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 1.8 2003/09/15 23:38:14 hsu 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>
121 #endif /* FAST_IPSEC */
123 #include <net/ip6fw/ip6_fw.h>
125 #include <netinet6/ip6protosw.h>
127 /* we need it for NLOOP. */
128 #include "use_loop.h"
130 #include <net/net_osdep.h>
132 extern int (*fr_checkp) (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 (void *);
159 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *);
160 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *);
162 static struct mbuf *ip6_pullexthdr (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 netisr_register(NETISR_IPV6, ip6_input, &ip6intrq);
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);
235 extern struct route_in6 ip6_forward_rt;
242 int off = sizeof(struct ip6_hdr), nest;
244 u_int32_t rtalert = ~0;
246 struct ifnet *deliverifp = NULL;
250 * should the inner packet be considered authentic?
251 * see comment in ah4_input().
254 m->m_flags &= ~M_AUTHIPHDR;
255 m->m_flags &= ~M_AUTHIPDGM;
260 * make sure we don't have onion peering information into m_aux.
267 if (m->m_flags & M_EXT) {
269 ip6stat.ip6s_mext2m++;
271 ip6stat.ip6s_mext1++;
273 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
275 if (m->m_flags & M_LOOP) {
276 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
277 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
278 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
280 ip6stat.ip6s_m2m[0]++;
286 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
287 ip6stat.ip6s_total++;
289 #ifndef PULLDOWN_TEST
291 * L2 bridge code and some other code can return mbuf chain
292 * that does not conform to KAME requirement. too bad.
293 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
295 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
298 MGETHDR(n, M_DONTWAIT, MT_HEADER);
301 if (n && n->m_pkthdr.len > MHLEN) {
302 MCLGET(n, M_DONTWAIT);
303 if ((n->m_flags & M_EXT) == 0) {
313 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
314 n->m_len = n->m_pkthdr.len;
318 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
321 if (m->m_len < sizeof(struct ip6_hdr)) {
323 inifp = m->m_pkthdr.rcvif;
324 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
325 ip6stat.ip6s_toosmall++;
326 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
331 ip6 = mtod(m, struct ip6_hdr *);
333 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
334 ip6stat.ip6s_badvers++;
335 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
340 * Check if we want to allow this packet to be processed.
341 * Consider it to be bad if not.
346 if ((*fr_checkp)((struct ip *)ip6, sizeof(*ip6),
347 m->m_pkthdr.rcvif, 0, &m1) != 0)
352 ip6 = mtod(m, struct ip6_hdr *);
355 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
358 * Check with the firewall...
360 if (ip6_fw_enable && ip6_fw_chk_ptr) {
362 /* If ipfw says divert, we have to just drop packet */
363 /* use port as a dummy argument */
364 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
373 * Check against address spoofing/corruption.
375 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
376 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
378 * XXX: "badscope" is not very suitable for a multicast source.
380 ip6stat.ip6s_badscope++;
381 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
384 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
385 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) &&
386 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
387 ip6stat.ip6s_badscope++;
388 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
393 * The following check is not documented in specs. A malicious
394 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
395 * and bypass security checks (act as if it was from 127.0.0.1 by using
396 * IPv6 src ::ffff:127.0.0.1). Be cautious.
398 * This check chokes if we are in an SIIT cloud. As none of BSDs
399 * support IPv4-less kernel compilation, we cannot support SIIT
400 * environment at all. So, it makes more sense for us to reject any
401 * malicious packets for non-SIIT environment, than try to do a
402 * partical support for SIIT environment.
404 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
405 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
406 ip6stat.ip6s_badscope++;
407 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
412 * Reject packets with IPv4 compatible addresses (auto tunnel).
414 * The code forbids auto tunnel relay case in RFC1933 (the check is
415 * stronger than RFC1933). We may want to re-enable it if mech-xx
416 * is revised to forbid relaying case.
418 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
419 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
420 ip6stat.ip6s_badscope++;
421 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
426 /* drop packets if interface ID portion is already filled */
427 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
428 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
429 ip6->ip6_src.s6_addr16[1]) {
430 ip6stat.ip6s_badscope++;
433 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
434 ip6->ip6_dst.s6_addr16[1]) {
435 ip6stat.ip6s_badscope++;
440 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
441 ip6->ip6_src.s6_addr16[1]
442 = htons(m->m_pkthdr.rcvif->if_index);
443 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
444 ip6->ip6_dst.s6_addr16[1]
445 = htons(m->m_pkthdr.rcvif->if_index);
447 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */
449 * We use rt->rt_ifp to determine if the address is ours or not.
450 * If rt_ifp is lo0, the address is ours.
451 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0,
452 * so any address under fe80::%lo0/64 will be mistakenly considered
453 * local. The special case is supplied to handle the case properly
454 * by actually looking at interface addresses
455 * (using in6ifa_ifpwithaddr).
457 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 &&
458 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
459 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) {
460 icmp6_error(m, ICMP6_DST_UNREACH,
461 ICMP6_DST_UNREACH_ADDR, 0);
462 /* m is already freed */
467 deliverifp = m->m_pkthdr.rcvif;
475 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
476 struct in6_multi *in6m = 0;
478 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
480 * See if we belong to the destination multicast group on the
483 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
486 else if (!ip6_mrouter) {
487 ip6stat.ip6s_notmember++;
488 ip6stat.ip6s_cantforward++;
489 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
492 deliverifp = m->m_pkthdr.rcvif;
499 switch (ip6_ours_check_algorithm) {
502 * XXX: I intentionally broke our indentation rule here,
503 * since this switch-case is just for measurement and
504 * therefore should soon be removed.
506 if (ip6_forward_rt.ro_rt != NULL &&
507 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
508 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
509 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
510 ip6stat.ip6s_forward_cachehit++;
512 struct sockaddr_in6 *dst6;
514 if (ip6_forward_rt.ro_rt) {
515 /* route is down or destination is different */
516 ip6stat.ip6s_forward_cachemiss++;
517 RTFREE(ip6_forward_rt.ro_rt);
518 ip6_forward_rt.ro_rt = 0;
521 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
522 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
523 dst6->sin6_len = sizeof(struct sockaddr_in6);
524 dst6->sin6_family = AF_INET6;
525 dst6->sin6_addr = ip6->ip6_dst;
527 ip6_forward_rt.ro_dst.sin6_scope_id =
528 in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst);
531 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
534 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
537 * Accept the packet if the forwarding interface to the destination
538 * according to the routing table is the loopback interface,
539 * unless the associated route has a gateway.
540 * Note that this approach causes to accept a packet if there is a
541 * route to the loopback interface for the destination of the packet.
542 * But we think it's even useful in some situations, e.g. when using
543 * a special daemon which wants to intercept the packet.
545 * XXX: some OSes automatically make a cloned route for the destination
546 * of an outgoing packet. If the outgoing interface of the packet
547 * is a loopback one, the kernel would consider the packet to be
548 * accepted, even if we have no such address assinged on the interface.
549 * We check the cloned flag of the route entry to reject such cases,
550 * assuming that route entries for our own addresses are not made by
551 * cloning (it should be true because in6_addloop explicitly installs
552 * the host route). However, we might have to do an explicit check
553 * while it would be less efficient. Or, should we rather install a
554 * reject route for such a case?
556 if (ip6_forward_rt.ro_rt &&
557 (ip6_forward_rt.ro_rt->rt_flags &
558 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
560 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
563 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
567 * The check below is redundant since the comparison of
568 * the destination and the key of the rtentry has
569 * already done through looking up the routing table.
571 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
572 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
574 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
575 struct in6_ifaddr *ia6 =
576 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
579 * record address information into m_aux.
581 (void)ip6_setdstifaddr(m, ia6);
584 * packets to a tentative, duplicated, or somehow invalid
585 * address must not be accepted.
587 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
588 /* this address is ready */
590 deliverifp = ia6->ia_ifp; /* correct? */
591 /* Count the packet in the ip address stats */
592 ia6->ia_ifa.if_ipackets++;
593 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
596 /* address is not ready, so discard the packet. */
598 "ip6_input: packet to an unready address %s->%s\n",
599 ip6_sprintf(&ip6->ip6_src),
600 ip6_sprintf(&ip6->ip6_dst)));
605 } /* XXX indentation (see above) */
608 * FAITH(Firewall Aided Internet Translator)
611 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
612 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
613 /* XXX do we need more sanity checks? */
615 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
621 * Now there is no reason to process the packet if it's not our own
622 * and we're not a router.
624 if (!ip6_forwarding) {
625 ip6stat.ip6s_cantforward++;
626 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
632 * record address information into m_aux, if we don't have one yet.
633 * note that we are unable to record it, if the address is not listed
634 * as our interface address (e.g. multicast addresses, addresses
635 * within FAITH prefixes and such).
637 if (deliverifp && !ip6_getdstifaddr(m)) {
638 struct in6_ifaddr *ia6;
640 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
642 if (!ip6_setdstifaddr(m, ia6)) {
644 * XXX maybe we should drop the packet here,
645 * as we could not provide enough information
646 * to the upper layers.
653 * Process Hop-by-Hop options header if it's contained.
654 * m may be modified in ip6_hopopts_input().
655 * If a JumboPayload option is included, plen will also be modified.
657 plen = (u_int32_t)ntohs(ip6->ip6_plen);
658 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
661 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
662 #if 0 /*touches NULL pointer*/
663 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
665 return; /* m have already been freed */
669 ip6 = mtod(m, struct ip6_hdr *);
672 * if the payload length field is 0 and the next header field
673 * indicates Hop-by-Hop Options header, then a Jumbo Payload
674 * option MUST be included.
676 if (ip6->ip6_plen == 0 && plen == 0) {
678 * Note that if a valid jumbo payload option is
679 * contained, ip6_hoptops_input() must set a valid
680 * (non-zero) payload length to the variable plen.
682 ip6stat.ip6s_badoptions++;
683 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
684 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
685 icmp6_error(m, ICMP6_PARAM_PROB,
686 ICMP6_PARAMPROB_HEADER,
687 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
690 #ifndef PULLDOWN_TEST
691 /* ip6_hopopts_input() ensures that mbuf is contiguous */
692 hbh = (struct ip6_hbh *)(ip6 + 1);
694 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
695 sizeof(struct ip6_hbh));
697 ip6stat.ip6s_tooshort++;
704 * accept the packet if a router alert option is included
705 * and we act as an IPv6 router.
707 if (rtalert != ~0 && ip6_forwarding)
713 * Check that the amount of data in the buffers
714 * is as at least much as the IPv6 header would have us expect.
715 * Trim mbufs if longer than we expect.
716 * Drop packet if shorter than we expect.
718 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
719 ip6stat.ip6s_tooshort++;
720 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
723 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
724 if (m->m_len == m->m_pkthdr.len) {
725 m->m_len = sizeof(struct ip6_hdr) + plen;
726 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
728 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
732 * Forward if desirable.
734 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
736 * If we are acting as a multicast router, all
737 * incoming multicast packets are passed to the
738 * kernel-level multicast forwarding function.
739 * The packet is returned (relatively) intact; if
740 * ip6_mforward() returns a non-zero value, the packet
741 * must be discarded, else it may be accepted below.
743 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
744 ip6stat.ip6s_cantforward++;
757 ip6 = mtod(m, struct ip6_hdr *);
760 * Malicious party may be able to use IPv4 mapped addr to confuse
761 * tcp/udp stack and bypass security checks (act as if it was from
762 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
764 * For SIIT end node behavior, you may want to disable the check.
765 * However, you will become vulnerable to attacks using IPv4 mapped
768 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
769 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
770 ip6stat.ip6s_badscope++;
771 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
776 * Tell launch routine the next header
778 ip6stat.ip6s_delivered++;
779 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
782 while (nxt != IPPROTO_DONE) {
783 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
784 ip6stat.ip6s_toomanyhdr++;
789 * protection against faulty packet - there should be
790 * more sanity checks in header chain processing.
792 if (m->m_pkthdr.len < off) {
793 ip6stat.ip6s_tooshort++;
794 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
800 * do we need to do it for every header? yeah, other
801 * functions can play with it (like re-allocate and copy).
803 mhist = ip6_addaux(m);
804 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) {
805 hist = mtod(mhist, caddr_t) + mhist->m_len;
806 bcopy(&nxt, hist, sizeof(nxt));
807 mhist->m_len += sizeof(nxt);
809 ip6stat.ip6s_toomanyhdr++;
816 * enforce IPsec policy checking if we are seeing last header.
817 * note that we do not visit this with protocols with pcb layer
818 * code - like udp/tcp/raw ip.
820 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
821 ipsec6_in_reject(m, NULL)) {
822 ipsec6stat.in_polvio++;
827 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
835 * set/grab in6_ifaddr correspond to IPv6 destination address.
836 * XXX backward compatibility wrapper
838 static struct ip6aux *
839 ip6_setdstifaddr(m, ia6)
841 struct in6_ifaddr *ia6;
847 n->ip6a_dstia6 = ia6;
848 return n; /* NULL if failed to set */
859 return n->ip6a_dstia6;
865 * Hop-by-Hop options header processing. If a valid jumbo payload option is
866 * included, the real payload length will be stored in plenp.
869 ip6_hopopts_input(plenp, rtalertp, mp, offp)
871 u_int32_t *rtalertp; /* XXX: should be stored more smart way */
875 struct mbuf *m = *mp;
876 int off = *offp, hbhlen;
880 /* validation of the length of the header */
881 #ifndef PULLDOWN_TEST
882 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
883 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
884 hbhlen = (hbh->ip6h_len + 1) << 3;
886 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
887 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
889 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
890 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
892 ip6stat.ip6s_tooshort++;
895 hbhlen = (hbh->ip6h_len + 1) << 3;
896 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
899 ip6stat.ip6s_tooshort++;
904 hbhlen -= sizeof(struct ip6_hbh);
905 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
907 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
908 hbhlen, rtalertp, plenp) < 0)
917 * Search header for all Hop-by-hop options and process each option.
918 * This function is separate from ip6_hopopts_input() in order to
919 * handle a case where the sending node itself process its hop-by-hop
920 * options header. In such a case, the function is called from ip6_output().
922 * The function assumes that hbh header is located right after the IPv6 header
923 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
924 * opthead + hbhlen is located in continuous memory region.
927 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
936 u_int8_t *opt = opthead;
937 u_int16_t rtalert_val;
939 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
941 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
947 if (hbhlen < IP6OPT_MINLEN) {
948 ip6stat.ip6s_toosmall++;
951 optlen = *(opt + 1) + 2;
954 /* XXX may need check for alignment */
955 if (hbhlen < IP6OPT_RTALERT_LEN) {
956 ip6stat.ip6s_toosmall++;
959 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
961 icmp6_error(m, ICMP6_PARAM_PROB,
962 ICMP6_PARAMPROB_HEADER,
963 erroff + opt + 1 - opthead);
966 optlen = IP6OPT_RTALERT_LEN;
967 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
968 *rtalertp = ntohs(rtalert_val);
971 /* XXX may need check for alignment */
972 if (hbhlen < IP6OPT_JUMBO_LEN) {
973 ip6stat.ip6s_toosmall++;
976 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
978 icmp6_error(m, ICMP6_PARAM_PROB,
979 ICMP6_PARAMPROB_HEADER,
980 erroff + opt + 1 - opthead);
983 optlen = IP6OPT_JUMBO_LEN;
986 * IPv6 packets that have non 0 payload length
987 * must not contain a jumbo payload option.
989 ip6 = mtod(m, struct ip6_hdr *);
991 ip6stat.ip6s_badoptions++;
992 icmp6_error(m, ICMP6_PARAM_PROB,
993 ICMP6_PARAMPROB_HEADER,
994 erroff + opt - opthead);
999 * We may see jumbolen in unaligned location, so
1000 * we'd need to perform bcopy().
1002 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1003 jumboplen = (u_int32_t)htonl(jumboplen);
1007 * if there are multiple jumbo payload options,
1008 * *plenp will be non-zero and the packet will be
1010 * the behavior may need some debate in ipngwg -
1011 * multiple options does not make sense, however,
1012 * there's no explicit mention in specification.
1015 ip6stat.ip6s_badoptions++;
1016 icmp6_error(m, ICMP6_PARAM_PROB,
1017 ICMP6_PARAMPROB_HEADER,
1018 erroff + opt + 2 - opthead);
1024 * jumbo payload length must be larger than 65535.
1026 if (jumboplen <= IPV6_MAXPACKET) {
1027 ip6stat.ip6s_badoptions++;
1028 icmp6_error(m, ICMP6_PARAM_PROB,
1029 ICMP6_PARAMPROB_HEADER,
1030 erroff + opt + 2 - opthead);
1036 default: /* unknown option */
1037 if (hbhlen < IP6OPT_MINLEN) {
1038 ip6stat.ip6s_toosmall++;
1041 optlen = ip6_unknown_opt(opt, m,
1042 erroff + opt - opthead);
1058 * Unknown option processing.
1059 * The third argument `off' is the offset from the IPv6 header to the option,
1060 * which is necessary if the IPv6 header the and option header and IPv6 header
1061 * is not continuous in order to return an ICMPv6 error.
1064 ip6_unknown_opt(optp, m, off)
1069 struct ip6_hdr *ip6;
1071 switch (IP6OPT_TYPE(*optp)) {
1072 case IP6OPT_TYPE_SKIP: /* ignore the option */
1073 return((int)*(optp + 1));
1074 case IP6OPT_TYPE_DISCARD: /* silently discard */
1077 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1078 ip6stat.ip6s_badoptions++;
1079 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1081 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1082 ip6stat.ip6s_badoptions++;
1083 ip6 = mtod(m, struct ip6_hdr *);
1084 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1085 (m->m_flags & (M_BCAST|M_MCAST)))
1088 icmp6_error(m, ICMP6_PARAM_PROB,
1089 ICMP6_PARAMPROB_OPTION, off);
1093 m_freem(m); /* XXX: NOTREACHED */
1098 * Create the "control" list for this pcb.
1099 * The function will not modify mbuf chain at all.
1101 * with KAME mbuf chain restriction:
1102 * The routine will be called from upper layer handlers like tcp6_input().
1103 * Thus the routine assumes that the caller (tcp6_input) have already
1104 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1105 * very first mbuf on the mbuf chain.
1108 ip6_savecontrol(in6p, mp, ip6, m)
1111 struct ip6_hdr *ip6;
1114 struct thread *td = curthread; /* XXX */
1116 int rthdr_exist = 0;
1123 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) {
1127 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1128 SCM_TIMESTAMP, SOL_SOCKET);
1130 mp = &(*mp)->m_next;
1135 /* RFC 2292 sec. 5 */
1136 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1137 struct in6_pktinfo pi6;
1138 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1139 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1140 pi6.ipi6_addr.s6_addr16[1] = 0;
1141 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1142 ? m->m_pkthdr.rcvif->if_index
1144 *mp = sbcreatecontrol((caddr_t) &pi6,
1145 sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1148 mp = &(*mp)->m_next;
1151 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) {
1152 int hlim = ip6->ip6_hlim & 0xff;
1153 *mp = sbcreatecontrol((caddr_t) &hlim,
1154 sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1156 mp = &(*mp)->m_next;
1160 * IPV6_HOPOPTS socket option. We require super-user privilege
1161 * for the option, but it might be too strict, since there might
1162 * be some hop-by-hop options which can be returned to normal user.
1163 * See RFC 2292 section 6.
1165 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1167 * Check if a hop-by-hop options header is contatined in the
1168 * received packet, and if so, store the options as ancillary
1169 * data. Note that a hop-by-hop options header must be
1170 * just after the IPv6 header, which fact is assured through
1171 * the IPv6 input processing.
1173 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1174 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1175 struct ip6_hbh *hbh;
1177 #ifdef PULLDOWN_TEST
1181 #ifndef PULLDOWN_TEST
1182 hbh = (struct ip6_hbh *)(ip6 + 1);
1183 hbhlen = (hbh->ip6h_len + 1) << 3;
1185 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1188 ip6stat.ip6s_tooshort++;
1191 hbh = mtod(ext, struct ip6_hbh *);
1192 hbhlen = (hbh->ip6h_len + 1) << 3;
1193 if (hbhlen != ext->m_len) {
1195 ip6stat.ip6s_tooshort++;
1201 * XXX: We copy whole the header even if a jumbo
1202 * payload option is included, which option is to
1203 * be removed before returning in the RFC 2292.
1204 * Note: this constraint is removed in 2292bis.
1206 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1207 IPV6_HOPOPTS, IPPROTO_IPV6);
1209 mp = &(*mp)->m_next;
1210 #ifdef PULLDOWN_TEST
1216 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1217 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1218 int proto, off, nxt;
1221 * go through the header chain to see if a routing header is
1222 * contained in the packet. We need this information to store
1223 * destination options headers (if any) properly.
1224 * XXX: performance issue. We should record this info when
1225 * processing extension headers in incoming routine.
1228 proto = IPPROTO_IPV6;
1234 newoff = ip6_nexthdr(m, off, proto, &nxt);
1237 if (newoff < off) /* invalid, check for safety */
1239 if ((proto = nxt) == IPPROTO_ROUTING) {
1247 if ((in6p->in6p_flags &
1248 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1249 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1250 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1253 * Search for destination options headers or routing
1254 * header(s) through the header chain, and stores each
1255 * header as ancillary data.
1256 * Note that the order of the headers remains in
1257 * the chain of ancillary data.
1259 while (1) { /* is explicit loop prevention necessary? */
1260 struct ip6_ext *ip6e = NULL;
1262 #ifdef PULLDOWN_TEST
1263 struct mbuf *ext = NULL;
1267 * if it is not an extension header, don't try to
1268 * pull it from the chain.
1271 case IPPROTO_DSTOPTS:
1272 case IPPROTO_ROUTING:
1273 case IPPROTO_HOPOPTS:
1274 case IPPROTO_AH: /* is it possible? */
1280 #ifndef PULLDOWN_TEST
1281 if (off + sizeof(*ip6e) > m->m_len)
1283 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1284 if (nxt == IPPROTO_AH)
1285 elen = (ip6e->ip6e_len + 2) << 2;
1287 elen = (ip6e->ip6e_len + 1) << 3;
1288 if (off + elen > m->m_len)
1291 ext = ip6_pullexthdr(m, off, nxt);
1293 ip6stat.ip6s_tooshort++;
1296 ip6e = mtod(ext, struct ip6_ext *);
1297 if (nxt == IPPROTO_AH)
1298 elen = (ip6e->ip6e_len + 2) << 2;
1300 elen = (ip6e->ip6e_len + 1) << 3;
1301 if (elen != ext->m_len) {
1303 ip6stat.ip6s_tooshort++;
1309 case IPPROTO_DSTOPTS:
1310 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0)
1314 * We also require super-user privilege for
1316 * See the comments on IN6_HOPOPTS.
1321 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1325 mp = &(*mp)->m_next;
1327 case IPPROTO_ROUTING:
1328 if (!in6p->in6p_flags & IN6P_RTHDR)
1331 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1335 mp = &(*mp)->m_next;
1337 case IPPROTO_HOPOPTS:
1338 case IPPROTO_AH: /* is it possible? */
1343 * other cases have been filtered in the above.
1344 * none will visit this case. here we supply
1345 * the code just in case (nxt overwritten or
1348 #ifdef PULLDOWN_TEST
1355 /* proceed with the next header. */
1357 nxt = ip6e->ip6e_nxt;
1359 #ifdef PULLDOWN_TEST
1370 #ifdef PULLDOWN_TEST
1372 * pull single extension header from mbuf chain. returns single mbuf that
1373 * contains the result, or NULL on error.
1375 static struct mbuf *
1376 ip6_pullexthdr(m, off, nxt)
1381 struct ip6_ext ip6e;
1387 case IPPROTO_DSTOPTS:
1388 case IPPROTO_ROUTING:
1389 case IPPROTO_HOPOPTS:
1390 case IPPROTO_AH: /* is it possible? */
1393 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1397 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1398 if (nxt == IPPROTO_AH)
1399 elen = (ip6e.ip6e_len + 2) << 2;
1401 elen = (ip6e.ip6e_len + 1) << 3;
1403 MGET(n, M_DONTWAIT, MT_DATA);
1404 if (n && elen >= MLEN) {
1405 MCLGET(n, M_DONTWAIT);
1406 if ((n->m_flags & M_EXT) == 0) {
1415 if (elen >= M_TRAILINGSPACE(n)) {
1420 m_copydata(m, off, elen, mtod(n, caddr_t));
1427 * Get pointer to the previous header followed by the header
1428 * currently processed.
1429 * XXX: This function supposes that
1430 * M includes all headers,
1431 * the next header field and the header length field of each header
1433 * the sum of each header length equals to OFF.
1434 * Because of these assumptions, this function must be called very
1435 * carefully. Moreover, it will not be used in the near future when
1436 * we develop `neater' mechanism to process extension headers.
1439 ip6_get_prevhdr(m, off)
1443 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1445 if (off == sizeof(struct ip6_hdr))
1446 return(&ip6->ip6_nxt);
1449 struct ip6_ext *ip6e = NULL;
1452 len = sizeof(struct ip6_hdr);
1454 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1457 case IPPROTO_FRAGMENT:
1458 len += sizeof(struct ip6_frag);
1461 len += (ip6e->ip6e_len + 2) << 2;
1464 len += (ip6e->ip6e_len + 1) << 3;
1467 nxt = ip6e->ip6e_nxt;
1470 return(&ip6e->ip6e_nxt);
1477 * get next header offset. m will be retained.
1480 ip6_nexthdr(m, off, proto, nxtp)
1487 struct ip6_ext ip6e;
1492 panic("ip6_nexthdr: m == NULL");
1493 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1498 if (m->m_pkthdr.len < off + sizeof(ip6))
1500 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1502 *nxtp = ip6.ip6_nxt;
1506 case IPPROTO_FRAGMENT:
1508 * terminate parsing if it is not the first fragment,
1509 * it does not make sense to parse through it.
1511 if (m->m_pkthdr.len < off + sizeof(fh))
1513 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1514 if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1517 *nxtp = fh.ip6f_nxt;
1518 off += sizeof(struct ip6_frag);
1522 if (m->m_pkthdr.len < off + sizeof(ip6e))
1524 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1526 *nxtp = ip6e.ip6e_nxt;
1527 off += (ip6e.ip6e_len + 2) << 2;
1530 case IPPROTO_HOPOPTS:
1531 case IPPROTO_ROUTING:
1532 case IPPROTO_DSTOPTS:
1533 if (m->m_pkthdr.len < off + sizeof(ip6e))
1535 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1537 *nxtp = ip6e.ip6e_nxt;
1538 off += (ip6e.ip6e_len + 1) << 3;
1543 case IPPROTO_IPCOMP:
1555 * get offset for the last header in the chain. m will be kept untainted.
1558 ip6_lasthdr(m, off, proto, nxtp)
1572 newoff = ip6_nexthdr(m, off, proto, nxtp);
1575 else if (newoff < off)
1576 return -1; /* invalid */
1577 else if (newoff == off)
1589 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1591 tag = m_tag_get(PACKET_TAG_IPV6_INPUT,
1592 sizeof (struct ip6aux),
1595 m_tag_prepend(m, tag);
1598 bzero(tag+1, sizeof (struct ip6aux));
1599 return tag ? (struct ip6aux*)(tag+1) : NULL;
1606 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1607 return tag ? (struct ip6aux*)(tag+1) : NULL;
1614 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1616 m_tag_delete(m, tag);
1620 * System control for IP6
1623 u_char inet6ctlerrmap[PRC_NCMDS] = {
1625 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1626 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1627 EMSGSIZE, EHOSTUNREACH, 0, 0,