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 **mp, uint32_t *plen, uint32_t *rtalert, int *off,
414 if (ip6_hopopts_input(plen, rtalert, mp, off)) {
415 #if 0 /*touches NULL pointer*/
416 in6_ifstat_inc((*mp)->m_pkthdr.rcvif, ifs6_in_discard);
418 goto out; /* m have already been freed */
423 ip6 = mtod(m, struct ip6_hdr *);
426 * if the payload length field is 0 and the next header field
427 * indicates Hop-by-Hop Options header, then a Jumbo Payload
428 * option MUST be included.
430 if (ip6->ip6_plen == 0 && *plen == 0) {
432 * Note that if a valid jumbo payload option is
433 * contained, ip6_hopopts_input() must set a valid
434 * (non-zero) payload length to the variable plen.
436 IP6STAT_INC(ip6s_badoptions);
437 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
438 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
439 icmp6_error(m, ICMP6_PARAM_PROB,
440 ICMP6_PARAMPROB_HEADER,
441 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
444 #ifndef PULLDOWN_TEST
445 /* ip6_hopopts_input() ensures that mbuf is contiguous */
446 hbh = (struct ip6_hbh *)(ip6 + 1);
448 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
449 sizeof(struct ip6_hbh));
451 IP6STAT_INC(ip6s_tooshort);
455 *nxt = hbh->ip6h_nxt;
458 * If we are acting as a router and the packet contains a
459 * router alert option, see if we know the option value.
460 * Currently, we only support the option value for MLD, in which
461 * case we should pass the packet to the multicast routing
464 if (*rtalert != ~0) {
466 case IP6OPT_RTALERT_MLD:
467 if (V_ip6_forwarding)
472 * RFC2711 requires unrecognized values must be
487 * IPv6 direct input routine.
489 * This is called when reinjecting completed fragments where
490 * all of the previous checking and book-keeping has been done.
493 ip6_direct_input(struct mbuf *m)
498 struct ip6_direct_ctx *ip6dc;
500 mtag = m_tag_locate(m, MTAG_ABI_IPV6, IPV6_TAG_DIRECT, NULL);
501 KASSERT(mtag != NULL, ("Reinjected packet w/o direct ctx tag!"));
503 ip6dc = (struct ip6_direct_ctx *)(mtag + 1);
504 nxt = ip6dc->ip6dc_nxt;
505 off = ip6dc->ip6dc_off;
509 m_tag_delete(m, mtag);
511 while (nxt != IPPROTO_DONE) {
512 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
513 IP6STAT_INC(ip6s_toomanyhdr);
518 * protection against faulty packet - there should be
519 * more sanity checks in header chain processing.
521 if (m->m_pkthdr.len < off) {
522 IP6STAT_INC(ip6s_tooshort);
523 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
527 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
528 if (IPSEC_ENABLED(ipv6)) {
529 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
534 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
543 ip6_input(struct mbuf *m)
545 struct in6_addr odst;
547 struct in6_ifaddr *ia;
550 u_int32_t rtalert = ~0;
551 int off = sizeof(struct ip6_hdr), nest;
556 * Drop the packet if IPv6 operation is disabled on the interface.
558 rcvif = m->m_pkthdr.rcvif;
559 if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED))
562 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
564 * should the inner packet be considered authentic?
565 * see comment in ah4_input().
566 * NB: m cannot be NULL when passed to the input routine
569 m->m_flags &= ~M_AUTHIPHDR;
570 m->m_flags &= ~M_AUTHIPDGM;
574 if (m->m_flags & M_FASTFWD_OURS) {
576 * Firewall changed destination to local.
578 ip6 = mtod(m, struct ip6_hdr *);
585 if (m->m_flags & M_EXT) {
587 IP6STAT_INC(ip6s_mext2m);
589 IP6STAT_INC(ip6s_mext1);
592 if (m->m_flags & M_LOOP) {
593 IP6STAT_INC(ip6s_m2m[V_loif->if_index]);
594 } else if (rcvif->if_index < IP6S_M2MMAX)
595 IP6STAT_INC(ip6s_m2m[rcvif->if_index]);
597 IP6STAT_INC(ip6s_m2m[0]);
599 IP6STAT_INC(ip6s_m1);
602 in6_ifstat_inc(rcvif, ifs6_in_receive);
603 IP6STAT_INC(ip6s_total);
605 #ifndef PULLDOWN_TEST
607 * L2 bridge code and some other code can return mbuf chain
608 * that does not conform to KAME requirement. too bad.
609 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
611 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
614 if (m->m_pkthdr.len > MHLEN)
615 n = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
617 n = m_gethdr(M_NOWAIT, MT_DATA);
622 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
623 n->m_len = n->m_pkthdr.len;
627 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
630 if (m->m_len < sizeof(struct ip6_hdr)) {
631 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
632 IP6STAT_INC(ip6s_toosmall);
633 in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
638 ip6 = mtod(m, struct ip6_hdr *);
639 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
640 IP6STAT_INC(ip6s_badvers);
641 in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
645 IP6STAT_INC(ip6s_nxthist[ip6->ip6_nxt]);
646 IP_PROBE(receive, NULL, NULL, ip6, rcvif, NULL, ip6);
649 * Check against address spoofing/corruption.
651 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
652 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
654 * XXX: "badscope" is not very suitable for a multicast source.
656 IP6STAT_INC(ip6s_badscope);
657 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
660 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
661 !(m->m_flags & M_LOOP)) {
663 * In this case, the packet should come from the loopback
664 * interface. However, we cannot just check the if_flags,
665 * because ip6_mloopback() passes the "actual" interface
666 * as the outgoing/incoming interface.
668 IP6STAT_INC(ip6s_badscope);
669 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
672 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
673 IPV6_ADDR_MC_SCOPE(&ip6->ip6_dst) == 0) {
676 * Nodes must not originate a packet to a multicast address
677 * whose scop field contains the reserved value 0; if such
678 * a packet is received, it must be silently dropped.
680 IP6STAT_INC(ip6s_badscope);
681 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
685 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
686 /* packet is dropped by traffic conditioner */
691 * The following check is not documented in specs. A malicious
692 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
693 * and bypass security checks (act as if it was from 127.0.0.1 by using
694 * IPv6 src ::ffff:127.0.0.1). Be cautious.
696 * This check chokes if we are in an SIIT cloud. As none of BSDs
697 * support IPv4-less kernel compilation, we cannot support SIIT
698 * environment at all. So, it makes more sense for us to reject any
699 * malicious packets for non-SIIT environment, than try to do a
700 * partial support for SIIT environment.
702 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
703 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
704 IP6STAT_INC(ip6s_badscope);
705 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
710 * Reject packets with IPv4 compatible addresses (auto tunnel).
712 * The code forbids auto tunnel relay case in RFC1933 (the check is
713 * stronger than RFC1933). We may want to re-enable it if mech-xx
714 * is revised to forbid relaying case.
716 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
717 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
718 IP6STAT_INC(ip6s_badscope);
719 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
724 * Try to forward the packet, but if we fail continue.
725 * ip6_tryforward() does not generate redirects, so fall
726 * through to normal processing if redirects are required.
727 * ip6_tryforward() does inbound and outbound packet firewall
728 * processing. If firewall has decided that destination becomes
729 * our local address, it sets M_FASTFWD_OURS flag. In this
730 * case skip another inbound firewall processing and update
733 if (V_ip6_forwarding != 0 && V_ip6_sendredirects == 0
734 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
735 && (!IPSEC_ENABLED(ipv6) ||
736 IPSEC_CAPS(ipv6, m, IPSEC_CAP_OPERABLE) == 0)
739 if ((m = ip6_tryforward(m)) == NULL)
741 if (m->m_flags & M_FASTFWD_OURS) {
742 ip6 = mtod(m, struct ip6_hdr *);
746 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
748 * Bypass packet filtering for packets previously handled by IPsec.
750 if (IPSEC_ENABLED(ipv6) &&
751 IPSEC_CAPS(ipv6, m, IPSEC_CAP_BYPASS_FILTER) != 0)
755 * Run through list of hooks for input packets.
757 * NB: Beware of the destination address changing
758 * (e.g. by NAT rewriting). When this happens,
759 * tell ip6_forward to do the right thing.
762 /* Jump over all PFIL processing if hooks are not active. */
763 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
767 if (pfil_run_hooks(&V_inet6_pfil_hook, &m,
768 m->m_pkthdr.rcvif, PFIL_IN, 0, NULL))
770 if (m == NULL) /* consumed by filter */
772 ip6 = mtod(m, struct ip6_hdr *);
773 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
774 if ((m->m_flags & (M_IP6_NEXTHOP | M_FASTFWD_OURS)) == M_IP6_NEXTHOP &&
775 m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
777 * Directly ship the packet on. This allows forwarding
778 * packets originally destined to us to some other directly
787 * Disambiguate address scope zones (if there is ambiguity).
788 * We first make sure that the original source or destination address
789 * is not in our internal form for scoped addresses. Such addresses
790 * are not necessarily invalid spec-wise, but we cannot accept them due
791 * to the usage conflict.
792 * in6_setscope() then also checks and rejects the cases where src or
793 * dst are the loopback address and the receiving interface
796 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
797 IP6STAT_INC(ip6s_badscope); /* XXX */
800 if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
801 in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
802 IP6STAT_INC(ip6s_badscope);
805 if (m->m_flags & M_FASTFWD_OURS) {
806 m->m_flags &= ~M_FASTFWD_OURS;
811 * Multicast check. Assume packet is for us to avoid
812 * prematurely taking locks.
814 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
816 in6_ifstat_inc(rcvif, ifs6_in_mcast);
821 * XXX: For now we keep link-local IPv6 addresses with embedded
822 * scope zone id, therefore we use zero zoneid here.
824 ia = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */);
826 if (ia->ia6_flags & IN6_IFF_NOTREADY) {
827 char ip6bufs[INET6_ADDRSTRLEN];
828 char ip6bufd[INET6_ADDRSTRLEN];
829 /* address is not ready, so discard the packet. */
831 "ip6_input: packet to an unready address %s->%s\n",
832 ip6_sprintf(ip6bufs, &ip6->ip6_src),
833 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
834 ifa_free(&ia->ia_ifa);
837 /* Count the packet in the ip address stats */
838 counter_u64_add(ia->ia_ifa.ifa_ipackets, 1);
839 counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len);
840 ifa_free(&ia->ia_ifa);
846 * Now there is no reason to process the packet if it's not our own
847 * and we're not a router.
849 if (!V_ip6_forwarding) {
850 IP6STAT_INC(ip6s_cantforward);
856 * Process Hop-by-Hop options header if it's contained.
857 * m may be modified in ip6_hopopts_input().
858 * If a JumboPayload option is included, plen will also be modified.
860 plen = (u_int32_t)ntohs(ip6->ip6_plen);
861 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
862 if (ip6_input_hbh(&m, &plen, &rtalert, &off, &nxt, &ours) != 0)
868 * Use mbuf flags to propagate Router Alert option to
869 * ICMPv6 layer, as hop-by-hop options have been stripped.
872 m->m_flags |= M_RTALERT_MLD;
875 * Check that the amount of data in the buffers
876 * is as at least much as the IPv6 header would have us expect.
877 * Trim mbufs if longer than we expect.
878 * Drop packet if shorter than we expect.
880 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
881 IP6STAT_INC(ip6s_tooshort);
882 in6_ifstat_inc(rcvif, ifs6_in_truncated);
885 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
886 if (m->m_len == m->m_pkthdr.len) {
887 m->m_len = sizeof(struct ip6_hdr) + plen;
888 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
890 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
894 * Forward if desirable.
897 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
899 * If we are acting as a multicast router, all
900 * incoming multicast packets are passed to the
901 * kernel-level multicast forwarding function.
902 * The packet is returned (relatively) intact; if
903 * ip6_mforward() returns a non-zero value, the packet
904 * must be discarded, else it may be accepted below.
906 * XXX TODO: Check hlim and multicast scope here to avoid
907 * unnecessarily calling into ip6_mforward().
909 if (ip6_mforward && ip6_mforward(ip6, rcvif, m)) {
910 IP6STAT_INC(ip6s_cantforward);
914 ip6_forward(m, srcrt);
918 ip6 = mtod(m, struct ip6_hdr *);
921 * Malicious party may be able to use IPv4 mapped addr to confuse
922 * tcp/udp stack and bypass security checks (act as if it was from
923 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
925 * For SIIT end node behavior, you may want to disable the check.
926 * However, you will become vulnerable to attacks using IPv4 mapped
929 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
930 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
931 IP6STAT_INC(ip6s_badscope);
932 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
937 * Tell launch routine the next header
939 IP6STAT_INC(ip6s_delivered);
940 in6_ifstat_inc(rcvif, ifs6_in_deliver);
943 while (nxt != IPPROTO_DONE) {
944 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
945 IP6STAT_INC(ip6s_toomanyhdr);
950 * protection against faulty packet - there should be
951 * more sanity checks in header chain processing.
953 if (m->m_pkthdr.len < off) {
954 IP6STAT_INC(ip6s_tooshort);
955 in6_ifstat_inc(rcvif, ifs6_in_truncated);
959 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
960 if (IPSEC_ENABLED(ipv6)) {
961 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
966 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
970 in6_ifstat_inc(rcvif, ifs6_in_discard);
976 * Hop-by-Hop options header processing. If a valid jumbo payload option is
977 * included, the real payload length will be stored in plenp.
979 * rtalertp - XXX: should be stored more smart way
982 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
983 struct mbuf **mp, int *offp)
985 struct mbuf *m = *mp;
986 int off = *offp, hbhlen;
989 /* validation of the length of the header */
990 #ifndef PULLDOWN_TEST
991 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
992 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
993 hbhlen = (hbh->ip6h_len + 1) << 3;
995 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
996 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
998 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
999 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1001 IP6STAT_INC(ip6s_tooshort);
1004 hbhlen = (hbh->ip6h_len + 1) << 3;
1005 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
1008 IP6STAT_INC(ip6s_tooshort);
1013 hbhlen -= sizeof(struct ip6_hbh);
1014 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
1015 hbhlen, rtalertp, plenp) < 0)
1024 * Search header for all Hop-by-hop options and process each option.
1025 * This function is separate from ip6_hopopts_input() in order to
1026 * handle a case where the sending node itself process its hop-by-hop
1027 * options header. In such a case, the function is called from ip6_output().
1029 * The function assumes that hbh header is located right after the IPv6 header
1030 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1031 * opthead + hbhlen is located in contiguous memory region.
1034 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
1035 u_int32_t *rtalertp, u_int32_t *plenp)
1037 struct ip6_hdr *ip6;
1039 u_int8_t *opt = opthead;
1040 u_int16_t rtalert_val;
1041 u_int32_t jumboplen;
1042 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1044 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1050 if (hbhlen < IP6OPT_MINLEN) {
1051 IP6STAT_INC(ip6s_toosmall);
1054 optlen = *(opt + 1) + 2;
1056 case IP6OPT_ROUTER_ALERT:
1057 /* XXX may need check for alignment */
1058 if (hbhlen < IP6OPT_RTALERT_LEN) {
1059 IP6STAT_INC(ip6s_toosmall);
1062 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1064 icmp6_error(m, ICMP6_PARAM_PROB,
1065 ICMP6_PARAMPROB_HEADER,
1066 erroff + opt + 1 - opthead);
1069 optlen = IP6OPT_RTALERT_LEN;
1070 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1071 *rtalertp = ntohs(rtalert_val);
1074 /* XXX may need check for alignment */
1075 if (hbhlen < IP6OPT_JUMBO_LEN) {
1076 IP6STAT_INC(ip6s_toosmall);
1079 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1081 icmp6_error(m, ICMP6_PARAM_PROB,
1082 ICMP6_PARAMPROB_HEADER,
1083 erroff + opt + 1 - opthead);
1086 optlen = IP6OPT_JUMBO_LEN;
1089 * IPv6 packets that have non 0 payload length
1090 * must not contain a jumbo payload option.
1092 ip6 = mtod(m, struct ip6_hdr *);
1093 if (ip6->ip6_plen) {
1094 IP6STAT_INC(ip6s_badoptions);
1095 icmp6_error(m, ICMP6_PARAM_PROB,
1096 ICMP6_PARAMPROB_HEADER,
1097 erroff + opt - opthead);
1102 * We may see jumbolen in unaligned location, so
1103 * we'd need to perform bcopy().
1105 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1106 jumboplen = (u_int32_t)htonl(jumboplen);
1110 * if there are multiple jumbo payload options,
1111 * *plenp will be non-zero and the packet will be
1113 * the behavior may need some debate in ipngwg -
1114 * multiple options does not make sense, however,
1115 * there's no explicit mention in specification.
1118 IP6STAT_INC(ip6s_badoptions);
1119 icmp6_error(m, ICMP6_PARAM_PROB,
1120 ICMP6_PARAMPROB_HEADER,
1121 erroff + opt + 2 - opthead);
1127 * jumbo payload length must be larger than 65535.
1129 if (jumboplen <= IPV6_MAXPACKET) {
1130 IP6STAT_INC(ip6s_badoptions);
1131 icmp6_error(m, ICMP6_PARAM_PROB,
1132 ICMP6_PARAMPROB_HEADER,
1133 erroff + opt + 2 - opthead);
1139 default: /* unknown option */
1140 if (hbhlen < IP6OPT_MINLEN) {
1141 IP6STAT_INC(ip6s_toosmall);
1144 optlen = ip6_unknown_opt(opt, m,
1145 erroff + opt - opthead);
1161 * Unknown option processing.
1162 * The third argument `off' is the offset from the IPv6 header to the option,
1163 * which is necessary if the IPv6 header the and option header and IPv6 header
1164 * is not contiguous in order to return an ICMPv6 error.
1167 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1169 struct ip6_hdr *ip6;
1171 switch (IP6OPT_TYPE(*optp)) {
1172 case IP6OPT_TYPE_SKIP: /* ignore the option */
1173 return ((int)*(optp + 1));
1174 case IP6OPT_TYPE_DISCARD: /* silently discard */
1177 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1178 IP6STAT_INC(ip6s_badoptions);
1179 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1181 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1182 IP6STAT_INC(ip6s_badoptions);
1183 ip6 = mtod(m, struct ip6_hdr *);
1184 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1185 (m->m_flags & (M_BCAST|M_MCAST)))
1188 icmp6_error(m, ICMP6_PARAM_PROB,
1189 ICMP6_PARAMPROB_OPTION, off);
1193 m_freem(m); /* XXX: NOTREACHED */
1198 * Create the "control" list for this pcb.
1199 * These functions will not modify mbuf chain at all.
1201 * With KAME mbuf chain restriction:
1202 * The routine will be called from upper layer handlers like tcp6_input().
1203 * Thus the routine assumes that the caller (tcp6_input) have already
1204 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1205 * very first mbuf on the mbuf chain.
1207 * ip6_savecontrol_v4 will handle those options that are possible to be
1208 * set on a v4-mapped socket.
1209 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1210 * options and handle the v6-only ones itself.
1213 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1216 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1219 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1225 struct bintime boottimebin, bt1;
1226 struct timespec ts1;
1230 switch (inp->inp_socket->so_ts_clock) {
1231 case SO_TS_REALTIME_MICRO:
1232 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1234 mbuf_tstmp2timespec(m, &ts1);
1235 timespec2bintime(&ts1, &bt1);
1236 getboottimebin(&boottimebin);
1237 bintime_add(&bt1, &boottimebin);
1238 bintime2timeval(&bt1, &t.tv);
1242 *mp = sbcreatecontrol((caddr_t) &t.tv, sizeof(t.tv),
1243 SCM_TIMESTAMP, SOL_SOCKET);
1245 mp = &(*mp)->m_next;
1251 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1253 mbuf_tstmp2timespec(m, &ts1);
1254 timespec2bintime(&ts1, &t.bt);
1255 getboottimebin(&boottimebin);
1256 bintime_add(&t.bt, &boottimebin);
1260 *mp = sbcreatecontrol((caddr_t)&t.bt, sizeof(t.bt),
1261 SCM_BINTIME, SOL_SOCKET);
1263 mp = &(*mp)->m_next;
1268 case SO_TS_REALTIME:
1269 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1271 mbuf_tstmp2timespec(m, &t.ts);
1272 getboottimebin(&boottimebin);
1273 bintime2timespec(&boottimebin, &ts1);
1274 timespecadd(&t.ts, &ts1, &t.ts);
1278 *mp = sbcreatecontrol((caddr_t)&t.ts, sizeof(t.ts),
1279 SCM_REALTIME, SOL_SOCKET);
1281 mp = &(*mp)->m_next;
1286 case SO_TS_MONOTONIC:
1287 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1289 mbuf_tstmp2timespec(m, &t.ts);
1292 *mp = sbcreatecontrol((caddr_t)&t.ts, sizeof(t.ts),
1293 SCM_MONOTONIC, SOL_SOCKET);
1295 mp = &(*mp)->m_next;
1301 panic("unknown (corrupted) so_ts_clock");
1303 if (stamped && (m->m_flags & (M_PKTHDR | M_TSTMP)) ==
1304 (M_PKTHDR | M_TSTMP)) {
1305 struct sock_timestamp_info sti;
1307 bzero(&sti, sizeof(sti));
1308 sti.st_info_flags = ST_INFO_HW;
1309 if ((m->m_flags & M_TSTMP_HPREC) != 0)
1310 sti.st_info_flags |= ST_INFO_HW_HPREC;
1311 *mp = sbcreatecontrol((caddr_t)&sti, sizeof(sti),
1312 SCM_TIME_INFO, SOL_SOCKET);
1314 mp = &(*mp)->m_next;
1319 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1320 /* RFC 2292 sec. 5 */
1321 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1322 struct in6_pktinfo pi6;
1324 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1328 ip = mtod(m, struct ip *);
1329 pi6.ipi6_addr.s6_addr32[0] = 0;
1330 pi6.ipi6_addr.s6_addr32[1] = 0;
1331 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1332 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1334 /* We won't hit this code */
1335 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1338 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1339 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1342 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1344 *mp = sbcreatecontrol((caddr_t) &pi6,
1345 sizeof(struct in6_pktinfo),
1346 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1348 mp = &(*mp)->m_next;
1351 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1354 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1358 ip = mtod(m, struct ip *);
1361 /* We won't hit this code */
1365 hlim = ip6->ip6_hlim & 0xff;
1367 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1368 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1371 mp = &(*mp)->m_next;
1374 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1377 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1381 ip = mtod(m, struct ip *);
1382 tclass = ip->ip_tos;
1384 /* We won't hit this code */
1390 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1392 tclass = flowinfo & 0xff;
1394 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(int),
1395 IPV6_TCLASS, IPPROTO_IPV6);
1397 mp = &(*mp)->m_next;
1400 if (v4only != NULL) {
1401 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1412 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1414 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1417 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1422 * IPV6_HOPOPTS socket option. Recall that we required super-user
1423 * privilege for the option (see ip6_ctloutput), but it might be too
1424 * strict, since there might be some hop-by-hop options which can be
1425 * returned to normal user.
1426 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1428 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1430 * Check if a hop-by-hop options header is contatined in the
1431 * received packet, and if so, store the options as ancillary
1432 * data. Note that a hop-by-hop options header must be
1433 * just after the IPv6 header, which is assured through the
1434 * IPv6 input processing.
1436 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1437 struct ip6_hbh *hbh;
1439 #ifdef PULLDOWN_TEST
1443 #ifndef PULLDOWN_TEST
1444 hbh = (struct ip6_hbh *)(ip6 + 1);
1445 hbhlen = (hbh->ip6h_len + 1) << 3;
1447 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1450 IP6STAT_INC(ip6s_tooshort);
1453 hbh = mtod(ext, struct ip6_hbh *);
1454 hbhlen = (hbh->ip6h_len + 1) << 3;
1455 if (hbhlen != ext->m_len) {
1457 IP6STAT_INC(ip6s_tooshort);
1463 * XXX: We copy the whole header even if a
1464 * jumbo payload option is included, the option which
1465 * is to be removed before returning according to
1467 * Note: this constraint is removed in RFC3542
1469 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1470 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1473 mp = &(*mp)->m_next;
1474 #ifdef PULLDOWN_TEST
1480 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1481 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1484 * Search for destination options headers or routing
1485 * header(s) through the header chain, and stores each
1486 * header as ancillary data.
1487 * Note that the order of the headers remains in
1488 * the chain of ancillary data.
1490 while (1) { /* is explicit loop prevention necessary? */
1491 struct ip6_ext *ip6e = NULL;
1493 #ifdef PULLDOWN_TEST
1494 struct mbuf *ext = NULL;
1498 * if it is not an extension header, don't try to
1499 * pull it from the chain.
1502 case IPPROTO_DSTOPTS:
1503 case IPPROTO_ROUTING:
1504 case IPPROTO_HOPOPTS:
1505 case IPPROTO_AH: /* is it possible? */
1511 #ifndef PULLDOWN_TEST
1512 if (off + sizeof(*ip6e) > m->m_len)
1514 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1515 if (nxt == IPPROTO_AH)
1516 elen = (ip6e->ip6e_len + 2) << 2;
1518 elen = (ip6e->ip6e_len + 1) << 3;
1519 if (off + elen > m->m_len)
1522 ext = ip6_pullexthdr(m, off, nxt);
1524 IP6STAT_INC(ip6s_tooshort);
1527 ip6e = mtod(ext, struct ip6_ext *);
1528 if (nxt == IPPROTO_AH)
1529 elen = (ip6e->ip6e_len + 2) << 2;
1531 elen = (ip6e->ip6e_len + 1) << 3;
1532 if (elen != ext->m_len) {
1534 IP6STAT_INC(ip6s_tooshort);
1540 case IPPROTO_DSTOPTS:
1541 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1544 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1546 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1549 mp = &(*mp)->m_next;
1551 case IPPROTO_ROUTING:
1552 if (!(in6p->inp_flags & IN6P_RTHDR))
1555 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1556 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1559 mp = &(*mp)->m_next;
1561 case IPPROTO_HOPOPTS:
1562 case IPPROTO_AH: /* is it possible? */
1567 * other cases have been filtered in the above.
1568 * none will visit this case. here we supply
1569 * the code just in case (nxt overwritten or
1572 #ifdef PULLDOWN_TEST
1579 /* proceed with the next header. */
1581 nxt = ip6e->ip6e_nxt;
1583 #ifdef PULLDOWN_TEST
1592 if (in6p->inp_flags2 & INP_RECVFLOWID) {
1593 uint32_t flowid, flow_type;
1595 flowid = m->m_pkthdr.flowid;
1596 flow_type = M_HASHTYPE_GET(m);
1599 * XXX should handle the failure of one or the
1600 * other - don't populate both?
1602 *mp = sbcreatecontrol((caddr_t) &flowid,
1603 sizeof(uint32_t), IPV6_FLOWID, IPPROTO_IPV6);
1605 mp = &(*mp)->m_next;
1606 *mp = sbcreatecontrol((caddr_t) &flow_type,
1607 sizeof(uint32_t), IPV6_FLOWTYPE, IPPROTO_IPV6);
1609 mp = &(*mp)->m_next;
1613 if (in6p->inp_flags2 & INP_RECVRSSBUCKETID) {
1614 uint32_t flowid, flow_type;
1615 uint32_t rss_bucketid;
1617 flowid = m->m_pkthdr.flowid;
1618 flow_type = M_HASHTYPE_GET(m);
1620 if (rss_hash2bucket(flowid, flow_type, &rss_bucketid) == 0) {
1621 *mp = sbcreatecontrol((caddr_t) &rss_bucketid,
1622 sizeof(uint32_t), IPV6_RSSBUCKETID, IPPROTO_IPV6);
1624 mp = &(*mp)->m_next;
1633 ip6_notify_pmtu(struct inpcb *inp, struct sockaddr_in6 *dst, u_int32_t mtu)
1637 struct ip6_mtuinfo mtuctl;
1639 KASSERT(inp != NULL, ("%s: inp == NULL", __func__));
1641 * Notify the error by sending IPV6_PATHMTU ancillary data if
1642 * application wanted to know the MTU value.
1643 * NOTE: we notify disconnected sockets, because some udp
1644 * applications keep sending sockets disconnected.
1645 * NOTE: our implementation doesn't notify connected sockets that has
1646 * foreign address that is different than given destination addresses
1647 * (this is permitted by RFC 3542).
1649 if ((inp->inp_flags & IN6P_MTU) == 0 || (
1650 !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1651 !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &dst->sin6_addr)))
1654 mtuctl.ip6m_mtu = mtu;
1655 mtuctl.ip6m_addr = *dst;
1656 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1659 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1660 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1663 so = inp->inp_socket;
1664 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1667 /* XXX: should count statistics */
1672 #ifdef PULLDOWN_TEST
1674 * pull single extension header from mbuf chain. returns single mbuf that
1675 * contains the result, or NULL on error.
1677 static struct mbuf *
1678 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1680 struct ip6_ext ip6e;
1686 case IPPROTO_DSTOPTS:
1687 case IPPROTO_ROUTING:
1688 case IPPROTO_HOPOPTS:
1689 case IPPROTO_AH: /* is it possible? */
1692 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1696 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1697 if (nxt == IPPROTO_AH)
1698 elen = (ip6e.ip6e_len + 2) << 2;
1700 elen = (ip6e.ip6e_len + 1) << 3;
1703 n = m_getcl(M_NOWAIT, MT_DATA, 0);
1705 n = m_get(M_NOWAIT, MT_DATA);
1709 m_copydata(m, off, elen, mtod(n, caddr_t));
1716 * Get pointer to the previous header followed by the header
1717 * currently processed.
1720 ip6_get_prevhdr(const struct mbuf *m, int off)
1722 struct ip6_ext ip6e;
1723 struct ip6_hdr *ip6;
1726 if (off == sizeof(struct ip6_hdr))
1727 return (offsetof(struct ip6_hdr, ip6_nxt));
1728 if (off < sizeof(struct ip6_hdr))
1729 panic("%s: off < sizeof(struct ip6_hdr)", __func__);
1731 ip6 = mtod(m, struct ip6_hdr *);
1733 len = sizeof(struct ip6_hdr);
1736 m_copydata(m, len, sizeof(ip6e), (caddr_t)&ip6e);
1738 case IPPROTO_FRAGMENT:
1739 nlen = sizeof(struct ip6_frag);
1742 nlen = (ip6e.ip6e_len + 2) << 2;
1745 nlen = (ip6e.ip6e_len + 1) << 3;
1748 nxt = ip6e.ip6e_nxt;
1750 return (len - nlen);
1754 * get next header offset. m will be retained.
1757 ip6_nexthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1760 struct ip6_ext ip6e;
1765 panic("ip6_nexthdr: m == NULL");
1766 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1771 if (m->m_pkthdr.len < off + sizeof(ip6))
1773 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1775 *nxtp = ip6.ip6_nxt;
1779 case IPPROTO_FRAGMENT:
1781 * terminate parsing if it is not the first fragment,
1782 * it does not make sense to parse through it.
1784 if (m->m_pkthdr.len < off + sizeof(fh))
1786 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1787 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1788 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1791 *nxtp = fh.ip6f_nxt;
1792 off += sizeof(struct ip6_frag);
1796 if (m->m_pkthdr.len < off + sizeof(ip6e))
1798 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1800 *nxtp = ip6e.ip6e_nxt;
1801 off += (ip6e.ip6e_len + 2) << 2;
1804 case IPPROTO_HOPOPTS:
1805 case IPPROTO_ROUTING:
1806 case IPPROTO_DSTOPTS:
1807 if (m->m_pkthdr.len < off + sizeof(ip6e))
1809 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1811 *nxtp = ip6e.ip6e_nxt;
1812 off += (ip6e.ip6e_len + 1) << 3;
1817 case IPPROTO_IPCOMP:
1829 * get offset for the last header in the chain. m will be kept untainted.
1832 ip6_lasthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1842 newoff = ip6_nexthdr(m, off, proto, nxtp);
1845 else if (newoff < off)
1846 return -1; /* invalid */
1847 else if (newoff == off)
1856 * System control for IP6
1859 u_char inet6ctlerrmap[PRC_NCMDS] = {
1861 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1862 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1863 EMSGSIZE, EHOSTUNREACH, 0, 0,
1864 0, 0, EHOSTUNREACH, 0,
1865 ENOPROTOOPT, ECONNREFUSED