xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 3816d47b2c42fcd6e549e3407f842a5b1a1d23ad)
1 /*	$NetBSD: ip6_input.c,v 1.128 2009/09/16 15:23:05 pooka 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 1.128 2009/09/16 15:23:05 pooka Exp $");
66 
67 #include "opt_inet.h"
68 #include "opt_inet6.h"
69 #include "opt_ipsec.h"
70 #include "opt_pfil_hooks.h"
71 #include "opt_compat_netbsd.h"
72 
73 #include <sys/param.h>
74 #include <sys/systm.h>
75 #include <sys/malloc.h>
76 #include <sys/mbuf.h>
77 #include <sys/domain.h>
78 #include <sys/protosw.h>
79 #include <sys/socket.h>
80 #include <sys/socketvar.h>
81 #include <sys/errno.h>
82 #include <sys/time.h>
83 #include <sys/kernel.h>
84 #include <sys/syslog.h>
85 #include <sys/proc.h>
86 #include <sys/sysctl.h>
87 
88 #include <net/if.h>
89 #include <net/if_types.h>
90 #include <net/if_dl.h>
91 #include <net/route.h>
92 #include <net/netisr.h>
93 #ifdef PFIL_HOOKS
94 #include <net/pfil.h>
95 #endif
96 
97 #include <netinet/in.h>
98 #include <netinet/in_systm.h>
99 #ifdef INET
100 #include <netinet/ip.h>
101 #include <netinet/ip_icmp.h>
102 #endif /* INET */
103 #include <netinet/ip6.h>
104 #include <netinet6/in6_var.h>
105 #include <netinet6/ip6_var.h>
106 #include <netinet6/ip6_private.h>
107 #include <netinet6/in6_pcb.h>
108 #include <netinet/icmp6.h>
109 #include <netinet6/scope6_var.h>
110 #include <netinet6/in6_ifattach.h>
111 #include <netinet6/nd6.h>
112 
113 #ifdef IPSEC
114 #include <netinet6/ipsec.h>
115 #include <netinet6/ipsec_private.h>
116 #endif
117 
118 #ifdef FAST_IPSEC
119 #include <netipsec/ipsec.h>
120 #include <netipsec/ipsec6.h>
121 #include <netipsec/key.h>
122 #endif /* FAST_IPSEC */
123 
124 #ifdef COMPAT_50
125 #include <compat/sys/time.h>
126 #include <compat/sys/socket.h>
127 #endif
128 
129 #include <netinet6/ip6protosw.h>
130 
131 #include "faith.h"
132 #include "gif.h"
133 
134 #if NGIF > 0
135 #include <netinet6/in6_gif.h>
136 #endif
137 
138 #include <net/net_osdep.h>
139 
140 extern struct domain inet6domain;
141 
142 u_char ip6_protox[IPPROTO_MAX];
143 static int ip6qmaxlen = IFQ_MAXLEN;
144 struct in6_ifaddr *in6_ifaddr;
145 struct ifqueue ip6intrq;
146 
147 extern callout_t in6_tmpaddrtimer_ch;
148 
149 int ip6_forward_srcrt;			/* XXX */
150 int ip6_sourcecheck;			/* XXX */
151 int ip6_sourcecheck_interval;		/* XXX */
152 
153 #ifdef PFIL_HOOKS
154 struct pfil_head inet6_pfil_hook;
155 #endif
156 
157 percpu_t *ip6stat_percpu;
158 
159 static void ip6_init2(void *);
160 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
161 
162 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
163 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
164 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
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 	sysctl_net_inet6_ip6_setup(NULL);
177 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
178 	if (pr == 0)
179 		panic("ip6_init");
180 	for (i = 0; i < IPPROTO_MAX; i++)
181 		ip6_protox[i] = pr - inet6sw;
182 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
183 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
184 		if (pr->pr_domain->dom_family == PF_INET6 &&
185 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
186 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
187 	ip6intrq.ifq_maxlen = ip6qmaxlen;
188 	scope6_init();
189 	addrsel_policy_init();
190 	nd6_init();
191 	frag6_init();
192 	ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
193 
194 	ip6_init2((void *)0);
195 #ifdef GATEWAY
196 	ip6flow_init(ip6_hashsize);
197 #endif
198 
199 #ifdef PFIL_HOOKS
200 	/* Register our Packet Filter hook. */
201 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
202 	inet6_pfil_hook.ph_af   = AF_INET6;
203 	i = pfil_head_register(&inet6_pfil_hook);
204 	if (i != 0)
205 		printf("ip6_init: WARNING: unable to register pfil hook, "
206 		    "error %d\n", i);
207 #endif /* PFIL_HOOKS */
208 
209 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
210 }
211 
212 static void
213 ip6_init2(void *dummy)
214 {
215 
216 	/* nd6_timer_init */
217 	callout_init(&nd6_timer_ch, CALLOUT_MPSAFE);
218 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
219 
220 	/* timer for regeneranation of temporary addresses randomize ID */
221 	callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE);
222 	callout_reset(&in6_tmpaddrtimer_ch,
223 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
224 		       ip6_temp_regen_advance) * hz,
225 		      in6_tmpaddrtimer, NULL);
226 }
227 
228 /*
229  * IP6 input interrupt handling. Just pass the packet to ip6_input.
230  */
231 void
232 ip6intr(void)
233 {
234 	int s;
235 	struct mbuf *m;
236 
237 	mutex_enter(softnet_lock);
238 	KERNEL_LOCK(1, NULL);
239 	for (;;) {
240 		s = splnet();
241 		IF_DEQUEUE(&ip6intrq, m);
242 		splx(s);
243 		if (m == 0)
244 			break;
245 		/* drop the packet if IPv6 operation is disabled on the IF */
246 		if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
247 			m_freem(m);
248 			break;
249 		}
250 		ip6_input(m);
251 	}
252 	KERNEL_UNLOCK_ONE(NULL);
253 	mutex_exit(softnet_lock);
254 }
255 
256 extern struct	route ip6_forward_rt;
257 
258 void
259 ip6_input(struct mbuf *m)
260 {
261 	struct ip6_hdr *ip6;
262 	int hit, off = sizeof(struct ip6_hdr), nest;
263 	u_int32_t plen;
264 	u_int32_t rtalert = ~0;
265 	int nxt, ours = 0, rh_present = 0;
266 	struct ifnet *deliverifp = NULL;
267 	int srcrt = 0;
268 	const struct rtentry *rt;
269 	union {
270 		struct sockaddr		dst;
271 		struct sockaddr_in6	dst6;
272 	} u;
273 #ifdef FAST_IPSEC
274 	struct m_tag *mtag;
275 	struct tdb_ident *tdbi;
276 	struct secpolicy *sp;
277 	int s, error;
278 #endif
279 
280 #ifdef IPSEC
281 	/*
282 	 * should the inner packet be considered authentic?
283 	 * see comment in ah4_input().
284 	 */
285 	m->m_flags &= ~M_AUTHIPHDR;
286 	m->m_flags &= ~M_AUTHIPDGM;
287 #endif
288 
289 	/*
290 	 * make sure we don't have onion peering information into m_tag.
291 	 */
292 	ip6_delaux(m);
293 
294 	/*
295 	 * mbuf statistics
296 	 */
297 	if (m->m_flags & M_EXT) {
298 		if (m->m_next)
299 			IP6_STATINC(IP6_STAT_MEXT2M);
300 		else
301 			IP6_STATINC(IP6_STAT_MEXT1);
302 	} else {
303 #define M2MMAX	32
304 		if (m->m_next) {
305 			if (m->m_flags & M_LOOP) {
306 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
307 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX) {
308 				IP6_STATINC(IP6_STAT_M2M +
309 					    m->m_pkthdr.rcvif->if_index);
310 			} else
311 				IP6_STATINC(IP6_STAT_M2M);
312 		} else
313 			IP6_STATINC(IP6_STAT_M1);
314 #undef M2MMAX
315 	}
316 
317 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
318 	IP6_STATINC(IP6_STAT_TOTAL);
319 
320 	/*
321 	 * If the IPv6 header is not aligned, slurp it up into a new
322 	 * mbuf with space for link headers, in the event we forward
323 	 * it.  Otherwise, if it is aligned, make sure the entire base
324 	 * IPv6 header is in the first mbuf of the chain.
325 	 */
326 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
327 		struct ifnet *inifp = m->m_pkthdr.rcvif;
328 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
329 				  (max_linkhdr + 3) & ~3)) == NULL) {
330 			/* XXXJRT new stat, please */
331 			IP6_STATINC(IP6_STAT_TOOSMALL);
332 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
333 			return;
334 		}
335 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
336 		struct ifnet *inifp = m->m_pkthdr.rcvif;
337 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
338 			IP6_STATINC(IP6_STAT_TOOSMALL);
339 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
340 			return;
341 		}
342 	}
343 
344 	ip6 = mtod(m, struct ip6_hdr *);
345 
346 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
347 		IP6_STATINC(IP6_STAT_BADVERS);
348 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
349 		goto bad;
350 	}
351 
352 #if defined(IPSEC)
353 	/* IPv6 fast forwarding is not compatible with IPsec. */
354 	m->m_flags &= ~M_CANFASTFWD;
355 #else
356 	/*
357 	 * Assume that we can create a fast-forward IP flow entry
358 	 * based on this packet.
359 	 */
360 	m->m_flags |= M_CANFASTFWD;
361 #endif
362 
363 #ifdef PFIL_HOOKS
364 	/*
365 	 * Run through list of hooks for input packets.  If there are any
366 	 * filters which require that additional packets in the flow are
367 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
368 	 * Note that filters must _never_ set this flag, as another filter
369 	 * in the list may have previously cleared it.
370 	 */
371 	/*
372 	 * let ipfilter look at packet on the wire,
373 	 * not the decapsulated packet.
374 	 */
375 #ifdef IPSEC
376 	if (!ipsec_getnhist(m))
377 #elif defined(FAST_IPSEC)
378 	if (!ipsec_indone(m))
379 #else
380 	if (1)
381 #endif
382 	{
383 		struct in6_addr odst;
384 
385 		odst = ip6->ip6_dst;
386 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
387 				   PFIL_IN) != 0)
388 			return;
389 		if (m == NULL)
390 			return;
391 		ip6 = mtod(m, struct ip6_hdr *);
392 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
393 	}
394 #endif /* PFIL_HOOKS */
395 
396 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
397 
398 #ifdef ALTQ
399 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
400 		/* packet is dropped by traffic conditioner */
401 		return;
402 	}
403 #endif
404 
405 	/*
406 	 * Check against address spoofing/corruption.
407 	 */
408 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
409 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
410 		/*
411 		 * XXX: "badscope" is not very suitable for a multicast source.
412 		 */
413 		IP6_STATINC(IP6_STAT_BADSCOPE);
414 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
415 		goto bad;
416 	}
417 	/*
418 	 * The following check is not documented in specs.  A malicious
419 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
420 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
421 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
422 	 *
423 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
424 	 * support IPv4-less kernel compilation, we cannot support SIIT
425 	 * environment at all.  So, it makes more sense for us to reject any
426 	 * malicious packets for non-SIIT environment, than try to do a
427 	 * partial support for SIIT environment.
428 	 */
429 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
430 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
431 		IP6_STATINC(IP6_STAT_BADSCOPE);
432 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
433 		goto bad;
434 	}
435 #if 0
436 	/*
437 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
438 	 *
439 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
440 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
441 	 * is revised to forbid relaying case.
442 	 */
443 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
444 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
445 		IP6_STATINC(IP6_STAT_BADSCOPE);
446 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
447 		goto bad;
448 	}
449 #endif
450 
451 	/*
452 	 * Disambiguate address scope zones (if there is ambiguity).
453 	 * We first make sure that the original source or destination address
454 	 * is not in our internal form for scoped addresses.  Such addresses
455 	 * are not necessarily invalid spec-wise, but we cannot accept them due
456 	 * to the usage conflict.
457 	 * in6_setscope() then also checks and rejects the cases where src or
458 	 * dst are the loopback address and the receiving interface
459 	 * is not loopback.
460 	 */
461 	if (__predict_false(
462 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
463 		goto bad;
464 	ip6 = mtod(m, struct ip6_hdr *);
465 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
466 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
467 		goto bad;
468 	}
469 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
470 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
471 		IP6_STATINC(IP6_STAT_BADSCOPE);
472 		goto bad;
473 	}
474 
475 	/*
476 	 * Multicast check
477 	 */
478 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
479 	  	struct	in6_multi *in6m = 0;
480 
481 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
482 		/*
483 		 * See if we belong to the destination multicast group on the
484 		 * arrival interface.
485 		 */
486 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
487 		if (in6m)
488 			ours = 1;
489 		else if (!ip6_mrouter) {
490 			uint64_t *ip6s = IP6_STAT_GETREF();
491 			ip6s[IP6_STAT_NOTMEMBER]++;
492 			ip6s[IP6_STAT_CANTFORWARD]++;
493 			IP6_STAT_PUTREF();
494 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
495 			goto bad;
496 		}
497 		deliverifp = m->m_pkthdr.rcvif;
498 		goto hbhcheck;
499 	}
500 
501 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
502 
503 	/*
504 	 *  Unicast check
505 	 */
506 	rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit);
507 	if (hit)
508 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
509 	else
510 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
511 
512 #define rt6_getkey(__rt) satocsin6(rt_getkey(__rt))
513 
514 	/*
515 	 * Accept the packet if the forwarding interface to the destination
516 	 * according to the routing table is the loopback interface,
517 	 * unless the associated route has a gateway.
518 	 * Note that this approach causes to accept a packet if there is a
519 	 * route to the loopback interface for the destination of the packet.
520 	 * But we think it's even useful in some situations, e.g. when using
521 	 * a special daemon which wants to intercept the packet.
522 	 */
523 	if (rt != NULL &&
524 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
525 	    !(rt->rt_flags & RTF_CLONED) &&
526 #if 0
527 	    /*
528 	     * The check below is redundant since the comparison of
529 	     * the destination and the key of the rtentry has
530 	     * already done through looking up the routing table.
531 	     */
532 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(rt)->sin6_addr) &&
533 #endif
534 	    rt->rt_ifp->if_type == IFT_LOOP) {
535 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
536 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
537 			m->m_flags |= M_ANYCAST6;
538 		/*
539 		 * packets to a tentative, duplicated, or somehow invalid
540 		 * address must not be accepted.
541 		 */
542 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
543 			/* this address is ready */
544 			ours = 1;
545 			deliverifp = ia6->ia_ifp;	/* correct? */
546 			goto hbhcheck;
547 		} else {
548 			/* address is not ready, so discard the packet. */
549 			nd6log((LOG_INFO,
550 			    "ip6_input: packet to an unready address %s->%s\n",
551 			    ip6_sprintf(&ip6->ip6_src),
552 			    ip6_sprintf(&ip6->ip6_dst)));
553 
554 			goto bad;
555 		}
556 	}
557 
558 	/*
559 	 * FAITH (Firewall Aided Internet Translator)
560 	 */
561 #if defined(NFAITH) && 0 < NFAITH
562 	if (ip6_keepfaith) {
563 		if (rt != NULL && rt->rt_ifp != NULL &&
564 		    rt->rt_ifp->if_type == IFT_FAITH) {
565 			/* XXX do we need more sanity checks? */
566 			ours = 1;
567 			deliverifp = rt->rt_ifp; /* faith */
568 			goto hbhcheck;
569 		}
570 	}
571 #endif
572 
573 #if 0
574     {
575 	/*
576 	 * Last resort: check in6_ifaddr for incoming interface.
577 	 * The code is here until I update the "goto ours hack" code above
578 	 * working right.
579 	 */
580 	struct ifaddr *ifa;
581 	IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
582 		if (ifa->ifa_addr == NULL)
583 			continue;	/* just for safety */
584 		if (ifa->ifa_addr->sa_family != AF_INET6)
585 			continue;
586 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
587 			ours = 1;
588 			deliverifp = ifa->ifa_ifp;
589 			goto hbhcheck;
590 		}
591 	}
592     }
593 #endif
594 
595 	/*
596 	 * Now there is no reason to process the packet if it's not our own
597 	 * and we're not a router.
598 	 */
599 	if (!ip6_forwarding) {
600 		IP6_STATINC(IP6_STAT_CANTFORWARD);
601 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
602 		goto bad;
603 	}
604 
605   hbhcheck:
606 	/*
607 	 * record address information into m_tag, if we don't have one yet.
608 	 * note that we are unable to record it, if the address is not listed
609 	 * as our interface address (e.g. multicast addresses, addresses
610 	 * within FAITH prefixes and such).
611 	 */
612 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
613 		struct in6_ifaddr *ia6;
614 
615 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
616 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
617 			/*
618 			 * XXX maybe we should drop the packet here,
619 			 * as we could not provide enough information
620 			 * to the upper layers.
621 			 */
622 		}
623 	}
624 
625 	/*
626 	 * Process Hop-by-Hop options header if it's contained.
627 	 * m may be modified in ip6_hopopts_input().
628 	 * If a JumboPayload option is included, plen will also be modified.
629 	 */
630 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
631 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
632 		struct ip6_hbh *hbh;
633 
634 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
635 #if 0	/*touches NULL pointer*/
636 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
637 #endif
638 			return;	/* m have already been freed */
639 		}
640 
641 		/* adjust pointer */
642 		ip6 = mtod(m, struct ip6_hdr *);
643 
644 		/*
645 		 * if the payload length field is 0 and the next header field
646 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
647 		 * option MUST be included.
648 		 */
649 		if (ip6->ip6_plen == 0 && plen == 0) {
650 			/*
651 			 * Note that if a valid jumbo payload option is
652 			 * contained, ip6_hopopts_input() must set a valid
653 			 * (non-zero) payload length to the variable plen.
654 			 */
655 			IP6_STATINC(IP6_STAT_BADOPTIONS);
656 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
657 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
658 			icmp6_error(m, ICMP6_PARAM_PROB,
659 				    ICMP6_PARAMPROB_HEADER,
660 				    (char *)&ip6->ip6_plen - (char *)ip6);
661 			return;
662 		}
663 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
664 			sizeof(struct ip6_hbh));
665 		if (hbh == NULL) {
666 			IP6_STATINC(IP6_STAT_TOOSHORT);
667 			return;
668 		}
669 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
670 		nxt = hbh->ip6h_nxt;
671 
672 		/*
673 		 * accept the packet if a router alert option is included
674 		 * and we act as an IPv6 router.
675 		 */
676 		if (rtalert != ~0 && ip6_forwarding)
677 			ours = 1;
678 	} else
679 		nxt = ip6->ip6_nxt;
680 
681 	/*
682 	 * Check that the amount of data in the buffers
683 	 * is as at least much as the IPv6 header would have us expect.
684 	 * Trim mbufs if longer than we expect.
685 	 * Drop packet if shorter than we expect.
686 	 */
687 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
688 		IP6_STATINC(IP6_STAT_TOOSHORT);
689 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
690 		goto bad;
691 	}
692 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
693 		if (m->m_len == m->m_pkthdr.len) {
694 			m->m_len = sizeof(struct ip6_hdr) + plen;
695 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
696 		} else
697 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
698 	}
699 
700 	/*
701 	 * Forward if desirable.
702 	 */
703 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
704 		/*
705 		 * If we are acting as a multicast router, all
706 		 * incoming multicast packets are passed to the
707 		 * kernel-level multicast forwarding function.
708 		 * The packet is returned (relatively) intact; if
709 		 * ip6_mforward() returns a non-zero value, the packet
710 		 * must be discarded, else it may be accepted below.
711 		 */
712 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
713 			IP6_STATINC(IP6_STAT_CANTFORWARD);
714 			m_freem(m);
715 			return;
716 		}
717 		if (!ours) {
718 			m_freem(m);
719 			return;
720 		}
721 	} else if (!ours) {
722 		ip6_forward(m, srcrt);
723 		return;
724 	}
725 
726 	ip6 = mtod(m, struct ip6_hdr *);
727 
728 	/*
729 	 * Malicious party may be able to use IPv4 mapped addr to confuse
730 	 * tcp/udp stack and bypass security checks (act as if it was from
731 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
732 	 *
733 	 * For SIIT end node behavior, you may want to disable the check.
734 	 * However, you will  become vulnerable to attacks using IPv4 mapped
735 	 * source.
736 	 */
737 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
738 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
739 		IP6_STATINC(IP6_STAT_BADSCOPE);
740 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
741 		goto bad;
742 	}
743 
744 	/*
745 	 * Tell launch routine the next header
746 	 */
747 #ifdef IFA_STATS
748 	if (deliverifp != NULL) {
749 		struct in6_ifaddr *ia6;
750 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
751 		if (ia6)
752 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
753 	}
754 #endif
755 	IP6_STATINC(IP6_STAT_DELIVERED);
756 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
757 	nest = 0;
758 
759 	rh_present = 0;
760 	while (nxt != IPPROTO_DONE) {
761 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
762 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
763 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
764 			goto bad;
765 		}
766 
767 		/*
768 		 * protection against faulty packet - there should be
769 		 * more sanity checks in header chain processing.
770 		 */
771 		if (m->m_pkthdr.len < off) {
772 			IP6_STATINC(IP6_STAT_TOOSHORT);
773 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
774 			goto bad;
775 		}
776 
777 		if (nxt == IPPROTO_ROUTING) {
778 			if (rh_present++) {
779 				in6_ifstat_inc(m->m_pkthdr.rcvif,
780 				    ifs6_in_hdrerr);
781 				IP6_STATINC(IP6_STAT_BADOPTIONS);
782 				goto bad;
783 			}
784 		}
785 
786 #ifdef IPSEC
787 		/*
788 		 * enforce IPsec policy checking if we are seeing last header.
789 		 * note that we do not visit this with protocols with pcb layer
790 		 * code - like udp/tcp/raw ip.
791 		 */
792 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
793 		    ipsec6_in_reject(m, NULL)) {
794 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
795 			goto bad;
796 		}
797 #endif
798 #ifdef FAST_IPSEC
799 	/*
800 	 * enforce IPsec policy checking if we are seeing last header.
801 	 * note that we do not visit this with protocols with pcb layer
802 	 * code - like udp/tcp/raw ip.
803 	 */
804 	if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) {
805 		/*
806 		 * Check if the packet has already had IPsec processing
807 		 * done.  If so, then just pass it along.  This tag gets
808 		 * set during AH, ESP, etc. input handling, before the
809 		 * packet is returned to the ip input queue for delivery.
810 		 */
811 		mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
812 		s = splsoftnet();
813 		if (mtag != NULL) {
814 			tdbi = (struct tdb_ident *)(mtag + 1);
815 			sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
816 		} else {
817 			sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
818 									IP_FORWARDING, &error);
819 		}
820 		if (sp != NULL) {
821 			/*
822 			 * Check security policy against packet attributes.
823 			 */
824 			error = ipsec_in_reject(sp, m);
825 			KEY_FREESP(&sp);
826 		} else {
827 			/* XXX error stat??? */
828 			error = EINVAL;
829 			DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/
830 		}
831 		splx(s);
832 		if (error)
833 			goto bad;
834 	}
835 #endif /* FAST_IPSEC */
836 
837 
838 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
839 	}
840 	return;
841  bad:
842 	m_freem(m);
843 }
844 
845 /*
846  * set/grab in6_ifaddr correspond to IPv6 destination address.
847  */
848 static struct m_tag *
849 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
850 {
851 	struct m_tag *mtag;
852 
853 	mtag = ip6_addaux(m);
854 	if (mtag != NULL) {
855 		struct ip6aux *ip6a;
856 
857 		ip6a = (struct ip6aux *)(mtag + 1);
858 		in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id);
859 		ip6a->ip6a_src = ia->ia_addr.sin6_addr;
860 		ip6a->ip6a_flags = ia->ia6_flags;
861 	}
862 	return mtag;	/* NULL if failed to set */
863 }
864 
865 const struct ip6aux *
866 ip6_getdstifaddr(struct mbuf *m)
867 {
868 	struct m_tag *mtag;
869 
870 	mtag = ip6_findaux(m);
871 	if (mtag != NULL)
872 		return (struct ip6aux *)(mtag + 1);
873 	else
874 		return NULL;
875 }
876 
877 /*
878  * Hop-by-Hop options header processing. If a valid jumbo payload option is
879  * included, the real payload length will be stored in plenp.
880  *
881  * rtalertp - XXX: should be stored more smart way
882  */
883 static int
884 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
885 	struct mbuf **mp, int *offp)
886 {
887 	struct mbuf *m = *mp;
888 	int off = *offp, hbhlen;
889 	struct ip6_hbh *hbh;
890 
891 	/* validation of the length of the header */
892 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
893 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
894 	if (hbh == NULL) {
895 		IP6_STATINC(IP6_STAT_TOOSHORT);
896 		return -1;
897 	}
898 	hbhlen = (hbh->ip6h_len + 1) << 3;
899 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
900 		hbhlen);
901 	if (hbh == NULL) {
902 		IP6_STATINC(IP6_STAT_TOOSHORT);
903 		return -1;
904 	}
905 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
906 	off += hbhlen;
907 	hbhlen -= sizeof(struct ip6_hbh);
908 
909 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
910 				hbhlen, rtalertp, plenp) < 0)
911 		return (-1);
912 
913 	*offp = off;
914 	*mp = m;
915 	return (0);
916 }
917 
918 /*
919  * Search header for all Hop-by-hop options and process each option.
920  * This function is separate from ip6_hopopts_input() in order to
921  * handle a case where the sending node itself process its hop-by-hop
922  * options header. In such a case, the function is called from ip6_output().
923  *
924  * The function assumes that hbh header is located right after the IPv6 header
925  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
926  * opthead + hbhlen is located in continuous memory region.
927  */
928 int
929 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
930 	u_int32_t *rtalertp, u_int32_t *plenp)
931 {
932 	struct ip6_hdr *ip6;
933 	int optlen = 0;
934 	u_int8_t *opt = opthead;
935 	u_int16_t rtalert_val;
936 	u_int32_t jumboplen;
937 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
938 
939 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
940 		switch (*opt) {
941 		case IP6OPT_PAD1:
942 			optlen = 1;
943 			break;
944 		case IP6OPT_PADN:
945 			if (hbhlen < IP6OPT_MINLEN) {
946 				IP6_STATINC(IP6_STAT_TOOSMALL);
947 				goto bad;
948 			}
949 			optlen = *(opt + 1) + 2;
950 			break;
951 		case IP6OPT_RTALERT:
952 			/* XXX may need check for alignment */
953 			if (hbhlen < IP6OPT_RTALERT_LEN) {
954 				IP6_STATINC(IP6_STAT_TOOSMALL);
955 				goto bad;
956 			}
957 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
958 				/* XXX stat */
959 				icmp6_error(m, ICMP6_PARAM_PROB,
960 				    ICMP6_PARAMPROB_HEADER,
961 				    erroff + opt + 1 - opthead);
962 				return (-1);
963 			}
964 			optlen = IP6OPT_RTALERT_LEN;
965 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
966 			*rtalertp = ntohs(rtalert_val);
967 			break;
968 		case IP6OPT_JUMBO:
969 			/* XXX may need check for alignment */
970 			if (hbhlen < IP6OPT_JUMBO_LEN) {
971 				IP6_STATINC(IP6_STAT_TOOSMALL);
972 				goto bad;
973 			}
974 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
975 				/* XXX stat */
976 				icmp6_error(m, ICMP6_PARAM_PROB,
977 				    ICMP6_PARAMPROB_HEADER,
978 				    erroff + opt + 1 - opthead);
979 				return (-1);
980 			}
981 			optlen = IP6OPT_JUMBO_LEN;
982 
983 			/*
984 			 * IPv6 packets that have non 0 payload length
985 			 * must not contain a jumbo payload option.
986 			 */
987 			ip6 = mtod(m, struct ip6_hdr *);
988 			if (ip6->ip6_plen) {
989 				IP6_STATINC(IP6_STAT_BADOPTIONS);
990 				icmp6_error(m, ICMP6_PARAM_PROB,
991 				    ICMP6_PARAMPROB_HEADER,
992 				    erroff + opt - opthead);
993 				return (-1);
994 			}
995 
996 			/*
997 			 * We may see jumbolen in unaligned location, so
998 			 * we'd need to perform bcopy().
999 			 */
1000 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
1001 			jumboplen = (u_int32_t)htonl(jumboplen);
1002 
1003 #if 1
1004 			/*
1005 			 * if there are multiple jumbo payload options,
1006 			 * *plenp will be non-zero and the packet will be
1007 			 * rejected.
1008 			 * the behavior may need some debate in ipngwg -
1009 			 * multiple options does not make sense, however,
1010 			 * there's no explicit mention in specification.
1011 			 */
1012 			if (*plenp != 0) {
1013 				IP6_STATINC(IP6_STAT_BADOPTIONS);
1014 				icmp6_error(m, ICMP6_PARAM_PROB,
1015 				    ICMP6_PARAMPROB_HEADER,
1016 				    erroff + opt + 2 - opthead);
1017 				return (-1);
1018 			}
1019 #endif
1020 
1021 			/*
1022 			 * jumbo payload length must be larger than 65535.
1023 			 */
1024 			if (jumboplen <= IPV6_MAXPACKET) {
1025 				IP6_STATINC(IP6_STAT_BADOPTIONS);
1026 				icmp6_error(m, ICMP6_PARAM_PROB,
1027 				    ICMP6_PARAMPROB_HEADER,
1028 				    erroff + opt + 2 - opthead);
1029 				return (-1);
1030 			}
1031 			*plenp = jumboplen;
1032 
1033 			break;
1034 		default:		/* unknown option */
1035 			if (hbhlen < IP6OPT_MINLEN) {
1036 				IP6_STATINC(IP6_STAT_TOOSMALL);
1037 				goto bad;
1038 			}
1039 			optlen = ip6_unknown_opt(opt, m,
1040 			    erroff + opt - opthead);
1041 			if (optlen == -1)
1042 				return (-1);
1043 			optlen += 2;
1044 			break;
1045 		}
1046 	}
1047 
1048 	return (0);
1049 
1050   bad:
1051 	m_freem(m);
1052 	return (-1);
1053 }
1054 
1055 /*
1056  * Unknown option processing.
1057  * The third argument `off' is the offset from the IPv6 header to the option,
1058  * which is necessary if the IPv6 header the and option header and IPv6 header
1059  * is not continuous in order to return an ICMPv6 error.
1060  */
1061 int
1062 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1063 {
1064 	struct ip6_hdr *ip6;
1065 
1066 	switch (IP6OPT_TYPE(*optp)) {
1067 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1068 		return ((int)*(optp + 1));
1069 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1070 		m_freem(m);
1071 		return (-1);
1072 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1073 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1074 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1075 		return (-1);
1076 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1077 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1078 		ip6 = mtod(m, struct ip6_hdr *);
1079 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1080 		    (m->m_flags & (M_BCAST|M_MCAST)))
1081 			m_freem(m);
1082 		else
1083 			icmp6_error(m, ICMP6_PARAM_PROB,
1084 				    ICMP6_PARAMPROB_OPTION, off);
1085 		return (-1);
1086 	}
1087 
1088 	m_freem(m);		/* XXX: NOTREACHED */
1089 	return (-1);
1090 }
1091 
1092 /*
1093  * Create the "control" list for this pcb.
1094  *
1095  * The routine will be called from upper layer handlers like tcp6_input().
1096  * Thus the routine assumes that the caller (tcp6_input) have already
1097  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1098  * very first mbuf on the mbuf chain.
1099  * We may want to add some infinite loop prevention or sanity checks for safety.
1100  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1101  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1102  */
1103 void
1104 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
1105 	struct ip6_hdr *ip6, struct mbuf *m)
1106 {
1107 #ifdef RFC2292
1108 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1109 #else
1110 #define IS2292(x, y)	(y)
1111 #endif
1112 
1113 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP
1114 #ifdef SO_OTIMESTAMP
1115 	    || in6p->in6p_socket->so_options & SO_OTIMESTAMP
1116 #endif
1117 	) {
1118 		struct timeval tv;
1119 
1120 		microtime(&tv);
1121 #ifdef SO_OTIMESTAMP
1122 		if (in6p->in6p_socket->so_options & SO_OTIMESTAMP) {
1123 			struct timeval50 tv50;
1124 			timeval_to_timeval50(&tv, &tv50);
1125 			*mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
1126 			    SCM_OTIMESTAMP, SOL_SOCKET);
1127 		} else
1128 #endif
1129 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1130 		    SCM_TIMESTAMP, SOL_SOCKET);
1131 		if (*mp)
1132 			mp = &(*mp)->m_next;
1133 	}
1134 
1135 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
1136 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1137 		return;
1138 
1139 	/* RFC 2292 sec. 5 */
1140 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1141 		struct in6_pktinfo pi6;
1142 
1143 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
1144 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1145 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
1146 		    m->m_pkthdr.rcvif->if_index : 0;
1147 		*mp = sbcreatecontrol((void *) &pi6,
1148 		    sizeof(struct in6_pktinfo),
1149 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1150 		if (*mp)
1151 			mp = &(*mp)->m_next;
1152 	}
1153 
1154 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1155 		int hlim = ip6->ip6_hlim & 0xff;
1156 
1157 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
1158 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1159 		if (*mp)
1160 			mp = &(*mp)->m_next;
1161 	}
1162 
1163 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1164 		u_int32_t flowinfo;
1165 		int tclass;
1166 
1167 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1168 		flowinfo >>= 20;
1169 
1170 		tclass = flowinfo & 0xff;
1171 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
1172 		    IPV6_TCLASS, IPPROTO_IPV6);
1173 
1174 		if (*mp)
1175 			mp = &(*mp)->m_next;
1176 	}
1177 
1178 	/*
1179 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1180 	 * privilege for the option (see ip6_ctloutput), but it might be too
1181 	 * strict, since there might be some hop-by-hop options which can be
1182 	 * returned to normal user.
1183 	 * See also RFC3542 section 8 (or RFC2292 section 6).
1184 	 */
1185 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1186 		/*
1187 		 * Check if a hop-by-hop options header is contatined in the
1188 		 * received packet, and if so, store the options as ancillary
1189 		 * data. Note that a hop-by-hop options header must be
1190 		 * just after the IPv6 header, which fact is assured through
1191 		 * the IPv6 input processing.
1192 		 */
1193 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1194 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1195 			struct ip6_hbh *hbh;
1196 			int hbhlen;
1197 			struct mbuf *ext;
1198 
1199 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1200 			    xip6->ip6_nxt);
1201 			if (ext == NULL) {
1202 				IP6_STATINC(IP6_STAT_TOOSHORT);
1203 				return;
1204 			}
1205 			hbh = mtod(ext, struct ip6_hbh *);
1206 			hbhlen = (hbh->ip6h_len + 1) << 3;
1207 			if (hbhlen != ext->m_len) {
1208 				m_freem(ext);
1209 				IP6_STATINC(IP6_STAT_TOOSHORT);
1210 				return;
1211 			}
1212 
1213 			/*
1214 			 * XXX: We copy whole the header even if a jumbo
1215 			 * payload option is included, which option is to
1216 			 * be removed before returning in the RFC 2292.
1217 			 * Note: this constraint is removed in RFC3542.
1218 			 */
1219 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
1220 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1221 			    IPPROTO_IPV6);
1222 			if (*mp)
1223 				mp = &(*mp)->m_next;
1224 			m_freem(ext);
1225 		}
1226 	}
1227 
1228 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1229 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1230 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1231 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1232 
1233 		/*
1234 		 * Search for destination options headers or routing
1235 		 * header(s) through the header chain, and stores each
1236 		 * header as ancillary data.
1237 		 * Note that the order of the headers remains in
1238 		 * the chain of ancillary data.
1239 		 */
1240 		for (;;) {	/* is explicit loop prevention necessary? */
1241 			struct ip6_ext *ip6e = NULL;
1242 			int elen;
1243 			struct mbuf *ext = NULL;
1244 
1245 			/*
1246 			 * if it is not an extension header, don't try to
1247 			 * pull it from the chain.
1248 			 */
1249 			switch (nxt) {
1250 			case IPPROTO_DSTOPTS:
1251 			case IPPROTO_ROUTING:
1252 			case IPPROTO_HOPOPTS:
1253 			case IPPROTO_AH: /* is it possible? */
1254 				break;
1255 			default:
1256 				goto loopend;
1257 			}
1258 
1259 			ext = ip6_pullexthdr(m, off, nxt);
1260 			if (ext == NULL) {
1261 				IP6_STATINC(IP6_STAT_TOOSHORT);
1262 				return;
1263 			}
1264 			ip6e = mtod(ext, struct ip6_ext *);
1265 			if (nxt == IPPROTO_AH)
1266 				elen = (ip6e->ip6e_len + 2) << 2;
1267 			else
1268 				elen = (ip6e->ip6e_len + 1) << 3;
1269 			if (elen != ext->m_len) {
1270 				m_freem(ext);
1271 				IP6_STATINC(IP6_STAT_TOOSHORT);
1272 				return;
1273 			}
1274 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1275 
1276 			switch (nxt) {
1277 			case IPPROTO_DSTOPTS:
1278 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1279 					break;
1280 
1281 				*mp = sbcreatecontrol((void *)ip6e, elen,
1282 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1283 				    IPPROTO_IPV6);
1284 				if (*mp)
1285 					mp = &(*mp)->m_next;
1286 				break;
1287 
1288 			case IPPROTO_ROUTING:
1289 				if (!(in6p->in6p_flags & IN6P_RTHDR))
1290 					break;
1291 
1292 				*mp = sbcreatecontrol((void *)ip6e, elen,
1293 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1294 				    IPPROTO_IPV6);
1295 				if (*mp)
1296 					mp = &(*mp)->m_next;
1297 				break;
1298 
1299 			case IPPROTO_HOPOPTS:
1300 			case IPPROTO_AH: /* is it possible? */
1301 				break;
1302 
1303 			default:
1304 				/*
1305 			 	 * other cases have been filtered in the above.
1306 				 * none will visit this case.  here we supply
1307 				 * the code just in case (nxt overwritten or
1308 				 * other cases).
1309 				 */
1310 				m_freem(ext);
1311 				goto loopend;
1312 
1313 			}
1314 
1315 			/* proceed with the next header. */
1316 			off += elen;
1317 			nxt = ip6e->ip6e_nxt;
1318 			ip6e = NULL;
1319 			m_freem(ext);
1320 			ext = NULL;
1321 		}
1322 	  loopend:
1323 	  	;
1324 	}
1325 }
1326 #undef IS2292
1327 
1328 
1329 void
1330 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
1331     uint32_t *mtu)
1332 {
1333 	struct socket *so;
1334 	struct mbuf *m_mtu;
1335 	struct ip6_mtuinfo mtuctl;
1336 
1337 	so = in6p->in6p_socket;
1338 
1339 	if (mtu == NULL)
1340 		return;
1341 
1342 #ifdef DIAGNOSTIC
1343 	if (so == NULL)		/* I believe this is impossible */
1344 		panic("ip6_notify_pmtu: socket is NULL");
1345 #endif
1346 
1347 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
1348 	mtuctl.ip6m_mtu = *mtu;
1349 	mtuctl.ip6m_addr = *dst;
1350 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
1351 		return;
1352 
1353 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
1354 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1355 		return;
1356 
1357 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
1358 	    == 0) {
1359 		m_freem(m_mtu);
1360 		/* XXX: should count statistics */
1361 	} else
1362 		sorwakeup(so);
1363 
1364 	return;
1365 }
1366 
1367 /*
1368  * pull single extension header from mbuf chain.  returns single mbuf that
1369  * contains the result, or NULL on error.
1370  */
1371 static struct mbuf *
1372 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1373 {
1374 	struct ip6_ext ip6e;
1375 	size_t elen;
1376 	struct mbuf *n;
1377 
1378 #ifdef DIAGNOSTIC
1379 	switch (nxt) {
1380 	case IPPROTO_DSTOPTS:
1381 	case IPPROTO_ROUTING:
1382 	case IPPROTO_HOPOPTS:
1383 	case IPPROTO_AH: /* is it possible? */
1384 		break;
1385 	default:
1386 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1387 	}
1388 #endif
1389 
1390 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1391 	if (nxt == IPPROTO_AH)
1392 		elen = (ip6e.ip6e_len + 2) << 2;
1393 	else
1394 		elen = (ip6e.ip6e_len + 1) << 3;
1395 
1396 	MGET(n, M_DONTWAIT, MT_DATA);
1397 	if (n && elen >= MLEN) {
1398 		MCLGET(n, M_DONTWAIT);
1399 		if ((n->m_flags & M_EXT) == 0) {
1400 			m_free(n);
1401 			n = NULL;
1402 		}
1403 	}
1404 	if (!n)
1405 		return NULL;
1406 
1407 	n->m_len = 0;
1408 	if (elen >= M_TRAILINGSPACE(n)) {
1409 		m_free(n);
1410 		return NULL;
1411 	}
1412 
1413 	m_copydata(m, off, elen, mtod(n, void *));
1414 	n->m_len = elen;
1415 	return n;
1416 }
1417 
1418 /*
1419  * Get pointer to the previous header followed by the header
1420  * currently processed.
1421  * XXX: This function supposes that
1422  *	M includes all headers,
1423  *	the next header field and the header length field of each header
1424  *	are valid, and
1425  *	the sum of each header length equals to OFF.
1426  * Because of these assumptions, this function must be called very
1427  * carefully. Moreover, it will not be used in the near future when
1428  * we develop `neater' mechanism to process extension headers.
1429  */
1430 u_int8_t *
1431 ip6_get_prevhdr(struct mbuf *m, int off)
1432 {
1433 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1434 
1435 	if (off == sizeof(struct ip6_hdr))
1436 		return (&ip6->ip6_nxt);
1437 	else {
1438 		int len, nxt;
1439 		struct ip6_ext *ip6e = NULL;
1440 
1441 		nxt = ip6->ip6_nxt;
1442 		len = sizeof(struct ip6_hdr);
1443 		while (len < off) {
1444 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
1445 
1446 			switch (nxt) {
1447 			case IPPROTO_FRAGMENT:
1448 				len += sizeof(struct ip6_frag);
1449 				break;
1450 			case IPPROTO_AH:
1451 				len += (ip6e->ip6e_len + 2) << 2;
1452 				break;
1453 			default:
1454 				len += (ip6e->ip6e_len + 1) << 3;
1455 				break;
1456 			}
1457 			nxt = ip6e->ip6e_nxt;
1458 		}
1459 		if (ip6e)
1460 			return (&ip6e->ip6e_nxt);
1461 		else
1462 			return NULL;
1463 	}
1464 }
1465 
1466 /*
1467  * get next header offset.  m will be retained.
1468  */
1469 int
1470 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1471 {
1472 	struct ip6_hdr ip6;
1473 	struct ip6_ext ip6e;
1474 	struct ip6_frag fh;
1475 
1476 	/* just in case */
1477 	if (m == NULL)
1478 		panic("ip6_nexthdr: m == NULL");
1479 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1480 		return -1;
1481 
1482 	switch (proto) {
1483 	case IPPROTO_IPV6:
1484 		/* do not chase beyond intermediate IPv6 headers */
1485 		if (off != 0)
1486 			return -1;
1487 		if (m->m_pkthdr.len < off + sizeof(ip6))
1488 			return -1;
1489 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
1490 		if (nxtp)
1491 			*nxtp = ip6.ip6_nxt;
1492 		off += sizeof(ip6);
1493 		return off;
1494 
1495 	case IPPROTO_FRAGMENT:
1496 		/*
1497 		 * terminate parsing if it is not the first fragment,
1498 		 * it does not make sense to parse through it.
1499 		 */
1500 		if (m->m_pkthdr.len < off + sizeof(fh))
1501 			return -1;
1502 		m_copydata(m, off, sizeof(fh), (void *)&fh);
1503 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1504 			return -1;
1505 		if (nxtp)
1506 			*nxtp = fh.ip6f_nxt;
1507 		off += sizeof(struct ip6_frag);
1508 		return off;
1509 
1510 	case IPPROTO_AH:
1511 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1512 			return -1;
1513 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1514 		if (nxtp)
1515 			*nxtp = ip6e.ip6e_nxt;
1516 		off += (ip6e.ip6e_len + 2) << 2;
1517 		if (m->m_pkthdr.len < off)
1518 			return -1;
1519 		return off;
1520 
1521 	case IPPROTO_HOPOPTS:
1522 	case IPPROTO_ROUTING:
1523 	case IPPROTO_DSTOPTS:
1524 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1525 			return -1;
1526 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1527 		if (nxtp)
1528 			*nxtp = ip6e.ip6e_nxt;
1529 		off += (ip6e.ip6e_len + 1) << 3;
1530 		if (m->m_pkthdr.len < off)
1531 			return -1;
1532 		return off;
1533 
1534 	case IPPROTO_NONE:
1535 	case IPPROTO_ESP:
1536 	case IPPROTO_IPCOMP:
1537 		/* give up */
1538 		return -1;
1539 
1540 	default:
1541 		return -1;
1542 	}
1543 }
1544 
1545 /*
1546  * get offset for the last header in the chain.  m will be kept untainted.
1547  */
1548 int
1549 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1550 {
1551 	int newoff;
1552 	int nxt;
1553 
1554 	if (!nxtp) {
1555 		nxt = -1;
1556 		nxtp = &nxt;
1557 	}
1558 	for (;;) {
1559 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1560 		if (newoff < 0)
1561 			return off;
1562 		else if (newoff < off)
1563 			return -1;	/* invalid */
1564 		else if (newoff == off)
1565 			return newoff;
1566 
1567 		off = newoff;
1568 		proto = *nxtp;
1569 	}
1570 }
1571 
1572 struct m_tag *
1573 ip6_addaux(struct mbuf *m)
1574 {
1575 	struct m_tag *mtag;
1576 
1577 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1578 	if (!mtag) {
1579 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1580 		    M_NOWAIT);
1581 		if (mtag) {
1582 			m_tag_prepend(m, mtag);
1583 			memset(mtag + 1, 0, sizeof(struct ip6aux));
1584 		}
1585 	}
1586 	return mtag;
1587 }
1588 
1589 struct m_tag *
1590 ip6_findaux(struct mbuf *m)
1591 {
1592 	struct m_tag *mtag;
1593 
1594 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1595 	return mtag;
1596 }
1597 
1598 void
1599 ip6_delaux(struct mbuf *m)
1600 {
1601 	struct m_tag *mtag;
1602 
1603 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1604 	if (mtag)
1605 		m_tag_delete(m, mtag);
1606 }
1607 
1608 #ifdef GATEWAY
1609 /*
1610  * sysctl helper routine for net.inet.ip6.maxflows. Since
1611  * we could reduce this value, call ip6flow_reap();
1612  */
1613 static int
1614 sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS)
1615 {
1616 	int error;
1617 
1618 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1619 	if (error || newp == NULL)
1620 		return (error);
1621 
1622 	mutex_enter(softnet_lock);
1623 	KERNEL_LOCK(1, NULL);
1624 
1625 	ip6flow_reap(0);
1626 
1627 	KERNEL_UNLOCK_ONE(NULL);
1628 	mutex_exit(softnet_lock);
1629 
1630 	return (0);
1631 }
1632 
1633 static int
1634 sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS)
1635 {
1636 	int error, tmp;
1637 	struct sysctlnode node;
1638 
1639 	node = *rnode;
1640 	tmp = ip6_hashsize;
1641 	node.sysctl_data = &tmp;
1642 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1643 	if (error || newp == NULL)
1644 		return (error);
1645 
1646 	if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
1647 		/*
1648 		 * Can only fail due to malloc()
1649 		 */
1650 		mutex_enter(softnet_lock);
1651 		KERNEL_LOCK(1, NULL);
1652 
1653 		error = ip6flow_invalidate_all(tmp);
1654 
1655 		KERNEL_UNLOCK_ONE(NULL);
1656 		mutex_exit(softnet_lock);
1657 	} else {
1658 		/*
1659 		 * EINVAL if not a power of 2
1660 		 */
1661 		error = EINVAL;
1662 	}
1663 
1664 	return error;
1665 }
1666 #endif /* GATEWAY */
1667 
1668 /*
1669  * System control for IP6
1670  */
1671 
1672 const u_char inet6ctlerrmap[PRC_NCMDS] = {
1673 	0,		0,		0,		0,
1674 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1675 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1676 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1677 	0,		0,		0,		0,
1678 	ENOPROTOOPT
1679 };
1680 
1681 static int
1682 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
1683 {
1684 
1685 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
1686 }
1687 
1688 static void
1689 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
1690 {
1691 #ifdef RFC2292
1692 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1693 #else
1694 #define IS2292(x, y)	(y)
1695 #endif
1696 
1697 	sysctl_createv(clog, 0, NULL, NULL,
1698 		       CTLFLAG_PERMANENT,
1699 		       CTLTYPE_NODE, "net", NULL,
1700 		       NULL, 0, NULL, 0,
1701 		       CTL_NET, CTL_EOL);
1702 	sysctl_createv(clog, 0, NULL, NULL,
1703 		       CTLFLAG_PERMANENT,
1704 		       CTLTYPE_NODE, "inet6",
1705 		       SYSCTL_DESCR("PF_INET6 related settings"),
1706 		       NULL, 0, NULL, 0,
1707 		       CTL_NET, PF_INET6, CTL_EOL);
1708 	sysctl_createv(clog, 0, NULL, NULL,
1709 		       CTLFLAG_PERMANENT,
1710 		       CTLTYPE_NODE, "ip6",
1711 		       SYSCTL_DESCR("IPv6 related settings"),
1712 		       NULL, 0, NULL, 0,
1713 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1714 
1715 	sysctl_createv(clog, 0, NULL, NULL,
1716 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1717 		       CTLTYPE_INT, "forwarding",
1718 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1719 		       NULL, 0, &ip6_forwarding, 0,
1720 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1721 		       IPV6CTL_FORWARDING, CTL_EOL);
1722 	sysctl_createv(clog, 0, NULL, NULL,
1723 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1724 		       CTLTYPE_INT, "redirect",
1725 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1726 		       NULL, 0, &ip6_sendredirects, 0,
1727 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1728 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1729 	sysctl_createv(clog, 0, NULL, NULL,
1730 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1731 		       CTLTYPE_INT, "hlim",
1732 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1733 		       NULL, 0, &ip6_defhlim, 0,
1734 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1735 		       IPV6CTL_DEFHLIM, CTL_EOL);
1736 #ifdef notyet
1737 	sysctl_createv(clog, 0, NULL, NULL,
1738 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1739 		       CTLTYPE_INT, "mtu", NULL,
1740 		       NULL, 0, &, 0,
1741 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1742 		       IPV6CTL_DEFMTU, CTL_EOL);
1743 #endif
1744 #ifdef __no_idea__
1745 	sysctl_createv(clog, 0, NULL, NULL,
1746 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1747 		       CTLTYPE_INT, "forwsrcrt", NULL,
1748 		       NULL, 0, &?, 0,
1749 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1750 		       IPV6CTL_FORWSRCRT, CTL_EOL);
1751 	sysctl_createv(clog, 0, NULL, NULL,
1752 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1753 		       CTLTYPE_STRUCT, "mrtstats", NULL,
1754 		       NULL, 0, &?, sizeof(?),
1755 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1756 		       IPV6CTL_MRTSTATS, CTL_EOL);
1757 	sysctl_createv(clog, 0, NULL, NULL,
1758 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1759 		       CTLTYPE_?, "mrtproto", NULL,
1760 		       NULL, 0, &?, sizeof(?),
1761 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1762 		       IPV6CTL_MRTPROTO, CTL_EOL);
1763 #endif
1764 	sysctl_createv(clog, 0, NULL, NULL,
1765 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1766 		       CTLTYPE_INT, "maxfragpackets",
1767 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1768 				    "for reassembly"),
1769 		       NULL, 0, &ip6_maxfragpackets, 0,
1770 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1771 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1772 #ifdef __no_idea__
1773 	sysctl_createv(clog, 0, NULL, NULL,
1774 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1775 		       CTLTYPE_INT, "sourcecheck", NULL,
1776 		       NULL, 0, &?, 0,
1777 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1778 		       IPV6CTL_SOURCECHECK, CTL_EOL);
1779 	sysctl_createv(clog, 0, NULL, NULL,
1780 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1781 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
1782 		       NULL, 0, &?, 0,
1783 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1784 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
1785 #endif
1786 	sysctl_createv(clog, 0, NULL, NULL,
1787 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1788 		       CTLTYPE_INT, "accept_rtadv",
1789 		       SYSCTL_DESCR("Accept router advertisements"),
1790 		       NULL, 0, &ip6_accept_rtadv, 0,
1791 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1792 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1793 	sysctl_createv(clog, 0, NULL, NULL,
1794 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1795 		       CTLTYPE_INT, "keepfaith",
1796 		       SYSCTL_DESCR("Activate faith interface"),
1797 		       NULL, 0, &ip6_keepfaith, 0,
1798 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1799 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1800 	sysctl_createv(clog, 0, NULL, NULL,
1801 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1802 		       CTLTYPE_INT, "log_interval",
1803 		       SYSCTL_DESCR("Minumum interval between logging "
1804 				    "unroutable packets"),
1805 		       NULL, 0, &ip6_log_interval, 0,
1806 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1807 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1808 	sysctl_createv(clog, 0, NULL, NULL,
1809 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1810 		       CTLTYPE_INT, "hdrnestlimit",
1811 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1812 		       NULL, 0, &ip6_hdrnestlimit, 0,
1813 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1814 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1815 	sysctl_createv(clog, 0, NULL, NULL,
1816 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1817 		       CTLTYPE_INT, "dad_count",
1818 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1819 				    "probes to send"),
1820 		       NULL, 0, &ip6_dad_count, 0,
1821 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1822 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1823 	sysctl_createv(clog, 0, NULL, NULL,
1824 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1825 		       CTLTYPE_INT, "auto_flowlabel",
1826 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1827 		       NULL, 0, &ip6_auto_flowlabel, 0,
1828 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1829 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1830 	sysctl_createv(clog, 0, NULL, NULL,
1831 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1832 		       CTLTYPE_INT, "defmcasthlim",
1833 		       SYSCTL_DESCR("Default multicast hop limit"),
1834 		       NULL, 0, &ip6_defmcasthlim, 0,
1835 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1836 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1837 #if NGIF > 0
1838 	sysctl_createv(clog, 0, NULL, NULL,
1839 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1840 		       CTLTYPE_INT, "gifhlim",
1841 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
1842 		       NULL, 0, &ip6_gif_hlim, 0,
1843 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1844 		       IPV6CTL_GIF_HLIM, CTL_EOL);
1845 #endif /* NGIF */
1846 	sysctl_createv(clog, 0, NULL, NULL,
1847 		       CTLFLAG_PERMANENT,
1848 		       CTLTYPE_STRING, "kame_version",
1849 		       SYSCTL_DESCR("KAME Version"),
1850 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1851 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1852 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1853 	sysctl_createv(clog, 0, NULL, NULL,
1854 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1855 		       CTLTYPE_INT, "use_deprecated",
1856 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1857 				    "source addresses"),
1858 		       NULL, 0, &ip6_use_deprecated, 0,
1859 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1860 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1861 	sysctl_createv(clog, 0, NULL, NULL,
1862 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1863 		       CTLTYPE_INT, "rr_prune", NULL,
1864 		       NULL, 0, &ip6_rr_prune, 0,
1865 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1866 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1867 	sysctl_createv(clog, 0, NULL, NULL,
1868 		       CTLFLAG_PERMANENT
1869 #ifndef INET6_BINDV6ONLY
1870 		       |CTLFLAG_READWRITE,
1871 #endif
1872 		       CTLTYPE_INT, "v6only",
1873 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1874 				    "to PF_INET sockets"),
1875 		       NULL, 0, &ip6_v6only, 0,
1876 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1877 		       IPV6CTL_V6ONLY, CTL_EOL);
1878 	sysctl_createv(clog, 0, NULL, NULL,
1879 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1880 		       CTLTYPE_INT, "anonportmin",
1881 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1882 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1883 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1884 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1885 	sysctl_createv(clog, 0, NULL, NULL,
1886 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1887 		       CTLTYPE_INT, "anonportmax",
1888 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1889 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1890 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1891 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1892 #ifndef IPNOPRIVPORTS
1893 	sysctl_createv(clog, 0, NULL, NULL,
1894 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1895 		       CTLTYPE_INT, "lowportmin",
1896 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1897 				    "to assign"),
1898 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1899 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1900 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1901 	sysctl_createv(clog, 0, NULL, NULL,
1902 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1903 		       CTLTYPE_INT, "lowportmax",
1904 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1905 				    "to assign"),
1906 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1907 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1908 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1909 #endif /* IPNOPRIVPORTS */
1910 	sysctl_createv(clog, 0, NULL, NULL,
1911 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1912 		       CTLTYPE_INT, "use_tempaddr",
1913 		       SYSCTL_DESCR("Use temporary address"),
1914 		       NULL, 0, &ip6_use_tempaddr, 0,
1915 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1916 		       CTL_CREATE, CTL_EOL);
1917 	sysctl_createv(clog, 0, NULL, NULL,
1918 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1919 		       CTLTYPE_INT, "temppltime",
1920 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
1921 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
1922 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1923 		       CTL_CREATE, CTL_EOL);
1924 	sysctl_createv(clog, 0, NULL, NULL,
1925 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1926 		       CTLTYPE_INT, "tempvltime",
1927 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
1928 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
1929 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1930 		       CTL_CREATE, CTL_EOL);
1931 	sysctl_createv(clog, 0, NULL, NULL,
1932 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1933 		       CTLTYPE_INT, "maxfrags",
1934 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1935 		       NULL, 0, &ip6_maxfrags, 0,
1936 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1937 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1938 	sysctl_createv(clog, 0, NULL, NULL,
1939 		       CTLFLAG_PERMANENT,
1940 		       CTLTYPE_STRUCT, "stats",
1941 		       SYSCTL_DESCR("IPv6 statistics"),
1942 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
1943 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1944 		       IPV6CTL_STATS, CTL_EOL);
1945 	sysctl_createv(clog, 0, NULL, NULL,
1946 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1947 		       CTLTYPE_INT, "use_defaultzone",
1948 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1949 		       NULL, 0, &ip6_use_defzone, 0,
1950 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1951 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1952 	sysctl_createv(clog, 0, NULL, NULL,
1953 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1954 		       CTLTYPE_INT, "mcast_pmtu",
1955 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
1956 		       NULL, 0, &ip6_mcast_pmtu, 0,
1957 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1958 		       CTL_CREATE, CTL_EOL);
1959 #ifdef GATEWAY
1960 	sysctl_createv(clog, 0, NULL, NULL,
1961 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1962 			CTLTYPE_INT, "maxflows",
1963 			SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"),
1964 			sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0,
1965 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1966 			CTL_CREATE, CTL_EOL);
1967 	sysctl_createv(clog, 0, NULL, NULL,
1968 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1969 			CTLTYPE_INT, "hashsize",
1970 			SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"),
1971 			sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0,
1972 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1973 			CTL_CREATE, CTL_EOL);
1974 #endif
1975 }
1976 
1977 void
1978 ip6_statinc(u_int stat)
1979 {
1980 
1981 	KASSERT(stat < IP6_NSTATS);
1982 	IP6_STATINC(stat);
1983 }
1984