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_kdtrace.h"
71 #include "opt_route.h"
73 #include <sys/param.h>
74 #include <sys/systm.h>
75 #include <sys/malloc.h>
78 #include <sys/domain.h>
79 #include <sys/protosw.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/errno.h>
85 #include <sys/kernel.h>
86 #include <sys/syslog.h>
89 #include <net/if_types.h>
90 #include <net/if_dl.h>
91 #include <net/route.h>
92 #include <net/netisr.h>
96 #include <netinet/in.h>
97 #include <netinet/in_kdtrace.h>
98 #include <netinet/ip_var.h>
99 #include <netinet/in_systm.h>
100 #include <net/if_llatbl.h>
102 #include <netinet/ip.h>
103 #include <netinet/ip_icmp.h>
105 #include <netinet/ip6.h>
106 #include <netinet6/in6_var.h>
107 #include <netinet6/ip6_var.h>
108 #include <netinet/in_pcb.h>
109 #include <netinet/icmp6.h>
110 #include <netinet6/scope6_var.h>
111 #include <netinet6/in6_ifattach.h>
112 #include <netinet6/nd6.h>
115 #include <netipsec/ipsec.h>
116 #include <netinet6/ip6_ipsec.h>
117 #include <netipsec/ipsec6.h>
120 #include <netinet6/ip6protosw.h>
123 #include <net/flowtable.h>
124 VNET_DECLARE(int, ip6_output_flowtable_size);
125 #define V_ip6_output_flowtable_size VNET(ip6_output_flowtable_size)
128 extern struct domain inet6domain;
130 u_char ip6_protox[IPPROTO_MAX];
131 VNET_DEFINE(struct in6_ifaddrhead, in6_ifaddrhead);
132 VNET_DEFINE(struct in6_ifaddrlisthead *, in6_ifaddrhashtbl);
133 VNET_DEFINE(u_long, in6_ifaddrhmask);
135 static struct netisr_handler ip6_nh = {
137 .nh_handler = ip6_input,
138 .nh_proto = NETISR_IPV6,
139 .nh_policy = NETISR_POLICY_FLOW,
142 VNET_DECLARE(struct callout, in6_tmpaddrtimer_ch);
143 #define V_in6_tmpaddrtimer_ch VNET(in6_tmpaddrtimer_ch)
145 VNET_DEFINE(struct pfil_head, inet6_pfil_hook);
147 VNET_PCPUSTAT_DEFINE(struct ip6stat, ip6stat);
148 VNET_PCPUSTAT_SYSINIT(ip6stat);
150 VNET_PCPUSTAT_SYSUNINIT(ip6stat);
153 struct rwlock in6_ifaddr_lock;
154 RW_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock");
156 static void ip6_init2(void *);
157 static struct ip6aux *ip6_setdstifaddr(struct mbuf *, struct in6_ifaddr *);
158 static struct ip6aux *ip6_addaux(struct mbuf *);
159 static struct ip6aux *ip6_findaux(struct mbuf *m);
160 static void ip6_delaux (struct mbuf *);
161 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
163 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;
176 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal",
177 &V_ip6_auto_linklocal);
178 TUNABLE_INT_FETCH("net.inet6.ip6.accept_rtadv", &V_ip6_accept_rtadv);
179 TUNABLE_INT_FETCH("net.inet6.ip6.no_radr", &V_ip6_no_radr);
181 TAILQ_INIT(&V_in6_ifaddrhead);
182 V_in6_ifaddrhashtbl = hashinit(IN6ADDR_NHASH, M_IFADDR,
185 /* Initialize packet filter hooks. */
186 V_inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
187 V_inet6_pfil_hook.ph_af = AF_INET6;
188 if ((i = pfil_head_register(&V_inet6_pfil_hook)) != 0)
189 printf("%s: WARNING: unable to register pfil hook, "
190 "error %d\n", __func__, i);
193 addrsel_policy_init();
198 if (TUNABLE_INT_FETCH("net.inet6.ip6.output_flowtable_size",
199 &V_ip6_output_flowtable_size)) {
200 if (V_ip6_output_flowtable_size < 256)
201 V_ip6_output_flowtable_size = 256;
202 if (!powerof2(V_ip6_output_flowtable_size)) {
203 printf("flowtable must be power of 2 size\n");
204 V_ip6_output_flowtable_size = 2048;
208 * round up to the next power of 2
210 V_ip6_output_flowtable_size = 1 << fls((1024 + maxusers * 64)-1);
212 V_ip6_ft = flowtable_alloc("ipv6", V_ip6_output_flowtable_size, FL_IPV6|FL_PCPU);
215 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
217 /* Skip global initialization stuff for non-default instances. */
218 if (!IS_DEFAULT_VNET(curvnet))
222 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
223 panic("sizeof(protosw) != sizeof(ip6protosw)");
225 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
229 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
230 for (i = 0; i < IPPROTO_MAX; i++)
231 ip6_protox[i] = pr - inet6sw;
233 * Cycle through IP protocols and put them into the appropriate place
236 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
237 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
238 if (pr->pr_domain->dom_family == PF_INET6 &&
239 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) {
240 /* Be careful to only index valid IP protocols. */
241 if (pr->pr_protocol < IPPROTO_MAX)
242 ip6_protox[pr->pr_protocol] = pr - inet6sw;
245 netisr_register(&ip6_nh);
249 * The protocol to be inserted into ip6_protox[] must be already registered
250 * in inet6sw[], either statically or through pf_proto_register().
253 ip6proto_register(short ip6proto)
255 struct ip6protosw *pr;
258 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
259 return (EPROTONOSUPPORT);
262 * The protocol slot must not be occupied by another protocol
263 * already. An index pointing to IPPROTO_RAW is unused.
265 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
267 return (EPFNOSUPPORT);
268 if (ip6_protox[ip6proto] != pr - inet6sw) /* IPPROTO_RAW */
272 * Find the protocol position in inet6sw[] and set the index.
274 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
275 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) {
276 if (pr->pr_domain->dom_family == PF_INET6 &&
277 pr->pr_protocol && pr->pr_protocol == ip6proto) {
278 ip6_protox[pr->pr_protocol] = pr - inet6sw;
282 return (EPROTONOSUPPORT);
286 ip6proto_unregister(short ip6proto)
288 struct ip6protosw *pr;
291 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
292 return (EPROTONOSUPPORT);
294 /* Check if the protocol was indeed registered. */
295 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
297 return (EPFNOSUPPORT);
298 if (ip6_protox[ip6proto] == pr - inet6sw) /* IPPROTO_RAW */
301 /* Reset the protocol slot to IPPROTO_RAW. */
302 ip6_protox[ip6proto] = pr - inet6sw;
312 if ((i = pfil_head_unregister(&V_inet6_pfil_hook)) != 0)
313 printf("%s: WARNING: unable to unregister pfil hook, "
314 "error %d\n", __func__, i);
315 hashdestroy(V_in6_ifaddrhashtbl, M_IFADDR, V_in6_ifaddrhmask);
317 callout_drain(&V_in6_tmpaddrtimer_ch);
322 ip6_init2_vnet(const void *unused __unused)
326 callout_init(&V_nd6_timer_ch, 0);
327 callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet);
329 /* timer for regeneranation of temporary addresses randomize ID */
330 callout_init(&V_in6_tmpaddrtimer_ch, 0);
331 callout_reset(&V_in6_tmpaddrtimer_ch,
332 (V_ip6_temp_preferred_lifetime - V_ip6_desync_factor -
333 V_ip6_temp_regen_advance) * hz,
334 in6_tmpaddrtimer, curvnet);
340 ip6_init2(void *dummy)
343 ip6_init2_vnet(NULL);
347 /* This must be after route_init(), which is now SI_ORDER_THIRD */
348 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
351 ip6_input_hbh(struct mbuf *m, uint32_t *plen, uint32_t *rtalert, int *off,
357 if (ip6_hopopts_input(plen, rtalert, &m, off)) {
358 #if 0 /*touches NULL pointer*/
359 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
361 goto out; /* m have already been freed */
365 ip6 = mtod(m, struct ip6_hdr *);
368 * if the payload length field is 0 and the next header field
369 * indicates Hop-by-Hop Options header, then a Jumbo Payload
370 * option MUST be included.
372 if (ip6->ip6_plen == 0 && *plen == 0) {
374 * Note that if a valid jumbo payload option is
375 * contained, ip6_hopopts_input() must set a valid
376 * (non-zero) payload length to the variable plen.
378 IP6STAT_INC(ip6s_badoptions);
379 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
380 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
381 icmp6_error(m, ICMP6_PARAM_PROB,
382 ICMP6_PARAMPROB_HEADER,
383 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
386 #ifndef PULLDOWN_TEST
387 /* ip6_hopopts_input() ensures that mbuf is contiguous */
388 hbh = (struct ip6_hbh *)(ip6 + 1);
390 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
391 sizeof(struct ip6_hbh));
393 IP6STAT_INC(ip6s_tooshort);
397 *nxt = hbh->ip6h_nxt;
400 * If we are acting as a router and the packet contains a
401 * router alert option, see if we know the option value.
402 * Currently, we only support the option value for MLD, in which
403 * case we should pass the packet to the multicast routing
406 if (*rtalert != ~0) {
408 case IP6OPT_RTALERT_MLD:
409 if (V_ip6_forwarding)
414 * RFC2711 requires unrecognized values must be
428 ip6_input(struct mbuf *m)
431 int off = sizeof(struct ip6_hdr), nest;
433 u_int32_t rtalert = ~0;
435 struct ifnet *deliverifp = NULL, *ifp = NULL;
436 struct in6_addr odst;
437 struct route_in6 rin6;
439 struct llentry *lle = NULL;
440 struct sockaddr_in6 dst6, *dst;
442 bzero(&rin6, sizeof(struct route_in6));
445 * should the inner packet be considered authentic?
446 * see comment in ah4_input().
447 * NB: m cannot be NULL when passed to the input routine
450 m->m_flags &= ~M_AUTHIPHDR;
451 m->m_flags &= ~M_AUTHIPDGM;
456 * make sure we don't have onion peering information into m_tag.
460 if (m->m_flags & M_FASTFWD_OURS) {
462 * Firewall changed destination to local.
464 m->m_flags &= ~M_FASTFWD_OURS;
466 deliverifp = m->m_pkthdr.rcvif;
467 ip6 = mtod(m, struct ip6_hdr *);
474 if (m->m_flags & M_EXT) {
476 IP6STAT_INC(ip6s_mext2m);
478 IP6STAT_INC(ip6s_mext1);
481 if (m->m_flags & M_LOOP) {
482 IP6STAT_INC(ip6s_m2m[V_loif->if_index]);
483 } else if (m->m_pkthdr.rcvif->if_index < IP6S_M2MMAX)
485 ip6s_m2m[m->m_pkthdr.rcvif->if_index]);
487 IP6STAT_INC(ip6s_m2m[0]);
489 IP6STAT_INC(ip6s_m1);
492 /* drop the packet if IPv6 operation is disabled on the IF */
493 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
498 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
499 IP6STAT_INC(ip6s_total);
501 #ifndef PULLDOWN_TEST
503 * L2 bridge code and some other code can return mbuf chain
504 * that does not conform to KAME requirement. too bad.
505 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
507 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
510 if (m->m_pkthdr.len > MHLEN)
511 n = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
513 n = m_gethdr(M_NOWAIT, MT_DATA);
516 return; /* ENOBUFS */
520 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
521 n->m_len = n->m_pkthdr.len;
525 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
528 if (m->m_len < sizeof(struct ip6_hdr)) {
530 inifp = m->m_pkthdr.rcvif;
531 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
532 IP6STAT_INC(ip6s_toosmall);
533 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
538 ip6 = mtod(m, struct ip6_hdr *);
540 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
541 IP6STAT_INC(ip6s_badvers);
542 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
546 IP6STAT_INC(ip6s_nxthist[ip6->ip6_nxt]);
548 IP_PROBE(receive, NULL, NULL, ip6, m->m_pkthdr.rcvif, NULL, ip6);
551 * Check against address spoofing/corruption.
553 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
554 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
556 * XXX: "badscope" is not very suitable for a multicast source.
558 IP6STAT_INC(ip6s_badscope);
559 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
562 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
563 !(m->m_flags & M_LOOP)) {
565 * In this case, the packet should come from the loopback
566 * interface. However, we cannot just check the if_flags,
567 * because ip6_mloopback() passes the "actual" interface
568 * as the outgoing/incoming interface.
570 IP6STAT_INC(ip6s_badscope);
571 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
576 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
577 /* packet is dropped by traffic conditioner */
582 * The following check is not documented in specs. A malicious
583 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
584 * and bypass security checks (act as if it was from 127.0.0.1 by using
585 * IPv6 src ::ffff:127.0.0.1). Be cautious.
587 * This check chokes if we are in an SIIT cloud. As none of BSDs
588 * support IPv4-less kernel compilation, we cannot support SIIT
589 * environment at all. So, it makes more sense for us to reject any
590 * malicious packets for non-SIIT environment, than try to do a
591 * partial support for SIIT environment.
593 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
594 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
595 IP6STAT_INC(ip6s_badscope);
596 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
601 * Reject packets with IPv4 compatible addresses (auto tunnel).
603 * The code forbids auto tunnel relay case in RFC1933 (the check is
604 * stronger than RFC1933). We may want to re-enable it if mech-xx
605 * is revised to forbid relaying case.
607 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
608 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
609 IP6STAT_INC(ip6s_badscope);
610 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
616 * Bypass packet filtering for packets previously handled by IPsec.
618 if (ip6_ipsec_filtertunnel(m))
623 * Run through list of hooks for input packets.
625 * NB: Beware of the destination address changing
626 * (e.g. by NAT rewriting). When this happens,
627 * tell ip6_forward to do the right thing.
631 /* Jump over all PFIL processing if hooks are not active. */
632 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
635 if (pfil_run_hooks(&V_inet6_pfil_hook, &m,
636 m->m_pkthdr.rcvif, PFIL_IN, NULL))
638 if (m == NULL) /* consumed by filter */
640 ip6 = mtod(m, struct ip6_hdr *);
641 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
643 if (m->m_flags & M_FASTFWD_OURS) {
644 m->m_flags &= ~M_FASTFWD_OURS;
646 deliverifp = m->m_pkthdr.rcvif;
649 if ((m->m_flags & M_IP6_NEXTHOP) &&
650 m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
652 * Directly ship the packet on. This allows forwarding
653 * packets originally destined to us to some other directly
662 * Disambiguate address scope zones (if there is ambiguity).
663 * We first make sure that the original source or destination address
664 * is not in our internal form for scoped addresses. Such addresses
665 * are not necessarily invalid spec-wise, but we cannot accept them due
666 * to the usage conflict.
667 * in6_setscope() then also checks and rejects the cases where src or
668 * dst are the loopback address and the receiving interface
671 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
672 IP6STAT_INC(ip6s_badscope); /* XXX */
675 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
676 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
677 IP6STAT_INC(ip6s_badscope);
682 * Multicast check. Assume packet is for us to avoid
683 * prematurely taking locks.
685 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
687 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
688 deliverifp = m->m_pkthdr.rcvif;
696 bzero(&dst6, sizeof(dst6));
697 dst6.sin6_family = AF_INET6;
698 dst6.sin6_len = sizeof(struct sockaddr_in6);
699 dst6.sin6_addr = ip6->ip6_dst;
700 ifp = m->m_pkthdr.rcvif;
701 IF_AFDATA_RLOCK(ifp);
702 lle = lla_lookup(LLTABLE6(ifp), 0,
703 (struct sockaddr *)&dst6);
704 IF_AFDATA_RUNLOCK(ifp);
705 if ((lle != NULL) && (lle->la_flags & LLE_IFADDR)) {
707 struct in6_ifaddr *ia6;
711 #define sa_equal(a1, a2) \
712 (bcmp((a1), (a2), ((a1))->sin6_len) == 0)
714 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
715 if (ifa->ifa_addr->sa_family != dst6.sin6_family)
717 if (sa_equal(&dst6, ifa->ifa_addr))
720 KASSERT(ifa != NULL, ("%s: ifa not found for lle %p",
724 ia6 = (struct in6_ifaddr *)ifa;
725 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
726 /* Count the packet in the ip address stats */
727 ia6->ia_ifa.if_ipackets++;
728 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
731 * record address information into m_tag.
733 (void)ip6_setdstifaddr(m, ia6);
737 char ip6bufs[INET6_ADDRSTRLEN];
738 char ip6bufd[INET6_ADDRSTRLEN];
739 /* address is not ready, so discard the packet. */
741 "ip6_input: packet to an unready address %s->%s\n",
742 ip6_sprintf(ip6bufs, &ip6->ip6_src),
743 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
745 IF_ADDR_RUNLOCK(ifp);
759 dst->sin6_len = sizeof(struct sockaddr_in6);
760 dst->sin6_family = AF_INET6;
761 dst->sin6_addr = ip6->ip6_dst;
762 rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
764 RT_UNLOCK(rin6.ro_rt);
766 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
769 * Accept the packet if the forwarding interface to the destination
770 * according to the routing table is the loopback interface,
771 * unless the associated route has a gateway.
772 * Note that this approach causes to accept a packet if there is a
773 * route to the loopback interface for the destination of the packet.
774 * But we think it's even useful in some situations, e.g. when using
775 * a special daemon which wants to intercept the packet.
777 * XXX: some OSes automatically make a cloned route for the destination
778 * of an outgoing packet. If the outgoing interface of the packet
779 * is a loopback one, the kernel would consider the packet to be
780 * accepted, even if we have no such address assinged on the interface.
781 * We check the cloned flag of the route entry to reject such cases,
782 * assuming that route entries for our own addresses are not made by
783 * cloning (it should be true because in6_addloop explicitly installs
784 * the host route). However, we might have to do an explicit check
785 * while it would be less efficient. Or, should we rather install a
786 * reject route for such a case?
789 (rin6.ro_rt->rt_flags &
790 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
792 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) &&
795 !(rin6.ro_rt->rt_flags & RTF_CLONED) &&
799 * The check below is redundant since the comparison of
800 * the destination and the key of the rtentry has
801 * already done through looking up the routing table.
803 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
804 &rt6_key(rin6.ro_rt)->sin6_addr)
806 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) {
808 struct in6_ifaddr *ia6;
811 * found the loopback route to the interface address
813 if (rin6.ro_rt->rt_gateway->sa_family == AF_LINK) {
814 struct sockaddr_in6 dest6;
816 bzero(&dest6, sizeof(dest6));
817 dest6.sin6_family = AF_INET6;
818 dest6.sin6_len = sizeof(dest6);
819 dest6.sin6_addr = ip6->ip6_dst;
820 ia6 = (struct in6_ifaddr *)
821 ifa_ifwithaddr((struct sockaddr *)&dest6);
827 ia6 = (struct in6_ifaddr *)rin6.ro_rt->rt_ifa;
830 * record address information into m_tag.
832 (void)ip6_setdstifaddr(m, ia6);
835 * packets to a tentative, duplicated, or somehow invalid
836 * address must not be accepted.
838 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
839 /* this address is ready */
841 deliverifp = ia6->ia_ifp; /* correct? */
842 /* Count the packet in the ip address stats */
843 ia6->ia_ifa.if_ipackets++;
844 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
845 if (ia6 != NULL && free_ia6 != 0)
846 ifa_free(&ia6->ia_ifa);
849 char ip6bufs[INET6_ADDRSTRLEN];
850 char ip6bufd[INET6_ADDRSTRLEN];
851 /* address is not ready, so discard the packet. */
853 "ip6_input: packet to an unready address %s->%s\n",
854 ip6_sprintf(ip6bufs, &ip6->ip6_src),
855 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
857 if (ia6 != NULL && free_ia6 != 0)
858 ifa_free(&ia6->ia_ifa);
864 * FAITH (Firewall Aided Internet Translator)
866 if (V_ip6_keepfaith) {
867 if (rin6.ro_rt && rin6.ro_rt->rt_ifp &&
868 rin6.ro_rt->rt_ifp->if_type == IFT_FAITH) {
869 /* XXX do we need more sanity checks? */
871 deliverifp = rin6.ro_rt->rt_ifp; /* faith */
877 * Now there is no reason to process the packet if it's not our own
878 * and we're not a router.
880 if (!V_ip6_forwarding) {
881 IP6STAT_INC(ip6s_cantforward);
882 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
888 * record address information into m_tag, if we don't have one yet.
889 * note that we are unable to record it, if the address is not listed
890 * as our interface address (e.g. multicast addresses, addresses
891 * within FAITH prefixes and such).
894 struct in6_ifaddr *ia6;
896 if ((ia6 = ip6_getdstifaddr(m)) != NULL) {
897 ifa_free(&ia6->ia_ifa);
899 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
901 if (!ip6_setdstifaddr(m, ia6)) {
903 * XXX maybe we should drop the packet here,
904 * as we could not provide enough information
905 * to the upper layers.
908 ifa_free(&ia6->ia_ifa);
914 * Process Hop-by-Hop options header if it's contained.
915 * m may be modified in ip6_hopopts_input().
916 * If a JumboPayload option is included, plen will also be modified.
918 plen = (u_int32_t)ntohs(ip6->ip6_plen);
919 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
922 error = ip6_input_hbh(m, &plen, &rtalert, &off, &nxt, &ours);
929 * Check that the amount of data in the buffers
930 * is as at least much as the IPv6 header would have us expect.
931 * Trim mbufs if longer than we expect.
932 * Drop packet if shorter than we expect.
934 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
935 IP6STAT_INC(ip6s_tooshort);
936 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
939 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
940 if (m->m_len == m->m_pkthdr.len) {
941 m->m_len = sizeof(struct ip6_hdr) + plen;
942 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
944 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
948 * Forward if desirable.
951 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
953 * If we are acting as a multicast router, all
954 * incoming multicast packets are passed to the
955 * kernel-level multicast forwarding function.
956 * The packet is returned (relatively) intact; if
957 * ip6_mforward() returns a non-zero value, the packet
958 * must be discarded, else it may be accepted below.
960 * XXX TODO: Check hlim and multicast scope here to avoid
961 * unnecessarily calling into ip6_mforward().
964 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
965 IP6STAT_INC(ip6s_cantforward);
966 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
970 ip6_forward(m, srcrt);
974 ip6 = mtod(m, struct ip6_hdr *);
977 * Malicious party may be able to use IPv4 mapped addr to confuse
978 * tcp/udp stack and bypass security checks (act as if it was from
979 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
981 * For SIIT end node behavior, you may want to disable the check.
982 * However, you will become vulnerable to attacks using IPv4 mapped
985 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
986 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
987 IP6STAT_INC(ip6s_badscope);
988 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
993 * Tell launch routine the next header
995 IP6STAT_INC(ip6s_delivered);
996 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
999 while (nxt != IPPROTO_DONE) {
1000 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
1001 IP6STAT_INC(ip6s_toomanyhdr);
1006 * protection against faulty packet - there should be
1007 * more sanity checks in header chain processing.
1009 if (m->m_pkthdr.len < off) {
1010 IP6STAT_INC(ip6s_tooshort);
1011 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
1017 * enforce IPsec policy checking if we are seeing last header.
1018 * note that we do not visit this with protocols with pcb layer
1019 * code - like udp/tcp/raw ip.
1021 if (ip6_ipsec_input(m, nxt))
1026 * Use mbuf flags to propagate Router Alert option to
1027 * ICMPv6 layer, as hop-by-hop options have been stripped.
1029 if (nxt == IPPROTO_ICMPV6 && rtalert != ~0)
1030 m->m_flags |= M_RTALERT_MLD;
1032 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
1043 * set/grab in6_ifaddr correspond to IPv6 destination address.
1044 * XXX backward compatibility wrapper
1046 * XXXRW: We should bump the refcount on ia6 before sticking it in the m_tag,
1047 * and then bump it when the tag is copied, and release it when the tag is
1048 * freed. Unfortunately, m_tags don't support deep copies (yet), so instead
1049 * we just bump the ia refcount when we receive it. This should be fixed.
1051 static struct ip6aux *
1052 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6)
1054 struct ip6aux *ip6a;
1056 ip6a = ip6_addaux(m);
1058 ip6a->ip6a_dstia6 = ia6;
1059 return ip6a; /* NULL if failed to set */
1063 ip6_getdstifaddr(struct mbuf *m)
1065 struct ip6aux *ip6a;
1066 struct in6_ifaddr *ia;
1068 ip6a = ip6_findaux(m);
1070 ia = ip6a->ip6a_dstia6;
1071 ifa_ref(&ia->ia_ifa);
1078 * Hop-by-Hop options header processing. If a valid jumbo payload option is
1079 * included, the real payload length will be stored in plenp.
1081 * rtalertp - XXX: should be stored more smart way
1084 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
1085 struct mbuf **mp, int *offp)
1087 struct mbuf *m = *mp;
1088 int off = *offp, hbhlen;
1089 struct ip6_hbh *hbh;
1092 /* validation of the length of the header */
1093 #ifndef PULLDOWN_TEST
1094 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
1095 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1096 hbhlen = (hbh->ip6h_len + 1) << 3;
1098 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
1099 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1101 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1102 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1104 IP6STAT_INC(ip6s_tooshort);
1107 hbhlen = (hbh->ip6h_len + 1) << 3;
1108 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
1111 IP6STAT_INC(ip6s_tooshort);
1116 hbhlen -= sizeof(struct ip6_hbh);
1117 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
1119 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
1120 hbhlen, rtalertp, plenp) < 0)
1129 * Search header for all Hop-by-hop options and process each option.
1130 * This function is separate from ip6_hopopts_input() in order to
1131 * handle a case where the sending node itself process its hop-by-hop
1132 * options header. In such a case, the function is called from ip6_output().
1134 * The function assumes that hbh header is located right after the IPv6 header
1135 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1136 * opthead + hbhlen is located in contiguous memory region.
1139 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
1140 u_int32_t *rtalertp, u_int32_t *plenp)
1142 struct ip6_hdr *ip6;
1144 u_int8_t *opt = opthead;
1145 u_int16_t rtalert_val;
1146 u_int32_t jumboplen;
1147 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1149 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1155 if (hbhlen < IP6OPT_MINLEN) {
1156 IP6STAT_INC(ip6s_toosmall);
1159 optlen = *(opt + 1) + 2;
1161 case IP6OPT_ROUTER_ALERT:
1162 /* XXX may need check for alignment */
1163 if (hbhlen < IP6OPT_RTALERT_LEN) {
1164 IP6STAT_INC(ip6s_toosmall);
1167 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1169 icmp6_error(m, ICMP6_PARAM_PROB,
1170 ICMP6_PARAMPROB_HEADER,
1171 erroff + opt + 1 - opthead);
1174 optlen = IP6OPT_RTALERT_LEN;
1175 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1176 *rtalertp = ntohs(rtalert_val);
1179 /* XXX may need check for alignment */
1180 if (hbhlen < IP6OPT_JUMBO_LEN) {
1181 IP6STAT_INC(ip6s_toosmall);
1184 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1186 icmp6_error(m, ICMP6_PARAM_PROB,
1187 ICMP6_PARAMPROB_HEADER,
1188 erroff + opt + 1 - opthead);
1191 optlen = IP6OPT_JUMBO_LEN;
1194 * IPv6 packets that have non 0 payload length
1195 * must not contain a jumbo payload option.
1197 ip6 = mtod(m, struct ip6_hdr *);
1198 if (ip6->ip6_plen) {
1199 IP6STAT_INC(ip6s_badoptions);
1200 icmp6_error(m, ICMP6_PARAM_PROB,
1201 ICMP6_PARAMPROB_HEADER,
1202 erroff + opt - opthead);
1207 * We may see jumbolen in unaligned location, so
1208 * we'd need to perform bcopy().
1210 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1211 jumboplen = (u_int32_t)htonl(jumboplen);
1215 * if there are multiple jumbo payload options,
1216 * *plenp will be non-zero and the packet will be
1218 * the behavior may need some debate in ipngwg -
1219 * multiple options does not make sense, however,
1220 * there's no explicit mention in specification.
1223 IP6STAT_INC(ip6s_badoptions);
1224 icmp6_error(m, ICMP6_PARAM_PROB,
1225 ICMP6_PARAMPROB_HEADER,
1226 erroff + opt + 2 - opthead);
1232 * jumbo payload length must be larger than 65535.
1234 if (jumboplen <= IPV6_MAXPACKET) {
1235 IP6STAT_INC(ip6s_badoptions);
1236 icmp6_error(m, ICMP6_PARAM_PROB,
1237 ICMP6_PARAMPROB_HEADER,
1238 erroff + opt + 2 - opthead);
1244 default: /* unknown option */
1245 if (hbhlen < IP6OPT_MINLEN) {
1246 IP6STAT_INC(ip6s_toosmall);
1249 optlen = ip6_unknown_opt(opt, m,
1250 erroff + opt - opthead);
1266 * Unknown option processing.
1267 * The third argument `off' is the offset from the IPv6 header to the option,
1268 * which is necessary if the IPv6 header the and option header and IPv6 header
1269 * is not contiguous in order to return an ICMPv6 error.
1272 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1274 struct ip6_hdr *ip6;
1276 switch (IP6OPT_TYPE(*optp)) {
1277 case IP6OPT_TYPE_SKIP: /* ignore the option */
1278 return ((int)*(optp + 1));
1279 case IP6OPT_TYPE_DISCARD: /* silently discard */
1282 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1283 IP6STAT_INC(ip6s_badoptions);
1284 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1286 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1287 IP6STAT_INC(ip6s_badoptions);
1288 ip6 = mtod(m, struct ip6_hdr *);
1289 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1290 (m->m_flags & (M_BCAST|M_MCAST)))
1293 icmp6_error(m, ICMP6_PARAM_PROB,
1294 ICMP6_PARAMPROB_OPTION, off);
1298 m_freem(m); /* XXX: NOTREACHED */
1303 * Create the "control" list for this pcb.
1304 * These functions will not modify mbuf chain at all.
1306 * With KAME mbuf chain restriction:
1307 * The routine will be called from upper layer handlers like tcp6_input().
1308 * Thus the routine assumes that the caller (tcp6_input) have already
1309 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1310 * very first mbuf on the mbuf chain.
1312 * ip6_savecontrol_v4 will handle those options that are possible to be
1313 * set on a v4-mapped socket.
1314 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1315 * options and handle the v6-only ones itself.
1318 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1321 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1324 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1328 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1329 SCM_TIMESTAMP, SOL_SOCKET);
1331 mp = &(*mp)->m_next;
1335 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1336 /* RFC 2292 sec. 5 */
1337 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1338 struct in6_pktinfo pi6;
1340 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1344 ip = mtod(m, struct ip *);
1345 pi6.ipi6_addr.s6_addr32[0] = 0;
1346 pi6.ipi6_addr.s6_addr32[1] = 0;
1347 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1348 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1350 /* We won't hit this code */
1351 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1354 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1355 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1358 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1360 *mp = sbcreatecontrol((caddr_t) &pi6,
1361 sizeof(struct in6_pktinfo),
1362 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1364 mp = &(*mp)->m_next;
1367 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1370 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1374 ip = mtod(m, struct ip *);
1377 /* We won't hit this code */
1381 hlim = ip6->ip6_hlim & 0xff;
1383 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1384 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1387 mp = &(*mp)->m_next;
1390 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1393 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1397 ip = mtod(m, struct ip *);
1398 tclass = ip->ip_tos;
1400 /* We won't hit this code */
1406 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1408 tclass = flowinfo & 0xff;
1410 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(int),
1411 IPV6_TCLASS, IPPROTO_IPV6);
1413 mp = &(*mp)->m_next;
1416 if (v4only != NULL) {
1417 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1428 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1430 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1433 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1438 * IPV6_HOPOPTS socket option. Recall that we required super-user
1439 * privilege for the option (see ip6_ctloutput), but it might be too
1440 * strict, since there might be some hop-by-hop options which can be
1441 * returned to normal user.
1442 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1444 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1446 * Check if a hop-by-hop options header is contatined in the
1447 * received packet, and if so, store the options as ancillary
1448 * data. Note that a hop-by-hop options header must be
1449 * just after the IPv6 header, which is assured through the
1450 * IPv6 input processing.
1452 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1453 struct ip6_hbh *hbh;
1455 #ifdef PULLDOWN_TEST
1459 #ifndef PULLDOWN_TEST
1460 hbh = (struct ip6_hbh *)(ip6 + 1);
1461 hbhlen = (hbh->ip6h_len + 1) << 3;
1463 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1466 IP6STAT_INC(ip6s_tooshort);
1469 hbh = mtod(ext, struct ip6_hbh *);
1470 hbhlen = (hbh->ip6h_len + 1) << 3;
1471 if (hbhlen != ext->m_len) {
1473 IP6STAT_INC(ip6s_tooshort);
1479 * XXX: We copy the whole header even if a
1480 * jumbo payload option is included, the option which
1481 * is to be removed before returning according to
1483 * Note: this constraint is removed in RFC3542
1485 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1486 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1489 mp = &(*mp)->m_next;
1490 #ifdef PULLDOWN_TEST
1496 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1497 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1500 * Search for destination options headers or routing
1501 * header(s) through the header chain, and stores each
1502 * header as ancillary data.
1503 * Note that the order of the headers remains in
1504 * the chain of ancillary data.
1506 while (1) { /* is explicit loop prevention necessary? */
1507 struct ip6_ext *ip6e = NULL;
1509 #ifdef PULLDOWN_TEST
1510 struct mbuf *ext = NULL;
1514 * if it is not an extension header, don't try to
1515 * pull it from the chain.
1518 case IPPROTO_DSTOPTS:
1519 case IPPROTO_ROUTING:
1520 case IPPROTO_HOPOPTS:
1521 case IPPROTO_AH: /* is it possible? */
1527 #ifndef PULLDOWN_TEST
1528 if (off + sizeof(*ip6e) > m->m_len)
1530 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1531 if (nxt == IPPROTO_AH)
1532 elen = (ip6e->ip6e_len + 2) << 2;
1534 elen = (ip6e->ip6e_len + 1) << 3;
1535 if (off + elen > m->m_len)
1538 ext = ip6_pullexthdr(m, off, nxt);
1540 IP6STAT_INC(ip6s_tooshort);
1543 ip6e = mtod(ext, struct ip6_ext *);
1544 if (nxt == IPPROTO_AH)
1545 elen = (ip6e->ip6e_len + 2) << 2;
1547 elen = (ip6e->ip6e_len + 1) << 3;
1548 if (elen != ext->m_len) {
1550 IP6STAT_INC(ip6s_tooshort);
1556 case IPPROTO_DSTOPTS:
1557 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1560 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1562 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1565 mp = &(*mp)->m_next;
1567 case IPPROTO_ROUTING:
1568 if (!(in6p->inp_flags & IN6P_RTHDR))
1571 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1572 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1575 mp = &(*mp)->m_next;
1577 case IPPROTO_HOPOPTS:
1578 case IPPROTO_AH: /* is it possible? */
1583 * other cases have been filtered in the above.
1584 * none will visit this case. here we supply
1585 * the code just in case (nxt overwritten or
1588 #ifdef PULLDOWN_TEST
1595 /* proceed with the next header. */
1597 nxt = ip6e->ip6e_nxt;
1599 #ifdef PULLDOWN_TEST
1611 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1615 struct ip6_mtuinfo mtuctl;
1617 so = in6p->inp_socket;
1623 if (so == NULL) /* I believe this is impossible */
1624 panic("ip6_notify_pmtu: socket is NULL");
1627 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1628 mtuctl.ip6m_mtu = *mtu;
1629 mtuctl.ip6m_addr = *dst;
1630 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1633 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1634 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1637 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1640 /* XXX: should count statistics */
1647 #ifdef PULLDOWN_TEST
1649 * pull single extension header from mbuf chain. returns single mbuf that
1650 * contains the result, or NULL on error.
1652 static struct mbuf *
1653 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1655 struct ip6_ext ip6e;
1661 case IPPROTO_DSTOPTS:
1662 case IPPROTO_ROUTING:
1663 case IPPROTO_HOPOPTS:
1664 case IPPROTO_AH: /* is it possible? */
1667 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1671 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1672 if (nxt == IPPROTO_AH)
1673 elen = (ip6e.ip6e_len + 2) << 2;
1675 elen = (ip6e.ip6e_len + 1) << 3;
1678 n = m_getcl(M_NOWAIT, MT_DATA, 0);
1680 n = m_get(M_NOWAIT, MT_DATA);
1684 m_copydata(m, off, elen, mtod(n, caddr_t));
1691 * Get pointer to the previous header followed by the header
1692 * currently processed.
1693 * XXX: This function supposes that
1694 * M includes all headers,
1695 * the next header field and the header length field of each header
1697 * the sum of each header length equals to OFF.
1698 * Because of these assumptions, this function must be called very
1699 * carefully. Moreover, it will not be used in the near future when
1700 * we develop `neater' mechanism to process extension headers.
1703 ip6_get_prevhdr(struct mbuf *m, int off)
1705 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1707 if (off == sizeof(struct ip6_hdr))
1708 return (&ip6->ip6_nxt);
1711 struct ip6_ext *ip6e = NULL;
1714 len = sizeof(struct ip6_hdr);
1716 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1719 case IPPROTO_FRAGMENT:
1720 len += sizeof(struct ip6_frag);
1723 len += (ip6e->ip6e_len + 2) << 2;
1726 len += (ip6e->ip6e_len + 1) << 3;
1729 nxt = ip6e->ip6e_nxt;
1732 return (&ip6e->ip6e_nxt);
1739 * get next header offset. m will be retained.
1742 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1745 struct ip6_ext ip6e;
1750 panic("ip6_nexthdr: m == NULL");
1751 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1756 if (m->m_pkthdr.len < off + sizeof(ip6))
1758 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1760 *nxtp = ip6.ip6_nxt;
1764 case IPPROTO_FRAGMENT:
1766 * terminate parsing if it is not the first fragment,
1767 * it does not make sense to parse through it.
1769 if (m->m_pkthdr.len < off + sizeof(fh))
1771 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1772 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1773 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1776 *nxtp = fh.ip6f_nxt;
1777 off += sizeof(struct ip6_frag);
1781 if (m->m_pkthdr.len < off + sizeof(ip6e))
1783 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1785 *nxtp = ip6e.ip6e_nxt;
1786 off += (ip6e.ip6e_len + 2) << 2;
1789 case IPPROTO_HOPOPTS:
1790 case IPPROTO_ROUTING:
1791 case IPPROTO_DSTOPTS:
1792 if (m->m_pkthdr.len < off + sizeof(ip6e))
1794 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1796 *nxtp = ip6e.ip6e_nxt;
1797 off += (ip6e.ip6e_len + 1) << 3;
1802 case IPPROTO_IPCOMP:
1814 * get offset for the last header in the chain. m will be kept untainted.
1817 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1827 newoff = ip6_nexthdr(m, off, proto, nxtp);
1830 else if (newoff < off)
1831 return -1; /* invalid */
1832 else if (newoff == off)
1840 static struct ip6aux *
1841 ip6_addaux(struct mbuf *m)
1845 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1847 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1850 m_tag_prepend(m, mtag);
1851 bzero(mtag + 1, sizeof(struct ip6aux));
1854 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1857 static struct ip6aux *
1858 ip6_findaux(struct mbuf *m)
1862 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1863 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1867 ip6_delaux(struct mbuf *m)
1871 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1873 m_tag_delete(m, mtag);
1877 * System control for IP6
1880 u_char inet6ctlerrmap[PRC_NCMDS] = {
1882 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1883 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1884 EMSGSIZE, EHOSTUNREACH, 0, 0,