xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 466a16a118933bd295a8a104f095714fadf9cf68)
1 /*	$NetBSD: ip6_input.c,v 1.122 2008/08/21 15:34:10 matt 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 1.122 2008/08/21 15:34:10 matt Exp $");
66 
67 #include "opt_inet.h"
68 #include "opt_inet6.h"
69 #include "opt_ipsec.h"
70 #include "opt_pfil_hooks.h"
71 
72 #include <sys/param.h>
73 #include <sys/systm.h>
74 #include <sys/malloc.h>
75 #include <sys/mbuf.h>
76 #include <sys/domain.h>
77 #include <sys/protosw.h>
78 #include <sys/socket.h>
79 #include <sys/socketvar.h>
80 #include <sys/errno.h>
81 #include <sys/time.h>
82 #include <sys/kernel.h>
83 #include <sys/syslog.h>
84 #include <sys/proc.h>
85 #include <sys/sysctl.h>
86 
87 #include <net/if.h>
88 #include <net/if_types.h>
89 #include <net/if_dl.h>
90 #include <net/route.h>
91 #include <net/netisr.h>
92 #ifdef PFIL_HOOKS
93 #include <net/pfil.h>
94 #endif
95 
96 #include <netinet/in.h>
97 #include <netinet/in_systm.h>
98 #ifdef INET
99 #include <netinet/ip.h>
100 #include <netinet/ip_icmp.h>
101 #endif /* INET */
102 #include <netinet/ip6.h>
103 #include <netinet6/in6_var.h>
104 #include <netinet6/ip6_var.h>
105 #include <netinet6/ip6_private.h>
106 #include <netinet6/in6_pcb.h>
107 #include <netinet/icmp6.h>
108 #include <netinet6/scope6_var.h>
109 #include <netinet6/in6_ifattach.h>
110 #include <netinet6/nd6.h>
111 
112 #ifdef IPSEC
113 #include <netinet6/ipsec.h>
114 #include <netinet6/ipsec_private.h>
115 #endif
116 
117 #ifdef FAST_IPSEC
118 #include <netipsec/ipsec.h>
119 #include <netipsec/ipsec6.h>
120 #include <netipsec/key.h>
121 #endif /* FAST_IPSEC */
122 
123 #include <netinet6/ip6protosw.h>
124 
125 #include "faith.h"
126 #include "gif.h"
127 
128 #if NGIF > 0
129 #include <netinet6/in6_gif.h>
130 #endif
131 
132 #include <net/net_osdep.h>
133 
134 extern struct domain inet6domain;
135 
136 u_char ip6_protox[IPPROTO_MAX];
137 static int ip6qmaxlen = IFQ_MAXLEN;
138 struct in6_ifaddr *in6_ifaddr;
139 struct ifqueue ip6intrq;
140 
141 extern callout_t in6_tmpaddrtimer_ch;
142 
143 int ip6_forward_srcrt;			/* XXX */
144 int ip6_sourcecheck;			/* XXX */
145 int ip6_sourcecheck_interval;		/* XXX */
146 
147 #ifdef PFIL_HOOKS
148 struct pfil_head inet6_pfil_hook;
149 #endif
150 
151 percpu_t *ip6stat_percpu;
152 
153 static void ip6_init2(void *);
154 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
155 
156 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
157 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
158 
159 /*
160  * IP6 initialization: fill in IP6 protocol switch table.
161  * All protocols not implemented in kernel go to raw IP6 protocol handler.
162  */
163 void
164 ip6_init(void)
165 {
166 	const struct ip6protosw *pr;
167 	int i;
168 
169 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
170 	if (pr == 0)
171 		panic("ip6_init");
172 	for (i = 0; i < IPPROTO_MAX; i++)
173 		ip6_protox[i] = pr - inet6sw;
174 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
175 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
176 		if (pr->pr_domain->dom_family == PF_INET6 &&
177 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
178 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
179 	ip6intrq.ifq_maxlen = ip6qmaxlen;
180 	scope6_init();
181 	addrsel_policy_init();
182 	nd6_init();
183 	frag6_init();
184 	ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
185 
186 	ip6_init2((void *)0);
187 #ifdef GATEWAY
188 	ip6flow_init(ip6_hashsize);
189 #endif
190 
191 #ifdef PFIL_HOOKS
192 	/* Register our Packet Filter hook. */
193 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
194 	inet6_pfil_hook.ph_af   = AF_INET6;
195 	i = pfil_head_register(&inet6_pfil_hook);
196 	if (i != 0)
197 		printf("ip6_init: WARNING: unable to register pfil hook, "
198 		    "error %d\n", i);
199 #endif /* PFIL_HOOKS */
200 
201 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
202 }
203 
204 static void
205 ip6_init2(void *dummy)
206 {
207 
208 	/* nd6_timer_init */
209 	callout_init(&nd6_timer_ch, CALLOUT_MPSAFE);
210 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
211 
212 	/* timer for regeneranation of temporary addresses randomize ID */
213 	callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE);
214 	callout_reset(&in6_tmpaddrtimer_ch,
215 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
216 		       ip6_temp_regen_advance) * hz,
217 		      in6_tmpaddrtimer, NULL);
218 }
219 
220 /*
221  * IP6 input interrupt handling. Just pass the packet to ip6_input.
222  */
223 void
224 ip6intr(void)
225 {
226 	int s;
227 	struct mbuf *m;
228 
229 	mutex_enter(softnet_lock);
230 	KERNEL_LOCK(1, NULL);
231 	for (;;) {
232 		s = splnet();
233 		IF_DEQUEUE(&ip6intrq, m);
234 		splx(s);
235 		if (m == 0)
236 			break;
237 		/* drop the packet if IPv6 operation is disabled on the IF */
238 		if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
239 			m_freem(m);
240 			break;
241 		}
242 		ip6_input(m);
243 	}
244 	KERNEL_UNLOCK_ONE(NULL);
245 	mutex_exit(softnet_lock);
246 }
247 
248 extern struct	route ip6_forward_rt;
249 
250 void
251 ip6_input(struct mbuf *m)
252 {
253 	struct ip6_hdr *ip6;
254 	int hit, off = sizeof(struct ip6_hdr), nest;
255 	u_int32_t plen;
256 	u_int32_t rtalert = ~0;
257 	int nxt, ours = 0, rh_present = 0;
258 	struct ifnet *deliverifp = NULL;
259 	int srcrt = 0;
260 	const struct rtentry *rt;
261 	union {
262 		struct sockaddr		dst;
263 		struct sockaddr_in6	dst6;
264 	} u;
265 #ifdef FAST_IPSEC
266 	struct m_tag *mtag;
267 	struct tdb_ident *tdbi;
268 	struct secpolicy *sp;
269 	int s, error;
270 #endif
271 
272 #ifdef IPSEC
273 	/*
274 	 * should the inner packet be considered authentic?
275 	 * see comment in ah4_input().
276 	 */
277 	m->m_flags &= ~M_AUTHIPHDR;
278 	m->m_flags &= ~M_AUTHIPDGM;
279 #endif
280 
281 	/*
282 	 * make sure we don't have onion peering information into m_tag.
283 	 */
284 	ip6_delaux(m);
285 
286 	/*
287 	 * mbuf statistics
288 	 */
289 	if (m->m_flags & M_EXT) {
290 		if (m->m_next)
291 			IP6_STATINC(IP6_STAT_MEXT2M);
292 		else
293 			IP6_STATINC(IP6_STAT_MEXT1);
294 	} else {
295 #define M2MMAX	32
296 		if (m->m_next) {
297 			if (m->m_flags & M_LOOP) {
298 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
299 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX) {
300 				IP6_STATINC(IP6_STAT_M2M +
301 					    m->m_pkthdr.rcvif->if_index);
302 			} else
303 				IP6_STATINC(IP6_STAT_M2M);
304 		} else
305 			IP6_STATINC(IP6_STAT_M1);
306 #undef M2MMAX
307 	}
308 
309 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
310 	IP6_STATINC(IP6_STAT_TOTAL);
311 
312 	/*
313 	 * If the IPv6 header is not aligned, slurp it up into a new
314 	 * mbuf with space for link headers, in the event we forward
315 	 * it.  Otherwise, if it is aligned, make sure the entire base
316 	 * IPv6 header is in the first mbuf of the chain.
317 	 */
318 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
319 		struct ifnet *inifp = m->m_pkthdr.rcvif;
320 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
321 				  (max_linkhdr + 3) & ~3)) == NULL) {
322 			/* XXXJRT new stat, please */
323 			IP6_STATINC(IP6_STAT_TOOSMALL);
324 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
325 			return;
326 		}
327 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
328 		struct ifnet *inifp = m->m_pkthdr.rcvif;
329 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
330 			IP6_STATINC(IP6_STAT_TOOSMALL);
331 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
332 			return;
333 		}
334 	}
335 
336 	ip6 = mtod(m, struct ip6_hdr *);
337 
338 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
339 		IP6_STATINC(IP6_STAT_BADVERS);
340 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
341 		goto bad;
342 	}
343 
344 #if defined(IPSEC)
345 	/* IPv6 fast forwarding is not compatible with IPsec. */
346 	m->m_flags &= ~M_CANFASTFWD;
347 #else
348 	/*
349 	 * Assume that we can create a fast-forward IP flow entry
350 	 * based on this packet.
351 	 */
352 	m->m_flags |= M_CANFASTFWD;
353 #endif
354 
355 #ifdef PFIL_HOOKS
356 	/*
357 	 * Run through list of hooks for input packets.  If there are any
358 	 * filters which require that additional packets in the flow are
359 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
360 	 * Note that filters must _never_ set this flag, as another filter
361 	 * in the list may have previously cleared it.
362 	 */
363 	/*
364 	 * let ipfilter look at packet on the wire,
365 	 * not the decapsulated packet.
366 	 */
367 #ifdef IPSEC
368 	if (!ipsec_getnhist(m))
369 #elif defined(FAST_IPSEC)
370 	if (!ipsec_indone(m))
371 #else
372 	if (1)
373 #endif
374 	{
375 		struct in6_addr odst;
376 
377 		odst = ip6->ip6_dst;
378 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
379 				   PFIL_IN) != 0)
380 			return;
381 		if (m == NULL)
382 			return;
383 		ip6 = mtod(m, struct ip6_hdr *);
384 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
385 	}
386 #endif /* PFIL_HOOKS */
387 
388 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
389 
390 #ifdef ALTQ
391 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
392 		/* packet is dropped by traffic conditioner */
393 		return;
394 	}
395 #endif
396 
397 	/*
398 	 * Check against address spoofing/corruption.
399 	 */
400 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
401 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
402 		/*
403 		 * XXX: "badscope" is not very suitable for a multicast source.
404 		 */
405 		IP6_STATINC(IP6_STAT_BADSCOPE);
406 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
407 		goto bad;
408 	}
409 	/*
410 	 * The following check is not documented in specs.  A malicious
411 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
412 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
413 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
414 	 *
415 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
416 	 * support IPv4-less kernel compilation, we cannot support SIIT
417 	 * environment at all.  So, it makes more sense for us to reject any
418 	 * malicious packets for non-SIIT environment, than try to do a
419 	 * partial support for SIIT environment.
420 	 */
421 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
422 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
423 		IP6_STATINC(IP6_STAT_BADSCOPE);
424 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
425 		goto bad;
426 	}
427 #if 0
428 	/*
429 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
430 	 *
431 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
432 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
433 	 * is revised to forbid relaying case.
434 	 */
435 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
436 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
437 		IP6_STATINC(IP6_STAT_BADSCOPE);
438 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
439 		goto bad;
440 	}
441 #endif
442 
443 	/*
444 	 * Disambiguate address scope zones (if there is ambiguity).
445 	 * We first make sure that the original source or destination address
446 	 * is not in our internal form for scoped addresses.  Such addresses
447 	 * are not necessarily invalid spec-wise, but we cannot accept them due
448 	 * to the usage conflict.
449 	 * in6_setscope() then also checks and rejects the cases where src or
450 	 * dst are the loopback address and the receiving interface
451 	 * is not loopback.
452 	 */
453 	if (__predict_false(
454 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
455 		goto bad;
456 	ip6 = mtod(m, struct ip6_hdr *);
457 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
458 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
459 		goto bad;
460 	}
461 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
462 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
463 		IP6_STATINC(IP6_STAT_BADSCOPE);
464 		goto bad;
465 	}
466 
467 	/*
468 	 * Multicast check
469 	 */
470 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
471 	  	struct	in6_multi *in6m = 0;
472 
473 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
474 		/*
475 		 * See if we belong to the destination multicast group on the
476 		 * arrival interface.
477 		 */
478 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
479 		if (in6m)
480 			ours = 1;
481 		else if (!ip6_mrouter) {
482 			uint64_t *ip6s = IP6_STAT_GETREF();
483 			ip6s[IP6_STAT_NOTMEMBER]++;
484 			ip6s[IP6_STAT_CANTFORWARD]++;
485 			IP6_STAT_PUTREF();
486 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
487 			goto bad;
488 		}
489 		deliverifp = m->m_pkthdr.rcvif;
490 		goto hbhcheck;
491 	}
492 
493 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
494 
495 	/*
496 	 *  Unicast check
497 	 */
498 	rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit);
499 	if (hit)
500 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
501 	else
502 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
503 
504 #define rt6_getkey(__rt) satocsin6(rt_getkey(__rt))
505 
506 	/*
507 	 * Accept the packet if the forwarding interface to the destination
508 	 * according to the routing table is the loopback interface,
509 	 * unless the associated route has a gateway.
510 	 * Note that this approach causes to accept a packet if there is a
511 	 * route to the loopback interface for the destination of the packet.
512 	 * But we think it's even useful in some situations, e.g. when using
513 	 * a special daemon which wants to intercept the packet.
514 	 */
515 	if (rt != NULL &&
516 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
517 	    !(rt->rt_flags & RTF_CLONED) &&
518 #if 0
519 	    /*
520 	     * The check below is redundant since the comparison of
521 	     * the destination and the key of the rtentry has
522 	     * already done through looking up the routing table.
523 	     */
524 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(rt)->sin6_addr) &&
525 #endif
526 	    rt->rt_ifp->if_type == IFT_LOOP) {
527 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
528 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
529 			m->m_flags |= M_ANYCAST6;
530 		/*
531 		 * packets to a tentative, duplicated, or somehow invalid
532 		 * address must not be accepted.
533 		 */
534 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
535 			/* this address is ready */
536 			ours = 1;
537 			deliverifp = ia6->ia_ifp;	/* correct? */
538 			goto hbhcheck;
539 		} else {
540 			/* address is not ready, so discard the packet. */
541 			nd6log((LOG_INFO,
542 			    "ip6_input: packet to an unready address %s->%s\n",
543 			    ip6_sprintf(&ip6->ip6_src),
544 			    ip6_sprintf(&ip6->ip6_dst)));
545 
546 			goto bad;
547 		}
548 	}
549 
550 	/*
551 	 * FAITH (Firewall Aided Internet Translator)
552 	 */
553 #if defined(NFAITH) && 0 < NFAITH
554 	if (ip6_keepfaith) {
555 		if (rt != NULL && rt->rt_ifp != NULL &&
556 		    rt->rt_ifp->if_type == IFT_FAITH) {
557 			/* XXX do we need more sanity checks? */
558 			ours = 1;
559 			deliverifp = rt->rt_ifp; /* faith */
560 			goto hbhcheck;
561 		}
562 	}
563 #endif
564 
565 #if 0
566     {
567 	/*
568 	 * Last resort: check in6_ifaddr for incoming interface.
569 	 * The code is here until I update the "goto ours hack" code above
570 	 * working right.
571 	 */
572 	struct ifaddr *ifa;
573 	IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
574 		if (ifa->ifa_addr == NULL)
575 			continue;	/* just for safety */
576 		if (ifa->ifa_addr->sa_family != AF_INET6)
577 			continue;
578 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
579 			ours = 1;
580 			deliverifp = ifa->ifa_ifp;
581 			goto hbhcheck;
582 		}
583 	}
584     }
585 #endif
586 
587 	/*
588 	 * Now there is no reason to process the packet if it's not our own
589 	 * and we're not a router.
590 	 */
591 	if (!ip6_forwarding) {
592 		IP6_STATINC(IP6_STAT_CANTFORWARD);
593 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
594 		goto bad;
595 	}
596 
597   hbhcheck:
598 	/*
599 	 * record address information into m_tag, if we don't have one yet.
600 	 * note that we are unable to record it, if the address is not listed
601 	 * as our interface address (e.g. multicast addresses, addresses
602 	 * within FAITH prefixes and such).
603 	 */
604 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
605 		struct in6_ifaddr *ia6;
606 
607 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
608 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
609 			/*
610 			 * XXX maybe we should drop the packet here,
611 			 * as we could not provide enough information
612 			 * to the upper layers.
613 			 */
614 		}
615 	}
616 
617 	/*
618 	 * Process Hop-by-Hop options header if it's contained.
619 	 * m may be modified in ip6_hopopts_input().
620 	 * If a JumboPayload option is included, plen will also be modified.
621 	 */
622 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
623 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
624 		struct ip6_hbh *hbh;
625 
626 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
627 #if 0	/*touches NULL pointer*/
628 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
629 #endif
630 			return;	/* m have already been freed */
631 		}
632 
633 		/* adjust pointer */
634 		ip6 = mtod(m, struct ip6_hdr *);
635 
636 		/*
637 		 * if the payload length field is 0 and the next header field
638 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
639 		 * option MUST be included.
640 		 */
641 		if (ip6->ip6_plen == 0 && plen == 0) {
642 			/*
643 			 * Note that if a valid jumbo payload option is
644 			 * contained, ip6_hopopts_input() must set a valid
645 			 * (non-zero) payload length to the variable plen.
646 			 */
647 			IP6_STATINC(IP6_STAT_BADOPTIONS);
648 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
649 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
650 			icmp6_error(m, ICMP6_PARAM_PROB,
651 				    ICMP6_PARAMPROB_HEADER,
652 				    (char *)&ip6->ip6_plen - (char *)ip6);
653 			return;
654 		}
655 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
656 			sizeof(struct ip6_hbh));
657 		if (hbh == NULL) {
658 			IP6_STATINC(IP6_STAT_TOOSHORT);
659 			return;
660 		}
661 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
662 		nxt = hbh->ip6h_nxt;
663 
664 		/*
665 		 * accept the packet if a router alert option is included
666 		 * and we act as an IPv6 router.
667 		 */
668 		if (rtalert != ~0 && ip6_forwarding)
669 			ours = 1;
670 	} else
671 		nxt = ip6->ip6_nxt;
672 
673 	/*
674 	 * Check that the amount of data in the buffers
675 	 * is as at least much as the IPv6 header would have us expect.
676 	 * Trim mbufs if longer than we expect.
677 	 * Drop packet if shorter than we expect.
678 	 */
679 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
680 		IP6_STATINC(IP6_STAT_TOOSHORT);
681 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
682 		goto bad;
683 	}
684 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
685 		if (m->m_len == m->m_pkthdr.len) {
686 			m->m_len = sizeof(struct ip6_hdr) + plen;
687 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
688 		} else
689 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
690 	}
691 
692 	/*
693 	 * Forward if desirable.
694 	 */
695 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
696 		/*
697 		 * If we are acting as a multicast router, all
698 		 * incoming multicast packets are passed to the
699 		 * kernel-level multicast forwarding function.
700 		 * The packet is returned (relatively) intact; if
701 		 * ip6_mforward() returns a non-zero value, the packet
702 		 * must be discarded, else it may be accepted below.
703 		 */
704 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
705 			IP6_STATINC(IP6_STAT_CANTFORWARD);
706 			m_freem(m);
707 			return;
708 		}
709 		if (!ours) {
710 			m_freem(m);
711 			return;
712 		}
713 	} else if (!ours) {
714 		ip6_forward(m, srcrt);
715 		return;
716 	}
717 
718 	ip6 = mtod(m, struct ip6_hdr *);
719 
720 	/*
721 	 * Malicious party may be able to use IPv4 mapped addr to confuse
722 	 * tcp/udp stack and bypass security checks (act as if it was from
723 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
724 	 *
725 	 * For SIIT end node behavior, you may want to disable the check.
726 	 * However, you will  become vulnerable to attacks using IPv4 mapped
727 	 * source.
728 	 */
729 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
730 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
731 		IP6_STATINC(IP6_STAT_BADSCOPE);
732 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
733 		goto bad;
734 	}
735 
736 	/*
737 	 * Tell launch routine the next header
738 	 */
739 #ifdef IFA_STATS
740 	if (deliverifp != NULL) {
741 		struct in6_ifaddr *ia6;
742 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
743 		if (ia6)
744 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
745 	}
746 #endif
747 	IP6_STATINC(IP6_STAT_DELIVERED);
748 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
749 	nest = 0;
750 
751 	rh_present = 0;
752 	while (nxt != IPPROTO_DONE) {
753 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
754 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
755 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
756 			goto bad;
757 		}
758 
759 		/*
760 		 * protection against faulty packet - there should be
761 		 * more sanity checks in header chain processing.
762 		 */
763 		if (m->m_pkthdr.len < off) {
764 			IP6_STATINC(IP6_STAT_TOOSHORT);
765 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
766 			goto bad;
767 		}
768 
769 		if (nxt == IPPROTO_ROUTING) {
770 			if (rh_present++) {
771 				in6_ifstat_inc(m->m_pkthdr.rcvif,
772 				    ifs6_in_hdrerr);
773 				IP6_STATINC(IP6_STAT_BADOPTIONS);
774 				goto bad;
775 			}
776 		}
777 
778 #ifdef IPSEC
779 		/*
780 		 * enforce IPsec policy checking if we are seeing last header.
781 		 * note that we do not visit this with protocols with pcb layer
782 		 * code - like udp/tcp/raw ip.
783 		 */
784 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
785 		    ipsec6_in_reject(m, NULL)) {
786 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
787 			goto bad;
788 		}
789 #endif
790 #ifdef FAST_IPSEC
791 	/*
792 	 * enforce IPsec policy checking if we are seeing last header.
793 	 * note that we do not visit this with protocols with pcb layer
794 	 * code - like udp/tcp/raw ip.
795 	 */
796 	if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) {
797 		/*
798 		 * Check if the packet has already had IPsec processing
799 		 * done.  If so, then just pass it along.  This tag gets
800 		 * set during AH, ESP, etc. input handling, before the
801 		 * packet is returned to the ip input queue for delivery.
802 		 */
803 		mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
804 		s = splsoftnet();
805 		if (mtag != NULL) {
806 			tdbi = (struct tdb_ident *)(mtag + 1);
807 			sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
808 		} else {
809 			sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
810 									IP_FORWARDING, &error);
811 		}
812 		if (sp != NULL) {
813 			/*
814 			 * Check security policy against packet attributes.
815 			 */
816 			error = ipsec_in_reject(sp, m);
817 			KEY_FREESP(&sp);
818 		} else {
819 			/* XXX error stat??? */
820 			error = EINVAL;
821 			DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/
822 		}
823 		splx(s);
824 		if (error)
825 			goto bad;
826 	}
827 #endif /* FAST_IPSEC */
828 
829 
830 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
831 	}
832 	return;
833  bad:
834 	m_freem(m);
835 }
836 
837 /*
838  * set/grab in6_ifaddr correspond to IPv6 destination address.
839  */
840 static struct m_tag *
841 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
842 {
843 	struct m_tag *mtag;
844 
845 	mtag = ip6_addaux(m);
846 	if (mtag != NULL) {
847 		struct ip6aux *ip6a;
848 
849 		ip6a = (struct ip6aux *)(mtag + 1);
850 		in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id);
851 		ip6a->ip6a_src = ia->ia_addr.sin6_addr;
852 		ip6a->ip6a_flags = ia->ia6_flags;
853 	}
854 	return mtag;	/* NULL if failed to set */
855 }
856 
857 const struct ip6aux *
858 ip6_getdstifaddr(struct mbuf *m)
859 {
860 	struct m_tag *mtag;
861 
862 	mtag = ip6_findaux(m);
863 	if (mtag != NULL)
864 		return (struct ip6aux *)(mtag + 1);
865 	else
866 		return NULL;
867 }
868 
869 /*
870  * Hop-by-Hop options header processing. If a valid jumbo payload option is
871  * included, the real payload length will be stored in plenp.
872  *
873  * rtalertp - XXX: should be stored more smart way
874  */
875 static int
876 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
877 	struct mbuf **mp, int *offp)
878 {
879 	struct mbuf *m = *mp;
880 	int off = *offp, hbhlen;
881 	struct ip6_hbh *hbh;
882 
883 	/* validation of the length of the header */
884 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
885 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
886 	if (hbh == NULL) {
887 		IP6_STATINC(IP6_STAT_TOOSHORT);
888 		return -1;
889 	}
890 	hbhlen = (hbh->ip6h_len + 1) << 3;
891 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
892 		hbhlen);
893 	if (hbh == NULL) {
894 		IP6_STATINC(IP6_STAT_TOOSHORT);
895 		return -1;
896 	}
897 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
898 	off += hbhlen;
899 	hbhlen -= sizeof(struct ip6_hbh);
900 
901 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
902 				hbhlen, rtalertp, plenp) < 0)
903 		return (-1);
904 
905 	*offp = off;
906 	*mp = m;
907 	return (0);
908 }
909 
910 /*
911  * Search header for all Hop-by-hop options and process each option.
912  * This function is separate from ip6_hopopts_input() in order to
913  * handle a case where the sending node itself process its hop-by-hop
914  * options header. In such a case, the function is called from ip6_output().
915  *
916  * The function assumes that hbh header is located right after the IPv6 header
917  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
918  * opthead + hbhlen is located in continuous memory region.
919  */
920 int
921 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
922 	u_int32_t *rtalertp, u_int32_t *plenp)
923 {
924 	struct ip6_hdr *ip6;
925 	int optlen = 0;
926 	u_int8_t *opt = opthead;
927 	u_int16_t rtalert_val;
928 	u_int32_t jumboplen;
929 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
930 
931 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
932 		switch (*opt) {
933 		case IP6OPT_PAD1:
934 			optlen = 1;
935 			break;
936 		case IP6OPT_PADN:
937 			if (hbhlen < IP6OPT_MINLEN) {
938 				IP6_STATINC(IP6_STAT_TOOSMALL);
939 				goto bad;
940 			}
941 			optlen = *(opt + 1) + 2;
942 			break;
943 		case IP6OPT_RTALERT:
944 			/* XXX may need check for alignment */
945 			if (hbhlen < IP6OPT_RTALERT_LEN) {
946 				IP6_STATINC(IP6_STAT_TOOSMALL);
947 				goto bad;
948 			}
949 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
950 				/* XXX stat */
951 				icmp6_error(m, ICMP6_PARAM_PROB,
952 				    ICMP6_PARAMPROB_HEADER,
953 				    erroff + opt + 1 - opthead);
954 				return (-1);
955 			}
956 			optlen = IP6OPT_RTALERT_LEN;
957 			bcopy((void *)(opt + 2), (void *)&rtalert_val, 2);
958 			*rtalertp = ntohs(rtalert_val);
959 			break;
960 		case IP6OPT_JUMBO:
961 			/* XXX may need check for alignment */
962 			if (hbhlen < IP6OPT_JUMBO_LEN) {
963 				IP6_STATINC(IP6_STAT_TOOSMALL);
964 				goto bad;
965 			}
966 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
967 				/* XXX stat */
968 				icmp6_error(m, ICMP6_PARAM_PROB,
969 				    ICMP6_PARAMPROB_HEADER,
970 				    erroff + opt + 1 - opthead);
971 				return (-1);
972 			}
973 			optlen = IP6OPT_JUMBO_LEN;
974 
975 			/*
976 			 * IPv6 packets that have non 0 payload length
977 			 * must not contain a jumbo payload option.
978 			 */
979 			ip6 = mtod(m, struct ip6_hdr *);
980 			if (ip6->ip6_plen) {
981 				IP6_STATINC(IP6_STAT_BADOPTIONS);
982 				icmp6_error(m, ICMP6_PARAM_PROB,
983 				    ICMP6_PARAMPROB_HEADER,
984 				    erroff + opt - opthead);
985 				return (-1);
986 			}
987 
988 			/*
989 			 * We may see jumbolen in unaligned location, so
990 			 * we'd need to perform bcopy().
991 			 */
992 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
993 			jumboplen = (u_int32_t)htonl(jumboplen);
994 
995 #if 1
996 			/*
997 			 * if there are multiple jumbo payload options,
998 			 * *plenp will be non-zero and the packet will be
999 			 * rejected.
1000 			 * the behavior may need some debate in ipngwg -
1001 			 * multiple options does not make sense, however,
1002 			 * there's no explicit mention in specification.
1003 			 */
1004 			if (*plenp != 0) {
1005 				IP6_STATINC(IP6_STAT_BADOPTIONS);
1006 				icmp6_error(m, ICMP6_PARAM_PROB,
1007 				    ICMP6_PARAMPROB_HEADER,
1008 				    erroff + opt + 2 - opthead);
1009 				return (-1);
1010 			}
1011 #endif
1012 
1013 			/*
1014 			 * jumbo payload length must be larger than 65535.
1015 			 */
1016 			if (jumboplen <= IPV6_MAXPACKET) {
1017 				IP6_STATINC(IP6_STAT_BADOPTIONS);
1018 				icmp6_error(m, ICMP6_PARAM_PROB,
1019 				    ICMP6_PARAMPROB_HEADER,
1020 				    erroff + opt + 2 - opthead);
1021 				return (-1);
1022 			}
1023 			*plenp = jumboplen;
1024 
1025 			break;
1026 		default:		/* unknown option */
1027 			if (hbhlen < IP6OPT_MINLEN) {
1028 				IP6_STATINC(IP6_STAT_TOOSMALL);
1029 				goto bad;
1030 			}
1031 			optlen = ip6_unknown_opt(opt, m,
1032 			    erroff + opt - opthead);
1033 			if (optlen == -1)
1034 				return (-1);
1035 			optlen += 2;
1036 			break;
1037 		}
1038 	}
1039 
1040 	return (0);
1041 
1042   bad:
1043 	m_freem(m);
1044 	return (-1);
1045 }
1046 
1047 /*
1048  * Unknown option processing.
1049  * The third argument `off' is the offset from the IPv6 header to the option,
1050  * which is necessary if the IPv6 header the and option header and IPv6 header
1051  * is not continuous in order to return an ICMPv6 error.
1052  */
1053 int
1054 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1055 {
1056 	struct ip6_hdr *ip6;
1057 
1058 	switch (IP6OPT_TYPE(*optp)) {
1059 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1060 		return ((int)*(optp + 1));
1061 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1062 		m_freem(m);
1063 		return (-1);
1064 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1065 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1066 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1067 		return (-1);
1068 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1069 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1070 		ip6 = mtod(m, struct ip6_hdr *);
1071 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1072 		    (m->m_flags & (M_BCAST|M_MCAST)))
1073 			m_freem(m);
1074 		else
1075 			icmp6_error(m, ICMP6_PARAM_PROB,
1076 				    ICMP6_PARAMPROB_OPTION, off);
1077 		return (-1);
1078 	}
1079 
1080 	m_freem(m);		/* XXX: NOTREACHED */
1081 	return (-1);
1082 }
1083 
1084 /*
1085  * Create the "control" list for this pcb.
1086  *
1087  * The routine will be called from upper layer handlers like tcp6_input().
1088  * Thus the routine assumes that the caller (tcp6_input) have already
1089  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1090  * very first mbuf on the mbuf chain.
1091  * We may want to add some infinite loop prevention or sanity checks for safety.
1092  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1093  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1094  */
1095 void
1096 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
1097 	struct ip6_hdr *ip6, struct mbuf *m)
1098 {
1099 #ifdef RFC2292
1100 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1101 #else
1102 #define IS2292(x, y)	(y)
1103 #endif
1104 
1105 #ifdef SO_TIMESTAMP
1106 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
1107 		struct timeval tv;
1108 
1109 		microtime(&tv);
1110 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1111 		    SCM_TIMESTAMP, SOL_SOCKET);
1112 		if (*mp)
1113 			mp = &(*mp)->m_next;
1114 	}
1115 #endif
1116 
1117 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
1118 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1119 		return;
1120 
1121 	/* RFC 2292 sec. 5 */
1122 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1123 		struct in6_pktinfo pi6;
1124 
1125 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1126 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1127 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
1128 		    m->m_pkthdr.rcvif->if_index : 0;
1129 		*mp = sbcreatecontrol((void *) &pi6,
1130 		    sizeof(struct in6_pktinfo),
1131 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1132 		if (*mp)
1133 			mp = &(*mp)->m_next;
1134 	}
1135 
1136 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1137 		int hlim = ip6->ip6_hlim & 0xff;
1138 
1139 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
1140 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1141 		if (*mp)
1142 			mp = &(*mp)->m_next;
1143 	}
1144 
1145 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1146 		u_int32_t flowinfo;
1147 		int tclass;
1148 
1149 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1150 		flowinfo >>= 20;
1151 
1152 		tclass = flowinfo & 0xff;
1153 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
1154 		    IPV6_TCLASS, IPPROTO_IPV6);
1155 
1156 		if (*mp)
1157 			mp = &(*mp)->m_next;
1158 	}
1159 
1160 	/*
1161 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1162 	 * privilege for the option (see ip6_ctloutput), but it might be too
1163 	 * strict, since there might be some hop-by-hop options which can be
1164 	 * returned to normal user.
1165 	 * See also RFC3542 section 8 (or RFC2292 section 6).
1166 	 */
1167 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1168 		/*
1169 		 * Check if a hop-by-hop options header is contatined in the
1170 		 * received packet, and if so, store the options as ancillary
1171 		 * data. Note that a hop-by-hop options header must be
1172 		 * just after the IPv6 header, which fact is assured through
1173 		 * the IPv6 input processing.
1174 		 */
1175 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1176 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1177 			struct ip6_hbh *hbh;
1178 			int hbhlen;
1179 			struct mbuf *ext;
1180 
1181 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1182 			    xip6->ip6_nxt);
1183 			if (ext == NULL) {
1184 				IP6_STATINC(IP6_STAT_TOOSHORT);
1185 				return;
1186 			}
1187 			hbh = mtod(ext, struct ip6_hbh *);
1188 			hbhlen = (hbh->ip6h_len + 1) << 3;
1189 			if (hbhlen != ext->m_len) {
1190 				m_freem(ext);
1191 				IP6_STATINC(IP6_STAT_TOOSHORT);
1192 				return;
1193 			}
1194 
1195 			/*
1196 			 * XXX: We copy whole the header even if a jumbo
1197 			 * payload option is included, which option is to
1198 			 * be removed before returning in the RFC 2292.
1199 			 * Note: this constraint is removed in RFC3542.
1200 			 */
1201 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
1202 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1203 			    IPPROTO_IPV6);
1204 			if (*mp)
1205 				mp = &(*mp)->m_next;
1206 			m_freem(ext);
1207 		}
1208 	}
1209 
1210 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1211 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1212 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1213 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1214 
1215 		/*
1216 		 * Search for destination options headers or routing
1217 		 * header(s) through the header chain, and stores each
1218 		 * header as ancillary data.
1219 		 * Note that the order of the headers remains in
1220 		 * the chain of ancillary data.
1221 		 */
1222 		for (;;) {	/* is explicit loop prevention necessary? */
1223 			struct ip6_ext *ip6e = NULL;
1224 			int elen;
1225 			struct mbuf *ext = NULL;
1226 
1227 			/*
1228 			 * if it is not an extension header, don't try to
1229 			 * pull it from the chain.
1230 			 */
1231 			switch (nxt) {
1232 			case IPPROTO_DSTOPTS:
1233 			case IPPROTO_ROUTING:
1234 			case IPPROTO_HOPOPTS:
1235 			case IPPROTO_AH: /* is it possible? */
1236 				break;
1237 			default:
1238 				goto loopend;
1239 			}
1240 
1241 			ext = ip6_pullexthdr(m, off, nxt);
1242 			if (ext == NULL) {
1243 				IP6_STATINC(IP6_STAT_TOOSHORT);
1244 				return;
1245 			}
1246 			ip6e = mtod(ext, struct ip6_ext *);
1247 			if (nxt == IPPROTO_AH)
1248 				elen = (ip6e->ip6e_len + 2) << 2;
1249 			else
1250 				elen = (ip6e->ip6e_len + 1) << 3;
1251 			if (elen != ext->m_len) {
1252 				m_freem(ext);
1253 				IP6_STATINC(IP6_STAT_TOOSHORT);
1254 				return;
1255 			}
1256 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1257 
1258 			switch (nxt) {
1259 			case IPPROTO_DSTOPTS:
1260 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
1261 					break;
1262 
1263 				*mp = sbcreatecontrol((void *)ip6e, elen,
1264 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1265 				    IPPROTO_IPV6);
1266 				if (*mp)
1267 					mp = &(*mp)->m_next;
1268 				break;
1269 
1270 			case IPPROTO_ROUTING:
1271 				if (!in6p->in6p_flags & IN6P_RTHDR)
1272 					break;
1273 
1274 				*mp = sbcreatecontrol((void *)ip6e, elen,
1275 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1276 				    IPPROTO_IPV6);
1277 				if (*mp)
1278 					mp = &(*mp)->m_next;
1279 				break;
1280 
1281 			case IPPROTO_HOPOPTS:
1282 			case IPPROTO_AH: /* is it possible? */
1283 				break;
1284 
1285 			default:
1286 				/*
1287 			 	 * other cases have been filtered in the above.
1288 				 * none will visit this case.  here we supply
1289 				 * the code just in case (nxt overwritten or
1290 				 * other cases).
1291 				 */
1292 				m_freem(ext);
1293 				goto loopend;
1294 
1295 			}
1296 
1297 			/* proceed with the next header. */
1298 			off += elen;
1299 			nxt = ip6e->ip6e_nxt;
1300 			ip6e = NULL;
1301 			m_freem(ext);
1302 			ext = NULL;
1303 		}
1304 	  loopend:
1305 	  	;
1306 	}
1307 }
1308 #undef IS2292
1309 
1310 
1311 void
1312 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
1313     uint32_t *mtu)
1314 {
1315 	struct socket *so;
1316 	struct mbuf *m_mtu;
1317 	struct ip6_mtuinfo mtuctl;
1318 
1319 	so = in6p->in6p_socket;
1320 
1321 	if (mtu == NULL)
1322 		return;
1323 
1324 #ifdef DIAGNOSTIC
1325 	if (so == NULL)		/* I believe this is impossible */
1326 		panic("ip6_notify_pmtu: socket is NULL");
1327 #endif
1328 
1329 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
1330 	mtuctl.ip6m_mtu = *mtu;
1331 	mtuctl.ip6m_addr = *dst;
1332 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
1333 		return;
1334 
1335 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
1336 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1337 		return;
1338 
1339 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
1340 	    == 0) {
1341 		m_freem(m_mtu);
1342 		/* XXX: should count statistics */
1343 	} else
1344 		sorwakeup(so);
1345 
1346 	return;
1347 }
1348 
1349 /*
1350  * pull single extension header from mbuf chain.  returns single mbuf that
1351  * contains the result, or NULL on error.
1352  */
1353 static struct mbuf *
1354 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1355 {
1356 	struct ip6_ext ip6e;
1357 	size_t elen;
1358 	struct mbuf *n;
1359 
1360 #ifdef DIAGNOSTIC
1361 	switch (nxt) {
1362 	case IPPROTO_DSTOPTS:
1363 	case IPPROTO_ROUTING:
1364 	case IPPROTO_HOPOPTS:
1365 	case IPPROTO_AH: /* is it possible? */
1366 		break;
1367 	default:
1368 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1369 	}
1370 #endif
1371 
1372 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1373 	if (nxt == IPPROTO_AH)
1374 		elen = (ip6e.ip6e_len + 2) << 2;
1375 	else
1376 		elen = (ip6e.ip6e_len + 1) << 3;
1377 
1378 	MGET(n, M_DONTWAIT, MT_DATA);
1379 	if (n && elen >= MLEN) {
1380 		MCLGET(n, M_DONTWAIT);
1381 		if ((n->m_flags & M_EXT) == 0) {
1382 			m_free(n);
1383 			n = NULL;
1384 		}
1385 	}
1386 	if (!n)
1387 		return NULL;
1388 
1389 	n->m_len = 0;
1390 	if (elen >= M_TRAILINGSPACE(n)) {
1391 		m_free(n);
1392 		return NULL;
1393 	}
1394 
1395 	m_copydata(m, off, elen, mtod(n, void *));
1396 	n->m_len = elen;
1397 	return n;
1398 }
1399 
1400 /*
1401  * Get pointer to the previous header followed by the header
1402  * currently processed.
1403  * XXX: This function supposes that
1404  *	M includes all headers,
1405  *	the next header field and the header length field of each header
1406  *	are valid, and
1407  *	the sum of each header length equals to OFF.
1408  * Because of these assumptions, this function must be called very
1409  * carefully. Moreover, it will not be used in the near future when
1410  * we develop `neater' mechanism to process extension headers.
1411  */
1412 u_int8_t *
1413 ip6_get_prevhdr(struct mbuf *m, int off)
1414 {
1415 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1416 
1417 	if (off == sizeof(struct ip6_hdr))
1418 		return (&ip6->ip6_nxt);
1419 	else {
1420 		int len, nxt;
1421 		struct ip6_ext *ip6e = NULL;
1422 
1423 		nxt = ip6->ip6_nxt;
1424 		len = sizeof(struct ip6_hdr);
1425 		while (len < off) {
1426 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
1427 
1428 			switch (nxt) {
1429 			case IPPROTO_FRAGMENT:
1430 				len += sizeof(struct ip6_frag);
1431 				break;
1432 			case IPPROTO_AH:
1433 				len += (ip6e->ip6e_len + 2) << 2;
1434 				break;
1435 			default:
1436 				len += (ip6e->ip6e_len + 1) << 3;
1437 				break;
1438 			}
1439 			nxt = ip6e->ip6e_nxt;
1440 		}
1441 		if (ip6e)
1442 			return (&ip6e->ip6e_nxt);
1443 		else
1444 			return NULL;
1445 	}
1446 }
1447 
1448 /*
1449  * get next header offset.  m will be retained.
1450  */
1451 int
1452 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1453 {
1454 	struct ip6_hdr ip6;
1455 	struct ip6_ext ip6e;
1456 	struct ip6_frag fh;
1457 
1458 	/* just in case */
1459 	if (m == NULL)
1460 		panic("ip6_nexthdr: m == NULL");
1461 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1462 		return -1;
1463 
1464 	switch (proto) {
1465 	case IPPROTO_IPV6:
1466 		/* do not chase beyond intermediate IPv6 headers */
1467 		if (off != 0)
1468 			return -1;
1469 		if (m->m_pkthdr.len < off + sizeof(ip6))
1470 			return -1;
1471 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
1472 		if (nxtp)
1473 			*nxtp = ip6.ip6_nxt;
1474 		off += sizeof(ip6);
1475 		return off;
1476 
1477 	case IPPROTO_FRAGMENT:
1478 		/*
1479 		 * terminate parsing if it is not the first fragment,
1480 		 * it does not make sense to parse through it.
1481 		 */
1482 		if (m->m_pkthdr.len < off + sizeof(fh))
1483 			return -1;
1484 		m_copydata(m, off, sizeof(fh), (void *)&fh);
1485 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1486 			return -1;
1487 		if (nxtp)
1488 			*nxtp = fh.ip6f_nxt;
1489 		off += sizeof(struct ip6_frag);
1490 		return off;
1491 
1492 	case IPPROTO_AH:
1493 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1494 			return -1;
1495 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1496 		if (nxtp)
1497 			*nxtp = ip6e.ip6e_nxt;
1498 		off += (ip6e.ip6e_len + 2) << 2;
1499 		if (m->m_pkthdr.len < off)
1500 			return -1;
1501 		return off;
1502 
1503 	case IPPROTO_HOPOPTS:
1504 	case IPPROTO_ROUTING:
1505 	case IPPROTO_DSTOPTS:
1506 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1507 			return -1;
1508 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1509 		if (nxtp)
1510 			*nxtp = ip6e.ip6e_nxt;
1511 		off += (ip6e.ip6e_len + 1) << 3;
1512 		if (m->m_pkthdr.len < off)
1513 			return -1;
1514 		return off;
1515 
1516 	case IPPROTO_NONE:
1517 	case IPPROTO_ESP:
1518 	case IPPROTO_IPCOMP:
1519 		/* give up */
1520 		return -1;
1521 
1522 	default:
1523 		return -1;
1524 	}
1525 }
1526 
1527 /*
1528  * get offset for the last header in the chain.  m will be kept untainted.
1529  */
1530 int
1531 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1532 {
1533 	int newoff;
1534 	int nxt;
1535 
1536 	if (!nxtp) {
1537 		nxt = -1;
1538 		nxtp = &nxt;
1539 	}
1540 	for (;;) {
1541 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1542 		if (newoff < 0)
1543 			return off;
1544 		else if (newoff < off)
1545 			return -1;	/* invalid */
1546 		else if (newoff == off)
1547 			return newoff;
1548 
1549 		off = newoff;
1550 		proto = *nxtp;
1551 	}
1552 }
1553 
1554 struct m_tag *
1555 ip6_addaux(struct mbuf *m)
1556 {
1557 	struct m_tag *mtag;
1558 
1559 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1560 	if (!mtag) {
1561 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1562 		    M_NOWAIT);
1563 		if (mtag) {
1564 			m_tag_prepend(m, mtag);
1565 			bzero(mtag + 1, sizeof(struct ip6aux));
1566 		}
1567 	}
1568 	return mtag;
1569 }
1570 
1571 struct m_tag *
1572 ip6_findaux(struct mbuf *m)
1573 {
1574 	struct m_tag *mtag;
1575 
1576 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1577 	return mtag;
1578 }
1579 
1580 void
1581 ip6_delaux(struct mbuf *m)
1582 {
1583 	struct m_tag *mtag;
1584 
1585 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1586 	if (mtag)
1587 		m_tag_delete(m, mtag);
1588 }
1589 
1590 #ifdef GATEWAY
1591 /*
1592  * sysctl helper routine for net.inet.ip6.maxflows. Since
1593  * we could reduce this value, call ip6flow_reap();
1594  */
1595 static int
1596 sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS)
1597 {
1598 	int error;
1599 
1600 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1601 	if (error || newp == NULL)
1602 		return (error);
1603 
1604 	mutex_enter(softnet_lock);
1605 	KERNEL_LOCK(1, NULL);
1606 
1607 	ip6flow_reap(0);
1608 
1609 	KERNEL_UNLOCK_ONE(NULL);
1610 	mutex_exit(softnet_lock);
1611 
1612 	return (0);
1613 }
1614 
1615 static int
1616 sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS)
1617 {
1618 	int error, tmp;
1619 	struct sysctlnode node;
1620 
1621 	node = *rnode;
1622 	tmp = ip6_hashsize;
1623 	node.sysctl_data = &tmp;
1624 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1625 	if (error || newp == NULL)
1626 		return (error);
1627 
1628 	if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
1629 		/*
1630 		 * Can only fail due to malloc()
1631 		 */
1632 		mutex_enter(softnet_lock);
1633 		KERNEL_LOCK(1, NULL);
1634 
1635 		error = ip6flow_invalidate_all(tmp);
1636 
1637 		KERNEL_UNLOCK_ONE(NULL);
1638 		mutex_exit(softnet_lock);
1639 	} else {
1640 		/*
1641 		 * EINVAL if not a power of 2
1642 		 */
1643 		error = EINVAL;
1644 	}
1645 
1646 	return error;
1647 }
1648 #endif /* GATEWAY */
1649 
1650 /*
1651  * System control for IP6
1652  */
1653 
1654 const u_char inet6ctlerrmap[PRC_NCMDS] = {
1655 	0,		0,		0,		0,
1656 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1657 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1658 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1659 	0,		0,		0,		0,
1660 	ENOPROTOOPT
1661 };
1662 
1663 static int
1664 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
1665 {
1666 
1667 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
1668 }
1669 
1670 SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup")
1671 {
1672 #ifdef RFC2292
1673 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1674 #else
1675 #define IS2292(x, y)	(y)
1676 #endif
1677 
1678 	sysctl_createv(clog, 0, NULL, NULL,
1679 		       CTLFLAG_PERMANENT,
1680 		       CTLTYPE_NODE, "net", NULL,
1681 		       NULL, 0, NULL, 0,
1682 		       CTL_NET, CTL_EOL);
1683 	sysctl_createv(clog, 0, NULL, NULL,
1684 		       CTLFLAG_PERMANENT,
1685 		       CTLTYPE_NODE, "inet6",
1686 		       SYSCTL_DESCR("PF_INET6 related settings"),
1687 		       NULL, 0, NULL, 0,
1688 		       CTL_NET, PF_INET6, CTL_EOL);
1689 	sysctl_createv(clog, 0, NULL, NULL,
1690 		       CTLFLAG_PERMANENT,
1691 		       CTLTYPE_NODE, "ip6",
1692 		       SYSCTL_DESCR("IPv6 related settings"),
1693 		       NULL, 0, NULL, 0,
1694 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1695 
1696 	sysctl_createv(clog, 0, NULL, NULL,
1697 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1698 		       CTLTYPE_INT, "forwarding",
1699 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1700 		       NULL, 0, &ip6_forwarding, 0,
1701 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1702 		       IPV6CTL_FORWARDING, CTL_EOL);
1703 	sysctl_createv(clog, 0, NULL, NULL,
1704 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1705 		       CTLTYPE_INT, "redirect",
1706 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1707 		       NULL, 0, &ip6_sendredirects, 0,
1708 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1709 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1710 	sysctl_createv(clog, 0, NULL, NULL,
1711 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1712 		       CTLTYPE_INT, "hlim",
1713 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1714 		       NULL, 0, &ip6_defhlim, 0,
1715 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1716 		       IPV6CTL_DEFHLIM, CTL_EOL);
1717 #ifdef notyet
1718 	sysctl_createv(clog, 0, NULL, NULL,
1719 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1720 		       CTLTYPE_INT, "mtu", NULL,
1721 		       NULL, 0, &, 0,
1722 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1723 		       IPV6CTL_DEFMTU, CTL_EOL);
1724 #endif
1725 #ifdef __no_idea__
1726 	sysctl_createv(clog, 0, NULL, NULL,
1727 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1728 		       CTLTYPE_INT, "forwsrcrt", NULL,
1729 		       NULL, 0, &?, 0,
1730 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1731 		       IPV6CTL_FORWSRCRT, CTL_EOL);
1732 	sysctl_createv(clog, 0, NULL, NULL,
1733 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1734 		       CTLTYPE_STRUCT, "mrtstats", NULL,
1735 		       NULL, 0, &?, sizeof(?),
1736 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1737 		       IPV6CTL_MRTSTATS, CTL_EOL);
1738 	sysctl_createv(clog, 0, NULL, NULL,
1739 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1740 		       CTLTYPE_?, "mrtproto", NULL,
1741 		       NULL, 0, &?, sizeof(?),
1742 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1743 		       IPV6CTL_MRTPROTO, CTL_EOL);
1744 #endif
1745 	sysctl_createv(clog, 0, NULL, NULL,
1746 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1747 		       CTLTYPE_INT, "maxfragpackets",
1748 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1749 				    "for reassembly"),
1750 		       NULL, 0, &ip6_maxfragpackets, 0,
1751 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1752 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1753 #ifdef __no_idea__
1754 	sysctl_createv(clog, 0, NULL, NULL,
1755 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1756 		       CTLTYPE_INT, "sourcecheck", NULL,
1757 		       NULL, 0, &?, 0,
1758 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1759 		       IPV6CTL_SOURCECHECK, CTL_EOL);
1760 	sysctl_createv(clog, 0, NULL, NULL,
1761 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1762 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
1763 		       NULL, 0, &?, 0,
1764 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1765 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
1766 #endif
1767 	sysctl_createv(clog, 0, NULL, NULL,
1768 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1769 		       CTLTYPE_INT, "accept_rtadv",
1770 		       SYSCTL_DESCR("Accept router advertisements"),
1771 		       NULL, 0, &ip6_accept_rtadv, 0,
1772 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1773 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1774 	sysctl_createv(clog, 0, NULL, NULL,
1775 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1776 		       CTLTYPE_INT, "keepfaith",
1777 		       SYSCTL_DESCR("Activate faith interface"),
1778 		       NULL, 0, &ip6_keepfaith, 0,
1779 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1780 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1781 	sysctl_createv(clog, 0, NULL, NULL,
1782 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1783 		       CTLTYPE_INT, "log_interval",
1784 		       SYSCTL_DESCR("Minumum interval between logging "
1785 				    "unroutable packets"),
1786 		       NULL, 0, &ip6_log_interval, 0,
1787 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1788 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1789 	sysctl_createv(clog, 0, NULL, NULL,
1790 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1791 		       CTLTYPE_INT, "hdrnestlimit",
1792 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1793 		       NULL, 0, &ip6_hdrnestlimit, 0,
1794 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1795 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1796 	sysctl_createv(clog, 0, NULL, NULL,
1797 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1798 		       CTLTYPE_INT, "dad_count",
1799 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1800 				    "probes to send"),
1801 		       NULL, 0, &ip6_dad_count, 0,
1802 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1803 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1804 	sysctl_createv(clog, 0, NULL, NULL,
1805 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1806 		       CTLTYPE_INT, "auto_flowlabel",
1807 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1808 		       NULL, 0, &ip6_auto_flowlabel, 0,
1809 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1810 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1811 	sysctl_createv(clog, 0, NULL, NULL,
1812 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1813 		       CTLTYPE_INT, "defmcasthlim",
1814 		       SYSCTL_DESCR("Default multicast hop limit"),
1815 		       NULL, 0, &ip6_defmcasthlim, 0,
1816 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1817 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1818 #if NGIF > 0
1819 	sysctl_createv(clog, 0, NULL, NULL,
1820 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1821 		       CTLTYPE_INT, "gifhlim",
1822 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
1823 		       NULL, 0, &ip6_gif_hlim, 0,
1824 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1825 		       IPV6CTL_GIF_HLIM, CTL_EOL);
1826 #endif /* NGIF */
1827 	sysctl_createv(clog, 0, NULL, NULL,
1828 		       CTLFLAG_PERMANENT,
1829 		       CTLTYPE_STRING, "kame_version",
1830 		       SYSCTL_DESCR("KAME Version"),
1831 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1832 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1833 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1834 	sysctl_createv(clog, 0, NULL, NULL,
1835 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1836 		       CTLTYPE_INT, "use_deprecated",
1837 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1838 				    "source addresses"),
1839 		       NULL, 0, &ip6_use_deprecated, 0,
1840 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1841 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1842 	sysctl_createv(clog, 0, NULL, NULL,
1843 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1844 		       CTLTYPE_INT, "rr_prune", NULL,
1845 		       NULL, 0, &ip6_rr_prune, 0,
1846 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1847 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1848 	sysctl_createv(clog, 0, NULL, NULL,
1849 		       CTLFLAG_PERMANENT
1850 #ifndef INET6_BINDV6ONLY
1851 		       |CTLFLAG_READWRITE,
1852 #endif
1853 		       CTLTYPE_INT, "v6only",
1854 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1855 				    "to PF_INET sockets"),
1856 		       NULL, 0, &ip6_v6only, 0,
1857 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1858 		       IPV6CTL_V6ONLY, CTL_EOL);
1859 	sysctl_createv(clog, 0, NULL, NULL,
1860 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1861 		       CTLTYPE_INT, "anonportmin",
1862 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1863 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1864 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1865 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1866 	sysctl_createv(clog, 0, NULL, NULL,
1867 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1868 		       CTLTYPE_INT, "anonportmax",
1869 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1870 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1871 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1872 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1873 #ifndef IPNOPRIVPORTS
1874 	sysctl_createv(clog, 0, NULL, NULL,
1875 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1876 		       CTLTYPE_INT, "lowportmin",
1877 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1878 				    "to assign"),
1879 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1880 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1881 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1882 	sysctl_createv(clog, 0, NULL, NULL,
1883 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1884 		       CTLTYPE_INT, "lowportmax",
1885 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1886 				    "to assign"),
1887 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1888 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1889 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1890 #endif /* IPNOPRIVPORTS */
1891 	sysctl_createv(clog, 0, NULL, NULL,
1892 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1893 		       CTLTYPE_INT, "use_tempaddr",
1894 		       SYSCTL_DESCR("Use temporary address"),
1895 		       NULL, 0, &ip6_use_tempaddr, 0,
1896 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1897 		       CTL_CREATE, CTL_EOL);
1898 	sysctl_createv(clog, 0, NULL, NULL,
1899 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1900 		       CTLTYPE_INT, "temppltime",
1901 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
1902 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
1903 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1904 		       CTL_CREATE, CTL_EOL);
1905 	sysctl_createv(clog, 0, NULL, NULL,
1906 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1907 		       CTLTYPE_INT, "tempvltime",
1908 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
1909 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
1910 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1911 		       CTL_CREATE, CTL_EOL);
1912 	sysctl_createv(clog, 0, NULL, NULL,
1913 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1914 		       CTLTYPE_INT, "maxfrags",
1915 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1916 		       NULL, 0, &ip6_maxfrags, 0,
1917 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1918 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1919 	sysctl_createv(clog, 0, NULL, NULL,
1920 		       CTLFLAG_PERMANENT,
1921 		       CTLTYPE_STRUCT, "stats",
1922 		       SYSCTL_DESCR("IPv6 statistics"),
1923 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
1924 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1925 		       IPV6CTL_STATS, CTL_EOL);
1926 	sysctl_createv(clog, 0, NULL, NULL,
1927 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1928 		       CTLTYPE_INT, "use_defaultzone",
1929 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1930 		       NULL, 0, &ip6_use_defzone, 0,
1931 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1932 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1933 	sysctl_createv(clog, 0, NULL, NULL,
1934 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1935 		       CTLTYPE_INT, "mcast_pmtu",
1936 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
1937 		       NULL, 0, &ip6_mcast_pmtu, 0,
1938 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1939 		       CTL_CREATE, CTL_EOL);
1940 #ifdef GATEWAY
1941 	sysctl_createv(clog, 0, NULL, NULL,
1942 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1943 			CTLTYPE_INT, "maxflows",
1944 			SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"),
1945 			sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 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, "hashsize",
1951 			SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"),
1952 			sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0,
1953 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1954 			CTL_CREATE, CTL_EOL);
1955 #endif
1956 }
1957 
1958 void
1959 ip6_statinc(u_int stat)
1960 {
1961 
1962 	KASSERT(stat < IP6_NSTATS);
1963 	IP6_STATINC(stat);
1964 }
1965