2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $
33 * Copyright (c) 1982, 1986, 1988, 1993
34 * The Regents of the University of California. All rights reserved.
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 4. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
63 #include <sys/cdefs.h>
64 __FBSDID("$FreeBSD$");
67 #include "opt_inet6.h"
69 #include "opt_ipsec.h"
70 #include "opt_route.h"
72 #include <sys/param.h>
73 #include <sys/systm.h>
74 #include <sys/malloc.h>
77 #include <sys/domain.h>
78 #include <sys/protosw.h>
79 #include <sys/socket.h>
80 #include <sys/socketvar.h>
81 #include <sys/errno.h>
83 #include <sys/kernel.h>
84 #include <sys/syslog.h>
87 #include <net/if_types.h>
88 #include <net/if_dl.h>
89 #include <net/route.h>
90 #include <net/netisr.h>
94 #include <netinet/in.h>
95 #include <netinet/ip_var.h>
96 #include <netinet/in_systm.h>
97 #include <net/if_llatbl.h>
99 #include <netinet/ip.h>
100 #include <netinet/ip_icmp.h>
102 #include <netinet/ip6.h>
103 #include <netinet6/in6_var.h>
104 #include <netinet6/ip6_var.h>
105 #include <netinet/in_pcb.h>
106 #include <netinet/icmp6.h>
107 #include <netinet6/scope6_var.h>
108 #include <netinet6/in6_ifattach.h>
109 #include <netinet6/nd6.h>
112 #include <netipsec/ipsec.h>
113 #include <netinet6/ip6_ipsec.h>
114 #include <netipsec/ipsec6.h>
117 #include <netinet6/ip6protosw.h>
120 #include <net/flowtable.h>
121 VNET_DECLARE(int, ip6_output_flowtable_size);
122 #define V_ip6_output_flowtable_size VNET(ip6_output_flowtable_size)
125 extern struct domain inet6domain;
127 u_char ip6_protox[IPPROTO_MAX];
128 VNET_DEFINE(struct in6_ifaddrhead, in6_ifaddrhead);
130 static struct netisr_handler ip6_nh = {
132 .nh_handler = ip6_input,
133 .nh_proto = NETISR_IPV6,
134 .nh_policy = NETISR_POLICY_FLOW,
137 VNET_DECLARE(struct callout, in6_tmpaddrtimer_ch);
138 #define V_in6_tmpaddrtimer_ch VNET(in6_tmpaddrtimer_ch)
140 VNET_DEFINE(struct pfil_head, inet6_pfil_hook);
142 VNET_DEFINE(struct ip6stat, ip6stat);
144 struct rwlock in6_ifaddr_lock;
145 RW_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock");
147 static void ip6_init2(void *);
148 static struct ip6aux *ip6_setdstifaddr(struct mbuf *, struct in6_ifaddr *);
149 static struct ip6aux *ip6_addaux(struct mbuf *);
150 static struct ip6aux *ip6_findaux(struct mbuf *m);
151 static void ip6_delaux (struct mbuf *);
152 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
154 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
158 * IP6 initialization: fill in IP6 protocol switch table.
159 * All protocols not implemented in kernel go to raw IP6 protocol handler.
164 struct ip6protosw *pr;
167 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal",
168 &V_ip6_auto_linklocal);
170 TAILQ_INIT(&V_in6_ifaddrhead);
172 /* Initialize packet filter hooks. */
173 V_inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
174 V_inet6_pfil_hook.ph_af = AF_INET6;
175 if ((i = pfil_head_register(&V_inet6_pfil_hook)) != 0)
176 printf("%s: WARNING: unable to register pfil hook, "
177 "error %d\n", __func__, i);
180 addrsel_policy_init();
185 if (TUNABLE_INT_FETCH("net.inet6.ip6.output_flowtable_size",
186 &V_ip6_output_flowtable_size)) {
187 if (V_ip6_output_flowtable_size < 256)
188 V_ip6_output_flowtable_size = 256;
189 if (!powerof2(V_ip6_output_flowtable_size)) {
190 printf("flowtable must be power of 2 size\n");
191 V_ip6_output_flowtable_size = 2048;
195 * round up to the next power of 2
197 V_ip6_output_flowtable_size = 1 << fls((1024 + maxusers * 64)-1);
199 V_ip6_ft = flowtable_alloc("ipv6", V_ip6_output_flowtable_size, FL_IPV6|FL_PCPU);
202 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
204 /* Skip global initialization stuff for non-default instances. */
205 if (!IS_DEFAULT_VNET(curvnet))
209 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
210 panic("sizeof(protosw) != sizeof(ip6protosw)");
212 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
216 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
217 for (i = 0; i < IPPROTO_MAX; i++)
218 ip6_protox[i] = pr - inet6sw;
220 * Cycle through IP protocols and put them into the appropriate place
223 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
224 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
225 if (pr->pr_domain->dom_family == PF_INET6 &&
226 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) {
227 /* Be careful to only index valid IP protocols. */
228 if (pr->pr_protocol < IPPROTO_MAX)
229 ip6_protox[pr->pr_protocol] = pr - inet6sw;
232 netisr_register(&ip6_nh);
236 * The protocol to be inserted into ip6_protox[] must be already registered
237 * in inet6sw[], either statically or through pf_proto_register().
240 ip6proto_register(short ip6proto)
242 struct ip6protosw *pr;
245 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
246 return (EPROTONOSUPPORT);
249 * The protocol slot must not be occupied by another protocol
250 * already. An index pointing to IPPROTO_RAW is unused.
252 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
254 return (EPFNOSUPPORT);
255 if (ip6_protox[ip6proto] != pr - inet6sw) /* IPPROTO_RAW */
259 * Find the protocol position in inet6sw[] and set the index.
261 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
262 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) {
263 if (pr->pr_domain->dom_family == PF_INET6 &&
264 pr->pr_protocol && pr->pr_protocol == ip6proto) {
265 ip6_protox[pr->pr_protocol] = pr - inet6sw;
269 return (EPROTONOSUPPORT);
273 ip6proto_unregister(short ip6proto)
275 struct ip6protosw *pr;
278 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
279 return (EPROTONOSUPPORT);
281 /* Check if the protocol was indeed registered. */
282 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
284 return (EPFNOSUPPORT);
285 if (ip6_protox[ip6proto] == pr - inet6sw) /* IPPROTO_RAW */
288 /* Reset the protocol slot to IPPROTO_RAW. */
289 ip6_protox[ip6proto] = pr - inet6sw;
299 callout_drain(&V_in6_tmpaddrtimer_ch);
304 ip6_init2_vnet(const void *unused __unused)
308 callout_init(&V_nd6_timer_ch, 0);
309 callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet);
311 /* timer for regeneranation of temporary addresses randomize ID */
312 callout_init(&V_in6_tmpaddrtimer_ch, 0);
313 callout_reset(&V_in6_tmpaddrtimer_ch,
314 (V_ip6_temp_preferred_lifetime - V_ip6_desync_factor -
315 V_ip6_temp_regen_advance) * hz,
316 in6_tmpaddrtimer, curvnet);
322 ip6_init2(void *dummy)
325 ip6_init2_vnet(NULL);
329 /* This must be after route_init(), which is now SI_ORDER_THIRD */
330 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
333 ip6_input_hbh(struct mbuf *m, uint32_t *plen, uint32_t *rtalert, int *off,
339 if (ip6_hopopts_input(plen, rtalert, &m, off)) {
340 #if 0 /*touches NULL pointer*/
341 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
343 goto out; /* m have already been freed */
347 ip6 = mtod(m, struct ip6_hdr *);
350 * if the payload length field is 0 and the next header field
351 * indicates Hop-by-Hop Options header, then a Jumbo Payload
352 * option MUST be included.
354 if (ip6->ip6_plen == 0 && *plen == 0) {
356 * Note that if a valid jumbo payload option is
357 * contained, ip6_hopopts_input() must set a valid
358 * (non-zero) payload length to the variable plen.
360 V_ip6stat.ip6s_badoptions++;
361 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
362 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
363 icmp6_error(m, ICMP6_PARAM_PROB,
364 ICMP6_PARAMPROB_HEADER,
365 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
368 #ifndef PULLDOWN_TEST
369 /* ip6_hopopts_input() ensures that mbuf is contiguous */
370 hbh = (struct ip6_hbh *)(ip6 + 1);
372 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
373 sizeof(struct ip6_hbh));
375 V_ip6stat.ip6s_tooshort++;
379 *nxt = hbh->ip6h_nxt;
382 * If we are acting as a router and the packet contains a
383 * router alert option, see if we know the option value.
384 * Currently, we only support the option value for MLD, in which
385 * case we should pass the packet to the multicast routing
388 if (*rtalert != ~0) {
390 case IP6OPT_RTALERT_MLD:
391 if (V_ip6_forwarding)
396 * RFC2711 requires unrecognized values must be
410 ip6_input(struct mbuf *m)
413 int off = sizeof(struct ip6_hdr), nest;
415 u_int32_t rtalert = ~0;
417 struct ifnet *deliverifp = NULL, *ifp = NULL;
418 struct in6_addr odst;
419 struct route_in6 rin6;
421 struct llentry *lle = NULL;
422 struct sockaddr_in6 dst6, *dst;
424 bzero(&rin6, sizeof(struct route_in6));
427 * should the inner packet be considered authentic?
428 * see comment in ah4_input().
429 * NB: m cannot be NULL when passed to the input routine
432 m->m_flags &= ~M_AUTHIPHDR;
433 m->m_flags &= ~M_AUTHIPDGM;
438 * make sure we don't have onion peering information into m_tag.
442 if (m->m_flags & M_FASTFWD_OURS) {
444 * Firewall changed destination to local.
446 m->m_flags &= ~M_FASTFWD_OURS;
448 deliverifp = m->m_pkthdr.rcvif;
449 ip6 = mtod(m, struct ip6_hdr *);
456 if (m->m_flags & M_EXT) {
458 V_ip6stat.ip6s_mext2m++;
460 V_ip6stat.ip6s_mext1++;
462 #define M2MMAX (sizeof(V_ip6stat.ip6s_m2m)/sizeof(V_ip6stat.ip6s_m2m[0]))
464 if (m->m_flags & M_LOOP) {
465 V_ip6stat.ip6s_m2m[V_loif->if_index]++;
466 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
467 V_ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
469 V_ip6stat.ip6s_m2m[0]++;
475 /* drop the packet if IPv6 operation is disabled on the IF */
476 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
481 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
482 V_ip6stat.ip6s_total++;
484 #ifndef PULLDOWN_TEST
486 * L2 bridge code and some other code can return mbuf chain
487 * that does not conform to KAME requirement. too bad.
488 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
490 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
493 MGETHDR(n, M_DONTWAIT, MT_HEADER);
496 if (n && n->m_pkthdr.len > MHLEN) {
497 MCLGET(n, M_DONTWAIT);
498 if ((n->m_flags & M_EXT) == 0) {
505 return; /* ENOBUFS */
508 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
509 n->m_len = n->m_pkthdr.len;
513 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
516 if (m->m_len < sizeof(struct ip6_hdr)) {
518 inifp = m->m_pkthdr.rcvif;
519 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
520 V_ip6stat.ip6s_toosmall++;
521 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
526 ip6 = mtod(m, struct ip6_hdr *);
528 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
529 V_ip6stat.ip6s_badvers++;
530 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
534 V_ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
537 * Check against address spoofing/corruption.
539 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
540 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
542 * XXX: "badscope" is not very suitable for a multicast source.
544 V_ip6stat.ip6s_badscope++;
545 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
548 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
549 !(m->m_flags & M_LOOP)) {
551 * In this case, the packet should come from the loopback
552 * interface. However, we cannot just check the if_flags,
553 * because ip6_mloopback() passes the "actual" interface
554 * as the outgoing/incoming interface.
556 V_ip6stat.ip6s_badscope++;
557 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
562 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
563 /* packet is dropped by traffic conditioner */
568 * The following check is not documented in specs. A malicious
569 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
570 * and bypass security checks (act as if it was from 127.0.0.1 by using
571 * IPv6 src ::ffff:127.0.0.1). Be cautious.
573 * This check chokes if we are in an SIIT cloud. As none of BSDs
574 * support IPv4-less kernel compilation, we cannot support SIIT
575 * environment at all. So, it makes more sense for us to reject any
576 * malicious packets for non-SIIT environment, than try to do a
577 * partial support for SIIT environment.
579 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
580 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
581 V_ip6stat.ip6s_badscope++;
582 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
587 * Reject packets with IPv4 compatible addresses (auto tunnel).
589 * The code forbids auto tunnel relay case in RFC1933 (the check is
590 * stronger than RFC1933). We may want to re-enable it if mech-xx
591 * is revised to forbid relaying case.
593 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
594 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
595 V_ip6stat.ip6s_badscope++;
596 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
602 * Bypass packet filtering for packets previously handled by IPsec.
604 if (ip6_ipsec_filtertunnel(m))
609 * Run through list of hooks for input packets.
611 * NB: Beware of the destination address changing
612 * (e.g. by NAT rewriting). When this happens,
613 * tell ip6_forward to do the right thing.
617 /* Jump over all PFIL processing if hooks are not active. */
618 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
621 if (pfil_run_hooks(&V_inet6_pfil_hook, &m,
622 m->m_pkthdr.rcvif, PFIL_IN, NULL))
624 if (m == NULL) /* consumed by filter */
626 ip6 = mtod(m, struct ip6_hdr *);
627 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
629 #ifdef IPFIREWALL_FORWARD
630 if (m->m_flags & M_FASTFWD_OURS) {
631 m->m_flags &= ~M_FASTFWD_OURS;
633 deliverifp = m->m_pkthdr.rcvif;
636 if (m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
638 * Directly ship the packet on. This allows forwarding
639 * packets originally destined to us to some other directly
645 #endif /* IPFIREWALL_FORWARD */
649 * Disambiguate address scope zones (if there is ambiguity).
650 * We first make sure that the original source or destination address
651 * is not in our internal form for scoped addresses. Such addresses
652 * are not necessarily invalid spec-wise, but we cannot accept them due
653 * to the usage conflict.
654 * in6_setscope() then also checks and rejects the cases where src or
655 * dst are the loopback address and the receiving interface
658 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
659 V_ip6stat.ip6s_badscope++; /* XXX */
662 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
663 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
664 V_ip6stat.ip6s_badscope++;
669 * Multicast check. Assume packet is for us to avoid
670 * prematurely taking locks.
672 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
674 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
675 deliverifp = m->m_pkthdr.rcvif;
683 bzero(&dst6, sizeof(dst6));
684 dst6.sin6_family = AF_INET6;
685 dst6.sin6_len = sizeof(struct sockaddr_in6);
686 dst6.sin6_addr = ip6->ip6_dst;
687 ifp = m->m_pkthdr.rcvif;
689 lle = lla_lookup(LLTABLE6(ifp), 0,
690 (struct sockaddr *)&dst6);
691 IF_AFDATA_UNLOCK(ifp);
692 if ((lle != NULL) && (lle->la_flags & LLE_IFADDR)) {
694 struct in6_ifaddr *ia6;
698 #define sa_equal(a1, a2) \
699 (bcmp((a1), (a2), ((a1))->sin6_len) == 0)
701 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
702 if (ifa->ifa_addr->sa_family != dst6.sin6_family)
704 if (sa_equal(&dst6, ifa->ifa_addr))
707 KASSERT(ifa != NULL, ("%s: ifa not found for lle %p",
711 ia6 = (struct in6_ifaddr *)ifa;
712 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
713 /* Count the packet in the ip address stats */
714 ia6->ia_ifa.if_ipackets++;
715 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
718 * record address information into m_tag.
720 (void)ip6_setdstifaddr(m, ia6);
724 char ip6bufs[INET6_ADDRSTRLEN];
725 char ip6bufd[INET6_ADDRSTRLEN];
726 /* address is not ready, so discard the packet. */
728 "ip6_input: packet to an unready address %s->%s\n",
729 ip6_sprintf(ip6bufs, &ip6->ip6_src),
730 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
732 IF_ADDR_RUNLOCK(ifp);
746 dst->sin6_len = sizeof(struct sockaddr_in6);
747 dst->sin6_family = AF_INET6;
748 dst->sin6_addr = ip6->ip6_dst;
749 rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
751 RT_UNLOCK(rin6.ro_rt);
753 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
756 * Accept the packet if the forwarding interface to the destination
757 * according to the routing table is the loopback interface,
758 * unless the associated route has a gateway.
759 * Note that this approach causes to accept a packet if there is a
760 * route to the loopback interface for the destination of the packet.
761 * But we think it's even useful in some situations, e.g. when using
762 * a special daemon which wants to intercept the packet.
764 * XXX: some OSes automatically make a cloned route for the destination
765 * of an outgoing packet. If the outgoing interface of the packet
766 * is a loopback one, the kernel would consider the packet to be
767 * accepted, even if we have no such address assinged on the interface.
768 * We check the cloned flag of the route entry to reject such cases,
769 * assuming that route entries for our own addresses are not made by
770 * cloning (it should be true because in6_addloop explicitly installs
771 * the host route). However, we might have to do an explicit check
772 * while it would be less efficient. Or, should we rather install a
773 * reject route for such a case?
776 (rin6.ro_rt->rt_flags &
777 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
779 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) &&
782 !(rin6.ro_rt->rt_flags & RTF_CLONED) &&
786 * The check below is redundant since the comparison of
787 * the destination and the key of the rtentry has
788 * already done through looking up the routing table.
790 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
791 &rt6_key(rin6.ro_rt)->sin6_addr)
793 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) {
795 struct in6_ifaddr *ia6;
798 * found the loopback route to the interface address
800 if (rin6.ro_rt->rt_gateway->sa_family == AF_LINK) {
801 struct sockaddr_in6 dest6;
803 bzero(&dest6, sizeof(dest6));
804 dest6.sin6_family = AF_INET6;
805 dest6.sin6_len = sizeof(dest6);
806 dest6.sin6_addr = ip6->ip6_dst;
807 ia6 = (struct in6_ifaddr *)
808 ifa_ifwithaddr((struct sockaddr *)&dest6);
814 ia6 = (struct in6_ifaddr *)rin6.ro_rt->rt_ifa;
817 * record address information into m_tag.
819 (void)ip6_setdstifaddr(m, ia6);
822 * packets to a tentative, duplicated, or somehow invalid
823 * address must not be accepted.
825 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
826 /* this address is ready */
828 deliverifp = ia6->ia_ifp; /* correct? */
829 /* Count the packet in the ip address stats */
830 ia6->ia_ifa.if_ipackets++;
831 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
832 if (ia6 != NULL && free_ia6 != 0)
833 ifa_free(&ia6->ia_ifa);
836 char ip6bufs[INET6_ADDRSTRLEN];
837 char ip6bufd[INET6_ADDRSTRLEN];
838 /* address is not ready, so discard the packet. */
840 "ip6_input: packet to an unready address %s->%s\n",
841 ip6_sprintf(ip6bufs, &ip6->ip6_src),
842 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
844 if (ia6 != NULL && free_ia6 != 0)
845 ifa_free(&ia6->ia_ifa);
851 * FAITH (Firewall Aided Internet Translator)
853 if (V_ip6_keepfaith) {
854 if (rin6.ro_rt && rin6.ro_rt->rt_ifp &&
855 rin6.ro_rt->rt_ifp->if_type == IFT_FAITH) {
856 /* XXX do we need more sanity checks? */
858 deliverifp = rin6.ro_rt->rt_ifp; /* faith */
864 * Now there is no reason to process the packet if it's not our own
865 * and we're not a router.
867 if (!V_ip6_forwarding) {
868 V_ip6stat.ip6s_cantforward++;
869 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
875 * record address information into m_tag, if we don't have one yet.
876 * note that we are unable to record it, if the address is not listed
877 * as our interface address (e.g. multicast addresses, addresses
878 * within FAITH prefixes and such).
881 struct in6_ifaddr *ia6;
883 if ((ia6 = ip6_getdstifaddr(m)) != NULL) {
884 ifa_free(&ia6->ia_ifa);
886 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
888 if (!ip6_setdstifaddr(m, ia6)) {
890 * XXX maybe we should drop the packet here,
891 * as we could not provide enough information
892 * to the upper layers.
895 ifa_free(&ia6->ia_ifa);
901 * Process Hop-by-Hop options header if it's contained.
902 * m may be modified in ip6_hopopts_input().
903 * If a JumboPayload option is included, plen will also be modified.
905 plen = (u_int32_t)ntohs(ip6->ip6_plen);
906 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
909 error = ip6_input_hbh(m, &plen, &rtalert, &off, &nxt, &ours);
916 * Check that the amount of data in the buffers
917 * is as at least much as the IPv6 header would have us expect.
918 * Trim mbufs if longer than we expect.
919 * Drop packet if shorter than we expect.
921 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
922 V_ip6stat.ip6s_tooshort++;
923 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
926 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
927 if (m->m_len == m->m_pkthdr.len) {
928 m->m_len = sizeof(struct ip6_hdr) + plen;
929 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
931 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
935 * Forward if desirable.
938 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
940 * If we are acting as a multicast router, all
941 * incoming multicast packets are passed to the
942 * kernel-level multicast forwarding function.
943 * The packet is returned (relatively) intact; if
944 * ip6_mforward() returns a non-zero value, the packet
945 * must be discarded, else it may be accepted below.
947 * XXX TODO: Check hlim and multicast scope here to avoid
948 * unnecessarily calling into ip6_mforward().
951 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
952 IP6STAT_INC(ip6s_cantforward);
953 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
957 ip6_forward(m, srcrt);
961 ip6 = mtod(m, struct ip6_hdr *);
964 * Malicious party may be able to use IPv4 mapped addr to confuse
965 * tcp/udp stack and bypass security checks (act as if it was from
966 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
968 * For SIIT end node behavior, you may want to disable the check.
969 * However, you will become vulnerable to attacks using IPv4 mapped
972 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
973 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
974 V_ip6stat.ip6s_badscope++;
975 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
980 * Tell launch routine the next header
982 V_ip6stat.ip6s_delivered++;
983 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
986 while (nxt != IPPROTO_DONE) {
987 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
988 V_ip6stat.ip6s_toomanyhdr++;
993 * protection against faulty packet - there should be
994 * more sanity checks in header chain processing.
996 if (m->m_pkthdr.len < off) {
997 V_ip6stat.ip6s_tooshort++;
998 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
1004 * enforce IPsec policy checking if we are seeing last header.
1005 * note that we do not visit this with protocols with pcb layer
1006 * code - like udp/tcp/raw ip.
1008 if (ip6_ipsec_input(m, nxt))
1013 * Use mbuf flags to propagate Router Alert option to
1014 * ICMPv6 layer, as hop-by-hop options have been stripped.
1016 if (nxt == IPPROTO_ICMPV6 && rtalert != ~0)
1017 m->m_flags |= M_RTALERT_MLD;
1019 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
1030 * set/grab in6_ifaddr correspond to IPv6 destination address.
1031 * XXX backward compatibility wrapper
1033 * XXXRW: We should bump the refcount on ia6 before sticking it in the m_tag,
1034 * and then bump it when the tag is copied, and release it when the tag is
1035 * freed. Unfortunately, m_tags don't support deep copies (yet), so instead
1036 * we just bump the ia refcount when we receive it. This should be fixed.
1038 static struct ip6aux *
1039 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6)
1041 struct ip6aux *ip6a;
1043 ip6a = ip6_addaux(m);
1045 ip6a->ip6a_dstia6 = ia6;
1046 return ip6a; /* NULL if failed to set */
1050 ip6_getdstifaddr(struct mbuf *m)
1052 struct ip6aux *ip6a;
1053 struct in6_ifaddr *ia;
1055 ip6a = ip6_findaux(m);
1057 ia = ip6a->ip6a_dstia6;
1058 ifa_ref(&ia->ia_ifa);
1065 * Hop-by-Hop options header processing. If a valid jumbo payload option is
1066 * included, the real payload length will be stored in plenp.
1068 * rtalertp - XXX: should be stored more smart way
1071 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
1072 struct mbuf **mp, int *offp)
1074 struct mbuf *m = *mp;
1075 int off = *offp, hbhlen;
1076 struct ip6_hbh *hbh;
1079 /* validation of the length of the header */
1080 #ifndef PULLDOWN_TEST
1081 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
1082 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1083 hbhlen = (hbh->ip6h_len + 1) << 3;
1085 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
1086 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1088 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1089 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1091 V_ip6stat.ip6s_tooshort++;
1094 hbhlen = (hbh->ip6h_len + 1) << 3;
1095 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
1098 V_ip6stat.ip6s_tooshort++;
1103 hbhlen -= sizeof(struct ip6_hbh);
1104 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
1106 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
1107 hbhlen, rtalertp, plenp) < 0)
1116 * Search header for all Hop-by-hop options and process each option.
1117 * This function is separate from ip6_hopopts_input() in order to
1118 * handle a case where the sending node itself process its hop-by-hop
1119 * options header. In such a case, the function is called from ip6_output().
1121 * The function assumes that hbh header is located right after the IPv6 header
1122 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1123 * opthead + hbhlen is located in contiguous memory region.
1126 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
1127 u_int32_t *rtalertp, u_int32_t *plenp)
1129 struct ip6_hdr *ip6;
1131 u_int8_t *opt = opthead;
1132 u_int16_t rtalert_val;
1133 u_int32_t jumboplen;
1134 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1136 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1142 if (hbhlen < IP6OPT_MINLEN) {
1143 V_ip6stat.ip6s_toosmall++;
1146 optlen = *(opt + 1) + 2;
1148 case IP6OPT_ROUTER_ALERT:
1149 /* XXX may need check for alignment */
1150 if (hbhlen < IP6OPT_RTALERT_LEN) {
1151 V_ip6stat.ip6s_toosmall++;
1154 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1156 icmp6_error(m, ICMP6_PARAM_PROB,
1157 ICMP6_PARAMPROB_HEADER,
1158 erroff + opt + 1 - opthead);
1161 optlen = IP6OPT_RTALERT_LEN;
1162 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1163 *rtalertp = ntohs(rtalert_val);
1166 /* XXX may need check for alignment */
1167 if (hbhlen < IP6OPT_JUMBO_LEN) {
1168 V_ip6stat.ip6s_toosmall++;
1171 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1173 icmp6_error(m, ICMP6_PARAM_PROB,
1174 ICMP6_PARAMPROB_HEADER,
1175 erroff + opt + 1 - opthead);
1178 optlen = IP6OPT_JUMBO_LEN;
1181 * IPv6 packets that have non 0 payload length
1182 * must not contain a jumbo payload option.
1184 ip6 = mtod(m, struct ip6_hdr *);
1185 if (ip6->ip6_plen) {
1186 V_ip6stat.ip6s_badoptions++;
1187 icmp6_error(m, ICMP6_PARAM_PROB,
1188 ICMP6_PARAMPROB_HEADER,
1189 erroff + opt - opthead);
1194 * We may see jumbolen in unaligned location, so
1195 * we'd need to perform bcopy().
1197 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1198 jumboplen = (u_int32_t)htonl(jumboplen);
1202 * if there are multiple jumbo payload options,
1203 * *plenp will be non-zero and the packet will be
1205 * the behavior may need some debate in ipngwg -
1206 * multiple options does not make sense, however,
1207 * there's no explicit mention in specification.
1210 V_ip6stat.ip6s_badoptions++;
1211 icmp6_error(m, ICMP6_PARAM_PROB,
1212 ICMP6_PARAMPROB_HEADER,
1213 erroff + opt + 2 - opthead);
1219 * jumbo payload length must be larger than 65535.
1221 if (jumboplen <= IPV6_MAXPACKET) {
1222 V_ip6stat.ip6s_badoptions++;
1223 icmp6_error(m, ICMP6_PARAM_PROB,
1224 ICMP6_PARAMPROB_HEADER,
1225 erroff + opt + 2 - opthead);
1231 default: /* unknown option */
1232 if (hbhlen < IP6OPT_MINLEN) {
1233 V_ip6stat.ip6s_toosmall++;
1236 optlen = ip6_unknown_opt(opt, m,
1237 erroff + opt - opthead);
1253 * Unknown option processing.
1254 * The third argument `off' is the offset from the IPv6 header to the option,
1255 * which is necessary if the IPv6 header the and option header and IPv6 header
1256 * is not contiguous in order to return an ICMPv6 error.
1259 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1261 struct ip6_hdr *ip6;
1263 switch (IP6OPT_TYPE(*optp)) {
1264 case IP6OPT_TYPE_SKIP: /* ignore the option */
1265 return ((int)*(optp + 1));
1266 case IP6OPT_TYPE_DISCARD: /* silently discard */
1269 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1270 V_ip6stat.ip6s_badoptions++;
1271 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1273 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1274 V_ip6stat.ip6s_badoptions++;
1275 ip6 = mtod(m, struct ip6_hdr *);
1276 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1277 (m->m_flags & (M_BCAST|M_MCAST)))
1280 icmp6_error(m, ICMP6_PARAM_PROB,
1281 ICMP6_PARAMPROB_OPTION, off);
1285 m_freem(m); /* XXX: NOTREACHED */
1290 * Create the "control" list for this pcb.
1291 * These functions will not modify mbuf chain at all.
1293 * With KAME mbuf chain restriction:
1294 * The routine will be called from upper layer handlers like tcp6_input().
1295 * Thus the routine assumes that the caller (tcp6_input) have already
1296 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1297 * very first mbuf on the mbuf chain.
1299 * ip6_savecontrol_v4 will handle those options that are possible to be
1300 * set on a v4-mapped socket.
1301 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1302 * options and handle the v6-only ones itself.
1305 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1308 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1311 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1315 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1316 SCM_TIMESTAMP, SOL_SOCKET);
1318 mp = &(*mp)->m_next;
1322 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1323 /* RFC 2292 sec. 5 */
1324 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1325 struct in6_pktinfo pi6;
1327 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1331 ip = mtod(m, struct ip *);
1332 pi6.ipi6_addr.s6_addr32[0] = 0;
1333 pi6.ipi6_addr.s6_addr32[1] = 0;
1334 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1335 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1337 /* We won't hit this code */
1338 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1341 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1342 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1345 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1347 *mp = sbcreatecontrol((caddr_t) &pi6,
1348 sizeof(struct in6_pktinfo),
1349 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1351 mp = &(*mp)->m_next;
1354 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1357 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1361 ip = mtod(m, struct ip *);
1364 /* We won't hit this code */
1368 hlim = ip6->ip6_hlim & 0xff;
1370 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1371 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1374 mp = &(*mp)->m_next;
1377 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1380 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1384 ip = mtod(m, struct ip *);
1385 tclass = ip->ip_tos;
1387 /* We won't hit this code */
1393 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1395 tclass = flowinfo & 0xff;
1397 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(int),
1398 IPV6_TCLASS, IPPROTO_IPV6);
1400 mp = &(*mp)->m_next;
1403 if (v4only != NULL) {
1404 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1415 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1417 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1420 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1425 * IPV6_HOPOPTS socket option. Recall that we required super-user
1426 * privilege for the option (see ip6_ctloutput), but it might be too
1427 * strict, since there might be some hop-by-hop options which can be
1428 * returned to normal user.
1429 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1431 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1433 * Check if a hop-by-hop options header is contatined in the
1434 * received packet, and if so, store the options as ancillary
1435 * data. Note that a hop-by-hop options header must be
1436 * just after the IPv6 header, which is assured through the
1437 * IPv6 input processing.
1439 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1440 struct ip6_hbh *hbh;
1442 #ifdef PULLDOWN_TEST
1446 #ifndef PULLDOWN_TEST
1447 hbh = (struct ip6_hbh *)(ip6 + 1);
1448 hbhlen = (hbh->ip6h_len + 1) << 3;
1450 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1453 V_ip6stat.ip6s_tooshort++;
1456 hbh = mtod(ext, struct ip6_hbh *);
1457 hbhlen = (hbh->ip6h_len + 1) << 3;
1458 if (hbhlen != ext->m_len) {
1460 V_ip6stat.ip6s_tooshort++;
1466 * XXX: We copy the whole header even if a
1467 * jumbo payload option is included, the option which
1468 * is to be removed before returning according to
1470 * Note: this constraint is removed in RFC3542
1472 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1473 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1476 mp = &(*mp)->m_next;
1477 #ifdef PULLDOWN_TEST
1483 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1484 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1487 * Search for destination options headers or routing
1488 * header(s) through the header chain, and stores each
1489 * header as ancillary data.
1490 * Note that the order of the headers remains in
1491 * the chain of ancillary data.
1493 while (1) { /* is explicit loop prevention necessary? */
1494 struct ip6_ext *ip6e = NULL;
1496 #ifdef PULLDOWN_TEST
1497 struct mbuf *ext = NULL;
1501 * if it is not an extension header, don't try to
1502 * pull it from the chain.
1505 case IPPROTO_DSTOPTS:
1506 case IPPROTO_ROUTING:
1507 case IPPROTO_HOPOPTS:
1508 case IPPROTO_AH: /* is it possible? */
1514 #ifndef PULLDOWN_TEST
1515 if (off + sizeof(*ip6e) > m->m_len)
1517 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1518 if (nxt == IPPROTO_AH)
1519 elen = (ip6e->ip6e_len + 2) << 2;
1521 elen = (ip6e->ip6e_len + 1) << 3;
1522 if (off + elen > m->m_len)
1525 ext = ip6_pullexthdr(m, off, nxt);
1527 V_ip6stat.ip6s_tooshort++;
1530 ip6e = mtod(ext, struct ip6_ext *);
1531 if (nxt == IPPROTO_AH)
1532 elen = (ip6e->ip6e_len + 2) << 2;
1534 elen = (ip6e->ip6e_len + 1) << 3;
1535 if (elen != ext->m_len) {
1537 V_ip6stat.ip6s_tooshort++;
1543 case IPPROTO_DSTOPTS:
1544 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1547 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1549 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1552 mp = &(*mp)->m_next;
1554 case IPPROTO_ROUTING:
1555 if (!(in6p->inp_flags & IN6P_RTHDR))
1558 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1559 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1562 mp = &(*mp)->m_next;
1564 case IPPROTO_HOPOPTS:
1565 case IPPROTO_AH: /* is it possible? */
1570 * other cases have been filtered in the above.
1571 * none will visit this case. here we supply
1572 * the code just in case (nxt overwritten or
1575 #ifdef PULLDOWN_TEST
1582 /* proceed with the next header. */
1584 nxt = ip6e->ip6e_nxt;
1586 #ifdef PULLDOWN_TEST
1598 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1602 struct ip6_mtuinfo mtuctl;
1604 so = in6p->inp_socket;
1610 if (so == NULL) /* I believe this is impossible */
1611 panic("ip6_notify_pmtu: socket is NULL");
1614 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1615 mtuctl.ip6m_mtu = *mtu;
1616 mtuctl.ip6m_addr = *dst;
1617 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1620 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1621 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1624 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1627 /* XXX: should count statistics */
1634 #ifdef PULLDOWN_TEST
1636 * pull single extension header from mbuf chain. returns single mbuf that
1637 * contains the result, or NULL on error.
1639 static struct mbuf *
1640 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1642 struct ip6_ext ip6e;
1648 case IPPROTO_DSTOPTS:
1649 case IPPROTO_ROUTING:
1650 case IPPROTO_HOPOPTS:
1651 case IPPROTO_AH: /* is it possible? */
1654 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1658 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1659 if (nxt == IPPROTO_AH)
1660 elen = (ip6e.ip6e_len + 2) << 2;
1662 elen = (ip6e.ip6e_len + 1) << 3;
1664 MGET(n, M_DONTWAIT, MT_DATA);
1665 if (n && elen >= MLEN) {
1666 MCLGET(n, M_DONTWAIT);
1667 if ((n->m_flags & M_EXT) == 0) {
1676 if (elen >= M_TRAILINGSPACE(n)) {
1681 m_copydata(m, off, elen, mtod(n, caddr_t));
1688 * Get pointer to the previous header followed by the header
1689 * currently processed.
1690 * XXX: This function supposes that
1691 * M includes all headers,
1692 * the next header field and the header length field of each header
1694 * the sum of each header length equals to OFF.
1695 * Because of these assumptions, this function must be called very
1696 * carefully. Moreover, it will not be used in the near future when
1697 * we develop `neater' mechanism to process extension headers.
1700 ip6_get_prevhdr(struct mbuf *m, int off)
1702 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1704 if (off == sizeof(struct ip6_hdr))
1705 return (&ip6->ip6_nxt);
1708 struct ip6_ext *ip6e = NULL;
1711 len = sizeof(struct ip6_hdr);
1713 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1716 case IPPROTO_FRAGMENT:
1717 len += sizeof(struct ip6_frag);
1720 len += (ip6e->ip6e_len + 2) << 2;
1723 len += (ip6e->ip6e_len + 1) << 3;
1726 nxt = ip6e->ip6e_nxt;
1729 return (&ip6e->ip6e_nxt);
1736 * get next header offset. m will be retained.
1739 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1742 struct ip6_ext ip6e;
1747 panic("ip6_nexthdr: m == NULL");
1748 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1753 if (m->m_pkthdr.len < off + sizeof(ip6))
1755 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1757 *nxtp = ip6.ip6_nxt;
1761 case IPPROTO_FRAGMENT:
1763 * terminate parsing if it is not the first fragment,
1764 * it does not make sense to parse through it.
1766 if (m->m_pkthdr.len < off + sizeof(fh))
1768 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1769 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1770 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1773 *nxtp = fh.ip6f_nxt;
1774 off += sizeof(struct ip6_frag);
1778 if (m->m_pkthdr.len < off + sizeof(ip6e))
1780 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1782 *nxtp = ip6e.ip6e_nxt;
1783 off += (ip6e.ip6e_len + 2) << 2;
1786 case IPPROTO_HOPOPTS:
1787 case IPPROTO_ROUTING:
1788 case IPPROTO_DSTOPTS:
1789 if (m->m_pkthdr.len < off + sizeof(ip6e))
1791 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1793 *nxtp = ip6e.ip6e_nxt;
1794 off += (ip6e.ip6e_len + 1) << 3;
1799 case IPPROTO_IPCOMP:
1811 * get offset for the last header in the chain. m will be kept untainted.
1814 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1824 newoff = ip6_nexthdr(m, off, proto, nxtp);
1827 else if (newoff < off)
1828 return -1; /* invalid */
1829 else if (newoff == off)
1837 static struct ip6aux *
1838 ip6_addaux(struct mbuf *m)
1842 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1844 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1847 m_tag_prepend(m, mtag);
1848 bzero(mtag + 1, sizeof(struct ip6aux));
1851 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1854 static struct ip6aux *
1855 ip6_findaux(struct mbuf *m)
1859 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1860 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1864 ip6_delaux(struct mbuf *m)
1868 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1870 m_tag_delete(m, mtag);
1874 * System control for IP6
1877 u_char inet6ctlerrmap[PRC_NCMDS] = {
1879 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1880 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1881 EMSGSIZE, EHOSTUNREACH, 0, 0,