xref: /netbsd-src/sys/netinet6/ip6_input.c (revision c38e7cc395b1472a774ff828e46123de44c628e9)
1 /*	$NetBSD: ip6_input.c,v 1.201 2018/05/01 07:21:39 maxv 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 1.201 2018/05/01 07:21:39 maxv Exp $");
66 
67 #ifdef _KERNEL_OPT
68 #include "opt_gateway.h"
69 #include "opt_inet.h"
70 #include "opt_inet6.h"
71 #include "opt_ipsec.h"
72 #include "opt_net_mpsafe.h"
73 #endif
74 
75 #include <sys/param.h>
76 #include <sys/systm.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 #include <sys/percpu.h>
90 
91 #include <net/if.h>
92 #include <net/if_types.h>
93 #include <net/if_dl.h>
94 #include <net/route.h>
95 #include <net/pktqueue.h>
96 #include <net/pfil.h>
97 
98 #include <netinet/in.h>
99 #include <netinet/in_systm.h>
100 #ifdef INET
101 #include <netinet/ip.h>
102 #include <netinet/ip_var.h>
103 #include <netinet/ip_icmp.h>
104 #endif /* INET */
105 #include <netinet/ip6.h>
106 #include <netinet/portalgo.h>
107 #include <netinet6/in6_var.h>
108 #include <netinet6/ip6_var.h>
109 #include <netinet6/ip6_private.h>
110 #include <netinet6/in6_pcb.h>
111 #include <netinet/icmp6.h>
112 #include <netinet6/scope6_var.h>
113 #include <netinet6/in6_ifattach.h>
114 #include <netinet6/nd6.h>
115 
116 #ifdef IPSEC
117 #include <netipsec/ipsec.h>
118 #include <netipsec/ipsec6.h>
119 #include <netipsec/key.h>
120 #endif /* IPSEC */
121 
122 #include <netinet6/ip6protosw.h>
123 
124 #include "faith.h"
125 
126 extern struct domain inet6domain;
127 
128 u_char ip6_protox[IPPROTO_MAX];
129 pktqueue_t *ip6_pktq __read_mostly;
130 
131 pfil_head_t *inet6_pfil_hook;
132 
133 percpu_t *ip6stat_percpu;
134 
135 percpu_t *ip6_forward_rt_percpu __cacheline_aligned;
136 
137 static void ip6_init2(void);
138 static void ip6intr(void *);
139 static bool ip6_badaddr(struct ip6_hdr *);
140 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
141 
142 static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *,
143     u_int32_t *);
144 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
145 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
146 
147 #ifdef NET_MPSAFE
148 #define	SOFTNET_LOCK()		mutex_enter(softnet_lock)
149 #define	SOFTNET_UNLOCK()	mutex_exit(softnet_lock)
150 #else
151 #define	SOFTNET_LOCK()		KASSERT(mutex_owned(softnet_lock))
152 #define	SOFTNET_UNLOCK()	KASSERT(mutex_owned(softnet_lock))
153 #endif
154 
155 /*
156  * IP6 initialization: fill in IP6 protocol switch table.
157  * All protocols not implemented in kernel go to raw IP6 protocol handler.
158  */
159 void
160 ip6_init(void)
161 {
162 	const struct ip6protosw *pr;
163 	int i;
164 
165 	in6_init();
166 
167 	sysctl_net_inet6_ip6_setup(NULL);
168 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
169 	if (pr == 0)
170 		panic("ip6_init");
171 	for (i = 0; i < IPPROTO_MAX; i++)
172 		ip6_protox[i] = pr - inet6sw;
173 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
174 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
175 		if (pr->pr_domain->dom_family == PF_INET6 &&
176 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
177 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
178 
179 	ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL);
180 	KASSERT(ip6_pktq != NULL);
181 
182 	scope6_init();
183 	addrsel_policy_init();
184 	nd6_init();
185 	frag6_init();
186 	ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR;
187 
188 	ip6_init2();
189 #ifdef GATEWAY
190 	ip6flow_init(ip6_hashsize);
191 #endif
192 	/* Register our Packet Filter hook. */
193 	inet6_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET6);
194 	KASSERT(inet6_pfil_hook != NULL);
195 
196 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
197 	ip6_forward_rt_percpu = percpu_alloc(sizeof(struct route));
198 }
199 
200 static void
201 ip6_init2(void)
202 {
203 
204 	/* timer for regeneration of temporary addresses randomize ID */
205 	callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE);
206 	callout_reset(&in6_tmpaddrtimer_ch,
207 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
208 		       ip6_temp_regen_advance) * hz,
209 		      in6_tmpaddrtimer, NULL);
210 }
211 
212 /*
213  * IP6 input interrupt handling. Just pass the packet to ip6_input.
214  */
215 static void
216 ip6intr(void *arg __unused)
217 {
218 	struct mbuf *m;
219 
220 	SOFTNET_KERNEL_LOCK_UNLESS_NET_MPSAFE();
221 	while ((m = pktq_dequeue(ip6_pktq)) != NULL) {
222 		struct psref psref;
223 		struct ifnet *rcvif = m_get_rcvif_psref(m, &psref);
224 
225 		if (rcvif == NULL) {
226 			m_freem(m);
227 			continue;
228 		}
229 		/*
230 		 * Drop the packet if IPv6 is disabled on the interface.
231 		 */
232 		if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED)) {
233 			m_put_rcvif_psref(rcvif, &psref);
234 			m_freem(m);
235 			continue;
236 		}
237 		ip6_input(m, rcvif);
238 		m_put_rcvif_psref(rcvif, &psref);
239 	}
240 	SOFTNET_KERNEL_UNLOCK_UNLESS_NET_MPSAFE();
241 }
242 
243 void
244 ip6_input(struct mbuf *m, struct ifnet *rcvif)
245 {
246 	struct ip6_hdr *ip6;
247 	int hit, off = sizeof(struct ip6_hdr), nest;
248 	u_int32_t plen;
249 	u_int32_t rtalert = ~0;
250 	int nxt, ours = 0, rh_present = 0, frg_present;
251 	struct ifnet *deliverifp = NULL;
252 	int srcrt = 0;
253 	struct rtentry *rt = NULL;
254 	union {
255 		struct sockaddr		dst;
256 		struct sockaddr_in6	dst6;
257 	} u;
258 	struct route *ro;
259 
260 	KASSERT(rcvif != NULL);
261 
262 	/*
263 	 * make sure we don't have onion peering information into m_tag.
264 	 */
265 	ip6_delaux(m);
266 
267 	/*
268 	 * mbuf statistics
269 	 */
270 	if (m->m_flags & M_EXT) {
271 		if (m->m_next)
272 			IP6_STATINC(IP6_STAT_MEXT2M);
273 		else
274 			IP6_STATINC(IP6_STAT_MEXT1);
275 	} else {
276 #define M2MMAX	32
277 		if (m->m_next) {
278 			if (m->m_flags & M_LOOP)
279 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
280 			else if (rcvif->if_index < M2MMAX)
281 				IP6_STATINC(IP6_STAT_M2M + rcvif->if_index);
282 			else
283 				IP6_STATINC(IP6_STAT_M2M);
284 		} else
285 			IP6_STATINC(IP6_STAT_M1);
286 #undef M2MMAX
287 	}
288 
289 	in6_ifstat_inc(rcvif, ifs6_in_receive);
290 	IP6_STATINC(IP6_STAT_TOTAL);
291 
292 	/*
293 	 * If the IPv6 header is not aligned, slurp it up into a new
294 	 * mbuf with space for link headers, in the event we forward
295 	 * it.  Otherwise, if it is aligned, make sure the entire base
296 	 * IPv6 header is in the first mbuf of the chain.
297 	 */
298 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
299 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
300 		    (max_linkhdr + 3) & ~3)) == NULL) {
301 			/* XXXJRT new stat, please */
302 			IP6_STATINC(IP6_STAT_TOOSMALL);
303 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
304 			return;
305 		}
306 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
307 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
308 			IP6_STATINC(IP6_STAT_TOOSMALL);
309 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
310 			return;
311 		}
312 	}
313 
314 	ip6 = mtod(m, struct ip6_hdr *);
315 
316 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
317 		IP6_STATINC(IP6_STAT_BADVERS);
318 		in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
319 		goto bad;
320 	}
321 
322 	if (ip6_badaddr(ip6)) {
323 		IP6_STATINC(IP6_STAT_BADSCOPE);
324 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
325 		goto bad;
326 	}
327 
328 	/*
329 	 * Assume that we can create a fast-forward IP flow entry
330 	 * based on this packet.
331 	 */
332 	m->m_flags |= M_CANFASTFWD;
333 
334 	/*
335 	 * Run through list of hooks for input packets.  If there are any
336 	 * filters which require that additional packets in the flow are
337 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
338 	 * Note that filters must _never_ set this flag, as another filter
339 	 * in the list may have previously cleared it.
340 	 *
341 	 * Don't call hooks if the packet has already been processed by
342 	 * IPsec (encapsulated, tunnel mode).
343 	 */
344 #if defined(IPSEC)
345 	if (!ipsec_used || !ipsec_indone(m))
346 #else
347 	if (1)
348 #endif
349 	{
350 		struct in6_addr odst;
351 
352 		odst = ip6->ip6_dst;
353 		if (pfil_run_hooks(inet6_pfil_hook, &m, rcvif, PFIL_IN) != 0)
354 			return;
355 		if (m == NULL)
356 			return;
357 		ip6 = mtod(m, struct ip6_hdr *);
358 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
359 	}
360 
361 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
362 
363 #ifdef ALTQ
364 	if (altq_input != NULL) {
365 		SOFTNET_LOCK();
366 		if ((*altq_input)(m, AF_INET6) == 0) {
367 			SOFTNET_UNLOCK();
368 			/* packet is dropped by traffic conditioner */
369 			return;
370 		}
371 		SOFTNET_UNLOCK();
372 	}
373 #endif
374 
375 	/*
376 	 * Disambiguate address scope zones (if there is ambiguity).
377 	 * We first make sure that the original source or destination address
378 	 * is not in our internal form for scoped addresses.  Such addresses
379 	 * are not necessarily invalid spec-wise, but we cannot accept them due
380 	 * to the usage conflict.
381 	 * in6_setscope() then also checks and rejects the cases where src or
382 	 * dst are the loopback address and the receiving interface
383 	 * is not loopback.
384 	 */
385 	if (__predict_false(
386 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
387 		goto bad;
388 	ip6 = mtod(m, struct ip6_hdr *);
389 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
390 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
391 		goto bad;
392 	}
393 	if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
394 	    in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
395 		IP6_STATINC(IP6_STAT_BADSCOPE);
396 		goto bad;
397 	}
398 
399 	ro = percpu_getref(ip6_forward_rt_percpu);
400 
401 	/*
402 	 * Multicast check
403 	 */
404 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
405 		bool ingroup;
406 
407 		in6_ifstat_inc(rcvif, ifs6_in_mcast);
408 		/*
409 		 * See if we belong to the destination multicast group on the
410 		 * arrival interface.
411 		 */
412 		ingroup = in6_multi_group(&ip6->ip6_dst, rcvif);
413 		if (ingroup) {
414 			ours = 1;
415 		} else if (!ip6_mrouter) {
416 			uint64_t *ip6s = IP6_STAT_GETREF();
417 			ip6s[IP6_STAT_NOTMEMBER]++;
418 			ip6s[IP6_STAT_CANTFORWARD]++;
419 			IP6_STAT_PUTREF();
420 			in6_ifstat_inc(rcvif, ifs6_in_discard);
421 			goto bad_unref;
422 		}
423 		deliverifp = rcvif;
424 		goto hbhcheck;
425 	}
426 
427 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
428 
429 	/*
430 	 * Unicast check
431 	 */
432 	rt = rtcache_lookup2(ro, &u.dst, 1, &hit);
433 	if (hit)
434 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
435 	else
436 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
437 
438 	/*
439 	 * Accept the packet if the forwarding interface to the destination
440 	 * (according to the routing table) is the loopback interface,
441 	 * unless the associated route has a gateway.
442 	 *
443 	 * We don't explicitly match ip6_dst against an interface here. It
444 	 * is already done in rtcache_lookup2: rt->rt_ifp->if_type will be
445 	 * IFT_LOOP if the packet is for us.
446 	 *
447 	 * Note that this approach causes to accept a packet if there is a
448 	 * route to the loopback interface for the destination of the packet.
449 	 * But we think it's even useful in some situations, e.g. when using
450 	 * a special daemon which wants to intercept the packet.
451 	 */
452 	if (rt != NULL &&
453 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
454 	    rt->rt_ifp->if_type == IFT_LOOP) {
455 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
456 		int addrok;
457 
458 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
459 			m->m_flags |= M_ANYCAST6;
460 		/*
461 		 * packets to a tentative, duplicated, or somehow invalid
462 		 * address must not be accepted.
463 		 */
464 		if (ia6->ia6_flags & IN6_IFF_NOTREADY)
465 			addrok = 0;
466 		else if (ia6->ia6_flags & IN6_IFF_DETACHED &&
467 		    !IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src))
468 		{
469 			/* Allow internal traffic to DETACHED addresses */
470 			struct sockaddr_in6 sin6;
471 			int s;
472 
473 			memset(&sin6, 0, sizeof(sin6));
474 			sin6.sin6_family = AF_INET6;
475 			sin6.sin6_len = sizeof(sin6);
476 			sin6.sin6_addr = ip6->ip6_src;
477 			s = pserialize_read_enter();
478 			addrok = (ifa_ifwithaddr(sin6tosa(&sin6)) != NULL);
479 			pserialize_read_exit(s);
480 		} else
481 			addrok = 1;
482 		if (addrok) {
483 			/* this address is ready */
484 			ours = 1;
485 			deliverifp = ia6->ia_ifp;	/* correct? */
486 			goto hbhcheck;
487 		} else {
488 			/* address is not ready, so discard the packet. */
489 			char ip6bufs[INET6_ADDRSTRLEN];
490 			char ip6bufd[INET6_ADDRSTRLEN];
491 			nd6log(LOG_INFO, "packet to an unready address %s->%s\n",
492 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
493 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
494 
495 			goto bad_unref;
496 		}
497 	}
498 
499 	/*
500 	 * FAITH (Firewall Aided Internet Translator)
501 	 */
502 #if defined(NFAITH) && 0 < NFAITH
503 	if (ip6_keepfaith) {
504 		if (rt != NULL && rt->rt_ifp != NULL &&
505 		    rt->rt_ifp->if_type == IFT_FAITH) {
506 			/* XXX do we need more sanity checks? */
507 			ours = 1;
508 			deliverifp = rt->rt_ifp; /* faith */
509 			goto hbhcheck;
510 		}
511 	}
512 #endif
513 
514 	/*
515 	 * Now there is no reason to process the packet if it's not our own
516 	 * and we're not a router.
517 	 */
518 	if (!ip6_forwarding) {
519 		IP6_STATINC(IP6_STAT_CANTFORWARD);
520 		in6_ifstat_inc(rcvif, ifs6_in_discard);
521 		goto bad_unref;
522 	}
523 
524 hbhcheck:
525 	/*
526 	 * Record address information into m_tag, if we don't have one yet.
527 	 * Note that we are unable to record it, if the address is not listed
528 	 * as our interface address (e.g. multicast addresses, addresses
529 	 * within FAITH prefixes and such).
530 	 */
531 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
532 		struct in6_ifaddr *ia6;
533 		int s = pserialize_read_enter();
534 
535 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
536 		/* Depends on ip6_setdstifaddr never sleep */
537 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
538 			/*
539 			 * XXX maybe we should drop the packet here,
540 			 * as we could not provide enough information
541 			 * to the upper layers.
542 			 */
543 		}
544 		pserialize_read_exit(s);
545 	}
546 
547 	/*
548 	 * Process Hop-by-Hop options header if it's contained.
549 	 * m may be modified in ip6_hopopts_input().
550 	 * If a JumboPayload option is included, plen will also be modified.
551 	 */
552 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
553 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
554 		struct ip6_hbh *hbh;
555 
556 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
557 			/* m already freed */
558 			in6_ifstat_inc(rcvif, ifs6_in_discard);
559 			rtcache_unref(rt, ro);
560 			percpu_putref(ip6_forward_rt_percpu);
561 			return;
562 		}
563 
564 		/* adjust pointer */
565 		ip6 = mtod(m, struct ip6_hdr *);
566 
567 		/*
568 		 * if the payload length field is 0 and the next header field
569 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
570 		 * option MUST be included.
571 		 */
572 		if (ip6->ip6_plen == 0 && plen == 0) {
573 			/*
574 			 * Note that if a valid jumbo payload option is
575 			 * contained, ip6_hopopts_input() must set a valid
576 			 * (non-zero) payload length to the variable plen.
577 			 */
578 			IP6_STATINC(IP6_STAT_BADOPTIONS);
579 			in6_ifstat_inc(rcvif, ifs6_in_discard);
580 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
581 			icmp6_error(m, ICMP6_PARAM_PROB,
582 				    ICMP6_PARAMPROB_HEADER,
583 				    (char *)&ip6->ip6_plen - (char *)ip6);
584 			rtcache_unref(rt, ro);
585 			percpu_putref(ip6_forward_rt_percpu);
586 			return;
587 		}
588 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
589 			sizeof(struct ip6_hbh));
590 		if (hbh == NULL) {
591 			IP6_STATINC(IP6_STAT_TOOSHORT);
592 			rtcache_unref(rt, ro);
593 			percpu_putref(ip6_forward_rt_percpu);
594 			return;
595 		}
596 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
597 		nxt = hbh->ip6h_nxt;
598 
599 		/*
600 		 * accept the packet if a router alert option is included
601 		 * and we act as an IPv6 router.
602 		 */
603 		if (rtalert != ~0 && ip6_forwarding)
604 			ours = 1;
605 	} else
606 		nxt = ip6->ip6_nxt;
607 
608 	/*
609 	 * Check that the amount of data in the buffers is at least much as
610 	 * the IPv6 header would have us expect. Trim mbufs if longer than we
611 	 * expect. Drop packet if shorter than we expect.
612 	 */
613 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
614 		IP6_STATINC(IP6_STAT_TOOSHORT);
615 		in6_ifstat_inc(rcvif, ifs6_in_truncated);
616 		goto bad_unref;
617 	}
618 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
619 		if (m->m_len == m->m_pkthdr.len) {
620 			m->m_len = sizeof(struct ip6_hdr) + plen;
621 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
622 		} else
623 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
624 	}
625 
626 	/*
627 	 * Forward if desirable.
628 	 */
629 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
630 		/*
631 		 * If we are acting as a multicast router, all
632 		 * incoming multicast packets are passed to the
633 		 * kernel-level multicast forwarding function.
634 		 * The packet is returned (relatively) intact; if
635 		 * ip6_mforward() returns a non-zero value, the packet
636 		 * must be discarded, else it may be accepted below.
637 		 */
638 		if (ip6_mrouter != NULL) {
639 			int error;
640 
641 			SOFTNET_LOCK();
642 			error = ip6_mforward(ip6, rcvif, m);
643 			SOFTNET_UNLOCK();
644 
645 			if (error != 0) {
646 				rtcache_unref(rt, ro);
647 				percpu_putref(ip6_forward_rt_percpu);
648 				IP6_STATINC(IP6_STAT_CANTFORWARD);
649 				goto bad;
650 			}
651 		}
652 		if (!ours)
653 			goto bad_unref;
654 	} else if (!ours) {
655 		rtcache_unref(rt, ro);
656 		percpu_putref(ip6_forward_rt_percpu);
657 		ip6_forward(m, srcrt);
658 		return;
659 	}
660 
661 	ip6 = mtod(m, struct ip6_hdr *);
662 
663 	/*
664 	 * Malicious party may be able to use IPv4 mapped addr to confuse
665 	 * tcp/udp stack and bypass security checks (act as if it was from
666 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
667 	 *
668 	 * For SIIT end node behavior, you may want to disable the check.
669 	 * However, you will  become vulnerable to attacks using IPv4 mapped
670 	 * source.
671 	 */
672 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
673 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
674 		IP6_STATINC(IP6_STAT_BADSCOPE);
675 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
676 		goto bad_unref;
677 	}
678 
679 #ifdef IFA_STATS
680 	if (deliverifp != NULL) {
681 		struct in6_ifaddr *ia6;
682 		int s = pserialize_read_enter();
683 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
684 		if (ia6)
685 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
686 		pserialize_read_exit(s);
687 	}
688 #endif
689 	IP6_STATINC(IP6_STAT_DELIVERED);
690 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
691 	nest = 0;
692 
693 	if (rt != NULL) {
694 		rtcache_unref(rt, ro);
695 		rt = NULL;
696 	}
697 	percpu_putref(ip6_forward_rt_percpu);
698 
699 	rh_present = 0;
700 	frg_present = 0;
701 	while (nxt != IPPROTO_DONE) {
702 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
703 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
704 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
705 			goto bad;
706 		}
707 
708 		M_VERIFY_PACKET(m);
709 
710 		/*
711 		 * protection against faulty packet - there should be
712 		 * more sanity checks in header chain processing.
713 		 */
714 		if (m->m_pkthdr.len < off) {
715 			IP6_STATINC(IP6_STAT_TOOSHORT);
716 			in6_ifstat_inc(rcvif, ifs6_in_truncated);
717 			goto bad;
718 		}
719 
720 		if (nxt == IPPROTO_ROUTING) {
721 			if (rh_present++) {
722 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
723 				IP6_STATINC(IP6_STAT_BADOPTIONS);
724 				goto bad;
725 			}
726 		} else if (nxt == IPPROTO_FRAGMENT) {
727 			if (frg_present++) {
728 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
729 				IP6_STATINC(IP6_STAT_BADOPTIONS);
730 				goto bad;
731 			}
732 		}
733 
734 #ifdef IPSEC
735 		if (ipsec_used) {
736 			/*
737 			 * Enforce IPsec policy checking if we are seeing last
738 			 * header. Note that we do not visit this with
739 			 * protocols with pcb layer code - like udp/tcp/raw ip.
740 			 */
741 			if ((inet6sw[ip_protox[nxt]].pr_flags
742 			    & PR_LASTHDR) != 0) {
743 				int error;
744 
745 				error = ipsec6_input(m);
746 				if (error)
747 					goto bad;
748 			}
749 		}
750 #endif
751 
752 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
753 	}
754 	return;
755 
756 bad_unref:
757 	rtcache_unref(rt, ro);
758 	percpu_putref(ip6_forward_rt_percpu);
759 bad:
760 	m_freem(m);
761 	return;
762 }
763 
764 static bool
765 ip6_badaddr(struct ip6_hdr *ip6)
766 {
767 	/* Check against address spoofing/corruption. */
768 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
769 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
770 		return true;
771 	}
772 
773 	/*
774 	 * The following check is not documented in specs.  A malicious
775 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
776 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
777 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
778 	 *
779 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
780 	 * support IPv4-less kernel compilation, we cannot support SIIT
781 	 * environment at all.  So, it makes more sense for us to reject any
782 	 * malicious packets for non-SIIT environment, than try to do a
783 	 * partial support for SIIT environment.
784 	 */
785 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
786 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
787 		return true;
788 	}
789 
790 	/*
791 	 * Reject packets with IPv4-compatible IPv6 addresses (RFC4291).
792 	 */
793 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
794 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
795 		return true;
796 	}
797 
798 	return false;
799 }
800 
801 /*
802  * set/grab in6_ifaddr correspond to IPv6 destination address.
803  */
804 static struct m_tag *
805 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
806 {
807 	struct m_tag *mtag;
808 	struct ip6aux *ip6a;
809 
810 	mtag = ip6_addaux(m);
811 	if (mtag == NULL)
812 		return NULL;
813 
814 	ip6a = (struct ip6aux *)(mtag + 1);
815 	if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) {
816 		IP6_STATINC(IP6_STAT_BADSCOPE);
817 		return NULL;
818 	}
819 
820 	ip6a->ip6a_src = ia->ia_addr.sin6_addr;
821 	ip6a->ip6a_flags = ia->ia6_flags;
822 	return mtag;
823 }
824 
825 const struct ip6aux *
826 ip6_getdstifaddr(struct mbuf *m)
827 {
828 	struct m_tag *mtag;
829 
830 	mtag = ip6_findaux(m);
831 	if (mtag != NULL)
832 		return (struct ip6aux *)(mtag + 1);
833 	else
834 		return NULL;
835 }
836 
837 /*
838  * Hop-by-Hop options header processing. If a valid jumbo payload option is
839  * included, the real payload length will be stored in plenp.
840  *
841  * rtalertp - XXX: should be stored more smart way
842  */
843 int
844 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
845 	struct mbuf **mp, int *offp)
846 {
847 	struct mbuf *m = *mp;
848 	int off = *offp, hbhlen;
849 	struct ip6_hbh *hbh;
850 
851 	/* validation of the length of the header */
852 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
853 	    sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
854 	if (hbh == NULL) {
855 		IP6_STATINC(IP6_STAT_TOOSHORT);
856 		return -1;
857 	}
858 	hbhlen = (hbh->ip6h_len + 1) << 3;
859 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
860 	    hbhlen);
861 	if (hbh == NULL) {
862 		IP6_STATINC(IP6_STAT_TOOSHORT);
863 		return -1;
864 	}
865 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
866 	off += hbhlen;
867 	hbhlen -= sizeof(struct ip6_hbh);
868 
869 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
870 	    hbhlen, rtalertp, plenp) < 0)
871 		return -1;
872 
873 	*offp = off;
874 	*mp = m;
875 	return 0;
876 }
877 
878 /*
879  * Search header for all Hop-by-hop options and process each option.
880  * This function is separate from ip6_hopopts_input() in order to
881  * handle a case where the sending node itself process its hop-by-hop
882  * options header. In such a case, the function is called from ip6_output().
883  *
884  * The function assumes that hbh header is located right after the IPv6 header
885  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
886  * opthead + hbhlen is located in continuous memory region.
887  */
888 static int
889 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
890 	u_int32_t *rtalertp, u_int32_t *plenp)
891 {
892 	struct ip6_hdr *ip6;
893 	int optlen = 0;
894 	u_int8_t *opt = opthead;
895 	u_int16_t rtalert_val;
896 	u_int32_t jumboplen;
897 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
898 
899 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
900 		switch (*opt) {
901 		case IP6OPT_PAD1:
902 			optlen = 1;
903 			break;
904 		case IP6OPT_PADN:
905 			if (hbhlen < IP6OPT_MINLEN) {
906 				IP6_STATINC(IP6_STAT_TOOSMALL);
907 				goto bad;
908 			}
909 			optlen = *(opt + 1) + 2;
910 			break;
911 		case IP6OPT_RTALERT:
912 			/* XXX may need check for alignment */
913 			if (hbhlen < IP6OPT_RTALERT_LEN) {
914 				IP6_STATINC(IP6_STAT_TOOSMALL);
915 				goto bad;
916 			}
917 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
918 				/* XXX stat */
919 				icmp6_error(m, ICMP6_PARAM_PROB,
920 				    ICMP6_PARAMPROB_HEADER,
921 				    erroff + opt + 1 - opthead);
922 				return (-1);
923 			}
924 			optlen = IP6OPT_RTALERT_LEN;
925 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
926 			*rtalertp = ntohs(rtalert_val);
927 			break;
928 		case IP6OPT_JUMBO:
929 			/* XXX may need check for alignment */
930 			if (hbhlen < IP6OPT_JUMBO_LEN) {
931 				IP6_STATINC(IP6_STAT_TOOSMALL);
932 				goto bad;
933 			}
934 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
935 				/* XXX stat */
936 				icmp6_error(m, ICMP6_PARAM_PROB,
937 				    ICMP6_PARAMPROB_HEADER,
938 				    erroff + opt + 1 - opthead);
939 				return (-1);
940 			}
941 			optlen = IP6OPT_JUMBO_LEN;
942 
943 			/*
944 			 * IPv6 packets that have non 0 payload length
945 			 * must not contain a jumbo payload option.
946 			 */
947 			ip6 = mtod(m, struct ip6_hdr *);
948 			if (ip6->ip6_plen) {
949 				IP6_STATINC(IP6_STAT_BADOPTIONS);
950 				icmp6_error(m, ICMP6_PARAM_PROB,
951 				    ICMP6_PARAMPROB_HEADER,
952 				    erroff + opt - opthead);
953 				return (-1);
954 			}
955 
956 			/*
957 			 * We may see jumbolen in unaligned location, so
958 			 * we'd need to perform memcpy().
959 			 */
960 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
961 			jumboplen = (u_int32_t)htonl(jumboplen);
962 
963 #if 1
964 			/*
965 			 * if there are multiple jumbo payload options,
966 			 * *plenp will be non-zero and the packet will be
967 			 * rejected.
968 			 * the behavior may need some debate in ipngwg -
969 			 * multiple options does not make sense, however,
970 			 * there's no explicit mention in specification.
971 			 */
972 			if (*plenp != 0) {
973 				IP6_STATINC(IP6_STAT_BADOPTIONS);
974 				icmp6_error(m, ICMP6_PARAM_PROB,
975 				    ICMP6_PARAMPROB_HEADER,
976 				    erroff + opt + 2 - opthead);
977 				return (-1);
978 			}
979 #endif
980 
981 			/*
982 			 * jumbo payload length must be larger than 65535.
983 			 */
984 			if (jumboplen <= IPV6_MAXPACKET) {
985 				IP6_STATINC(IP6_STAT_BADOPTIONS);
986 				icmp6_error(m, ICMP6_PARAM_PROB,
987 				    ICMP6_PARAMPROB_HEADER,
988 				    erroff + opt + 2 - opthead);
989 				return (-1);
990 			}
991 			*plenp = jumboplen;
992 
993 			break;
994 		default:		/* unknown option */
995 			if (hbhlen < IP6OPT_MINLEN) {
996 				IP6_STATINC(IP6_STAT_TOOSMALL);
997 				goto bad;
998 			}
999 			optlen = ip6_unknown_opt(opt, m,
1000 			    erroff + opt - opthead);
1001 			if (optlen == -1)
1002 				return (-1);
1003 			optlen += 2;
1004 			break;
1005 		}
1006 	}
1007 
1008 	return (0);
1009 
1010   bad:
1011 	m_freem(m);
1012 	return (-1);
1013 }
1014 
1015 /*
1016  * Unknown option processing.
1017  * The third argument `off' is the offset from the IPv6 header to the option,
1018  * which is necessary if the IPv6 header the and option header and IPv6 header
1019  * is not continuous in order to return an ICMPv6 error.
1020  */
1021 int
1022 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1023 {
1024 	struct ip6_hdr *ip6;
1025 
1026 	switch (IP6OPT_TYPE(*optp)) {
1027 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1028 		return ((int)*(optp + 1));
1029 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1030 		m_freem(m);
1031 		return (-1);
1032 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1033 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1034 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1035 		return (-1);
1036 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1037 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1038 		ip6 = mtod(m, struct ip6_hdr *);
1039 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1040 		    (m->m_flags & (M_BCAST|M_MCAST)))
1041 			m_freem(m);
1042 		else
1043 			icmp6_error(m, ICMP6_PARAM_PROB,
1044 				    ICMP6_PARAMPROB_OPTION, off);
1045 		return (-1);
1046 	}
1047 
1048 	m_freem(m);		/* XXX: NOTREACHED */
1049 	return (-1);
1050 }
1051 
1052 /*
1053  * Create the "control" list for this pcb.
1054  *
1055  * The routine will be called from upper layer handlers like tcp6_input().
1056  * Thus the routine assumes that the caller (tcp6_input) have already
1057  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1058  * very first mbuf on the mbuf chain.
1059  * We may want to add some infinite loop prevention or sanity checks for safety.
1060  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1061  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1062  */
1063 void
1064 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
1065 	struct ip6_hdr *ip6, struct mbuf *m)
1066 {
1067 	struct socket *so = in6p->in6p_socket;
1068 #ifdef RFC2292
1069 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1070 #else
1071 #define IS2292(x, y)	(y)
1072 #endif
1073 
1074 	if (SOOPT_TIMESTAMP(so->so_options))
1075 		mp = sbsavetimestamp(so->so_options, mp);
1076 
1077 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
1078 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1079 		return;
1080 
1081 	/* RFC 2292 sec. 5 */
1082 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1083 		struct in6_pktinfo pi6;
1084 
1085 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
1086 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1087 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index;
1088 		*mp = sbcreatecontrol(&pi6, sizeof(pi6),
1089 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1090 		if (*mp)
1091 			mp = &(*mp)->m_next;
1092 	}
1093 
1094 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1095 		int hlim = ip6->ip6_hlim & 0xff;
1096 
1097 		*mp = sbcreatecontrol(&hlim, sizeof(hlim),
1098 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1099 		if (*mp)
1100 			mp = &(*mp)->m_next;
1101 	}
1102 
1103 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1104 		u_int32_t flowinfo;
1105 		int tclass;
1106 
1107 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1108 		flowinfo >>= 20;
1109 
1110 		tclass = flowinfo & 0xff;
1111 		*mp = sbcreatecontrol(&tclass, sizeof(tclass),
1112 		    IPV6_TCLASS, IPPROTO_IPV6);
1113 
1114 		if (*mp)
1115 			mp = &(*mp)->m_next;
1116 	}
1117 
1118 	/*
1119 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1120 	 * privilege for the option (see ip6_ctloutput), but it might be too
1121 	 * strict, since there might be some hop-by-hop options which can be
1122 	 * returned to normal user.
1123 	 * See also RFC3542 section 8 (or RFC2292 section 6).
1124 	 */
1125 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1126 		/*
1127 		 * Check if a hop-by-hop options header is contatined in the
1128 		 * received packet, and if so, store the options as ancillary
1129 		 * data. Note that a hop-by-hop options header must be
1130 		 * just after the IPv6 header, which fact is assured through
1131 		 * the IPv6 input processing.
1132 		 */
1133 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1134 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1135 			struct ip6_hbh *hbh;
1136 			int hbhlen;
1137 			struct mbuf *ext;
1138 
1139 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1140 			    xip6->ip6_nxt);
1141 			if (ext == NULL) {
1142 				IP6_STATINC(IP6_STAT_TOOSHORT);
1143 				return;
1144 			}
1145 			hbh = mtod(ext, struct ip6_hbh *);
1146 			hbhlen = (hbh->ip6h_len + 1) << 3;
1147 			if (hbhlen != ext->m_len) {
1148 				m_freem(ext);
1149 				IP6_STATINC(IP6_STAT_TOOSHORT);
1150 				return;
1151 			}
1152 
1153 			/*
1154 			 * XXX: We copy whole the header even if a jumbo
1155 			 * payload option is included, which option is to
1156 			 * be removed before returning in the RFC 2292.
1157 			 * Note: this constraint is removed in RFC3542.
1158 			 */
1159 			*mp = sbcreatecontrol(hbh, hbhlen,
1160 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1161 			    IPPROTO_IPV6);
1162 			if (*mp)
1163 				mp = &(*mp)->m_next;
1164 			m_freem(ext);
1165 		}
1166 	}
1167 
1168 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1169 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1170 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1171 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1172 
1173 		/*
1174 		 * Search for destination options headers or routing
1175 		 * header(s) through the header chain, and stores each
1176 		 * header as ancillary data.
1177 		 * Note that the order of the headers remains in
1178 		 * the chain of ancillary data.
1179 		 */
1180 		for (;;) {	/* is explicit loop prevention necessary? */
1181 			struct ip6_ext *ip6e = NULL;
1182 			int elen;
1183 			struct mbuf *ext = NULL;
1184 
1185 			/*
1186 			 * if it is not an extension header, don't try to
1187 			 * pull it from the chain.
1188 			 */
1189 			switch (nxt) {
1190 			case IPPROTO_DSTOPTS:
1191 			case IPPROTO_ROUTING:
1192 			case IPPROTO_HOPOPTS:
1193 			case IPPROTO_AH: /* is it possible? */
1194 				break;
1195 			default:
1196 				goto loopend;
1197 			}
1198 
1199 			ext = ip6_pullexthdr(m, off, nxt);
1200 			if (ext == NULL) {
1201 				IP6_STATINC(IP6_STAT_TOOSHORT);
1202 				return;
1203 			}
1204 			ip6e = mtod(ext, struct ip6_ext *);
1205 			if (nxt == IPPROTO_AH)
1206 				elen = (ip6e->ip6e_len + 2) << 2;
1207 			else
1208 				elen = (ip6e->ip6e_len + 1) << 3;
1209 			if (elen != ext->m_len) {
1210 				m_freem(ext);
1211 				IP6_STATINC(IP6_STAT_TOOSHORT);
1212 				return;
1213 			}
1214 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1215 
1216 			switch (nxt) {
1217 			case IPPROTO_DSTOPTS:
1218 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1219 					break;
1220 
1221 				*mp = sbcreatecontrol(ip6e, elen,
1222 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1223 				    IPPROTO_IPV6);
1224 				if (*mp)
1225 					mp = &(*mp)->m_next;
1226 				break;
1227 
1228 			case IPPROTO_ROUTING:
1229 				if (!(in6p->in6p_flags & IN6P_RTHDR))
1230 					break;
1231 
1232 				*mp = sbcreatecontrol(ip6e, elen,
1233 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1234 				    IPPROTO_IPV6);
1235 				if (*mp)
1236 					mp = &(*mp)->m_next;
1237 				break;
1238 
1239 			case IPPROTO_HOPOPTS:
1240 			case IPPROTO_AH: /* is it possible? */
1241 				break;
1242 
1243 			default:
1244 				/*
1245 			 	 * other cases have been filtered in the above.
1246 				 * none will visit this case.  here we supply
1247 				 * the code just in case (nxt overwritten or
1248 				 * other cases).
1249 				 */
1250 				m_freem(ext);
1251 				goto loopend;
1252 
1253 			}
1254 
1255 			/* proceed with the next header. */
1256 			off += elen;
1257 			nxt = ip6e->ip6e_nxt;
1258 			ip6e = NULL;
1259 			m_freem(ext);
1260 			ext = NULL;
1261 		}
1262 	  loopend:
1263 	  	;
1264 	}
1265 }
1266 #undef IS2292
1267 
1268 
1269 void
1270 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
1271     uint32_t *mtu)
1272 {
1273 	struct socket *so;
1274 	struct mbuf *m_mtu;
1275 	struct ip6_mtuinfo mtuctl;
1276 
1277 	so = in6p->in6p_socket;
1278 
1279 	if (mtu == NULL)
1280 		return;
1281 
1282 	KASSERT(so != NULL);
1283 
1284 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
1285 	mtuctl.ip6m_mtu = *mtu;
1286 	mtuctl.ip6m_addr = *dst;
1287 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
1288 		return;
1289 
1290 	if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl),
1291 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1292 		return;
1293 
1294 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
1295 	    == 0) {
1296 		soroverflow(so);
1297 		m_freem(m_mtu);
1298 	} else
1299 		sorwakeup(so);
1300 
1301 	return;
1302 }
1303 
1304 /*
1305  * pull single extension header from mbuf chain.  returns single mbuf that
1306  * contains the result, or NULL on error.
1307  */
1308 static struct mbuf *
1309 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1310 {
1311 	struct ip6_ext ip6e;
1312 	size_t elen;
1313 	struct mbuf *n;
1314 
1315 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1316 	if (nxt == IPPROTO_AH)
1317 		elen = (ip6e.ip6e_len + 2) << 2;
1318 	else
1319 		elen = (ip6e.ip6e_len + 1) << 3;
1320 
1321 	MGET(n, M_DONTWAIT, MT_DATA);
1322 	if (n && elen >= MLEN) {
1323 		MCLGET(n, M_DONTWAIT);
1324 		if ((n->m_flags & M_EXT) == 0) {
1325 			m_free(n);
1326 			n = NULL;
1327 		}
1328 	}
1329 	if (!n)
1330 		return NULL;
1331 
1332 	n->m_len = 0;
1333 	if (elen >= M_TRAILINGSPACE(n)) {
1334 		m_free(n);
1335 		return NULL;
1336 	}
1337 
1338 	m_copydata(m, off, elen, mtod(n, void *));
1339 	n->m_len = elen;
1340 	return n;
1341 }
1342 
1343 /*
1344  * Get offset to the previous header followed by the header
1345  * currently processed.
1346  */
1347 int
1348 ip6_get_prevhdr(struct mbuf *m, int off)
1349 {
1350 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1351 
1352 	if (off == sizeof(struct ip6_hdr)) {
1353 		return offsetof(struct ip6_hdr, ip6_nxt);
1354 	} else if (off < sizeof(struct ip6_hdr)) {
1355 		panic("%s: off < sizeof(struct ip6_hdr)", __func__);
1356 	} else {
1357 		int len, nlen, nxt;
1358 		struct ip6_ext ip6e;
1359 
1360 		nxt = ip6->ip6_nxt;
1361 		len = sizeof(struct ip6_hdr);
1362 		nlen = 0;
1363 		while (len < off) {
1364 			m_copydata(m, len, sizeof(ip6e), &ip6e);
1365 
1366 			switch (nxt) {
1367 			case IPPROTO_FRAGMENT:
1368 				nlen = sizeof(struct ip6_frag);
1369 				break;
1370 			case IPPROTO_AH:
1371 				nlen = (ip6e.ip6e_len + 2) << 2;
1372 				break;
1373 			default:
1374 				nlen = (ip6e.ip6e_len + 1) << 3;
1375 				break;
1376 			}
1377 			len += nlen;
1378 			nxt = ip6e.ip6e_nxt;
1379 		}
1380 
1381 		return (len - nlen);
1382 	}
1383 }
1384 
1385 /*
1386  * get next header offset.  m will be retained.
1387  */
1388 int
1389 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1390 {
1391 	struct ip6_hdr ip6;
1392 	struct ip6_ext ip6e;
1393 	struct ip6_frag fh;
1394 
1395 	/* just in case */
1396 	if (m == NULL)
1397 		panic("%s: m == NULL", __func__);
1398 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1399 		return -1;
1400 
1401 	switch (proto) {
1402 	case IPPROTO_IPV6:
1403 		/* do not chase beyond intermediate IPv6 headers */
1404 		if (off != 0)
1405 			return -1;
1406 		if (m->m_pkthdr.len < off + sizeof(ip6))
1407 			return -1;
1408 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
1409 		if (nxtp)
1410 			*nxtp = ip6.ip6_nxt;
1411 		off += sizeof(ip6);
1412 		return off;
1413 
1414 	case IPPROTO_FRAGMENT:
1415 		/*
1416 		 * terminate parsing if it is not the first fragment,
1417 		 * it does not make sense to parse through it.
1418 		 */
1419 		if (m->m_pkthdr.len < off + sizeof(fh))
1420 			return -1;
1421 		m_copydata(m, off, sizeof(fh), (void *)&fh);
1422 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1423 			return -1;
1424 		if (nxtp)
1425 			*nxtp = fh.ip6f_nxt;
1426 		off += sizeof(struct ip6_frag);
1427 		return off;
1428 
1429 	case IPPROTO_AH:
1430 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1431 			return -1;
1432 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1433 		if (nxtp)
1434 			*nxtp = ip6e.ip6e_nxt;
1435 		off += (ip6e.ip6e_len + 2) << 2;
1436 		if (m->m_pkthdr.len < off)
1437 			return -1;
1438 		return off;
1439 
1440 	case IPPROTO_HOPOPTS:
1441 	case IPPROTO_ROUTING:
1442 	case IPPROTO_DSTOPTS:
1443 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1444 			return -1;
1445 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1446 		if (nxtp)
1447 			*nxtp = ip6e.ip6e_nxt;
1448 		off += (ip6e.ip6e_len + 1) << 3;
1449 		if (m->m_pkthdr.len < off)
1450 			return -1;
1451 		return off;
1452 
1453 	case IPPROTO_NONE:
1454 	case IPPROTO_ESP:
1455 	case IPPROTO_IPCOMP:
1456 		/* give up */
1457 		return -1;
1458 
1459 	default:
1460 		return -1;
1461 	}
1462 }
1463 
1464 /*
1465  * get offset for the last header in the chain.  m will be kept untainted.
1466  */
1467 int
1468 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1469 {
1470 	int newoff;
1471 	int nxt;
1472 
1473 	if (!nxtp) {
1474 		nxt = -1;
1475 		nxtp = &nxt;
1476 	}
1477 	for (;;) {
1478 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1479 		if (newoff < 0)
1480 			return off;
1481 		else if (newoff < off)
1482 			return -1;	/* invalid */
1483 		else if (newoff == off)
1484 			return newoff;
1485 
1486 		off = newoff;
1487 		proto = *nxtp;
1488 	}
1489 }
1490 
1491 struct m_tag *
1492 ip6_addaux(struct mbuf *m)
1493 {
1494 	struct m_tag *mtag;
1495 
1496 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1497 	if (!mtag) {
1498 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1499 		    M_NOWAIT);
1500 		if (mtag) {
1501 			m_tag_prepend(m, mtag);
1502 			memset(mtag + 1, 0, sizeof(struct ip6aux));
1503 		}
1504 	}
1505 	return mtag;
1506 }
1507 
1508 struct m_tag *
1509 ip6_findaux(struct mbuf *m)
1510 {
1511 	struct m_tag *mtag;
1512 
1513 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1514 	return mtag;
1515 }
1516 
1517 void
1518 ip6_delaux(struct mbuf *m)
1519 {
1520 	struct m_tag *mtag;
1521 
1522 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
1523 	if (mtag)
1524 		m_tag_delete(m, mtag);
1525 }
1526 
1527 /*
1528  * System control for IP6
1529  */
1530 
1531 const u_char inet6ctlerrmap[PRC_NCMDS] = {
1532 	0,		0,		0,		0,
1533 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1534 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1535 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1536 	0,		0,		0,		0,
1537 	ENOPROTOOPT
1538 };
1539 
1540 extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS);
1541 
1542 static int
1543 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
1544 {
1545 
1546 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
1547 }
1548 
1549 static void
1550 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
1551 {
1552 
1553 	sysctl_createv(clog, 0, NULL, NULL,
1554 		       CTLFLAG_PERMANENT,
1555 		       CTLTYPE_NODE, "inet6",
1556 		       SYSCTL_DESCR("PF_INET6 related settings"),
1557 		       NULL, 0, NULL, 0,
1558 		       CTL_NET, PF_INET6, CTL_EOL);
1559 	sysctl_createv(clog, 0, NULL, NULL,
1560 		       CTLFLAG_PERMANENT,
1561 		       CTLTYPE_NODE, "ip6",
1562 		       SYSCTL_DESCR("IPv6 related settings"),
1563 		       NULL, 0, NULL, 0,
1564 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1565 
1566 	sysctl_createv(clog, 0, NULL, NULL,
1567 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1568 		       CTLTYPE_INT, "forwarding",
1569 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1570 		       NULL, 0, &ip6_forwarding, 0,
1571 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1572 		       IPV6CTL_FORWARDING, CTL_EOL);
1573 	sysctl_createv(clog, 0, NULL, NULL,
1574 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1575 		       CTLTYPE_INT, "redirect",
1576 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1577 		       NULL, 0, &ip6_sendredirects, 0,
1578 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1579 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1580 	sysctl_createv(clog, 0, NULL, NULL,
1581 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1582 		       CTLTYPE_INT, "hlim",
1583 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1584 		       NULL, 0, &ip6_defhlim, 0,
1585 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1586 		       IPV6CTL_DEFHLIM, CTL_EOL);
1587 	sysctl_createv(clog, 0, NULL, NULL,
1588 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1589 		       CTLTYPE_INT, "maxfragpackets",
1590 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1591 				    "for reassembly"),
1592 		       NULL, 0, &ip6_maxfragpackets, 0,
1593 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1594 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1595 	sysctl_createv(clog, 0, NULL, NULL,
1596 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1597 		       CTLTYPE_INT, "accept_rtadv",
1598 		       SYSCTL_DESCR("Accept router advertisements"),
1599 		       NULL, 0, &ip6_accept_rtadv, 0,
1600 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1601 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1602 	sysctl_createv(clog, 0, NULL, NULL,
1603 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1604 		       CTLTYPE_INT, "rtadv_maxroutes",
1605 		       SYSCTL_DESCR("Maximum number of routes accepted via router advertisements"),
1606 		       NULL, 0, &ip6_rtadv_maxroutes, 0,
1607 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1608 		       IPV6CTL_RTADV_MAXROUTES, CTL_EOL);
1609 	sysctl_createv(clog, 0, NULL, NULL,
1610 		       CTLFLAG_PERMANENT,
1611 		       CTLTYPE_INT, "rtadv_numroutes",
1612 		       SYSCTL_DESCR("Current number of routes accepted via router advertisements"),
1613 		       NULL, 0, &nd6_numroutes, 0,
1614 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1615 		       IPV6CTL_RTADV_NUMROUTES, CTL_EOL);
1616 	sysctl_createv(clog, 0, NULL, NULL,
1617 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1618 		       CTLTYPE_INT, "keepfaith",
1619 		       SYSCTL_DESCR("Activate faith interface"),
1620 		       NULL, 0, &ip6_keepfaith, 0,
1621 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1622 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1623 	sysctl_createv(clog, 0, NULL, NULL,
1624 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1625 		       CTLTYPE_INT, "log_interval",
1626 		       SYSCTL_DESCR("Minimum interval between logging "
1627 				    "unroutable packets"),
1628 		       NULL, 0, &ip6_log_interval, 0,
1629 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1630 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1631 	sysctl_createv(clog, 0, NULL, NULL,
1632 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1633 		       CTLTYPE_INT, "hdrnestlimit",
1634 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1635 		       NULL, 0, &ip6_hdrnestlimit, 0,
1636 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1637 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1638 	sysctl_createv(clog, 0, NULL, NULL,
1639 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1640 		       CTLTYPE_INT, "dad_count",
1641 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1642 				    "probes to send"),
1643 		       NULL, 0, &ip6_dad_count, 0,
1644 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1645 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1646 	sysctl_createv(clog, 0, NULL, NULL,
1647 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1648 		       CTLTYPE_INT, "auto_flowlabel",
1649 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1650 		       NULL, 0, &ip6_auto_flowlabel, 0,
1651 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1652 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1653 	sysctl_createv(clog, 0, NULL, NULL,
1654 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1655 		       CTLTYPE_INT, "defmcasthlim",
1656 		       SYSCTL_DESCR("Default multicast hop limit"),
1657 		       NULL, 0, &ip6_defmcasthlim, 0,
1658 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1659 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1660 	sysctl_createv(clog, 0, NULL, NULL,
1661 		       CTLFLAG_PERMANENT,
1662 		       CTLTYPE_STRING, "kame_version",
1663 		       SYSCTL_DESCR("KAME Version"),
1664 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1665 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1666 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1667 	sysctl_createv(clog, 0, NULL, NULL,
1668 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1669 		       CTLTYPE_INT, "use_deprecated",
1670 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1671 				    "source addresses"),
1672 		       NULL, 0, &ip6_use_deprecated, 0,
1673 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1674 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1675 	sysctl_createv(clog, 0, NULL, NULL,
1676 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1677 		       CTLTYPE_INT, "rr_prune", NULL,
1678 		       NULL, 0, &ip6_rr_prune, 0,
1679 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1680 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1681 	sysctl_createv(clog, 0, NULL, NULL,
1682 		       CTLFLAG_PERMANENT
1683 #ifndef INET6_BINDV6ONLY
1684 		       |CTLFLAG_READWRITE,
1685 #endif
1686 		       CTLTYPE_INT, "v6only",
1687 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1688 				    "to PF_INET sockets"),
1689 		       NULL, 0, &ip6_v6only, 0,
1690 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1691 		       IPV6CTL_V6ONLY, CTL_EOL);
1692 	sysctl_createv(clog, 0, NULL, NULL,
1693 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1694 		       CTLTYPE_INT, "anonportmin",
1695 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1696 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1697 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1698 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1699 	sysctl_createv(clog, 0, NULL, NULL,
1700 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1701 		       CTLTYPE_INT, "anonportmax",
1702 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1703 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1704 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1705 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1706 #ifndef IPNOPRIVPORTS
1707 	sysctl_createv(clog, 0, NULL, NULL,
1708 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1709 		       CTLTYPE_INT, "lowportmin",
1710 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1711 				    "to assign"),
1712 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1713 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1714 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1715 	sysctl_createv(clog, 0, NULL, NULL,
1716 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1717 		       CTLTYPE_INT, "lowportmax",
1718 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1719 				    "to assign"),
1720 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1721 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1722 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1723 #endif /* IPNOPRIVPORTS */
1724 	sysctl_createv(clog, 0, NULL, NULL,
1725 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1726 		       CTLTYPE_INT, "auto_linklocal",
1727 		       SYSCTL_DESCR("Default value of per-interface flag for "
1728 		                    "adding an IPv6 link-local address to "
1729 				    "interfaces when attached"),
1730 		       NULL, 0, &ip6_auto_linklocal, 0,
1731 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1732 		       IPV6CTL_AUTO_LINKLOCAL, CTL_EOL);
1733 	sysctl_createv(clog, 0, NULL, NULL,
1734 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
1735 		       CTLTYPE_STRUCT, "addctlpolicy",
1736 		       SYSCTL_DESCR("Return the current address control"
1737 			   " policy"),
1738 		       sysctl_net_inet6_addrctlpolicy, 0, NULL, 0,
1739 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1740 		       IPV6CTL_ADDRCTLPOLICY, CTL_EOL);
1741 	sysctl_createv(clog, 0, NULL, NULL,
1742 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1743 		       CTLTYPE_INT, "use_tempaddr",
1744 		       SYSCTL_DESCR("Use temporary address"),
1745 		       NULL, 0, &ip6_use_tempaddr, 0,
1746 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1747 		       CTL_CREATE, CTL_EOL);
1748 	sysctl_createv(clog, 0, NULL, NULL,
1749 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1750 		       CTLTYPE_INT, "prefer_tempaddr",
1751 		       SYSCTL_DESCR("Prefer temporary address as source "
1752 		                    "address"),
1753 		       NULL, 0, &ip6_prefer_tempaddr, 0,
1754 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1755 		       CTL_CREATE, CTL_EOL);
1756 	sysctl_createv(clog, 0, NULL, NULL,
1757 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1758 		       CTLTYPE_INT, "temppltime",
1759 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
1760 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
1761 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1762 		       CTL_CREATE, CTL_EOL);
1763 	sysctl_createv(clog, 0, NULL, NULL,
1764 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1765 		       CTLTYPE_INT, "tempvltime",
1766 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
1767 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
1768 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1769 		       CTL_CREATE, CTL_EOL);
1770 	sysctl_createv(clog, 0, NULL, NULL,
1771 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1772 		       CTLTYPE_INT, "maxfrags",
1773 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1774 		       NULL, 0, &ip6_maxfrags, 0,
1775 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1776 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1777 	sysctl_createv(clog, 0, NULL, NULL,
1778 		       CTLFLAG_PERMANENT,
1779 		       CTLTYPE_STRUCT, "stats",
1780 		       SYSCTL_DESCR("IPv6 statistics"),
1781 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
1782 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1783 		       IPV6CTL_STATS, CTL_EOL);
1784 	sysctl_createv(clog, 0, NULL, NULL,
1785 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1786 		       CTLTYPE_INT, "use_defaultzone",
1787 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1788 		       NULL, 0, &ip6_use_defzone, 0,
1789 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1790 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1791 	sysctl_createv(clog, 0, NULL, NULL,
1792 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1793 		       CTLTYPE_INT, "mcast_pmtu",
1794 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
1795 		       NULL, 0, &ip6_mcast_pmtu, 0,
1796 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1797 		       CTL_CREATE, CTL_EOL);
1798 	/* anonportalgo RFC6056 subtree */
1799 	const struct sysctlnode *portalgo_node;
1800 	sysctl_createv(clog, 0, NULL, &portalgo_node,
1801 		       CTLFLAG_PERMANENT,
1802 		       CTLTYPE_NODE, "anonportalgo",
1803 		       SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"),
1804 	    	       NULL, 0, NULL, 0,
1805 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL);
1806 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1807 		       CTLFLAG_PERMANENT,
1808 		       CTLTYPE_STRING, "available",
1809 		       SYSCTL_DESCR("available algorithms"),
1810 		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1811 		       CTL_CREATE, CTL_EOL);
1812 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1813 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1814 		       CTLTYPE_STRING, "selected",
1815 		       SYSCTL_DESCR("selected algorithm"),
1816 	               sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN,
1817 		       CTL_CREATE, CTL_EOL);
1818 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1819 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1820 		       CTLTYPE_STRUCT, "reserve",
1821 		       SYSCTL_DESCR("bitmap of reserved ports"),
1822 		       sysctl_portalgo_reserve6, 0, NULL, 0,
1823 		       CTL_CREATE, CTL_EOL);
1824 	sysctl_createv(clog, 0, NULL, NULL,
1825 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1826 		       CTLTYPE_INT, "neighborgcthresh",
1827 		       SYSCTL_DESCR("Maximum number of entries in neighbor"
1828 			" cache"),
1829 		       NULL, 1, &ip6_neighborgcthresh, 0,
1830 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1831 		       CTL_CREATE, CTL_EOL);
1832 	sysctl_createv(clog, 0, NULL, NULL,
1833 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1834 		       CTLTYPE_INT, "maxifprefixes",
1835 		       SYSCTL_DESCR("Maximum number of prefixes created by"
1836 			   " route advertisement per interface"),
1837 		       NULL, 1, &ip6_maxifprefixes, 0,
1838 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1839 		       CTL_CREATE, CTL_EOL);
1840 	sysctl_createv(clog, 0, NULL, NULL,
1841 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1842 		       CTLTYPE_INT, "maxifdefrouters",
1843 		       SYSCTL_DESCR("Maximum number of default routers created"
1844 			   " by route advertisement per interface"),
1845 		       NULL, 1, &ip6_maxifdefrouters, 0,
1846 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1847 		       CTL_CREATE, CTL_EOL);
1848 	sysctl_createv(clog, 0, NULL, NULL,
1849 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1850 		       CTLTYPE_INT, "maxdynroutes",
1851 		       SYSCTL_DESCR("Maximum number of routes created via"
1852 			   " redirect"),
1853 		       NULL, 1, &ip6_maxdynroutes, 0,
1854 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1855 		       CTL_CREATE, CTL_EOL);
1856 }
1857 
1858 void
1859 ip6_statinc(u_int stat)
1860 {
1861 
1862 	KASSERT(stat < IP6_NSTATS);
1863 	IP6_STATINC(stat);
1864 }
1865