xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 8b0f9554ff8762542c4defc4f70e1eb76fb508fa)
1 /*	$NetBSD: ip6_input.c,v 1.113 2007/12/04 10:27:34 dyoung Exp $	*/
2 /*	$KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $	*/
3 
4 /*
5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the project nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 /*
34  * Copyright (c) 1982, 1986, 1988, 1993
35  *	The Regents of the University of California.  All rights reserved.
36  *
37  * Redistribution and use in source and binary forms, with or without
38  * modification, are permitted provided that the following conditions
39  * are met:
40  * 1. Redistributions of source code must retain the above copyright
41  *    notice, this list of conditions and the following disclaimer.
42  * 2. Redistributions in binary form must reproduce the above copyright
43  *    notice, this list of conditions and the following disclaimer in the
44  *    documentation and/or other materials provided with the distribution.
45  * 3. Neither the name of the University nor the names of its contributors
46  *    may be used to endorse or promote products derived from this software
47  *    without specific prior written permission.
48  *
49  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59  * SUCH DAMAGE.
60  *
61  *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
62  */
63 
64 #include <sys/cdefs.h>
65 __KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.113 2007/12/04 10:27:34 dyoung Exp $");
66 
67 #include "opt_inet.h"
68 #include "opt_inet6.h"
69 #include "opt_ipsec.h"
70 #include "opt_pfil_hooks.h"
71 
72 #include <sys/param.h>
73 #include <sys/systm.h>
74 #include <sys/malloc.h>
75 #include <sys/mbuf.h>
76 #include <sys/domain.h>
77 #include <sys/protosw.h>
78 #include <sys/socket.h>
79 #include <sys/socketvar.h>
80 #include <sys/errno.h>
81 #include <sys/time.h>
82 #include <sys/kernel.h>
83 #include <sys/syslog.h>
84 #include <sys/proc.h>
85 #include <sys/sysctl.h>
86 
87 #include <net/if.h>
88 #include <net/if_types.h>
89 #include <net/if_dl.h>
90 #include <net/route.h>
91 #include <net/netisr.h>
92 #ifdef PFIL_HOOKS
93 #include <net/pfil.h>
94 #endif
95 
96 #include <netinet/in.h>
97 #include <netinet/in_systm.h>
98 #ifdef INET
99 #include <netinet/ip.h>
100 #include <netinet/ip_icmp.h>
101 #endif /* INET */
102 #include <netinet/ip6.h>
103 #include <netinet6/in6_var.h>
104 #include <netinet6/ip6_var.h>
105 #include <netinet6/in6_pcb.h>
106 #include <netinet/icmp6.h>
107 #include <netinet6/scope6_var.h>
108 #include <netinet6/in6_ifattach.h>
109 #include <netinet6/nd6.h>
110 
111 #ifdef IPSEC
112 #include <netinet6/ipsec.h>
113 #endif
114 
115 #ifdef FAST_IPSEC
116 #include <netipsec/ipsec.h>
117 #include <netipsec/ipsec6.h>
118 #include <netipsec/key.h>
119 #endif /* FAST_IPSEC */
120 
121 #include <netinet6/ip6protosw.h>
122 
123 #include "faith.h"
124 #include "gif.h"
125 
126 #if NGIF > 0
127 #include <netinet6/in6_gif.h>
128 #endif
129 
130 #include <net/net_osdep.h>
131 
132 extern struct domain inet6domain;
133 
134 u_char ip6_protox[IPPROTO_MAX];
135 static int ip6qmaxlen = IFQ_MAXLEN;
136 struct in6_ifaddr *in6_ifaddr;
137 struct ifqueue ip6intrq;
138 
139 extern callout_t in6_tmpaddrtimer_ch;
140 
141 int ip6_forward_srcrt;			/* XXX */
142 int ip6_sourcecheck;			/* XXX */
143 int ip6_sourcecheck_interval;		/* XXX */
144 
145 #ifdef PFIL_HOOKS
146 struct pfil_head inet6_pfil_hook;
147 #endif
148 
149 struct ip6stat ip6stat;
150 
151 static void ip6_init2(void *);
152 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
153 
154 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
155 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
156 
157 /*
158  * IP6 initialization: fill in IP6 protocol switch table.
159  * All protocols not implemented in kernel go to raw IP6 protocol handler.
160  */
161 void
162 ip6_init()
163 {
164 	const struct ip6protosw *pr;
165 	int i;
166 
167 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
168 	if (pr == 0)
169 		panic("ip6_init");
170 	for (i = 0; i < IPPROTO_MAX; i++)
171 		ip6_protox[i] = pr - inet6sw;
172 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
173 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
174 		if (pr->pr_domain->dom_family == PF_INET6 &&
175 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
176 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
177 	ip6intrq.ifq_maxlen = ip6qmaxlen;
178 	scope6_init();
179 	addrsel_policy_init();
180 	nd6_init();
181 	frag6_init();
182 	ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
183 
184 	ip6_init2((void *)0);
185 #ifdef GATEWAY
186 	ip6flow_init(ip6_hashsize);
187 #endif
188 
189 #ifdef PFIL_HOOKS
190 	/* Register our Packet Filter hook. */
191 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
192 	inet6_pfil_hook.ph_af   = AF_INET6;
193 	i = pfil_head_register(&inet6_pfil_hook);
194 	if (i != 0)
195 		printf("ip6_init: WARNING: unable to register pfil hook, "
196 		    "error %d\n", i);
197 #endif /* PFIL_HOOKS */
198 }
199 
200 static void
201 ip6_init2(void *dummy)
202 {
203 
204 	/* nd6_timer_init */
205 	callout_init(&nd6_timer_ch, 0);
206 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
207 
208 	/* timer for regeneranation of temporary addresses randomize ID */
209 	callout_init(&in6_tmpaddrtimer_ch, 0);
210 	callout_reset(&in6_tmpaddrtimer_ch,
211 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
212 		       ip6_temp_regen_advance) * hz,
213 		      in6_tmpaddrtimer, NULL);
214 }
215 
216 /*
217  * IP6 input interrupt handling. Just pass the packet to ip6_input.
218  */
219 void
220 ip6intr()
221 {
222 	int s;
223 	struct mbuf *m;
224 
225 	for (;;) {
226 		s = splnet();
227 		IF_DEQUEUE(&ip6intrq, m);
228 		splx(s);
229 		if (m == 0)
230 			return;
231 		/* drop the packet if IPv6 operation is disabled on the IF */
232 		if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
233 			m_freem(m);
234 			return;
235 		}
236 		ip6_input(m);
237 	}
238 }
239 
240 extern struct	route ip6_forward_rt;
241 
242 void
243 ip6_input(struct mbuf *m)
244 {
245 	struct ip6_hdr *ip6;
246 	int hit, off = sizeof(struct ip6_hdr), nest;
247 	u_int32_t plen;
248 	u_int32_t rtalert = ~0;
249 	int nxt, ours = 0, rh_present = 0;
250 	struct ifnet *deliverifp = NULL;
251 	int srcrt = 0;
252 	const struct rtentry *rt;
253 	union {
254 		struct sockaddr		dst;
255 		struct sockaddr_in6	dst6;
256 	} u;
257 #ifdef FAST_IPSEC
258 	struct m_tag *mtag;
259 	struct tdb_ident *tdbi;
260 	struct secpolicy *sp;
261 	int s, error;
262 #endif
263 
264 #ifdef IPSEC
265 	/*
266 	 * should the inner packet be considered authentic?
267 	 * see comment in ah4_input().
268 	 */
269 	m->m_flags &= ~M_AUTHIPHDR;
270 	m->m_flags &= ~M_AUTHIPDGM;
271 #endif
272 
273 	/*
274 	 * make sure we don't have onion peering information into m_tag.
275 	 */
276 	ip6_delaux(m);
277 
278 	/*
279 	 * mbuf statistics
280 	 */
281 	if (m->m_flags & M_EXT) {
282 		if (m->m_next)
283 			ip6stat.ip6s_mext2m++;
284 		else
285 			ip6stat.ip6s_mext1++;
286 	} else {
287 #define M2MMAX	__arraycount(ip6stat.ip6s_m2m)
288 		if (m->m_next) {
289 			if (m->m_flags & M_LOOP) {
290 				ip6stat.ip6s_m2m[lo0ifp->if_index]++; /* XXX */
291 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
292 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
293 			else
294 				ip6stat.ip6s_m2m[0]++;
295 		} else
296 			ip6stat.ip6s_m1++;
297 #undef M2MMAX
298 	}
299 
300 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
301 	ip6stat.ip6s_total++;
302 
303 	/*
304 	 * If the IPv6 header is not aligned, slurp it up into a new
305 	 * mbuf with space for link headers, in the event we forward
306 	 * it.  Otherwise, if it is aligned, make sure the entire base
307 	 * IPv6 header is in the first mbuf of the chain.
308 	 */
309 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
310 		struct ifnet *inifp = m->m_pkthdr.rcvif;
311 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
312 				  (max_linkhdr + 3) & ~3)) == NULL) {
313 			/* XXXJRT new stat, please */
314 			ip6stat.ip6s_toosmall++;
315 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
316 			return;
317 		}
318 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
319 		struct ifnet *inifp = m->m_pkthdr.rcvif;
320 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
321 			ip6stat.ip6s_toosmall++;
322 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
323 			return;
324 		}
325 	}
326 
327 	ip6 = mtod(m, struct ip6_hdr *);
328 
329 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
330 		ip6stat.ip6s_badvers++;
331 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
332 		goto bad;
333 	}
334 
335 #if defined(IPSEC)
336 	/* IPv6 fast forwarding is not compatible with IPsec. */
337 	m->m_flags &= ~M_CANFASTFWD;
338 #else
339 	/*
340 	 * Assume that we can create a fast-forward IP flow entry
341 	 * based on this packet.
342 	 */
343 	m->m_flags |= M_CANFASTFWD;
344 #endif
345 
346 #ifdef PFIL_HOOKS
347 	/*
348 	 * Run through list of hooks for input packets.  If there are any
349 	 * filters which require that additional packets in the flow are
350 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
351 	 * Note that filters must _never_ set this flag, as another filter
352 	 * in the list may have previously cleared it.
353 	 */
354 	/*
355 	 * let ipfilter look at packet on the wire,
356 	 * not the decapsulated packet.
357 	 */
358 #ifdef IPSEC
359 	if (!ipsec_getnhist(m))
360 #elif defined(FAST_IPSEC)
361 	if (!ipsec_indone(m))
362 #else
363 	if (1)
364 #endif
365 	{
366 		struct in6_addr odst;
367 
368 		odst = ip6->ip6_dst;
369 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
370 				   PFIL_IN) != 0)
371 			return;
372 		if (m == NULL)
373 			return;
374 		ip6 = mtod(m, struct ip6_hdr *);
375 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
376 	}
377 #endif /* PFIL_HOOKS */
378 
379 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
380 
381 #ifdef ALTQ
382 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
383 		/* packet is dropped by traffic conditioner */
384 		return;
385 	}
386 #endif
387 
388 	/*
389 	 * Check against address spoofing/corruption.
390 	 */
391 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
392 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
393 		/*
394 		 * XXX: "badscope" is not very suitable for a multicast source.
395 		 */
396 		ip6stat.ip6s_badscope++;
397 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
398 		goto bad;
399 	}
400 	/*
401 	 * The following check is not documented in specs.  A malicious
402 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
403 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
404 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
405 	 *
406 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
407 	 * support IPv4-less kernel compilation, we cannot support SIIT
408 	 * environment at all.  So, it makes more sense for us to reject any
409 	 * malicious packets for non-SIIT environment, than try to do a
410 	 * partial support for SIIT environment.
411 	 */
412 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
413 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
414 		ip6stat.ip6s_badscope++;
415 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
416 		goto bad;
417 	}
418 #if 0
419 	/*
420 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
421 	 *
422 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
423 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
424 	 * is revised to forbid relaying case.
425 	 */
426 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
427 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
428 		ip6stat.ip6s_badscope++;
429 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
430 		goto bad;
431 	}
432 #endif
433 
434 	/*
435 	 * Disambiguate address scope zones (if there is ambiguity).
436 	 * We first make sure that the original source or destination address
437 	 * is not in our internal form for scoped addresses.  Such addresses
438 	 * are not necessarily invalid spec-wise, but we cannot accept them due
439 	 * to the usage conflict.
440 	 * in6_setscope() then also checks and rejects the cases where src or
441 	 * dst are the loopback address and the receiving interface
442 	 * is not loopback.
443 	 */
444 	if (__predict_false(
445 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
446 		goto bad;
447 	ip6 = mtod(m, struct ip6_hdr *);
448 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
449 		ip6stat.ip6s_badscope++; /* XXX */
450 		goto bad;
451 	}
452 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
453 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
454 		ip6stat.ip6s_badscope++;
455 		goto bad;
456 	}
457 
458 	/*
459 	 * Multicast check
460 	 */
461 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
462 	  	struct	in6_multi *in6m = 0;
463 
464 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
465 		/*
466 		 * See if we belong to the destination multicast group on the
467 		 * arrival interface.
468 		 */
469 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
470 		if (in6m)
471 			ours = 1;
472 		else if (!ip6_mrouter) {
473 			ip6stat.ip6s_notmember++;
474 			ip6stat.ip6s_cantforward++;
475 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
476 			goto bad;
477 		}
478 		deliverifp = m->m_pkthdr.rcvif;
479 		goto hbhcheck;
480 	}
481 
482 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
483 
484 	/*
485 	 *  Unicast check
486 	 */
487 	rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit);
488 	if (hit)
489 		ip6stat.ip6s_forward_cachehit++;
490 	else
491 		ip6stat.ip6s_forward_cachemiss++;
492 
493 #define rt6_getkey(__rt) satocsin6(rt_getkey(__rt))
494 
495 	/*
496 	 * Accept the packet if the forwarding interface to the destination
497 	 * according to the routing table is the loopback interface,
498 	 * unless the associated route has a gateway.
499 	 * Note that this approach causes to accept a packet if there is a
500 	 * route to the loopback interface for the destination of the packet.
501 	 * But we think it's even useful in some situations, e.g. when using
502 	 * a special daemon which wants to intercept the packet.
503 	 */
504 	if (rt != NULL &&
505 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
506 	    !(rt->rt_flags & RTF_CLONED) &&
507 #if 0
508 	    /*
509 	     * The check below is redundant since the comparison of
510 	     * the destination and the key of the rtentry has
511 	     * already done through looking up the routing table.
512 	     */
513 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(rt)->sin6_addr) &&
514 #endif
515 	    rt->rt_ifp->if_type == IFT_LOOP) {
516 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
517 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
518 			m->m_flags |= M_ANYCAST6;
519 		/*
520 		 * packets to a tentative, duplicated, or somehow invalid
521 		 * address must not be accepted.
522 		 */
523 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
524 			/* this address is ready */
525 			ours = 1;
526 			deliverifp = ia6->ia_ifp;	/* correct? */
527 			goto hbhcheck;
528 		} else {
529 			/* address is not ready, so discard the packet. */
530 			nd6log((LOG_INFO,
531 			    "ip6_input: packet to an unready address %s->%s\n",
532 			    ip6_sprintf(&ip6->ip6_src),
533 			    ip6_sprintf(&ip6->ip6_dst)));
534 
535 			goto bad;
536 		}
537 	}
538 
539 	/*
540 	 * FAITH (Firewall Aided Internet Translator)
541 	 */
542 #if defined(NFAITH) && 0 < NFAITH
543 	if (ip6_keepfaith) {
544 		if (rt != NULL && rt->rt_ifp != NULL &&
545 		    rt->rt_ifp->if_type == IFT_FAITH) {
546 			/* XXX do we need more sanity checks? */
547 			ours = 1;
548 			deliverifp = rt->rt_ifp; /* faith */
549 			goto hbhcheck;
550 		}
551 	}
552 #endif
553 
554 #if 0
555     {
556 	/*
557 	 * Last resort: check in6_ifaddr for incoming interface.
558 	 * The code is here until I update the "goto ours hack" code above
559 	 * working right.
560 	 */
561 	struct ifaddr *ifa;
562 	IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
563 		if (ifa->ifa_addr == NULL)
564 			continue;	/* just for safety */
565 		if (ifa->ifa_addr->sa_family != AF_INET6)
566 			continue;
567 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
568 			ours = 1;
569 			deliverifp = ifa->ifa_ifp;
570 			goto hbhcheck;
571 		}
572 	}
573     }
574 #endif
575 
576 	/*
577 	 * Now there is no reason to process the packet if it's not our own
578 	 * and we're not a router.
579 	 */
580 	if (!ip6_forwarding) {
581 		ip6stat.ip6s_cantforward++;
582 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
583 		goto bad;
584 	}
585 
586   hbhcheck:
587 	/*
588 	 * record address information into m_tag, if we don't have one yet.
589 	 * note that we are unable to record it, if the address is not listed
590 	 * as our interface address (e.g. multicast addresses, addresses
591 	 * within FAITH prefixes and such).
592 	 */
593 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
594 		struct in6_ifaddr *ia6;
595 
596 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
597 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
598 			/*
599 			 * XXX maybe we should drop the packet here,
600 			 * as we could not provide enough information
601 			 * to the upper layers.
602 			 */
603 		}
604 	}
605 
606 	/*
607 	 * Process Hop-by-Hop options header if it's contained.
608 	 * m may be modified in ip6_hopopts_input().
609 	 * If a JumboPayload option is included, plen will also be modified.
610 	 */
611 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
612 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
613 		struct ip6_hbh *hbh;
614 
615 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
616 #if 0	/*touches NULL pointer*/
617 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
618 #endif
619 			return;	/* m have already been freed */
620 		}
621 
622 		/* adjust pointer */
623 		ip6 = mtod(m, struct ip6_hdr *);
624 
625 		/*
626 		 * if the payload length field is 0 and the next header field
627 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
628 		 * option MUST be included.
629 		 */
630 		if (ip6->ip6_plen == 0 && plen == 0) {
631 			/*
632 			 * Note that if a valid jumbo payload option is
633 			 * contained, ip6_hopopts_input() must set a valid
634 			 * (non-zero) payload length to the variable plen.
635 			 */
636 			ip6stat.ip6s_badoptions++;
637 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
638 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
639 			icmp6_error(m, ICMP6_PARAM_PROB,
640 				    ICMP6_PARAMPROB_HEADER,
641 				    (char *)&ip6->ip6_plen - (char *)ip6);
642 			return;
643 		}
644 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
645 			sizeof(struct ip6_hbh));
646 		if (hbh == NULL) {
647 			ip6stat.ip6s_tooshort++;
648 			return;
649 		}
650 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
651 		nxt = hbh->ip6h_nxt;
652 
653 		/*
654 		 * accept the packet if a router alert option is included
655 		 * and we act as an IPv6 router.
656 		 */
657 		if (rtalert != ~0 && ip6_forwarding)
658 			ours = 1;
659 	} else
660 		nxt = ip6->ip6_nxt;
661 
662 	/*
663 	 * Check that the amount of data in the buffers
664 	 * is as at least much as the IPv6 header would have us expect.
665 	 * Trim mbufs if longer than we expect.
666 	 * Drop packet if shorter than we expect.
667 	 */
668 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
669 		ip6stat.ip6s_tooshort++;
670 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
671 		goto bad;
672 	}
673 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
674 		if (m->m_len == m->m_pkthdr.len) {
675 			m->m_len = sizeof(struct ip6_hdr) + plen;
676 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
677 		} else
678 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
679 	}
680 
681 	/*
682 	 * Forward if desirable.
683 	 */
684 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
685 		/*
686 		 * If we are acting as a multicast router, all
687 		 * incoming multicast packets are passed to the
688 		 * kernel-level multicast forwarding function.
689 		 * The packet is returned (relatively) intact; if
690 		 * ip6_mforward() returns a non-zero value, the packet
691 		 * must be discarded, else it may be accepted below.
692 		 */
693 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
694 			ip6stat.ip6s_cantforward++;
695 			m_freem(m);
696 			return;
697 		}
698 		if (!ours) {
699 			m_freem(m);
700 			return;
701 		}
702 	} else if (!ours) {
703 		ip6_forward(m, srcrt);
704 		return;
705 	}
706 
707 	ip6 = mtod(m, struct ip6_hdr *);
708 
709 	/*
710 	 * Malicious party may be able to use IPv4 mapped addr to confuse
711 	 * tcp/udp stack and bypass security checks (act as if it was from
712 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
713 	 *
714 	 * For SIIT end node behavior, you may want to disable the check.
715 	 * However, you will  become vulnerable to attacks using IPv4 mapped
716 	 * source.
717 	 */
718 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
719 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
720 		ip6stat.ip6s_badscope++;
721 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
722 		goto bad;
723 	}
724 
725 	/*
726 	 * Tell launch routine the next header
727 	 */
728 #ifdef IFA_STATS
729 	if (deliverifp != NULL) {
730 		struct in6_ifaddr *ia6;
731 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
732 		if (ia6)
733 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
734 	}
735 #endif
736 	ip6stat.ip6s_delivered++;
737 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
738 	nest = 0;
739 
740 	rh_present = 0;
741 	while (nxt != IPPROTO_DONE) {
742 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
743 			ip6stat.ip6s_toomanyhdr++;
744 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
745 			goto bad;
746 		}
747 
748 		/*
749 		 * protection against faulty packet - there should be
750 		 * more sanity checks in header chain processing.
751 		 */
752 		if (m->m_pkthdr.len < off) {
753 			ip6stat.ip6s_tooshort++;
754 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
755 			goto bad;
756 		}
757 
758 		if (nxt == IPPROTO_ROUTING) {
759 			if (rh_present++) {
760 				in6_ifstat_inc(m->m_pkthdr.rcvif,
761 				    ifs6_in_hdrerr);
762 				ip6stat.ip6s_badoptions++;
763 				goto bad;
764 			}
765 		}
766 
767 #ifdef IPSEC
768 		/*
769 		 * enforce IPsec policy checking if we are seeing last header.
770 		 * note that we do not visit this with protocols with pcb layer
771 		 * code - like udp/tcp/raw ip.
772 		 */
773 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
774 		    ipsec6_in_reject(m, NULL)) {
775 			ipsec6stat.in_polvio++;
776 			goto bad;
777 		}
778 #endif
779 #ifdef FAST_IPSEC
780 	/*
781 	 * enforce IPsec policy checking if we are seeing last header.
782 	 * note that we do not visit this with protocols with pcb layer
783 	 * code - like udp/tcp/raw ip.
784 	 */
785 	if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) {
786 		/*
787 		 * Check if the packet has already had IPsec processing
788 		 * done.  If so, then just pass it along.  This tag gets
789 		 * set during AH, ESP, etc. input handling, before the
790 		 * packet is returned to the ip input queue for delivery.
791 		 */
792 		mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
793 		s = splsoftnet();
794 		if (mtag != NULL) {
795 			tdbi = (struct tdb_ident *)(mtag + 1);
796 			sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
797 		} else {
798 			sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
799 									IP_FORWARDING, &error);
800 		}
801 		if (sp != NULL) {
802 			/*
803 			 * Check security policy against packet attributes.
804 			 */
805 			error = ipsec_in_reject(sp, m);
806 			KEY_FREESP(&sp);
807 		} else {
808 			/* XXX error stat??? */
809 			error = EINVAL;
810 			DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/
811 		}
812 		splx(s);
813 		if (error)
814 			goto bad;
815 	}
816 #endif /* FAST_IPSEC */
817 
818 
819 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
820 	}
821 	return;
822  bad:
823 	m_freem(m);
824 }
825 
826 /*
827  * set/grab in6_ifaddr correspond to IPv6 destination address.
828  */
829 static struct m_tag *
830 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
831 {
832 	struct m_tag *mtag;
833 
834 	mtag = ip6_addaux(m);
835 	if (mtag != NULL) {
836 		struct ip6aux *ip6a;
837 
838 		ip6a = (struct ip6aux *)(mtag + 1);
839 		in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id);
840 		ip6a->ip6a_src = ia->ia_addr.sin6_addr;
841 		ip6a->ip6a_flags = ia->ia6_flags;
842 	}
843 	return mtag;	/* NULL if failed to set */
844 }
845 
846 const struct ip6aux *
847 ip6_getdstifaddr(struct mbuf *m)
848 {
849 	struct m_tag *mtag;
850 
851 	mtag = ip6_findaux(m);
852 	if (mtag != NULL)
853 		return (struct ip6aux *)(mtag + 1);
854 	else
855 		return NULL;
856 }
857 
858 /*
859  * Hop-by-Hop options header processing. If a valid jumbo payload option is
860  * included, the real payload length will be stored in plenp.
861  *
862  * rtalertp - XXX: should be stored more smart way
863  */
864 static int
865 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
866 	struct mbuf **mp, int *offp)
867 {
868 	struct mbuf *m = *mp;
869 	int off = *offp, hbhlen;
870 	struct ip6_hbh *hbh;
871 
872 	/* validation of the length of the header */
873 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
874 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
875 	if (hbh == NULL) {
876 		ip6stat.ip6s_tooshort++;
877 		return -1;
878 	}
879 	hbhlen = (hbh->ip6h_len + 1) << 3;
880 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
881 		hbhlen);
882 	if (hbh == NULL) {
883 		ip6stat.ip6s_tooshort++;
884 		return -1;
885 	}
886 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
887 	off += hbhlen;
888 	hbhlen -= sizeof(struct ip6_hbh);
889 
890 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
891 				hbhlen, rtalertp, plenp) < 0)
892 		return (-1);
893 
894 	*offp = off;
895 	*mp = m;
896 	return (0);
897 }
898 
899 /*
900  * Search header for all Hop-by-hop options and process each option.
901  * This function is separate from ip6_hopopts_input() in order to
902  * handle a case where the sending node itself process its hop-by-hop
903  * options header. In such a case, the function is called from ip6_output().
904  *
905  * The function assumes that hbh header is located right after the IPv6 header
906  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
907  * opthead + hbhlen is located in continuous memory region.
908  */
909 int
910 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
911 	u_int32_t *rtalertp, u_int32_t *plenp)
912 {
913 	struct ip6_hdr *ip6;
914 	int optlen = 0;
915 	u_int8_t *opt = opthead;
916 	u_int16_t rtalert_val;
917 	u_int32_t jumboplen;
918 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
919 
920 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
921 		switch (*opt) {
922 		case IP6OPT_PAD1:
923 			optlen = 1;
924 			break;
925 		case IP6OPT_PADN:
926 			if (hbhlen < IP6OPT_MINLEN) {
927 				ip6stat.ip6s_toosmall++;
928 				goto bad;
929 			}
930 			optlen = *(opt + 1) + 2;
931 			break;
932 		case IP6OPT_RTALERT:
933 			/* XXX may need check for alignment */
934 			if (hbhlen < IP6OPT_RTALERT_LEN) {
935 				ip6stat.ip6s_toosmall++;
936 				goto bad;
937 			}
938 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
939 				/* XXX stat */
940 				icmp6_error(m, ICMP6_PARAM_PROB,
941 				    ICMP6_PARAMPROB_HEADER,
942 				    erroff + opt + 1 - opthead);
943 				return (-1);
944 			}
945 			optlen = IP6OPT_RTALERT_LEN;
946 			bcopy((void *)(opt + 2), (void *)&rtalert_val, 2);
947 			*rtalertp = ntohs(rtalert_val);
948 			break;
949 		case IP6OPT_JUMBO:
950 			/* XXX may need check for alignment */
951 			if (hbhlen < IP6OPT_JUMBO_LEN) {
952 				ip6stat.ip6s_toosmall++;
953 				goto bad;
954 			}
955 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
956 				/* XXX stat */
957 				icmp6_error(m, ICMP6_PARAM_PROB,
958 				    ICMP6_PARAMPROB_HEADER,
959 				    erroff + opt + 1 - opthead);
960 				return (-1);
961 			}
962 			optlen = IP6OPT_JUMBO_LEN;
963 
964 			/*
965 			 * IPv6 packets that have non 0 payload length
966 			 * must not contain a jumbo payload option.
967 			 */
968 			ip6 = mtod(m, struct ip6_hdr *);
969 			if (ip6->ip6_plen) {
970 				ip6stat.ip6s_badoptions++;
971 				icmp6_error(m, ICMP6_PARAM_PROB,
972 				    ICMP6_PARAMPROB_HEADER,
973 				    erroff + opt - opthead);
974 				return (-1);
975 			}
976 
977 			/*
978 			 * We may see jumbolen in unaligned location, so
979 			 * we'd need to perform bcopy().
980 			 */
981 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
982 			jumboplen = (u_int32_t)htonl(jumboplen);
983 
984 #if 1
985 			/*
986 			 * if there are multiple jumbo payload options,
987 			 * *plenp will be non-zero and the packet will be
988 			 * rejected.
989 			 * the behavior may need some debate in ipngwg -
990 			 * multiple options does not make sense, however,
991 			 * there's no explicit mention in specification.
992 			 */
993 			if (*plenp != 0) {
994 				ip6stat.ip6s_badoptions++;
995 				icmp6_error(m, ICMP6_PARAM_PROB,
996 				    ICMP6_PARAMPROB_HEADER,
997 				    erroff + opt + 2 - opthead);
998 				return (-1);
999 			}
1000 #endif
1001 
1002 			/*
1003 			 * jumbo payload length must be larger than 65535.
1004 			 */
1005 			if (jumboplen <= IPV6_MAXPACKET) {
1006 				ip6stat.ip6s_badoptions++;
1007 				icmp6_error(m, ICMP6_PARAM_PROB,
1008 				    ICMP6_PARAMPROB_HEADER,
1009 				    erroff + opt + 2 - opthead);
1010 				return (-1);
1011 			}
1012 			*plenp = jumboplen;
1013 
1014 			break;
1015 		default:		/* unknown option */
1016 			if (hbhlen < IP6OPT_MINLEN) {
1017 				ip6stat.ip6s_toosmall++;
1018 				goto bad;
1019 			}
1020 			optlen = ip6_unknown_opt(opt, m,
1021 			    erroff + opt - opthead);
1022 			if (optlen == -1)
1023 				return (-1);
1024 			optlen += 2;
1025 			break;
1026 		}
1027 	}
1028 
1029 	return (0);
1030 
1031   bad:
1032 	m_freem(m);
1033 	return (-1);
1034 }
1035 
1036 /*
1037  * Unknown option processing.
1038  * The third argument `off' is the offset from the IPv6 header to the option,
1039  * which is necessary if the IPv6 header the and option header and IPv6 header
1040  * is not continuous in order to return an ICMPv6 error.
1041  */
1042 int
1043 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1044 {
1045 	struct ip6_hdr *ip6;
1046 
1047 	switch (IP6OPT_TYPE(*optp)) {
1048 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1049 		return ((int)*(optp + 1));
1050 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1051 		m_freem(m);
1052 		return (-1);
1053 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1054 		ip6stat.ip6s_badoptions++;
1055 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1056 		return (-1);
1057 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1058 		ip6stat.ip6s_badoptions++;
1059 		ip6 = mtod(m, struct ip6_hdr *);
1060 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1061 		    (m->m_flags & (M_BCAST|M_MCAST)))
1062 			m_freem(m);
1063 		else
1064 			icmp6_error(m, ICMP6_PARAM_PROB,
1065 				    ICMP6_PARAMPROB_OPTION, off);
1066 		return (-1);
1067 	}
1068 
1069 	m_freem(m);		/* XXX: NOTREACHED */
1070 	return (-1);
1071 }
1072 
1073 /*
1074  * Create the "control" list for this pcb.
1075  *
1076  * The routine will be called from upper layer handlers like tcp6_input().
1077  * Thus the routine assumes that the caller (tcp6_input) have already
1078  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1079  * very first mbuf on the mbuf chain.
1080  * We may want to add some infinite loop prevention or sanity checks for safety.
1081  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1082  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1083  */
1084 void
1085 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
1086 	struct ip6_hdr *ip6, struct mbuf *m)
1087 {
1088 #ifdef RFC2292
1089 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1090 #else
1091 #define IS2292(x, y)	(y)
1092 #endif
1093 
1094 #ifdef SO_TIMESTAMP
1095 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
1096 		struct timeval tv;
1097 
1098 		microtime(&tv);
1099 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1100 		    SCM_TIMESTAMP, SOL_SOCKET);
1101 		if (*mp)
1102 			mp = &(*mp)->m_next;
1103 	}
1104 #endif
1105 
1106 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
1107 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1108 		return;
1109 
1110 	/* RFC 2292 sec. 5 */
1111 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1112 		struct in6_pktinfo pi6;
1113 
1114 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1115 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1116 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
1117 		    m->m_pkthdr.rcvif->if_index : 0;
1118 		*mp = sbcreatecontrol((void *) &pi6,
1119 		    sizeof(struct in6_pktinfo),
1120 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1121 		if (*mp)
1122 			mp = &(*mp)->m_next;
1123 	}
1124 
1125 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1126 		int hlim = ip6->ip6_hlim & 0xff;
1127 
1128 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
1129 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1130 		if (*mp)
1131 			mp = &(*mp)->m_next;
1132 	}
1133 
1134 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1135 		u_int32_t flowinfo;
1136 		int tclass;
1137 
1138 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1139 		flowinfo >>= 20;
1140 
1141 		tclass = flowinfo & 0xff;
1142 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
1143 		    IPV6_TCLASS, IPPROTO_IPV6);
1144 
1145 		if (*mp)
1146 			mp = &(*mp)->m_next;
1147 	}
1148 
1149 	/*
1150 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1151 	 * privilege for the option (see ip6_ctloutput), but it might be too
1152 	 * strict, since there might be some hop-by-hop options which can be
1153 	 * returned to normal user.
1154 	 * See also RFC3542 section 8 (or RFC2292 section 6).
1155 	 */
1156 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1157 		/*
1158 		 * Check if a hop-by-hop options header is contatined in the
1159 		 * received packet, and if so, store the options as ancillary
1160 		 * data. Note that a hop-by-hop options header must be
1161 		 * just after the IPv6 header, which fact is assured through
1162 		 * the IPv6 input processing.
1163 		 */
1164 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1165 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1166 			struct ip6_hbh *hbh;
1167 			int hbhlen;
1168 			struct mbuf *ext;
1169 
1170 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1171 			    xip6->ip6_nxt);
1172 			if (ext == NULL) {
1173 				ip6stat.ip6s_tooshort++;
1174 				return;
1175 			}
1176 			hbh = mtod(ext, struct ip6_hbh *);
1177 			hbhlen = (hbh->ip6h_len + 1) << 3;
1178 			if (hbhlen != ext->m_len) {
1179 				m_freem(ext);
1180 				ip6stat.ip6s_tooshort++;
1181 				return;
1182 			}
1183 
1184 			/*
1185 			 * XXX: We copy whole the header even if a jumbo
1186 			 * payload option is included, which option is to
1187 			 * be removed before returning in the RFC 2292.
1188 			 * Note: this constraint is removed in RFC3542.
1189 			 */
1190 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
1191 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1192 			    IPPROTO_IPV6);
1193 			if (*mp)
1194 				mp = &(*mp)->m_next;
1195 			m_freem(ext);
1196 		}
1197 	}
1198 
1199 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1200 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1201 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1202 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1203 
1204 		/*
1205 		 * Search for destination options headers or routing
1206 		 * header(s) through the header chain, and stores each
1207 		 * header as ancillary data.
1208 		 * Note that the order of the headers remains in
1209 		 * the chain of ancillary data.
1210 		 */
1211 		for (;;) {	/* is explicit loop prevention necessary? */
1212 			struct ip6_ext *ip6e = NULL;
1213 			int elen;
1214 			struct mbuf *ext = NULL;
1215 
1216 			/*
1217 			 * if it is not an extension header, don't try to
1218 			 * pull it from the chain.
1219 			 */
1220 			switch (nxt) {
1221 			case IPPROTO_DSTOPTS:
1222 			case IPPROTO_ROUTING:
1223 			case IPPROTO_HOPOPTS:
1224 			case IPPROTO_AH: /* is it possible? */
1225 				break;
1226 			default:
1227 				goto loopend;
1228 			}
1229 
1230 			ext = ip6_pullexthdr(m, off, nxt);
1231 			if (ext == NULL) {
1232 				ip6stat.ip6s_tooshort++;
1233 				return;
1234 			}
1235 			ip6e = mtod(ext, struct ip6_ext *);
1236 			if (nxt == IPPROTO_AH)
1237 				elen = (ip6e->ip6e_len + 2) << 2;
1238 			else
1239 				elen = (ip6e->ip6e_len + 1) << 3;
1240 			if (elen != ext->m_len) {
1241 				m_freem(ext);
1242 				ip6stat.ip6s_tooshort++;
1243 				return;
1244 			}
1245 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1246 
1247 			switch (nxt) {
1248 			case IPPROTO_DSTOPTS:
1249 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
1250 					break;
1251 
1252 				*mp = sbcreatecontrol((void *)ip6e, elen,
1253 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1254 				    IPPROTO_IPV6);
1255 				if (*mp)
1256 					mp = &(*mp)->m_next;
1257 				break;
1258 
1259 			case IPPROTO_ROUTING:
1260 				if (!in6p->in6p_flags & IN6P_RTHDR)
1261 					break;
1262 
1263 				*mp = sbcreatecontrol((void *)ip6e, elen,
1264 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1265 				    IPPROTO_IPV6);
1266 				if (*mp)
1267 					mp = &(*mp)->m_next;
1268 				break;
1269 
1270 			case IPPROTO_HOPOPTS:
1271 			case IPPROTO_AH: /* is it possible? */
1272 				break;
1273 
1274 			default:
1275 				/*
1276 			 	 * other cases have been filtered in the above.
1277 				 * none will visit this case.  here we supply
1278 				 * the code just in case (nxt overwritten or
1279 				 * other cases).
1280 				 */
1281 				m_freem(ext);
1282 				goto loopend;
1283 
1284 			}
1285 
1286 			/* proceed with the next header. */
1287 			off += elen;
1288 			nxt = ip6e->ip6e_nxt;
1289 			ip6e = NULL;
1290 			m_freem(ext);
1291 			ext = NULL;
1292 		}
1293 	  loopend:
1294 	  	;
1295 	}
1296 }
1297 #undef IS2292
1298 
1299 
1300 void
1301 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
1302     uint32_t *mtu)
1303 {
1304 	struct socket *so;
1305 	struct mbuf *m_mtu;
1306 	struct ip6_mtuinfo mtuctl;
1307 
1308 	so = in6p->in6p_socket;
1309 
1310 	if (mtu == NULL)
1311 		return;
1312 
1313 #ifdef DIAGNOSTIC
1314 	if (so == NULL)		/* I believe this is impossible */
1315 		panic("ip6_notify_pmtu: socket is NULL");
1316 #endif
1317 
1318 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
1319 	mtuctl.ip6m_mtu = *mtu;
1320 	mtuctl.ip6m_addr = *dst;
1321 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
1322 		return;
1323 
1324 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
1325 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1326 		return;
1327 
1328 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
1329 	    == 0) {
1330 		m_freem(m_mtu);
1331 		/* XXX: should count statistics */
1332 	} else
1333 		sorwakeup(so);
1334 
1335 	return;
1336 }
1337 
1338 /*
1339  * pull single extension header from mbuf chain.  returns single mbuf that
1340  * contains the result, or NULL on error.
1341  */
1342 static struct mbuf *
1343 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1344 {
1345 	struct ip6_ext ip6e;
1346 	size_t elen;
1347 	struct mbuf *n;
1348 
1349 #ifdef DIAGNOSTIC
1350 	switch (nxt) {
1351 	case IPPROTO_DSTOPTS:
1352 	case IPPROTO_ROUTING:
1353 	case IPPROTO_HOPOPTS:
1354 	case IPPROTO_AH: /* is it possible? */
1355 		break;
1356 	default:
1357 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1358 	}
1359 #endif
1360 
1361 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1362 	if (nxt == IPPROTO_AH)
1363 		elen = (ip6e.ip6e_len + 2) << 2;
1364 	else
1365 		elen = (ip6e.ip6e_len + 1) << 3;
1366 
1367 	MGET(n, M_DONTWAIT, MT_DATA);
1368 	if (n && elen >= MLEN) {
1369 		MCLGET(n, M_DONTWAIT);
1370 		if ((n->m_flags & M_EXT) == 0) {
1371 			m_free(n);
1372 			n = NULL;
1373 		}
1374 	}
1375 	if (!n)
1376 		return NULL;
1377 
1378 	n->m_len = 0;
1379 	if (elen >= M_TRAILINGSPACE(n)) {
1380 		m_free(n);
1381 		return NULL;
1382 	}
1383 
1384 	m_copydata(m, off, elen, mtod(n, void *));
1385 	n->m_len = elen;
1386 	return n;
1387 }
1388 
1389 /*
1390  * Get pointer to the previous header followed by the header
1391  * currently processed.
1392  * XXX: This function supposes that
1393  *	M includes all headers,
1394  *	the next header field and the header length field of each header
1395  *	are valid, and
1396  *	the sum of each header length equals to OFF.
1397  * Because of these assumptions, this function must be called very
1398  * carefully. Moreover, it will not be used in the near future when
1399  * we develop `neater' mechanism to process extension headers.
1400  */
1401 u_int8_t *
1402 ip6_get_prevhdr(struct mbuf *m, int off)
1403 {
1404 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1405 
1406 	if (off == sizeof(struct ip6_hdr))
1407 		return (&ip6->ip6_nxt);
1408 	else {
1409 		int len, nxt;
1410 		struct ip6_ext *ip6e = NULL;
1411 
1412 		nxt = ip6->ip6_nxt;
1413 		len = sizeof(struct ip6_hdr);
1414 		while (len < off) {
1415 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
1416 
1417 			switch (nxt) {
1418 			case IPPROTO_FRAGMENT:
1419 				len += sizeof(struct ip6_frag);
1420 				break;
1421 			case IPPROTO_AH:
1422 				len += (ip6e->ip6e_len + 2) << 2;
1423 				break;
1424 			default:
1425 				len += (ip6e->ip6e_len + 1) << 3;
1426 				break;
1427 			}
1428 			nxt = ip6e->ip6e_nxt;
1429 		}
1430 		if (ip6e)
1431 			return (&ip6e->ip6e_nxt);
1432 		else
1433 			return NULL;
1434 	}
1435 }
1436 
1437 /*
1438  * get next header offset.  m will be retained.
1439  */
1440 int
1441 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1442 {
1443 	struct ip6_hdr ip6;
1444 	struct ip6_ext ip6e;
1445 	struct ip6_frag fh;
1446 
1447 	/* just in case */
1448 	if (m == NULL)
1449 		panic("ip6_nexthdr: m == NULL");
1450 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1451 		return -1;
1452 
1453 	switch (proto) {
1454 	case IPPROTO_IPV6:
1455 		/* do not chase beyond intermediate IPv6 headers */
1456 		if (off != 0)
1457 			return -1;
1458 		if (m->m_pkthdr.len < off + sizeof(ip6))
1459 			return -1;
1460 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
1461 		if (nxtp)
1462 			*nxtp = ip6.ip6_nxt;
1463 		off += sizeof(ip6);
1464 		return off;
1465 
1466 	case IPPROTO_FRAGMENT:
1467 		/*
1468 		 * terminate parsing if it is not the first fragment,
1469 		 * it does not make sense to parse through it.
1470 		 */
1471 		if (m->m_pkthdr.len < off + sizeof(fh))
1472 			return -1;
1473 		m_copydata(m, off, sizeof(fh), (void *)&fh);
1474 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1475 			return -1;
1476 		if (nxtp)
1477 			*nxtp = fh.ip6f_nxt;
1478 		off += sizeof(struct ip6_frag);
1479 		return off;
1480 
1481 	case IPPROTO_AH:
1482 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1483 			return -1;
1484 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1485 		if (nxtp)
1486 			*nxtp = ip6e.ip6e_nxt;
1487 		off += (ip6e.ip6e_len + 2) << 2;
1488 		if (m->m_pkthdr.len < off)
1489 			return -1;
1490 		return off;
1491 
1492 	case IPPROTO_HOPOPTS:
1493 	case IPPROTO_ROUTING:
1494 	case IPPROTO_DSTOPTS:
1495 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1496 			return -1;
1497 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1498 		if (nxtp)
1499 			*nxtp = ip6e.ip6e_nxt;
1500 		off += (ip6e.ip6e_len + 1) << 3;
1501 		if (m->m_pkthdr.len < off)
1502 			return -1;
1503 		return off;
1504 
1505 	case IPPROTO_NONE:
1506 	case IPPROTO_ESP:
1507 	case IPPROTO_IPCOMP:
1508 		/* give up */
1509 		return -1;
1510 
1511 	default:
1512 		return -1;
1513 	}
1514 }
1515 
1516 /*
1517  * get offset for the last header in the chain.  m will be kept untainted.
1518  */
1519 int
1520 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1521 {
1522 	int newoff;
1523 	int nxt;
1524 
1525 	if (!nxtp) {
1526 		nxt = -1;
1527 		nxtp = &nxt;
1528 	}
1529 	for (;;) {
1530 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1531 		if (newoff < 0)
1532 			return off;
1533 		else if (newoff < off)
1534 			return -1;	/* invalid */
1535 		else if (newoff == off)
1536 			return newoff;
1537 
1538 		off = newoff;
1539 		proto = *nxtp;
1540 	}
1541 }
1542 
1543 struct m_tag *
1544 ip6_addaux(struct mbuf *m)
1545 {
1546 	struct m_tag *mtag;
1547 
1548 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1549 	if (!mtag) {
1550 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1551 		    M_NOWAIT);
1552 		if (mtag) {
1553 			m_tag_prepend(m, mtag);
1554 			bzero(mtag + 1, sizeof(struct ip6aux));
1555 		}
1556 	}
1557 	return mtag;
1558 }
1559 
1560 struct m_tag *
1561 ip6_findaux(struct mbuf *m)
1562 {
1563 	struct m_tag *mtag;
1564 
1565 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1566 	return mtag;
1567 }
1568 
1569 void
1570 ip6_delaux(struct mbuf *m)
1571 {
1572 	struct m_tag *mtag;
1573 
1574 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1575 	if (mtag)
1576 		m_tag_delete(m, mtag);
1577 }
1578 
1579 #ifdef GATEWAY
1580 /*
1581  * sysctl helper routine for net.inet.ip6.maxflows. Since
1582  * we could reduce this value, call ip6flow_reap();
1583  */
1584 static int
1585 sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS)
1586 {
1587         int s;
1588 
1589         s = sysctl_lookup(SYSCTLFN_CALL(rnode));
1590         if (s || newp == NULL)
1591                 return (s);
1592 
1593         s = splsoftnet();
1594         ip6flow_reap(0);
1595         splx(s);
1596 
1597         return (0);
1598 }
1599 
1600 static int
1601 sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS)
1602 {
1603 	int error, tmp;
1604 	struct sysctlnode node;
1605 
1606 	node = *rnode;
1607 	tmp = ip6_hashsize;
1608 	node.sysctl_data = &tmp;
1609 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1610 	if (error || newp == NULL)
1611 		return (error);
1612 
1613 	if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
1614 		/*
1615 		 * Can only fail due to malloc()
1616 		 */
1617 		if (ip6flow_invalidate_all(tmp))
1618 			return ENOMEM;
1619 	} else {
1620 		/*
1621 		 * EINVAL if not a power of 2
1622 	         */
1623 		return EINVAL;
1624 	}
1625 
1626 	return (0);
1627 }
1628 #endif /* GATEWAY */
1629 
1630 /*
1631  * System control for IP6
1632  */
1633 
1634 u_char	inet6ctlerrmap[PRC_NCMDS] = {
1635 	0,		0,		0,		0,
1636 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1637 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1638 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1639 	0,		0,		0,		0,
1640 	ENOPROTOOPT
1641 };
1642 
1643 SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup")
1644 {
1645 #ifdef RFC2292
1646 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1647 #else
1648 #define IS2292(x, y)	(y)
1649 #endif
1650 
1651 	sysctl_createv(clog, 0, NULL, NULL,
1652 		       CTLFLAG_PERMANENT,
1653 		       CTLTYPE_NODE, "net", NULL,
1654 		       NULL, 0, NULL, 0,
1655 		       CTL_NET, CTL_EOL);
1656 	sysctl_createv(clog, 0, NULL, NULL,
1657 		       CTLFLAG_PERMANENT,
1658 		       CTLTYPE_NODE, "inet6",
1659 		       SYSCTL_DESCR("PF_INET6 related settings"),
1660 		       NULL, 0, NULL, 0,
1661 		       CTL_NET, PF_INET6, CTL_EOL);
1662 	sysctl_createv(clog, 0, NULL, NULL,
1663 		       CTLFLAG_PERMANENT,
1664 		       CTLTYPE_NODE, "ip6",
1665 		       SYSCTL_DESCR("IPv6 related settings"),
1666 		       NULL, 0, NULL, 0,
1667 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1668 
1669 	sysctl_createv(clog, 0, NULL, NULL,
1670 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1671 		       CTLTYPE_INT, "forwarding",
1672 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1673 		       NULL, 0, &ip6_forwarding, 0,
1674 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1675 		       IPV6CTL_FORWARDING, CTL_EOL);
1676 	sysctl_createv(clog, 0, NULL, NULL,
1677 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1678 		       CTLTYPE_INT, "redirect",
1679 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1680 		       NULL, 0, &ip6_sendredirects, 0,
1681 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1682 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1683 	sysctl_createv(clog, 0, NULL, NULL,
1684 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1685 		       CTLTYPE_INT, "hlim",
1686 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1687 		       NULL, 0, &ip6_defhlim, 0,
1688 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1689 		       IPV6CTL_DEFHLIM, CTL_EOL);
1690 #ifdef notyet
1691 	sysctl_createv(clog, 0, NULL, NULL,
1692 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1693 		       CTLTYPE_INT, "mtu", NULL,
1694 		       NULL, 0, &, 0,
1695 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1696 		       IPV6CTL_DEFMTU, CTL_EOL);
1697 #endif
1698 #ifdef __no_idea__
1699 	sysctl_createv(clog, 0, NULL, NULL,
1700 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1701 		       CTLTYPE_INT, "forwsrcrt", NULL,
1702 		       NULL, 0, &?, 0,
1703 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1704 		       IPV6CTL_FORWSRCRT, CTL_EOL);
1705 	sysctl_createv(clog, 0, NULL, NULL,
1706 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1707 		       CTLTYPE_STRUCT, "mrtstats", NULL,
1708 		       NULL, 0, &?, sizeof(?),
1709 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1710 		       IPV6CTL_MRTSTATS, CTL_EOL);
1711 	sysctl_createv(clog, 0, NULL, NULL,
1712 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1713 		       CTLTYPE_?, "mrtproto", NULL,
1714 		       NULL, 0, &?, sizeof(?),
1715 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1716 		       IPV6CTL_MRTPROTO, CTL_EOL);
1717 #endif
1718 	sysctl_createv(clog, 0, NULL, NULL,
1719 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1720 		       CTLTYPE_INT, "maxfragpackets",
1721 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1722 				    "for reassembly"),
1723 		       NULL, 0, &ip6_maxfragpackets, 0,
1724 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1725 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1726 #ifdef __no_idea__
1727 	sysctl_createv(clog, 0, NULL, NULL,
1728 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1729 		       CTLTYPE_INT, "sourcecheck", NULL,
1730 		       NULL, 0, &?, 0,
1731 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1732 		       IPV6CTL_SOURCECHECK, CTL_EOL);
1733 	sysctl_createv(clog, 0, NULL, NULL,
1734 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1735 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
1736 		       NULL, 0, &?, 0,
1737 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1738 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
1739 #endif
1740 	sysctl_createv(clog, 0, NULL, NULL,
1741 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1742 		       CTLTYPE_INT, "accept_rtadv",
1743 		       SYSCTL_DESCR("Accept router advertisements"),
1744 		       NULL, 0, &ip6_accept_rtadv, 0,
1745 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1746 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1747 	sysctl_createv(clog, 0, NULL, NULL,
1748 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1749 		       CTLTYPE_INT, "keepfaith",
1750 		       SYSCTL_DESCR("Activate faith interface"),
1751 		       NULL, 0, &ip6_keepfaith, 0,
1752 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1753 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1754 	sysctl_createv(clog, 0, NULL, NULL,
1755 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1756 		       CTLTYPE_INT, "log_interval",
1757 		       SYSCTL_DESCR("Minumum interval between logging "
1758 				    "unroutable packets"),
1759 		       NULL, 0, &ip6_log_interval, 0,
1760 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1761 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1762 	sysctl_createv(clog, 0, NULL, NULL,
1763 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1764 		       CTLTYPE_INT, "hdrnestlimit",
1765 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1766 		       NULL, 0, &ip6_hdrnestlimit, 0,
1767 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1768 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1769 	sysctl_createv(clog, 0, NULL, NULL,
1770 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1771 		       CTLTYPE_INT, "dad_count",
1772 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1773 				    "probes to send"),
1774 		       NULL, 0, &ip6_dad_count, 0,
1775 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1776 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1777 	sysctl_createv(clog, 0, NULL, NULL,
1778 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1779 		       CTLTYPE_INT, "auto_flowlabel",
1780 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1781 		       NULL, 0, &ip6_auto_flowlabel, 0,
1782 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1783 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1784 	sysctl_createv(clog, 0, NULL, NULL,
1785 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1786 		       CTLTYPE_INT, "defmcasthlim",
1787 		       SYSCTL_DESCR("Default multicast hop limit"),
1788 		       NULL, 0, &ip6_defmcasthlim, 0,
1789 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1790 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1791 #if NGIF > 0
1792 	sysctl_createv(clog, 0, NULL, NULL,
1793 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1794 		       CTLTYPE_INT, "gifhlim",
1795 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
1796 		       NULL, 0, &ip6_gif_hlim, 0,
1797 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1798 		       IPV6CTL_GIF_HLIM, CTL_EOL);
1799 #endif /* NGIF */
1800 	sysctl_createv(clog, 0, NULL, NULL,
1801 		       CTLFLAG_PERMANENT,
1802 		       CTLTYPE_STRING, "kame_version",
1803 		       SYSCTL_DESCR("KAME Version"),
1804 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1805 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1806 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1807 	sysctl_createv(clog, 0, NULL, NULL,
1808 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1809 		       CTLTYPE_INT, "use_deprecated",
1810 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1811 				    "source addresses"),
1812 		       NULL, 0, &ip6_use_deprecated, 0,
1813 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1814 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1815 	sysctl_createv(clog, 0, NULL, NULL,
1816 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1817 		       CTLTYPE_INT, "rr_prune", NULL,
1818 		       NULL, 0, &ip6_rr_prune, 0,
1819 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1820 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1821 	sysctl_createv(clog, 0, NULL, NULL,
1822 		       CTLFLAG_PERMANENT
1823 #ifndef INET6_BINDV6ONLY
1824 		       |CTLFLAG_READWRITE,
1825 #endif
1826 		       CTLTYPE_INT, "v6only",
1827 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1828 				    "to PF_INET sockets"),
1829 		       NULL, 0, &ip6_v6only, 0,
1830 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1831 		       IPV6CTL_V6ONLY, CTL_EOL);
1832 	sysctl_createv(clog, 0, NULL, NULL,
1833 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1834 		       CTLTYPE_INT, "anonportmin",
1835 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1836 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1837 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1838 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1839 	sysctl_createv(clog, 0, NULL, NULL,
1840 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1841 		       CTLTYPE_INT, "anonportmax",
1842 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1843 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1844 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1845 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1846 #ifndef IPNOPRIVPORTS
1847 	sysctl_createv(clog, 0, NULL, NULL,
1848 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1849 		       CTLTYPE_INT, "lowportmin",
1850 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1851 				    "to assign"),
1852 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1853 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1854 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1855 	sysctl_createv(clog, 0, NULL, NULL,
1856 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1857 		       CTLTYPE_INT, "lowportmax",
1858 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1859 				    "to assign"),
1860 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1861 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1862 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1863 #endif /* IPNOPRIVPORTS */
1864 	sysctl_createv(clog, 0, NULL, NULL,
1865 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1866 		       CTLTYPE_INT, "use_tempaddr",
1867 		       SYSCTL_DESCR("Use temporary address"),
1868 		       NULL, 0, &ip6_use_tempaddr, 0,
1869 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1870 		       CTL_CREATE, CTL_EOL);
1871 	sysctl_createv(clog, 0, NULL, NULL,
1872 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1873 		       CTLTYPE_INT, "temppltime",
1874 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
1875 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
1876 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1877 		       CTL_CREATE, CTL_EOL);
1878 	sysctl_createv(clog, 0, NULL, NULL,
1879 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1880 		       CTLTYPE_INT, "tempvltime",
1881 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
1882 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
1883 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1884 		       CTL_CREATE, CTL_EOL);
1885 	sysctl_createv(clog, 0, NULL, NULL,
1886 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1887 		       CTLTYPE_INT, "maxfrags",
1888 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1889 		       NULL, 0, &ip6_maxfrags, 0,
1890 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1891 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1892 	sysctl_createv(clog, 0, NULL, NULL,
1893 		       CTLFLAG_PERMANENT,
1894 		       CTLTYPE_STRUCT, "stats",
1895 		       SYSCTL_DESCR("IPv6 statistics"),
1896 		       NULL, 0, &ip6stat, sizeof(ip6stat),
1897 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1898 		       IPV6CTL_STATS, CTL_EOL);
1899 	sysctl_createv(clog, 0, NULL, NULL,
1900 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1901 		       CTLTYPE_INT, "use_defaultzone",
1902 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1903 		       NULL, 0, &ip6_use_defzone, 0,
1904 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1905 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1906 	sysctl_createv(clog, 0, NULL, NULL,
1907 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1908 		       CTLTYPE_INT, "mcast_pmtu",
1909 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
1910 		       NULL, 0, &ip6_mcast_pmtu, 0,
1911 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1912 		       CTL_CREATE, CTL_EOL);
1913 #ifdef GATEWAY
1914 	sysctl_createv(clog, 0, NULL, NULL,
1915 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1916 			CTLTYPE_INT, "maxflows",
1917 			SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"),
1918 			sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0,
1919 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1920 			CTL_CREATE, CTL_EOL);
1921 	sysctl_createv(clog, 0, NULL, NULL,
1922 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1923 			CTLTYPE_INT, "hashsize",
1924 			SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"),
1925 			sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0,
1926 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1927 			CTL_CREATE, CTL_EOL);
1928 #endif
1929 }
1930