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>
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>
88 #include <net/if_var.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>
122 extern struct domain inet6domain;
124 u_char ip6_protox[IPPROTO_MAX];
125 VNET_DEFINE(struct in6_ifaddrhead, in6_ifaddrhead);
126 VNET_DEFINE(struct in6_ifaddrlisthead *, in6_ifaddrhashtbl);
127 VNET_DEFINE(u_long, in6_ifaddrhmask);
129 static struct netisr_handler ip6_nh = {
131 .nh_handler = ip6_input,
132 .nh_proto = NETISR_IPV6,
133 .nh_policy = NETISR_POLICY_FLOW,
136 VNET_DECLARE(struct callout, in6_tmpaddrtimer_ch);
137 #define V_in6_tmpaddrtimer_ch VNET(in6_tmpaddrtimer_ch)
139 VNET_DEFINE(struct pfil_head, inet6_pfil_hook);
141 VNET_PCPUSTAT_DEFINE(struct ip6stat, ip6stat);
142 VNET_PCPUSTAT_SYSINIT(ip6stat);
144 VNET_PCPUSTAT_SYSUNINIT(ip6stat);
147 struct rwlock in6_ifaddr_lock;
148 RW_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock");
150 static void ip6_init2(void *);
151 static struct ip6aux *ip6_setdstifaddr(struct mbuf *, struct in6_ifaddr *);
152 static struct ip6aux *ip6_addaux(struct mbuf *);
153 static struct ip6aux *ip6_findaux(struct mbuf *m);
154 static void ip6_delaux (struct mbuf *);
155 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
157 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
161 * IP6 initialization: fill in IP6 protocol switch table.
162 * All protocols not implemented in kernel go to raw IP6 protocol handler.
167 struct ip6protosw *pr;
170 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal",
171 &V_ip6_auto_linklocal);
172 TUNABLE_INT_FETCH("net.inet6.ip6.accept_rtadv", &V_ip6_accept_rtadv);
173 TUNABLE_INT_FETCH("net.inet6.ip6.no_radr", &V_ip6_no_radr);
175 TAILQ_INIT(&V_in6_ifaddrhead);
176 V_in6_ifaddrhashtbl = hashinit(IN6ADDR_NHASH, M_IFADDR,
179 /* Initialize packet filter hooks. */
180 V_inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
181 V_inet6_pfil_hook.ph_af = AF_INET6;
182 if ((i = pfil_head_register(&V_inet6_pfil_hook)) != 0)
183 printf("%s: WARNING: unable to register pfil hook, "
184 "error %d\n", __func__, i);
187 addrsel_policy_init();
191 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
193 /* Skip global initialization stuff for non-default instances. */
194 if (!IS_DEFAULT_VNET(curvnet))
198 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
199 panic("sizeof(protosw) != sizeof(ip6protosw)");
201 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
205 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
206 for (i = 0; i < IPPROTO_MAX; i++)
207 ip6_protox[i] = pr - inet6sw;
209 * Cycle through IP protocols and put them into the appropriate place
212 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
213 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
214 if (pr->pr_domain->dom_family == PF_INET6 &&
215 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) {
216 /* Be careful to only index valid IP protocols. */
217 if (pr->pr_protocol < IPPROTO_MAX)
218 ip6_protox[pr->pr_protocol] = pr - inet6sw;
221 netisr_register(&ip6_nh);
225 * The protocol to be inserted into ip6_protox[] must be already registered
226 * in inet6sw[], either statically or through pf_proto_register().
229 ip6proto_register(short ip6proto)
231 struct ip6protosw *pr;
234 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
235 return (EPROTONOSUPPORT);
238 * The protocol slot must not be occupied by another protocol
239 * already. An index pointing to IPPROTO_RAW is unused.
241 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
243 return (EPFNOSUPPORT);
244 if (ip6_protox[ip6proto] != pr - inet6sw) /* IPPROTO_RAW */
248 * Find the protocol position in inet6sw[] and set the index.
250 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
251 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) {
252 if (pr->pr_domain->dom_family == PF_INET6 &&
253 pr->pr_protocol && pr->pr_protocol == ip6proto) {
254 ip6_protox[pr->pr_protocol] = pr - inet6sw;
258 return (EPROTONOSUPPORT);
262 ip6proto_unregister(short ip6proto)
264 struct ip6protosw *pr;
267 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
268 return (EPROTONOSUPPORT);
270 /* Check if the protocol was indeed registered. */
271 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
273 return (EPFNOSUPPORT);
274 if (ip6_protox[ip6proto] == pr - inet6sw) /* IPPROTO_RAW */
277 /* Reset the protocol slot to IPPROTO_RAW. */
278 ip6_protox[ip6proto] = pr - inet6sw;
288 if ((i = pfil_head_unregister(&V_inet6_pfil_hook)) != 0)
289 printf("%s: WARNING: unable to unregister pfil hook, "
290 "error %d\n", __func__, i);
291 hashdestroy(V_in6_ifaddrhashtbl, M_IFADDR, V_in6_ifaddrhmask);
293 callout_drain(&V_in6_tmpaddrtimer_ch);
298 ip6_init2_vnet(const void *unused __unused)
302 callout_init(&V_nd6_timer_ch, 0);
303 callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet);
305 /* timer for regeneranation of temporary addresses randomize ID */
306 callout_init(&V_in6_tmpaddrtimer_ch, 0);
307 callout_reset(&V_in6_tmpaddrtimer_ch,
308 (V_ip6_temp_preferred_lifetime - V_ip6_desync_factor -
309 V_ip6_temp_regen_advance) * hz,
310 in6_tmpaddrtimer, curvnet);
316 ip6_init2(void *dummy)
319 ip6_init2_vnet(NULL);
323 /* This must be after route_init(), which is now SI_ORDER_THIRD */
324 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
327 ip6_input_hbh(struct mbuf *m, uint32_t *plen, uint32_t *rtalert, int *off,
333 if (ip6_hopopts_input(plen, rtalert, &m, off)) {
334 #if 0 /*touches NULL pointer*/
335 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
337 goto out; /* m have already been freed */
341 ip6 = mtod(m, struct ip6_hdr *);
344 * if the payload length field is 0 and the next header field
345 * indicates Hop-by-Hop Options header, then a Jumbo Payload
346 * option MUST be included.
348 if (ip6->ip6_plen == 0 && *plen == 0) {
350 * Note that if a valid jumbo payload option is
351 * contained, ip6_hopopts_input() must set a valid
352 * (non-zero) payload length to the variable plen.
354 IP6STAT_INC(ip6s_badoptions);
355 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
356 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
357 icmp6_error(m, ICMP6_PARAM_PROB,
358 ICMP6_PARAMPROB_HEADER,
359 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
362 #ifndef PULLDOWN_TEST
363 /* ip6_hopopts_input() ensures that mbuf is contiguous */
364 hbh = (struct ip6_hbh *)(ip6 + 1);
366 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
367 sizeof(struct ip6_hbh));
369 IP6STAT_INC(ip6s_tooshort);
373 *nxt = hbh->ip6h_nxt;
376 * If we are acting as a router and the packet contains a
377 * router alert option, see if we know the option value.
378 * Currently, we only support the option value for MLD, in which
379 * case we should pass the packet to the multicast routing
382 if (*rtalert != ~0) {
384 case IP6OPT_RTALERT_MLD:
385 if (V_ip6_forwarding)
390 * RFC2711 requires unrecognized values must be
404 ip6_input(struct mbuf *m)
407 int off = sizeof(struct ip6_hdr), nest;
409 u_int32_t rtalert = ~0;
411 struct ifnet *deliverifp = NULL, *ifp = NULL;
412 struct in6_addr odst;
413 struct route_in6 rin6;
415 struct llentry *lle = NULL;
416 struct sockaddr_in6 dst6, *dst;
418 bzero(&rin6, sizeof(struct route_in6));
421 * should the inner packet be considered authentic?
422 * see comment in ah4_input().
423 * NB: m cannot be NULL when passed to the input routine
426 m->m_flags &= ~M_AUTHIPHDR;
427 m->m_flags &= ~M_AUTHIPDGM;
432 * make sure we don't have onion peering information into m_tag.
436 if (m->m_flags & M_FASTFWD_OURS) {
438 * Firewall changed destination to local.
440 m->m_flags &= ~M_FASTFWD_OURS;
442 deliverifp = m->m_pkthdr.rcvif;
443 ip6 = mtod(m, struct ip6_hdr *);
450 if (m->m_flags & M_EXT) {
452 IP6STAT_INC(ip6s_mext2m);
454 IP6STAT_INC(ip6s_mext1);
457 if (m->m_flags & M_LOOP) {
458 IP6STAT_INC(ip6s_m2m[V_loif->if_index]);
459 } else if (m->m_pkthdr.rcvif->if_index < IP6S_M2MMAX)
461 ip6s_m2m[m->m_pkthdr.rcvif->if_index]);
463 IP6STAT_INC(ip6s_m2m[0]);
465 IP6STAT_INC(ip6s_m1);
468 /* drop the packet if IPv6 operation is disabled on the IF */
469 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
474 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
475 IP6STAT_INC(ip6s_total);
477 #ifndef PULLDOWN_TEST
479 * L2 bridge code and some other code can return mbuf chain
480 * that does not conform to KAME requirement. too bad.
481 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
483 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
486 if (m->m_pkthdr.len > MHLEN)
487 n = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
489 n = m_gethdr(M_NOWAIT, MT_DATA);
492 return; /* ENOBUFS */
496 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
497 n->m_len = n->m_pkthdr.len;
501 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
504 if (m->m_len < sizeof(struct ip6_hdr)) {
506 inifp = m->m_pkthdr.rcvif;
507 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
508 IP6STAT_INC(ip6s_toosmall);
509 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
514 ip6 = mtod(m, struct ip6_hdr *);
516 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
517 IP6STAT_INC(ip6s_badvers);
518 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
522 IP6STAT_INC(ip6s_nxthist[ip6->ip6_nxt]);
524 IP_PROBE(receive, NULL, NULL, ip6, m->m_pkthdr.rcvif, NULL, ip6);
527 * Check against address spoofing/corruption.
529 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
530 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
532 * XXX: "badscope" is not very suitable for a multicast source.
534 IP6STAT_INC(ip6s_badscope);
535 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
538 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
539 !(m->m_flags & M_LOOP)) {
541 * In this case, the packet should come from the loopback
542 * interface. However, we cannot just check the if_flags,
543 * because ip6_mloopback() passes the "actual" interface
544 * as the outgoing/incoming interface.
546 IP6STAT_INC(ip6s_badscope);
547 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
550 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
551 IPV6_ADDR_MC_SCOPE(&ip6->ip6_dst) == 0) {
554 * Nodes must not originate a packet to a multicast address
555 * whose scop field contains the reserved value 0; if such
556 * a packet is received, it must be silently dropped.
558 IP6STAT_INC(ip6s_badscope);
559 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
563 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
564 /* packet is dropped by traffic conditioner */
569 * The following check is not documented in specs. A malicious
570 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
571 * and bypass security checks (act as if it was from 127.0.0.1 by using
572 * IPv6 src ::ffff:127.0.0.1). Be cautious.
574 * This check chokes if we are in an SIIT cloud. As none of BSDs
575 * support IPv4-less kernel compilation, we cannot support SIIT
576 * environment at all. So, it makes more sense for us to reject any
577 * malicious packets for non-SIIT environment, than try to do a
578 * partial support for SIIT environment.
580 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
581 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
582 IP6STAT_INC(ip6s_badscope);
583 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
588 * Reject packets with IPv4 compatible addresses (auto tunnel).
590 * The code forbids auto tunnel relay case in RFC1933 (the check is
591 * stronger than RFC1933). We may want to re-enable it if mech-xx
592 * is revised to forbid relaying case.
594 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
595 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
596 IP6STAT_INC(ip6s_badscope);
597 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
603 * Bypass packet filtering for packets previously handled by IPsec.
605 if (ip6_ipsec_filtertunnel(m))
610 * Run through list of hooks for input packets.
612 * NB: Beware of the destination address changing
613 * (e.g. by NAT rewriting). When this happens,
614 * tell ip6_forward to do the right thing.
618 /* Jump over all PFIL processing if hooks are not active. */
619 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
622 if (pfil_run_hooks(&V_inet6_pfil_hook, &m,
623 m->m_pkthdr.rcvif, PFIL_IN, NULL))
625 if (m == NULL) /* consumed by filter */
627 ip6 = mtod(m, struct ip6_hdr *);
628 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
630 if (m->m_flags & M_FASTFWD_OURS) {
631 m->m_flags &= ~M_FASTFWD_OURS;
633 deliverifp = m->m_pkthdr.rcvif;
636 if ((m->m_flags & M_IP6_NEXTHOP) &&
637 m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
639 * Directly ship the packet on. This allows forwarding
640 * packets originally destined to us to some other directly
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 IP6STAT_INC(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 IP6STAT_INC(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;
688 IF_AFDATA_RLOCK(ifp);
689 lle = lla_lookup(LLTABLE6(ifp), 0,
690 (struct sockaddr *)&dst6);
691 IF_AFDATA_RUNLOCK(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 counter_u64_add(ia6->ia_ifa.ifa_ipackets, 1);
715 counter_u64_add(ia6->ia_ifa.ifa_ibytes,
719 * record address information into m_tag.
721 (void)ip6_setdstifaddr(m, ia6);
725 char ip6bufs[INET6_ADDRSTRLEN];
726 char ip6bufd[INET6_ADDRSTRLEN];
727 /* address is not ready, so discard the packet. */
729 "ip6_input: packet to an unready address %s->%s\n",
730 ip6_sprintf(ip6bufs, &ip6->ip6_src),
731 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
733 IF_ADDR_RUNLOCK(ifp);
747 dst->sin6_len = sizeof(struct sockaddr_in6);
748 dst->sin6_family = AF_INET6;
749 dst->sin6_addr = ip6->ip6_dst;
750 rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
752 RT_UNLOCK(rin6.ro_rt);
754 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
757 * Accept the packet if the forwarding interface to the destination
758 * according to the routing table is the loopback interface,
759 * unless the associated route has a gateway.
760 * Note that this approach causes to accept a packet if there is a
761 * route to the loopback interface for the destination of the packet.
762 * But we think it's even useful in some situations, e.g. when using
763 * a special daemon which wants to intercept the packet.
765 * XXX: some OSes automatically make a cloned route for the destination
766 * of an outgoing packet. If the outgoing interface of the packet
767 * is a loopback one, the kernel would consider the packet to be
768 * accepted, even if we have no such address assinged on the interface.
769 * We check the cloned flag of the route entry to reject such cases,
770 * assuming that route entries for our own addresses are not made by
771 * cloning (it should be true because in6_addloop explicitly installs
772 * the host route). However, we might have to do an explicit check
773 * while it would be less efficient. Or, should we rather install a
774 * reject route for such a case?
777 (rin6.ro_rt->rt_flags &
778 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
780 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) &&
783 !(rin6.ro_rt->rt_flags & RTF_CLONED) &&
787 * The check below is redundant since the comparison of
788 * the destination and the key of the rtentry has
789 * already done through looking up the routing table.
791 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
792 &rt6_key(rin6.ro_rt)->sin6_addr)
794 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) {
796 struct in6_ifaddr *ia6;
799 * found the loopback route to the interface address
801 if (rin6.ro_rt->rt_gateway->sa_family == AF_LINK) {
802 struct sockaddr_in6 dest6;
804 bzero(&dest6, sizeof(dest6));
805 dest6.sin6_family = AF_INET6;
806 dest6.sin6_len = sizeof(dest6);
807 dest6.sin6_addr = ip6->ip6_dst;
808 ia6 = (struct in6_ifaddr *)
809 ifa_ifwithaddr((struct sockaddr *)&dest6);
815 ia6 = (struct in6_ifaddr *)rin6.ro_rt->rt_ifa;
818 * record address information into m_tag.
820 (void)ip6_setdstifaddr(m, ia6);
823 * packets to a tentative, duplicated, or somehow invalid
824 * address must not be accepted.
826 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
827 /* this address is ready */
829 deliverifp = ia6->ia_ifp; /* correct? */
830 /* Count the packet in the ip address stats */
831 counter_u64_add(ia6->ia_ifa.ifa_ipackets, 1);
832 counter_u64_add(ia6->ia_ifa.ifa_ibytes,
835 ifa_free(&ia6->ia_ifa);
838 char ip6bufs[INET6_ADDRSTRLEN];
839 char ip6bufd[INET6_ADDRSTRLEN];
840 /* address is not ready, so discard the packet. */
842 "ip6_input: packet to an unready address %s->%s\n",
843 ip6_sprintf(ip6bufs, &ip6->ip6_src),
844 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
847 ifa_free(&ia6->ia_ifa);
853 * FAITH (Firewall Aided Internet Translator)
855 if (V_ip6_keepfaith) {
856 if (rin6.ro_rt && rin6.ro_rt->rt_ifp &&
857 rin6.ro_rt->rt_ifp->if_type == IFT_FAITH) {
858 /* XXX do we need more sanity checks? */
860 deliverifp = rin6.ro_rt->rt_ifp; /* faith */
866 * Now there is no reason to process the packet if it's not our own
867 * and we're not a router.
869 if (!V_ip6_forwarding) {
870 IP6STAT_INC(ip6s_cantforward);
871 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
877 * record address information into m_tag, if we don't have one yet.
878 * note that we are unable to record it, if the address is not listed
879 * as our interface address (e.g. multicast addresses, addresses
880 * within FAITH prefixes and such).
883 struct in6_ifaddr *ia6;
885 if ((ia6 = ip6_getdstifaddr(m)) != NULL) {
886 ifa_free(&ia6->ia_ifa);
888 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
890 if (!ip6_setdstifaddr(m, ia6)) {
892 * XXX maybe we should drop the packet here,
893 * as we could not provide enough information
894 * to the upper layers.
897 ifa_free(&ia6->ia_ifa);
903 * Process Hop-by-Hop options header if it's contained.
904 * m may be modified in ip6_hopopts_input().
905 * If a JumboPayload option is included, plen will also be modified.
907 plen = (u_int32_t)ntohs(ip6->ip6_plen);
908 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
911 error = ip6_input_hbh(m, &plen, &rtalert, &off, &nxt, &ours);
918 * Check that the amount of data in the buffers
919 * is as at least much as the IPv6 header would have us expect.
920 * Trim mbufs if longer than we expect.
921 * Drop packet if shorter than we expect.
923 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
924 IP6STAT_INC(ip6s_tooshort);
925 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
928 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
929 if (m->m_len == m->m_pkthdr.len) {
930 m->m_len = sizeof(struct ip6_hdr) + plen;
931 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
933 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
937 * Forward if desirable.
940 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
942 * If we are acting as a multicast router, all
943 * incoming multicast packets are passed to the
944 * kernel-level multicast forwarding function.
945 * The packet is returned (relatively) intact; if
946 * ip6_mforward() returns a non-zero value, the packet
947 * must be discarded, else it may be accepted below.
949 * XXX TODO: Check hlim and multicast scope here to avoid
950 * unnecessarily calling into ip6_mforward().
953 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
954 IP6STAT_INC(ip6s_cantforward);
955 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
959 ip6_forward(m, srcrt);
963 ip6 = mtod(m, struct ip6_hdr *);
966 * Malicious party may be able to use IPv4 mapped addr to confuse
967 * tcp/udp stack and bypass security checks (act as if it was from
968 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
970 * For SIIT end node behavior, you may want to disable the check.
971 * However, you will become vulnerable to attacks using IPv4 mapped
974 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
975 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
976 IP6STAT_INC(ip6s_badscope);
977 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
982 * Tell launch routine the next header
984 IP6STAT_INC(ip6s_delivered);
985 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
988 while (nxt != IPPROTO_DONE) {
989 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
990 IP6STAT_INC(ip6s_toomanyhdr);
995 * protection against faulty packet - there should be
996 * more sanity checks in header chain processing.
998 if (m->m_pkthdr.len < off) {
999 IP6STAT_INC(ip6s_tooshort);
1000 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
1006 * enforce IPsec policy checking if we are seeing last header.
1007 * note that we do not visit this with protocols with pcb layer
1008 * code - like udp/tcp/raw ip.
1010 if (ip6_ipsec_input(m, nxt))
1015 * Use mbuf flags to propagate Router Alert option to
1016 * ICMPv6 layer, as hop-by-hop options have been stripped.
1018 if (nxt == IPPROTO_ICMPV6 && rtalert != ~0)
1019 m->m_flags |= M_RTALERT_MLD;
1021 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
1032 * set/grab in6_ifaddr correspond to IPv6 destination address.
1033 * XXX backward compatibility wrapper
1035 * XXXRW: We should bump the refcount on ia6 before sticking it in the m_tag,
1036 * and then bump it when the tag is copied, and release it when the tag is
1037 * freed. Unfortunately, m_tags don't support deep copies (yet), so instead
1038 * we just bump the ia refcount when we receive it. This should be fixed.
1040 static struct ip6aux *
1041 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6)
1043 struct ip6aux *ip6a;
1045 ip6a = ip6_addaux(m);
1047 ip6a->ip6a_dstia6 = ia6;
1048 return ip6a; /* NULL if failed to set */
1052 ip6_getdstifaddr(struct mbuf *m)
1054 struct ip6aux *ip6a;
1055 struct in6_ifaddr *ia;
1057 ip6a = ip6_findaux(m);
1059 ia = ip6a->ip6a_dstia6;
1060 ifa_ref(&ia->ia_ifa);
1067 * Hop-by-Hop options header processing. If a valid jumbo payload option is
1068 * included, the real payload length will be stored in plenp.
1070 * rtalertp - XXX: should be stored more smart way
1073 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
1074 struct mbuf **mp, int *offp)
1076 struct mbuf *m = *mp;
1077 int off = *offp, hbhlen;
1078 struct ip6_hbh *hbh;
1081 /* validation of the length of the header */
1082 #ifndef PULLDOWN_TEST
1083 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
1084 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1085 hbhlen = (hbh->ip6h_len + 1) << 3;
1087 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
1088 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1090 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1091 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1093 IP6STAT_INC(ip6s_tooshort);
1096 hbhlen = (hbh->ip6h_len + 1) << 3;
1097 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
1100 IP6STAT_INC(ip6s_tooshort);
1105 hbhlen -= sizeof(struct ip6_hbh);
1106 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
1108 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
1109 hbhlen, rtalertp, plenp) < 0)
1118 * Search header for all Hop-by-hop options and process each option.
1119 * This function is separate from ip6_hopopts_input() in order to
1120 * handle a case where the sending node itself process its hop-by-hop
1121 * options header. In such a case, the function is called from ip6_output().
1123 * The function assumes that hbh header is located right after the IPv6 header
1124 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1125 * opthead + hbhlen is located in contiguous memory region.
1128 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
1129 u_int32_t *rtalertp, u_int32_t *plenp)
1131 struct ip6_hdr *ip6;
1133 u_int8_t *opt = opthead;
1134 u_int16_t rtalert_val;
1135 u_int32_t jumboplen;
1136 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1138 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1144 if (hbhlen < IP6OPT_MINLEN) {
1145 IP6STAT_INC(ip6s_toosmall);
1148 optlen = *(opt + 1) + 2;
1150 case IP6OPT_ROUTER_ALERT:
1151 /* XXX may need check for alignment */
1152 if (hbhlen < IP6OPT_RTALERT_LEN) {
1153 IP6STAT_INC(ip6s_toosmall);
1156 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1158 icmp6_error(m, ICMP6_PARAM_PROB,
1159 ICMP6_PARAMPROB_HEADER,
1160 erroff + opt + 1 - opthead);
1163 optlen = IP6OPT_RTALERT_LEN;
1164 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1165 *rtalertp = ntohs(rtalert_val);
1168 /* XXX may need check for alignment */
1169 if (hbhlen < IP6OPT_JUMBO_LEN) {
1170 IP6STAT_INC(ip6s_toosmall);
1173 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1175 icmp6_error(m, ICMP6_PARAM_PROB,
1176 ICMP6_PARAMPROB_HEADER,
1177 erroff + opt + 1 - opthead);
1180 optlen = IP6OPT_JUMBO_LEN;
1183 * IPv6 packets that have non 0 payload length
1184 * must not contain a jumbo payload option.
1186 ip6 = mtod(m, struct ip6_hdr *);
1187 if (ip6->ip6_plen) {
1188 IP6STAT_INC(ip6s_badoptions);
1189 icmp6_error(m, ICMP6_PARAM_PROB,
1190 ICMP6_PARAMPROB_HEADER,
1191 erroff + opt - opthead);
1196 * We may see jumbolen in unaligned location, so
1197 * we'd need to perform bcopy().
1199 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1200 jumboplen = (u_int32_t)htonl(jumboplen);
1204 * if there are multiple jumbo payload options,
1205 * *plenp will be non-zero and the packet will be
1207 * the behavior may need some debate in ipngwg -
1208 * multiple options does not make sense, however,
1209 * there's no explicit mention in specification.
1212 IP6STAT_INC(ip6s_badoptions);
1213 icmp6_error(m, ICMP6_PARAM_PROB,
1214 ICMP6_PARAMPROB_HEADER,
1215 erroff + opt + 2 - opthead);
1221 * jumbo payload length must be larger than 65535.
1223 if (jumboplen <= IPV6_MAXPACKET) {
1224 IP6STAT_INC(ip6s_badoptions);
1225 icmp6_error(m, ICMP6_PARAM_PROB,
1226 ICMP6_PARAMPROB_HEADER,
1227 erroff + opt + 2 - opthead);
1233 default: /* unknown option */
1234 if (hbhlen < IP6OPT_MINLEN) {
1235 IP6STAT_INC(ip6s_toosmall);
1238 optlen = ip6_unknown_opt(opt, m,
1239 erroff + opt - opthead);
1255 * Unknown option processing.
1256 * The third argument `off' is the offset from the IPv6 header to the option,
1257 * which is necessary if the IPv6 header the and option header and IPv6 header
1258 * is not contiguous in order to return an ICMPv6 error.
1261 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1263 struct ip6_hdr *ip6;
1265 switch (IP6OPT_TYPE(*optp)) {
1266 case IP6OPT_TYPE_SKIP: /* ignore the option */
1267 return ((int)*(optp + 1));
1268 case IP6OPT_TYPE_DISCARD: /* silently discard */
1271 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1272 IP6STAT_INC(ip6s_badoptions);
1273 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1275 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1276 IP6STAT_INC(ip6s_badoptions);
1277 ip6 = mtod(m, struct ip6_hdr *);
1278 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1279 (m->m_flags & (M_BCAST|M_MCAST)))
1282 icmp6_error(m, ICMP6_PARAM_PROB,
1283 ICMP6_PARAMPROB_OPTION, off);
1287 m_freem(m); /* XXX: NOTREACHED */
1292 * Create the "control" list for this pcb.
1293 * These functions will not modify mbuf chain at all.
1295 * With KAME mbuf chain restriction:
1296 * The routine will be called from upper layer handlers like tcp6_input().
1297 * Thus the routine assumes that the caller (tcp6_input) have already
1298 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1299 * very first mbuf on the mbuf chain.
1301 * ip6_savecontrol_v4 will handle those options that are possible to be
1302 * set on a v4-mapped socket.
1303 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1304 * options and handle the v6-only ones itself.
1307 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1310 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1313 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1317 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1318 SCM_TIMESTAMP, SOL_SOCKET);
1320 mp = &(*mp)->m_next;
1324 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1325 /* RFC 2292 sec. 5 */
1326 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1327 struct in6_pktinfo pi6;
1329 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1333 ip = mtod(m, struct ip *);
1334 pi6.ipi6_addr.s6_addr32[0] = 0;
1335 pi6.ipi6_addr.s6_addr32[1] = 0;
1336 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1337 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1339 /* We won't hit this code */
1340 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1343 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1344 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1347 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1349 *mp = sbcreatecontrol((caddr_t) &pi6,
1350 sizeof(struct in6_pktinfo),
1351 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1353 mp = &(*mp)->m_next;
1356 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1359 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1363 ip = mtod(m, struct ip *);
1366 /* We won't hit this code */
1370 hlim = ip6->ip6_hlim & 0xff;
1372 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1373 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1376 mp = &(*mp)->m_next;
1379 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1382 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1386 ip = mtod(m, struct ip *);
1387 tclass = ip->ip_tos;
1389 /* We won't hit this code */
1395 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1397 tclass = flowinfo & 0xff;
1399 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(int),
1400 IPV6_TCLASS, IPPROTO_IPV6);
1402 mp = &(*mp)->m_next;
1405 if (v4only != NULL) {
1406 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1417 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1419 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1422 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1427 * IPV6_HOPOPTS socket option. Recall that we required super-user
1428 * privilege for the option (see ip6_ctloutput), but it might be too
1429 * strict, since there might be some hop-by-hop options which can be
1430 * returned to normal user.
1431 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1433 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1435 * Check if a hop-by-hop options header is contatined in the
1436 * received packet, and if so, store the options as ancillary
1437 * data. Note that a hop-by-hop options header must be
1438 * just after the IPv6 header, which is assured through the
1439 * IPv6 input processing.
1441 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1442 struct ip6_hbh *hbh;
1444 #ifdef PULLDOWN_TEST
1448 #ifndef PULLDOWN_TEST
1449 hbh = (struct ip6_hbh *)(ip6 + 1);
1450 hbhlen = (hbh->ip6h_len + 1) << 3;
1452 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1455 IP6STAT_INC(ip6s_tooshort);
1458 hbh = mtod(ext, struct ip6_hbh *);
1459 hbhlen = (hbh->ip6h_len + 1) << 3;
1460 if (hbhlen != ext->m_len) {
1462 IP6STAT_INC(ip6s_tooshort);
1468 * XXX: We copy the whole header even if a
1469 * jumbo payload option is included, the option which
1470 * is to be removed before returning according to
1472 * Note: this constraint is removed in RFC3542
1474 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1475 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1478 mp = &(*mp)->m_next;
1479 #ifdef PULLDOWN_TEST
1485 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1486 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1489 * Search for destination options headers or routing
1490 * header(s) through the header chain, and stores each
1491 * header as ancillary data.
1492 * Note that the order of the headers remains in
1493 * the chain of ancillary data.
1495 while (1) { /* is explicit loop prevention necessary? */
1496 struct ip6_ext *ip6e = NULL;
1498 #ifdef PULLDOWN_TEST
1499 struct mbuf *ext = NULL;
1503 * if it is not an extension header, don't try to
1504 * pull it from the chain.
1507 case IPPROTO_DSTOPTS:
1508 case IPPROTO_ROUTING:
1509 case IPPROTO_HOPOPTS:
1510 case IPPROTO_AH: /* is it possible? */
1516 #ifndef PULLDOWN_TEST
1517 if (off + sizeof(*ip6e) > m->m_len)
1519 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1520 if (nxt == IPPROTO_AH)
1521 elen = (ip6e->ip6e_len + 2) << 2;
1523 elen = (ip6e->ip6e_len + 1) << 3;
1524 if (off + elen > m->m_len)
1527 ext = ip6_pullexthdr(m, off, nxt);
1529 IP6STAT_INC(ip6s_tooshort);
1532 ip6e = mtod(ext, struct ip6_ext *);
1533 if (nxt == IPPROTO_AH)
1534 elen = (ip6e->ip6e_len + 2) << 2;
1536 elen = (ip6e->ip6e_len + 1) << 3;
1537 if (elen != ext->m_len) {
1539 IP6STAT_INC(ip6s_tooshort);
1545 case IPPROTO_DSTOPTS:
1546 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1549 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1551 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1554 mp = &(*mp)->m_next;
1556 case IPPROTO_ROUTING:
1557 if (!(in6p->inp_flags & IN6P_RTHDR))
1560 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1561 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1564 mp = &(*mp)->m_next;
1566 case IPPROTO_HOPOPTS:
1567 case IPPROTO_AH: /* is it possible? */
1572 * other cases have been filtered in the above.
1573 * none will visit this case. here we supply
1574 * the code just in case (nxt overwritten or
1577 #ifdef PULLDOWN_TEST
1584 /* proceed with the next header. */
1586 nxt = ip6e->ip6e_nxt;
1588 #ifdef PULLDOWN_TEST
1600 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1604 struct ip6_mtuinfo mtuctl;
1606 so = in6p->inp_socket;
1612 if (so == NULL) /* I believe this is impossible */
1613 panic("ip6_notify_pmtu: socket is NULL");
1616 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1617 mtuctl.ip6m_mtu = *mtu;
1618 mtuctl.ip6m_addr = *dst;
1619 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1622 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1623 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1626 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1629 /* XXX: should count statistics */
1636 #ifdef PULLDOWN_TEST
1638 * pull single extension header from mbuf chain. returns single mbuf that
1639 * contains the result, or NULL on error.
1641 static struct mbuf *
1642 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1644 struct ip6_ext ip6e;
1650 case IPPROTO_DSTOPTS:
1651 case IPPROTO_ROUTING:
1652 case IPPROTO_HOPOPTS:
1653 case IPPROTO_AH: /* is it possible? */
1656 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1660 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1661 if (nxt == IPPROTO_AH)
1662 elen = (ip6e.ip6e_len + 2) << 2;
1664 elen = (ip6e.ip6e_len + 1) << 3;
1667 n = m_getcl(M_NOWAIT, MT_DATA, 0);
1669 n = m_get(M_NOWAIT, MT_DATA);
1673 m_copydata(m, off, elen, mtod(n, caddr_t));
1680 * Get pointer to the previous header followed by the header
1681 * currently processed.
1682 * XXX: This function supposes that
1683 * M includes all headers,
1684 * the next header field and the header length field of each header
1686 * the sum of each header length equals to OFF.
1687 * Because of these assumptions, this function must be called very
1688 * carefully. Moreover, it will not be used in the near future when
1689 * we develop `neater' mechanism to process extension headers.
1692 ip6_get_prevhdr(struct mbuf *m, int off)
1694 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1696 if (off == sizeof(struct ip6_hdr))
1697 return (&ip6->ip6_nxt);
1700 struct ip6_ext *ip6e = NULL;
1703 len = sizeof(struct ip6_hdr);
1705 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1708 case IPPROTO_FRAGMENT:
1709 len += sizeof(struct ip6_frag);
1712 len += (ip6e->ip6e_len + 2) << 2;
1715 len += (ip6e->ip6e_len + 1) << 3;
1718 nxt = ip6e->ip6e_nxt;
1721 return (&ip6e->ip6e_nxt);
1728 * get next header offset. m will be retained.
1731 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1734 struct ip6_ext ip6e;
1739 panic("ip6_nexthdr: m == NULL");
1740 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1745 if (m->m_pkthdr.len < off + sizeof(ip6))
1747 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1749 *nxtp = ip6.ip6_nxt;
1753 case IPPROTO_FRAGMENT:
1755 * terminate parsing if it is not the first fragment,
1756 * it does not make sense to parse through it.
1758 if (m->m_pkthdr.len < off + sizeof(fh))
1760 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1761 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1762 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1765 *nxtp = fh.ip6f_nxt;
1766 off += sizeof(struct ip6_frag);
1770 if (m->m_pkthdr.len < off + sizeof(ip6e))
1772 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1774 *nxtp = ip6e.ip6e_nxt;
1775 off += (ip6e.ip6e_len + 2) << 2;
1778 case IPPROTO_HOPOPTS:
1779 case IPPROTO_ROUTING:
1780 case IPPROTO_DSTOPTS:
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 + 1) << 3;
1791 case IPPROTO_IPCOMP:
1803 * get offset for the last header in the chain. m will be kept untainted.
1806 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1816 newoff = ip6_nexthdr(m, off, proto, nxtp);
1819 else if (newoff < off)
1820 return -1; /* invalid */
1821 else if (newoff == off)
1829 static struct ip6aux *
1830 ip6_addaux(struct mbuf *m)
1834 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1836 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1839 m_tag_prepend(m, mtag);
1840 bzero(mtag + 1, sizeof(struct ip6aux));
1843 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1846 static struct ip6aux *
1847 ip6_findaux(struct mbuf *m)
1851 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1852 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1856 ip6_delaux(struct mbuf *m)
1860 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1862 m_tag_delete(m, mtag);
1866 * System control for IP6
1869 u_char inet6ctlerrmap[PRC_NCMDS] = {
1871 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1872 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1873 EMSGSIZE, EHOSTUNREACH, 0, 0,