Home | History | Annotate | Line # | Download | only in netinet6
ip6_input.c revision 1.97
      1 /*	$NetBSD: ip6_input.c,v 1.97 2007/03/04 06:03:26 christos 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 1.97 2007/03/04 06:03:26 christos Exp $");
     66 
     67 #include "opt_inet.h"
     68 #include "opt_inet6.h"
     69 #include "opt_ipsec.h"
     70 #include "opt_pfil_hooks.h"
     71 
     72 #include <sys/param.h>
     73 #include <sys/systm.h>
     74 #include <sys/malloc.h>
     75 #include <sys/mbuf.h>
     76 #include <sys/domain.h>
     77 #include <sys/protosw.h>
     78 #include <sys/socket.h>
     79 #include <sys/socketvar.h>
     80 #include <sys/errno.h>
     81 #include <sys/time.h>
     82 #include <sys/kernel.h>
     83 #include <sys/syslog.h>
     84 #include <sys/proc.h>
     85 #include <sys/sysctl.h>
     86 
     87 #include <net/if.h>
     88 #include <net/if_types.h>
     89 #include <net/if_dl.h>
     90 #include <net/route.h>
     91 #include <net/netisr.h>
     92 #ifdef PFIL_HOOKS
     93 #include <net/pfil.h>
     94 #endif
     95 
     96 #include <netinet/in.h>
     97 #include <netinet/in_systm.h>
     98 #ifdef INET
     99 #include <netinet/ip.h>
    100 #include <netinet/ip_icmp.h>
    101 #endif /* INET */
    102 #include <netinet/ip6.h>
    103 #include <netinet6/in6_var.h>
    104 #include <netinet6/ip6_var.h>
    105 #include <netinet6/in6_pcb.h>
    106 #include <netinet/icmp6.h>
    107 #include <netinet6/scope6_var.h>
    108 #include <netinet6/in6_ifattach.h>
    109 #include <netinet6/nd6.h>
    110 
    111 #ifdef IPSEC
    112 #include <netinet6/ipsec.h>
    113 #endif
    114 
    115 #ifdef FAST_IPSEC
    116 #include <netipsec/ipsec.h>
    117 #include <netipsec/ipsec6.h>
    118 #include <netipsec/key.h>
    119 #endif /* FAST_IPSEC */
    120 
    121 #include <netinet6/ip6protosw.h>
    122 
    123 #include "faith.h"
    124 #include "gif.h"
    125 
    126 #if NGIF > 0
    127 #include <netinet6/in6_gif.h>
    128 #endif
    129 
    130 #include <net/net_osdep.h>
    131 
    132 extern struct domain inet6domain;
    133 
    134 u_char ip6_protox[IPPROTO_MAX];
    135 static int ip6qmaxlen = IFQ_MAXLEN;
    136 struct in6_ifaddr *in6_ifaddr;
    137 struct ifqueue ip6intrq;
    138 
    139 extern struct callout in6_tmpaddrtimer_ch;
    140 
    141 int ip6_forward_srcrt;			/* XXX */
    142 int ip6_sourcecheck;			/* XXX */
    143 int ip6_sourcecheck_interval;		/* XXX */
    144 
    145 #ifdef PFIL_HOOKS
    146 struct pfil_head inet6_pfil_hook;
    147 #endif
    148 
    149 struct ip6stat ip6stat;
    150 
    151 static void ip6_init2 __P((void *));
    152 static struct m_tag *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *));
    153 
    154 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
    155 static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int));
    156 
    157 /*
    158  * IP6 initialization: fill in IP6 protocol switch table.
    159  * All protocols not implemented in kernel go to raw IP6 protocol handler.
    160  */
    161 void
    162 ip6_init()
    163 {
    164 	const struct ip6protosw *pr;
    165 	int i;
    166 
    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 	ip6intrq.ifq_maxlen = ip6qmaxlen;
    178 	scope6_init();
    179 	addrsel_policy_init();
    180 	nd6_init();
    181 	frag6_init();
    182 	ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
    183 
    184 	ip6_init2((void *)0);
    185 
    186 #ifdef PFIL_HOOKS
    187 	/* Register our Packet Filter hook. */
    188 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
    189 	inet6_pfil_hook.ph_af   = AF_INET6;
    190 	i = pfil_head_register(&inet6_pfil_hook);
    191 	if (i != 0)
    192 		printf("ip6_init: WARNING: unable to register pfil hook, "
    193 		    "error %d\n", i);
    194 #endif /* PFIL_HOOKS */
    195 }
    196 
    197 static void
    198 ip6_init2(void *dummy)
    199 {
    200 
    201 	/* nd6_timer_init */
    202 	callout_init(&nd6_timer_ch);
    203 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
    204 
    205 	/* timer for regeneranation of temporary addresses randomize ID */
    206 	callout_init(&in6_tmpaddrtimer_ch);
    207 	callout_reset(&in6_tmpaddrtimer_ch,
    208 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
    209 		       ip6_temp_regen_advance) * hz,
    210 		      in6_tmpaddrtimer, NULL);
    211 }
    212 
    213 /*
    214  * IP6 input interrupt handling. Just pass the packet to ip6_input.
    215  */
    216 void
    217 ip6intr()
    218 {
    219 	int s;
    220 	struct mbuf *m;
    221 
    222 	for (;;) {
    223 		s = splnet();
    224 		IF_DEQUEUE(&ip6intrq, m);
    225 		splx(s);
    226 		if (m == 0)
    227 			return;
    228 		/* drop the packet if IPv6 operation is disabled on the IF */
    229 		if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
    230 			m_freem(m);
    231 			return;
    232 		}
    233 		ip6_input(m);
    234 	}
    235 }
    236 
    237 extern struct	route_in6 ip6_forward_rt;
    238 
    239 void
    240 ip6_input(struct mbuf *m)
    241 {
    242 	struct ip6_hdr *ip6;
    243 	int off = sizeof(struct ip6_hdr), nest;
    244 	u_int32_t plen;
    245 	u_int32_t rtalert = ~0;
    246 	int nxt, ours = 0;
    247 	struct ifnet *deliverifp = NULL;
    248 	int srcrt = 0;
    249 #ifdef FAST_IPSEC
    250 	struct m_tag *mtag;
    251 	struct tdb_ident *tdbi;
    252 	struct secpolicy *sp;
    253 	int s, error;
    254 #endif
    255 
    256 #ifdef IPSEC
    257 	/*
    258 	 * should the inner packet be considered authentic?
    259 	 * see comment in ah4_input().
    260 	 */
    261 	m->m_flags &= ~M_AUTHIPHDR;
    262 	m->m_flags &= ~M_AUTHIPDGM;
    263 #endif
    264 
    265 	/*
    266 	 * make sure we don't have onion peering information into m_tag.
    267 	 */
    268 	ip6_delaux(m);
    269 
    270 	/*
    271 	 * mbuf statistics
    272 	 */
    273 	if (m->m_flags & M_EXT) {
    274 		if (m->m_next)
    275 			ip6stat.ip6s_mext2m++;
    276 		else
    277 			ip6stat.ip6s_mext1++;
    278 	} else {
    279 #define M2MMAX	__arraycount(ip6stat.ip6s_m2m)
    280 		if (m->m_next) {
    281 			if (m->m_flags & M_LOOP) {
    282 				ip6stat.ip6s_m2m[lo0ifp->if_index]++; /* XXX */
    283 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
    284 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
    285 			else
    286 				ip6stat.ip6s_m2m[0]++;
    287 		} else
    288 			ip6stat.ip6s_m1++;
    289 #undef M2MMAX
    290 	}
    291 
    292 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
    293 	ip6stat.ip6s_total++;
    294 
    295 	/*
    296 	 * If the IPv6 header is not aligned, slurp it up into a new
    297 	 * mbuf with space for link headers, in the event we forward
    298 	 * it.  Otherwise, if it is aligned, make sure the entire base
    299 	 * IPv6 header is in the first mbuf of the chain.
    300 	 */
    301 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
    302 		struct ifnet *inifp = m->m_pkthdr.rcvif;
    303 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
    304 				  (max_linkhdr + 3) & ~3)) == NULL) {
    305 			/* XXXJRT new stat, please */
    306 			ip6stat.ip6s_toosmall++;
    307 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    308 			return;
    309 		}
    310 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
    311 		struct ifnet *inifp = m->m_pkthdr.rcvif;
    312 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    313 			ip6stat.ip6s_toosmall++;
    314 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    315 			return;
    316 		}
    317 	}
    318 
    319 	ip6 = mtod(m, struct ip6_hdr *);
    320 
    321 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    322 		ip6stat.ip6s_badvers++;
    323 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    324 		goto bad;
    325 	}
    326 
    327 #ifdef PFIL_HOOKS
    328 	/*
    329 	 * Run through list of hooks for input packets.  If there are any
    330 	 * filters which require that additional packets in the flow are
    331 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
    332 	 * Note that filters must _never_ set this flag, as another filter
    333 	 * in the list may have previously cleared it.
    334 	 */
    335 	/*
    336 	 * let ipfilter look at packet on the wire,
    337 	 * not the decapsulated packet.
    338 	 */
    339 #ifdef IPSEC
    340 	if (!ipsec_getnhist(m))
    341 #elif defined(FAST_IPSEC)
    342 	if (!ipsec_indone(m))
    343 #else
    344 	if (1)
    345 #endif
    346 	{
    347 		struct in6_addr odst;
    348 
    349 		odst = ip6->ip6_dst;
    350 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
    351 				   PFIL_IN) != 0)
    352 			return;
    353 		if (m == NULL)
    354 			return;
    355 		ip6 = mtod(m, struct ip6_hdr *);
    356 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
    357 	}
    358 #endif /* PFIL_HOOKS */
    359 
    360 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
    361 
    362 #ifdef ALTQ
    363 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
    364 		/* packet is dropped by traffic conditioner */
    365 		return;
    366 	}
    367 #endif
    368 
    369 	/*
    370 	 * Check against address spoofing/corruption.
    371 	 */
    372 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
    373 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
    374 		/*
    375 		 * XXX: "badscope" is not very suitable for a multicast source.
    376 		 */
    377 		ip6stat.ip6s_badscope++;
    378 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    379 		goto bad;
    380 	}
    381 	/*
    382 	 * The following check is not documented in specs.  A malicious
    383 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
    384 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
    385 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
    386 	 *
    387 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
    388 	 * support IPv4-less kernel compilation, we cannot support SIIT
    389 	 * environment at all.  So, it makes more sense for us to reject any
    390 	 * malicious packets for non-SIIT environment, than try to do a
    391 	 * partial support for SIIT environment.
    392 	 */
    393 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    394 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    395 		ip6stat.ip6s_badscope++;
    396 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    397 		goto bad;
    398 	}
    399 #if 0
    400 	/*
    401 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
    402 	 *
    403 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
    404 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
    405 	 * is revised to forbid relaying case.
    406 	 */
    407 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
    408 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
    409 		ip6stat.ip6s_badscope++;
    410 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    411 		goto bad;
    412 	}
    413 #endif
    414 
    415 	/*
    416 	 * Disambiguate address scope zones (if there is ambiguity).
    417 	 * We first make sure that the original source or destination address
    418 	 * is not in our internal form for scoped addresses.  Such addresses
    419 	 * are not necessarily invalid spec-wise, but we cannot accept them due
    420 	 * to the usage conflict.
    421 	 * in6_setscope() then also checks and rejects the cases where src or
    422 	 * dst are the loopback address and the receiving interface
    423 	 * is not loopback.
    424 	 */
    425 	if (__predict_false(
    426 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
    427 		goto bad;
    428 	ip6 = mtod(m, struct ip6_hdr *);
    429 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
    430 		ip6stat.ip6s_badscope++; /* XXX */
    431 		goto bad;
    432 	}
    433 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
    434 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
    435 		ip6stat.ip6s_badscope++;
    436 		goto bad;
    437 	}
    438 
    439 	/*
    440 	 * Multicast check
    441 	 */
    442 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    443 	  	struct	in6_multi *in6m = 0;
    444 
    445 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
    446 		/*
    447 		 * See if we belong to the destination multicast group on the
    448 		 * arrival interface.
    449 		 */
    450 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
    451 		if (in6m)
    452 			ours = 1;
    453 		else if (!ip6_mrouter) {
    454 			ip6stat.ip6s_notmember++;
    455 			ip6stat.ip6s_cantforward++;
    456 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    457 			goto bad;
    458 		}
    459 		deliverifp = m->m_pkthdr.rcvif;
    460 		goto hbhcheck;
    461 	}
    462 
    463 	/*
    464 	 *  Unicast check
    465 	 */
    466 	if (!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
    467 	    &((const struct sockaddr_in6 *)rtcache_getdst((const struct route *)&ip6_forward_rt))->sin6_addr))
    468 		rtcache_free((struct route *)&ip6_forward_rt);
    469 	else
    470 		rtcache_check((struct route *)&ip6_forward_rt);
    471 	if (ip6_forward_rt.ro_rt != NULL) {
    472 		/* XXX Revalidated route is accounted wrongly. */
    473 		ip6stat.ip6s_forward_cachehit++;
    474 	} else {
    475 		struct sockaddr_in6 *dst6;
    476 
    477 		ip6stat.ip6s_forward_cachemiss++;
    478 
    479 		dst6 = &ip6_forward_rt.ro_dst;
    480 		memset(dst6, 0, sizeof(*dst6));
    481 		dst6->sin6_len = sizeof(struct sockaddr_in6);
    482 		dst6->sin6_family = AF_INET6;
    483 		dst6->sin6_addr = ip6->ip6_dst;
    484 
    485 		rtcache_init((struct route *)&ip6_forward_rt);
    486 	}
    487 
    488 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
    489 
    490 	/*
    491 	 * Accept the packet if the forwarding interface to the destination
    492 	 * according to the routing table is the loopback interface,
    493 	 * unless the associated route has a gateway.
    494 	 * Note that this approach causes to accept a packet if there is a
    495 	 * route to the loopback interface for the destination of the packet.
    496 	 * But we think it's even useful in some situations, e.g. when using
    497 	 * a special daemon which wants to intercept the packet.
    498 	 */
    499 	if (ip6_forward_rt.ro_rt != NULL &&
    500 	    (ip6_forward_rt.ro_rt->rt_flags &
    501 	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
    502 	    !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
    503 #if 0
    504 	    /*
    505 	     * The check below is redundant since the comparison of
    506 	     * the destination and the key of the rtentry has
    507 	     * already done through looking up the routing table.
    508 	     */
    509 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
    510 	    &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
    511 #endif
    512 	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
    513 		struct in6_ifaddr *ia6 =
    514 			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
    515 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
    516 			m->m_flags |= M_ANYCAST6;
    517 		/*
    518 		 * packets to a tentative, duplicated, or somehow invalid
    519 		 * address must not be accepted.
    520 		 */
    521 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
    522 			/* this address is ready */
    523 			ours = 1;
    524 			deliverifp = ia6->ia_ifp;	/* correct? */
    525 			goto hbhcheck;
    526 		} else {
    527 			/* address is not ready, so discard the packet. */
    528 			nd6log((LOG_INFO,
    529 			    "ip6_input: packet to an unready address %s->%s\n",
    530 			    ip6_sprintf(&ip6->ip6_src),
    531 			    ip6_sprintf(&ip6->ip6_dst)));
    532 
    533 			goto bad;
    534 		}
    535 	}
    536 
    537 	/*
    538 	 * FAITH (Firewall Aided Internet Translator)
    539 	 */
    540 #if defined(NFAITH) && 0 < NFAITH
    541 	if (ip6_keepfaith) {
    542 		if (ip6_forward_rt.ro_rt != NULL &&
    543 		    ip6_forward_rt.ro_rt->rt_ifp != NULL &&
    544 		    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
    545 			/* XXX do we need more sanity checks? */
    546 			ours = 1;
    547 			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
    548 			goto hbhcheck;
    549 		}
    550 	}
    551 #endif
    552 
    553 #if 0
    554     {
    555 	/*
    556 	 * Last resort: check in6_ifaddr for incoming interface.
    557 	 * The code is here until I update the "goto ours hack" code above
    558 	 * working right.
    559 	 */
    560 	struct ifaddr *ifa;
    561 	TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) {
    562 		if (ifa->ifa_addr == NULL)
    563 			continue;	/* just for safety */
    564 		if (ifa->ifa_addr->sa_family != AF_INET6)
    565 			continue;
    566 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
    567 			ours = 1;
    568 			deliverifp = ifa->ifa_ifp;
    569 			goto hbhcheck;
    570 		}
    571 	}
    572     }
    573 #endif
    574 
    575 	/*
    576 	 * Now there is no reason to process the packet if it's not our own
    577 	 * and we're not a router.
    578 	 */
    579 	if (!ip6_forwarding) {
    580 		ip6stat.ip6s_cantforward++;
    581 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    582 		goto bad;
    583 	}
    584 
    585   hbhcheck:
    586 	/*
    587 	 * record address information into m_tag, if we don't have one yet.
    588 	 * note that we are unable to record it, if the address is not listed
    589 	 * as our interface address (e.g. multicast addresses, addresses
    590 	 * within FAITH prefixes and such).
    591 	 */
    592 	if (deliverifp && !ip6_getdstifaddr(m)) {
    593 		struct in6_ifaddr *ia6;
    594 
    595 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    596 		if (ia6) {
    597 			if (!ip6_setdstifaddr(m, ia6)) {
    598 				/*
    599 				 * XXX maybe we should drop the packet here,
    600 				 * as we could not provide enough information
    601 				 * to the upper layers.
    602 				 */
    603 			}
    604 		}
    605 	}
    606 
    607 	/*
    608 	 * Process Hop-by-Hop options header if it's contained.
    609 	 * m may be modified in ip6_hopopts_input().
    610 	 * If a JumboPayload option is included, plen will also be modified.
    611 	 */
    612 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
    613 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
    614 		struct ip6_hbh *hbh;
    615 
    616 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
    617 #if 0	/*touches NULL pointer*/
    618 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    619 #endif
    620 			return;	/* m have already been freed */
    621 		}
    622 
    623 		/* adjust pointer */
    624 		ip6 = mtod(m, struct ip6_hdr *);
    625 
    626 		/*
    627 		 * if the payload length field is 0 and the next header field
    628 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
    629 		 * option MUST be included.
    630 		 */
    631 		if (ip6->ip6_plen == 0 && plen == 0) {
    632 			/*
    633 			 * Note that if a valid jumbo payload option is
    634 			 * contained, ip6_hopopts_input() must set a valid
    635 			 * (non-zero) payload length to the variable plen.
    636 			 */
    637 			ip6stat.ip6s_badoptions++;
    638 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    639 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    640 			icmp6_error(m, ICMP6_PARAM_PROB,
    641 				    ICMP6_PARAMPROB_HEADER,
    642 				    (char *)&ip6->ip6_plen - (char *)ip6);
    643 			return;
    644 		}
    645 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    646 			sizeof(struct ip6_hbh));
    647 		if (hbh == NULL) {
    648 			ip6stat.ip6s_tooshort++;
    649 			return;
    650 		}
    651 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
    652 		nxt = hbh->ip6h_nxt;
    653 
    654 		/*
    655 		 * accept the packet if a router alert option is included
    656 		 * and we act as an IPv6 router.
    657 		 */
    658 		if (rtalert != ~0 && ip6_forwarding)
    659 			ours = 1;
    660 	} else
    661 		nxt = ip6->ip6_nxt;
    662 
    663 	/*
    664 	 * Check that the amount of data in the buffers
    665 	 * is as at least much as the IPv6 header would have us expect.
    666 	 * Trim mbufs if longer than we expect.
    667 	 * Drop packet if shorter than we expect.
    668 	 */
    669 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
    670 		ip6stat.ip6s_tooshort++;
    671 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    672 		goto bad;
    673 	}
    674 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
    675 		if (m->m_len == m->m_pkthdr.len) {
    676 			m->m_len = sizeof(struct ip6_hdr) + plen;
    677 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
    678 		} else
    679 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
    680 	}
    681 
    682 	/*
    683 	 * Forward if desirable.
    684 	 */
    685 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    686 		/*
    687 		 * If we are acting as a multicast router, all
    688 		 * incoming multicast packets are passed to the
    689 		 * kernel-level multicast forwarding function.
    690 		 * The packet is returned (relatively) intact; if
    691 		 * ip6_mforward() returns a non-zero value, the packet
    692 		 * must be discarded, else it may be accepted below.
    693 		 */
    694 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
    695 			ip6stat.ip6s_cantforward++;
    696 			m_freem(m);
    697 			return;
    698 		}
    699 		if (!ours) {
    700 			m_freem(m);
    701 			return;
    702 		}
    703 	} else if (!ours) {
    704 		ip6_forward(m, srcrt);
    705 		return;
    706 	}
    707 
    708 	ip6 = mtod(m, struct ip6_hdr *);
    709 
    710 	/*
    711 	 * Malicious party may be able to use IPv4 mapped addr to confuse
    712 	 * tcp/udp stack and bypass security checks (act as if it was from
    713 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
    714 	 *
    715 	 * For SIIT end node behavior, you may want to disable the check.
    716 	 * However, you will  become vulnerable to attacks using IPv4 mapped
    717 	 * source.
    718 	 */
    719 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    720 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    721 		ip6stat.ip6s_badscope++;
    722 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    723 		goto bad;
    724 	}
    725 
    726 	/*
    727 	 * Tell launch routine the next header
    728 	 */
    729 #ifdef IFA_STATS
    730 	if (deliverifp != NULL) {
    731 		struct in6_ifaddr *ia6;
    732 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    733 		if (ia6)
    734 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
    735 	}
    736 #endif
    737 	ip6stat.ip6s_delivered++;
    738 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
    739 	nest = 0;
    740 
    741 	while (nxt != IPPROTO_DONE) {
    742 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
    743 			ip6stat.ip6s_toomanyhdr++;
    744 			goto bad;
    745 		}
    746 
    747 		/*
    748 		 * protection against faulty packet - there should be
    749 		 * more sanity checks in header chain processing.
    750 		 */
    751 		if (m->m_pkthdr.len < off) {
    752 			ip6stat.ip6s_tooshort++;
    753 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    754 			goto bad;
    755 		}
    756 
    757 #ifdef IPSEC
    758 		/*
    759 		 * enforce IPsec policy checking if we are seeing last header.
    760 		 * note that we do not visit this with protocols with pcb layer
    761 		 * code - like udp/tcp/raw ip.
    762 		 */
    763 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
    764 		    ipsec6_in_reject(m, NULL)) {
    765 			ipsec6stat.in_polvio++;
    766 			goto bad;
    767 		}
    768 #endif
    769 #ifdef FAST_IPSEC
    770 	/*
    771 	 * enforce IPsec policy checking if we are seeing last header.
    772 	 * note that we do not visit this with protocols with pcb layer
    773 	 * code - like udp/tcp/raw ip.
    774 	 */
    775 	if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) {
    776 		/*
    777 		 * Check if the packet has already had IPsec processing
    778 		 * done.  If so, then just pass it along.  This tag gets
    779 		 * set during AH, ESP, etc. input handling, before the
    780 		 * packet is returned to the ip input queue for delivery.
    781 		 */
    782 		mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
    783 		s = splsoftnet();
    784 		if (mtag != NULL) {
    785 			tdbi = (struct tdb_ident *)(mtag + 1);
    786 			sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
    787 		} else {
    788 			sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
    789 									IP_FORWARDING, &error);
    790 		}
    791 		if (sp != NULL) {
    792 			/*
    793 			 * Check security policy against packet attributes.
    794 			 */
    795 			error = ipsec_in_reject(sp, m);
    796 			KEY_FREESP(&sp);
    797 		} else {
    798 			/* XXX error stat??? */
    799 			error = EINVAL;
    800 			DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/
    801 			goto bad;
    802 		}
    803 		splx(s);
    804 		if (error)
    805 			goto bad;
    806 	}
    807 #endif /* FAST_IPSEC */
    808 
    809 
    810 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
    811 	}
    812 	return;
    813  bad:
    814 	m_freem(m);
    815 }
    816 
    817 /*
    818  * set/grab in6_ifaddr correspond to IPv6 destination address.
    819  */
    820 static struct m_tag *
    821 ip6_setdstifaddr(m, ia6)
    822 	struct mbuf *m;
    823 	struct in6_ifaddr *ia6;
    824 {
    825 	struct m_tag *mtag;
    826 
    827 	mtag = ip6_addaux(m);
    828 	if (mtag)
    829 		((struct ip6aux *)(mtag + 1))->ip6a_dstia6 = ia6;
    830 	return mtag;	/* NULL if failed to set */
    831 }
    832 
    833 struct in6_ifaddr *
    834 ip6_getdstifaddr(m)
    835 	struct mbuf *m;
    836 {
    837 	struct m_tag *mtag;
    838 
    839 	mtag = ip6_findaux(m);
    840 	if (mtag)
    841 		return ((struct ip6aux *)(mtag + 1))->ip6a_dstia6;
    842 	else
    843 		return NULL;
    844 }
    845 
    846 /*
    847  * Hop-by-Hop options header processing. If a valid jumbo payload option is
    848  * included, the real payload length will be stored in plenp.
    849  */
    850 static int
    851 ip6_hopopts_input(plenp, rtalertp, mp, offp)
    852 	u_int32_t *plenp;
    853 	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
    854 	struct mbuf **mp;
    855 	int *offp;
    856 {
    857 	struct mbuf *m = *mp;
    858 	int off = *offp, hbhlen;
    859 	struct ip6_hbh *hbh;
    860 
    861 	/* validation of the length of the header */
    862 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
    863 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
    864 	if (hbh == NULL) {
    865 		ip6stat.ip6s_tooshort++;
    866 		return -1;
    867 	}
    868 	hbhlen = (hbh->ip6h_len + 1) << 3;
    869 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    870 		hbhlen);
    871 	if (hbh == NULL) {
    872 		ip6stat.ip6s_tooshort++;
    873 		return -1;
    874 	}
    875 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
    876 	off += hbhlen;
    877 	hbhlen -= sizeof(struct ip6_hbh);
    878 
    879 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
    880 				hbhlen, rtalertp, plenp) < 0)
    881 		return (-1);
    882 
    883 	*offp = off;
    884 	*mp = m;
    885 	return (0);
    886 }
    887 
    888 /*
    889  * Search header for all Hop-by-hop options and process each option.
    890  * This function is separate from ip6_hopopts_input() in order to
    891  * handle a case where the sending node itself process its hop-by-hop
    892  * options header. In such a case, the function is called from ip6_output().
    893  *
    894  * The function assumes that hbh header is located right after the IPv6 header
    895  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
    896  * opthead + hbhlen is located in continuous memory region.
    897  */
    898 int
    899 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
    900 	struct mbuf *m;
    901 	u_int8_t *opthead;
    902 	int hbhlen;
    903 	u_int32_t *rtalertp;
    904 	u_int32_t *plenp;
    905 {
    906 	struct ip6_hdr *ip6;
    907 	int optlen = 0;
    908 	u_int8_t *opt = opthead;
    909 	u_int16_t rtalert_val;
    910 	u_int32_t jumboplen;
    911 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
    912 
    913 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
    914 		switch (*opt) {
    915 		case IP6OPT_PAD1:
    916 			optlen = 1;
    917 			break;
    918 		case IP6OPT_PADN:
    919 			if (hbhlen < IP6OPT_MINLEN) {
    920 				ip6stat.ip6s_toosmall++;
    921 				goto bad;
    922 			}
    923 			optlen = *(opt + 1) + 2;
    924 			break;
    925 		case IP6OPT_RTALERT:
    926 			/* XXX may need check for alignment */
    927 			if (hbhlen < IP6OPT_RTALERT_LEN) {
    928 				ip6stat.ip6s_toosmall++;
    929 				goto bad;
    930 			}
    931 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
    932 				/* XXX stat */
    933 				icmp6_error(m, ICMP6_PARAM_PROB,
    934 				    ICMP6_PARAMPROB_HEADER,
    935 				    erroff + opt + 1 - opthead);
    936 				return (-1);
    937 			}
    938 			optlen = IP6OPT_RTALERT_LEN;
    939 			bcopy((void *)(opt + 2), (void *)&rtalert_val, 2);
    940 			*rtalertp = ntohs(rtalert_val);
    941 			break;
    942 		case IP6OPT_JUMBO:
    943 			/* XXX may need check for alignment */
    944 			if (hbhlen < IP6OPT_JUMBO_LEN) {
    945 				ip6stat.ip6s_toosmall++;
    946 				goto bad;
    947 			}
    948 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
    949 				/* XXX stat */
    950 				icmp6_error(m, ICMP6_PARAM_PROB,
    951 				    ICMP6_PARAMPROB_HEADER,
    952 				    erroff + opt + 1 - opthead);
    953 				return (-1);
    954 			}
    955 			optlen = IP6OPT_JUMBO_LEN;
    956 
    957 			/*
    958 			 * IPv6 packets that have non 0 payload length
    959 			 * must not contain a jumbo payload option.
    960 			 */
    961 			ip6 = mtod(m, struct ip6_hdr *);
    962 			if (ip6->ip6_plen) {
    963 				ip6stat.ip6s_badoptions++;
    964 				icmp6_error(m, ICMP6_PARAM_PROB,
    965 				    ICMP6_PARAMPROB_HEADER,
    966 				    erroff + opt - opthead);
    967 				return (-1);
    968 			}
    969 
    970 			/*
    971 			 * We may see jumbolen in unaligned location, so
    972 			 * we'd need to perform bcopy().
    973 			 */
    974 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
    975 			jumboplen = (u_int32_t)htonl(jumboplen);
    976 
    977 #if 1
    978 			/*
    979 			 * if there are multiple jumbo payload options,
    980 			 * *plenp will be non-zero and the packet will be
    981 			 * rejected.
    982 			 * the behavior may need some debate in ipngwg -
    983 			 * multiple options does not make sense, however,
    984 			 * there's no explicit mention in specification.
    985 			 */
    986 			if (*plenp != 0) {
    987 				ip6stat.ip6s_badoptions++;
    988 				icmp6_error(m, ICMP6_PARAM_PROB,
    989 				    ICMP6_PARAMPROB_HEADER,
    990 				    erroff + opt + 2 - opthead);
    991 				return (-1);
    992 			}
    993 #endif
    994 
    995 			/*
    996 			 * jumbo payload length must be larger than 65535.
    997 			 */
    998 			if (jumboplen <= IPV6_MAXPACKET) {
    999 				ip6stat.ip6s_badoptions++;
   1000 				icmp6_error(m, ICMP6_PARAM_PROB,
   1001 				    ICMP6_PARAMPROB_HEADER,
   1002 				    erroff + opt + 2 - opthead);
   1003 				return (-1);
   1004 			}
   1005 			*plenp = jumboplen;
   1006 
   1007 			break;
   1008 		default:		/* unknown option */
   1009 			if (hbhlen < IP6OPT_MINLEN) {
   1010 				ip6stat.ip6s_toosmall++;
   1011 				goto bad;
   1012 			}
   1013 			optlen = ip6_unknown_opt(opt, m,
   1014 			    erroff + opt - opthead);
   1015 			if (optlen == -1)
   1016 				return (-1);
   1017 			optlen += 2;
   1018 			break;
   1019 		}
   1020 	}
   1021 
   1022 	return (0);
   1023 
   1024   bad:
   1025 	m_freem(m);
   1026 	return (-1);
   1027 }
   1028 
   1029 /*
   1030  * Unknown option processing.
   1031  * The third argument `off' is the offset from the IPv6 header to the option,
   1032  * which is necessary if the IPv6 header the and option header and IPv6 header
   1033  * is not continuous in order to return an ICMPv6 error.
   1034  */
   1035 int
   1036 ip6_unknown_opt(optp, m, off)
   1037 	u_int8_t *optp;
   1038 	struct mbuf *m;
   1039 	int off;
   1040 {
   1041 	struct ip6_hdr *ip6;
   1042 
   1043 	switch (IP6OPT_TYPE(*optp)) {
   1044 	case IP6OPT_TYPE_SKIP: /* ignore the option */
   1045 		return ((int)*(optp + 1));
   1046 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
   1047 		m_freem(m);
   1048 		return (-1);
   1049 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
   1050 		ip6stat.ip6s_badoptions++;
   1051 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
   1052 		return (-1);
   1053 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
   1054 		ip6stat.ip6s_badoptions++;
   1055 		ip6 = mtod(m, struct ip6_hdr *);
   1056 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
   1057 		    (m->m_flags & (M_BCAST|M_MCAST)))
   1058 			m_freem(m);
   1059 		else
   1060 			icmp6_error(m, ICMP6_PARAM_PROB,
   1061 				    ICMP6_PARAMPROB_OPTION, off);
   1062 		return (-1);
   1063 	}
   1064 
   1065 	m_freem(m);		/* XXX: NOTREACHED */
   1066 	return (-1);
   1067 }
   1068 
   1069 /*
   1070  * Create the "control" list for this pcb.
   1071  *
   1072  * The routine will be called from upper layer handlers like tcp6_input().
   1073  * Thus the routine assumes that the caller (tcp6_input) have already
   1074  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
   1075  * very first mbuf on the mbuf chain.
   1076  * We may want to add some infinite loop prevention or sanity checks for safety.
   1077  * (This applies only when you are using KAME mbuf chain restriction, i.e.
   1078  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
   1079  */
   1080 void
   1081 ip6_savecontrol(in6p, mp, ip6, m)
   1082 	struct in6pcb *in6p;
   1083 	struct mbuf **mp;
   1084 	struct ip6_hdr *ip6;
   1085 	struct mbuf *m;
   1086 {
   1087 #ifdef RFC2292
   1088 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1089 #else
   1090 #define IS2292(x, y)	(y)
   1091 #endif
   1092 
   1093 #ifdef SO_TIMESTAMP
   1094 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
   1095 		struct timeval tv;
   1096 
   1097 		microtime(&tv);
   1098 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
   1099 		    SCM_TIMESTAMP, SOL_SOCKET);
   1100 		if (*mp)
   1101 			mp = &(*mp)->m_next;
   1102 	}
   1103 #endif
   1104 
   1105 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
   1106 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
   1107 		return;
   1108 
   1109 	/* RFC 2292 sec. 5 */
   1110 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
   1111 		struct in6_pktinfo pi6;
   1112 
   1113 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
   1114 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
   1115 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
   1116 		    m->m_pkthdr.rcvif->if_index : 0;
   1117 		*mp = sbcreatecontrol((void *) &pi6,
   1118 		    sizeof(struct in6_pktinfo),
   1119 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
   1120 		if (*mp)
   1121 			mp = &(*mp)->m_next;
   1122 	}
   1123 
   1124 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
   1125 		int hlim = ip6->ip6_hlim & 0xff;
   1126 
   1127 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
   1128 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
   1129 		if (*mp)
   1130 			mp = &(*mp)->m_next;
   1131 	}
   1132 
   1133 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
   1134 		u_int32_t flowinfo;
   1135 		int tclass;
   1136 
   1137 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
   1138 		flowinfo >>= 20;
   1139 
   1140 		tclass = flowinfo & 0xff;
   1141 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
   1142 		    IPV6_TCLASS, IPPROTO_IPV6);
   1143 
   1144 		if (*mp)
   1145 			mp = &(*mp)->m_next;
   1146 	}
   1147 
   1148 	/*
   1149 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
   1150 	 * privilege for the option (see ip6_ctloutput), but it might be too
   1151 	 * strict, since there might be some hop-by-hop options which can be
   1152 	 * returned to normal user.
   1153 	 * See also RFC3542 section 8 (or RFC2292 section 6).
   1154 	 */
   1155 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
   1156 		/*
   1157 		 * Check if a hop-by-hop options header is contatined in the
   1158 		 * received packet, and if so, store the options as ancillary
   1159 		 * data. Note that a hop-by-hop options header must be
   1160 		 * just after the IPv6 header, which fact is assured through
   1161 		 * the IPv6 input processing.
   1162 		 */
   1163 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1164 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
   1165 			struct ip6_hbh *hbh;
   1166 			int hbhlen;
   1167 			struct mbuf *ext;
   1168 
   1169 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
   1170 			    xip6->ip6_nxt);
   1171 			if (ext == NULL) {
   1172 				ip6stat.ip6s_tooshort++;
   1173 				return;
   1174 			}
   1175 			hbh = mtod(ext, struct ip6_hbh *);
   1176 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1177 			if (hbhlen != ext->m_len) {
   1178 				m_freem(ext);
   1179 				ip6stat.ip6s_tooshort++;
   1180 				return;
   1181 			}
   1182 
   1183 			/*
   1184 			 * XXX: We copy whole the header even if a jumbo
   1185 			 * payload option is included, which option is to
   1186 			 * be removed before returning in the RFC 2292.
   1187 			 * Note: this constraint is removed in RFC3542.
   1188 			 */
   1189 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
   1190 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
   1191 			    IPPROTO_IPV6);
   1192 			if (*mp)
   1193 				mp = &(*mp)->m_next;
   1194 			m_freem(ext);
   1195 		}
   1196 	}
   1197 
   1198 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
   1199 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
   1200 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1201 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
   1202 
   1203 		/*
   1204 		 * Search for destination options headers or routing
   1205 		 * header(s) through the header chain, and stores each
   1206 		 * header as ancillary data.
   1207 		 * Note that the order of the headers remains in
   1208 		 * the chain of ancillary data.
   1209 		 */
   1210 		for (;;) {	/* is explicit loop prevention necessary? */
   1211 			struct ip6_ext *ip6e = NULL;
   1212 			int elen;
   1213 			struct mbuf *ext = NULL;
   1214 
   1215 			/*
   1216 			 * if it is not an extension header, don't try to
   1217 			 * pull it from the chain.
   1218 			 */
   1219 			switch (nxt) {
   1220 			case IPPROTO_DSTOPTS:
   1221 			case IPPROTO_ROUTING:
   1222 			case IPPROTO_HOPOPTS:
   1223 			case IPPROTO_AH: /* is it possible? */
   1224 				break;
   1225 			default:
   1226 				goto loopend;
   1227 			}
   1228 
   1229 			ext = ip6_pullexthdr(m, off, nxt);
   1230 			if (ext == NULL) {
   1231 				ip6stat.ip6s_tooshort++;
   1232 				return;
   1233 			}
   1234 			ip6e = mtod(ext, struct ip6_ext *);
   1235 			if (nxt == IPPROTO_AH)
   1236 				elen = (ip6e->ip6e_len + 2) << 2;
   1237 			else
   1238 				elen = (ip6e->ip6e_len + 1) << 3;
   1239 			if (elen != ext->m_len) {
   1240 				m_freem(ext);
   1241 				ip6stat.ip6s_tooshort++;
   1242 				return;
   1243 			}
   1244 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
   1245 
   1246 			switch (nxt) {
   1247 			case IPPROTO_DSTOPTS:
   1248 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
   1249 					break;
   1250 
   1251 				*mp = sbcreatecontrol((void *)ip6e, elen,
   1252 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
   1253 				    IPPROTO_IPV6);
   1254 				if (*mp)
   1255 					mp = &(*mp)->m_next;
   1256 				break;
   1257 
   1258 			case IPPROTO_ROUTING:
   1259 				if (!in6p->in6p_flags & IN6P_RTHDR)
   1260 					break;
   1261 
   1262 				*mp = sbcreatecontrol((void *)ip6e, elen,
   1263 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
   1264 				    IPPROTO_IPV6);
   1265 				if (*mp)
   1266 					mp = &(*mp)->m_next;
   1267 				break;
   1268 
   1269 			case IPPROTO_HOPOPTS:
   1270 			case IPPROTO_AH: /* is it possible? */
   1271 				break;
   1272 
   1273 			default:
   1274 				/*
   1275 			 	 * other cases have been filtered in the above.
   1276 				 * none will visit this case.  here we supply
   1277 				 * the code just in case (nxt overwritten or
   1278 				 * other cases).
   1279 				 */
   1280 				m_freem(ext);
   1281 				goto loopend;
   1282 
   1283 			}
   1284 
   1285 			/* proceed with the next header. */
   1286 			off += elen;
   1287 			nxt = ip6e->ip6e_nxt;
   1288 			ip6e = NULL;
   1289 			m_freem(ext);
   1290 			ext = NULL;
   1291 		}
   1292 	  loopend:
   1293 	  	;
   1294 	}
   1295 }
   1296 #undef IS2292
   1297 
   1298 
   1299 void
   1300 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
   1301     uint32_t *mtu)
   1302 {
   1303 	struct socket *so;
   1304 	struct mbuf *m_mtu;
   1305 	struct ip6_mtuinfo mtuctl;
   1306 
   1307 	so = in6p->in6p_socket;
   1308 
   1309 	if (mtu == NULL)
   1310 		return;
   1311 
   1312 #ifdef DIAGNOSTIC
   1313 	if (so == NULL)		/* I believe this is impossible */
   1314 		panic("ip6_notify_pmtu: socket is NULL");
   1315 #endif
   1316 
   1317 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
   1318 	mtuctl.ip6m_mtu = *mtu;
   1319 	mtuctl.ip6m_addr = *dst;
   1320 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
   1321 		return;
   1322 
   1323 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
   1324 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
   1325 		return;
   1326 
   1327 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
   1328 	    == 0) {
   1329 		m_freem(m_mtu);
   1330 		/* XXX: should count statistics */
   1331 	} else
   1332 		sorwakeup(so);
   1333 
   1334 	return;
   1335 }
   1336 
   1337 /*
   1338  * pull single extension header from mbuf chain.  returns single mbuf that
   1339  * contains the result, or NULL on error.
   1340  */
   1341 static struct mbuf *
   1342 ip6_pullexthdr(m, off, nxt)
   1343 	struct mbuf *m;
   1344 	size_t off;
   1345 	int nxt;
   1346 {
   1347 	struct ip6_ext ip6e;
   1348 	size_t elen;
   1349 	struct mbuf *n;
   1350 
   1351 #ifdef DIAGNOSTIC
   1352 	switch (nxt) {
   1353 	case IPPROTO_DSTOPTS:
   1354 	case IPPROTO_ROUTING:
   1355 	case IPPROTO_HOPOPTS:
   1356 	case IPPROTO_AH: /* is it possible? */
   1357 		break;
   1358 	default:
   1359 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
   1360 	}
   1361 #endif
   1362 
   1363 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1364 	if (nxt == IPPROTO_AH)
   1365 		elen = (ip6e.ip6e_len + 2) << 2;
   1366 	else
   1367 		elen = (ip6e.ip6e_len + 1) << 3;
   1368 
   1369 	MGET(n, M_DONTWAIT, MT_DATA);
   1370 	if (n && elen >= MLEN) {
   1371 		MCLGET(n, M_DONTWAIT);
   1372 		if ((n->m_flags & M_EXT) == 0) {
   1373 			m_free(n);
   1374 			n = NULL;
   1375 		}
   1376 	}
   1377 	if (!n)
   1378 		return NULL;
   1379 
   1380 	n->m_len = 0;
   1381 	if (elen >= M_TRAILINGSPACE(n)) {
   1382 		m_free(n);
   1383 		return NULL;
   1384 	}
   1385 
   1386 	m_copydata(m, off, elen, mtod(n, void *));
   1387 	n->m_len = elen;
   1388 	return n;
   1389 }
   1390 
   1391 /*
   1392  * Get pointer to the previous header followed by the header
   1393  * currently processed.
   1394  * XXX: This function supposes that
   1395  *	M includes all headers,
   1396  *	the next header field and the header length field of each header
   1397  *	are valid, and
   1398  *	the sum of each header length equals to OFF.
   1399  * Because of these assumptions, this function must be called very
   1400  * carefully. Moreover, it will not be used in the near future when
   1401  * we develop `neater' mechanism to process extension headers.
   1402  */
   1403 u_int8_t *
   1404 ip6_get_prevhdr(m, off)
   1405 	struct mbuf *m;
   1406 	int off;
   1407 {
   1408 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1409 
   1410 	if (off == sizeof(struct ip6_hdr))
   1411 		return (&ip6->ip6_nxt);
   1412 	else {
   1413 		int len, nxt;
   1414 		struct ip6_ext *ip6e = NULL;
   1415 
   1416 		nxt = ip6->ip6_nxt;
   1417 		len = sizeof(struct ip6_hdr);
   1418 		while (len < off) {
   1419 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
   1420 
   1421 			switch (nxt) {
   1422 			case IPPROTO_FRAGMENT:
   1423 				len += sizeof(struct ip6_frag);
   1424 				break;
   1425 			case IPPROTO_AH:
   1426 				len += (ip6e->ip6e_len + 2) << 2;
   1427 				break;
   1428 			default:
   1429 				len += (ip6e->ip6e_len + 1) << 3;
   1430 				break;
   1431 			}
   1432 			nxt = ip6e->ip6e_nxt;
   1433 		}
   1434 		if (ip6e)
   1435 			return (&ip6e->ip6e_nxt);
   1436 		else
   1437 			return NULL;
   1438 	}
   1439 }
   1440 
   1441 /*
   1442  * get next header offset.  m will be retained.
   1443  */
   1444 int
   1445 ip6_nexthdr(m, off, proto, nxtp)
   1446 	struct mbuf *m;
   1447 	int off;
   1448 	int proto;
   1449 	int *nxtp;
   1450 {
   1451 	struct ip6_hdr ip6;
   1452 	struct ip6_ext ip6e;
   1453 	struct ip6_frag fh;
   1454 
   1455 	/* just in case */
   1456 	if (m == NULL)
   1457 		panic("ip6_nexthdr: m == NULL");
   1458 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
   1459 		return -1;
   1460 
   1461 	switch (proto) {
   1462 	case IPPROTO_IPV6:
   1463 		/* do not chase beyond intermediate IPv6 headers */
   1464 		if (off != 0)
   1465 			return -1;
   1466 		if (m->m_pkthdr.len < off + sizeof(ip6))
   1467 			return -1;
   1468 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
   1469 		if (nxtp)
   1470 			*nxtp = ip6.ip6_nxt;
   1471 		off += sizeof(ip6);
   1472 		return off;
   1473 
   1474 	case IPPROTO_FRAGMENT:
   1475 		/*
   1476 		 * terminate parsing if it is not the first fragment,
   1477 		 * it does not make sense to parse through it.
   1478 		 */
   1479 		if (m->m_pkthdr.len < off + sizeof(fh))
   1480 			return -1;
   1481 		m_copydata(m, off, sizeof(fh), (void *)&fh);
   1482 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
   1483 			return -1;
   1484 		if (nxtp)
   1485 			*nxtp = fh.ip6f_nxt;
   1486 		off += sizeof(struct ip6_frag);
   1487 		return off;
   1488 
   1489 	case IPPROTO_AH:
   1490 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1491 			return -1;
   1492 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1493 		if (nxtp)
   1494 			*nxtp = ip6e.ip6e_nxt;
   1495 		off += (ip6e.ip6e_len + 2) << 2;
   1496 		if (m->m_pkthdr.len < off)
   1497 			return -1;
   1498 		return off;
   1499 
   1500 	case IPPROTO_HOPOPTS:
   1501 	case IPPROTO_ROUTING:
   1502 	case IPPROTO_DSTOPTS:
   1503 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1504 			return -1;
   1505 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1506 		if (nxtp)
   1507 			*nxtp = ip6e.ip6e_nxt;
   1508 		off += (ip6e.ip6e_len + 1) << 3;
   1509 		if (m->m_pkthdr.len < off)
   1510 			return -1;
   1511 		return off;
   1512 
   1513 	case IPPROTO_NONE:
   1514 	case IPPROTO_ESP:
   1515 	case IPPROTO_IPCOMP:
   1516 		/* give up */
   1517 		return -1;
   1518 
   1519 	default:
   1520 		return -1;
   1521 	}
   1522 }
   1523 
   1524 /*
   1525  * get offset for the last header in the chain.  m will be kept untainted.
   1526  */
   1527 int
   1528 ip6_lasthdr(m, off, proto, nxtp)
   1529 	struct mbuf *m;
   1530 	int off;
   1531 	int proto;
   1532 	int *nxtp;
   1533 {
   1534 	int newoff;
   1535 	int nxt;
   1536 
   1537 	if (!nxtp) {
   1538 		nxt = -1;
   1539 		nxtp = &nxt;
   1540 	}
   1541 	for (;;) {
   1542 		newoff = ip6_nexthdr(m, off, proto, nxtp);
   1543 		if (newoff < 0)
   1544 			return off;
   1545 		else if (newoff < off)
   1546 			return -1;	/* invalid */
   1547 		else if (newoff == off)
   1548 			return newoff;
   1549 
   1550 		off = newoff;
   1551 		proto = *nxtp;
   1552 	}
   1553 }
   1554 
   1555 struct m_tag *
   1556 ip6_addaux(m)
   1557 	struct mbuf *m;
   1558 {
   1559 	struct m_tag *mtag;
   1560 
   1561 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1562 	if (!mtag) {
   1563 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
   1564 		    M_NOWAIT);
   1565 		if (mtag) {
   1566 			m_tag_prepend(m, mtag);
   1567 			bzero(mtag + 1, sizeof(struct ip6aux));
   1568 		}
   1569 	}
   1570 	return mtag;
   1571 }
   1572 
   1573 struct m_tag *
   1574 ip6_findaux(m)
   1575 	struct mbuf *m;
   1576 {
   1577 	struct m_tag *mtag;
   1578 
   1579 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1580 	return mtag;
   1581 }
   1582 
   1583 void
   1584 ip6_delaux(m)
   1585 	struct mbuf *m;
   1586 {
   1587 	struct m_tag *mtag;
   1588 
   1589 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1590 	if (mtag)
   1591 		m_tag_delete(m, mtag);
   1592 }
   1593 
   1594 /*
   1595  * System control for IP6
   1596  */
   1597 
   1598 u_char	inet6ctlerrmap[PRC_NCMDS] = {
   1599 	0,		0,		0,		0,
   1600 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
   1601 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
   1602 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
   1603 	0,		0,		0,		0,
   1604 	ENOPROTOOPT
   1605 };
   1606 
   1607 SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup")
   1608 {
   1609 #ifdef RFC2292
   1610 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1611 #else
   1612 #define IS2292(x, y)	(y)
   1613 #endif
   1614 
   1615 	sysctl_createv(clog, 0, NULL, NULL,
   1616 		       CTLFLAG_PERMANENT,
   1617 		       CTLTYPE_NODE, "net", NULL,
   1618 		       NULL, 0, NULL, 0,
   1619 		       CTL_NET, CTL_EOL);
   1620 	sysctl_createv(clog, 0, NULL, NULL,
   1621 		       CTLFLAG_PERMANENT,
   1622 		       CTLTYPE_NODE, "inet6",
   1623 		       SYSCTL_DESCR("PF_INET6 related settings"),
   1624 		       NULL, 0, NULL, 0,
   1625 		       CTL_NET, PF_INET6, CTL_EOL);
   1626 	sysctl_createv(clog, 0, NULL, NULL,
   1627 		       CTLFLAG_PERMANENT,
   1628 		       CTLTYPE_NODE, "ip6",
   1629 		       SYSCTL_DESCR("IPv6 related settings"),
   1630 		       NULL, 0, NULL, 0,
   1631 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
   1632 
   1633 	sysctl_createv(clog, 0, NULL, NULL,
   1634 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1635 		       CTLTYPE_INT, "forwarding",
   1636 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
   1637 		       NULL, 0, &ip6_forwarding, 0,
   1638 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1639 		       IPV6CTL_FORWARDING, CTL_EOL);
   1640 	sysctl_createv(clog, 0, NULL, NULL,
   1641 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1642 		       CTLTYPE_INT, "redirect",
   1643 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
   1644 		       NULL, 0, &ip6_sendredirects, 0,
   1645 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1646 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
   1647 	sysctl_createv(clog, 0, NULL, NULL,
   1648 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1649 		       CTLTYPE_INT, "hlim",
   1650 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
   1651 		       NULL, 0, &ip6_defhlim, 0,
   1652 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1653 		       IPV6CTL_DEFHLIM, CTL_EOL);
   1654 #ifdef notyet
   1655 	sysctl_createv(clog, 0, NULL, NULL,
   1656 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1657 		       CTLTYPE_INT, "mtu", NULL,
   1658 		       NULL, 0, &, 0,
   1659 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1660 		       IPV6CTL_DEFMTU, CTL_EOL);
   1661 #endif
   1662 #ifdef __no_idea__
   1663 	sysctl_createv(clog, 0, NULL, NULL,
   1664 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1665 		       CTLTYPE_INT, "forwsrcrt", NULL,
   1666 		       NULL, 0, &?, 0,
   1667 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1668 		       IPV6CTL_FORWSRCRT, CTL_EOL);
   1669 	sysctl_createv(clog, 0, NULL, NULL,
   1670 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1671 		       CTLTYPE_STRUCT, "mrtstats", NULL,
   1672 		       NULL, 0, &?, sizeof(?),
   1673 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1674 		       IPV6CTL_MRTSTATS, CTL_EOL);
   1675 	sysctl_createv(clog, 0, NULL, NULL,
   1676 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1677 		       CTLTYPE_?, "mrtproto", NULL,
   1678 		       NULL, 0, &?, sizeof(?),
   1679 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1680 		       IPV6CTL_MRTPROTO, CTL_EOL);
   1681 #endif
   1682 	sysctl_createv(clog, 0, NULL, NULL,
   1683 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1684 		       CTLTYPE_INT, "maxfragpackets",
   1685 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
   1686 				    "for reassembly"),
   1687 		       NULL, 0, &ip6_maxfragpackets, 0,
   1688 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1689 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
   1690 #ifdef __no_idea__
   1691 	sysctl_createv(clog, 0, NULL, NULL,
   1692 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1693 		       CTLTYPE_INT, "sourcecheck", NULL,
   1694 		       NULL, 0, &?, 0,
   1695 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1696 		       IPV6CTL_SOURCECHECK, CTL_EOL);
   1697 	sysctl_createv(clog, 0, NULL, NULL,
   1698 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1699 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
   1700 		       NULL, 0, &?, 0,
   1701 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1702 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
   1703 #endif
   1704 	sysctl_createv(clog, 0, NULL, NULL,
   1705 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1706 		       CTLTYPE_INT, "accept_rtadv",
   1707 		       SYSCTL_DESCR("Accept router advertisements"),
   1708 		       NULL, 0, &ip6_accept_rtadv, 0,
   1709 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1710 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
   1711 	sysctl_createv(clog, 0, NULL, NULL,
   1712 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1713 		       CTLTYPE_INT, "keepfaith",
   1714 		       SYSCTL_DESCR("Activate faith interface"),
   1715 		       NULL, 0, &ip6_keepfaith, 0,
   1716 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1717 		       IPV6CTL_KEEPFAITH, CTL_EOL);
   1718 	sysctl_createv(clog, 0, NULL, NULL,
   1719 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1720 		       CTLTYPE_INT, "log_interval",
   1721 		       SYSCTL_DESCR("Minumum interval between logging "
   1722 				    "unroutable packets"),
   1723 		       NULL, 0, &ip6_log_interval, 0,
   1724 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1725 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
   1726 	sysctl_createv(clog, 0, NULL, NULL,
   1727 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1728 		       CTLTYPE_INT, "hdrnestlimit",
   1729 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
   1730 		       NULL, 0, &ip6_hdrnestlimit, 0,
   1731 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1732 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
   1733 	sysctl_createv(clog, 0, NULL, NULL,
   1734 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1735 		       CTLTYPE_INT, "dad_count",
   1736 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
   1737 				    "probes to send"),
   1738 		       NULL, 0, &ip6_dad_count, 0,
   1739 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1740 		       IPV6CTL_DAD_COUNT, CTL_EOL);
   1741 	sysctl_createv(clog, 0, NULL, NULL,
   1742 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1743 		       CTLTYPE_INT, "auto_flowlabel",
   1744 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
   1745 		       NULL, 0, &ip6_auto_flowlabel, 0,
   1746 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1747 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
   1748 	sysctl_createv(clog, 0, NULL, NULL,
   1749 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1750 		       CTLTYPE_INT, "defmcasthlim",
   1751 		       SYSCTL_DESCR("Default multicast hop limit"),
   1752 		       NULL, 0, &ip6_defmcasthlim, 0,
   1753 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1754 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
   1755 #if NGIF > 0
   1756 	sysctl_createv(clog, 0, NULL, NULL,
   1757 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1758 		       CTLTYPE_INT, "gifhlim",
   1759 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
   1760 		       NULL, 0, &ip6_gif_hlim, 0,
   1761 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1762 		       IPV6CTL_GIF_HLIM, CTL_EOL);
   1763 #endif /* NGIF */
   1764 	sysctl_createv(clog, 0, NULL, NULL,
   1765 		       CTLFLAG_PERMANENT,
   1766 		       CTLTYPE_STRING, "kame_version",
   1767 		       SYSCTL_DESCR("KAME Version"),
   1768 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
   1769 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1770 		       IPV6CTL_KAME_VERSION, CTL_EOL);
   1771 	sysctl_createv(clog, 0, NULL, NULL,
   1772 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1773 		       CTLTYPE_INT, "use_deprecated",
   1774 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
   1775 				    "source addresses"),
   1776 		       NULL, 0, &ip6_use_deprecated, 0,
   1777 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1778 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
   1779 	sysctl_createv(clog, 0, NULL, NULL,
   1780 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1781 		       CTLTYPE_INT, "rr_prune", NULL,
   1782 		       NULL, 0, &ip6_rr_prune, 0,
   1783 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1784 		       IPV6CTL_RR_PRUNE, CTL_EOL);
   1785 	sysctl_createv(clog, 0, NULL, NULL,
   1786 		       CTLFLAG_PERMANENT
   1787 #ifndef INET6_BINDV6ONLY
   1788 		       |CTLFLAG_READWRITE,
   1789 #endif
   1790 		       CTLTYPE_INT, "v6only",
   1791 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
   1792 				    "to PF_INET sockets"),
   1793 		       NULL, 0, &ip6_v6only, 0,
   1794 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1795 		       IPV6CTL_V6ONLY, CTL_EOL);
   1796 	sysctl_createv(clog, 0, NULL, NULL,
   1797 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1798 		       CTLTYPE_INT, "anonportmin",
   1799 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
   1800 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
   1801 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1802 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
   1803 	sysctl_createv(clog, 0, NULL, NULL,
   1804 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1805 		       CTLTYPE_INT, "anonportmax",
   1806 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
   1807 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
   1808 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1809 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
   1810 #ifndef IPNOPRIVPORTS
   1811 	sysctl_createv(clog, 0, NULL, NULL,
   1812 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1813 		       CTLTYPE_INT, "lowportmin",
   1814 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
   1815 				    "to assign"),
   1816 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
   1817 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1818 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
   1819 	sysctl_createv(clog, 0, NULL, NULL,
   1820 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1821 		       CTLTYPE_INT, "lowportmax",
   1822 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
   1823 				    "to assign"),
   1824 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
   1825 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1826 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
   1827 #endif /* IPNOPRIVPORTS */
   1828 	sysctl_createv(clog, 0, NULL, NULL,
   1829 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1830 		       CTLTYPE_INT, "use_tempaddr",
   1831 		       SYSCTL_DESCR("Use temporary address"),
   1832 		       NULL, 0, &ip6_use_tempaddr, 0,
   1833 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1834 		       CTL_CREATE, CTL_EOL);
   1835 	sysctl_createv(clog, 0, NULL, NULL,
   1836 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1837 		       CTLTYPE_INT, "temppltime",
   1838 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
   1839 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
   1840 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1841 		       CTL_CREATE, CTL_EOL);
   1842 	sysctl_createv(clog, 0, NULL, NULL,
   1843 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1844 		       CTLTYPE_INT, "tempvltime",
   1845 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
   1846 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
   1847 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1848 		       CTL_CREATE, CTL_EOL);
   1849 	sysctl_createv(clog, 0, NULL, NULL,
   1850 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1851 		       CTLTYPE_INT, "maxfrags",
   1852 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
   1853 		       NULL, 0, &ip6_maxfrags, 0,
   1854 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1855 		       IPV6CTL_MAXFRAGS, CTL_EOL);
   1856 	sysctl_createv(clog, 0, NULL, NULL,
   1857 		       CTLFLAG_PERMANENT,
   1858 		       CTLTYPE_STRUCT, "stats",
   1859 		       SYSCTL_DESCR("IPv6 statistics"),
   1860 		       NULL, 0, &ip6stat, sizeof(ip6stat),
   1861 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1862 		       IPV6CTL_STATS, CTL_EOL);
   1863 	sysctl_createv(clog, 0, NULL, NULL,
   1864 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1865 		       CTLTYPE_INT, "use_defaultzone",
   1866 		       SYSCTL_DESCR("Whether to use the default scope zones"),
   1867 		       NULL, 0, &ip6_use_defzone, 0,
   1868 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1869 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
   1870 	sysctl_createv(clog, 0, NULL, NULL,
   1871 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1872 		       CTLTYPE_INT, "mcast_pmtu",
   1873 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
   1874 		       NULL, 0, &ip6_mcast_pmtu, 0,
   1875 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1876 		       CTL_CREATE, CTL_EOL);
   1877 }
   1878