2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the project nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $
35 * Copyright (c) 1982, 1986, 1988, 1993
36 * The Regents of the University of California. All rights reserved.
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. Neither the name of the University nor the names of its contributors
47 * may be used to endorse or promote products derived from this software
48 * without specific prior written permission.
50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
65 #include <sys/cdefs.h>
66 __FBSDID("$FreeBSD$");
69 #include "opt_inet6.h"
70 #include "opt_ipsec.h"
71 #include "opt_route.h"
74 #include <sys/param.h>
75 #include <sys/systm.h>
76 #include <sys/hhook.h>
77 #include <sys/malloc.h>
80 #include <sys/domain.h>
81 #include <sys/protosw.h>
83 #include <sys/socket.h>
84 #include <sys/socketvar.h>
85 #include <sys/errno.h>
87 #include <sys/kernel.h>
89 #include <sys/rmlock.h>
90 #include <sys/syslog.h>
91 #include <sys/sysctl.h>
94 #include <net/if_var.h>
95 #include <net/if_types.h>
96 #include <net/if_dl.h>
97 #include <net/route.h>
98 #include <net/netisr.h>
99 #include <net/rss_config.h>
100 #include <net/pfil.h>
101 #include <net/vnet.h>
103 #include <netinet/in.h>
104 #include <netinet/in_kdtrace.h>
105 #include <netinet/ip_var.h>
106 #include <netinet/in_systm.h>
107 #include <net/if_llatbl.h>
109 #include <netinet/ip.h>
110 #include <netinet/ip_icmp.h>
112 #include <netinet/ip6.h>
113 #include <netinet6/in6_var.h>
114 #include <netinet6/ip6_var.h>
115 #include <netinet/in_pcb.h>
116 #include <netinet/icmp6.h>
117 #include <netinet6/scope6_var.h>
118 #include <netinet6/in6_ifattach.h>
119 #include <netinet6/mld6_var.h>
120 #include <netinet6/nd6.h>
121 #include <netinet6/in6_rss.h>
123 #include <netipsec/ipsec_support.h>
125 #include <netinet6/ip6protosw.h>
127 extern struct domain inet6domain;
129 u_char ip6_protox[IPPROTO_MAX];
130 VNET_DEFINE(struct in6_ifaddrhead, in6_ifaddrhead);
131 VNET_DEFINE(struct in6_ifaddrlisthead *, in6_ifaddrhashtbl);
132 VNET_DEFINE(u_long, in6_ifaddrhmask);
134 static struct netisr_handler ip6_nh = {
136 .nh_handler = ip6_input,
137 .nh_proto = NETISR_IPV6,
139 .nh_m2cpuid = rss_soft_m2cpuid_v6,
140 .nh_policy = NETISR_POLICY_CPU,
141 .nh_dispatch = NETISR_DISPATCH_HYBRID,
143 .nh_policy = NETISR_POLICY_FLOW,
148 sysctl_netinet6_intr_queue_maxlen(SYSCTL_HANDLER_ARGS)
152 netisr_getqlimit(&ip6_nh, &qlimit);
153 error = sysctl_handle_int(oidp, &qlimit, 0, req);
154 if (error || !req->newptr)
158 return (netisr_setqlimit(&ip6_nh, qlimit));
160 SYSCTL_DECL(_net_inet6_ip6);
161 SYSCTL_PROC(_net_inet6_ip6, IPV6CTL_INTRQMAXLEN, intr_queue_maxlen,
162 CTLTYPE_INT|CTLFLAG_RW, 0, 0, sysctl_netinet6_intr_queue_maxlen, "I",
163 "Maximum size of the IPv6 input queue");
166 static struct netisr_handler ip6_direct_nh = {
167 .nh_name = "ip6_direct",
168 .nh_handler = ip6_direct_input,
169 .nh_proto = NETISR_IPV6_DIRECT,
170 .nh_m2cpuid = rss_soft_m2cpuid_v6,
171 .nh_policy = NETISR_POLICY_CPU,
172 .nh_dispatch = NETISR_DISPATCH_HYBRID,
176 sysctl_netinet6_intr_direct_queue_maxlen(SYSCTL_HANDLER_ARGS)
180 netisr_getqlimit(&ip6_direct_nh, &qlimit);
181 error = sysctl_handle_int(oidp, &qlimit, 0, req);
182 if (error || !req->newptr)
186 return (netisr_setqlimit(&ip6_direct_nh, qlimit));
188 SYSCTL_PROC(_net_inet6_ip6, IPV6CTL_INTRDQMAXLEN, intr_direct_queue_maxlen,
189 CTLTYPE_INT|CTLFLAG_RW, 0, 0, sysctl_netinet6_intr_direct_queue_maxlen,
190 "I", "Maximum size of the IPv6 direct input queue");
194 VNET_DEFINE(struct pfil_head, inet6_pfil_hook);
196 VNET_PCPUSTAT_DEFINE(struct ip6stat, ip6stat);
197 VNET_PCPUSTAT_SYSINIT(ip6stat);
199 VNET_PCPUSTAT_SYSUNINIT(ip6stat);
202 struct rmlock in6_ifaddr_lock;
203 RM_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock");
205 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
207 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
211 * IP6 initialization: fill in IP6 protocol switch table.
212 * All protocols not implemented in kernel go to raw IP6 protocol handler.
220 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal",
221 &V_ip6_auto_linklocal);
222 TUNABLE_INT_FETCH("net.inet6.ip6.accept_rtadv", &V_ip6_accept_rtadv);
223 TUNABLE_INT_FETCH("net.inet6.ip6.no_radr", &V_ip6_no_radr);
225 CK_STAILQ_INIT(&V_in6_ifaddrhead);
226 V_in6_ifaddrhashtbl = hashinit(IN6ADDR_NHASH, M_IFADDR,
229 /* Initialize packet filter hooks. */
230 V_inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
231 V_inet6_pfil_hook.ph_af = AF_INET6;
232 if ((i = pfil_head_register(&V_inet6_pfil_hook)) != 0)
233 printf("%s: WARNING: unable to register pfil hook, "
234 "error %d\n", __func__, i);
236 if (hhook_head_register(HHOOK_TYPE_IPSEC_IN, AF_INET6,
237 &V_ipsec_hhh_in[HHOOK_IPSEC_INET6],
238 HHOOK_WAITOK | HHOOK_HEADISINVNET) != 0)
239 printf("%s: WARNING: unable to register input helper hook\n",
241 if (hhook_head_register(HHOOK_TYPE_IPSEC_OUT, AF_INET6,
242 &V_ipsec_hhh_out[HHOOK_IPSEC_INET6],
243 HHOOK_WAITOK | HHOOK_HEADISINVNET) != 0)
244 printf("%s: WARNING: unable to register output helper hook\n",
248 addrsel_policy_init();
252 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
254 /* Skip global initialization stuff for non-default instances. */
256 if (!IS_DEFAULT_VNET(curvnet)) {
257 netisr_register_vnet(&ip6_nh);
259 netisr_register_vnet(&ip6_direct_nh);
265 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
269 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
270 for (i = 0; i < IPPROTO_MAX; i++)
271 ip6_protox[i] = pr - inet6sw;
273 * Cycle through IP protocols and put them into the appropriate place
276 for (pr = inet6domain.dom_protosw;
277 pr < inet6domain.dom_protoswNPROTOSW; pr++)
278 if (pr->pr_domain->dom_family == PF_INET6 &&
279 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) {
280 /* Be careful to only index valid IP protocols. */
281 if (pr->pr_protocol < IPPROTO_MAX)
282 ip6_protox[pr->pr_protocol] = pr - inet6sw;
285 netisr_register(&ip6_nh);
287 netisr_register(&ip6_direct_nh);
292 * The protocol to be inserted into ip6_protox[] must be already registered
293 * in inet6sw[], either statically or through pf_proto_register().
296 ip6proto_register(short ip6proto)
301 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
302 return (EPROTONOSUPPORT);
305 * The protocol slot must not be occupied by another protocol
306 * already. An index pointing to IPPROTO_RAW is unused.
308 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
310 return (EPFNOSUPPORT);
311 if (ip6_protox[ip6proto] != pr - inet6sw) /* IPPROTO_RAW */
315 * Find the protocol position in inet6sw[] and set the index.
317 for (pr = inet6domain.dom_protosw;
318 pr < inet6domain.dom_protoswNPROTOSW; pr++) {
319 if (pr->pr_domain->dom_family == PF_INET6 &&
320 pr->pr_protocol && pr->pr_protocol == ip6proto) {
321 ip6_protox[pr->pr_protocol] = pr - inet6sw;
325 return (EPROTONOSUPPORT);
329 ip6proto_unregister(short ip6proto)
334 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
335 return (EPROTONOSUPPORT);
337 /* Check if the protocol was indeed registered. */
338 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
340 return (EPFNOSUPPORT);
341 if (ip6_protox[ip6proto] == pr - inet6sw) /* IPPROTO_RAW */
344 /* Reset the protocol slot to IPPROTO_RAW. */
345 ip6_protox[ip6proto] = pr - inet6sw;
351 ip6_destroy(void *unused __unused)
353 struct ifaddr *ifa, *nifa;
358 netisr_unregister_vnet(&ip6_direct_nh);
360 netisr_unregister_vnet(&ip6_nh);
362 if ((error = pfil_head_unregister(&V_inet6_pfil_hook)) != 0)
363 printf("%s: WARNING: unable to unregister pfil hook, "
364 "error %d\n", __func__, error);
365 error = hhook_head_deregister(V_ipsec_hhh_in[HHOOK_IPSEC_INET6]);
367 printf("%s: WARNING: unable to deregister input helper hook "
368 "type HHOOK_TYPE_IPSEC_IN, id HHOOK_IPSEC_INET6: "
369 "error %d returned\n", __func__, error);
371 error = hhook_head_deregister(V_ipsec_hhh_out[HHOOK_IPSEC_INET6]);
373 printf("%s: WARNING: unable to deregister output helper hook "
374 "type HHOOK_TYPE_IPSEC_OUT, id HHOOK_IPSEC_INET6: "
375 "error %d returned\n", __func__, error);
378 /* Cleanup addresses. */
380 CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
381 /* Cannot lock here - lock recursion. */
382 /* IF_ADDR_LOCK(ifp); */
383 CK_STAILQ_FOREACH_SAFE(ifa, &ifp->if_addrhead, ifa_link, nifa) {
385 if (ifa->ifa_addr->sa_family != AF_INET6)
389 /* IF_ADDR_UNLOCK(ifp); */
390 in6_ifdetach_destroy(ifp);
391 mld_domifdetach(ifp);
392 /* Make sure any routes are gone as well. */
393 rt_flushifroutes_af(ifp, AF_INET6);
398 in6_ifattach_destroy();
400 hashdestroy(V_in6_ifaddrhashtbl, M_IFADDR, V_in6_ifaddrhmask);
403 VNET_SYSUNINIT(inet6, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, ip6_destroy, NULL);
407 ip6_input_hbh(struct mbuf *m, uint32_t *plen, uint32_t *rtalert, int *off,
413 if (ip6_hopopts_input(plen, rtalert, &m, off)) {
414 #if 0 /*touches NULL pointer*/
415 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
417 goto out; /* m have already been freed */
421 ip6 = mtod(m, struct ip6_hdr *);
424 * if the payload length field is 0 and the next header field
425 * indicates Hop-by-Hop Options header, then a Jumbo Payload
426 * option MUST be included.
428 if (ip6->ip6_plen == 0 && *plen == 0) {
430 * Note that if a valid jumbo payload option is
431 * contained, ip6_hopopts_input() must set a valid
432 * (non-zero) payload length to the variable plen.
434 IP6STAT_INC(ip6s_badoptions);
435 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
436 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
437 icmp6_error(m, ICMP6_PARAM_PROB,
438 ICMP6_PARAMPROB_HEADER,
439 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
442 #ifndef PULLDOWN_TEST
443 /* ip6_hopopts_input() ensures that mbuf is contiguous */
444 hbh = (struct ip6_hbh *)(ip6 + 1);
446 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
447 sizeof(struct ip6_hbh));
449 IP6STAT_INC(ip6s_tooshort);
453 *nxt = hbh->ip6h_nxt;
456 * If we are acting as a router and the packet contains a
457 * router alert option, see if we know the option value.
458 * Currently, we only support the option value for MLD, in which
459 * case we should pass the packet to the multicast routing
462 if (*rtalert != ~0) {
464 case IP6OPT_RTALERT_MLD:
465 if (V_ip6_forwarding)
470 * RFC2711 requires unrecognized values must be
485 * IPv6 direct input routine.
487 * This is called when reinjecting completed fragments where
488 * all of the previous checking and book-keeping has been done.
491 ip6_direct_input(struct mbuf *m)
496 struct ip6_direct_ctx *ip6dc;
498 mtag = m_tag_locate(m, MTAG_ABI_IPV6, IPV6_TAG_DIRECT, NULL);
499 KASSERT(mtag != NULL, ("Reinjected packet w/o direct ctx tag!"));
501 ip6dc = (struct ip6_direct_ctx *)(mtag + 1);
502 nxt = ip6dc->ip6dc_nxt;
503 off = ip6dc->ip6dc_off;
507 m_tag_delete(m, mtag);
509 while (nxt != IPPROTO_DONE) {
510 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
511 IP6STAT_INC(ip6s_toomanyhdr);
516 * protection against faulty packet - there should be
517 * more sanity checks in header chain processing.
519 if (m->m_pkthdr.len < off) {
520 IP6STAT_INC(ip6s_tooshort);
521 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
525 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
526 if (IPSEC_ENABLED(ipv6)) {
527 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
532 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
541 ip6_input(struct mbuf *m)
543 struct in6_addr odst;
545 struct in6_ifaddr *ia;
548 u_int32_t rtalert = ~0;
549 int off = sizeof(struct ip6_hdr), nest;
554 * Drop the packet if IPv6 operation is disabled on the interface.
556 rcvif = m->m_pkthdr.rcvif;
557 if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED))
560 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
562 * should the inner packet be considered authentic?
563 * see comment in ah4_input().
564 * NB: m cannot be NULL when passed to the input routine
567 m->m_flags &= ~M_AUTHIPHDR;
568 m->m_flags &= ~M_AUTHIPDGM;
572 if (m->m_flags & M_FASTFWD_OURS) {
574 * Firewall changed destination to local.
576 ip6 = mtod(m, struct ip6_hdr *);
583 if (m->m_flags & M_EXT) {
585 IP6STAT_INC(ip6s_mext2m);
587 IP6STAT_INC(ip6s_mext1);
590 if (m->m_flags & M_LOOP) {
591 IP6STAT_INC(ip6s_m2m[V_loif->if_index]);
592 } else if (rcvif->if_index < IP6S_M2MMAX)
593 IP6STAT_INC(ip6s_m2m[rcvif->if_index]);
595 IP6STAT_INC(ip6s_m2m[0]);
597 IP6STAT_INC(ip6s_m1);
600 in6_ifstat_inc(rcvif, ifs6_in_receive);
601 IP6STAT_INC(ip6s_total);
603 #ifndef PULLDOWN_TEST
605 * L2 bridge code and some other code can return mbuf chain
606 * that does not conform to KAME requirement. too bad.
607 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
609 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
612 if (m->m_pkthdr.len > MHLEN)
613 n = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
615 n = m_gethdr(M_NOWAIT, MT_DATA);
620 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
621 n->m_len = n->m_pkthdr.len;
625 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
628 if (m->m_len < sizeof(struct ip6_hdr)) {
629 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
630 IP6STAT_INC(ip6s_toosmall);
631 in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
636 ip6 = mtod(m, struct ip6_hdr *);
637 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
638 IP6STAT_INC(ip6s_badvers);
639 in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
643 IP6STAT_INC(ip6s_nxthist[ip6->ip6_nxt]);
644 IP_PROBE(receive, NULL, NULL, ip6, rcvif, NULL, ip6);
647 * Check against address spoofing/corruption.
649 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
650 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
652 * XXX: "badscope" is not very suitable for a multicast source.
654 IP6STAT_INC(ip6s_badscope);
655 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
658 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
659 !(m->m_flags & M_LOOP)) {
661 * In this case, the packet should come from the loopback
662 * interface. However, we cannot just check the if_flags,
663 * because ip6_mloopback() passes the "actual" interface
664 * as the outgoing/incoming interface.
666 IP6STAT_INC(ip6s_badscope);
667 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
670 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
671 IPV6_ADDR_MC_SCOPE(&ip6->ip6_dst) == 0) {
674 * Nodes must not originate a packet to a multicast address
675 * whose scop field contains the reserved value 0; if such
676 * a packet is received, it must be silently dropped.
678 IP6STAT_INC(ip6s_badscope);
679 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
683 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
684 /* packet is dropped by traffic conditioner */
689 * The following check is not documented in specs. A malicious
690 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
691 * and bypass security checks (act as if it was from 127.0.0.1 by using
692 * IPv6 src ::ffff:127.0.0.1). Be cautious.
694 * This check chokes if we are in an SIIT cloud. As none of BSDs
695 * support IPv4-less kernel compilation, we cannot support SIIT
696 * environment at all. So, it makes more sense for us to reject any
697 * malicious packets for non-SIIT environment, than try to do a
698 * partial support for SIIT environment.
700 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
701 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
702 IP6STAT_INC(ip6s_badscope);
703 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
708 * Reject packets with IPv4 compatible addresses (auto tunnel).
710 * The code forbids auto tunnel relay case in RFC1933 (the check is
711 * stronger than RFC1933). We may want to re-enable it if mech-xx
712 * is revised to forbid relaying case.
714 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
715 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
716 IP6STAT_INC(ip6s_badscope);
717 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
722 * Try to forward the packet, but if we fail continue.
723 * ip6_tryforward() does inbound and outbound packet firewall
724 * processing. If firewall has decided that destination becomes
725 * our local address, it sets M_FASTFWD_OURS flag. In this
726 * case skip another inbound firewall processing and update
729 if (V_ip6_forwarding != 0
730 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
731 && (!IPSEC_ENABLED(ipv6) ||
732 IPSEC_CAPS(ipv6, m, IPSEC_CAP_OPERABLE) == 0)
735 if ((m = ip6_tryforward(m)) == NULL)
737 if (m->m_flags & M_FASTFWD_OURS) {
738 ip6 = mtod(m, struct ip6_hdr *);
742 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
744 * Bypass packet filtering for packets previously handled by IPsec.
746 if (IPSEC_ENABLED(ipv6) &&
747 IPSEC_CAPS(ipv6, m, IPSEC_CAP_BYPASS_FILTER) != 0)
751 * Run through list of hooks for input packets.
753 * NB: Beware of the destination address changing
754 * (e.g. by NAT rewriting). When this happens,
755 * tell ip6_forward to do the right thing.
758 /* Jump over all PFIL processing if hooks are not active. */
759 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
763 if (pfil_run_hooks(&V_inet6_pfil_hook, &m,
764 m->m_pkthdr.rcvif, PFIL_IN, 0, NULL))
766 if (m == NULL) /* consumed by filter */
768 ip6 = mtod(m, struct ip6_hdr *);
769 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
770 if ((m->m_flags & (M_IP6_NEXTHOP | M_FASTFWD_OURS)) == M_IP6_NEXTHOP &&
771 m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
773 * Directly ship the packet on. This allows forwarding
774 * packets originally destined to us to some other directly
783 * Disambiguate address scope zones (if there is ambiguity).
784 * We first make sure that the original source or destination address
785 * is not in our internal form for scoped addresses. Such addresses
786 * are not necessarily invalid spec-wise, but we cannot accept them due
787 * to the usage conflict.
788 * in6_setscope() then also checks and rejects the cases where src or
789 * dst are the loopback address and the receiving interface
792 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
793 IP6STAT_INC(ip6s_badscope); /* XXX */
796 if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
797 in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
798 IP6STAT_INC(ip6s_badscope);
801 if (m->m_flags & M_FASTFWD_OURS) {
802 m->m_flags &= ~M_FASTFWD_OURS;
807 * Multicast check. Assume packet is for us to avoid
808 * prematurely taking locks.
810 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
812 in6_ifstat_inc(rcvif, ifs6_in_mcast);
817 * XXX: For now we keep link-local IPv6 addresses with embedded
818 * scope zone id, therefore we use zero zoneid here.
820 ia = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */);
822 if (ia->ia6_flags & IN6_IFF_NOTREADY) {
823 char ip6bufs[INET6_ADDRSTRLEN];
824 char ip6bufd[INET6_ADDRSTRLEN];
825 /* address is not ready, so discard the packet. */
827 "ip6_input: packet to an unready address %s->%s\n",
828 ip6_sprintf(ip6bufs, &ip6->ip6_src),
829 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
830 ifa_free(&ia->ia_ifa);
833 /* Count the packet in the ip address stats */
834 counter_u64_add(ia->ia_ifa.ifa_ipackets, 1);
835 counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len);
836 ifa_free(&ia->ia_ifa);
842 * Now there is no reason to process the packet if it's not our own
843 * and we're not a router.
845 if (!V_ip6_forwarding) {
846 IP6STAT_INC(ip6s_cantforward);
852 * Process Hop-by-Hop options header if it's contained.
853 * m may be modified in ip6_hopopts_input().
854 * If a JumboPayload option is included, plen will also be modified.
856 plen = (u_int32_t)ntohs(ip6->ip6_plen);
857 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
858 if (ip6_input_hbh(m, &plen, &rtalert, &off, &nxt, &ours) != 0)
864 * Use mbuf flags to propagate Router Alert option to
865 * ICMPv6 layer, as hop-by-hop options have been stripped.
868 m->m_flags |= M_RTALERT_MLD;
871 * Check that the amount of data in the buffers
872 * is as at least much as the IPv6 header would have us expect.
873 * Trim mbufs if longer than we expect.
874 * Drop packet if shorter than we expect.
876 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
877 IP6STAT_INC(ip6s_tooshort);
878 in6_ifstat_inc(rcvif, ifs6_in_truncated);
881 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
882 if (m->m_len == m->m_pkthdr.len) {
883 m->m_len = sizeof(struct ip6_hdr) + plen;
884 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
886 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
890 * Forward if desirable.
893 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
895 * If we are acting as a multicast router, all
896 * incoming multicast packets are passed to the
897 * kernel-level multicast forwarding function.
898 * The packet is returned (relatively) intact; if
899 * ip6_mforward() returns a non-zero value, the packet
900 * must be discarded, else it may be accepted below.
902 * XXX TODO: Check hlim and multicast scope here to avoid
903 * unnecessarily calling into ip6_mforward().
905 if (ip6_mforward && ip6_mforward(ip6, rcvif, m)) {
906 IP6STAT_INC(ip6s_cantforward);
910 ip6_forward(m, srcrt);
914 ip6 = mtod(m, struct ip6_hdr *);
917 * Malicious party may be able to use IPv4 mapped addr to confuse
918 * tcp/udp stack and bypass security checks (act as if it was from
919 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
921 * For SIIT end node behavior, you may want to disable the check.
922 * However, you will become vulnerable to attacks using IPv4 mapped
925 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
926 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
927 IP6STAT_INC(ip6s_badscope);
928 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
933 * Tell launch routine the next header
935 IP6STAT_INC(ip6s_delivered);
936 in6_ifstat_inc(rcvif, ifs6_in_deliver);
939 while (nxt != IPPROTO_DONE) {
940 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
941 IP6STAT_INC(ip6s_toomanyhdr);
946 * protection against faulty packet - there should be
947 * more sanity checks in header chain processing.
949 if (m->m_pkthdr.len < off) {
950 IP6STAT_INC(ip6s_tooshort);
951 in6_ifstat_inc(rcvif, ifs6_in_truncated);
955 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
956 if (IPSEC_ENABLED(ipv6)) {
957 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
962 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
966 in6_ifstat_inc(rcvif, ifs6_in_discard);
972 * Hop-by-Hop options header processing. If a valid jumbo payload option is
973 * included, the real payload length will be stored in plenp.
975 * rtalertp - XXX: should be stored more smart way
978 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
979 struct mbuf **mp, int *offp)
981 struct mbuf *m = *mp;
982 int off = *offp, hbhlen;
985 /* validation of the length of the header */
986 #ifndef PULLDOWN_TEST
987 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
988 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
989 hbhlen = (hbh->ip6h_len + 1) << 3;
991 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
992 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
994 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
995 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
997 IP6STAT_INC(ip6s_tooshort);
1000 hbhlen = (hbh->ip6h_len + 1) << 3;
1001 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
1004 IP6STAT_INC(ip6s_tooshort);
1009 hbhlen -= sizeof(struct ip6_hbh);
1010 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
1011 hbhlen, rtalertp, plenp) < 0)
1020 * Search header for all Hop-by-hop options and process each option.
1021 * This function is separate from ip6_hopopts_input() in order to
1022 * handle a case where the sending node itself process its hop-by-hop
1023 * options header. In such a case, the function is called from ip6_output().
1025 * The function assumes that hbh header is located right after the IPv6 header
1026 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1027 * opthead + hbhlen is located in contiguous memory region.
1030 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
1031 u_int32_t *rtalertp, u_int32_t *plenp)
1033 struct ip6_hdr *ip6;
1035 u_int8_t *opt = opthead;
1036 u_int16_t rtalert_val;
1037 u_int32_t jumboplen;
1038 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1040 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1046 if (hbhlen < IP6OPT_MINLEN) {
1047 IP6STAT_INC(ip6s_toosmall);
1050 optlen = *(opt + 1) + 2;
1052 case IP6OPT_ROUTER_ALERT:
1053 /* XXX may need check for alignment */
1054 if (hbhlen < IP6OPT_RTALERT_LEN) {
1055 IP6STAT_INC(ip6s_toosmall);
1058 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1060 icmp6_error(m, ICMP6_PARAM_PROB,
1061 ICMP6_PARAMPROB_HEADER,
1062 erroff + opt + 1 - opthead);
1065 optlen = IP6OPT_RTALERT_LEN;
1066 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1067 *rtalertp = ntohs(rtalert_val);
1070 /* XXX may need check for alignment */
1071 if (hbhlen < IP6OPT_JUMBO_LEN) {
1072 IP6STAT_INC(ip6s_toosmall);
1075 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1077 icmp6_error(m, ICMP6_PARAM_PROB,
1078 ICMP6_PARAMPROB_HEADER,
1079 erroff + opt + 1 - opthead);
1082 optlen = IP6OPT_JUMBO_LEN;
1085 * IPv6 packets that have non 0 payload length
1086 * must not contain a jumbo payload option.
1088 ip6 = mtod(m, struct ip6_hdr *);
1089 if (ip6->ip6_plen) {
1090 IP6STAT_INC(ip6s_badoptions);
1091 icmp6_error(m, ICMP6_PARAM_PROB,
1092 ICMP6_PARAMPROB_HEADER,
1093 erroff + opt - opthead);
1098 * We may see jumbolen in unaligned location, so
1099 * we'd need to perform bcopy().
1101 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1102 jumboplen = (u_int32_t)htonl(jumboplen);
1106 * if there are multiple jumbo payload options,
1107 * *plenp will be non-zero and the packet will be
1109 * the behavior may need some debate in ipngwg -
1110 * multiple options does not make sense, however,
1111 * there's no explicit mention in specification.
1114 IP6STAT_INC(ip6s_badoptions);
1115 icmp6_error(m, ICMP6_PARAM_PROB,
1116 ICMP6_PARAMPROB_HEADER,
1117 erroff + opt + 2 - opthead);
1123 * jumbo payload length must be larger than 65535.
1125 if (jumboplen <= IPV6_MAXPACKET) {
1126 IP6STAT_INC(ip6s_badoptions);
1127 icmp6_error(m, ICMP6_PARAM_PROB,
1128 ICMP6_PARAMPROB_HEADER,
1129 erroff + opt + 2 - opthead);
1135 default: /* unknown option */
1136 if (hbhlen < IP6OPT_MINLEN) {
1137 IP6STAT_INC(ip6s_toosmall);
1140 optlen = ip6_unknown_opt(opt, m,
1141 erroff + opt - opthead);
1157 * Unknown option processing.
1158 * The third argument `off' is the offset from the IPv6 header to the option,
1159 * which is necessary if the IPv6 header the and option header and IPv6 header
1160 * is not contiguous in order to return an ICMPv6 error.
1163 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1165 struct ip6_hdr *ip6;
1167 switch (IP6OPT_TYPE(*optp)) {
1168 case IP6OPT_TYPE_SKIP: /* ignore the option */
1169 return ((int)*(optp + 1));
1170 case IP6OPT_TYPE_DISCARD: /* silently discard */
1173 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1174 IP6STAT_INC(ip6s_badoptions);
1175 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1177 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1178 IP6STAT_INC(ip6s_badoptions);
1179 ip6 = mtod(m, struct ip6_hdr *);
1180 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1181 (m->m_flags & (M_BCAST|M_MCAST)))
1184 icmp6_error(m, ICMP6_PARAM_PROB,
1185 ICMP6_PARAMPROB_OPTION, off);
1189 m_freem(m); /* XXX: NOTREACHED */
1194 * Create the "control" list for this pcb.
1195 * These functions will not modify mbuf chain at all.
1197 * With KAME mbuf chain restriction:
1198 * The routine will be called from upper layer handlers like tcp6_input().
1199 * Thus the routine assumes that the caller (tcp6_input) have already
1200 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1201 * very first mbuf on the mbuf chain.
1203 * ip6_savecontrol_v4 will handle those options that are possible to be
1204 * set on a v4-mapped socket.
1205 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1206 * options and handle the v6-only ones itself.
1209 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1212 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1215 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1221 struct bintime boottimebin, bt1;
1222 struct timespec ts1;
1226 switch (inp->inp_socket->so_ts_clock) {
1227 case SO_TS_REALTIME_MICRO:
1228 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1230 mbuf_tstmp2timespec(m, &ts1);
1231 timespec2bintime(&ts1, &bt1);
1232 getboottimebin(&boottimebin);
1233 bintime_add(&bt1, &boottimebin);
1234 bintime2timeval(&bt1, &t.tv);
1238 *mp = sbcreatecontrol((caddr_t) &t.tv, sizeof(t.tv),
1239 SCM_TIMESTAMP, SOL_SOCKET);
1241 mp = &(*mp)->m_next;
1247 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1249 mbuf_tstmp2timespec(m, &ts1);
1250 timespec2bintime(&ts1, &t.bt);
1251 getboottimebin(&boottimebin);
1252 bintime_add(&t.bt, &boottimebin);
1256 *mp = sbcreatecontrol((caddr_t)&t.bt, sizeof(t.bt),
1257 SCM_BINTIME, SOL_SOCKET);
1259 mp = &(*mp)->m_next;
1264 case SO_TS_REALTIME:
1265 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1267 mbuf_tstmp2timespec(m, &t.ts);
1268 getboottimebin(&boottimebin);
1269 bintime2timespec(&boottimebin, &ts1);
1270 timespecadd(&t.ts, &ts1);
1274 *mp = sbcreatecontrol((caddr_t)&t.ts, sizeof(t.ts),
1275 SCM_REALTIME, SOL_SOCKET);
1277 mp = &(*mp)->m_next;
1282 case SO_TS_MONOTONIC:
1283 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1285 mbuf_tstmp2timespec(m, &t.ts);
1288 *mp = sbcreatecontrol((caddr_t)&t.ts, sizeof(t.ts),
1289 SCM_MONOTONIC, SOL_SOCKET);
1291 mp = &(*mp)->m_next;
1297 panic("unknown (corrupted) so_ts_clock");
1299 if (stamped && (m->m_flags & (M_PKTHDR | M_TSTMP)) ==
1300 (M_PKTHDR | M_TSTMP)) {
1301 struct sock_timestamp_info sti;
1303 bzero(&sti, sizeof(sti));
1304 sti.st_info_flags = ST_INFO_HW;
1305 if ((m->m_flags & M_TSTMP_HPREC) != 0)
1306 sti.st_info_flags |= ST_INFO_HW_HPREC;
1307 *mp = sbcreatecontrol((caddr_t)&sti, sizeof(sti),
1308 SCM_TIME_INFO, SOL_SOCKET);
1310 mp = &(*mp)->m_next;
1315 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1316 /* RFC 2292 sec. 5 */
1317 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1318 struct in6_pktinfo pi6;
1320 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1324 ip = mtod(m, struct ip *);
1325 pi6.ipi6_addr.s6_addr32[0] = 0;
1326 pi6.ipi6_addr.s6_addr32[1] = 0;
1327 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1328 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1330 /* We won't hit this code */
1331 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1334 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1335 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1338 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1340 *mp = sbcreatecontrol((caddr_t) &pi6,
1341 sizeof(struct in6_pktinfo),
1342 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1344 mp = &(*mp)->m_next;
1347 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1350 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1354 ip = mtod(m, struct ip *);
1357 /* We won't hit this code */
1361 hlim = ip6->ip6_hlim & 0xff;
1363 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1364 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1367 mp = &(*mp)->m_next;
1370 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1373 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1377 ip = mtod(m, struct ip *);
1378 tclass = ip->ip_tos;
1380 /* We won't hit this code */
1386 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1388 tclass = flowinfo & 0xff;
1390 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(int),
1391 IPV6_TCLASS, IPPROTO_IPV6);
1393 mp = &(*mp)->m_next;
1396 if (v4only != NULL) {
1397 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1408 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1410 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1413 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1418 * IPV6_HOPOPTS socket option. Recall that we required super-user
1419 * privilege for the option (see ip6_ctloutput), but it might be too
1420 * strict, since there might be some hop-by-hop options which can be
1421 * returned to normal user.
1422 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1424 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1426 * Check if a hop-by-hop options header is contatined in the
1427 * received packet, and if so, store the options as ancillary
1428 * data. Note that a hop-by-hop options header must be
1429 * just after the IPv6 header, which is assured through the
1430 * IPv6 input processing.
1432 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1433 struct ip6_hbh *hbh;
1435 #ifdef PULLDOWN_TEST
1439 #ifndef PULLDOWN_TEST
1440 hbh = (struct ip6_hbh *)(ip6 + 1);
1441 hbhlen = (hbh->ip6h_len + 1) << 3;
1443 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1446 IP6STAT_INC(ip6s_tooshort);
1449 hbh = mtod(ext, struct ip6_hbh *);
1450 hbhlen = (hbh->ip6h_len + 1) << 3;
1451 if (hbhlen != ext->m_len) {
1453 IP6STAT_INC(ip6s_tooshort);
1459 * XXX: We copy the whole header even if a
1460 * jumbo payload option is included, the option which
1461 * is to be removed before returning according to
1463 * Note: this constraint is removed in RFC3542
1465 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1466 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1469 mp = &(*mp)->m_next;
1470 #ifdef PULLDOWN_TEST
1476 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1477 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1480 * Search for destination options headers or routing
1481 * header(s) through the header chain, and stores each
1482 * header as ancillary data.
1483 * Note that the order of the headers remains in
1484 * the chain of ancillary data.
1486 while (1) { /* is explicit loop prevention necessary? */
1487 struct ip6_ext *ip6e = NULL;
1489 #ifdef PULLDOWN_TEST
1490 struct mbuf *ext = NULL;
1494 * if it is not an extension header, don't try to
1495 * pull it from the chain.
1498 case IPPROTO_DSTOPTS:
1499 case IPPROTO_ROUTING:
1500 case IPPROTO_HOPOPTS:
1501 case IPPROTO_AH: /* is it possible? */
1507 #ifndef PULLDOWN_TEST
1508 if (off + sizeof(*ip6e) > m->m_len)
1510 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1511 if (nxt == IPPROTO_AH)
1512 elen = (ip6e->ip6e_len + 2) << 2;
1514 elen = (ip6e->ip6e_len + 1) << 3;
1515 if (off + elen > m->m_len)
1518 ext = ip6_pullexthdr(m, off, nxt);
1520 IP6STAT_INC(ip6s_tooshort);
1523 ip6e = mtod(ext, struct ip6_ext *);
1524 if (nxt == IPPROTO_AH)
1525 elen = (ip6e->ip6e_len + 2) << 2;
1527 elen = (ip6e->ip6e_len + 1) << 3;
1528 if (elen != ext->m_len) {
1530 IP6STAT_INC(ip6s_tooshort);
1536 case IPPROTO_DSTOPTS:
1537 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1540 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1542 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1545 mp = &(*mp)->m_next;
1547 case IPPROTO_ROUTING:
1548 if (!(in6p->inp_flags & IN6P_RTHDR))
1551 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1552 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1555 mp = &(*mp)->m_next;
1557 case IPPROTO_HOPOPTS:
1558 case IPPROTO_AH: /* is it possible? */
1563 * other cases have been filtered in the above.
1564 * none will visit this case. here we supply
1565 * the code just in case (nxt overwritten or
1568 #ifdef PULLDOWN_TEST
1575 /* proceed with the next header. */
1577 nxt = ip6e->ip6e_nxt;
1579 #ifdef PULLDOWN_TEST
1588 if (in6p->inp_flags2 & INP_RECVFLOWID) {
1589 uint32_t flowid, flow_type;
1591 flowid = m->m_pkthdr.flowid;
1592 flow_type = M_HASHTYPE_GET(m);
1595 * XXX should handle the failure of one or the
1596 * other - don't populate both?
1598 *mp = sbcreatecontrol((caddr_t) &flowid,
1599 sizeof(uint32_t), IPV6_FLOWID, IPPROTO_IPV6);
1601 mp = &(*mp)->m_next;
1602 *mp = sbcreatecontrol((caddr_t) &flow_type,
1603 sizeof(uint32_t), IPV6_FLOWTYPE, IPPROTO_IPV6);
1605 mp = &(*mp)->m_next;
1609 if (in6p->inp_flags2 & INP_RECVRSSBUCKETID) {
1610 uint32_t flowid, flow_type;
1611 uint32_t rss_bucketid;
1613 flowid = m->m_pkthdr.flowid;
1614 flow_type = M_HASHTYPE_GET(m);
1616 if (rss_hash2bucket(flowid, flow_type, &rss_bucketid) == 0) {
1617 *mp = sbcreatecontrol((caddr_t) &rss_bucketid,
1618 sizeof(uint32_t), IPV6_RSSBUCKETID, IPPROTO_IPV6);
1620 mp = &(*mp)->m_next;
1629 ip6_notify_pmtu(struct inpcb *inp, struct sockaddr_in6 *dst, u_int32_t mtu)
1633 struct ip6_mtuinfo mtuctl;
1635 KASSERT(inp != NULL, ("%s: inp == NULL", __func__));
1637 * Notify the error by sending IPV6_PATHMTU ancillary data if
1638 * application wanted to know the MTU value.
1639 * NOTE: we notify disconnected sockets, because some udp
1640 * applications keep sending sockets disconnected.
1641 * NOTE: our implementation doesn't notify connected sockets that has
1642 * foreign address that is different than given destination addresses
1643 * (this is permitted by RFC 3542).
1645 if ((inp->inp_flags & IN6P_MTU) == 0 || (
1646 !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1647 !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &dst->sin6_addr)))
1650 mtuctl.ip6m_mtu = mtu;
1651 mtuctl.ip6m_addr = *dst;
1652 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1655 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1656 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1659 so = inp->inp_socket;
1660 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1663 /* XXX: should count statistics */
1668 #ifdef PULLDOWN_TEST
1670 * pull single extension header from mbuf chain. returns single mbuf that
1671 * contains the result, or NULL on error.
1673 static struct mbuf *
1674 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1676 struct ip6_ext ip6e;
1682 case IPPROTO_DSTOPTS:
1683 case IPPROTO_ROUTING:
1684 case IPPROTO_HOPOPTS:
1685 case IPPROTO_AH: /* is it possible? */
1688 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1692 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1693 if (nxt == IPPROTO_AH)
1694 elen = (ip6e.ip6e_len + 2) << 2;
1696 elen = (ip6e.ip6e_len + 1) << 3;
1699 n = m_getcl(M_NOWAIT, MT_DATA, 0);
1701 n = m_get(M_NOWAIT, MT_DATA);
1705 m_copydata(m, off, elen, mtod(n, caddr_t));
1712 * Get pointer to the previous header followed by the header
1713 * currently processed.
1716 ip6_get_prevhdr(const struct mbuf *m, int off)
1718 struct ip6_ext ip6e;
1719 struct ip6_hdr *ip6;
1722 if (off == sizeof(struct ip6_hdr))
1723 return (offsetof(struct ip6_hdr, ip6_nxt));
1724 if (off < sizeof(struct ip6_hdr))
1725 panic("%s: off < sizeof(struct ip6_hdr)", __func__);
1727 ip6 = mtod(m, struct ip6_hdr *);
1729 len = sizeof(struct ip6_hdr);
1732 m_copydata(m, len, sizeof(ip6e), (caddr_t)&ip6e);
1734 case IPPROTO_FRAGMENT:
1735 nlen = sizeof(struct ip6_frag);
1738 nlen = (ip6e.ip6e_len + 2) << 2;
1741 nlen = (ip6e.ip6e_len + 1) << 3;
1744 nxt = ip6e.ip6e_nxt;
1746 return (len - nlen);
1750 * get next header offset. m will be retained.
1753 ip6_nexthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1756 struct ip6_ext ip6e;
1761 panic("ip6_nexthdr: m == NULL");
1762 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1767 if (m->m_pkthdr.len < off + sizeof(ip6))
1769 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1771 *nxtp = ip6.ip6_nxt;
1775 case IPPROTO_FRAGMENT:
1777 * terminate parsing if it is not the first fragment,
1778 * it does not make sense to parse through it.
1780 if (m->m_pkthdr.len < off + sizeof(fh))
1782 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1783 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1784 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1787 *nxtp = fh.ip6f_nxt;
1788 off += sizeof(struct ip6_frag);
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 + 2) << 2;
1800 case IPPROTO_HOPOPTS:
1801 case IPPROTO_ROUTING:
1802 case IPPROTO_DSTOPTS:
1803 if (m->m_pkthdr.len < off + sizeof(ip6e))
1805 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1807 *nxtp = ip6e.ip6e_nxt;
1808 off += (ip6e.ip6e_len + 1) << 3;
1813 case IPPROTO_IPCOMP:
1825 * get offset for the last header in the chain. m will be kept untainted.
1828 ip6_lasthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1838 newoff = ip6_nexthdr(m, off, proto, nxtp);
1841 else if (newoff < off)
1842 return -1; /* invalid */
1843 else if (newoff == off)
1852 * System control for IP6
1855 u_char inet6ctlerrmap[PRC_NCMDS] = {
1857 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1858 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1859 EMSGSIZE, EHOSTUNREACH, 0, 0,
1860 0, 0, EHOSTUNREACH, 0,
1861 ENOPROTOOPT, ECONNREFUSED