Home | History | Annotate | Line # | Download | only in netinet6
ip6_input.c revision 1.212
      1 /*	$NetBSD: ip6_input.c,v 1.212 2019/10/16 07:40:40 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 1.212 2019/10/16 07:40:40 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 ip6intr(void *);
    138 static bool ip6_badaddr(struct ip6_hdr *);
    139 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
    140 
    141 static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *,
    142     u_int32_t *);
    143 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
    144 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
    145 
    146 #ifdef NET_MPSAFE
    147 #define	SOFTNET_LOCK()		mutex_enter(softnet_lock)
    148 #define	SOFTNET_UNLOCK()	mutex_exit(softnet_lock)
    149 #else
    150 #define	SOFTNET_LOCK()		KASSERT(mutex_owned(softnet_lock))
    151 #define	SOFTNET_UNLOCK()	KASSERT(mutex_owned(softnet_lock))
    152 #endif
    153 
    154 /*
    155  * IP6 initialization: fill in IP6 protocol switch table.
    156  * All protocols not implemented in kernel go to raw IP6 protocol handler.
    157  */
    158 void
    159 ip6_init(void)
    160 {
    161 	const struct ip6protosw *pr;
    162 	int i;
    163 
    164 	in6_init();
    165 
    166 	sysctl_net_inet6_ip6_setup(NULL);
    167 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
    168 	if (pr == 0)
    169 		panic("ip6_init");
    170 	for (i = 0; i < IPPROTO_MAX; i++)
    171 		ip6_protox[i] = pr - inet6sw;
    172 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
    173 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
    174 		if (pr->pr_domain->dom_family == PF_INET6 &&
    175 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
    176 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
    177 
    178 	ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL);
    179 	KASSERT(ip6_pktq != NULL);
    180 
    181 	scope6_init();
    182 	addrsel_policy_init();
    183 	nd6_init();
    184 	frag6_init();
    185 	ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR;
    186 
    187 	in6_tmpaddrtimer_init();
    188 #ifdef GATEWAY
    189 	ip6flow_init(ip6_hashsize);
    190 #endif
    191 	/* Register our Packet Filter hook. */
    192 	inet6_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET6);
    193 	KASSERT(inet6_pfil_hook != NULL);
    194 
    195 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
    196 	ip6_forward_rt_percpu = rtcache_percpu_alloc();
    197 }
    198 
    199 /*
    200  * IP6 input interrupt handling. Just pass the packet to ip6_input.
    201  */
    202 static void
    203 ip6intr(void *arg __unused)
    204 {
    205 	struct mbuf *m;
    206 
    207 	SOFTNET_KERNEL_LOCK_UNLESS_NET_MPSAFE();
    208 	while ((m = pktq_dequeue(ip6_pktq)) != NULL) {
    209 		struct psref psref;
    210 		struct ifnet *rcvif = m_get_rcvif_psref(m, &psref);
    211 
    212 		if (rcvif == NULL) {
    213 			m_freem(m);
    214 			continue;
    215 		}
    216 		/*
    217 		 * Drop the packet if IPv6 is disabled on the interface.
    218 		 */
    219 		if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED)) {
    220 			m_put_rcvif_psref(rcvif, &psref);
    221 			m_freem(m);
    222 			continue;
    223 		}
    224 		ip6_input(m, rcvif);
    225 		m_put_rcvif_psref(rcvif, &psref);
    226 	}
    227 	SOFTNET_KERNEL_UNLOCK_UNLESS_NET_MPSAFE();
    228 }
    229 
    230 void
    231 ip6_input(struct mbuf *m, struct ifnet *rcvif)
    232 {
    233 	struct ip6_hdr *ip6;
    234 	int hit, off = sizeof(struct ip6_hdr), nest;
    235 	u_int32_t plen;
    236 	u_int32_t rtalert = ~0;
    237 	int nxt, ours = 0, rh_present = 0, frg_present;
    238 	struct ifnet *deliverifp = NULL;
    239 	int srcrt = 0;
    240 	struct rtentry *rt = NULL;
    241 	union {
    242 		struct sockaddr		dst;
    243 		struct sockaddr_in6	dst6;
    244 	} u;
    245 	struct route *ro;
    246 
    247 	KASSERT(rcvif != NULL);
    248 
    249 	/*
    250 	 * make sure we don't have onion peering information into m_tag.
    251 	 */
    252 	ip6_delaux(m);
    253 
    254 	/*
    255 	 * mbuf statistics
    256 	 */
    257 	if (m->m_flags & M_EXT) {
    258 		if (m->m_next)
    259 			IP6_STATINC(IP6_STAT_MEXT2M);
    260 		else
    261 			IP6_STATINC(IP6_STAT_MEXT1);
    262 	} else {
    263 #define M2MMAX	32
    264 		if (m->m_next) {
    265 			if (m->m_flags & M_LOOP)
    266 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
    267 			else if (rcvif->if_index < M2MMAX)
    268 				IP6_STATINC(IP6_STAT_M2M + rcvif->if_index);
    269 			else
    270 				IP6_STATINC(IP6_STAT_M2M);
    271 		} else
    272 			IP6_STATINC(IP6_STAT_M1);
    273 #undef M2MMAX
    274 	}
    275 
    276 	in6_ifstat_inc(rcvif, ifs6_in_receive);
    277 	IP6_STATINC(IP6_STAT_TOTAL);
    278 
    279 	/*
    280 	 * If the IPv6 header is not aligned, slurp it up into a new
    281 	 * mbuf with space for link headers, in the event we forward
    282 	 * it.  Otherwise, if it is aligned, make sure the entire base
    283 	 * IPv6 header is in the first mbuf of the chain.
    284 	 */
    285 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
    286 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
    287 		    (max_linkhdr + 3) & ~3)) == NULL) {
    288 			/* XXXJRT new stat, please */
    289 			IP6_STATINC(IP6_STAT_TOOSMALL);
    290 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    291 			return;
    292 		}
    293 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
    294 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    295 			IP6_STATINC(IP6_STAT_TOOSMALL);
    296 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    297 			return;
    298 		}
    299 	}
    300 
    301 	ip6 = mtod(m, struct ip6_hdr *);
    302 
    303 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    304 		IP6_STATINC(IP6_STAT_BADVERS);
    305 		in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    306 		goto bad;
    307 	}
    308 
    309 	if (ip6_badaddr(ip6)) {
    310 		IP6_STATINC(IP6_STAT_BADSCOPE);
    311 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
    312 		goto bad;
    313 	}
    314 
    315 	/*
    316 	 * Assume that we can create a fast-forward IP flow entry
    317 	 * based on this packet.
    318 	 */
    319 	m->m_flags |= M_CANFASTFWD;
    320 
    321 	/*
    322 	 * Run through list of hooks for input packets.  If there are any
    323 	 * filters which require that additional packets in the flow are
    324 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
    325 	 * Note that filters must _never_ set this flag, as another filter
    326 	 * in the list may have previously cleared it.
    327 	 *
    328 	 * Don't call hooks if the packet has already been processed by
    329 	 * IPsec (encapsulated, tunnel mode).
    330 	 */
    331 #if defined(IPSEC)
    332 	if (!ipsec_used || !ipsec_skip_pfil(m))
    333 #else
    334 	if (1)
    335 #endif
    336 	{
    337 		struct in6_addr odst;
    338 		int error;
    339 
    340 		odst = ip6->ip6_dst;
    341 		error = pfil_run_hooks(inet6_pfil_hook, &m, rcvif, PFIL_IN);
    342 		if (error != 0 || m == NULL) {
    343 			IP6_STATINC(IP6_STAT_PFILDROP_IN);
    344 			return;
    345 		}
    346 		if (m->m_len < sizeof(struct ip6_hdr)) {
    347 			if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    348 				IP6_STATINC(IP6_STAT_TOOSMALL);
    349 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    350 				return;
    351 			}
    352 		}
    353 		ip6 = mtod(m, struct ip6_hdr *);
    354 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
    355 	}
    356 
    357 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
    358 
    359 #ifdef ALTQ
    360 	if (altq_input != NULL) {
    361 		SOFTNET_LOCK();
    362 		if ((*altq_input)(m, AF_INET6) == 0) {
    363 			SOFTNET_UNLOCK();
    364 			/* packet is dropped by traffic conditioner */
    365 			return;
    366 		}
    367 		SOFTNET_UNLOCK();
    368 	}
    369 #endif
    370 
    371 	/*
    372 	 * Disambiguate address scope zones (if there is ambiguity).
    373 	 * We first make sure that the original source or destination address
    374 	 * is not in our internal form for scoped addresses.  Such addresses
    375 	 * are not necessarily invalid spec-wise, but we cannot accept them due
    376 	 * to the usage conflict.
    377 	 * in6_setscope() then also checks and rejects the cases where src or
    378 	 * dst are the loopback address and the receiving interface
    379 	 * is not loopback.
    380 	 */
    381 	if (__predict_false(
    382 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
    383 		goto bad;
    384 	ip6 = mtod(m, struct ip6_hdr *);
    385 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
    386 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
    387 		goto bad;
    388 	}
    389 	if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
    390 	    in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
    391 		IP6_STATINC(IP6_STAT_BADSCOPE);
    392 		goto bad;
    393 	}
    394 
    395 	ro = rtcache_percpu_getref(ip6_forward_rt_percpu);
    396 
    397 	/*
    398 	 * Multicast check
    399 	 */
    400 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    401 		bool ingroup;
    402 
    403 		in6_ifstat_inc(rcvif, ifs6_in_mcast);
    404 		/*
    405 		 * See if we belong to the destination multicast group on the
    406 		 * arrival interface.
    407 		 */
    408 		ingroup = in6_multi_group(&ip6->ip6_dst, rcvif);
    409 		if (ingroup) {
    410 			ours = 1;
    411 		} else if (!ip6_mrouter) {
    412 			uint64_t *ip6s = IP6_STAT_GETREF();
    413 			ip6s[IP6_STAT_NOTMEMBER]++;
    414 			ip6s[IP6_STAT_CANTFORWARD]++;
    415 			IP6_STAT_PUTREF();
    416 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    417 			goto bad_unref;
    418 		}
    419 		deliverifp = rcvif;
    420 		goto hbhcheck;
    421 	}
    422 
    423 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
    424 
    425 	/*
    426 	 * Unicast check
    427 	 */
    428 	rt = rtcache_lookup2(ro, &u.dst, 1, &hit);
    429 	if (hit)
    430 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
    431 	else
    432 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
    433 
    434 	/*
    435 	 * Accept the packet if the forwarding interface to the destination
    436 	 * (according to the routing table) is the loopback interface,
    437 	 * unless the associated route has a gateway.
    438 	 *
    439 	 * We don't explicitly match ip6_dst against an interface here. It
    440 	 * is already done in rtcache_lookup2: rt->rt_ifp->if_type will be
    441 	 * IFT_LOOP if the packet is for us.
    442 	 *
    443 	 * Note that this approach causes to accept a packet if there is a
    444 	 * route to the loopback interface for the destination of the packet.
    445 	 * But we think it's even useful in some situations, e.g. when using
    446 	 * a special daemon which wants to intercept the packet.
    447 	 */
    448 	if (rt != NULL &&
    449 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
    450 	    rt->rt_ifp->if_type == IFT_LOOP) {
    451 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
    452 		int addrok;
    453 
    454 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
    455 			m->m_flags |= M_ANYCAST6;
    456 		/*
    457 		 * packets to a tentative, duplicated, or somehow invalid
    458 		 * address must not be accepted.
    459 		 */
    460 		if (ia6->ia6_flags & IN6_IFF_NOTREADY)
    461 			addrok = 0;
    462 		else if (ia6->ia6_flags & IN6_IFF_DETACHED &&
    463 		    !IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src))
    464 		{
    465 			/* Allow internal traffic to DETACHED addresses */
    466 			struct sockaddr_in6 sin6;
    467 			int s;
    468 
    469 			memset(&sin6, 0, sizeof(sin6));
    470 			sin6.sin6_family = AF_INET6;
    471 			sin6.sin6_len = sizeof(sin6);
    472 			sin6.sin6_addr = ip6->ip6_src;
    473 			s = pserialize_read_enter();
    474 			addrok = (ifa_ifwithaddr(sin6tosa(&sin6)) != NULL);
    475 			pserialize_read_exit(s);
    476 		} else
    477 			addrok = 1;
    478 		if (addrok) {
    479 			/* this address is ready */
    480 			ours = 1;
    481 			deliverifp = ia6->ia_ifp;	/* correct? */
    482 			goto hbhcheck;
    483 		} else {
    484 			/* address is not ready, so discard the packet. */
    485 			char ip6bufs[INET6_ADDRSTRLEN];
    486 			char ip6bufd[INET6_ADDRSTRLEN];
    487 			nd6log(LOG_INFO, "packet to an unready address %s->%s\n",
    488 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
    489 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
    490 
    491 			goto bad_unref;
    492 		}
    493 	}
    494 
    495 	/*
    496 	 * FAITH (Firewall Aided Internet Translator)
    497 	 */
    498 #if defined(NFAITH) && 0 < NFAITH
    499 	if (ip6_keepfaith) {
    500 		if (rt != NULL && rt->rt_ifp != NULL &&
    501 		    rt->rt_ifp->if_type == IFT_FAITH) {
    502 			/* XXX do we need more sanity checks? */
    503 			ours = 1;
    504 			deliverifp = rt->rt_ifp; /* faith */
    505 			goto hbhcheck;
    506 		}
    507 	}
    508 #endif
    509 
    510 	/*
    511 	 * Now there is no reason to process the packet if it's not our own
    512 	 * and we're not a router.
    513 	 */
    514 	if (!ip6_forwarding) {
    515 		IP6_STATINC(IP6_STAT_CANTFORWARD);
    516 		in6_ifstat_inc(rcvif, ifs6_in_discard);
    517 		goto bad_unref;
    518 	}
    519 
    520 hbhcheck:
    521 	/*
    522 	 * Record address information into m_tag, if we don't have one yet.
    523 	 * Note that we are unable to record it, if the address is not listed
    524 	 * as our interface address (e.g. multicast addresses, addresses
    525 	 * within FAITH prefixes and such).
    526 	 */
    527 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
    528 		struct in6_ifaddr *ia6;
    529 		int s = pserialize_read_enter();
    530 
    531 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    532 		/* Depends on ip6_setdstifaddr never sleep */
    533 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
    534 			/*
    535 			 * XXX maybe we should drop the packet here,
    536 			 * as we could not provide enough information
    537 			 * to the upper layers.
    538 			 */
    539 		}
    540 		pserialize_read_exit(s);
    541 	}
    542 
    543 	/*
    544 	 * Process Hop-by-Hop options header if it's contained.
    545 	 * m may be modified in ip6_hopopts_input().
    546 	 * If a JumboPayload option is included, plen will also be modified.
    547 	 */
    548 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
    549 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
    550 		struct ip6_hbh *hbh;
    551 
    552 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
    553 			/* m already freed */
    554 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    555 			rtcache_unref(rt, ro);
    556 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    557 			return;
    558 		}
    559 
    560 		/* adjust pointer */
    561 		ip6 = mtod(m, struct ip6_hdr *);
    562 
    563 		/*
    564 		 * if the payload length field is 0 and the next header field
    565 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
    566 		 * option MUST be included.
    567 		 */
    568 		if (ip6->ip6_plen == 0 && plen == 0) {
    569 			/*
    570 			 * Note that if a valid jumbo payload option is
    571 			 * contained, ip6_hopopts_input() must set a valid
    572 			 * (non-zero) payload length to the variable plen.
    573 			 */
    574 			IP6_STATINC(IP6_STAT_BADOPTIONS);
    575 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    576 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    577 			icmp6_error(m, ICMP6_PARAM_PROB,
    578 				    ICMP6_PARAMPROB_HEADER,
    579 				    (char *)&ip6->ip6_plen - (char *)ip6);
    580 			rtcache_unref(rt, ro);
    581 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    582 			return;
    583 		}
    584 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    585 			sizeof(struct ip6_hbh));
    586 		if (hbh == NULL) {
    587 			IP6_STATINC(IP6_STAT_TOOSHORT);
    588 			rtcache_unref(rt, ro);
    589 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    590 			return;
    591 		}
    592 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
    593 		nxt = hbh->ip6h_nxt;
    594 
    595 		/*
    596 		 * accept the packet if a router alert option is included
    597 		 * and we act as an IPv6 router.
    598 		 */
    599 		if (rtalert != ~0 && ip6_forwarding)
    600 			ours = 1;
    601 	} else
    602 		nxt = ip6->ip6_nxt;
    603 
    604 	/*
    605 	 * Check that the amount of data in the buffers is at least much as
    606 	 * the IPv6 header would have us expect. Trim mbufs if longer than we
    607 	 * expect. Drop packet if shorter than we expect.
    608 	 */
    609 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
    610 		IP6_STATINC(IP6_STAT_TOOSHORT);
    611 		in6_ifstat_inc(rcvif, ifs6_in_truncated);
    612 		goto bad_unref;
    613 	}
    614 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
    615 		if (m->m_len == m->m_pkthdr.len) {
    616 			m->m_len = sizeof(struct ip6_hdr) + plen;
    617 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
    618 		} else
    619 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
    620 	}
    621 
    622 	/*
    623 	 * Forward if desirable.
    624 	 */
    625 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    626 		/*
    627 		 * If we are acting as a multicast router, all
    628 		 * incoming multicast packets are passed to the
    629 		 * kernel-level multicast forwarding function.
    630 		 * The packet is returned (relatively) intact; if
    631 		 * ip6_mforward() returns a non-zero value, the packet
    632 		 * must be discarded, else it may be accepted below.
    633 		 */
    634 		if (ip6_mrouter != NULL) {
    635 			int error;
    636 
    637 			SOFTNET_LOCK();
    638 			error = ip6_mforward(ip6, rcvif, m);
    639 			SOFTNET_UNLOCK();
    640 
    641 			if (error != 0) {
    642 				rtcache_unref(rt, ro);
    643 				rtcache_percpu_putref(ip6_forward_rt_percpu);
    644 				IP6_STATINC(IP6_STAT_CANTFORWARD);
    645 				goto bad;
    646 			}
    647 		}
    648 		if (!ours)
    649 			goto bad_unref;
    650 	} else if (!ours) {
    651 		rtcache_unref(rt, ro);
    652 		rtcache_percpu_putref(ip6_forward_rt_percpu);
    653 		ip6_forward(m, srcrt);
    654 		return;
    655 	}
    656 
    657 	ip6 = mtod(m, struct ip6_hdr *);
    658 
    659 	/*
    660 	 * Malicious party may be able to use IPv4 mapped addr to confuse
    661 	 * tcp/udp stack and bypass security checks (act as if it was from
    662 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
    663 	 *
    664 	 * For SIIT end node behavior, you may want to disable the check.
    665 	 * However, you will  become vulnerable to attacks using IPv4 mapped
    666 	 * source.
    667 	 */
    668 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    669 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    670 		IP6_STATINC(IP6_STAT_BADSCOPE);
    671 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
    672 		goto bad_unref;
    673 	}
    674 
    675 #ifdef IFA_STATS
    676 	if (deliverifp != NULL) {
    677 		struct in6_ifaddr *ia6;
    678 		int s = pserialize_read_enter();
    679 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    680 		if (ia6)
    681 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
    682 		pserialize_read_exit(s);
    683 	}
    684 #endif
    685 	IP6_STATINC(IP6_STAT_DELIVERED);
    686 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
    687 	nest = 0;
    688 
    689 	if (rt != NULL) {
    690 		rtcache_unref(rt, ro);
    691 		rt = NULL;
    692 	}
    693 	rtcache_percpu_putref(ip6_forward_rt_percpu);
    694 
    695 	rh_present = 0;
    696 	frg_present = 0;
    697 	while (nxt != IPPROTO_DONE) {
    698 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
    699 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
    700 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    701 			goto bad;
    702 		}
    703 
    704 		M_VERIFY_PACKET(m);
    705 
    706 		/*
    707 		 * protection against faulty packet - there should be
    708 		 * more sanity checks in header chain processing.
    709 		 */
    710 		if (m->m_pkthdr.len < off) {
    711 			IP6_STATINC(IP6_STAT_TOOSHORT);
    712 			in6_ifstat_inc(rcvif, ifs6_in_truncated);
    713 			goto bad;
    714 		}
    715 
    716 		if (nxt == IPPROTO_ROUTING) {
    717 			if (rh_present++) {
    718 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    719 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    720 				goto bad;
    721 			}
    722 		} else if (nxt == IPPROTO_FRAGMENT) {
    723 			if (frg_present++) {
    724 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    725 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    726 				goto bad;
    727 			}
    728 		}
    729 
    730 #ifdef IPSEC
    731 		if (ipsec_used) {
    732 			/*
    733 			 * Enforce IPsec policy checking if we are seeing last
    734 			 * header. Note that we do not visit this with
    735 			 * protocols with pcb layer code - like udp/tcp/raw ip.
    736 			 */
    737 			if ((inet6sw[ip6_protox[nxt]].pr_flags
    738 			    & PR_LASTHDR) != 0) {
    739 				int error;
    740 
    741 				error = ipsec_ip_input(m, false);
    742 				if (error)
    743 					goto bad;
    744 			}
    745 		}
    746 #endif
    747 
    748 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
    749 	}
    750 	return;
    751 
    752 bad_unref:
    753 	rtcache_unref(rt, ro);
    754 	rtcache_percpu_putref(ip6_forward_rt_percpu);
    755 bad:
    756 	m_freem(m);
    757 	return;
    758 }
    759 
    760 static bool
    761 ip6_badaddr(struct ip6_hdr *ip6)
    762 {
    763 	/* Check against address spoofing/corruption. */
    764 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
    765 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
    766 		return true;
    767 	}
    768 
    769 	/*
    770 	 * The following check is not documented in specs.  A malicious
    771 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
    772 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
    773 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
    774 	 *
    775 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
    776 	 * support IPv4-less kernel compilation, we cannot support SIIT
    777 	 * environment at all.  So, it makes more sense for us to reject any
    778 	 * malicious packets for non-SIIT environment, than try to do a
    779 	 * partial support for SIIT environment.
    780 	 */
    781 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    782 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    783 		return true;
    784 	}
    785 
    786 	/*
    787 	 * Reject packets with IPv4-compatible IPv6 addresses (RFC4291).
    788 	 */
    789 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
    790 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
    791 		return true;
    792 	}
    793 
    794 	return false;
    795 }
    796 
    797 /*
    798  * set/grab in6_ifaddr correspond to IPv6 destination address.
    799  */
    800 static struct m_tag *
    801 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
    802 {
    803 	struct m_tag *mtag;
    804 	struct ip6aux *ip6a;
    805 
    806 	mtag = ip6_addaux(m);
    807 	if (mtag == NULL)
    808 		return NULL;
    809 
    810 	ip6a = (struct ip6aux *)(mtag + 1);
    811 	if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) {
    812 		IP6_STATINC(IP6_STAT_BADSCOPE);
    813 		return NULL;
    814 	}
    815 
    816 	ip6a->ip6a_src = ia->ia_addr.sin6_addr;
    817 	ip6a->ip6a_flags = ia->ia6_flags;
    818 	return mtag;
    819 }
    820 
    821 const struct ip6aux *
    822 ip6_getdstifaddr(struct mbuf *m)
    823 {
    824 	struct m_tag *mtag;
    825 
    826 	mtag = ip6_findaux(m);
    827 	if (mtag != NULL)
    828 		return (struct ip6aux *)(mtag + 1);
    829 	else
    830 		return NULL;
    831 }
    832 
    833 /*
    834  * Hop-by-Hop options header processing. If a valid jumbo payload option is
    835  * included, the real payload length will be stored in plenp.
    836  *
    837  * rtalertp - XXX: should be stored more smart way
    838  */
    839 int
    840 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
    841 	struct mbuf **mp, int *offp)
    842 {
    843 	struct mbuf *m = *mp;
    844 	int off = *offp, hbhlen;
    845 	struct ip6_hbh *hbh;
    846 
    847 	/* validation of the length of the header */
    848 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
    849 	    sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
    850 	if (hbh == NULL) {
    851 		IP6_STATINC(IP6_STAT_TOOSHORT);
    852 		return -1;
    853 	}
    854 	hbhlen = (hbh->ip6h_len + 1) << 3;
    855 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    856 	    hbhlen);
    857 	if (hbh == NULL) {
    858 		IP6_STATINC(IP6_STAT_TOOSHORT);
    859 		return -1;
    860 	}
    861 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
    862 	off += hbhlen;
    863 	hbhlen -= sizeof(struct ip6_hbh);
    864 
    865 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
    866 	    hbhlen, rtalertp, plenp) < 0)
    867 		return -1;
    868 
    869 	*offp = off;
    870 	*mp = m;
    871 	return 0;
    872 }
    873 
    874 /*
    875  * Search header for all Hop-by-hop options and process each option.
    876  * This function is separate from ip6_hopopts_input() in order to
    877  * handle a case where the sending node itself process its hop-by-hop
    878  * options header. In such a case, the function is called from ip6_output().
    879  *
    880  * The function assumes that hbh header is located right after the IPv6 header
    881  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
    882  * opthead + hbhlen is located in continuous memory region.
    883  */
    884 static int
    885 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
    886 	u_int32_t *rtalertp, u_int32_t *plenp)
    887 {
    888 	struct ip6_hdr *ip6;
    889 	int optlen = 0;
    890 	u_int8_t *opt = opthead;
    891 	u_int16_t rtalert_val;
    892 	u_int32_t jumboplen;
    893 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
    894 
    895 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
    896 		switch (*opt) {
    897 		case IP6OPT_PAD1:
    898 			optlen = 1;
    899 			break;
    900 		case IP6OPT_PADN:
    901 			if (hbhlen < IP6OPT_MINLEN) {
    902 				IP6_STATINC(IP6_STAT_TOOSMALL);
    903 				goto bad;
    904 			}
    905 			optlen = *(opt + 1) + 2;
    906 			break;
    907 		case IP6OPT_RTALERT:
    908 			/* XXX may need check for alignment */
    909 			if (hbhlen < IP6OPT_RTALERT_LEN) {
    910 				IP6_STATINC(IP6_STAT_TOOSMALL);
    911 				goto bad;
    912 			}
    913 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
    914 				/* XXX stat */
    915 				icmp6_error(m, ICMP6_PARAM_PROB,
    916 				    ICMP6_PARAMPROB_HEADER,
    917 				    erroff + opt + 1 - opthead);
    918 				return (-1);
    919 			}
    920 			optlen = IP6OPT_RTALERT_LEN;
    921 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
    922 			*rtalertp = ntohs(rtalert_val);
    923 			break;
    924 		case IP6OPT_JUMBO:
    925 			/* XXX may need check for alignment */
    926 			if (hbhlen < IP6OPT_JUMBO_LEN) {
    927 				IP6_STATINC(IP6_STAT_TOOSMALL);
    928 				goto bad;
    929 			}
    930 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
    931 				/* XXX stat */
    932 				icmp6_error(m, ICMP6_PARAM_PROB,
    933 				    ICMP6_PARAMPROB_HEADER,
    934 				    erroff + opt + 1 - opthead);
    935 				return (-1);
    936 			}
    937 			optlen = IP6OPT_JUMBO_LEN;
    938 
    939 			/*
    940 			 * IPv6 packets that have non 0 payload length
    941 			 * must not contain a jumbo payload option.
    942 			 */
    943 			ip6 = mtod(m, struct ip6_hdr *);
    944 			if (ip6->ip6_plen) {
    945 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    946 				icmp6_error(m, ICMP6_PARAM_PROB,
    947 				    ICMP6_PARAMPROB_HEADER,
    948 				    erroff + opt - opthead);
    949 				return (-1);
    950 			}
    951 
    952 			/*
    953 			 * We may see jumbolen in unaligned location, so
    954 			 * we'd need to perform memcpy().
    955 			 */
    956 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
    957 			jumboplen = (u_int32_t)htonl(jumboplen);
    958 
    959 #if 1
    960 			/*
    961 			 * if there are multiple jumbo payload options,
    962 			 * *plenp will be non-zero and the packet will be
    963 			 * rejected.
    964 			 * the behavior may need some debate in ipngwg -
    965 			 * multiple options does not make sense, however,
    966 			 * there's no explicit mention in specification.
    967 			 */
    968 			if (*plenp != 0) {
    969 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    970 				icmp6_error(m, ICMP6_PARAM_PROB,
    971 				    ICMP6_PARAMPROB_HEADER,
    972 				    erroff + opt + 2 - opthead);
    973 				return (-1);
    974 			}
    975 #endif
    976 
    977 			/*
    978 			 * jumbo payload length must be larger than 65535.
    979 			 */
    980 			if (jumboplen <= IPV6_MAXPACKET) {
    981 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    982 				icmp6_error(m, ICMP6_PARAM_PROB,
    983 				    ICMP6_PARAMPROB_HEADER,
    984 				    erroff + opt + 2 - opthead);
    985 				return (-1);
    986 			}
    987 			*plenp = jumboplen;
    988 
    989 			break;
    990 		default:		/* unknown option */
    991 			if (hbhlen < IP6OPT_MINLEN) {
    992 				IP6_STATINC(IP6_STAT_TOOSMALL);
    993 				goto bad;
    994 			}
    995 			optlen = ip6_unknown_opt(opt, m,
    996 			    erroff + opt - opthead);
    997 			if (optlen == -1)
    998 				return (-1);
    999 			optlen += 2;
   1000 			break;
   1001 		}
   1002 	}
   1003 
   1004 	return (0);
   1005 
   1006   bad:
   1007 	m_freem(m);
   1008 	return (-1);
   1009 }
   1010 
   1011 /*
   1012  * Unknown option processing.
   1013  * The third argument `off' is the offset from the IPv6 header to the option,
   1014  * which is necessary if the IPv6 header the and option header and IPv6 header
   1015  * is not continuous in order to return an ICMPv6 error.
   1016  */
   1017 int
   1018 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
   1019 {
   1020 	struct ip6_hdr *ip6;
   1021 
   1022 	switch (IP6OPT_TYPE(*optp)) {
   1023 	case IP6OPT_TYPE_SKIP: /* ignore the option */
   1024 		return ((int)*(optp + 1));
   1025 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
   1026 		m_freem(m);
   1027 		return (-1);
   1028 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
   1029 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1030 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
   1031 		return (-1);
   1032 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
   1033 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1034 		ip6 = mtod(m, struct ip6_hdr *);
   1035 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
   1036 		    (m->m_flags & (M_BCAST|M_MCAST)))
   1037 			m_freem(m);
   1038 		else
   1039 			icmp6_error(m, ICMP6_PARAM_PROB,
   1040 				    ICMP6_PARAMPROB_OPTION, off);
   1041 		return (-1);
   1042 	}
   1043 
   1044 	m_freem(m);		/* XXX: NOTREACHED */
   1045 	return (-1);
   1046 }
   1047 
   1048 void
   1049 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
   1050 	struct ip6_hdr *ip6, struct mbuf *m)
   1051 {
   1052 	struct socket *so = in6p->in6p_socket;
   1053 #ifdef RFC2292
   1054 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1055 #else
   1056 #define IS2292(x, y)	(y)
   1057 #endif
   1058 
   1059 	if (SOOPT_TIMESTAMP(so->so_options))
   1060 		mp = sbsavetimestamp(so->so_options, mp);
   1061 
   1062 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
   1063 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
   1064 		return;
   1065 
   1066 	/* RFC 2292 sec. 5 */
   1067 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
   1068 		struct in6_pktinfo pi6;
   1069 
   1070 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
   1071 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
   1072 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index;
   1073 		*mp = sbcreatecontrol(&pi6, sizeof(pi6),
   1074 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
   1075 		if (*mp)
   1076 			mp = &(*mp)->m_next;
   1077 	}
   1078 
   1079 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
   1080 		int hlim = ip6->ip6_hlim & 0xff;
   1081 
   1082 		*mp = sbcreatecontrol(&hlim, sizeof(hlim),
   1083 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
   1084 		if (*mp)
   1085 			mp = &(*mp)->m_next;
   1086 	}
   1087 
   1088 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
   1089 		u_int32_t flowinfo;
   1090 		int tclass;
   1091 
   1092 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
   1093 		flowinfo >>= 20;
   1094 
   1095 		tclass = flowinfo & 0xff;
   1096 		*mp = sbcreatecontrol(&tclass, sizeof(tclass),
   1097 		    IPV6_TCLASS, IPPROTO_IPV6);
   1098 
   1099 		if (*mp)
   1100 			mp = &(*mp)->m_next;
   1101 	}
   1102 
   1103 	/*
   1104 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
   1105 	 * privilege for the option (see ip6_ctloutput), but it might be too
   1106 	 * strict, since there might be some hop-by-hop options which can be
   1107 	 * returned to normal user.
   1108 	 * See also RFC3542 section 8 (or RFC2292 section 6).
   1109 	 */
   1110 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
   1111 		/*
   1112 		 * Check if a hop-by-hop options header is contatined in the
   1113 		 * received packet, and if so, store the options as ancillary
   1114 		 * data. Note that a hop-by-hop options header must be
   1115 		 * just after the IPv6 header, which fact is assured through
   1116 		 * the IPv6 input processing.
   1117 		 */
   1118 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1119 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
   1120 			struct ip6_hbh *hbh;
   1121 			int hbhlen;
   1122 			struct mbuf *ext;
   1123 
   1124 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
   1125 			    xip6->ip6_nxt);
   1126 			if (ext == NULL) {
   1127 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1128 				return;
   1129 			}
   1130 			hbh = mtod(ext, struct ip6_hbh *);
   1131 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1132 			if (hbhlen != ext->m_len) {
   1133 				m_freem(ext);
   1134 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1135 				return;
   1136 			}
   1137 
   1138 			/*
   1139 			 * XXX: We copy whole the header even if a jumbo
   1140 			 * payload option is included, which option is to
   1141 			 * be removed before returning in the RFC 2292.
   1142 			 * Note: this constraint is removed in RFC3542.
   1143 			 */
   1144 			*mp = sbcreatecontrol(hbh, hbhlen,
   1145 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
   1146 			    IPPROTO_IPV6);
   1147 			if (*mp)
   1148 				mp = &(*mp)->m_next;
   1149 			m_freem(ext);
   1150 		}
   1151 	}
   1152 
   1153 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
   1154 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
   1155 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1156 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
   1157 
   1158 		/*
   1159 		 * Search for destination options headers or routing
   1160 		 * header(s) through the header chain, and stores each
   1161 		 * header as ancillary data.
   1162 		 * Note that the order of the headers remains in
   1163 		 * the chain of ancillary data.
   1164 		 */
   1165 		for (;;) {	/* is explicit loop prevention necessary? */
   1166 			struct ip6_ext *ip6e = NULL;
   1167 			int elen;
   1168 			struct mbuf *ext = NULL;
   1169 
   1170 			/*
   1171 			 * if it is not an extension header, don't try to
   1172 			 * pull it from the chain.
   1173 			 */
   1174 			switch (nxt) {
   1175 			case IPPROTO_DSTOPTS:
   1176 			case IPPROTO_ROUTING:
   1177 			case IPPROTO_HOPOPTS:
   1178 			case IPPROTO_AH: /* is it possible? */
   1179 				break;
   1180 			default:
   1181 				goto loopend;
   1182 			}
   1183 
   1184 			ext = ip6_pullexthdr(m, off, nxt);
   1185 			if (ext == NULL) {
   1186 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1187 				return;
   1188 			}
   1189 			ip6e = mtod(ext, struct ip6_ext *);
   1190 			if (nxt == IPPROTO_AH)
   1191 				elen = (ip6e->ip6e_len + 2) << 2;
   1192 			else
   1193 				elen = (ip6e->ip6e_len + 1) << 3;
   1194 			if (elen != ext->m_len) {
   1195 				m_freem(ext);
   1196 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1197 				return;
   1198 			}
   1199 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
   1200 
   1201 			switch (nxt) {
   1202 			case IPPROTO_DSTOPTS:
   1203 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
   1204 					break;
   1205 
   1206 				*mp = sbcreatecontrol(ip6e, elen,
   1207 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
   1208 				    IPPROTO_IPV6);
   1209 				if (*mp)
   1210 					mp = &(*mp)->m_next;
   1211 				break;
   1212 
   1213 			case IPPROTO_ROUTING:
   1214 				if (!(in6p->in6p_flags & IN6P_RTHDR))
   1215 					break;
   1216 
   1217 				*mp = sbcreatecontrol(ip6e, elen,
   1218 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
   1219 				    IPPROTO_IPV6);
   1220 				if (*mp)
   1221 					mp = &(*mp)->m_next;
   1222 				break;
   1223 
   1224 			case IPPROTO_HOPOPTS:
   1225 			case IPPROTO_AH: /* is it possible? */
   1226 				break;
   1227 
   1228 			default:
   1229 				/*
   1230 			 	 * other cases have been filtered in the above.
   1231 				 * none will visit this case.  here we supply
   1232 				 * the code just in case (nxt overwritten or
   1233 				 * other cases).
   1234 				 */
   1235 				m_freem(ext);
   1236 				goto loopend;
   1237 
   1238 			}
   1239 
   1240 			/* proceed with the next header. */
   1241 			off += elen;
   1242 			nxt = ip6e->ip6e_nxt;
   1243 			ip6e = NULL;
   1244 			m_freem(ext);
   1245 			ext = NULL;
   1246 		}
   1247 	  loopend:
   1248 	  	;
   1249 	}
   1250 }
   1251 #undef IS2292
   1252 
   1253 
   1254 void
   1255 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
   1256     uint32_t *mtu)
   1257 {
   1258 	struct socket *so;
   1259 	struct mbuf *m_mtu;
   1260 	struct ip6_mtuinfo mtuctl;
   1261 
   1262 	so = in6p->in6p_socket;
   1263 
   1264 	if (mtu == NULL)
   1265 		return;
   1266 
   1267 	KASSERT(so != NULL);
   1268 
   1269 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
   1270 	mtuctl.ip6m_mtu = *mtu;
   1271 	mtuctl.ip6m_addr = *dst;
   1272 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
   1273 		return;
   1274 
   1275 	if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl),
   1276 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
   1277 		return;
   1278 
   1279 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
   1280 	    == 0) {
   1281 		soroverflow(so);
   1282 		m_freem(m_mtu);
   1283 	} else
   1284 		sorwakeup(so);
   1285 
   1286 	return;
   1287 }
   1288 
   1289 /*
   1290  * pull single extension header from mbuf chain.  returns single mbuf that
   1291  * contains the result, or NULL on error.
   1292  */
   1293 static struct mbuf *
   1294 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
   1295 {
   1296 	struct ip6_ext ip6e;
   1297 	size_t elen;
   1298 	struct mbuf *n;
   1299 
   1300 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1301 	if (nxt == IPPROTO_AH)
   1302 		elen = (ip6e.ip6e_len + 2) << 2;
   1303 	else
   1304 		elen = (ip6e.ip6e_len + 1) << 3;
   1305 
   1306 	MGET(n, M_DONTWAIT, MT_DATA);
   1307 	if (n && elen >= MLEN) {
   1308 		MCLGET(n, M_DONTWAIT);
   1309 		if ((n->m_flags & M_EXT) == 0) {
   1310 			m_free(n);
   1311 			n = NULL;
   1312 		}
   1313 	}
   1314 	if (!n)
   1315 		return NULL;
   1316 
   1317 	n->m_len = 0;
   1318 	if (elen >= M_TRAILINGSPACE(n)) {
   1319 		m_free(n);
   1320 		return NULL;
   1321 	}
   1322 
   1323 	m_copydata(m, off, elen, mtod(n, void *));
   1324 	n->m_len = elen;
   1325 	return n;
   1326 }
   1327 
   1328 /*
   1329  * Get offset to the previous header followed by the header
   1330  * currently processed.
   1331  */
   1332 int
   1333 ip6_get_prevhdr(struct mbuf *m, int off)
   1334 {
   1335 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1336 
   1337 	if (off == sizeof(struct ip6_hdr)) {
   1338 		return offsetof(struct ip6_hdr, ip6_nxt);
   1339 	} else if (off < sizeof(struct ip6_hdr)) {
   1340 		panic("%s: off < sizeof(struct ip6_hdr)", __func__);
   1341 	} else {
   1342 		int len, nlen, nxt;
   1343 		struct ip6_ext ip6e;
   1344 
   1345 		nxt = ip6->ip6_nxt;
   1346 		len = sizeof(struct ip6_hdr);
   1347 		nlen = 0;
   1348 		while (len < off) {
   1349 			m_copydata(m, len, sizeof(ip6e), &ip6e);
   1350 
   1351 			switch (nxt) {
   1352 			case IPPROTO_FRAGMENT:
   1353 				nlen = sizeof(struct ip6_frag);
   1354 				break;
   1355 			case IPPROTO_AH:
   1356 				nlen = (ip6e.ip6e_len + 2) << 2;
   1357 				break;
   1358 			default:
   1359 				nlen = (ip6e.ip6e_len + 1) << 3;
   1360 				break;
   1361 			}
   1362 			len += nlen;
   1363 			nxt = ip6e.ip6e_nxt;
   1364 		}
   1365 
   1366 		return (len - nlen);
   1367 	}
   1368 }
   1369 
   1370 /*
   1371  * get next header offset.  m will be retained.
   1372  */
   1373 int
   1374 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1375 {
   1376 	struct ip6_hdr ip6;
   1377 	struct ip6_ext ip6e;
   1378 	struct ip6_frag fh;
   1379 
   1380 	/* just in case */
   1381 	if (m == NULL)
   1382 		panic("%s: m == NULL", __func__);
   1383 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
   1384 		return -1;
   1385 
   1386 	switch (proto) {
   1387 	case IPPROTO_IPV6:
   1388 		/* do not chase beyond intermediate IPv6 headers */
   1389 		if (off != 0)
   1390 			return -1;
   1391 		if (m->m_pkthdr.len < off + sizeof(ip6))
   1392 			return -1;
   1393 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
   1394 		if (nxtp)
   1395 			*nxtp = ip6.ip6_nxt;
   1396 		off += sizeof(ip6);
   1397 		return off;
   1398 
   1399 	case IPPROTO_FRAGMENT:
   1400 		/*
   1401 		 * terminate parsing if it is not the first fragment,
   1402 		 * it does not make sense to parse through it.
   1403 		 */
   1404 		if (m->m_pkthdr.len < off + sizeof(fh))
   1405 			return -1;
   1406 		m_copydata(m, off, sizeof(fh), (void *)&fh);
   1407 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
   1408 			return -1;
   1409 		if (nxtp)
   1410 			*nxtp = fh.ip6f_nxt;
   1411 		off += sizeof(struct ip6_frag);
   1412 		return off;
   1413 
   1414 	case IPPROTO_AH:
   1415 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1416 			return -1;
   1417 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1418 		if (nxtp)
   1419 			*nxtp = ip6e.ip6e_nxt;
   1420 		off += (ip6e.ip6e_len + 2) << 2;
   1421 		if (m->m_pkthdr.len < off)
   1422 			return -1;
   1423 		return off;
   1424 
   1425 	case IPPROTO_HOPOPTS:
   1426 	case IPPROTO_ROUTING:
   1427 	case IPPROTO_DSTOPTS:
   1428 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1429 			return -1;
   1430 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1431 		if (nxtp)
   1432 			*nxtp = ip6e.ip6e_nxt;
   1433 		off += (ip6e.ip6e_len + 1) << 3;
   1434 		if (m->m_pkthdr.len < off)
   1435 			return -1;
   1436 		return off;
   1437 
   1438 	case IPPROTO_NONE:
   1439 	case IPPROTO_ESP:
   1440 	case IPPROTO_IPCOMP:
   1441 		/* give up */
   1442 		return -1;
   1443 
   1444 	default:
   1445 		return -1;
   1446 	}
   1447 }
   1448 
   1449 /*
   1450  * get offset for the last header in the chain.  m will be kept untainted.
   1451  */
   1452 int
   1453 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1454 {
   1455 	int newoff;
   1456 	int nxt;
   1457 
   1458 	if (!nxtp) {
   1459 		nxt = -1;
   1460 		nxtp = &nxt;
   1461 	}
   1462 	for (;;) {
   1463 		newoff = ip6_nexthdr(m, off, proto, nxtp);
   1464 		if (newoff < 0)
   1465 			return off;
   1466 		else if (newoff < off)
   1467 			return -1;	/* invalid */
   1468 		else if (newoff == off)
   1469 			return newoff;
   1470 
   1471 		off = newoff;
   1472 		proto = *nxtp;
   1473 	}
   1474 }
   1475 
   1476 struct m_tag *
   1477 ip6_addaux(struct mbuf *m)
   1478 {
   1479 	struct m_tag *mtag;
   1480 
   1481 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1482 	if (!mtag) {
   1483 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
   1484 		    M_NOWAIT);
   1485 		if (mtag) {
   1486 			m_tag_prepend(m, mtag);
   1487 			memset(mtag + 1, 0, sizeof(struct ip6aux));
   1488 		}
   1489 	}
   1490 	return mtag;
   1491 }
   1492 
   1493 struct m_tag *
   1494 ip6_findaux(struct mbuf *m)
   1495 {
   1496 	struct m_tag *mtag;
   1497 
   1498 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1499 	return mtag;
   1500 }
   1501 
   1502 void
   1503 ip6_delaux(struct mbuf *m)
   1504 {
   1505 	struct m_tag *mtag;
   1506 
   1507 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1508 	if (mtag)
   1509 		m_tag_delete(m, mtag);
   1510 }
   1511 
   1512 /*
   1513  * System control for IP6
   1514  */
   1515 
   1516 const u_char inet6ctlerrmap[PRC_NCMDS] = {
   1517 	0,		0,		0,		0,
   1518 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
   1519 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
   1520 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
   1521 	0,		0,		0,		0,
   1522 	ENOPROTOOPT
   1523 };
   1524 
   1525 extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS);
   1526 
   1527 static int
   1528 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
   1529 {
   1530 
   1531 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
   1532 }
   1533 
   1534 static void
   1535 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
   1536 {
   1537 
   1538 	sysctl_createv(clog, 0, NULL, NULL,
   1539 		       CTLFLAG_PERMANENT,
   1540 		       CTLTYPE_NODE, "inet6",
   1541 		       SYSCTL_DESCR("PF_INET6 related settings"),
   1542 		       NULL, 0, NULL, 0,
   1543 		       CTL_NET, PF_INET6, CTL_EOL);
   1544 	sysctl_createv(clog, 0, NULL, NULL,
   1545 		       CTLFLAG_PERMANENT,
   1546 		       CTLTYPE_NODE, "ip6",
   1547 		       SYSCTL_DESCR("IPv6 related settings"),
   1548 		       NULL, 0, NULL, 0,
   1549 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
   1550 
   1551 	sysctl_createv(clog, 0, NULL, NULL,
   1552 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1553 		       CTLTYPE_INT, "forwarding",
   1554 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
   1555 		       NULL, 0, &ip6_forwarding, 0,
   1556 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1557 		       IPV6CTL_FORWARDING, CTL_EOL);
   1558 	sysctl_createv(clog, 0, NULL, NULL,
   1559 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1560 		       CTLTYPE_INT, "redirect",
   1561 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
   1562 		       NULL, 0, &ip6_sendredirects, 0,
   1563 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1564 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
   1565 	sysctl_createv(clog, 0, NULL, NULL,
   1566 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1567 		       CTLTYPE_INT, "hlim",
   1568 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
   1569 		       NULL, 0, &ip6_defhlim, 0,
   1570 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1571 		       IPV6CTL_DEFHLIM, CTL_EOL);
   1572 	sysctl_createv(clog, 0, NULL, NULL,
   1573 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1574 		       CTLTYPE_INT, "maxfragpackets",
   1575 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
   1576 				    "for reassembly"),
   1577 		       NULL, 0, &ip6_maxfragpackets, 0,
   1578 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1579 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
   1580 	sysctl_createv(clog, 0, NULL, NULL,
   1581 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1582 		       CTLTYPE_INT, "accept_rtadv",
   1583 		       SYSCTL_DESCR("Accept router advertisements"),
   1584 		       NULL, 0, &ip6_accept_rtadv, 0,
   1585 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1586 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
   1587 	sysctl_createv(clog, 0, NULL, NULL,
   1588 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1589 		       CTLTYPE_INT, "rtadv_maxroutes",
   1590 		       SYSCTL_DESCR("Maximum number of routes accepted via router advertisements"),
   1591 		       NULL, 0, &ip6_rtadv_maxroutes, 0,
   1592 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1593 		       IPV6CTL_RTADV_MAXROUTES, CTL_EOL);
   1594 	sysctl_createv(clog, 0, NULL, NULL,
   1595 		       CTLFLAG_PERMANENT,
   1596 		       CTLTYPE_INT, "rtadv_numroutes",
   1597 		       SYSCTL_DESCR("Current number of routes accepted via router advertisements"),
   1598 		       NULL, 0, &nd6_numroutes, 0,
   1599 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1600 		       IPV6CTL_RTADV_NUMROUTES, CTL_EOL);
   1601 	sysctl_createv(clog, 0, NULL, NULL,
   1602 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1603 		       CTLTYPE_INT, "keepfaith",
   1604 		       SYSCTL_DESCR("Activate faith interface"),
   1605 		       NULL, 0, &ip6_keepfaith, 0,
   1606 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1607 		       IPV6CTL_KEEPFAITH, CTL_EOL);
   1608 	sysctl_createv(clog, 0, NULL, NULL,
   1609 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1610 		       CTLTYPE_INT, "log_interval",
   1611 		       SYSCTL_DESCR("Minimum interval between logging "
   1612 				    "unroutable packets"),
   1613 		       NULL, 0, &ip6_log_interval, 0,
   1614 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1615 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
   1616 	sysctl_createv(clog, 0, NULL, NULL,
   1617 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1618 		       CTLTYPE_INT, "hdrnestlimit",
   1619 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
   1620 		       NULL, 0, &ip6_hdrnestlimit, 0,
   1621 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1622 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
   1623 	sysctl_createv(clog, 0, NULL, NULL,
   1624 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1625 		       CTLTYPE_INT, "dad_count",
   1626 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
   1627 				    "probes to send"),
   1628 		       NULL, 0, &ip6_dad_count, 0,
   1629 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1630 		       IPV6CTL_DAD_COUNT, CTL_EOL);
   1631 	sysctl_createv(clog, 0, NULL, NULL,
   1632 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1633 		       CTLTYPE_INT, "auto_flowlabel",
   1634 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
   1635 		       NULL, 0, &ip6_auto_flowlabel, 0,
   1636 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1637 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
   1638 	sysctl_createv(clog, 0, NULL, NULL,
   1639 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1640 		       CTLTYPE_INT, "defmcasthlim",
   1641 		       SYSCTL_DESCR("Default multicast hop limit"),
   1642 		       NULL, 0, &ip6_defmcasthlim, 0,
   1643 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1644 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
   1645 	sysctl_createv(clog, 0, NULL, NULL,
   1646 		       CTLFLAG_PERMANENT,
   1647 		       CTLTYPE_STRING, "kame_version",
   1648 		       SYSCTL_DESCR("KAME Version"),
   1649 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
   1650 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1651 		       IPV6CTL_KAME_VERSION, CTL_EOL);
   1652 	sysctl_createv(clog, 0, NULL, NULL,
   1653 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1654 		       CTLTYPE_INT, "use_deprecated",
   1655 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
   1656 				    "source addresses"),
   1657 		       NULL, 0, &ip6_use_deprecated, 0,
   1658 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1659 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
   1660 	sysctl_createv(clog, 0, NULL, NULL,
   1661 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1662 		       CTLTYPE_INT, "rr_prune", NULL,
   1663 		       NULL, 0, &ip6_rr_prune, 0,
   1664 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1665 		       IPV6CTL_RR_PRUNE, CTL_EOL);
   1666 	sysctl_createv(clog, 0, NULL, NULL,
   1667 		       CTLFLAG_PERMANENT
   1668 #ifndef INET6_BINDV6ONLY
   1669 		       |CTLFLAG_READWRITE,
   1670 #endif
   1671 		       CTLTYPE_INT, "v6only",
   1672 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
   1673 				    "to PF_INET sockets"),
   1674 		       NULL, 0, &ip6_v6only, 0,
   1675 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1676 		       IPV6CTL_V6ONLY, CTL_EOL);
   1677 	sysctl_createv(clog, 0, NULL, NULL,
   1678 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1679 		       CTLTYPE_INT, "anonportmin",
   1680 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
   1681 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
   1682 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1683 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
   1684 	sysctl_createv(clog, 0, NULL, NULL,
   1685 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1686 		       CTLTYPE_INT, "anonportmax",
   1687 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
   1688 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
   1689 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1690 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
   1691 #ifndef IPNOPRIVPORTS
   1692 	sysctl_createv(clog, 0, NULL, NULL,
   1693 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1694 		       CTLTYPE_INT, "lowportmin",
   1695 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
   1696 				    "to assign"),
   1697 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
   1698 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1699 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
   1700 	sysctl_createv(clog, 0, NULL, NULL,
   1701 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1702 		       CTLTYPE_INT, "lowportmax",
   1703 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
   1704 				    "to assign"),
   1705 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
   1706 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1707 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
   1708 #endif /* IPNOPRIVPORTS */
   1709 	sysctl_createv(clog, 0, NULL, NULL,
   1710 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1711 		       CTLTYPE_INT, "auto_linklocal",
   1712 		       SYSCTL_DESCR("Default value of per-interface flag for "
   1713 		                    "adding an IPv6 link-local address to "
   1714 				    "interfaces when attached"),
   1715 		       NULL, 0, &ip6_auto_linklocal, 0,
   1716 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1717 		       IPV6CTL_AUTO_LINKLOCAL, CTL_EOL);
   1718 	sysctl_createv(clog, 0, NULL, NULL,
   1719 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
   1720 		       CTLTYPE_STRUCT, "addctlpolicy",
   1721 		       SYSCTL_DESCR("Return the current address control"
   1722 			   " policy"),
   1723 		       sysctl_net_inet6_addrctlpolicy, 0, NULL, 0,
   1724 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1725 		       IPV6CTL_ADDRCTLPOLICY, CTL_EOL);
   1726 	sysctl_createv(clog, 0, NULL, NULL,
   1727 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1728 		       CTLTYPE_INT, "use_tempaddr",
   1729 		       SYSCTL_DESCR("Use temporary address"),
   1730 		       NULL, 0, &ip6_use_tempaddr, 0,
   1731 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1732 		       CTL_CREATE, CTL_EOL);
   1733 	sysctl_createv(clog, 0, NULL, NULL,
   1734 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1735 		       CTLTYPE_INT, "prefer_tempaddr",
   1736 		       SYSCTL_DESCR("Prefer temporary address as source "
   1737 		                    "address"),
   1738 		       NULL, 0, &ip6_prefer_tempaddr, 0,
   1739 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1740 		       CTL_CREATE, CTL_EOL);
   1741 	sysctl_createv(clog, 0, NULL, NULL,
   1742 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1743 		       CTLTYPE_INT, "temppltime",
   1744 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
   1745 		       NULL, 0, &ip6_temp_preferred_lifetime, 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, "tempvltime",
   1751 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
   1752 		       NULL, 0, &ip6_temp_valid_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, "maxfrags",
   1758 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
   1759 		       NULL, 0, &ip6_maxfrags, 0,
   1760 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1761 		       IPV6CTL_MAXFRAGS, CTL_EOL);
   1762 	sysctl_createv(clog, 0, NULL, NULL,
   1763 		       CTLFLAG_PERMANENT,
   1764 		       CTLTYPE_STRUCT, "stats",
   1765 		       SYSCTL_DESCR("IPv6 statistics"),
   1766 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
   1767 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1768 		       IPV6CTL_STATS, CTL_EOL);
   1769 	sysctl_createv(clog, 0, NULL, NULL,
   1770 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1771 		       CTLTYPE_INT, "use_defaultzone",
   1772 		       SYSCTL_DESCR("Whether to use the default scope zones"),
   1773 		       NULL, 0, &ip6_use_defzone, 0,
   1774 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1775 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
   1776 	sysctl_createv(clog, 0, NULL, NULL,
   1777 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1778 		       CTLTYPE_INT, "mcast_pmtu",
   1779 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
   1780 		       NULL, 0, &ip6_mcast_pmtu, 0,
   1781 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1782 		       CTL_CREATE, CTL_EOL);
   1783 	/* anonportalgo RFC6056 subtree */
   1784 	const struct sysctlnode *portalgo_node;
   1785 	sysctl_createv(clog, 0, NULL, &portalgo_node,
   1786 		       CTLFLAG_PERMANENT,
   1787 		       CTLTYPE_NODE, "anonportalgo",
   1788 		       SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"),
   1789 	    	       NULL, 0, NULL, 0,
   1790 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL);
   1791 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1792 		       CTLFLAG_PERMANENT,
   1793 		       CTLTYPE_STRING, "available",
   1794 		       SYSCTL_DESCR("available algorithms"),
   1795 		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
   1796 		       CTL_CREATE, CTL_EOL);
   1797 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1798 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1799 		       CTLTYPE_STRING, "selected",
   1800 		       SYSCTL_DESCR("selected algorithm"),
   1801 	               sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN,
   1802 		       CTL_CREATE, CTL_EOL);
   1803 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1804 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1805 		       CTLTYPE_STRUCT, "reserve",
   1806 		       SYSCTL_DESCR("bitmap of reserved ports"),
   1807 		       sysctl_portalgo_reserve6, 0, NULL, 0,
   1808 		       CTL_CREATE, CTL_EOL);
   1809 	sysctl_createv(clog, 0, NULL, NULL,
   1810 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1811 		       CTLTYPE_INT, "neighborgcthresh",
   1812 		       SYSCTL_DESCR("Maximum number of entries in neighbor"
   1813 			" cache"),
   1814 		       NULL, 1, &ip6_neighborgcthresh, 0,
   1815 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1816 		       CTL_CREATE, CTL_EOL);
   1817 	sysctl_createv(clog, 0, NULL, NULL,
   1818 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1819 		       CTLTYPE_INT, "maxifprefixes",
   1820 		       SYSCTL_DESCR("Maximum number of prefixes created by"
   1821 			   " route advertisement per interface"),
   1822 		       NULL, 1, &ip6_maxifprefixes, 0,
   1823 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1824 		       CTL_CREATE, CTL_EOL);
   1825 	sysctl_createv(clog, 0, NULL, NULL,
   1826 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1827 		       CTLTYPE_INT, "maxifdefrouters",
   1828 		       SYSCTL_DESCR("Maximum number of default routers created"
   1829 			   " by route advertisement per interface"),
   1830 		       NULL, 1, &ip6_maxifdefrouters, 0,
   1831 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1832 		       CTL_CREATE, CTL_EOL);
   1833 	sysctl_createv(clog, 0, NULL, NULL,
   1834 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1835 		       CTLTYPE_INT, "maxdynroutes",
   1836 		       SYSCTL_DESCR("Maximum number of routes created via"
   1837 			   " redirect"),
   1838 		       NULL, 1, &ip6_maxdynroutes, 0,
   1839 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1840 		       CTL_CREATE, CTL_EOL);
   1841 }
   1842 
   1843 void
   1844 ip6_statinc(u_int stat)
   1845 {
   1846 
   1847 	KASSERT(stat < IP6_NSTATS);
   1848 	IP6_STATINC(stat);
   1849 }
   1850