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: icmp6.c,v 1.211 2001/04/04 05:56:20 itojun 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_icmp.c 8.2 (Berkeley) 1/4/94
65 #include <sys/cdefs.h>
66 __FBSDID("$FreeBSD$");
68 #define MBUF_PRIVATE /* XXXRW: Optimisation tries to avoid M_EXT mbufs */
71 #include "opt_inet6.h"
73 #include <sys/param.h>
74 #include <sys/domain.h>
76 #include <sys/kernel.h>
78 #include <sys/malloc.h>
81 #include <sys/protosw.h>
82 #include <sys/signalvar.h>
83 #include <sys/socket.h>
84 #include <sys/socketvar.h>
86 #include <sys/syslog.h>
87 #include <sys/systm.h>
91 #include <net/if_var.h>
92 #include <net/if_dl.h>
93 #include <net/if_llatbl.h>
94 #include <net/if_types.h>
95 #include <net/route.h>
96 #include <net/route/route_ctl.h>
97 #include <net/route/nhop.h>
100 #include <netinet/in.h>
101 #include <netinet/in_pcb.h>
102 #include <netinet/in_var.h>
103 #include <netinet/ip6.h>
104 #include <netinet/icmp6.h>
105 #include <netinet/tcp_var.h>
107 #include <netinet6/in6_fib.h>
108 #include <netinet6/in6_ifattach.h>
109 #include <netinet6/in6_pcb.h>
110 #include <netinet6/ip6protosw.h>
111 #include <netinet6/ip6_var.h>
112 #include <netinet6/scope6_var.h>
113 #include <netinet6/mld6_var.h>
114 #include <netinet6/nd6.h>
115 #include <netinet6/send.h>
117 extern struct domain inet6domain;
119 VNET_PCPUSTAT_DEFINE(struct icmp6stat, icmp6stat);
120 VNET_PCPUSTAT_SYSINIT(icmp6stat);
123 VNET_PCPUSTAT_SYSUNINIT(icmp6stat);
126 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
127 VNET_DECLARE(struct inpcbhead, ripcb);
128 VNET_DECLARE(int, icmp6errppslim);
129 VNET_DEFINE_STATIC(int, icmp6errpps_count) = 0;
130 VNET_DEFINE_STATIC(struct timeval, icmp6errppslim_last);
131 VNET_DECLARE(int, icmp6_nodeinfo);
133 #define V_ripcbinfo VNET(ripcbinfo)
134 #define V_ripcb VNET(ripcb)
135 #define V_icmp6errppslim VNET(icmp6errppslim)
136 #define V_icmp6errpps_count VNET(icmp6errpps_count)
137 #define V_icmp6errppslim_last VNET(icmp6errppslim_last)
138 #define V_icmp6_nodeinfo VNET(icmp6_nodeinfo)
140 static void icmp6_errcount(int, int);
141 static int icmp6_rip6_input(struct mbuf **, int);
142 static void icmp6_reflect(struct mbuf *, size_t);
143 static int icmp6_ratelimit(const struct in6_addr *, const int, const int);
144 static const char *icmp6_redirect_diag(struct in6_addr *,
145 struct in6_addr *, struct in6_addr *);
146 static struct mbuf *ni6_input(struct mbuf *, int, struct prison *);
147 static struct mbuf *ni6_nametodns(const char *, int, int);
148 static int ni6_dnsmatch(const char *, int, const char *, int);
149 static int ni6_addrs(struct icmp6_nodeinfo *, struct mbuf *,
150 struct ifnet **, struct in6_addr *);
151 static int ni6_store_addrs(struct icmp6_nodeinfo *, struct icmp6_nodeinfo *,
152 struct ifnet *, int);
153 static int icmp6_notify_error(struct mbuf **, int, int, int);
156 * Kernel module interface for updating icmp6stat. The argument is an index
157 * into icmp6stat treated as an array of u_quad_t. While this encodes the
158 * general layout of icmp6stat into the caller, it doesn't encode its
159 * location, so that future changes to add, for example, per-CPU stats
160 * support won't cause binary compatibility problems for kernel modules.
163 kmod_icmp6stat_inc(int statnum)
166 counter_u64_add(VNET(icmp6stat)[statnum], 1);
170 icmp6_errcount(int type, int code)
173 case ICMP6_DST_UNREACH:
175 case ICMP6_DST_UNREACH_NOROUTE:
176 ICMP6STAT_INC(icp6s_odst_unreach_noroute);
178 case ICMP6_DST_UNREACH_ADMIN:
179 ICMP6STAT_INC(icp6s_odst_unreach_admin);
181 case ICMP6_DST_UNREACH_BEYONDSCOPE:
182 ICMP6STAT_INC(icp6s_odst_unreach_beyondscope);
184 case ICMP6_DST_UNREACH_ADDR:
185 ICMP6STAT_INC(icp6s_odst_unreach_addr);
187 case ICMP6_DST_UNREACH_NOPORT:
188 ICMP6STAT_INC(icp6s_odst_unreach_noport);
192 case ICMP6_PACKET_TOO_BIG:
193 ICMP6STAT_INC(icp6s_opacket_too_big);
195 case ICMP6_TIME_EXCEEDED:
197 case ICMP6_TIME_EXCEED_TRANSIT:
198 ICMP6STAT_INC(icp6s_otime_exceed_transit);
200 case ICMP6_TIME_EXCEED_REASSEMBLY:
201 ICMP6STAT_INC(icp6s_otime_exceed_reassembly);
205 case ICMP6_PARAM_PROB:
207 case ICMP6_PARAMPROB_HEADER:
208 ICMP6STAT_INC(icp6s_oparamprob_header);
210 case ICMP6_PARAMPROB_NEXTHEADER:
211 ICMP6STAT_INC(icp6s_oparamprob_nextheader);
213 case ICMP6_PARAMPROB_OPTION:
214 ICMP6STAT_INC(icp6s_oparamprob_option);
219 ICMP6STAT_INC(icp6s_oredirect);
222 ICMP6STAT_INC(icp6s_ounknown);
226 * A wrapper function for icmp6_error() necessary when the erroneous packet
227 * may not contain enough scope zone information.
230 icmp6_error2(struct mbuf *m, int type, int code, int param,
238 if (m->m_len < sizeof(struct ip6_hdr)) {
239 m = m_pullup(m, sizeof(struct ip6_hdr));
241 IP6STAT_INC(ip6s_exthdrtoolong);
245 ip6 = mtod(m, struct ip6_hdr *);
247 if (in6_setscope(&ip6->ip6_src, ifp, NULL) != 0)
249 if (in6_setscope(&ip6->ip6_dst, ifp, NULL) != 0)
252 icmp6_error(m, type, code, param);
256 * Generate an error packet of type error in response to bad IP6 packet.
259 icmp6_error(struct mbuf *m, int type, int code, int param)
261 struct ip6_hdr *oip6, *nip6;
262 struct icmp6_hdr *icmp6;
263 struct epoch_tracker et;
268 ICMP6STAT_INC(icp6s_error);
270 /* count per-type-code statistics */
271 icmp6_errcount(type, code);
273 #ifdef M_DECRYPTED /*not openbsd*/
274 if (m->m_flags & M_DECRYPTED) {
275 ICMP6STAT_INC(icp6s_canterror);
280 if (m->m_len < sizeof(struct ip6_hdr)) {
281 m = m_pullup(m, sizeof(struct ip6_hdr));
283 IP6STAT_INC(ip6s_exthdrtoolong);
287 oip6 = mtod(m, struct ip6_hdr *);
290 * If the destination address of the erroneous packet is a multicast
291 * address, or the packet was sent using link-layer multicast,
292 * we should basically suppress sending an error (RFC 2463, Section
294 * We have two exceptions (the item e.2 in that section):
295 * - the Packet Too Big message can be sent for path MTU discovery.
296 * - the Parameter Problem Message that can be allowed an icmp6 error
297 * in the option type field. This check has been done in
298 * ip6_unknown_opt(), so we can just check the type and code.
300 if ((m->m_flags & (M_BCAST|M_MCAST) ||
301 IN6_IS_ADDR_MULTICAST(&oip6->ip6_dst)) &&
302 (type != ICMP6_PACKET_TOO_BIG &&
303 (type != ICMP6_PARAM_PROB ||
304 code != ICMP6_PARAMPROB_OPTION)))
308 * RFC 2463, 2.4 (e.5): source address check.
309 * XXX: the case of anycast source?
311 if (IN6_IS_ADDR_UNSPECIFIED(&oip6->ip6_src) ||
312 IN6_IS_ADDR_MULTICAST(&oip6->ip6_src))
316 * If we are about to send ICMPv6 against ICMPv6 error/redirect,
320 off = ip6_lasthdr(m, 0, IPPROTO_IPV6, &nxt);
321 if (off >= 0 && nxt == IPPROTO_ICMPV6) {
322 struct icmp6_hdr *icp;
324 if (m->m_len < off + sizeof(struct icmp6_hdr)) {
325 m = m_pullup(m, off + sizeof(struct icmp6_hdr));
327 IP6STAT_INC(ip6s_exthdrtoolong);
331 oip6 = mtod(m, struct ip6_hdr *);
332 icp = (struct icmp6_hdr *)(mtod(m, caddr_t) + off);
334 if (icp->icmp6_type < ICMP6_ECHO_REQUEST ||
335 icp->icmp6_type == ND_REDIRECT) {
338 * Special case: for redirect (which is
339 * informational) we must not send icmp6 error.
341 ICMP6STAT_INC(icp6s_canterror);
344 /* ICMPv6 informational - send the error */
347 /* non-ICMPv6 - send the error */
350 /* Finally, do rate limitation check. */
351 if (icmp6_ratelimit(&oip6->ip6_src, type, code)) {
352 ICMP6STAT_INC(icp6s_toofreq);
357 * OK, ICMP6 can be generated.
360 if (m->m_pkthdr.len >= ICMPV6_PLD_MAXLEN)
361 m_adj(m, ICMPV6_PLD_MAXLEN - m->m_pkthdr.len);
363 preplen = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr);
364 M_PREPEND(m, preplen, M_NOWAIT); /* FIB is also copied over. */
366 nd6log((LOG_DEBUG, "ENOBUFS in icmp6_error %d\n", __LINE__));
370 nip6 = mtod(m, struct ip6_hdr *);
371 nip6->ip6_src = oip6->ip6_src;
372 nip6->ip6_dst = oip6->ip6_dst;
374 in6_clearscope(&oip6->ip6_src);
375 in6_clearscope(&oip6->ip6_dst);
377 icmp6 = (struct icmp6_hdr *)(nip6 + 1);
378 icmp6->icmp6_type = type;
379 icmp6->icmp6_code = code;
380 icmp6->icmp6_pptr = htonl((u_int32_t)param);
382 ICMP6STAT_INC(icp6s_outhist[type]);
384 icmp6_reflect(m, sizeof(struct ip6_hdr)); /* header order: IPv6 - ICMPv6 */
391 * If we can't tell whether or not we can generate ICMP6, free it.
397 * Process a received ICMP6 message.
400 icmp6_input(struct mbuf **mp, int *offp, int proto)
404 struct ip6_hdr *ip6, *nip6;
405 struct icmp6_hdr *icmp6, *nicmp6;
406 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
407 int code, error, icmp6len, ip6len, noff, off, sum;
414 if (m->m_len < off + sizeof(struct icmp6_hdr)) {
415 m = m_pullup(m, off + sizeof(struct icmp6_hdr));
417 IP6STAT_INC(ip6s_exthdrtoolong);
419 return (IPPROTO_DONE);
424 * Locate icmp6 structure in mbuf, and check
425 * that not corrupted and of at least minimum length
428 icmp6len = m->m_pkthdr.len - off;
429 if (icmp6len < sizeof(struct icmp6_hdr)) {
430 ICMP6STAT_INC(icp6s_tooshort);
434 ip6 = mtod(m, struct ip6_hdr *);
435 ifp = m->m_pkthdr.rcvif;
437 * Check multicast group membership.
438 * Note: SSM filters are not applied for ICMPv6 traffic.
440 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
441 struct in6_multi *inm;
443 inm = in6m_lookup(ifp, &ip6->ip6_dst);
445 IP6STAT_INC(ip6s_notmember);
446 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
451 /* Calculate the checksum. */
452 icmp6 = (struct icmp6_hdr *)((caddr_t)ip6 + off);
453 code = icmp6->icmp6_code;
454 if ((sum = in6_cksum(m, IPPROTO_ICMPV6, off, icmp6len)) != 0) {
456 "ICMP6 checksum error(%d|%x) %s\n",
457 icmp6->icmp6_type, sum,
458 ip6_sprintf(ip6bufs, &ip6->ip6_src)));
459 ICMP6STAT_INC(icp6s_checksum);
463 ICMP6STAT_INC(icp6s_inhist[icmp6->icmp6_type]);
464 icmp6_ifstat_inc(ifp, ifs6_in_msg);
465 if (icmp6->icmp6_type < ICMP6_INFOMSG_MASK)
466 icmp6_ifstat_inc(ifp, ifs6_in_error);
468 ip6len = sizeof(struct ip6_hdr) + ntohs(ip6->ip6_plen);
469 switch (icmp6->icmp6_type) {
470 case ICMP6_DST_UNREACH:
471 icmp6_ifstat_inc(ifp, ifs6_in_dstunreach);
473 case ICMP6_DST_UNREACH_NOROUTE:
474 case ICMP6_DST_UNREACH_ADDR: /* PRC_HOSTDEAD is a DOS */
475 code = PRC_UNREACH_NET;
477 case ICMP6_DST_UNREACH_ADMIN:
478 icmp6_ifstat_inc(ifp, ifs6_in_adminprohib);
479 code = PRC_UNREACH_ADMIN_PROHIB;
481 case ICMP6_DST_UNREACH_BEYONDSCOPE:
482 /* I mean "source address was incorrect." */
483 code = PRC_PARAMPROB;
485 case ICMP6_DST_UNREACH_NOPORT:
486 code = PRC_UNREACH_PORT;
494 case ICMP6_PACKET_TOO_BIG:
495 icmp6_ifstat_inc(ifp, ifs6_in_pkttoobig);
497 /* validation is made in icmp6_mtudisc_update */
502 * Updating the path MTU will be done after examining
503 * intermediate extension headers.
508 case ICMP6_TIME_EXCEEDED:
509 icmp6_ifstat_inc(ifp, ifs6_in_timeexceed);
511 case ICMP6_TIME_EXCEED_TRANSIT:
512 code = PRC_TIMXCEED_INTRANS;
514 case ICMP6_TIME_EXCEED_REASSEMBLY:
515 code = PRC_TIMXCEED_REASS;
523 case ICMP6_PARAM_PROB:
524 icmp6_ifstat_inc(ifp, ifs6_in_paramprob);
526 case ICMP6_PARAMPROB_NEXTHEADER:
527 code = PRC_UNREACH_PROTOCOL;
529 case ICMP6_PARAMPROB_HEADER:
530 case ICMP6_PARAMPROB_OPTION:
531 code = PRC_PARAMPROB;
539 case ICMP6_ECHO_REQUEST:
540 icmp6_ifstat_inc(ifp, ifs6_in_echo);
543 if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL) {
548 || n->m_len < off + sizeof(struct icmp6_hdr)) {
552 CTASSERT(sizeof(*nip6) + sizeof(*nicmp6) <= MHLEN);
553 n = m_gethdr(M_NOWAIT, n0->m_type);
560 m_move_pkthdr(n, n0); /* FIB copied. */
561 n0len = n0->m_pkthdr.len; /* save for use below */
563 * Copy IPv6 and ICMPv6 only.
565 nip6 = mtod(n, struct ip6_hdr *);
566 bcopy(ip6, nip6, sizeof(struct ip6_hdr));
567 nicmp6 = (struct icmp6_hdr *)(nip6 + 1);
568 bcopy(icmp6, nicmp6, sizeof(struct icmp6_hdr));
569 noff = sizeof(struct ip6_hdr);
570 /* new mbuf contains only ipv6+icmpv6 headers */
571 n->m_len = noff + sizeof(struct icmp6_hdr);
573 * Adjust mbuf. ip6_plen will be adjusted in
576 m_adj(n0, off + sizeof(struct icmp6_hdr));
577 /* recalculate complete packet size */
578 n->m_pkthdr.len = n0len + (noff - off);
581 if (n->m_len < off + sizeof(*nicmp6)) {
582 n = m_pullup(n, off + sizeof(*nicmp6));
584 IP6STAT_INC(ip6s_exthdrtoolong);
588 nicmp6 = (struct icmp6_hdr *)(mtod(n, caddr_t) + off);
592 nicmp6->icmp6_type = ICMP6_ECHO_REPLY;
593 nicmp6->icmp6_code = 0;
594 ICMP6STAT_INC(icp6s_reflect);
595 ICMP6STAT_INC(icp6s_outhist[ICMP6_ECHO_REPLY]);
596 icmp6_reflect(n, noff);
600 case ICMP6_ECHO_REPLY:
601 icmp6_ifstat_inc(ifp, ifs6_in_echoreply);
606 case MLD_LISTENER_QUERY:
607 case MLD_LISTENER_REPORT:
608 case MLD_LISTENER_DONE:
609 case MLDV2_LISTENER_REPORT:
611 * Drop MLD traffic which is not link-local, has a hop limit
612 * of greater than 1 hop, or which does not have the
613 * IPv6 HBH Router Alert option.
614 * As IPv6 HBH options are stripped in ip6_input() we must
615 * check an mbuf header flag.
616 * XXX Should we also sanity check that these messages
617 * were directed to a link-local multicast prefix?
619 if ((ip6->ip6_hlim != 1) || (m->m_flags & M_RTALERT_MLD) == 0)
621 if (mld_input(&m, off, icmp6len) != 0) {
623 return (IPPROTO_DONE);
628 case ICMP6_WRUREQUEST: /* ICMP6_FQDN_QUERY */
630 enum { WRU, FQDN } mode;
633 if (!V_icmp6_nodeinfo)
636 if (icmp6len == sizeof(struct icmp6_hdr) + 4)
638 else if (icmp6len >= sizeof(struct icmp6_nodeinfo))
644 sx_slock(&allprison_lock);
645 TAILQ_FOREACH(pr, &allprison, pr_list)
646 if (pr->pr_vnet == ifp->if_vnet)
648 sx_sunlock(&allprison_lock);
650 pr = curthread->td_ucred->cr_prison;
652 if (m->m_len < off + sizeof(struct icmp6_nodeinfo)) {
653 m = m_pullup(m, off +
654 sizeof(struct icmp6_nodeinfo));
656 IP6STAT_INC(ip6s_exthdrtoolong);
658 return (IPPROTO_DONE);
661 n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
663 n = ni6_input(n, off, pr);
664 /* XXX meaningless if n == NULL */
665 noff = sizeof(struct ip6_hdr);
671 * XXX: this combination of flags is pointless,
672 * but should we keep this for compatibility?
674 if ((V_icmp6_nodeinfo & (ICMP6_NODEINFO_FQDNOK |
675 ICMP6_NODEINFO_TMPADDROK)) !=
676 (ICMP6_NODEINFO_FQDNOK | ICMP6_NODEINFO_TMPADDROK))
682 CTASSERT(sizeof(*nip6) + sizeof(*nicmp6) + 4 <= MHLEN);
683 n = m_gethdr(M_NOWAIT, m->m_type);
688 if (!m_dup_pkthdr(n, m, M_NOWAIT)) {
690 * Previous code did a blind M_COPY_PKTHDR
691 * and said "just for rcvif". If true, then
692 * we could tolerate the dup failing (due to
693 * the deep copy of the tag chain). For now
694 * be conservative and just fail.
701 * Copy IPv6 and ICMPv6 only.
703 nip6 = mtod(n, struct ip6_hdr *);
704 bcopy(ip6, nip6, sizeof(struct ip6_hdr));
705 nicmp6 = (struct icmp6_hdr *)(nip6 + 1);
706 bcopy(icmp6, nicmp6, sizeof(struct icmp6_hdr));
707 p = (u_char *)(nicmp6 + 1);
710 maxhlen = M_TRAILINGSPACE(n) -
711 (sizeof(*nip6) + sizeof(*nicmp6) + 4);
712 mtx_lock(&pr->pr_mtx);
713 hlen = strlen(pr->pr_hostname);
716 /* meaningless TTL */
717 bcopy(pr->pr_hostname, p + 4, maxhlen);
718 mtx_unlock(&pr->pr_mtx);
719 noff = sizeof(struct ip6_hdr);
720 n->m_pkthdr.len = n->m_len = sizeof(struct ip6_hdr) +
721 sizeof(struct icmp6_hdr) + 4 + maxhlen;
722 nicmp6->icmp6_type = ICMP6_WRUREPLY;
723 nicmp6->icmp6_code = 0;
726 ICMP6STAT_INC(icp6s_reflect);
727 ICMP6STAT_INC(icp6s_outhist[ICMP6_WRUREPLY]);
728 icmp6_reflect(n, noff);
738 case ND_ROUTER_SOLICIT:
739 icmp6_ifstat_inc(ifp, ifs6_in_routersolicit);
742 if (icmp6len < sizeof(struct nd_router_solicit))
744 if (send_sendso_input_hook != NULL) {
745 if (m->m_len < off + icmp6len) {
746 m = m_pullup(m, off + icmp6len);
748 IP6STAT_INC(ip6s_exthdrtoolong);
750 return (IPPROTO_DONE);
753 error = send_sendso_input_hook(m, ifp, SND_IN, ip6len);
759 n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
760 nd6_rs_input(m, off, icmp6len);
766 case ND_ROUTER_ADVERT:
767 icmp6_ifstat_inc(ifp, ifs6_in_routeradvert);
770 if (icmp6len < sizeof(struct nd_router_advert))
772 if (send_sendso_input_hook != NULL) {
773 error = send_sendso_input_hook(m, ifp, SND_IN, ip6len);
779 n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
780 nd6_ra_input(m, off, icmp6len);
786 case ND_NEIGHBOR_SOLICIT:
787 icmp6_ifstat_inc(ifp, ifs6_in_neighborsolicit);
790 if (icmp6len < sizeof(struct nd_neighbor_solicit))
792 if (send_sendso_input_hook != NULL) {
793 error = send_sendso_input_hook(m, ifp, SND_IN, ip6len);
799 n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
800 nd6_ns_input(m, off, icmp6len);
806 case ND_NEIGHBOR_ADVERT:
807 icmp6_ifstat_inc(ifp, ifs6_in_neighboradvert);
810 if (icmp6len < sizeof(struct nd_neighbor_advert))
812 if (send_sendso_input_hook != NULL) {
813 error = send_sendso_input_hook(m, ifp, SND_IN, ip6len);
819 n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
820 nd6_na_input(m, off, icmp6len);
827 icmp6_ifstat_inc(ifp, ifs6_in_redirect);
830 if (icmp6len < sizeof(struct nd_redirect))
832 if (send_sendso_input_hook != NULL) {
833 error = send_sendso_input_hook(m, ifp, SND_IN, ip6len);
839 n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
840 icmp6_redirect_input(m, off);
846 case ICMP6_ROUTER_RENUMBERING:
847 if (code != ICMP6_ROUTER_RENUMBERING_COMMAND &&
848 code != ICMP6_ROUTER_RENUMBERING_RESULT)
850 if (icmp6len < sizeof(struct icmp6_router_renum))
856 "icmp6_input: unknown type %d(src=%s, dst=%s, ifid=%d)\n",
857 icmp6->icmp6_type, ip6_sprintf(ip6bufs, &ip6->ip6_src),
858 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
859 ifp ? ifp->if_index : 0));
860 if (icmp6->icmp6_type < ICMP6_ECHO_REQUEST) {
861 /* ICMPv6 error: MUST deliver it by spec... */
865 /* ICMPv6 informational: MUST not deliver */
869 if (icmp6_notify_error(&m, off, icmp6len, code) != 0) {
870 /* In this case, m should've been freed. */
872 return (IPPROTO_DONE);
877 ICMP6STAT_INC(icp6s_badcode);
881 ICMP6STAT_INC(icp6s_badlen);
885 /* deliver the packet to appropriate sockets */
886 icmp6_rip6_input(&m, *offp);
889 return (IPPROTO_DONE);
894 return (IPPROTO_DONE);
898 icmp6_notify_error(struct mbuf **mp, int off, int icmp6len, int code)
901 struct icmp6_hdr *icmp6;
902 struct ip6_hdr *eip6;
904 struct sockaddr_in6 icmp6src, icmp6dst;
908 if (icmp6len < sizeof(struct icmp6_hdr) + sizeof(struct ip6_hdr)) {
909 ICMP6STAT_INC(icp6s_tooshort);
913 if (m->m_len < off + sizeof(*icmp6) + sizeof(struct ip6_hdr)) {
914 m = m_pullup(m, off + sizeof(*icmp6) + sizeof(struct ip6_hdr));
916 IP6STAT_INC(ip6s_exthdrtoolong);
921 icmp6 = (struct icmp6_hdr *)(mtod(m, caddr_t) + off);
922 eip6 = (struct ip6_hdr *)(icmp6 + 1);
923 bzero(&icmp6dst, sizeof(icmp6dst));
925 /* Detect the upper level protocol */
927 void (*ctlfunc)(int, struct sockaddr *, void *);
928 u_int8_t nxt = eip6->ip6_nxt;
929 int eoff = off + sizeof(struct icmp6_hdr) +
930 sizeof(struct ip6_hdr);
931 struct ip6ctlparam ip6cp;
932 int icmp6type = icmp6->icmp6_type;
934 struct ip6_rthdr *rth;
935 struct ip6_rthdr0 *rth0;
938 while (1) { /* XXX: should avoid infinite loop explicitly? */
942 case IPPROTO_HOPOPTS:
943 case IPPROTO_DSTOPTS:
945 if (m->m_len < eoff + sizeof(struct ip6_ext)) {
946 m = m_pullup(m, eoff +
947 sizeof(struct ip6_ext));
949 IP6STAT_INC(ip6s_exthdrtoolong);
954 eh = (struct ip6_ext *)
955 (mtod(m, caddr_t) + eoff);
956 if (nxt == IPPROTO_AH)
957 eoff += (eh->ip6e_len + 2) << 2;
959 eoff += (eh->ip6e_len + 1) << 3;
962 case IPPROTO_ROUTING:
964 * When the erroneous packet contains a
965 * routing header, we should examine the
966 * header to determine the final destination.
967 * Otherwise, we can't properly update
968 * information that depends on the final
969 * destination (e.g. path MTU).
971 if (m->m_len < eoff + sizeof(*rth)) {
972 m = m_pullup(m, eoff + sizeof(*rth));
974 IP6STAT_INC(ip6s_exthdrtoolong);
979 rth = (struct ip6_rthdr *)
980 (mtod(m, caddr_t) + eoff);
981 rthlen = (rth->ip6r_len + 1) << 3;
983 * XXX: currently there is no
984 * officially defined type other
986 * Note that if the segment left field
987 * is 0, all intermediate hops must
990 if (rth->ip6r_segleft &&
991 rth->ip6r_type == IPV6_RTHDR_TYPE_0) {
994 if (m->m_len < eoff + rthlen) {
995 m = m_pullup(m, eoff + rthlen);
1003 rth0 = (struct ip6_rthdr0 *)
1004 (mtod(m, caddr_t) + eoff);
1006 /* just ignore a bogus header */
1007 if ((rth0->ip6r0_len % 2) == 0 &&
1008 (hops = rth0->ip6r0_len/2))
1009 icmp6dst.sin6_addr = *((struct in6_addr *)(rth0 + 1) + (hops - 1));
1012 nxt = rth->ip6r_nxt;
1014 case IPPROTO_FRAGMENT:
1015 if (m->m_len < eoff + sizeof(struct ip6_frag)) {
1016 m = m_pullup(m, eoff +
1017 sizeof(struct ip6_frag));
1019 IP6STAT_INC(ip6s_exthdrtoolong);
1024 fh = (struct ip6_frag *)(mtod(m, caddr_t) +
1027 * Data after a fragment header is meaningless
1028 * unless it is the first fragment, but
1029 * we'll go to the notify label for path MTU
1032 if (fh->ip6f_offlg & IP6F_OFF_MASK)
1035 eoff += sizeof(struct ip6_frag);
1040 * This case includes ESP and the No Next
1041 * Header. In such cases going to the notify
1042 * label does not have any meaning
1043 * (i.e. ctlfunc will be NULL), but we go
1044 * anyway since we might have to update
1045 * path MTU information.
1051 icmp6 = (struct icmp6_hdr *)(mtod(m, caddr_t) + off);
1054 * retrieve parameters from the inner IPv6 header, and convert
1055 * them into sockaddr structures.
1056 * XXX: there is no guarantee that the source or destination
1057 * addresses of the inner packet are in the same scope as
1058 * the addresses of the icmp packet. But there is no other
1059 * way to determine the zone.
1061 eip6 = (struct ip6_hdr *)(icmp6 + 1);
1063 icmp6dst.sin6_len = sizeof(struct sockaddr_in6);
1064 icmp6dst.sin6_family = AF_INET6;
1065 if (IN6_IS_ADDR_UNSPECIFIED(&icmp6dst.sin6_addr))
1066 icmp6dst.sin6_addr = eip6->ip6_dst;
1067 if (in6_setscope(&icmp6dst.sin6_addr, m->m_pkthdr.rcvif, NULL))
1069 bzero(&icmp6src, sizeof(icmp6src));
1070 icmp6src.sin6_len = sizeof(struct sockaddr_in6);
1071 icmp6src.sin6_family = AF_INET6;
1072 icmp6src.sin6_addr = eip6->ip6_src;
1073 if (in6_setscope(&icmp6src.sin6_addr, m->m_pkthdr.rcvif, NULL))
1075 icmp6src.sin6_flowinfo =
1076 (eip6->ip6_flow & IPV6_FLOWLABEL_MASK);
1079 ip6cp.ip6c_icmp6 = icmp6;
1080 ip6cp.ip6c_ip6 = (struct ip6_hdr *)(icmp6 + 1);
1081 ip6cp.ip6c_off = eoff;
1082 ip6cp.ip6c_finaldst = &icmp6dst.sin6_addr;
1083 ip6cp.ip6c_src = &icmp6src;
1084 ip6cp.ip6c_nxt = nxt;
1086 if (icmp6type == ICMP6_PACKET_TOO_BIG) {
1087 notifymtu = ntohl(icmp6->icmp6_mtu);
1088 ip6cp.ip6c_cmdarg = (void *)¬ifymtu;
1089 icmp6_mtudisc_update(&ip6cp, 1); /*XXX*/
1092 ctlfunc = (void (*)(int, struct sockaddr *, void *))
1093 (inet6sw[ip6_protox[nxt]].pr_ctlinput);
1095 (void) (*ctlfunc)(code, (struct sockaddr *)&icmp6dst,
1109 icmp6_mtudisc_update(struct ip6ctlparam *ip6cp, int validated)
1111 struct in6_addr *dst = ip6cp->ip6c_finaldst;
1112 struct icmp6_hdr *icmp6 = ip6cp->ip6c_icmp6;
1113 struct mbuf *m = ip6cp->ip6c_m; /* will be necessary for scope issue */
1114 u_int mtu = ntohl(icmp6->icmp6_mtu);
1115 struct in_conninfo inc;
1120 * RFC2460 section 5, last paragraph.
1121 * even though minimum link MTU for IPv6 is IPV6_MMTU,
1122 * we may see ICMPv6 too big with mtu < IPV6_MMTU
1123 * due to packet translator in the middle.
1124 * see ip6_output() and ip6_getpmtu() "alwaysfrag" case for
1127 if (mtu < IPV6_MMTU)
1132 * we reject ICMPv6 too big with abnormally small value.
1133 * XXX what is the good definition of "abnormally small"?
1135 if (mtu < sizeof(struct ip6_hdr) + sizeof(struct ip6_frag) + 8)
1142 * In case the suggested mtu is less than IPV6_MMTU, we
1143 * only need to remember that it was for above mentioned
1144 * "alwaysfrag" case.
1145 * Try to be as close to the spec as possible.
1147 if (mtu < IPV6_MMTU)
1148 mtu = IPV6_MMTU - 8;
1150 bzero(&inc, sizeof(inc));
1151 inc.inc_fibnum = M_GETFIB(m);
1152 inc.inc_flags |= INC_ISIPV6;
1153 inc.inc6_faddr = *dst;
1154 if (in6_setscope(&inc.inc6_faddr, m->m_pkthdr.rcvif, NULL))
1157 max_mtu = tcp_hc_getmtu(&inc);
1159 max_mtu = tcp_maxmtu6(&inc, NULL);
1161 if (mtu < max_mtu) {
1162 tcp_hc_updatemtu(&inc, mtu);
1163 ICMP6STAT_INC(icp6s_pmtuchg);
1168 * Process a Node Information Query packet, based on
1169 * draft-ietf-ipngwg-icmp-name-lookups-07.
1171 * Spec incompatibilities:
1172 * - IPv6 Subject address handling
1173 * - IPv4 Subject address handling support missing
1174 * - Proxy reply (answer even if it's not for me)
1175 * - joins NI group address at in6_ifattach() time only, does not cope
1176 * with hostname changes by sethostname(3)
1178 static struct mbuf *
1179 ni6_input(struct mbuf *m, int off, struct prison *pr)
1181 struct icmp6_nodeinfo *ni6, *nni6;
1182 struct mbuf *n = NULL;
1185 int replylen = sizeof(struct ip6_hdr) + sizeof(struct icmp6_nodeinfo);
1186 struct ni_reply_fqdn *fqdn;
1187 int addrs; /* for NI_QTYPE_NODEADDR */
1188 struct ifnet *ifp = NULL; /* for NI_QTYPE_NODEADDR */
1189 struct in6_addr in6_subj; /* subject address */
1190 struct ip6_hdr *ip6;
1191 int oldfqdn = 0; /* if 1, return pascal string (03 draft) */
1193 struct in6_ifaddr *ia6 = NULL;
1195 ip6 = mtod(m, struct ip6_hdr *);
1196 ni6 = (struct icmp6_nodeinfo *)(mtod(m, caddr_t) + off);
1199 * Validate IPv6 source address.
1200 * The default configuration MUST be to refuse answering queries from
1201 * global-scope addresses according to RFC4602.
1203 * - it's not very clear what "refuse" means; this implementation
1205 * - it's not very easy to identify global-scope (unicast) addresses
1206 * since there are many prefixes for them. It should be safer
1207 * and in practice sufficient to check "all" but loopback and
1208 * link-local (note that site-local unicast was deprecated and
1209 * ULA is defined as global scope-wise)
1211 if ((V_icmp6_nodeinfo & ICMP6_NODEINFO_GLOBALOK) == 0 &&
1212 !IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) &&
1213 !IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_src))
1217 * Validate IPv6 destination address.
1219 * The Responder must discard the Query without further processing
1220 * unless it is one of the Responder's unicast or anycast addresses, or
1221 * a link-local scope multicast address which the Responder has joined.
1222 * [RFC4602, Section 5.]
1224 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
1225 if (!IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1227 /* else it's a link-local multicast, fine */
1228 } else { /* unicast or anycast */
1229 ia6 = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */, false);
1231 goto bad; /* XXX impossible */
1233 if ((ia6->ia6_flags & IN6_IFF_TEMPORARY) &&
1234 !(V_icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK)) {
1235 nd6log((LOG_DEBUG, "ni6_input: ignore node info to "
1236 "a temporary address in %s:%d",
1237 __FILE__, __LINE__));
1242 /* validate query Subject field. */
1243 qtype = ntohs(ni6->ni_qtype);
1244 subjlen = m->m_pkthdr.len - off - sizeof(struct icmp6_nodeinfo);
1247 case NI_QTYPE_SUPTYPES:
1249 if (ni6->ni_code == ICMP6_NI_SUBJ_FQDN && subjlen == 0)
1253 case NI_QTYPE_NODEADDR:
1254 case NI_QTYPE_IPV4ADDR:
1255 switch (ni6->ni_code) {
1256 case ICMP6_NI_SUBJ_IPV6:
1257 #if ICMP6_NI_SUBJ_IPV6 != 0
1261 * backward compatibility - try to accept 03 draft
1262 * format, where no Subject is present.
1264 if (qtype == NI_QTYPE_FQDN && ni6->ni_code == 0 &&
1269 #if ICMP6_NI_SUBJ_IPV6 != 0
1270 if (ni6->ni_code != ICMP6_NI_SUBJ_IPV6)
1274 if (subjlen != sizeof(struct in6_addr))
1278 * Validate Subject address.
1280 * Not sure what exactly "address belongs to the node"
1281 * means in the spec, is it just unicast, or what?
1283 * At this moment we consider Subject address as
1284 * "belong to the node" if the Subject address equals
1285 * to the IPv6 destination address; validation for
1286 * IPv6 destination address should have done enough
1289 * We do not do proxy at this moment.
1291 m_copydata(m, off + sizeof(struct icmp6_nodeinfo),
1292 subjlen, (caddr_t)&in6_subj);
1293 if (in6_setscope(&in6_subj, m->m_pkthdr.rcvif, NULL))
1296 subj = (char *)&in6_subj;
1297 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &in6_subj))
1301 * XXX if we are to allow other cases, we should really
1302 * be careful about scope here.
1303 * basically, we should disallow queries toward IPv6
1304 * destination X with subject Y,
1305 * if scope(X) > scope(Y).
1306 * if we allow scope(X) > scope(Y), it will result in
1307 * information leakage across scope boundary.
1311 case ICMP6_NI_SUBJ_FQDN:
1313 * Validate Subject name with gethostname(3).
1315 * The behavior may need some debate, since:
1316 * - we are not sure if the node has FQDN as
1317 * hostname (returned by gethostname(3)).
1318 * - the code does wildcard match for truncated names.
1319 * however, we are not sure if we want to perform
1320 * wildcard match, if gethostname(3) side has
1321 * truncated hostname.
1323 mtx_lock(&pr->pr_mtx);
1324 n = ni6_nametodns(pr->pr_hostname,
1325 strlen(pr->pr_hostname), 0);
1326 mtx_unlock(&pr->pr_mtx);
1327 if (!n || n->m_next || n->m_len == 0)
1329 if (m->m_len < off + sizeof(struct icmp6_nodeinfo) +
1331 m = m_pullup(m, off +
1332 sizeof(struct icmp6_nodeinfo) + subjlen);
1334 IP6STAT_INC(ip6s_exthdrtoolong);
1338 /* ip6 possibly invalid but not used after. */
1339 ni6 = (struct icmp6_nodeinfo *)(mtod(m, caddr_t) + off);
1340 subj = (char *)(mtod(m, caddr_t) + off +
1341 sizeof(struct icmp6_nodeinfo));
1342 if (!ni6_dnsmatch(subj, subjlen, mtod(n, const char *),
1350 case ICMP6_NI_SUBJ_IPV4: /* XXX: to be implemented? */
1357 /* refuse based on configuration. XXX ICMP6_NI_REFUSED? */
1360 if ((V_icmp6_nodeinfo & ICMP6_NODEINFO_FQDNOK) == 0)
1363 case NI_QTYPE_NODEADDR:
1364 case NI_QTYPE_IPV4ADDR:
1365 if ((V_icmp6_nodeinfo & ICMP6_NODEINFO_NODEADDROK) == 0)
1370 /* guess reply length */
1373 break; /* no reply data */
1374 case NI_QTYPE_SUPTYPES:
1375 replylen += sizeof(u_int32_t);
1378 /* XXX will append an mbuf */
1379 replylen += offsetof(struct ni_reply_fqdn, ni_fqdn_namelen);
1381 case NI_QTYPE_NODEADDR:
1382 addrs = ni6_addrs(ni6, m, &ifp, (struct in6_addr *)subj);
1383 if ((replylen += addrs * (sizeof(struct in6_addr) +
1384 sizeof(u_int32_t))) > MCLBYTES)
1385 replylen = MCLBYTES; /* XXX: will truncate pkt later */
1387 case NI_QTYPE_IPV4ADDR:
1388 /* unsupported - should respond with unknown Qtype? */
1392 * XXX: We must return a reply with the ICMP6 code
1393 * `unknown Qtype' in this case. However we regard the case
1394 * as an FQDN query for backward compatibility.
1395 * Older versions set a random value to this field,
1396 * so it rarely varies in the defined qtypes.
1397 * But the mechanism is not reliable...
1398 * maybe we should obsolete older versions.
1400 qtype = NI_QTYPE_FQDN;
1401 /* XXX will append an mbuf */
1402 replylen += offsetof(struct ni_reply_fqdn, ni_fqdn_namelen);
1407 /* Allocate an mbuf to reply. */
1408 if (replylen > MCLBYTES) {
1410 * XXX: should we try to allocate more? But MCLBYTES
1411 * is probably much larger than IPV6_MMTU...
1415 if (replylen > MHLEN)
1416 n = m_getcl(M_NOWAIT, m->m_type, M_PKTHDR);
1418 n = m_gethdr(M_NOWAIT, m->m_type);
1423 m_move_pkthdr(n, m); /* just for recvif and FIB */
1424 n->m_pkthdr.len = n->m_len = replylen;
1426 /* copy mbuf header and IPv6 + Node Information base headers */
1427 bcopy(mtod(m, caddr_t), mtod(n, caddr_t), sizeof(struct ip6_hdr));
1428 nni6 = (struct icmp6_nodeinfo *)(mtod(n, struct ip6_hdr *) + 1);
1429 bcopy((caddr_t)ni6, (caddr_t)nni6, sizeof(struct icmp6_nodeinfo));
1431 /* qtype dependent procedure */
1434 nni6->ni_code = ICMP6_NI_SUCCESS;
1437 case NI_QTYPE_SUPTYPES:
1440 nni6->ni_code = ICMP6_NI_SUCCESS;
1441 nni6->ni_flags = htons(0x0000); /* raw bitmap */
1442 /* supports NOOP, SUPTYPES, FQDN, and NODEADDR */
1443 v = (u_int32_t)htonl(0x0000000f);
1444 bcopy(&v, nni6 + 1, sizeof(u_int32_t));
1448 nni6->ni_code = ICMP6_NI_SUCCESS;
1449 fqdn = (struct ni_reply_fqdn *)(mtod(n, caddr_t) +
1450 sizeof(struct ip6_hdr) + sizeof(struct icmp6_nodeinfo));
1451 nni6->ni_flags = 0; /* XXX: meaningless TTL */
1452 fqdn->ni_fqdn_ttl = 0; /* ditto. */
1454 * XXX do we really have FQDN in hostname?
1456 mtx_lock(&pr->pr_mtx);
1457 n->m_next = ni6_nametodns(pr->pr_hostname,
1458 strlen(pr->pr_hostname), oldfqdn);
1459 mtx_unlock(&pr->pr_mtx);
1460 if (n->m_next == NULL)
1462 /* XXX we assume that n->m_next is not a chain */
1463 if (n->m_next->m_next != NULL)
1465 n->m_pkthdr.len += n->m_next->m_len;
1467 case NI_QTYPE_NODEADDR:
1471 nni6->ni_code = ICMP6_NI_SUCCESS;
1472 n->m_pkthdr.len = n->m_len =
1473 sizeof(struct ip6_hdr) + sizeof(struct icmp6_nodeinfo);
1474 lenlim = M_TRAILINGSPACE(n);
1475 copied = ni6_store_addrs(ni6, nni6, ifp, lenlim);
1476 /* XXX: reset mbuf length */
1477 n->m_pkthdr.len = n->m_len = sizeof(struct ip6_hdr) +
1478 sizeof(struct icmp6_nodeinfo) + copied;
1482 break; /* XXX impossible! */
1485 nni6->ni_type = ICMP6_NI_REPLY;
1497 * make a mbuf with DNS-encoded string. no compression support.
1499 * XXX names with less than 2 dots (like "foo" or "foo.section") will be
1500 * treated as truncated name (two \0 at the end). this is a wild guess.
1502 * old - return pascal string if non-zero
1504 static struct mbuf *
1505 ni6_nametodns(const char *name, int namelen, int old)
1517 /* Because MAXHOSTNAMELEN is usually 256, we use cluster mbuf. */
1519 m = m_getcl(M_NOWAIT, MT_DATA, 0);
1521 m = m_get(M_NOWAIT, MT_DATA);
1527 *mtod(m, char *) = namelen;
1528 bcopy(name, mtod(m, char *) + 1, namelen);
1532 cp = mtod(m, char *);
1533 ep = mtod(m, char *) + M_TRAILINGSPACE(m);
1535 /* if not certain about my name, return empty buffer */
1540 * guess if it looks like shortened hostname, or FQDN.
1541 * shortened hostname needs two trailing "\0".
1544 for (p = name; p < name + namelen; p++) {
1545 if (*p && *p == '.')
1554 while (cp < ep && p < name + namelen) {
1556 for (q = p; q < name + namelen && *q && *q != '.'; q++)
1558 /* result does not fit into mbuf */
1559 if (cp + i + 1 >= ep)
1562 * DNS label length restriction, RFC1035 page 8.
1563 * "i == 0" case is included here to avoid returning
1564 * 0-length label on "foo..bar".
1566 if (i <= 0 || i >= 64)
1572 if (p < name + namelen && *p == '.')
1576 if (cp + nterm >= ep)
1580 m->m_len = cp - mtod(m, char *);
1584 panic("should not reach here");
1594 * check if two DNS-encoded string matches. takes care of truncated
1595 * form (with \0\0 at the end). no compression support.
1596 * XXX upper/lowercase match (see RFC2065)
1599 ni6_dnsmatch(const char *a, int alen, const char *b, int blen)
1601 const char *a0, *b0;
1604 /* simplest case - need validation? */
1605 if (alen == blen && bcmp(a, b, alen) == 0)
1611 /* termination is mandatory */
1612 if (alen < 2 || blen < 2)
1614 if (a0[alen - 1] != '\0' || b0[blen - 1] != '\0')
1619 while (a - a0 < alen && b - b0 < blen) {
1620 if (a - a0 + 1 > alen || b - b0 + 1 > blen)
1623 if ((signed char)a[0] < 0 || (signed char)b[0] < 0)
1625 /* we don't support compression yet */
1626 if (a[0] >= 64 || b[0] >= 64)
1629 /* truncated case */
1630 if (a[0] == 0 && a - a0 == alen - 1)
1632 if (b[0] == 0 && b - b0 == blen - 1)
1634 if (a[0] == 0 || b[0] == 0)
1640 if (a - a0 + 1 + l > alen || b - b0 + 1 + l > blen)
1642 if (bcmp(a + 1, b + 1, l) != 0)
1649 if (a - a0 == alen && b - b0 == blen)
1656 * calculate the number of addresses to be returned in the node info reply.
1659 ni6_addrs(struct icmp6_nodeinfo *ni6, struct mbuf *m, struct ifnet **ifpp,
1660 struct in6_addr *subj)
1663 struct in6_ifaddr *ifa6;
1665 int addrs = 0, addrsofif, iffound = 0;
1666 int niflags = ni6->ni_flags;
1670 if ((niflags & NI_NODEADDR_FLAG_ALL) == 0) {
1671 switch (ni6->ni_code) {
1672 case ICMP6_NI_SUBJ_IPV6:
1673 if (subj == NULL) /* must be impossible... */
1678 * XXX: we only support IPv6 subject address for
1685 CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
1687 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1688 if (ifa->ifa_addr->sa_family != AF_INET6)
1690 ifa6 = (struct in6_ifaddr *)ifa;
1692 if ((niflags & NI_NODEADDR_FLAG_ALL) == 0 &&
1693 IN6_ARE_ADDR_EQUAL(subj, &ifa6->ia_addr.sin6_addr))
1697 * IPv4-mapped addresses can only be returned by a
1698 * Node Information proxy, since they represent
1699 * addresses of IPv4-only nodes, which perforce do
1700 * not implement this protocol.
1701 * [icmp-name-lookups-07, Section 5.4]
1702 * So we don't support NI_NODEADDR_FLAG_COMPAT in
1703 * this function at this moment.
1706 /* What do we have to do about ::1? */
1707 switch (in6_addrscope(&ifa6->ia_addr.sin6_addr)) {
1708 case IPV6_ADDR_SCOPE_LINKLOCAL:
1709 if ((niflags & NI_NODEADDR_FLAG_LINKLOCAL) == 0)
1712 case IPV6_ADDR_SCOPE_SITELOCAL:
1713 if ((niflags & NI_NODEADDR_FLAG_SITELOCAL) == 0)
1716 case IPV6_ADDR_SCOPE_GLOBAL:
1717 if ((niflags & NI_NODEADDR_FLAG_GLOBAL) == 0)
1725 * check if anycast is okay.
1726 * XXX: just experimental. not in the spec.
1728 if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0 &&
1729 (niflags & NI_NODEADDR_FLAG_ANYCAST) == 0)
1730 continue; /* we need only unicast addresses */
1731 if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0 &&
1732 (V_icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK) == 0) {
1735 addrsofif++; /* count the address */
1749 ni6_store_addrs(struct icmp6_nodeinfo *ni6, struct icmp6_nodeinfo *nni6,
1750 struct ifnet *ifp0, int resid)
1753 struct in6_ifaddr *ifa6;
1755 struct ifnet *ifp_dep = NULL;
1756 int copied = 0, allow_deprecated = 0;
1757 u_char *cp = (u_char *)(nni6 + 1);
1758 int niflags = ni6->ni_flags;
1763 if (ifp0 == NULL && !(niflags & NI_NODEADDR_FLAG_ALL))
1764 return (0); /* needless to copy */
1766 ifp = ifp0 ? ifp0 : CK_STAILQ_FIRST(&V_ifnet);
1769 for (; ifp; ifp = CK_STAILQ_NEXT(ifp, if_link)) {
1770 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1771 if (ifa->ifa_addr->sa_family != AF_INET6)
1773 ifa6 = (struct in6_ifaddr *)ifa;
1775 if ((ifa6->ia6_flags & IN6_IFF_DEPRECATED) != 0 &&
1776 allow_deprecated == 0) {
1778 * prefererred address should be put before
1779 * deprecated addresses.
1782 /* record the interface for later search */
1783 if (ifp_dep == NULL)
1787 } else if ((ifa6->ia6_flags & IN6_IFF_DEPRECATED) == 0 &&
1788 allow_deprecated != 0)
1789 continue; /* we now collect deprecated addrs */
1791 /* What do we have to do about ::1? */
1792 switch (in6_addrscope(&ifa6->ia_addr.sin6_addr)) {
1793 case IPV6_ADDR_SCOPE_LINKLOCAL:
1794 if ((niflags & NI_NODEADDR_FLAG_LINKLOCAL) == 0)
1797 case IPV6_ADDR_SCOPE_SITELOCAL:
1798 if ((niflags & NI_NODEADDR_FLAG_SITELOCAL) == 0)
1801 case IPV6_ADDR_SCOPE_GLOBAL:
1802 if ((niflags & NI_NODEADDR_FLAG_GLOBAL) == 0)
1810 * check if anycast is okay.
1811 * XXX: just experimental. not in the spec.
1813 if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0 &&
1814 (niflags & NI_NODEADDR_FLAG_ANYCAST) == 0)
1816 if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0 &&
1817 (V_icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK) == 0) {
1821 /* now we can copy the address */
1822 if (resid < sizeof(struct in6_addr) +
1823 sizeof(u_int32_t)) {
1825 * We give up much more copy.
1826 * Set the truncate flag and return.
1828 nni6->ni_flags |= NI_NODEADDR_FLAG_TRUNCATE;
1833 * Set the TTL of the address.
1834 * The TTL value should be one of the following
1835 * according to the specification:
1837 * 1. The remaining lifetime of a DHCP lease on the
1839 * 2. The remaining Valid Lifetime of a prefix from
1840 * which the address was derived through Stateless
1841 * Autoconfiguration.
1843 * Note that we currently do not support stateful
1844 * address configuration by DHCPv6, so the former
1845 * case can't happen.
1847 if (ifa6->ia6_lifetime.ia6t_expire == 0)
1848 ltime = ND6_INFINITE_LIFETIME;
1850 if (ifa6->ia6_lifetime.ia6t_expire >
1852 ltime = htonl(ifa6->ia6_lifetime.ia6t_expire - time_uptime);
1857 bcopy(<ime, cp, sizeof(u_int32_t));
1858 cp += sizeof(u_int32_t);
1860 /* copy the address itself */
1861 bcopy(&ifa6->ia_addr.sin6_addr, cp,
1862 sizeof(struct in6_addr));
1863 in6_clearscope((struct in6_addr *)cp); /* XXX */
1864 cp += sizeof(struct in6_addr);
1866 resid -= (sizeof(struct in6_addr) + sizeof(u_int32_t));
1867 copied += (sizeof(struct in6_addr) + sizeof(u_int32_t));
1869 if (ifp0) /* we need search only on the specified IF */
1873 if (allow_deprecated == 0 && ifp_dep != NULL) {
1875 allow_deprecated = 1;
1884 * XXX almost dup'ed code with rip6_input.
1887 icmp6_rip6_input(struct mbuf **mp, int off)
1889 struct mbuf *m = *mp;
1890 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1892 struct inpcb *last = NULL;
1893 struct sockaddr_in6 fromsa;
1894 struct icmp6_hdr *icmp6;
1895 struct mbuf *opts = NULL;
1899 /* This is assumed to be safe; icmp6_input() does a pullup. */
1900 icmp6 = (struct icmp6_hdr *)((caddr_t)ip6 + off);
1903 * XXX: the address may have embedded scope zone ID, which should be
1904 * hidden from applications.
1906 bzero(&fromsa, sizeof(fromsa));
1907 fromsa.sin6_family = AF_INET6;
1908 fromsa.sin6_len = sizeof(struct sockaddr_in6);
1909 fromsa.sin6_addr = ip6->ip6_src;
1910 if (sa6_recoverscope(&fromsa)) {
1913 return (IPPROTO_DONE);
1916 CK_LIST_FOREACH(inp, &V_ripcb, inp_list) {
1917 if ((inp->inp_vflag & INP_IPV6) == 0)
1919 if (inp->inp_ip_p != IPPROTO_ICMPV6)
1921 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
1922 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
1924 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1925 !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src))
1928 if (__predict_false(inp->inp_flags2 & INP_FREED)) {
1932 if (ICMP6_FILTER_WILLBLOCK(icmp6->icmp6_type,
1933 inp->in6p_icmp6filt)) {
1938 struct mbuf *n = NULL;
1941 * Recent network drivers tend to allocate a single
1942 * mbuf cluster, rather than to make a couple of
1943 * mbufs without clusters. Also, since the IPv6 code
1944 * path tries to avoid m_pullup(), it is highly
1945 * probable that we still have an mbuf cluster here
1946 * even though the necessary length can be stored in an
1947 * mbuf's internal buffer.
1948 * Meanwhile, the default size of the receive socket
1949 * buffer for raw sockets is not so large. This means
1950 * the possibility of packet loss is relatively higher
1951 * than before. To avoid this scenario, we copy the
1952 * received data to a separate mbuf that does not use
1953 * a cluster, if possible.
1954 * XXX: it is better to copy the data after stripping
1955 * intermediate headers.
1957 if ((m->m_flags & M_EXT) && m->m_next == NULL &&
1958 m->m_len <= MHLEN) {
1959 n = m_get(M_NOWAIT, m->m_type);
1961 if (m_dup_pkthdr(n, m, M_NOWAIT)) {
1962 bcopy(m->m_data, n->m_data,
1964 n->m_len = m->m_len;
1972 (n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) != NULL) {
1973 if (last->inp_flags & INP_CONTROLOPTS)
1974 ip6_savecontrol(last, n, &opts);
1975 /* strip intermediate headers */
1977 SOCKBUF_LOCK(&last->inp_socket->so_rcv);
1978 if (sbappendaddr_locked(
1979 &last->inp_socket->so_rcv,
1980 (struct sockaddr *)&fromsa, n, opts)
1982 soroverflow_locked(last->inp_socket);
1988 sorwakeup_locked(last->inp_socket);
1996 if (last->inp_flags & INP_CONTROLOPTS)
1997 ip6_savecontrol(last, m, &opts);
1998 /* strip intermediate headers */
2001 /* avoid using mbuf clusters if possible (see above) */
2002 if ((m->m_flags & M_EXT) && m->m_next == NULL &&
2003 m->m_len <= MHLEN) {
2006 n = m_get(M_NOWAIT, m->m_type);
2008 if (m_dup_pkthdr(n, m, M_NOWAIT)) {
2009 bcopy(m->m_data, n->m_data, m->m_len);
2010 n->m_len = m->m_len;
2020 SOCKBUF_LOCK(&last->inp_socket->so_rcv);
2021 if (sbappendaddr_locked(&last->inp_socket->so_rcv,
2022 (struct sockaddr *)&fromsa, m, opts) == 0) {
2026 soroverflow_locked(last->inp_socket);
2028 sorwakeup_locked(last->inp_socket);
2032 IP6STAT_DEC(ip6s_delivered);
2035 return (IPPROTO_DONE);
2039 * Reflect the ip6 packet back to the source.
2040 * OFF points to the icmp6 header, counted from the top of the mbuf.
2043 icmp6_reflect(struct mbuf *m, size_t off)
2045 struct in6_addr src6, *srcp;
2046 struct ip6_hdr *ip6;
2047 struct icmp6_hdr *icmp6;
2048 struct in6_ifaddr *ia = NULL;
2049 struct ifnet *outif = NULL;
2051 int type, code, hlim;
2053 /* too short to reflect */
2054 if (off < sizeof(struct ip6_hdr)) {
2056 "sanity fail: off=%lx, sizeof(ip6)=%lx in %s:%d\n",
2057 (u_long)off, (u_long)sizeof(struct ip6_hdr),
2058 __FILE__, __LINE__));
2063 * If there are extra headers between IPv6 and ICMPv6, strip
2064 * off that header first.
2067 if (sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr) > MHLEN)
2068 panic("assumption failed in icmp6_reflect");
2070 if (off > sizeof(struct ip6_hdr)) {
2072 struct ip6_hdr nip6;
2074 l = off - sizeof(struct ip6_hdr);
2075 m_copydata(m, 0, sizeof(nip6), (caddr_t)&nip6);
2077 l = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr);
2079 if ((m = m_pullup(m, l)) == NULL)
2082 bcopy((caddr_t)&nip6, mtod(m, caddr_t), sizeof(nip6));
2083 } else /* off == sizeof(struct ip6_hdr) */ {
2085 l = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr);
2087 if ((m = m_pullup(m, l)) == NULL)
2091 plen = m->m_pkthdr.len - sizeof(struct ip6_hdr);
2092 ip6 = mtod(m, struct ip6_hdr *);
2093 ip6->ip6_nxt = IPPROTO_ICMPV6;
2094 icmp6 = (struct icmp6_hdr *)(ip6 + 1);
2095 type = icmp6->icmp6_type; /* keep type for statistics */
2096 code = icmp6->icmp6_code; /* ditto. */
2101 * If the incoming packet was addressed directly to us (i.e. unicast),
2102 * use dst as the src for the reply.
2103 * The IN6_IFF_NOTREADY case should be VERY rare, but is possible
2104 * (for example) when we encounter an error while forwarding procedure
2105 * destined to a duplicated address of ours.
2107 if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
2108 ia = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */, false);
2109 if (ia != NULL && !(ia->ia6_flags &
2110 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY))) {
2111 src6 = ia->ia_addr.sin6_addr;
2114 if (m->m_pkthdr.rcvif != NULL) {
2115 /* XXX: This may not be the outgoing interface */
2116 hlim = ND_IFINFO(m->m_pkthdr.rcvif)->chlim;
2118 hlim = V_ip6_defhlim;
2124 struct in6_addr dst6;
2128 * This case matches to multicasts, our anycast, or unicasts
2129 * that we do not own. Select a source address based on the
2130 * source address of the erroneous packet.
2132 in6_splitscope(&ip6->ip6_src, &dst6, &scopeid);
2133 error = in6_selectsrc_addr(M_GETFIB(m), &dst6,
2134 scopeid, NULL, &src6, &hlim);
2137 char ip6buf[INET6_ADDRSTRLEN];
2139 "icmp6_reflect: source can't be determined: "
2140 "dst=%s, error=%d\n",
2141 ip6_sprintf(ip6buf, &ip6->ip6_dst), error));
2147 * ip6_input() drops a packet if its src is multicast.
2148 * So, the src is never multicast.
2150 ip6->ip6_dst = ip6->ip6_src;
2151 ip6->ip6_src = *srcp;
2153 ip6->ip6_vfc &= ~IPV6_VERSION_MASK;
2154 ip6->ip6_vfc |= IPV6_VERSION;
2155 ip6->ip6_nxt = IPPROTO_ICMPV6;
2156 ip6->ip6_hlim = hlim;
2158 icmp6->icmp6_cksum = 0;
2159 icmp6->icmp6_cksum = in6_cksum(m, IPPROTO_ICMPV6,
2160 sizeof(struct ip6_hdr), plen);
2163 * XXX option handling
2166 m->m_flags &= ~(M_BCAST|M_MCAST);
2167 m->m_pkthdr.rcvif = NULL;
2168 ip6_output(m, NULL, NULL, 0, NULL, &outif, NULL);
2170 icmp6_ifoutstat_inc(outif, type, code);
2180 icmp6_fasttimo(void)
2187 icmp6_slowtimo(void)
2194 icmp6_redirect_diag(struct in6_addr *src6, struct in6_addr *dst6,
2195 struct in6_addr *tgt6)
2197 static char buf[1024];
2198 char ip6bufs[INET6_ADDRSTRLEN];
2199 char ip6bufd[INET6_ADDRSTRLEN];
2200 char ip6buft[INET6_ADDRSTRLEN];
2201 snprintf(buf, sizeof(buf), "(src=%s dst=%s tgt=%s)",
2202 ip6_sprintf(ip6bufs, src6), ip6_sprintf(ip6bufd, dst6),
2203 ip6_sprintf(ip6buft, tgt6));
2208 icmp6_redirect_input(struct mbuf *m, int off)
2211 struct ip6_hdr *ip6;
2212 struct nd_redirect *nd_rd;
2213 struct in6_addr src6, redtgt6, reddst6;
2214 union nd_opts ndopts;
2215 char ip6buf[INET6_ADDRSTRLEN];
2217 int icmp6len, is_onlink, is_router, lladdrlen;
2220 KASSERT(m->m_pkthdr.rcvif != NULL, ("%s: no rcvif", __func__));
2222 /* XXX if we are router, we don't update route by icmp6 redirect */
2223 if (V_ip6_forwarding)
2225 if (!V_icmp6_rediraccept)
2228 /* RFC 6980: Nodes MUST silently ignore fragments */
2229 if(m->m_flags & M_FRAGMENTED)
2232 ip6 = mtod(m, struct ip6_hdr *);
2233 icmp6len = ntohs(ip6->ip6_plen);
2234 if (m->m_len < off + icmp6len) {
2235 m = m_pullup(m, off + icmp6len);
2237 IP6STAT_INC(ip6s_exthdrtoolong);
2241 ip6 = mtod(m, struct ip6_hdr *);
2242 nd_rd = (struct nd_redirect *)((caddr_t)ip6 + off);
2244 ifp = m->m_pkthdr.rcvif;
2245 redtgt6 = nd_rd->nd_rd_target;
2246 reddst6 = nd_rd->nd_rd_dst;
2248 if (in6_setscope(&redtgt6, ifp, NULL) ||
2249 in6_setscope(&reddst6, ifp, NULL)) {
2254 src6 = ip6->ip6_src;
2255 if (!IN6_IS_ADDR_LINKLOCAL(&src6)) {
2257 "ICMP6 redirect sent from %s rejected; "
2258 "must be from linklocal\n",
2259 ip6_sprintf(ip6buf, &src6)));
2262 if (__predict_false(ip6->ip6_hlim != 255)) {
2263 ICMP6STAT_INC(icp6s_invlhlim);
2265 "ICMP6 redirect sent from %s rejected; "
2266 "hlim=%d (must be 255)\n",
2267 ip6_sprintf(ip6buf, &src6), ip6->ip6_hlim));
2271 /* ip6->ip6_src must be equal to gw for icmp6->icmp6_reddst */
2272 struct nhop_object *nh;
2273 struct in6_addr kdst;
2276 in6_splitscope(&reddst6, &kdst, &scopeid);
2278 nh = fib6_lookup(ifp->if_fib, &kdst, scopeid, 0, 0);
2280 struct in6_addr nh_addr;
2281 nh_addr = ifatoia6(nh->nh_ifa)->ia_addr.sin6_addr;
2282 if ((nh->nh_flags & NHF_GATEWAY) == 0) {
2284 "ICMP6 redirect rejected; no route "
2285 "with inet6 gateway found for redirect dst: %s\n",
2286 icmp6_redirect_diag(&src6, &reddst6, &redtgt6)));
2291 * Embed scope zone id into next hop address.
2293 nh_addr = nh->gw6_sa.sin6_addr;
2295 if (IN6_ARE_ADDR_EQUAL(&src6, &nh_addr) == 0) {
2297 "ICMP6 redirect rejected; "
2298 "not equal to gw-for-src=%s (must be same): "
2300 ip6_sprintf(ip6buf, &nh_addr),
2301 icmp6_redirect_diag(&src6, &reddst6, &redtgt6)));
2306 "ICMP6 redirect rejected; "
2307 "no route found for redirect dst: %s\n",
2308 icmp6_redirect_diag(&src6, &reddst6, &redtgt6)));
2312 if (IN6_IS_ADDR_MULTICAST(&reddst6)) {
2314 "ICMP6 redirect rejected; "
2315 "redirect dst must be unicast: %s\n",
2316 icmp6_redirect_diag(&src6, &reddst6, &redtgt6)));
2320 is_router = is_onlink = 0;
2321 if (IN6_IS_ADDR_LINKLOCAL(&redtgt6))
2322 is_router = 1; /* router case */
2323 if (bcmp(&redtgt6, &reddst6, sizeof(redtgt6)) == 0)
2324 is_onlink = 1; /* on-link destination case */
2325 if (!is_router && !is_onlink) {
2327 "ICMP6 redirect rejected; "
2328 "neither router case nor onlink case: %s\n",
2329 icmp6_redirect_diag(&src6, &reddst6, &redtgt6)));
2333 icmp6len -= sizeof(*nd_rd);
2334 nd6_option_init(nd_rd + 1, icmp6len, &ndopts);
2335 if (nd6_options(&ndopts) < 0) {
2336 nd6log((LOG_INFO, "%s: invalid ND option, rejected: %s\n",
2337 __func__, icmp6_redirect_diag(&src6, &reddst6, &redtgt6)));
2338 /* nd6_options have incremented stats */
2344 if (ndopts.nd_opts_tgt_lladdr) {
2345 lladdr = (char *)(ndopts.nd_opts_tgt_lladdr + 1);
2346 lladdrlen = ndopts.nd_opts_tgt_lladdr->nd_opt_len << 3;
2349 if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
2350 nd6log((LOG_INFO, "%s: lladdrlen mismatch for %s "
2351 "(if %d, icmp6 packet %d): %s\n",
2352 __func__, ip6_sprintf(ip6buf, &redtgt6),
2353 ifp->if_addrlen, lladdrlen - 2,
2354 icmp6_redirect_diag(&src6, &reddst6, &redtgt6)));
2358 /* Validation passed. */
2361 nd6_cache_lladdr(ifp, &redtgt6, lladdr, lladdrlen, ND_REDIRECT,
2362 is_onlink ? ND_REDIRECT_ONLINK : ND_REDIRECT_ROUTER);
2365 * Install a gateway route in the better-router case or an interface
2366 * route in the on-link-destination case.
2369 struct sockaddr_in6 sdst;
2370 struct sockaddr_in6 sgw;
2371 struct sockaddr_in6 ssrc;
2372 struct sockaddr *gw;
2376 bzero(&sdst, sizeof(sdst));
2377 bzero(&ssrc, sizeof(ssrc));
2378 sdst.sin6_family = ssrc.sin6_family = AF_INET6;
2379 sdst.sin6_len = ssrc.sin6_len = sizeof(struct sockaddr_in6);
2380 bcopy(&reddst6, &sdst.sin6_addr, sizeof(struct in6_addr));
2381 bcopy(&src6, &ssrc.sin6_addr, sizeof(struct in6_addr));
2384 bzero(&sgw, sizeof(sgw));
2385 sgw.sin6_family = AF_INET6;
2386 sgw.sin6_len = sizeof(struct sockaddr_in6);
2387 bcopy(&redtgt6, &sgw.sin6_addr,
2388 sizeof(struct in6_addr));
2389 gw = (struct sockaddr *)&sgw;
2390 rt_flags |= RTF_GATEWAY;
2392 gw = ifp->if_addr->ifa_addr;
2393 for (fibnum = 0; fibnum < rt_numfibs; fibnum++)
2394 rib_add_redirect(fibnum, (struct sockaddr *)&sdst, gw,
2395 (struct sockaddr *)&ssrc, ifp, rt_flags,
2396 V_icmp6_redirtimeout);
2398 /* finally update cached route in each socket via pfctlinput */
2400 struct sockaddr_in6 sdst;
2402 bzero(&sdst, sizeof(sdst));
2403 sdst.sin6_family = AF_INET6;
2404 sdst.sin6_len = sizeof(struct sockaddr_in6);
2405 bcopy(&reddst6, &sdst.sin6_addr, sizeof(struct in6_addr));
2406 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&sdst);
2414 ICMP6STAT_INC(icp6s_badredirect);
2419 icmp6_redirect_output(struct mbuf *m0, struct nhop_object *nh)
2421 struct ifnet *ifp; /* my outgoing interface */
2422 struct in6_addr *ifp_ll6;
2423 struct in6_addr *router_ll6;
2424 struct ip6_hdr *sip6; /* m0 as struct ip6_hdr */
2425 struct mbuf *m = NULL; /* newly allocated one */
2427 struct ip6_hdr *ip6; /* m as struct ip6_hdr */
2428 struct nd_redirect *nd_rd;
2429 struct llentry *ln = NULL;
2432 struct ifnet *outif = NULL;
2433 struct sockaddr_in6 src_sa;
2435 icmp6_errcount(ND_REDIRECT, 0);
2437 /* if we are not router, we don't send icmp6 redirect */
2438 if (!V_ip6_forwarding)
2442 if (!m0 || !nh || !(NH_IS_VALID(nh)) || !(ifp = nh->nh_ifp))
2447 * the source address must identify a neighbor, and
2448 * the destination address must not be a multicast address
2449 * [RFC 2461, sec 8.2]
2451 sip6 = mtod(m0, struct ip6_hdr *);
2452 bzero(&src_sa, sizeof(src_sa));
2453 src_sa.sin6_family = AF_INET6;
2454 src_sa.sin6_len = sizeof(src_sa);
2455 src_sa.sin6_addr = sip6->ip6_src;
2456 if (nd6_is_addr_neighbor(&src_sa, ifp) == 0)
2458 if (IN6_IS_ADDR_MULTICAST(&sip6->ip6_dst))
2459 goto fail; /* what should we do here? */
2462 if (icmp6_ratelimit(&sip6->ip6_src, ND_REDIRECT, 0))
2466 * Since we are going to append up to 1280 bytes (= IPV6_MMTU),
2467 * we almost always ask for an mbuf cluster for simplicity.
2468 * (MHLEN < IPV6_MMTU is almost always true)
2470 #if IPV6_MMTU >= MCLBYTES
2471 # error assumption failed about IPV6_MMTU and MCLBYTES
2473 m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
2476 M_SETFIB(m, M_GETFIB(m0));
2477 maxlen = M_TRAILINGSPACE(m);
2478 maxlen = min(IPV6_MMTU, maxlen);
2479 /* just for safety */
2480 if (maxlen < sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr) +
2481 ((sizeof(struct nd_opt_hdr) + ifp->if_addrlen + 7) & ~7)) {
2486 /* get ip6 linklocal address for ifp(my outgoing interface). */
2487 struct in6_ifaddr *ia;
2488 if ((ia = in6ifa_ifpforlinklocal(ifp,
2490 IN6_IFF_ANYCAST)) == NULL)
2492 ifp_ll6 = &ia->ia_addr.sin6_addr;
2493 /* XXXRW: reference released prematurely. */
2494 ifa_free(&ia->ia_ifa);
2497 /* get ip6 linklocal address for the router. */
2498 if (nh->nh_flags & NHF_GATEWAY) {
2499 struct sockaddr_in6 *sin6;
2501 router_ll6 = &sin6->sin6_addr;
2502 if (!IN6_IS_ADDR_LINKLOCAL(router_ll6))
2503 router_ll6 = (struct in6_addr *)NULL;
2505 router_ll6 = (struct in6_addr *)NULL;
2508 ip6 = mtod(m, struct ip6_hdr *);
2510 ip6->ip6_vfc &= ~IPV6_VERSION_MASK;
2511 ip6->ip6_vfc |= IPV6_VERSION;
2512 /* ip6->ip6_plen will be set later */
2513 ip6->ip6_nxt = IPPROTO_ICMPV6;
2514 ip6->ip6_hlim = 255;
2515 /* ip6->ip6_src must be linklocal addr for my outgoing if. */
2516 bcopy(ifp_ll6, &ip6->ip6_src, sizeof(struct in6_addr));
2517 bcopy(&sip6->ip6_src, &ip6->ip6_dst, sizeof(struct in6_addr));
2520 nd_rd = (struct nd_redirect *)(ip6 + 1);
2521 nd_rd->nd_rd_type = ND_REDIRECT;
2522 nd_rd->nd_rd_code = 0;
2523 nd_rd->nd_rd_reserved = 0;
2524 if (nh->nh_flags & NHF_GATEWAY) {
2526 * nd_rd->nd_rd_target must be a link-local address in
2527 * better router cases.
2531 bcopy(router_ll6, &nd_rd->nd_rd_target,
2532 sizeof(nd_rd->nd_rd_target));
2533 bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_dst,
2534 sizeof(nd_rd->nd_rd_dst));
2536 /* make sure redtgt == reddst */
2537 bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_target,
2538 sizeof(nd_rd->nd_rd_target));
2539 bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_dst,
2540 sizeof(nd_rd->nd_rd_dst));
2543 p = (u_char *)(nd_rd + 1);
2549 /* target lladdr option */
2551 struct nd_opt_hdr *nd_opt;
2554 ln = nd6_lookup(router_ll6, LLE_SF(AF_INET6, 0), ifp);
2558 len = sizeof(*nd_opt) + ifp->if_addrlen;
2559 len = (len + 7) & ~7; /* round by 8 */
2561 if (len + (p - (u_char *)ip6) > maxlen)
2564 if (ln->la_flags & LLE_VALID) {
2565 nd_opt = (struct nd_opt_hdr *)p;
2566 nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR;
2567 nd_opt->nd_opt_len = len >> 3;
2568 lladdr = (char *)(nd_opt + 1);
2569 bcopy(ln->ll_addr, lladdr, ifp->if_addrlen);
2577 m->m_pkthdr.len = m->m_len = p - (u_char *)ip6;
2579 /* just to be safe */
2580 #ifdef M_DECRYPTED /*not openbsd*/
2581 if (m0->m_flags & M_DECRYPTED)
2584 if (p - (u_char *)ip6 > maxlen)
2588 /* redirected header option */
2590 struct nd_opt_rd_hdr *nd_opt_rh;
2593 * compute the maximum size for icmp6 redirect header option.
2594 * XXX room for auth header?
2596 len = maxlen - (p - (u_char *)ip6);
2599 /* This is just for simplicity. */
2600 if (m0->m_pkthdr.len != m0->m_len) {
2602 m_freem(m0->m_next);
2605 m0->m_pkthdr.len = m0->m_len;
2609 * Redirected header option spec (RFC2461 4.6.3) talks nothing
2610 * about padding/truncate rule for the original IP packet.
2611 * From the discussion on IPv6imp in Feb 1999,
2612 * the consensus was:
2613 * - "attach as much as possible" is the goal
2614 * - pad if not aligned (original size can be guessed by
2615 * original ip6 header)
2616 * Following code adds the padding if it is simple enough,
2617 * and truncates if not.
2619 if (m0->m_next || m0->m_pkthdr.len != m0->m_len)
2620 panic("assumption failed in %s:%d", __FILE__,
2623 if (len - sizeof(*nd_opt_rh) < m0->m_pkthdr.len) {
2624 /* not enough room, truncate */
2625 m0->m_pkthdr.len = m0->m_len = len -
2628 /* enough room, pad or truncate */
2631 extra = m0->m_pkthdr.len % 8;
2633 /* pad if easy enough, truncate if not */
2634 if (8 - extra <= M_TRAILINGSPACE(m0)) {
2636 m0->m_len += (8 - extra);
2637 m0->m_pkthdr.len += (8 - extra);
2640 m0->m_pkthdr.len -= extra;
2644 len = m0->m_pkthdr.len + sizeof(*nd_opt_rh);
2645 m0->m_pkthdr.len = m0->m_len = len -
2649 nd_opt_rh = (struct nd_opt_rd_hdr *)p;
2650 bzero(nd_opt_rh, sizeof(*nd_opt_rh));
2651 nd_opt_rh->nd_opt_rh_type = ND_OPT_REDIRECTED_HEADER;
2652 nd_opt_rh->nd_opt_rh_len = len >> 3;
2653 p += sizeof(*nd_opt_rh);
2654 m->m_pkthdr.len = m->m_len = p - (u_char *)ip6;
2656 /* connect m0 to m */
2657 m_tag_delete_chain(m0, NULL);
2658 m0->m_flags &= ~M_PKTHDR;
2660 m->m_pkthdr.len = m->m_len + m0->m_len;
2669 /* XXX: clear embedded link IDs in the inner header */
2670 in6_clearscope(&sip6->ip6_src);
2671 in6_clearscope(&sip6->ip6_dst);
2672 in6_clearscope(&nd_rd->nd_rd_target);
2673 in6_clearscope(&nd_rd->nd_rd_dst);
2675 ip6->ip6_plen = htons(m->m_pkthdr.len - sizeof(struct ip6_hdr));
2677 nd_rd->nd_rd_cksum = 0;
2678 nd_rd->nd_rd_cksum = in6_cksum(m, IPPROTO_ICMPV6,
2679 sizeof(*ip6), ntohs(ip6->ip6_plen));
2681 if (send_sendso_input_hook != NULL) {
2682 mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short),
2686 *(unsigned short *)(mtag + 1) = nd_rd->nd_rd_type;
2687 m_tag_prepend(m, mtag);
2690 /* send the packet to outside... */
2691 ip6_output(m, NULL, NULL, 0, NULL, &outif, NULL);
2693 icmp6_ifstat_inc(outif, ifs6_out_msg);
2694 icmp6_ifstat_inc(outif, ifs6_out_redirect);
2696 ICMP6STAT_INC(icp6s_outhist[ND_REDIRECT]);
2708 * ICMPv6 socket option processing.
2711 icmp6_ctloutput(struct socket *so, struct sockopt *sopt)
2715 struct inpcb *inp = sotoinpcb(so);
2716 int level, op, optname;
2719 level = sopt->sopt_level;
2720 op = sopt->sopt_dir;
2721 optname = sopt->sopt_name;
2722 optlen = sopt->sopt_valsize;
2724 level = op = optname = optlen = 0;
2726 if (level != IPPROTO_ICMPV6) {
2735 struct icmp6_filter ic6f;
2737 if (optlen != sizeof(ic6f)) {
2741 error = sooptcopyin(sopt, &ic6f, optlen, optlen);
2744 *inp->in6p_icmp6filt = ic6f;
2751 error = ENOPROTOOPT;
2760 struct icmp6_filter ic6f;
2763 ic6f = *inp->in6p_icmp6filt;
2765 error = sooptcopyout(sopt, &ic6f, sizeof(ic6f));
2770 error = ENOPROTOOPT;
2780 * Perform rate limit check.
2781 * Returns 0 if it is okay to send the icmp6 packet.
2782 * Returns 1 if the router SHOULD NOT send this icmp6 packet due to rate
2785 * XXX per-destination/type check necessary?
2787 * dst - not used at this moment
2788 * type - not used at this moment
2789 * code - not used at this moment
2792 icmp6_ratelimit(const struct in6_addr *dst, const int type,
2797 ret = 0; /* okay to send */
2800 if (!ppsratecheck(&V_icmp6errppslim_last, &V_icmp6errpps_count,
2801 V_icmp6errppslim)) {
2802 /* The packet is subject to rate limit */