xref: /netbsd-src/sys/netinet6/ip6_input.c (revision ba65fde2d7fefa7d39838fa5fa855e62bd606b5e)
1 /*	$NetBSD: ip6_input.c,v 1.141 2012/11/29 02:07:20 christos 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 1.141 2012/11/29 02:07:20 christos Exp $");
66 
67 #include "opt_gateway.h"
68 #include "opt_inet.h"
69 #include "opt_inet6.h"
70 #include "opt_ipsec.h"
71 #include "opt_pfil_hooks.h"
72 #include "opt_compat_netbsd.h"
73 
74 #include <sys/param.h>
75 #include <sys/systm.h>
76 #include <sys/malloc.h>
77 #include <sys/mbuf.h>
78 #include <sys/domain.h>
79 #include <sys/protosw.h>
80 #include <sys/socket.h>
81 #include <sys/socketvar.h>
82 #include <sys/errno.h>
83 #include <sys/time.h>
84 #include <sys/kernel.h>
85 #include <sys/syslog.h>
86 #include <sys/proc.h>
87 #include <sys/sysctl.h>
88 #include <sys/cprng.h>
89 
90 #include <net/if.h>
91 #include <net/if_types.h>
92 #include <net/if_dl.h>
93 #include <net/route.h>
94 #include <net/netisr.h>
95 #ifdef PFIL_HOOKS
96 #include <net/pfil.h>
97 #endif
98 
99 #include <netinet/in.h>
100 #include <netinet/in_systm.h>
101 #ifdef INET
102 #include <netinet/ip.h>
103 #include <netinet/ip_icmp.h>
104 #endif /* INET */
105 #include <netinet/ip6.h>
106 #include <netinet/portalgo.h>
107 #include <netinet6/in6_var.h>
108 #include <netinet6/ip6_var.h>
109 #include <netinet6/ip6_private.h>
110 #include <netinet6/in6_pcb.h>
111 #include <netinet/icmp6.h>
112 #include <netinet6/scope6_var.h>
113 #include <netinet6/in6_ifattach.h>
114 #include <netinet6/nd6.h>
115 
116 #ifdef FAST_IPSEC
117 #include <netipsec/ipsec.h>
118 #include <netipsec/ipsec6.h>
119 #include <netipsec/key.h>
120 #endif /* FAST_IPSEC */
121 
122 #ifdef COMPAT_50
123 #include <compat/sys/time.h>
124 #include <compat/sys/socket.h>
125 #endif
126 
127 #include <netinet6/ip6protosw.h>
128 
129 #include "faith.h"
130 #include "gif.h"
131 
132 #if NGIF > 0
133 #include <netinet6/in6_gif.h>
134 #endif
135 
136 #include <net/net_osdep.h>
137 
138 extern struct domain inet6domain;
139 
140 u_char ip6_protox[IPPROTO_MAX];
141 static int ip6qmaxlen = IFQ_MAXLEN;
142 struct in6_ifaddr *in6_ifaddr;
143 struct ifqueue ip6intrq;
144 
145 extern callout_t in6_tmpaddrtimer_ch;
146 
147 int ip6_forward_srcrt;			/* XXX */
148 int ip6_sourcecheck;			/* XXX */
149 int ip6_sourcecheck_interval;		/* XXX */
150 
151 #ifdef PFIL_HOOKS
152 struct pfil_head inet6_pfil_hook;
153 #endif
154 
155 percpu_t *ip6stat_percpu;
156 
157 static void ip6_init2(void *);
158 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
159 
160 static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *,
161 	u_int32_t *);
162 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
163 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
164 
165 /*
166  * IP6 initialization: fill in IP6 protocol switch table.
167  * All protocols not implemented in kernel go to raw IP6 protocol handler.
168  */
169 void
170 ip6_init(void)
171 {
172 	const struct ip6protosw *pr;
173 	int i;
174 
175 	sysctl_net_inet6_ip6_setup(NULL);
176 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
177 	if (pr == 0)
178 		panic("ip6_init");
179 	for (i = 0; i < IPPROTO_MAX; i++)
180 		ip6_protox[i] = pr - inet6sw;
181 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
182 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
183 		if (pr->pr_domain->dom_family == PF_INET6 &&
184 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
185 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
186 	ip6intrq.ifq_maxlen = ip6qmaxlen;
187 	scope6_init();
188 	addrsel_policy_init();
189 	nd6_init();
190 	frag6_init();
191 	ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR;
192 
193 	ip6_init2(NULL);
194 #ifdef GATEWAY
195 	ip6flow_init(ip6_hashsize);
196 #endif
197 
198 #ifdef PFIL_HOOKS
199 	/* Register our Packet Filter hook. */
200 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
201 	inet6_pfil_hook.ph_af   = AF_INET6;
202 	i = pfil_head_register(&inet6_pfil_hook);
203 	if (i != 0)
204 		printf("ip6_init: WARNING: unable to register pfil hook, "
205 		    "error %d\n", i);
206 #endif /* PFIL_HOOKS */
207 
208 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
209 }
210 
211 static void
212 ip6_init2(void *dummy)
213 {
214 
215 	/* nd6_timer_init */
216 	callout_init(&nd6_timer_ch, CALLOUT_MPSAFE);
217 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
218 
219 	/* timer for regeneranation of temporary addresses randomize ID */
220 	callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE);
221 	callout_reset(&in6_tmpaddrtimer_ch,
222 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
223 		       ip6_temp_regen_advance) * hz,
224 		      in6_tmpaddrtimer, NULL);
225 }
226 
227 /*
228  * IP6 input interrupt handling. Just pass the packet to ip6_input.
229  */
230 void
231 ip6intr(void)
232 {
233 	int s;
234 	struct mbuf *m;
235 
236 	mutex_enter(softnet_lock);
237 	KERNEL_LOCK(1, NULL);
238 	for (;;) {
239 		s = splnet();
240 		IF_DEQUEUE(&ip6intrq, m);
241 		splx(s);
242 		if (m == 0)
243 			break;
244 		/* drop the packet if IPv6 operation is disabled on the IF */
245 		if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
246 			m_freem(m);
247 			break;
248 		}
249 		ip6_input(m);
250 	}
251 	KERNEL_UNLOCK_ONE(NULL);
252 	mutex_exit(softnet_lock);
253 }
254 
255 extern struct	route ip6_forward_rt;
256 
257 void
258 ip6_input(struct mbuf *m)
259 {
260 	struct ip6_hdr *ip6;
261 	int hit, off = sizeof(struct ip6_hdr), nest;
262 	u_int32_t plen;
263 	u_int32_t rtalert = ~0;
264 	int nxt, ours = 0, rh_present = 0;
265 	struct ifnet *deliverifp = NULL;
266 	int srcrt = 0;
267 	const struct rtentry *rt;
268 	union {
269 		struct sockaddr		dst;
270 		struct sockaddr_in6	dst6;
271 	} u;
272 #ifdef FAST_IPSEC
273 	struct m_tag *mtag;
274 	struct tdb_ident *tdbi;
275 	struct secpolicy *sp;
276 	int s, error;
277 #endif
278 
279 	/*
280 	 * make sure we don't have onion peering information into m_tag.
281 	 */
282 	ip6_delaux(m);
283 
284 	/*
285 	 * mbuf statistics
286 	 */
287 	if (m->m_flags & M_EXT) {
288 		if (m->m_next)
289 			IP6_STATINC(IP6_STAT_MEXT2M);
290 		else
291 			IP6_STATINC(IP6_STAT_MEXT1);
292 	} else {
293 #define M2MMAX	32
294 		if (m->m_next) {
295 			if (m->m_flags & M_LOOP) {
296 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
297 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX) {
298 				IP6_STATINC(IP6_STAT_M2M +
299 					    m->m_pkthdr.rcvif->if_index);
300 			} else
301 				IP6_STATINC(IP6_STAT_M2M);
302 		} else
303 			IP6_STATINC(IP6_STAT_M1);
304 #undef M2MMAX
305 	}
306 
307 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
308 	IP6_STATINC(IP6_STAT_TOTAL);
309 
310 	/*
311 	 * If the IPv6 header is not aligned, slurp it up into a new
312 	 * mbuf with space for link headers, in the event we forward
313 	 * it.  Otherwise, if it is aligned, make sure the entire base
314 	 * IPv6 header is in the first mbuf of the chain.
315 	 */
316 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
317 		struct ifnet *inifp = m->m_pkthdr.rcvif;
318 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
319 				  (max_linkhdr + 3) & ~3)) == NULL) {
320 			/* XXXJRT new stat, please */
321 			IP6_STATINC(IP6_STAT_TOOSMALL);
322 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
323 			return;
324 		}
325 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
326 		struct ifnet *inifp = m->m_pkthdr.rcvif;
327 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
328 			IP6_STATINC(IP6_STAT_TOOSMALL);
329 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
330 			return;
331 		}
332 	}
333 
334 	ip6 = mtod(m, struct ip6_hdr *);
335 
336 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
337 		IP6_STATINC(IP6_STAT_BADVERS);
338 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
339 		goto bad;
340 	}
341 
342 	/*
343 	 * Assume that we can create a fast-forward IP flow entry
344 	 * based on this packet.
345 	 */
346 	m->m_flags |= M_CANFASTFWD;
347 
348 #ifdef PFIL_HOOKS
349 	/*
350 	 * Run through list of hooks for input packets.  If there are any
351 	 * filters which require that additional packets in the flow are
352 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
353 	 * Note that filters must _never_ set this flag, as another filter
354 	 * in the list may have previously cleared it.
355 	 */
356 	/*
357 	 * let ipfilter look at packet on the wire,
358 	 * not the decapsulated packet.
359 	 */
360 #if defined(FAST_IPSEC)
361 	if (!ipsec_indone(m))
362 #else
363 	if (1)
364 #endif
365 	{
366 		struct in6_addr odst;
367 
368 		odst = ip6->ip6_dst;
369 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
370 				   PFIL_IN) != 0)
371 			return;
372 		if (m == NULL)
373 			return;
374 		ip6 = mtod(m, struct ip6_hdr *);
375 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
376 	}
377 #endif /* PFIL_HOOKS */
378 
379 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
380 
381 #ifdef ALTQ
382 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
383 		/* packet is dropped by traffic conditioner */
384 		return;
385 	}
386 #endif
387 
388 	/*
389 	 * Check against address spoofing/corruption.
390 	 */
391 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
392 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
393 		/*
394 		 * XXX: "badscope" is not very suitable for a multicast source.
395 		 */
396 		IP6_STATINC(IP6_STAT_BADSCOPE);
397 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
398 		goto bad;
399 	}
400 	/*
401 	 * The following check is not documented in specs.  A malicious
402 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
403 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
404 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
405 	 *
406 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
407 	 * support IPv4-less kernel compilation, we cannot support SIIT
408 	 * environment at all.  So, it makes more sense for us to reject any
409 	 * malicious packets for non-SIIT environment, than try to do a
410 	 * partial support for SIIT environment.
411 	 */
412 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
413 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
414 		IP6_STATINC(IP6_STAT_BADSCOPE);
415 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
416 		goto bad;
417 	}
418 #if 0
419 	/*
420 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
421 	 *
422 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
423 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
424 	 * is revised to forbid relaying case.
425 	 */
426 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
427 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
428 		IP6_STATINC(IP6_STAT_BADSCOPE);
429 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
430 		goto bad;
431 	}
432 #endif
433 
434 	/*
435 	 * Disambiguate address scope zones (if there is ambiguity).
436 	 * We first make sure that the original source or destination address
437 	 * is not in our internal form for scoped addresses.  Such addresses
438 	 * are not necessarily invalid spec-wise, but we cannot accept them due
439 	 * to the usage conflict.
440 	 * in6_setscope() then also checks and rejects the cases where src or
441 	 * dst are the loopback address and the receiving interface
442 	 * is not loopback.
443 	 */
444 	if (__predict_false(
445 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
446 		goto bad;
447 	ip6 = mtod(m, struct ip6_hdr *);
448 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
449 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
450 		goto bad;
451 	}
452 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
453 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
454 		IP6_STATINC(IP6_STAT_BADSCOPE);
455 		goto bad;
456 	}
457 
458 	/*
459 	 * Multicast check
460 	 */
461 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
462 	  	struct	in6_multi *in6m = 0;
463 
464 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
465 		/*
466 		 * See if we belong to the destination multicast group on the
467 		 * arrival interface.
468 		 */
469 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
470 		if (in6m)
471 			ours = 1;
472 		else if (!ip6_mrouter) {
473 			uint64_t *ip6s = IP6_STAT_GETREF();
474 			ip6s[IP6_STAT_NOTMEMBER]++;
475 			ip6s[IP6_STAT_CANTFORWARD]++;
476 			IP6_STAT_PUTREF();
477 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
478 			goto bad;
479 		}
480 		deliverifp = m->m_pkthdr.rcvif;
481 		goto hbhcheck;
482 	}
483 
484 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
485 
486 	/*
487 	 *  Unicast check
488 	 */
489 	rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit);
490 	if (hit)
491 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
492 	else
493 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
494 
495 #define rt6_getkey(__rt) satocsin6(rt_getkey(__rt))
496 
497 	/*
498 	 * Accept the packet if the forwarding interface to the destination
499 	 * according to the routing table is the loopback interface,
500 	 * unless the associated route has a gateway.
501 	 * Note that this approach causes to accept a packet if there is a
502 	 * route to the loopback interface for the destination of the packet.
503 	 * But we think it's even useful in some situations, e.g. when using
504 	 * a special daemon which wants to intercept the packet.
505 	 */
506 	if (rt != NULL &&
507 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
508 	    !(rt->rt_flags & RTF_CLONED) &&
509 #if 0
510 	    /*
511 	     * The check below is redundant since the comparison of
512 	     * the destination and the key of the rtentry has
513 	     * already done through looking up the routing table.
514 	     */
515 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(rt)->sin6_addr) &&
516 #endif
517 	    rt->rt_ifp->if_type == IFT_LOOP) {
518 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
519 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
520 			m->m_flags |= M_ANYCAST6;
521 		/*
522 		 * packets to a tentative, duplicated, or somehow invalid
523 		 * address must not be accepted.
524 		 */
525 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
526 			/* this address is ready */
527 			ours = 1;
528 			deliverifp = ia6->ia_ifp;	/* correct? */
529 			goto hbhcheck;
530 		} else {
531 			/* address is not ready, so discard the packet. */
532 			nd6log((LOG_INFO,
533 			    "ip6_input: packet to an unready address %s->%s\n",
534 			    ip6_sprintf(&ip6->ip6_src),
535 			    ip6_sprintf(&ip6->ip6_dst)));
536 
537 			goto bad;
538 		}
539 	}
540 
541 	/*
542 	 * FAITH (Firewall Aided Internet Translator)
543 	 */
544 #if defined(NFAITH) && 0 < NFAITH
545 	if (ip6_keepfaith) {
546 		if (rt != NULL && rt->rt_ifp != NULL &&
547 		    rt->rt_ifp->if_type == IFT_FAITH) {
548 			/* XXX do we need more sanity checks? */
549 			ours = 1;
550 			deliverifp = rt->rt_ifp; /* faith */
551 			goto hbhcheck;
552 		}
553 	}
554 #endif
555 
556 #if 0
557     {
558 	/*
559 	 * Last resort: check in6_ifaddr for incoming interface.
560 	 * The code is here until I update the "goto ours hack" code above
561 	 * working right.
562 	 */
563 	struct ifaddr *ifa;
564 	IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
565 		if (ifa->ifa_addr == NULL)
566 			continue;	/* just for safety */
567 		if (ifa->ifa_addr->sa_family != AF_INET6)
568 			continue;
569 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
570 			ours = 1;
571 			deliverifp = ifa->ifa_ifp;
572 			goto hbhcheck;
573 		}
574 	}
575     }
576 #endif
577 
578 	/*
579 	 * Now there is no reason to process the packet if it's not our own
580 	 * and we're not a router.
581 	 */
582 	if (!ip6_forwarding) {
583 		IP6_STATINC(IP6_STAT_CANTFORWARD);
584 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
585 		goto bad;
586 	}
587 
588   hbhcheck:
589 	/*
590 	 * record address information into m_tag, if we don't have one yet.
591 	 * note that we are unable to record it, if the address is not listed
592 	 * as our interface address (e.g. multicast addresses, addresses
593 	 * within FAITH prefixes and such).
594 	 */
595 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
596 		struct in6_ifaddr *ia6;
597 
598 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
599 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
600 			/*
601 			 * XXX maybe we should drop the packet here,
602 			 * as we could not provide enough information
603 			 * to the upper layers.
604 			 */
605 		}
606 	}
607 
608 	/*
609 	 * Process Hop-by-Hop options header if it's contained.
610 	 * m may be modified in ip6_hopopts_input().
611 	 * If a JumboPayload option is included, plen will also be modified.
612 	 */
613 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
614 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
615 		struct ip6_hbh *hbh;
616 
617 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
618 #if 0	/*touches NULL pointer*/
619 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
620 #endif
621 			return;	/* m have already been freed */
622 		}
623 
624 		/* adjust pointer */
625 		ip6 = mtod(m, struct ip6_hdr *);
626 
627 		/*
628 		 * if the payload length field is 0 and the next header field
629 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
630 		 * option MUST be included.
631 		 */
632 		if (ip6->ip6_plen == 0 && plen == 0) {
633 			/*
634 			 * Note that if a valid jumbo payload option is
635 			 * contained, ip6_hopopts_input() must set a valid
636 			 * (non-zero) payload length to the variable plen.
637 			 */
638 			IP6_STATINC(IP6_STAT_BADOPTIONS);
639 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
640 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
641 			icmp6_error(m, ICMP6_PARAM_PROB,
642 				    ICMP6_PARAMPROB_HEADER,
643 				    (char *)&ip6->ip6_plen - (char *)ip6);
644 			return;
645 		}
646 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
647 			sizeof(struct ip6_hbh));
648 		if (hbh == NULL) {
649 			IP6_STATINC(IP6_STAT_TOOSHORT);
650 			return;
651 		}
652 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
653 		nxt = hbh->ip6h_nxt;
654 
655 		/*
656 		 * accept the packet if a router alert option is included
657 		 * and we act as an IPv6 router.
658 		 */
659 		if (rtalert != ~0 && ip6_forwarding)
660 			ours = 1;
661 	} else
662 		nxt = ip6->ip6_nxt;
663 
664 	/*
665 	 * Check that the amount of data in the buffers
666 	 * is as at least much as the IPv6 header would have us expect.
667 	 * Trim mbufs if longer than we expect.
668 	 * Drop packet if shorter than we expect.
669 	 */
670 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
671 		IP6_STATINC(IP6_STAT_TOOSHORT);
672 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
673 		goto bad;
674 	}
675 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
676 		if (m->m_len == m->m_pkthdr.len) {
677 			m->m_len = sizeof(struct ip6_hdr) + plen;
678 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
679 		} else
680 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
681 	}
682 
683 	/*
684 	 * Forward if desirable.
685 	 */
686 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
687 		/*
688 		 * If we are acting as a multicast router, all
689 		 * incoming multicast packets are passed to the
690 		 * kernel-level multicast forwarding function.
691 		 * The packet is returned (relatively) intact; if
692 		 * ip6_mforward() returns a non-zero value, the packet
693 		 * must be discarded, else it may be accepted below.
694 		 */
695 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
696 			IP6_STATINC(IP6_STAT_CANTFORWARD);
697 			m_freem(m);
698 			return;
699 		}
700 		if (!ours) {
701 			m_freem(m);
702 			return;
703 		}
704 	} else if (!ours) {
705 		ip6_forward(m, srcrt);
706 		return;
707 	}
708 
709 	ip6 = mtod(m, struct ip6_hdr *);
710 
711 	/*
712 	 * Malicious party may be able to use IPv4 mapped addr to confuse
713 	 * tcp/udp stack and bypass security checks (act as if it was from
714 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
715 	 *
716 	 * For SIIT end node behavior, you may want to disable the check.
717 	 * However, you will  become vulnerable to attacks using IPv4 mapped
718 	 * source.
719 	 */
720 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
721 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
722 		IP6_STATINC(IP6_STAT_BADSCOPE);
723 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
724 		goto bad;
725 	}
726 
727 	/*
728 	 * Tell launch routine the next header
729 	 */
730 #ifdef IFA_STATS
731 	if (deliverifp != NULL) {
732 		struct in6_ifaddr *ia6;
733 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
734 		if (ia6)
735 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
736 	}
737 #endif
738 	IP6_STATINC(IP6_STAT_DELIVERED);
739 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
740 	nest = 0;
741 
742 	rh_present = 0;
743 	while (nxt != IPPROTO_DONE) {
744 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
745 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
746 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
747 			goto bad;
748 		}
749 
750 		/*
751 		 * protection against faulty packet - there should be
752 		 * more sanity checks in header chain processing.
753 		 */
754 		if (m->m_pkthdr.len < off) {
755 			IP6_STATINC(IP6_STAT_TOOSHORT);
756 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
757 			goto bad;
758 		}
759 
760 		if (nxt == IPPROTO_ROUTING) {
761 			if (rh_present++) {
762 				in6_ifstat_inc(m->m_pkthdr.rcvif,
763 				    ifs6_in_hdrerr);
764 				IP6_STATINC(IP6_STAT_BADOPTIONS);
765 				goto bad;
766 			}
767 		}
768 
769 #ifdef FAST_IPSEC
770 	/*
771 	 * enforce IPsec policy checking if we are seeing last header.
772 	 * note that we do not visit this with protocols with pcb layer
773 	 * code - like udp/tcp/raw ip.
774 	 */
775 	if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) {
776 		/*
777 		 * Check if the packet has already had IPsec processing
778 		 * done.  If so, then just pass it along.  This tag gets
779 		 * set during AH, ESP, etc. input handling, before the
780 		 * packet is returned to the ip input queue for delivery.
781 		 */
782 		mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
783 		s = splsoftnet();
784 		if (mtag != NULL) {
785 			tdbi = (struct tdb_ident *)(mtag + 1);
786 			sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
787 		} else {
788 			sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
789 									IP_FORWARDING, &error);
790 		}
791 		if (sp != NULL) {
792 			/*
793 			 * Check security policy against packet attributes.
794 			 */
795 			error = ipsec_in_reject(sp, m);
796 			KEY_FREESP(&sp);
797 		} else {
798 			/* XXX error stat??? */
799 			error = EINVAL;
800 			DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/
801 		}
802 		splx(s);
803 		if (error)
804 			goto bad;
805 	}
806 #endif /* FAST_IPSEC */
807 
808 
809 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
810 	}
811 	return;
812  bad:
813 	m_freem(m);
814 }
815 
816 /*
817  * set/grab in6_ifaddr correspond to IPv6 destination address.
818  */
819 static struct m_tag *
820 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
821 {
822 	struct m_tag *mtag;
823 
824 	mtag = ip6_addaux(m);
825 	if (mtag != NULL) {
826 		struct ip6aux *ip6a;
827 
828 		ip6a = (struct ip6aux *)(mtag + 1);
829 		in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id);
830 		ip6a->ip6a_src = ia->ia_addr.sin6_addr;
831 		ip6a->ip6a_flags = ia->ia6_flags;
832 	}
833 	return mtag;	/* NULL if failed to set */
834 }
835 
836 const struct ip6aux *
837 ip6_getdstifaddr(struct mbuf *m)
838 {
839 	struct m_tag *mtag;
840 
841 	mtag = ip6_findaux(m);
842 	if (mtag != NULL)
843 		return (struct ip6aux *)(mtag + 1);
844 	else
845 		return NULL;
846 }
847 
848 /*
849  * Hop-by-Hop options header processing. If a valid jumbo payload option is
850  * included, the real payload length will be stored in plenp.
851  *
852  * rtalertp - XXX: should be stored more smart way
853  */
854 int
855 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
856 	struct mbuf **mp, int *offp)
857 {
858 	struct mbuf *m = *mp;
859 	int off = *offp, hbhlen;
860 	struct ip6_hbh *hbh;
861 
862 	/* validation of the length of the header */
863 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
864 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
865 	if (hbh == NULL) {
866 		IP6_STATINC(IP6_STAT_TOOSHORT);
867 		return -1;
868 	}
869 	hbhlen = (hbh->ip6h_len + 1) << 3;
870 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
871 		hbhlen);
872 	if (hbh == NULL) {
873 		IP6_STATINC(IP6_STAT_TOOSHORT);
874 		return -1;
875 	}
876 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
877 	off += hbhlen;
878 	hbhlen -= sizeof(struct ip6_hbh);
879 
880 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
881 				hbhlen, rtalertp, plenp) < 0)
882 		return (-1);
883 
884 	*offp = off;
885 	*mp = m;
886 	return (0);
887 }
888 
889 /*
890  * Search header for all Hop-by-hop options and process each option.
891  * This function is separate from ip6_hopopts_input() in order to
892  * handle a case where the sending node itself process its hop-by-hop
893  * options header. In such a case, the function is called from ip6_output().
894  *
895  * The function assumes that hbh header is located right after the IPv6 header
896  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
897  * opthead + hbhlen is located in continuous memory region.
898  */
899 static int
900 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
901 	u_int32_t *rtalertp, u_int32_t *plenp)
902 {
903 	struct ip6_hdr *ip6;
904 	int optlen = 0;
905 	u_int8_t *opt = opthead;
906 	u_int16_t rtalert_val;
907 	u_int32_t jumboplen;
908 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
909 
910 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
911 		switch (*opt) {
912 		case IP6OPT_PAD1:
913 			optlen = 1;
914 			break;
915 		case IP6OPT_PADN:
916 			if (hbhlen < IP6OPT_MINLEN) {
917 				IP6_STATINC(IP6_STAT_TOOSMALL);
918 				goto bad;
919 			}
920 			optlen = *(opt + 1) + 2;
921 			break;
922 		case IP6OPT_RTALERT:
923 			/* XXX may need check for alignment */
924 			if (hbhlen < IP6OPT_RTALERT_LEN) {
925 				IP6_STATINC(IP6_STAT_TOOSMALL);
926 				goto bad;
927 			}
928 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
929 				/* XXX stat */
930 				icmp6_error(m, ICMP6_PARAM_PROB,
931 				    ICMP6_PARAMPROB_HEADER,
932 				    erroff + opt + 1 - opthead);
933 				return (-1);
934 			}
935 			optlen = IP6OPT_RTALERT_LEN;
936 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
937 			*rtalertp = ntohs(rtalert_val);
938 			break;
939 		case IP6OPT_JUMBO:
940 			/* XXX may need check for alignment */
941 			if (hbhlen < IP6OPT_JUMBO_LEN) {
942 				IP6_STATINC(IP6_STAT_TOOSMALL);
943 				goto bad;
944 			}
945 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
946 				/* XXX stat */
947 				icmp6_error(m, ICMP6_PARAM_PROB,
948 				    ICMP6_PARAMPROB_HEADER,
949 				    erroff + opt + 1 - opthead);
950 				return (-1);
951 			}
952 			optlen = IP6OPT_JUMBO_LEN;
953 
954 			/*
955 			 * IPv6 packets that have non 0 payload length
956 			 * must not contain a jumbo payload option.
957 			 */
958 			ip6 = mtod(m, struct ip6_hdr *);
959 			if (ip6->ip6_plen) {
960 				IP6_STATINC(IP6_STAT_BADOPTIONS);
961 				icmp6_error(m, ICMP6_PARAM_PROB,
962 				    ICMP6_PARAMPROB_HEADER,
963 				    erroff + opt - opthead);
964 				return (-1);
965 			}
966 
967 			/*
968 			 * We may see jumbolen in unaligned location, so
969 			 * we'd need to perform bcopy().
970 			 */
971 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
972 			jumboplen = (u_int32_t)htonl(jumboplen);
973 
974 #if 1
975 			/*
976 			 * if there are multiple jumbo payload options,
977 			 * *plenp will be non-zero and the packet will be
978 			 * rejected.
979 			 * the behavior may need some debate in ipngwg -
980 			 * multiple options does not make sense, however,
981 			 * there's no explicit mention in specification.
982 			 */
983 			if (*plenp != 0) {
984 				IP6_STATINC(IP6_STAT_BADOPTIONS);
985 				icmp6_error(m, ICMP6_PARAM_PROB,
986 				    ICMP6_PARAMPROB_HEADER,
987 				    erroff + opt + 2 - opthead);
988 				return (-1);
989 			}
990 #endif
991 
992 			/*
993 			 * jumbo payload length must be larger than 65535.
994 			 */
995 			if (jumboplen <= IPV6_MAXPACKET) {
996 				IP6_STATINC(IP6_STAT_BADOPTIONS);
997 				icmp6_error(m, ICMP6_PARAM_PROB,
998 				    ICMP6_PARAMPROB_HEADER,
999 				    erroff + opt + 2 - opthead);
1000 				return (-1);
1001 			}
1002 			*plenp = jumboplen;
1003 
1004 			break;
1005 		default:		/* unknown option */
1006 			if (hbhlen < IP6OPT_MINLEN) {
1007 				IP6_STATINC(IP6_STAT_TOOSMALL);
1008 				goto bad;
1009 			}
1010 			optlen = ip6_unknown_opt(opt, m,
1011 			    erroff + opt - opthead);
1012 			if (optlen == -1)
1013 				return (-1);
1014 			optlen += 2;
1015 			break;
1016 		}
1017 	}
1018 
1019 	return (0);
1020 
1021   bad:
1022 	m_freem(m);
1023 	return (-1);
1024 }
1025 
1026 /*
1027  * Unknown option processing.
1028  * The third argument `off' is the offset from the IPv6 header to the option,
1029  * which is necessary if the IPv6 header the and option header and IPv6 header
1030  * is not continuous in order to return an ICMPv6 error.
1031  */
1032 int
1033 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1034 {
1035 	struct ip6_hdr *ip6;
1036 
1037 	switch (IP6OPT_TYPE(*optp)) {
1038 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1039 		return ((int)*(optp + 1));
1040 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1041 		m_freem(m);
1042 		return (-1);
1043 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1044 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1045 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1046 		return (-1);
1047 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1048 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1049 		ip6 = mtod(m, struct ip6_hdr *);
1050 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1051 		    (m->m_flags & (M_BCAST|M_MCAST)))
1052 			m_freem(m);
1053 		else
1054 			icmp6_error(m, ICMP6_PARAM_PROB,
1055 				    ICMP6_PARAMPROB_OPTION, off);
1056 		return (-1);
1057 	}
1058 
1059 	m_freem(m);		/* XXX: NOTREACHED */
1060 	return (-1);
1061 }
1062 
1063 /*
1064  * Create the "control" list for this pcb.
1065  *
1066  * The routine will be called from upper layer handlers like tcp6_input().
1067  * Thus the routine assumes that the caller (tcp6_input) have already
1068  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1069  * very first mbuf on the mbuf chain.
1070  * We may want to add some infinite loop prevention or sanity checks for safety.
1071  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1072  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1073  */
1074 void
1075 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
1076 	struct ip6_hdr *ip6, struct mbuf *m)
1077 {
1078 #ifdef RFC2292
1079 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1080 #else
1081 #define IS2292(x, y)	(y)
1082 #endif
1083 
1084 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP
1085 #ifdef SO_OTIMESTAMP
1086 	    || in6p->in6p_socket->so_options & SO_OTIMESTAMP
1087 #endif
1088 	) {
1089 		struct timeval tv;
1090 
1091 		microtime(&tv);
1092 #ifdef SO_OTIMESTAMP
1093 		if (in6p->in6p_socket->so_options & SO_OTIMESTAMP) {
1094 			struct timeval50 tv50;
1095 			timeval_to_timeval50(&tv, &tv50);
1096 			*mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
1097 			    SCM_OTIMESTAMP, SOL_SOCKET);
1098 		} else
1099 #endif
1100 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1101 		    SCM_TIMESTAMP, SOL_SOCKET);
1102 		if (*mp)
1103 			mp = &(*mp)->m_next;
1104 	}
1105 
1106 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
1107 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1108 		return;
1109 
1110 	/* RFC 2292 sec. 5 */
1111 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1112 		struct in6_pktinfo pi6;
1113 
1114 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
1115 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1116 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
1117 		    m->m_pkthdr.rcvif->if_index : 0;
1118 		*mp = sbcreatecontrol((void *) &pi6,
1119 		    sizeof(struct in6_pktinfo),
1120 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1121 		if (*mp)
1122 			mp = &(*mp)->m_next;
1123 	}
1124 
1125 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1126 		int hlim = ip6->ip6_hlim & 0xff;
1127 
1128 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
1129 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1130 		if (*mp)
1131 			mp = &(*mp)->m_next;
1132 	}
1133 
1134 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1135 		u_int32_t flowinfo;
1136 		int tclass;
1137 
1138 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1139 		flowinfo >>= 20;
1140 
1141 		tclass = flowinfo & 0xff;
1142 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
1143 		    IPV6_TCLASS, IPPROTO_IPV6);
1144 
1145 		if (*mp)
1146 			mp = &(*mp)->m_next;
1147 	}
1148 
1149 	/*
1150 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1151 	 * privilege for the option (see ip6_ctloutput), but it might be too
1152 	 * strict, since there might be some hop-by-hop options which can be
1153 	 * returned to normal user.
1154 	 * See also RFC3542 section 8 (or RFC2292 section 6).
1155 	 */
1156 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1157 		/*
1158 		 * Check if a hop-by-hop options header is contatined in the
1159 		 * received packet, and if so, store the options as ancillary
1160 		 * data. Note that a hop-by-hop options header must be
1161 		 * just after the IPv6 header, which fact is assured through
1162 		 * the IPv6 input processing.
1163 		 */
1164 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1165 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1166 			struct ip6_hbh *hbh;
1167 			int hbhlen;
1168 			struct mbuf *ext;
1169 
1170 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1171 			    xip6->ip6_nxt);
1172 			if (ext == NULL) {
1173 				IP6_STATINC(IP6_STAT_TOOSHORT);
1174 				return;
1175 			}
1176 			hbh = mtod(ext, struct ip6_hbh *);
1177 			hbhlen = (hbh->ip6h_len + 1) << 3;
1178 			if (hbhlen != ext->m_len) {
1179 				m_freem(ext);
1180 				IP6_STATINC(IP6_STAT_TOOSHORT);
1181 				return;
1182 			}
1183 
1184 			/*
1185 			 * XXX: We copy whole the header even if a jumbo
1186 			 * payload option is included, which option is to
1187 			 * be removed before returning in the RFC 2292.
1188 			 * Note: this constraint is removed in RFC3542.
1189 			 */
1190 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
1191 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1192 			    IPPROTO_IPV6);
1193 			if (*mp)
1194 				mp = &(*mp)->m_next;
1195 			m_freem(ext);
1196 		}
1197 	}
1198 
1199 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1200 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1201 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1202 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1203 
1204 		/*
1205 		 * Search for destination options headers or routing
1206 		 * header(s) through the header chain, and stores each
1207 		 * header as ancillary data.
1208 		 * Note that the order of the headers remains in
1209 		 * the chain of ancillary data.
1210 		 */
1211 		for (;;) {	/* is explicit loop prevention necessary? */
1212 			struct ip6_ext *ip6e = NULL;
1213 			int elen;
1214 			struct mbuf *ext = NULL;
1215 
1216 			/*
1217 			 * if it is not an extension header, don't try to
1218 			 * pull it from the chain.
1219 			 */
1220 			switch (nxt) {
1221 			case IPPROTO_DSTOPTS:
1222 			case IPPROTO_ROUTING:
1223 			case IPPROTO_HOPOPTS:
1224 			case IPPROTO_AH: /* is it possible? */
1225 				break;
1226 			default:
1227 				goto loopend;
1228 			}
1229 
1230 			ext = ip6_pullexthdr(m, off, nxt);
1231 			if (ext == NULL) {
1232 				IP6_STATINC(IP6_STAT_TOOSHORT);
1233 				return;
1234 			}
1235 			ip6e = mtod(ext, struct ip6_ext *);
1236 			if (nxt == IPPROTO_AH)
1237 				elen = (ip6e->ip6e_len + 2) << 2;
1238 			else
1239 				elen = (ip6e->ip6e_len + 1) << 3;
1240 			if (elen != ext->m_len) {
1241 				m_freem(ext);
1242 				IP6_STATINC(IP6_STAT_TOOSHORT);
1243 				return;
1244 			}
1245 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1246 
1247 			switch (nxt) {
1248 			case IPPROTO_DSTOPTS:
1249 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1250 					break;
1251 
1252 				*mp = sbcreatecontrol((void *)ip6e, elen,
1253 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1254 				    IPPROTO_IPV6);
1255 				if (*mp)
1256 					mp = &(*mp)->m_next;
1257 				break;
1258 
1259 			case IPPROTO_ROUTING:
1260 				if (!(in6p->in6p_flags & IN6P_RTHDR))
1261 					break;
1262 
1263 				*mp = sbcreatecontrol((void *)ip6e, elen,
1264 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1265 				    IPPROTO_IPV6);
1266 				if (*mp)
1267 					mp = &(*mp)->m_next;
1268 				break;
1269 
1270 			case IPPROTO_HOPOPTS:
1271 			case IPPROTO_AH: /* is it possible? */
1272 				break;
1273 
1274 			default:
1275 				/*
1276 			 	 * other cases have been filtered in the above.
1277 				 * none will visit this case.  here we supply
1278 				 * the code just in case (nxt overwritten or
1279 				 * other cases).
1280 				 */
1281 				m_freem(ext);
1282 				goto loopend;
1283 
1284 			}
1285 
1286 			/* proceed with the next header. */
1287 			off += elen;
1288 			nxt = ip6e->ip6e_nxt;
1289 			ip6e = NULL;
1290 			m_freem(ext);
1291 			ext = NULL;
1292 		}
1293 	  loopend:
1294 	  	;
1295 	}
1296 }
1297 #undef IS2292
1298 
1299 
1300 void
1301 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
1302     uint32_t *mtu)
1303 {
1304 	struct socket *so;
1305 	struct mbuf *m_mtu;
1306 	struct ip6_mtuinfo mtuctl;
1307 
1308 	so = in6p->in6p_socket;
1309 
1310 	if (mtu == NULL)
1311 		return;
1312 
1313 #ifdef DIAGNOSTIC
1314 	if (so == NULL)		/* I believe this is impossible */
1315 		panic("ip6_notify_pmtu: socket is NULL");
1316 #endif
1317 
1318 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
1319 	mtuctl.ip6m_mtu = *mtu;
1320 	mtuctl.ip6m_addr = *dst;
1321 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
1322 		return;
1323 
1324 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
1325 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1326 		return;
1327 
1328 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
1329 	    == 0) {
1330 		m_freem(m_mtu);
1331 		/* XXX: should count statistics */
1332 	} else
1333 		sorwakeup(so);
1334 
1335 	return;
1336 }
1337 
1338 /*
1339  * pull single extension header from mbuf chain.  returns single mbuf that
1340  * contains the result, or NULL on error.
1341  */
1342 static struct mbuf *
1343 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1344 {
1345 	struct ip6_ext ip6e;
1346 	size_t elen;
1347 	struct mbuf *n;
1348 
1349 #ifdef DIAGNOSTIC
1350 	switch (nxt) {
1351 	case IPPROTO_DSTOPTS:
1352 	case IPPROTO_ROUTING:
1353 	case IPPROTO_HOPOPTS:
1354 	case IPPROTO_AH: /* is it possible? */
1355 		break;
1356 	default:
1357 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1358 	}
1359 #endif
1360 
1361 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1362 	if (nxt == IPPROTO_AH)
1363 		elen = (ip6e.ip6e_len + 2) << 2;
1364 	else
1365 		elen = (ip6e.ip6e_len + 1) << 3;
1366 
1367 	MGET(n, M_DONTWAIT, MT_DATA);
1368 	if (n && elen >= MLEN) {
1369 		MCLGET(n, M_DONTWAIT);
1370 		if ((n->m_flags & M_EXT) == 0) {
1371 			m_free(n);
1372 			n = NULL;
1373 		}
1374 	}
1375 	if (!n)
1376 		return NULL;
1377 
1378 	n->m_len = 0;
1379 	if (elen >= M_TRAILINGSPACE(n)) {
1380 		m_free(n);
1381 		return NULL;
1382 	}
1383 
1384 	m_copydata(m, off, elen, mtod(n, void *));
1385 	n->m_len = elen;
1386 	return n;
1387 }
1388 
1389 /*
1390  * Get pointer to the previous header followed by the header
1391  * currently processed.
1392  * XXX: This function supposes that
1393  *	M includes all headers,
1394  *	the next header field and the header length field of each header
1395  *	are valid, and
1396  *	the sum of each header length equals to OFF.
1397  * Because of these assumptions, this function must be called very
1398  * carefully. Moreover, it will not be used in the near future when
1399  * we develop `neater' mechanism to process extension headers.
1400  */
1401 u_int8_t *
1402 ip6_get_prevhdr(struct mbuf *m, int off)
1403 {
1404 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1405 
1406 	if (off == sizeof(struct ip6_hdr))
1407 		return (&ip6->ip6_nxt);
1408 	else {
1409 		int len, nxt;
1410 		struct ip6_ext *ip6e = NULL;
1411 
1412 		nxt = ip6->ip6_nxt;
1413 		len = sizeof(struct ip6_hdr);
1414 		while (len < off) {
1415 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
1416 
1417 			switch (nxt) {
1418 			case IPPROTO_FRAGMENT:
1419 				len += sizeof(struct ip6_frag);
1420 				break;
1421 			case IPPROTO_AH:
1422 				len += (ip6e->ip6e_len + 2) << 2;
1423 				break;
1424 			default:
1425 				len += (ip6e->ip6e_len + 1) << 3;
1426 				break;
1427 			}
1428 			nxt = ip6e->ip6e_nxt;
1429 		}
1430 		if (ip6e)
1431 			return (&ip6e->ip6e_nxt);
1432 		else
1433 			return NULL;
1434 	}
1435 }
1436 
1437 /*
1438  * get next header offset.  m will be retained.
1439  */
1440 int
1441 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1442 {
1443 	struct ip6_hdr ip6;
1444 	struct ip6_ext ip6e;
1445 	struct ip6_frag fh;
1446 
1447 	/* just in case */
1448 	if (m == NULL)
1449 		panic("ip6_nexthdr: m == NULL");
1450 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1451 		return -1;
1452 
1453 	switch (proto) {
1454 	case IPPROTO_IPV6:
1455 		/* do not chase beyond intermediate IPv6 headers */
1456 		if (off != 0)
1457 			return -1;
1458 		if (m->m_pkthdr.len < off + sizeof(ip6))
1459 			return -1;
1460 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
1461 		if (nxtp)
1462 			*nxtp = ip6.ip6_nxt;
1463 		off += sizeof(ip6);
1464 		return off;
1465 
1466 	case IPPROTO_FRAGMENT:
1467 		/*
1468 		 * terminate parsing if it is not the first fragment,
1469 		 * it does not make sense to parse through it.
1470 		 */
1471 		if (m->m_pkthdr.len < off + sizeof(fh))
1472 			return -1;
1473 		m_copydata(m, off, sizeof(fh), (void *)&fh);
1474 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1475 			return -1;
1476 		if (nxtp)
1477 			*nxtp = fh.ip6f_nxt;
1478 		off += sizeof(struct ip6_frag);
1479 		return off;
1480 
1481 	case IPPROTO_AH:
1482 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1483 			return -1;
1484 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1485 		if (nxtp)
1486 			*nxtp = ip6e.ip6e_nxt;
1487 		off += (ip6e.ip6e_len + 2) << 2;
1488 		if (m->m_pkthdr.len < off)
1489 			return -1;
1490 		return off;
1491 
1492 	case IPPROTO_HOPOPTS:
1493 	case IPPROTO_ROUTING:
1494 	case IPPROTO_DSTOPTS:
1495 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1496 			return -1;
1497 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1498 		if (nxtp)
1499 			*nxtp = ip6e.ip6e_nxt;
1500 		off += (ip6e.ip6e_len + 1) << 3;
1501 		if (m->m_pkthdr.len < off)
1502 			return -1;
1503 		return off;
1504 
1505 	case IPPROTO_NONE:
1506 	case IPPROTO_ESP:
1507 	case IPPROTO_IPCOMP:
1508 		/* give up */
1509 		return -1;
1510 
1511 	default:
1512 		return -1;
1513 	}
1514 }
1515 
1516 /*
1517  * get offset for the last header in the chain.  m will be kept untainted.
1518  */
1519 int
1520 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1521 {
1522 	int newoff;
1523 	int nxt;
1524 
1525 	if (!nxtp) {
1526 		nxt = -1;
1527 		nxtp = &nxt;
1528 	}
1529 	for (;;) {
1530 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1531 		if (newoff < 0)
1532 			return off;
1533 		else if (newoff < off)
1534 			return -1;	/* invalid */
1535 		else if (newoff == off)
1536 			return newoff;
1537 
1538 		off = newoff;
1539 		proto = *nxtp;
1540 	}
1541 }
1542 
1543 struct m_tag *
1544 ip6_addaux(struct mbuf *m)
1545 {
1546 	struct m_tag *mtag;
1547 
1548 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1549 	if (!mtag) {
1550 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1551 		    M_NOWAIT);
1552 		if (mtag) {
1553 			m_tag_prepend(m, mtag);
1554 			memset(mtag + 1, 0, sizeof(struct ip6aux));
1555 		}
1556 	}
1557 	return mtag;
1558 }
1559 
1560 struct m_tag *
1561 ip6_findaux(struct mbuf *m)
1562 {
1563 	struct m_tag *mtag;
1564 
1565 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1566 	return mtag;
1567 }
1568 
1569 void
1570 ip6_delaux(struct mbuf *m)
1571 {
1572 	struct m_tag *mtag;
1573 
1574 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1575 	if (mtag)
1576 		m_tag_delete(m, mtag);
1577 }
1578 
1579 #ifdef GATEWAY
1580 /*
1581  * sysctl helper routine for net.inet.ip6.maxflows. Since
1582  * we could reduce this value, call ip6flow_reap();
1583  */
1584 static int
1585 sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS)
1586 {
1587 	int error;
1588 
1589 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1590 	if (error || newp == NULL)
1591 		return (error);
1592 
1593 	mutex_enter(softnet_lock);
1594 	KERNEL_LOCK(1, NULL);
1595 
1596 	ip6flow_reap(0);
1597 
1598 	KERNEL_UNLOCK_ONE(NULL);
1599 	mutex_exit(softnet_lock);
1600 
1601 	return (0);
1602 }
1603 
1604 static int
1605 sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS)
1606 {
1607 	int error, tmp;
1608 	struct sysctlnode node;
1609 
1610 	node = *rnode;
1611 	tmp = ip6_hashsize;
1612 	node.sysctl_data = &tmp;
1613 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1614 	if (error || newp == NULL)
1615 		return (error);
1616 
1617 	if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
1618 		/*
1619 		 * Can only fail due to malloc()
1620 		 */
1621 		mutex_enter(softnet_lock);
1622 		KERNEL_LOCK(1, NULL);
1623 
1624 		error = ip6flow_invalidate_all(tmp);
1625 
1626 		KERNEL_UNLOCK_ONE(NULL);
1627 		mutex_exit(softnet_lock);
1628 	} else {
1629 		/*
1630 		 * EINVAL if not a power of 2
1631 		 */
1632 		error = EINVAL;
1633 	}
1634 
1635 	return error;
1636 }
1637 #endif /* GATEWAY */
1638 
1639 /*
1640  * System control for IP6
1641  */
1642 
1643 const u_char inet6ctlerrmap[PRC_NCMDS] = {
1644 	0,		0,		0,		0,
1645 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1646 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1647 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1648 	0,		0,		0,		0,
1649 	ENOPROTOOPT
1650 };
1651 
1652 static int
1653 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
1654 {
1655 
1656 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
1657 }
1658 
1659 static void
1660 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
1661 {
1662 #ifdef RFC2292
1663 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1664 #else
1665 #define IS2292(x, y)	(y)
1666 #endif
1667 
1668 	sysctl_createv(clog, 0, NULL, NULL,
1669 		       CTLFLAG_PERMANENT,
1670 		       CTLTYPE_NODE, "net", NULL,
1671 		       NULL, 0, NULL, 0,
1672 		       CTL_NET, CTL_EOL);
1673 	sysctl_createv(clog, 0, NULL, NULL,
1674 		       CTLFLAG_PERMANENT,
1675 		       CTLTYPE_NODE, "inet6",
1676 		       SYSCTL_DESCR("PF_INET6 related settings"),
1677 		       NULL, 0, NULL, 0,
1678 		       CTL_NET, PF_INET6, CTL_EOL);
1679 	sysctl_createv(clog, 0, NULL, NULL,
1680 		       CTLFLAG_PERMANENT,
1681 		       CTLTYPE_NODE, "ip6",
1682 		       SYSCTL_DESCR("IPv6 related settings"),
1683 		       NULL, 0, NULL, 0,
1684 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1685 
1686 	sysctl_createv(clog, 0, NULL, NULL,
1687 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1688 		       CTLTYPE_INT, "forwarding",
1689 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1690 		       NULL, 0, &ip6_forwarding, 0,
1691 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1692 		       IPV6CTL_FORWARDING, CTL_EOL);
1693 	sysctl_createv(clog, 0, NULL, NULL,
1694 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1695 		       CTLTYPE_INT, "redirect",
1696 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1697 		       NULL, 0, &ip6_sendredirects, 0,
1698 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1699 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1700 	sysctl_createv(clog, 0, NULL, NULL,
1701 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1702 		       CTLTYPE_INT, "hlim",
1703 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1704 		       NULL, 0, &ip6_defhlim, 0,
1705 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1706 		       IPV6CTL_DEFHLIM, CTL_EOL);
1707 #ifdef notyet
1708 	sysctl_createv(clog, 0, NULL, NULL,
1709 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1710 		       CTLTYPE_INT, "mtu", NULL,
1711 		       NULL, 0, &, 0,
1712 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1713 		       IPV6CTL_DEFMTU, CTL_EOL);
1714 #endif
1715 #ifdef __no_idea__
1716 	sysctl_createv(clog, 0, NULL, NULL,
1717 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1718 		       CTLTYPE_INT, "forwsrcrt", NULL,
1719 		       NULL, 0, &?, 0,
1720 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1721 		       IPV6CTL_FORWSRCRT, CTL_EOL);
1722 	sysctl_createv(clog, 0, NULL, NULL,
1723 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1724 		       CTLTYPE_STRUCT, "mrtstats", NULL,
1725 		       NULL, 0, &?, sizeof(?),
1726 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1727 		       IPV6CTL_MRTSTATS, CTL_EOL);
1728 	sysctl_createv(clog, 0, NULL, NULL,
1729 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1730 		       CTLTYPE_?, "mrtproto", NULL,
1731 		       NULL, 0, &?, sizeof(?),
1732 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1733 		       IPV6CTL_MRTPROTO, CTL_EOL);
1734 #endif
1735 	sysctl_createv(clog, 0, NULL, NULL,
1736 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1737 		       CTLTYPE_INT, "maxfragpackets",
1738 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1739 				    "for reassembly"),
1740 		       NULL, 0, &ip6_maxfragpackets, 0,
1741 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1742 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1743 #ifdef __no_idea__
1744 	sysctl_createv(clog, 0, NULL, NULL,
1745 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1746 		       CTLTYPE_INT, "sourcecheck", NULL,
1747 		       NULL, 0, &?, 0,
1748 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1749 		       IPV6CTL_SOURCECHECK, CTL_EOL);
1750 	sysctl_createv(clog, 0, NULL, NULL,
1751 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1752 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
1753 		       NULL, 0, &?, 0,
1754 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1755 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
1756 #endif
1757 	sysctl_createv(clog, 0, NULL, NULL,
1758 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1759 		       CTLTYPE_INT, "accept_rtadv",
1760 		       SYSCTL_DESCR("Accept router advertisements"),
1761 		       NULL, 0, &ip6_accept_rtadv, 0,
1762 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1763 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1764 	sysctl_createv(clog, 0, NULL, NULL,
1765 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1766 		       CTLTYPE_INT, "rtadv_maxroutes",
1767 		       SYSCTL_DESCR("Maximum number of routes accepted via router advertisements"),
1768 		       NULL, 0, &ip6_rtadv_maxroutes, 0,
1769 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1770 		       IPV6CTL_RTADV_MAXROUTES, CTL_EOL);
1771 	sysctl_createv(clog, 0, NULL, NULL,
1772 		       CTLFLAG_PERMANENT,
1773 		       CTLTYPE_INT, "rtadv_numroutes",
1774 		       SYSCTL_DESCR("Current number of routes accepted via router advertisements"),
1775 		       NULL, 0, &nd6_numroutes, 0,
1776 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1777 		       IPV6CTL_RTADV_NUMROUTES, CTL_EOL);
1778 	sysctl_createv(clog, 0, NULL, NULL,
1779 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1780 		       CTLTYPE_INT, "keepfaith",
1781 		       SYSCTL_DESCR("Activate faith interface"),
1782 		       NULL, 0, &ip6_keepfaith, 0,
1783 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1784 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1785 	sysctl_createv(clog, 0, NULL, NULL,
1786 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1787 		       CTLTYPE_INT, "log_interval",
1788 		       SYSCTL_DESCR("Minumum interval between logging "
1789 				    "unroutable packets"),
1790 		       NULL, 0, &ip6_log_interval, 0,
1791 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1792 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1793 	sysctl_createv(clog, 0, NULL, NULL,
1794 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1795 		       CTLTYPE_INT, "hdrnestlimit",
1796 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1797 		       NULL, 0, &ip6_hdrnestlimit, 0,
1798 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1799 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1800 	sysctl_createv(clog, 0, NULL, NULL,
1801 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1802 		       CTLTYPE_INT, "dad_count",
1803 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1804 				    "probes to send"),
1805 		       NULL, 0, &ip6_dad_count, 0,
1806 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1807 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1808 	sysctl_createv(clog, 0, NULL, NULL,
1809 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1810 		       CTLTYPE_INT, "auto_flowlabel",
1811 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1812 		       NULL, 0, &ip6_auto_flowlabel, 0,
1813 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1814 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1815 	sysctl_createv(clog, 0, NULL, NULL,
1816 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1817 		       CTLTYPE_INT, "defmcasthlim",
1818 		       SYSCTL_DESCR("Default multicast hop limit"),
1819 		       NULL, 0, &ip6_defmcasthlim, 0,
1820 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1821 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1822 #if NGIF > 0
1823 	sysctl_createv(clog, 0, NULL, NULL,
1824 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1825 		       CTLTYPE_INT, "gifhlim",
1826 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
1827 		       NULL, 0, &ip6_gif_hlim, 0,
1828 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1829 		       IPV6CTL_GIF_HLIM, CTL_EOL);
1830 #endif /* NGIF */
1831 	sysctl_createv(clog, 0, NULL, NULL,
1832 		       CTLFLAG_PERMANENT,
1833 		       CTLTYPE_STRING, "kame_version",
1834 		       SYSCTL_DESCR("KAME Version"),
1835 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1836 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1837 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1838 	sysctl_createv(clog, 0, NULL, NULL,
1839 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1840 		       CTLTYPE_INT, "use_deprecated",
1841 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1842 				    "source addresses"),
1843 		       NULL, 0, &ip6_use_deprecated, 0,
1844 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1845 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1846 	sysctl_createv(clog, 0, NULL, NULL,
1847 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1848 		       CTLTYPE_INT, "rr_prune", NULL,
1849 		       NULL, 0, &ip6_rr_prune, 0,
1850 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1851 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1852 	sysctl_createv(clog, 0, NULL, NULL,
1853 		       CTLFLAG_PERMANENT
1854 #ifndef INET6_BINDV6ONLY
1855 		       |CTLFLAG_READWRITE,
1856 #endif
1857 		       CTLTYPE_INT, "v6only",
1858 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1859 				    "to PF_INET sockets"),
1860 		       NULL, 0, &ip6_v6only, 0,
1861 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1862 		       IPV6CTL_V6ONLY, CTL_EOL);
1863 	sysctl_createv(clog, 0, NULL, NULL,
1864 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1865 		       CTLTYPE_INT, "anonportmin",
1866 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1867 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1868 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1869 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1870 	sysctl_createv(clog, 0, NULL, NULL,
1871 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1872 		       CTLTYPE_INT, "anonportmax",
1873 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1874 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1875 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1876 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1877 #ifndef IPNOPRIVPORTS
1878 	sysctl_createv(clog, 0, NULL, NULL,
1879 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1880 		       CTLTYPE_INT, "lowportmin",
1881 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1882 				    "to assign"),
1883 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1884 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1885 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1886 	sysctl_createv(clog, 0, NULL, NULL,
1887 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1888 		       CTLTYPE_INT, "lowportmax",
1889 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1890 				    "to assign"),
1891 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1892 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1893 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1894 #endif /* IPNOPRIVPORTS */
1895 	sysctl_createv(clog, 0, NULL, NULL,
1896 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1897 		       CTLTYPE_INT, "use_tempaddr",
1898 		       SYSCTL_DESCR("Use temporary address"),
1899 		       NULL, 0, &ip6_use_tempaddr, 0,
1900 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1901 		       CTL_CREATE, CTL_EOL);
1902 	sysctl_createv(clog, 0, NULL, NULL,
1903 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1904 		       CTLTYPE_INT, "temppltime",
1905 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
1906 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
1907 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1908 		       CTL_CREATE, CTL_EOL);
1909 	sysctl_createv(clog, 0, NULL, NULL,
1910 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1911 		       CTLTYPE_INT, "tempvltime",
1912 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
1913 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
1914 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1915 		       CTL_CREATE, CTL_EOL);
1916 	sysctl_createv(clog, 0, NULL, NULL,
1917 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1918 		       CTLTYPE_INT, "maxfrags",
1919 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1920 		       NULL, 0, &ip6_maxfrags, 0,
1921 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1922 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1923 	sysctl_createv(clog, 0, NULL, NULL,
1924 		       CTLFLAG_PERMANENT,
1925 		       CTLTYPE_STRUCT, "stats",
1926 		       SYSCTL_DESCR("IPv6 statistics"),
1927 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
1928 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1929 		       IPV6CTL_STATS, CTL_EOL);
1930 	sysctl_createv(clog, 0, NULL, NULL,
1931 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1932 		       CTLTYPE_INT, "use_defaultzone",
1933 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1934 		       NULL, 0, &ip6_use_defzone, 0,
1935 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1936 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1937 	sysctl_createv(clog, 0, NULL, NULL,
1938 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1939 		       CTLTYPE_INT, "mcast_pmtu",
1940 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
1941 		       NULL, 0, &ip6_mcast_pmtu, 0,
1942 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1943 		       CTL_CREATE, CTL_EOL);
1944 #ifdef GATEWAY
1945 	sysctl_createv(clog, 0, NULL, NULL,
1946 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1947 			CTLTYPE_INT, "maxflows",
1948 			SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"),
1949 			sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0,
1950 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1951 			CTL_CREATE, CTL_EOL);
1952 	sysctl_createv(clog, 0, NULL, NULL,
1953 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1954 			CTLTYPE_INT, "hashsize",
1955 			SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"),
1956 			sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0,
1957 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1958 			CTL_CREATE, CTL_EOL);
1959 #endif
1960 	/* anonportalgo RFC6056 subtree */
1961 	const struct sysctlnode *portalgo_node;
1962 	sysctl_createv(clog, 0, NULL, &portalgo_node,
1963 		       CTLFLAG_PERMANENT,
1964 		       CTLTYPE_NODE, "anonportalgo",
1965 		       SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"),
1966 	    	       NULL, 0, NULL, 0,
1967 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL);
1968 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1969 		       CTLFLAG_PERMANENT,
1970 		       CTLTYPE_STRING, "available",
1971 		       SYSCTL_DESCR("available algorithms"),
1972 		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1973 		       CTL_CREATE, CTL_EOL);
1974 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1975 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1976 		       CTLTYPE_STRING, "selected",
1977 		       SYSCTL_DESCR("selected algorithm"),
1978 	               sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN,
1979 		       CTL_CREATE, CTL_EOL);
1980 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1981 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1982 		       CTLTYPE_STRUCT, "reserve",
1983 		       SYSCTL_DESCR("bitmap of reserved ports"),
1984 		       sysctl_portalgo_reserve6, 0, NULL, 0,
1985 		       CTL_CREATE, CTL_EOL);
1986 	sysctl_createv(clog, 0, NULL, NULL,
1987 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1988 		       CTLTYPE_INT, "neighborgcthresh",
1989 		       SYSCTL_DESCR("Maximum number of entries in neighbor"
1990 			" cache"),
1991 		       NULL, 1, &ip6_neighborgcthresh, 0,
1992 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1993 		       CTL_CREATE, CTL_EOL);
1994 	sysctl_createv(clog, 0, NULL, NULL,
1995 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1996 		       CTLTYPE_INT, "maxifprefixes",
1997 		       SYSCTL_DESCR("Maximum number of prefixes created by"
1998 			   " route advertisement per interface"),
1999 		       NULL, 1, &ip6_maxifprefixes, 0,
2000 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
2001 		       CTL_CREATE, CTL_EOL);
2002 	sysctl_createv(clog, 0, NULL, NULL,
2003 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2004 		       CTLTYPE_INT, "maxifdefrouters",
2005 		       SYSCTL_DESCR("Maximum number of default routers created"
2006 			   " by route advertisement per interface"),
2007 		       NULL, 1, &ip6_maxifdefrouters, 0,
2008 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
2009 		       CTL_CREATE, CTL_EOL);
2010 	sysctl_createv(clog, 0, NULL, NULL,
2011 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2012 		       CTLTYPE_INT, "maxdynroutes",
2013 		       SYSCTL_DESCR("Maximum number of routes created via"
2014 			   " redirect"),
2015 		       NULL, 1, &ip6_maxdynroutes, 0,
2016 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
2017 		       CTL_CREATE, CTL_EOL);
2018 }
2019 
2020 void
2021 ip6_statinc(u_int stat)
2022 {
2023 
2024 	KASSERT(stat < IP6_NSTATS);
2025 	IP6_STATINC(stat);
2026 }
2027