xref: /netbsd-src/sys/netinet6/ip6_input.c (revision f3cfa6f6ce31685c6c4a758bc430e69eb99f50a4)
1 /*	$NetBSD: ip6_input.c,v 1.208 2019/05/13 07:47:59 ozaki-r 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 1.208 2019/05/13 07:47:59 ozaki-r 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_skip_pfil(m))
346 #else
347 	if (1)
348 #endif
349 	{
350 		struct in6_addr odst;
351 		int error;
352 
353 		odst = ip6->ip6_dst;
354 		error = pfil_run_hooks(inet6_pfil_hook, &m, rcvif, PFIL_IN);
355 		if (error != 0 || m == NULL) {
356 			IP6_STATINC(IP6_STAT_PFILDROP_IN);
357 			return;
358 		}
359 		KASSERT(m->m_len >= sizeof(struct ip6_hdr));
360 		ip6 = mtod(m, struct ip6_hdr *);
361 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
362 	}
363 
364 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
365 
366 #ifdef ALTQ
367 	if (altq_input != NULL) {
368 		SOFTNET_LOCK();
369 		if ((*altq_input)(m, AF_INET6) == 0) {
370 			SOFTNET_UNLOCK();
371 			/* packet is dropped by traffic conditioner */
372 			return;
373 		}
374 		SOFTNET_UNLOCK();
375 	}
376 #endif
377 
378 	/*
379 	 * Disambiguate address scope zones (if there is ambiguity).
380 	 * We first make sure that the original source or destination address
381 	 * is not in our internal form for scoped addresses.  Such addresses
382 	 * are not necessarily invalid spec-wise, but we cannot accept them due
383 	 * to the usage conflict.
384 	 * in6_setscope() then also checks and rejects the cases where src or
385 	 * dst are the loopback address and the receiving interface
386 	 * is not loopback.
387 	 */
388 	if (__predict_false(
389 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
390 		goto bad;
391 	ip6 = mtod(m, struct ip6_hdr *);
392 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
393 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
394 		goto bad;
395 	}
396 	if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
397 	    in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
398 		IP6_STATINC(IP6_STAT_BADSCOPE);
399 		goto bad;
400 	}
401 
402 	ro = percpu_getref(ip6_forward_rt_percpu);
403 
404 	/*
405 	 * Multicast check
406 	 */
407 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
408 		bool ingroup;
409 
410 		in6_ifstat_inc(rcvif, ifs6_in_mcast);
411 		/*
412 		 * See if we belong to the destination multicast group on the
413 		 * arrival interface.
414 		 */
415 		ingroup = in6_multi_group(&ip6->ip6_dst, rcvif);
416 		if (ingroup) {
417 			ours = 1;
418 		} else if (!ip6_mrouter) {
419 			uint64_t *ip6s = IP6_STAT_GETREF();
420 			ip6s[IP6_STAT_NOTMEMBER]++;
421 			ip6s[IP6_STAT_CANTFORWARD]++;
422 			IP6_STAT_PUTREF();
423 			in6_ifstat_inc(rcvif, ifs6_in_discard);
424 			goto bad_unref;
425 		}
426 		deliverifp = rcvif;
427 		goto hbhcheck;
428 	}
429 
430 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
431 
432 	/*
433 	 * Unicast check
434 	 */
435 	rt = rtcache_lookup2(ro, &u.dst, 1, &hit);
436 	if (hit)
437 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
438 	else
439 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
440 
441 	/*
442 	 * Accept the packet if the forwarding interface to the destination
443 	 * (according to the routing table) is the loopback interface,
444 	 * unless the associated route has a gateway.
445 	 *
446 	 * We don't explicitly match ip6_dst against an interface here. It
447 	 * is already done in rtcache_lookup2: rt->rt_ifp->if_type will be
448 	 * IFT_LOOP if the packet is for us.
449 	 *
450 	 * Note that this approach causes to accept a packet if there is a
451 	 * route to the loopback interface for the destination of the packet.
452 	 * But we think it's even useful in some situations, e.g. when using
453 	 * a special daemon which wants to intercept the packet.
454 	 */
455 	if (rt != NULL &&
456 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
457 	    rt->rt_ifp->if_type == IFT_LOOP) {
458 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
459 		int addrok;
460 
461 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
462 			m->m_flags |= M_ANYCAST6;
463 		/*
464 		 * packets to a tentative, duplicated, or somehow invalid
465 		 * address must not be accepted.
466 		 */
467 		if (ia6->ia6_flags & IN6_IFF_NOTREADY)
468 			addrok = 0;
469 		else if (ia6->ia6_flags & IN6_IFF_DETACHED &&
470 		    !IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src))
471 		{
472 			/* Allow internal traffic to DETACHED addresses */
473 			struct sockaddr_in6 sin6;
474 			int s;
475 
476 			memset(&sin6, 0, sizeof(sin6));
477 			sin6.sin6_family = AF_INET6;
478 			sin6.sin6_len = sizeof(sin6);
479 			sin6.sin6_addr = ip6->ip6_src;
480 			s = pserialize_read_enter();
481 			addrok = (ifa_ifwithaddr(sin6tosa(&sin6)) != NULL);
482 			pserialize_read_exit(s);
483 		} else
484 			addrok = 1;
485 		if (addrok) {
486 			/* this address is ready */
487 			ours = 1;
488 			deliverifp = ia6->ia_ifp;	/* correct? */
489 			goto hbhcheck;
490 		} else {
491 			/* address is not ready, so discard the packet. */
492 			char ip6bufs[INET6_ADDRSTRLEN];
493 			char ip6bufd[INET6_ADDRSTRLEN];
494 			nd6log(LOG_INFO, "packet to an unready address %s->%s\n",
495 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
496 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
497 
498 			goto bad_unref;
499 		}
500 	}
501 
502 	/*
503 	 * FAITH (Firewall Aided Internet Translator)
504 	 */
505 #if defined(NFAITH) && 0 < NFAITH
506 	if (ip6_keepfaith) {
507 		if (rt != NULL && rt->rt_ifp != NULL &&
508 		    rt->rt_ifp->if_type == IFT_FAITH) {
509 			/* XXX do we need more sanity checks? */
510 			ours = 1;
511 			deliverifp = rt->rt_ifp; /* faith */
512 			goto hbhcheck;
513 		}
514 	}
515 #endif
516 
517 	/*
518 	 * Now there is no reason to process the packet if it's not our own
519 	 * and we're not a router.
520 	 */
521 	if (!ip6_forwarding) {
522 		IP6_STATINC(IP6_STAT_CANTFORWARD);
523 		in6_ifstat_inc(rcvif, ifs6_in_discard);
524 		goto bad_unref;
525 	}
526 
527 hbhcheck:
528 	/*
529 	 * Record address information into m_tag, if we don't have one yet.
530 	 * Note that we are unable to record it, if the address is not listed
531 	 * as our interface address (e.g. multicast addresses, addresses
532 	 * within FAITH prefixes and such).
533 	 */
534 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
535 		struct in6_ifaddr *ia6;
536 		int s = pserialize_read_enter();
537 
538 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
539 		/* Depends on ip6_setdstifaddr never sleep */
540 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
541 			/*
542 			 * XXX maybe we should drop the packet here,
543 			 * as we could not provide enough information
544 			 * to the upper layers.
545 			 */
546 		}
547 		pserialize_read_exit(s);
548 	}
549 
550 	/*
551 	 * Process Hop-by-Hop options header if it's contained.
552 	 * m may be modified in ip6_hopopts_input().
553 	 * If a JumboPayload option is included, plen will also be modified.
554 	 */
555 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
556 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
557 		struct ip6_hbh *hbh;
558 
559 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
560 			/* m already freed */
561 			in6_ifstat_inc(rcvif, ifs6_in_discard);
562 			rtcache_unref(rt, ro);
563 			percpu_putref(ip6_forward_rt_percpu);
564 			return;
565 		}
566 
567 		/* adjust pointer */
568 		ip6 = mtod(m, struct ip6_hdr *);
569 
570 		/*
571 		 * if the payload length field is 0 and the next header field
572 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
573 		 * option MUST be included.
574 		 */
575 		if (ip6->ip6_plen == 0 && plen == 0) {
576 			/*
577 			 * Note that if a valid jumbo payload option is
578 			 * contained, ip6_hopopts_input() must set a valid
579 			 * (non-zero) payload length to the variable plen.
580 			 */
581 			IP6_STATINC(IP6_STAT_BADOPTIONS);
582 			in6_ifstat_inc(rcvif, ifs6_in_discard);
583 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
584 			icmp6_error(m, ICMP6_PARAM_PROB,
585 				    ICMP6_PARAMPROB_HEADER,
586 				    (char *)&ip6->ip6_plen - (char *)ip6);
587 			rtcache_unref(rt, ro);
588 			percpu_putref(ip6_forward_rt_percpu);
589 			return;
590 		}
591 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
592 			sizeof(struct ip6_hbh));
593 		if (hbh == NULL) {
594 			IP6_STATINC(IP6_STAT_TOOSHORT);
595 			rtcache_unref(rt, ro);
596 			percpu_putref(ip6_forward_rt_percpu);
597 			return;
598 		}
599 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
600 		nxt = hbh->ip6h_nxt;
601 
602 		/*
603 		 * accept the packet if a router alert option is included
604 		 * and we act as an IPv6 router.
605 		 */
606 		if (rtalert != ~0 && ip6_forwarding)
607 			ours = 1;
608 	} else
609 		nxt = ip6->ip6_nxt;
610 
611 	/*
612 	 * Check that the amount of data in the buffers is at least much as
613 	 * the IPv6 header would have us expect. Trim mbufs if longer than we
614 	 * expect. Drop packet if shorter than we expect.
615 	 */
616 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
617 		IP6_STATINC(IP6_STAT_TOOSHORT);
618 		in6_ifstat_inc(rcvif, ifs6_in_truncated);
619 		goto bad_unref;
620 	}
621 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
622 		if (m->m_len == m->m_pkthdr.len) {
623 			m->m_len = sizeof(struct ip6_hdr) + plen;
624 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
625 		} else
626 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
627 	}
628 
629 	/*
630 	 * Forward if desirable.
631 	 */
632 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
633 		/*
634 		 * If we are acting as a multicast router, all
635 		 * incoming multicast packets are passed to the
636 		 * kernel-level multicast forwarding function.
637 		 * The packet is returned (relatively) intact; if
638 		 * ip6_mforward() returns a non-zero value, the packet
639 		 * must be discarded, else it may be accepted below.
640 		 */
641 		if (ip6_mrouter != NULL) {
642 			int error;
643 
644 			SOFTNET_LOCK();
645 			error = ip6_mforward(ip6, rcvif, m);
646 			SOFTNET_UNLOCK();
647 
648 			if (error != 0) {
649 				rtcache_unref(rt, ro);
650 				percpu_putref(ip6_forward_rt_percpu);
651 				IP6_STATINC(IP6_STAT_CANTFORWARD);
652 				goto bad;
653 			}
654 		}
655 		if (!ours)
656 			goto bad_unref;
657 	} else if (!ours) {
658 		rtcache_unref(rt, ro);
659 		percpu_putref(ip6_forward_rt_percpu);
660 		ip6_forward(m, srcrt);
661 		return;
662 	}
663 
664 	ip6 = mtod(m, struct ip6_hdr *);
665 
666 	/*
667 	 * Malicious party may be able to use IPv4 mapped addr to confuse
668 	 * tcp/udp stack and bypass security checks (act as if it was from
669 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
670 	 *
671 	 * For SIIT end node behavior, you may want to disable the check.
672 	 * However, you will  become vulnerable to attacks using IPv4 mapped
673 	 * source.
674 	 */
675 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
676 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
677 		IP6_STATINC(IP6_STAT_BADSCOPE);
678 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
679 		goto bad_unref;
680 	}
681 
682 #ifdef IFA_STATS
683 	if (deliverifp != NULL) {
684 		struct in6_ifaddr *ia6;
685 		int s = pserialize_read_enter();
686 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
687 		if (ia6)
688 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
689 		pserialize_read_exit(s);
690 	}
691 #endif
692 	IP6_STATINC(IP6_STAT_DELIVERED);
693 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
694 	nest = 0;
695 
696 	if (rt != NULL) {
697 		rtcache_unref(rt, ro);
698 		rt = NULL;
699 	}
700 	percpu_putref(ip6_forward_rt_percpu);
701 
702 	rh_present = 0;
703 	frg_present = 0;
704 	while (nxt != IPPROTO_DONE) {
705 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
706 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
707 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
708 			goto bad;
709 		}
710 
711 		M_VERIFY_PACKET(m);
712 
713 		/*
714 		 * protection against faulty packet - there should be
715 		 * more sanity checks in header chain processing.
716 		 */
717 		if (m->m_pkthdr.len < off) {
718 			IP6_STATINC(IP6_STAT_TOOSHORT);
719 			in6_ifstat_inc(rcvif, ifs6_in_truncated);
720 			goto bad;
721 		}
722 
723 		if (nxt == IPPROTO_ROUTING) {
724 			if (rh_present++) {
725 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
726 				IP6_STATINC(IP6_STAT_BADOPTIONS);
727 				goto bad;
728 			}
729 		} else if (nxt == IPPROTO_FRAGMENT) {
730 			if (frg_present++) {
731 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
732 				IP6_STATINC(IP6_STAT_BADOPTIONS);
733 				goto bad;
734 			}
735 		}
736 
737 #ifdef IPSEC
738 		if (ipsec_used) {
739 			/*
740 			 * Enforce IPsec policy checking if we are seeing last
741 			 * header. Note that we do not visit this with
742 			 * protocols with pcb layer code - like udp/tcp/raw ip.
743 			 */
744 			if ((inet6sw[ip6_protox[nxt]].pr_flags
745 			    & PR_LASTHDR) != 0) {
746 				int error;
747 
748 				error = ipsec_ip_input(m, false);
749 				if (error)
750 					goto bad;
751 			}
752 		}
753 #endif
754 
755 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
756 	}
757 	return;
758 
759 bad_unref:
760 	rtcache_unref(rt, ro);
761 	percpu_putref(ip6_forward_rt_percpu);
762 bad:
763 	m_freem(m);
764 	return;
765 }
766 
767 static bool
768 ip6_badaddr(struct ip6_hdr *ip6)
769 {
770 	/* Check against address spoofing/corruption. */
771 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
772 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
773 		return true;
774 	}
775 
776 	/*
777 	 * The following check is not documented in specs.  A malicious
778 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
779 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
780 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
781 	 *
782 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
783 	 * support IPv4-less kernel compilation, we cannot support SIIT
784 	 * environment at all.  So, it makes more sense for us to reject any
785 	 * malicious packets for non-SIIT environment, than try to do a
786 	 * partial support for SIIT environment.
787 	 */
788 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
789 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
790 		return true;
791 	}
792 
793 	/*
794 	 * Reject packets with IPv4-compatible IPv6 addresses (RFC4291).
795 	 */
796 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
797 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
798 		return true;
799 	}
800 
801 	return false;
802 }
803 
804 /*
805  * set/grab in6_ifaddr correspond to IPv6 destination address.
806  */
807 static struct m_tag *
808 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
809 {
810 	struct m_tag *mtag;
811 	struct ip6aux *ip6a;
812 
813 	mtag = ip6_addaux(m);
814 	if (mtag == NULL)
815 		return NULL;
816 
817 	ip6a = (struct ip6aux *)(mtag + 1);
818 	if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) {
819 		IP6_STATINC(IP6_STAT_BADSCOPE);
820 		return NULL;
821 	}
822 
823 	ip6a->ip6a_src = ia->ia_addr.sin6_addr;
824 	ip6a->ip6a_flags = ia->ia6_flags;
825 	return mtag;
826 }
827 
828 const struct ip6aux *
829 ip6_getdstifaddr(struct mbuf *m)
830 {
831 	struct m_tag *mtag;
832 
833 	mtag = ip6_findaux(m);
834 	if (mtag != NULL)
835 		return (struct ip6aux *)(mtag + 1);
836 	else
837 		return NULL;
838 }
839 
840 /*
841  * Hop-by-Hop options header processing. If a valid jumbo payload option is
842  * included, the real payload length will be stored in plenp.
843  *
844  * rtalertp - XXX: should be stored more smart way
845  */
846 int
847 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
848 	struct mbuf **mp, int *offp)
849 {
850 	struct mbuf *m = *mp;
851 	int off = *offp, hbhlen;
852 	struct ip6_hbh *hbh;
853 
854 	/* validation of the length of the header */
855 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
856 	    sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
857 	if (hbh == NULL) {
858 		IP6_STATINC(IP6_STAT_TOOSHORT);
859 		return -1;
860 	}
861 	hbhlen = (hbh->ip6h_len + 1) << 3;
862 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
863 	    hbhlen);
864 	if (hbh == NULL) {
865 		IP6_STATINC(IP6_STAT_TOOSHORT);
866 		return -1;
867 	}
868 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
869 	off += hbhlen;
870 	hbhlen -= sizeof(struct ip6_hbh);
871 
872 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
873 	    hbhlen, rtalertp, plenp) < 0)
874 		return -1;
875 
876 	*offp = off;
877 	*mp = m;
878 	return 0;
879 }
880 
881 /*
882  * Search header for all Hop-by-hop options and process each option.
883  * This function is separate from ip6_hopopts_input() in order to
884  * handle a case where the sending node itself process its hop-by-hop
885  * options header. In such a case, the function is called from ip6_output().
886  *
887  * The function assumes that hbh header is located right after the IPv6 header
888  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
889  * opthead + hbhlen is located in continuous memory region.
890  */
891 static int
892 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
893 	u_int32_t *rtalertp, u_int32_t *plenp)
894 {
895 	struct ip6_hdr *ip6;
896 	int optlen = 0;
897 	u_int8_t *opt = opthead;
898 	u_int16_t rtalert_val;
899 	u_int32_t jumboplen;
900 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
901 
902 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
903 		switch (*opt) {
904 		case IP6OPT_PAD1:
905 			optlen = 1;
906 			break;
907 		case IP6OPT_PADN:
908 			if (hbhlen < IP6OPT_MINLEN) {
909 				IP6_STATINC(IP6_STAT_TOOSMALL);
910 				goto bad;
911 			}
912 			optlen = *(opt + 1) + 2;
913 			break;
914 		case IP6OPT_RTALERT:
915 			/* XXX may need check for alignment */
916 			if (hbhlen < IP6OPT_RTALERT_LEN) {
917 				IP6_STATINC(IP6_STAT_TOOSMALL);
918 				goto bad;
919 			}
920 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
921 				/* XXX stat */
922 				icmp6_error(m, ICMP6_PARAM_PROB,
923 				    ICMP6_PARAMPROB_HEADER,
924 				    erroff + opt + 1 - opthead);
925 				return (-1);
926 			}
927 			optlen = IP6OPT_RTALERT_LEN;
928 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
929 			*rtalertp = ntohs(rtalert_val);
930 			break;
931 		case IP6OPT_JUMBO:
932 			/* XXX may need check for alignment */
933 			if (hbhlen < IP6OPT_JUMBO_LEN) {
934 				IP6_STATINC(IP6_STAT_TOOSMALL);
935 				goto bad;
936 			}
937 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
938 				/* XXX stat */
939 				icmp6_error(m, ICMP6_PARAM_PROB,
940 				    ICMP6_PARAMPROB_HEADER,
941 				    erroff + opt + 1 - opthead);
942 				return (-1);
943 			}
944 			optlen = IP6OPT_JUMBO_LEN;
945 
946 			/*
947 			 * IPv6 packets that have non 0 payload length
948 			 * must not contain a jumbo payload option.
949 			 */
950 			ip6 = mtod(m, struct ip6_hdr *);
951 			if (ip6->ip6_plen) {
952 				IP6_STATINC(IP6_STAT_BADOPTIONS);
953 				icmp6_error(m, ICMP6_PARAM_PROB,
954 				    ICMP6_PARAMPROB_HEADER,
955 				    erroff + opt - opthead);
956 				return (-1);
957 			}
958 
959 			/*
960 			 * We may see jumbolen in unaligned location, so
961 			 * we'd need to perform memcpy().
962 			 */
963 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
964 			jumboplen = (u_int32_t)htonl(jumboplen);
965 
966 #if 1
967 			/*
968 			 * if there are multiple jumbo payload options,
969 			 * *plenp will be non-zero and the packet will be
970 			 * rejected.
971 			 * the behavior may need some debate in ipngwg -
972 			 * multiple options does not make sense, however,
973 			 * there's no explicit mention in specification.
974 			 */
975 			if (*plenp != 0) {
976 				IP6_STATINC(IP6_STAT_BADOPTIONS);
977 				icmp6_error(m, ICMP6_PARAM_PROB,
978 				    ICMP6_PARAMPROB_HEADER,
979 				    erroff + opt + 2 - opthead);
980 				return (-1);
981 			}
982 #endif
983 
984 			/*
985 			 * jumbo payload length must be larger than 65535.
986 			 */
987 			if (jumboplen <= IPV6_MAXPACKET) {
988 				IP6_STATINC(IP6_STAT_BADOPTIONS);
989 				icmp6_error(m, ICMP6_PARAM_PROB,
990 				    ICMP6_PARAMPROB_HEADER,
991 				    erroff + opt + 2 - opthead);
992 				return (-1);
993 			}
994 			*plenp = jumboplen;
995 
996 			break;
997 		default:		/* unknown option */
998 			if (hbhlen < IP6OPT_MINLEN) {
999 				IP6_STATINC(IP6_STAT_TOOSMALL);
1000 				goto bad;
1001 			}
1002 			optlen = ip6_unknown_opt(opt, m,
1003 			    erroff + opt - opthead);
1004 			if (optlen == -1)
1005 				return (-1);
1006 			optlen += 2;
1007 			break;
1008 		}
1009 	}
1010 
1011 	return (0);
1012 
1013   bad:
1014 	m_freem(m);
1015 	return (-1);
1016 }
1017 
1018 /*
1019  * Unknown option processing.
1020  * The third argument `off' is the offset from the IPv6 header to the option,
1021  * which is necessary if the IPv6 header the and option header and IPv6 header
1022  * is not continuous in order to return an ICMPv6 error.
1023  */
1024 int
1025 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1026 {
1027 	struct ip6_hdr *ip6;
1028 
1029 	switch (IP6OPT_TYPE(*optp)) {
1030 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1031 		return ((int)*(optp + 1));
1032 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1033 		m_freem(m);
1034 		return (-1);
1035 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1036 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1037 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1038 		return (-1);
1039 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1040 		IP6_STATINC(IP6_STAT_BADOPTIONS);
1041 		ip6 = mtod(m, struct ip6_hdr *);
1042 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1043 		    (m->m_flags & (M_BCAST|M_MCAST)))
1044 			m_freem(m);
1045 		else
1046 			icmp6_error(m, ICMP6_PARAM_PROB,
1047 				    ICMP6_PARAMPROB_OPTION, off);
1048 		return (-1);
1049 	}
1050 
1051 	m_freem(m);		/* XXX: NOTREACHED */
1052 	return (-1);
1053 }
1054 
1055 void
1056 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
1057 	struct ip6_hdr *ip6, struct mbuf *m)
1058 {
1059 	struct socket *so = in6p->in6p_socket;
1060 #ifdef RFC2292
1061 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1062 #else
1063 #define IS2292(x, y)	(y)
1064 #endif
1065 
1066 	if (SOOPT_TIMESTAMP(so->so_options))
1067 		mp = sbsavetimestamp(so->so_options, mp);
1068 
1069 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
1070 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1071 		return;
1072 
1073 	/* RFC 2292 sec. 5 */
1074 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1075 		struct in6_pktinfo pi6;
1076 
1077 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
1078 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
1079 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index;
1080 		*mp = sbcreatecontrol(&pi6, sizeof(pi6),
1081 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1082 		if (*mp)
1083 			mp = &(*mp)->m_next;
1084 	}
1085 
1086 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1087 		int hlim = ip6->ip6_hlim & 0xff;
1088 
1089 		*mp = sbcreatecontrol(&hlim, sizeof(hlim),
1090 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1091 		if (*mp)
1092 			mp = &(*mp)->m_next;
1093 	}
1094 
1095 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1096 		u_int32_t flowinfo;
1097 		int tclass;
1098 
1099 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1100 		flowinfo >>= 20;
1101 
1102 		tclass = flowinfo & 0xff;
1103 		*mp = sbcreatecontrol(&tclass, sizeof(tclass),
1104 		    IPV6_TCLASS, IPPROTO_IPV6);
1105 
1106 		if (*mp)
1107 			mp = &(*mp)->m_next;
1108 	}
1109 
1110 	/*
1111 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1112 	 * privilege for the option (see ip6_ctloutput), but it might be too
1113 	 * strict, since there might be some hop-by-hop options which can be
1114 	 * returned to normal user.
1115 	 * See also RFC3542 section 8 (or RFC2292 section 6).
1116 	 */
1117 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1118 		/*
1119 		 * Check if a hop-by-hop options header is contatined in the
1120 		 * received packet, and if so, store the options as ancillary
1121 		 * data. Note that a hop-by-hop options header must be
1122 		 * just after the IPv6 header, which fact is assured through
1123 		 * the IPv6 input processing.
1124 		 */
1125 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1126 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
1127 			struct ip6_hbh *hbh;
1128 			int hbhlen;
1129 			struct mbuf *ext;
1130 
1131 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1132 			    xip6->ip6_nxt);
1133 			if (ext == NULL) {
1134 				IP6_STATINC(IP6_STAT_TOOSHORT);
1135 				return;
1136 			}
1137 			hbh = mtod(ext, struct ip6_hbh *);
1138 			hbhlen = (hbh->ip6h_len + 1) << 3;
1139 			if (hbhlen != ext->m_len) {
1140 				m_freem(ext);
1141 				IP6_STATINC(IP6_STAT_TOOSHORT);
1142 				return;
1143 			}
1144 
1145 			/*
1146 			 * XXX: We copy whole the header even if a jumbo
1147 			 * payload option is included, which option is to
1148 			 * be removed before returning in the RFC 2292.
1149 			 * Note: this constraint is removed in RFC3542.
1150 			 */
1151 			*mp = sbcreatecontrol(hbh, hbhlen,
1152 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1153 			    IPPROTO_IPV6);
1154 			if (*mp)
1155 				mp = &(*mp)->m_next;
1156 			m_freem(ext);
1157 		}
1158 	}
1159 
1160 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1161 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1162 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
1163 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1164 
1165 		/*
1166 		 * Search for destination options headers or routing
1167 		 * header(s) through the header chain, and stores each
1168 		 * header as ancillary data.
1169 		 * Note that the order of the headers remains in
1170 		 * the chain of ancillary data.
1171 		 */
1172 		for (;;) {	/* is explicit loop prevention necessary? */
1173 			struct ip6_ext *ip6e = NULL;
1174 			int elen;
1175 			struct mbuf *ext = NULL;
1176 
1177 			/*
1178 			 * if it is not an extension header, don't try to
1179 			 * pull it from the chain.
1180 			 */
1181 			switch (nxt) {
1182 			case IPPROTO_DSTOPTS:
1183 			case IPPROTO_ROUTING:
1184 			case IPPROTO_HOPOPTS:
1185 			case IPPROTO_AH: /* is it possible? */
1186 				break;
1187 			default:
1188 				goto loopend;
1189 			}
1190 
1191 			ext = ip6_pullexthdr(m, off, nxt);
1192 			if (ext == NULL) {
1193 				IP6_STATINC(IP6_STAT_TOOSHORT);
1194 				return;
1195 			}
1196 			ip6e = mtod(ext, struct ip6_ext *);
1197 			if (nxt == IPPROTO_AH)
1198 				elen = (ip6e->ip6e_len + 2) << 2;
1199 			else
1200 				elen = (ip6e->ip6e_len + 1) << 3;
1201 			if (elen != ext->m_len) {
1202 				m_freem(ext);
1203 				IP6_STATINC(IP6_STAT_TOOSHORT);
1204 				return;
1205 			}
1206 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
1207 
1208 			switch (nxt) {
1209 			case IPPROTO_DSTOPTS:
1210 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1211 					break;
1212 
1213 				*mp = sbcreatecontrol(ip6e, elen,
1214 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1215 				    IPPROTO_IPV6);
1216 				if (*mp)
1217 					mp = &(*mp)->m_next;
1218 				break;
1219 
1220 			case IPPROTO_ROUTING:
1221 				if (!(in6p->in6p_flags & IN6P_RTHDR))
1222 					break;
1223 
1224 				*mp = sbcreatecontrol(ip6e, elen,
1225 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1226 				    IPPROTO_IPV6);
1227 				if (*mp)
1228 					mp = &(*mp)->m_next;
1229 				break;
1230 
1231 			case IPPROTO_HOPOPTS:
1232 			case IPPROTO_AH: /* is it possible? */
1233 				break;
1234 
1235 			default:
1236 				/*
1237 			 	 * other cases have been filtered in the above.
1238 				 * none will visit this case.  here we supply
1239 				 * the code just in case (nxt overwritten or
1240 				 * other cases).
1241 				 */
1242 				m_freem(ext);
1243 				goto loopend;
1244 
1245 			}
1246 
1247 			/* proceed with the next header. */
1248 			off += elen;
1249 			nxt = ip6e->ip6e_nxt;
1250 			ip6e = NULL;
1251 			m_freem(ext);
1252 			ext = NULL;
1253 		}
1254 	  loopend:
1255 	  	;
1256 	}
1257 }
1258 #undef IS2292
1259 
1260 
1261 void
1262 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
1263     uint32_t *mtu)
1264 {
1265 	struct socket *so;
1266 	struct mbuf *m_mtu;
1267 	struct ip6_mtuinfo mtuctl;
1268 
1269 	so = in6p->in6p_socket;
1270 
1271 	if (mtu == NULL)
1272 		return;
1273 
1274 	KASSERT(so != NULL);
1275 
1276 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
1277 	mtuctl.ip6m_mtu = *mtu;
1278 	mtuctl.ip6m_addr = *dst;
1279 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
1280 		return;
1281 
1282 	if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl),
1283 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1284 		return;
1285 
1286 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
1287 	    == 0) {
1288 		soroverflow(so);
1289 		m_freem(m_mtu);
1290 	} else
1291 		sorwakeup(so);
1292 
1293 	return;
1294 }
1295 
1296 /*
1297  * pull single extension header from mbuf chain.  returns single mbuf that
1298  * contains the result, or NULL on error.
1299  */
1300 static struct mbuf *
1301 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1302 {
1303 	struct ip6_ext ip6e;
1304 	size_t elen;
1305 	struct mbuf *n;
1306 
1307 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1308 	if (nxt == IPPROTO_AH)
1309 		elen = (ip6e.ip6e_len + 2) << 2;
1310 	else
1311 		elen = (ip6e.ip6e_len + 1) << 3;
1312 
1313 	MGET(n, M_DONTWAIT, MT_DATA);
1314 	if (n && elen >= MLEN) {
1315 		MCLGET(n, M_DONTWAIT);
1316 		if ((n->m_flags & M_EXT) == 0) {
1317 			m_free(n);
1318 			n = NULL;
1319 		}
1320 	}
1321 	if (!n)
1322 		return NULL;
1323 
1324 	n->m_len = 0;
1325 	if (elen >= M_TRAILINGSPACE(n)) {
1326 		m_free(n);
1327 		return NULL;
1328 	}
1329 
1330 	m_copydata(m, off, elen, mtod(n, void *));
1331 	n->m_len = elen;
1332 	return n;
1333 }
1334 
1335 /*
1336  * Get offset to the previous header followed by the header
1337  * currently processed.
1338  */
1339 int
1340 ip6_get_prevhdr(struct mbuf *m, int off)
1341 {
1342 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1343 
1344 	if (off == sizeof(struct ip6_hdr)) {
1345 		return offsetof(struct ip6_hdr, ip6_nxt);
1346 	} else if (off < sizeof(struct ip6_hdr)) {
1347 		panic("%s: off < sizeof(struct ip6_hdr)", __func__);
1348 	} else {
1349 		int len, nlen, nxt;
1350 		struct ip6_ext ip6e;
1351 
1352 		nxt = ip6->ip6_nxt;
1353 		len = sizeof(struct ip6_hdr);
1354 		nlen = 0;
1355 		while (len < off) {
1356 			m_copydata(m, len, sizeof(ip6e), &ip6e);
1357 
1358 			switch (nxt) {
1359 			case IPPROTO_FRAGMENT:
1360 				nlen = sizeof(struct ip6_frag);
1361 				break;
1362 			case IPPROTO_AH:
1363 				nlen = (ip6e.ip6e_len + 2) << 2;
1364 				break;
1365 			default:
1366 				nlen = (ip6e.ip6e_len + 1) << 3;
1367 				break;
1368 			}
1369 			len += nlen;
1370 			nxt = ip6e.ip6e_nxt;
1371 		}
1372 
1373 		return (len - nlen);
1374 	}
1375 }
1376 
1377 /*
1378  * get next header offset.  m will be retained.
1379  */
1380 int
1381 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1382 {
1383 	struct ip6_hdr ip6;
1384 	struct ip6_ext ip6e;
1385 	struct ip6_frag fh;
1386 
1387 	/* just in case */
1388 	if (m == NULL)
1389 		panic("%s: m == NULL", __func__);
1390 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1391 		return -1;
1392 
1393 	switch (proto) {
1394 	case IPPROTO_IPV6:
1395 		/* do not chase beyond intermediate IPv6 headers */
1396 		if (off != 0)
1397 			return -1;
1398 		if (m->m_pkthdr.len < off + sizeof(ip6))
1399 			return -1;
1400 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
1401 		if (nxtp)
1402 			*nxtp = ip6.ip6_nxt;
1403 		off += sizeof(ip6);
1404 		return off;
1405 
1406 	case IPPROTO_FRAGMENT:
1407 		/*
1408 		 * terminate parsing if it is not the first fragment,
1409 		 * it does not make sense to parse through it.
1410 		 */
1411 		if (m->m_pkthdr.len < off + sizeof(fh))
1412 			return -1;
1413 		m_copydata(m, off, sizeof(fh), (void *)&fh);
1414 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
1415 			return -1;
1416 		if (nxtp)
1417 			*nxtp = fh.ip6f_nxt;
1418 		off += sizeof(struct ip6_frag);
1419 		return off;
1420 
1421 	case IPPROTO_AH:
1422 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1423 			return -1;
1424 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1425 		if (nxtp)
1426 			*nxtp = ip6e.ip6e_nxt;
1427 		off += (ip6e.ip6e_len + 2) << 2;
1428 		if (m->m_pkthdr.len < off)
1429 			return -1;
1430 		return off;
1431 
1432 	case IPPROTO_HOPOPTS:
1433 	case IPPROTO_ROUTING:
1434 	case IPPROTO_DSTOPTS:
1435 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1436 			return -1;
1437 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
1438 		if (nxtp)
1439 			*nxtp = ip6e.ip6e_nxt;
1440 		off += (ip6e.ip6e_len + 1) << 3;
1441 		if (m->m_pkthdr.len < off)
1442 			return -1;
1443 		return off;
1444 
1445 	case IPPROTO_NONE:
1446 	case IPPROTO_ESP:
1447 	case IPPROTO_IPCOMP:
1448 		/* give up */
1449 		return -1;
1450 
1451 	default:
1452 		return -1;
1453 	}
1454 }
1455 
1456 /*
1457  * get offset for the last header in the chain.  m will be kept untainted.
1458  */
1459 int
1460 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1461 {
1462 	int newoff;
1463 	int nxt;
1464 
1465 	if (!nxtp) {
1466 		nxt = -1;
1467 		nxtp = &nxt;
1468 	}
1469 	for (;;) {
1470 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1471 		if (newoff < 0)
1472 			return off;
1473 		else if (newoff < off)
1474 			return -1;	/* invalid */
1475 		else if (newoff == off)
1476 			return newoff;
1477 
1478 		off = newoff;
1479 		proto = *nxtp;
1480 	}
1481 }
1482 
1483 struct m_tag *
1484 ip6_addaux(struct mbuf *m)
1485 {
1486 	struct m_tag *mtag;
1487 
1488 	mtag = m_tag_find(m, PACKET_TAG_INET6);
1489 	if (!mtag) {
1490 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
1491 		    M_NOWAIT);
1492 		if (mtag) {
1493 			m_tag_prepend(m, mtag);
1494 			memset(mtag + 1, 0, sizeof(struct ip6aux));
1495 		}
1496 	}
1497 	return mtag;
1498 }
1499 
1500 struct m_tag *
1501 ip6_findaux(struct mbuf *m)
1502 {
1503 	struct m_tag *mtag;
1504 
1505 	mtag = m_tag_find(m, PACKET_TAG_INET6);
1506 	return mtag;
1507 }
1508 
1509 void
1510 ip6_delaux(struct mbuf *m)
1511 {
1512 	struct m_tag *mtag;
1513 
1514 	mtag = m_tag_find(m, PACKET_TAG_INET6);
1515 	if (mtag)
1516 		m_tag_delete(m, mtag);
1517 }
1518 
1519 /*
1520  * System control for IP6
1521  */
1522 
1523 const u_char inet6ctlerrmap[PRC_NCMDS] = {
1524 	0,		0,		0,		0,
1525 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1526 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1527 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1528 	0,		0,		0,		0,
1529 	ENOPROTOOPT
1530 };
1531 
1532 extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS);
1533 
1534 static int
1535 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
1536 {
1537 
1538 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
1539 }
1540 
1541 static void
1542 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
1543 {
1544 
1545 	sysctl_createv(clog, 0, NULL, NULL,
1546 		       CTLFLAG_PERMANENT,
1547 		       CTLTYPE_NODE, "inet6",
1548 		       SYSCTL_DESCR("PF_INET6 related settings"),
1549 		       NULL, 0, NULL, 0,
1550 		       CTL_NET, PF_INET6, CTL_EOL);
1551 	sysctl_createv(clog, 0, NULL, NULL,
1552 		       CTLFLAG_PERMANENT,
1553 		       CTLTYPE_NODE, "ip6",
1554 		       SYSCTL_DESCR("IPv6 related settings"),
1555 		       NULL, 0, NULL, 0,
1556 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
1557 
1558 	sysctl_createv(clog, 0, NULL, NULL,
1559 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1560 		       CTLTYPE_INT, "forwarding",
1561 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
1562 		       NULL, 0, &ip6_forwarding, 0,
1563 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1564 		       IPV6CTL_FORWARDING, CTL_EOL);
1565 	sysctl_createv(clog, 0, NULL, NULL,
1566 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1567 		       CTLTYPE_INT, "redirect",
1568 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
1569 		       NULL, 0, &ip6_sendredirects, 0,
1570 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1571 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
1572 	sysctl_createv(clog, 0, NULL, NULL,
1573 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1574 		       CTLTYPE_INT, "hlim",
1575 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
1576 		       NULL, 0, &ip6_defhlim, 0,
1577 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1578 		       IPV6CTL_DEFHLIM, CTL_EOL);
1579 	sysctl_createv(clog, 0, NULL, NULL,
1580 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1581 		       CTLTYPE_INT, "maxfragpackets",
1582 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
1583 				    "for reassembly"),
1584 		       NULL, 0, &ip6_maxfragpackets, 0,
1585 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1586 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
1587 	sysctl_createv(clog, 0, NULL, NULL,
1588 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1589 		       CTLTYPE_INT, "accept_rtadv",
1590 		       SYSCTL_DESCR("Accept router advertisements"),
1591 		       NULL, 0, &ip6_accept_rtadv, 0,
1592 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1593 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
1594 	sysctl_createv(clog, 0, NULL, NULL,
1595 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1596 		       CTLTYPE_INT, "rtadv_maxroutes",
1597 		       SYSCTL_DESCR("Maximum number of routes accepted via router advertisements"),
1598 		       NULL, 0, &ip6_rtadv_maxroutes, 0,
1599 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1600 		       IPV6CTL_RTADV_MAXROUTES, CTL_EOL);
1601 	sysctl_createv(clog, 0, NULL, NULL,
1602 		       CTLFLAG_PERMANENT,
1603 		       CTLTYPE_INT, "rtadv_numroutes",
1604 		       SYSCTL_DESCR("Current number of routes accepted via router advertisements"),
1605 		       NULL, 0, &nd6_numroutes, 0,
1606 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1607 		       IPV6CTL_RTADV_NUMROUTES, CTL_EOL);
1608 	sysctl_createv(clog, 0, NULL, NULL,
1609 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1610 		       CTLTYPE_INT, "keepfaith",
1611 		       SYSCTL_DESCR("Activate faith interface"),
1612 		       NULL, 0, &ip6_keepfaith, 0,
1613 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1614 		       IPV6CTL_KEEPFAITH, CTL_EOL);
1615 	sysctl_createv(clog, 0, NULL, NULL,
1616 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1617 		       CTLTYPE_INT, "log_interval",
1618 		       SYSCTL_DESCR("Minimum interval between logging "
1619 				    "unroutable packets"),
1620 		       NULL, 0, &ip6_log_interval, 0,
1621 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1622 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
1623 	sysctl_createv(clog, 0, NULL, NULL,
1624 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1625 		       CTLTYPE_INT, "hdrnestlimit",
1626 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
1627 		       NULL, 0, &ip6_hdrnestlimit, 0,
1628 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1629 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
1630 	sysctl_createv(clog, 0, NULL, NULL,
1631 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1632 		       CTLTYPE_INT, "dad_count",
1633 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1634 				    "probes to send"),
1635 		       NULL, 0, &ip6_dad_count, 0,
1636 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1637 		       IPV6CTL_DAD_COUNT, CTL_EOL);
1638 	sysctl_createv(clog, 0, NULL, NULL,
1639 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1640 		       CTLTYPE_INT, "auto_flowlabel",
1641 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
1642 		       NULL, 0, &ip6_auto_flowlabel, 0,
1643 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1644 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
1645 	sysctl_createv(clog, 0, NULL, NULL,
1646 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1647 		       CTLTYPE_INT, "defmcasthlim",
1648 		       SYSCTL_DESCR("Default multicast hop limit"),
1649 		       NULL, 0, &ip6_defmcasthlim, 0,
1650 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1651 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
1652 	sysctl_createv(clog, 0, NULL, NULL,
1653 		       CTLFLAG_PERMANENT,
1654 		       CTLTYPE_STRING, "kame_version",
1655 		       SYSCTL_DESCR("KAME Version"),
1656 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
1657 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1658 		       IPV6CTL_KAME_VERSION, CTL_EOL);
1659 	sysctl_createv(clog, 0, NULL, NULL,
1660 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1661 		       CTLTYPE_INT, "use_deprecated",
1662 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
1663 				    "source addresses"),
1664 		       NULL, 0, &ip6_use_deprecated, 0,
1665 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1666 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
1667 	sysctl_createv(clog, 0, NULL, NULL,
1668 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1669 		       CTLTYPE_INT, "rr_prune", NULL,
1670 		       NULL, 0, &ip6_rr_prune, 0,
1671 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1672 		       IPV6CTL_RR_PRUNE, CTL_EOL);
1673 	sysctl_createv(clog, 0, NULL, NULL,
1674 		       CTLFLAG_PERMANENT
1675 #ifndef INET6_BINDV6ONLY
1676 		       |CTLFLAG_READWRITE,
1677 #endif
1678 		       CTLTYPE_INT, "v6only",
1679 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
1680 				    "to PF_INET sockets"),
1681 		       NULL, 0, &ip6_v6only, 0,
1682 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1683 		       IPV6CTL_V6ONLY, CTL_EOL);
1684 	sysctl_createv(clog, 0, NULL, NULL,
1685 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1686 		       CTLTYPE_INT, "anonportmin",
1687 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1688 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
1689 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1690 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
1691 	sysctl_createv(clog, 0, NULL, NULL,
1692 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1693 		       CTLTYPE_INT, "anonportmax",
1694 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1695 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
1696 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1697 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
1698 #ifndef IPNOPRIVPORTS
1699 	sysctl_createv(clog, 0, NULL, NULL,
1700 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1701 		       CTLTYPE_INT, "lowportmin",
1702 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1703 				    "to assign"),
1704 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
1705 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1706 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
1707 	sysctl_createv(clog, 0, NULL, NULL,
1708 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1709 		       CTLTYPE_INT, "lowportmax",
1710 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1711 				    "to assign"),
1712 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
1713 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1714 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
1715 #endif /* IPNOPRIVPORTS */
1716 	sysctl_createv(clog, 0, NULL, NULL,
1717 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1718 		       CTLTYPE_INT, "auto_linklocal",
1719 		       SYSCTL_DESCR("Default value of per-interface flag for "
1720 		                    "adding an IPv6 link-local address to "
1721 				    "interfaces when attached"),
1722 		       NULL, 0, &ip6_auto_linklocal, 0,
1723 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1724 		       IPV6CTL_AUTO_LINKLOCAL, CTL_EOL);
1725 	sysctl_createv(clog, 0, NULL, NULL,
1726 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
1727 		       CTLTYPE_STRUCT, "addctlpolicy",
1728 		       SYSCTL_DESCR("Return the current address control"
1729 			   " policy"),
1730 		       sysctl_net_inet6_addrctlpolicy, 0, NULL, 0,
1731 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1732 		       IPV6CTL_ADDRCTLPOLICY, CTL_EOL);
1733 	sysctl_createv(clog, 0, NULL, NULL,
1734 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1735 		       CTLTYPE_INT, "use_tempaddr",
1736 		       SYSCTL_DESCR("Use temporary address"),
1737 		       NULL, 0, &ip6_use_tempaddr, 0,
1738 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1739 		       CTL_CREATE, CTL_EOL);
1740 	sysctl_createv(clog, 0, NULL, NULL,
1741 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1742 		       CTLTYPE_INT, "prefer_tempaddr",
1743 		       SYSCTL_DESCR("Prefer temporary address as source "
1744 		                    "address"),
1745 		       NULL, 0, &ip6_prefer_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, "temppltime",
1751 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
1752 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
1753 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1754 		       CTL_CREATE, CTL_EOL);
1755 	sysctl_createv(clog, 0, NULL, NULL,
1756 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1757 		       CTLTYPE_INT, "tempvltime",
1758 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
1759 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
1760 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1761 		       CTL_CREATE, CTL_EOL);
1762 	sysctl_createv(clog, 0, NULL, NULL,
1763 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1764 		       CTLTYPE_INT, "maxfrags",
1765 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
1766 		       NULL, 0, &ip6_maxfrags, 0,
1767 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1768 		       IPV6CTL_MAXFRAGS, CTL_EOL);
1769 	sysctl_createv(clog, 0, NULL, NULL,
1770 		       CTLFLAG_PERMANENT,
1771 		       CTLTYPE_STRUCT, "stats",
1772 		       SYSCTL_DESCR("IPv6 statistics"),
1773 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
1774 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1775 		       IPV6CTL_STATS, CTL_EOL);
1776 	sysctl_createv(clog, 0, NULL, NULL,
1777 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1778 		       CTLTYPE_INT, "use_defaultzone",
1779 		       SYSCTL_DESCR("Whether to use the default scope zones"),
1780 		       NULL, 0, &ip6_use_defzone, 0,
1781 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1782 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
1783 	sysctl_createv(clog, 0, NULL, NULL,
1784 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1785 		       CTLTYPE_INT, "mcast_pmtu",
1786 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
1787 		       NULL, 0, &ip6_mcast_pmtu, 0,
1788 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1789 		       CTL_CREATE, CTL_EOL);
1790 	/* anonportalgo RFC6056 subtree */
1791 	const struct sysctlnode *portalgo_node;
1792 	sysctl_createv(clog, 0, NULL, &portalgo_node,
1793 		       CTLFLAG_PERMANENT,
1794 		       CTLTYPE_NODE, "anonportalgo",
1795 		       SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"),
1796 	    	       NULL, 0, NULL, 0,
1797 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL);
1798 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1799 		       CTLFLAG_PERMANENT,
1800 		       CTLTYPE_STRING, "available",
1801 		       SYSCTL_DESCR("available algorithms"),
1802 		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1803 		       CTL_CREATE, CTL_EOL);
1804 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1805 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1806 		       CTLTYPE_STRING, "selected",
1807 		       SYSCTL_DESCR("selected algorithm"),
1808 	               sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN,
1809 		       CTL_CREATE, CTL_EOL);
1810 	sysctl_createv(clog, 0, &portalgo_node, NULL,
1811 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1812 		       CTLTYPE_STRUCT, "reserve",
1813 		       SYSCTL_DESCR("bitmap of reserved ports"),
1814 		       sysctl_portalgo_reserve6, 0, NULL, 0,
1815 		       CTL_CREATE, CTL_EOL);
1816 	sysctl_createv(clog, 0, NULL, NULL,
1817 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1818 		       CTLTYPE_INT, "neighborgcthresh",
1819 		       SYSCTL_DESCR("Maximum number of entries in neighbor"
1820 			" cache"),
1821 		       NULL, 1, &ip6_neighborgcthresh, 0,
1822 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1823 		       CTL_CREATE, CTL_EOL);
1824 	sysctl_createv(clog, 0, NULL, NULL,
1825 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1826 		       CTLTYPE_INT, "maxifprefixes",
1827 		       SYSCTL_DESCR("Maximum number of prefixes created by"
1828 			   " route advertisement per interface"),
1829 		       NULL, 1, &ip6_maxifprefixes, 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, "maxifdefrouters",
1835 		       SYSCTL_DESCR("Maximum number of default routers created"
1836 			   " by route advertisement per interface"),
1837 		       NULL, 1, &ip6_maxifdefrouters, 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, "maxdynroutes",
1843 		       SYSCTL_DESCR("Maximum number of routes created via"
1844 			   " redirect"),
1845 		       NULL, 1, &ip6_maxdynroutes, 0,
1846 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
1847 		       CTL_CREATE, CTL_EOL);
1848 }
1849 
1850 void
1851 ip6_statinc(u_int stat)
1852 {
1853 
1854 	KASSERT(stat < IP6_NSTATS);
1855 	IP6_STATINC(stat);
1856 }
1857