Home | History | Annotate | Line # | Download | only in netinet6
      1 /*	$NetBSD: ip6_input.c,v 1.228 2024/06/29 13:00:44 riastradh 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 1.228 2024/06/29 13:00:44 riastradh 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 ip6intr(void *);
    138 static void ip6_input(struct mbuf *, struct ifnet *);
    139 static bool ip6_badaddr(struct ip6_hdr *);
    140 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
    141 
    142 static struct m_tag *ip6_addaux(struct mbuf *);
    143 static struct m_tag *ip6_findaux(struct mbuf *);
    144 static void ip6_delaux(struct mbuf *);
    145 
    146 static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *,
    147     u_int32_t *);
    148 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
    149 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
    150 
    151 #ifdef NET_MPSAFE
    152 #define	SOFTNET_LOCK()		mutex_enter(softnet_lock)
    153 #define	SOFTNET_UNLOCK()	mutex_exit(softnet_lock)
    154 #else
    155 #define	SOFTNET_LOCK()		KASSERT(mutex_owned(softnet_lock))
    156 #define	SOFTNET_UNLOCK()	KASSERT(mutex_owned(softnet_lock))
    157 #endif
    158 
    159 /* Ensure that non packed structures are the desired size. */
    160 __CTASSERT(sizeof(struct ip6_hdr) == 40);
    161 __CTASSERT(sizeof(struct ip6_ext) == 2);
    162 __CTASSERT(sizeof(struct ip6_hbh) == 2);
    163 __CTASSERT(sizeof(struct ip6_dest) == 2);
    164 __CTASSERT(sizeof(struct ip6_opt) == 2);
    165 __CTASSERT(sizeof(struct ip6_opt_jumbo) == 6);
    166 __CTASSERT(sizeof(struct ip6_opt_nsap) == 4);
    167 __CTASSERT(sizeof(struct ip6_opt_tunnel) == 3);
    168 __CTASSERT(sizeof(struct ip6_opt_router) == 4);
    169 __CTASSERT(sizeof(struct ip6_rthdr) == 4);
    170 __CTASSERT(sizeof(struct ip6_rthdr0) == 8);
    171 __CTASSERT(sizeof(struct ip6_frag) == 8);
    172 
    173 /*
    174  * IP6 initialization: fill in IP6 protocol switch table.
    175  * All protocols not implemented in kernel go to raw IP6 protocol handler.
    176  */
    177 void
    178 ip6_init(void)
    179 {
    180 	const struct ip6protosw *pr;
    181 	int i;
    182 
    183 	in6_init();
    184 
    185 	ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL);
    186 	KASSERT(ip6_pktq != NULL);
    187 
    188 	sysctl_net_inet6_ip6_setup(NULL);
    189 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
    190 	if (pr == 0)
    191 		panic("ip6_init");
    192 	for (i = 0; i < IPPROTO_MAX; i++)
    193 		ip6_protox[i] = pr - inet6sw;
    194 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
    195 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
    196 		if (pr->pr_domain->dom_family == PF_INET6 &&
    197 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
    198 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
    199 
    200 	scope6_init();
    201 	addrsel_policy_init();
    202 	nd6_init();
    203 	frag6_init();
    204 
    205 #ifdef GATEWAY
    206 	ip6flow_init(ip6_hashsize);
    207 #endif
    208 	/* Register our Packet Filter hook. */
    209 	inet6_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET6);
    210 	KASSERT(inet6_pfil_hook != NULL);
    211 
    212 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
    213 	ip6_forward_rt_percpu = rtcache_percpu_alloc();
    214 }
    215 
    216 /*
    217  * IP6 input interrupt handling. Just pass the packet to ip6_input.
    218  */
    219 static void
    220 ip6intr(void *arg __unused)
    221 {
    222 	struct mbuf *m;
    223 
    224 	SOFTNET_KERNEL_LOCK_UNLESS_NET_MPSAFE();
    225 	while ((m = pktq_dequeue(ip6_pktq)) != NULL) {
    226 		struct psref psref;
    227 		struct ifnet *rcvif = m_get_rcvif_psref(m, &psref);
    228 
    229 		if (rcvif == NULL) {
    230 			IP6_STATINC(IP6_STAT_IFDROP);
    231 			m_freem(m);
    232 			continue;
    233 		}
    234 		/*
    235 		 * Drop the packet if IPv6 is disabled on the interface.
    236 		 */
    237 		if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED)) {
    238 			m_put_rcvif_psref(rcvif, &psref);
    239 			IP6_STATINC(IP6_STAT_IFDROP);
    240 			m_freem(m);
    241 			continue;
    242 		}
    243 		ip6_input(m, rcvif);
    244 		m_put_rcvif_psref(rcvif, &psref);
    245 	}
    246 	SOFTNET_KERNEL_UNLOCK_UNLESS_NET_MPSAFE();
    247 }
    248 
    249 static void
    250 ip6_input(struct mbuf *m, struct ifnet *rcvif)
    251 {
    252 	struct ip6_hdr *ip6;
    253 	int hit, off = sizeof(struct ip6_hdr), nest;
    254 	u_int32_t plen;
    255 	u_int32_t rtalert = ~0;
    256 	int nxt, ours = 0, rh_present = 0, frg_present;
    257 	struct ifnet *deliverifp = NULL;
    258 	int srcrt = 0;
    259 	struct rtentry *rt = NULL;
    260 	union {
    261 		struct sockaddr		dst;
    262 		struct sockaddr_in6	dst6;
    263 	} u;
    264 	struct route *ro;
    265 
    266 	KASSERT(rcvif != NULL);
    267 
    268 	/*
    269 	 * make sure we don't have onion peering information into m_tag.
    270 	 */
    271 	ip6_delaux(m);
    272 
    273 	/*
    274 	 * mbuf statistics
    275 	 */
    276 	if (m->m_flags & M_EXT) {
    277 		if (m->m_next)
    278 			IP6_STATINC(IP6_STAT_MEXT2M);
    279 		else
    280 			IP6_STATINC(IP6_STAT_MEXT1);
    281 	} else {
    282 #define M2MMAX	32
    283 		if (m->m_next) {
    284 			if (m->m_flags & M_LOOP)
    285 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
    286 			else if (rcvif->if_index < M2MMAX)
    287 				IP6_STATINC(IP6_STAT_M2M + rcvif->if_index);
    288 			else
    289 				IP6_STATINC(IP6_STAT_M2M);
    290 		} else
    291 			IP6_STATINC(IP6_STAT_M1);
    292 #undef M2MMAX
    293 	}
    294 
    295 	in6_ifstat_inc(rcvif, ifs6_in_receive);
    296 	IP6_STATINC(IP6_STAT_TOTAL);
    297 
    298 	/*
    299 	 * If the IPv6 header is not aligned, slurp it up into a new
    300 	 * mbuf with space for link headers, in the event we forward
    301 	 * it.  Otherwise, if it is aligned, make sure the entire base
    302 	 * IPv6 header is in the first mbuf of the chain.
    303 	 */
    304 	if (M_GET_ALIGNED_HDR(&m, struct ip6_hdr, true) != 0) {
    305 		/* XXXJRT new stat, please */
    306 		IP6_STATINC(IP6_STAT_TOOSMALL);
    307 		in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    308 		return;
    309 	}
    310 
    311 	ip6 = mtod(m, struct ip6_hdr *);
    312 
    313 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    314 		IP6_STATINC(IP6_STAT_BADVERS);
    315 		in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    316 		goto bad;
    317 	}
    318 
    319 	if (ip6_badaddr(ip6)) {
    320 		IP6_STATINC(IP6_STAT_BADSCOPE);
    321 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
    322 		goto bad;
    323 	}
    324 
    325 	/*
    326 	 * Assume that we can create a fast-forward IP flow entry
    327 	 * based on this packet.
    328 	 */
    329 	m->m_flags |= M_CANFASTFWD;
    330 
    331 	/*
    332 	 * Run through list of hooks for input packets.  If there are any
    333 	 * filters which require that additional packets in the flow are
    334 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
    335 	 * Note that filters must _never_ set this flag, as another filter
    336 	 * in the list may have previously cleared it.
    337 	 *
    338 	 * Don't call hooks if the packet has already been processed by
    339 	 * IPsec (encapsulated, tunnel mode).
    340 	 */
    341 #if defined(IPSEC)
    342 	if (!ipsec_used || !ipsec_skip_pfil(m))
    343 #else
    344 	if (1)
    345 #endif
    346 	{
    347 		struct in6_addr odst;
    348 		int error;
    349 
    350 		odst = ip6->ip6_dst;
    351 		error = pfil_run_hooks(inet6_pfil_hook, &m, rcvif, PFIL_IN);
    352 		if (error != 0 || m == NULL) {
    353 			IP6_STATINC(IP6_STAT_PFILDROP_IN);
    354 			return;
    355 		}
    356 		if (m->m_len < sizeof(struct ip6_hdr)) {
    357 			if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    358 				IP6_STATINC(IP6_STAT_TOOSMALL);
    359 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    360 				return;
    361 			}
    362 		}
    363 		ip6 = mtod(m, struct ip6_hdr *);
    364 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
    365 	}
    366 
    367 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
    368 
    369 #ifdef ALTQ
    370 	if (altq_input != NULL) {
    371 		SOFTNET_LOCK();
    372 		if ((*altq_input)(m, AF_INET6) == 0) {
    373 			SOFTNET_UNLOCK();
    374 			/* packet is dropped by traffic conditioner */
    375 			return;
    376 		}
    377 		SOFTNET_UNLOCK();
    378 	}
    379 #endif
    380 
    381 	/*
    382 	 * Disambiguate address scope zones (if there is ambiguity).
    383 	 * We first make sure that the original source or destination address
    384 	 * is not in our internal form for scoped addresses.  Such addresses
    385 	 * are not necessarily invalid spec-wise, but we cannot accept them due
    386 	 * to the usage conflict.
    387 	 * in6_setscope() then also checks and rejects the cases where src or
    388 	 * dst are the loopback address and the receiving interface
    389 	 * is not loopback.
    390 	 */
    391 	if (__predict_false(
    392 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT))) {
    393 		IP6_STATINC(IP6_STAT_IDROPPED);
    394 		goto bad;
    395 	}
    396 	ip6 = mtod(m, struct ip6_hdr *);
    397 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
    398 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
    399 		goto bad;
    400 	}
    401 	if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
    402 	    in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
    403 		IP6_STATINC(IP6_STAT_BADSCOPE);
    404 		goto bad;
    405 	}
    406 
    407 	ro = rtcache_percpu_getref(ip6_forward_rt_percpu);
    408 
    409 	/*
    410 	 * Multicast check
    411 	 */
    412 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    413 		bool ingroup;
    414 
    415 		in6_ifstat_inc(rcvif, ifs6_in_mcast);
    416 		/*
    417 		 * See if we belong to the destination multicast group on the
    418 		 * arrival interface.
    419 		 */
    420 		ingroup = in6_multi_group(&ip6->ip6_dst, rcvif);
    421 		if (ingroup) {
    422 			ours = 1;
    423 		} else if (!ip6_mrouter) {
    424 			net_stat_ref_t ip6s = IP6_STAT_GETREF();
    425 			_NET_STATINC_REF(ip6s, IP6_STAT_NOTMEMBER);
    426 			_NET_STATINC_REF(ip6s, IP6_STAT_CANTFORWARD);
    427 			IP6_STAT_PUTREF();
    428 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    429 			goto bad_unref;
    430 		}
    431 		deliverifp = rcvif;
    432 		goto hbhcheck;
    433 	}
    434 
    435 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
    436 
    437 	/*
    438 	 * Unicast check
    439 	 */
    440 	rt = rtcache_lookup2(ro, &u.dst, 1, &hit);
    441 	if (hit)
    442 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
    443 	else
    444 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
    445 
    446 	/*
    447 	 * Accept the packet if the forwarding interface to the destination
    448 	 * (according to the routing table) is the loopback interface,
    449 	 * unless the associated route has a gateway.
    450 	 *
    451 	 * We don't explicitly match ip6_dst against an interface here. It
    452 	 * is already done in rtcache_lookup2: rt->rt_ifp->if_type will be
    453 	 * IFT_LOOP if the packet is for us.
    454 	 *
    455 	 * Note that this approach causes to accept a packet if there is a
    456 	 * route to the loopback interface for the destination of the packet.
    457 	 * But we think it's even useful in some situations, e.g. when using
    458 	 * a special daemon which wants to intercept the packet.
    459 	 */
    460 	if (rt != NULL &&
    461 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
    462 	    rt->rt_ifp->if_type == IFT_LOOP) {
    463 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
    464 		int addrok;
    465 
    466 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
    467 			m->m_flags |= M_ANYCAST6;
    468 		/*
    469 		 * packets to a tentative, duplicated, or somehow invalid
    470 		 * address must not be accepted.
    471 		 */
    472 		if (ia6->ia6_flags & IN6_IFF_NOTREADY)
    473 			addrok = 0;
    474 		else if (ia6->ia6_flags & IN6_IFF_DETACHED &&
    475 		    !IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src))
    476 		{
    477 			/* Allow internal traffic to DETACHED addresses */
    478 			struct sockaddr_in6 sin6;
    479 			int s;
    480 
    481 			memset(&sin6, 0, sizeof(sin6));
    482 			sin6.sin6_family = AF_INET6;
    483 			sin6.sin6_len = sizeof(sin6);
    484 			sin6.sin6_addr = ip6->ip6_src;
    485 			s = pserialize_read_enter();
    486 			addrok = (ifa_ifwithaddr(sin6tosa(&sin6)) != NULL);
    487 			pserialize_read_exit(s);
    488 		} else
    489 			addrok = 1;
    490 		if (addrok) {
    491 			/* this address is ready */
    492 			ours = 1;
    493 			deliverifp = ia6->ia_ifp;	/* correct? */
    494 			goto hbhcheck;
    495 		} else {
    496 			/* address is not ready, so discard the packet. */
    497 			char ip6bufs[INET6_ADDRSTRLEN];
    498 			char ip6bufd[INET6_ADDRSTRLEN];
    499 			nd6log(LOG_INFO, "packet to an unready address %s->%s\n",
    500 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
    501 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
    502 
    503 			IP6_STATINC(IP6_STAT_IDROPPED);
    504 			goto bad_unref;
    505 		}
    506 	}
    507 
    508 	/*
    509 	 * FAITH (Firewall Aided Internet Translator)
    510 	 */
    511 #if defined(NFAITH) && 0 < NFAITH
    512 	if (ip6_keepfaith) {
    513 		if (rt != NULL && rt->rt_ifp != NULL &&
    514 		    rt->rt_ifp->if_type == IFT_FAITH) {
    515 			/* XXX do we need more sanity checks? */
    516 			ours = 1;
    517 			deliverifp = rt->rt_ifp; /* faith */
    518 			goto hbhcheck;
    519 		}
    520 	}
    521 #endif
    522 
    523 	/*
    524 	 * Now there is no reason to process the packet if it's not our own
    525 	 * and we're not a router.
    526 	 */
    527 	if (!ip6_forwarding) {
    528 		IP6_STATINC(IP6_STAT_CANTFORWARD);
    529 		in6_ifstat_inc(rcvif, ifs6_in_discard);
    530 		goto bad_unref;
    531 	}
    532 
    533 hbhcheck:
    534 	/*
    535 	 * Record address information into m_tag, if we don't have one yet.
    536 	 * Note that we are unable to record it, if the address is not listed
    537 	 * as our interface address (e.g. multicast addresses, addresses
    538 	 * within FAITH prefixes and such).
    539 	 */
    540 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
    541 		struct in6_ifaddr *ia6;
    542 		int s = pserialize_read_enter();
    543 
    544 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    545 		/* Depends on ip6_setdstifaddr never sleep */
    546 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
    547 			/*
    548 			 * XXX maybe we should drop the packet here,
    549 			 * as we could not provide enough information
    550 			 * to the upper layers.
    551 			 */
    552 		}
    553 		pserialize_read_exit(s);
    554 	}
    555 
    556 	/*
    557 	 * Process Hop-by-Hop options header if it's contained.
    558 	 * m may be modified in ip6_hopopts_input().
    559 	 * If a JumboPayload option is included, plen will also be modified.
    560 	 */
    561 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
    562 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
    563 		struct ip6_hbh *hbh;
    564 
    565 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
    566 			/* m already freed */
    567 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    568 			rtcache_unref(rt, ro);
    569 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    570 			return;
    571 		}
    572 
    573 		/* adjust pointer */
    574 		ip6 = mtod(m, struct ip6_hdr *);
    575 
    576 		/*
    577 		 * if the payload length field is 0 and the next header field
    578 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
    579 		 * option MUST be included.
    580 		 */
    581 		if (ip6->ip6_plen == 0 && plen == 0) {
    582 			/*
    583 			 * Note that if a valid jumbo payload option is
    584 			 * contained, ip6_hopopts_input() must set a valid
    585 			 * (non-zero) payload length to the variable plen.
    586 			 */
    587 			IP6_STATINC(IP6_STAT_BADOPTIONS);
    588 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    589 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    590 			icmp6_error(m, ICMP6_PARAM_PROB,
    591 				    ICMP6_PARAMPROB_HEADER,
    592 				    (char *)&ip6->ip6_plen - (char *)ip6);
    593 			rtcache_unref(rt, ro);
    594 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    595 			return;
    596 		}
    597 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    598 			sizeof(struct ip6_hbh));
    599 		if (hbh == NULL) {
    600 			IP6_STATINC(IP6_STAT_TOOSHORT);
    601 			rtcache_unref(rt, ro);
    602 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    603 			return;
    604 		}
    605 		KASSERT(ACCESSIBLE_POINTER(hbh, struct ip6_hdr));
    606 		nxt = hbh->ip6h_nxt;
    607 
    608 		/*
    609 		 * accept the packet if a router alert option is included
    610 		 * and we act as an IPv6 router.
    611 		 */
    612 		if (rtalert != ~0 && ip6_forwarding)
    613 			ours = 1;
    614 	} else
    615 		nxt = ip6->ip6_nxt;
    616 
    617 	/*
    618 	 * Check that the amount of data in the buffers is at least much as
    619 	 * the IPv6 header would have us expect. Trim mbufs if longer than we
    620 	 * expect. Drop packet if shorter than we expect.
    621 	 */
    622 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
    623 		IP6_STATINC(IP6_STAT_TOOSHORT);
    624 		in6_ifstat_inc(rcvif, ifs6_in_truncated);
    625 		goto bad_unref;
    626 	}
    627 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
    628 		if (m->m_len == m->m_pkthdr.len) {
    629 			m->m_len = sizeof(struct ip6_hdr) + plen;
    630 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
    631 		} else
    632 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
    633 	}
    634 
    635 	/*
    636 	 * Forward if desirable.
    637 	 */
    638 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    639 		/*
    640 		 * If we are acting as a multicast router, all
    641 		 * incoming multicast packets are passed to the
    642 		 * kernel-level multicast forwarding function.
    643 		 * The packet is returned (relatively) intact; if
    644 		 * ip6_mforward() returns a non-zero value, the packet
    645 		 * must be discarded, else it may be accepted below.
    646 		 */
    647 		if (ip6_mrouter != NULL) {
    648 			int error;
    649 
    650 			SOFTNET_LOCK();
    651 			error = ip6_mforward(ip6, rcvif, m);
    652 			SOFTNET_UNLOCK();
    653 
    654 			if (error != 0) {
    655 				rtcache_unref(rt, ro);
    656 				rtcache_percpu_putref(ip6_forward_rt_percpu);
    657 				IP6_STATINC(IP6_STAT_CANTFORWARD);
    658 				goto bad;
    659 			}
    660 		}
    661 		if (!ours) {
    662 			IP6_STATINC(IP6_STAT_CANTFORWARD);
    663 			goto bad_unref;
    664 		}
    665 	} else if (!ours) {
    666 		rtcache_unref(rt, ro);
    667 		rtcache_percpu_putref(ip6_forward_rt_percpu);
    668 		ip6_forward(m, srcrt, rcvif);
    669 		return;
    670 	}
    671 
    672 	ip6 = mtod(m, struct ip6_hdr *);
    673 
    674 	/*
    675 	 * Malicious party may be able to use IPv4 mapped addr to confuse
    676 	 * tcp/udp stack and bypass security checks (act as if it was from
    677 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
    678 	 *
    679 	 * For SIIT end node behavior, you may want to disable the check.
    680 	 * However, you will  become vulnerable to attacks using IPv4 mapped
    681 	 * source.
    682 	 */
    683 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    684 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    685 		IP6_STATINC(IP6_STAT_BADSCOPE);
    686 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
    687 		goto bad_unref;
    688 	}
    689 
    690 #ifdef IFA_STATS
    691 	if (deliverifp != NULL) {
    692 		struct in6_ifaddr *ia6;
    693 		int s = pserialize_read_enter();
    694 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    695 		if (ia6)
    696 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
    697 		pserialize_read_exit(s);
    698 	}
    699 #endif
    700 	IP6_STATINC(IP6_STAT_DELIVERED);
    701 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
    702 	nest = 0;
    703 
    704 	if (rt != NULL) {
    705 		rtcache_unref(rt, ro);
    706 		rt = NULL;
    707 	}
    708 	rtcache_percpu_putref(ip6_forward_rt_percpu);
    709 
    710 	rh_present = 0;
    711 	frg_present = 0;
    712 	while (nxt != IPPROTO_DONE) {
    713 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
    714 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
    715 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    716 			goto bad;
    717 		}
    718 
    719 		M_VERIFY_PACKET(m);
    720 
    721 		/*
    722 		 * protection against faulty packet - there should be
    723 		 * more sanity checks in header chain processing.
    724 		 */
    725 		if (m->m_pkthdr.len < off) {
    726 			IP6_STATINC(IP6_STAT_TOOSHORT);
    727 			in6_ifstat_inc(rcvif, ifs6_in_truncated);
    728 			goto bad;
    729 		}
    730 
    731 		if (nxt == IPPROTO_ROUTING) {
    732 			if (rh_present++) {
    733 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    734 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    735 				goto bad;
    736 			}
    737 		} else if (nxt == IPPROTO_FRAGMENT) {
    738 			if (frg_present++) {
    739 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    740 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    741 				goto bad;
    742 			}
    743 		}
    744 
    745 #ifdef IPSEC
    746 		if (ipsec_used) {
    747 			/*
    748 			 * Enforce IPsec policy checking if we are seeing last
    749 			 * header. Note that we do not visit this with
    750 			 * protocols with pcb layer code - like udp/tcp/raw ip.
    751 			 */
    752 			if ((inet6sw[ip6_protox[nxt]].pr_flags
    753 			    & PR_LASTHDR) != 0) {
    754 				int error;
    755 
    756 				error = ipsec_ip_input_checkpolicy(m, false);
    757 				if (error) {
    758 					IP6_STATINC(IP6_STAT_IPSECDROP_IN);
    759 					goto bad;
    760 				}
    761 			}
    762 		}
    763 #endif
    764 
    765 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
    766 	}
    767 	return;
    768 
    769 bad_unref:
    770 	rtcache_unref(rt, ro);
    771 	rtcache_percpu_putref(ip6_forward_rt_percpu);
    772 bad:
    773 	m_freem(m);
    774 	return;
    775 }
    776 
    777 static bool
    778 ip6_badaddr(struct ip6_hdr *ip6)
    779 {
    780 	/* Check against address spoofing/corruption. */
    781 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
    782 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
    783 		return true;
    784 	}
    785 
    786 	/*
    787 	 * The following check is not documented in specs.  A malicious
    788 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
    789 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
    790 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
    791 	 *
    792 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
    793 	 * support IPv4-less kernel compilation, we cannot support SIIT
    794 	 * environment at all.  So, it makes more sense for us to reject any
    795 	 * malicious packets for non-SIIT environment, than try to do a
    796 	 * partial support for SIIT environment.
    797 	 */
    798 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    799 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    800 		return true;
    801 	}
    802 
    803 	/*
    804 	 * Reject packets with IPv4-compatible IPv6 addresses (RFC4291).
    805 	 */
    806 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
    807 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
    808 		return true;
    809 	}
    810 
    811 	return false;
    812 }
    813 
    814 /*
    815  * set/grab in6_ifaddr correspond to IPv6 destination address.
    816  */
    817 static struct m_tag *
    818 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
    819 {
    820 	struct m_tag *mtag;
    821 	struct ip6aux *ip6a;
    822 
    823 	mtag = ip6_addaux(m);
    824 	if (mtag == NULL)
    825 		return NULL;
    826 
    827 	ip6a = (struct ip6aux *)(mtag + 1);
    828 	if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) {
    829 		IP6_STATINC(IP6_STAT_BADSCOPE);
    830 		return NULL;
    831 	}
    832 
    833 	ip6a->ip6a_src = ia->ia_addr.sin6_addr;
    834 	ip6a->ip6a_flags = ia->ia6_flags;
    835 	return mtag;
    836 }
    837 
    838 const struct ip6aux *
    839 ip6_getdstifaddr(struct mbuf *m)
    840 {
    841 	struct m_tag *mtag;
    842 
    843 	mtag = ip6_findaux(m);
    844 	if (mtag != NULL)
    845 		return (struct ip6aux *)(mtag + 1);
    846 	else
    847 		return NULL;
    848 }
    849 
    850 /*
    851  * Hop-by-Hop options header processing. If a valid jumbo payload option is
    852  * included, the real payload length will be stored in plenp.
    853  *
    854  * rtalertp - XXX: should be stored more smart way
    855  */
    856 int
    857 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
    858 	struct mbuf **mp, int *offp)
    859 {
    860 	struct mbuf *m = *mp;
    861 	int off = *offp, hbhlen;
    862 	struct ip6_hbh *hbh;
    863 
    864 	/* validation of the length of the header */
    865 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
    866 	    sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
    867 	if (hbh == NULL) {
    868 		IP6_STATINC(IP6_STAT_TOOSHORT);
    869 		return -1;
    870 	}
    871 	hbhlen = (hbh->ip6h_len + 1) << 3;
    872 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    873 	    hbhlen);
    874 	if (hbh == NULL) {
    875 		IP6_STATINC(IP6_STAT_TOOSHORT);
    876 		return -1;
    877 	}
    878 	KASSERT(ACCESSIBLE_POINTER(hbh, struct ip6_hdr));
    879 	off += hbhlen;
    880 	hbhlen -= sizeof(struct ip6_hbh);
    881 
    882 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
    883 	    hbhlen, rtalertp, plenp) < 0)
    884 		return -1;
    885 
    886 	*offp = off;
    887 	*mp = m;
    888 	return 0;
    889 }
    890 
    891 /*
    892  * Search header for all Hop-by-hop options and process each option.
    893  * This function is separate from ip6_hopopts_input() in order to
    894  * handle a case where the sending node itself process its hop-by-hop
    895  * options header. In such a case, the function is called from ip6_output().
    896  *
    897  * The function assumes that hbh header is located right after the IPv6 header
    898  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
    899  * opthead + hbhlen is located in continuous memory region.
    900  */
    901 static int
    902 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
    903 	u_int32_t *rtalertp, u_int32_t *plenp)
    904 {
    905 	struct ip6_hdr *ip6;
    906 	int optlen = 0;
    907 	u_int8_t *opt = opthead;
    908 	u_int16_t rtalert_val;
    909 	u_int32_t jumboplen;
    910 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
    911 
    912 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
    913 		switch (*opt) {
    914 		case IP6OPT_PAD1:
    915 			optlen = 1;
    916 			break;
    917 		case IP6OPT_PADN:
    918 			if (hbhlen < IP6OPT_MINLEN) {
    919 				IP6_STATINC(IP6_STAT_TOOSMALL);
    920 				goto bad;
    921 			}
    922 			optlen = *(opt + 1) + 2;
    923 			break;
    924 		case IP6OPT_RTALERT:
    925 			/* XXX may need check for alignment */
    926 			if (hbhlen < IP6OPT_RTALERT_LEN) {
    927 				IP6_STATINC(IP6_STAT_TOOSMALL);
    928 				goto bad;
    929 			}
    930 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
    931 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    932 				icmp6_error(m, ICMP6_PARAM_PROB,
    933 				    ICMP6_PARAMPROB_HEADER,
    934 				    erroff + opt + 1 - opthead);
    935 				return (-1);
    936 			}
    937 			optlen = IP6OPT_RTALERT_LEN;
    938 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
    939 			*rtalertp = ntohs(rtalert_val);
    940 			break;
    941 		case IP6OPT_JUMBO:
    942 			/* XXX may need check for alignment */
    943 			if (hbhlen < IP6OPT_JUMBO_LEN) {
    944 				IP6_STATINC(IP6_STAT_TOOSMALL);
    945 				goto bad;
    946 			}
    947 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
    948 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    949 				icmp6_error(m, ICMP6_PARAM_PROB,
    950 				    ICMP6_PARAMPROB_HEADER,
    951 				    erroff + opt + 1 - opthead);
    952 				return (-1);
    953 			}
    954 			optlen = IP6OPT_JUMBO_LEN;
    955 
    956 			/*
    957 			 * IPv6 packets that have non 0 payload length
    958 			 * must not contain a jumbo payload option.
    959 			 */
    960 			ip6 = mtod(m, struct ip6_hdr *);
    961 			if (ip6->ip6_plen) {
    962 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    963 				icmp6_error(m, ICMP6_PARAM_PROB,
    964 				    ICMP6_PARAMPROB_HEADER,
    965 				    erroff + opt - opthead);
    966 				return (-1);
    967 			}
    968 
    969 			/*
    970 			 * We may see jumbolen in unaligned location, so
    971 			 * we'd need to perform memcpy().
    972 			 */
    973 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
    974 			jumboplen = (u_int32_t)htonl(jumboplen);
    975 
    976 #if 1
    977 			/*
    978 			 * if there are multiple jumbo payload options,
    979 			 * *plenp will be non-zero and the packet will be
    980 			 * rejected.
    981 			 * the behavior may need some debate in ipngwg -
    982 			 * multiple options does not make sense, however,
    983 			 * there's no explicit mention in specification.
    984 			 */
    985 			if (*plenp != 0) {
    986 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    987 				icmp6_error(m, ICMP6_PARAM_PROB,
    988 				    ICMP6_PARAMPROB_HEADER,
    989 				    erroff + opt + 2 - opthead);
    990 				return (-1);
    991 			}
    992 #endif
    993 
    994 			/*
    995 			 * jumbo payload length must be larger than 65535.
    996 			 */
    997 			if (jumboplen <= IPV6_MAXPACKET) {
    998 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    999 				icmp6_error(m, ICMP6_PARAM_PROB,
   1000 				    ICMP6_PARAMPROB_HEADER,
   1001 				    erroff + opt + 2 - opthead);
   1002 				return (-1);
   1003 			}
   1004 			*plenp = jumboplen;
   1005 
   1006 			break;
   1007 		default:		/* unknown option */
   1008 			if (hbhlen < IP6OPT_MINLEN) {
   1009 				IP6_STATINC(IP6_STAT_TOOSMALL);
   1010 				goto bad;
   1011 			}
   1012 			optlen = ip6_unknown_opt(opt, m,
   1013 			    erroff + opt - opthead);
   1014 			if (optlen == -1)
   1015 				return (-1);
   1016 			optlen += 2;
   1017 			break;
   1018 		}
   1019 	}
   1020 
   1021 	return (0);
   1022 
   1023   bad:
   1024 	m_freem(m);
   1025 	return (-1);
   1026 }
   1027 
   1028 /*
   1029  * Unknown option processing.
   1030  * The third argument `off' is the offset from the IPv6 header to the option,
   1031  * which is necessary if the IPv6 header the and option header and IPv6 header
   1032  * is not continuous in order to return an ICMPv6 error.
   1033  */
   1034 int
   1035 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
   1036 {
   1037 	struct ip6_hdr *ip6;
   1038 
   1039 	switch (IP6OPT_TYPE(*optp)) {
   1040 	case IP6OPT_TYPE_SKIP: /* ignore the option */
   1041 		return ((int)*(optp + 1));
   1042 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
   1043 		m_freem(m);
   1044 		return (-1);
   1045 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
   1046 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1047 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
   1048 		return (-1);
   1049 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
   1050 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1051 		ip6 = mtod(m, struct ip6_hdr *);
   1052 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
   1053 		    (m->m_flags & (M_BCAST|M_MCAST)))
   1054 			m_freem(m);
   1055 		else
   1056 			icmp6_error(m, ICMP6_PARAM_PROB,
   1057 				    ICMP6_PARAMPROB_OPTION, off);
   1058 		return (-1);
   1059 	}
   1060 
   1061 	m_freem(m);		/* XXX: NOTREACHED */
   1062 	return (-1);
   1063 }
   1064 
   1065 void
   1066 ip6_savecontrol(struct inpcb *inp, struct mbuf **mp,
   1067 	struct ip6_hdr *ip6, struct mbuf *m)
   1068 {
   1069 	struct socket *so = inp->inp_socket;
   1070 #ifdef RFC2292
   1071 #define IS2292(x, y)	((inp->inp_flags & IN6P_RFC2292) ? (x) : (y))
   1072 #else
   1073 #define IS2292(x, y)	(y)
   1074 #endif
   1075 
   1076 	KASSERT(m->m_flags & M_PKTHDR);
   1077 
   1078 	if (SOOPT_TIMESTAMP(so->so_options))
   1079 		mp = sbsavetimestamp(so->so_options, mp);
   1080 
   1081 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
   1082 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
   1083 		return;
   1084 
   1085 	/* RFC 2292 sec. 5 */
   1086 	if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
   1087 		struct in6_pktinfo pi6;
   1088 
   1089 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
   1090 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
   1091 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index;
   1092 		*mp = sbcreatecontrol(&pi6, sizeof(pi6),
   1093 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
   1094 		if (*mp)
   1095 			mp = &(*mp)->m_next;
   1096 	}
   1097 
   1098 	if (inp->inp_flags & IN6P_HOPLIMIT) {
   1099 		int hlim = ip6->ip6_hlim & 0xff;
   1100 
   1101 		*mp = sbcreatecontrol(&hlim, sizeof(hlim),
   1102 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
   1103 		if (*mp)
   1104 			mp = &(*mp)->m_next;
   1105 	}
   1106 
   1107 	if ((inp->inp_flags & IN6P_TCLASS) != 0) {
   1108 		u_int32_t flowinfo;
   1109 		int tclass;
   1110 
   1111 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
   1112 		flowinfo >>= 20;
   1113 
   1114 		tclass = flowinfo & 0xff;
   1115 		*mp = sbcreatecontrol(&tclass, sizeof(tclass),
   1116 		    IPV6_TCLASS, IPPROTO_IPV6);
   1117 
   1118 		if (*mp)
   1119 			mp = &(*mp)->m_next;
   1120 	}
   1121 
   1122 	/*
   1123 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
   1124 	 * privilege for the option (see ip6_ctloutput), but it might be too
   1125 	 * strict, since there might be some hop-by-hop options which can be
   1126 	 * returned to normal user.
   1127 	 * See also RFC3542 section 8 (or RFC2292 section 6).
   1128 	 */
   1129 	if ((inp->inp_flags & IN6P_HOPOPTS) != 0) {
   1130 		/*
   1131 		 * Check if a hop-by-hop options header is contatined in the
   1132 		 * received packet, and if so, store the options as ancillary
   1133 		 * data. Note that a hop-by-hop options header must be
   1134 		 * just after the IPv6 header, which fact is assured through
   1135 		 * the IPv6 input processing.
   1136 		 */
   1137 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1138 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
   1139 			struct ip6_hbh *hbh;
   1140 			int hbhlen;
   1141 			struct mbuf *ext;
   1142 
   1143 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
   1144 			    xip6->ip6_nxt);
   1145 			if (ext == NULL) {
   1146 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1147 				return;
   1148 			}
   1149 			hbh = mtod(ext, struct ip6_hbh *);
   1150 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1151 			if (hbhlen != ext->m_len) {
   1152 				m_freem(ext);
   1153 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1154 				return;
   1155 			}
   1156 
   1157 			/*
   1158 			 * XXX: We copy whole the header even if a jumbo
   1159 			 * payload option is included, which option is to
   1160 			 * be removed before returning in the RFC 2292.
   1161 			 * Note: this constraint is removed in RFC3542.
   1162 			 */
   1163 			*mp = sbcreatecontrol(hbh, hbhlen,
   1164 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
   1165 			    IPPROTO_IPV6);
   1166 			if (*mp)
   1167 				mp = &(*mp)->m_next;
   1168 			m_freem(ext);
   1169 		}
   1170 	}
   1171 
   1172 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
   1173 	if (inp->inp_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
   1174 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1175 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
   1176 
   1177 		/*
   1178 		 * Search for destination options headers or routing
   1179 		 * header(s) through the header chain, and stores each
   1180 		 * header as ancillary data.
   1181 		 * Note that the order of the headers remains in
   1182 		 * the chain of ancillary data.
   1183 		 */
   1184 		for (;;) {	/* is explicit loop prevention necessary? */
   1185 			struct ip6_ext *ip6e = NULL;
   1186 			int elen;
   1187 			struct mbuf *ext = NULL;
   1188 
   1189 			/*
   1190 			 * if it is not an extension header, don't try to
   1191 			 * pull it from the chain.
   1192 			 */
   1193 			switch (nxt) {
   1194 			case IPPROTO_DSTOPTS:
   1195 			case IPPROTO_ROUTING:
   1196 			case IPPROTO_HOPOPTS:
   1197 			case IPPROTO_AH: /* is it possible? */
   1198 				break;
   1199 			default:
   1200 				goto loopend;
   1201 			}
   1202 
   1203 			ext = ip6_pullexthdr(m, off, nxt);
   1204 			if (ext == NULL) {
   1205 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1206 				return;
   1207 			}
   1208 			ip6e = mtod(ext, struct ip6_ext *);
   1209 			if (nxt == IPPROTO_AH)
   1210 				elen = (ip6e->ip6e_len + 2) << 2;
   1211 			else
   1212 				elen = (ip6e->ip6e_len + 1) << 3;
   1213 			if (elen != ext->m_len) {
   1214 				m_freem(ext);
   1215 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1216 				return;
   1217 			}
   1218 			KASSERT(ACCESSIBLE_POINTER(ip6e, struct ip6_hdr));
   1219 
   1220 			switch (nxt) {
   1221 			case IPPROTO_DSTOPTS:
   1222 				if (!(inp->inp_flags & IN6P_DSTOPTS))
   1223 					break;
   1224 
   1225 				*mp = sbcreatecontrol(ip6e, elen,
   1226 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
   1227 				    IPPROTO_IPV6);
   1228 				if (*mp)
   1229 					mp = &(*mp)->m_next;
   1230 				break;
   1231 
   1232 			case IPPROTO_ROUTING:
   1233 				if (!(inp->inp_flags & IN6P_RTHDR))
   1234 					break;
   1235 
   1236 				*mp = sbcreatecontrol(ip6e, elen,
   1237 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
   1238 				    IPPROTO_IPV6);
   1239 				if (*mp)
   1240 					mp = &(*mp)->m_next;
   1241 				break;
   1242 
   1243 			case IPPROTO_HOPOPTS:
   1244 			case IPPROTO_AH: /* is it possible? */
   1245 				break;
   1246 
   1247 			default:
   1248 				/*
   1249 			 	 * other cases have been filtered in the above.
   1250 				 * none will visit this case.  here we supply
   1251 				 * the code just in case (nxt overwritten or
   1252 				 * other cases).
   1253 				 */
   1254 				m_freem(ext);
   1255 				goto loopend;
   1256 
   1257 			}
   1258 
   1259 			/* proceed with the next header. */
   1260 			off += elen;
   1261 			nxt = ip6e->ip6e_nxt;
   1262 			ip6e = NULL;
   1263 			m_freem(ext);
   1264 			ext = NULL;
   1265 		}
   1266 	  loopend:
   1267 	  	;
   1268 	}
   1269 }
   1270 #undef IS2292
   1271 
   1272 
   1273 void
   1274 ip6_notify_pmtu(struct inpcb *inp, const struct sockaddr_in6 *dst,
   1275     uint32_t *mtu)
   1276 {
   1277 	struct socket *so;
   1278 	struct mbuf *m_mtu;
   1279 	struct ip6_mtuinfo mtuctl;
   1280 
   1281 	so = inp->inp_socket;
   1282 
   1283 	if (mtu == NULL)
   1284 		return;
   1285 
   1286 	KASSERT(so != NULL);
   1287 
   1288 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
   1289 	mtuctl.ip6m_mtu = *mtu;
   1290 	mtuctl.ip6m_addr = *dst;
   1291 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
   1292 		return;
   1293 
   1294 	if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl),
   1295 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
   1296 		return;
   1297 
   1298 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
   1299 	    == 0) {
   1300 		soroverflow(so);
   1301 		m_freem(m_mtu);
   1302 	} else
   1303 		sorwakeup(so);
   1304 
   1305 	return;
   1306 }
   1307 
   1308 /*
   1309  * pull single extension header from mbuf chain.  returns single mbuf that
   1310  * contains the result, or NULL on error.
   1311  */
   1312 static struct mbuf *
   1313 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
   1314 {
   1315 	struct ip6_ext ip6e;
   1316 	size_t elen;
   1317 	struct mbuf *n;
   1318 
   1319 	if (off + sizeof(ip6e) > m->m_pkthdr.len)
   1320 		return NULL;
   1321 
   1322 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1323 	if (nxt == IPPROTO_AH)
   1324 		elen = (ip6e.ip6e_len + 2) << 2;
   1325 	else
   1326 		elen = (ip6e.ip6e_len + 1) << 3;
   1327 
   1328 	if (off + elen > m->m_pkthdr.len)
   1329 		return NULL;
   1330 
   1331 	MGET(n, M_DONTWAIT, MT_DATA);
   1332 	if (n && elen >= MLEN) {
   1333 		MCLGET(n, M_DONTWAIT);
   1334 		if ((n->m_flags & M_EXT) == 0) {
   1335 			m_free(n);
   1336 			n = NULL;
   1337 		}
   1338 	}
   1339 	if (!n)
   1340 		return NULL;
   1341 
   1342 	n->m_len = 0;
   1343 	if (elen >= M_TRAILINGSPACE(n)) {
   1344 		m_free(n);
   1345 		return NULL;
   1346 	}
   1347 
   1348 	m_copydata(m, off, elen, mtod(n, void *));
   1349 	n->m_len = elen;
   1350 	return n;
   1351 }
   1352 
   1353 /*
   1354  * Get offset to the previous header followed by the header
   1355  * currently processed.
   1356  */
   1357 int
   1358 ip6_get_prevhdr(struct mbuf *m, int off)
   1359 {
   1360 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1361 
   1362 	if (off == sizeof(struct ip6_hdr)) {
   1363 		return offsetof(struct ip6_hdr, ip6_nxt);
   1364 	} else if (off < sizeof(struct ip6_hdr)) {
   1365 		panic("%s: off < sizeof(struct ip6_hdr)", __func__);
   1366 	} else {
   1367 		int len, nlen, nxt;
   1368 		struct ip6_ext ip6e;
   1369 
   1370 		nxt = ip6->ip6_nxt;
   1371 		len = sizeof(struct ip6_hdr);
   1372 		nlen = 0;
   1373 		while (len < off) {
   1374 			m_copydata(m, len, sizeof(ip6e), &ip6e);
   1375 
   1376 			switch (nxt) {
   1377 			case IPPROTO_FRAGMENT:
   1378 				nlen = sizeof(struct ip6_frag);
   1379 				break;
   1380 			case IPPROTO_AH:
   1381 				nlen = (ip6e.ip6e_len + 2) << 2;
   1382 				break;
   1383 			default:
   1384 				nlen = (ip6e.ip6e_len + 1) << 3;
   1385 				break;
   1386 			}
   1387 			len += nlen;
   1388 			nxt = ip6e.ip6e_nxt;
   1389 		}
   1390 
   1391 		return (len - nlen);
   1392 	}
   1393 }
   1394 
   1395 /*
   1396  * get next header offset.  m will be retained.
   1397  */
   1398 int
   1399 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1400 {
   1401 	struct ip6_hdr ip6;
   1402 	struct ip6_ext ip6e;
   1403 	struct ip6_frag fh;
   1404 
   1405 	/* just in case */
   1406 	if (m == NULL)
   1407 		panic("%s: m == NULL", __func__);
   1408 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
   1409 		return -1;
   1410 
   1411 	switch (proto) {
   1412 	case IPPROTO_IPV6:
   1413 		/* do not chase beyond intermediate IPv6 headers */
   1414 		if (off != 0)
   1415 			return -1;
   1416 		if (m->m_pkthdr.len < off + sizeof(ip6))
   1417 			return -1;
   1418 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
   1419 		if (nxtp)
   1420 			*nxtp = ip6.ip6_nxt;
   1421 		off += sizeof(ip6);
   1422 		return off;
   1423 
   1424 	case IPPROTO_FRAGMENT:
   1425 		/*
   1426 		 * terminate parsing if it is not the first fragment,
   1427 		 * it does not make sense to parse through it.
   1428 		 */
   1429 		if (m->m_pkthdr.len < off + sizeof(fh))
   1430 			return -1;
   1431 		m_copydata(m, off, sizeof(fh), (void *)&fh);
   1432 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
   1433 			return -1;
   1434 		if (nxtp)
   1435 			*nxtp = fh.ip6f_nxt;
   1436 		off += sizeof(struct ip6_frag);
   1437 		return off;
   1438 
   1439 	case IPPROTO_AH:
   1440 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1441 			return -1;
   1442 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1443 		if (nxtp)
   1444 			*nxtp = ip6e.ip6e_nxt;
   1445 		off += (ip6e.ip6e_len + 2) << 2;
   1446 		if (m->m_pkthdr.len < off)
   1447 			return -1;
   1448 		return off;
   1449 
   1450 	case IPPROTO_HOPOPTS:
   1451 	case IPPROTO_ROUTING:
   1452 	case IPPROTO_DSTOPTS:
   1453 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1454 			return -1;
   1455 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1456 		if (nxtp)
   1457 			*nxtp = ip6e.ip6e_nxt;
   1458 		off += (ip6e.ip6e_len + 1) << 3;
   1459 		if (m->m_pkthdr.len < off)
   1460 			return -1;
   1461 		return off;
   1462 
   1463 	case IPPROTO_NONE:
   1464 	case IPPROTO_ESP:
   1465 	case IPPROTO_IPCOMP:
   1466 		/* give up */
   1467 		return -1;
   1468 
   1469 	default:
   1470 		return -1;
   1471 	}
   1472 }
   1473 
   1474 /*
   1475  * get offset for the last header in the chain.  m will be kept untainted.
   1476  */
   1477 int
   1478 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1479 {
   1480 	int newoff;
   1481 	int nxt;
   1482 
   1483 	if (!nxtp) {
   1484 		nxt = -1;
   1485 		nxtp = &nxt;
   1486 	}
   1487 	for (;;) {
   1488 		newoff = ip6_nexthdr(m, off, proto, nxtp);
   1489 		if (newoff < 0)
   1490 			return off;
   1491 		else if (newoff < off)
   1492 			return -1;	/* invalid */
   1493 		else if (newoff == off)
   1494 			return newoff;
   1495 
   1496 		off = newoff;
   1497 		proto = *nxtp;
   1498 	}
   1499 }
   1500 
   1501 static struct m_tag *
   1502 ip6_addaux(struct mbuf *m)
   1503 {
   1504 	struct m_tag *mtag;
   1505 
   1506 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1507 	if (!mtag) {
   1508 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
   1509 		    M_NOWAIT);
   1510 		if (mtag) {
   1511 			m_tag_prepend(m, mtag);
   1512 			memset(mtag + 1, 0, sizeof(struct ip6aux));
   1513 		}
   1514 	}
   1515 	return mtag;
   1516 }
   1517 
   1518 static struct m_tag *
   1519 ip6_findaux(struct mbuf *m)
   1520 {
   1521 	struct m_tag *mtag;
   1522 
   1523 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1524 	return mtag;
   1525 }
   1526 
   1527 static void
   1528 ip6_delaux(struct mbuf *m)
   1529 {
   1530 	struct m_tag *mtag;
   1531 
   1532 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1533 	if (mtag)
   1534 		m_tag_delete(m, mtag);
   1535 }
   1536 
   1537 /*
   1538  * System control for IP6
   1539  */
   1540 
   1541 const u_char inet6ctlerrmap[PRC_NCMDS] = {
   1542 	0,		0,		0,		0,
   1543 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
   1544 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
   1545 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
   1546 	0,		0,		0,		0,
   1547 	ENOPROTOOPT
   1548 };
   1549 
   1550 extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS);
   1551 
   1552 static int
   1553 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
   1554 {
   1555 
   1556 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
   1557 }
   1558 
   1559 static void
   1560 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
   1561 {
   1562 	const struct sysctlnode *ip6_node;
   1563 
   1564 	sysctl_createv(clog, 0, NULL, NULL,
   1565 		       CTLFLAG_PERMANENT,
   1566 		       CTLTYPE_NODE, "inet6",
   1567 		       SYSCTL_DESCR("PF_INET6 related settings"),
   1568 		       NULL, 0, NULL, 0,
   1569 		       CTL_NET, PF_INET6, CTL_EOL);
   1570 	sysctl_createv(clog, 0, NULL, &ip6_node,
   1571 		       CTLFLAG_PERMANENT,
   1572 		       CTLTYPE_NODE, "ip6",
   1573 		       SYSCTL_DESCR("IPv6 related settings"),
   1574 		       NULL, 0, NULL, 0,
   1575 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
   1576 
   1577 	sysctl_createv(clog, 0, NULL, NULL,
   1578 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1579 		       CTLTYPE_INT, "forwarding",
   1580 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
   1581 		       NULL, 0, &ip6_forwarding, 0,
   1582 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1583 		       IPV6CTL_FORWARDING, CTL_EOL);
   1584 	sysctl_createv(clog, 0, NULL, NULL,
   1585 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1586 		       CTLTYPE_INT, "redirect",
   1587 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
   1588 		       NULL, 0, &ip6_sendredirects, 0,
   1589 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1590 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
   1591 	sysctl_createv(clog, 0, NULL, NULL,
   1592 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1593 		       CTLTYPE_INT, "hlim",
   1594 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
   1595 		       NULL, 0, &ip6_defhlim, 0,
   1596 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1597 		       IPV6CTL_DEFHLIM, CTL_EOL);
   1598 	sysctl_createv(clog, 0, NULL, NULL,
   1599 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1600 		       CTLTYPE_INT, "maxfragpackets",
   1601 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
   1602 				    "for reassembly"),
   1603 		       NULL, 0, &ip6_maxfragpackets, 0,
   1604 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1605 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
   1606 
   1607 	pktq_sysctl_setup(ip6_pktq, clog, ip6_node, IPV6CTL_IFQ);
   1608 
   1609 	sysctl_createv(clog, 0, NULL, NULL,
   1610 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1611 		       CTLTYPE_INT, "keepfaith",
   1612 		       SYSCTL_DESCR("Activate faith interface"),
   1613 		       NULL, 0, &ip6_keepfaith, 0,
   1614 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1615 		       IPV6CTL_KEEPFAITH, CTL_EOL);
   1616 	sysctl_createv(clog, 0, NULL, NULL,
   1617 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1618 		       CTLTYPE_INT, "log_interval",
   1619 		       SYSCTL_DESCR("Minimum interval between logging "
   1620 				    "unroutable packets"),
   1621 		       NULL, 0, &ip6_log_interval, 0,
   1622 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1623 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
   1624 	sysctl_createv(clog, 0, NULL, NULL,
   1625 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1626 		       CTLTYPE_INT, "hdrnestlimit",
   1627 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
   1628 		       NULL, 0, &ip6_hdrnestlimit, 0,
   1629 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1630 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
   1631 	sysctl_createv(clog, 0, NULL, NULL,
   1632 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1633 		       CTLTYPE_INT, "dad_count",
   1634 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
   1635 				    "probes to send"),
   1636 		       NULL, 0, &ip6_dad_count, 0,
   1637 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1638 		       IPV6CTL_DAD_COUNT, CTL_EOL);
   1639 	sysctl_createv(clog, 0, NULL, NULL,
   1640 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1641 		       CTLTYPE_INT, "auto_flowlabel",
   1642 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
   1643 		       NULL, 0, &ip6_auto_flowlabel, 0,
   1644 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1645 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
   1646 	sysctl_createv(clog, 0, NULL, NULL,
   1647 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1648 		       CTLTYPE_INT, "defmcasthlim",
   1649 		       SYSCTL_DESCR("Default multicast hop limit"),
   1650 		       NULL, 0, &ip6_defmcasthlim, 0,
   1651 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1652 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
   1653 	sysctl_createv(clog, 0, NULL, NULL,
   1654 		       CTLFLAG_PERMANENT,
   1655 		       CTLTYPE_STRING, "kame_version",
   1656 		       SYSCTL_DESCR("KAME Version"),
   1657 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
   1658 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1659 		       IPV6CTL_KAME_VERSION, CTL_EOL);
   1660 	sysctl_createv(clog, 0, NULL, NULL,
   1661 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1662 		       CTLTYPE_INT, "use_deprecated",
   1663 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
   1664 				    "source addresses"),
   1665 		       NULL, 0, &ip6_use_deprecated, 0,
   1666 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1667 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
   1668 	sysctl_createv(clog, 0, NULL, NULL,
   1669 		       CTLFLAG_PERMANENT
   1670 #ifndef INET6_BINDV6ONLY
   1671 		       |CTLFLAG_READWRITE,
   1672 #endif
   1673 		       CTLTYPE_INT, "v6only",
   1674 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
   1675 				    "to PF_INET sockets"),
   1676 		       NULL, 0, &ip6_v6only, 0,
   1677 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1678 		       IPV6CTL_V6ONLY, CTL_EOL);
   1679 	sysctl_createv(clog, 0, NULL, NULL,
   1680 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1681 		       CTLTYPE_INT, "anonportmin",
   1682 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
   1683 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
   1684 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1685 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
   1686 	sysctl_createv(clog, 0, NULL, NULL,
   1687 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1688 		       CTLTYPE_INT, "anonportmax",
   1689 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
   1690 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
   1691 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1692 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
   1693 #ifndef IPNOPRIVPORTS
   1694 	sysctl_createv(clog, 0, NULL, NULL,
   1695 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1696 		       CTLTYPE_INT, "lowportmin",
   1697 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
   1698 				    "to assign"),
   1699 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
   1700 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1701 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
   1702 	sysctl_createv(clog, 0, NULL, NULL,
   1703 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1704 		       CTLTYPE_INT, "lowportmax",
   1705 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
   1706 				    "to assign"),
   1707 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
   1708 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1709 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
   1710 #endif /* IPNOPRIVPORTS */
   1711 	sysctl_createv(clog, 0, NULL, NULL,
   1712 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1713 		       CTLTYPE_INT, "auto_linklocal",
   1714 		       SYSCTL_DESCR("Default value of per-interface flag for "
   1715 		                    "adding an IPv6 link-local address to "
   1716 				    "interfaces when attached"),
   1717 		       NULL, 0, &ip6_auto_linklocal, 0,
   1718 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1719 		       IPV6CTL_AUTO_LINKLOCAL, CTL_EOL);
   1720 	sysctl_createv(clog, 0, NULL, NULL,
   1721 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
   1722 		       CTLTYPE_STRUCT, "addctlpolicy",
   1723 		       SYSCTL_DESCR("Return the current address control"
   1724 			   " policy"),
   1725 		       sysctl_net_inet6_addrctlpolicy, 0, NULL, 0,
   1726 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1727 		       IPV6CTL_ADDRCTLPOLICY, CTL_EOL);
   1728 	sysctl_createv(clog, 0, NULL, NULL,
   1729 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1730 		       CTLTYPE_INT, "prefer_tempaddr",
   1731 		       SYSCTL_DESCR("Prefer temporary address as source "
   1732 		                    "address"),
   1733 		       NULL, 0, &ip6_prefer_tempaddr, 0,
   1734 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1735 		       CTL_CREATE, CTL_EOL);
   1736 	sysctl_createv(clog, 0, NULL, NULL,
   1737 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1738 		       CTLTYPE_INT, "maxfrags",
   1739 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
   1740 		       NULL, 0, &ip6_maxfrags, 0,
   1741 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1742 		       IPV6CTL_MAXFRAGS, CTL_EOL);
   1743 	sysctl_createv(clog, 0, NULL, NULL,
   1744 		       CTLFLAG_PERMANENT,
   1745 		       CTLTYPE_STRUCT, "stats",
   1746 		       SYSCTL_DESCR("IPv6 statistics"),
   1747 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
   1748 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1749 		       IPV6CTL_STATS, CTL_EOL);
   1750 	sysctl_createv(clog, 0, NULL, NULL,
   1751 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1752 		       CTLTYPE_INT, "use_defaultzone",
   1753 		       SYSCTL_DESCR("Whether to use the default scope zones"),
   1754 		       NULL, 0, &ip6_use_defzone, 0,
   1755 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1756 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
   1757 	sysctl_createv(clog, 0, NULL, NULL,
   1758 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1759 		       CTLTYPE_INT, "mcast_pmtu",
   1760 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
   1761 		       NULL, 0, &ip6_mcast_pmtu, 0,
   1762 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1763 		       CTL_CREATE, CTL_EOL);
   1764 	/* anonportalgo RFC6056 subtree */
   1765 	const struct sysctlnode *portalgo_node;
   1766 	sysctl_createv(clog, 0, NULL, &portalgo_node,
   1767 		       CTLFLAG_PERMANENT,
   1768 		       CTLTYPE_NODE, "anonportalgo",
   1769 		       SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"),
   1770 	    	       NULL, 0, NULL, 0,
   1771 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL);
   1772 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1773 		       CTLFLAG_PERMANENT,
   1774 		       CTLTYPE_STRING, "available",
   1775 		       SYSCTL_DESCR("available algorithms"),
   1776 		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
   1777 		       CTL_CREATE, CTL_EOL);
   1778 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1779 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1780 		       CTLTYPE_STRING, "selected",
   1781 		       SYSCTL_DESCR("selected algorithm"),
   1782 	               sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN,
   1783 		       CTL_CREATE, CTL_EOL);
   1784 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1785 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1786 		       CTLTYPE_STRUCT, "reserve",
   1787 		       SYSCTL_DESCR("bitmap of reserved ports"),
   1788 		       sysctl_portalgo_reserve6, 0, NULL, 0,
   1789 		       CTL_CREATE, CTL_EOL);
   1790 	sysctl_createv(clog, 0, NULL, NULL,
   1791 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1792 		       CTLTYPE_INT, "neighborgcthresh",
   1793 		       SYSCTL_DESCR("Maximum number of entries in neighbor"
   1794 			" cache"),
   1795 		       NULL, 1, &ip6_neighborgcthresh, 0,
   1796 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1797 		       CTL_CREATE, CTL_EOL);
   1798 	sysctl_createv(clog, 0, NULL, NULL,
   1799 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1800 		       CTLTYPE_INT, "maxdynroutes",
   1801 		       SYSCTL_DESCR("Maximum number of routes created via"
   1802 			   " redirect"),
   1803 		       NULL, 1, &ip6_maxdynroutes, 0,
   1804 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1805 		       CTL_CREATE, CTL_EOL);
   1806 	sysctl_createv(clog, 0, NULL, NULL,
   1807 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1808 		       CTLTYPE_INT, "param_rt_msg",
   1809 		       SYSCTL_DESCR("How to send parameter changing"
   1810 			   " routing message"),
   1811 		       NULL, 0, &ip6_param_rt_msg, 0,
   1812 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1813 		       CTL_CREATE, CTL_EOL);
   1814 }
   1815 
   1816 void
   1817 ip6_statinc(u_int stat)
   1818 {
   1819 
   1820 	KASSERT(stat < IP6_NSTATS);
   1821 	IP6_STATINC(stat);
   1822 }
   1823