ip6e))
1518 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1520 *nxtp = ip6e.ip6e_nxt;
1521 off += (ip6e.ip6e_len + 2) << 2;
1524 case IPPROTO_HOPOPTS:
1525 case IPPROTO_ROUTING:
1526 case IPPROTO_DSTOPTS:
1527 if (m->m_pkthdr.len < off + sizeof(ip6e))
1529 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1531 *nxtp = ip6e.ip6e_nxt;
1532 off += (ip6e.ip6e_len + 1) << 3;
1537 case IPPROTO_IPCOMP:
1549 * get offset for the last header in the chain. m will be kept untainted.
1552 ip6_lasthdr(m, off, proto, nxtp)
1566 newoff = ip6_nexthdr(m, off, proto, nxtp);
1569 else if (newoff < off)
1570 return -1; /* invalid */
1571 else if (newoff == off)
1583 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1585 tag = m_tag_get(PACKET_TAG_IPV6_INPUT,
1586 sizeof (struct ip6aux),
1589 m_tag_prepend(m, tag);
1592 bzero(tag+1, sizeof (struct ip6aux));
1593 return tag ? (struct ip6aux*)(tag+1) : NULL;
1600 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1601 return tag ? (struct ip6aux*)(tag+1) : NULL;
1608 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1610 m_tag_delete(m, tag);
1614 * System control for IP6
1617 u_char inet6ctlerrmap[PRC_NCMDS] = {
1619 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1620 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1621 EMSGSIZE, EHOSTUNREACH, 0, 0,