Home | History | Annotate | Line # | Download | only in netinet6
ip6_input.c revision 1.126
      1 /*	$NetBSD: ip6_input.c,v 1.126 2009/04/18 14:58:05 tsutsui 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 1.126 2009/04/18 14:58:05 tsutsui Exp $");
     66 
     67 #include "opt_inet.h"
     68 #include "opt_inet6.h"
     69 #include "opt_ipsec.h"
     70 #include "opt_pfil_hooks.h"
     71 #include "opt_compat_netbsd.h"
     72 
     73 #include <sys/param.h>
     74 #include <sys/systm.h>
     75 #include <sys/malloc.h>
     76 #include <sys/mbuf.h>
     77 #include <sys/domain.h>
     78 #include <sys/protosw.h>
     79 #include <sys/socket.h>
     80 #include <sys/socketvar.h>
     81 #include <sys/errno.h>
     82 #include <sys/time.h>
     83 #include <sys/kernel.h>
     84 #include <sys/syslog.h>
     85 #include <sys/proc.h>
     86 #include <sys/sysctl.h>
     87 
     88 #include <net/if.h>
     89 #include <net/if_types.h>
     90 #include <net/if_dl.h>
     91 #include <net/route.h>
     92 #include <net/netisr.h>
     93 #ifdef PFIL_HOOKS
     94 #include <net/pfil.h>
     95 #endif
     96 
     97 #include <netinet/in.h>
     98 #include <netinet/in_systm.h>
     99 #ifdef INET
    100 #include <netinet/ip.h>
    101 #include <netinet/ip_icmp.h>
    102 #endif /* INET */
    103 #include <netinet/ip6.h>
    104 #include <netinet6/in6_var.h>
    105 #include <netinet6/ip6_var.h>
    106 #include <netinet6/ip6_private.h>
    107 #include <netinet6/in6_pcb.h>
    108 #include <netinet/icmp6.h>
    109 #include <netinet6/scope6_var.h>
    110 #include <netinet6/in6_ifattach.h>
    111 #include <netinet6/nd6.h>
    112 
    113 #ifdef IPSEC
    114 #include <netinet6/ipsec.h>
    115 #include <netinet6/ipsec_private.h>
    116 #endif
    117 
    118 #ifdef FAST_IPSEC
    119 #include <netipsec/ipsec.h>
    120 #include <netipsec/ipsec6.h>
    121 #include <netipsec/key.h>
    122 #endif /* FAST_IPSEC */
    123 
    124 #ifdef COMPAT_50
    125 #include <compat/sys/time.h>
    126 #include <compat/sys/socket.h>
    127 #endif
    128 
    129 #include <netinet6/ip6protosw.h>
    130 
    131 #include "faith.h"
    132 #include "gif.h"
    133 
    134 #if NGIF > 0
    135 #include <netinet6/in6_gif.h>
    136 #endif
    137 
    138 #include <net/net_osdep.h>
    139 
    140 extern struct domain inet6domain;
    141 
    142 u_char ip6_protox[IPPROTO_MAX];
    143 static int ip6qmaxlen = IFQ_MAXLEN;
    144 struct in6_ifaddr *in6_ifaddr;
    145 struct ifqueue ip6intrq;
    146 
    147 extern callout_t in6_tmpaddrtimer_ch;
    148 
    149 int ip6_forward_srcrt;			/* XXX */
    150 int ip6_sourcecheck;			/* XXX */
    151 int ip6_sourcecheck_interval;		/* XXX */
    152 
    153 #ifdef PFIL_HOOKS
    154 struct pfil_head inet6_pfil_hook;
    155 #endif
    156 
    157 percpu_t *ip6stat_percpu;
    158 
    159 static void ip6_init2(void *);
    160 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
    161 
    162 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
    163 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
    164 
    165 /*
    166  * IP6 initialization: fill in IP6 protocol switch table.
    167  * All protocols not implemented in kernel go to raw IP6 protocol handler.
    168  */
    169 void
    170 ip6_init(void)
    171 {
    172 	const struct ip6protosw *pr;
    173 	int i;
    174 
    175 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
    176 	if (pr == 0)
    177 		panic("ip6_init");
    178 	for (i = 0; i < IPPROTO_MAX; i++)
    179 		ip6_protox[i] = pr - inet6sw;
    180 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
    181 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
    182 		if (pr->pr_domain->dom_family == PF_INET6 &&
    183 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
    184 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
    185 	ip6intrq.ifq_maxlen = ip6qmaxlen;
    186 	scope6_init();
    187 	addrsel_policy_init();
    188 	nd6_init();
    189 	frag6_init();
    190 	ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
    191 
    192 	ip6_init2((void *)0);
    193 #ifdef GATEWAY
    194 	ip6flow_init(ip6_hashsize);
    195 #endif
    196 
    197 #ifdef PFIL_HOOKS
    198 	/* Register our Packet Filter hook. */
    199 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
    200 	inet6_pfil_hook.ph_af   = AF_INET6;
    201 	i = pfil_head_register(&inet6_pfil_hook);
    202 	if (i != 0)
    203 		printf("ip6_init: WARNING: unable to register pfil hook, "
    204 		    "error %d\n", i);
    205 #endif /* PFIL_HOOKS */
    206 
    207 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
    208 }
    209 
    210 static void
    211 ip6_init2(void *dummy)
    212 {
    213 
    214 	/* nd6_timer_init */
    215 	callout_init(&nd6_timer_ch, CALLOUT_MPSAFE);
    216 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
    217 
    218 	/* timer for regeneranation of temporary addresses randomize ID */
    219 	callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE);
    220 	callout_reset(&in6_tmpaddrtimer_ch,
    221 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
    222 		       ip6_temp_regen_advance) * hz,
    223 		      in6_tmpaddrtimer, NULL);
    224 }
    225 
    226 /*
    227  * IP6 input interrupt handling. Just pass the packet to ip6_input.
    228  */
    229 void
    230 ip6intr(void)
    231 {
    232 	int s;
    233 	struct mbuf *m;
    234 
    235 	mutex_enter(softnet_lock);
    236 	KERNEL_LOCK(1, NULL);
    237 	for (;;) {
    238 		s = splnet();
    239 		IF_DEQUEUE(&ip6intrq, m);
    240 		splx(s);
    241 		if (m == 0)
    242 			break;
    243 		/* drop the packet if IPv6 operation is disabled on the IF */
    244 		if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
    245 			m_freem(m);
    246 			break;
    247 		}
    248 		ip6_input(m);
    249 	}
    250 	KERNEL_UNLOCK_ONE(NULL);
    251 	mutex_exit(softnet_lock);
    252 }
    253 
    254 extern struct	route ip6_forward_rt;
    255 
    256 void
    257 ip6_input(struct mbuf *m)
    258 {
    259 	struct ip6_hdr *ip6;
    260 	int hit, off = sizeof(struct ip6_hdr), nest;
    261 	u_int32_t plen;
    262 	u_int32_t rtalert = ~0;
    263 	int nxt, ours = 0, rh_present = 0;
    264 	struct ifnet *deliverifp = NULL;
    265 	int srcrt = 0;
    266 	const struct rtentry *rt;
    267 	union {
    268 		struct sockaddr		dst;
    269 		struct sockaddr_in6	dst6;
    270 	} u;
    271 #ifdef FAST_IPSEC
    272 	struct m_tag *mtag;
    273 	struct tdb_ident *tdbi;
    274 	struct secpolicy *sp;
    275 	int s, error;
    276 #endif
    277 
    278 #ifdef IPSEC
    279 	/*
    280 	 * should the inner packet be considered authentic?
    281 	 * see comment in ah4_input().
    282 	 */
    283 	m->m_flags &= ~M_AUTHIPHDR;
    284 	m->m_flags &= ~M_AUTHIPDGM;
    285 #endif
    286 
    287 	/*
    288 	 * make sure we don't have onion peering information into m_tag.
    289 	 */
    290 	ip6_delaux(m);
    291 
    292 	/*
    293 	 * mbuf statistics
    294 	 */
    295 	if (m->m_flags & M_EXT) {
    296 		if (m->m_next)
    297 			IP6_STATINC(IP6_STAT_MEXT2M);
    298 		else
    299 			IP6_STATINC(IP6_STAT_MEXT1);
    300 	} else {
    301 #define M2MMAX	32
    302 		if (m->m_next) {
    303 			if (m->m_flags & M_LOOP) {
    304 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
    305 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX) {
    306 				IP6_STATINC(IP6_STAT_M2M +
    307 					    m->m_pkthdr.rcvif->if_index);
    308 			} else
    309 				IP6_STATINC(IP6_STAT_M2M);
    310 		} else
    311 			IP6_STATINC(IP6_STAT_M1);
    312 #undef M2MMAX
    313 	}
    314 
    315 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
    316 	IP6_STATINC(IP6_STAT_TOTAL);
    317 
    318 	/*
    319 	 * If the IPv6 header is not aligned, slurp it up into a new
    320 	 * mbuf with space for link headers, in the event we forward
    321 	 * it.  Otherwise, if it is aligned, make sure the entire base
    322 	 * IPv6 header is in the first mbuf of the chain.
    323 	 */
    324 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
    325 		struct ifnet *inifp = m->m_pkthdr.rcvif;
    326 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
    327 				  (max_linkhdr + 3) & ~3)) == NULL) {
    328 			/* XXXJRT new stat, please */
    329 			IP6_STATINC(IP6_STAT_TOOSMALL);
    330 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    331 			return;
    332 		}
    333 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
    334 		struct ifnet *inifp = m->m_pkthdr.rcvif;
    335 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    336 			IP6_STATINC(IP6_STAT_TOOSMALL);
    337 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    338 			return;
    339 		}
    340 	}
    341 
    342 	ip6 = mtod(m, struct ip6_hdr *);
    343 
    344 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    345 		IP6_STATINC(IP6_STAT_BADVERS);
    346 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    347 		goto bad;
    348 	}
    349 
    350 #if defined(IPSEC)
    351 	/* IPv6 fast forwarding is not compatible with IPsec. */
    352 	m->m_flags &= ~M_CANFASTFWD;
    353 #else
    354 	/*
    355 	 * Assume that we can create a fast-forward IP flow entry
    356 	 * based on this packet.
    357 	 */
    358 	m->m_flags |= M_CANFASTFWD;
    359 #endif
    360 
    361 #ifdef PFIL_HOOKS
    362 	/*
    363 	 * Run through list of hooks for input packets.  If there are any
    364 	 * filters which require that additional packets in the flow are
    365 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
    366 	 * Note that filters must _never_ set this flag, as another filter
    367 	 * in the list may have previously cleared it.
    368 	 */
    369 	/*
    370 	 * let ipfilter look at packet on the wire,
    371 	 * not the decapsulated packet.
    372 	 */
    373 #ifdef IPSEC
    374 	if (!ipsec_getnhist(m))
    375 #elif defined(FAST_IPSEC)
    376 	if (!ipsec_indone(m))
    377 #else
    378 	if (1)
    379 #endif
    380 	{
    381 		struct in6_addr odst;
    382 
    383 		odst = ip6->ip6_dst;
    384 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
    385 				   PFIL_IN) != 0)
    386 			return;
    387 		if (m == NULL)
    388 			return;
    389 		ip6 = mtod(m, struct ip6_hdr *);
    390 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
    391 	}
    392 #endif /* PFIL_HOOKS */
    393 
    394 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
    395 
    396 #ifdef ALTQ
    397 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
    398 		/* packet is dropped by traffic conditioner */
    399 		return;
    400 	}
    401 #endif
    402 
    403 	/*
    404 	 * Check against address spoofing/corruption.
    405 	 */
    406 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
    407 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
    408 		/*
    409 		 * XXX: "badscope" is not very suitable for a multicast source.
    410 		 */
    411 		IP6_STATINC(IP6_STAT_BADSCOPE);
    412 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    413 		goto bad;
    414 	}
    415 	/*
    416 	 * The following check is not documented in specs.  A malicious
    417 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
    418 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
    419 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
    420 	 *
    421 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
    422 	 * support IPv4-less kernel compilation, we cannot support SIIT
    423 	 * environment at all.  So, it makes more sense for us to reject any
    424 	 * malicious packets for non-SIIT environment, than try to do a
    425 	 * partial support for SIIT environment.
    426 	 */
    427 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    428 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    429 		IP6_STATINC(IP6_STAT_BADSCOPE);
    430 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    431 		goto bad;
    432 	}
    433 #if 0
    434 	/*
    435 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
    436 	 *
    437 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
    438 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
    439 	 * is revised to forbid relaying case.
    440 	 */
    441 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
    442 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
    443 		IP6_STATINC(IP6_STAT_BADSCOPE);
    444 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    445 		goto bad;
    446 	}
    447 #endif
    448 
    449 	/*
    450 	 * Disambiguate address scope zones (if there is ambiguity).
    451 	 * We first make sure that the original source or destination address
    452 	 * is not in our internal form for scoped addresses.  Such addresses
    453 	 * are not necessarily invalid spec-wise, but we cannot accept them due
    454 	 * to the usage conflict.
    455 	 * in6_setscope() then also checks and rejects the cases where src or
    456 	 * dst are the loopback address and the receiving interface
    457 	 * is not loopback.
    458 	 */
    459 	if (__predict_false(
    460 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
    461 		goto bad;
    462 	ip6 = mtod(m, struct ip6_hdr *);
    463 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
    464 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
    465 		goto bad;
    466 	}
    467 	if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
    468 	    in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
    469 		IP6_STATINC(IP6_STAT_BADSCOPE);
    470 		goto bad;
    471 	}
    472 
    473 	/*
    474 	 * Multicast check
    475 	 */
    476 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    477 	  	struct	in6_multi *in6m = 0;
    478 
    479 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
    480 		/*
    481 		 * See if we belong to the destination multicast group on the
    482 		 * arrival interface.
    483 		 */
    484 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
    485 		if (in6m)
    486 			ours = 1;
    487 		else if (!ip6_mrouter) {
    488 			uint64_t *ip6s = IP6_STAT_GETREF();
    489 			ip6s[IP6_STAT_NOTMEMBER]++;
    490 			ip6s[IP6_STAT_CANTFORWARD]++;
    491 			IP6_STAT_PUTREF();
    492 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    493 			goto bad;
    494 		}
    495 		deliverifp = m->m_pkthdr.rcvif;
    496 		goto hbhcheck;
    497 	}
    498 
    499 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
    500 
    501 	/*
    502 	 *  Unicast check
    503 	 */
    504 	rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit);
    505 	if (hit)
    506 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
    507 	else
    508 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
    509 
    510 #define rt6_getkey(__rt) satocsin6(rt_getkey(__rt))
    511 
    512 	/*
    513 	 * Accept the packet if the forwarding interface to the destination
    514 	 * according to the routing table is the loopback interface,
    515 	 * unless the associated route has a gateway.
    516 	 * Note that this approach causes to accept a packet if there is a
    517 	 * route to the loopback interface for the destination of the packet.
    518 	 * But we think it's even useful in some situations, e.g. when using
    519 	 * a special daemon which wants to intercept the packet.
    520 	 */
    521 	if (rt != NULL &&
    522 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
    523 	    !(rt->rt_flags & RTF_CLONED) &&
    524 #if 0
    525 	    /*
    526 	     * The check below is redundant since the comparison of
    527 	     * the destination and the key of the rtentry has
    528 	     * already done through looking up the routing table.
    529 	     */
    530 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(rt)->sin6_addr) &&
    531 #endif
    532 	    rt->rt_ifp->if_type == IFT_LOOP) {
    533 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
    534 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
    535 			m->m_flags |= M_ANYCAST6;
    536 		/*
    537 		 * packets to a tentative, duplicated, or somehow invalid
    538 		 * address must not be accepted.
    539 		 */
    540 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
    541 			/* this address is ready */
    542 			ours = 1;
    543 			deliverifp = ia6->ia_ifp;	/* correct? */
    544 			goto hbhcheck;
    545 		} else {
    546 			/* address is not ready, so discard the packet. */
    547 			nd6log((LOG_INFO,
    548 			    "ip6_input: packet to an unready address %s->%s\n",
    549 			    ip6_sprintf(&ip6->ip6_src),
    550 			    ip6_sprintf(&ip6->ip6_dst)));
    551 
    552 			goto bad;
    553 		}
    554 	}
    555 
    556 	/*
    557 	 * FAITH (Firewall Aided Internet Translator)
    558 	 */
    559 #if defined(NFAITH) && 0 < NFAITH
    560 	if (ip6_keepfaith) {
    561 		if (rt != NULL && rt->rt_ifp != NULL &&
    562 		    rt->rt_ifp->if_type == IFT_FAITH) {
    563 			/* XXX do we need more sanity checks? */
    564 			ours = 1;
    565 			deliverifp = rt->rt_ifp; /* faith */
    566 			goto hbhcheck;
    567 		}
    568 	}
    569 #endif
    570 
    571 #if 0
    572     {
    573 	/*
    574 	 * Last resort: check in6_ifaddr for incoming interface.
    575 	 * The code is here until I update the "goto ours hack" code above
    576 	 * working right.
    577 	 */
    578 	struct ifaddr *ifa;
    579 	IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
    580 		if (ifa->ifa_addr == NULL)
    581 			continue;	/* just for safety */
    582 		if (ifa->ifa_addr->sa_family != AF_INET6)
    583 			continue;
    584 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
    585 			ours = 1;
    586 			deliverifp = ifa->ifa_ifp;
    587 			goto hbhcheck;
    588 		}
    589 	}
    590     }
    591 #endif
    592 
    593 	/*
    594 	 * Now there is no reason to process the packet if it's not our own
    595 	 * and we're not a router.
    596 	 */
    597 	if (!ip6_forwarding) {
    598 		IP6_STATINC(IP6_STAT_CANTFORWARD);
    599 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    600 		goto bad;
    601 	}
    602 
    603   hbhcheck:
    604 	/*
    605 	 * record address information into m_tag, if we don't have one yet.
    606 	 * note that we are unable to record it, if the address is not listed
    607 	 * as our interface address (e.g. multicast addresses, addresses
    608 	 * within FAITH prefixes and such).
    609 	 */
    610 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
    611 		struct in6_ifaddr *ia6;
    612 
    613 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    614 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
    615 			/*
    616 			 * XXX maybe we should drop the packet here,
    617 			 * as we could not provide enough information
    618 			 * to the upper layers.
    619 			 */
    620 		}
    621 	}
    622 
    623 	/*
    624 	 * Process Hop-by-Hop options header if it's contained.
    625 	 * m may be modified in ip6_hopopts_input().
    626 	 * If a JumboPayload option is included, plen will also be modified.
    627 	 */
    628 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
    629 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
    630 		struct ip6_hbh *hbh;
    631 
    632 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
    633 #if 0	/*touches NULL pointer*/
    634 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    635 #endif
    636 			return;	/* m have already been freed */
    637 		}
    638 
    639 		/* adjust pointer */
    640 		ip6 = mtod(m, struct ip6_hdr *);
    641 
    642 		/*
    643 		 * if the payload length field is 0 and the next header field
    644 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
    645 		 * option MUST be included.
    646 		 */
    647 		if (ip6->ip6_plen == 0 && plen == 0) {
    648 			/*
    649 			 * Note that if a valid jumbo payload option is
    650 			 * contained, ip6_hopopts_input() must set a valid
    651 			 * (non-zero) payload length to the variable plen.
    652 			 */
    653 			IP6_STATINC(IP6_STAT_BADOPTIONS);
    654 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    655 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    656 			icmp6_error(m, ICMP6_PARAM_PROB,
    657 				    ICMP6_PARAMPROB_HEADER,
    658 				    (char *)&ip6->ip6_plen - (char *)ip6);
    659 			return;
    660 		}
    661 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    662 			sizeof(struct ip6_hbh));
    663 		if (hbh == NULL) {
    664 			IP6_STATINC(IP6_STAT_TOOSHORT);
    665 			return;
    666 		}
    667 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
    668 		nxt = hbh->ip6h_nxt;
    669 
    670 		/*
    671 		 * accept the packet if a router alert option is included
    672 		 * and we act as an IPv6 router.
    673 		 */
    674 		if (rtalert != ~0 && ip6_forwarding)
    675 			ours = 1;
    676 	} else
    677 		nxt = ip6->ip6_nxt;
    678 
    679 	/*
    680 	 * Check that the amount of data in the buffers
    681 	 * is as at least much as the IPv6 header would have us expect.
    682 	 * Trim mbufs if longer than we expect.
    683 	 * Drop packet if shorter than we expect.
    684 	 */
    685 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
    686 		IP6_STATINC(IP6_STAT_TOOSHORT);
    687 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    688 		goto bad;
    689 	}
    690 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
    691 		if (m->m_len == m->m_pkthdr.len) {
    692 			m->m_len = sizeof(struct ip6_hdr) + plen;
    693 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
    694 		} else
    695 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
    696 	}
    697 
    698 	/*
    699 	 * Forward if desirable.
    700 	 */
    701 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    702 		/*
    703 		 * If we are acting as a multicast router, all
    704 		 * incoming multicast packets are passed to the
    705 		 * kernel-level multicast forwarding function.
    706 		 * The packet is returned (relatively) intact; if
    707 		 * ip6_mforward() returns a non-zero value, the packet
    708 		 * must be discarded, else it may be accepted below.
    709 		 */
    710 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
    711 			IP6_STATINC(IP6_STAT_CANTFORWARD);
    712 			m_freem(m);
    713 			return;
    714 		}
    715 		if (!ours) {
    716 			m_freem(m);
    717 			return;
    718 		}
    719 	} else if (!ours) {
    720 		ip6_forward(m, srcrt);
    721 		return;
    722 	}
    723 
    724 	ip6 = mtod(m, struct ip6_hdr *);
    725 
    726 	/*
    727 	 * Malicious party may be able to use IPv4 mapped addr to confuse
    728 	 * tcp/udp stack and bypass security checks (act as if it was from
    729 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
    730 	 *
    731 	 * For SIIT end node behavior, you may want to disable the check.
    732 	 * However, you will  become vulnerable to attacks using IPv4 mapped
    733 	 * source.
    734 	 */
    735 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    736 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    737 		IP6_STATINC(IP6_STAT_BADSCOPE);
    738 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    739 		goto bad;
    740 	}
    741 
    742 	/*
    743 	 * Tell launch routine the next header
    744 	 */
    745 #ifdef IFA_STATS
    746 	if (deliverifp != NULL) {
    747 		struct in6_ifaddr *ia6;
    748 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    749 		if (ia6)
    750 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
    751 	}
    752 #endif
    753 	IP6_STATINC(IP6_STAT_DELIVERED);
    754 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
    755 	nest = 0;
    756 
    757 	rh_present = 0;
    758 	while (nxt != IPPROTO_DONE) {
    759 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
    760 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
    761 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    762 			goto bad;
    763 		}
    764 
    765 		/*
    766 		 * protection against faulty packet - there should be
    767 		 * more sanity checks in header chain processing.
    768 		 */
    769 		if (m->m_pkthdr.len < off) {
    770 			IP6_STATINC(IP6_STAT_TOOSHORT);
    771 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    772 			goto bad;
    773 		}
    774 
    775 		if (nxt == IPPROTO_ROUTING) {
    776 			if (rh_present++) {
    777 				in6_ifstat_inc(m->m_pkthdr.rcvif,
    778 				    ifs6_in_hdrerr);
    779 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    780 				goto bad;
    781 			}
    782 		}
    783 
    784 #ifdef IPSEC
    785 		/*
    786 		 * enforce IPsec policy checking if we are seeing last header.
    787 		 * note that we do not visit this with protocols with pcb layer
    788 		 * code - like udp/tcp/raw ip.
    789 		 */
    790 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
    791 		    ipsec6_in_reject(m, NULL)) {
    792 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
    793 			goto bad;
    794 		}
    795 #endif
    796 #ifdef FAST_IPSEC
    797 	/*
    798 	 * enforce IPsec policy checking if we are seeing last header.
    799 	 * note that we do not visit this with protocols with pcb layer
    800 	 * code - like udp/tcp/raw ip.
    801 	 */
    802 	if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) {
    803 		/*
    804 		 * Check if the packet has already had IPsec processing
    805 		 * done.  If so, then just pass it along.  This tag gets
    806 		 * set during AH, ESP, etc. input handling, before the
    807 		 * packet is returned to the ip input queue for delivery.
    808 		 */
    809 		mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
    810 		s = splsoftnet();
    811 		if (mtag != NULL) {
    812 			tdbi = (struct tdb_ident *)(mtag + 1);
    813 			sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
    814 		} else {
    815 			sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
    816 									IP_FORWARDING, &error);
    817 		}
    818 		if (sp != NULL) {
    819 			/*
    820 			 * Check security policy against packet attributes.
    821 			 */
    822 			error = ipsec_in_reject(sp, m);
    823 			KEY_FREESP(&sp);
    824 		} else {
    825 			/* XXX error stat??? */
    826 			error = EINVAL;
    827 			DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/
    828 		}
    829 		splx(s);
    830 		if (error)
    831 			goto bad;
    832 	}
    833 #endif /* FAST_IPSEC */
    834 
    835 
    836 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
    837 	}
    838 	return;
    839  bad:
    840 	m_freem(m);
    841 }
    842 
    843 /*
    844  * set/grab in6_ifaddr correspond to IPv6 destination address.
    845  */
    846 static struct m_tag *
    847 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
    848 {
    849 	struct m_tag *mtag;
    850 
    851 	mtag = ip6_addaux(m);
    852 	if (mtag != NULL) {
    853 		struct ip6aux *ip6a;
    854 
    855 		ip6a = (struct ip6aux *)(mtag + 1);
    856 		in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id);
    857 		ip6a->ip6a_src = ia->ia_addr.sin6_addr;
    858 		ip6a->ip6a_flags = ia->ia6_flags;
    859 	}
    860 	return mtag;	/* NULL if failed to set */
    861 }
    862 
    863 const struct ip6aux *
    864 ip6_getdstifaddr(struct mbuf *m)
    865 {
    866 	struct m_tag *mtag;
    867 
    868 	mtag = ip6_findaux(m);
    869 	if (mtag != NULL)
    870 		return (struct ip6aux *)(mtag + 1);
    871 	else
    872 		return NULL;
    873 }
    874 
    875 /*
    876  * Hop-by-Hop options header processing. If a valid jumbo payload option is
    877  * included, the real payload length will be stored in plenp.
    878  *
    879  * rtalertp - XXX: should be stored more smart way
    880  */
    881 static int
    882 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
    883 	struct mbuf **mp, int *offp)
    884 {
    885 	struct mbuf *m = *mp;
    886 	int off = *offp, hbhlen;
    887 	struct ip6_hbh *hbh;
    888 
    889 	/* validation of the length of the header */
    890 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
    891 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
    892 	if (hbh == NULL) {
    893 		IP6_STATINC(IP6_STAT_TOOSHORT);
    894 		return -1;
    895 	}
    896 	hbhlen = (hbh->ip6h_len + 1) << 3;
    897 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    898 		hbhlen);
    899 	if (hbh == NULL) {
    900 		IP6_STATINC(IP6_STAT_TOOSHORT);
    901 		return -1;
    902 	}
    903 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
    904 	off += hbhlen;
    905 	hbhlen -= sizeof(struct ip6_hbh);
    906 
    907 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
    908 				hbhlen, rtalertp, plenp) < 0)
    909 		return (-1);
    910 
    911 	*offp = off;
    912 	*mp = m;
    913 	return (0);
    914 }
    915 
    916 /*
    917  * Search header for all Hop-by-hop options and process each option.
    918  * This function is separate from ip6_hopopts_input() in order to
    919  * handle a case where the sending node itself process its hop-by-hop
    920  * options header. In such a case, the function is called from ip6_output().
    921  *
    922  * The function assumes that hbh header is located right after the IPv6 header
    923  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
    924  * opthead + hbhlen is located in continuous memory region.
    925  */
    926 int
    927 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
    928 	u_int32_t *rtalertp, u_int32_t *plenp)
    929 {
    930 	struct ip6_hdr *ip6;
    931 	int optlen = 0;
    932 	u_int8_t *opt = opthead;
    933 	u_int16_t rtalert_val;
    934 	u_int32_t jumboplen;
    935 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
    936 
    937 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
    938 		switch (*opt) {
    939 		case IP6OPT_PAD1:
    940 			optlen = 1;
    941 			break;
    942 		case IP6OPT_PADN:
    943 			if (hbhlen < IP6OPT_MINLEN) {
    944 				IP6_STATINC(IP6_STAT_TOOSMALL);
    945 				goto bad;
    946 			}
    947 			optlen = *(opt + 1) + 2;
    948 			break;
    949 		case IP6OPT_RTALERT:
    950 			/* XXX may need check for alignment */
    951 			if (hbhlen < IP6OPT_RTALERT_LEN) {
    952 				IP6_STATINC(IP6_STAT_TOOSMALL);
    953 				goto bad;
    954 			}
    955 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
    956 				/* XXX stat */
    957 				icmp6_error(m, ICMP6_PARAM_PROB,
    958 				    ICMP6_PARAMPROB_HEADER,
    959 				    erroff + opt + 1 - opthead);
    960 				return (-1);
    961 			}
    962 			optlen = IP6OPT_RTALERT_LEN;
    963 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
    964 			*rtalertp = ntohs(rtalert_val);
    965 			break;
    966 		case IP6OPT_JUMBO:
    967 			/* XXX may need check for alignment */
    968 			if (hbhlen < IP6OPT_JUMBO_LEN) {
    969 				IP6_STATINC(IP6_STAT_TOOSMALL);
    970 				goto bad;
    971 			}
    972 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
    973 				/* XXX stat */
    974 				icmp6_error(m, ICMP6_PARAM_PROB,
    975 				    ICMP6_PARAMPROB_HEADER,
    976 				    erroff + opt + 1 - opthead);
    977 				return (-1);
    978 			}
    979 			optlen = IP6OPT_JUMBO_LEN;
    980 
    981 			/*
    982 			 * IPv6 packets that have non 0 payload length
    983 			 * must not contain a jumbo payload option.
    984 			 */
    985 			ip6 = mtod(m, struct ip6_hdr *);
    986 			if (ip6->ip6_plen) {
    987 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    988 				icmp6_error(m, ICMP6_PARAM_PROB,
    989 				    ICMP6_PARAMPROB_HEADER,
    990 				    erroff + opt - opthead);
    991 				return (-1);
    992 			}
    993 
    994 			/*
    995 			 * We may see jumbolen in unaligned location, so
    996 			 * we'd need to perform bcopy().
    997 			 */
    998 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
    999 			jumboplen = (u_int32_t)htonl(jumboplen);
   1000 
   1001 #if 1
   1002 			/*
   1003 			 * if there are multiple jumbo payload options,
   1004 			 * *plenp will be non-zero and the packet will be
   1005 			 * rejected.
   1006 			 * the behavior may need some debate in ipngwg -
   1007 			 * multiple options does not make sense, however,
   1008 			 * there's no explicit mention in specification.
   1009 			 */
   1010 			if (*plenp != 0) {
   1011 				IP6_STATINC(IP6_STAT_BADOPTIONS);
   1012 				icmp6_error(m, ICMP6_PARAM_PROB,
   1013 				    ICMP6_PARAMPROB_HEADER,
   1014 				    erroff + opt + 2 - opthead);
   1015 				return (-1);
   1016 			}
   1017 #endif
   1018 
   1019 			/*
   1020 			 * jumbo payload length must be larger than 65535.
   1021 			 */
   1022 			if (jumboplen <= IPV6_MAXPACKET) {
   1023 				IP6_STATINC(IP6_STAT_BADOPTIONS);
   1024 				icmp6_error(m, ICMP6_PARAM_PROB,
   1025 				    ICMP6_PARAMPROB_HEADER,
   1026 				    erroff + opt + 2 - opthead);
   1027 				return (-1);
   1028 			}
   1029 			*plenp = jumboplen;
   1030 
   1031 			break;
   1032 		default:		/* unknown option */
   1033 			if (hbhlen < IP6OPT_MINLEN) {
   1034 				IP6_STATINC(IP6_STAT_TOOSMALL);
   1035 				goto bad;
   1036 			}
   1037 			optlen = ip6_unknown_opt(opt, m,
   1038 			    erroff + opt - opthead);
   1039 			if (optlen == -1)
   1040 				return (-1);
   1041 			optlen += 2;
   1042 			break;
   1043 		}
   1044 	}
   1045 
   1046 	return (0);
   1047 
   1048   bad:
   1049 	m_freem(m);
   1050 	return (-1);
   1051 }
   1052 
   1053 /*
   1054  * Unknown option processing.
   1055  * The third argument `off' is the offset from the IPv6 header to the option,
   1056  * which is necessary if the IPv6 header the and option header and IPv6 header
   1057  * is not continuous in order to return an ICMPv6 error.
   1058  */
   1059 int
   1060 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
   1061 {
   1062 	struct ip6_hdr *ip6;
   1063 
   1064 	switch (IP6OPT_TYPE(*optp)) {
   1065 	case IP6OPT_TYPE_SKIP: /* ignore the option */
   1066 		return ((int)*(optp + 1));
   1067 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
   1068 		m_freem(m);
   1069 		return (-1);
   1070 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
   1071 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1072 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
   1073 		return (-1);
   1074 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
   1075 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1076 		ip6 = mtod(m, struct ip6_hdr *);
   1077 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
   1078 		    (m->m_flags & (M_BCAST|M_MCAST)))
   1079 			m_freem(m);
   1080 		else
   1081 			icmp6_error(m, ICMP6_PARAM_PROB,
   1082 				    ICMP6_PARAMPROB_OPTION, off);
   1083 		return (-1);
   1084 	}
   1085 
   1086 	m_freem(m);		/* XXX: NOTREACHED */
   1087 	return (-1);
   1088 }
   1089 
   1090 /*
   1091  * Create the "control" list for this pcb.
   1092  *
   1093  * The routine will be called from upper layer handlers like tcp6_input().
   1094  * Thus the routine assumes that the caller (tcp6_input) have already
   1095  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
   1096  * very first mbuf on the mbuf chain.
   1097  * We may want to add some infinite loop prevention or sanity checks for safety.
   1098  * (This applies only when you are using KAME mbuf chain restriction, i.e.
   1099  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
   1100  */
   1101 void
   1102 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
   1103 	struct ip6_hdr *ip6, struct mbuf *m)
   1104 {
   1105 #ifdef RFC2292
   1106 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1107 #else
   1108 #define IS2292(x, y)	(y)
   1109 #endif
   1110 
   1111 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP
   1112 #ifdef SO_OTIMESTAMP
   1113 	    || in6p->in6p_socket->so_options & SO_OTIMESTAMP
   1114 #endif
   1115 	) {
   1116 		struct timeval tv;
   1117 
   1118 		microtime(&tv);
   1119 #ifdef SO_OTIMESTAMP
   1120 		if (in6p->in6p_socket->so_options & SO_OTIMESTAMP) {
   1121 			struct timeval50 tv50;
   1122 			timeval_to_timeval50(&tv, &tv50);
   1123 			*mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
   1124 			    SCM_OTIMESTAMP, SOL_SOCKET);
   1125 		} else
   1126 #endif
   1127 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
   1128 		    SCM_TIMESTAMP, SOL_SOCKET);
   1129 		if (*mp)
   1130 			mp = &(*mp)->m_next;
   1131 	}
   1132 
   1133 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
   1134 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
   1135 		return;
   1136 
   1137 	/* RFC 2292 sec. 5 */
   1138 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
   1139 		struct in6_pktinfo pi6;
   1140 
   1141 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
   1142 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
   1143 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
   1144 		    m->m_pkthdr.rcvif->if_index : 0;
   1145 		*mp = sbcreatecontrol((void *) &pi6,
   1146 		    sizeof(struct in6_pktinfo),
   1147 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
   1148 		if (*mp)
   1149 			mp = &(*mp)->m_next;
   1150 	}
   1151 
   1152 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
   1153 		int hlim = ip6->ip6_hlim & 0xff;
   1154 
   1155 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
   1156 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
   1157 		if (*mp)
   1158 			mp = &(*mp)->m_next;
   1159 	}
   1160 
   1161 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
   1162 		u_int32_t flowinfo;
   1163 		int tclass;
   1164 
   1165 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
   1166 		flowinfo >>= 20;
   1167 
   1168 		tclass = flowinfo & 0xff;
   1169 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
   1170 		    IPV6_TCLASS, IPPROTO_IPV6);
   1171 
   1172 		if (*mp)
   1173 			mp = &(*mp)->m_next;
   1174 	}
   1175 
   1176 	/*
   1177 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
   1178 	 * privilege for the option (see ip6_ctloutput), but it might be too
   1179 	 * strict, since there might be some hop-by-hop options which can be
   1180 	 * returned to normal user.
   1181 	 * See also RFC3542 section 8 (or RFC2292 section 6).
   1182 	 */
   1183 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
   1184 		/*
   1185 		 * Check if a hop-by-hop options header is contatined in the
   1186 		 * received packet, and if so, store the options as ancillary
   1187 		 * data. Note that a hop-by-hop options header must be
   1188 		 * just after the IPv6 header, which fact is assured through
   1189 		 * the IPv6 input processing.
   1190 		 */
   1191 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1192 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
   1193 			struct ip6_hbh *hbh;
   1194 			int hbhlen;
   1195 			struct mbuf *ext;
   1196 
   1197 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
   1198 			    xip6->ip6_nxt);
   1199 			if (ext == NULL) {
   1200 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1201 				return;
   1202 			}
   1203 			hbh = mtod(ext, struct ip6_hbh *);
   1204 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1205 			if (hbhlen != ext->m_len) {
   1206 				m_freem(ext);
   1207 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1208 				return;
   1209 			}
   1210 
   1211 			/*
   1212 			 * XXX: We copy whole the header even if a jumbo
   1213 			 * payload option is included, which option is to
   1214 			 * be removed before returning in the RFC 2292.
   1215 			 * Note: this constraint is removed in RFC3542.
   1216 			 */
   1217 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
   1218 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
   1219 			    IPPROTO_IPV6);
   1220 			if (*mp)
   1221 				mp = &(*mp)->m_next;
   1222 			m_freem(ext);
   1223 		}
   1224 	}
   1225 
   1226 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
   1227 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
   1228 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1229 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
   1230 
   1231 		/*
   1232 		 * Search for destination options headers or routing
   1233 		 * header(s) through the header chain, and stores each
   1234 		 * header as ancillary data.
   1235 		 * Note that the order of the headers remains in
   1236 		 * the chain of ancillary data.
   1237 		 */
   1238 		for (;;) {	/* is explicit loop prevention necessary? */
   1239 			struct ip6_ext *ip6e = NULL;
   1240 			int elen;
   1241 			struct mbuf *ext = NULL;
   1242 
   1243 			/*
   1244 			 * if it is not an extension header, don't try to
   1245 			 * pull it from the chain.
   1246 			 */
   1247 			switch (nxt) {
   1248 			case IPPROTO_DSTOPTS:
   1249 			case IPPROTO_ROUTING:
   1250 			case IPPROTO_HOPOPTS:
   1251 			case IPPROTO_AH: /* is it possible? */
   1252 				break;
   1253 			default:
   1254 				goto loopend;
   1255 			}
   1256 
   1257 			ext = ip6_pullexthdr(m, off, nxt);
   1258 			if (ext == NULL) {
   1259 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1260 				return;
   1261 			}
   1262 			ip6e = mtod(ext, struct ip6_ext *);
   1263 			if (nxt == IPPROTO_AH)
   1264 				elen = (ip6e->ip6e_len + 2) << 2;
   1265 			else
   1266 				elen = (ip6e->ip6e_len + 1) << 3;
   1267 			if (elen != ext->m_len) {
   1268 				m_freem(ext);
   1269 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1270 				return;
   1271 			}
   1272 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
   1273 
   1274 			switch (nxt) {
   1275 			case IPPROTO_DSTOPTS:
   1276 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
   1277 					break;
   1278 
   1279 				*mp = sbcreatecontrol((void *)ip6e, elen,
   1280 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
   1281 				    IPPROTO_IPV6);
   1282 				if (*mp)
   1283 					mp = &(*mp)->m_next;
   1284 				break;
   1285 
   1286 			case IPPROTO_ROUTING:
   1287 				if (!in6p->in6p_flags & IN6P_RTHDR)
   1288 					break;
   1289 
   1290 				*mp = sbcreatecontrol((void *)ip6e, elen,
   1291 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
   1292 				    IPPROTO_IPV6);
   1293 				if (*mp)
   1294 					mp = &(*mp)->m_next;
   1295 				break;
   1296 
   1297 			case IPPROTO_HOPOPTS:
   1298 			case IPPROTO_AH: /* is it possible? */
   1299 				break;
   1300 
   1301 			default:
   1302 				/*
   1303 			 	 * other cases have been filtered in the above.
   1304 				 * none will visit this case.  here we supply
   1305 				 * the code just in case (nxt overwritten or
   1306 				 * other cases).
   1307 				 */
   1308 				m_freem(ext);
   1309 				goto loopend;
   1310 
   1311 			}
   1312 
   1313 			/* proceed with the next header. */
   1314 			off += elen;
   1315 			nxt = ip6e->ip6e_nxt;
   1316 			ip6e = NULL;
   1317 			m_freem(ext);
   1318 			ext = NULL;
   1319 		}
   1320 	  loopend:
   1321 	  	;
   1322 	}
   1323 }
   1324 #undef IS2292
   1325 
   1326 
   1327 void
   1328 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
   1329     uint32_t *mtu)
   1330 {
   1331 	struct socket *so;
   1332 	struct mbuf *m_mtu;
   1333 	struct ip6_mtuinfo mtuctl;
   1334 
   1335 	so = in6p->in6p_socket;
   1336 
   1337 	if (mtu == NULL)
   1338 		return;
   1339 
   1340 #ifdef DIAGNOSTIC
   1341 	if (so == NULL)		/* I believe this is impossible */
   1342 		panic("ip6_notify_pmtu: socket is NULL");
   1343 #endif
   1344 
   1345 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
   1346 	mtuctl.ip6m_mtu = *mtu;
   1347 	mtuctl.ip6m_addr = *dst;
   1348 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
   1349 		return;
   1350 
   1351 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
   1352 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
   1353 		return;
   1354 
   1355 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
   1356 	    == 0) {
   1357 		m_freem(m_mtu);
   1358 		/* XXX: should count statistics */
   1359 	} else
   1360 		sorwakeup(so);
   1361 
   1362 	return;
   1363 }
   1364 
   1365 /*
   1366  * pull single extension header from mbuf chain.  returns single mbuf that
   1367  * contains the result, or NULL on error.
   1368  */
   1369 static struct mbuf *
   1370 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
   1371 {
   1372 	struct ip6_ext ip6e;
   1373 	size_t elen;
   1374 	struct mbuf *n;
   1375 
   1376 #ifdef DIAGNOSTIC
   1377 	switch (nxt) {
   1378 	case IPPROTO_DSTOPTS:
   1379 	case IPPROTO_ROUTING:
   1380 	case IPPROTO_HOPOPTS:
   1381 	case IPPROTO_AH: /* is it possible? */
   1382 		break;
   1383 	default:
   1384 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
   1385 	}
   1386 #endif
   1387 
   1388 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1389 	if (nxt == IPPROTO_AH)
   1390 		elen = (ip6e.ip6e_len + 2) << 2;
   1391 	else
   1392 		elen = (ip6e.ip6e_len + 1) << 3;
   1393 
   1394 	MGET(n, M_DONTWAIT, MT_DATA);
   1395 	if (n && elen >= MLEN) {
   1396 		MCLGET(n, M_DONTWAIT);
   1397 		if ((n->m_flags & M_EXT) == 0) {
   1398 			m_free(n);
   1399 			n = NULL;
   1400 		}
   1401 	}
   1402 	if (!n)
   1403 		return NULL;
   1404 
   1405 	n->m_len = 0;
   1406 	if (elen >= M_TRAILINGSPACE(n)) {
   1407 		m_free(n);
   1408 		return NULL;
   1409 	}
   1410 
   1411 	m_copydata(m, off, elen, mtod(n, void *));
   1412 	n->m_len = elen;
   1413 	return n;
   1414 }
   1415 
   1416 /*
   1417  * Get pointer to the previous header followed by the header
   1418  * currently processed.
   1419  * XXX: This function supposes that
   1420  *	M includes all headers,
   1421  *	the next header field and the header length field of each header
   1422  *	are valid, and
   1423  *	the sum of each header length equals to OFF.
   1424  * Because of these assumptions, this function must be called very
   1425  * carefully. Moreover, it will not be used in the near future when
   1426  * we develop `neater' mechanism to process extension headers.
   1427  */
   1428 u_int8_t *
   1429 ip6_get_prevhdr(struct mbuf *m, int off)
   1430 {
   1431 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1432 
   1433 	if (off == sizeof(struct ip6_hdr))
   1434 		return (&ip6->ip6_nxt);
   1435 	else {
   1436 		int len, nxt;
   1437 		struct ip6_ext *ip6e = NULL;
   1438 
   1439 		nxt = ip6->ip6_nxt;
   1440 		len = sizeof(struct ip6_hdr);
   1441 		while (len < off) {
   1442 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
   1443 
   1444 			switch (nxt) {
   1445 			case IPPROTO_FRAGMENT:
   1446 				len += sizeof(struct ip6_frag);
   1447 				break;
   1448 			case IPPROTO_AH:
   1449 				len += (ip6e->ip6e_len + 2) << 2;
   1450 				break;
   1451 			default:
   1452 				len += (ip6e->ip6e_len + 1) << 3;
   1453 				break;
   1454 			}
   1455 			nxt = ip6e->ip6e_nxt;
   1456 		}
   1457 		if (ip6e)
   1458 			return (&ip6e->ip6e_nxt);
   1459 		else
   1460 			return NULL;
   1461 	}
   1462 }
   1463 
   1464 /*
   1465  * get next header offset.  m will be retained.
   1466  */
   1467 int
   1468 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1469 {
   1470 	struct ip6_hdr ip6;
   1471 	struct ip6_ext ip6e;
   1472 	struct ip6_frag fh;
   1473 
   1474 	/* just in case */
   1475 	if (m == NULL)
   1476 		panic("ip6_nexthdr: m == NULL");
   1477 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
   1478 		return -1;
   1479 
   1480 	switch (proto) {
   1481 	case IPPROTO_IPV6:
   1482 		/* do not chase beyond intermediate IPv6 headers */
   1483 		if (off != 0)
   1484 			return -1;
   1485 		if (m->m_pkthdr.len < off + sizeof(ip6))
   1486 			return -1;
   1487 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
   1488 		if (nxtp)
   1489 			*nxtp = ip6.ip6_nxt;
   1490 		off += sizeof(ip6);
   1491 		return off;
   1492 
   1493 	case IPPROTO_FRAGMENT:
   1494 		/*
   1495 		 * terminate parsing if it is not the first fragment,
   1496 		 * it does not make sense to parse through it.
   1497 		 */
   1498 		if (m->m_pkthdr.len < off + sizeof(fh))
   1499 			return -1;
   1500 		m_copydata(m, off, sizeof(fh), (void *)&fh);
   1501 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
   1502 			return -1;
   1503 		if (nxtp)
   1504 			*nxtp = fh.ip6f_nxt;
   1505 		off += sizeof(struct ip6_frag);
   1506 		return off;
   1507 
   1508 	case IPPROTO_AH:
   1509 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1510 			return -1;
   1511 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1512 		if (nxtp)
   1513 			*nxtp = ip6e.ip6e_nxt;
   1514 		off += (ip6e.ip6e_len + 2) << 2;
   1515 		if (m->m_pkthdr.len < off)
   1516 			return -1;
   1517 		return off;
   1518 
   1519 	case IPPROTO_HOPOPTS:
   1520 	case IPPROTO_ROUTING:
   1521 	case IPPROTO_DSTOPTS:
   1522 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1523 			return -1;
   1524 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1525 		if (nxtp)
   1526 			*nxtp = ip6e.ip6e_nxt;
   1527 		off += (ip6e.ip6e_len + 1) << 3;
   1528 		if (m->m_pkthdr.len < off)
   1529 			return -1;
   1530 		return off;
   1531 
   1532 	case IPPROTO_NONE:
   1533 	case IPPROTO_ESP:
   1534 	case IPPROTO_IPCOMP:
   1535 		/* give up */
   1536 		return -1;
   1537 
   1538 	default:
   1539 		return -1;
   1540 	}
   1541 }
   1542 
   1543 /*
   1544  * get offset for the last header in the chain.  m will be kept untainted.
   1545  */
   1546 int
   1547 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1548 {
   1549 	int newoff;
   1550 	int nxt;
   1551 
   1552 	if (!nxtp) {
   1553 		nxt = -1;
   1554 		nxtp = &nxt;
   1555 	}
   1556 	for (;;) {
   1557 		newoff = ip6_nexthdr(m, off, proto, nxtp);
   1558 		if (newoff < 0)
   1559 			return off;
   1560 		else if (newoff < off)
   1561 			return -1;	/* invalid */
   1562 		else if (newoff == off)
   1563 			return newoff;
   1564 
   1565 		off = newoff;
   1566 		proto = *nxtp;
   1567 	}
   1568 }
   1569 
   1570 struct m_tag *
   1571 ip6_addaux(struct mbuf *m)
   1572 {
   1573 	struct m_tag *mtag;
   1574 
   1575 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1576 	if (!mtag) {
   1577 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
   1578 		    M_NOWAIT);
   1579 		if (mtag) {
   1580 			m_tag_prepend(m, mtag);
   1581 			memset(mtag + 1, 0, sizeof(struct ip6aux));
   1582 		}
   1583 	}
   1584 	return mtag;
   1585 }
   1586 
   1587 struct m_tag *
   1588 ip6_findaux(struct mbuf *m)
   1589 {
   1590 	struct m_tag *mtag;
   1591 
   1592 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1593 	return mtag;
   1594 }
   1595 
   1596 void
   1597 ip6_delaux(struct mbuf *m)
   1598 {
   1599 	struct m_tag *mtag;
   1600 
   1601 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1602 	if (mtag)
   1603 		m_tag_delete(m, mtag);
   1604 }
   1605 
   1606 #ifdef GATEWAY
   1607 /*
   1608  * sysctl helper routine for net.inet.ip6.maxflows. Since
   1609  * we could reduce this value, call ip6flow_reap();
   1610  */
   1611 static int
   1612 sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS)
   1613 {
   1614 	int error;
   1615 
   1616 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
   1617 	if (error || newp == NULL)
   1618 		return (error);
   1619 
   1620 	mutex_enter(softnet_lock);
   1621 	KERNEL_LOCK(1, NULL);
   1622 
   1623 	ip6flow_reap(0);
   1624 
   1625 	KERNEL_UNLOCK_ONE(NULL);
   1626 	mutex_exit(softnet_lock);
   1627 
   1628 	return (0);
   1629 }
   1630 
   1631 static int
   1632 sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS)
   1633 {
   1634 	int error, tmp;
   1635 	struct sysctlnode node;
   1636 
   1637 	node = *rnode;
   1638 	tmp = ip6_hashsize;
   1639 	node.sysctl_data = &tmp;
   1640 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1641 	if (error || newp == NULL)
   1642 		return (error);
   1643 
   1644 	if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
   1645 		/*
   1646 		 * Can only fail due to malloc()
   1647 		 */
   1648 		mutex_enter(softnet_lock);
   1649 		KERNEL_LOCK(1, NULL);
   1650 
   1651 		error = ip6flow_invalidate_all(tmp);
   1652 
   1653 		KERNEL_UNLOCK_ONE(NULL);
   1654 		mutex_exit(softnet_lock);
   1655 	} else {
   1656 		/*
   1657 		 * EINVAL if not a power of 2
   1658 		 */
   1659 		error = EINVAL;
   1660 	}
   1661 
   1662 	return error;
   1663 }
   1664 #endif /* GATEWAY */
   1665 
   1666 /*
   1667  * System control for IP6
   1668  */
   1669 
   1670 const u_char inet6ctlerrmap[PRC_NCMDS] = {
   1671 	0,		0,		0,		0,
   1672 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
   1673 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
   1674 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
   1675 	0,		0,		0,		0,
   1676 	ENOPROTOOPT
   1677 };
   1678 
   1679 static int
   1680 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
   1681 {
   1682 
   1683 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
   1684 }
   1685 
   1686 SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup")
   1687 {
   1688 #ifdef RFC2292
   1689 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1690 #else
   1691 #define IS2292(x, y)	(y)
   1692 #endif
   1693 
   1694 	sysctl_createv(clog, 0, NULL, NULL,
   1695 		       CTLFLAG_PERMANENT,
   1696 		       CTLTYPE_NODE, "net", NULL,
   1697 		       NULL, 0, NULL, 0,
   1698 		       CTL_NET, CTL_EOL);
   1699 	sysctl_createv(clog, 0, NULL, NULL,
   1700 		       CTLFLAG_PERMANENT,
   1701 		       CTLTYPE_NODE, "inet6",
   1702 		       SYSCTL_DESCR("PF_INET6 related settings"),
   1703 		       NULL, 0, NULL, 0,
   1704 		       CTL_NET, PF_INET6, CTL_EOL);
   1705 	sysctl_createv(clog, 0, NULL, NULL,
   1706 		       CTLFLAG_PERMANENT,
   1707 		       CTLTYPE_NODE, "ip6",
   1708 		       SYSCTL_DESCR("IPv6 related settings"),
   1709 		       NULL, 0, NULL, 0,
   1710 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
   1711 
   1712 	sysctl_createv(clog, 0, NULL, NULL,
   1713 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1714 		       CTLTYPE_INT, "forwarding",
   1715 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
   1716 		       NULL, 0, &ip6_forwarding, 0,
   1717 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1718 		       IPV6CTL_FORWARDING, CTL_EOL);
   1719 	sysctl_createv(clog, 0, NULL, NULL,
   1720 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1721 		       CTLTYPE_INT, "redirect",
   1722 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
   1723 		       NULL, 0, &ip6_sendredirects, 0,
   1724 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1725 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
   1726 	sysctl_createv(clog, 0, NULL, NULL,
   1727 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1728 		       CTLTYPE_INT, "hlim",
   1729 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
   1730 		       NULL, 0, &ip6_defhlim, 0,
   1731 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1732 		       IPV6CTL_DEFHLIM, CTL_EOL);
   1733 #ifdef notyet
   1734 	sysctl_createv(clog, 0, NULL, NULL,
   1735 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1736 		       CTLTYPE_INT, "mtu", NULL,
   1737 		       NULL, 0, &, 0,
   1738 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1739 		       IPV6CTL_DEFMTU, CTL_EOL);
   1740 #endif
   1741 #ifdef __no_idea__
   1742 	sysctl_createv(clog, 0, NULL, NULL,
   1743 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1744 		       CTLTYPE_INT, "forwsrcrt", NULL,
   1745 		       NULL, 0, &?, 0,
   1746 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1747 		       IPV6CTL_FORWSRCRT, CTL_EOL);
   1748 	sysctl_createv(clog, 0, NULL, NULL,
   1749 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1750 		       CTLTYPE_STRUCT, "mrtstats", NULL,
   1751 		       NULL, 0, &?, sizeof(?),
   1752 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1753 		       IPV6CTL_MRTSTATS, CTL_EOL);
   1754 	sysctl_createv(clog, 0, NULL, NULL,
   1755 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1756 		       CTLTYPE_?, "mrtproto", NULL,
   1757 		       NULL, 0, &?, sizeof(?),
   1758 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1759 		       IPV6CTL_MRTPROTO, CTL_EOL);
   1760 #endif
   1761 	sysctl_createv(clog, 0, NULL, NULL,
   1762 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1763 		       CTLTYPE_INT, "maxfragpackets",
   1764 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
   1765 				    "for reassembly"),
   1766 		       NULL, 0, &ip6_maxfragpackets, 0,
   1767 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1768 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
   1769 #ifdef __no_idea__
   1770 	sysctl_createv(clog, 0, NULL, NULL,
   1771 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1772 		       CTLTYPE_INT, "sourcecheck", NULL,
   1773 		       NULL, 0, &?, 0,
   1774 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1775 		       IPV6CTL_SOURCECHECK, CTL_EOL);
   1776 	sysctl_createv(clog, 0, NULL, NULL,
   1777 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1778 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
   1779 		       NULL, 0, &?, 0,
   1780 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1781 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
   1782 #endif
   1783 	sysctl_createv(clog, 0, NULL, NULL,
   1784 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1785 		       CTLTYPE_INT, "accept_rtadv",
   1786 		       SYSCTL_DESCR("Accept router advertisements"),
   1787 		       NULL, 0, &ip6_accept_rtadv, 0,
   1788 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1789 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
   1790 	sysctl_createv(clog, 0, NULL, NULL,
   1791 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1792 		       CTLTYPE_INT, "keepfaith",
   1793 		       SYSCTL_DESCR("Activate faith interface"),
   1794 		       NULL, 0, &ip6_keepfaith, 0,
   1795 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1796 		       IPV6CTL_KEEPFAITH, CTL_EOL);
   1797 	sysctl_createv(clog, 0, NULL, NULL,
   1798 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1799 		       CTLTYPE_INT, "log_interval",
   1800 		       SYSCTL_DESCR("Minumum interval between logging "
   1801 				    "unroutable packets"),
   1802 		       NULL, 0, &ip6_log_interval, 0,
   1803 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1804 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
   1805 	sysctl_createv(clog, 0, NULL, NULL,
   1806 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1807 		       CTLTYPE_INT, "hdrnestlimit",
   1808 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
   1809 		       NULL, 0, &ip6_hdrnestlimit, 0,
   1810 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1811 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
   1812 	sysctl_createv(clog, 0, NULL, NULL,
   1813 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1814 		       CTLTYPE_INT, "dad_count",
   1815 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
   1816 				    "probes to send"),
   1817 		       NULL, 0, &ip6_dad_count, 0,
   1818 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1819 		       IPV6CTL_DAD_COUNT, CTL_EOL);
   1820 	sysctl_createv(clog, 0, NULL, NULL,
   1821 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1822 		       CTLTYPE_INT, "auto_flowlabel",
   1823 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
   1824 		       NULL, 0, &ip6_auto_flowlabel, 0,
   1825 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1826 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
   1827 	sysctl_createv(clog, 0, NULL, NULL,
   1828 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1829 		       CTLTYPE_INT, "defmcasthlim",
   1830 		       SYSCTL_DESCR("Default multicast hop limit"),
   1831 		       NULL, 0, &ip6_defmcasthlim, 0,
   1832 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1833 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
   1834 #if NGIF > 0
   1835 	sysctl_createv(clog, 0, NULL, NULL,
   1836 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1837 		       CTLTYPE_INT, "gifhlim",
   1838 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
   1839 		       NULL, 0, &ip6_gif_hlim, 0,
   1840 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1841 		       IPV6CTL_GIF_HLIM, CTL_EOL);
   1842 #endif /* NGIF */
   1843 	sysctl_createv(clog, 0, NULL, NULL,
   1844 		       CTLFLAG_PERMANENT,
   1845 		       CTLTYPE_STRING, "kame_version",
   1846 		       SYSCTL_DESCR("KAME Version"),
   1847 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
   1848 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1849 		       IPV6CTL_KAME_VERSION, CTL_EOL);
   1850 	sysctl_createv(clog, 0, NULL, NULL,
   1851 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1852 		       CTLTYPE_INT, "use_deprecated",
   1853 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
   1854 				    "source addresses"),
   1855 		       NULL, 0, &ip6_use_deprecated, 0,
   1856 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1857 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
   1858 	sysctl_createv(clog, 0, NULL, NULL,
   1859 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1860 		       CTLTYPE_INT, "rr_prune", NULL,
   1861 		       NULL, 0, &ip6_rr_prune, 0,
   1862 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1863 		       IPV6CTL_RR_PRUNE, CTL_EOL);
   1864 	sysctl_createv(clog, 0, NULL, NULL,
   1865 		       CTLFLAG_PERMANENT
   1866 #ifndef INET6_BINDV6ONLY
   1867 		       |CTLFLAG_READWRITE,
   1868 #endif
   1869 		       CTLTYPE_INT, "v6only",
   1870 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
   1871 				    "to PF_INET sockets"),
   1872 		       NULL, 0, &ip6_v6only, 0,
   1873 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1874 		       IPV6CTL_V6ONLY, CTL_EOL);
   1875 	sysctl_createv(clog, 0, NULL, NULL,
   1876 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1877 		       CTLTYPE_INT, "anonportmin",
   1878 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
   1879 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
   1880 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1881 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
   1882 	sysctl_createv(clog, 0, NULL, NULL,
   1883 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1884 		       CTLTYPE_INT, "anonportmax",
   1885 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
   1886 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
   1887 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1888 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
   1889 #ifndef IPNOPRIVPORTS
   1890 	sysctl_createv(clog, 0, NULL, NULL,
   1891 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1892 		       CTLTYPE_INT, "lowportmin",
   1893 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
   1894 				    "to assign"),
   1895 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
   1896 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1897 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
   1898 	sysctl_createv(clog, 0, NULL, NULL,
   1899 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1900 		       CTLTYPE_INT, "lowportmax",
   1901 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
   1902 				    "to assign"),
   1903 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
   1904 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1905 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
   1906 #endif /* IPNOPRIVPORTS */
   1907 	sysctl_createv(clog, 0, NULL, NULL,
   1908 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1909 		       CTLTYPE_INT, "use_tempaddr",
   1910 		       SYSCTL_DESCR("Use temporary address"),
   1911 		       NULL, 0, &ip6_use_tempaddr, 0,
   1912 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1913 		       CTL_CREATE, CTL_EOL);
   1914 	sysctl_createv(clog, 0, NULL, NULL,
   1915 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1916 		       CTLTYPE_INT, "temppltime",
   1917 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
   1918 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
   1919 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1920 		       CTL_CREATE, CTL_EOL);
   1921 	sysctl_createv(clog, 0, NULL, NULL,
   1922 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1923 		       CTLTYPE_INT, "tempvltime",
   1924 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
   1925 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
   1926 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1927 		       CTL_CREATE, CTL_EOL);
   1928 	sysctl_createv(clog, 0, NULL, NULL,
   1929 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1930 		       CTLTYPE_INT, "maxfrags",
   1931 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
   1932 		       NULL, 0, &ip6_maxfrags, 0,
   1933 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1934 		       IPV6CTL_MAXFRAGS, CTL_EOL);
   1935 	sysctl_createv(clog, 0, NULL, NULL,
   1936 		       CTLFLAG_PERMANENT,
   1937 		       CTLTYPE_STRUCT, "stats",
   1938 		       SYSCTL_DESCR("IPv6 statistics"),
   1939 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
   1940 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1941 		       IPV6CTL_STATS, CTL_EOL);
   1942 	sysctl_createv(clog, 0, NULL, NULL,
   1943 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1944 		       CTLTYPE_INT, "use_defaultzone",
   1945 		       SYSCTL_DESCR("Whether to use the default scope zones"),
   1946 		       NULL, 0, &ip6_use_defzone, 0,
   1947 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1948 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
   1949 	sysctl_createv(clog, 0, NULL, NULL,
   1950 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1951 		       CTLTYPE_INT, "mcast_pmtu",
   1952 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
   1953 		       NULL, 0, &ip6_mcast_pmtu, 0,
   1954 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1955 		       CTL_CREATE, CTL_EOL);
   1956 #ifdef GATEWAY
   1957 	sysctl_createv(clog, 0, NULL, NULL,
   1958 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1959 			CTLTYPE_INT, "maxflows",
   1960 			SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"),
   1961 			sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0,
   1962 			CTL_NET, PF_INET6, IPPROTO_IPV6,
   1963 			CTL_CREATE, CTL_EOL);
   1964 	sysctl_createv(clog, 0, NULL, NULL,
   1965 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1966 			CTLTYPE_INT, "hashsize",
   1967 			SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"),
   1968 			sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0,
   1969 			CTL_NET, PF_INET6, IPPROTO_IPV6,
   1970 			CTL_CREATE, CTL_EOL);
   1971 #endif
   1972 }
   1973 
   1974 void
   1975 ip6_statinc(u_int stat)
   1976 {
   1977 
   1978 	KASSERT(stat < IP6_NSTATS);
   1979 	IP6_STATINC(stat);
   1980 }
   1981