Home | History | Annotate | Line # | Download | only in netinet6
raw_ip6.c revision 1.83
      1 /*	$NetBSD: raw_ip6.c,v 1.83 2007/02/22 09:30:33 dyoung Exp $	*/
      2 /*	$KAME: raw_ip6.c,v 1.82 2001/07/23 18:57:56 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. 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  *	@(#)raw_ip.c	8.2 (Berkeley) 1/4/94
     62  */
     63 
     64 #include <sys/cdefs.h>
     65 __KERNEL_RCSID(0, "$NetBSD: raw_ip6.c,v 1.83 2007/02/22 09:30:33 dyoung Exp $");
     66 
     67 #include "opt_ipsec.h"
     68 
     69 #include <sys/param.h>
     70 #include <sys/sysctl.h>
     71 #include <sys/malloc.h>
     72 #include <sys/mbuf.h>
     73 #include <sys/socket.h>
     74 #include <sys/protosw.h>
     75 #include <sys/socketvar.h>
     76 #include <sys/errno.h>
     77 #include <sys/systm.h>
     78 #include <sys/proc.h>
     79 #include <sys/kauth.h>
     80 
     81 #include <net/if.h>
     82 #include <net/route.h>
     83 #include <net/if_types.h>
     84 
     85 #include <netinet/in.h>
     86 #include <netinet/in_var.h>
     87 #include <netinet/ip6.h>
     88 #include <netinet6/ip6_var.h>
     89 #include <netinet6/ip6_mroute.h>
     90 #include <netinet/icmp6.h>
     91 #include <netinet6/in6_pcb.h>
     92 #include <netinet6/nd6.h>
     93 #include <netinet6/ip6protosw.h>
     94 #include <netinet6/scope6_var.h>
     95 #include <netinet6/raw_ip6.h>
     96 
     97 #ifdef IPSEC
     98 #include <netinet6/ipsec.h>
     99 #endif /* IPSEC */
    100 
    101 #ifdef FAST_IPSEC
    102 #include <netipsec/ipsec.h>
    103 #include <netipsec/ipsec_var.h> /* XXX ipsecstat namespace */
    104 #include <netipsec/ipsec6.h>
    105 #endif
    106 
    107 #include <machine/stdarg.h>
    108 
    109 #include "faith.h"
    110 #if defined(NFAITH) && 0 < NFAITH
    111 #include <net/if_faith.h>
    112 #endif
    113 
    114 extern struct inpcbtable rawcbtable;
    115 struct	inpcbtable raw6cbtable;
    116 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
    117 
    118 /*
    119  * Raw interface to IP6 protocol.
    120  */
    121 
    122 struct rip6stat rip6stat;
    123 
    124 /*
    125  * Initialize raw connection block queue.
    126  */
    127 void
    128 rip6_init()
    129 {
    130 
    131 	in6_pcbinit(&raw6cbtable, 1, 1);
    132 }
    133 
    134 /*
    135  * Setup generic address and protocol structures
    136  * for raw_input routine, then pass them along with
    137  * mbuf chain.
    138  */
    139 int
    140 rip6_input(mp, offp, proto)
    141 	struct	mbuf **mp;
    142 	int	*offp, proto;
    143 {
    144 	struct mbuf *m = *mp;
    145 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
    146 	struct inpcb_hdr *inph;
    147 	struct in6pcb *in6p;
    148 	struct in6pcb *last = NULL;
    149 	struct sockaddr_in6 rip6src;
    150 	struct mbuf *opts = NULL;
    151 
    152 	rip6stat.rip6s_ipackets++;
    153 
    154 #if defined(NFAITH) && 0 < NFAITH
    155 	if (faithprefix(&ip6->ip6_dst)) {
    156 		/* send icmp6 host unreach? */
    157 		m_freem(m);
    158 		return IPPROTO_DONE;
    159 	}
    160 #endif
    161 
    162 	/* Be proactive about malicious use of IPv4 mapped address */
    163 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    164 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    165 		/* XXX stat */
    166 		m_freem(m);
    167 		return IPPROTO_DONE;
    168 	}
    169 
    170 	bzero(&rip6src, sizeof(rip6src));
    171 	rip6src.sin6_len = sizeof(struct sockaddr_in6);
    172 	rip6src.sin6_family = AF_INET6;
    173 	rip6src.sin6_addr = ip6->ip6_src;
    174 	if (sa6_recoverscope(&rip6src) != 0) {
    175 		/* XXX: should be impossible. */
    176 		m_freem(m);
    177 		return IPPROTO_DONE;
    178 	}
    179 
    180 	CIRCLEQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
    181 		in6p = (struct in6pcb *)inph;
    182 		if (in6p->in6p_af != AF_INET6)
    183 			continue;
    184 		if (in6p->in6p_ip6.ip6_nxt &&
    185 		    in6p->in6p_ip6.ip6_nxt != proto)
    186 			continue;
    187 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
    188 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
    189 			continue;
    190 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
    191 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
    192 			continue;
    193 		if (in6p->in6p_cksum != -1) {
    194 			rip6stat.rip6s_isum++;
    195 			if (in6_cksum(m, proto, *offp,
    196 			    m->m_pkthdr.len - *offp)) {
    197 				rip6stat.rip6s_badsum++;
    198 				continue;
    199 			}
    200 		}
    201 		if (last) {
    202 			struct	mbuf *n;
    203 
    204 #ifdef IPSEC
    205 			/*
    206 			 * Check AH/ESP integrity.
    207 			 */
    208 			if (ipsec6_in_reject(m, last)) {
    209 				ipsec6stat.in_polvio++;
    210 				/* do not inject data into pcb */
    211 			} else
    212 #endif /* IPSEC */
    213 #ifdef FAST_IPSEC
    214 			/*
    215 			 * Check AH/ESP integrity
    216 			 */
    217 			if (!ipsec6_in_reject(m,last))
    218 #endif /* FAST_IPSEC */
    219 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
    220 				if (last->in6p_flags & IN6P_CONTROLOPTS)
    221 					ip6_savecontrol(last, &opts, ip6, n);
    222 				/* strip intermediate headers */
    223 				m_adj(n, *offp);
    224 				if (sbappendaddr(&last->in6p_socket->so_rcv,
    225 				    (struct sockaddr *)&rip6src, n, opts) == 0) {
    226 					/* should notify about lost packet */
    227 					m_freem(n);
    228 					if (opts)
    229 						m_freem(opts);
    230 					rip6stat.rip6s_fullsock++;
    231 				} else
    232 					sorwakeup(last->in6p_socket);
    233 				opts = NULL;
    234 			}
    235 		}
    236 		last = in6p;
    237 	}
    238 #ifdef IPSEC
    239 	/*
    240 	 * Check AH/ESP integrity.
    241 	 */
    242 	if (last && ipsec6_in_reject(m, last)) {
    243 		m_freem(m);
    244 		ipsec6stat.in_polvio++;
    245 		ip6stat.ip6s_delivered--;
    246 		/* do not inject data into pcb */
    247 	} else
    248 #endif /* IPSEC */
    249 #ifdef FAST_IPSEC
    250 	if (last && ipsec6_in_reject(m, last)) {
    251 		m_freem(m);
    252 		/*
    253 		 * XXX ipsec6_in_reject update stat if there is an error
    254 		 * so we just need to update stats by hand in the case of last is
    255 		 * NULL
    256 		 */
    257 		if (!last)
    258 			ipsec6stat.in_polvio++;
    259 			ip6stat.ip6s_delivered--;
    260 			/* do not inject data into pcb */
    261 		} else
    262 #endif /* FAST_IPSEC */
    263 	if (last) {
    264 		if (last->in6p_flags & IN6P_CONTROLOPTS)
    265 			ip6_savecontrol(last, &opts, ip6, m);
    266 		/* strip intermediate headers */
    267 		m_adj(m, *offp);
    268 		if (sbappendaddr(&last->in6p_socket->so_rcv,
    269 		    (struct sockaddr *)&rip6src, m, opts) == 0) {
    270 			m_freem(m);
    271 			if (opts)
    272 				m_freem(opts);
    273 			rip6stat.rip6s_fullsock++;
    274 		} else
    275 			sorwakeup(last->in6p_socket);
    276 	} else {
    277 		rip6stat.rip6s_nosock++;
    278 		if (m->m_flags & M_MCAST)
    279 			rip6stat.rip6s_nosockmcast++;
    280 		if (proto == IPPROTO_NONE)
    281 			m_freem(m);
    282 		else {
    283 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
    284 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_protounknown);
    285 			icmp6_error(m, ICMP6_PARAM_PROB,
    286 			    ICMP6_PARAMPROB_NEXTHEADER,
    287 			    prvnxtp - mtod(m, u_int8_t *));
    288 		}
    289 		ip6stat.ip6s_delivered--;
    290 	}
    291 	return IPPROTO_DONE;
    292 }
    293 
    294 void
    295 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
    296 {
    297 	struct ip6_hdr *ip6;
    298 	struct ip6ctlparam *ip6cp = NULL;
    299 	const struct sockaddr_in6 *sa6_src = NULL;
    300 	void *cmdarg;
    301 	void (*notify) __P((struct in6pcb *, int)) = in6_rtchange;
    302 	int nxt;
    303 
    304 	if (sa->sa_family != AF_INET6 ||
    305 	    sa->sa_len != sizeof(struct sockaddr_in6))
    306 		return;
    307 
    308 	if ((unsigned)cmd >= PRC_NCMDS)
    309 		return;
    310 	if (PRC_IS_REDIRECT(cmd))
    311 		notify = in6_rtchange, d = NULL;
    312 	else if (cmd == PRC_HOSTDEAD)
    313 		d = NULL;
    314 	else if (cmd == PRC_MSGSIZE)
    315 		; /* special code is present, see below */
    316 	else if (inet6ctlerrmap[cmd] == 0)
    317 		return;
    318 
    319 	/* if the parameter is from icmp6, decode it. */
    320 	if (d != NULL) {
    321 		ip6cp = (struct ip6ctlparam *)d;
    322 		ip6 = ip6cp->ip6c_ip6;
    323 		cmdarg = ip6cp->ip6c_cmdarg;
    324 		sa6_src = ip6cp->ip6c_src;
    325 		nxt = ip6cp->ip6c_nxt;
    326 	} else {
    327 		ip6 = NULL;
    328 		cmdarg = NULL;
    329 		sa6_src = &sa6_any;
    330 		nxt = -1;
    331 	}
    332 
    333 	if (ip6 && cmd == PRC_MSGSIZE) {
    334 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
    335 		int valid = 0;
    336 		struct in6pcb *in6p;
    337 
    338 		/*
    339 		 * Check to see if we have a valid raw IPv6 socket
    340 		 * corresponding to the address in the ICMPv6 message
    341 		 * payload, and the protocol (ip6_nxt) meets the socket.
    342 		 * XXX chase extension headers, or pass final nxt value
    343 		 * from icmp6_notify_error()
    344 		 */
    345 		in6p = NULL;
    346 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
    347 		    (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0);
    348 #if 0
    349 		if (!in6p) {
    350 			/*
    351 			 * As the use of sendto(2) is fairly popular,
    352 			 * we may want to allow non-connected pcb too.
    353 			 * But it could be too weak against attacks...
    354 			 * We should at least check if the local
    355 			 * address (= s) is really ours.
    356 			 */
    357 			in6p = in6_pcblookup_bind(&raw6cbtable,
    358 			    &sa6->sin6_addr, 0, 0);
    359 		}
    360 #endif
    361 
    362 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
    363 		    in6p->in6p_ip6.ip6_nxt == nxt)
    364 			valid++;
    365 
    366 		/*
    367 		 * Depending on the value of "valid" and routing table
    368 		 * size (mtudisc_{hi,lo}wat), we will:
    369 		 * - recalculate the new MTU and create the
    370 		 *   corresponding routing entry, or
    371 		 * - ignore the MTU change notification.
    372 		 */
    373 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
    374 
    375 		/*
    376 		 * regardless of if we called icmp6_mtudisc_update(),
    377 		 * we need to call in6_pcbnotify(), to notify path MTU
    378 		 * change to the userland (RFC3542), because some
    379 		 * unconnected sockets may share the same destination
    380 		 * and want to know the path MTU.
    381 		 */
    382 	}
    383 
    384 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
    385 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
    386 }
    387 
    388 /*
    389  * Generate IPv6 header and pass packet to ip6_output.
    390  * Tack on options user may have setup with control call.
    391  */
    392 int
    393 #if __STDC__
    394 rip6_output(struct mbuf *m, ...)
    395 #else
    396 rip6_output(m, va_alist)
    397 	struct mbuf *m;
    398 	va_dcl
    399 #endif
    400 {
    401 	struct socket *so;
    402 	struct sockaddr_in6 *dstsock;
    403 	struct mbuf *control;
    404 	struct in6_addr *dst;
    405 	struct ip6_hdr *ip6;
    406 	struct in6pcb *in6p;
    407 	u_int	plen = m->m_pkthdr.len;
    408 	int error = 0;
    409 	struct ip6_pktopts opt, *optp = NULL;
    410 	struct ifnet *oifp = NULL;
    411 	int type, code;		/* for ICMPv6 output statistics only */
    412 	int priv = 0;
    413 	int scope_ambiguous = 0;
    414 	struct in6_addr *in6a;
    415 	va_list ap;
    416 
    417 	va_start(ap, m);
    418 	so = va_arg(ap, struct socket *);
    419 	dstsock = va_arg(ap, struct sockaddr_in6 *);
    420 	control = va_arg(ap, struct mbuf *);
    421 	va_end(ap);
    422 
    423 	in6p = sotoin6pcb(so);
    424 
    425 	priv = 0;
    426 	if (curlwp && !kauth_authorize_generic(curlwp->l_cred,
    427 	    KAUTH_GENERIC_ISSUSER, NULL))
    428 		priv = 1;
    429 
    430 	dst = &dstsock->sin6_addr;
    431 	if (control) {
    432 		if ((error = ip6_setpktopts(control, &opt,
    433 		    in6p->in6p_outputopts,
    434 		    priv, so->so_proto->pr_protocol)) != 0) {
    435 			goto bad;
    436 		}
    437 		optp = &opt;
    438 	} else
    439 		optp = in6p->in6p_outputopts;
    440 
    441 	/*
    442 	 * Check and convert scope zone ID into internal form.
    443 	 * XXX: we may still need to determine the zone later.
    444 	 */
    445 	if (!(so->so_state & SS_ISCONNECTED)) {
    446 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
    447 			scope_ambiguous = 1;
    448 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
    449 			goto bad;
    450 	}
    451 
    452 	/*
    453 	 * For an ICMPv6 packet, we should know its type and code
    454 	 * to update statistics.
    455 	 */
    456 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    457 		struct icmp6_hdr *icmp6;
    458 		if (m->m_len < sizeof(struct icmp6_hdr) &&
    459 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
    460 			error = ENOBUFS;
    461 			goto bad;
    462 		}
    463 		icmp6 = mtod(m, struct icmp6_hdr *);
    464 		type = icmp6->icmp6_type;
    465 		code = icmp6->icmp6_code;
    466 	} else {
    467 		type = 0;
    468 		code = 0;
    469 	}
    470 
    471 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
    472 	if (!m) {
    473 		error = ENOBUFS;
    474 		goto bad;
    475 	}
    476 	ip6 = mtod(m, struct ip6_hdr *);
    477 
    478 	/*
    479 	 * Next header might not be ICMP6 but use its pseudo header anyway.
    480 	 */
    481 	ip6->ip6_dst = *dst;
    482 
    483 	/*
    484 	 * Source address selection.
    485 	 */
    486 	if ((in6a = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
    487 	    (struct route *)&in6p->in6p_route, &in6p->in6p_laddr, &oifp,
    488 	    &error)) == 0) {
    489 		if (error == 0)
    490 			error = EADDRNOTAVAIL;
    491 		goto bad;
    492 	}
    493 	ip6->ip6_src = *in6a;
    494 
    495 	if (oifp && scope_ambiguous) {
    496 		/*
    497 		 * Application should provide a proper zone ID or the use of
    498 		 * default zone IDs should be enabled.  Unfortunately, some
    499 		 * applications do not behave as it should, so we need a
    500 		 * workaround.  Even if an appropriate ID is not determined
    501 		 * (when it's required), if we can determine the outgoing
    502 		 * interface. determine the zone ID based on the interface.
    503 		 */
    504 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
    505 		if (error != 0)
    506 			goto bad;
    507 	}
    508 	ip6->ip6_dst = dstsock->sin6_addr;
    509 
    510 	/* fill in the rest of the IPv6 header fields */
    511 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
    512 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
    513 	ip6->ip6_vfc  |= IPV6_VERSION;
    514 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
    515 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
    516 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
    517 
    518 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
    519 	    in6p->in6p_cksum != -1) {
    520 		int off;
    521 		u_int16_t sum;
    522 
    523 		/* compute checksum */
    524 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
    525 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
    526 		else
    527 			off = in6p->in6p_cksum;
    528 		if (plen < off + 1) {
    529 			error = EINVAL;
    530 			goto bad;
    531 		}
    532 		off += sizeof(struct ip6_hdr);
    533 
    534 		sum = 0;
    535 		m = m_copyback_cow(m, off, sizeof(sum), (caddr_t)&sum,
    536 		    M_DONTWAIT);
    537 		if (m == NULL) {
    538 			error = ENOBUFS;
    539 			goto bad;
    540 		}
    541 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
    542 		m = m_copyback_cow(m, off, sizeof(sum), (caddr_t)&sum,
    543 		    M_DONTWAIT);
    544 		if (m == NULL) {
    545 			error = ENOBUFS;
    546 			goto bad;
    547 		}
    548 	}
    549 
    550 	error = ip6_output(m, optp, &in6p->in6p_route, 0,
    551 	    in6p->in6p_moptions, so, &oifp);
    552 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    553 		if (oifp)
    554 			icmp6_ifoutstat_inc(oifp, type, code);
    555 		icmp6stat.icp6s_outhist[type]++;
    556 	} else
    557 		rip6stat.rip6s_opackets++;
    558 
    559 	goto freectl;
    560 
    561  bad:
    562 	if (m)
    563 		m_freem(m);
    564 
    565  freectl:
    566 	if (control) {
    567 		ip6_clearpktopts(&opt, -1);
    568 		m_freem(control);
    569 	}
    570 	return error;
    571 }
    572 
    573 /*
    574  * Raw IPv6 socket option processing.
    575  */
    576 int
    577 rip6_ctloutput(int op, struct socket *so, int level, int optname,
    578     struct mbuf **mp)
    579 {
    580 	int error = 0;
    581 
    582 	switch (level) {
    583 	case IPPROTO_IPV6:
    584 		switch (optname) {
    585 		case MRT6_INIT:
    586 		case MRT6_DONE:
    587 		case MRT6_ADD_MIF:
    588 		case MRT6_DEL_MIF:
    589 		case MRT6_ADD_MFC:
    590 		case MRT6_DEL_MFC:
    591 		case MRT6_PIM:
    592 			if (op == PRCO_SETOPT) {
    593 				error = ip6_mrouter_set(optname, so, *mp);
    594 				if (*mp)
    595 					(void)m_free(*mp);
    596 			} else if (op == PRCO_GETOPT)
    597 				error = ip6_mrouter_get(optname, so, mp);
    598 			else
    599 				error = EINVAL;
    600 			return error;
    601 		case IPV6_CHECKSUM:
    602 			return ip6_raw_ctloutput(op, so, level, optname, mp);
    603 		default:
    604 			return ip6_ctloutput(op, so, level, optname, mp);
    605 		}
    606 
    607 	case IPPROTO_ICMPV6:
    608 		/*
    609 		 * XXX: is it better to call icmp6_ctloutput() directly
    610 		 * from protosw?
    611 		 */
    612 		return icmp6_ctloutput(op, so, level, optname, mp);
    613 
    614 	default:
    615 		if (op == PRCO_SETOPT && *mp)
    616 			m_free(*mp);
    617 		return EINVAL;
    618 	}
    619 }
    620 
    621 extern	u_long rip6_sendspace;
    622 extern	u_long rip6_recvspace;
    623 
    624 int
    625 rip6_usrreq(so, req, m, nam, control, l)
    626 	struct socket *so;
    627 	int req;
    628 	struct mbuf *m, *nam, *control;
    629 	struct lwp *l;
    630 {
    631 	struct in6pcb *in6p = sotoin6pcb(so);
    632 	int s;
    633 	int error = 0;
    634 	int priv;
    635 
    636 	priv = 0;
    637 	if (l && !kauth_authorize_generic(l->l_cred,
    638 	    KAUTH_GENERIC_ISSUSER, NULL))
    639 		priv++;
    640 
    641 	if (req == PRU_CONTROL)
    642 		return in6_control(so, (u_long)m, (caddr_t)nam,
    643 		    (struct ifnet *)control, l);
    644 
    645 	if (req == PRU_PURGEIF) {
    646 		in6_pcbpurgeif0(&raw6cbtable, (struct ifnet *)control);
    647 		in6_purgeif((struct ifnet *)control);
    648 		in6_pcbpurgeif(&raw6cbtable, (struct ifnet *)control);
    649 		return 0;
    650 	}
    651 
    652 	switch (req) {
    653 	case PRU_ATTACH:
    654 		if (in6p != NULL)
    655 			panic("rip6_attach");
    656 		if (!priv) {
    657 			error = EACCES;
    658 			break;
    659 		}
    660 		s = splsoftnet();
    661 		error = soreserve(so, rip6_sendspace, rip6_recvspace);
    662 		if (error != 0) {
    663 			splx(s);
    664 			break;
    665 		}
    666 		if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
    667 			splx(s);
    668 			break;
    669 		}
    670 		splx(s);
    671 		in6p = sotoin6pcb(so);
    672 		in6p->in6p_ip6.ip6_nxt = (long)nam;
    673 		in6p->in6p_cksum = -1;
    674 
    675 		MALLOC(in6p->in6p_icmp6filt, struct icmp6_filter *,
    676 		    sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
    677 		if (in6p->in6p_icmp6filt == NULL) {
    678 			in6_pcbdetach(in6p);
    679 			error = ENOMEM;
    680 			break;
    681 		}
    682 		ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
    683 		break;
    684 
    685 	case PRU_DISCONNECT:
    686 		if ((so->so_state & SS_ISCONNECTED) == 0) {
    687 			error = ENOTCONN;
    688 			break;
    689 		}
    690 		in6p->in6p_faddr = in6addr_any;
    691 		so->so_state &= ~SS_ISCONNECTED;	/* XXX */
    692 		break;
    693 
    694 	case PRU_ABORT:
    695 		soisdisconnected(so);
    696 		/* Fallthrough */
    697 	case PRU_DETACH:
    698 		if (in6p == NULL)
    699 			panic("rip6_detach");
    700 		if (so == ip6_mrouter)
    701 			ip6_mrouter_done();
    702 		/* xxx: RSVP */
    703 		if (in6p->in6p_icmp6filt != NULL) {
    704 			FREE(in6p->in6p_icmp6filt, M_PCB);
    705 			in6p->in6p_icmp6filt = NULL;
    706 		}
    707 		in6_pcbdetach(in6p);
    708 		break;
    709 
    710 	case PRU_BIND:
    711 	    {
    712 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
    713 		struct ifaddr *ia = NULL;
    714 
    715 		if (nam->m_len != sizeof(*addr)) {
    716 			error = EINVAL;
    717 			break;
    718 		}
    719 		if (TAILQ_EMPTY(&ifnet) || addr->sin6_family != AF_INET6) {
    720 			error = EADDRNOTAVAIL;
    721 			break;
    722 		}
    723 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    724 			break;
    725 
    726 		/*
    727 		 * we don't support mapped address here, it would confuse
    728 		 * users so reject it
    729 		 */
    730 		if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr)) {
    731 			error = EADDRNOTAVAIL;
    732 			break;
    733 		}
    734 		if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
    735 		    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0) {
    736 			error = EADDRNOTAVAIL;
    737 			break;
    738 		}
    739 		if (ia && ((struct in6_ifaddr *)ia)->ia6_flags &
    740 		    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
    741 		     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
    742 			error = EADDRNOTAVAIL;
    743 			break;
    744 		}
    745 		in6p->in6p_laddr = addr->sin6_addr;
    746 		break;
    747 	    }
    748 
    749 	case PRU_CONNECT:
    750 	{
    751 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
    752 		struct in6_addr *in6a = NULL;
    753 		struct ifnet *ifp = NULL;
    754 		int scope_ambiguous = 0;
    755 
    756 		if (nam->m_len != sizeof(*addr)) {
    757 			error = EINVAL;
    758 			break;
    759 		}
    760 		if (TAILQ_EMPTY(&ifnet)) {
    761 			error = EADDRNOTAVAIL;
    762 			break;
    763 		}
    764 		if (addr->sin6_family != AF_INET6) {
    765 			error = EAFNOSUPPORT;
    766 			break;
    767 		}
    768 
    769 		/*
    770 		 * Application should provide a proper zone ID or the use of
    771 		 * default zone IDs should be enabled.  Unfortunately, some
    772 		 * applications do not behave as it should, so we need a
    773 		 * workaround.  Even if an appropriate ID is not determined,
    774 		 * we'll see if we can determine the outgoing interface.  If we
    775 		 * can, determine the zone ID based on the interface below.
    776 		 */
    777 		if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
    778 			scope_ambiguous = 1;
    779 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    780 			return error;
    781 
    782 		/* Source address selection. XXX: need pcblookup? */
    783 		in6a = in6_selectsrc(addr, in6p->in6p_outputopts,
    784 		    in6p->in6p_moptions, (struct route *)&in6p->in6p_route,
    785 		    &in6p->in6p_laddr, &ifp, &error);
    786 		if (in6a == NULL) {
    787 			if (error == 0)
    788 				error = EADDRNOTAVAIL;
    789 			break;
    790 		}
    791 		/* XXX: see above */
    792 		if (ifp && scope_ambiguous &&
    793 		    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
    794 			break;
    795 		}
    796 		in6p->in6p_laddr = *in6a;
    797 		in6p->in6p_faddr = addr->sin6_addr;
    798 		soisconnected(so);
    799 		break;
    800 	}
    801 
    802 	case PRU_CONNECT2:
    803 		error = EOPNOTSUPP;
    804 		break;
    805 
    806 	/*
    807 	 * Mark the connection as being incapable of futther input.
    808 	 */
    809 	case PRU_SHUTDOWN:
    810 		socantsendmore(so);
    811 		break;
    812 	/*
    813 	 * Ship a packet out. The appropriate raw output
    814 	 * routine handles any messaging necessary.
    815 	 */
    816 	case PRU_SEND:
    817 	{
    818 		struct sockaddr_in6 tmp;
    819 		struct sockaddr_in6 *dst;
    820 
    821 		/* always copy sockaddr to avoid overwrites */
    822 		if (so->so_state & SS_ISCONNECTED) {
    823 			if (nam) {
    824 				error = EISCONN;
    825 				break;
    826 			}
    827 			/* XXX */
    828 			bzero(&tmp, sizeof(tmp));
    829 			tmp.sin6_family = AF_INET6;
    830 			tmp.sin6_len = sizeof(struct sockaddr_in6);
    831 			bcopy(&in6p->in6p_faddr, &tmp.sin6_addr,
    832 			    sizeof(struct in6_addr));
    833 			dst = &tmp;
    834 		} else {
    835 			if (nam == NULL) {
    836 				error = ENOTCONN;
    837 				break;
    838 			}
    839 			if (nam->m_len != sizeof(tmp)) {
    840 				error = EINVAL;
    841 				break;
    842 			}
    843 
    844 			tmp = *mtod(nam, struct sockaddr_in6 *);
    845 			dst = &tmp;
    846 
    847 			if (dst->sin6_family != AF_INET6) {
    848 				error = EAFNOSUPPORT;
    849 				break;
    850 			}
    851 		}
    852 		error = rip6_output(m, so, dst, control);
    853 		m = NULL;
    854 		break;
    855 	}
    856 
    857 	case PRU_SENSE:
    858 		/*
    859 		 * stat: don't bother with a blocksize
    860 		 */
    861 		return 0;
    862 	/*
    863 	 * Not supported.
    864 	 */
    865 	case PRU_RCVOOB:
    866 	case PRU_RCVD:
    867 	case PRU_LISTEN:
    868 	case PRU_ACCEPT:
    869 	case PRU_SENDOOB:
    870 		error = EOPNOTSUPP;
    871 		break;
    872 
    873 	case PRU_SOCKADDR:
    874 		in6_setsockaddr(in6p, nam);
    875 		break;
    876 
    877 	case PRU_PEERADDR:
    878 		in6_setpeeraddr(in6p, nam);
    879 		break;
    880 
    881 	default:
    882 		panic("rip6_usrreq");
    883 	}
    884 	if (m != NULL)
    885 		m_freem(m);
    886 	return error;
    887 }
    888 
    889 SYSCTL_SETUP(sysctl_net_inet6_raw6_setup, "sysctl net.inet6.raw6 subtree setup")
    890 {
    891 
    892 	sysctl_createv(clog, 0, NULL, NULL,
    893 		       CTLFLAG_PERMANENT,
    894 		       CTLTYPE_NODE, "net", NULL,
    895 		       NULL, 0, NULL, 0,
    896 		       CTL_NET, CTL_EOL);
    897 	sysctl_createv(clog, 0, NULL, NULL,
    898 		       CTLFLAG_PERMANENT,
    899 		       CTLTYPE_NODE, "inet6", NULL,
    900 		       NULL, 0, NULL, 0,
    901 		       CTL_NET, PF_INET6, CTL_EOL);
    902 	sysctl_createv(clog, 0, NULL, NULL,
    903 		       CTLFLAG_PERMANENT,
    904 		       CTLTYPE_NODE, "raw6",
    905 		       SYSCTL_DESCR("Raw IPv6 settings"),
    906 		       NULL, 0, NULL, 0,
    907 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
    908 
    909 	sysctl_createv(clog, 0, NULL, NULL,
    910 		       CTLFLAG_PERMANENT,
    911 		       CTLTYPE_STRUCT, "pcblist",
    912 		       SYSCTL_DESCR("Raw IPv6 control block list"),
    913 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
    914 		       CTL_NET, PF_INET6, IPPROTO_RAW,
    915 		       CTL_CREATE, CTL_EOL);
    916 	sysctl_createv(clog, 0, NULL, NULL,
    917 		       CTLFLAG_PERMANENT,
    918 		       CTLTYPE_STRUCT, "stats",
    919 		       SYSCTL_DESCR("Raw IPv6 statistics"),
    920 		       NULL, 0, &rip6stat, sizeof(rip6stat),
    921 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
    922 		       CTL_EOL);
    923 }
    924