xref: /netbsd-src/sys/netinet6/ip6_input.c (revision b757af438b42b93f8c6571f026d8b8ef3eaf5fc9)
1 /*	$NetBSD: ip6_input.c,v 1.137 2012/03/22 20:34:40 drochner 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 1.137 2012/03/22 20:34:40 drochner 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 <netinet6/in6_var.h>
107 #include <netinet6/ip6_var.h>
108 #include <netinet6/ip6_private.h>
109 #include <netinet6/in6_pcb.h>
110 #include <netinet/icmp6.h>
111 #include <netinet6/scope6_var.h>
112 #include <netinet6/in6_ifattach.h>
113 #include <netinet6/nd6.h>
114 
115 #ifdef FAST_IPSEC
116 #include <netipsec/ipsec.h>
117 #include <netipsec/ipsec6.h>
118 #include <netipsec/key.h>
119 #endif /* FAST_IPSEC */
120 
121 #ifdef COMPAT_50
122 #include <compat/sys/time.h>
123 #include <compat/sys/socket.h>
124 #endif
125 
126 #include <netinet6/ip6protosw.h>
127 
128 #include "faith.h"
129 #include "gif.h"
130 
131 #if NGIF > 0
132 #include <netinet6/in6_gif.h>
133 #endif
134 
135 #include <net/net_osdep.h>
136 
137 extern struct domain inet6domain;
138 
139 u_char ip6_protox[IPPROTO_MAX];
140 static int ip6qmaxlen = IFQ_MAXLEN;
141 struct in6_ifaddr *in6_ifaddr;
142 struct ifqueue ip6intrq;
143 
144 extern callout_t in6_tmpaddrtimer_ch;
145 
146 int ip6_forward_srcrt;			/* XXX */
147 int ip6_sourcecheck;			/* XXX */
148 int ip6_sourcecheck_interval;		/* XXX */
149 
150 #ifdef PFIL_HOOKS
151 struct pfil_head inet6_pfil_hook;
152 #endif
153 
154 percpu_t *ip6stat_percpu;
155 
156 static void ip6_init2(void *);
157 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
158 
159 static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *,
160 	u_int32_t *);
161 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
162 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
163 
164 /*
165  * IP6 initialization: fill in IP6 protocol switch table.
166  * All protocols not implemented in kernel go to raw IP6 protocol handler.
167  */
168 void
169 ip6_init(void)
170 {
171 	const struct ip6protosw *pr;
172 	int i;
173 
174 	sysctl_net_inet6_ip6_setup(NULL);
175 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
176 	if (pr == 0)
177 		panic("ip6_init");
178 	for (i = 0; i < IPPROTO_MAX; i++)
179 		ip6_protox[i] = pr - inet6sw;
180 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
181 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
182 		if (pr->pr_domain->dom_family == PF_INET6 &&
183 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
184 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
185 	ip6intrq.ifq_maxlen = ip6qmaxlen;
186 	scope6_init();
187 	addrsel_policy_init();
188 	nd6_init();
189 	frag6_init();
190 	ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR;
191 
192 	ip6_init2(NULL);
193 #ifdef GATEWAY
194 	ip6flow_init(ip6_hashsize);
195 #endif
196 
197 #ifdef PFIL_HOOKS
198 	/* Register our Packet Filter hook. */
199 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
200 	inet6_pfil_hook.ph_af   = AF_INET6;
201 	i = pfil_head_register(&inet6_pfil_hook);
202 	if (i != 0)
203 		printf("ip6_init: WARNING: unable to register pfil hook, "
204 		    "error %d\n", i);
205 #endif /* PFIL_HOOKS */
206 
207 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
208 }
209 
210 static void
211 ip6_init2(void *dummy)
212 {
213 
214 	/* nd6_timer_init */
215 	callout_init(&nd6_timer_ch, CALLOUT_MPSAFE);
216 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
217 
218 	/* timer for regeneranation of temporary addresses randomize ID */
219 	callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE);
220 	callout_reset(&in6_tmpaddrtimer_ch,
221 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
222 		       ip6_temp_regen_advance) * hz,
223 		      in6_tmpaddrtimer, NULL);
224 }
225 
226 /*
227  * IP6 input interrupt handling. Just pass the packet to ip6_input.
228  */
229 void
230 ip6intr(void)
231 {
232 	int s;
233 	struct mbuf *m;
234 
235 	mutex_enter(softnet_lock);
236 	KERNEL_LOCK(1, NULL);
237 	for (;;) {
238 		s = splnet();
239 		IF_DEQUEUE(&ip6intrq, m);
240 		splx(s);
241 		if (m == 0)
242 			break;
243 		/* drop the packet if IPv6 operation is disabled on the IF */
244 		if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
245 			m_freem(m);
246 			break;
247 		}
248 		ip6_input(m);
249 	}
250 	KERNEL_UNLOCK_ONE(NULL);
251 	mutex_exit(softnet_lock);
252 }
253 
254 extern struct	route ip6_forward_rt;
255 
256 void
257 ip6_input(struct mbuf *m)
258 {
259 	struct ip6_hdr *ip6;
260 	int hit, off = sizeof(struct ip6_hdr), nest;
261 	u_int32_t plen;
262 	u_int32_t rtalert = ~0;
263 	int nxt, ours = 0, rh_present = 0;
264 	struct ifnet *deliverifp = NULL;
265 	int srcrt = 0;
266 	const struct rtentry *rt;
267 	union {
268 		struct sockaddr		dst;
269 		struct sockaddr_in6	dst6;
270 	} u;
271 #ifdef FAST_IPSEC
272 	struct m_tag *mtag;
273 	struct tdb_ident *tdbi;
274 	struct secpolicy *sp;
275 	int s, error;
276 #endif
277 
278 	/*
279 	 * make sure we don't have onion peering information into m_tag.
280 	 */
281 	ip6_delaux(m);
282 
283 	/*
284 	 * mbuf statistics
285 	 */
286 	if (m->m_flags & M_EXT) {
287 		if (m->m_next)
288 			IP6_STATINC(IP6_STAT_MEXT2M);
289 		else
290 			IP6_STATINC(IP6_STAT_MEXT1);
291 	} else {
292 #define M2MMAX	32
293 		if (m->m_next) {
294 			if (m->m_flags & M_LOOP) {
295 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
296 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX) {
297 				IP6_STATINC(IP6_STAT_M2M +
298 					    m->m_pkthdr.rcvif->if_index);
299 			} else
300 				IP6_STATINC(IP6_STAT_M2M);
301 		} else
302 			IP6_STATINC(IP6_STAT_M1);
303 #undef M2MMAX
304 	}
305 
306 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
307 	IP6_STATINC(IP6_STAT_TOTAL);
308 
309 	/*
310 	 * If the IPv6 header is not aligned, slurp it up into a new
311 	 * mbuf with space for link headers, in the event we forward
312 	 * it.  Otherwise, if it is aligned, make sure the entire base
313 	 * IPv6 header is in the first mbuf of the chain.
314 	 */
315 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
316 		struct ifnet *inifp = m->m_pkthdr.rcvif;
317 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
318 				  (max_linkhdr + 3) & ~3)) == NULL) {
319 			/* XXXJRT new stat, please */
320 			IP6_STATINC(IP6_STAT_TOOSMALL);
321 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
322 			return;
323 		}
324 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
325 		struct ifnet *inifp = m->m_pkthdr.rcvif;
326 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
327 			IP6_STATINC(IP6_STAT_TOOSMALL);
328 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
329 			return;
330 		}
331 	}
332 
333 	ip6 = mtod(m, struct ip6_hdr *);
334 
335 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
336 		IP6_STATINC(IP6_STAT_BADVERS);
337 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
338 		goto bad;
339 	}
340 
341 	/*
342 	 * Assume that we can create a fast-forward IP flow entry
343 	 * based on this packet.
344 	 */
345 	m->m_flags |= M_CANFASTFWD;
346 
347 #ifdef PFIL_HOOKS
348 	/*
349 	 * Run through list of hooks for input packets.  If there are any
350 	 * filters which require that additional packets in the flow are
351 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
352 	 * Note that filters must _never_ set this flag, as another filter
353 	 * in the list may have previously cleared it.
354 	 */
355 	/*
356 	 * let ipfilter look at packet on the wire,
357 	 * not the decapsulated packet.
358 	 */
359 #if defined(FAST_IPSEC)
360 	if (!ipsec_indone(m))
361 #else
362 	if (1)
363 #endif
364 	{
365 		struct in6_addr odst;
366 
367 		odst = ip6->ip6_dst;
368 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
369 				   PFIL_IN) != 0)
370 			return;
371 		if (m == NULL)
372 			return;
373 		ip6 = mtod(m, struct ip6_hdr *);
374 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
375 	}
376 #endif /* PFIL_HOOKS */
377 
378 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
379 
380 #ifdef ALTQ
381 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
382 		/* packet is dropped by traffic conditioner */
383 		return;
384 	}
385 #endif
386 
387 	/*
388 	 * Check against address spoofing/corruption.
389 	 */
390 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
391 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
392 		/*
393 		 * XXX: "badscope" is not very suitable for a multicast source.
394 		 */
395 		IP6_STATINC(IP6_STAT_BADSCOPE);
396 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
397 		goto bad;
398 	}
399 	/*
400 	 * The following check is not documented in specs.  A malicious
401 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
402 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
403 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
404 	 *
405 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
406 	 * support IPv4-less kernel compilation, we cannot support SIIT
407 	 * environment at all.  So, it makes more sense for us to reject any
408 	 * malicious packets for non-SIIT environment, than try to do a
409 	 * partial support for SIIT environment.
410 	 */
411 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
412 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
413 		IP6_STATINC(IP6_STAT_BADSCOPE);
414 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
415 		goto bad;
416 	}
417 #if 0
418 	/*
419 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
420 	 *
421 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
422 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
423 	 * is revised to forbid relaying case.
424 	 */
425 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
426 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
427 		IP6_STATINC(IP6_STAT_BADSCOPE);
428 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
429 		goto bad;
430 	}
431 #endif
432 
433 	/*
434 	 * Disambiguate address scope zones (if there is ambiguity).
435 	 * We first make sure that the original source or destination address
436 	 * is not in our internal form for scoped addresses.  Such addresses
437 	 * are not necessarily invalid spec-wise, but we cannot accept them due
438 	 * to the usage conflict.
439 	 * in6_setscope() then also checks and rejects the cases where src or
440 	 * dst are the loopback address and the receiving interface
441 	 * is not loopback.
442 	 */
443 	if (__predict_false(
444 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
445 		goto bad;
446 	ip6 = mtod(m, struct ip6_hdr *);
447 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
448 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
449 		goto bad;
450 	}
451 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
452 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
453 		IP6_STATINC(IP6_STAT_BADSCOPE);
454 		goto bad;
455 	}
456 
457 	/*
458 	 * Multicast check
459 	 */
460 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
461 	  	struct	in6_multi *in6m = 0;
462 
463 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
464 		/*
465 		 * See if we belong to the destination multicast group on the
466 		 * arrival interface.
467 		 */
468 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
469 		if (in6m)
470 			ours = 1;
471 		else if (!ip6_mrouter) {
472 			uint64_t *ip6s = IP6_STAT_GETREF();
473 			ip6s[IP6_STAT_NOTMEMBER]++;
474 			ip6s[IP6_STAT_CANTFORWARD]++;
475 			IP6_STAT_PUTREF();
476 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
477 			goto bad;
478 		}
479 		deliverifp = m->m_pkthdr.rcvif;
480 		goto hbhcheck;
481 	}
482 
483 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
484 
485 	/*
486 	 *  Unicast check
487 	 */
488 	rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit);
489 	if (hit)
490 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
491 	else
492 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
493 
494 #define rt6_getkey(__rt) satocsin6(rt_getkey(__rt))
495 
496 	/*
497 	 * Accept the packet if the forwarding interface to the destination
498 	 * according to the routing table is the loopback interface,
499 	 * unless the associated route has a gateway.
500 	 * Note that this approach causes to accept a packet if there is a
501 	 * route to the loopback interface for the destination of the packet.
502 	 * But we think it's even useful in some situations, e.g. when using
503 	 * a special daemon which wants to intercept the packet.
504 	 */
505 	if (rt != NULL &&
506 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
507 	    !(rt->rt_flags & RTF_CLONED) &&
508 #if 0
509 	    /*
510 	     * The check below is redundant since the comparison of
511 	     * the destination and the key of the rtentry has
512 	     * already done through looking up the routing table.
513 	     */
514 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(rt)->sin6_addr) &&
515 #endif
516 	    rt->rt_ifp->if_type == IFT_LOOP) {
517 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
518 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
519 			m->m_flags |= M_ANYCAST6;
520 		/*
521 		 * packets to a tentative, duplicated, or somehow invalid
522 		 * address must not be accepted.
523 		 */
524 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
525 			/* this address is ready */
526 			ours = 1;
527 			deliverifp = ia6->ia_ifp;	/* correct? */
528 			goto hbhcheck;
529 		} else {
530 			/* address is not ready, so discard the packet. */
531 			nd6log((LOG_INFO,
532 			    "ip6_input: packet to an unready address %s->%s\n",
533 			    ip6_sprintf(&ip6->ip6_src),
534 			    ip6_sprintf(&ip6->ip6_dst)));
535 
536 			goto bad;
537 		}
538 	}
539 
540 	/*
541 	 * FAITH (Firewall Aided Internet Translator)
542 	 */
543 #if defined(NFAITH) && 0 < NFAITH
544 	if (ip6_keepfaith) {
545 		if (rt != NULL && rt->rt_ifp != NULL &&
546 		    rt->rt_ifp->if_type == IFT_FAITH) {
547 			/* XXX do we need more sanity checks? */
548 			ours = 1;
549 			deliverifp = rt->rt_ifp; /* faith */
550 			goto hbhcheck;
551 		}
552 	}
553 #endif
554 
555 #if 0
556     {
557 	/*
558 	 * Last resort: check in6_ifaddr for incoming interface.
559 	 * The code is here until I update the "goto ours hack" code above
560 	 * working right.
561 	 */
562 	struct ifaddr *ifa;
563 	IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
564 		if (ifa->ifa_addr == NULL)
565 			continue;	/* just for safety */
566 		if (ifa->ifa_addr->sa_family != AF_INET6)
567 			continue;
568 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
569 			ours = 1;
570 			deliverifp = ifa->ifa_ifp;
571 			goto hbhcheck;
572 		}
573 	}
574     }
575 #endif
576 
577 	/*
578 	 * Now there is no reason to process the packet if it's not our own
579 	 * and we're not a router.
580 	 */
581 	if (!ip6_forwarding) {
582 		IP6_STATINC(IP6_STAT_CANTFORWARD);
583 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
584 		goto bad;
585 	}
586 
587   hbhcheck:
588 	/*
589 	 * record address information into m_tag, if we don't have one yet.
590 	 * note that we are unable to record it, if the address is not listed
591 	 * as our interface address (e.g. multicast addresses, addresses
592 	 * within FAITH prefixes and such).
593 	 */
594 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
595 		struct in6_ifaddr *ia6;
596 
597 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
598 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
599 			/*
600 			 * XXX maybe we should drop the packet here,
601 			 * as we could not provide enough information
602 			 * to the upper layers.
603 			 */
604 		}
605 	}
606 
607 	/*
608 	 * Process Hop-by-Hop options header if it's contained.
609 	 * m may be modified in ip6_hopopts_input().
610 	 * If a JumboPayload option is included, plen will also be modified.
611 	 */
612 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
613 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
614 		struct ip6_hbh *hbh;
615 
616 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
617 #if 0	/*touches NULL pointer*/
618 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
619 #endif
620 			return;	/* m have already been freed */
621 		}
622 
623 		/* adjust pointer */
624 		ip6 = mtod(m, struct ip6_hdr *);
625 
626 		/*
627 		 * if the payload length field is 0 and the next header field
628 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
629 		 * option MUST be included.
630 		 */
631 		if (ip6->ip6_plen == 0 && plen == 0) {
632 			/*
633 			 * Note that if a valid jumbo payload option is
634 			 * contained, ip6_hopopts_input() must set a valid
635 			 * (non-zero) payload length to the variable plen.
636 			 */
637 			IP6_STATINC(IP6_STAT_BADOPTIONS);
638 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
639 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
640 			icmp6_error(m, ICMP6_PARAM_PROB,
641 				    ICMP6_PARAMPROB_HEADER,
642 				    (char *)&ip6->ip6_plen - (char *)ip6);
643 			return;
644 		}
645 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
646 			sizeof(struct ip6_hbh));
647 		if (hbh == NULL) {
648 			IP6_STATINC(IP6_STAT_TOOSHORT);
649 			return;
650 		}
651 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
652 		nxt = hbh->ip6h_nxt;
653 
654 		/*
655 		 * accept the packet if a router alert option is included
656 		 * and we act as an IPv6 router.
657 		 */
658 		if (rtalert != ~0 && ip6_forwarding)
659 			ours = 1;
660 	} else
661 		nxt = ip6->ip6_nxt;
662 
663 	/*
664 	 * Check that the amount of data in the buffers
665 	 * is as at least much as the IPv6 header would have us expect.
666 	 * Trim mbufs if longer than we expect.
667 	 * Drop packet if shorter than we expect.
668 	 */
669 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
670 		IP6_STATINC(IP6_STAT_TOOSHORT);
671 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
672 		goto bad;
673 	}
674 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
675 		if (m->m_len == m->m_pkthdr.len) {
676 			m->m_len = sizeof(struct ip6_hdr) + plen;
677 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
678 		} else
679 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
680 	}
681 
682 	/*
683 	 * Forward if desirable.
684 	 */
685 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
686 		/*
687 		 * If we are acting as a multicast router, all
688 		 * incoming multicast packets are passed to the
689 		 * kernel-level multicast forwarding function.
690 		 * The packet is returned (relatively) intact; if
691 		 * ip6_mforward() returns a non-zero value, the packet
692 		 * must be discarded, else it may be accepted below.
693 		 */
694 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
695 			IP6_STATINC(IP6_STAT_CANTFORWARD);
696 			m_freem(m);
697 			return;
698 		}
699 		if (!ours) {
700 			m_freem(m);
701 			return;
702 		}
703 	} else if (!ours) {
704 		ip6_forward(m, srcrt);
705 		return;
706 	}
707 
708 	ip6 = mtod(m, struct ip6_hdr *);
709 
710 	/*
711 	 * Malicious party may be able to use IPv4 mapped addr to confuse
712 	 * tcp/udp stack and bypass security checks (act as if it was from
713 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
714 	 *
715 	 * For SIIT end node behavior, you may want to disable the check.
716 	 * However, you will  become vulnerable to attacks using IPv4 mapped
717 	 * source.
718 	 */
719 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
720 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
721 		IP6_STATINC(IP6_STAT_BADSCOPE);
722 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
723 		goto bad;
724 	}
725 
726 	/*
727 	 * Tell launch routine the next header
728 	 */
729 #ifdef IFA_STATS
730 	if (deliverifp != NULL) {
731 		struct in6_ifaddr *ia6;
732 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
733 		if (ia6)
734 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
735 	}
736 #endif
737 	IP6_STATINC(IP6_STAT_DELIVERED);
738 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
739 	nest = 0;
740 
741 	rh_present = 0;
742 	while (nxt != IPPROTO_DONE) {
743 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
744 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
745 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
746 			goto bad;
747 		}
748 
749 		/*
750 		 * protection against faulty packet - there should be
751 		 * more sanity checks in header chain processing.
752 		 */
753 		if (m->m_pkthdr.len < off) {
754 			IP6_STATINC(IP6_STAT_TOOSHORT);
755 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
756 			goto bad;
757 		}
758 
759 		if (nxt == IPPROTO_ROUTING) {
760 			if (rh_present++) {
761 				in6_ifstat_inc(m->m_pkthdr.rcvif,
762 				    ifs6_in_hdrerr);
763 				IP6_STATINC(IP6_STAT_BADOPTIONS);
764 				goto bad;
765 			}
766 		}
767 
768 #ifdef FAST_IPSEC
769 	/*
770 	 * enforce IPsec policy checking if we are seeing last header.
771 	 * note that we do not visit this with protocols with pcb layer
772 	 * code - like udp/tcp/raw ip.
773 	 */
774 	if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) {
775 		/*
776 		 * Check if the packet has already had IPsec processing
777 		 * done.  If so, then just pass it along.  This tag gets
778 		 * set during AH, ESP, etc. input handling, before the
779 		 * packet is returned to the ip input queue for delivery.
780 		 */
781 		mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
782 		s = splsoftnet();
783 		if (mtag != NULL) {
784 			tdbi = (struct tdb_ident *)(mtag + 1);
785 			sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
786 		} else {
787 			sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
788 									IP_FORWARDING, &error);
789 		}
790 		if (sp != NULL) {
791 			/*
792 			 * Check security policy against packet attributes.
793 			 */
794 			error = ipsec_in_reject(sp, m);
795 			KEY_FREESP(&sp);
796 		} else {
797 			/* XXX error stat??? */
798 			error = EINVAL;
799 			DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/
800 		}
801 		splx(s);
802 		if (error)
803 			goto bad;
804 	}
805 #endif /* FAST_IPSEC */
806 
807 
808 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
809 	}
810 	return;
811  bad:
812 	m_freem(m);
813 }
814 
815 /*
816  * set/grab in6_ifaddr correspond to IPv6 destination address.
817  */
818 static struct m_tag *
819 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
820 {
821 	struct m_tag *mtag;
822 
823 	mtag = ip6_addaux(m);
824 	if (mtag != NULL) {
825 		struct ip6aux *ip6a;
826 
827 		ip6a = (struct ip6aux *)(mtag + 1);
828 		in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id);
829 		ip6a->ip6a_src = ia->ia_addr.sin6_addr;
830 		ip6a->ip6a_flags = ia->ia6_flags;
831 	}
832 	return mtag;	/* NULL if failed to set */
833 }
834 
835 const struct ip6aux *
836 ip6_getdstifaddr(struct mbuf *m)
837 {
838 	struct m_tag *mtag;
839 
840 	mtag = ip6_findaux(m);
841 	if (mtag != NULL)
842 		return (struct ip6aux *)(mtag + 1);
843 	else
844 		return NULL;
845 }
846 
847 /*
848  * Hop-by-Hop options header processing. If a valid jumbo payload option is
849  * included, the real payload length will be stored in plenp.
850  *
851  * rtalertp - XXX: should be stored more smart way
852  */
853 int
854 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
855 	struct mbuf **mp, int *offp)
856 {
857 	struct mbuf *m = *mp;
858 	int off = *offp, hbhlen;
859 	struct ip6_hbh *hbh;
860 
861 	/* validation of the length of the header */
862 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
863 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
864 	if (hbh == NULL) {
865 		IP6_STATINC(IP6_STAT_TOOSHORT);
866 		return -1;
867 	}
868 	hbhlen = (hbh->ip6h_len + 1) << 3;
869 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
870 		hbhlen);
871 	if (hbh == NULL) {
872 		IP6_STATINC(IP6_STAT_TOOSHORT);
873 		return -1;
874 	}
875 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
876 	off += hbhlen;
877 	hbhlen -= sizeof(struct ip6_hbh);
878 
879 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
880 				hbhlen, rtalertp, plenp) < 0)
881 		return (-1);
882 
883 	*offp = off;
884 	*mp = m;
885 	return (0);
886 }
887 
888 /*
889  * Search header for all Hop-by-hop options and process each option.
890  * This function is separate from ip6_hopopts_input() in order to
891  * handle a case where the sending node itself process its hop-by-hop
892  * options header. In such a case, the function is called from ip6_output().
893  *
894  * The function assumes that hbh header is located right after the IPv6 header
895  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
896  * opthead + hbhlen is located in continuous memory region.
897  */
898 static int
899 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
900 	u_int32_t *rtalertp, u_int32_t *plenp)
901 {
902 	struct ip6_hdr *ip6;
903 	int optlen = 0;
904 	u_int8_t *opt = opthead;
905 	u_int16_t rtalert_val;
906 	u_int32_t jumboplen;
907 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
908 
909 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
910 		switch (*opt) {
911 		case IP6OPT_PAD1:
912 			optlen = 1;
913 			break;
914 		case IP6OPT_PADN:
915 			if (hbhlen < IP6OPT_MINLEN) {
916 				IP6_STATINC(IP6_STAT_TOOSMALL);
917 				goto bad;
918 			}
919 			optlen = *(opt + 1) + 2;
920 			break;
921 		case IP6OPT_RTALERT:
922 			/* XXX may need check for alignment */
923 			if (hbhlen < IP6OPT_RTALERT_LEN) {
924 				IP6_STATINC(IP6_STAT_TOOSMALL);
925 				goto bad;
926 			}
927 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
928 				/* XXX stat */
929 				icmp6_error(m, ICMP6_PARAM_PROB,
930 				    ICMP6_PARAMPROB_HEADER,
931 				    erroff + opt + 1 - opthead);
932 				return (-1);
933 			}
934 			optlen = IP6OPT_RTALERT_LEN;
935 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
936 			*rtalertp = ntohs(rtalert_val);
937 			break;
938 		case IP6OPT_JUMBO:
939 			/* XXX may need check for alignment */
940 			if (hbhlen < IP6OPT_JUMBO_LEN) {
941 				IP6_STATINC(IP6_STAT_TOOSMALL);
942 				goto bad;
943 			}
944 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
945 				/* XXX stat */
946 				icmp6_error(m, ICMP6_PARAM_PROB,
947 				    ICMP6_PARAMPROB_HEADER,
948 				    erroff + opt + 1 - opthead);
949 				return (-1);
950 			}
951 			optlen = IP6OPT_JUMBO_LEN;
952 
953 			/*
954 			 * IPv6 packets that have non 0 payload length
955 			 * must not contain a jumbo payload option.
956 			 */
957 			ip6 = mtod(m, struct ip6_hdr *);
958 			if (ip6->ip6_plen) {
959 				IP6_STATINC(IP6_STAT_BADOPTIONS);
960 				icmp6_error(m, ICMP6_PARAM_PROB,
961 				    ICMP6_PARAMPROB_HEADER,
962 				    erroff + opt - opthead);
963 				return (-1);
964 			}
965 
966 			/*
967 			 * We may see jumbolen in unaligned location, so
968 			 * we'd need to perform bcopy().
969 			 */
970 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
971 			jumboplen = (u_int32_t)htonl(jumboplen);
972 
973 #if 1
974 			/*
975 			 * if there are multiple jumbo payload options,
976 			 * *plenp will be non-zero and the packet will be
977 			 * rejected.
978 			 * the behavior may need some debate in ipngwg -
979 			 * multiple options does not make sense, however,
980 			 * there's no explicit mention in specification.
981 			 */
982 			if (*plenp != 0) {
983 				IP6_STATINC(IP6_STAT_BADOPTIONS);
984 				icmp6_error(m, ICMP6_PARAM_PROB,
985 				    ICMP6_PARAMPROB_HEADER,
986 				    erroff + opt + 2 - opthead);
987 				return (-1);
988 			}
989 #endif
990 
991 			/*
992 			 * jumbo payload length must be larger than 65535.
993 			 */
994 			if (jumboplen <= IPV6_MAXPACKET) {
995 				IP6_STATINC(IP6_STAT_BADOPTIONS);
996 				icmp6_error(m, ICMP6_PARAM_PROB,
997 				    ICMP6_PARAMPROB_HEADER,
998 				    erroff + opt + 2 - opthead);
999 				return (-1);
1000 			}
1001 			*plenp = jumboplen;
1002 
1003 			break;
1004 		default:		/* unknown option */
1005 			if (hbhlen < IP6OPT_MINLEN) {
1006 				IP6_STATINC(IP6_STAT_TOOSMALL);
1007 				goto bad;
1008 			}
1009 			optlen = ip6_unknown_opt(opt, m,
1010 			    erroff + opt - opthead);
1011 			if (optlen == -1)
1012 				return (-1);
1013 			optlen += 2;
1014 			break;
1015 		}
1016 	}
1017 
1018 	return (0);
1019 
1020   bad:
1021 	m_freem(m);
1022 	return (-1);
1023 }
1024 
1025 /*
1026  * Unknown option processing.
1027  * The third argument `off' is the offset from the IPv6 header to the option,
1028  * which is necessary if the IPv6 header the and option header and IPv6 header
1029  * is not continuous in order to return an ICMPv6 error.
1030  */
1031 int
1032 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1033 {
1034 	struct ip6_hdr *ip6;
1035 
1036 	switch (IP6OPT_TYPE(*optp)) {
1037 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1038 		return ((int)*(optp + 1));
1039 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1040 		m_freem(m);
1041 		return (-1);
1042 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1043 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1044 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1045 		return (-1);
1046 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1047 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1048 		ip6 = mtod(m, struct ip6_hdr *);
1049 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1050 		    (m->m_flags & (M_BCAST|M_MCAST)))
1051 			m_freem(m);
1052 		else
1053 			icmp6_error(m, ICMP6_PARAM_PROB,
1054 				    ICMP6_PARAMPROB_OPTION, off);
1055 		return (-1);
1056 	}
1057 
1058 	m_freem(m);		/* XXX: NOTREACHED */
1059 	return (-1);
1060 }
1061 
1062 /*
1063  * Create the "control" list for this pcb.
1064  *
1065  * The routine will be called from upper layer handlers like tcp6_input().
1066  * Thus the routine assumes that the caller (tcp6_input) have already
1067  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1068  * very first mbuf on the mbuf chain.
1069  * We may want to add some infinite loop prevention or sanity checks for safety.
1070  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1071  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1072  */
1073 void
1074 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
1075 	struct ip6_hdr *ip6, struct mbuf *m)
1076 {
1077 #ifdef RFC2292
1078 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1079 #else
1080 #define IS2292(x, y)	(y)
1081 #endif
1082 
1083 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP
1084 #ifdef SO_OTIMESTAMP
1085 	    || in6p->in6p_socket->so_options & SO_OTIMESTAMP
1086 #endif
1087 	) {
1088 		struct timeval tv;
1089 
1090 		microtime(&tv);
1091 #ifdef SO_OTIMESTAMP
1092 		if (in6p->in6p_socket->so_options & SO_OTIMESTAMP) {
1093 			struct timeval50 tv50;
1094 			timeval_to_timeval50(&tv, &tv50);
1095 			*mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
1096 			    SCM_OTIMESTAMP, SOL_SOCKET);
1097 		} else
1098 #endif
1099 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1100 		    SCM_TIMESTAMP, SOL_SOCKET);
1101 		if (*mp)
1102 			mp = &(*mp)->m_next;
1103 	}
1104 
1105 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
1106 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1107 		return;
1108 
1109 	/* RFC 2292 sec. 5 */
1110 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1111 		struct in6_pktinfo pi6;
1112 
1113 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
1114 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1115 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
1116 		    m->m_pkthdr.rcvif->if_index : 0;
1117 		*mp = sbcreatecontrol((void *) &pi6,
1118 		    sizeof(struct in6_pktinfo),
1119 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1120 		if (*mp)
1121 			mp = &(*mp)->m_next;
1122 	}
1123 
1124 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1125 		int hlim = ip6->ip6_hlim & 0xff;
1126 
1127 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
1128 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1129 		if (*mp)
1130 			mp = &(*mp)->m_next;
1131 	}
1132 
1133 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1134 		u_int32_t flowinfo;
1135 		int tclass;
1136 
1137 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1138 		flowinfo >>= 20;
1139 
1140 		tclass = flowinfo & 0xff;
1141 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
1142 		    IPV6_TCLASS, IPPROTO_IPV6);
1143 
1144 		if (*mp)
1145 			mp = &(*mp)->m_next;
1146 	}
1147 
1148 	/*
1149 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1150 	 * privilege for the option (see ip6_ctloutput), but it might be too
1151 	 * strict, since there might be some hop-by-hop options which can be
1152 	 * returned to normal user.
1153 	 * See also RFC3542 section 8 (or RFC2292 section 6).
1154 	 */
1155 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1156 		/*
1157 		 * Check if a hop-by-hop options header is contatined in the
1158 		 * received packet, and if so, store the options as ancillary
1159 		 * data. Note that a hop-by-hop options header must be
1160 		 * just after the IPv6 header, which fact is assured through
1161 		 * the IPv6 input processing.
1162 		 */
1163 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1164 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1165 			struct ip6_hbh *hbh;
1166 			int hbhlen;
1167 			struct mbuf *ext;
1168 
1169 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1170 			    xip6->ip6_nxt);
1171 			if (ext == NULL) {
1172 				IP6_STATINC(IP6_STAT_TOOSHORT);
1173 				return;
1174 			}
1175 			hbh = mtod(ext, struct ip6_hbh *);
1176 			hbhlen = (hbh->ip6h_len + 1) << 3;
1177 			if (hbhlen != ext->m_len) {
1178 				m_freem(ext);
1179 				IP6_STATINC(IP6_STAT_TOOSHORT);
1180 				return;
1181 			}
1182 
1183 			/*
1184 			 * XXX: We copy whole the header even if a jumbo
1185 			 * payload option is included, which option is to
1186 			 * be removed before returning in the RFC 2292.
1187 			 * Note: this constraint is removed in RFC3542.
1188 			 */
1189 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
1190 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1191 			    IPPROTO_IPV6);
1192 			if (*mp)
1193 				mp = &(*mp)->m_next;
1194 			m_freem(ext);
1195 		}
1196 	}
1197 
1198 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1199 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1200 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1201 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1202 
1203 		/*
1204 		 * Search for destination options headers or routing
1205 		 * header(s) through the header chain, and stores each
1206 		 * header as ancillary data.
1207 		 * Note that the order of the headers remains in
1208 		 * the chain of ancillary data.
1209 		 */
1210 		for (;;) {	/* is explicit loop prevention necessary? */
1211 			struct ip6_ext *ip6e = NULL;
1212 			int elen;
1213 			struct mbuf *ext = NULL;
1214 
1215 			/*
1216 			 * if it is not an extension header, don't try to
1217 			 * pull it from the chain.
1218 			 */
1219 			switch (nxt) {
1220 			case IPPROTO_DSTOPTS:
1221 			case IPPROTO_ROUTING:
1222 			case IPPROTO_HOPOPTS:
1223 			case IPPROTO_AH: /* is it possible? */
1224 				break;
1225 			default:
1226 				goto loopend;
1227 			}
1228 
1229 			ext = ip6_pullexthdr(m, off, nxt);
1230 			if (ext == NULL) {
1231 				IP6_STATINC(IP6_STAT_TOOSHORT);
1232 				return;
1233 			}
1234 			ip6e = mtod(ext, struct ip6_ext *);
1235 			if (nxt == IPPROTO_AH)
1236 				elen = (ip6e->ip6e_len + 2) << 2;
1237 			else
1238 				elen = (ip6e->ip6e_len + 1) << 3;
1239 			if (elen != ext->m_len) {
1240 				m_freem(ext);
1241 				IP6_STATINC(IP6_STAT_TOOSHORT);
1242 				return;
1243 			}
1244 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1245 
1246 			switch (nxt) {
1247 			case IPPROTO_DSTOPTS:
1248 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1249 					break;
1250 
1251 				*mp = sbcreatecontrol((void *)ip6e, elen,
1252 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1253 				    IPPROTO_IPV6);
1254 				if (*mp)
1255 					mp = &(*mp)->m_next;
1256 				break;
1257 
1258 			case IPPROTO_ROUTING:
1259 				if (!(in6p->in6p_flags & IN6P_RTHDR))
1260 					break;
1261 
1262 				*mp = sbcreatecontrol((void *)ip6e, elen,
1263 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1264 				    IPPROTO_IPV6);
1265 				if (*mp)
1266 					mp = &(*mp)->m_next;
1267 				break;
1268 
1269 			case IPPROTO_HOPOPTS:
1270 			case IPPROTO_AH: /* is it possible? */
1271 				break;
1272 
1273 			default:
1274 				/*
1275 			 	 * other cases have been filtered in the above.
1276 				 * none will visit this case.  here we supply
1277 				 * the code just in case (nxt overwritten or
1278 				 * other cases).
1279 				 */
1280 				m_freem(ext);
1281 				goto loopend;
1282 
1283 			}
1284 
1285 			/* proceed with the next header. */
1286 			off += elen;
1287 			nxt = ip6e->ip6e_nxt;
1288 			ip6e = NULL;
1289 			m_freem(ext);
1290 			ext = NULL;
1291 		}
1292 	  loopend:
1293 	  	;
1294 	}
1295 }
1296 #undef IS2292
1297 
1298 
1299 void
1300 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
1301     uint32_t *mtu)
1302 {
1303 	struct socket *so;
1304 	struct mbuf *m_mtu;
1305 	struct ip6_mtuinfo mtuctl;
1306 
1307 	so = in6p->in6p_socket;
1308 
1309 	if (mtu == NULL)
1310 		return;
1311 
1312 #ifdef DIAGNOSTIC
1313 	if (so == NULL)		/* I believe this is impossible */
1314 		panic("ip6_notify_pmtu: socket is NULL");
1315 #endif
1316 
1317 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
1318 	mtuctl.ip6m_mtu = *mtu;
1319 	mtuctl.ip6m_addr = *dst;
1320 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
1321 		return;
1322 
1323 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
1324 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1325 		return;
1326 
1327 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
1328 	    == 0) {
1329 		m_freem(m_mtu);
1330 		/* XXX: should count statistics */
1331 	} else
1332 		sorwakeup(so);
1333 
1334 	return;
1335 }
1336 
1337 /*
1338  * pull single extension header from mbuf chain.  returns single mbuf that
1339  * contains the result, or NULL on error.
1340  */
1341 static struct mbuf *
1342 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1343 {
1344 	struct ip6_ext ip6e;
1345 	size_t elen;
1346 	struct mbuf *n;
1347 
1348 #ifdef DIAGNOSTIC
1349 	switch (nxt) {
1350 	case IPPROTO_DSTOPTS:
1351 	case IPPROTO_ROUTING:
1352 	case IPPROTO_HOPOPTS:
1353 	case IPPROTO_AH: /* is it possible? */
1354 		break;
1355 	default:
1356 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1357 	}
1358 #endif
1359 
1360 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1361 	if (nxt == IPPROTO_AH)
1362 		elen = (ip6e.ip6e_len + 2) << 2;
1363 	else
1364 		elen = (ip6e.ip6e_len + 1) << 3;
1365 
1366 	MGET(n, M_DONTWAIT, MT_DATA);
1367 	if (n && elen >= MLEN) {
1368 		MCLGET(n, M_DONTWAIT);
1369 		if ((n->m_flags & M_EXT) == 0) {
1370 			m_free(n);
1371 			n = NULL;
1372 		}
1373 	}
1374 	if (!n)
1375 		return NULL;
1376 
1377 	n->m_len = 0;
1378 	if (elen >= M_TRAILINGSPACE(n)) {
1379 		m_free(n);
1380 		return NULL;
1381 	}
1382 
1383 	m_copydata(m, off, elen, mtod(n, void *));
1384 	n->m_len = elen;
1385 	return n;
1386 }
1387 
1388 /*
1389  * Get pointer to the previous header followed by the header
1390  * currently processed.
1391  * XXX: This function supposes that
1392  *	M includes all headers,
1393  *	the next header field and the header length field of each header
1394  *	are valid, and
1395  *	the sum of each header length equals to OFF.
1396  * Because of these assumptions, this function must be called very
1397  * carefully. Moreover, it will not be used in the near future when
1398  * we develop `neater' mechanism to process extension headers.
1399  */
1400 u_int8_t *
1401 ip6_get_prevhdr(struct mbuf *m, int off)
1402 {
1403 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1404 
1405 	if (off == sizeof(struct ip6_hdr))
1406 		return (&ip6->ip6_nxt);
1407 	else {
1408 		int len, nxt;
1409 		struct ip6_ext *ip6e = NULL;
1410 
1411 		nxt = ip6->ip6_nxt;
1412 		len = sizeof(struct ip6_hdr);
1413 		while (len < off) {
1414 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
1415 
1416 			switch (nxt) {
1417 			case IPPROTO_FRAGMENT:
1418 				len += sizeof(struct ip6_frag);
1419 				break;
1420 			case IPPROTO_AH:
1421 				len += (ip6e->ip6e_len + 2) << 2;
1422 				break;
1423 			default:
1424 				len += (ip6e->ip6e_len + 1) << 3;
1425 				break;
1426 			}
1427 			nxt = ip6e->ip6e_nxt;
1428 		}
1429 		if (ip6e)
1430 			return (&ip6e->ip6e_nxt);
1431 		else
1432 			return NULL;
1433 	}
1434 }
1435 
1436 /*
1437  * get next header offset.  m will be retained.
1438  */
1439 int
1440 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1441 {
1442 	struct ip6_hdr ip6;
1443 	struct ip6_ext ip6e;
1444 	struct ip6_frag fh;
1445 
1446 	/* just in case */
1447 	if (m == NULL)
1448 		panic("ip6_nexthdr: m == NULL");
1449 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1450 		return -1;
1451 
1452 	switch (proto) {
1453 	case IPPROTO_IPV6:
1454 		/* do not chase beyond intermediate IPv6 headers */
1455 		if (off != 0)
1456 			return -1;
1457 		if (m->m_pkthdr.len < off + sizeof(ip6))
1458 			return -1;
1459 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
1460 		if (nxtp)
1461 			*nxtp = ip6.ip6_nxt;
1462 		off += sizeof(ip6);
1463 		return off;
1464 
1465 	case IPPROTO_FRAGMENT:
1466 		/*
1467 		 * terminate parsing if it is not the first fragment,
1468 		 * it does not make sense to parse through it.
1469 		 */
1470 		if (m->m_pkthdr.len < off + sizeof(fh))
1471 			return -1;
1472 		m_copydata(m, off, sizeof(fh), (void *)&fh);
1473 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1474 			return -1;
1475 		if (nxtp)
1476 			*nxtp = fh.ip6f_nxt;
1477 		off += sizeof(struct ip6_frag);
1478 		return off;
1479 
1480 	case IPPROTO_AH:
1481 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1482 			return -1;
1483 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1484 		if (nxtp)
1485 			*nxtp = ip6e.ip6e_nxt;
1486 		off += (ip6e.ip6e_len + 2) << 2;
1487 		if (m->m_pkthdr.len < off)
1488 			return -1;
1489 		return off;
1490 
1491 	case IPPROTO_HOPOPTS:
1492 	case IPPROTO_ROUTING:
1493 	case IPPROTO_DSTOPTS:
1494 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1495 			return -1;
1496 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1497 		if (nxtp)
1498 			*nxtp = ip6e.ip6e_nxt;
1499 		off += (ip6e.ip6e_len + 1) << 3;
1500 		if (m->m_pkthdr.len < off)
1501 			return -1;
1502 		return off;
1503 
1504 	case IPPROTO_NONE:
1505 	case IPPROTO_ESP:
1506 	case IPPROTO_IPCOMP:
1507 		/* give up */
1508 		return -1;
1509 
1510 	default:
1511 		return -1;
1512 	}
1513 }
1514 
1515 /*
1516  * get offset for the last header in the chain.  m will be kept untainted.
1517  */
1518 int
1519 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1520 {
1521 	int newoff;
1522 	int nxt;
1523 
1524 	if (!nxtp) {
1525 		nxt = -1;
1526 		nxtp = &nxt;
1527 	}
1528 	for (;;) {
1529 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1530 		if (newoff < 0)
1531 			return off;
1532 		else if (newoff < off)
1533 			return -1;	/* invalid */
1534 		else if (newoff == off)
1535 			return newoff;
1536 
1537 		off = newoff;
1538 		proto = *nxtp;
1539 	}
1540 }
1541 
1542 struct m_tag *
1543 ip6_addaux(struct mbuf *m)
1544 {
1545 	struct m_tag *mtag;
1546 
1547 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1548 	if (!mtag) {
1549 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1550 		    M_NOWAIT);
1551 		if (mtag) {
1552 			m_tag_prepend(m, mtag);
1553 			memset(mtag + 1, 0, sizeof(struct ip6aux));
1554 		}
1555 	}
1556 	return mtag;
1557 }
1558 
1559 struct m_tag *
1560 ip6_findaux(struct mbuf *m)
1561 {
1562 	struct m_tag *mtag;
1563 
1564 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1565 	return mtag;
1566 }
1567 
1568 void
1569 ip6_delaux(struct mbuf *m)
1570 {
1571 	struct m_tag *mtag;
1572 
1573 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1574 	if (mtag)
1575 		m_tag_delete(m, mtag);
1576 }
1577 
1578 #ifdef GATEWAY
1579 /*
1580  * sysctl helper routine for net.inet.ip6.maxflows. Since
1581  * we could reduce this value, call ip6flow_reap();
1582  */
1583 static int
1584 sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS)
1585 {
1586 	int error;
1587 
1588 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1589 	if (error || newp == NULL)
1590 		return (error);
1591 
1592 	mutex_enter(softnet_lock);
1593 	KERNEL_LOCK(1, NULL);
1594 
1595 	ip6flow_reap(0);
1596 
1597 	KERNEL_UNLOCK_ONE(NULL);
1598 	mutex_exit(softnet_lock);
1599 
1600 	return (0);
1601 }
1602 
1603 static int
1604 sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS)
1605 {
1606 	int error, tmp;
1607 	struct sysctlnode node;
1608 
1609 	node = *rnode;
1610 	tmp = ip6_hashsize;
1611 	node.sysctl_data = &tmp;
1612 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1613 	if (error || newp == NULL)
1614 		return (error);
1615 
1616 	if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
1617 		/*
1618 		 * Can only fail due to malloc()
1619 		 */
1620 		mutex_enter(softnet_lock);
1621 		KERNEL_LOCK(1, NULL);
1622 
1623 		error = ip6flow_invalidate_all(tmp);
1624 
1625 		KERNEL_UNLOCK_ONE(NULL);
1626 		mutex_exit(softnet_lock);
1627 	} else {
1628 		/*
1629 		 * EINVAL if not a power of 2
1630 		 */
1631 		error = EINVAL;
1632 	}
1633 
1634 	return error;
1635 }
1636 #endif /* GATEWAY */
1637 
1638 /*
1639  * System control for IP6
1640  */
1641 
1642 const u_char inet6ctlerrmap[PRC_NCMDS] = {
1643 	0,		0,		0,		0,
1644 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1645 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1646 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1647 	0,		0,		0,		0,
1648 	ENOPROTOOPT
1649 };
1650 
1651 static int
1652 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
1653 {
1654 
1655 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
1656 }
1657 
1658 static void
1659 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
1660 {
1661 #ifdef RFC2292
1662 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1663 #else
1664 #define IS2292(x, y)	(y)
1665 #endif
1666 
1667 	sysctl_createv(clog, 0, NULL, NULL,
1668 		       CTLFLAG_PERMANENT,
1669 		       CTLTYPE_NODE, "net", NULL,
1670 		       NULL, 0, NULL, 0,
1671 		       CTL_NET, CTL_EOL);
1672 	sysctl_createv(clog, 0, NULL, NULL,
1673 		       CTLFLAG_PERMANENT,
1674 		       CTLTYPE_NODE, "inet6",
1675 		       SYSCTL_DESCR("PF_INET6 related settings"),
1676 		       NULL, 0, NULL, 0,
1677 		       CTL_NET, PF_INET6, CTL_EOL);
1678 	sysctl_createv(clog, 0, NULL, NULL,
1679 		       CTLFLAG_PERMANENT,
1680 		       CTLTYPE_NODE, "ip6",
1681 		       SYSCTL_DESCR("IPv6 related settings"),
1682 		       NULL, 0, NULL, 0,
1683 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1684 
1685 	sysctl_createv(clog, 0, NULL, NULL,
1686 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1687 		       CTLTYPE_INT, "forwarding",
1688 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1689 		       NULL, 0, &ip6_forwarding, 0,
1690 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1691 		       IPV6CTL_FORWARDING, CTL_EOL);
1692 	sysctl_createv(clog, 0, NULL, NULL,
1693 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1694 		       CTLTYPE_INT, "redirect",
1695 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1696 		       NULL, 0, &ip6_sendredirects, 0,
1697 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1698 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1699 	sysctl_createv(clog, 0, NULL, NULL,
1700 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1701 		       CTLTYPE_INT, "hlim",
1702 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1703 		       NULL, 0, &ip6_defhlim, 0,
1704 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1705 		       IPV6CTL_DEFHLIM, CTL_EOL);
1706 #ifdef notyet
1707 	sysctl_createv(clog, 0, NULL, NULL,
1708 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1709 		       CTLTYPE_INT, "mtu", NULL,
1710 		       NULL, 0, &, 0,
1711 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1712 		       IPV6CTL_DEFMTU, CTL_EOL);
1713 #endif
1714 #ifdef __no_idea__
1715 	sysctl_createv(clog, 0, NULL, NULL,
1716 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1717 		       CTLTYPE_INT, "forwsrcrt", NULL,
1718 		       NULL, 0, &?, 0,
1719 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1720 		       IPV6CTL_FORWSRCRT, CTL_EOL);
1721 	sysctl_createv(clog, 0, NULL, NULL,
1722 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1723 		       CTLTYPE_STRUCT, "mrtstats", NULL,
1724 		       NULL, 0, &?, sizeof(?),
1725 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1726 		       IPV6CTL_MRTSTATS, CTL_EOL);
1727 	sysctl_createv(clog, 0, NULL, NULL,
1728 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1729 		       CTLTYPE_?, "mrtproto", NULL,
1730 		       NULL, 0, &?, sizeof(?),
1731 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1732 		       IPV6CTL_MRTPROTO, CTL_EOL);
1733 #endif
1734 	sysctl_createv(clog, 0, NULL, NULL,
1735 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1736 		       CTLTYPE_INT, "maxfragpackets",
1737 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1738 				    "for reassembly"),
1739 		       NULL, 0, &ip6_maxfragpackets, 0,
1740 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1741 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1742 #ifdef __no_idea__
1743 	sysctl_createv(clog, 0, NULL, NULL,
1744 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1745 		       CTLTYPE_INT, "sourcecheck", NULL,
1746 		       NULL, 0, &?, 0,
1747 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1748 		       IPV6CTL_SOURCECHECK, CTL_EOL);
1749 	sysctl_createv(clog, 0, NULL, NULL,
1750 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1751 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
1752 		       NULL, 0, &?, 0,
1753 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1754 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
1755 #endif
1756 	sysctl_createv(clog, 0, NULL, NULL,
1757 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1758 		       CTLTYPE_INT, "accept_rtadv",
1759 		       SYSCTL_DESCR("Accept router advertisements"),
1760 		       NULL, 0, &ip6_accept_rtadv, 0,
1761 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1762 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1763 	sysctl_createv(clog, 0, NULL, NULL,
1764 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1765 		       CTLTYPE_INT, "rtadv_maxroutes",
1766 		       SYSCTL_DESCR("Maximum number of routes accepted via router advertisements"),
1767 		       NULL, 0, &ip6_rtadv_maxroutes, 0,
1768 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1769 		       IPV6CTL_RTADV_MAXROUTES, CTL_EOL);
1770 	sysctl_createv(clog, 0, NULL, NULL,
1771 		       CTLFLAG_PERMANENT,
1772 		       CTLTYPE_INT, "rtadv_numroutes",
1773 		       SYSCTL_DESCR("Current number of routes accepted via router advertisements"),
1774 		       NULL, 0, &nd6_numroutes, 0,
1775 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1776 		       IPV6CTL_RTADV_NUMROUTES, CTL_EOL);
1777 	sysctl_createv(clog, 0, NULL, NULL,
1778 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1779 		       CTLTYPE_INT, "keepfaith",
1780 		       SYSCTL_DESCR("Activate faith interface"),
1781 		       NULL, 0, &ip6_keepfaith, 0,
1782 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1783 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1784 	sysctl_createv(clog, 0, NULL, NULL,
1785 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1786 		       CTLTYPE_INT, "log_interval",
1787 		       SYSCTL_DESCR("Minumum interval between logging "
1788 				    "unroutable packets"),
1789 		       NULL, 0, &ip6_log_interval, 0,
1790 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1791 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1792 	sysctl_createv(clog, 0, NULL, NULL,
1793 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1794 		       CTLTYPE_INT, "hdrnestlimit",
1795 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1796 		       NULL, 0, &ip6_hdrnestlimit, 0,
1797 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1798 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1799 	sysctl_createv(clog, 0, NULL, NULL,
1800 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1801 		       CTLTYPE_INT, "dad_count",
1802 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1803 				    "probes to send"),
1804 		       NULL, 0, &ip6_dad_count, 0,
1805 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1806 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1807 	sysctl_createv(clog, 0, NULL, NULL,
1808 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1809 		       CTLTYPE_INT, "auto_flowlabel",
1810 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1811 		       NULL, 0, &ip6_auto_flowlabel, 0,
1812 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1813 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1814 	sysctl_createv(clog, 0, NULL, NULL,
1815 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1816 		       CTLTYPE_INT, "defmcasthlim",
1817 		       SYSCTL_DESCR("Default multicast hop limit"),
1818 		       NULL, 0, &ip6_defmcasthlim, 0,
1819 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1820 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1821 #if NGIF > 0
1822 	sysctl_createv(clog, 0, NULL, NULL,
1823 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1824 		       CTLTYPE_INT, "gifhlim",
1825 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
1826 		       NULL, 0, &ip6_gif_hlim, 0,
1827 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1828 		       IPV6CTL_GIF_HLIM, CTL_EOL);
1829 #endif /* NGIF */
1830 	sysctl_createv(clog, 0, NULL, NULL,
1831 		       CTLFLAG_PERMANENT,
1832 		       CTLTYPE_STRING, "kame_version",
1833 		       SYSCTL_DESCR("KAME Version"),
1834 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1835 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1836 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1837 	sysctl_createv(clog, 0, NULL, NULL,
1838 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1839 		       CTLTYPE_INT, "use_deprecated",
1840 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1841 				    "source addresses"),
1842 		       NULL, 0, &ip6_use_deprecated, 0,
1843 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1844 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1845 	sysctl_createv(clog, 0, NULL, NULL,
1846 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1847 		       CTLTYPE_INT, "rr_prune", NULL,
1848 		       NULL, 0, &ip6_rr_prune, 0,
1849 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1850 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1851 	sysctl_createv(clog, 0, NULL, NULL,
1852 		       CTLFLAG_PERMANENT
1853 #ifndef INET6_BINDV6ONLY
1854 		       |CTLFLAG_READWRITE,
1855 #endif
1856 		       CTLTYPE_INT, "v6only",
1857 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1858 				    "to PF_INET sockets"),
1859 		       NULL, 0, &ip6_v6only, 0,
1860 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1861 		       IPV6CTL_V6ONLY, CTL_EOL);
1862 	sysctl_createv(clog, 0, NULL, NULL,
1863 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1864 		       CTLTYPE_INT, "anonportmin",
1865 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1866 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1867 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1868 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1869 	sysctl_createv(clog, 0, NULL, NULL,
1870 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1871 		       CTLTYPE_INT, "anonportmax",
1872 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1873 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1874 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1875 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1876 #ifndef IPNOPRIVPORTS
1877 	sysctl_createv(clog, 0, NULL, NULL,
1878 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1879 		       CTLTYPE_INT, "lowportmin",
1880 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1881 				    "to assign"),
1882 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1883 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1884 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1885 	sysctl_createv(clog, 0, NULL, NULL,
1886 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1887 		       CTLTYPE_INT, "lowportmax",
1888 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1889 				    "to assign"),
1890 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1891 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1892 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1893 #endif /* IPNOPRIVPORTS */
1894 	sysctl_createv(clog, 0, NULL, NULL,
1895 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1896 		       CTLTYPE_INT, "use_tempaddr",
1897 		       SYSCTL_DESCR("Use temporary address"),
1898 		       NULL, 0, &ip6_use_tempaddr, 0,
1899 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1900 		       CTL_CREATE, CTL_EOL);
1901 	sysctl_createv(clog, 0, NULL, NULL,
1902 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1903 		       CTLTYPE_INT, "temppltime",
1904 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
1905 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
1906 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1907 		       CTL_CREATE, CTL_EOL);
1908 	sysctl_createv(clog, 0, NULL, NULL,
1909 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1910 		       CTLTYPE_INT, "tempvltime",
1911 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
1912 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
1913 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1914 		       CTL_CREATE, CTL_EOL);
1915 	sysctl_createv(clog, 0, NULL, NULL,
1916 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1917 		       CTLTYPE_INT, "maxfrags",
1918 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1919 		       NULL, 0, &ip6_maxfrags, 0,
1920 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1921 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1922 	sysctl_createv(clog, 0, NULL, NULL,
1923 		       CTLFLAG_PERMANENT,
1924 		       CTLTYPE_STRUCT, "stats",
1925 		       SYSCTL_DESCR("IPv6 statistics"),
1926 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
1927 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1928 		       IPV6CTL_STATS, CTL_EOL);
1929 	sysctl_createv(clog, 0, NULL, NULL,
1930 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1931 		       CTLTYPE_INT, "use_defaultzone",
1932 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1933 		       NULL, 0, &ip6_use_defzone, 0,
1934 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1935 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1936 	sysctl_createv(clog, 0, NULL, NULL,
1937 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1938 		       CTLTYPE_INT, "mcast_pmtu",
1939 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
1940 		       NULL, 0, &ip6_mcast_pmtu, 0,
1941 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1942 		       CTL_CREATE, CTL_EOL);
1943 #ifdef GATEWAY
1944 	sysctl_createv(clog, 0, NULL, NULL,
1945 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1946 			CTLTYPE_INT, "maxflows",
1947 			SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"),
1948 			sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0,
1949 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1950 			CTL_CREATE, CTL_EOL);
1951 	sysctl_createv(clog, 0, NULL, NULL,
1952 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1953 			CTLTYPE_INT, "hashsize",
1954 			SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"),
1955 			sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0,
1956 			CTL_NET, PF_INET6, IPPROTO_IPV6,
1957 			CTL_CREATE, CTL_EOL);
1958 #endif
1959 }
1960 
1961 void
1962 ip6_statinc(u_int stat)
1963 {
1964 
1965 	KASSERT(stat < IP6_NSTATS);
1966 	IP6_STATINC(stat);
1967 }
1968