xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 7e30e94394d0994ab9534f68a8f91665045c91ce)
1 /*	$NetBSD: ip6_input.c,v 1.177 2017/03/14 04:25:10 ozaki-r 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 1.177 2017/03/14 04:25:10 ozaki-r Exp $");
66 
67 #ifdef _KERNEL_OPT
68 #include "opt_gateway.h"
69 #include "opt_inet.h"
70 #include "opt_inet6.h"
71 #include "opt_ipsec.h"
72 #include "opt_compat_netbsd.h"
73 #include "opt_net_mpsafe.h"
74 #endif
75 
76 #include <sys/param.h>
77 #include <sys/systm.h>
78 #include <sys/mbuf.h>
79 #include <sys/domain.h>
80 #include <sys/protosw.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/errno.h>
84 #include <sys/time.h>
85 #include <sys/kernel.h>
86 #include <sys/syslog.h>
87 #include <sys/proc.h>
88 #include <sys/sysctl.h>
89 #include <sys/cprng.h>
90 #include <sys/percpu.h>
91 
92 #include <net/if.h>
93 #include <net/if_types.h>
94 #include <net/if_dl.h>
95 #include <net/route.h>
96 #include <net/pktqueue.h>
97 #include <net/pfil.h>
98 
99 #include <netinet/in.h>
100 #include <netinet/in_systm.h>
101 #ifdef INET
102 #include <netinet/ip.h>
103 #include <netinet/ip_var.h>
104 #include <netinet/ip_icmp.h>
105 #endif /* INET */
106 #include <netinet/ip6.h>
107 #include <netinet/portalgo.h>
108 #include <netinet6/in6_var.h>
109 #include <netinet6/ip6_var.h>
110 #include <netinet6/ip6_private.h>
111 #include <netinet6/in6_pcb.h>
112 #include <netinet/icmp6.h>
113 #include <netinet6/scope6_var.h>
114 #include <netinet6/in6_ifattach.h>
115 #include <netinet6/nd6.h>
116 
117 #ifdef IPSEC
118 #include <netipsec/ipsec.h>
119 #include <netipsec/ipsec6.h>
120 #include <netipsec/key.h>
121 #endif /* IPSEC */
122 
123 #ifdef COMPAT_50
124 #include <compat/sys/time.h>
125 #include <compat/sys/socket.h>
126 #endif
127 
128 #include <netinet6/ip6protosw.h>
129 
130 #include "faith.h"
131 
132 #include <net/net_osdep.h>
133 
134 extern struct domain inet6domain;
135 
136 u_char ip6_protox[IPPROTO_MAX];
137 pktqueue_t *ip6_pktq __read_mostly;
138 
139 int ip6_forward_srcrt;			/* XXX */
140 int ip6_sourcecheck;			/* XXX */
141 int ip6_sourcecheck_interval;		/* XXX */
142 
143 pfil_head_t *inet6_pfil_hook;
144 
145 percpu_t *ip6stat_percpu;
146 
147 percpu_t *ip6_forward_rt_percpu __cacheline_aligned;
148 
149 static void ip6_init2(void);
150 static void ip6intr(void *);
151 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
152 
153 static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *,
154 	u_int32_t *);
155 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
156 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
157 
158 #ifdef NET_MPSAFE
159 #define	SOFTNET_LOCK()		mutex_enter(softnet_lock)
160 #define	SOFTNET_UNLOCK()	mutex_exit(softnet_lock)
161 #else
162 #define	SOFTNET_LOCK()		KASSERT(mutex_owned(softnet_lock))
163 #define	SOFTNET_UNLOCK()	KASSERT(mutex_owned(softnet_lock))
164 #endif
165 
166 /*
167  * IP6 initialization: fill in IP6 protocol switch table.
168  * All protocols not implemented in kernel go to raw IP6 protocol handler.
169  */
170 void
171 ip6_init(void)
172 {
173 	const struct ip6protosw *pr;
174 	int i;
175 
176 	in6_init();
177 
178 	sysctl_net_inet6_ip6_setup(NULL);
179 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
180 	if (pr == 0)
181 		panic("ip6_init");
182 	for (i = 0; i < IPPROTO_MAX; i++)
183 		ip6_protox[i] = pr - inet6sw;
184 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
185 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
186 		if (pr->pr_domain->dom_family == PF_INET6 &&
187 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
188 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
189 
190 	ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL);
191 	KASSERT(ip6_pktq != NULL);
192 
193 	scope6_init();
194 	addrsel_policy_init();
195 	nd6_init();
196 	frag6_init();
197 	ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR;
198 
199 	ip6_init2();
200 #ifdef GATEWAY
201 	ip6flow_init(ip6_hashsize);
202 #endif
203 	/* Register our Packet Filter hook. */
204 	inet6_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET6);
205 	KASSERT(inet6_pfil_hook != NULL);
206 
207 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
208 
209 	ip6_forward_rt_percpu = percpu_alloc(sizeof(struct route));
210 	if (ip6_forward_rt_percpu == NULL)
211 		panic("failed to alllocate ip6_forward_rt_percpu");
212 }
213 
214 static void
215 ip6_init2(void)
216 {
217 
218 	/* timer for regeneranation of temporary addresses randomize ID */
219 	callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE);
220 	callout_reset(&in6_tmpaddrtimer_ch,
221 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
222 		       ip6_temp_regen_advance) * hz,
223 		      in6_tmpaddrtimer, NULL);
224 }
225 
226 /*
227  * IP6 input interrupt handling. Just pass the packet to ip6_input.
228  */
229 static void
230 ip6intr(void *arg __unused)
231 {
232 	struct mbuf *m;
233 
234 #ifndef NET_MPSAFE
235 	mutex_enter(softnet_lock);
236 #endif
237 	while ((m = pktq_dequeue(ip6_pktq)) != NULL) {
238 		struct psref psref;
239 		struct ifnet *rcvif = m_get_rcvif_psref(m, &psref);
240 
241 		if (rcvif == NULL) {
242 			m_freem(m);
243 			continue;
244 		}
245 		/*
246 		 * Drop the packet if IPv6 is disabled on the interface.
247 		 */
248 		if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED)) {
249 			m_put_rcvif_psref(rcvif, &psref);
250 			m_freem(m);
251 			continue;
252 		}
253 		ip6_input(m, rcvif);
254 		m_put_rcvif_psref(rcvif, &psref);
255 	}
256 #ifndef NET_MPSAFE
257 	mutex_exit(softnet_lock);
258 #endif
259 }
260 
261 void
262 ip6_input(struct mbuf *m, struct ifnet *rcvif)
263 {
264 	struct ip6_hdr *ip6;
265 	int hit, off = sizeof(struct ip6_hdr), nest;
266 	u_int32_t plen;
267 	u_int32_t rtalert = ~0;
268 	int nxt, ours = 0, rh_present = 0;
269 	struct ifnet *deliverifp = NULL;
270 	int srcrt = 0;
271 	struct rtentry *rt = NULL;
272 	union {
273 		struct sockaddr		dst;
274 		struct sockaddr_in6	dst6;
275 	} u;
276 	struct route *ro;
277 
278 	/*
279 	 * make sure we don't have onion peering information into m_tag.
280 	 */
281 	ip6_delaux(m);
282 
283 	/*
284 	 * mbuf statistics
285 	 */
286 	if (m->m_flags & M_EXT) {
287 		if (m->m_next)
288 			IP6_STATINC(IP6_STAT_MEXT2M);
289 		else
290 			IP6_STATINC(IP6_STAT_MEXT1);
291 	} else {
292 #define M2MMAX	32
293 		if (m->m_next) {
294 			if (m->m_flags & M_LOOP)
295 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
296 			else if (rcvif->if_index < M2MMAX)
297 				IP6_STATINC(IP6_STAT_M2M + rcvif->if_index);
298 			else
299 				IP6_STATINC(IP6_STAT_M2M);
300 		} else
301 			IP6_STATINC(IP6_STAT_M1);
302 #undef M2MMAX
303 	}
304 
305 	in6_ifstat_inc(rcvif, ifs6_in_receive);
306 	IP6_STATINC(IP6_STAT_TOTAL);
307 
308 	/*
309 	 * If the IPv6 header is not aligned, slurp it up into a new
310 	 * mbuf with space for link headers, in the event we forward
311 	 * it.  Otherwise, if it is aligned, make sure the entire base
312 	 * IPv6 header is in the first mbuf of the chain.
313 	 */
314 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
315 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
316 				  (max_linkhdr + 3) & ~3)) == NULL) {
317 			/* XXXJRT new stat, please */
318 			IP6_STATINC(IP6_STAT_TOOSMALL);
319 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
320 			return;
321 		}
322 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
323 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
324 			IP6_STATINC(IP6_STAT_TOOSMALL);
325 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
326 			return;
327 		}
328 	}
329 
330 	ip6 = mtod(m, struct ip6_hdr *);
331 
332 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
333 		IP6_STATINC(IP6_STAT_BADVERS);
334 		in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
335 		goto bad;
336 	}
337 
338 	/*
339 	 * Assume that we can create a fast-forward IP flow entry
340 	 * based on this packet.
341 	 */
342 	m->m_flags |= M_CANFASTFWD;
343 
344 	/*
345 	 * Run through list of hooks for input packets.  If there are any
346 	 * filters which require that additional packets in the flow are
347 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
348 	 * Note that filters must _never_ set this flag, as another filter
349 	 * in the list may have previously cleared it.
350 	 */
351 	/*
352 	 * let ipfilter look at packet on the wire,
353 	 * not the decapsulated packet.
354 	 */
355 #if defined(IPSEC)
356 	if (!ipsec_used || !ipsec_indone(m))
357 #else
358 	if (1)
359 #endif
360 	{
361 		struct in6_addr odst;
362 
363 		odst = ip6->ip6_dst;
364 		if (pfil_run_hooks(inet6_pfil_hook, &m, rcvif, PFIL_IN) != 0)
365 			return;
366 		if (m == NULL)
367 			return;
368 		ip6 = mtod(m, struct ip6_hdr *);
369 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
370 	}
371 
372 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
373 
374 #ifdef ALTQ
375 	if (altq_input != NULL) {
376 		SOFTNET_LOCK();
377 		if ((*altq_input)(m, AF_INET6) == 0) {
378 			SOFTNET_UNLOCK();
379 			/* packet is dropped by traffic conditioner */
380 			return;
381 		}
382 		SOFTNET_UNLOCK();
383 	}
384 #endif
385 
386 	/*
387 	 * Check against address spoofing/corruption.
388 	 */
389 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
390 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
391 		/*
392 		 * XXX: "badscope" is not very suitable for a multicast source.
393 		 */
394 		IP6_STATINC(IP6_STAT_BADSCOPE);
395 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
396 		goto bad;
397 	}
398 	/*
399 	 * The following check is not documented in specs.  A malicious
400 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
401 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
402 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
403 	 *
404 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
405 	 * support IPv4-less kernel compilation, we cannot support SIIT
406 	 * environment at all.  So, it makes more sense for us to reject any
407 	 * malicious packets for non-SIIT environment, than try to do a
408 	 * partial support for SIIT environment.
409 	 */
410 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
411 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
412 		IP6_STATINC(IP6_STAT_BADSCOPE);
413 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
414 		goto bad;
415 	}
416 #if 0
417 	/*
418 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
419 	 *
420 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
421 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
422 	 * is revised to forbid relaying case.
423 	 */
424 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
425 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
426 		IP6_STATINC(IP6_STAT_BADSCOPE);
427 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
428 		goto bad;
429 	}
430 #endif
431 
432 	/*
433 	 * Disambiguate address scope zones (if there is ambiguity).
434 	 * We first make sure that the original source or destination address
435 	 * is not in our internal form for scoped addresses.  Such addresses
436 	 * are not necessarily invalid spec-wise, but we cannot accept them due
437 	 * to the usage conflict.
438 	 * in6_setscope() then also checks and rejects the cases where src or
439 	 * dst are the loopback address and the receiving interface
440 	 * is not loopback.
441 	 */
442 	if (__predict_false(
443 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
444 		goto bad;
445 	ip6 = mtod(m, struct ip6_hdr *);
446 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
447 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
448 		goto bad;
449 	}
450 	if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
451 	    in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
452 		IP6_STATINC(IP6_STAT_BADSCOPE);
453 		goto bad;
454 	}
455 
456 	ro = percpu_getref(ip6_forward_rt_percpu);
457 	/*
458 	 * Multicast check
459 	 */
460 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
461 		bool ingroup;
462 
463 		in6_ifstat_inc(rcvif, ifs6_in_mcast);
464 		/*
465 		 * See if we belong to the destination multicast group on the
466 		 * arrival interface.
467 		 */
468 		ingroup = in6_multi_group(&ip6->ip6_dst, rcvif);
469 		if (ingroup)
470 			ours = 1;
471 		else if (!ip6_mrouter) {
472 			uint64_t *ip6s = IP6_STAT_GETREF();
473 			ip6s[IP6_STAT_NOTMEMBER]++;
474 			ip6s[IP6_STAT_CANTFORWARD]++;
475 			IP6_STAT_PUTREF();
476 			in6_ifstat_inc(rcvif, ifs6_in_discard);
477 			goto bad_unref;
478 		}
479 		deliverifp = rcvif;
480 		goto hbhcheck;
481 	}
482 
483 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
484 
485 	/*
486 	 *  Unicast check
487 	 */
488 	rt = rtcache_lookup2(ro, &u.dst, 1, &hit);
489 	if (hit)
490 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
491 	else
492 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
493 
494 #define rt6_getkey(__rt) satocsin6(rt_getkey(__rt))
495 
496 	/*
497 	 * Accept the packet if the forwarding interface to the destination
498 	 * according to the routing table is the loopback interface,
499 	 * unless the associated route has a gateway.
500 	 * Note that this approach causes to accept a packet if there is a
501 	 * route to the loopback interface for the destination of the packet.
502 	 * But we think it's even useful in some situations, e.g. when using
503 	 * a special daemon which wants to intercept the packet.
504 	 */
505 	if (rt != NULL &&
506 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
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 | IN6_IFF_DETACHED))) {
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 			char ip6bufs[INET6_ADDRSTRLEN];
531 			char ip6bufd[INET6_ADDRSTRLEN];
532 			nd6log(LOG_INFO, "packet to an unready address %s->%s\n",
533 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
534 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
535 
536 			goto bad_unref;
537 		}
538 	}
539 
540 	/*
541 	 * FAITH (Firewall Aided Internet Translator)
542 	 */
543 #if defined(NFAITH) && 0 < NFAITH
544 	if (ip6_keepfaith) {
545 		if (rt != NULL && rt->rt_ifp != NULL &&
546 		    rt->rt_ifp->if_type == IFT_FAITH) {
547 			/* XXX do we need more sanity checks? */
548 			ours = 1;
549 			deliverifp = rt->rt_ifp; /* faith */
550 			goto hbhcheck;
551 		}
552 	}
553 #endif
554 
555 #if 0
556     {
557 	/*
558 	 * Last resort: check in6_ifaddr for incoming interface.
559 	 * The code is here until I update the "goto ours hack" code above
560 	 * working right.
561 	 */
562 	struct ifaddr *ifa;
563 	IFADDR_READER_FOREACH(ifa, rcvif) {
564 		if (ifa->ifa_addr->sa_family != AF_INET6)
565 			continue;
566 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
567 			ours = 1;
568 			deliverifp = ifa->ifa_ifp;
569 			goto hbhcheck;
570 		}
571 	}
572     }
573 #endif
574 
575 	/*
576 	 * Now there is no reason to process the packet if it's not our own
577 	 * and we're not a router.
578 	 */
579 	if (!ip6_forwarding) {
580 		IP6_STATINC(IP6_STAT_CANTFORWARD);
581 		in6_ifstat_inc(rcvif, ifs6_in_discard);
582 		goto bad_unref;
583 	}
584 
585   hbhcheck:
586 	/*
587 	 * record address information into m_tag, if we don't have one yet.
588 	 * note that we are unable to record it, if the address is not listed
589 	 * as our interface address (e.g. multicast addresses, addresses
590 	 * within FAITH prefixes and such).
591 	 */
592 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
593 		struct in6_ifaddr *ia6;
594 		int s = pserialize_read_enter();
595 
596 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
597 		/* Depends on ip6_setdstifaddr never sleep */
598 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
599 			/*
600 			 * XXX maybe we should drop the packet here,
601 			 * as we could not provide enough information
602 			 * to the upper layers.
603 			 */
604 		}
605 		pserialize_read_exit(s);
606 	}
607 
608 	/*
609 	 * Process Hop-by-Hop options header if it's contained.
610 	 * m may be modified in ip6_hopopts_input().
611 	 * If a JumboPayload option is included, plen will also be modified.
612 	 */
613 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
614 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
615 		struct ip6_hbh *hbh;
616 
617 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
618 #if 0	/*touches NULL pointer*/
619 			in6_ifstat_inc(rcvif, ifs6_in_discard);
620 #endif
621 			rtcache_unref(rt, ro);
622 			percpu_putref(ip6_forward_rt_percpu);
623 			return;	/* m have already been freed */
624 		}
625 
626 		/* adjust pointer */
627 		ip6 = mtod(m, struct ip6_hdr *);
628 
629 		/*
630 		 * if the payload length field is 0 and the next header field
631 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
632 		 * option MUST be included.
633 		 */
634 		if (ip6->ip6_plen == 0 && plen == 0) {
635 			/*
636 			 * Note that if a valid jumbo payload option is
637 			 * contained, ip6_hopopts_input() must set a valid
638 			 * (non-zero) payload length to the variable plen.
639 			 */
640 			IP6_STATINC(IP6_STAT_BADOPTIONS);
641 			in6_ifstat_inc(rcvif, ifs6_in_discard);
642 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
643 			icmp6_error(m, ICMP6_PARAM_PROB,
644 				    ICMP6_PARAMPROB_HEADER,
645 				    (char *)&ip6->ip6_plen - (char *)ip6);
646 			rtcache_unref(rt, ro);
647 			percpu_putref(ip6_forward_rt_percpu);
648 			return;
649 		}
650 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
651 			sizeof(struct ip6_hbh));
652 		if (hbh == NULL) {
653 			IP6_STATINC(IP6_STAT_TOOSHORT);
654 			rtcache_unref(rt, ro);
655 			percpu_putref(ip6_forward_rt_percpu);
656 			return;
657 		}
658 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
659 		nxt = hbh->ip6h_nxt;
660 
661 		/*
662 		 * accept the packet if a router alert option is included
663 		 * and we act as an IPv6 router.
664 		 */
665 		if (rtalert != ~0 && ip6_forwarding)
666 			ours = 1;
667 	} else
668 		nxt = ip6->ip6_nxt;
669 
670 	/*
671 	 * Check that the amount of data in the buffers
672 	 * is as at least much as the IPv6 header would have us expect.
673 	 * Trim mbufs if longer than we expect.
674 	 * Drop packet if shorter than we expect.
675 	 */
676 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
677 		IP6_STATINC(IP6_STAT_TOOSHORT);
678 		in6_ifstat_inc(rcvif, ifs6_in_truncated);
679 		goto bad_unref;
680 	}
681 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
682 		if (m->m_len == m->m_pkthdr.len) {
683 			m->m_len = sizeof(struct ip6_hdr) + plen;
684 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
685 		} else
686 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
687 	}
688 
689 	/*
690 	 * Forward if desirable.
691 	 */
692 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
693 		/*
694 		 * If we are acting as a multicast router, all
695 		 * incoming multicast packets are passed to the
696 		 * kernel-level multicast forwarding function.
697 		 * The packet is returned (relatively) intact; if
698 		 * ip6_mforward() returns a non-zero value, the packet
699 		 * must be discarded, else it may be accepted below.
700 		 */
701 		if (ip6_mrouter != NULL) {
702 			int error;
703 
704 			SOFTNET_LOCK();
705 			error = ip6_mforward(ip6, rcvif, m);
706 			SOFTNET_UNLOCK();
707 
708 			if (error != 0) {
709 				rtcache_unref(rt, ro);
710 				percpu_putref(ip6_forward_rt_percpu);
711 				IP6_STATINC(IP6_STAT_CANTFORWARD);
712 				goto bad;
713 			}
714 		}
715 		if (!ours)
716 			goto bad_unref;
717 	} else if (!ours) {
718 		rtcache_unref(rt, ro);
719 		percpu_putref(ip6_forward_rt_percpu);
720 		ip6_forward(m, srcrt);
721 		return;
722 	}
723 
724 	ip6 = mtod(m, struct ip6_hdr *);
725 
726 	/*
727 	 * Malicious party may be able to use IPv4 mapped addr to confuse
728 	 * tcp/udp stack and bypass security checks (act as if it was from
729 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
730 	 *
731 	 * For SIIT end node behavior, you may want to disable the check.
732 	 * However, you will  become vulnerable to attacks using IPv4 mapped
733 	 * source.
734 	 */
735 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
736 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
737 		IP6_STATINC(IP6_STAT_BADSCOPE);
738 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
739 		goto bad_unref;
740 	}
741 
742 	/*
743 	 * Tell launch routine the next header
744 	 */
745 #ifdef IFA_STATS
746 	if (deliverifp != NULL) {
747 		struct in6_ifaddr *ia6;
748 		int s = pserialize_read_enter();
749 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
750 		if (ia6)
751 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
752 		pserialize_read_exit(s);
753 	}
754 #endif
755 	IP6_STATINC(IP6_STAT_DELIVERED);
756 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
757 	nest = 0;
758 
759 	if (rt != NULL) {
760 		rtcache_unref(rt, ro);
761 		rt = NULL;
762 	}
763 	percpu_putref(ip6_forward_rt_percpu);
764 
765 	rh_present = 0;
766 	while (nxt != IPPROTO_DONE) {
767 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
768 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
769 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
770 			goto bad;
771 		}
772 
773 		/*
774 		 * protection against faulty packet - there should be
775 		 * more sanity checks in header chain processing.
776 		 */
777 		if (m->m_pkthdr.len < off) {
778 			IP6_STATINC(IP6_STAT_TOOSHORT);
779 			in6_ifstat_inc(rcvif, ifs6_in_truncated);
780 			goto bad;
781 		}
782 
783 		if (nxt == IPPROTO_ROUTING) {
784 			if (rh_present++) {
785 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
786 				IP6_STATINC(IP6_STAT_BADOPTIONS);
787 				goto bad;
788 			}
789 		}
790 
791 #ifdef IPSEC
792 		if (ipsec_used) {
793 			/*
794 			 * enforce IPsec policy checking if we are seeing last
795 			 * header. note that we do not visit this with
796 			 * protocols with pcb layer code - like udp/tcp/raw ip.
797 			 */
798 			if ((inet6sw[ip_protox[nxt]].pr_flags
799 			    & PR_LASTHDR) != 0) {
800 				int error;
801 
802 				SOFTNET_LOCK();
803 				error = ipsec6_input(m);
804 				SOFTNET_UNLOCK();
805 				if (error)
806 					goto bad;
807 			}
808 		}
809 #endif /* IPSEC */
810 
811 		SOFTNET_LOCK();
812 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
813 		SOFTNET_UNLOCK();
814 	}
815 	return;
816 
817  bad_unref:
818 	rtcache_unref(rt, ro);
819 	percpu_putref(ip6_forward_rt_percpu);
820  bad:
821 	m_freem(m);
822 	return;
823 }
824 
825 /*
826  * set/grab in6_ifaddr correspond to IPv6 destination address.
827  */
828 static struct m_tag *
829 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
830 {
831 	struct m_tag *mtag;
832 	struct ip6aux *ip6a;
833 
834 	mtag = ip6_addaux(m);
835 	if (mtag == NULL)
836 		return NULL;
837 
838 	ip6a = (struct ip6aux *)(mtag + 1);
839 	if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) {
840 		IP6_STATINC(IP6_STAT_BADSCOPE);
841 		return NULL;
842 	}
843 
844 	ip6a->ip6a_src = ia->ia_addr.sin6_addr;
845 	ip6a->ip6a_flags = ia->ia6_flags;
846 	return mtag;
847 }
848 
849 const struct ip6aux *
850 ip6_getdstifaddr(struct mbuf *m)
851 {
852 	struct m_tag *mtag;
853 
854 	mtag = ip6_findaux(m);
855 	if (mtag != NULL)
856 		return (struct ip6aux *)(mtag + 1);
857 	else
858 		return NULL;
859 }
860 
861 /*
862  * Hop-by-Hop options header processing. If a valid jumbo payload option is
863  * included, the real payload length will be stored in plenp.
864  *
865  * rtalertp - XXX: should be stored more smart way
866  */
867 int
868 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
869 	struct mbuf **mp, int *offp)
870 {
871 	struct mbuf *m = *mp;
872 	int off = *offp, hbhlen;
873 	struct ip6_hbh *hbh;
874 
875 	/* validation of the length of the header */
876 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
877 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
878 	if (hbh == NULL) {
879 		IP6_STATINC(IP6_STAT_TOOSHORT);
880 		return -1;
881 	}
882 	hbhlen = (hbh->ip6h_len + 1) << 3;
883 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
884 		hbhlen);
885 	if (hbh == NULL) {
886 		IP6_STATINC(IP6_STAT_TOOSHORT);
887 		return -1;
888 	}
889 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
890 	off += hbhlen;
891 	hbhlen -= sizeof(struct ip6_hbh);
892 
893 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
894 				hbhlen, rtalertp, plenp) < 0)
895 		return (-1);
896 
897 	*offp = off;
898 	*mp = m;
899 	return (0);
900 }
901 
902 /*
903  * Search header for all Hop-by-hop options and process each option.
904  * This function is separate from ip6_hopopts_input() in order to
905  * handle a case where the sending node itself process its hop-by-hop
906  * options header. In such a case, the function is called from ip6_output().
907  *
908  * The function assumes that hbh header is located right after the IPv6 header
909  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
910  * opthead + hbhlen is located in continuous memory region.
911  */
912 static int
913 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
914 	u_int32_t *rtalertp, u_int32_t *plenp)
915 {
916 	struct ip6_hdr *ip6;
917 	int optlen = 0;
918 	u_int8_t *opt = opthead;
919 	u_int16_t rtalert_val;
920 	u_int32_t jumboplen;
921 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
922 
923 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
924 		switch (*opt) {
925 		case IP6OPT_PAD1:
926 			optlen = 1;
927 			break;
928 		case IP6OPT_PADN:
929 			if (hbhlen < IP6OPT_MINLEN) {
930 				IP6_STATINC(IP6_STAT_TOOSMALL);
931 				goto bad;
932 			}
933 			optlen = *(opt + 1) + 2;
934 			break;
935 		case IP6OPT_RTALERT:
936 			/* XXX may need check for alignment */
937 			if (hbhlen < IP6OPT_RTALERT_LEN) {
938 				IP6_STATINC(IP6_STAT_TOOSMALL);
939 				goto bad;
940 			}
941 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
942 				/* XXX stat */
943 				icmp6_error(m, ICMP6_PARAM_PROB,
944 				    ICMP6_PARAMPROB_HEADER,
945 				    erroff + opt + 1 - opthead);
946 				return (-1);
947 			}
948 			optlen = IP6OPT_RTALERT_LEN;
949 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
950 			*rtalertp = ntohs(rtalert_val);
951 			break;
952 		case IP6OPT_JUMBO:
953 			/* XXX may need check for alignment */
954 			if (hbhlen < IP6OPT_JUMBO_LEN) {
955 				IP6_STATINC(IP6_STAT_TOOSMALL);
956 				goto bad;
957 			}
958 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
959 				/* XXX stat */
960 				icmp6_error(m, ICMP6_PARAM_PROB,
961 				    ICMP6_PARAMPROB_HEADER,
962 				    erroff + opt + 1 - opthead);
963 				return (-1);
964 			}
965 			optlen = IP6OPT_JUMBO_LEN;
966 
967 			/*
968 			 * IPv6 packets that have non 0 payload length
969 			 * must not contain a jumbo payload option.
970 			 */
971 			ip6 = mtod(m, struct ip6_hdr *);
972 			if (ip6->ip6_plen) {
973 				IP6_STATINC(IP6_STAT_BADOPTIONS);
974 				icmp6_error(m, ICMP6_PARAM_PROB,
975 				    ICMP6_PARAMPROB_HEADER,
976 				    erroff + opt - opthead);
977 				return (-1);
978 			}
979 
980 			/*
981 			 * We may see jumbolen in unaligned location, so
982 			 * we'd need to perform bcopy().
983 			 */
984 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
985 			jumboplen = (u_int32_t)htonl(jumboplen);
986 
987 #if 1
988 			/*
989 			 * if there are multiple jumbo payload options,
990 			 * *plenp will be non-zero and the packet will be
991 			 * rejected.
992 			 * the behavior may need some debate in ipngwg -
993 			 * multiple options does not make sense, however,
994 			 * there's no explicit mention in specification.
995 			 */
996 			if (*plenp != 0) {
997 				IP6_STATINC(IP6_STAT_BADOPTIONS);
998 				icmp6_error(m, ICMP6_PARAM_PROB,
999 				    ICMP6_PARAMPROB_HEADER,
1000 				    erroff + opt + 2 - opthead);
1001 				return (-1);
1002 			}
1003 #endif
1004 
1005 			/*
1006 			 * jumbo payload length must be larger than 65535.
1007 			 */
1008 			if (jumboplen <= IPV6_MAXPACKET) {
1009 				IP6_STATINC(IP6_STAT_BADOPTIONS);
1010 				icmp6_error(m, ICMP6_PARAM_PROB,
1011 				    ICMP6_PARAMPROB_HEADER,
1012 				    erroff + opt + 2 - opthead);
1013 				return (-1);
1014 			}
1015 			*plenp = jumboplen;
1016 
1017 			break;
1018 		default:		/* unknown option */
1019 			if (hbhlen < IP6OPT_MINLEN) {
1020 				IP6_STATINC(IP6_STAT_TOOSMALL);
1021 				goto bad;
1022 			}
1023 			optlen = ip6_unknown_opt(opt, m,
1024 			    erroff + opt - opthead);
1025 			if (optlen == -1)
1026 				return (-1);
1027 			optlen += 2;
1028 			break;
1029 		}
1030 	}
1031 
1032 	return (0);
1033 
1034   bad:
1035 	m_freem(m);
1036 	return (-1);
1037 }
1038 
1039 /*
1040  * Unknown option processing.
1041  * The third argument `off' is the offset from the IPv6 header to the option,
1042  * which is necessary if the IPv6 header the and option header and IPv6 header
1043  * is not continuous in order to return an ICMPv6 error.
1044  */
1045 int
1046 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1047 {
1048 	struct ip6_hdr *ip6;
1049 
1050 	switch (IP6OPT_TYPE(*optp)) {
1051 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1052 		return ((int)*(optp + 1));
1053 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1054 		m_freem(m);
1055 		return (-1);
1056 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1057 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1058 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1059 		return (-1);
1060 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1061 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1062 		ip6 = mtod(m, struct ip6_hdr *);
1063 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1064 		    (m->m_flags & (M_BCAST|M_MCAST)))
1065 			m_freem(m);
1066 		else
1067 			icmp6_error(m, ICMP6_PARAM_PROB,
1068 				    ICMP6_PARAMPROB_OPTION, off);
1069 		return (-1);
1070 	}
1071 
1072 	m_freem(m);		/* XXX: NOTREACHED */
1073 	return (-1);
1074 }
1075 
1076 /*
1077  * Create the "control" list for this pcb.
1078  *
1079  * The routine will be called from upper layer handlers like tcp6_input().
1080  * Thus the routine assumes that the caller (tcp6_input) have already
1081  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1082  * very first mbuf on the mbuf chain.
1083  * We may want to add some infinite loop prevention or sanity checks for safety.
1084  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1085  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1086  */
1087 void
1088 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
1089 	struct ip6_hdr *ip6, struct mbuf *m)
1090 {
1091 #ifdef RFC2292
1092 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1093 #else
1094 #define IS2292(x, y)	(y)
1095 #endif
1096 
1097 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP
1098 #ifdef SO_OTIMESTAMP
1099 	    || in6p->in6p_socket->so_options & SO_OTIMESTAMP
1100 #endif
1101 	) {
1102 		struct timeval tv;
1103 
1104 		microtime(&tv);
1105 #ifdef SO_OTIMESTAMP
1106 		if (in6p->in6p_socket->so_options & SO_OTIMESTAMP) {
1107 			struct timeval50 tv50;
1108 			timeval_to_timeval50(&tv, &tv50);
1109 			*mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
1110 			    SCM_OTIMESTAMP, SOL_SOCKET);
1111 		} else
1112 #endif
1113 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1114 		    SCM_TIMESTAMP, SOL_SOCKET);
1115 		if (*mp)
1116 			mp = &(*mp)->m_next;
1117 	}
1118 
1119 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
1120 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1121 		return;
1122 
1123 	/* RFC 2292 sec. 5 */
1124 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1125 		struct in6_pktinfo pi6;
1126 
1127 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
1128 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1129 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index;
1130 		*mp = sbcreatecontrol((void *) &pi6,
1131 		    sizeof(struct in6_pktinfo),
1132 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1133 		if (*mp)
1134 			mp = &(*mp)->m_next;
1135 	}
1136 
1137 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1138 		int hlim = ip6->ip6_hlim & 0xff;
1139 
1140 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
1141 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1142 		if (*mp)
1143 			mp = &(*mp)->m_next;
1144 	}
1145 
1146 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1147 		u_int32_t flowinfo;
1148 		int tclass;
1149 
1150 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1151 		flowinfo >>= 20;
1152 
1153 		tclass = flowinfo & 0xff;
1154 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
1155 		    IPV6_TCLASS, IPPROTO_IPV6);
1156 
1157 		if (*mp)
1158 			mp = &(*mp)->m_next;
1159 	}
1160 
1161 	/*
1162 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1163 	 * privilege for the option (see ip6_ctloutput), but it might be too
1164 	 * strict, since there might be some hop-by-hop options which can be
1165 	 * returned to normal user.
1166 	 * See also RFC3542 section 8 (or RFC2292 section 6).
1167 	 */
1168 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1169 		/*
1170 		 * Check if a hop-by-hop options header is contatined in the
1171 		 * received packet, and if so, store the options as ancillary
1172 		 * data. Note that a hop-by-hop options header must be
1173 		 * just after the IPv6 header, which fact is assured through
1174 		 * the IPv6 input processing.
1175 		 */
1176 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1177 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1178 			struct ip6_hbh *hbh;
1179 			int hbhlen;
1180 			struct mbuf *ext;
1181 
1182 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1183 			    xip6->ip6_nxt);
1184 			if (ext == NULL) {
1185 				IP6_STATINC(IP6_STAT_TOOSHORT);
1186 				return;
1187 			}
1188 			hbh = mtod(ext, struct ip6_hbh *);
1189 			hbhlen = (hbh->ip6h_len + 1) << 3;
1190 			if (hbhlen != ext->m_len) {
1191 				m_freem(ext);
1192 				IP6_STATINC(IP6_STAT_TOOSHORT);
1193 				return;
1194 			}
1195 
1196 			/*
1197 			 * XXX: We copy whole the header even if a jumbo
1198 			 * payload option is included, which option is to
1199 			 * be removed before returning in the RFC 2292.
1200 			 * Note: this constraint is removed in RFC3542.
1201 			 */
1202 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
1203 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1204 			    IPPROTO_IPV6);
1205 			if (*mp)
1206 				mp = &(*mp)->m_next;
1207 			m_freem(ext);
1208 		}
1209 	}
1210 
1211 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1212 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1213 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1214 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1215 
1216 		/*
1217 		 * Search for destination options headers or routing
1218 		 * header(s) through the header chain, and stores each
1219 		 * header as ancillary data.
1220 		 * Note that the order of the headers remains in
1221 		 * the chain of ancillary data.
1222 		 */
1223 		for (;;) {	/* is explicit loop prevention necessary? */
1224 			struct ip6_ext *ip6e = NULL;
1225 			int elen;
1226 			struct mbuf *ext = NULL;
1227 
1228 			/*
1229 			 * if it is not an extension header, don't try to
1230 			 * pull it from the chain.
1231 			 */
1232 			switch (nxt) {
1233 			case IPPROTO_DSTOPTS:
1234 			case IPPROTO_ROUTING:
1235 			case IPPROTO_HOPOPTS:
1236 			case IPPROTO_AH: /* is it possible? */
1237 				break;
1238 			default:
1239 				goto loopend;
1240 			}
1241 
1242 			ext = ip6_pullexthdr(m, off, nxt);
1243 			if (ext == NULL) {
1244 				IP6_STATINC(IP6_STAT_TOOSHORT);
1245 				return;
1246 			}
1247 			ip6e = mtod(ext, struct ip6_ext *);
1248 			if (nxt == IPPROTO_AH)
1249 				elen = (ip6e->ip6e_len + 2) << 2;
1250 			else
1251 				elen = (ip6e->ip6e_len + 1) << 3;
1252 			if (elen != ext->m_len) {
1253 				m_freem(ext);
1254 				IP6_STATINC(IP6_STAT_TOOSHORT);
1255 				return;
1256 			}
1257 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1258 
1259 			switch (nxt) {
1260 			case IPPROTO_DSTOPTS:
1261 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1262 					break;
1263 
1264 				*mp = sbcreatecontrol((void *)ip6e, elen,
1265 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1266 				    IPPROTO_IPV6);
1267 				if (*mp)
1268 					mp = &(*mp)->m_next;
1269 				break;
1270 
1271 			case IPPROTO_ROUTING:
1272 				if (!(in6p->in6p_flags & IN6P_RTHDR))
1273 					break;
1274 
1275 				*mp = sbcreatecontrol((void *)ip6e, elen,
1276 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1277 				    IPPROTO_IPV6);
1278 				if (*mp)
1279 					mp = &(*mp)->m_next;
1280 				break;
1281 
1282 			case IPPROTO_HOPOPTS:
1283 			case IPPROTO_AH: /* is it possible? */
1284 				break;
1285 
1286 			default:
1287 				/*
1288 			 	 * other cases have been filtered in the above.
1289 				 * none will visit this case.  here we supply
1290 				 * the code just in case (nxt overwritten or
1291 				 * other cases).
1292 				 */
1293 				m_freem(ext);
1294 				goto loopend;
1295 
1296 			}
1297 
1298 			/* proceed with the next header. */
1299 			off += elen;
1300 			nxt = ip6e->ip6e_nxt;
1301 			ip6e = NULL;
1302 			m_freem(ext);
1303 			ext = NULL;
1304 		}
1305 	  loopend:
1306 	  	;
1307 	}
1308 }
1309 #undef IS2292
1310 
1311 
1312 void
1313 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
1314     uint32_t *mtu)
1315 {
1316 	struct socket *so;
1317 	struct mbuf *m_mtu;
1318 	struct ip6_mtuinfo mtuctl;
1319 
1320 	so = in6p->in6p_socket;
1321 
1322 	if (mtu == NULL)
1323 		return;
1324 
1325 	KASSERT(so != NULL);
1326 
1327 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
1328 	mtuctl.ip6m_mtu = *mtu;
1329 	mtuctl.ip6m_addr = *dst;
1330 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
1331 		return;
1332 
1333 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
1334 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1335 		return;
1336 
1337 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
1338 	    == 0) {
1339 		m_freem(m_mtu);
1340 		/* XXX: should count statistics */
1341 	} else
1342 		sorwakeup(so);
1343 
1344 	return;
1345 }
1346 
1347 /*
1348  * pull single extension header from mbuf chain.  returns single mbuf that
1349  * contains the result, or NULL on error.
1350  */
1351 static struct mbuf *
1352 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1353 {
1354 	struct ip6_ext ip6e;
1355 	size_t elen;
1356 	struct mbuf *n;
1357 
1358 #ifdef DIAGNOSTIC
1359 	switch (nxt) {
1360 	case IPPROTO_DSTOPTS:
1361 	case IPPROTO_ROUTING:
1362 	case IPPROTO_HOPOPTS:
1363 	case IPPROTO_AH: /* is it possible? */
1364 		break;
1365 	default:
1366 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1367 	}
1368 #endif
1369 
1370 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1371 	if (nxt == IPPROTO_AH)
1372 		elen = (ip6e.ip6e_len + 2) << 2;
1373 	else
1374 		elen = (ip6e.ip6e_len + 1) << 3;
1375 
1376 	MGET(n, M_DONTWAIT, MT_DATA);
1377 	if (n && elen >= MLEN) {
1378 		MCLGET(n, M_DONTWAIT);
1379 		if ((n->m_flags & M_EXT) == 0) {
1380 			m_free(n);
1381 			n = NULL;
1382 		}
1383 	}
1384 	if (!n)
1385 		return NULL;
1386 
1387 	n->m_len = 0;
1388 	if (elen >= M_TRAILINGSPACE(n)) {
1389 		m_free(n);
1390 		return NULL;
1391 	}
1392 
1393 	m_copydata(m, off, elen, mtod(n, void *));
1394 	n->m_len = elen;
1395 	return n;
1396 }
1397 
1398 /*
1399  * Get pointer to the previous header followed by the header
1400  * currently processed.
1401  * XXX: This function supposes that
1402  *	M includes all headers,
1403  *	the next header field and the header length field of each header
1404  *	are valid, and
1405  *	the sum of each header length equals to OFF.
1406  * Because of these assumptions, this function must be called very
1407  * carefully. Moreover, it will not be used in the near future when
1408  * we develop `neater' mechanism to process extension headers.
1409  */
1410 u_int8_t *
1411 ip6_get_prevhdr(struct mbuf *m, int off)
1412 {
1413 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1414 
1415 	if (off == sizeof(struct ip6_hdr))
1416 		return (&ip6->ip6_nxt);
1417 	else {
1418 		int len, nxt;
1419 		struct ip6_ext *ip6e = NULL;
1420 
1421 		nxt = ip6->ip6_nxt;
1422 		len = sizeof(struct ip6_hdr);
1423 		while (len < off) {
1424 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
1425 
1426 			switch (nxt) {
1427 			case IPPROTO_FRAGMENT:
1428 				len += sizeof(struct ip6_frag);
1429 				break;
1430 			case IPPROTO_AH:
1431 				len += (ip6e->ip6e_len + 2) << 2;
1432 				break;
1433 			default:
1434 				len += (ip6e->ip6e_len + 1) << 3;
1435 				break;
1436 			}
1437 			nxt = ip6e->ip6e_nxt;
1438 		}
1439 		if (ip6e)
1440 			return (&ip6e->ip6e_nxt);
1441 		else
1442 			return NULL;
1443 	}
1444 }
1445 
1446 /*
1447  * get next header offset.  m will be retained.
1448  */
1449 int
1450 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1451 {
1452 	struct ip6_hdr ip6;
1453 	struct ip6_ext ip6e;
1454 	struct ip6_frag fh;
1455 
1456 	/* just in case */
1457 	if (m == NULL)
1458 		panic("ip6_nexthdr: m == NULL");
1459 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1460 		return -1;
1461 
1462 	switch (proto) {
1463 	case IPPROTO_IPV6:
1464 		/* do not chase beyond intermediate IPv6 headers */
1465 		if (off != 0)
1466 			return -1;
1467 		if (m->m_pkthdr.len < off + sizeof(ip6))
1468 			return -1;
1469 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
1470 		if (nxtp)
1471 			*nxtp = ip6.ip6_nxt;
1472 		off += sizeof(ip6);
1473 		return off;
1474 
1475 	case IPPROTO_FRAGMENT:
1476 		/*
1477 		 * terminate parsing if it is not the first fragment,
1478 		 * it does not make sense to parse through it.
1479 		 */
1480 		if (m->m_pkthdr.len < off + sizeof(fh))
1481 			return -1;
1482 		m_copydata(m, off, sizeof(fh), (void *)&fh);
1483 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1484 			return -1;
1485 		if (nxtp)
1486 			*nxtp = fh.ip6f_nxt;
1487 		off += sizeof(struct ip6_frag);
1488 		return off;
1489 
1490 	case IPPROTO_AH:
1491 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1492 			return -1;
1493 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1494 		if (nxtp)
1495 			*nxtp = ip6e.ip6e_nxt;
1496 		off += (ip6e.ip6e_len + 2) << 2;
1497 		if (m->m_pkthdr.len < off)
1498 			return -1;
1499 		return off;
1500 
1501 	case IPPROTO_HOPOPTS:
1502 	case IPPROTO_ROUTING:
1503 	case IPPROTO_DSTOPTS:
1504 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1505 			return -1;
1506 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1507 		if (nxtp)
1508 			*nxtp = ip6e.ip6e_nxt;
1509 		off += (ip6e.ip6e_len + 1) << 3;
1510 		if (m->m_pkthdr.len < off)
1511 			return -1;
1512 		return off;
1513 
1514 	case IPPROTO_NONE:
1515 	case IPPROTO_ESP:
1516 	case IPPROTO_IPCOMP:
1517 		/* give up */
1518 		return -1;
1519 
1520 	default:
1521 		return -1;
1522 	}
1523 }
1524 
1525 /*
1526  * get offset for the last header in the chain.  m will be kept untainted.
1527  */
1528 int
1529 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1530 {
1531 	int newoff;
1532 	int nxt;
1533 
1534 	if (!nxtp) {
1535 		nxt = -1;
1536 		nxtp = &nxt;
1537 	}
1538 	for (;;) {
1539 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1540 		if (newoff < 0)
1541 			return off;
1542 		else if (newoff < off)
1543 			return -1;	/* invalid */
1544 		else if (newoff == off)
1545 			return newoff;
1546 
1547 		off = newoff;
1548 		proto = *nxtp;
1549 	}
1550 }
1551 
1552 struct m_tag *
1553 ip6_addaux(struct mbuf *m)
1554 {
1555 	struct m_tag *mtag;
1556 
1557 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1558 	if (!mtag) {
1559 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1560 		    M_NOWAIT);
1561 		if (mtag) {
1562 			m_tag_prepend(m, mtag);
1563 			memset(mtag + 1, 0, sizeof(struct ip6aux));
1564 		}
1565 	}
1566 	return mtag;
1567 }
1568 
1569 struct m_tag *
1570 ip6_findaux(struct mbuf *m)
1571 {
1572 	struct m_tag *mtag;
1573 
1574 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1575 	return mtag;
1576 }
1577 
1578 void
1579 ip6_delaux(struct mbuf *m)
1580 {
1581 	struct m_tag *mtag;
1582 
1583 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1584 	if (mtag)
1585 		m_tag_delete(m, mtag);
1586 }
1587 
1588 /*
1589  * System control for IP6
1590  */
1591 
1592 const u_char inet6ctlerrmap[PRC_NCMDS] = {
1593 	0,		0,		0,		0,
1594 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1595 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1596 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1597 	0,		0,		0,		0,
1598 	ENOPROTOOPT
1599 };
1600 
1601 extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS);
1602 
1603 static int
1604 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
1605 {
1606 
1607 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
1608 }
1609 
1610 static void
1611 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
1612 {
1613 #ifdef RFC2292
1614 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1615 #else
1616 #define IS2292(x, y)	(y)
1617 #endif
1618 
1619 	sysctl_createv(clog, 0, NULL, NULL,
1620 		       CTLFLAG_PERMANENT,
1621 		       CTLTYPE_NODE, "inet6",
1622 		       SYSCTL_DESCR("PF_INET6 related settings"),
1623 		       NULL, 0, NULL, 0,
1624 		       CTL_NET, PF_INET6, CTL_EOL);
1625 	sysctl_createv(clog, 0, NULL, NULL,
1626 		       CTLFLAG_PERMANENT,
1627 		       CTLTYPE_NODE, "ip6",
1628 		       SYSCTL_DESCR("IPv6 related settings"),
1629 		       NULL, 0, NULL, 0,
1630 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1631 
1632 	sysctl_createv(clog, 0, NULL, NULL,
1633 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1634 		       CTLTYPE_INT, "forwarding",
1635 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1636 		       NULL, 0, &ip6_forwarding, 0,
1637 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1638 		       IPV6CTL_FORWARDING, CTL_EOL);
1639 	sysctl_createv(clog, 0, NULL, NULL,
1640 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1641 		       CTLTYPE_INT, "redirect",
1642 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1643 		       NULL, 0, &ip6_sendredirects, 0,
1644 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1645 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1646 	sysctl_createv(clog, 0, NULL, NULL,
1647 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1648 		       CTLTYPE_INT, "hlim",
1649 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1650 		       NULL, 0, &ip6_defhlim, 0,
1651 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1652 		       IPV6CTL_DEFHLIM, CTL_EOL);
1653 #ifdef notyet
1654 	sysctl_createv(clog, 0, NULL, NULL,
1655 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1656 		       CTLTYPE_INT, "mtu", NULL,
1657 		       NULL, 0, &, 0,
1658 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1659 		       IPV6CTL_DEFMTU, CTL_EOL);
1660 #endif
1661 #ifdef __no_idea__
1662 	sysctl_createv(clog, 0, NULL, NULL,
1663 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1664 		       CTLTYPE_INT, "forwsrcrt", NULL,
1665 		       NULL, 0, &?, 0,
1666 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1667 		       IPV6CTL_FORWSRCRT, CTL_EOL);
1668 	sysctl_createv(clog, 0, NULL, NULL,
1669 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1670 		       CTLTYPE_STRUCT, "mrtstats", NULL,
1671 		       NULL, 0, &?, sizeof(?),
1672 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1673 		       IPV6CTL_MRTSTATS, CTL_EOL);
1674 	sysctl_createv(clog, 0, NULL, NULL,
1675 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1676 		       CTLTYPE_?, "mrtproto", NULL,
1677 		       NULL, 0, &?, sizeof(?),
1678 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1679 		       IPV6CTL_MRTPROTO, CTL_EOL);
1680 #endif
1681 	sysctl_createv(clog, 0, NULL, NULL,
1682 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1683 		       CTLTYPE_INT, "maxfragpackets",
1684 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1685 				    "for reassembly"),
1686 		       NULL, 0, &ip6_maxfragpackets, 0,
1687 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1688 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1689 #ifdef __no_idea__
1690 	sysctl_createv(clog, 0, NULL, NULL,
1691 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1692 		       CTLTYPE_INT, "sourcecheck", NULL,
1693 		       NULL, 0, &?, 0,
1694 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1695 		       IPV6CTL_SOURCECHECK, CTL_EOL);
1696 	sysctl_createv(clog, 0, NULL, NULL,
1697 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1698 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
1699 		       NULL, 0, &?, 0,
1700 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1701 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
1702 #endif
1703 	sysctl_createv(clog, 0, NULL, NULL,
1704 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1705 		       CTLTYPE_INT, "accept_rtadv",
1706 		       SYSCTL_DESCR("Accept router advertisements"),
1707 		       NULL, 0, &ip6_accept_rtadv, 0,
1708 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1709 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1710 	sysctl_createv(clog, 0, NULL, NULL,
1711 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1712 		       CTLTYPE_INT, "rtadv_maxroutes",
1713 		       SYSCTL_DESCR("Maximum number of routes accepted via router advertisements"),
1714 		       NULL, 0, &ip6_rtadv_maxroutes, 0,
1715 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1716 		       IPV6CTL_RTADV_MAXROUTES, CTL_EOL);
1717 	sysctl_createv(clog, 0, NULL, NULL,
1718 		       CTLFLAG_PERMANENT,
1719 		       CTLTYPE_INT, "rtadv_numroutes",
1720 		       SYSCTL_DESCR("Current number of routes accepted via router advertisements"),
1721 		       NULL, 0, &nd6_numroutes, 0,
1722 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1723 		       IPV6CTL_RTADV_NUMROUTES, CTL_EOL);
1724 	sysctl_createv(clog, 0, NULL, NULL,
1725 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1726 		       CTLTYPE_INT, "keepfaith",
1727 		       SYSCTL_DESCR("Activate faith interface"),
1728 		       NULL, 0, &ip6_keepfaith, 0,
1729 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1730 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1731 	sysctl_createv(clog, 0, NULL, NULL,
1732 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1733 		       CTLTYPE_INT, "log_interval",
1734 		       SYSCTL_DESCR("Minumum interval between logging "
1735 				    "unroutable packets"),
1736 		       NULL, 0, &ip6_log_interval, 0,
1737 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1738 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1739 	sysctl_createv(clog, 0, NULL, NULL,
1740 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1741 		       CTLTYPE_INT, "hdrnestlimit",
1742 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1743 		       NULL, 0, &ip6_hdrnestlimit, 0,
1744 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1745 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1746 	sysctl_createv(clog, 0, NULL, NULL,
1747 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1748 		       CTLTYPE_INT, "dad_count",
1749 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1750 				    "probes to send"),
1751 		       NULL, 0, &ip6_dad_count, 0,
1752 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1753 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1754 	sysctl_createv(clog, 0, NULL, NULL,
1755 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1756 		       CTLTYPE_INT, "auto_flowlabel",
1757 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1758 		       NULL, 0, &ip6_auto_flowlabel, 0,
1759 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1760 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1761 	sysctl_createv(clog, 0, NULL, NULL,
1762 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1763 		       CTLTYPE_INT, "defmcasthlim",
1764 		       SYSCTL_DESCR("Default multicast hop limit"),
1765 		       NULL, 0, &ip6_defmcasthlim, 0,
1766 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1767 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1768 	sysctl_createv(clog, 0, NULL, NULL,
1769 		       CTLFLAG_PERMANENT,
1770 		       CTLTYPE_STRING, "kame_version",
1771 		       SYSCTL_DESCR("KAME Version"),
1772 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1773 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1774 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1775 	sysctl_createv(clog, 0, NULL, NULL,
1776 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1777 		       CTLTYPE_INT, "use_deprecated",
1778 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1779 				    "source addresses"),
1780 		       NULL, 0, &ip6_use_deprecated, 0,
1781 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1782 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1783 	sysctl_createv(clog, 0, NULL, NULL,
1784 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1785 		       CTLTYPE_INT, "rr_prune", NULL,
1786 		       NULL, 0, &ip6_rr_prune, 0,
1787 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1788 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1789 	sysctl_createv(clog, 0, NULL, NULL,
1790 		       CTLFLAG_PERMANENT
1791 #ifndef INET6_BINDV6ONLY
1792 		       |CTLFLAG_READWRITE,
1793 #endif
1794 		       CTLTYPE_INT, "v6only",
1795 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1796 				    "to PF_INET sockets"),
1797 		       NULL, 0, &ip6_v6only, 0,
1798 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1799 		       IPV6CTL_V6ONLY, CTL_EOL);
1800 	sysctl_createv(clog, 0, NULL, NULL,
1801 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1802 		       CTLTYPE_INT, "anonportmin",
1803 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1804 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1805 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1806 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1807 	sysctl_createv(clog, 0, NULL, NULL,
1808 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1809 		       CTLTYPE_INT, "anonportmax",
1810 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1811 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1812 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1813 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1814 #ifndef IPNOPRIVPORTS
1815 	sysctl_createv(clog, 0, NULL, NULL,
1816 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1817 		       CTLTYPE_INT, "lowportmin",
1818 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1819 				    "to assign"),
1820 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1821 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1822 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1823 	sysctl_createv(clog, 0, NULL, NULL,
1824 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1825 		       CTLTYPE_INT, "lowportmax",
1826 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1827 				    "to assign"),
1828 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1829 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1830 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1831 #endif /* IPNOPRIVPORTS */
1832 	sysctl_createv(clog, 0, NULL, NULL,
1833 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1834 		       CTLTYPE_INT, "auto_linklocal",
1835 		       SYSCTL_DESCR("Default value of per-interface flag for "
1836 		                    "adding an IPv6 link-local address to "
1837 				    "interfaces when attached"),
1838 		       NULL, 0, &ip6_auto_linklocal, 0,
1839 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1840 		       IPV6CTL_AUTO_LINKLOCAL, CTL_EOL);
1841 	sysctl_createv(clog, 0, NULL, NULL,
1842 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
1843 		       CTLTYPE_STRUCT, "addctlpolicy",
1844 		       SYSCTL_DESCR("Return the current address control"
1845 			   " policy"),
1846 		       sysctl_net_inet6_addrctlpolicy, 0, NULL, 0,
1847 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1848 		       IPV6CTL_ADDRCTLPOLICY, CTL_EOL);
1849 	sysctl_createv(clog, 0, NULL, NULL,
1850 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1851 		       CTLTYPE_INT, "use_tempaddr",
1852 		       SYSCTL_DESCR("Use temporary address"),
1853 		       NULL, 0, &ip6_use_tempaddr, 0,
1854 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1855 		       CTL_CREATE, CTL_EOL);
1856 	sysctl_createv(clog, 0, NULL, NULL,
1857 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1858 		       CTLTYPE_INT, "prefer_tempaddr",
1859 		       SYSCTL_DESCR("Prefer temporary address as source "
1860 		                    "address"),
1861 		       NULL, 0, &ip6_prefer_tempaddr, 0,
1862 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1863 		       CTL_CREATE, CTL_EOL);
1864 	sysctl_createv(clog, 0, NULL, NULL,
1865 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1866 		       CTLTYPE_INT, "temppltime",
1867 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
1868 		       NULL, 0, &ip6_temp_preferred_lifetime, 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, "tempvltime",
1874 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
1875 		       NULL, 0, &ip6_temp_valid_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, "maxfrags",
1881 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1882 		       NULL, 0, &ip6_maxfrags, 0,
1883 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1884 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1885 	sysctl_createv(clog, 0, NULL, NULL,
1886 		       CTLFLAG_PERMANENT,
1887 		       CTLTYPE_STRUCT, "stats",
1888 		       SYSCTL_DESCR("IPv6 statistics"),
1889 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
1890 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1891 		       IPV6CTL_STATS, CTL_EOL);
1892 	sysctl_createv(clog, 0, NULL, NULL,
1893 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1894 		       CTLTYPE_INT, "use_defaultzone",
1895 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1896 		       NULL, 0, &ip6_use_defzone, 0,
1897 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1898 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1899 	sysctl_createv(clog, 0, NULL, NULL,
1900 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1901 		       CTLTYPE_INT, "mcast_pmtu",
1902 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
1903 		       NULL, 0, &ip6_mcast_pmtu, 0,
1904 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1905 		       CTL_CREATE, CTL_EOL);
1906 	/* anonportalgo RFC6056 subtree */
1907 	const struct sysctlnode *portalgo_node;
1908 	sysctl_createv(clog, 0, NULL, &portalgo_node,
1909 		       CTLFLAG_PERMANENT,
1910 		       CTLTYPE_NODE, "anonportalgo",
1911 		       SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"),
1912 	    	       NULL, 0, NULL, 0,
1913 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL);
1914 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1915 		       CTLFLAG_PERMANENT,
1916 		       CTLTYPE_STRING, "available",
1917 		       SYSCTL_DESCR("available algorithms"),
1918 		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1919 		       CTL_CREATE, CTL_EOL);
1920 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1921 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1922 		       CTLTYPE_STRING, "selected",
1923 		       SYSCTL_DESCR("selected algorithm"),
1924 	               sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN,
1925 		       CTL_CREATE, CTL_EOL);
1926 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1927 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1928 		       CTLTYPE_STRUCT, "reserve",
1929 		       SYSCTL_DESCR("bitmap of reserved ports"),
1930 		       sysctl_portalgo_reserve6, 0, NULL, 0,
1931 		       CTL_CREATE, CTL_EOL);
1932 	sysctl_createv(clog, 0, NULL, NULL,
1933 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1934 		       CTLTYPE_INT, "neighborgcthresh",
1935 		       SYSCTL_DESCR("Maximum number of entries in neighbor"
1936 			" cache"),
1937 		       NULL, 1, &ip6_neighborgcthresh, 0,
1938 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1939 		       CTL_CREATE, CTL_EOL);
1940 	sysctl_createv(clog, 0, NULL, NULL,
1941 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1942 		       CTLTYPE_INT, "maxifprefixes",
1943 		       SYSCTL_DESCR("Maximum number of prefixes created by"
1944 			   " route advertisement per interface"),
1945 		       NULL, 1, &ip6_maxifprefixes, 0,
1946 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1947 		       CTL_CREATE, CTL_EOL);
1948 	sysctl_createv(clog, 0, NULL, NULL,
1949 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1950 		       CTLTYPE_INT, "maxifdefrouters",
1951 		       SYSCTL_DESCR("Maximum number of default routers created"
1952 			   " by route advertisement per interface"),
1953 		       NULL, 1, &ip6_maxifdefrouters, 0,
1954 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1955 		       CTL_CREATE, CTL_EOL);
1956 	sysctl_createv(clog, 0, NULL, NULL,
1957 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1958 		       CTLTYPE_INT, "maxdynroutes",
1959 		       SYSCTL_DESCR("Maximum number of routes created via"
1960 			   " redirect"),
1961 		       NULL, 1, &ip6_maxdynroutes, 0,
1962 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1963 		       CTL_CREATE, CTL_EOL);
1964 }
1965 
1966 void
1967 ip6_statinc(u_int stat)
1968 {
1969 
1970 	KASSERT(stat < IP6_NSTATS);
1971 	IP6_STATINC(stat);
1972 }
1973