Home | History | Annotate | Line # | Download | only in netinet6
ip6_input.c revision 1.153
      1 /*	$NetBSD: ip6_input.c,v 1.153 2015/12/12 23:34:25 christos Exp $	*/
      2 /*	$KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $	*/
      3 
      4 /*
      5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
      6  * All rights reserved.
      7  *
      8  * Redistribution and use in source and binary forms, with or without
      9  * modification, are permitted provided that the following conditions
     10  * are met:
     11  * 1. Redistributions of source code must retain the above copyright
     12  *    notice, this list of conditions and the following disclaimer.
     13  * 2. Redistributions in binary form must reproduce the above copyright
     14  *    notice, this list of conditions and the following disclaimer in the
     15  *    documentation and/or other materials provided with the distribution.
     16  * 3. Neither the name of the project nor the names of its contributors
     17  *    may be used to endorse or promote products derived from this software
     18  *    without specific prior written permission.
     19  *
     20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
     21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
     24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     30  * SUCH DAMAGE.
     31  */
     32 
     33 /*
     34  * Copyright (c) 1982, 1986, 1988, 1993
     35  *	The Regents of the University of California.  All rights reserved.
     36  *
     37  * Redistribution and use in source and binary forms, with or without
     38  * modification, are permitted provided that the following conditions
     39  * are met:
     40  * 1. Redistributions of source code must retain the above copyright
     41  *    notice, this list of conditions and the following disclaimer.
     42  * 2. Redistributions in binary form must reproduce the above copyright
     43  *    notice, this list of conditions and the following disclaimer in the
     44  *    documentation and/or other materials provided with the distribution.
     45  * 3. Neither the name of the University nor the names of its contributors
     46  *    may be used to endorse or promote products derived from this software
     47  *    without specific prior written permission.
     48  *
     49  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     50  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     51  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     52  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     53  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     54  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     55  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     56  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     57  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     58  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     59  * SUCH DAMAGE.
     60  *
     61  *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
     62  */
     63 
     64 #include <sys/cdefs.h>
     65 __KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.153 2015/12/12 23:34:25 christos Exp $");
     66 
     67 #ifdef _KERNEL_OPT
     68 #include "opt_gateway.h"
     69 #include "opt_inet.h"
     70 #include "opt_inet6.h"
     71 #include "opt_ipsec.h"
     72 #include "opt_compat_netbsd.h"
     73 #endif
     74 
     75 #include <sys/param.h>
     76 #include <sys/systm.h>
     77 #include <sys/malloc.h>
     78 #include <sys/mbuf.h>
     79 #include <sys/domain.h>
     80 #include <sys/protosw.h>
     81 #include <sys/socket.h>
     82 #include <sys/socketvar.h>
     83 #include <sys/errno.h>
     84 #include <sys/time.h>
     85 #include <sys/kernel.h>
     86 #include <sys/syslog.h>
     87 #include <sys/proc.h>
     88 #include <sys/sysctl.h>
     89 #include <sys/cprng.h>
     90 
     91 #include <net/if.h>
     92 #include <net/if_types.h>
     93 #include <net/if_dl.h>
     94 #include <net/route.h>
     95 #include <net/pktqueue.h>
     96 #include <net/pfil.h>
     97 
     98 #include <netinet/in.h>
     99 #include <netinet/in_systm.h>
    100 #ifdef INET
    101 #include <netinet/ip.h>
    102 #include <netinet/ip_var.h>
    103 #include <netinet/ip_icmp.h>
    104 #endif /* INET */
    105 #include <netinet/ip6.h>
    106 #include <netinet/portalgo.h>
    107 #include <netinet6/in6_var.h>
    108 #include <netinet6/ip6_var.h>
    109 #include <netinet6/ip6_private.h>
    110 #include <netinet6/in6_pcb.h>
    111 #include <netinet/icmp6.h>
    112 #include <netinet6/scope6_var.h>
    113 #include <netinet6/in6_ifattach.h>
    114 #include <netinet6/nd6.h>
    115 
    116 #ifdef IPSEC
    117 #include <netipsec/ipsec.h>
    118 #include <netipsec/ipsec6.h>
    119 #include <netipsec/key.h>
    120 #endif /* IPSEC */
    121 
    122 #ifdef COMPAT_50
    123 #include <compat/sys/time.h>
    124 #include <compat/sys/socket.h>
    125 #endif
    126 
    127 #include <netinet6/ip6protosw.h>
    128 
    129 #include "faith.h"
    130 #include "gif.h"
    131 
    132 #if NGIF > 0
    133 #include <netinet6/in6_gif.h>
    134 #endif
    135 
    136 #include <net/net_osdep.h>
    137 
    138 extern struct domain inet6domain;
    139 
    140 u_char ip6_protox[IPPROTO_MAX];
    141 struct in6_ifaddr *in6_ifaddr;
    142 pktqueue_t *ip6_pktq __read_mostly;
    143 
    144 extern callout_t in6_tmpaddrtimer_ch;
    145 
    146 int ip6_forward_srcrt;			/* XXX */
    147 int ip6_sourcecheck;			/* XXX */
    148 int ip6_sourcecheck_interval;		/* XXX */
    149 
    150 pfil_head_t *inet6_pfil_hook;
    151 
    152 percpu_t *ip6stat_percpu;
    153 
    154 static void ip6_init2(void *);
    155 static void ip6intr(void *);
    156 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
    157 
    158 static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *,
    159 	u_int32_t *);
    160 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
    161 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
    162 
    163 /*
    164  * IP6 initialization: fill in IP6 protocol switch table.
    165  * All protocols not implemented in kernel go to raw IP6 protocol handler.
    166  */
    167 void
    168 ip6_init(void)
    169 {
    170 	const struct ip6protosw *pr;
    171 	int i;
    172 
    173 	sysctl_net_inet6_ip6_setup(NULL);
    174 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
    175 	if (pr == 0)
    176 		panic("ip6_init");
    177 	for (i = 0; i < IPPROTO_MAX; i++)
    178 		ip6_protox[i] = pr - inet6sw;
    179 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
    180 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
    181 		if (pr->pr_domain->dom_family == PF_INET6 &&
    182 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
    183 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
    184 
    185 	ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL);
    186 	KASSERT(ip6_pktq != NULL);
    187 
    188 	scope6_init();
    189 	addrsel_policy_init();
    190 	nd6_init();
    191 	frag6_init();
    192 	ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR;
    193 
    194 	ip6_init2(NULL);
    195 #ifdef GATEWAY
    196 	ip6flow_init(ip6_hashsize);
    197 #endif
    198 	/* Register our Packet Filter hook. */
    199 	inet6_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET6);
    200 	KASSERT(inet6_pfil_hook != NULL);
    201 
    202 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
    203 }
    204 
    205 static void
    206 ip6_init2(void *dummy)
    207 {
    208 
    209 	/* nd6_timer_init */
    210 	callout_init(&nd6_timer_ch, CALLOUT_MPSAFE);
    211 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
    212 
    213 	/* timer for regeneranation of temporary addresses randomize ID */
    214 	callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE);
    215 	callout_reset(&in6_tmpaddrtimer_ch,
    216 		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
    217 		       ip6_temp_regen_advance) * hz,
    218 		      in6_tmpaddrtimer, NULL);
    219 }
    220 
    221 /*
    222  * IP6 input interrupt handling. Just pass the packet to ip6_input.
    223  */
    224 static void
    225 ip6intr(void *arg __unused)
    226 {
    227 	struct mbuf *m;
    228 
    229 	mutex_enter(softnet_lock);
    230 	while ((m = pktq_dequeue(ip6_pktq)) != NULL) {
    231 		const ifnet_t *ifp = m->m_pkthdr.rcvif;
    232 
    233 		/*
    234 		 * Drop the packet if IPv6 is disabled on the interface.
    235 		 */
    236 		if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)) {
    237 			m_freem(m);
    238 			continue;
    239 		}
    240 		ip6_input(m);
    241 	}
    242 	mutex_exit(softnet_lock);
    243 }
    244 
    245 extern struct	route ip6_forward_rt;
    246 
    247 void
    248 ip6_input(struct mbuf *m)
    249 {
    250 	struct ip6_hdr *ip6;
    251 	int hit, off = sizeof(struct ip6_hdr), nest;
    252 	u_int32_t plen;
    253 	u_int32_t rtalert = ~0;
    254 	int nxt, ours = 0, rh_present = 0;
    255 	struct ifnet *deliverifp = NULL;
    256 	int srcrt = 0;
    257 	const struct rtentry *rt;
    258 	union {
    259 		struct sockaddr		dst;
    260 		struct sockaddr_in6	dst6;
    261 	} u;
    262 
    263 	/*
    264 	 * make sure we don't have onion peering information into m_tag.
    265 	 */
    266 	ip6_delaux(m);
    267 
    268 	/*
    269 	 * mbuf statistics
    270 	 */
    271 	if (m->m_flags & M_EXT) {
    272 		if (m->m_next)
    273 			IP6_STATINC(IP6_STAT_MEXT2M);
    274 		else
    275 			IP6_STATINC(IP6_STAT_MEXT1);
    276 	} else {
    277 #define M2MMAX	32
    278 		if (m->m_next) {
    279 			if (m->m_flags & M_LOOP) {
    280 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
    281 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX) {
    282 				IP6_STATINC(IP6_STAT_M2M +
    283 					    m->m_pkthdr.rcvif->if_index);
    284 			} else
    285 				IP6_STATINC(IP6_STAT_M2M);
    286 		} else
    287 			IP6_STATINC(IP6_STAT_M1);
    288 #undef M2MMAX
    289 	}
    290 
    291 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
    292 	IP6_STATINC(IP6_STAT_TOTAL);
    293 
    294 	/*
    295 	 * If the IPv6 header is not aligned, slurp it up into a new
    296 	 * mbuf with space for link headers, in the event we forward
    297 	 * it.  Otherwise, if it is aligned, make sure the entire base
    298 	 * IPv6 header is in the first mbuf of the chain.
    299 	 */
    300 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
    301 		struct ifnet *inifp = m->m_pkthdr.rcvif;
    302 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
    303 				  (max_linkhdr + 3) & ~3)) == NULL) {
    304 			/* XXXJRT new stat, please */
    305 			IP6_STATINC(IP6_STAT_TOOSMALL);
    306 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    307 			return;
    308 		}
    309 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
    310 		struct ifnet *inifp = m->m_pkthdr.rcvif;
    311 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    312 			IP6_STATINC(IP6_STAT_TOOSMALL);
    313 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    314 			return;
    315 		}
    316 	}
    317 
    318 	ip6 = mtod(m, struct ip6_hdr *);
    319 
    320 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    321 		IP6_STATINC(IP6_STAT_BADVERS);
    322 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    323 		goto bad;
    324 	}
    325 
    326 	/*
    327 	 * Assume that we can create a fast-forward IP flow entry
    328 	 * based on this packet.
    329 	 */
    330 	m->m_flags |= M_CANFASTFWD;
    331 
    332 	/*
    333 	 * Run through list of hooks for input packets.  If there are any
    334 	 * filters which require that additional packets in the flow are
    335 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
    336 	 * Note that filters must _never_ set this flag, as another filter
    337 	 * in the list may have previously cleared it.
    338 	 */
    339 	/*
    340 	 * let ipfilter look at packet on the wire,
    341 	 * not the decapsulated packet.
    342 	 */
    343 #if defined(IPSEC)
    344 	if (!ipsec_used || !ipsec_indone(m))
    345 #else
    346 	if (1)
    347 #endif
    348 	{
    349 		struct in6_addr odst;
    350 
    351 		odst = ip6->ip6_dst;
    352 		if (pfil_run_hooks(inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
    353 				   PFIL_IN) != 0)
    354 			return;
    355 		if (m == NULL)
    356 			return;
    357 		ip6 = mtod(m, struct ip6_hdr *);
    358 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
    359 	}
    360 
    361 	IP6_STATINC(IP6_STAT_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 		IP6_STATINC(IP6_STAT_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 		IP6_STATINC(IP6_STAT_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 		IP6_STATINC(IP6_STAT_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 		IP6_STATINC(IP6_STAT_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 		IP6_STATINC(IP6_STAT_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 			uint64_t *ip6s = IP6_STAT_GETREF();
    456 			ip6s[IP6_STAT_NOTMEMBER]++;
    457 			ip6s[IP6_STAT_CANTFORWARD]++;
    458 			IP6_STAT_PUTREF();
    459 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    460 			goto bad;
    461 		}
    462 		deliverifp = m->m_pkthdr.rcvif;
    463 		goto hbhcheck;
    464 	}
    465 
    466 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
    467 
    468 	/*
    469 	 *  Unicast check
    470 	 */
    471 	rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit);
    472 	if (hit)
    473 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
    474 	else
    475 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
    476 
    477 #define rt6_getkey(__rt) satocsin6(rt_getkey(__rt))
    478 
    479 	/*
    480 	 * Accept the packet if the forwarding interface to the destination
    481 	 * according to the routing table is the loopback interface,
    482 	 * unless the associated route has a gateway.
    483 	 * Note that this approach causes to accept a packet if there is a
    484 	 * route to the loopback interface for the destination of the packet.
    485 	 * But we think it's even useful in some situations, e.g. when using
    486 	 * a special daemon which wants to intercept the packet.
    487 	 */
    488 	if (rt != NULL &&
    489 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
    490 	    !(rt->rt_flags & RTF_CLONED) &&
    491 #if 0
    492 	    /*
    493 	     * The check below is redundant since the comparison of
    494 	     * the destination and the key of the rtentry has
    495 	     * already done through looking up the routing table.
    496 	     */
    497 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(rt)->sin6_addr) &&
    498 #endif
    499 	    rt->rt_ifp->if_type == IFT_LOOP) {
    500 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
    501 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
    502 			m->m_flags |= M_ANYCAST6;
    503 		/*
    504 		 * packets to a tentative, duplicated, or somehow invalid
    505 		 * address must not be accepted.
    506 		 */
    507 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
    508 			/* this address is ready */
    509 			ours = 1;
    510 			deliverifp = ia6->ia_ifp;	/* correct? */
    511 			goto hbhcheck;
    512 		} else {
    513 			/* address is not ready, so discard the packet. */
    514 			nd6log((LOG_INFO,
    515 			    "ip6_input: packet to an unready address %s->%s\n",
    516 			    ip6_sprintf(&ip6->ip6_src),
    517 			    ip6_sprintf(&ip6->ip6_dst)));
    518 
    519 			goto bad;
    520 		}
    521 	}
    522 
    523 	/*
    524 	 * FAITH (Firewall Aided Internet Translator)
    525 	 */
    526 #if defined(NFAITH) && 0 < NFAITH
    527 	if (ip6_keepfaith) {
    528 		if (rt != NULL && rt->rt_ifp != NULL &&
    529 		    rt->rt_ifp->if_type == IFT_FAITH) {
    530 			/* XXX do we need more sanity checks? */
    531 			ours = 1;
    532 			deliverifp = rt->rt_ifp; /* faith */
    533 			goto hbhcheck;
    534 		}
    535 	}
    536 #endif
    537 
    538 #if 0
    539     {
    540 	/*
    541 	 * Last resort: check in6_ifaddr for incoming interface.
    542 	 * The code is here until I update the "goto ours hack" code above
    543 	 * working right.
    544 	 */
    545 	struct ifaddr *ifa;
    546 	IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
    547 		if (ifa->ifa_addr == NULL)
    548 			continue;	/* just for safety */
    549 		if (ifa->ifa_addr->sa_family != AF_INET6)
    550 			continue;
    551 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
    552 			ours = 1;
    553 			deliverifp = ifa->ifa_ifp;
    554 			goto hbhcheck;
    555 		}
    556 	}
    557     }
    558 #endif
    559 
    560 	/*
    561 	 * Now there is no reason to process the packet if it's not our own
    562 	 * and we're not a router.
    563 	 */
    564 	if (!ip6_forwarding) {
    565 		IP6_STATINC(IP6_STAT_CANTFORWARD);
    566 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    567 		goto bad;
    568 	}
    569 
    570   hbhcheck:
    571 	/*
    572 	 * record address information into m_tag, if we don't have one yet.
    573 	 * note that we are unable to record it, if the address is not listed
    574 	 * as our interface address (e.g. multicast addresses, addresses
    575 	 * within FAITH prefixes and such).
    576 	 */
    577 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
    578 		struct in6_ifaddr *ia6;
    579 
    580 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    581 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
    582 			/*
    583 			 * XXX maybe we should drop the packet here,
    584 			 * as we could not provide enough information
    585 			 * to the upper layers.
    586 			 */
    587 		}
    588 	}
    589 
    590 	/*
    591 	 * Process Hop-by-Hop options header if it's contained.
    592 	 * m may be modified in ip6_hopopts_input().
    593 	 * If a JumboPayload option is included, plen will also be modified.
    594 	 */
    595 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
    596 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
    597 		struct ip6_hbh *hbh;
    598 
    599 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
    600 #if 0	/*touches NULL pointer*/
    601 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    602 #endif
    603 			return;	/* m have already been freed */
    604 		}
    605 
    606 		/* adjust pointer */
    607 		ip6 = mtod(m, struct ip6_hdr *);
    608 
    609 		/*
    610 		 * if the payload length field is 0 and the next header field
    611 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
    612 		 * option MUST be included.
    613 		 */
    614 		if (ip6->ip6_plen == 0 && plen == 0) {
    615 			/*
    616 			 * Note that if a valid jumbo payload option is
    617 			 * contained, ip6_hopopts_input() must set a valid
    618 			 * (non-zero) payload length to the variable plen.
    619 			 */
    620 			IP6_STATINC(IP6_STAT_BADOPTIONS);
    621 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    622 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    623 			icmp6_error(m, ICMP6_PARAM_PROB,
    624 				    ICMP6_PARAMPROB_HEADER,
    625 				    (char *)&ip6->ip6_plen - (char *)ip6);
    626 			return;
    627 		}
    628 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    629 			sizeof(struct ip6_hbh));
    630 		if (hbh == NULL) {
    631 			IP6_STATINC(IP6_STAT_TOOSHORT);
    632 			return;
    633 		}
    634 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
    635 		nxt = hbh->ip6h_nxt;
    636 
    637 		/*
    638 		 * accept the packet if a router alert option is included
    639 		 * and we act as an IPv6 router.
    640 		 */
    641 		if (rtalert != ~0 && ip6_forwarding)
    642 			ours = 1;
    643 	} else
    644 		nxt = ip6->ip6_nxt;
    645 
    646 	/*
    647 	 * Check that the amount of data in the buffers
    648 	 * is as at least much as the IPv6 header would have us expect.
    649 	 * Trim mbufs if longer than we expect.
    650 	 * Drop packet if shorter than we expect.
    651 	 */
    652 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
    653 		IP6_STATINC(IP6_STAT_TOOSHORT);
    654 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    655 		goto bad;
    656 	}
    657 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
    658 		if (m->m_len == m->m_pkthdr.len) {
    659 			m->m_len = sizeof(struct ip6_hdr) + plen;
    660 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
    661 		} else
    662 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
    663 	}
    664 
    665 	/*
    666 	 * Forward if desirable.
    667 	 */
    668 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    669 		/*
    670 		 * If we are acting as a multicast router, all
    671 		 * incoming multicast packets are passed to the
    672 		 * kernel-level multicast forwarding function.
    673 		 * The packet is returned (relatively) intact; if
    674 		 * ip6_mforward() returns a non-zero value, the packet
    675 		 * must be discarded, else it may be accepted below.
    676 		 */
    677 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
    678 			IP6_STATINC(IP6_STAT_CANTFORWARD);
    679 			m_freem(m);
    680 			return;
    681 		}
    682 		if (!ours) {
    683 			m_freem(m);
    684 			return;
    685 		}
    686 	} else if (!ours) {
    687 		ip6_forward(m, srcrt);
    688 		return;
    689 	}
    690 
    691 	ip6 = mtod(m, struct ip6_hdr *);
    692 
    693 	/*
    694 	 * Malicious party may be able to use IPv4 mapped addr to confuse
    695 	 * tcp/udp stack and bypass security checks (act as if it was from
    696 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
    697 	 *
    698 	 * For SIIT end node behavior, you may want to disable the check.
    699 	 * However, you will  become vulnerable to attacks using IPv4 mapped
    700 	 * source.
    701 	 */
    702 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    703 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    704 		IP6_STATINC(IP6_STAT_BADSCOPE);
    705 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    706 		goto bad;
    707 	}
    708 
    709 	/*
    710 	 * Tell launch routine the next header
    711 	 */
    712 #ifdef IFA_STATS
    713 	if (deliverifp != NULL) {
    714 		struct in6_ifaddr *ia6;
    715 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    716 		if (ia6)
    717 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
    718 	}
    719 #endif
    720 	IP6_STATINC(IP6_STAT_DELIVERED);
    721 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
    722 	nest = 0;
    723 
    724 	rh_present = 0;
    725 	while (nxt != IPPROTO_DONE) {
    726 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
    727 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
    728 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    729 			goto bad;
    730 		}
    731 
    732 		/*
    733 		 * protection against faulty packet - there should be
    734 		 * more sanity checks in header chain processing.
    735 		 */
    736 		if (m->m_pkthdr.len < off) {
    737 			IP6_STATINC(IP6_STAT_TOOSHORT);
    738 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    739 			goto bad;
    740 		}
    741 
    742 		if (nxt == IPPROTO_ROUTING) {
    743 			if (rh_present++) {
    744 				in6_ifstat_inc(m->m_pkthdr.rcvif,
    745 				    ifs6_in_hdrerr);
    746 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    747 				goto bad;
    748 			}
    749 		}
    750 
    751 #ifdef IPSEC
    752 		if (ipsec_used) {
    753 			/*
    754 			 * enforce IPsec policy checking if we are seeing last
    755 			 * header. note that we do not visit this with
    756 			 * protocols with pcb layer code - like udp/tcp/raw ip.
    757 			 */
    758 			if ((inet6sw[ip_protox[nxt]].pr_flags
    759 			    & PR_LASTHDR) != 0) {
    760 				int error = ipsec6_input(m);
    761 				if (error)
    762 					goto bad;
    763 			}
    764 		}
    765 #endif /* IPSEC */
    766 
    767 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
    768 	}
    769 	return;
    770  bad:
    771 	m_freem(m);
    772 }
    773 
    774 /*
    775  * set/grab in6_ifaddr correspond to IPv6 destination address.
    776  */
    777 static struct m_tag *
    778 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
    779 {
    780 	struct m_tag *mtag;
    781 	struct ip6aux *ip6a;
    782 
    783 	mtag = ip6_addaux(m);
    784 	if (mtag == NULL)
    785 		return NULL;
    786 
    787 	ip6a = (struct ip6aux *)(mtag + 1);
    788 	if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) {
    789 		IP6_STATINC(IP6_STAT_BADSCOPE);
    790 		return NULL;
    791 	}
    792 
    793 	ip6a->ip6a_src = ia->ia_addr.sin6_addr;
    794 	ip6a->ip6a_flags = ia->ia6_flags;
    795 	return mtag;
    796 }
    797 
    798 const struct ip6aux *
    799 ip6_getdstifaddr(struct mbuf *m)
    800 {
    801 	struct m_tag *mtag;
    802 
    803 	mtag = ip6_findaux(m);
    804 	if (mtag != NULL)
    805 		return (struct ip6aux *)(mtag + 1);
    806 	else
    807 		return NULL;
    808 }
    809 
    810 /*
    811  * Hop-by-Hop options header processing. If a valid jumbo payload option is
    812  * included, the real payload length will be stored in plenp.
    813  *
    814  * rtalertp - XXX: should be stored more smart way
    815  */
    816 int
    817 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
    818 	struct mbuf **mp, int *offp)
    819 {
    820 	struct mbuf *m = *mp;
    821 	int off = *offp, hbhlen;
    822 	struct ip6_hbh *hbh;
    823 
    824 	/* validation of the length of the header */
    825 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
    826 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
    827 	if (hbh == NULL) {
    828 		IP6_STATINC(IP6_STAT_TOOSHORT);
    829 		return -1;
    830 	}
    831 	hbhlen = (hbh->ip6h_len + 1) << 3;
    832 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    833 		hbhlen);
    834 	if (hbh == NULL) {
    835 		IP6_STATINC(IP6_STAT_TOOSHORT);
    836 		return -1;
    837 	}
    838 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
    839 	off += hbhlen;
    840 	hbhlen -= sizeof(struct ip6_hbh);
    841 
    842 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
    843 				hbhlen, rtalertp, plenp) < 0)
    844 		return (-1);
    845 
    846 	*offp = off;
    847 	*mp = m;
    848 	return (0);
    849 }
    850 
    851 /*
    852  * Search header for all Hop-by-hop options and process each option.
    853  * This function is separate from ip6_hopopts_input() in order to
    854  * handle a case where the sending node itself process its hop-by-hop
    855  * options header. In such a case, the function is called from ip6_output().
    856  *
    857  * The function assumes that hbh header is located right after the IPv6 header
    858  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
    859  * opthead + hbhlen is located in continuous memory region.
    860  */
    861 static int
    862 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
    863 	u_int32_t *rtalertp, u_int32_t *plenp)
    864 {
    865 	struct ip6_hdr *ip6;
    866 	int optlen = 0;
    867 	u_int8_t *opt = opthead;
    868 	u_int16_t rtalert_val;
    869 	u_int32_t jumboplen;
    870 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
    871 
    872 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
    873 		switch (*opt) {
    874 		case IP6OPT_PAD1:
    875 			optlen = 1;
    876 			break;
    877 		case IP6OPT_PADN:
    878 			if (hbhlen < IP6OPT_MINLEN) {
    879 				IP6_STATINC(IP6_STAT_TOOSMALL);
    880 				goto bad;
    881 			}
    882 			optlen = *(opt + 1) + 2;
    883 			break;
    884 		case IP6OPT_RTALERT:
    885 			/* XXX may need check for alignment */
    886 			if (hbhlen < IP6OPT_RTALERT_LEN) {
    887 				IP6_STATINC(IP6_STAT_TOOSMALL);
    888 				goto bad;
    889 			}
    890 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
    891 				/* XXX stat */
    892 				icmp6_error(m, ICMP6_PARAM_PROB,
    893 				    ICMP6_PARAMPROB_HEADER,
    894 				    erroff + opt + 1 - opthead);
    895 				return (-1);
    896 			}
    897 			optlen = IP6OPT_RTALERT_LEN;
    898 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
    899 			*rtalertp = ntohs(rtalert_val);
    900 			break;
    901 		case IP6OPT_JUMBO:
    902 			/* XXX may need check for alignment */
    903 			if (hbhlen < IP6OPT_JUMBO_LEN) {
    904 				IP6_STATINC(IP6_STAT_TOOSMALL);
    905 				goto bad;
    906 			}
    907 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
    908 				/* XXX stat */
    909 				icmp6_error(m, ICMP6_PARAM_PROB,
    910 				    ICMP6_PARAMPROB_HEADER,
    911 				    erroff + opt + 1 - opthead);
    912 				return (-1);
    913 			}
    914 			optlen = IP6OPT_JUMBO_LEN;
    915 
    916 			/*
    917 			 * IPv6 packets that have non 0 payload length
    918 			 * must not contain a jumbo payload option.
    919 			 */
    920 			ip6 = mtod(m, struct ip6_hdr *);
    921 			if (ip6->ip6_plen) {
    922 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    923 				icmp6_error(m, ICMP6_PARAM_PROB,
    924 				    ICMP6_PARAMPROB_HEADER,
    925 				    erroff + opt - opthead);
    926 				return (-1);
    927 			}
    928 
    929 			/*
    930 			 * We may see jumbolen in unaligned location, so
    931 			 * we'd need to perform bcopy().
    932 			 */
    933 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
    934 			jumboplen = (u_int32_t)htonl(jumboplen);
    935 
    936 #if 1
    937 			/*
    938 			 * if there are multiple jumbo payload options,
    939 			 * *plenp will be non-zero and the packet will be
    940 			 * rejected.
    941 			 * the behavior may need some debate in ipngwg -
    942 			 * multiple options does not make sense, however,
    943 			 * there's no explicit mention in specification.
    944 			 */
    945 			if (*plenp != 0) {
    946 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    947 				icmp6_error(m, ICMP6_PARAM_PROB,
    948 				    ICMP6_PARAMPROB_HEADER,
    949 				    erroff + opt + 2 - opthead);
    950 				return (-1);
    951 			}
    952 #endif
    953 
    954 			/*
    955 			 * jumbo payload length must be larger than 65535.
    956 			 */
    957 			if (jumboplen <= IPV6_MAXPACKET) {
    958 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    959 				icmp6_error(m, ICMP6_PARAM_PROB,
    960 				    ICMP6_PARAMPROB_HEADER,
    961 				    erroff + opt + 2 - opthead);
    962 				return (-1);
    963 			}
    964 			*plenp = jumboplen;
    965 
    966 			break;
    967 		default:		/* unknown option */
    968 			if (hbhlen < IP6OPT_MINLEN) {
    969 				IP6_STATINC(IP6_STAT_TOOSMALL);
    970 				goto bad;
    971 			}
    972 			optlen = ip6_unknown_opt(opt, m,
    973 			    erroff + opt - opthead);
    974 			if (optlen == -1)
    975 				return (-1);
    976 			optlen += 2;
    977 			break;
    978 		}
    979 	}
    980 
    981 	return (0);
    982 
    983   bad:
    984 	m_freem(m);
    985 	return (-1);
    986 }
    987 
    988 /*
    989  * Unknown option processing.
    990  * The third argument `off' is the offset from the IPv6 header to the option,
    991  * which is necessary if the IPv6 header the and option header and IPv6 header
    992  * is not continuous in order to return an ICMPv6 error.
    993  */
    994 int
    995 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
    996 {
    997 	struct ip6_hdr *ip6;
    998 
    999 	switch (IP6OPT_TYPE(*optp)) {
   1000 	case IP6OPT_TYPE_SKIP: /* ignore the option */
   1001 		return ((int)*(optp + 1));
   1002 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
   1003 		m_freem(m);
   1004 		return (-1);
   1005 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
   1006 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1007 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
   1008 		return (-1);
   1009 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
   1010 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1011 		ip6 = mtod(m, struct ip6_hdr *);
   1012 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
   1013 		    (m->m_flags & (M_BCAST|M_MCAST)))
   1014 			m_freem(m);
   1015 		else
   1016 			icmp6_error(m, ICMP6_PARAM_PROB,
   1017 				    ICMP6_PARAMPROB_OPTION, off);
   1018 		return (-1);
   1019 	}
   1020 
   1021 	m_freem(m);		/* XXX: NOTREACHED */
   1022 	return (-1);
   1023 }
   1024 
   1025 /*
   1026  * Create the "control" list for this pcb.
   1027  *
   1028  * The routine will be called from upper layer handlers like tcp6_input().
   1029  * Thus the routine assumes that the caller (tcp6_input) have already
   1030  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
   1031  * very first mbuf on the mbuf chain.
   1032  * We may want to add some infinite loop prevention or sanity checks for safety.
   1033  * (This applies only when you are using KAME mbuf chain restriction, i.e.
   1034  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
   1035  */
   1036 void
   1037 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
   1038 	struct ip6_hdr *ip6, struct mbuf *m)
   1039 {
   1040 #ifdef RFC2292
   1041 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1042 #else
   1043 #define IS2292(x, y)	(y)
   1044 #endif
   1045 
   1046 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP
   1047 #ifdef SO_OTIMESTAMP
   1048 	    || in6p->in6p_socket->so_options & SO_OTIMESTAMP
   1049 #endif
   1050 	) {
   1051 		struct timeval tv;
   1052 
   1053 		microtime(&tv);
   1054 #ifdef SO_OTIMESTAMP
   1055 		if (in6p->in6p_socket->so_options & SO_OTIMESTAMP) {
   1056 			struct timeval50 tv50;
   1057 			timeval_to_timeval50(&tv, &tv50);
   1058 			*mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
   1059 			    SCM_OTIMESTAMP, SOL_SOCKET);
   1060 		} else
   1061 #endif
   1062 		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
   1063 		    SCM_TIMESTAMP, SOL_SOCKET);
   1064 		if (*mp)
   1065 			mp = &(*mp)->m_next;
   1066 	}
   1067 
   1068 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
   1069 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
   1070 		return;
   1071 
   1072 	/* RFC 2292 sec. 5 */
   1073 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
   1074 		struct in6_pktinfo pi6;
   1075 
   1076 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
   1077 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
   1078 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif ?
   1079 		    m->m_pkthdr.rcvif->if_index : 0;
   1080 		*mp = sbcreatecontrol((void *) &pi6,
   1081 		    sizeof(struct in6_pktinfo),
   1082 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
   1083 		if (*mp)
   1084 			mp = &(*mp)->m_next;
   1085 	}
   1086 
   1087 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
   1088 		int hlim = ip6->ip6_hlim & 0xff;
   1089 
   1090 		*mp = sbcreatecontrol((void *) &hlim, sizeof(int),
   1091 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
   1092 		if (*mp)
   1093 			mp = &(*mp)->m_next;
   1094 	}
   1095 
   1096 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
   1097 		u_int32_t flowinfo;
   1098 		int tclass;
   1099 
   1100 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
   1101 		flowinfo >>= 20;
   1102 
   1103 		tclass = flowinfo & 0xff;
   1104 		*mp = sbcreatecontrol((void *)&tclass, sizeof(tclass),
   1105 		    IPV6_TCLASS, IPPROTO_IPV6);
   1106 
   1107 		if (*mp)
   1108 			mp = &(*mp)->m_next;
   1109 	}
   1110 
   1111 	/*
   1112 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
   1113 	 * privilege for the option (see ip6_ctloutput), but it might be too
   1114 	 * strict, since there might be some hop-by-hop options which can be
   1115 	 * returned to normal user.
   1116 	 * See also RFC3542 section 8 (or RFC2292 section 6).
   1117 	 */
   1118 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
   1119 		/*
   1120 		 * Check if a hop-by-hop options header is contatined in the
   1121 		 * received packet, and if so, store the options as ancillary
   1122 		 * data. Note that a hop-by-hop options header must be
   1123 		 * just after the IPv6 header, which fact is assured through
   1124 		 * the IPv6 input processing.
   1125 		 */
   1126 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1127 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
   1128 			struct ip6_hbh *hbh;
   1129 			int hbhlen;
   1130 			struct mbuf *ext;
   1131 
   1132 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
   1133 			    xip6->ip6_nxt);
   1134 			if (ext == NULL) {
   1135 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1136 				return;
   1137 			}
   1138 			hbh = mtod(ext, struct ip6_hbh *);
   1139 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1140 			if (hbhlen != ext->m_len) {
   1141 				m_freem(ext);
   1142 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1143 				return;
   1144 			}
   1145 
   1146 			/*
   1147 			 * XXX: We copy whole the header even if a jumbo
   1148 			 * payload option is included, which option is to
   1149 			 * be removed before returning in the RFC 2292.
   1150 			 * Note: this constraint is removed in RFC3542.
   1151 			 */
   1152 			*mp = sbcreatecontrol((void *)hbh, hbhlen,
   1153 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
   1154 			    IPPROTO_IPV6);
   1155 			if (*mp)
   1156 				mp = &(*mp)->m_next;
   1157 			m_freem(ext);
   1158 		}
   1159 	}
   1160 
   1161 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
   1162 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
   1163 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1164 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
   1165 
   1166 		/*
   1167 		 * Search for destination options headers or routing
   1168 		 * header(s) through the header chain, and stores each
   1169 		 * header as ancillary data.
   1170 		 * Note that the order of the headers remains in
   1171 		 * the chain of ancillary data.
   1172 		 */
   1173 		for (;;) {	/* is explicit loop prevention necessary? */
   1174 			struct ip6_ext *ip6e = NULL;
   1175 			int elen;
   1176 			struct mbuf *ext = NULL;
   1177 
   1178 			/*
   1179 			 * if it is not an extension header, don't try to
   1180 			 * pull it from the chain.
   1181 			 */
   1182 			switch (nxt) {
   1183 			case IPPROTO_DSTOPTS:
   1184 			case IPPROTO_ROUTING:
   1185 			case IPPROTO_HOPOPTS:
   1186 			case IPPROTO_AH: /* is it possible? */
   1187 				break;
   1188 			default:
   1189 				goto loopend;
   1190 			}
   1191 
   1192 			ext = ip6_pullexthdr(m, off, nxt);
   1193 			if (ext == NULL) {
   1194 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1195 				return;
   1196 			}
   1197 			ip6e = mtod(ext, struct ip6_ext *);
   1198 			if (nxt == IPPROTO_AH)
   1199 				elen = (ip6e->ip6e_len + 2) << 2;
   1200 			else
   1201 				elen = (ip6e->ip6e_len + 1) << 3;
   1202 			if (elen != ext->m_len) {
   1203 				m_freem(ext);
   1204 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1205 				return;
   1206 			}
   1207 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
   1208 
   1209 			switch (nxt) {
   1210 			case IPPROTO_DSTOPTS:
   1211 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
   1212 					break;
   1213 
   1214 				*mp = sbcreatecontrol((void *)ip6e, elen,
   1215 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
   1216 				    IPPROTO_IPV6);
   1217 				if (*mp)
   1218 					mp = &(*mp)->m_next;
   1219 				break;
   1220 
   1221 			case IPPROTO_ROUTING:
   1222 				if (!(in6p->in6p_flags & IN6P_RTHDR))
   1223 					break;
   1224 
   1225 				*mp = sbcreatecontrol((void *)ip6e, elen,
   1226 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
   1227 				    IPPROTO_IPV6);
   1228 				if (*mp)
   1229 					mp = &(*mp)->m_next;
   1230 				break;
   1231 
   1232 			case IPPROTO_HOPOPTS:
   1233 			case IPPROTO_AH: /* is it possible? */
   1234 				break;
   1235 
   1236 			default:
   1237 				/*
   1238 			 	 * other cases have been filtered in the above.
   1239 				 * none will visit this case.  here we supply
   1240 				 * the code just in case (nxt overwritten or
   1241 				 * other cases).
   1242 				 */
   1243 				m_freem(ext);
   1244 				goto loopend;
   1245 
   1246 			}
   1247 
   1248 			/* proceed with the next header. */
   1249 			off += elen;
   1250 			nxt = ip6e->ip6e_nxt;
   1251 			ip6e = NULL;
   1252 			m_freem(ext);
   1253 			ext = NULL;
   1254 		}
   1255 	  loopend:
   1256 	  	;
   1257 	}
   1258 }
   1259 #undef IS2292
   1260 
   1261 
   1262 void
   1263 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
   1264     uint32_t *mtu)
   1265 {
   1266 	struct socket *so;
   1267 	struct mbuf *m_mtu;
   1268 	struct ip6_mtuinfo mtuctl;
   1269 
   1270 	so = in6p->in6p_socket;
   1271 
   1272 	if (mtu == NULL)
   1273 		return;
   1274 
   1275 #ifdef DIAGNOSTIC
   1276 	if (so == NULL)		/* I believe this is impossible */
   1277 		panic("ip6_notify_pmtu: socket is NULL");
   1278 #endif
   1279 
   1280 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
   1281 	mtuctl.ip6m_mtu = *mtu;
   1282 	mtuctl.ip6m_addr = *dst;
   1283 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
   1284 		return;
   1285 
   1286 	if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl),
   1287 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
   1288 		return;
   1289 
   1290 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
   1291 	    == 0) {
   1292 		m_freem(m_mtu);
   1293 		/* XXX: should count statistics */
   1294 	} else
   1295 		sorwakeup(so);
   1296 
   1297 	return;
   1298 }
   1299 
   1300 /*
   1301  * pull single extension header from mbuf chain.  returns single mbuf that
   1302  * contains the result, or NULL on error.
   1303  */
   1304 static struct mbuf *
   1305 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
   1306 {
   1307 	struct ip6_ext ip6e;
   1308 	size_t elen;
   1309 	struct mbuf *n;
   1310 
   1311 #ifdef DIAGNOSTIC
   1312 	switch (nxt) {
   1313 	case IPPROTO_DSTOPTS:
   1314 	case IPPROTO_ROUTING:
   1315 	case IPPROTO_HOPOPTS:
   1316 	case IPPROTO_AH: /* is it possible? */
   1317 		break;
   1318 	default:
   1319 		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
   1320 	}
   1321 #endif
   1322 
   1323 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1324 	if (nxt == IPPROTO_AH)
   1325 		elen = (ip6e.ip6e_len + 2) << 2;
   1326 	else
   1327 		elen = (ip6e.ip6e_len + 1) << 3;
   1328 
   1329 	MGET(n, M_DONTWAIT, MT_DATA);
   1330 	if (n && elen >= MLEN) {
   1331 		MCLGET(n, M_DONTWAIT);
   1332 		if ((n->m_flags & M_EXT) == 0) {
   1333 			m_free(n);
   1334 			n = NULL;
   1335 		}
   1336 	}
   1337 	if (!n)
   1338 		return NULL;
   1339 
   1340 	n->m_len = 0;
   1341 	if (elen >= M_TRAILINGSPACE(n)) {
   1342 		m_free(n);
   1343 		return NULL;
   1344 	}
   1345 
   1346 	m_copydata(m, off, elen, mtod(n, void *));
   1347 	n->m_len = elen;
   1348 	return n;
   1349 }
   1350 
   1351 /*
   1352  * Get pointer to the previous header followed by the header
   1353  * currently processed.
   1354  * XXX: This function supposes that
   1355  *	M includes all headers,
   1356  *	the next header field and the header length field of each header
   1357  *	are valid, and
   1358  *	the sum of each header length equals to OFF.
   1359  * Because of these assumptions, this function must be called very
   1360  * carefully. Moreover, it will not be used in the near future when
   1361  * we develop `neater' mechanism to process extension headers.
   1362  */
   1363 u_int8_t *
   1364 ip6_get_prevhdr(struct mbuf *m, int off)
   1365 {
   1366 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1367 
   1368 	if (off == sizeof(struct ip6_hdr))
   1369 		return (&ip6->ip6_nxt);
   1370 	else {
   1371 		int len, nxt;
   1372 		struct ip6_ext *ip6e = NULL;
   1373 
   1374 		nxt = ip6->ip6_nxt;
   1375 		len = sizeof(struct ip6_hdr);
   1376 		while (len < off) {
   1377 			ip6e = (struct ip6_ext *)(mtod(m, char *) + len);
   1378 
   1379 			switch (nxt) {
   1380 			case IPPROTO_FRAGMENT:
   1381 				len += sizeof(struct ip6_frag);
   1382 				break;
   1383 			case IPPROTO_AH:
   1384 				len += (ip6e->ip6e_len + 2) << 2;
   1385 				break;
   1386 			default:
   1387 				len += (ip6e->ip6e_len + 1) << 3;
   1388 				break;
   1389 			}
   1390 			nxt = ip6e->ip6e_nxt;
   1391 		}
   1392 		if (ip6e)
   1393 			return (&ip6e->ip6e_nxt);
   1394 		else
   1395 			return NULL;
   1396 	}
   1397 }
   1398 
   1399 /*
   1400  * get next header offset.  m will be retained.
   1401  */
   1402 int
   1403 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1404 {
   1405 	struct ip6_hdr ip6;
   1406 	struct ip6_ext ip6e;
   1407 	struct ip6_frag fh;
   1408 
   1409 	/* just in case */
   1410 	if (m == NULL)
   1411 		panic("ip6_nexthdr: m == NULL");
   1412 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
   1413 		return -1;
   1414 
   1415 	switch (proto) {
   1416 	case IPPROTO_IPV6:
   1417 		/* do not chase beyond intermediate IPv6 headers */
   1418 		if (off != 0)
   1419 			return -1;
   1420 		if (m->m_pkthdr.len < off + sizeof(ip6))
   1421 			return -1;
   1422 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
   1423 		if (nxtp)
   1424 			*nxtp = ip6.ip6_nxt;
   1425 		off += sizeof(ip6);
   1426 		return off;
   1427 
   1428 	case IPPROTO_FRAGMENT:
   1429 		/*
   1430 		 * terminate parsing if it is not the first fragment,
   1431 		 * it does not make sense to parse through it.
   1432 		 */
   1433 		if (m->m_pkthdr.len < off + sizeof(fh))
   1434 			return -1;
   1435 		m_copydata(m, off, sizeof(fh), (void *)&fh);
   1436 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
   1437 			return -1;
   1438 		if (nxtp)
   1439 			*nxtp = fh.ip6f_nxt;
   1440 		off += sizeof(struct ip6_frag);
   1441 		return off;
   1442 
   1443 	case IPPROTO_AH:
   1444 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1445 			return -1;
   1446 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1447 		if (nxtp)
   1448 			*nxtp = ip6e.ip6e_nxt;
   1449 		off += (ip6e.ip6e_len + 2) << 2;
   1450 		if (m->m_pkthdr.len < off)
   1451 			return -1;
   1452 		return off;
   1453 
   1454 	case IPPROTO_HOPOPTS:
   1455 	case IPPROTO_ROUTING:
   1456 	case IPPROTO_DSTOPTS:
   1457 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1458 			return -1;
   1459 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1460 		if (nxtp)
   1461 			*nxtp = ip6e.ip6e_nxt;
   1462 		off += (ip6e.ip6e_len + 1) << 3;
   1463 		if (m->m_pkthdr.len < off)
   1464 			return -1;
   1465 		return off;
   1466 
   1467 	case IPPROTO_NONE:
   1468 	case IPPROTO_ESP:
   1469 	case IPPROTO_IPCOMP:
   1470 		/* give up */
   1471 		return -1;
   1472 
   1473 	default:
   1474 		return -1;
   1475 	}
   1476 }
   1477 
   1478 /*
   1479  * get offset for the last header in the chain.  m will be kept untainted.
   1480  */
   1481 int
   1482 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1483 {
   1484 	int newoff;
   1485 	int nxt;
   1486 
   1487 	if (!nxtp) {
   1488 		nxt = -1;
   1489 		nxtp = &nxt;
   1490 	}
   1491 	for (;;) {
   1492 		newoff = ip6_nexthdr(m, off, proto, nxtp);
   1493 		if (newoff < 0)
   1494 			return off;
   1495 		else if (newoff < off)
   1496 			return -1;	/* invalid */
   1497 		else if (newoff == off)
   1498 			return newoff;
   1499 
   1500 		off = newoff;
   1501 		proto = *nxtp;
   1502 	}
   1503 }
   1504 
   1505 struct m_tag *
   1506 ip6_addaux(struct mbuf *m)
   1507 {
   1508 	struct m_tag *mtag;
   1509 
   1510 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1511 	if (!mtag) {
   1512 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
   1513 		    M_NOWAIT);
   1514 		if (mtag) {
   1515 			m_tag_prepend(m, mtag);
   1516 			memset(mtag + 1, 0, sizeof(struct ip6aux));
   1517 		}
   1518 	}
   1519 	return mtag;
   1520 }
   1521 
   1522 struct m_tag *
   1523 ip6_findaux(struct mbuf *m)
   1524 {
   1525 	struct m_tag *mtag;
   1526 
   1527 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1528 	return mtag;
   1529 }
   1530 
   1531 void
   1532 ip6_delaux(struct mbuf *m)
   1533 {
   1534 	struct m_tag *mtag;
   1535 
   1536 	mtag = m_tag_find(m, PACKET_TAG_INET6, NULL);
   1537 	if (mtag)
   1538 		m_tag_delete(m, mtag);
   1539 }
   1540 
   1541 #ifdef GATEWAY
   1542 /*
   1543  * sysctl helper routine for net.inet.ip6.maxflows. Since
   1544  * we could reduce this value, call ip6flow_reap();
   1545  */
   1546 static int
   1547 sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS)
   1548 {
   1549 	int error;
   1550 
   1551 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
   1552 	if (error || newp == NULL)
   1553 		return (error);
   1554 
   1555 	mutex_enter(softnet_lock);
   1556 	KERNEL_LOCK(1, NULL);
   1557 
   1558 	ip6flow_reap(0);
   1559 
   1560 	KERNEL_UNLOCK_ONE(NULL);
   1561 	mutex_exit(softnet_lock);
   1562 
   1563 	return (0);
   1564 }
   1565 
   1566 static int
   1567 sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS)
   1568 {
   1569 	int error, tmp;
   1570 	struct sysctlnode node;
   1571 
   1572 	node = *rnode;
   1573 	tmp = ip6_hashsize;
   1574 	node.sysctl_data = &tmp;
   1575 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1576 	if (error || newp == NULL)
   1577 		return (error);
   1578 
   1579 	if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
   1580 		/*
   1581 		 * Can only fail due to malloc()
   1582 		 */
   1583 		mutex_enter(softnet_lock);
   1584 		KERNEL_LOCK(1, NULL);
   1585 
   1586 		error = ip6flow_invalidate_all(tmp);
   1587 
   1588 		KERNEL_UNLOCK_ONE(NULL);
   1589 		mutex_exit(softnet_lock);
   1590 	} else {
   1591 		/*
   1592 		 * EINVAL if not a power of 2
   1593 		 */
   1594 		error = EINVAL;
   1595 	}
   1596 
   1597 	return error;
   1598 }
   1599 #endif /* GATEWAY */
   1600 
   1601 /*
   1602  * System control for IP6
   1603  */
   1604 
   1605 const u_char inet6ctlerrmap[PRC_NCMDS] = {
   1606 	0,		0,		0,		0,
   1607 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
   1608 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
   1609 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
   1610 	0,		0,		0,		0,
   1611 	ENOPROTOOPT
   1612 };
   1613 
   1614 extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS);
   1615 
   1616 static int
   1617 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
   1618 {
   1619 
   1620 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
   1621 }
   1622 
   1623 static void
   1624 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
   1625 {
   1626 #ifdef RFC2292
   1627 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1628 #else
   1629 #define IS2292(x, y)	(y)
   1630 #endif
   1631 
   1632 	sysctl_createv(clog, 0, NULL, NULL,
   1633 		       CTLFLAG_PERMANENT,
   1634 		       CTLTYPE_NODE, "inet6",
   1635 		       SYSCTL_DESCR("PF_INET6 related settings"),
   1636 		       NULL, 0, NULL, 0,
   1637 		       CTL_NET, PF_INET6, CTL_EOL);
   1638 	sysctl_createv(clog, 0, NULL, NULL,
   1639 		       CTLFLAG_PERMANENT,
   1640 		       CTLTYPE_NODE, "ip6",
   1641 		       SYSCTL_DESCR("IPv6 related settings"),
   1642 		       NULL, 0, NULL, 0,
   1643 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
   1644 
   1645 	sysctl_createv(clog, 0, NULL, NULL,
   1646 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1647 		       CTLTYPE_INT, "forwarding",
   1648 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
   1649 		       NULL, 0, &ip6_forwarding, 0,
   1650 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1651 		       IPV6CTL_FORWARDING, CTL_EOL);
   1652 	sysctl_createv(clog, 0, NULL, NULL,
   1653 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1654 		       CTLTYPE_INT, "redirect",
   1655 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
   1656 		       NULL, 0, &ip6_sendredirects, 0,
   1657 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1658 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
   1659 	sysctl_createv(clog, 0, NULL, NULL,
   1660 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1661 		       CTLTYPE_INT, "hlim",
   1662 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
   1663 		       NULL, 0, &ip6_defhlim, 0,
   1664 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1665 		       IPV6CTL_DEFHLIM, CTL_EOL);
   1666 #ifdef notyet
   1667 	sysctl_createv(clog, 0, NULL, NULL,
   1668 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1669 		       CTLTYPE_INT, "mtu", NULL,
   1670 		       NULL, 0, &, 0,
   1671 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1672 		       IPV6CTL_DEFMTU, CTL_EOL);
   1673 #endif
   1674 #ifdef __no_idea__
   1675 	sysctl_createv(clog, 0, NULL, NULL,
   1676 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1677 		       CTLTYPE_INT, "forwsrcrt", NULL,
   1678 		       NULL, 0, &?, 0,
   1679 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1680 		       IPV6CTL_FORWSRCRT, CTL_EOL);
   1681 	sysctl_createv(clog, 0, NULL, NULL,
   1682 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1683 		       CTLTYPE_STRUCT, "mrtstats", NULL,
   1684 		       NULL, 0, &?, sizeof(?),
   1685 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1686 		       IPV6CTL_MRTSTATS, CTL_EOL);
   1687 	sysctl_createv(clog, 0, NULL, NULL,
   1688 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1689 		       CTLTYPE_?, "mrtproto", NULL,
   1690 		       NULL, 0, &?, sizeof(?),
   1691 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1692 		       IPV6CTL_MRTPROTO, CTL_EOL);
   1693 #endif
   1694 	sysctl_createv(clog, 0, NULL, NULL,
   1695 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1696 		       CTLTYPE_INT, "maxfragpackets",
   1697 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
   1698 				    "for reassembly"),
   1699 		       NULL, 0, &ip6_maxfragpackets, 0,
   1700 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1701 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
   1702 #ifdef __no_idea__
   1703 	sysctl_createv(clog, 0, NULL, NULL,
   1704 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1705 		       CTLTYPE_INT, "sourcecheck", NULL,
   1706 		       NULL, 0, &?, 0,
   1707 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1708 		       IPV6CTL_SOURCECHECK, CTL_EOL);
   1709 	sysctl_createv(clog, 0, NULL, NULL,
   1710 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1711 		       CTLTYPE_INT, "sourcecheck_logint", NULL,
   1712 		       NULL, 0, &?, 0,
   1713 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1714 		       IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL);
   1715 #endif
   1716 	sysctl_createv(clog, 0, NULL, NULL,
   1717 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1718 		       CTLTYPE_INT, "accept_rtadv",
   1719 		       SYSCTL_DESCR("Accept router advertisements"),
   1720 		       NULL, 0, &ip6_accept_rtadv, 0,
   1721 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1722 		       IPV6CTL_ACCEPT_RTADV, CTL_EOL);
   1723 	sysctl_createv(clog, 0, NULL, NULL,
   1724 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1725 		       CTLTYPE_INT, "rtadv_maxroutes",
   1726 		       SYSCTL_DESCR("Maximum number of routes accepted via router advertisements"),
   1727 		       NULL, 0, &ip6_rtadv_maxroutes, 0,
   1728 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1729 		       IPV6CTL_RTADV_MAXROUTES, CTL_EOL);
   1730 	sysctl_createv(clog, 0, NULL, NULL,
   1731 		       CTLFLAG_PERMANENT,
   1732 		       CTLTYPE_INT, "rtadv_numroutes",
   1733 		       SYSCTL_DESCR("Current number of routes accepted via router advertisements"),
   1734 		       NULL, 0, &nd6_numroutes, 0,
   1735 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1736 		       IPV6CTL_RTADV_NUMROUTES, CTL_EOL);
   1737 	sysctl_createv(clog, 0, NULL, NULL,
   1738 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1739 		       CTLTYPE_INT, "keepfaith",
   1740 		       SYSCTL_DESCR("Activate faith interface"),
   1741 		       NULL, 0, &ip6_keepfaith, 0,
   1742 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1743 		       IPV6CTL_KEEPFAITH, CTL_EOL);
   1744 	sysctl_createv(clog, 0, NULL, NULL,
   1745 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1746 		       CTLTYPE_INT, "log_interval",
   1747 		       SYSCTL_DESCR("Minumum interval between logging "
   1748 				    "unroutable packets"),
   1749 		       NULL, 0, &ip6_log_interval, 0,
   1750 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1751 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
   1752 	sysctl_createv(clog, 0, NULL, NULL,
   1753 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1754 		       CTLTYPE_INT, "hdrnestlimit",
   1755 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
   1756 		       NULL, 0, &ip6_hdrnestlimit, 0,
   1757 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1758 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
   1759 	sysctl_createv(clog, 0, NULL, NULL,
   1760 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1761 		       CTLTYPE_INT, "dad_count",
   1762 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
   1763 				    "probes to send"),
   1764 		       NULL, 0, &ip6_dad_count, 0,
   1765 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1766 		       IPV6CTL_DAD_COUNT, CTL_EOL);
   1767 	sysctl_createv(clog, 0, NULL, NULL,
   1768 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1769 		       CTLTYPE_INT, "auto_flowlabel",
   1770 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
   1771 		       NULL, 0, &ip6_auto_flowlabel, 0,
   1772 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1773 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
   1774 	sysctl_createv(clog, 0, NULL, NULL,
   1775 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1776 		       CTLTYPE_INT, "defmcasthlim",
   1777 		       SYSCTL_DESCR("Default multicast hop limit"),
   1778 		       NULL, 0, &ip6_defmcasthlim, 0,
   1779 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1780 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
   1781 #if NGIF > 0
   1782 	sysctl_createv(clog, 0, NULL, NULL,
   1783 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1784 		       CTLTYPE_INT, "gifhlim",
   1785 		       SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"),
   1786 		       NULL, 0, &ip6_gif_hlim, 0,
   1787 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1788 		       IPV6CTL_GIF_HLIM, CTL_EOL);
   1789 #endif /* NGIF */
   1790 	sysctl_createv(clog, 0, NULL, NULL,
   1791 		       CTLFLAG_PERMANENT,
   1792 		       CTLTYPE_STRING, "kame_version",
   1793 		       SYSCTL_DESCR("KAME Version"),
   1794 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
   1795 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1796 		       IPV6CTL_KAME_VERSION, CTL_EOL);
   1797 	sysctl_createv(clog, 0, NULL, NULL,
   1798 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1799 		       CTLTYPE_INT, "use_deprecated",
   1800 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
   1801 				    "source addresses"),
   1802 		       NULL, 0, &ip6_use_deprecated, 0,
   1803 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1804 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
   1805 	sysctl_createv(clog, 0, NULL, NULL,
   1806 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1807 		       CTLTYPE_INT, "rr_prune", NULL,
   1808 		       NULL, 0, &ip6_rr_prune, 0,
   1809 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1810 		       IPV6CTL_RR_PRUNE, CTL_EOL);
   1811 	sysctl_createv(clog, 0, NULL, NULL,
   1812 		       CTLFLAG_PERMANENT
   1813 #ifndef INET6_BINDV6ONLY
   1814 		       |CTLFLAG_READWRITE,
   1815 #endif
   1816 		       CTLTYPE_INT, "v6only",
   1817 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
   1818 				    "to PF_INET sockets"),
   1819 		       NULL, 0, &ip6_v6only, 0,
   1820 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1821 		       IPV6CTL_V6ONLY, CTL_EOL);
   1822 	sysctl_createv(clog, 0, NULL, NULL,
   1823 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1824 		       CTLTYPE_INT, "anonportmin",
   1825 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
   1826 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
   1827 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1828 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
   1829 	sysctl_createv(clog, 0, NULL, NULL,
   1830 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1831 		       CTLTYPE_INT, "anonportmax",
   1832 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
   1833 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
   1834 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1835 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
   1836 #ifndef IPNOPRIVPORTS
   1837 	sysctl_createv(clog, 0, NULL, NULL,
   1838 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1839 		       CTLTYPE_INT, "lowportmin",
   1840 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
   1841 				    "to assign"),
   1842 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
   1843 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1844 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
   1845 	sysctl_createv(clog, 0, NULL, NULL,
   1846 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1847 		       CTLTYPE_INT, "lowportmax",
   1848 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
   1849 				    "to assign"),
   1850 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
   1851 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1852 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
   1853 #endif /* IPNOPRIVPORTS */
   1854 	sysctl_createv(clog, 0, NULL, NULL,
   1855 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1856 		       CTLTYPE_INT, "auto_linklocal",
   1857 		       SYSCTL_DESCR("Default value of per-interface flag for "
   1858 		                    "adding an IPv6 link-local address to "
   1859 				    "interfaces when attached"),
   1860 		       NULL, 0, &ip6_auto_linklocal, 0,
   1861 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1862 		       IPV6CTL_AUTO_LINKLOCAL, CTL_EOL);
   1863 	sysctl_createv(clog, 0, NULL, NULL,
   1864 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
   1865 		       CTLTYPE_STRUCT, "addctlpolicy",
   1866 		       SYSCTL_DESCR("Return the current address control"
   1867 			   " policy"),
   1868 		       sysctl_net_inet6_addrctlpolicy, 0, NULL, 0,
   1869 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1870 		       IPV6CTL_ADDRCTLPOLICY, CTL_EOL);
   1871 	sysctl_createv(clog, 0, NULL, NULL,
   1872 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1873 		       CTLTYPE_INT, "use_tempaddr",
   1874 		       SYSCTL_DESCR("Use temporary address"),
   1875 		       NULL, 0, &ip6_use_tempaddr, 0,
   1876 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1877 		       CTL_CREATE, CTL_EOL);
   1878 	sysctl_createv(clog, 0, NULL, NULL,
   1879 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1880 		       CTLTYPE_INT, "prefer_tempaddr",
   1881 		       SYSCTL_DESCR("Prefer temporary address as source "
   1882 		                    "address"),
   1883 		       NULL, 0, &ip6_prefer_tempaddr, 0,
   1884 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1885 		       CTL_CREATE, CTL_EOL);
   1886 	sysctl_createv(clog, 0, NULL, NULL,
   1887 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1888 		       CTLTYPE_INT, "temppltime",
   1889 		       SYSCTL_DESCR("preferred lifetime of a temporary address"),
   1890 		       NULL, 0, &ip6_temp_preferred_lifetime, 0,
   1891 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1892 		       CTL_CREATE, CTL_EOL);
   1893 	sysctl_createv(clog, 0, NULL, NULL,
   1894 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1895 		       CTLTYPE_INT, "tempvltime",
   1896 		       SYSCTL_DESCR("valid lifetime of a temporary address"),
   1897 		       NULL, 0, &ip6_temp_valid_lifetime, 0,
   1898 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1899 		       CTL_CREATE, CTL_EOL);
   1900 	sysctl_createv(clog, 0, NULL, NULL,
   1901 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1902 		       CTLTYPE_INT, "maxfrags",
   1903 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
   1904 		       NULL, 0, &ip6_maxfrags, 0,
   1905 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1906 		       IPV6CTL_MAXFRAGS, CTL_EOL);
   1907 	sysctl_createv(clog, 0, NULL, NULL,
   1908 		       CTLFLAG_PERMANENT,
   1909 		       CTLTYPE_STRUCT, "stats",
   1910 		       SYSCTL_DESCR("IPv6 statistics"),
   1911 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
   1912 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1913 		       IPV6CTL_STATS, CTL_EOL);
   1914 	sysctl_createv(clog, 0, NULL, NULL,
   1915 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1916 		       CTLTYPE_INT, "use_defaultzone",
   1917 		       SYSCTL_DESCR("Whether to use the default scope zones"),
   1918 		       NULL, 0, &ip6_use_defzone, 0,
   1919 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1920 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
   1921 	sysctl_createv(clog, 0, NULL, NULL,
   1922 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1923 		       CTLTYPE_INT, "mcast_pmtu",
   1924 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
   1925 		       NULL, 0, &ip6_mcast_pmtu, 0,
   1926 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1927 		       CTL_CREATE, CTL_EOL);
   1928 #ifdef GATEWAY
   1929 	sysctl_createv(clog, 0, NULL, NULL,
   1930 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1931 			CTLTYPE_INT, "maxflows",
   1932 			SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"),
   1933 			sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0,
   1934 			CTL_NET, PF_INET6, IPPROTO_IPV6,
   1935 			CTL_CREATE, CTL_EOL);
   1936 	sysctl_createv(clog, 0, NULL, NULL,
   1937 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1938 			CTLTYPE_INT, "hashsize",
   1939 			SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"),
   1940 			sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0,
   1941 			CTL_NET, PF_INET6, IPPROTO_IPV6,
   1942 			CTL_CREATE, CTL_EOL);
   1943 #endif
   1944 	/* anonportalgo RFC6056 subtree */
   1945 	const struct sysctlnode *portalgo_node;
   1946 	sysctl_createv(clog, 0, NULL, &portalgo_node,
   1947 		       CTLFLAG_PERMANENT,
   1948 		       CTLTYPE_NODE, "anonportalgo",
   1949 		       SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"),
   1950 	    	       NULL, 0, NULL, 0,
   1951 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL);
   1952 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1953 		       CTLFLAG_PERMANENT,
   1954 		       CTLTYPE_STRING, "available",
   1955 		       SYSCTL_DESCR("available algorithms"),
   1956 		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
   1957 		       CTL_CREATE, CTL_EOL);
   1958 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1959 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1960 		       CTLTYPE_STRING, "selected",
   1961 		       SYSCTL_DESCR("selected algorithm"),
   1962 	               sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN,
   1963 		       CTL_CREATE, CTL_EOL);
   1964 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1965 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1966 		       CTLTYPE_STRUCT, "reserve",
   1967 		       SYSCTL_DESCR("bitmap of reserved ports"),
   1968 		       sysctl_portalgo_reserve6, 0, NULL, 0,
   1969 		       CTL_CREATE, CTL_EOL);
   1970 	sysctl_createv(clog, 0, NULL, NULL,
   1971 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1972 		       CTLTYPE_INT, "neighborgcthresh",
   1973 		       SYSCTL_DESCR("Maximum number of entries in neighbor"
   1974 			" cache"),
   1975 		       NULL, 1, &ip6_neighborgcthresh, 0,
   1976 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1977 		       CTL_CREATE, CTL_EOL);
   1978 	sysctl_createv(clog, 0, NULL, NULL,
   1979 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1980 		       CTLTYPE_INT, "maxifprefixes",
   1981 		       SYSCTL_DESCR("Maximum number of prefixes created by"
   1982 			   " route advertisement per interface"),
   1983 		       NULL, 1, &ip6_maxifprefixes, 0,
   1984 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1985 		       CTL_CREATE, CTL_EOL);
   1986 	sysctl_createv(clog, 0, NULL, NULL,
   1987 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1988 		       CTLTYPE_INT, "maxifdefrouters",
   1989 		       SYSCTL_DESCR("Maximum number of default routers created"
   1990 			   " by route advertisement per interface"),
   1991 		       NULL, 1, &ip6_maxifdefrouters, 0,
   1992 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1993 		       CTL_CREATE, CTL_EOL);
   1994 	sysctl_createv(clog, 0, NULL, NULL,
   1995 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1996 		       CTLTYPE_INT, "maxdynroutes",
   1997 		       SYSCTL_DESCR("Maximum number of routes created via"
   1998 			   " redirect"),
   1999 		       NULL, 1, &ip6_maxdynroutes, 0,
   2000 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   2001 		       CTL_CREATE, CTL_EOL);
   2002 }
   2003 
   2004 void
   2005 ip6_statinc(u_int stat)
   2006 {
   2007 
   2008 	KASSERT(stat < IP6_NSTATS);
   2009 	IP6_STATINC(stat);
   2010 }
   2011