Home | History | Annotate | Line # | Download | only in netinet6
ip6_input.c revision 1.94
      1 /*	$NetBSD: ip6_input.c,v 1.94 2007/02/10 09:43:05 degroote 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 1.94 2007/02/10 09:43:05 degroote 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(m)
    241 	struct mbuf *m;
    242 {
    243 	struct ip6_hdr *ip6;
    244 	int off = sizeof(struct ip6_hdr), nest;
    245 	u_int32_t plen;
    246 	u_int32_t rtalert = ~0;
    247 	int nxt, ours = 0;
    248 	struct ifnet *deliverifp = NULL;
    249 	int srcrt = 0;
    250 #ifdef FAST_IPSEC
    251 	struct m_tag *mtag;
    252 	struct tdb_ident *tdbi;
    253 	struct secpolicy *sp;
    254 	int s, error;
    255 #endif
    256 
    257 #ifdef IPSEC
    258 	/*
    259 	 * should the inner packet be considered authentic?
    260 	 * see comment in ah4_input().
    261 	 */
    262 	m->m_flags &= ~M_AUTHIPHDR;
    263 	m->m_flags &= ~M_AUTHIPDGM;
    264 #endif
    265 
    266 	/*
    267 	 * make sure we don't have onion peering information into m_tag.
    268 	 */
    269 	ip6_delaux(m);
    270 
    271 	/*
    272 	 * mbuf statistics
    273 	 */
    274 	if (m->m_flags & M_EXT) {
    275 		if (m->m_next)
    276 			ip6stat.ip6s_mext2m++;
    277 		else
    278 			ip6stat.ip6s_mext1++;
    279 	} else {
    280 #define M2MMAX	(sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
    281 		if (m->m_next) {
    282 			if (m->m_flags & M_LOOP) {
    283 				ip6stat.ip6s_m2m[lo0ifp->if_index]++; /* XXX */
    284 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
    285 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
    286 			else
    287 				ip6stat.ip6s_m2m[0]++;
    288 		} else
    289 			ip6stat.ip6s_m1++;
    290 #undef M2MMAX
    291 	}
    292 
    293 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
    294 	ip6stat.ip6s_total++;
    295 
    296 	/*
    297 	 * If the IPv6 header is not aligned, slurp it up into a new
    298 	 * mbuf with space for link headers, in the event we forward
    299 	 * it.  Otherwise, if it is aligned, make sure the entire base
    300 	 * IPv6 header is in the first mbuf of the chain.
    301 	 */
    302 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
    303 		struct ifnet *inifp = m->m_pkthdr.rcvif;
    304 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
    305 				  (max_linkhdr + 3) & ~3)) == NULL) {
    306 			/* XXXJRT new stat, please */
    307 			ip6stat.ip6s_toosmall++;
    308 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    309 			return;
    310 		}
    311 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
    312 		struct ifnet *inifp = m->m_pkthdr.rcvif;
    313 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    314 			ip6stat.ip6s_toosmall++;
    315 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    316 			return;
    317 		}
    318 	}
    319 
    320 	ip6 = mtod(m, struct ip6_hdr *);
    321 
    322 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    323 		ip6stat.ip6s_badvers++;
    324 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    325 		goto bad;
    326 	}
    327 
    328 #ifdef PFIL_HOOKS
    329 	/*
    330 	 * Run through list of hooks for input packets.  If there are any
    331 	 * filters which require that additional packets in the flow are
    332 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
    333 	 * Note that filters must _never_ set this flag, as another filter
    334 	 * in the list may have previously cleared it.
    335 	 */
    336 	/*
    337 	 * let ipfilter look at packet on the wire,
    338 	 * not the decapsulated packet.
    339 	 */
    340 #ifdef IPSEC
    341 	if (!ipsec_getnhist(m))
    342 #elif defined(FAST_IPSEC)
    343 	if (!ipsec_indone(m))
    344 #else
    345 	if (1)
    346 #endif
    347 	{
    348 		struct in6_addr odst;
    349 
    350 		odst = ip6->ip6_dst;
    351 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
    352 				   PFIL_IN) != 0)
    353 			return;
    354 		if (m == NULL)
    355 			return;
    356 		ip6 = mtod(m, struct ip6_hdr *);
    357 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
    358 	}
    359 #endif /* PFIL_HOOKS */
    360 
    361 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
    362 
    363 #ifdef ALTQ
    364 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
    365 		/* packet is dropped by traffic conditioner */
    366 		return;
    367 	}
    368 #endif
    369 
    370 	/*
    371 	 * Check against address spoofing/corruption.
    372 	 */
    373 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
    374 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
    375 		/*
    376 		 * XXX: "badscope" is not very suitable for a multicast source.
    377 		 */
    378 		ip6stat.ip6s_badscope++;
    379 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    380 		goto bad;
    381 	}
    382 	/*
    383 	 * The following check is not documented in specs.  A malicious
    384 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
    385 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
    386 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
    387 	 *
    388 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
    389 	 * support IPv4-less kernel compilation, we cannot support SIIT
    390 	 * environment at all.  So, it makes more sense for us to reject any
    391 	 * malicious packets for non-SIIT environment, than try to do a
    392 	 * partial support for SIIT environment.
    393 	 */
    394 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    395 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    396 		ip6stat.ip6s_badscope++;
    397 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    398 		goto bad;
    399 	}
    400 #if 0
    401 	/*
    402 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
    403 	 *
    404 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
    405 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
    406 	 * is revised to forbid relaying case.
    407 	 */
    408 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
    409 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
    410 		ip6stat.ip6s_badscope++;
    411 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    412 		goto bad;
    413 	}
    414 #endif
    415 
    416 	/*
    417 	 * Disambiguate address scope zones (if there is ambiguity).
    418 	 * We first make sure that the original source or destination address
    419 	 * is not in our internal form for scoped addresses.  Such addresses
    420 	 * are not necessarily invalid spec-wise, but we cannot accept them due
    421 	 * to the usage conflict.
    422 	 * in6_setscope() then also checks and rejects the cases where src or
    423 	 * dst are the loopback address and the receiving interface
    424 	 * is not loopback.
    425 	 */
    426 	if (__predict_false(
    427 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
    428 		goto bad;
    429 	ip6 = mtod(m, struct ip6_hdr *);
    430 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
    431 		ip6stat.ip6s_badscope++; /* XXX */
    432 		goto bad;
    433 	}
    434 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
    435 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
    436 		ip6stat.ip6s_badscope++;
    437 		goto bad;
    438 	}
    439 
    440 	/*
    441 	 * Multicast check
    442 	 */
    443 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    444 	  	struct	in6_multi *in6m = 0;
    445 
    446 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
    447 		/*
    448 		 * See if we belong to the destination multicast group on the
    449 		 * arrival interface.
    450 		 */
    451 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
    452 		if (in6m)
    453 			ours = 1;
    454 		else if (!ip6_mrouter) {
    455 			ip6stat.ip6s_notmember++;
    456 			ip6stat.ip6s_cantforward++;
    457 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    458 			goto bad;
    459 		}
    460 		deliverifp = m->m_pkthdr.rcvif;
    461 		goto hbhcheck;
    462 	}
    463 
    464 	/*
    465 	 *  Unicast check
    466 	 */
    467 	if (!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
    468 	    &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
    469 		rtcache_free((struct route *)&ip6_forward_rt);
    470 	else
    471 		rtcache_check((struct route *)&ip6_forward_rt);
    472 	if (ip6_forward_rt.ro_rt != NULL) {
    473 		/* XXX Revalidated route is accounted wrongly. */
    474 		ip6stat.ip6s_forward_cachehit++;
    475 	} else {
    476 		struct sockaddr_in6 *dst6;
    477 
    478 		ip6stat.ip6s_forward_cachemiss++;
    479 
    480 		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
    481 		dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
    482 		dst6->sin6_len = sizeof(struct sockaddr_in6);
    483 		dst6->sin6_family = AF_INET6;
    484 		dst6->sin6_addr = ip6->ip6_dst;
    485 
    486 		rtcache_init((struct route *)&ip6_forward_rt);
    487 	}
    488 
    489 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
    490 
    491 	/*
    492 	 * Accept the packet if the forwarding interface to the destination
    493 	 * according to the routing table is the loopback interface,
    494 	 * unless the associated route has a gateway.
    495 	 * Note that this approach causes to accept a packet if there is a
    496 	 * route to the loopback interface for the destination of the packet.
    497 	 * But we think it's even useful in some situations, e.g. when using
    498 	 * a special daemon which wants to intercept the packet.
    499 	 */
    500 	if (ip6_forward_rt.ro_rt != NULL &&
    501 	    (ip6_forward_rt.ro_rt->rt_flags &
    502 	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
    503 	    !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
    504 #if 0
    505 	    /*
    506 	     * The check below is redundant since the comparison of
    507 	     * the destination and the key of the rtentry has
    508 	     * already done through looking up the routing table.
    509 	     */
    510 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
    511 	    &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
    512 #endif
    513 	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
    514 		struct in6_ifaddr *ia6 =
    515 			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
    516 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
    517 			m->m_flags |= M_ANYCAST6;
    518 		/*
    519 		 * packets to a tentative, duplicated, or somehow invalid
    520 		 * address must not be accepted.
    521 		 */
    522 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
    523 			/* this address is ready */
    524 			ours = 1;
    525 			deliverifp = ia6->ia_ifp;	/* correct? */
    526 			goto hbhcheck;
    527 		} else {
    528 			/* address is not ready, so discard the packet. */
    529 			nd6log((LOG_INFO,
    530 			    "ip6_input: packet to an unready address %s->%s\n",
    531 			    ip6_sprintf(&ip6->ip6_src),
    532 			    ip6_sprintf(&ip6->ip6_dst)));
    533 
    534 			goto bad;
    535 		}
    536 	}
    537 
    538 	/*
    539 	 * FAITH (Firewall Aided Internet Translator)
    540 	 */
    541 #if defined(NFAITH) && 0 < NFAITH
    542 	if (ip6_keepfaith) {
    543 		if (ip6_forward_rt.ro_rt != NULL &&
    544 		    ip6_forward_rt.ro_rt->rt_ifp != NULL &&
    545 		    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
    546 			/* XXX do we need more sanity checks? */
    547 			ours = 1;
    548 			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
    549 			goto hbhcheck;
    550 		}
    551 	}
    552 #endif
    553 
    554 #if 0
    555     {
    556 	/*
    557 	 * Last resort: check in6_ifaddr for incoming interface.
    558 	 * The code is here until I update the "goto ours hack" code above
    559 	 * working right.
    560 	 */
    561 	struct ifaddr *ifa;
    562 	TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) {
    563 		if (ifa->ifa_addr == NULL)
    564 			continue;	/* just for safety */
    565 		if (ifa->ifa_addr->sa_family != AF_INET6)
    566 			continue;
    567 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
    568 			ours = 1;
    569 			deliverifp = ifa->ifa_ifp;
    570 			goto hbhcheck;
    571 		}
    572 	}
    573     }
    574 #endif
    575 
    576 	/*
    577 	 * Now there is no reason to process the packet if it's not our own
    578 	 * and we're not a router.
    579 	 */
    580 	if (!ip6_forwarding) {
    581 		ip6stat.ip6s_cantforward++;
    582 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    583 		goto bad;
    584 	}
    585 
    586   hbhcheck:
    587 	/*
    588 	 * record address information into m_tag, if we don't have one yet.
    589 	 * note that we are unable to record it, if the address is not listed
    590 	 * as our interface address (e.g. multicast addresses, addresses
    591 	 * within FAITH prefixes and such).
    592 	 */
    593 	if (deliverifp && !ip6_getdstifaddr(m)) {
    594 		struct in6_ifaddr *ia6;
    595 
    596 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    597 		if (ia6) {
    598 			if (!ip6_setdstifaddr(m, ia6)) {
    599 				/*
    600 				 * XXX maybe we should drop the packet here,
    601 				 * as we could not provide enough information
    602 				 * to the upper layers.
    603 				 */
    604 			}
    605 		}
    606 	}
    607 
    608 	/*
    609 	 * Process Hop-by-Hop options header if it's contained.
    610 	 * m may be modified in ip6_hopopts_input().
    611 	 * If a JumboPayload option is included, plen will also be modified.
    612 	 */
    613 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
    614 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
    615 		struct ip6_hbh *hbh;
    616 
    617 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
    618 #if 0	/*touches NULL pointer*/
    619 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    620 #endif
    621 			return;	/* m have already been freed */
    622 		}
    623 
    624 		/* adjust pointer */
    625 		ip6 = mtod(m, struct ip6_hdr *);
    626 
    627 		/*
    628 		 * if the payload length field is 0 and the next header field
    629 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
    630 		 * option MUST be included.
    631 		 */
    632 		if (ip6->ip6_plen == 0 && plen == 0) {
    633 			/*
    634 			 * Note that if a valid jumbo payload option is
    635 			 * contained, ip6_hopopts_input() must set a valid
    636 			 * (non-zero) payload length to the variable plen.
    637 			 */
    638 			ip6stat.ip6s_badoptions++;
    639 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    640 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    641 			icmp6_error(m, ICMP6_PARAM_PROB,
    642 				    ICMP6_PARAMPROB_HEADER,
    643 				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
    644 			return;
    645 		}
    646 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    647 			sizeof(struct ip6_hbh));
    648 		if (hbh == NULL) {
    649 			ip6stat.ip6s_tooshort++;
    650 			return;
    651 		}
    652 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
    653 		nxt = hbh->ip6h_nxt;
    654 
    655 		/*
    656 		 * accept the packet if a router alert option is included
    657 		 * and we act as an IPv6 router.
    658 		 */
    659 		if (rtalert != ~0 && ip6_forwarding)
    660 			ours = 1;
    661 	} else
    662 		nxt = ip6->ip6_nxt;
    663 
    664 	/*
    665 	 * Check that the amount of data in the buffers
    666 	 * is as at least much as the IPv6 header would have us expect.
    667 	 * Trim mbufs if longer than we expect.
    668 	 * Drop packet if shorter than we expect.
    669 	 */
    670 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
    671 		ip6stat.ip6s_tooshort++;
    672 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    673 		goto bad;
    674 	}
    675 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
    676 		if (m->m_len == m->m_pkthdr.len) {
    677 			m->m_len = sizeof(struct ip6_hdr) + plen;
    678 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
    679 		} else
    680 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
    681 	}
    682 
    683 	/*
    684 	 * Forward if desirable.
    685 	 */
    686 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    687 		/*
    688 		 * If we are acting as a multicast router, all
    689 		 * incoming multicast packets are passed to the
    690 		 * kernel-level multicast forwarding function.
    691 		 * The packet is returned (relatively) intact; if
    692 		 * ip6_mforward() returns a non-zero value, the packet
    693 		 * must be discarded, else it may be accepted below.
    694 		 */
    695 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
    696 			ip6stat.ip6s_cantforward++;
    697 			m_freem(m);
    698 			return;
    699 		}
    700 		if (!ours) {
    701 			m_freem(m);
    702 			return;
    703 		}
    704 	} else if (!ours) {
    705 		ip6_forward(m, srcrt);
    706 		return;
    707 	}
    708 
    709 	ip6 = mtod(m, struct ip6_hdr *);
    710 
    711 	/*
    712 	 * Malicious party may be able to use IPv4 mapped addr to confuse
    713 	 * tcp/udp stack and bypass security checks (act as if it was from
    714 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
    715 	 *
    716 	 * For SIIT end node behavior, you may want to disable the check.
    717 	 * However, you will  become vulnerable to attacks using IPv4 mapped
    718 	 * source.
    719 	 */
    720 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    721 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    722 		ip6stat.ip6s_badscope++;
    723 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    724 		goto bad;
    725 	}
    726 
    727 	/*
    728 	 * Tell launch routine the next header
    729 	 */
    730 #ifdef IFA_STATS
    731 	if (deliverifp != NULL) {
    732 		struct in6_ifaddr *ia6;
    733 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    734 		if (ia6)
    735 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
    736 	}
    737 #endif
    738 	ip6stat.ip6s_delivered++;
    739 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
    740 	nest = 0;
    741 
    742 	while (nxt != IPPROTO_DONE) {
    743 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
    744 			ip6stat.ip6s_toomanyhdr++;
    745 			goto bad;
    746 		}
    747 
    748 		/*
    749 		 * protection against faulty packet - there should be
    750 		 * more sanity checks in header chain processing.
    751 		 */
    752 		if (m->m_pkthdr.len < off) {
    753 			ip6stat.ip6s_tooshort++;
    754 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    755 			goto bad;
    756 		}
    757 
    758 #ifdef IPSEC
    759 		/*
    760 		 * enforce IPsec policy checking if we are seeing last header.
    761 		 * note that we do not visit this with protocols with pcb layer
    762 		 * code - like udp/tcp/raw ip.
    763 		 */
    764 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
    765 		    ipsec6_in_reject(m, NULL)) {
    766 			ipsec6stat.in_polvio++;
    767 			goto bad;
    768 		}
    769 #endif
    770 #ifdef FAST_IPSEC
    771 	/*
    772 	 * enforce IPsec policy checking if we are seeing last header.
    773 	 * note that we do not visit this with protocols with pcb layer
    774 	 * code - like udp/tcp/raw ip.
    775 	 */
    776 	if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) {
    777 		/*
    778 		 * Check if the packet has already had IPsec processing
    779 		 * done.  If so, then just pass it along.  This tag gets
    780 		 * set during AH, ESP, etc. input handling, before the
    781 		 * packet is returned to the ip input queue for delivery.
    782 		 */
    783 		mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
    784 		s = splsoftnet();
    785 		if (mtag != NULL) {
    786 			tdbi = (struct tdb_ident *)(mtag + 1);
    787 			sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
    788 		} else {
    789 			sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
    790 									IP_FORWARDING, &error);
    791 		}
    792 		if (sp != NULL) {
    793 			/*
    794 			 * Check security policy against packet attributes.
    795 			 */
    796 			error = ipsec_in_reject(sp, m);
    797 			KEY_FREESP(&sp);
    798 		} else {
    799 			/* XXX error stat??? */
    800 			error = EINVAL;
    801 			DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/
    802 			goto bad;
    803 		}
    804 		splx(s);
    805 		if (error)
    806 			goto bad;
    807 	}
    808 #endif /* FAST_IPSEC */
    809 
    810 
    811 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
    812 	}
    813 	return;
    814  bad:
    815 	m_freem(m);
    816 }
    817 
    818 /*
    819  * set/grab in6_ifaddr correspond to IPv6 destination address.
    820  */
    821 static struct m_tag *
    822 ip6_setdstifaddr(m, ia6)
    823 	struct mbuf *m;
    824 	struct in6_ifaddr *ia6;
    825 {
    826 	struct m_tag *mtag;
    827 
    828 	mtag = ip6_addaux(m);
    829 	if (mtag)
    830 		((struct ip6aux *)(mtag + 1))->ip6a_dstia6 = ia6;
    831 	return mtag;	/* NULL if failed to set */
    832 }
    833 
    834 struct in6_ifaddr *
    835 ip6_getdstifaddr(m)
    836 	struct mbuf *m;
    837 {
    838 	struct m_tag *mtag;
    839 
    840 	mtag = ip6_findaux(m);
    841 	if (mtag)
    842 		return ((struct ip6aux *)(mtag + 1))->ip6a_dstia6;
    843 	else
    844 		return NULL;
    845 }
    846 
    847 /*
    848  * Hop-by-Hop options header processing. If a valid jumbo payload option is
    849  * included, the real payload length will be stored in plenp.
    850  */
    851 static int
    852 ip6_hopopts_input(plenp, rtalertp, mp, offp)
    853 	u_int32_t *plenp;
    854 	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
    855 	struct mbuf **mp;
    856 	int *offp;
    857 {
    858 	struct mbuf *m = *mp;
    859 	int off = *offp, hbhlen;
    860 	struct ip6_hbh *hbh;
    861 
    862 	/* validation of the length of the header */
    863 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
    864 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
    865 	if (hbh == NULL) {
    866 		ip6stat.ip6s_tooshort++;
    867 		return -1;
    868 	}
    869 	hbhlen = (hbh->ip6h_len + 1) << 3;
    870 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    871 		hbhlen);
    872 	if (hbh == NULL) {
    873 		ip6stat.ip6s_tooshort++;
    874 		return -1;
    875 	}
    876 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
    877 	off += hbhlen;
    878 	hbhlen -= sizeof(struct ip6_hbh);
    879 
    880 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
    881 				hbhlen, rtalertp, plenp) < 0)
    882 		return (-1);
    883 
    884 	*offp = off;
    885 	*mp = m;
    886 	return (0);
    887 }
    888 
    889 /*
    890  * Search header for all Hop-by-hop options and process each option.
    891  * This function is separate from ip6_hopopts_input() in order to
    892  * handle a case where the sending node itself process its hop-by-hop
    893  * options header. In such a case, the function is called from ip6_output().
    894  *
    895  * The function assumes that hbh header is located right after the IPv6 header
    896  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
    897  * opthead + hbhlen is located in continuous memory region.
    898  */
    899 int
    900 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
    901 	struct mbuf *m;
    902 	u_int8_t *opthead;
    903 	int hbhlen;
    904 	u_int32_t *rtalertp;
    905 	u_int32_t *plenp;
    906 {
    907 	struct ip6_hdr *ip6;
    908 	int optlen = 0;
    909 	u_int8_t *opt = opthead;
    910 	u_int16_t rtalert_val;
    911 	u_int32_t jumboplen;
    912 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
    913 
    914 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
    915 		switch (*opt) {
    916 		case IP6OPT_PAD1:
    917 			optlen = 1;
    918 			break;
    919 		case IP6OPT_PADN:
    920 			if (hbhlen < IP6OPT_MINLEN) {
    921 				ip6stat.ip6s_toosmall++;
    922 				goto bad;
    923 			}
    924 			optlen = *(opt + 1) + 2;
    925 			break;
    926 		case IP6OPT_RTALERT:
    927 			/* XXX may need check for alignment */
    928 			if (hbhlen < IP6OPT_RTALERT_LEN) {
    929 				ip6stat.ip6s_toosmall++;
    930 				goto bad;
    931 			}
    932 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
    933 				/* XXX stat */
    934 				icmp6_error(m, ICMP6_PARAM_PROB,
    935 				    ICMP6_PARAMPROB_HEADER,
    936 				    erroff + opt + 1 - opthead);
    937 				return (-1);
    938 			}
    939 			optlen = IP6OPT_RTALERT_LEN;
    940 			bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
    941 			*rtalertp = ntohs(rtalert_val);
    942 			break;
    943 		case IP6OPT_JUMBO:
    944 			/* XXX may need check for alignment */
    945 			if (hbhlen < IP6OPT_JUMBO_LEN) {
    946 				ip6stat.ip6s_toosmall++;
    947 				goto bad;
    948 			}
    949 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
    950 				/* XXX stat */
    951 				icmp6_error(m, ICMP6_PARAM_PROB,
    952 				    ICMP6_PARAMPROB_HEADER,
    953 				    erroff + opt + 1 - opthead);
    954 				return (-1);
    955 			}
    956 			optlen = IP6OPT_JUMBO_LEN;
    957 
    958 			/*
    959 			 * IPv6 packets that have non 0 payload length
    960 			 * must not contain a jumbo payload option.
    961 			 */
    962 			ip6 = mtod(m, struct ip6_hdr *);
    963 			if (ip6->ip6_plen) {
    964 				ip6stat.ip6s_badoptions++;
    965 				icmp6_error(m, ICMP6_PARAM_PROB,
    966 				    ICMP6_PARAMPROB_HEADER,
    967 				    erroff + opt - opthead);
    968 				return (-1);
    969 			}
    970 
    971 			/*
    972 			 * We may see jumbolen in unaligned location, so
    973 			 * we'd need to perform bcopy().
    974 			 */
    975 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
    976 			jumboplen = (u_int32_t)htonl(jumboplen);
    977 
    978 #if 1
    979 			/*
    980 			 * if there are multiple jumbo payload options,
    981 			 * *plenp will be non-zero and the packet will be
    982 			 * rejected.
    983 			 * the behavior may need some debate in ipngwg -
    984 			 * multiple options does not make sense, however,
    985 			 * there's no explicit mention in specification.
    986 			 */
    987 			if (*plenp != 0) {
    988 				ip6stat.ip6s_badoptions++;
    989 				icmp6_error(m, ICMP6_PARAM_PROB,
    990 				    ICMP6_PARAMPROB_HEADER,
    991 				    erroff + opt + 2 - opthead);
    992 				return (-1);
    993 			}
    994 #endif
    995 
    996 			/*
    997 			 * jumbo payload length must be larger than 65535.
    998 			 */
    999 			if (jumboplen <= IPV6_MAXPACKET) {
   1000 				ip6stat.ip6s_badoptions++;
   1001 				icmp6_error(m, ICMP6_PARAM_PROB,
   1002 				    ICMP6_PARAMPROB_HEADER,
   1003 				    erroff + opt + 2 - opthead);
   1004 				return (-1);
   1005 			}
   1006 			*plenp = jumboplen;
   1007 
   1008 			break;
   1009 		default:		/* unknown option */
   1010 			if (hbhlen < IP6OPT_MINLEN) {
   1011 				ip6stat.ip6s_toosmall++;
   1012 				goto bad;
   1013 			}
   1014 			optlen = ip6_unknown_opt(opt, m,
   1015 			    erroff + opt - opthead);
   1016 			if (optlen == -1)
   1017 				return (-1);
   1018 			optlen += 2;
   1019 			break;
   1020 		}
   1021 	}
   1022 
   1023 	return (0);
   1024 
   1025   bad:
   1026 	m_freem(m);
   1027 	return (-1);
   1028 }
   1029 
   1030 /*
   1031  * Unknown option processing.
   1032  * The third argument `off' is the offset from the IPv6 header to the option,
   1033  * which is necessary if the IPv6 header the and option header and IPv6 header
   1034  * is not continuous in order to return an ICMPv6 error.
   1035  */
   1036 int
   1037 ip6_unknown_opt(optp, m, off)
   1038 	u_int8_t *optp;
   1039 	struct mbuf *m;
   1040 	int off;
   1041 {
   1042 	struct ip6_hdr *ip6;
   1043 
   1044 	switch (IP6OPT_TYPE(*optp)) {
   1045 	case IP6OPT_TYPE_SKIP: /* ignore the option */
   1046 		return ((int)*(optp + 1));
   1047 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
   1048 		m_freem(m);
   1049 		return (-1);
   1050 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
   1051 		ip6stat.ip6s_badoptions++;
   1052 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
   1053 		return (-1);
   1054 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
   1055 		ip6stat.ip6s_badoptions++;
   1056 		ip6 = mtod(m, struct ip6_hdr *);
   1057 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
   1058 		    (m->m_flags & (M_BCAST|M_MCAST)))
   1059 			m_freem(m);
   1060 		else
   1061 			icmp6_error(m, ICMP6_PARAM_PROB,
   1062 				    ICMP6_PARAMPROB_OPTION, off);
   1063 		return (-1);
   1064 	}
   1065 
   1066 	m_freem(m);		/* XXX: NOTREACHED */
   1067 	return (-1);
   1068 }
   1069 
   1070 /*
   1071  * Create the "control" list for this pcb.
   1072  *
   1073  * The routine will be called from upper layer handlers like tcp6_input().
   1074  * Thus the routine assumes that the caller (tcp6_input) have already
   1075  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
   1076  * very first mbuf on the mbuf chain.
   1077  * We may want to add some infinite loop prevention or sanity checks for safety.
   1078  * (This applies only when you are using KAME mbuf chain restriction, i.e.
   1079  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
   1080  */
   1081 void
   1082 ip6_savecontrol(in6p, mp, ip6, m)
   1083 	struct in6pcb *in6p;
   1084 	struct mbuf **mp;
   1085 	struct ip6_hdr *ip6;
   1086 	struct mbuf *m;
   1087 {
   1088 #ifdef RFC2292
   1089 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1090 #else
   1091 #define IS2292(x, y)	(y)
   1092 #endif
   1093 
   1094 #ifdef SO_TIMESTAMP
   1095 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
   1096 		struct timeval tv;
   1097 
   1098 		microtime(&tv);
   1099 		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
   1100 		    SCM_TIMESTAMP, SOL_SOCKET);
   1101 		if (*mp)
   1102 			mp = &(*mp)->m_next;
   1103 	}
   1104 #endif
   1105 
   1106 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
   1107 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
   1108 		return;
   1109 
   1110 	/* RFC 2292 sec. 5 */
   1111 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
   1112 		struct in6_pktinfo pi6;
   1113 
   1114 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
   1115 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
   1116 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
   1117 		    m->m_pkthdr.rcvif->if_index : 0;
   1118 		*mp = sbcreatecontrol((caddr_t) &pi6,
   1119 		    sizeof(struct in6_pktinfo),
   1120 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
   1121 		if (*mp)
   1122 			mp = &(*mp)->m_next;
   1123 	}
   1124 
   1125 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
   1126 		int hlim = ip6->ip6_hlim & 0xff;
   1127 
   1128 		*mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
   1129 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
   1130 		if (*mp)
   1131 			mp = &(*mp)->m_next;
   1132 	}
   1133 
   1134 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
   1135 		u_int32_t flowinfo;
   1136 		int tclass;
   1137 
   1138 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
   1139 		flowinfo >>= 20;
   1140 
   1141 		tclass = flowinfo & 0xff;
   1142 		*mp = sbcreatecontrol((caddr_t)&tclass, sizeof(tclass),
   1143 		    IPV6_TCLASS, IPPROTO_IPV6);
   1144 
   1145 		if (*mp)
   1146 			mp = &(*mp)->m_next;
   1147 	}
   1148 
   1149 	/*
   1150 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
   1151 	 * privilege for the option (see ip6_ctloutput), but it might be too
   1152 	 * strict, since there might be some hop-by-hop options which can be
   1153 	 * returned to normal user.
   1154 	 * See also RFC3542 section 8 (or RFC2292 section 6).
   1155 	 */
   1156 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
   1157 		/*
   1158 		 * Check if a hop-by-hop options header is contatined in the
   1159 		 * received packet, and if so, store the options as ancillary
   1160 		 * data. Note that a hop-by-hop options header must be
   1161 		 * just after the IPv6 header, which fact is assured through
   1162 		 * the IPv6 input processing.
   1163 		 */
   1164 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1165 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
   1166 			struct ip6_hbh *hbh;
   1167 			int hbhlen;
   1168 			struct mbuf *ext;
   1169 
   1170 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
   1171 			    xip6->ip6_nxt);
   1172 			if (ext == NULL) {
   1173 				ip6stat.ip6s_tooshort++;
   1174 				return;
   1175 			}
   1176 			hbh = mtod(ext, struct ip6_hbh *);
   1177 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1178 			if (hbhlen != ext->m_len) {
   1179 				m_freem(ext);
   1180 				ip6stat.ip6s_tooshort++;
   1181 				return;
   1182 			}
   1183 
   1184 			/*
   1185 			 * XXX: We copy whole the header even if a jumbo
   1186 			 * payload option is included, which option is to
   1187 			 * be removed before returning in the RFC 2292.
   1188 			 * Note: this constraint is removed in RFC3542.
   1189 			 */
   1190 			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
   1191 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
   1192 			    IPPROTO_IPV6);
   1193 			if (*mp)
   1194 				mp = &(*mp)->m_next;
   1195 			m_freem(ext);
   1196 		}
   1197 	}
   1198 
   1199 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
   1200 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
   1201 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1202 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
   1203 
   1204 		/*
   1205 		 * Search for destination options headers or routing
   1206 		 * header(s) through the header chain, and stores each
   1207 		 * header as ancillary data.
   1208 		 * Note that the order of the headers remains in
   1209 		 * the chain of ancillary data.
   1210 		 */
   1211 		for (;;) {	/* is explicit loop prevention necessary? */
   1212 			struct ip6_ext *ip6e = NULL;
   1213 			int elen;
   1214 			struct mbuf *ext = NULL;
   1215 
   1216 			/*
   1217 			 * if it is not an extension header, don't try to
   1218 			 * pull it from the chain.
   1219 			 */
   1220 			switch (nxt) {
   1221 			case IPPROTO_DSTOPTS:
   1222 			case IPPROTO_ROUTING:
   1223 			case IPPROTO_HOPOPTS:
   1224 			case IPPROTO_AH: /* is it possible? */
   1225 				break;
   1226 			default:
   1227 				goto loopend;
   1228 			}
   1229 
   1230 			ext = ip6_pullexthdr(m, off, nxt);
   1231 			if (ext == NULL) {
   1232 				ip6stat.ip6s_tooshort++;
   1233 				return;
   1234 			}
   1235 			ip6e = mtod(ext, struct ip6_ext *);
   1236 			if (nxt == IPPROTO_AH)
   1237 				elen = (ip6e->ip6e_len + 2) << 2;
   1238 			else
   1239 				elen = (ip6e->ip6e_len + 1) << 3;
   1240 			if (elen != ext->m_len) {
   1241 				m_freem(ext);
   1242 				ip6stat.ip6s_tooshort++;
   1243 				return;
   1244 			}
   1245 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
   1246 
   1247 			switch (nxt) {
   1248 			case IPPROTO_DSTOPTS:
   1249 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
   1250 					break;
   1251 
   1252 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
   1253 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
   1254 				    IPPROTO_IPV6);
   1255 				if (*mp)
   1256 					mp = &(*mp)->m_next;
   1257 				break;
   1258 
   1259 			case IPPROTO_ROUTING:
   1260 				if (!in6p->in6p_flags & IN6P_RTHDR)
   1261 					break;
   1262 
   1263 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
   1264 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
   1265 				    IPPROTO_IPV6);
   1266 				if (*mp)
   1267 					mp = &(*mp)->m_next;
   1268 				break;
   1269 
   1270 			case IPPROTO_HOPOPTS:
   1271 			case IPPROTO_AH: /* is it possible? */
   1272 				break;
   1273 
   1274 			default:
   1275 				/*
   1276 			 	 * other cases have been filtered in the above.
   1277 				 * none will visit this case.  here we supply
   1278 				 * the code just in case (nxt overwritten or
   1279 				 * other cases).
   1280 				 */
   1281 				m_freem(ext);
   1282 				goto loopend;
   1283 
   1284 			}
   1285 
   1286 			/* proceed with the next header. */
   1287 			off += elen;
   1288 			nxt = ip6e->ip6e_nxt;
   1289 			ip6e = NULL;
   1290 			m_freem(ext);
   1291 			ext = NULL;
   1292 		}
   1293 	  loopend:
   1294 	  	;
   1295 	}
   1296 }
   1297 #undef IS2292
   1298 
   1299 
   1300 void
   1301 ip6_notify_pmtu(struct in6pcb *in6p, struct sockaddr_in6 *dst, 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((caddr_t)&mtuctl, sizeof(mtuctl),
   1324 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
   1325 		return;
   1326 
   1327 	if (sbappendaddr(&so->so_rcv, (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), (caddr_t)&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, caddr_t));
   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, caddr_t) + 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), (caddr_t)&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), (caddr_t)&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), (caddr_t)&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), (caddr_t)&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