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(pfil_head_t, inet6_pfil_head);
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.
217 struct pfil_head_args args;
221 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal",
222 &V_ip6_auto_linklocal);
223 TUNABLE_INT_FETCH("net.inet6.ip6.accept_rtadv", &V_ip6_accept_rtadv);
224 TUNABLE_INT_FETCH("net.inet6.ip6.no_radr", &V_ip6_no_radr);
226 CK_STAILQ_INIT(&V_in6_ifaddrhead);
227 V_in6_ifaddrhashtbl = hashinit(IN6ADDR_NHASH, M_IFADDR,
230 /* Initialize packet filter hooks. */
231 args.pa_version = PFIL_VERSION;
232 args.pa_flags = PFIL_IN | PFIL_OUT;
233 args.pa_type = PFIL_TYPE_IP6;
234 args.pa_headname = PFIL_INET6_NAME;
235 V_inet6_pfil_head = pfil_head_register(&args);
237 if (hhook_head_register(HHOOK_TYPE_IPSEC_IN, AF_INET6,
238 &V_ipsec_hhh_in[HHOOK_IPSEC_INET6],
239 HHOOK_WAITOK | HHOOK_HEADISINVNET) != 0)
240 printf("%s: WARNING: unable to register input helper hook\n",
242 if (hhook_head_register(HHOOK_TYPE_IPSEC_OUT, AF_INET6,
243 &V_ipsec_hhh_out[HHOOK_IPSEC_INET6],
244 HHOOK_WAITOK | HHOOK_HEADISINVNET) != 0)
245 printf("%s: WARNING: unable to register output helper hook\n",
249 addrsel_policy_init();
253 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
255 /* Skip global initialization stuff for non-default instances. */
257 if (!IS_DEFAULT_VNET(curvnet)) {
258 netisr_register_vnet(&ip6_nh);
260 netisr_register_vnet(&ip6_direct_nh);
266 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
270 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
271 for (i = 0; i < IPPROTO_MAX; i++)
272 ip6_protox[i] = pr - inet6sw;
274 * Cycle through IP protocols and put them into the appropriate place
277 for (pr = inet6domain.dom_protosw;
278 pr < inet6domain.dom_protoswNPROTOSW; pr++)
279 if (pr->pr_domain->dom_family == PF_INET6 &&
280 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) {
281 /* Be careful to only index valid IP protocols. */
282 if (pr->pr_protocol < IPPROTO_MAX)
283 ip6_protox[pr->pr_protocol] = pr - inet6sw;
286 netisr_register(&ip6_nh);
288 netisr_register(&ip6_direct_nh);
293 * The protocol to be inserted into ip6_protox[] must be already registered
294 * in inet6sw[], either statically or through pf_proto_register().
297 ip6proto_register(short ip6proto)
302 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
303 return (EPROTONOSUPPORT);
306 * The protocol slot must not be occupied by another protocol
307 * already. An index pointing to IPPROTO_RAW is unused.
309 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
311 return (EPFNOSUPPORT);
312 if (ip6_protox[ip6proto] != pr - inet6sw) /* IPPROTO_RAW */
316 * Find the protocol position in inet6sw[] and set the index.
318 for (pr = inet6domain.dom_protosw;
319 pr < inet6domain.dom_protoswNPROTOSW; pr++) {
320 if (pr->pr_domain->dom_family == PF_INET6 &&
321 pr->pr_protocol && pr->pr_protocol == ip6proto) {
322 ip6_protox[pr->pr_protocol] = pr - inet6sw;
326 return (EPROTONOSUPPORT);
330 ip6proto_unregister(short ip6proto)
335 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX)
336 return (EPROTONOSUPPORT);
338 /* Check if the protocol was indeed registered. */
339 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
341 return (EPFNOSUPPORT);
342 if (ip6_protox[ip6proto] == pr - inet6sw) /* IPPROTO_RAW */
345 /* Reset the protocol slot to IPPROTO_RAW. */
346 ip6_protox[ip6proto] = pr - inet6sw;
352 ip6_destroy(void *unused __unused)
354 struct ifaddr *ifa, *nifa;
359 netisr_unregister_vnet(&ip6_direct_nh);
361 netisr_unregister_vnet(&ip6_nh);
363 pfil_head_unregister(V_inet6_pfil_head);
364 error = hhook_head_deregister(V_ipsec_hhh_in[HHOOK_IPSEC_INET6]);
366 printf("%s: WARNING: unable to deregister input helper hook "
367 "type HHOOK_TYPE_IPSEC_IN, id HHOOK_IPSEC_INET6: "
368 "error %d returned\n", __func__, error);
370 error = hhook_head_deregister(V_ipsec_hhh_out[HHOOK_IPSEC_INET6]);
372 printf("%s: WARNING: unable to deregister output helper hook "
373 "type HHOOK_TYPE_IPSEC_OUT, id HHOOK_IPSEC_INET6: "
374 "error %d returned\n", __func__, error);
377 /* Cleanup addresses. */
379 CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
380 /* Cannot lock here - lock recursion. */
381 /* IF_ADDR_LOCK(ifp); */
382 CK_STAILQ_FOREACH_SAFE(ifa, &ifp->if_addrhead, ifa_link, nifa) {
384 if (ifa->ifa_addr->sa_family != AF_INET6)
388 /* IF_ADDR_UNLOCK(ifp); */
389 in6_ifdetach_destroy(ifp);
390 mld_domifdetach(ifp);
391 /* Make sure any routes are gone as well. */
392 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_IN(V_inet6_pfil_head))
767 if (pfil_run_hooks(V_inet6_pfil_head, &m, m->m_pkthdr.rcvif, PFIL_IN,
770 ip6 = mtod(m, struct ip6_hdr *);
771 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
772 if ((m->m_flags & (M_IP6_NEXTHOP | M_FASTFWD_OURS)) == M_IP6_NEXTHOP &&
773 m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
775 * Directly ship the packet on. This allows forwarding
776 * packets originally destined to us to some other directly
785 * Disambiguate address scope zones (if there is ambiguity).
786 * We first make sure that the original source or destination address
787 * is not in our internal form for scoped addresses. Such addresses
788 * are not necessarily invalid spec-wise, but we cannot accept them due
789 * to the usage conflict.
790 * in6_setscope() then also checks and rejects the cases where src or
791 * dst are the loopback address and the receiving interface
794 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
795 IP6STAT_INC(ip6s_badscope); /* XXX */
798 if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
799 in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
800 IP6STAT_INC(ip6s_badscope);
803 if (m->m_flags & M_FASTFWD_OURS) {
804 m->m_flags &= ~M_FASTFWD_OURS;
809 * Multicast check. Assume packet is for us to avoid
810 * prematurely taking locks.
812 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
814 in6_ifstat_inc(rcvif, ifs6_in_mcast);
819 * XXX: For now we keep link-local IPv6 addresses with embedded
820 * scope zone id, therefore we use zero zoneid here.
822 ia = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */);
824 if (ia->ia6_flags & IN6_IFF_NOTREADY) {
825 char ip6bufs[INET6_ADDRSTRLEN];
826 char ip6bufd[INET6_ADDRSTRLEN];
827 /* address is not ready, so discard the packet. */
829 "ip6_input: packet to an unready address %s->%s\n",
830 ip6_sprintf(ip6bufs, &ip6->ip6_src),
831 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
832 ifa_free(&ia->ia_ifa);
835 /* Count the packet in the ip address stats */
836 counter_u64_add(ia->ia_ifa.ifa_ipackets, 1);
837 counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len);
838 ifa_free(&ia->ia_ifa);
844 * Now there is no reason to process the packet if it's not our own
845 * and we're not a router.
847 if (!V_ip6_forwarding) {
848 IP6STAT_INC(ip6s_cantforward);
854 * Process Hop-by-Hop options header if it's contained.
855 * m may be modified in ip6_hopopts_input().
856 * If a JumboPayload option is included, plen will also be modified.
858 plen = (u_int32_t)ntohs(ip6->ip6_plen);
859 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
860 if (ip6_input_hbh(&m, &plen, &rtalert, &off, &nxt, &ours) != 0)
866 * Use mbuf flags to propagate Router Alert option to
867 * ICMPv6 layer, as hop-by-hop options have been stripped.
870 m->m_flags |= M_RTALERT_MLD;
873 * Check that the amount of data in the buffers
874 * is as at least much as the IPv6 header would have us expect.
875 * Trim mbufs if longer than we expect.
876 * Drop packet if shorter than we expect.
878 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
879 IP6STAT_INC(ip6s_tooshort);
880 in6_ifstat_inc(rcvif, ifs6_in_truncated);
883 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
884 if (m->m_len == m->m_pkthdr.len) {
885 m->m_len = sizeof(struct ip6_hdr) + plen;
886 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
888 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
892 * Forward if desirable.
895 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
897 * If we are acting as a multicast router, all
898 * incoming multicast packets are passed to the
899 * kernel-level multicast forwarding function.
900 * The packet is returned (relatively) intact; if
901 * ip6_mforward() returns a non-zero value, the packet
902 * must be discarded, else it may be accepted below.
904 * XXX TODO: Check hlim and multicast scope here to avoid
905 * unnecessarily calling into ip6_mforward().
907 if (ip6_mforward && ip6_mforward(ip6, rcvif, m)) {
908 IP6STAT_INC(ip6s_cantforward);
912 ip6_forward(m, srcrt);
916 ip6 = mtod(m, struct ip6_hdr *);
919 * Malicious party may be able to use IPv4 mapped addr to confuse
920 * tcp/udp stack and bypass security checks (act as if it was from
921 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
923 * For SIIT end node behavior, you may want to disable the check.
924 * However, you will become vulnerable to attacks using IPv4 mapped
927 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
928 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
929 IP6STAT_INC(ip6s_badscope);
930 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
935 * Tell launch routine the next header
937 IP6STAT_INC(ip6s_delivered);
938 in6_ifstat_inc(rcvif, ifs6_in_deliver);
941 while (nxt != IPPROTO_DONE) {
942 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
943 IP6STAT_INC(ip6s_toomanyhdr);
948 * protection against faulty packet - there should be
949 * more sanity checks in header chain processing.
951 if (m->m_pkthdr.len < off) {
952 IP6STAT_INC(ip6s_tooshort);
953 in6_ifstat_inc(rcvif, ifs6_in_truncated);
957 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
958 if (IPSEC_ENABLED(ipv6)) {
959 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
964 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
968 in6_ifstat_inc(rcvif, ifs6_in_discard);
974 * Hop-by-Hop options header processing. If a valid jumbo payload option is
975 * included, the real payload length will be stored in plenp.
977 * rtalertp - XXX: should be stored more smart way
980 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
981 struct mbuf **mp, int *offp)
983 struct mbuf *m = *mp;
984 int off = *offp, hbhlen;
987 /* validation of the length of the header */
988 #ifndef PULLDOWN_TEST
989 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
990 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
991 hbhlen = (hbh->ip6h_len + 1) << 3;
993 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
994 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
996 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
997 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
999 IP6STAT_INC(ip6s_tooshort);
1002 hbhlen = (hbh->ip6h_len + 1) << 3;
1003 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
1006 IP6STAT_INC(ip6s_tooshort);
1011 hbhlen -= sizeof(struct ip6_hbh);
1012 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
1013 hbhlen, rtalertp, plenp) < 0)
1022 * Search header for all Hop-by-hop options and process each option.
1023 * This function is separate from ip6_hopopts_input() in order to
1024 * handle a case where the sending node itself process its hop-by-hop
1025 * options header. In such a case, the function is called from ip6_output().
1027 * The function assumes that hbh header is located right after the IPv6 header
1028 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1029 * opthead + hbhlen is located in contiguous memory region.
1032 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
1033 u_int32_t *rtalertp, u_int32_t *plenp)
1035 struct ip6_hdr *ip6;
1037 u_int8_t *opt = opthead;
1038 u_int16_t rtalert_val;
1039 u_int32_t jumboplen;
1040 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1042 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1048 if (hbhlen < IP6OPT_MINLEN) {
1049 IP6STAT_INC(ip6s_toosmall);
1052 optlen = *(opt + 1) + 2;
1054 case IP6OPT_ROUTER_ALERT:
1055 /* XXX may need check for alignment */
1056 if (hbhlen < IP6OPT_RTALERT_LEN) {
1057 IP6STAT_INC(ip6s_toosmall);
1060 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1062 icmp6_error(m, ICMP6_PARAM_PROB,
1063 ICMP6_PARAMPROB_HEADER,
1064 erroff + opt + 1 - opthead);
1067 optlen = IP6OPT_RTALERT_LEN;
1068 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1069 *rtalertp = ntohs(rtalert_val);
1072 /* XXX may need check for alignment */
1073 if (hbhlen < IP6OPT_JUMBO_LEN) {
1074 IP6STAT_INC(ip6s_toosmall);
1077 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1079 icmp6_error(m, ICMP6_PARAM_PROB,
1080 ICMP6_PARAMPROB_HEADER,
1081 erroff + opt + 1 - opthead);
1084 optlen = IP6OPT_JUMBO_LEN;
1087 * IPv6 packets that have non 0 payload length
1088 * must not contain a jumbo payload option.
1090 ip6 = mtod(m, struct ip6_hdr *);
1091 if (ip6->ip6_plen) {
1092 IP6STAT_INC(ip6s_badoptions);
1093 icmp6_error(m, ICMP6_PARAM_PROB,
1094 ICMP6_PARAMPROB_HEADER,
1095 erroff + opt - opthead);
1100 * We may see jumbolen in unaligned location, so
1101 * we'd need to perform bcopy().
1103 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1104 jumboplen = (u_int32_t)htonl(jumboplen);
1108 * if there are multiple jumbo payload options,
1109 * *plenp will be non-zero and the packet will be
1111 * the behavior may need some debate in ipngwg -
1112 * multiple options does not make sense, however,
1113 * there's no explicit mention in specification.
1116 IP6STAT_INC(ip6s_badoptions);
1117 icmp6_error(m, ICMP6_PARAM_PROB,
1118 ICMP6_PARAMPROB_HEADER,
1119 erroff + opt + 2 - opthead);
1125 * jumbo payload length must be larger than 65535.
1127 if (jumboplen <= IPV6_MAXPACKET) {
1128 IP6STAT_INC(ip6s_badoptions);
1129 icmp6_error(m, ICMP6_PARAM_PROB,
1130 ICMP6_PARAMPROB_HEADER,
1131 erroff + opt + 2 - opthead);
1137 default: /* unknown option */
1138 if (hbhlen < IP6OPT_MINLEN) {
1139 IP6STAT_INC(ip6s_toosmall);
1142 optlen = ip6_unknown_opt(opt, m,
1143 erroff + opt - opthead);
1159 * Unknown option processing.
1160 * The third argument `off' is the offset from the IPv6 header to the option,
1161 * which is necessary if the IPv6 header the and option header and IPv6 header
1162 * is not contiguous in order to return an ICMPv6 error.
1165 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1167 struct ip6_hdr *ip6;
1169 switch (IP6OPT_TYPE(*optp)) {
1170 case IP6OPT_TYPE_SKIP: /* ignore the option */
1171 return ((int)*(optp + 1));
1172 case IP6OPT_TYPE_DISCARD: /* silently discard */
1175 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1176 IP6STAT_INC(ip6s_badoptions);
1177 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1179 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1180 IP6STAT_INC(ip6s_badoptions);
1181 ip6 = mtod(m, struct ip6_hdr *);
1182 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1183 (m->m_flags & (M_BCAST|M_MCAST)))
1186 icmp6_error(m, ICMP6_PARAM_PROB,
1187 ICMP6_PARAMPROB_OPTION, off);
1191 m_freem(m); /* XXX: NOTREACHED */
1196 * Create the "control" list for this pcb.
1197 * These functions will not modify mbuf chain at all.
1199 * With KAME mbuf chain restriction:
1200 * The routine will be called from upper layer handlers like tcp6_input().
1201 * Thus the routine assumes that the caller (tcp6_input) have already
1202 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1203 * very first mbuf on the mbuf chain.
1205 * ip6_savecontrol_v4 will handle those options that are possible to be
1206 * set on a v4-mapped socket.
1207 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1208 * options and handle the v6-only ones itself.
1211 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1214 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1217 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1223 struct bintime boottimebin, bt1;
1224 struct timespec ts1;
1228 switch (inp->inp_socket->so_ts_clock) {
1229 case SO_TS_REALTIME_MICRO:
1230 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1232 mbuf_tstmp2timespec(m, &ts1);
1233 timespec2bintime(&ts1, &bt1);
1234 getboottimebin(&boottimebin);
1235 bintime_add(&bt1, &boottimebin);
1236 bintime2timeval(&bt1, &t.tv);
1240 *mp = sbcreatecontrol((caddr_t) &t.tv, sizeof(t.tv),
1241 SCM_TIMESTAMP, SOL_SOCKET);
1243 mp = &(*mp)->m_next;
1249 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1251 mbuf_tstmp2timespec(m, &ts1);
1252 timespec2bintime(&ts1, &t.bt);
1253 getboottimebin(&boottimebin);
1254 bintime_add(&t.bt, &boottimebin);
1258 *mp = sbcreatecontrol((caddr_t)&t.bt, sizeof(t.bt),
1259 SCM_BINTIME, SOL_SOCKET);
1261 mp = &(*mp)->m_next;
1266 case SO_TS_REALTIME:
1267 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1269 mbuf_tstmp2timespec(m, &t.ts);
1270 getboottimebin(&boottimebin);
1271 bintime2timespec(&boottimebin, &ts1);
1272 timespecadd(&t.ts, &ts1, &t.ts);
1276 *mp = sbcreatecontrol((caddr_t)&t.ts, sizeof(t.ts),
1277 SCM_REALTIME, SOL_SOCKET);
1279 mp = &(*mp)->m_next;
1284 case SO_TS_MONOTONIC:
1285 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1287 mbuf_tstmp2timespec(m, &t.ts);
1290 *mp = sbcreatecontrol((caddr_t)&t.ts, sizeof(t.ts),
1291 SCM_MONOTONIC, SOL_SOCKET);
1293 mp = &(*mp)->m_next;
1299 panic("unknown (corrupted) so_ts_clock");
1301 if (stamped && (m->m_flags & (M_PKTHDR | M_TSTMP)) ==
1302 (M_PKTHDR | M_TSTMP)) {
1303 struct sock_timestamp_info sti;
1305 bzero(&sti, sizeof(sti));
1306 sti.st_info_flags = ST_INFO_HW;
1307 if ((m->m_flags & M_TSTMP_HPREC) != 0)
1308 sti.st_info_flags |= ST_INFO_HW_HPREC;
1309 *mp = sbcreatecontrol((caddr_t)&sti, sizeof(sti),
1310 SCM_TIME_INFO, SOL_SOCKET);
1312 mp = &(*mp)->m_next;
1317 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1318 /* RFC 2292 sec. 5 */
1319 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1320 struct in6_pktinfo pi6;
1322 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1326 ip = mtod(m, struct ip *);
1327 pi6.ipi6_addr.s6_addr32[0] = 0;
1328 pi6.ipi6_addr.s6_addr32[1] = 0;
1329 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1330 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1332 /* We won't hit this code */
1333 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1336 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1337 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1340 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1342 *mp = sbcreatecontrol((caddr_t) &pi6,
1343 sizeof(struct in6_pktinfo),
1344 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1346 mp = &(*mp)->m_next;
1349 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1352 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1356 ip = mtod(m, struct ip *);
1359 /* We won't hit this code */
1363 hlim = ip6->ip6_hlim & 0xff;
1365 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1366 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1369 mp = &(*mp)->m_next;
1372 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1375 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1379 ip = mtod(m, struct ip *);
1380 tclass = ip->ip_tos;
1382 /* We won't hit this code */
1388 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1390 tclass = flowinfo & 0xff;
1392 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(int),
1393 IPV6_TCLASS, IPPROTO_IPV6);
1395 mp = &(*mp)->m_next;
1398 if (v4only != NULL) {
1399 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1410 ip6_savecontrol(struct inpcb *inp, struct mbuf *m, struct mbuf **mp)
1412 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1415 mp = ip6_savecontrol_v4(inp, m, mp, &v4only);
1420 * IPV6_HOPOPTS socket option. Recall that we required super-user
1421 * privilege for the option (see ip6_ctloutput), but it might be too
1422 * strict, since there might be some hop-by-hop options which can be
1423 * returned to normal user.
1424 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1426 if ((inp->inp_flags & IN6P_HOPOPTS) != 0) {
1428 * Check if a hop-by-hop options header is contatined in the
1429 * received packet, and if so, store the options as ancillary
1430 * data. Note that a hop-by-hop options header must be
1431 * just after the IPv6 header, which is assured through the
1432 * IPv6 input processing.
1434 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1435 struct ip6_hbh *hbh;
1437 #ifdef PULLDOWN_TEST
1441 #ifndef PULLDOWN_TEST
1442 hbh = (struct ip6_hbh *)(ip6 + 1);
1443 hbhlen = (hbh->ip6h_len + 1) << 3;
1445 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1448 IP6STAT_INC(ip6s_tooshort);
1451 hbh = mtod(ext, struct ip6_hbh *);
1452 hbhlen = (hbh->ip6h_len + 1) << 3;
1453 if (hbhlen != ext->m_len) {
1455 IP6STAT_INC(ip6s_tooshort);
1461 * XXX: We copy the whole header even if a
1462 * jumbo payload option is included, the option which
1463 * is to be removed before returning according to
1465 * Note: this constraint is removed in RFC3542
1467 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1468 IS2292(inp, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1471 mp = &(*mp)->m_next;
1472 #ifdef PULLDOWN_TEST
1478 if ((inp->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1479 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1482 * Search for destination options headers or routing
1483 * header(s) through the header chain, and stores each
1484 * header as ancillary data.
1485 * Note that the order of the headers remains in
1486 * the chain of ancillary data.
1488 while (1) { /* is explicit loop prevention necessary? */
1489 struct ip6_ext *ip6e = NULL;
1491 #ifdef PULLDOWN_TEST
1492 struct mbuf *ext = NULL;
1496 * if it is not an extension header, don't try to
1497 * pull it from the chain.
1500 case IPPROTO_DSTOPTS:
1501 case IPPROTO_ROUTING:
1502 case IPPROTO_HOPOPTS:
1503 case IPPROTO_AH: /* is it possible? */
1509 #ifndef PULLDOWN_TEST
1510 if (off + sizeof(*ip6e) > m->m_len)
1512 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1513 if (nxt == IPPROTO_AH)
1514 elen = (ip6e->ip6e_len + 2) << 2;
1516 elen = (ip6e->ip6e_len + 1) << 3;
1517 if (off + elen > m->m_len)
1520 ext = ip6_pullexthdr(m, off, nxt);
1522 IP6STAT_INC(ip6s_tooshort);
1525 ip6e = mtod(ext, struct ip6_ext *);
1526 if (nxt == IPPROTO_AH)
1527 elen = (ip6e->ip6e_len + 2) << 2;
1529 elen = (ip6e->ip6e_len + 1) << 3;
1530 if (elen != ext->m_len) {
1532 IP6STAT_INC(ip6s_tooshort);
1538 case IPPROTO_DSTOPTS:
1539 if (!(inp->inp_flags & IN6P_DSTOPTS))
1542 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1544 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1547 mp = &(*mp)->m_next;
1549 case IPPROTO_ROUTING:
1550 if (!(inp->inp_flags & IN6P_RTHDR))
1553 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1554 IS2292(inp, IPV6_2292RTHDR, IPV6_RTHDR),
1557 mp = &(*mp)->m_next;
1559 case IPPROTO_HOPOPTS:
1560 case IPPROTO_AH: /* is it possible? */
1565 * other cases have been filtered in the above.
1566 * none will visit this case. here we supply
1567 * the code just in case (nxt overwritten or
1570 #ifdef PULLDOWN_TEST
1577 /* proceed with the next header. */
1579 nxt = ip6e->ip6e_nxt;
1581 #ifdef PULLDOWN_TEST
1590 if (inp->inp_flags2 & INP_RECVFLOWID) {
1591 uint32_t flowid, flow_type;
1593 flowid = m->m_pkthdr.flowid;
1594 flow_type = M_HASHTYPE_GET(m);
1597 * XXX should handle the failure of one or the
1598 * other - don't populate both?
1600 *mp = sbcreatecontrol((caddr_t) &flowid,
1601 sizeof(uint32_t), IPV6_FLOWID, IPPROTO_IPV6);
1603 mp = &(*mp)->m_next;
1604 *mp = sbcreatecontrol((caddr_t) &flow_type,
1605 sizeof(uint32_t), IPV6_FLOWTYPE, IPPROTO_IPV6);
1607 mp = &(*mp)->m_next;
1611 if (inp->inp_flags2 & INP_RECVRSSBUCKETID) {
1612 uint32_t flowid, flow_type;
1613 uint32_t rss_bucketid;
1615 flowid = m->m_pkthdr.flowid;
1616 flow_type = M_HASHTYPE_GET(m);
1618 if (rss_hash2bucket(flowid, flow_type, &rss_bucketid) == 0) {
1619 *mp = sbcreatecontrol((caddr_t) &rss_bucketid,
1620 sizeof(uint32_t), IPV6_RSSBUCKETID, IPPROTO_IPV6);
1622 mp = &(*mp)->m_next;
1631 ip6_notify_pmtu(struct inpcb *inp, struct sockaddr_in6 *dst, u_int32_t mtu)
1635 struct ip6_mtuinfo mtuctl;
1637 KASSERT(inp != NULL, ("%s: inp == NULL", __func__));
1639 * Notify the error by sending IPV6_PATHMTU ancillary data if
1640 * application wanted to know the MTU value.
1641 * NOTE: we notify disconnected sockets, because some udp
1642 * applications keep sending sockets disconnected.
1643 * NOTE: our implementation doesn't notify connected sockets that has
1644 * foreign address that is different than given destination addresses
1645 * (this is permitted by RFC 3542).
1647 if ((inp->inp_flags & IN6P_MTU) == 0 || (
1648 !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1649 !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &dst->sin6_addr)))
1652 mtuctl.ip6m_mtu = mtu;
1653 mtuctl.ip6m_addr = *dst;
1654 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1657 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1658 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1661 so = inp->inp_socket;
1662 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1665 /* XXX: should count statistics */
1670 #ifdef PULLDOWN_TEST
1672 * pull single extension header from mbuf chain. returns single mbuf that
1673 * contains the result, or NULL on error.
1675 static struct mbuf *
1676 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1678 struct ip6_ext ip6e;
1684 case IPPROTO_DSTOPTS:
1685 case IPPROTO_ROUTING:
1686 case IPPROTO_HOPOPTS:
1687 case IPPROTO_AH: /* is it possible? */
1690 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1694 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1695 if (nxt == IPPROTO_AH)
1696 elen = (ip6e.ip6e_len + 2) << 2;
1698 elen = (ip6e.ip6e_len + 1) << 3;
1701 n = m_getcl(M_NOWAIT, MT_DATA, 0);
1703 n = m_get(M_NOWAIT, MT_DATA);
1707 m_copydata(m, off, elen, mtod(n, caddr_t));
1714 * Get pointer to the previous header followed by the header
1715 * currently processed.
1718 ip6_get_prevhdr(const struct mbuf *m, int off)
1720 struct ip6_ext ip6e;
1721 struct ip6_hdr *ip6;
1724 if (off == sizeof(struct ip6_hdr))
1725 return (offsetof(struct ip6_hdr, ip6_nxt));
1726 if (off < sizeof(struct ip6_hdr))
1727 panic("%s: off < sizeof(struct ip6_hdr)", __func__);
1729 ip6 = mtod(m, struct ip6_hdr *);
1731 len = sizeof(struct ip6_hdr);
1734 m_copydata(m, len, sizeof(ip6e), (caddr_t)&ip6e);
1736 case IPPROTO_FRAGMENT:
1737 nlen = sizeof(struct ip6_frag);
1740 nlen = (ip6e.ip6e_len + 2) << 2;
1743 nlen = (ip6e.ip6e_len + 1) << 3;
1746 nxt = ip6e.ip6e_nxt;
1748 return (len - nlen);
1752 * get next header offset. m will be retained.
1755 ip6_nexthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1758 struct ip6_ext ip6e;
1763 panic("ip6_nexthdr: m == NULL");
1764 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1769 if (m->m_pkthdr.len < off + sizeof(ip6))
1771 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1773 *nxtp = ip6.ip6_nxt;
1777 case IPPROTO_FRAGMENT:
1779 * terminate parsing if it is not the first fragment,
1780 * it does not make sense to parse through it.
1782 if (m->m_pkthdr.len < off + sizeof(fh))
1784 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1785 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1786 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1789 *nxtp = fh.ip6f_nxt;
1790 off += sizeof(struct ip6_frag);
1794 if (m->m_pkthdr.len < off + sizeof(ip6e))
1796 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1798 *nxtp = ip6e.ip6e_nxt;
1799 off += (ip6e.ip6e_len + 2) << 2;
1802 case IPPROTO_HOPOPTS:
1803 case IPPROTO_ROUTING:
1804 case IPPROTO_DSTOPTS:
1805 if (m->m_pkthdr.len < off + sizeof(ip6e))
1807 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1809 *nxtp = ip6e.ip6e_nxt;
1810 off += (ip6e.ip6e_len + 1) << 3;
1815 case IPPROTO_IPCOMP:
1827 * get offset for the last header in the chain. m will be kept untainted.
1830 ip6_lasthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1840 newoff = ip6_nexthdr(m, off, proto, nxtp);
1843 else if (newoff < off)
1844 return -1; /* invalid */
1845 else if (newoff == off)
1854 * System control for IP6
1857 u_char inet6ctlerrmap[PRC_NCMDS] = {
1859 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1860 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1861 EMSGSIZE, EHOSTUNREACH, 0, 0,
1862 0, 0, EHOSTUNREACH, 0,
1863 ENOPROTOOPT, ECONNREFUSED