Home | History | Annotate | Line # | Download | only in netinet6
ip6_input.c revision 1.36
      1 /*	$NetBSD: ip6_input.c,v 1.36 2001/02/24 00:02:16 cgd Exp $	*/
      2 /*	$KAME: ip6_input.c,v 1.174 2001/02/09 06:17:41 jinmei 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. All advertising materials mentioning features or use of this software
     46  *    must display the following acknowledgement:
     47  *	This product includes software developed by the University of
     48  *	California, Berkeley and its contributors.
     49  * 4. Neither the name of the University nor the names of its contributors
     50  *    may be used to endorse or promote products derived from this software
     51  *    without specific prior written permission.
     52  *
     53  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     54  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     55  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     56  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     57  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     58  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     59  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     60  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     61  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     62  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     63  * SUCH DAMAGE.
     64  *
     65  *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
     66  */
     67 
     68 #include "opt_inet.h"
     69 #include "opt_ipsec.h"
     70 #include "opt_pfil_hooks.h"
     71 
     72 #include <sys/param.h>
     73 #include <sys/systm.h>
     74 #include <sys/malloc.h>
     75 #include <sys/mbuf.h>
     76 #include <sys/domain.h>
     77 #include <sys/protosw.h>
     78 #include <sys/socket.h>
     79 #include <sys/socketvar.h>
     80 #include <sys/errno.h>
     81 #include <sys/time.h>
     82 #include <sys/kernel.h>
     83 #include <sys/syslog.h>
     84 #include <sys/proc.h>
     85 
     86 #include <net/if.h>
     87 #include <net/if_types.h>
     88 #include <net/if_dl.h>
     89 #include <net/route.h>
     90 #include <net/netisr.h>
     91 #ifdef PFIL_HOOKS
     92 #include <net/pfil.h>
     93 #endif
     94 
     95 #include <netinet/in.h>
     96 #include <netinet/in_systm.h>
     97 #ifdef INET
     98 #include <netinet/ip.h>
     99 #include <netinet/ip_icmp.h>
    100 #endif /*INET*/
    101 #include <netinet/ip6.h>
    102 #include <netinet6/in6_var.h>
    103 #include <netinet6/ip6_var.h>
    104 #include <netinet6/in6_pcb.h>
    105 #include <netinet/icmp6.h>
    106 #include <netinet6/in6_ifattach.h>
    107 #include <netinet6/nd6.h>
    108 #include <netinet6/in6_prefix.h>
    109 
    110 #include <netinet6/ip6protosw.h>
    111 
    112 /* we need it for NLOOP. */
    113 #include "loop.h"
    114 #include "faith.h"
    115 
    116 #include "gif.h"
    117 #include "bpfilter.h"
    118 
    119 #include <net/net_osdep.h>
    120 
    121 extern struct domain inet6domain;
    122 
    123 u_char ip6_protox[IPPROTO_MAX];
    124 static int ip6qmaxlen = IFQ_MAXLEN;
    125 struct in6_ifaddr *in6_ifaddr;
    126 struct ifqueue ip6intrq;
    127 
    128 extern struct ifnet loif[NLOOP];
    129 int ip6_forward_srcrt;			/* XXX */
    130 int ip6_sourcecheck;			/* XXX */
    131 int ip6_sourcecheck_interval;		/* XXX */
    132 
    133 #ifdef PFIL_HOOKS
    134 struct pfil_head inet6_pfil_hook;
    135 #endif
    136 
    137 struct ip6stat ip6stat;
    138 
    139 static void ip6_init2 __P((void *));
    140 
    141 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
    142 
    143 /*
    144  * IP6 initialization: fill in IP6 protocol switch table.
    145  * All protocols not implemented in kernel go to raw IP6 protocol handler.
    146  */
    147 void
    148 ip6_init()
    149 {
    150 	struct ip6protosw *pr;
    151 	int i;
    152 	struct timeval tv;
    153 
    154 	pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
    155 	if (pr == 0)
    156 		panic("ip6_init");
    157 	for (i = 0; i < IPPROTO_MAX; i++)
    158 		ip6_protox[i] = pr - inet6sw;
    159 	for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
    160 	    pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
    161 		if (pr->pr_domain->dom_family == PF_INET6 &&
    162 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
    163 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
    164 	ip6intrq.ifq_maxlen = ip6qmaxlen;
    165 	nd6_init();
    166 	frag6_init();
    167 	/*
    168 	 * in many cases, random() here does NOT return random number
    169 	 * as initialization during bootstrap time occur in fixed order.
    170 	 */
    171 	microtime(&tv);
    172 	ip6_flow_seq = random() ^ tv.tv_usec;
    173 
    174 	ip6_init2((void *)0);
    175 
    176 #ifdef PFIL_HOOKS
    177 	/* Register our Packet Filter hook. */
    178 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
    179 	inet6_pfil_hook.ph_af   = AF_INET6;
    180 	i = pfil_head_register(&inet6_pfil_hook);
    181 	if (i != 0)
    182 		printf("ip6_init: WARNING: unable to register pfil hook, "
    183 		    "error %d\n", i);
    184 #endif /* PFIL_HOOKS */
    185 }
    186 
    187 static void
    188 ip6_init2(dummy)
    189 	void *dummy;
    190 {
    191 	/*
    192 	 * to route local address of p2p link to loopback,
    193 	 * assign loopback address first.
    194 	 */
    195 	in6_ifattach(&loif[0], NULL);
    196 
    197 	/* nd6_timer_init */
    198 	callout_init(&nd6_timer_ch);
    199 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
    200 	/* router renumbering prefix list maintenance */
    201 	callout_init(&in6_rr_timer_ch);
    202 	callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
    203 }
    204 
    205 /*
    206  * IP6 input interrupt handling. Just pass the packet to ip6_input.
    207  */
    208 void
    209 ip6intr()
    210 {
    211 	int s;
    212 	struct mbuf *m;
    213 
    214 	for (;;) {
    215 		s = splimp();
    216 		IF_DEQUEUE(&ip6intrq, m);
    217 		splx(s);
    218 		if (m == 0)
    219 			return;
    220 		ip6_input(m);
    221 	}
    222 }
    223 
    224 extern struct	route_in6 ip6_forward_rt;
    225 
    226 void
    227 ip6_input(m)
    228 	struct mbuf *m;
    229 {
    230 	struct ip6_hdr *ip6;
    231 	int off = sizeof(struct ip6_hdr), nest;
    232 	u_int32_t plen;
    233 	u_int32_t rtalert = ~0;
    234 	int nxt, ours = 0;
    235 	struct ifnet *deliverifp = NULL;
    236 
    237 #ifdef IPSEC
    238 	/*
    239 	 * should the inner packet be considered authentic?
    240 	 * see comment in ah4_input().
    241 	 */
    242 	if (m) {
    243 		m->m_flags &= ~M_AUTHIPHDR;
    244 		m->m_flags &= ~M_AUTHIPDGM;
    245 	}
    246 #endif
    247 
    248 	/*
    249 	 * mbuf statistics by kazu
    250 	 */
    251 	if (m->m_flags & M_EXT) {
    252 		if (m->m_next)
    253 			ip6stat.ip6s_mext2m++;
    254 		else
    255 			ip6stat.ip6s_mext1++;
    256 	} else {
    257 		if (m->m_next) {
    258 			if (m->m_flags & M_LOOP) {
    259 				ip6stat.ip6s_m2m[loif[0].if_index]++;	/*XXX*/
    260 			} else if (m->m_pkthdr.rcvif->if_index <= 31)
    261 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
    262 			else
    263 				ip6stat.ip6s_m2m[0]++;
    264 		} else
    265 			ip6stat.ip6s_m1++;
    266 	}
    267 
    268 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
    269 	ip6stat.ip6s_total++;
    270 
    271 #ifndef PULLDOWN_TEST
    272 	/* XXX is the line really necessary? */
    273 	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
    274 #endif
    275 
    276 	if (m->m_len < sizeof(struct ip6_hdr)) {
    277 		struct ifnet *inifp;
    278 		inifp = m->m_pkthdr.rcvif;
    279 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
    280 			ip6stat.ip6s_toosmall++;
    281 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
    282 			return;
    283 		}
    284 	}
    285 
    286 	ip6 = mtod(m, struct ip6_hdr *);
    287 
    288 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    289 		ip6stat.ip6s_badvers++;
    290 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    291 		goto bad;
    292 	}
    293 
    294 #ifdef PFIL_HOOKS
    295 	/*
    296 	 * Run through list of hooks for input packets.  If there are any
    297 	 * filters which require that additional packets in the flow are
    298 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
    299 	 * Note that filters must _never_ set this flag, as another filter
    300 	 * in the list may have previously cleared it.
    301 	 */
    302 	if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
    303 			   PFIL_IN) != 0)
    304 		return;
    305 	if (m == NULL)
    306 		return;
    307 	ip6 = mtod(m, struct ip6_hdr *);
    308 #endif /* PFIL_HOOKS */
    309 
    310 
    311 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
    312 
    313 #ifdef ALTQ
    314 	/* XXX Temporary until ALTQ is changed to use a pfil hook */
    315 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
    316 		/* packet is dropped by traffic conditioner */
    317 		return;
    318 	}
    319 #endif
    320 
    321 	/*
    322 	 * Scope check
    323 	 */
    324 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
    325 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
    326 		ip6stat.ip6s_badscope++;
    327 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    328 		goto bad;
    329 	}
    330 	/*
    331 	 * The following check is not documented in the spec.  Malicious party
    332 	 * may be able to use IPv4 mapped addr to confuse tcp/udp stack and
    333 	 * bypass security checks (act as if it was from 127.0.0.1 by using
    334 	 * IPv6 src ::ffff:127.0.0.1).	Be cautious.
    335 	 *
    336 	 * This check chokes if we are in SIIT cloud.  As none of BSDs support
    337 	 * IPv4-less kernel compilation, we cannot support SIIT environment
    338 	 * at all.  So, it makes more sense for us to reject any malicious
    339 	 * packets for non-SIIT environment, than try to do a partical support
    340 	 * for SIIT environment.
    341 	 */
    342 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    343 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    344 		ip6stat.ip6s_badscope++;
    345 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    346 		goto bad;
    347 	}
    348 #if 0
    349 	/*
    350 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
    351 	 *
    352 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
    353 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
    354 	 * is revised to forbid relaying case.
    355 	 */
    356 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
    357 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
    358 		ip6stat.ip6s_badscope++;
    359 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    360 		goto bad;
    361 	}
    362 #endif
    363 
    364 	if (IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
    365 	    IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) {
    366 		if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) {
    367 			ours = 1;
    368 			deliverifp = m->m_pkthdr.rcvif;
    369 			goto hbhcheck;
    370 		} else {
    371 			ip6stat.ip6s_badscope++;
    372 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    373 			goto bad;
    374 		}
    375 	}
    376 
    377 #ifndef FAKE_LOOPBACK_IF
    378 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0)
    379 #else
    380 	if (1)
    381 #endif
    382 	{
    383 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
    384 			ip6->ip6_src.s6_addr16[1]
    385 				= htons(m->m_pkthdr.rcvif->if_index);
    386 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
    387 			ip6->ip6_dst.s6_addr16[1]
    388 				= htons(m->m_pkthdr.rcvif->if_index);
    389 	}
    390 
    391 	/*
    392 	 * XXX we need this since we do not have "goto ours" hack route
    393 	 * for some of our ifaddrs on loopback interface.
    394 	 * we should correct it by changing in6_ifattach to install
    395 	 * "goto ours" hack route.
    396 	 */
    397 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0) {
    398 		if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
    399 			ours = 1;
    400 			deliverifp = m->m_pkthdr.rcvif;
    401 			goto hbhcheck;
    402 		}
    403 	}
    404 
    405 	/*
    406 	 * Multicast check
    407 	 */
    408 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    409 	  	struct	in6_multi *in6m = 0;
    410 
    411 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
    412 		/*
    413 		 * See if we belong to the destination multicast group on the
    414 		 * arrival interface.
    415 		 */
    416 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
    417 		if (in6m)
    418 			ours = 1;
    419 		else if (!ip6_mrouter) {
    420 			ip6stat.ip6s_notmember++;
    421 			ip6stat.ip6s_cantforward++;
    422 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    423 			goto bad;
    424 		}
    425 		deliverifp = m->m_pkthdr.rcvif;
    426 		goto hbhcheck;
    427 	}
    428 
    429 	/*
    430 	 *  Unicast check
    431 	 */
    432 	if (ip6_forward_rt.ro_rt != NULL &&
    433 	    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
    434 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
    435 			       &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
    436 		ip6stat.ip6s_forward_cachehit++;
    437 	else {
    438 		struct sockaddr_in6 *dst6;
    439 
    440 		if (ip6_forward_rt.ro_rt) {
    441 			/* route is down or destination is different */
    442 			ip6stat.ip6s_forward_cachemiss++;
    443 			RTFREE(ip6_forward_rt.ro_rt);
    444 			ip6_forward_rt.ro_rt = 0;
    445 		}
    446 
    447 		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
    448 		dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
    449 		dst6->sin6_len = sizeof(struct sockaddr_in6);
    450 		dst6->sin6_family = AF_INET6;
    451 		dst6->sin6_addr = ip6->ip6_dst;
    452 
    453 		rtalloc((struct route *)&ip6_forward_rt);
    454 	}
    455 
    456 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
    457 
    458 	/*
    459 	 * Accept the packet if the forwarding interface to the destination
    460 	 * according to the routing table is the loopback interface,
    461 	 * unless the associated route has a gateway.
    462 	 * Note that this approach causes to accept a packet if there is a
    463 	 * route to the loopback interface for the destination of the packet.
    464 	 * But we think it's even useful in some situations, e.g. when using
    465 	 * a special daemon which wants to intercept the packet.
    466 	 */
    467 	if (ip6_forward_rt.ro_rt &&
    468 	    (ip6_forward_rt.ro_rt->rt_flags &
    469 	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
    470 #if 0
    471 	    /*
    472 	     * The check below is redundant since the comparison of
    473 	     * the destination and the key of the rtentry has
    474 	     * already done through looking up the routing table.
    475 	     */
    476 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
    477 			       &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
    478 #endif
    479 	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
    480 		struct in6_ifaddr *ia6 =
    481 			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
    482 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
    483 			m->m_flags |= M_ANYCAST6;
    484 		/*
    485 		 * packets to a tentative, duplicated, or somehow invalid
    486 		 * address must not be accepted.
    487 		 */
    488 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
    489 			/* this address is ready */
    490 			ours = 1;
    491 			deliverifp = ia6->ia_ifp;	/* correct? */
    492 			goto hbhcheck;
    493 		} else {
    494 			/* address is not ready, so discard the packet. */
    495 			nd6log((LOG_INFO,
    496 			    "ip6_input: packet to an unready address %s->%s\n",
    497 			    ip6_sprintf(&ip6->ip6_src),
    498 			    ip6_sprintf(&ip6->ip6_dst)));
    499 
    500 			goto bad;
    501 		}
    502 	}
    503 
    504 	/*
    505 	 * FAITH(Firewall Aided Internet Translator)
    506 	 */
    507 #if defined(NFAITH) && 0 < NFAITH
    508 	if (ip6_keepfaith) {
    509 		if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
    510 		 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
    511 			/* XXX do we need more sanity checks? */
    512 			ours = 1;
    513 			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /*faith*/
    514 			goto hbhcheck;
    515 		}
    516 	}
    517 #endif
    518 
    519 #if 0
    520     {
    521 	/*
    522 	 * Last resort: check in6_ifaddr for incoming interface.
    523 	 * The code is here until I update the "goto ours hack" code above
    524 	 * working right.
    525 	 */
    526 	struct ifaddr *ifa;
    527 	for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
    528 	     ifa;
    529 	     ifa = ifa->ifa_list.tqe_next) {
    530 		if (ifa->ifa_addr == NULL)
    531 			continue;	/* just for safety */
    532 		if (ifa->ifa_addr->sa_family != AF_INET6)
    533 			continue;
    534 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
    535 			ours = 1;
    536 			deliverifp = ifa->ifa_ifp;
    537 			goto hbhcheck;
    538 		}
    539 	}
    540     }
    541 #endif
    542 
    543 	/*
    544 	 * Now there is no reason to process the packet if it's not our own
    545 	 * and we're not a router.
    546 	 */
    547 	if (!ip6_forwarding) {
    548 		ip6stat.ip6s_cantforward++;
    549 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    550 		goto bad;
    551 	}
    552 
    553   hbhcheck:
    554 	/*
    555 	 * Process Hop-by-Hop options header if it's contained.
    556 	 * m may be modified in ip6_hopopts_input().
    557 	 * If a JumboPayload option is included, plen will also be modified.
    558 	 */
    559 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
    560 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
    561 		struct ip6_hbh *hbh;
    562 
    563 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
    564 #if 0	/*touches NULL pointer*/
    565 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    566 #endif
    567 			return;	/* m have already been freed */
    568 		}
    569 
    570 		/* adjust pointer */
    571 		ip6 = mtod(m, struct ip6_hdr *);
    572 
    573 		/*
    574 		 * if the payload length field is 0 and the next header field
    575 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
    576 		 * option MUST be included.
    577 		 */
    578 		if (ip6->ip6_plen == 0 && plen == 0) {
    579 			/*
    580 			 * Note that if a valid jumbo payload option is
    581 			 * contained, ip6_hoptops_input() must set a valid
    582 			 * (non-zero) payload length to the variable plen.
    583 			 */
    584 			ip6stat.ip6s_badoptions++;
    585 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
    586 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
    587 			icmp6_error(m, ICMP6_PARAM_PROB,
    588 				    ICMP6_PARAMPROB_HEADER,
    589 				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
    590 			return;
    591 		}
    592 #ifndef PULLDOWN_TEST
    593 		/* ip6_hopopts_input() ensures that mbuf is contiguous */
    594 		hbh = (struct ip6_hbh *)(ip6 + 1);
    595 #else
    596 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    597 			sizeof(struct ip6_hbh));
    598 		if (hbh == NULL) {
    599 			ip6stat.ip6s_tooshort++;
    600 			return;
    601 		}
    602 #endif
    603 		nxt = hbh->ip6h_nxt;
    604 
    605 		/*
    606 		 * accept the packet if a router alert option is included
    607 		 * and we act as an IPv6 router.
    608 		 */
    609 		if (rtalert != ~0 && ip6_forwarding)
    610 			ours = 1;
    611 	} else
    612 		nxt = ip6->ip6_nxt;
    613 
    614 	/*
    615 	 * Check that the amount of data in the buffers
    616 	 * is as at least much as the IPv6 header would have us expect.
    617 	 * Trim mbufs if longer than we expect.
    618 	 * Drop packet if shorter than we expect.
    619 	 */
    620 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
    621 		ip6stat.ip6s_tooshort++;
    622 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    623 		goto bad;
    624 	}
    625 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
    626 		if (m->m_len == m->m_pkthdr.len) {
    627 			m->m_len = sizeof(struct ip6_hdr) + plen;
    628 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
    629 		} else
    630 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
    631 	}
    632 
    633 	/*
    634 	 * Forward if desirable.
    635 	 */
    636 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    637 		/*
    638 		 * If we are acting as a multicast router, all
    639 		 * incoming multicast packets are passed to the
    640 		 * kernel-level multicast forwarding function.
    641 		 * The packet is returned (relatively) intact; if
    642 		 * ip6_mforward() returns a non-zero value, the packet
    643 		 * must be discarded, else it may be accepted below.
    644 		 */
    645 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
    646 			ip6stat.ip6s_cantforward++;
    647 			m_freem(m);
    648 			return;
    649 		}
    650 		if (!ours) {
    651 			m_freem(m);
    652 			return;
    653 		}
    654 	} else if (!ours) {
    655 		ip6_forward(m, 0);
    656 		return;
    657 	}
    658 
    659 	ip6 = mtod(m, struct ip6_hdr *);
    660 
    661 	/*
    662 	 * Malicious party may be able to use IPv4 mapped addr to confuse
    663 	 * tcp/udp stack and bypass security checks (act as if it was from
    664 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
    665 	 *
    666 	 * For SIIT end node behavior, you may want to disable the check.
    667 	 * However, you will  become vulnerable to attacks using IPv4 mapped
    668 	 * source.
    669 	 */
    670 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    671 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    672 		ip6stat.ip6s_badscope++;
    673 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
    674 		goto bad;
    675 	}
    676 
    677 	/*
    678 	 * Tell launch routine the next header
    679 	 */
    680 #ifdef IFA_STATS
    681 	if (deliverifp != NULL) {
    682 		struct in6_ifaddr *ia6;
    683 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    684 		if (ia6)
    685 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
    686 	}
    687 #endif
    688 	ip6stat.ip6s_delivered++;
    689 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
    690 	nest = 0;
    691 	while (nxt != IPPROTO_DONE) {
    692 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
    693 			ip6stat.ip6s_toomanyhdr++;
    694 			goto bad;
    695 		}
    696 
    697 		/*
    698 		 * protection against faulty packet - there should be
    699 		 * more sanity checks in header chain processing.
    700 		 */
    701 		if (m->m_pkthdr.len < off) {
    702 			ip6stat.ip6s_tooshort++;
    703 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
    704 			goto bad;
    705 		}
    706 
    707 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
    708 	}
    709 	return;
    710  bad:
    711 	m_freem(m);
    712 }
    713 
    714 /*
    715  * Hop-by-Hop options header processing. If a valid jumbo payload option is
    716  * included, the real payload length will be stored in plenp.
    717  */
    718 static int
    719 ip6_hopopts_input(plenp, rtalertp, mp, offp)
    720 	u_int32_t *plenp;
    721 	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
    722 	struct mbuf **mp;
    723 	int *offp;
    724 {
    725 	struct mbuf *m = *mp;
    726 	int off = *offp, hbhlen;
    727 	struct ip6_hbh *hbh;
    728 	u_int8_t *opt;
    729 
    730 	/* validation of the length of the header */
    731 #ifndef PULLDOWN_TEST
    732 	IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
    733 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
    734 	hbhlen = (hbh->ip6h_len + 1) << 3;
    735 
    736 	IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
    737 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
    738 #else
    739 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
    740 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
    741 	if (hbh == NULL) {
    742 		ip6stat.ip6s_tooshort++;
    743 		return -1;
    744 	}
    745 	hbhlen = (hbh->ip6h_len + 1) << 3;
    746 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    747 		hbhlen);
    748 	if (hbh == NULL) {
    749 		ip6stat.ip6s_tooshort++;
    750 		return -1;
    751 	}
    752 #endif
    753 	off += hbhlen;
    754 	hbhlen -= sizeof(struct ip6_hbh);
    755 	opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
    756 
    757 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
    758 				hbhlen, rtalertp, plenp) < 0)
    759 		return(-1);
    760 
    761 	*offp = off;
    762 	*mp = m;
    763 	return(0);
    764 }
    765 
    766 /*
    767  * Search header for all Hop-by-hop options and process each option.
    768  * This function is separate from ip6_hopopts_input() in order to
    769  * handle a case where the sending node itself process its hop-by-hop
    770  * options header. In such a case, the function is called from ip6_output().
    771  */
    772 int
    773 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
    774 	struct mbuf *m;
    775 	u_int8_t *opthead;
    776 	int hbhlen;
    777 	u_int32_t *rtalertp;
    778 	u_int32_t *plenp;
    779 {
    780 	struct ip6_hdr *ip6;
    781 	int optlen = 0;
    782 	u_int8_t *opt = opthead;
    783 	u_int16_t rtalert_val;
    784 	u_int32_t jumboplen;
    785 
    786 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
    787 		switch (*opt) {
    788 		case IP6OPT_PAD1:
    789 			optlen = 1;
    790 			break;
    791 		case IP6OPT_PADN:
    792 			if (hbhlen < IP6OPT_MINLEN) {
    793 				ip6stat.ip6s_toosmall++;
    794 				goto bad;
    795 			}
    796 			optlen = *(opt + 1) + 2;
    797 			break;
    798 		case IP6OPT_RTALERT:
    799 			/* XXX may need check for alignment */
    800 			if (hbhlen < IP6OPT_RTALERT_LEN) {
    801 				ip6stat.ip6s_toosmall++;
    802 				goto bad;
    803 			}
    804 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
    805 				 /* XXX: should we discard the packet? */
    806 				log(LOG_ERR, "length of router alert opt is inconsitent(%d)",
    807 				    *(opt + 1));
    808 			}
    809 			optlen = IP6OPT_RTALERT_LEN;
    810 			bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
    811 			*rtalertp = ntohs(rtalert_val);
    812 			break;
    813 		case IP6OPT_JUMBO:
    814 			/* XXX may need check for alignment */
    815 			if (hbhlen < IP6OPT_JUMBO_LEN) {
    816 				ip6stat.ip6s_toosmall++;
    817 				goto bad;
    818 			}
    819 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
    820 				/* XXX: should we discard the packet? */
    821 				log(LOG_ERR, "length of jumbopayload opt "
    822 				    "is inconsistent(%d)\n",
    823 				    *(opt + 1));
    824 			}
    825 			optlen = IP6OPT_JUMBO_LEN;
    826 
    827 			/*
    828 			 * IPv6 packets that have non 0 payload length
    829 			 * must not contain a jumbo payload option.
    830 			 */
    831 			ip6 = mtod(m, struct ip6_hdr *);
    832 			if (ip6->ip6_plen) {
    833 				ip6stat.ip6s_badoptions++;
    834 				icmp6_error(m, ICMP6_PARAM_PROB,
    835 					    ICMP6_PARAMPROB_HEADER,
    836 					    sizeof(struct ip6_hdr) +
    837 					    sizeof(struct ip6_hbh) +
    838 					    opt - opthead);
    839 				return(-1);
    840 			}
    841 
    842 			/*
    843 			 * We may see jumbolen in unaligned location, so
    844 			 * we'd need to perform bcopy().
    845 			 */
    846 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
    847 			jumboplen = (u_int32_t)htonl(jumboplen);
    848 
    849 #if 1
    850 			/*
    851 			 * if there are multiple jumbo payload options,
    852 			 * *plenp will be non-zero and the packet will be
    853 			 * rejected.
    854 			 * the behavior may need some debate in ipngwg -
    855 			 * multiple options does not make sense, however,
    856 			 * there's no explicit mention in specification.
    857 			 */
    858 			if (*plenp != 0) {
    859 				ip6stat.ip6s_badoptions++;
    860 				icmp6_error(m, ICMP6_PARAM_PROB,
    861 					    ICMP6_PARAMPROB_HEADER,
    862 					    sizeof(struct ip6_hdr) +
    863 					    sizeof(struct ip6_hbh) +
    864 					    opt + 2 - opthead);
    865 				return(-1);
    866 			}
    867 #endif
    868 
    869 			/*
    870 			 * jumbo payload length must be larger than 65535.
    871 			 */
    872 			if (jumboplen <= IPV6_MAXPACKET) {
    873 				ip6stat.ip6s_badoptions++;
    874 				icmp6_error(m, ICMP6_PARAM_PROB,
    875 					    ICMP6_PARAMPROB_HEADER,
    876 					    sizeof(struct ip6_hdr) +
    877 					    sizeof(struct ip6_hbh) +
    878 					    opt + 2 - opthead);
    879 				return(-1);
    880 			}
    881 			*plenp = jumboplen;
    882 
    883 			break;
    884 		default:		/* unknown option */
    885 			if (hbhlen < IP6OPT_MINLEN) {
    886 				ip6stat.ip6s_toosmall++;
    887 				goto bad;
    888 			}
    889 			if ((optlen = ip6_unknown_opt(opt, m,
    890 						      sizeof(struct ip6_hdr) +
    891 						      sizeof(struct ip6_hbh) +
    892 						      opt - opthead)) == -1)
    893 				return(-1);
    894 			optlen += 2;
    895 			break;
    896 		}
    897 	}
    898 
    899 	return(0);
    900 
    901   bad:
    902 	m_freem(m);
    903 	return(-1);
    904 }
    905 
    906 /*
    907  * Unknown option processing.
    908  * The third argument `off' is the offset from the IPv6 header to the option,
    909  * which is necessary if the IPv6 header the and option header and IPv6 header
    910  * is not continuous in order to return an ICMPv6 error.
    911  */
    912 int
    913 ip6_unknown_opt(optp, m, off)
    914 	u_int8_t *optp;
    915 	struct mbuf *m;
    916 	int off;
    917 {
    918 	struct ip6_hdr *ip6;
    919 
    920 	switch (IP6OPT_TYPE(*optp)) {
    921 	case IP6OPT_TYPE_SKIP: /* ignore the option */
    922 		return((int)*(optp + 1));
    923 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
    924 		m_freem(m);
    925 		return(-1);
    926 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
    927 		ip6stat.ip6s_badoptions++;
    928 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
    929 		return(-1);
    930 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
    931 		ip6stat.ip6s_badoptions++;
    932 		ip6 = mtod(m, struct ip6_hdr *);
    933 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
    934 		    (m->m_flags & (M_BCAST|M_MCAST)))
    935 			m_freem(m);
    936 		else
    937 			icmp6_error(m, ICMP6_PARAM_PROB,
    938 				    ICMP6_PARAMPROB_OPTION, off);
    939 		return(-1);
    940 	}
    941 
    942 	m_freem(m);		/* XXX: NOTREACHED */
    943 	return(-1);
    944 }
    945 
    946 /*
    947  * Create the "control" list for this pcb.
    948  *
    949  * The routine will be called from upper layer handlers like tcp6_input().
    950  * Thus the routine assumes that the caller (tcp6_input) have already
    951  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
    952  * very first mbuf on the mbuf chain.
    953  * We may want to add some infinite loop prevention or sanity checks for safety.
    954  * (This applies only when you are using KAME mbuf chain restriction, i.e.
    955  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
    956  */
    957 void
    958 ip6_savecontrol(in6p, mp, ip6, m)
    959 	struct in6pcb *in6p;
    960 	struct mbuf **mp;
    961 	struct ip6_hdr *ip6;
    962 	struct mbuf *m;
    963 {
    964 	struct proc *p = curproc;	/* XXX */
    965 	int privileged;
    966 
    967 	privileged = 0;
    968 	if (p && !suser(p->p_ucred, &p->p_acflag))
    969 		privileged++;
    970 
    971 #ifdef SO_TIMESTAMP
    972 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
    973 		struct timeval tv;
    974 
    975 		microtime(&tv);
    976 		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
    977 			SCM_TIMESTAMP, SOL_SOCKET);
    978 		if (*mp)
    979 			mp = &(*mp)->m_next;
    980 	}
    981 #endif
    982 	if (in6p->in6p_flags & IN6P_RECVDSTADDR) {
    983 		*mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst,
    984 			sizeof(struct in6_addr), IPV6_RECVDSTADDR,
    985 			IPPROTO_IPV6);
    986 		if (*mp)
    987 			mp = &(*mp)->m_next;
    988 	}
    989 
    990 #ifdef noyet
    991 	/* options were tossed above */
    992 	if (in6p->in6p_flags & IN6P_RECVOPTS)
    993 		/* broken */
    994 	/* ip6_srcroute doesn't do what we want here, need to fix */
    995 	if (in6p->in6p_flags & IPV6P_RECVRETOPTS)
    996 		/* broken */
    997 #endif
    998 
    999 	/* RFC 2292 sec. 5 */
   1000 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
   1001 		struct in6_pktinfo pi6;
   1002 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
   1003 		if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
   1004 			pi6.ipi6_addr.s6_addr16[1] = 0;
   1005 		pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
   1006 					? m->m_pkthdr.rcvif->if_index
   1007 					: 0;
   1008 		*mp = sbcreatecontrol((caddr_t) &pi6,
   1009 			sizeof(struct in6_pktinfo), IPV6_PKTINFO,
   1010 			IPPROTO_IPV6);
   1011 		if (*mp)
   1012 			mp = &(*mp)->m_next;
   1013 	}
   1014 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
   1015 		int hlim = ip6->ip6_hlim & 0xff;
   1016 		*mp = sbcreatecontrol((caddr_t) &hlim,
   1017 			sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
   1018 		if (*mp)
   1019 			mp = &(*mp)->m_next;
   1020 	}
   1021 	/* IN6P_NEXTHOP - for outgoing packet only */
   1022 
   1023 	/*
   1024 	 * IPV6_HOPOPTS socket option. We require super-user privilege
   1025 	 * for the option, but it might be too strict, since there might
   1026 	 * be some hop-by-hop options which can be returned to normal user.
   1027 	 * See RFC 2292 section 6.
   1028 	 */
   1029 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
   1030 		/*
   1031 		 * Check if a hop-by-hop options header is contatined in the
   1032 		 * received packet, and if so, store the options as ancillary
   1033 		 * data. Note that a hop-by-hop options header must be
   1034 		 * just after the IPv6 header, which fact is assured through
   1035 		 * the IPv6 input processing.
   1036 		 */
   1037 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1038 		if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
   1039 			struct ip6_hbh *hbh;
   1040 			int hbhlen;
   1041 
   1042 #ifndef PULLDOWN_TEST
   1043 			hbh = (struct ip6_hbh *)(ip6 + 1);
   1044 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1045 #else
   1046 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
   1047 				sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
   1048 			if (hbh == NULL) {
   1049 				ip6stat.ip6s_tooshort++;
   1050 				return;
   1051 			}
   1052 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1053 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
   1054 				sizeof(struct ip6_hdr), hbhlen);
   1055 			if (hbh == NULL) {
   1056 				ip6stat.ip6s_tooshort++;
   1057 				return;
   1058 			}
   1059 #endif
   1060 
   1061 			/*
   1062 			 * XXX: We copy whole the header even if a jumbo
   1063 			 * payload option is included, which option is to
   1064 			 * be removed before returning in the RFC 2292.
   1065 			 * But it's too painful operation...
   1066 			 */
   1067 			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
   1068 					      IPV6_HOPOPTS, IPPROTO_IPV6);
   1069 			if (*mp)
   1070 				mp = &(*mp)->m_next;
   1071 		}
   1072 	}
   1073 
   1074 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
   1075 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
   1076 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1077 		int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);;
   1078 
   1079 		/*
   1080 		 * Search for destination options headers or routing
   1081 		 * header(s) through the header chain, and stores each
   1082 		 * header as ancillary data.
   1083 		 * Note that the order of the headers remains in
   1084 		 * the chain of ancillary data.
   1085 		 */
   1086 		while (1) {	/* is explicit loop prevention necessary? */
   1087 			struct ip6_ext *ip6e;
   1088 			int elen;
   1089 
   1090 #ifndef PULLDOWN_TEST
   1091 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
   1092 			if (nxt == IPPROTO_AH)
   1093 				elen = (ip6e->ip6e_len + 2) << 2;
   1094 			else
   1095 				elen = (ip6e->ip6e_len + 1) << 3;
   1096 #else
   1097 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off,
   1098 				sizeof(struct ip6_ext));
   1099 			if (ip6e == NULL) {
   1100 				ip6stat.ip6s_tooshort++;
   1101 				return;
   1102 			}
   1103 			if (nxt == IPPROTO_AH)
   1104 				elen = (ip6e->ip6e_len + 2) << 2;
   1105 			else
   1106 				elen = (ip6e->ip6e_len + 1) << 3;
   1107 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off, elen);
   1108 			if (ip6e == NULL) {
   1109 				ip6stat.ip6s_tooshort++;
   1110 				return;
   1111 			}
   1112 #endif
   1113 
   1114 			switch (nxt) {
   1115 		        case IPPROTO_DSTOPTS:
   1116 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
   1117 					break;
   1118 
   1119 				/*
   1120 				 * We also require super-user privilege for
   1121 				 * the option.
   1122 				 * See the comments on IN6_HOPOPTS.
   1123 				 */
   1124 				if (!privileged)
   1125 					break;
   1126 
   1127 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
   1128 						      IPV6_DSTOPTS,
   1129 						      IPPROTO_IPV6);
   1130 				if (*mp)
   1131 					mp = &(*mp)->m_next;
   1132 				break;
   1133 
   1134 			case IPPROTO_ROUTING:
   1135 				if (!in6p->in6p_flags & IN6P_RTHDR)
   1136 					break;
   1137 
   1138 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
   1139 						      IPV6_RTHDR,
   1140 						      IPPROTO_IPV6);
   1141 				if (*mp)
   1142 					mp = &(*mp)->m_next;
   1143 				break;
   1144 
   1145 			case IPPROTO_UDP:
   1146 			case IPPROTO_TCP:
   1147 			case IPPROTO_ICMPV6:
   1148 			default:
   1149 				/*
   1150 				 * stop search if we encounter an upper
   1151 				 * layer protocol headers.
   1152 				 */
   1153 				goto loopend;
   1154 
   1155 			case IPPROTO_HOPOPTS:
   1156 			case IPPROTO_AH: /* is it possible? */
   1157 				break;
   1158 			}
   1159 
   1160 			/* proceed with the next header. */
   1161 			off += elen;
   1162 			nxt = ip6e->ip6e_nxt;
   1163 		}
   1164 	  loopend:
   1165 		;
   1166 	}
   1167 	if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) {
   1168 		/* to be done */
   1169 	}
   1170 	if ((in6p->in6p_flags & IN6P_DSTOPTS) && privileged) {
   1171 		/* to be done */
   1172 	}
   1173 	/* IN6P_RTHDR - to be done */
   1174 
   1175 }
   1176 
   1177 /*
   1178  * Get pointer to the previous header followed by the header
   1179  * currently processed.
   1180  * XXX: This function supposes that
   1181  *	M includes all headers,
   1182  *	the next header field and the header length field of each header
   1183  *	are valid, and
   1184  *	the sum of each header length equals to OFF.
   1185  * Because of these assumptions, this function must be called very
   1186  * carefully. Moreover, it will not be used in the near future when
   1187  * we develop `neater' mechanism to process extension headers.
   1188  */
   1189 char *
   1190 ip6_get_prevhdr(m, off)
   1191 	struct mbuf *m;
   1192 	int off;
   1193 {
   1194 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1195 
   1196 	if (off == sizeof(struct ip6_hdr))
   1197 		return(&ip6->ip6_nxt);
   1198 	else {
   1199 		int len, nxt;
   1200 		struct ip6_ext *ip6e = NULL;
   1201 
   1202 		nxt = ip6->ip6_nxt;
   1203 		len = sizeof(struct ip6_hdr);
   1204 		while (len < off) {
   1205 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
   1206 
   1207 			switch (nxt) {
   1208 			case IPPROTO_FRAGMENT:
   1209 				len += sizeof(struct ip6_frag);
   1210 				break;
   1211 			case IPPROTO_AH:
   1212 				len += (ip6e->ip6e_len + 2) << 2;
   1213 				break;
   1214 			default:
   1215 				len += (ip6e->ip6e_len + 1) << 3;
   1216 				break;
   1217 			}
   1218 			nxt = ip6e->ip6e_nxt;
   1219 		}
   1220 		if (ip6e)
   1221 			return(&ip6e->ip6e_nxt);
   1222 		else
   1223 			return NULL;
   1224 	}
   1225 }
   1226 
   1227 /*
   1228  * get next header offset.  m will be retained.
   1229  */
   1230 int
   1231 ip6_nexthdr(m, off, proto, nxtp)
   1232 	struct mbuf *m;
   1233 	int off;
   1234 	int proto;
   1235 	int *nxtp;
   1236 {
   1237 	struct ip6_hdr ip6;
   1238 	struct ip6_ext ip6e;
   1239 	struct ip6_frag fh;
   1240 
   1241 	/* just in case */
   1242 	if (m == NULL)
   1243 		panic("ip6_nexthdr: m == NULL");
   1244 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
   1245 		return -1;
   1246 
   1247 	switch (proto) {
   1248 	case IPPROTO_IPV6:
   1249 		if (m->m_pkthdr.len < off + sizeof(ip6))
   1250 			return -1;
   1251 		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
   1252 		if (nxtp)
   1253 			*nxtp = ip6.ip6_nxt;
   1254 		off += sizeof(ip6);
   1255 		return off;
   1256 
   1257 	case IPPROTO_FRAGMENT:
   1258 		/*
   1259 		 * terminate parsing if it is not the first fragment,
   1260 		 * it does not make sense to parse through it.
   1261 		 */
   1262 		if (m->m_pkthdr.len < off + sizeof(fh))
   1263 			return -1;
   1264 		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
   1265 		if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
   1266 			return -1;
   1267 		if (nxtp)
   1268 			*nxtp = fh.ip6f_nxt;
   1269 		off += sizeof(struct ip6_frag);
   1270 		return off;
   1271 
   1272 	case IPPROTO_AH:
   1273 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1274 			return -1;
   1275 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
   1276 		if (nxtp)
   1277 			*nxtp = ip6e.ip6e_nxt;
   1278 		off += (ip6e.ip6e_len + 2) << 2;
   1279 		return off;
   1280 
   1281 	case IPPROTO_HOPOPTS:
   1282 	case IPPROTO_ROUTING:
   1283 	case IPPROTO_DSTOPTS:
   1284 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1285 			return -1;
   1286 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
   1287 		if (nxtp)
   1288 			*nxtp = ip6e.ip6e_nxt;
   1289 		off += (ip6e.ip6e_len + 1) << 3;
   1290 		return off;
   1291 
   1292 	case IPPROTO_NONE:
   1293 	case IPPROTO_ESP:
   1294 	case IPPROTO_IPCOMP:
   1295 		/* give up */
   1296 		return -1;
   1297 
   1298 	default:
   1299 		return -1;
   1300 	}
   1301 
   1302 	return -1;
   1303 }
   1304 
   1305 /*
   1306  * get offset for the last header in the chain.  m will be kept untainted.
   1307  */
   1308 int
   1309 ip6_lasthdr(m, off, proto, nxtp)
   1310 	struct mbuf *m;
   1311 	int off;
   1312 	int proto;
   1313 	int *nxtp;
   1314 {
   1315 	int newoff;
   1316 	int nxt;
   1317 
   1318 	if (!nxtp) {
   1319 		nxt = -1;
   1320 		nxtp = &nxt;
   1321 	}
   1322 	while (1) {
   1323 		newoff = ip6_nexthdr(m, off, proto, nxtp);
   1324 		if (newoff < 0)
   1325 			return off;
   1326 		else if (newoff < off)
   1327 			return -1;	/* invalid */
   1328 		else if (newoff == off)
   1329 			return newoff;
   1330 
   1331 		off = newoff;
   1332 		proto = *nxtp;
   1333 	}
   1334 }
   1335 
   1336 /*
   1337  * System control for IP6
   1338  */
   1339 
   1340 u_char	inet6ctlerrmap[PRC_NCMDS] = {
   1341 	0,		0,		0,		0,
   1342 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
   1343 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
   1344 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
   1345 	0,		0,		0,		0,
   1346 	ENOPROTOOPT
   1347 };
   1348 
   1349 #include <uvm/uvm_extern.h>
   1350 #include <sys/sysctl.h>
   1351 
   1352 int
   1353 ip6_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
   1354 	int *name;
   1355 	u_int namelen;
   1356 	void *oldp;
   1357 	size_t *oldlenp;
   1358 	void *newp;
   1359 	size_t newlen;
   1360 {
   1361 	int old, error;
   1362 
   1363 	/* All sysctl names at this level are terminal. */
   1364 	if (namelen != 1)
   1365 		return ENOTDIR;
   1366 
   1367 	switch (name[0]) {
   1368 
   1369 	case IPV6CTL_FORWARDING:
   1370 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1371 				  &ip6_forwarding);
   1372 	case IPV6CTL_SENDREDIRECTS:
   1373 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1374 				&ip6_sendredirects);
   1375 	case IPV6CTL_DEFHLIM:
   1376 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_defhlim);
   1377 	case IPV6CTL_MAXFRAGPACKETS:
   1378 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1379 				&ip6_maxfragpackets);
   1380 	case IPV6CTL_ACCEPT_RTADV:
   1381 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1382 				&ip6_accept_rtadv);
   1383 	case IPV6CTL_KEEPFAITH:
   1384 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_keepfaith);
   1385 	case IPV6CTL_LOG_INTERVAL:
   1386 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1387 				&ip6_log_interval);
   1388 	case IPV6CTL_HDRNESTLIMIT:
   1389 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1390 				&ip6_hdrnestlimit);
   1391 	case IPV6CTL_DAD_COUNT:
   1392 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_dad_count);
   1393 	case IPV6CTL_AUTO_FLOWLABEL:
   1394 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1395 				&ip6_auto_flowlabel);
   1396 	case IPV6CTL_DEFMCASTHLIM:
   1397 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1398 				&ip6_defmcasthlim);
   1399 	case IPV6CTL_GIF_HLIM:
   1400 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1401 				&ip6_gif_hlim);
   1402 	case IPV6CTL_KAME_VERSION:
   1403 		return sysctl_rdstring(oldp, oldlenp, newp, __KAME_VERSION);
   1404 	case IPV6CTL_USE_DEPRECATED:
   1405 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1406 				&ip6_use_deprecated);
   1407 	case IPV6CTL_RR_PRUNE:
   1408 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_rr_prune);
   1409 #ifndef INET6_BINDV6ONLY
   1410 	case IPV6CTL_BINDV6ONLY:
   1411 		return sysctl_int(oldp, oldlenp, newp, newlen,
   1412 				&ip6_bindv6only);
   1413 #endif
   1414 	case IPV6CTL_ANONPORTMIN:
   1415 		old = ip6_anonportmin;
   1416 		error = sysctl_int(oldp, oldlenp, newp, newlen,
   1417 		    &ip6_anonportmin);
   1418 		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmin < 0 ||
   1419 		    ip6_anonportmin > 65535
   1420 #ifndef IPNOPRIVPORTS
   1421 		    || ip6_anonportmin < IPV6PORT_RESERVED
   1422 #endif
   1423 		    ) {
   1424 			ip6_anonportmin = old;
   1425 			return (EINVAL);
   1426 		}
   1427 		return (error);
   1428 	case IPV6CTL_ANONPORTMAX:
   1429 		old = ip6_anonportmax;
   1430 		error = sysctl_int(oldp, oldlenp, newp, newlen,
   1431 		    &ip6_anonportmax);
   1432 		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmax < 0 ||
   1433 		    ip6_anonportmax > 65535
   1434 #ifndef IPNOPRIVPORTS
   1435 		    || ip6_anonportmax < IPV6PORT_RESERVED
   1436 #endif
   1437 		    ) {
   1438 			ip6_anonportmax = old;
   1439 			return (EINVAL);
   1440 		}
   1441 		return (error);
   1442 #ifndef IPNOPRIVPORTS
   1443 	case IPV6CTL_LOWPORTMIN:
   1444 		old = ip6_lowportmin;
   1445 		error = sysctl_int(oldp, oldlenp, newp, newlen,
   1446 		    &ip6_lowportmin);
   1447 		if (ip6_lowportmin >= ip6_lowportmax ||
   1448 		    ip6_lowportmin > IPV6PORT_RESERVEDMAX ||
   1449 		    ip6_lowportmin < IPV6PORT_RESERVEDMIN) {
   1450 			ip6_lowportmin = old;
   1451 			return (EINVAL);
   1452 		}
   1453 		return (error);
   1454 	case IPV6CTL_LOWPORTMAX:
   1455 		old = ip6_lowportmax;
   1456 		error = sysctl_int(oldp, oldlenp, newp, newlen,
   1457 		    &ip6_lowportmax);
   1458 		if (ip6_lowportmin >= ip6_lowportmax ||
   1459 		    ip6_lowportmax > IPV6PORT_RESERVEDMAX ||
   1460 		    ip6_lowportmax < IPV6PORT_RESERVEDMIN) {
   1461 			ip6_lowportmax = old;
   1462 			return (EINVAL);
   1463 		}
   1464 		return (error);
   1465 #endif
   1466 	default:
   1467 		return EOPNOTSUPP;
   1468 	}
   1469 	/* NOTREACHED */
   1470 }
   1471