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);
169 TUNABLE_INT_FETCH("net.inet6.ip6.accept_rtadv", &V_ip6_accept_rtadv);
170 TUNABLE_INT_FETCH("net.inet6.ip6.no_radr", &V_ip6_no_radr);
172 TAILQ_INIT(&V_in6_ifaddrhead);
174 /* Initialize packet filter hooks. */
175 V_inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
176 V_inet6_pfil_hook.ph_af = AF_INET6;
177 if ((i = pfil_head_register(&V_inet6_pfil_hook)) != 0)
178 printf("%s: WARNING: unable to register pfil hook, "
179 "error %d\n", __func__, i);
182 addrsel_policy_init();
187 if (TUNABLE_INT_FETCH("net.inet6.ip6.output_flowtable_size",
188 &V_ip6_output_flowtable_size)) {
189 if (V_ip6_output_flowtable_size < 256)
190 V_ip6_output_flowtable_size = 256;
191 if (!powerof2(V_ip6_output_flowtable_size)) {
192 printf("flowtable must be power of 2 size\n");
193 V_ip6_output_flowtable_size = 2048;
197 * round up to the next power of 2
199 V_ip6_output_flowtable_size = 1 << fls((1024 + maxusers * 64)-1);
201 V_ip6_ft = flowtable_alloc("ipv6", V_ip6_output_flowtable_size, FL_IPV6|FL_PCPU);
204 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
206 /* Skip global initialization stuff for non-default instances. */
207 if (!IS_DEFAULT_VNET(curvnet))
211 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
212 panic("sizeof(protosw) != sizeof(ip6protosw)");
214 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
218 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
219 for (i = 0; i < IPPROTO_MAX; i++)
220 ip6_protox[i] = pr - inet6sw;
222 * Cycle through IP protocols and put them into the appropriate place
225 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
226 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
227 if (pr->pr_domain->dom_family == PF_INET6 &&
228 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) {
229 /* Be careful to only index valid IP protocols. */
230 if (pr->pr_protocol < IPPROTO_MAX)
231 ip6_protox[pr->pr_protocol] = pr - inet6sw;
234 netisr_register(&ip6_nh);
238 * The protocol to be inserted into ip6_protox[] must be already registered
239 * in inet6sw[], either statically or through pf_proto_register().
242 ip6proto_register(short ip6proto)
244 struct ip6protosw *pr;
247 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
248 return (EPROTONOSUPPORT);
251 * The protocol slot must not be occupied by another protocol
252 * already. An index pointing to IPPROTO_RAW is unused.
254 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
256 return (EPFNOSUPPORT);
257 if (ip6_protox[ip6proto] != pr - inet6sw) /* IPPROTO_RAW */
261 * Find the protocol position in inet6sw[] and set the index.
263 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
264 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) {
265 if (pr->pr_domain->dom_family == PF_INET6 &&
266 pr->pr_protocol && pr->pr_protocol == ip6proto) {
267 ip6_protox[pr->pr_protocol] = pr - inet6sw;
271 return (EPROTONOSUPPORT);
275 ip6proto_unregister(short ip6proto)
277 struct ip6protosw *pr;
280 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
281 return (EPROTONOSUPPORT);
283 /* Check if the protocol was indeed registered. */
284 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
286 return (EPFNOSUPPORT);
287 if (ip6_protox[ip6proto] == pr - inet6sw) /* IPPROTO_RAW */
290 /* Reset the protocol slot to IPPROTO_RAW. */
291 ip6_protox[ip6proto] = pr - inet6sw;
301 callout_drain(&V_in6_tmpaddrtimer_ch);
306 ip6_init2_vnet(const void *unused __unused)
310 callout_init(&V_nd6_timer_ch, 0);
311 callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet);
313 /* timer for regeneranation of temporary addresses randomize ID */
314 callout_init(&V_in6_tmpaddrtimer_ch, 0);
315 callout_reset(&V_in6_tmpaddrtimer_ch,
316 (V_ip6_temp_preferred_lifetime - V_ip6_desync_factor -
317 V_ip6_temp_regen_advance) * hz,
318 in6_tmpaddrtimer, curvnet);
324 ip6_init2(void *dummy)
327 ip6_init2_vnet(NULL);
331 /* This must be after route_init(), which is now SI_ORDER_THIRD */
332 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
335 ip6_input_hbh(struct mbuf *m, uint32_t *plen, uint32_t *rtalert, int *off,
341 if (ip6_hopopts_input(plen, rtalert, &m, off)) {
342 #if 0 /*touches NULL pointer*/
343 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
345 goto out; /* m have already been freed */
349 ip6 = mtod(m, struct ip6_hdr *);
352 * if the payload length field is 0 and the next header field
353 * indicates Hop-by-Hop Options header, then a Jumbo Payload
354 * option MUST be included.
356 if (ip6->ip6_plen == 0 && *plen == 0) {
358 * Note that if a valid jumbo payload option is
359 * contained, ip6_hopopts_input() must set a valid
360 * (non-zero) payload length to the variable plen.
362 IP6STAT_INC(ip6s_badoptions);
363 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
364 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
365 icmp6_error(m, ICMP6_PARAM_PROB,
366 ICMP6_PARAMPROB_HEADER,
367 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
370 #ifndef PULLDOWN_TEST
371 /* ip6_hopopts_input() ensures that mbuf is contiguous */
372 hbh = (struct ip6_hbh *)(ip6 + 1);
374 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
375 sizeof(struct ip6_hbh));
377 IP6STAT_INC(ip6s_tooshort);
381 *nxt = hbh->ip6h_nxt;
384 * If we are acting as a router and the packet contains a
385 * router alert option, see if we know the option value.
386 * Currently, we only support the option value for MLD, in which
387 * case we should pass the packet to the multicast routing
390 if (*rtalert != ~0) {
392 case IP6OPT_RTALERT_MLD:
393 if (V_ip6_forwarding)
398 * RFC2711 requires unrecognized values must be
412 ip6_input(struct mbuf *m)
415 int off = sizeof(struct ip6_hdr), nest;
417 u_int32_t rtalert = ~0;
419 struct ifnet *deliverifp = NULL, *ifp = NULL;
420 struct in6_addr odst;
421 struct route_in6 rin6;
423 struct llentry *lle = NULL;
424 struct sockaddr_in6 dst6, *dst;
426 bzero(&rin6, sizeof(struct route_in6));
429 * should the inner packet be considered authentic?
430 * see comment in ah4_input().
431 * NB: m cannot be NULL when passed to the input routine
434 m->m_flags &= ~M_AUTHIPHDR;
435 m->m_flags &= ~M_AUTHIPDGM;
440 * make sure we don't have onion peering information into m_tag.
444 if (m->m_flags & M_FASTFWD_OURS) {
446 * Firewall changed destination to local.
448 m->m_flags &= ~M_FASTFWD_OURS;
450 deliverifp = m->m_pkthdr.rcvif;
451 ip6 = mtod(m, struct ip6_hdr *);
458 if (m->m_flags & M_EXT) {
460 IP6STAT_INC(ip6s_mext2m);
462 IP6STAT_INC(ip6s_mext1);
465 if (m->m_flags & M_LOOP) {
466 IP6STAT_INC(ip6s_m2m[V_loif->if_index]);
467 } else if (m->m_pkthdr.rcvif->if_index < IP6S_M2MMAX)
469 ip6s_m2m[m->m_pkthdr.rcvif->if_index]);
471 IP6STAT_INC(ip6s_m2m[0]);
473 IP6STAT_INC(ip6s_m1);
476 /* drop the packet if IPv6 operation is disabled on the IF */
477 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
482 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
483 IP6STAT_INC(ip6s_total);
485 #ifndef PULLDOWN_TEST
487 * L2 bridge code and some other code can return mbuf chain
488 * that does not conform to KAME requirement. too bad.
489 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
491 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
494 MGETHDR(n, M_DONTWAIT, MT_HEADER);
497 if (n && n->m_pkthdr.len > MHLEN) {
498 MCLGET(n, M_DONTWAIT);
499 if ((n->m_flags & M_EXT) == 0) {
506 return; /* ENOBUFS */
509 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
510 n->m_len = n->m_pkthdr.len;
514 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
517 if (m->m_len < sizeof(struct ip6_hdr)) {
519 inifp = m->m_pkthdr.rcvif;
520 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
521 IP6STAT_INC(ip6s_toosmall);
522 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
527 ip6 = mtod(m, struct ip6_hdr *);
529 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
530 IP6STAT_INC(ip6s_badvers);
531 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
535 IP6STAT_INC(ip6s_nxthist[ip6->ip6_nxt]);
538 * Check against address spoofing/corruption.
540 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
541 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
543 * XXX: "badscope" is not very suitable for a multicast source.
545 IP6STAT_INC(ip6s_badscope);
546 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
549 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
550 !(m->m_flags & M_LOOP)) {
552 * In this case, the packet should come from the loopback
553 * interface. However, we cannot just check the if_flags,
554 * because ip6_mloopback() passes the "actual" interface
555 * as the outgoing/incoming interface.
557 IP6STAT_INC(ip6s_badscope);
558 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
561 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
562 IPV6_ADDR_MC_SCOPE(&ip6->ip6_dst) == 0) {
565 * Nodes must not originate a packet to a multicast address
566 * whose scop field contains the reserved value 0; if such
567 * a packet is received, it must be silently dropped.
569 IP6STAT_INC(ip6s_badscope);
570 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
574 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
575 /* packet is dropped by traffic conditioner */
580 * The following check is not documented in specs. A malicious
581 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
582 * and bypass security checks (act as if it was from 127.0.0.1 by using
583 * IPv6 src ::ffff:127.0.0.1). Be cautious.
585 * This check chokes if we are in an SIIT cloud. As none of BSDs
586 * support IPv4-less kernel compilation, we cannot support SIIT
587 * environment at all. So, it makes more sense for us to reject any
588 * malicious packets for non-SIIT environment, than try to do a
589 * partial support for SIIT environment.
591 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
592 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
593 IP6STAT_INC(ip6s_badscope);
594 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
599 * Reject packets with IPv4 compatible addresses (auto tunnel).
601 * The code forbids auto tunnel relay case in RFC1933 (the check is
602 * stronger than RFC1933). We may want to re-enable it if mech-xx
603 * is revised to forbid relaying case.
605 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
606 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
607 IP6STAT_INC(ip6s_badscope);
608 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
614 * Bypass packet filtering for packets previously handled by IPsec.
616 if (ip6_ipsec_filtertunnel(m))
621 * Run through list of hooks for input packets.
623 * NB: Beware of the destination address changing
624 * (e.g. by NAT rewriting). When this happens,
625 * tell ip6_forward to do the right thing.
629 /* Jump over all PFIL processing if hooks are not active. */
630 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
633 if (pfil_run_hooks(&V_inet6_pfil_hook, &m,
634 m->m_pkthdr.rcvif, PFIL_IN, NULL))
636 if (m == NULL) /* consumed by filter */
638 ip6 = mtod(m, struct ip6_hdr *);
639 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
641 if (m->m_flags & M_FASTFWD_OURS) {
642 m->m_flags &= ~M_FASTFWD_OURS;
644 deliverifp = m->m_pkthdr.rcvif;
647 if ((m->m_flags & M_IP6_NEXTHOP) &&
648 m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
650 * Directly ship the packet on. This allows forwarding
651 * packets originally destined to us to some other directly
660 * Disambiguate address scope zones (if there is ambiguity).
661 * We first make sure that the original source or destination address
662 * is not in our internal form for scoped addresses. Such addresses
663 * are not necessarily invalid spec-wise, but we cannot accept them due
664 * to the usage conflict.
665 * in6_setscope() then also checks and rejects the cases where src or
666 * dst are the loopback address and the receiving interface
669 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
670 IP6STAT_INC(ip6s_badscope); /* XXX */
673 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
674 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
675 IP6STAT_INC(ip6s_badscope);
680 * Multicast check. Assume packet is for us to avoid
681 * prematurely taking locks.
683 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
685 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
686 deliverifp = m->m_pkthdr.rcvif;
694 bzero(&dst6, sizeof(dst6));
695 dst6.sin6_family = AF_INET6;
696 dst6.sin6_len = sizeof(struct sockaddr_in6);
697 dst6.sin6_addr = ip6->ip6_dst;
698 ifp = m->m_pkthdr.rcvif;
699 IF_AFDATA_RLOCK(ifp);
700 lle = lla_lookup(LLTABLE6(ifp), 0,
701 (struct sockaddr *)&dst6);
702 IF_AFDATA_RUNLOCK(ifp);
703 if ((lle != NULL) && (lle->la_flags & LLE_IFADDR)) {
705 struct in6_ifaddr *ia6;
709 #define sa_equal(a1, a2) \
710 (bcmp((a1), (a2), ((a1))->sin6_len) == 0)
712 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
713 if (ifa->ifa_addr->sa_family != dst6.sin6_family)
715 if (sa_equal(&dst6, ifa->ifa_addr))
718 KASSERT(ifa != NULL, ("%s: ifa not found for lle %p",
722 ia6 = (struct in6_ifaddr *)ifa;
723 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
724 /* Count the packet in the ip address stats */
725 ia6->ia_ifa.if_ipackets++;
726 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
729 * record address information into m_tag.
731 (void)ip6_setdstifaddr(m, ia6);
735 char ip6bufs[INET6_ADDRSTRLEN];
736 char ip6bufd[INET6_ADDRSTRLEN];
737 /* address is not ready, so discard the packet. */
739 "ip6_input: packet to an unready address %s->%s\n",
740 ip6_sprintf(ip6bufs, &ip6->ip6_src),
741 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
743 IF_ADDR_RUNLOCK(ifp);
757 dst->sin6_len = sizeof(struct sockaddr_in6);
758 dst->sin6_family = AF_INET6;
759 dst->sin6_addr = ip6->ip6_dst;
760 rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
762 RT_UNLOCK(rin6.ro_rt);
764 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
767 * Accept the packet if the forwarding interface to the destination
768 * according to the routing table is the loopback interface,
769 * unless the associated route has a gateway.
770 * Note that this approach causes to accept a packet if there is a
771 * route to the loopback interface for the destination of the packet.
772 * But we think it's even useful in some situations, e.g. when using
773 * a special daemon which wants to intercept the packet.
775 * XXX: some OSes automatically make a cloned route for the destination
776 * of an outgoing packet. If the outgoing interface of the packet
777 * is a loopback one, the kernel would consider the packet to be
778 * accepted, even if we have no such address assinged on the interface.
779 * We check the cloned flag of the route entry to reject such cases,
780 * assuming that route entries for our own addresses are not made by
781 * cloning (it should be true because in6_addloop explicitly installs
782 * the host route). However, we might have to do an explicit check
783 * while it would be less efficient. Or, should we rather install a
784 * reject route for such a case?
787 (rin6.ro_rt->rt_flags &
788 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
790 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) &&
793 !(rin6.ro_rt->rt_flags & RTF_CLONED) &&
797 * The check below is redundant since the comparison of
798 * the destination and the key of the rtentry has
799 * already done through looking up the routing table.
801 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
802 &rt6_key(rin6.ro_rt)->sin6_addr)
804 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) {
806 struct in6_ifaddr *ia6;
809 * found the loopback route to the interface address
811 if (rin6.ro_rt->rt_gateway->sa_family == AF_LINK) {
812 struct sockaddr_in6 dest6;
814 bzero(&dest6, sizeof(dest6));
815 dest6.sin6_family = AF_INET6;
816 dest6.sin6_len = sizeof(dest6);
817 dest6.sin6_addr = ip6->ip6_dst;
818 ia6 = (struct in6_ifaddr *)
819 ifa_ifwithaddr((struct sockaddr *)&dest6);
825 ia6 = (struct in6_ifaddr *)rin6.ro_rt->rt_ifa;
828 * record address information into m_tag.
830 (void)ip6_setdstifaddr(m, ia6);
833 * packets to a tentative, duplicated, or somehow invalid
834 * address must not be accepted.
836 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
837 /* this address is ready */
839 deliverifp = ia6->ia_ifp; /* correct? */
840 /* Count the packet in the ip address stats */
841 ia6->ia_ifa.if_ipackets++;
842 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
843 if (ia6 != NULL && free_ia6 != 0)
844 ifa_free(&ia6->ia_ifa);
847 char ip6bufs[INET6_ADDRSTRLEN];
848 char ip6bufd[INET6_ADDRSTRLEN];
849 /* address is not ready, so discard the packet. */
851 "ip6_input: packet to an unready address %s->%s\n",
852 ip6_sprintf(ip6bufs, &ip6->ip6_src),
853 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
855 if (ia6 != NULL && free_ia6 != 0)
856 ifa_free(&ia6->ia_ifa);
862 * FAITH (Firewall Aided Internet Translator)
864 if (V_ip6_keepfaith) {
865 if (rin6.ro_rt && rin6.ro_rt->rt_ifp &&
866 rin6.ro_rt->rt_ifp->if_type == IFT_FAITH) {
867 /* XXX do we need more sanity checks? */
869 deliverifp = rin6.ro_rt->rt_ifp; /* faith */
875 * Now there is no reason to process the packet if it's not our own
876 * and we're not a router.
878 if (!V_ip6_forwarding) {
879 IP6STAT_INC(ip6s_cantforward);
880 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
886 * record address information into m_tag, if we don't have one yet.
887 * note that we are unable to record it, if the address is not listed
888 * as our interface address (e.g. multicast addresses, addresses
889 * within FAITH prefixes and such).
892 struct in6_ifaddr *ia6;
894 if ((ia6 = ip6_getdstifaddr(m)) != NULL) {
895 ifa_free(&ia6->ia_ifa);
897 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
899 if (!ip6_setdstifaddr(m, ia6)) {
901 * XXX maybe we should drop the packet here,
902 * as we could not provide enough information
903 * to the upper layers.
906 ifa_free(&ia6->ia_ifa);
912 * Process Hop-by-Hop options header if it's contained.
913 * m may be modified in ip6_hopopts_input().
914 * If a JumboPayload option is included, plen will also be modified.
916 plen = (u_int32_t)ntohs(ip6->ip6_plen);
917 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
920 error = ip6_input_hbh(m, &plen, &rtalert, &off, &nxt, &ours);
927 * Check that the amount of data in the buffers
928 * is as at least much as the IPv6 header would have us expect.
929 * Trim mbufs if longer than we expect.
930 * Drop packet if shorter than we expect.
932 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
933 IP6STAT_INC(ip6s_tooshort);
934 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
937 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
938 if (m->m_len == m->m_pkthdr.len) {
939 m->m_len = sizeof(struct ip6_hdr) + plen;
940 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
942 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
946 * Forward if desirable.
949 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
951 * If we are acting as a multicast router, all
952 * incoming multicast packets are passed to the
953 * kernel-level multicast forwarding function.
954 * The packet is returned (relatively) intact; if
955 * ip6_mforward() returns a non-zero value, the packet
956 * must be discarded, else it may be accepted below.
958 * XXX TODO: Check hlim and multicast scope here to avoid
959 * unnecessarily calling into ip6_mforward().
962 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
963 IP6STAT_INC(ip6s_cantforward);
964 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
968 ip6_forward(m, srcrt);
972 ip6 = mtod(m, struct ip6_hdr *);
975 * Malicious party may be able to use IPv4 mapped addr to confuse
976 * tcp/udp stack and bypass security checks (act as if it was from
977 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
979 * For SIIT end node behavior, you may want to disable the check.
980 * However, you will become vulnerable to attacks using IPv4 mapped
983 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
984 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
985 IP6STAT_INC(ip6s_badscope);
986 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
991 * Tell launch routine the next header
993 IP6STAT_INC(ip6s_delivered);
994 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
997 while (nxt != IPPROTO_DONE) {
998 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
999 IP6STAT_INC(ip6s_toomanyhdr);
1004 * protection against faulty packet - there should be
1005 * more sanity checks in header chain processing.
1007 if (m->m_pkthdr.len < off) {
1008 IP6STAT_INC(ip6s_tooshort);
1009 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
1015 * enforce IPsec policy checking if we are seeing last header.
1016 * note that we do not visit this with protocols with pcb layer
1017 * code - like udp/tcp/raw ip.
1019 if (ip6_ipsec_input(m, nxt))
1024 * Use mbuf flags to propagate Router Alert option to
1025 * ICMPv6 layer, as hop-by-hop options have been stripped.
1027 if (nxt == IPPROTO_ICMPV6 && rtalert != ~0)
1028 m->m_flags |= M_RTALERT_MLD;
1030 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
1041 * set/grab in6_ifaddr correspond to IPv6 destination address.
1042 * XXX backward compatibility wrapper
1044 * XXXRW: We should bump the refcount on ia6 before sticking it in the m_tag,
1045 * and then bump it when the tag is copied, and release it when the tag is
1046 * freed. Unfortunately, m_tags don't support deep copies (yet), so instead
1047 * we just bump the ia refcount when we receive it. This should be fixed.
1049 static struct ip6aux *
1050 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6)
1052 struct ip6aux *ip6a;
1054 ip6a = ip6_addaux(m);
1056 ip6a->ip6a_dstia6 = ia6;
1057 return ip6a; /* NULL if failed to set */
1061 ip6_getdstifaddr(struct mbuf *m)
1063 struct ip6aux *ip6a;
1064 struct in6_ifaddr *ia;
1066 ip6a = ip6_findaux(m);
1068 ia = ip6a->ip6a_dstia6;
1069 ifa_ref(&ia->ia_ifa);
1076 * Hop-by-Hop options header processing. If a valid jumbo payload option is
1077 * included, the real payload length will be stored in plenp.
1079 * rtalertp - XXX: should be stored more smart way
1082 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
1083 struct mbuf **mp, int *offp)
1085 struct mbuf *m = *mp;
1086 int off = *offp, hbhlen;
1087 struct ip6_hbh *hbh;
1089 /* validation of the length of the header */
1090 #ifndef PULLDOWN_TEST
1091 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
1092 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1093 hbhlen = (hbh->ip6h_len + 1) << 3;
1095 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
1096 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1098 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1099 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1101 IP6STAT_INC(ip6s_tooshort);
1104 hbhlen = (hbh->ip6h_len + 1) << 3;
1105 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
1108 IP6STAT_INC(ip6s_tooshort);
1113 hbhlen -= sizeof(struct ip6_hbh);
1114 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
1115 hbhlen, rtalertp, plenp) < 0)
1124 * Search header for all Hop-by-hop options and process each option.
1125 * This function is separate from ip6_hopopts_input() in order to
1126 * handle a case where the sending node itself process its hop-by-hop
1127 * options header. In such a case, the function is called from ip6_output().
1129 * The function assumes that hbh header is located right after the IPv6 header
1130 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1131 * opthead + hbhlen is located in contiguous memory region.
1134 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
1135 u_int32_t *rtalertp, u_int32_t *plenp)
1137 struct ip6_hdr *ip6;
1139 u_int8_t *opt = opthead;
1140 u_int16_t rtalert_val;
1141 u_int32_t jumboplen;
1142 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1144 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1150 if (hbhlen < IP6OPT_MINLEN) {
1151 IP6STAT_INC(ip6s_toosmall);
1154 optlen = *(opt + 1) + 2;
1156 case IP6OPT_ROUTER_ALERT:
1157 /* XXX may need check for alignment */
1158 if (hbhlen < IP6OPT_RTALERT_LEN) {
1159 IP6STAT_INC(ip6s_toosmall);
1162 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1164 icmp6_error(m, ICMP6_PARAM_PROB,
1165 ICMP6_PARAMPROB_HEADER,
1166 erroff + opt + 1 - opthead);
1169 optlen = IP6OPT_RTALERT_LEN;
1170 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1171 *rtalertp = ntohs(rtalert_val);
1174 /* XXX may need check for alignment */
1175 if (hbhlen < IP6OPT_JUMBO_LEN) {
1176 IP6STAT_INC(ip6s_toosmall);
1179 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1181 icmp6_error(m, ICMP6_PARAM_PROB,
1182 ICMP6_PARAMPROB_HEADER,
1183 erroff + opt + 1 - opthead);
1186 optlen = IP6OPT_JUMBO_LEN;
1189 * IPv6 packets that have non 0 payload length
1190 * must not contain a jumbo payload option.
1192 ip6 = mtod(m, struct ip6_hdr *);
1193 if (ip6->ip6_plen) {
1194 IP6STAT_INC(ip6s_badoptions);
1195 icmp6_error(m, ICMP6_PARAM_PROB,
1196 ICMP6_PARAMPROB_HEADER,
1197 erroff + opt - opthead);
1202 * We may see jumbolen in unaligned location, so
1203 * we'd need to perform bcopy().
1205 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1206 jumboplen = (u_int32_t)htonl(jumboplen);
1210 * if there are multiple jumbo payload options,
1211 * *plenp will be non-zero and the packet will be
1213 * the behavior may need some debate in ipngwg -
1214 * multiple options does not make sense, however,
1215 * there's no explicit mention in specification.
1218 IP6STAT_INC(ip6s_badoptions);
1219 icmp6_error(m, ICMP6_PARAM_PROB,
1220 ICMP6_PARAMPROB_HEADER,
1221 erroff + opt + 2 - opthead);
1227 * jumbo payload length must be larger than 65535.
1229 if (jumboplen <= IPV6_MAXPACKET) {
1230 IP6STAT_INC(ip6s_badoptions);
1231 icmp6_error(m, ICMP6_PARAM_PROB,
1232 ICMP6_PARAMPROB_HEADER,
1233 erroff + opt + 2 - opthead);
1239 default: /* unknown option */
1240 if (hbhlen < IP6OPT_MINLEN) {
1241 IP6STAT_INC(ip6s_toosmall);
1244 optlen = ip6_unknown_opt(opt, m,
1245 erroff + opt - opthead);
1261 * Unknown option processing.
1262 * The third argument `off' is the offset from the IPv6 header to the option,
1263 * which is necessary if the IPv6 header the and option header and IPv6 header
1264 * is not contiguous in order to return an ICMPv6 error.
1267 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1269 struct ip6_hdr *ip6;
1271 switch (IP6OPT_TYPE(*optp)) {
1272 case IP6OPT_TYPE_SKIP: /* ignore the option */
1273 return ((int)*(optp + 1));
1274 case IP6OPT_TYPE_DISCARD: /* silently discard */
1277 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1278 IP6STAT_INC(ip6s_badoptions);
1279 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1281 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1282 IP6STAT_INC(ip6s_badoptions);
1283 ip6 = mtod(m, struct ip6_hdr *);
1284 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1285 (m->m_flags & (M_BCAST|M_MCAST)))
1288 icmp6_error(m, ICMP6_PARAM_PROB,
1289 ICMP6_PARAMPROB_OPTION, off);
1293 m_freem(m); /* XXX: NOTREACHED */
1298 * Create the "control" list for this pcb.
1299 * These functions will not modify mbuf chain at all.
1301 * With KAME mbuf chain restriction:
1302 * The routine will be called from upper layer handlers like tcp6_input().
1303 * Thus the routine assumes that the caller (tcp6_input) have already
1304 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1305 * very first mbuf on the mbuf chain.
1307 * ip6_savecontrol_v4 will handle those options that are possible to be
1308 * set on a v4-mapped socket.
1309 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1310 * options and handle the v6-only ones itself.
1313 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1316 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1319 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1323 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1324 SCM_TIMESTAMP, SOL_SOCKET);
1326 mp = &(*mp)->m_next;
1330 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1331 /* RFC 2292 sec. 5 */
1332 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1333 struct in6_pktinfo pi6;
1335 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1339 ip = mtod(m, struct ip *);
1340 pi6.ipi6_addr.s6_addr32[0] = 0;
1341 pi6.ipi6_addr.s6_addr32[1] = 0;
1342 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1343 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1345 /* We won't hit this code */
1346 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1349 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1350 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1353 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1355 *mp = sbcreatecontrol((caddr_t) &pi6,
1356 sizeof(struct in6_pktinfo),
1357 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1359 mp = &(*mp)->m_next;
1362 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1365 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1369 ip = mtod(m, struct ip *);
1372 /* We won't hit this code */
1376 hlim = ip6->ip6_hlim & 0xff;
1378 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1379 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1382 mp = &(*mp)->m_next;
1385 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1388 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1392 ip = mtod(m, struct ip *);
1393 tclass = ip->ip_tos;
1395 /* We won't hit this code */
1401 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1403 tclass = flowinfo & 0xff;
1405 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(int),
1406 IPV6_TCLASS, IPPROTO_IPV6);
1408 mp = &(*mp)->m_next;
1411 if (v4only != NULL) {
1412 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1423 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1425 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1428 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1433 * IPV6_HOPOPTS socket option. Recall that we required super-user
1434 * privilege for the option (see ip6_ctloutput), but it might be too
1435 * strict, since there might be some hop-by-hop options which can be
1436 * returned to normal user.
1437 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1439 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1441 * Check if a hop-by-hop options header is contatined in the
1442 * received packet, and if so, store the options as ancillary
1443 * data. Note that a hop-by-hop options header must be
1444 * just after the IPv6 header, which is assured through the
1445 * IPv6 input processing.
1447 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1448 struct ip6_hbh *hbh;
1450 #ifdef PULLDOWN_TEST
1454 #ifndef PULLDOWN_TEST
1455 hbh = (struct ip6_hbh *)(ip6 + 1);
1456 hbhlen = (hbh->ip6h_len + 1) << 3;
1458 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1461 IP6STAT_INC(ip6s_tooshort);
1464 hbh = mtod(ext, struct ip6_hbh *);
1465 hbhlen = (hbh->ip6h_len + 1) << 3;
1466 if (hbhlen != ext->m_len) {
1468 IP6STAT_INC(ip6s_tooshort);
1474 * XXX: We copy the whole header even if a
1475 * jumbo payload option is included, the option which
1476 * is to be removed before returning according to
1478 * Note: this constraint is removed in RFC3542
1480 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1481 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1484 mp = &(*mp)->m_next;
1485 #ifdef PULLDOWN_TEST
1491 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1492 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1495 * Search for destination options headers or routing
1496 * header(s) through the header chain, and stores each
1497 * header as ancillary data.
1498 * Note that the order of the headers remains in
1499 * the chain of ancillary data.
1501 while (1) { /* is explicit loop prevention necessary? */
1502 struct ip6_ext *ip6e = NULL;
1504 #ifdef PULLDOWN_TEST
1505 struct mbuf *ext = NULL;
1509 * if it is not an extension header, don't try to
1510 * pull it from the chain.
1513 case IPPROTO_DSTOPTS:
1514 case IPPROTO_ROUTING:
1515 case IPPROTO_HOPOPTS:
1516 case IPPROTO_AH: /* is it possible? */
1522 #ifndef PULLDOWN_TEST
1523 if (off + sizeof(*ip6e) > m->m_len)
1525 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1526 if (nxt == IPPROTO_AH)
1527 elen = (ip6e->ip6e_len + 2) << 2;
1529 elen = (ip6e->ip6e_len + 1) << 3;
1530 if (off + elen > m->m_len)
1533 ext = ip6_pullexthdr(m, off, nxt);
1535 IP6STAT_INC(ip6s_tooshort);
1538 ip6e = mtod(ext, struct ip6_ext *);
1539 if (nxt == IPPROTO_AH)
1540 elen = (ip6e->ip6e_len + 2) << 2;
1542 elen = (ip6e->ip6e_len + 1) << 3;
1543 if (elen != ext->m_len) {
1545 IP6STAT_INC(ip6s_tooshort);
1551 case IPPROTO_DSTOPTS:
1552 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1555 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1557 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1560 mp = &(*mp)->m_next;
1562 case IPPROTO_ROUTING:
1563 if (!(in6p->inp_flags & IN6P_RTHDR))
1566 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1567 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1570 mp = &(*mp)->m_next;
1572 case IPPROTO_HOPOPTS:
1573 case IPPROTO_AH: /* is it possible? */
1578 * other cases have been filtered in the above.
1579 * none will visit this case. here we supply
1580 * the code just in case (nxt overwritten or
1583 #ifdef PULLDOWN_TEST
1590 /* proceed with the next header. */
1592 nxt = ip6e->ip6e_nxt;
1594 #ifdef PULLDOWN_TEST
1606 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1610 struct ip6_mtuinfo mtuctl;
1612 so = in6p->inp_socket;
1618 if (so == NULL) /* I believe this is impossible */
1619 panic("ip6_notify_pmtu: socket is NULL");
1622 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1623 mtuctl.ip6m_mtu = *mtu;
1624 mtuctl.ip6m_addr = *dst;
1625 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1628 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1629 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1632 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1635 /* XXX: should count statistics */
1642 #ifdef PULLDOWN_TEST
1644 * pull single extension header from mbuf chain. returns single mbuf that
1645 * contains the result, or NULL on error.
1647 static struct mbuf *
1648 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1650 struct ip6_ext ip6e;
1656 case IPPROTO_DSTOPTS:
1657 case IPPROTO_ROUTING:
1658 case IPPROTO_HOPOPTS:
1659 case IPPROTO_AH: /* is it possible? */
1662 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1666 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1667 if (nxt == IPPROTO_AH)
1668 elen = (ip6e.ip6e_len + 2) << 2;
1670 elen = (ip6e.ip6e_len + 1) << 3;
1672 MGET(n, M_DONTWAIT, MT_DATA);
1673 if (n && elen >= MLEN) {
1674 MCLGET(n, M_DONTWAIT);
1675 if ((n->m_flags & M_EXT) == 0) {
1684 if (elen >= M_TRAILINGSPACE(n)) {
1689 m_copydata(m, off, elen, mtod(n, caddr_t));
1696 * Get pointer to the previous header followed by the header
1697 * currently processed.
1698 * XXX: This function supposes that
1699 * M includes all headers,
1700 * the next header field and the header length field of each header
1702 * the sum of each header length equals to OFF.
1703 * Because of these assumptions, this function must be called very
1704 * carefully. Moreover, it will not be used in the near future when
1705 * we develop `neater' mechanism to process extension headers.
1708 ip6_get_prevhdr(struct mbuf *m, int off)
1710 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1712 if (off == sizeof(struct ip6_hdr))
1713 return (&ip6->ip6_nxt);
1716 struct ip6_ext *ip6e = NULL;
1719 len = sizeof(struct ip6_hdr);
1721 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1724 case IPPROTO_FRAGMENT:
1725 len += sizeof(struct ip6_frag);
1728 len += (ip6e->ip6e_len + 2) << 2;
1731 len += (ip6e->ip6e_len + 1) << 3;
1734 nxt = ip6e->ip6e_nxt;
1737 return (&ip6e->ip6e_nxt);
1744 * get next header offset. m will be retained.
1747 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1750 struct ip6_ext ip6e;
1755 panic("ip6_nexthdr: m == NULL");
1756 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1761 if (m->m_pkthdr.len < off + sizeof(ip6))
1763 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1765 *nxtp = ip6.ip6_nxt;
1769 case IPPROTO_FRAGMENT:
1771 * terminate parsing if it is not the first fragment,
1772 * it does not make sense to parse through it.
1774 if (m->m_pkthdr.len < off + sizeof(fh))
1776 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1777 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1778 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1781 *nxtp = fh.ip6f_nxt;
1782 off += sizeof(struct ip6_frag);
1786 if (m->m_pkthdr.len < off + sizeof(ip6e))
1788 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1790 *nxtp = ip6e.ip6e_nxt;
1791 off += (ip6e.ip6e_len + 2) << 2;
1794 case IPPROTO_HOPOPTS:
1795 case IPPROTO_ROUTING:
1796 case IPPROTO_DSTOPTS:
1797 if (m->m_pkthdr.len < off + sizeof(ip6e))
1799 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1801 *nxtp = ip6e.ip6e_nxt;
1802 off += (ip6e.ip6e_len + 1) << 3;
1807 case IPPROTO_IPCOMP:
1819 * get offset for the last header in the chain. m will be kept untainted.
1822 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1832 newoff = ip6_nexthdr(m, off, proto, nxtp);
1835 else if (newoff < off)
1836 return -1; /* invalid */
1837 else if (newoff == off)
1845 static struct ip6aux *
1846 ip6_addaux(struct mbuf *m)
1850 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1852 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1855 m_tag_prepend(m, mtag);
1856 bzero(mtag + 1, sizeof(struct ip6aux));
1859 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1862 static struct ip6aux *
1863 ip6_findaux(struct mbuf *m)
1867 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1868 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1872 ip6_delaux(struct mbuf *m)
1876 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1878 m_tag_delete(m, mtag);
1882 * System control for IP6
1885 u_char inet6ctlerrmap[PRC_NCMDS] = {
1887 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1888 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1889 EMSGSIZE, EHOSTUNREACH, 0, 0,