icmp6.c revision 1.18
1/*	$NetBSD: icmp6.c,v 1.18 2000/01/16 18:06:03 itojun Exp $	*/
2
3/*
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
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.
18 *
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
29 * SUCH DAMAGE.
30 */
31
32/*
33 * Copyright (c) 1982, 1986, 1988, 1993
34 *	The Regents of the University of California.  All rights reserved.
35 *
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
38 * are met:
39 * 1. Redistributions of source code must retain the above copyright
40 *    notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 *    notice, this list of conditions and the following disclaimer in the
43 *    documentation and/or other materials provided with the distribution.
44 * 3. All advertising materials mentioning features or use of this software
45 *    must display the following acknowledgement:
46 *	This product includes software developed by the University of
47 *	California, Berkeley and its contributors.
48 * 4. Neither the name of the University nor the names of its contributors
49 *    may be used to endorse or promote products derived from this software
50 *    without specific prior written permission.
51 *
52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * SUCH DAMAGE.
63 *
64 *	@(#)ip_icmp.c	8.2 (Berkeley) 1/4/94
65 */
66
67#include "opt_inet.h"
68#include "opt_ipsec.h"
69
70#include <sys/param.h>
71#include <sys/systm.h>
72#include <sys/malloc.h>
73#include <sys/mbuf.h>
74#include <sys/protosw.h>
75#include <sys/socket.h>
76#include <sys/socketvar.h>
77#include <sys/time.h>
78#include <sys/kernel.h>
79#include <sys/syslog.h>
80#include <sys/domain.h>
81
82#include <net/if.h>
83#include <net/route.h>
84#include <net/if_dl.h>
85#include <net/if_types.h>
86
87#include <netinet/in.h>
88#include <netinet/in_var.h>
89#include <netinet6/ip6.h>
90#include <netinet6/ip6_var.h>
91#include <netinet6/icmp6.h>
92#include <netinet6/mld6_var.h>
93#include <netinet6/in6_pcb.h>
94#include <netinet6/nd6.h>
95#include <netinet6/in6_ifattach.h>
96#include <netinet6/ip6protosw.h>
97
98
99#ifdef IPSEC
100#include <netinet6/ipsec.h>
101#include <netkey/key.h>
102#include <netkey/key_debug.h>
103#endif
104
105#include "faith.h"
106
107#include <net/net_osdep.h>
108
109extern struct domain inet6domain;
110extern struct ip6protosw inet6sw[];
111extern u_char ip6_protox[];
112
113struct icmp6stat icmp6stat;
114
115extern struct in6pcb rawin6pcb;
116extern u_int icmp6errratelim;
117extern int icmp6_nodeinfo;
118static struct rttimer_queue *icmp6_mtudisc_timeout_q = NULL;
119extern int pmtu_expire;
120
121static int icmp6_rip6_input __P((struct mbuf **, int));
122static int icmp6_ratelimit __P((const struct in6_addr *, const int, const int));
123static const char *icmp6_redirect_diag __P((struct in6_addr *,
124	struct in6_addr *, struct in6_addr *));
125static struct mbuf * ni6_input __P((struct mbuf *, int));
126static int ni6_addrs __P((struct icmp6_nodeinfo *, struct mbuf *,
127			  struct ifnet **));
128static int ni6_store_addrs __P((struct icmp6_nodeinfo *, struct icmp6_nodeinfo *,
129				struct ifnet *, int));
130static struct rtentry *icmp6_mtudisc_clone __P((struct sockaddr *));
131static void icmp6_mtudisc_timeout __P((struct rtentry *, struct rttimer *));
132
133#ifdef COMPAT_RFC1885
134static struct route_in6 icmp6_reflect_rt;
135#endif
136static struct timeval icmp6_nextsend = {0, 0};
137
138void
139icmp6_init()
140{
141	mld6_init();
142	icmp6_mtudisc_timeout_q = rt_timer_queue_create(pmtu_expire);
143}
144
145/*
146 * Generate an error packet of type error in response to bad IP6 packet.
147 */
148void
149icmp6_error(m, type, code, param)
150	struct mbuf *m;
151	int type, code, param;
152{
153	struct ip6_hdr *oip6, *nip6;
154	struct icmp6_hdr *icmp6;
155	u_int preplen;
156	int off;
157	u_char nxt;
158
159	icmp6stat.icp6s_error++;
160
161#ifdef M_DECRYPTED	/*not openbsd*/
162	if (m->m_flags & M_DECRYPTED)
163		goto freeit;
164#endif
165
166#ifndef PULLDOWN_TEST
167	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), );
168#else
169	if (m->m_len < sizeof(struct ip6_hdr)) {
170		m = m_pullup(m, sizeof(struct ip6_hdr));
171		if (m == NULL)
172			return;
173	}
174#endif
175	oip6 = mtod(m, struct ip6_hdr *);
176
177	/*
178	 * Multicast destination check. For unrecognized option errors,
179	 * this check has already done in ip6_unknown_opt(), so we can
180	 * check only for other errors.
181	 */
182	if ((m->m_flags & (M_BCAST|M_MCAST) ||
183	     IN6_IS_ADDR_MULTICAST(&oip6->ip6_dst)) &&
184	    (type != ICMP6_PACKET_TOO_BIG &&
185	     (type != ICMP6_PARAM_PROB ||
186	      code != ICMP6_PARAMPROB_OPTION)))
187		goto freeit;
188
189	/* Source address check. XXX: the case of anycast source? */
190	if (IN6_IS_ADDR_UNSPECIFIED(&oip6->ip6_src) ||
191	    IN6_IS_ADDR_MULTICAST(&oip6->ip6_src))
192		goto freeit;
193
194	/*
195	 * If the erroneous packet is also an ICMP error, discard it.
196	 */
197	off = sizeof(struct ip6_hdr);
198	nxt = oip6->ip6_nxt;
199	while (1) {		/* XXX: should avoid inf. loop explicitly? */
200		struct ip6_ext *ip6e;
201		struct icmp6_hdr *icp;
202
203		switch(nxt) {
204		case IPPROTO_IPV6:
205		case IPPROTO_IPV4:
206		case IPPROTO_UDP:
207		case IPPROTO_TCP:
208		case IPPROTO_ESP:
209		case IPPROTO_IPCOMP:
210		case IPPROTO_FRAGMENT:
211			/*
212			 * ICMPv6 error must not be fragmented.
213			 * XXX: but can we trust the sender?
214			 */
215		default:
216			/* What if unknown header followed by ICMP error? */
217			goto generate;
218		case IPPROTO_ICMPV6:
219#ifndef PULLDOWN_TEST
220			IP6_EXTHDR_CHECK(m, 0, off + sizeof(struct icmp6_hdr), );
221			icp = (struct icmp6_hdr *)(mtod(m, caddr_t) + off);
222#else
223			IP6_EXTHDR_GET(icp, struct icmp6_hdr *, m, off,
224				sizeof(*icp));
225			if (icp == NULL) {
226				icmp6stat.icp6s_tooshort++;
227				return;
228			}
229#endif
230			if (icp->icmp6_type < ICMP6_ECHO_REQUEST
231			 || icp->icmp6_type == ND_REDIRECT) {
232				/*
233				 * ICMPv6 error
234				 * Special case: for redirect (which is
235				 * informational) we must not send icmp6 error.
236				 */
237				icmp6stat.icp6s_canterror++;
238				goto freeit;
239			} else {
240				/* ICMPv6 informational */
241				goto generate;
242			}
243		case IPPROTO_HOPOPTS:
244		case IPPROTO_DSTOPTS:
245		case IPPROTO_ROUTING:
246		case IPPROTO_AH:
247#ifndef PULLDOWN_TEST
248			IP6_EXTHDR_CHECK(m, 0, off + sizeof(struct ip6_ext), );
249			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
250#else
251			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off,
252				sizeof(*ip6e));
253			if (ip6e == NULL) {
254				/*XXX stat */
255				return;
256			}
257#endif
258			if (nxt == IPPROTO_AH)
259				off += (ip6e->ip6e_len + 2) << 2;
260			else
261				off += (ip6e->ip6e_len + 1) << 3;
262			nxt = ip6e->ip6e_nxt;
263			break;
264		}
265	}
266
267  freeit:
268	/*
269	 * If we can't tell wheter or not we can generate ICMP6, free it.
270	 */
271	m_freem(m);
272	return;
273
274  generate:
275	oip6 = mtod(m, struct ip6_hdr *); /* adjust pointer */
276
277	/* Finally, do rate limitation check. */
278	if (icmp6_ratelimit(&oip6->ip6_src, type, code)) {
279		icmp6stat.icp6s_toofreq++;
280		goto freeit;
281	}
282
283	/*
284	 * OK, ICMP6 can be generated.
285	 */
286
287	if (m->m_pkthdr.len >= ICMPV6_PLD_MAXLEN)
288		m_adj(m, ICMPV6_PLD_MAXLEN - m->m_pkthdr.len);
289
290	preplen = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr);
291	M_PREPEND(m, preplen, M_DONTWAIT);
292	if (m && m->m_len < preplen)
293		m = m_pullup(m, preplen);
294	if (m == NULL) {
295		printf("ENOBUFS in icmp6_error %d\n", __LINE__);
296		return;
297	}
298
299	nip6 = mtod(m, struct ip6_hdr *);
300	nip6->ip6_src  = oip6->ip6_src;
301	nip6->ip6_dst  = oip6->ip6_dst;
302
303	if (IN6_IS_SCOPE_LINKLOCAL(&oip6->ip6_src))
304		oip6->ip6_src.s6_addr16[1] = 0;
305	if (IN6_IS_SCOPE_LINKLOCAL(&oip6->ip6_dst))
306		oip6->ip6_dst.s6_addr16[1] = 0;
307
308	icmp6 = (struct icmp6_hdr *)(nip6 + 1);
309	icmp6->icmp6_type = type;
310	icmp6->icmp6_code = code;
311	icmp6->icmp6_pptr = htonl((u_int32_t)param);
312
313	icmp6stat.icp6s_outhist[type]++;
314	icmp6_reflect(m, sizeof(struct ip6_hdr)); /*header order: IPv6 - ICMPv6*/
315}
316
317/*
318 * Process a received ICMP6 message.
319 */
320int
321icmp6_input(mp, offp, proto)
322	struct mbuf **mp;
323	int *offp, proto;
324{
325	struct mbuf *m = *mp, *n;
326	struct ip6_hdr *ip6, *nip6;
327	struct icmp6_hdr *icmp6, *nicmp6;
328	int off = *offp;
329	int icmp6len = m->m_pkthdr.len - *offp;
330	int code, sum, noff;
331	struct sockaddr_in6 icmp6src;
332
333#ifndef PULLDOWN_TEST
334	IP6_EXTHDR_CHECK(m, off, sizeof(struct icmp6_hdr), IPPROTO_DONE);
335	/* m might change if M_LOOP. So, call mtod after this */
336#endif
337
338	/*
339	 * Locate icmp6 structure in mbuf, and check
340	 * that not corrupted and of at least minimum length
341	 */
342
343	ip6 = mtod(m, struct ip6_hdr *);
344	if (icmp6len < sizeof(struct icmp6_hdr)) {
345		icmp6stat.icp6s_tooshort++;
346		goto freeit;
347	}
348
349	/*
350	 * calculate the checksum
351	 */
352#ifndef PULLDOWN_TEST
353	icmp6 = (struct icmp6_hdr *)((caddr_t)ip6 + off);
354#else
355	IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off, sizeof(*icmp6));
356	if (icmp6 == NULL) {
357		icmp6stat.icp6s_tooshort++;
358		return IPPROTO_DONE;
359	}
360#endif
361	code = icmp6->icmp6_code;
362
363	if ((sum = in6_cksum(m, IPPROTO_ICMPV6, off, icmp6len)) != 0) {
364		log(LOG_ERR,
365		    "ICMP6 checksum error(%d|%x) %s\n",
366		    icmp6->icmp6_type,
367		    sum,
368		    ip6_sprintf(&ip6->ip6_src));
369		icmp6stat.icp6s_checksum++;
370		goto freeit;
371	}
372
373#if defined(NFAITH) && 0 < NFAITH
374	if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
375		/*
376		 * Deliver very specific ICMP6 type only.
377		 * This is important to deilver TOOBIG.  Otherwise PMTUD
378		 * will not work.
379		 */
380		switch (icmp6->icmp6_type) {
381		case ICMP6_DST_UNREACH:
382		case ICMP6_PACKET_TOO_BIG:
383		case ICMP6_TIME_EXCEEDED:
384			break;
385		default:
386			goto freeit;
387		}
388	}
389#endif
390
391#ifdef IPSEC
392	/* drop it if it does not match the default policy */
393	if (ipsec6_in_reject(m, NULL)) {
394		ipsecstat.in_polvio++;
395		goto freeit;
396	}
397#endif
398
399	icmp6stat.icp6s_inhist[icmp6->icmp6_type]++;
400	icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_msg);
401	if (icmp6->icmp6_type < ICMP6_INFOMSG_MASK)
402		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_error);
403
404	switch (icmp6->icmp6_type) {
405
406	case ICMP6_DST_UNREACH:
407		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_dstunreach);
408		switch (code) {
409		case ICMP6_DST_UNREACH_NOROUTE:
410			code = PRC_UNREACH_NET;
411			break;
412		case ICMP6_DST_UNREACH_ADMIN:
413			icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_adminprohib);
414		case ICMP6_DST_UNREACH_ADDR:
415			code = PRC_UNREACH_HOST;
416			break;
417		case ICMP6_DST_UNREACH_NOTNEIGHBOR:
418			code = PRC_UNREACH_SRCFAIL;
419			break;
420		case ICMP6_DST_UNREACH_NOPORT:
421			code = PRC_UNREACH_PORT;
422			break;
423		default:
424			goto badcode;
425		}
426		goto deliver;
427		break;
428
429	case ICMP6_PACKET_TOO_BIG:
430		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_pkttoobig);
431		if (code != 0)
432			goto badcode;
433	    {
434		u_int mtu = ntohl(icmp6->icmp6_mtu);
435		struct rtentry *rt = NULL;
436		struct sockaddr_in6 sin6;
437
438		if (icmp6len < sizeof(struct icmp6_hdr) + sizeof(struct ip6_hdr)) {
439			icmp6stat.icp6s_tooshort++;
440			goto freeit;
441		}
442#ifndef PULLDOWN_TEST
443		IP6_EXTHDR_CHECK(m, off,
444			sizeof(struct icmp6_hdr) + sizeof(struct ip6_hdr),
445			IPPROTO_DONE);
446		icmp6 = (struct icmp6_hdr *)(mtod(m, caddr_t) + off);
447#else
448		IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off,
449			sizeof(*icmp6) + sizeof(struct ip6_hdr));
450		if (icmp6 == NULL) {
451			icmp6stat.icp6s_tooshort++;
452			return IPPROTO_DONE;
453		}
454#endif
455		code = PRC_MSGSIZE;
456		bzero(&sin6, sizeof(sin6));
457		sin6.sin6_family = PF_INET6;
458		sin6.sin6_len = sizeof(struct sockaddr_in6);
459		sin6.sin6_addr = ((struct ip6_hdr *)(icmp6 + 1))->ip6_dst;
460		rt = rtalloc1((struct sockaddr *)&sin6, 1);	/*clone*/
461		if (!rt || (rt->rt_flags & RTF_HOST) == 0) {
462			if (rt)
463				RTFREE(rt);
464			rt = icmp6_mtudisc_clone((struct sockaddr *)&sin6);
465		}
466
467		if (rt && (rt->rt_flags & RTF_HOST)
468		    && !(rt->rt_rmx.rmx_locks & RTV_MTU)) {
469			if (mtu < IPV6_MMTU) {
470				/* xxx */
471				rt->rt_rmx.rmx_locks |= RTV_MTU;
472			} else if (mtu < rt->rt_ifp->if_mtu &&
473				   rt->rt_rmx.rmx_mtu > mtu) {
474				rt->rt_rmx.rmx_mtu = mtu;
475			}
476		}
477		if (rt)
478			RTFREE(rt);
479
480		goto deliver;
481	    }
482		break;
483
484	case ICMP6_TIME_EXCEEDED:
485		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_timeexceed);
486		switch (code) {
487		case ICMP6_TIME_EXCEED_TRANSIT:
488		case ICMP6_TIME_EXCEED_REASSEMBLY:
489			code += PRC_TIMXCEED_INTRANS;
490			break;
491		default:
492			goto badcode;
493		}
494		goto deliver;
495		break;
496
497	case ICMP6_PARAM_PROB:
498		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_paramprob);
499		switch (code) {
500		case ICMP6_PARAMPROB_NEXTHEADER:
501			code = PRC_UNREACH_PROTOCOL;
502			break;
503		case ICMP6_PARAMPROB_HEADER:
504		case ICMP6_PARAMPROB_OPTION:
505			code = PRC_PARAMPROB;
506			break;
507		default:
508			goto badcode;
509		}
510		goto deliver;
511		break;
512
513	case ICMP6_ECHO_REQUEST:
514		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_echo);
515		if (code != 0)
516			goto badcode;
517		if ((n = m_copy(m, 0, M_COPYALL)) == NULL) {
518			/* Give up remote */
519			break;
520		}
521		if ((n->m_flags & M_EXT) != 0
522		 || n->m_len < off + sizeof(struct icmp6_hdr)) {
523			struct mbuf *n0 = n;
524
525			/*
526			 * Prepare an internal mbuf. m_pullup() doesn't
527			 * always copy the length we specified.
528			 */
529			MGETHDR(n, M_DONTWAIT, n0->m_type);
530			if (n == NULL) {
531				/* Give up remote */
532				m_freem(n0);
533				break;
534			}
535			M_COPY_PKTHDR(n, n0);
536			/*
537			 * Copy IPv6 and ICMPv6 only.
538			 */
539			nip6 = mtod(n, struct ip6_hdr *);
540			bcopy(ip6, nip6, sizeof(struct ip6_hdr));
541			nicmp6 = (struct icmp6_hdr *)(nip6 + 1);
542			bcopy(icmp6, nicmp6, sizeof(struct icmp6_hdr));
543			noff = sizeof(struct ip6_hdr);
544			n->m_pkthdr.len = n->m_len =
545				noff + sizeof(struct icmp6_hdr);
546			/*
547			 * Adjust mbuf. ip6_plen will be adjusted in
548			 * ip6_output().
549			 */
550			m_adj(n0, off + sizeof(struct icmp6_hdr));
551			n->m_pkthdr.len += n0->m_pkthdr.len;
552			n->m_next = n0;
553			n0->m_flags &= ~M_PKTHDR;
554		} else {
555			nip6 = mtod(n, struct ip6_hdr *);
556			nicmp6 = (struct icmp6_hdr *)((caddr_t)nip6 + off);
557			noff = off;
558		}
559		nicmp6->icmp6_type = ICMP6_ECHO_REPLY;
560		nicmp6->icmp6_code = 0;
561		if (n) {
562			icmp6stat.icp6s_reflect++;
563			icmp6stat.icp6s_outhist[ICMP6_ECHO_REPLY]++;
564			icmp6_reflect(n, noff);
565		}
566		break;
567
568	case ICMP6_ECHO_REPLY:
569		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_echoreply);
570		if (code != 0)
571			goto badcode;
572		break;
573
574	case MLD6_LISTENER_QUERY:
575	case MLD6_LISTENER_REPORT:
576		if (icmp6len < sizeof(struct mld6_hdr))
577			goto badlen;
578		if (icmp6->icmp6_type == MLD6_LISTENER_QUERY) /* XXX: ugly... */
579			icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mldquery);
580		else
581			icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mldreport);
582		IP6_EXTHDR_CHECK(m, off, icmp6len, IPPROTO_DONE);
583		mld6_input(m, off);
584		/* m stays. */
585		break;
586
587	case MLD6_LISTENER_DONE:
588		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mlddone);
589		if (icmp6len < sizeof(struct mld6_hdr))	/* necessary? */
590			goto badlen;
591		break;		/* nothing to be done in kernel */
592
593	case MLD6_MTRACE_RESP:
594	case MLD6_MTRACE:
595		/* XXX: these two are experimental. not officially defind. */
596		/* XXX: per-interface statistics? */
597		break;		/* just pass it to the userland daemon */
598
599	case ICMP6_WRUREQUEST:	/* ICMP6_FQDN_QUERY */
600	    {
601		enum { WRU, FQDN } mode;
602
603		if (code != 0)
604			goto badcode;
605		if (!icmp6_nodeinfo)
606			break;
607
608		if (icmp6len == sizeof(struct icmp6_hdr) + 4)
609			mode = WRU;
610		else if (icmp6len >= sizeof(struct icmp6_hdr) + 8) /* XXX */
611			mode = FQDN;
612		else
613			goto badlen;
614
615		if (mode == FQDN) {
616#ifndef PULLDOWN_TEST
617			IP6_EXTHDR_CHECK(m, off, sizeof(struct icmp6_nodeinfo),
618					 IPPROTO_DONE);
619#endif
620			n = m_copy(m, 0, M_COPYALL);
621			if (n)
622				n = ni6_input(n, off);
623			if (n)
624				noff = sizeof(struct ip6_hdr);
625		} else {
626			u_char *p;
627
628			MGETHDR(n, M_DONTWAIT, m->m_type);
629			if (n == NULL) {
630				/* Give up remote */
631				break;
632			}
633			/*
634			 * Copy IPv6 and ICMPv6 only.
635			 */
636			nip6 = mtod(n, struct ip6_hdr *);
637			bcopy(ip6, nip6, sizeof(struct ip6_hdr));
638			nicmp6 = (struct icmp6_hdr *)(nip6 + 1);
639			bcopy(icmp6, nicmp6, sizeof(struct icmp6_hdr));
640			p = (u_char *)(nicmp6 + 1);
641			bzero(p, 4);
642			bcopy(hostname, p + 4, hostnamelen);
643			noff = sizeof(struct ip6_hdr);
644			M_COPY_PKTHDR(n, m); /* just for recvif */
645			n->m_pkthdr.len = n->m_len = sizeof(struct ip6_hdr) +
646				sizeof(struct icmp6_hdr) + 4 + hostnamelen;
647			nicmp6->icmp6_type = ICMP6_WRUREPLY;
648			nicmp6->icmp6_code = 0;
649		}
650#undef hostnamelen
651		if (n) {
652			icmp6stat.icp6s_reflect++;
653			icmp6stat.icp6s_outhist[ICMP6_WRUREPLY]++;
654			icmp6_reflect(n, noff);
655		}
656		break;
657	    }
658
659	case ICMP6_WRUREPLY:
660		if (code != 0)
661			goto badcode;
662		break;
663
664	case ND_ROUTER_SOLICIT:
665		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_routersolicit);
666		if (code != 0)
667			goto badcode;
668		if (icmp6len < sizeof(struct nd_router_solicit))
669			goto badlen;
670		IP6_EXTHDR_CHECK(m, off, icmp6len, IPPROTO_DONE);
671		nd6_rs_input(m, off, icmp6len);
672		/* m stays. */
673		break;
674
675	case ND_ROUTER_ADVERT:
676		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_routeradvert);
677		if (code != 0)
678			goto badcode;
679		if (icmp6len < sizeof(struct nd_router_advert))
680			goto badlen;
681		IP6_EXTHDR_CHECK(m, off, icmp6len, IPPROTO_DONE);
682		nd6_ra_input(m, off, icmp6len);
683		/* m stays. */
684		break;
685
686	case ND_NEIGHBOR_SOLICIT:
687		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_neighborsolicit);
688		if (code != 0)
689			goto badcode;
690		if (icmp6len < sizeof(struct nd_neighbor_solicit))
691			goto badlen;
692		IP6_EXTHDR_CHECK(m, off, icmp6len, IPPROTO_DONE);
693		nd6_ns_input(m, off, icmp6len);
694		/* m stays. */
695		break;
696
697	case ND_NEIGHBOR_ADVERT:
698		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_neighboradvert);
699		if (code != 0)
700			goto badcode;
701		if (icmp6len < sizeof(struct nd_neighbor_advert))
702			goto badlen;
703		IP6_EXTHDR_CHECK(m, off, icmp6len, IPPROTO_DONE);
704		nd6_na_input(m, off, icmp6len);
705		/* m stays. */
706		break;
707
708	case ND_REDIRECT:
709		icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_redirect);
710		if (code != 0)
711			goto badcode;
712		if (icmp6len < sizeof(struct nd_redirect))
713			goto badlen;
714		icmp6_redirect_input(m, off);
715		/* m stays. */
716		break;
717
718	case ICMP6_ROUTER_RENUMBERING:
719		if (code != ICMP6_ROUTER_RENUMBERING_COMMAND &&
720		    code != ICMP6_ROUTER_RENUMBERING_RESULT)
721			goto badcode;
722		if (icmp6len < sizeof(struct icmp6_router_renum))
723			goto badlen;
724		break;
725
726	default:
727		printf("icmp6_input: unknown type %d(src=%s, dst=%s, ifid=%d)\n",
728		       icmp6->icmp6_type, ip6_sprintf(&ip6->ip6_src),
729		       ip6_sprintf(&ip6->ip6_dst),
730		       m->m_pkthdr.rcvif ? m->m_pkthdr.rcvif->if_index : 0);
731		if (icmp6->icmp6_type < ICMP6_ECHO_REQUEST) {
732			/* ICMPv6 error: MUST deliver it by spec... */
733			code = PRC_NCMDS;
734			/* deliver */
735		} else {
736			/* ICMPv6 informational: MUST not deliver */
737			break;
738		}
739	deliver:
740		if (icmp6len < sizeof(struct icmp6_hdr) + sizeof(struct ip6_hdr)) {
741			icmp6stat.icp6s_tooshort++;
742			goto freeit;
743		}
744#ifndef PULLDOWN_TEST
745		IP6_EXTHDR_CHECK(m, off,
746			sizeof(struct icmp6_hdr) + sizeof(struct ip6_hdr),
747			IPPROTO_DONE);
748		icmp6 = (struct icmp6_hdr *)(mtod(m, caddr_t) + off);
749#else
750		IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off,
751			sizeof(*icmp6) + sizeof(struct ip6_hdr));
752		if (icmp6 == NULL) {
753			icmp6stat.icp6s_tooshort++;
754			return IPPROTO_DONE;
755		}
756#endif
757		bzero(&icmp6src, sizeof(icmp6src));
758		icmp6src.sin6_len = sizeof(struct sockaddr_in6);
759		icmp6src.sin6_family = AF_INET6;
760		icmp6src.sin6_addr = ((struct ip6_hdr *)(icmp6 + 1))->ip6_dst;
761
762		/* Detect the upper level protocol */
763	    {
764		void (*ctlfunc) __P((int, struct sockaddr *, void *));
765		struct ip6_hdr *eip6 = (struct ip6_hdr *)(icmp6 + 1);
766		u_int8_t nxt = eip6->ip6_nxt;
767		int eoff = off + sizeof(struct icmp6_hdr) +
768			sizeof(struct ip6_hdr);
769		struct ip6ctlparam ip6cp;
770
771		while (1) { /* XXX: should avoid inf. loop explicitly? */
772			struct ip6_ext *eh;
773
774			switch(nxt) {
775			case IPPROTO_ESP:
776			case IPPROTO_NONE:
777				goto passit;
778			case IPPROTO_HOPOPTS:
779			case IPPROTO_DSTOPTS:
780			case IPPROTO_ROUTING:
781			case IPPROTO_AH:
782			case IPPROTO_FRAGMENT:
783#ifndef PULLDOWN_TEST
784				IP6_EXTHDR_CHECK(m, 0, eoff +
785						 sizeof(struct ip6_ext),
786						 IPPROTO_DONE);
787				eh = (struct ip6_ext *)(mtod(m, caddr_t)
788							+ eoff);
789#else
790				IP6_EXTHDR_GET(eh, struct ip6_ext *, m,
791					eoff, sizeof(*eh));
792				if (eh == NULL) {
793					icmp6stat.icp6s_tooshort++;
794					return IPPROTO_DONE;
795				}
796#endif
797				if (nxt == IPPROTO_AH)
798					eoff += (eh->ip6e_len + 2) << 2;
799				else if (nxt == IPPROTO_FRAGMENT)
800					eoff += sizeof(struct ip6_frag);
801				else
802					eoff += (eh->ip6e_len + 1) << 3;
803				nxt = eh->ip6e_nxt;
804				break;
805			default:
806				goto notify;
807			}
808		}
809	    notify:
810#ifndef PULLDOWN_TEST
811		icmp6 = (struct icmp6_hdr *)(mtod(m, caddr_t) + off);
812#else
813		IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off,
814			sizeof(*icmp6) + sizeof(struct ip6_hdr));
815		if (icmp6 == NULL) {
816			icmp6stat.icp6s_tooshort++;
817			return IPPROTO_DONE;
818		}
819#endif
820		ctlfunc = (void (*) __P((int, struct sockaddr *, void *)))
821			(inet6sw[ip6_protox[nxt]].pr_ctlinput);
822		if (ctlfunc) {
823			ip6cp.ip6c_m = m;
824			ip6cp.ip6c_ip6 = (struct ip6_hdr *)(icmp6 + 1);
825			ip6cp.ip6c_off = eoff;
826			(*ctlfunc)(code, (struct sockaddr *)&icmp6src, &ip6cp);
827		}
828	    }
829		break;
830
831	badcode:
832		icmp6stat.icp6s_badcode++;
833		break;
834
835	badlen:
836		icmp6stat.icp6s_badlen++;
837		break;
838	}
839
840 passit:
841	icmp6_rip6_input(&m, *offp);
842	return IPPROTO_DONE;
843
844 freeit:
845	m_freem(m);
846	return IPPROTO_DONE;
847}
848
849/*
850 * Process a Node Information Query
851 */
852#ifndef offsetof		/* XXX */
853#define	offsetof(type, member)	((size_t)(&((type *)0)->member))
854#endif
855
856static struct mbuf *
857ni6_input(m, off)
858	struct mbuf *m;
859	int off;
860{
861	struct icmp6_nodeinfo *ni6, *nni6;
862	struct mbuf *n = NULL;
863	u_int16_t qtype;
864	int replylen = sizeof(struct ip6_hdr) + sizeof(struct icmp6_nodeinfo);
865	struct ni_reply_fqdn *fqdn;
866	int addrs;		/* for NI_QTYPE_NODEADDR */
867	struct ifnet *ifp = NULL; /* for NI_QTYPE_NODEADDR */
868
869#ifndef PULLDOWN_TEST
870	ni6 = (struct icmp6_nodeinfo *)(mtod(m, caddr_t) + off);
871#else
872	IP6_EXTHDR_GET(ni6, struct icmp6_nodeinfo *, m, off, sizeof(*ni6));
873	if (ni6 == NULL) {
874		/* m is already reclaimed */
875		return NULL;
876	}
877#endif
878	qtype = ntohs(ni6->ni_qtype);
879
880	switch(qtype) {
881	 case NI_QTYPE_NOOP:
882		 break;		/* no reply data */
883	 case NI_QTYPE_SUPTYPES:
884		 goto bad;	/* xxx: to be implemented */
885		 break;
886	 case NI_QTYPE_FQDN:
887		 replylen += offsetof(struct ni_reply_fqdn, ni_fqdn_name) +
888			 hostnamelen;
889		 break;
890	 case NI_QTYPE_NODEADDR:
891		 addrs = ni6_addrs(ni6, m, &ifp);
892		 if ((replylen += addrs * sizeof(struct in6_addr)) > MCLBYTES)
893			 replylen = MCLBYTES; /* XXX: we'll truncate later */
894
895		 break;
896	 default:
897		 /*
898		  * XXX: We must return a reply with the ICMP6 code
899		  * `unknown Qtype' in this case. However we regard the case
900		  * as an FQDN query for backward compatibility.
901		  * Older versions set a random value to this field,
902		  * so it rarely varies in the defined qtypes.
903		  * But the mechanism is not reliable...
904		  * maybe we should obsolete older versions.
905		  */
906		 qtype = NI_QTYPE_FQDN;
907		 replylen += offsetof(struct ni_reply_fqdn, ni_fqdn_name) +
908			 hostnamelen;
909		 break;
910	}
911
912	/* allocate a mbuf to reply. */
913	MGETHDR(n, M_DONTWAIT, m->m_type);
914	if (n == NULL) {
915		m_freem(m);
916		return(NULL);
917	}
918	M_COPY_PKTHDR(n, m); /* just for recvif */
919	if (replylen > MHLEN) {
920		if (replylen > MCLBYTES)
921			 /*
922			  * XXX: should we try to allocate more? But MCLBYTES is
923			  * probably much larger than IPV6_MMTU...
924			  */
925			goto bad;
926		MCLGET(n, M_DONTWAIT);
927		if ((n->m_flags & M_EXT) == 0) {
928			goto bad;
929		}
930	}
931	n->m_pkthdr.len = n->m_len = replylen;
932
933	/* copy mbuf header and IPv6 + Node Information base headers */
934	bcopy(mtod(m, caddr_t), mtod(n, caddr_t), sizeof(struct ip6_hdr));
935	nni6 = (struct icmp6_nodeinfo *)(mtod(n, struct ip6_hdr *) + 1);
936	bcopy((caddr_t)ni6, (caddr_t)nni6, sizeof(struct icmp6_nodeinfo));
937
938	/* qtype dependent procedure */
939	switch (qtype) {
940	 case NI_QTYPE_NOOP:
941		 nni6->ni_flags = 0;
942		 break;
943	 case NI_QTYPE_SUPTYPES:
944		 goto bad;	/* xxx: to be implemented */
945		 break;
946	 case NI_QTYPE_FQDN:
947		 if (hostnamelen > 255) { /* XXX: rare case, but may happen */
948			 printf("ni6_input: "
949				"hostname length(%d) is too large for reply\n",
950				hostnamelen);
951			 goto bad;
952		 }
953		 fqdn = (struct ni_reply_fqdn *)(mtod(n, caddr_t) +
954						 sizeof(struct ip6_hdr) +
955						 sizeof(struct icmp6_nodeinfo));
956		 nni6->ni_flags = 0; /* XXX: meaningless TTL */
957		 fqdn->ni_fqdn_ttl = 0;	/* ditto. */
958		 fqdn->ni_fqdn_namelen = hostnamelen;
959		 bcopy(hostname, &fqdn->ni_fqdn_name[0], hostnamelen);
960		 break;
961	 case NI_QTYPE_NODEADDR:
962	 {
963		 int lenlim, copied;
964
965		 if (n->m_flags & M_EXT)
966			 lenlim = MCLBYTES - sizeof(struct ip6_hdr) -
967				 sizeof(struct icmp6_nodeinfo);
968		 else
969			 lenlim = MHLEN - sizeof(struct ip6_hdr) -
970				 sizeof(struct icmp6_nodeinfo);
971		 copied = ni6_store_addrs(ni6, nni6, ifp, lenlim);
972		 /* XXX: reset mbuf length */
973		 n->m_pkthdr.len = n->m_len = sizeof(struct ip6_hdr) +
974			 sizeof(struct icmp6_nodeinfo) + copied;
975		 break;
976	 }
977	 default:
978		 break;		/* XXX impossible! */
979	}
980
981	nni6->ni_type = ICMP6_NI_REPLY;
982	nni6->ni_code = ICMP6_NI_SUCESS;
983	m_freem(m);
984	return(n);
985
986  bad:
987	m_freem(m);
988	if (n)
989		m_freem(n);
990	return(NULL);
991}
992#undef hostnamelen
993
994/*
995 * calculate the number of addresses to be returned in the node info reply.
996 */
997static int
998ni6_addrs(ni6, m, ifpp)
999	struct icmp6_nodeinfo *ni6;
1000	struct mbuf *m;
1001	struct ifnet **ifpp;
1002{
1003	register struct ifnet *ifp;
1004	register struct in6_ifaddr *ifa6;
1005	register struct ifaddr *ifa;
1006	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1007	int addrs = 0, addrsofif, iffound = 0;
1008
1009	for (ifp = TAILQ_FIRST(&ifnet); ifp; ifp = TAILQ_NEXT(ifp, if_list))
1010	{
1011		addrsofif = 0;
1012		for (ifa = ifp->if_addrlist.tqh_first; ifa;
1013		     ifa = ifa->ifa_list.tqe_next)
1014		{
1015			if (ifa->ifa_addr->sa_family != AF_INET6)
1016				continue;
1017			ifa6 = (struct in6_ifaddr *)ifa;
1018
1019			if (!(ni6->ni_flags & NI_NODEADDR_FLAG_ALL) &&
1020			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1021					       &ifa6->ia_addr.sin6_addr))
1022				iffound = 1;
1023
1024			if (ifa6->ia6_flags & IN6_IFF_ANYCAST)
1025				continue; /* we need only unicast addresses */
1026
1027			if ((ni6->ni_flags & (NI_NODEADDR_FLAG_LINKLOCAL |
1028					      NI_NODEADDR_FLAG_SITELOCAL |
1029					      NI_NODEADDR_FLAG_GLOBAL)) == 0)
1030				continue;
1031
1032			/* What do we have to do about ::1? */
1033			switch(in6_addrscope(&ifa6->ia_addr.sin6_addr)) {
1034			 case IPV6_ADDR_SCOPE_LINKLOCAL:
1035				if (ni6->ni_flags & NI_NODEADDR_FLAG_LINKLOCAL)
1036					addrsofif++;
1037				break;
1038			 case IPV6_ADDR_SCOPE_SITELOCAL:
1039				if (ni6->ni_flags & NI_NODEADDR_FLAG_SITELOCAL)
1040					addrsofif++;
1041				break;
1042			 case IPV6_ADDR_SCOPE_GLOBAL:
1043				 if (ni6->ni_flags & NI_NODEADDR_FLAG_GLOBAL)
1044					 addrsofif++;
1045				 break;
1046			 default:
1047				 continue;
1048			}
1049		}
1050		if (iffound) {
1051			*ifpp = ifp;
1052			return(addrsofif);
1053		}
1054
1055		addrs += addrsofif;
1056	}
1057
1058	return(addrs);
1059}
1060
1061static int
1062ni6_store_addrs(ni6, nni6, ifp0, resid)
1063	struct icmp6_nodeinfo *ni6, *nni6;
1064	struct ifnet *ifp0;
1065	int resid;
1066{
1067	register struct ifnet *ifp = ifp0 ? ifp0 : TAILQ_FIRST(&ifnet);
1068	register struct in6_ifaddr *ifa6;
1069	register struct ifaddr *ifa;
1070	int docopy, copied = 0;
1071	u_char *cp = (u_char *)(nni6 + 1);
1072
1073	if (ifp0 == NULL && !(ni6->ni_flags & NI_NODEADDR_FLAG_ALL))
1074		return(0);	/* needless to copy */
1075
1076	for (; ifp; ifp = TAILQ_NEXT(ifp, if_list))
1077	{
1078		for (ifa = ifp->if_addrlist.tqh_first; ifa;
1079		     ifa = ifa->ifa_list.tqe_next)
1080		{
1081			docopy = 0;
1082
1083			if (ifa->ifa_addr->sa_family != AF_INET6)
1084				continue;
1085			ifa6 = (struct in6_ifaddr *)ifa;
1086
1087			if (ifa6->ia6_flags & IN6_IFF_ANYCAST) {
1088				/* just experimental. not in the spec. */
1089				if (ni6->ni_flags & NI_NODEADDR_FLAG_ANYCAST)
1090					docopy = 1;
1091				else
1092					continue;
1093			}
1094			else {	/* unicast address */
1095				if (ni6->ni_flags & NI_NODEADDR_FLAG_ANYCAST)
1096					continue;
1097				else
1098					docopy = 1;
1099			}
1100
1101			/* What do we have to do about ::1? */
1102			switch(in6_addrscope(&ifa6->ia_addr.sin6_addr)) {
1103			 case IPV6_ADDR_SCOPE_LINKLOCAL:
1104				if (ni6->ni_flags & NI_NODEADDR_FLAG_LINKLOCAL)
1105					docopy = 1;
1106				break;
1107			 case IPV6_ADDR_SCOPE_SITELOCAL:
1108				if (ni6->ni_flags & NI_NODEADDR_FLAG_SITELOCAL)
1109					docopy = 1;
1110				break;
1111			 case IPV6_ADDR_SCOPE_GLOBAL:
1112				 if (ni6->ni_flags & NI_NODEADDR_FLAG_GLOBAL)
1113					 docopy = 1;
1114				 break;
1115			 default:
1116				 continue;
1117			}
1118
1119			if (docopy) {
1120				if (resid < sizeof(struct in6_addr)) {
1121					/*
1122					 * We give up much more copy.
1123					 * Set the truncate flag and return.
1124					 */
1125					nni6->ni_flags |=
1126						NI_NODEADDR_FLAG_TRUNCATE;
1127					return(copied);
1128				}
1129				bcopy(&ifa6->ia_addr.sin6_addr, cp,
1130				      sizeof(struct in6_addr));
1131				/* XXX: KAME link-local hack; remove ifindex */
1132				if (IN6_IS_ADDR_LINKLOCAL(&ifa6->ia_addr.sin6_addr))
1133					((struct in6_addr *)cp)->s6_addr16[1] = 0;
1134				cp += sizeof(struct in6_addr);
1135				resid -= sizeof(struct in6_addr);
1136				copied += sizeof(struct in6_addr);
1137			}
1138		}
1139		if (ifp0)	/* we need search only on the specified IF */
1140			break;
1141	}
1142
1143	return(copied);
1144}
1145
1146/*
1147 * XXX almost dup'ed code with rip6_input.
1148 */
1149static int
1150icmp6_rip6_input(mp, off)
1151	struct	mbuf **mp;
1152	int	off;
1153{
1154	struct mbuf *m = *mp;
1155	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1156	register struct in6pcb *in6p;
1157	struct in6pcb *last = NULL;
1158	struct sockaddr_in6 rip6src;
1159	struct icmp6_hdr *icmp6;
1160	struct mbuf *opts = NULL;
1161
1162	/* this is assumed to be safe. */
1163	icmp6 = (struct icmp6_hdr *)((caddr_t)ip6 + off);
1164
1165	bzero(&rip6src, sizeof(rip6src));
1166	rip6src.sin6_len = sizeof(struct sockaddr_in6);
1167	rip6src.sin6_family = AF_INET6;
1168	rip6src.sin6_addr = ip6->ip6_src;
1169	if (IN6_IS_SCOPE_LINKLOCAL(&rip6src.sin6_addr))
1170		rip6src.sin6_addr.s6_addr16[1] = 0;
1171	if (m->m_pkthdr.rcvif) {
1172		if (IN6_IS_SCOPE_LINKLOCAL(&rip6src.sin6_addr))
1173			rip6src.sin6_scope_id = m->m_pkthdr.rcvif->if_index;
1174		else
1175			rip6src.sin6_scope_id = 0;
1176	} else
1177		rip6src.sin6_scope_id = 0;
1178
1179	for (in6p = rawin6pcb.in6p_next;
1180	     in6p != &rawin6pcb; in6p = in6p->in6p_next)
1181	{
1182		if (in6p->in6p_ip6_nxt != IPPROTO_ICMPV6)
1183			continue;
1184		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
1185		   !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
1186			continue;
1187		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
1188		   !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
1189			continue;
1190		if (in6p->in6p_icmp6filt
1191		    && ICMP6_FILTER_WILLBLOCK(icmp6->icmp6_type,
1192				 in6p->in6p_icmp6filt))
1193			continue;
1194		if (last) {
1195			struct	mbuf *n;
1196			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
1197				if (last->in6p_flags & IN6P_CONTROLOPTS)
1198					ip6_savecontrol(last, &opts, ip6, n);
1199				/* strip intermediate headers */
1200				m_adj(n, off);
1201				if (sbappendaddr(&last->in6p_socket->so_rcv,
1202						 (struct sockaddr *)&rip6src,
1203						 n, opts) == 0) {
1204					/* should notify about lost packet */
1205					m_freem(n);
1206					if (opts)
1207						m_freem(opts);
1208				} else
1209					sorwakeup(last->in6p_socket);
1210				opts = NULL;
1211			}
1212		}
1213		last = in6p;
1214	}
1215	if (last) {
1216		if (last->in6p_flags & IN6P_CONTROLOPTS)
1217			ip6_savecontrol(last, &opts, ip6, m);
1218		/* strip intermediate headers */
1219		m_adj(m, off);
1220		if (sbappendaddr(&last->in6p_socket->so_rcv,
1221				(struct sockaddr *)&rip6src, m, opts) == 0) {
1222			m_freem(m);
1223			if (opts)
1224				m_freem(opts);
1225		} else
1226			sorwakeup(last->in6p_socket);
1227	} else {
1228		m_freem(m);
1229		ip6stat.ip6s_delivered--;
1230	}
1231	return IPPROTO_DONE;
1232}
1233
1234/*
1235 * Reflect the ip6 packet back to the source.
1236 * OFF points to the icmp6 header, counted from the top of the mbuf.
1237 */
1238void
1239icmp6_reflect(m, off)
1240	struct	mbuf *m;
1241	size_t off;
1242{
1243	struct ip6_hdr *ip6;
1244	struct icmp6_hdr *icmp6;
1245	struct in6_ifaddr *ia;
1246	struct in6_addr t, *src = 0;
1247	int plen;
1248	int type, code;
1249	struct ifnet *outif = NULL;
1250#ifdef COMPAT_RFC1885
1251	int mtu = IPV6_MMTU;
1252	struct sockaddr_in6 *sin6 = &icmp6_reflect_rt.ro_dst;
1253#endif
1254
1255	/* too short to reflect */
1256	if (off < sizeof(struct ip6_hdr)) {
1257		printf("sanity fail: off=%lx, sizeof(ip6)=%lx in %s:%d\n",
1258		       (u_long)off, (u_long)sizeof(struct ip6_hdr),
1259		       __FILE__, __LINE__);
1260		goto bad;
1261	}
1262
1263	/*
1264	 * If there are extra headers between IPv6 and ICMPv6, strip
1265	 * off that header first.
1266	 */
1267	if (off > sizeof(struct ip6_hdr)) {
1268		size_t l;
1269		struct ip6_hdr nip6;
1270
1271		l = off - sizeof(struct ip6_hdr);
1272		m_copydata(m, 0, sizeof(nip6), (caddr_t)&nip6);
1273		m_adj(m, l);
1274		l = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr);
1275		if (m->m_len < l) {
1276			if ((m = m_pullup(m, l)) == NULL)
1277				return;
1278		}
1279		bcopy((caddr_t)&nip6, mtod(m, caddr_t), sizeof(nip6));
1280	} else /* off == sizeof(struct ip6_hdr) */ {
1281		size_t l;
1282		l = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr);
1283		if (m->m_len < l) {
1284			if ((m = m_pullup(m, l)) == NULL)
1285				return;
1286		}
1287	}
1288	plen = m->m_pkthdr.len - sizeof(struct ip6_hdr);
1289	ip6 = mtod(m, struct ip6_hdr *);
1290	ip6->ip6_nxt = IPPROTO_ICMPV6;
1291	icmp6 = (struct icmp6_hdr *)(ip6 + 1);
1292	type = icmp6->icmp6_type; /* keep type for statistics */
1293	code = icmp6->icmp6_code; /* ditto. */
1294
1295	t = ip6->ip6_dst;
1296	/*
1297	 * ip6_input() drops a packet if its src is multicast.
1298	 * So, the src is never multicast.
1299	 */
1300	ip6->ip6_dst = ip6->ip6_src;
1301
1302	/* XXX hack for link-local addresses */
1303	if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst))
1304		ip6->ip6_dst.s6_addr16[1] =
1305			htons(m->m_pkthdr.rcvif->if_index);
1306	if (IN6_IS_ADDR_LINKLOCAL(&t))
1307		t.s6_addr16[1] = htons(m->m_pkthdr.rcvif->if_index);
1308
1309#ifdef COMPAT_RFC1885
1310	/*
1311	 * xxx guess MTU
1312	 * RFC 1885 requires that echo reply should be truncated if it
1313	 * does not fit in with (return) path MTU, but the description was
1314	 * removed in the new spec.
1315	 */
1316	if (icmp6_reflect_rt.ro_rt == 0 ||
1317	    ! (IN6_ARE_ADDR_EQUAL(&sin6->sin6_addr, &ip6->ip6_dst))) {
1318		if (icmp6_reflect_rt.ro_rt) {
1319			icmp6_reflect_rt.ro_rt = 0;
1320		}
1321		bzero(sin6, sizeof(*sin6));
1322		sin6->sin6_family = PF_INET6;
1323		sin6->sin6_len = sizeof(struct sockaddr_in6);
1324		sin6->sin6_addr = ip6->ip6_dst;
1325
1326		rtalloc((struct route *)&icmp6_reflect_rt.ro_rt);
1327	}
1328
1329	if (icmp6_reflect_rt.ro_rt == 0)
1330		goto bad;
1331
1332	if ((icmp6_reflect_rt.ro_rt->rt_flags & RTF_HOST)
1333	    && mtu < icmp6_reflect_rt.ro_rt->rt_ifp->if_mtu)
1334		mtu = icmp6_reflect_rt.ro_rt->rt_rmx.rmx_mtu;
1335
1336	if (mtu < m->m_pkthdr.len) {
1337		plen -= (m->m_pkthdr.len - mtu);
1338		m_adj(m, mtu - m->m_pkthdr.len);
1339	}
1340#endif
1341	/*
1342	 * If the incoming packet was addressed directly to us(i.e. unicast),
1343	 * use dst as the src for the reply.
1344	 */
1345	for (ia = in6_ifaddr; ia; ia = ia->ia_next)
1346		if (IN6_ARE_ADDR_EQUAL(&t, &ia->ia_addr.sin6_addr) &&
1347		    (ia->ia6_flags & IN6_IFF_ANYCAST) == 0) {
1348			src = &t;
1349			break;
1350		}
1351	if (ia == NULL && IN6_IS_ADDR_LINKLOCAL(&t) && (m->m_flags & M_LOOP)) {
1352		/*
1353		 * This is the case if the dst is our link-local address
1354		 * and the sender is also ourseleves.
1355		 */
1356		src = &t;
1357	}
1358
1359	if (src == 0)
1360		/*
1361		 * We have not multicast routing yet. So this case matches
1362		 * to our multicast, our anycast or not to our unicast.
1363		 * Select a source address which has the same scope.
1364		 */
1365		if ((ia = in6_ifawithscope(m->m_pkthdr.rcvif, &t)) != 0)
1366			src = &IA6_SIN6(ia)->sin6_addr;
1367
1368	if (src == 0)
1369		goto bad;
1370
1371	ip6->ip6_src = *src;
1372
1373	ip6->ip6_flow = 0;
1374	ip6->ip6_vfc &= ~IPV6_VERSION_MASK;
1375	ip6->ip6_vfc |= IPV6_VERSION;
1376	ip6->ip6_nxt = IPPROTO_ICMPV6;
1377	if (m->m_pkthdr.rcvif) {
1378		/* XXX: This may not be the outgoing interface */
1379		ip6->ip6_hlim = nd_ifinfo[m->m_pkthdr.rcvif->if_index].chlim;
1380	}
1381
1382	icmp6->icmp6_cksum = 0;
1383	icmp6->icmp6_cksum = in6_cksum(m, IPPROTO_ICMPV6,
1384					sizeof(struct ip6_hdr), plen);
1385
1386	/*
1387	 * xxx option handling
1388	 */
1389
1390	m->m_flags &= ~(M_BCAST|M_MCAST);
1391#ifdef IPSEC
1392	m->m_pkthdr.rcvif = NULL;
1393#endif /*IPSEC*/
1394
1395#ifdef COMPAT_RFC1885
1396	ip6_output(m, NULL, &icmp6_reflect_rt, 0, NULL, &outif);
1397#else
1398	ip6_output(m, NULL, NULL, 0, NULL, &outif);
1399#endif
1400	if (outif)
1401		icmp6_ifoutstat_inc(outif, type, code);
1402
1403	return;
1404
1405 bad:
1406	m_freem(m);
1407	return;
1408}
1409
1410void
1411icmp6_fasttimo()
1412{
1413	mld6_fasttimeo();
1414}
1415
1416static const char *
1417icmp6_redirect_diag(src6, dst6, tgt6)
1418	struct in6_addr *src6;
1419	struct in6_addr *dst6;
1420	struct in6_addr *tgt6;
1421{
1422	static char buf[1024];
1423	snprintf(buf, sizeof(buf), "(src=%s dst=%s tgt=%s)",
1424		ip6_sprintf(src6), ip6_sprintf(dst6), ip6_sprintf(tgt6));
1425	return buf;
1426}
1427
1428void
1429icmp6_redirect_input(m, off)
1430	register struct mbuf *m;
1431	int off;
1432{
1433	struct ifnet *ifp = m->m_pkthdr.rcvif;
1434	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1435	struct nd_redirect *nd_rd = (struct nd_redirect *)((caddr_t)ip6 + off);
1436	int icmp6len = ntohs(ip6->ip6_plen);
1437	char *lladdr = NULL;
1438	int lladdrlen = 0;
1439	u_char *redirhdr = NULL;
1440	int redirhdrlen = 0;
1441	struct rtentry *rt = NULL;
1442	int is_router;
1443	int is_onlink;
1444	struct in6_addr src6 = ip6->ip6_src;
1445	struct in6_addr redtgt6 = nd_rd->nd_rd_target;
1446	struct in6_addr reddst6 = nd_rd->nd_rd_dst;
1447	union nd_opts ndopts;
1448
1449	if (!m || !ifp)
1450		return;
1451
1452	/* XXX if we are router, we don't update route by icmp6 redirect */
1453	if (ip6_forwarding)
1454		return;
1455	if (!icmp6_rediraccept)
1456		return;
1457
1458	if (IN6_IS_ADDR_LINKLOCAL(&redtgt6))
1459		redtgt6.s6_addr16[1] = htons(ifp->if_index);
1460	if (IN6_IS_ADDR_LINKLOCAL(&reddst6))
1461		reddst6.s6_addr16[1] = htons(ifp->if_index);
1462
1463	/* validation */
1464	if (!IN6_IS_ADDR_LINKLOCAL(&src6)) {
1465		log(LOG_ERR,
1466			"ICMP6 redirect sent from %s rejected; "
1467			"must be from linklocal\n", ip6_sprintf(&src6));
1468		return;
1469	}
1470	if (ip6->ip6_hlim != 255) {
1471		log(LOG_ERR,
1472			"ICMP6 redirect sent from %s rejected; "
1473			"hlim=%d (must be 255)\n",
1474			ip6_sprintf(&src6), ip6->ip6_hlim);
1475		return;
1476	}
1477    {
1478	/* ip6->ip6_src must be equal to gw for icmp6->icmp6_reddst */
1479	struct sockaddr_in6 sin6;
1480	struct in6_addr *gw6;
1481
1482	bzero(&sin6, sizeof(sin6));
1483	sin6.sin6_family = AF_INET6;
1484	sin6.sin6_len = sizeof(struct sockaddr_in6);
1485	bcopy(&reddst6, &sin6.sin6_addr, sizeof(reddst6));
1486	rt = rtalloc1((struct sockaddr *)&sin6, 0
1487		      );
1488	if (rt) {
1489		gw6 = &(((struct sockaddr_in6 *)rt->rt_gateway)->sin6_addr);
1490		if (bcmp(&src6, gw6, sizeof(struct in6_addr)) != 0) {
1491			log(LOG_ERR,
1492				"ICMP6 redirect rejected; "
1493				"not equal to gw-for-src=%s (must be same): "
1494				"%s\n",
1495				ip6_sprintf(gw6),
1496				icmp6_redirect_diag(&src6, &reddst6, &redtgt6));
1497			RTFREE(rt);
1498			return;
1499		}
1500	} else {
1501		log(LOG_ERR,
1502			"ICMP6 redirect rejected; "
1503			"no route found for redirect dst: %s\n",
1504			icmp6_redirect_diag(&src6, &reddst6, &redtgt6));
1505		return;
1506	}
1507	RTFREE(rt);
1508	rt = NULL;
1509    }
1510	if (IN6_IS_ADDR_MULTICAST(&reddst6)) {
1511		log(LOG_ERR,
1512			"ICMP6 redirect rejected; "
1513			"redirect dst must be unicast: %s\n",
1514			icmp6_redirect_diag(&src6, &reddst6, &redtgt6));
1515		return;
1516	}
1517
1518	is_router = is_onlink = 0;
1519	if (IN6_IS_ADDR_LINKLOCAL(&redtgt6))
1520		is_router = 1;	/* router case */
1521	if (bcmp(&redtgt6, &reddst6, sizeof(redtgt6)) == 0)
1522		is_onlink = 1;	/* on-link destination case */
1523	if (!is_router && !is_onlink) {
1524		log(LOG_ERR,
1525			"ICMP6 redirect rejected; "
1526			"neither router case nor onlink case: %s\n",
1527			icmp6_redirect_diag(&src6, &reddst6, &redtgt6));
1528		return;
1529	}
1530	/* validation passed */
1531
1532	icmp6len -= sizeof(*nd_rd);
1533	nd6_option_init(nd_rd + 1, icmp6len, &ndopts);
1534	if (nd6_options(&ndopts) < 0) {
1535		log(LOG_INFO, "icmp6_redirect_input: "
1536			"invalid ND option, rejected: %s\n",
1537			icmp6_redirect_diag(&src6, &reddst6, &redtgt6));
1538		return;
1539	}
1540
1541	if (ndopts.nd_opts_tgt_lladdr) {
1542		lladdr = (char *)(ndopts.nd_opts_tgt_lladdr + 1);
1543		lladdrlen = ndopts.nd_opts_tgt_lladdr->nd_opt_len << 3;
1544	}
1545
1546	if (ndopts.nd_opts_rh) {
1547		redirhdrlen = ndopts.nd_opts_rh->nd_opt_rh_len;
1548		redirhdr = (u_char *)(ndopts.nd_opts_rh + 1); /* xxx */
1549	}
1550
1551	if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
1552		log(LOG_INFO,
1553			"icmp6_redirect_input: lladdrlen mismatch for %s "
1554			"(if %d, icmp6 packet %d): %s\n",
1555			ip6_sprintf(&redtgt6), ifp->if_addrlen, lladdrlen - 2,
1556			icmp6_redirect_diag(&src6, &reddst6, &redtgt6));
1557	}
1558
1559	/* RFC 2461 8.3 */
1560	nd6_cache_lladdr(ifp, &redtgt6, lladdr, lladdrlen, ND_REDIRECT,
1561			 is_onlink ? ND_REDIRECT_ONLINK : ND_REDIRECT_ROUTER);
1562
1563	if (!is_onlink) {	/* better router case. perform rtredirect. */
1564		/* perform rtredirect */
1565		struct sockaddr_in6 sdst;
1566		struct sockaddr_in6 sgw;
1567		struct sockaddr_in6 ssrc;
1568
1569		bzero(&sdst, sizeof(sdst));
1570		bzero(&sgw, sizeof(sgw));
1571		bzero(&ssrc, sizeof(ssrc));
1572		sdst.sin6_family = sgw.sin6_family = ssrc.sin6_family = AF_INET6;
1573		sdst.sin6_len = sgw.sin6_len = ssrc.sin6_len =
1574			sizeof(struct sockaddr_in6);
1575		bcopy(&redtgt6, &sgw.sin6_addr, sizeof(struct in6_addr));
1576		bcopy(&reddst6, &sdst.sin6_addr, sizeof(struct in6_addr));
1577		bcopy(&src6, &ssrc.sin6_addr, sizeof(struct in6_addr));
1578		rtredirect((struct sockaddr *)&sdst, (struct sockaddr *)&sgw,
1579			   (struct sockaddr *)NULL, RTF_GATEWAY | RTF_HOST,
1580			   (struct sockaddr *)&ssrc,
1581			   (struct rtentry **)NULL
1582			   );
1583	}
1584	/* finally update cached route in each socket via pfctlinput */
1585    {
1586	struct sockaddr_in6 sdst;
1587#if 1
1588#else
1589	struct ip6protosw *pr;
1590#endif
1591
1592	bzero(&sdst, sizeof(sdst));
1593	sdst.sin6_family = AF_INET6;
1594	sdst.sin6_len = sizeof(struct sockaddr_in6);
1595	bcopy(&reddst6, &sdst.sin6_addr, sizeof(struct in6_addr));
1596#if 1
1597	pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&sdst);
1598#else
1599	/*
1600	 * do not use pfctlinput() here, we have different prototype for
1601	 * xx_ctlinput() in ip6proto.
1602	 */
1603	for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
1604	     pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW;
1605	     pr++) {
1606		if (pr->pr_ctlinput) {
1607			(*pr->pr_ctlinput)(PRC_REDIRECT_HOST,
1608				(struct sockaddr *)&sdst, NULL, NULL, 0);
1609		}
1610	}
1611#endif
1612#ifdef IPSEC
1613	key_sa_routechange((struct sockaddr *)&sdst);
1614#endif
1615    }
1616}
1617
1618void
1619icmp6_redirect_output(m0, rt)
1620	struct mbuf *m0;
1621	struct rtentry *rt;
1622{
1623	struct ifnet *ifp;	/* my outgoing interface */
1624	struct in6_addr *ifp_ll6;
1625	struct in6_addr *router_ll6;
1626	struct ip6_hdr *sip6;	/* m0 as struct ip6_hdr */
1627	struct mbuf *m = NULL;	/* newly allocated one */
1628	struct ip6_hdr *ip6;	/* m as struct ip6_hdr */
1629	struct nd_redirect *nd_rd;
1630	size_t maxlen;
1631	u_char *p;
1632	struct ifnet *outif = NULL;
1633
1634	/* if we are not router, we don't send icmp6 redirect */
1635	if (!ip6_forwarding || ip6_accept_rtadv)
1636		goto fail;
1637
1638	/* sanity check */
1639	if (!m0 || !rt || !(rt->rt_flags & RTF_UP) || !(ifp = rt->rt_ifp))
1640		goto fail;
1641
1642	/*
1643	 * Address check:
1644	 *  the source address must identify a neighbor, and
1645	 *  the destination address must not be a multicast address
1646	 *  [RFC 2461, sec 8.2]
1647	 */
1648	sip6 = mtod(m0, struct ip6_hdr *);
1649	if (nd6_is_addr_neighbor(&sip6->ip6_src, ifp) == 0)
1650		goto fail;
1651	if (IN6_IS_ADDR_MULTICAST(&sip6->ip6_dst))
1652		goto fail;	/* what should we do here? */
1653
1654	/* rate limit */
1655	if (icmp6_ratelimit(&sip6->ip6_src, ND_REDIRECT, 0))
1656		goto fail;
1657
1658	/*
1659	 * Since we are going to append up to 1280 bytes (= IPV6_MMTU),
1660	 * we almost always ask for an mbuf cluster for simplicity.
1661	 * (MHLEN < IPV6_MMTU is almost always true)
1662	 */
1663	MGETHDR(m, M_DONTWAIT, MT_HEADER);
1664	if (!m)
1665		goto fail;
1666	if (MHLEN < IPV6_MMTU)
1667		MCLGET(m, M_DONTWAIT);
1668	maxlen = (m->m_flags & M_EXT) ? MCLBYTES : MHLEN;
1669	maxlen = min(IPV6_MMTU, maxlen);
1670	/* just for safety */
1671	if (maxlen < sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr))
1672		goto fail;
1673
1674	{
1675		/* get ip6 linklocal address for ifp(my outgoing interface). */
1676		struct in6_ifaddr *ia = in6ifa_ifpforlinklocal(ifp);
1677		if (ia == NULL)
1678			goto fail;
1679		ifp_ll6 = &ia->ia_addr.sin6_addr;
1680	}
1681
1682	/* get ip6 linklocal address for the router. */
1683	if (rt->rt_gateway && (rt->rt_flags & RTF_GATEWAY)) {
1684		struct sockaddr_in6 *sin6;
1685		sin6 = (struct sockaddr_in6 *)rt->rt_gateway;
1686		router_ll6 = &sin6->sin6_addr;
1687		if (!IN6_IS_ADDR_LINKLOCAL(router_ll6))
1688			router_ll6 = (struct in6_addr *)NULL;
1689	} else
1690		router_ll6 = (struct in6_addr *)NULL;
1691
1692	/* ip6 */
1693	ip6 = mtod(m, struct ip6_hdr *);
1694	ip6->ip6_flow = 0;
1695	ip6->ip6_vfc &= ~IPV6_VERSION_MASK;
1696	ip6->ip6_vfc |= IPV6_VERSION;
1697	/* ip6->ip6_plen will be set later */
1698	ip6->ip6_nxt = IPPROTO_ICMPV6;
1699	ip6->ip6_hlim = 255;
1700	/* ip6->ip6_src must be linklocal addr for my outgoing if. */
1701	bcopy(ifp_ll6, &ip6->ip6_src, sizeof(struct in6_addr));
1702	bcopy(&sip6->ip6_src, &ip6->ip6_dst, sizeof(struct in6_addr));
1703
1704	/* ND Redirect */
1705	nd_rd = (struct nd_redirect *)(ip6 + 1);
1706	nd_rd->nd_rd_type = ND_REDIRECT;
1707	nd_rd->nd_rd_code = 0;
1708	nd_rd->nd_rd_reserved = 0;
1709	if (rt->rt_flags & RTF_GATEWAY) {
1710		/*
1711		 * nd_rd->nd_rd_target must be a link-local address in
1712		 * better router cases.
1713		 */
1714		if (!router_ll6)
1715			goto fail;
1716		bcopy(router_ll6, &nd_rd->nd_rd_target,
1717		      sizeof(nd_rd->nd_rd_target));
1718		bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_dst,
1719		      sizeof(nd_rd->nd_rd_dst));
1720	} else {
1721		/* make sure redtgt == reddst */
1722		bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_target,
1723		      sizeof(nd_rd->nd_rd_target));
1724		bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_dst,
1725		      sizeof(nd_rd->nd_rd_dst));
1726	}
1727
1728	p = (u_char *)(nd_rd + 1);
1729
1730	if (!router_ll6)
1731		goto nolladdropt;
1732
1733    {
1734	/* target lladdr option */
1735	struct rtentry *rt_router = NULL;
1736	int len;
1737	struct sockaddr_dl *sdl;
1738	struct nd_opt_hdr *nd_opt;
1739	char *lladdr;
1740
1741	rt_router = nd6_lookup(router_ll6, 0, ifp);
1742	if (!rt_router)
1743		goto nolladdropt;
1744	if (!(rt_router->rt_flags & RTF_GATEWAY) &&
1745	    (rt_router->rt_flags & RTF_LLINFO) &&
1746	    (rt_router->rt_gateway->sa_family == AF_LINK) &&
1747	    (sdl = (struct sockaddr_dl *)rt_router->rt_gateway) &&
1748	    sdl->sdl_alen) {
1749		nd_opt = (struct nd_opt_hdr *)p;
1750		nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR;
1751		len = 2 + ifp->if_addrlen;
1752		len = (len + 7) & ~7;	/*round by 8*/
1753		nd_opt->nd_opt_len = len >> 3;
1754		p += len;
1755		lladdr = (char *)(nd_opt + 1);
1756		bcopy(LLADDR(sdl), lladdr, ifp->if_addrlen);
1757	}
1758    }
1759nolladdropt:;
1760
1761	m->m_pkthdr.len = m->m_len = p - (u_char *)ip6;
1762
1763	/* just to be safe */
1764#ifdef M_DECRYPTED	/*not openbsd*/
1765	if (m0->m_flags & M_DECRYPTED)
1766		goto noredhdropt;
1767#endif
1768
1769    {
1770	/* redirected header option */
1771	int len;
1772	struct nd_opt_rd_hdr *nd_opt_rh;
1773
1774	/*
1775	 * compute the maximum size for icmp6 redirect header option.
1776	 * XXX room for auth header?
1777	 */
1778	len = maxlen - (p - (u_char *)ip6);
1779	len &= ~7;
1780
1781	/* This is just for simplicity. */
1782	if (m0->m_pkthdr.len != m0->m_len) {
1783		if (m0->m_next) {
1784			m_freem(m0->m_next);
1785			m0->m_next = NULL;
1786		}
1787		m0->m_pkthdr.len = m0->m_len;
1788	}
1789
1790	/*
1791	 * Redirected header option spec (RFC2461 4.6.3) talks nothing
1792	 * about padding/truncate rule for the original IP packet.
1793	 * From the discussion on IPv6imp in Feb 1999, the consensus was:
1794	 * - "attach as much as possible" is the goal
1795	 * - pad if not aligned (original size can be guessed by original
1796	 *   ip6 header)
1797	 * Following code adds the padding if it is simple enough,
1798	 * and truncates if not.
1799	 */
1800	if (m0->m_next || m0->m_pkthdr.len != m0->m_len)
1801		panic("assumption failed in %s:%d\n", __FILE__, __LINE__);
1802
1803	if (len - sizeof(*nd_opt_rh) < m0->m_pkthdr.len) {
1804		/* not enough room, truncate */
1805		m0->m_pkthdr.len = m0->m_len = len - sizeof(*nd_opt_rh);
1806	} else {
1807		/* enough room, pad or truncate */
1808		size_t extra;
1809
1810		extra = m0->m_pkthdr.len % 8;
1811		if (extra) {
1812			/* pad if easy enough, truncate if not */
1813			if (8 - extra <= M_TRAILINGSPACE(m0)) {
1814				/* pad */
1815				m0->m_len += (8 - extra);
1816				m0->m_pkthdr.len += (8 - extra);
1817			} else {
1818				/* truncate */
1819				m0->m_pkthdr.len -= extra;
1820				m0->m_len -= extra;
1821			}
1822		}
1823		len = m0->m_pkthdr.len + sizeof(*nd_opt_rh);
1824		m0->m_pkthdr.len = m0->m_len = len - sizeof(*nd_opt_rh);
1825	}
1826
1827	nd_opt_rh = (struct nd_opt_rd_hdr *)p;
1828	bzero(nd_opt_rh, sizeof(*nd_opt_rh));
1829	nd_opt_rh->nd_opt_rh_type = ND_OPT_REDIRECTED_HEADER;
1830	nd_opt_rh->nd_opt_rh_len = len >> 3;
1831	p += sizeof(*nd_opt_rh);
1832	m->m_pkthdr.len = m->m_len = p - (u_char *)ip6;
1833
1834	/* connect m0 to m */
1835	m->m_next = m0;
1836	m->m_pkthdr.len = m->m_len + m0->m_len;
1837    }
1838#ifdef M_DECRYPTED	/*not openbsd*/
1839noredhdropt:;
1840#endif
1841
1842	if (IN6_IS_ADDR_LINKLOCAL(&sip6->ip6_src))
1843		sip6->ip6_src.s6_addr16[1] = 0;
1844	if (IN6_IS_ADDR_LINKLOCAL(&sip6->ip6_dst))
1845		sip6->ip6_dst.s6_addr16[1] = 0;
1846#if 0
1847	if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_src))
1848		ip6->ip6_src.s6_addr16[1] = 0;
1849	if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst))
1850		ip6->ip6_dst.s6_addr16[1] = 0;
1851#endif
1852	if (IN6_IS_ADDR_LINKLOCAL(&nd_rd->nd_rd_target))
1853		nd_rd->nd_rd_target.s6_addr16[1] = 0;
1854	if (IN6_IS_ADDR_LINKLOCAL(&nd_rd->nd_rd_dst))
1855		nd_rd->nd_rd_dst.s6_addr16[1] = 0;
1856
1857	ip6->ip6_plen = htons(m->m_pkthdr.len - sizeof(struct ip6_hdr));
1858
1859	nd_rd->nd_rd_cksum = 0;
1860	nd_rd->nd_rd_cksum
1861		= in6_cksum(m, IPPROTO_ICMPV6, sizeof(*ip6), ntohs(ip6->ip6_plen));
1862
1863	/* send the packet to outside... */
1864#ifdef IPSEC
1865	m->m_pkthdr.rcvif = NULL;
1866#endif /*IPSEC*/
1867	ip6_output(m, NULL, NULL, 0, NULL, &outif);
1868	if (outif) {
1869		icmp6_ifstat_inc(outif, ifs6_out_msg);
1870		icmp6_ifstat_inc(outif, ifs6_out_redirect);
1871	}
1872	icmp6stat.icp6s_outhist[ND_REDIRECT]++;
1873
1874	return;
1875
1876fail:
1877	if (m)
1878		m_freem(m);
1879	if (m0)
1880		m_freem(m0);
1881}
1882
1883/*
1884 * ICMPv6 socket option processing.
1885 */
1886int
1887icmp6_ctloutput(op, so, level, optname, mp)
1888	int op;
1889	struct socket *so;
1890	int level, optname;
1891	struct mbuf **mp;
1892{
1893	register struct in6pcb *in6p = sotoin6pcb(so);
1894	register struct mbuf *m = *mp;
1895	int error = 0;
1896
1897	if (level != IPPROTO_ICMPV6) {
1898		error = EINVAL;
1899		if (op == PRCO_SETOPT && m)
1900			(void)m_free(m);
1901	} else switch(op) {
1902	 case PRCO_SETOPT:
1903		 switch (optname) {
1904		  case ICMP6_FILTER:
1905		  {
1906			  struct icmp6_filter *p;
1907
1908			  p = mtod(m, struct icmp6_filter *);
1909			  if (!p || !in6p->in6p_icmp6filt) {
1910				  error = EINVAL;
1911				  break;
1912			  }
1913			  bcopy(p, in6p->in6p_icmp6filt,
1914				sizeof(struct icmp6_filter));
1915			  error = 0;
1916			  break;
1917		  }
1918
1919		  default:
1920			  error = ENOPROTOOPT;
1921			  break;
1922		 }
1923		 if (m)
1924			 (void)m_free(m);
1925		 break;
1926
1927	 case PRCO_GETOPT:
1928		 switch (optname) {
1929		  case ICMP6_FILTER:
1930		  {
1931			  struct icmp6_filter *p;
1932
1933			  if (!in6p->in6p_icmp6filt) {
1934				  error = EINVAL;
1935				  break;
1936			  }
1937			  *mp = m = m_get(M_WAIT, MT_SOOPTS);
1938			  m->m_len = sizeof(struct icmp6_filter);
1939			  p = mtod(m, struct icmp6_filter *);
1940			  bcopy(in6p->in6p_icmp6filt, p,
1941				sizeof(struct icmp6_filter));
1942			  error = 0;
1943			  break;
1944		  }
1945
1946		  default:
1947			  error = ENOPROTOOPT;
1948			  break;
1949		 }
1950		 break;
1951	}
1952
1953	return(error);
1954}
1955
1956/*
1957 * Perform rate limit check.
1958 * Returns 0 if it is okay to send the icmp6 packet.
1959 * Returns 1 if the router SHOULD NOT send this icmp6 packet due to rate
1960 * limitation.
1961 *
1962 * XXX per-destination/type check necessary?
1963 */
1964static int
1965icmp6_ratelimit(dst, type, code)
1966	const struct in6_addr *dst;	/* not used at this moment */
1967	const int type;			/* not used at this moment */
1968	const int code;			/* not used at this moment */
1969{
1970	struct timeval tp;
1971	long sec_diff, usec_diff;
1972
1973	/* If we are not doing rate limitation, it is always okay to send */
1974	if (!icmp6errratelim)
1975		return 0;
1976
1977	tp = time;
1978	if (tp.tv_sec < icmp6_nextsend.tv_sec
1979	 || (tp.tv_sec == icmp6_nextsend.tv_sec
1980	  && tp.tv_usec < icmp6_nextsend.tv_usec)) {
1981		/* The packet is subject to rate limit */
1982		return 1;
1983	}
1984	sec_diff = icmp6errratelim / 1000000;
1985	usec_diff = icmp6errratelim % 1000000;
1986	icmp6_nextsend.tv_sec = tp.tv_sec + sec_diff;
1987	if ((tp.tv_usec = tp.tv_usec + usec_diff) >= 1000000) {
1988		icmp6_nextsend.tv_sec++;
1989		icmp6_nextsend.tv_usec -= 1000000;
1990	}
1991
1992	/* it is okay to send this */
1993	return 0;
1994}
1995
1996static struct rtentry *
1997icmp6_mtudisc_clone(dst)
1998	struct sockaddr *dst;
1999{
2000	struct rtentry *rt;
2001	int    error;
2002
2003	rt = rtalloc1(dst, 1);
2004	if (rt == 0)
2005		return NULL;
2006
2007	/* If we didn't get a host route, allocate one */
2008	if ((rt->rt_flags & RTF_HOST) == 0) {
2009		struct rtentry *nrt;
2010
2011		error = rtrequest((int) RTM_ADD, dst,
2012		    (struct sockaddr *) rt->rt_gateway,
2013		    (struct sockaddr *) 0,
2014		    RTF_GATEWAY | RTF_HOST | RTF_DYNAMIC, &nrt);
2015		if (error) {
2016			rtfree(rt);
2017			rtfree(nrt);
2018			return NULL;
2019		}
2020		nrt->rt_rmx = rt->rt_rmx;
2021		rtfree(rt);
2022		rt = nrt;
2023	}
2024	error = rt_timer_add(rt, icmp6_mtudisc_timeout,
2025			icmp6_mtudisc_timeout_q);
2026	if (error) {
2027		rtfree(rt);
2028		return NULL;
2029	}
2030
2031	return rt;	/* caller need to call rtfree() */
2032}
2033
2034static void
2035icmp6_mtudisc_timeout(rt, r)
2036	struct rtentry *rt;
2037	struct rttimer *r;
2038{
2039	if (rt == NULL)
2040		panic("icmp6_mtudisc_timeout: bad route to timeout");
2041	if ((rt->rt_flags & (RTF_DYNAMIC | RTF_HOST)) ==
2042	    (RTF_DYNAMIC | RTF_HOST)) {
2043		rtrequest((int) RTM_DELETE, (struct sockaddr *)rt_key(rt),
2044		    rt->rt_gateway, rt_mask(rt), rt->rt_flags, 0);
2045	} else {
2046		if ((rt->rt_rmx.rmx_locks & RTV_MTU) == 0) {
2047			rt->rt_rmx.rmx_mtu = 0;
2048		}
2049	}
2050}
2051
2052#include <vm/vm.h>
2053#include <sys/sysctl.h>
2054int
2055icmp6_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
2056	int *name;
2057	u_int namelen;
2058	void *oldp;
2059	size_t *oldlenp;
2060	void *newp;
2061	size_t newlen;
2062{
2063
2064	/* All sysctl names at this level are terminal. */
2065	if (namelen != 1)
2066		return ENOTDIR;
2067
2068	switch (name[0]) {
2069
2070	case ICMPV6CTL_REDIRACCEPT:
2071		return sysctl_int(oldp, oldlenp, newp, newlen,
2072				&icmp6_rediraccept);
2073	case ICMPV6CTL_REDIRTIMEOUT:
2074		return sysctl_int(oldp, oldlenp, newp, newlen,
2075				&icmp6_redirtimeout);
2076	case ICMPV6CTL_STATS:
2077		return sysctl_rdstruct(oldp, oldlenp, newp,
2078				&icmp6stat, sizeof(icmp6stat));
2079	case ICMPV6CTL_ERRRATELIMIT:
2080		return sysctl_int(oldp, oldlenp, newp, newlen,
2081				  &icmp6errratelim);
2082	case ICMPV6CTL_ND6_PRUNE:
2083		return sysctl_int(oldp, oldlenp, newp, newlen, &nd6_prune);
2084	case ICMPV6CTL_ND6_DELAY:
2085		return sysctl_int(oldp, oldlenp, newp, newlen, &nd6_delay);
2086	case ICMPV6CTL_ND6_UMAXTRIES:
2087		return sysctl_int(oldp, oldlenp, newp, newlen, &nd6_umaxtries);
2088	case ICMPV6CTL_ND6_MMAXTRIES:
2089		return sysctl_int(oldp, oldlenp, newp, newlen, &nd6_mmaxtries);
2090	case ICMPV6CTL_ND6_USELOOPBACK:
2091		return sysctl_int(oldp, oldlenp, newp, newlen,
2092				&nd6_useloopback);
2093	case ICMPV6CTL_ND6_PROXYALL:
2094		return sysctl_int(oldp, oldlenp, newp, newlen, &nd6_proxyall);
2095	case ICMPV6CTL_NODEINFO:
2096		return sysctl_int(oldp, oldlenp, newp, newlen, &icmp6_nodeinfo);
2097	default:
2098		return ENOPROTOOPT;
2099	}
2100	/* NOTREACHED */
2101}
2102