xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 27527e67bbdf8d9ec84fd58803048ed6d181ece2)
1 /*	$NetBSD: ip6_input.c,v 1.82 2006/01/23 23:01:40 yamt 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 1.82 2006/01/23 23:01:40 yamt Exp $");
66 
67 #include "opt_inet.h"
68 #include "opt_ipsec.h"
69 #include "opt_pfil_hooks.h"
70 
71 #include <sys/param.h>
72 #include <sys/systm.h>
73 #include <sys/malloc.h>
74 #include <sys/mbuf.h>
75 #include <sys/domain.h>
76 #include <sys/protosw.h>
77 #include <sys/socket.h>
78 #include <sys/socketvar.h>
79 #include <sys/errno.h>
80 #include <sys/time.h>
81 #include <sys/kernel.h>
82 #include <sys/syslog.h>
83 #include <sys/proc.h>
84 #include <sys/sysctl.h>
85 
86 #include <net/if.h>
87 #include <net/if_types.h>
88 #include <net/if_dl.h>
89 #include <net/route.h>
90 #include <net/netisr.h>
91 #ifdef PFIL_HOOKS
92 #include <net/pfil.h>
93 #endif
94 
95 #include <netinet/in.h>
96 #include <netinet/in_systm.h>
97 #ifdef INET
98 #include <netinet/ip.h>
99 #include <netinet/ip_icmp.h>
100 #endif /* INET */
101 #include <netinet/ip6.h>
102 #include <netinet6/in6_var.h>
103 #include <netinet6/ip6_var.h>
104 #include <netinet6/in6_pcb.h>
105 #include <netinet/icmp6.h>
106 #include <netinet6/scope6_var.h>
107 #include <netinet6/in6_ifattach.h>
108 #include <netinet6/nd6.h>
109 
110 #ifdef IPSEC
111 #include <netinet6/ipsec.h>
112 #endif
113 
114 #include <netinet6/ip6protosw.h>
115 
116 #include "faith.h"
117 #include "gif.h"
118 
119 #if NGIF > 0
120 #include <netinet6/in6_gif.h>
121 #endif
122 
123 #include <net/net_osdep.h>
124 
125 extern struct domain inet6domain;
126 
127 u_char ip6_protox[IPPROTO_MAX];
128 static int ip6qmaxlen = IFQ_MAXLEN;
129 struct in6_ifaddr *in6_ifaddr;
130 struct ifqueue ip6intrq;
131 
132 int ip6_forward_srcrt;			/* XXX */
133 int ip6_sourcecheck;			/* XXX */
134 int ip6_sourcecheck_interval;		/* XXX */
135 
136 #ifdef PFIL_HOOKS
137 struct pfil_head inet6_pfil_hook;
138 #endif
139 
140 struct ip6stat ip6stat;
141 
142 static void ip6_init2 __P((void *));
143 static struct m_tag *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *));
144 
145 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
146 static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int));
147 
148 /*
149  * IP6 initialization: fill in IP6 protocol switch table.
150  * All protocols not implemented in kernel go to raw IP6 protocol handler.
151  */
152 void
153 ip6_init()
154 {
155 	const struct ip6protosw *pr;
156 	int i;
157 
158 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
159 	if (pr == 0)
160 		panic("ip6_init");
161 	for (i = 0; i < IPPROTO_MAX; i++)
162 		ip6_protox[i] = pr - inet6sw;
163 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
164 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
165 		if (pr->pr_domain->dom_family == PF_INET6 &&
166 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
167 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
168 	ip6intrq.ifq_maxlen = ip6qmaxlen;
169 	scope6_init();
170 	addrsel_policy_init();
171 	nd6_init();
172 	frag6_init();
173 
174 	ip6_init2((void *)0);
175 
176 #ifdef PFIL_HOOKS
177 	/* Register our Packet Filter hook. */
178 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
179 	inet6_pfil_hook.ph_af   = AF_INET6;
180 	i = pfil_head_register(&inet6_pfil_hook);
181 	if (i != 0)
182 		printf("ip6_init: WARNING: unable to register pfil hook, "
183 		    "error %d\n", i);
184 #endif /* PFIL_HOOKS */
185 }
186 
187 static void
188 ip6_init2(dummy)
189 	void *dummy;
190 {
191 
192 	/* nd6_timer_init */
193 	callout_init(&nd6_timer_ch);
194 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
195 }
196 
197 /*
198  * IP6 input interrupt handling. Just pass the packet to ip6_input.
199  */
200 void
201 ip6intr()
202 {
203 	int s;
204 	struct mbuf *m;
205 
206 	for (;;) {
207 		s = splnet();
208 		IF_DEQUEUE(&ip6intrq, m);
209 		splx(s);
210 		if (m == 0)
211 			return;
212 		ip6_input(m);
213 	}
214 }
215 
216 extern struct	route_in6 ip6_forward_rt;
217 
218 void
219 ip6_input(m)
220 	struct mbuf *m;
221 {
222 	struct ip6_hdr *ip6;
223 	int off = sizeof(struct ip6_hdr), nest;
224 	u_int32_t plen;
225 	u_int32_t rtalert = ~0;
226 	int nxt, ours = 0;
227 	struct ifnet *deliverifp = NULL;
228 	int srcrt = 0;
229 
230 #ifdef IPSEC
231 	/*
232 	 * should the inner packet be considered authentic?
233 	 * see comment in ah4_input().
234 	 */
235 	m->m_flags &= ~M_AUTHIPHDR;
236 	m->m_flags &= ~M_AUTHIPDGM;
237 #endif
238 
239 	/*
240 	 * make sure we don't have onion peering information into m_tag.
241 	 */
242 	ip6_delaux(m);
243 
244 	/*
245 	 * mbuf statistics
246 	 */
247 	if (m->m_flags & M_EXT) {
248 		if (m->m_next)
249 			ip6stat.ip6s_mext2m++;
250 		else
251 			ip6stat.ip6s_mext1++;
252 	} else {
253 #define M2MMAX	(sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
254 		if (m->m_next) {
255 			if (m->m_flags & M_LOOP) {
256 				ip6stat.ip6s_m2m[lo0ifp->if_index]++; /* XXX */
257 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
258 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
259 			else
260 				ip6stat.ip6s_m2m[0]++;
261 		} else
262 			ip6stat.ip6s_m1++;
263 #undef M2MMAX
264 	}
265 
266 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
267 	ip6stat.ip6s_total++;
268 
269 	/*
270 	 * If the IPv6 header is not aligned, slurp it up into a new
271 	 * mbuf with space for link headers, in the event we forward
272 	 * it.  Otherwise, if it is aligned, make sure the entire base
273 	 * IPv6 header is in the first mbuf of the chain.
274 	 */
275 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
276 		struct ifnet *inifp = m->m_pkthdr.rcvif;
277 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
278 				  (max_linkhdr + 3) & ~3)) == NULL) {
279 			/* XXXJRT new stat, please */
280 			ip6stat.ip6s_toosmall++;
281 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
282 			return;
283 		}
284 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
285 		struct ifnet *inifp = m->m_pkthdr.rcvif;
286 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
287 			ip6stat.ip6s_toosmall++;
288 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
289 			return;
290 		}
291 	}
292 
293 	ip6 = mtod(m, struct ip6_hdr *);
294 
295 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
296 		ip6stat.ip6s_badvers++;
297 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
298 		goto bad;
299 	}
300 
301 #ifdef PFIL_HOOKS
302 	/*
303 	 * Run through list of hooks for input packets.  If there are any
304 	 * filters which require that additional packets in the flow are
305 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
306 	 * Note that filters must _never_ set this flag, as another filter
307 	 * in the list may have previously cleared it.
308 	 */
309 	/*
310 	 * let ipfilter look at packet on the wire,
311 	 * not the decapsulated packet.
312 	 */
313 #ifdef IPSEC
314 	if (!ipsec_getnhist(m))
315 #else
316 	if (1)
317 #endif
318 	{
319 		struct in6_addr odst;
320 
321 		odst = ip6->ip6_dst;
322 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
323 				   PFIL_IN) != 0)
324 			return;
325 		if (m == NULL)
326 			return;
327 		ip6 = mtod(m, struct ip6_hdr *);
328 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
329 	}
330 #endif /* PFIL_HOOKS */
331 
332 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
333 
334 #ifdef ALTQ
335 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
336 		/* packet is dropped by traffic conditioner */
337 		return;
338 	}
339 #endif
340 
341 	/*
342 	 * Check against address spoofing/corruption.
343 	 */
344 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
345 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
346 		/*
347 		 * XXX: "badscope" is not very suitable for a multicast source.
348 		 */
349 		ip6stat.ip6s_badscope++;
350 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
351 		goto bad;
352 	}
353 	/*
354 	 * The following check is not documented in specs.  A malicious
355 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
356 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
357 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
358 	 *
359 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
360 	 * support IPv4-less kernel compilation, we cannot support SIIT
361 	 * environment at all.  So, it makes more sense for us to reject any
362 	 * malicious packets for non-SIIT environment, than try to do a
363 	 * partial support for SIIT environment.
364 	 */
365 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
366 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
367 		ip6stat.ip6s_badscope++;
368 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
369 		goto bad;
370 	}
371 #if 0
372 	/*
373 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
374 	 *
375 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
376 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
377 	 * is revised to forbid relaying case.
378 	 */
379 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
380 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
381 		ip6stat.ip6s_badscope++;
382 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
383 		goto bad;
384 	}
385 #endif
386 
387 	/*
388 	 * Disambiguate address scope zones (if there is ambiguity).
389 	 * We first make sure that the original source or destination address
390 	 * is not in our internal form for scoped addresses.  Such addresses
391 	 * are not necessarily invalid spec-wise, but we cannot accept them due
392 	 * to the usage conflict.
393 	 * in6_setscope() then also checks and rejects the cases where src or
394 	 * dst are the loopback address and the receiving interface
395 	 * is not loopback.
396 	 */
397 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
398 		ip6stat.ip6s_badscope++; /* XXX */
399 		goto bad;
400 	}
401 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
402 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
403 		ip6stat.ip6s_badscope++;
404 		goto bad;
405 	}
406 
407 	/*
408 	 * Multicast check
409 	 */
410 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
411 	  	struct	in6_multi *in6m = 0;
412 
413 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
414 		/*
415 		 * See if we belong to the destination multicast group on the
416 		 * arrival interface.
417 		 */
418 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
419 		if (in6m)
420 			ours = 1;
421 		else if (!ip6_mrouter) {
422 			ip6stat.ip6s_notmember++;
423 			ip6stat.ip6s_cantforward++;
424 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
425 			goto bad;
426 		}
427 		deliverifp = m->m_pkthdr.rcvif;
428 		goto hbhcheck;
429 	}
430 
431 	/*
432 	 *  Unicast check
433 	 */
434 	if (ip6_forward_rt.ro_rt != NULL &&
435 	    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
436 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
437 	    &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
438 		ip6stat.ip6s_forward_cachehit++;
439 	else {
440 		struct sockaddr_in6 *dst6;
441 
442 		if (ip6_forward_rt.ro_rt) {
443 			/* route is down or destination is different */
444 			ip6stat.ip6s_forward_cachemiss++;
445 			RTFREE(ip6_forward_rt.ro_rt);
446 			ip6_forward_rt.ro_rt = 0;
447 		}
448 
449 		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
450 		dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
451 		dst6->sin6_len = sizeof(struct sockaddr_in6);
452 		dst6->sin6_family = AF_INET6;
453 		dst6->sin6_addr = ip6->ip6_dst;
454 
455 		rtalloc((struct route *)&ip6_forward_rt);
456 	}
457 
458 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
459 
460 	/*
461 	 * Accept the packet if the forwarding interface to the destination
462 	 * according to the routing table is the loopback interface,
463 	 * unless the associated route has a gateway.
464 	 * Note that this approach causes to accept a packet if there is a
465 	 * route to the loopback interface for the destination of the packet.
466 	 * But we think it's even useful in some situations, e.g. when using
467 	 * a special daemon which wants to intercept the packet.
468 	 */
469 	if (ip6_forward_rt.ro_rt &&
470 	    (ip6_forward_rt.ro_rt->rt_flags &
471 	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
472 	    !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
473 #if 0
474 	    /*
475 	     * The check below is redundant since the comparison of
476 	     * the destination and the key of the rtentry has
477 	     * already done through looking up the routing table.
478 	     */
479 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
480 	    &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
481 #endif
482 	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
483 		struct in6_ifaddr *ia6 =
484 			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
485 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
486 			m->m_flags |= M_ANYCAST6;
487 		/*
488 		 * packets to a tentative, duplicated, or somehow invalid
489 		 * address must not be accepted.
490 		 */
491 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
492 			/* this address is ready */
493 			ours = 1;
494 			deliverifp = ia6->ia_ifp;	/* correct? */
495 			goto hbhcheck;
496 		} else {
497 			/* address is not ready, so discard the packet. */
498 			nd6log((LOG_INFO,
499 			    "ip6_input: packet to an unready address %s->%s\n",
500 			    ip6_sprintf(&ip6->ip6_src),
501 			    ip6_sprintf(&ip6->ip6_dst)));
502 
503 			goto bad;
504 		}
505 	}
506 
507 	/*
508 	 * FAITH (Firewall Aided Internet Translator)
509 	 */
510 #if defined(NFAITH) && 0 < NFAITH
511 	if (ip6_keepfaith) {
512 		if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp &&
513 		    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
514 			/* XXX do we need more sanity checks? */
515 			ours = 1;
516 			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
517 			goto hbhcheck;
518 		}
519 	}
520 #endif
521 
522 #if 0
523     {
524 	/*
525 	 * Last resort: check in6_ifaddr for incoming interface.
526 	 * The code is here until I update the "goto ours hack" code above
527 	 * working right.
528 	 */
529 	struct ifaddr *ifa;
530 	for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
531 	     ifa;
532 	     ifa = ifa->ifa_list.tqe_next) {
533 		if (ifa->ifa_addr == NULL)
534 			continue;	/* just for safety */
535 		if (ifa->ifa_addr->sa_family != AF_INET6)
536 			continue;
537 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
538 			ours = 1;
539 			deliverifp = ifa->ifa_ifp;
540 			goto hbhcheck;
541 		}
542 	}
543     }
544 #endif
545 
546 	/*
547 	 * Now there is no reason to process the packet if it's not our own
548 	 * and we're not a router.
549 	 */
550 	if (!ip6_forwarding) {
551 		ip6stat.ip6s_cantforward++;
552 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
553 		goto bad;
554 	}
555 
556   hbhcheck:
557 	/*
558 	 * record address information into m_tag, if we don't have one yet.
559 	 * note that we are unable to record it, if the address is not listed
560 	 * as our interface address (e.g. multicast addresses, addresses
561 	 * within FAITH prefixes and such).
562 	 */
563 	if (deliverifp && !ip6_getdstifaddr(m)) {
564 		struct in6_ifaddr *ia6;
565 
566 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
567 		if (ia6) {
568 			if (!ip6_setdstifaddr(m, ia6)) {
569 				/*
570 				 * XXX maybe we should drop the packet here,
571 				 * as we could not provide enough information
572 				 * to the upper layers.
573 				 */
574 			}
575 		}
576 	}
577 
578 	/*
579 	 * Process Hop-by-Hop options header if it's contained.
580 	 * m may be modified in ip6_hopopts_input().
581 	 * If a JumboPayload option is included, plen will also be modified.
582 	 */
583 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
584 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
585 		struct ip6_hbh *hbh;
586 
587 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
588 #if 0	/*touches NULL pointer*/
589 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
590 #endif
591 			return;	/* m have already been freed */
592 		}
593 
594 		/* adjust pointer */
595 		ip6 = mtod(m, struct ip6_hdr *);
596 
597 		/*
598 		 * if the payload length field is 0 and the next header field
599 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
600 		 * option MUST be included.
601 		 */
602 		if (ip6->ip6_plen == 0 && plen == 0) {
603 			/*
604 			 * Note that if a valid jumbo payload option is
605 			 * contained, ip6_hoptops_input() must set a valid
606 			 * (non-zero) payload length to the variable plen.
607 			 */
608 			ip6stat.ip6s_badoptions++;
609 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
610 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
611 			icmp6_error(m, ICMP6_PARAM_PROB,
612 				    ICMP6_PARAMPROB_HEADER,
613 				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
614 			return;
615 		}
616 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
617 			sizeof(struct ip6_hbh));
618 		if (hbh == NULL) {
619 			ip6stat.ip6s_tooshort++;
620 			return;
621 		}
622 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
623 		nxt = hbh->ip6h_nxt;
624 
625 		/*
626 		 * accept the packet if a router alert option is included
627 		 * and we act as an IPv6 router.
628 		 */
629 		if (rtalert != ~0 && ip6_forwarding)
630 			ours = 1;
631 	} else
632 		nxt = ip6->ip6_nxt;
633 
634 	/*
635 	 * Check that the amount of data in the buffers
636 	 * is as at least much as the IPv6 header would have us expect.
637 	 * Trim mbufs if longer than we expect.
638 	 * Drop packet if shorter than we expect.
639 	 */
640 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
641 		ip6stat.ip6s_tooshort++;
642 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
643 		goto bad;
644 	}
645 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
646 		if (m->m_len == m->m_pkthdr.len) {
647 			m->m_len = sizeof(struct ip6_hdr) + plen;
648 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
649 		} else
650 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
651 	}
652 
653 	/*
654 	 * Forward if desirable.
655 	 */
656 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
657 		/*
658 		 * If we are acting as a multicast router, all
659 		 * incoming multicast packets are passed to the
660 		 * kernel-level multicast forwarding function.
661 		 * The packet is returned (relatively) intact; if
662 		 * ip6_mforward() returns a non-zero value, the packet
663 		 * must be discarded, else it may be accepted below.
664 		 */
665 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
666 			ip6stat.ip6s_cantforward++;
667 			m_freem(m);
668 			return;
669 		}
670 		if (!ours) {
671 			m_freem(m);
672 			return;
673 		}
674 	} else if (!ours) {
675 		ip6_forward(m, srcrt);
676 		return;
677 	}
678 
679 	ip6 = mtod(m, struct ip6_hdr *);
680 
681 	/*
682 	 * Malicious party may be able to use IPv4 mapped addr to confuse
683 	 * tcp/udp stack and bypass security checks (act as if it was from
684 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
685 	 *
686 	 * For SIIT end node behavior, you may want to disable the check.
687 	 * However, you will  become vulnerable to attacks using IPv4 mapped
688 	 * source.
689 	 */
690 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
691 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
692 		ip6stat.ip6s_badscope++;
693 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
694 		goto bad;
695 	}
696 
697 	/*
698 	 * Tell launch routine the next header
699 	 */
700 #ifdef IFA_STATS
701 	if (deliverifp != NULL) {
702 		struct in6_ifaddr *ia6;
703 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
704 		if (ia6)
705 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
706 	}
707 #endif
708 	ip6stat.ip6s_delivered++;
709 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
710 	nest = 0;
711 
712 	while (nxt != IPPROTO_DONE) {
713 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
714 			ip6stat.ip6s_toomanyhdr++;
715 			goto bad;
716 		}
717 
718 		/*
719 		 * protection against faulty packet - there should be
720 		 * more sanity checks in header chain processing.
721 		 */
722 		if (m->m_pkthdr.len < off) {
723 			ip6stat.ip6s_tooshort++;
724 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
725 			goto bad;
726 		}
727 
728 #ifdef IPSEC
729 		/*
730 		 * enforce IPsec policy checking if we are seeing last header.
731 		 * note that we do not visit this with protocols with pcb layer
732 		 * code - like udp/tcp/raw ip.
733 		 */
734 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
735 		    ipsec6_in_reject(m, NULL)) {
736 			ipsec6stat.in_polvio++;
737 			goto bad;
738 		}
739 #endif
740 
741 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
742 	}
743 	return;
744  bad:
745 	m_freem(m);
746 }
747 
748 /*
749  * set/grab in6_ifaddr correspond to IPv6 destination address.
750  */
751 static struct m_tag *
752 ip6_setdstifaddr(m, ia6)
753 	struct mbuf *m;
754 	struct in6_ifaddr *ia6;
755 {
756 	struct m_tag *mtag;
757 
758 	mtag = ip6_addaux(m);
759 	if (mtag)
760 		((struct ip6aux *)(mtag + 1))->ip6a_dstia6 = ia6;
761 	return mtag;	/* NULL if failed to set */
762 }
763 
764 struct in6_ifaddr *
765 ip6_getdstifaddr(m)
766 	struct mbuf *m;
767 {
768 	struct m_tag *mtag;
769 
770 	mtag = ip6_findaux(m);
771 	if (mtag)
772 		return ((struct ip6aux *)(mtag + 1))->ip6a_dstia6;
773 	else
774 		return NULL;
775 }
776 
777 /*
778  * Hop-by-Hop options header processing. If a valid jumbo payload option is
779  * included, the real payload length will be stored in plenp.
780  */
781 static int
782 ip6_hopopts_input(plenp, rtalertp, mp, offp)
783 	u_int32_t *plenp;
784 	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
785 	struct mbuf **mp;
786 	int *offp;
787 {
788 	struct mbuf *m = *mp;
789 	int off = *offp, hbhlen;
790 	struct ip6_hbh *hbh;
791 
792 	/* validation of the length of the header */
793 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
794 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
795 	if (hbh == NULL) {
796 		ip6stat.ip6s_tooshort++;
797 		return -1;
798 	}
799 	hbhlen = (hbh->ip6h_len + 1) << 3;
800 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
801 		hbhlen);
802 	if (hbh == NULL) {
803 		ip6stat.ip6s_tooshort++;
804 		return -1;
805 	}
806 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
807 	off += hbhlen;
808 	hbhlen -= sizeof(struct ip6_hbh);
809 
810 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
811 				hbhlen, rtalertp, plenp) < 0)
812 		return (-1);
813 
814 	*offp = off;
815 	*mp = m;
816 	return (0);
817 }
818 
819 /*
820  * Search header for all Hop-by-hop options and process each option.
821  * This function is separate from ip6_hopopts_input() in order to
822  * handle a case where the sending node itself process its hop-by-hop
823  * options header. In such a case, the function is called from ip6_output().
824  *
825  * The function assumes that hbh header is located right after the IPv6 header
826  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
827  * opthead + hbhlen is located in continuous memory region.
828  */
829 int
830 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
831 	struct mbuf *m;
832 	u_int8_t *opthead;
833 	int hbhlen;
834 	u_int32_t *rtalertp;
835 	u_int32_t *plenp;
836 {
837 	struct ip6_hdr *ip6;
838 	int optlen = 0;
839 	u_int8_t *opt = opthead;
840 	u_int16_t rtalert_val;
841 	u_int32_t jumboplen;
842 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
843 
844 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
845 		switch (*opt) {
846 		case IP6OPT_PAD1:
847 			optlen = 1;
848 			break;
849 		case IP6OPT_PADN:
850 			if (hbhlen < IP6OPT_MINLEN) {
851 				ip6stat.ip6s_toosmall++;
852 				goto bad;
853 			}
854 			optlen = *(opt + 1) + 2;
855 			break;
856 		case IP6OPT_RTALERT:
857 			/* XXX may need check for alignment */
858 			if (hbhlen < IP6OPT_RTALERT_LEN) {
859 				ip6stat.ip6s_toosmall++;
860 				goto bad;
861 			}
862 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
863 				/* XXX stat */
864 				icmp6_error(m, ICMP6_PARAM_PROB,
865 				    ICMP6_PARAMPROB_HEADER,
866 				    erroff + opt + 1 - opthead);
867 				return (-1);
868 			}
869 			optlen = IP6OPT_RTALERT_LEN;
870 			bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
871 			*rtalertp = ntohs(rtalert_val);
872 			break;
873 		case IP6OPT_JUMBO:
874 			/* XXX may need check for alignment */
875 			if (hbhlen < IP6OPT_JUMBO_LEN) {
876 				ip6stat.ip6s_toosmall++;
877 				goto bad;
878 			}
879 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
880 				/* XXX stat */
881 				icmp6_error(m, ICMP6_PARAM_PROB,
882 				    ICMP6_PARAMPROB_HEADER,
883 				    erroff + opt + 1 - opthead);
884 				return (-1);
885 			}
886 			optlen = IP6OPT_JUMBO_LEN;
887 
888 			/*
889 			 * IPv6 packets that have non 0 payload length
890 			 * must not contain a jumbo payload option.
891 			 */
892 			ip6 = mtod(m, struct ip6_hdr *);
893 			if (ip6->ip6_plen) {
894 				ip6stat.ip6s_badoptions++;
895 				icmp6_error(m, ICMP6_PARAM_PROB,
896 				    ICMP6_PARAMPROB_HEADER,
897 				    erroff + opt - opthead);
898 				return (-1);
899 			}
900 
901 			/*
902 			 * We may see jumbolen in unaligned location, so
903 			 * we'd need to perform bcopy().
904 			 */
905 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
906 			jumboplen = (u_int32_t)htonl(jumboplen);
907 
908 #if 1
909 			/*
910 			 * if there are multiple jumbo payload options,
911 			 * *plenp will be non-zero and the packet will be
912 			 * rejected.
913 			 * the behavior may need some debate in ipngwg -
914 			 * multiple options does not make sense, however,
915 			 * there's no explicit mention in specification.
916 			 */
917 			if (*plenp != 0) {
918 				ip6stat.ip6s_badoptions++;
919 				icmp6_error(m, ICMP6_PARAM_PROB,
920 				    ICMP6_PARAMPROB_HEADER,
921 				    erroff + opt + 2 - opthead);
922 				return (-1);
923 			}
924 #endif
925 
926 			/*
927 			 * jumbo payload length must be larger than 65535.
928 			 */
929 			if (jumboplen <= IPV6_MAXPACKET) {
930 				ip6stat.ip6s_badoptions++;
931 				icmp6_error(m, ICMP6_PARAM_PROB,
932 				    ICMP6_PARAMPROB_HEADER,
933 				    erroff + opt + 2 - opthead);
934 				return (-1);
935 			}
936 			*plenp = jumboplen;
937 
938 			break;
939 		default:		/* unknown option */
940 			if (hbhlen < IP6OPT_MINLEN) {
941 				ip6stat.ip6s_toosmall++;
942 				goto bad;
943 			}
944 			optlen = ip6_unknown_opt(opt, m,
945 			    erroff + opt - opthead);
946 			if (optlen == -1)
947 				return (-1);
948 			optlen += 2;
949 			break;
950 		}
951 	}
952 
953 	return (0);
954 
955   bad:
956 	m_freem(m);
957 	return (-1);
958 }
959 
960 /*
961  * Unknown option processing.
962  * The third argument `off' is the offset from the IPv6 header to the option,
963  * which is necessary if the IPv6 header the and option header and IPv6 header
964  * is not continuous in order to return an ICMPv6 error.
965  */
966 int
967 ip6_unknown_opt(optp, m, off)
968 	u_int8_t *optp;
969 	struct mbuf *m;
970 	int off;
971 {
972 	struct ip6_hdr *ip6;
973 
974 	switch (IP6OPT_TYPE(*optp)) {
975 	case IP6OPT_TYPE_SKIP: /* ignore the option */
976 		return ((int)*(optp + 1));
977 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
978 		m_freem(m);
979 		return (-1);
980 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
981 		ip6stat.ip6s_badoptions++;
982 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
983 		return (-1);
984 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
985 		ip6stat.ip6s_badoptions++;
986 		ip6 = mtod(m, struct ip6_hdr *);
987 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
988 		    (m->m_flags & (M_BCAST|M_MCAST)))
989 			m_freem(m);
990 		else
991 			icmp6_error(m, ICMP6_PARAM_PROB,
992 				    ICMP6_PARAMPROB_OPTION, off);
993 		return (-1);
994 	}
995 
996 	m_freem(m);		/* XXX: NOTREACHED */
997 	return (-1);
998 }
999 
1000 /*
1001  * Create the "control" list for this pcb.
1002  *
1003  * The routine will be called from upper layer handlers like tcp6_input().
1004  * Thus the routine assumes that the caller (tcp6_input) have already
1005  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1006  * very first mbuf on the mbuf chain.
1007  * We may want to add some infinite loop prevention or sanity checks for safety.
1008  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1009  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1010  */
1011 void
1012 ip6_savecontrol(in6p, mp, ip6, m)
1013 	struct in6pcb *in6p;
1014 	struct mbuf **mp;
1015 	struct ip6_hdr *ip6;
1016 	struct mbuf *m;
1017 {
1018 
1019 #ifdef SO_TIMESTAMP
1020 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
1021 		struct timeval tv;
1022 
1023 		microtime(&tv);
1024 		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1025 		    SCM_TIMESTAMP, SOL_SOCKET);
1026 		if (*mp)
1027 			mp = &(*mp)->m_next;
1028 	}
1029 #endif
1030 	if (in6p->in6p_flags & IN6P_RECVDSTADDR) {
1031 		*mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst,
1032 		    sizeof(struct in6_addr), IPV6_RECVDSTADDR, IPPROTO_IPV6);
1033 		if (*mp)
1034 			mp = &(*mp)->m_next;
1035 	}
1036 
1037 #ifdef noyet
1038 	/* options were tossed above */
1039 	if (in6p->in6p_flags & IN6P_RECVOPTS)
1040 		/* broken */
1041 	/* ip6_srcroute doesn't do what we want here, need to fix */
1042 	if (in6p->in6p_flags & IPV6P_RECVRETOPTS)
1043 		/* broken */
1044 #endif
1045 
1046 	/* RFC 2292 sec. 5 */
1047 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1048 		struct in6_pktinfo pi6;
1049 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1050 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1051 		pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1052 					? m->m_pkthdr.rcvif->if_index
1053 					: 0;
1054 		*mp = sbcreatecontrol((caddr_t) &pi6,
1055 		    sizeof(struct in6_pktinfo), IPV6_PKTINFO, IPPROTO_IPV6);
1056 		if (*mp)
1057 			mp = &(*mp)->m_next;
1058 	}
1059 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1060 		int hlim = ip6->ip6_hlim & 0xff;
1061 		*mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1062 		    IPV6_HOPLIMIT, IPPROTO_IPV6);
1063 		if (*mp)
1064 			mp = &(*mp)->m_next;
1065 	}
1066 	/* IN6P_NEXTHOP - for outgoing packet only */
1067 
1068 	/*
1069 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1070 	 * privilege for the option (see ip6_ctloutput), but it might be too
1071 	 * strict, since there might be some hop-by-hop options which can be
1072 	 * returned to normal user.
1073 	 * See also RFC 2292 section 6.
1074 	 */
1075 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1076 		/*
1077 		 * Check if a hop-by-hop options header is contatined in the
1078 		 * received packet, and if so, store the options as ancillary
1079 		 * data. Note that a hop-by-hop options header must be
1080 		 * just after the IPv6 header, which fact is assured through
1081 		 * the IPv6 input processing.
1082 		 */
1083 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1084 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1085 			struct ip6_hbh *hbh;
1086 			int hbhlen;
1087 			struct mbuf *ext;
1088 
1089 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1090 			    xip6->ip6_nxt);
1091 			if (ext == NULL) {
1092 				ip6stat.ip6s_tooshort++;
1093 				return;
1094 			}
1095 			hbh = mtod(ext, struct ip6_hbh *);
1096 			hbhlen = (hbh->ip6h_len + 1) << 3;
1097 			if (hbhlen != ext->m_len) {
1098 				m_freem(ext);
1099 				ip6stat.ip6s_tooshort++;
1100 				return;
1101 			}
1102 
1103 			/*
1104 			 * XXX: We copy whole the header even if a jumbo
1105 			 * payload option is included, which option is to
1106 			 * be removed before returning in the RFC 2292.
1107 			 * But it's too painful operation...
1108 			 */
1109 			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1110 			    IPV6_HOPOPTS, IPPROTO_IPV6);
1111 			if (*mp)
1112 				mp = &(*mp)->m_next;
1113 			m_freem(ext);
1114 		}
1115 	}
1116 
1117 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1118 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1119 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1120 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1121 
1122 		/*
1123 		 * Search for destination options headers or routing
1124 		 * header(s) through the header chain, and stores each
1125 		 * header as ancillary data.
1126 		 * Note that the order of the headers remains in
1127 		 * the chain of ancillary data.
1128 		 */
1129 		while (1) {	/* is explicit loop prevention necessary? */
1130 			struct ip6_ext *ip6e = NULL;
1131 			int elen;
1132 			struct mbuf *ext = NULL;
1133 
1134 			/*
1135 			 * if it is not an extension header, don't try to
1136 			 * pull it from the chain.
1137 			 */
1138 			switch (nxt) {
1139 			case IPPROTO_DSTOPTS:
1140 			case IPPROTO_ROUTING:
1141 			case IPPROTO_HOPOPTS:
1142 			case IPPROTO_AH: /* is it possible? */
1143 				break;
1144 			default:
1145 				goto loopend;
1146 			}
1147 
1148 			ext = ip6_pullexthdr(m, off, nxt);
1149 			if (ext == NULL) {
1150 				ip6stat.ip6s_tooshort++;
1151 				return;
1152 			}
1153 			ip6e = mtod(ext, struct ip6_ext *);
1154 			if (nxt == IPPROTO_AH)
1155 				elen = (ip6e->ip6e_len + 2) << 2;
1156 			else
1157 				elen = (ip6e->ip6e_len + 1) << 3;
1158 			if (elen != ext->m_len) {
1159 				m_freem(ext);
1160 				ip6stat.ip6s_tooshort++;
1161 				return;
1162 			}
1163 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1164 
1165 			switch (nxt) {
1166 			case IPPROTO_DSTOPTS:
1167 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
1168 					break;
1169 
1170 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1171 				    IPV6_DSTOPTS, IPPROTO_IPV6);
1172 				if (*mp)
1173 					mp = &(*mp)->m_next;
1174 				break;
1175 
1176 			case IPPROTO_ROUTING:
1177 				if (!in6p->in6p_flags & IN6P_RTHDR)
1178 					break;
1179 
1180 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1181 				    IPV6_RTHDR, IPPROTO_IPV6);
1182 				if (*mp)
1183 					mp = &(*mp)->m_next;
1184 				break;
1185 
1186 			case IPPROTO_HOPOPTS:
1187 			case IPPROTO_AH: /* is it possible? */
1188 				break;
1189 
1190 			default:
1191 				/*
1192 			 	 * other cases have been filtered in the above.
1193 				 * none will visit this case.  here we supply
1194 				 * the code just in case (nxt overwritten or
1195 				 * other cases).
1196 				 */
1197 				m_freem(ext);
1198 				goto loopend;
1199 
1200 			}
1201 
1202 			/* proceed with the next header. */
1203 			off += elen;
1204 			nxt = ip6e->ip6e_nxt;
1205 			ip6e = NULL;
1206 			m_freem(ext);
1207 			ext = NULL;
1208 		}
1209 	  loopend:
1210 	  	;
1211 	}
1212 }
1213 
1214 /*
1215  * pull single extension header from mbuf chain.  returns single mbuf that
1216  * contains the result, or NULL on error.
1217  */
1218 static struct mbuf *
1219 ip6_pullexthdr(m, off, nxt)
1220 	struct mbuf *m;
1221 	size_t off;
1222 	int nxt;
1223 {
1224 	struct ip6_ext ip6e;
1225 	size_t elen;
1226 	struct mbuf *n;
1227 
1228 #ifdef DIAGNOSTIC
1229 	switch (nxt) {
1230 	case IPPROTO_DSTOPTS:
1231 	case IPPROTO_ROUTING:
1232 	case IPPROTO_HOPOPTS:
1233 	case IPPROTO_AH: /* is it possible? */
1234 		break;
1235 	default:
1236 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1237 	}
1238 #endif
1239 
1240 	m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1241 	if (nxt == IPPROTO_AH)
1242 		elen = (ip6e.ip6e_len + 2) << 2;
1243 	else
1244 		elen = (ip6e.ip6e_len + 1) << 3;
1245 
1246 	MGET(n, M_DONTWAIT, MT_DATA);
1247 	if (n && elen >= MLEN) {
1248 		MCLGET(n, M_DONTWAIT);
1249 		if ((n->m_flags & M_EXT) == 0) {
1250 			m_free(n);
1251 			n = NULL;
1252 		}
1253 	}
1254 	if (!n)
1255 		return NULL;
1256 
1257 	n->m_len = 0;
1258 	if (elen >= M_TRAILINGSPACE(n)) {
1259 		m_free(n);
1260 		return NULL;
1261 	}
1262 
1263 	m_copydata(m, off, elen, mtod(n, caddr_t));
1264 	n->m_len = elen;
1265 	return n;
1266 }
1267 
1268 /*
1269  * Get pointer to the previous header followed by the header
1270  * currently processed.
1271  * XXX: This function supposes that
1272  *	M includes all headers,
1273  *	the next header field and the header length field of each header
1274  *	are valid, and
1275  *	the sum of each header length equals to OFF.
1276  * Because of these assumptions, this function must be called very
1277  * carefully. Moreover, it will not be used in the near future when
1278  * we develop `neater' mechanism to process extension headers.
1279  */
1280 u_int8_t *
1281 ip6_get_prevhdr(m, off)
1282 	struct mbuf *m;
1283 	int off;
1284 {
1285 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1286 
1287 	if (off == sizeof(struct ip6_hdr))
1288 		return (&ip6->ip6_nxt);
1289 	else {
1290 		int len, nxt;
1291 		struct ip6_ext *ip6e = NULL;
1292 
1293 		nxt = ip6->ip6_nxt;
1294 		len = sizeof(struct ip6_hdr);
1295 		while (len < off) {
1296 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1297 
1298 			switch (nxt) {
1299 			case IPPROTO_FRAGMENT:
1300 				len += sizeof(struct ip6_frag);
1301 				break;
1302 			case IPPROTO_AH:
1303 				len += (ip6e->ip6e_len + 2) << 2;
1304 				break;
1305 			default:
1306 				len += (ip6e->ip6e_len + 1) << 3;
1307 				break;
1308 			}
1309 			nxt = ip6e->ip6e_nxt;
1310 		}
1311 		if (ip6e)
1312 			return (&ip6e->ip6e_nxt);
1313 		else
1314 			return NULL;
1315 	}
1316 }
1317 
1318 /*
1319  * get next header offset.  m will be retained.
1320  */
1321 int
1322 ip6_nexthdr(m, off, proto, nxtp)
1323 	struct mbuf *m;
1324 	int off;
1325 	int proto;
1326 	int *nxtp;
1327 {
1328 	struct ip6_hdr ip6;
1329 	struct ip6_ext ip6e;
1330 	struct ip6_frag fh;
1331 
1332 	/* just in case */
1333 	if (m == NULL)
1334 		panic("ip6_nexthdr: m == NULL");
1335 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1336 		return -1;
1337 
1338 	switch (proto) {
1339 	case IPPROTO_IPV6:
1340 		if (m->m_pkthdr.len < off + sizeof(ip6))
1341 			return -1;
1342 		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1343 		if (nxtp)
1344 			*nxtp = ip6.ip6_nxt;
1345 		off += sizeof(ip6);
1346 		return off;
1347 
1348 	case IPPROTO_FRAGMENT:
1349 		/*
1350 		 * terminate parsing if it is not the first fragment,
1351 		 * it does not make sense to parse through it.
1352 		 */
1353 		if (m->m_pkthdr.len < off + sizeof(fh))
1354 			return -1;
1355 		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1356 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1357 			return -1;
1358 		if (nxtp)
1359 			*nxtp = fh.ip6f_nxt;
1360 		off += sizeof(struct ip6_frag);
1361 		return off;
1362 
1363 	case IPPROTO_AH:
1364 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1365 			return -1;
1366 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1367 		if (nxtp)
1368 			*nxtp = ip6e.ip6e_nxt;
1369 		off += (ip6e.ip6e_len + 2) << 2;
1370 		if (m->m_pkthdr.len < off)
1371 			return -1;
1372 		return off;
1373 
1374 	case IPPROTO_HOPOPTS:
1375 	case IPPROTO_ROUTING:
1376 	case IPPROTO_DSTOPTS:
1377 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1378 			return -1;
1379 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1380 		if (nxtp)
1381 			*nxtp = ip6e.ip6e_nxt;
1382 		off += (ip6e.ip6e_len + 1) << 3;
1383 		if (m->m_pkthdr.len < off)
1384 			return -1;
1385 		return off;
1386 
1387 	case IPPROTO_NONE:
1388 	case IPPROTO_ESP:
1389 	case IPPROTO_IPCOMP:
1390 		/* give up */
1391 		return -1;
1392 
1393 	default:
1394 		return -1;
1395 	}
1396 }
1397 
1398 /*
1399  * get offset for the last header in the chain.  m will be kept untainted.
1400  */
1401 int
1402 ip6_lasthdr(m, off, proto, nxtp)
1403 	struct mbuf *m;
1404 	int off;
1405 	int proto;
1406 	int *nxtp;
1407 {
1408 	int newoff;
1409 	int nxt;
1410 
1411 	if (!nxtp) {
1412 		nxt = -1;
1413 		nxtp = &nxt;
1414 	}
1415 	while (1) {
1416 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1417 		if (newoff < 0)
1418 			return off;
1419 		else if (newoff < off)
1420 			return -1;	/* invalid */
1421 		else if (newoff == off)
1422 			return newoff;
1423 
1424 		off = newoff;
1425 		proto = *nxtp;
1426 	}
1427 }
1428 
1429 struct m_tag *
1430 ip6_addaux(m)
1431 	struct mbuf *m;
1432 {
1433 	struct m_tag *mtag;
1434 
1435 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1436 	if (!mtag) {
1437 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1438 		    M_NOWAIT);
1439 		if (mtag) {
1440 			m_tag_prepend(m, mtag);
1441 			bzero(mtag + 1, sizeof(struct ip6aux));
1442 		}
1443 	}
1444 	return mtag;
1445 }
1446 
1447 struct m_tag *
1448 ip6_findaux(m)
1449 	struct mbuf *m;
1450 {
1451 	struct m_tag *mtag;
1452 
1453 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1454 	return mtag;
1455 }
1456 
1457 void
1458 ip6_delaux(m)
1459 	struct mbuf *m;
1460 {
1461 	struct m_tag *mtag;
1462 
1463 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1464 	if (mtag)
1465 		m_tag_delete(m, mtag);
1466 }
1467 
1468 /*
1469  * System control for IP6
1470  */
1471 
1472 u_char	inet6ctlerrmap[PRC_NCMDS] = {
1473 	0,		0,		0,		0,
1474 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1475 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1476 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1477 	0,		0,		0,		0,
1478 	ENOPROTOOPT
1479 };
1480 
1481 SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup")
1482 {
1483 
1484 	sysctl_createv(clog, 0, NULL, NULL,
1485 		       CTLFLAG_PERMANENT,
1486 		       CTLTYPE_NODE, "net", NULL,
1487 		       NULL, 0, NULL, 0,
1488 		       CTL_NET, CTL_EOL);
1489 	sysctl_createv(clog, 0, NULL, NULL,
1490 		       CTLFLAG_PERMANENT,
1491 		       CTLTYPE_NODE, "inet6",
1492 		       SYSCTL_DESCR("PF_INET6 related settings"),
1493 		       NULL, 0, NULL, 0,
1494 		       CTL_NET, PF_INET6, CTL_EOL);
1495 	sysctl_createv(clog, 0, NULL, NULL,
1496 		       CTLFLAG_PERMANENT,
1497 		       CTLTYPE_NODE, "ip6",
1498 		       SYSCTL_DESCR("IPv6 related settings"),
1499 		       NULL, 0, NULL, 0,
1500 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1501 
1502 	sysctl_createv(clog, 0, NULL, NULL,
1503 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1504 		       CTLTYPE_INT, "forwarding",
1505 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1506 		       NULL, 0, &ip6_forwarding, 0,
1507 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1508 		       IPV6CTL_FORWARDING, CTL_EOL);
1509 	sysctl_createv(clog, 0, NULL, NULL,
1510 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1511 		       CTLTYPE_INT, "redirect",
1512 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1513 		       NULL, 0, &ip6_sendredirects, 0,
1514 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1515 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1516 	sysctl_createv(clog, 0, NULL, NULL,
1517 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1518 		       CTLTYPE_INT, "hlim",
1519 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1520 		       NULL, 0, &ip6_defhlim, 0,
1521 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1522 		       IPV6CTL_DEFHLIM, CTL_EOL);
1523 #ifdef notyet
1524 	sysctl_createv(clog, 0, NULL, NULL,
1525 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1526 		       CTLTYPE_INT, "mtu", NULL,
1527 		       NULL, 0, &, 0,
1528 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1529 		       IPV6CTL_DEFMTU, CTL_EOL);
1530 #endif
1531 #ifdef __no_idea__
1532 	sysctl_createv(clog, 0, NULL, NULL,
1533 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1534 		       CTLTYPE_INT, "forwsrcrt", NULL,
1535 		       NULL, 0, &?, 0,
1536 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1537 		       IPV6CTL_FORWSRCRT, CTL_EOL);
1538 	sysctl_createv(clog, 0, NULL, NULL,
1539 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1540 		       CTLTYPE_STRUCT, "mrtstats", NULL,
1541 		       NULL, 0, &?, sizeof(?),
1542 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1543 		       IPV6CTL_MRTSTATS, CTL_EOL);
1544 	sysctl_createv(clog, 0, NULL, NULL,
1545 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1546 		       CTLTYPE_?, "mrtproto", NULL,
1547 		       NULL, 0, &?, sizeof(?),
1548 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1549 		       IPV6CTL_MRTPROTO, CTL_EOL);
1550 #endif
1551 	sysctl_createv(clog, 0, NULL, NULL,
1552 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1553 		       CTLTYPE_INT, "maxfragpackets",
1554 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1555 				    "for reassembly"),
1556 		       NULL, 0, &ip6_maxfragpackets, 0,
1557 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1558 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1559 #ifdef __no_idea__
1560 	sysctl_createv(clog, 0, NULL, NULL,
1561 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1562 		       CTLTYPE_INT, "sourcecheck", NULL,
1563 		       NULL, 0, &?, 0,
1564 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1565 		       IPV6CTL_SOURCECHECK, CTL_EOL);
1566 	sysctl_createv(clog, 0, NULL, NULL,
1567 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1568 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
1569 		       NULL, 0, &?, 0,
1570 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1571 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
1572 #endif
1573 	sysctl_createv(clog, 0, NULL, NULL,
1574 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1575 		       CTLTYPE_INT, "accept_rtadv",
1576 		       SYSCTL_DESCR("Accept router advertisements"),
1577 		       NULL, 0, &ip6_accept_rtadv, 0,
1578 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1579 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1580 	sysctl_createv(clog, 0, NULL, NULL,
1581 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1582 		       CTLTYPE_INT, "keepfaith",
1583 		       SYSCTL_DESCR("Activate faith interface"),
1584 		       NULL, 0, &ip6_keepfaith, 0,
1585 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1586 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1587 	sysctl_createv(clog, 0, NULL, NULL,
1588 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1589 		       CTLTYPE_INT, "log_interval",
1590 		       SYSCTL_DESCR("Minumum interval between logging "
1591 				    "unroutable packets"),
1592 		       NULL, 0, &ip6_log_interval, 0,
1593 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1594 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1595 	sysctl_createv(clog, 0, NULL, NULL,
1596 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1597 		       CTLTYPE_INT, "hdrnestlimit",
1598 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1599 		       NULL, 0, &ip6_hdrnestlimit, 0,
1600 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1601 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1602 	sysctl_createv(clog, 0, NULL, NULL,
1603 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1604 		       CTLTYPE_INT, "dad_count",
1605 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1606 				    "probes to send"),
1607 		       NULL, 0, &ip6_dad_count, 0,
1608 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1609 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1610 	sysctl_createv(clog, 0, NULL, NULL,
1611 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1612 		       CTLTYPE_INT, "auto_flowlabel",
1613 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1614 		       NULL, 0, &ip6_auto_flowlabel, 0,
1615 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1616 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1617 	sysctl_createv(clog, 0, NULL, NULL,
1618 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1619 		       CTLTYPE_INT, "defmcasthlim",
1620 		       SYSCTL_DESCR("Default multicast hop limit"),
1621 		       NULL, 0, &ip6_defmcasthlim, 0,
1622 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1623 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1624 #if NGIF > 0
1625 	sysctl_createv(clog, 0, NULL, NULL,
1626 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1627 		       CTLTYPE_INT, "gifhlim",
1628 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
1629 		       NULL, 0, &ip6_gif_hlim, 0,
1630 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1631 		       IPV6CTL_GIF_HLIM, CTL_EOL);
1632 #endif /* NGIF */
1633 	sysctl_createv(clog, 0, NULL, NULL,
1634 		       CTLFLAG_PERMANENT,
1635 		       CTLTYPE_STRING, "kame_version",
1636 		       SYSCTL_DESCR("KAME Version"),
1637 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1638 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1639 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1640 	sysctl_createv(clog, 0, NULL, NULL,
1641 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1642 		       CTLTYPE_INT, "use_deprecated",
1643 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1644 				    "source addresses"),
1645 		       NULL, 0, &ip6_use_deprecated, 0,
1646 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1647 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1648 	sysctl_createv(clog, 0, NULL, NULL,
1649 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1650 		       CTLTYPE_INT, "rr_prune", NULL,
1651 		       NULL, 0, &ip6_rr_prune, 0,
1652 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1653 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1654 	sysctl_createv(clog, 0, NULL, NULL,
1655 		       CTLFLAG_PERMANENT
1656 #ifndef INET6_BINDV6ONLY
1657 		       |CTLFLAG_READWRITE,
1658 #endif
1659 		       CTLTYPE_INT, "v6only",
1660 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1661 				    "to PF_INET sockets"),
1662 		       NULL, 0, &ip6_v6only, 0,
1663 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1664 		       IPV6CTL_V6ONLY, CTL_EOL);
1665 	sysctl_createv(clog, 0, NULL, NULL,
1666 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1667 		       CTLTYPE_INT, "anonportmin",
1668 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1669 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1670 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1671 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1672 	sysctl_createv(clog, 0, NULL, NULL,
1673 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1674 		       CTLTYPE_INT, "anonportmax",
1675 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1676 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1677 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1678 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1679 #ifndef IPNOPRIVPORTS
1680 	sysctl_createv(clog, 0, NULL, NULL,
1681 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1682 		       CTLTYPE_INT, "lowportmin",
1683 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1684 				    "to assign"),
1685 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1686 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1687 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1688 	sysctl_createv(clog, 0, NULL, NULL,
1689 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1690 		       CTLTYPE_INT, "lowportmax",
1691 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1692 				    "to assign"),
1693 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1694 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1695 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1696 #endif /* IPNOPRIVPORTS */
1697 	sysctl_createv(clog, 0, NULL, NULL,
1698 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1699 		       CTLTYPE_INT, "maxfrags",
1700 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1701 		       NULL, 0, &ip6_maxfrags, 0,
1702 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1703 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1704 	sysctl_createv(clog, 0, NULL, NULL,
1705 		       CTLFLAG_PERMANENT,
1706 		       CTLTYPE_STRUCT, "stats",
1707 		       SYSCTL_DESCR("IPv6 statistics"),
1708 		       NULL, 0, &ip6stat, sizeof(ip6stat),
1709 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1710 		       IPV6CTL_STATS, CTL_EOL);
1711 	sysctl_createv(clog, 0, NULL, NULL,
1712 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1713 		       CTLTYPE_INT, "use_defaultzone",
1714 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1715 		       NULL, 0, &ip6_use_defzone, 0,
1716 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1717 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1718 }
1719