Home | History | Annotate | Line # | Download | only in netinet6
raw_ip6.c revision 1.76
      1 /*	$NetBSD: raw_ip6.c,v 1.76 2006/05/05 00:03:22 rpaulo 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 1.76 2006/05/05 00:03:22 rpaulo 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 
     80 #include <net/if.h>
     81 #include <net/route.h>
     82 #include <net/if_types.h>
     83 
     84 #include <netinet/in.h>
     85 #include <netinet/in_var.h>
     86 #include <netinet/ip6.h>
     87 #include <netinet6/ip6_var.h>
     88 #include <netinet6/ip6_mroute.h>
     89 #include <netinet/icmp6.h>
     90 #include <netinet6/in6_pcb.h>
     91 #include <netinet6/nd6.h>
     92 #include <netinet6/ip6protosw.h>
     93 #include <netinet6/scope6_var.h>
     94 #include <netinet6/raw_ip6.h>
     95 
     96 #ifdef IPSEC
     97 #include <netinet6/ipsec.h>
     98 #endif /* IPSEC */
     99 
    100 #include <machine/stdarg.h>
    101 
    102 #include "faith.h"
    103 #if defined(NFAITH) && 0 < NFAITH
    104 #include <net/if_faith.h>
    105 #endif
    106 
    107 extern struct inpcbtable rawcbtable;
    108 struct	inpcbtable raw6cbtable;
    109 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
    110 
    111 /*
    112  * Raw interface to IP6 protocol.
    113  */
    114 
    115 struct rip6stat rip6stat;
    116 
    117 /*
    118  * Initialize raw connection block queue.
    119  */
    120 void
    121 rip6_init()
    122 {
    123 
    124 	in6_pcbinit(&raw6cbtable, 1, 1);
    125 }
    126 
    127 /*
    128  * Setup generic address and protocol structures
    129  * for raw_input routine, then pass them along with
    130  * mbuf chain.
    131  */
    132 int
    133 rip6_input(mp, offp, proto)
    134 	struct	mbuf **mp;
    135 	int	*offp, proto;
    136 {
    137 	struct mbuf *m = *mp;
    138 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
    139 	struct inpcb_hdr *inph;
    140 	struct in6pcb *in6p;
    141 	struct in6pcb *last = NULL;
    142 	struct sockaddr_in6 rip6src;
    143 	struct mbuf *opts = NULL;
    144 
    145 	rip6stat.rip6s_ipackets++;
    146 
    147 #if defined(NFAITH) && 0 < NFAITH
    148 	if (faithprefix(&ip6->ip6_dst)) {
    149 		/* send icmp6 host unreach? */
    150 		m_freem(m);
    151 		return IPPROTO_DONE;
    152 	}
    153 #endif
    154 
    155 	/* Be proactive about malicious use of IPv4 mapped address */
    156 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    157 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    158 		/* XXX stat */
    159 		m_freem(m);
    160 		return IPPROTO_DONE;
    161 	}
    162 
    163 	bzero(&rip6src, sizeof(rip6src));
    164 	rip6src.sin6_len = sizeof(struct sockaddr_in6);
    165 	rip6src.sin6_family = AF_INET6;
    166 	rip6src.sin6_addr = ip6->ip6_src;
    167 	if (sa6_recoverscope(&rip6src) != 0) {
    168 		/* XXX: should be impossible. */
    169 		m_freem(m);
    170 		return IPPROTO_DONE;
    171 	}
    172 
    173 	CIRCLEQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
    174 		in6p = (struct in6pcb *)inph;
    175 		if (in6p->in6p_af != AF_INET6)
    176 			continue;
    177 		if (in6p->in6p_ip6.ip6_nxt &&
    178 		    in6p->in6p_ip6.ip6_nxt != proto)
    179 			continue;
    180 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
    181 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
    182 			continue;
    183 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
    184 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
    185 			continue;
    186 		if (in6p->in6p_cksum != -1) {
    187 			rip6stat.rip6s_isum++;
    188 			if (in6_cksum(m, proto, *offp,
    189 			    m->m_pkthdr.len - *offp)) {
    190 				rip6stat.rip6s_badsum++;
    191 				continue;
    192 			}
    193 		}
    194 		if (last) {
    195 			struct	mbuf *n;
    196 
    197 #ifdef IPSEC
    198 			/*
    199 			 * Check AH/ESP integrity.
    200 			 */
    201 			if (ipsec6_in_reject(m, last)) {
    202 				ipsec6stat.in_polvio++;
    203 				/* do not inject data into pcb */
    204 			} else
    205 #endif /* IPSEC */
    206 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
    207 				if (last->in6p_flags & IN6P_CONTROLOPTS)
    208 					ip6_savecontrol(last, &opts, ip6, n);
    209 				/* strip intermediate headers */
    210 				m_adj(n, *offp);
    211 				if (sbappendaddr(&last->in6p_socket->so_rcv,
    212 				    (struct sockaddr *)&rip6src, n, opts) == 0) {
    213 					/* should notify about lost packet */
    214 					m_freem(n);
    215 					if (opts)
    216 						m_freem(opts);
    217 					rip6stat.rip6s_fullsock++;
    218 				} else
    219 					sorwakeup(last->in6p_socket);
    220 				opts = NULL;
    221 			}
    222 		}
    223 		last = in6p;
    224 	}
    225 #ifdef IPSEC
    226 	/*
    227 	 * Check AH/ESP integrity.
    228 	 */
    229 	if (last && ipsec6_in_reject(m, last)) {
    230 		m_freem(m);
    231 		ipsec6stat.in_polvio++;
    232 		ip6stat.ip6s_delivered--;
    233 		/* do not inject data into pcb */
    234 	} else
    235 #endif /* IPSEC */
    236 	if (last) {
    237 		if (last->in6p_flags & IN6P_CONTROLOPTS)
    238 			ip6_savecontrol(last, &opts, ip6, m);
    239 		/* strip intermediate headers */
    240 		m_adj(m, *offp);
    241 		if (sbappendaddr(&last->in6p_socket->so_rcv,
    242 		    (struct sockaddr *)&rip6src, m, opts) == 0) {
    243 			m_freem(m);
    244 			if (opts)
    245 				m_freem(opts);
    246 			rip6stat.rip6s_fullsock++;
    247 		} else
    248 			sorwakeup(last->in6p_socket);
    249 	} else {
    250 		rip6stat.rip6s_nosock++;
    251 		if (m->m_flags & M_MCAST)
    252 			rip6stat.rip6s_nosockmcast++;
    253 		if (proto == IPPROTO_NONE)
    254 			m_freem(m);
    255 		else {
    256 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
    257 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_protounknown);
    258 			icmp6_error(m, ICMP6_PARAM_PROB,
    259 			    ICMP6_PARAMPROB_NEXTHEADER,
    260 			    prvnxtp - mtod(m, u_int8_t *));
    261 		}
    262 		ip6stat.ip6s_delivered--;
    263 	}
    264 	return IPPROTO_DONE;
    265 }
    266 
    267 void
    268 rip6_ctlinput(cmd, sa, d)
    269 	int cmd;
    270 	struct sockaddr *sa;
    271 	void *d;
    272 {
    273 	struct ip6_hdr *ip6;
    274 	struct ip6ctlparam *ip6cp = NULL;
    275 	const struct sockaddr_in6 *sa6_src = NULL;
    276 	void *cmdarg;
    277 	void (*notify) __P((struct in6pcb *, int)) = in6_rtchange;
    278 	int nxt;
    279 
    280 	if (sa->sa_family != AF_INET6 ||
    281 	    sa->sa_len != sizeof(struct sockaddr_in6))
    282 		return;
    283 
    284 	if ((unsigned)cmd >= PRC_NCMDS)
    285 		return;
    286 	if (PRC_IS_REDIRECT(cmd))
    287 		notify = in6_rtchange, d = NULL;
    288 	else if (cmd == PRC_HOSTDEAD)
    289 		d = NULL;
    290 	else if (cmd == PRC_MSGSIZE)
    291 		; /* special code is present, see below */
    292 	else if (inet6ctlerrmap[cmd] == 0)
    293 		return;
    294 
    295 	/* if the parameter is from icmp6, decode it. */
    296 	if (d != NULL) {
    297 		ip6cp = (struct ip6ctlparam *)d;
    298 		ip6 = ip6cp->ip6c_ip6;
    299 		cmdarg = ip6cp->ip6c_cmdarg;
    300 		sa6_src = ip6cp->ip6c_src;
    301 		nxt = ip6cp->ip6c_nxt;
    302 	} else {
    303 		ip6 = NULL;
    304 		cmdarg = NULL;
    305 		sa6_src = &sa6_any;
    306 		nxt = -1;
    307 	}
    308 
    309 	if (ip6 && cmd == PRC_MSGSIZE) {
    310 		struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)sa;
    311 		int valid = 0;
    312 		struct in6pcb *in6p;
    313 
    314 		/*
    315 		 * Check to see if we have a valid raw IPv6 socket
    316 		 * corresponding to the address in the ICMPv6 message
    317 		 * payload, and the protocol (ip6_nxt) meets the socket.
    318 		 * XXX chase extension headers, or pass final nxt value
    319 		 * from icmp6_notify_error()
    320 		 */
    321 		in6p = NULL;
    322 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
    323 		    (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0);
    324 #if 0
    325 		if (!in6p) {
    326 			/*
    327 			 * As the use of sendto(2) is fairly popular,
    328 			 * we may want to allow non-connected pcb too.
    329 			 * But it could be too weak against attacks...
    330 			 * We should at least check if the local
    331 			 * address (= s) is really ours.
    332 			 */
    333 			in6p = in6_pcblookup_bind(&raw6cbtable,
    334 			    &sa6->sin6_addr, 0, 0);
    335 		}
    336 #endif
    337 
    338 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
    339 		    in6p->in6p_ip6.ip6_nxt == nxt)
    340 			valid++;
    341 
    342 		/*
    343 		 * Depending on the value of "valid" and routing table
    344 		 * size (mtudisc_{hi,lo}wat), we will:
    345 		 * - recalculate the new MTU and create the
    346 		 *   corresponding routing entry, or
    347 		 * - ignore the MTU change notification.
    348 		 */
    349 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
    350 
    351 		/*
    352 		 * regardless of if we called icmp6_mtudisc_update(),
    353 		 * we need to call in6_pcbnotify(), to notify path MTU
    354 		 * change to the userland (RFC3542), because some
    355 		 * unconnected sockets may share the same destination
    356 		 * and want to know the path MTU.
    357 		 */
    358 	}
    359 
    360 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
    361 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
    362 }
    363 
    364 /*
    365  * Generate IPv6 header and pass packet to ip6_output.
    366  * Tack on options user may have setup with control call.
    367  */
    368 int
    369 #if __STDC__
    370 rip6_output(struct mbuf *m, ...)
    371 #else
    372 rip6_output(m, va_alist)
    373 	struct mbuf *m;
    374 	va_dcl
    375 #endif
    376 {
    377 	struct socket *so;
    378 	struct sockaddr_in6 *dstsock;
    379 	struct mbuf *control;
    380 	struct in6_addr *dst;
    381 	struct ip6_hdr *ip6;
    382 	struct in6pcb *in6p;
    383 	u_int	plen = m->m_pkthdr.len;
    384 	int error = 0;
    385 	struct ip6_pktopts opt, *optp = NULL;
    386 	struct ifnet *oifp = NULL;
    387 	int type, code;		/* for ICMPv6 output statistics only */
    388 	int priv = 0;
    389 	int scope_ambiguous = 0;
    390 	struct in6_addr *in6a;
    391 	va_list ap;
    392 
    393 	va_start(ap, m);
    394 	so = va_arg(ap, struct socket *);
    395 	dstsock = va_arg(ap, struct sockaddr_in6 *);
    396 	control = va_arg(ap, struct mbuf *);
    397 	va_end(ap);
    398 
    399 	in6p = sotoin6pcb(so);
    400 
    401 	priv = 0;
    402 	if (curproc && !suser(curproc->p_ucred, &curproc->p_acflag))
    403 		priv = 1;
    404 
    405 	dst = &dstsock->sin6_addr;
    406 	if (control) {
    407 		if ((error = ip6_setpktopts(control, &opt,
    408 		    in6p->in6p_outputopts,
    409 		    priv, so->so_proto->pr_protocol)) != 0) {
    410 			goto bad;
    411 		}
    412 		optp = &opt;
    413 	} else
    414 		optp = in6p->in6p_outputopts;
    415 
    416 	/*
    417 	 * Check and convert scope zone ID into internal form.
    418 	 * XXX: we may still need to determine the zone later.
    419 	 */
    420 	if (!(so->so_state & SS_ISCONNECTED)) {
    421 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
    422 			scope_ambiguous = 1;
    423 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
    424 			goto bad;
    425 	}
    426 
    427 	/*
    428 	 * For an ICMPv6 packet, we should know its type and code
    429 	 * to update statistics.
    430 	 */
    431 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    432 		struct icmp6_hdr *icmp6;
    433 		if (m->m_len < sizeof(struct icmp6_hdr) &&
    434 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
    435 			error = ENOBUFS;
    436 			goto bad;
    437 		}
    438 		icmp6 = mtod(m, struct icmp6_hdr *);
    439 		type = icmp6->icmp6_type;
    440 		code = icmp6->icmp6_code;
    441 	} else {
    442 		type = 0;
    443 		code = 0;
    444 	}
    445 
    446 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
    447 	if (!m) {
    448 		error = ENOBUFS;
    449 		goto bad;
    450 	}
    451 	ip6 = mtod(m, struct ip6_hdr *);
    452 
    453 	/*
    454 	 * Next header might not be ICMP6 but use its pseudo header anyway.
    455 	 */
    456 	ip6->ip6_dst = *dst;
    457 
    458 	/*
    459 	 * Source address selection.
    460 	 */
    461 	if ((in6a = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
    462 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp, &error)) == 0) {
    463 		if (error == 0)
    464 			error = EADDRNOTAVAIL;
    465 		goto bad;
    466 	}
    467 	ip6->ip6_src = *in6a;
    468 
    469 	if (oifp && scope_ambiguous) {
    470 		/*
    471 		 * Application should provide a proper zone ID or the use of
    472 		 * default zone IDs should be enabled.  Unfortunately, some
    473 		 * applications do not behave as it should, so we need a
    474 		 * workaround.  Even if an appropriate ID is not determined
    475 		 * (when it's required), if we can determine the outgoing
    476 		 * interface. determine the zone ID based on the interface.
    477 		 */
    478 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
    479 		if (error != 0)
    480 			goto bad;
    481 	}
    482 	ip6->ip6_dst = dstsock->sin6_addr;
    483 
    484 	/* fill in the rest of the IPv6 header fields */
    485 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
    486 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
    487 	ip6->ip6_vfc  |= IPV6_VERSION;
    488 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
    489 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
    490 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
    491 
    492 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
    493 	    in6p->in6p_cksum != -1) {
    494 		int off;
    495 		u_int16_t sum;
    496 
    497 		/* compute checksum */
    498 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
    499 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
    500 		else
    501 			off = in6p->in6p_cksum;
    502 		if (plen < off + 1) {
    503 			error = EINVAL;
    504 			goto bad;
    505 		}
    506 		off += sizeof(struct ip6_hdr);
    507 
    508 		sum = 0;
    509 		m = m_copyback_cow(m, off, sizeof(sum), (caddr_t)&sum,
    510 		    M_DONTWAIT);
    511 		if (m == NULL) {
    512 			error = ENOBUFS;
    513 			goto bad;
    514 		}
    515 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
    516 		m = m_copyback_cow(m, off, sizeof(sum), (caddr_t)&sum,
    517 		    M_DONTWAIT);
    518 		if (m == NULL) {
    519 			error = ENOBUFS;
    520 			goto bad;
    521 		}
    522 	}
    523 
    524 	error = ip6_output(m, optp, &in6p->in6p_route, 0,
    525 	    in6p->in6p_moptions, so, &oifp);
    526 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    527 		if (oifp)
    528 			icmp6_ifoutstat_inc(oifp, type, code);
    529 		icmp6stat.icp6s_outhist[type]++;
    530 	} else
    531 		rip6stat.rip6s_opackets++;
    532 
    533 	goto freectl;
    534 
    535  bad:
    536 	if (m)
    537 		m_freem(m);
    538 
    539  freectl:
    540 	if (control) {
    541 		ip6_clearpktopts(&opt, -1);
    542 		m_freem(control);
    543 	}
    544 	return (error);
    545 }
    546 
    547 /*
    548  * Raw IPv6 socket option processing.
    549  */
    550 int
    551 rip6_ctloutput(op, so, level, optname, mp)
    552 	int op;
    553 	struct socket *so;
    554 	int level, optname;
    555 	struct mbuf **mp;
    556 {
    557 	int error = 0;
    558 
    559 	switch (level) {
    560 	case IPPROTO_IPV6:
    561 		switch (optname) {
    562 		case MRT6_INIT:
    563 		case MRT6_DONE:
    564 		case MRT6_ADD_MIF:
    565 		case MRT6_DEL_MIF:
    566 		case MRT6_ADD_MFC:
    567 		case MRT6_DEL_MFC:
    568 		case MRT6_PIM:
    569 			if (op == PRCO_SETOPT) {
    570 				error = ip6_mrouter_set(optname, so, *mp);
    571 				if (*mp)
    572 					(void)m_free(*mp);
    573 			} else if (op == PRCO_GETOPT)
    574 				error = ip6_mrouter_get(optname, so, mp);
    575 			else
    576 				error = EINVAL;
    577 			return (error);
    578 		case IPV6_CHECKSUM:
    579 			return (ip6_raw_ctloutput(op, so, level, optname, mp));
    580 		default:
    581 			return (ip6_ctloutput(op, so, level, optname, mp));
    582 		}
    583 
    584 	case IPPROTO_ICMPV6:
    585 		/*
    586 		 * XXX: is it better to call icmp6_ctloutput() directly
    587 		 * from protosw?
    588 		 */
    589 		return (icmp6_ctloutput(op, so, level, optname, mp));
    590 
    591 	default:
    592 		if (op == PRCO_SETOPT && *mp)
    593 			m_free(*mp);
    594 		return EINVAL;
    595 	}
    596 }
    597 
    598 extern	u_long rip6_sendspace;
    599 extern	u_long rip6_recvspace;
    600 
    601 int
    602 rip6_usrreq(so, req, m, nam, control, l)
    603 	struct socket *so;
    604 	int req;
    605 	struct mbuf *m, *nam, *control;
    606 	struct lwp *l;
    607 {
    608 	struct in6pcb *in6p = sotoin6pcb(so);
    609 	struct proc *p;
    610 	int s;
    611 	int error = 0;
    612 	int priv;
    613 
    614 	priv = 0;
    615 	p = l ? l->l_proc : NULL;
    616 	if (p && !suser(p->p_ucred, &p->p_acflag))
    617 		priv++;
    618 
    619 	if (req == PRU_CONTROL)
    620 		return (in6_control(so, (u_long)m, (caddr_t)nam,
    621 		    (struct ifnet *)control, p));
    622 
    623 	if (req == PRU_PURGEIF) {
    624 		in6_pcbpurgeif0(&raw6cbtable, (struct ifnet *)control);
    625 		in6_purgeif((struct ifnet *)control);
    626 		in6_pcbpurgeif(&raw6cbtable, (struct ifnet *)control);
    627 		return (0);
    628 	}
    629 
    630 	switch (req) {
    631 	case PRU_ATTACH:
    632 		if (in6p)
    633 			panic("rip6_attach");
    634 		if (!priv) {
    635 			error = EACCES;
    636 			break;
    637 		}
    638 		s = splsoftnet();
    639 		if ((error = soreserve(so, rip6_sendspace, rip6_recvspace)) != 0) {
    640 			splx(s);
    641 			break;
    642 		}
    643 		if ((error = in6_pcballoc(so, &raw6cbtable)) != 0)
    644 		{
    645 			splx(s);
    646 			break;
    647 		}
    648 		splx(s);
    649 		in6p = sotoin6pcb(so);
    650 		in6p->in6p_ip6.ip6_nxt = (long)nam;
    651 		in6p->in6p_cksum = -1;
    652 
    653 		MALLOC(in6p->in6p_icmp6filt, struct icmp6_filter *,
    654 		    sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
    655 		if (in6p->in6p_icmp6filt == NULL) {
    656 			in6_pcbdetach(in6p);
    657 			error = ENOMEM;
    658 			break;
    659 		}
    660 		ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
    661 		break;
    662 
    663 	case PRU_DISCONNECT:
    664 		if ((so->so_state & SS_ISCONNECTED) == 0) {
    665 			error = ENOTCONN;
    666 			break;
    667 		}
    668 		in6p->in6p_faddr = in6addr_any;
    669 		so->so_state &= ~SS_ISCONNECTED;	/* XXX */
    670 		break;
    671 
    672 	case PRU_ABORT:
    673 		soisdisconnected(so);
    674 		/* Fallthrough */
    675 	case PRU_DETACH:
    676 		if (in6p == 0)
    677 			panic("rip6_detach");
    678 		if (so == ip6_mrouter)
    679 			ip6_mrouter_done();
    680 		/* xxx: RSVP */
    681 		if (in6p->in6p_icmp6filt) {
    682 			FREE(in6p->in6p_icmp6filt, M_PCB);
    683 			in6p->in6p_icmp6filt = NULL;
    684 		}
    685 		in6_pcbdetach(in6p);
    686 		break;
    687 
    688 	case PRU_BIND:
    689 	    {
    690 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
    691 		struct ifaddr *ia = NULL;
    692 
    693 		if (nam->m_len != sizeof(*addr)) {
    694 			error = EINVAL;
    695 			break;
    696 		}
    697 		if ((ifnet.tqh_first == 0) || (addr->sin6_family != AF_INET6)) {
    698 			error = EADDRNOTAVAIL;
    699 			break;
    700 		}
    701 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    702 			break;
    703 
    704 		/*
    705 		 * we don't support mapped address here, it would confuse
    706 		 * users so reject it
    707 		 */
    708 		if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr)) {
    709 			error = EADDRNOTAVAIL;
    710 			break;
    711 		}
    712 		if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
    713 		    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0) {
    714 			error = EADDRNOTAVAIL;
    715 			break;
    716 		}
    717 		if (ia && ((struct in6_ifaddr *)ia)->ia6_flags &
    718 		    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
    719 		     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
    720 			error = EADDRNOTAVAIL;
    721 			break;
    722 		}
    723 		in6p->in6p_laddr = addr->sin6_addr;
    724 		break;
    725 	    }
    726 
    727 	case PRU_CONNECT:
    728 	{
    729 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
    730 		struct in6_addr *in6a = NULL;
    731 		struct ifnet *ifp = NULL;
    732 		int scope_ambiguous = 0;
    733 
    734 		if (nam->m_len != sizeof(*addr)) {
    735 			error = EINVAL;
    736 			break;
    737 		}
    738 		if (ifnet.tqh_first == 0)
    739 		{
    740 			error = EADDRNOTAVAIL;
    741 			break;
    742 		}
    743 		if (addr->sin6_family != AF_INET6) {
    744 			error = EAFNOSUPPORT;
    745 			break;
    746 		}
    747 
    748 		/*
    749 		 * Application should provide a proper zone ID or the use of
    750 		 * default zone IDs should be enabled.  Unfortunately, some
    751 		 * applications do not behave as it should, so we need a
    752 		 * workaround.  Even if an appropriate ID is not determined,
    753 		 * we'll see if we can determine the outgoing interface.  If we
    754 		 * can, determine the zone ID based on the interface below.
    755 		 */
    756 		if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
    757 			scope_ambiguous = 1;
    758 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    759 			return(error);
    760 
    761 		/* Source address selection. XXX: need pcblookup? */
    762 		in6a = in6_selectsrc(addr, in6p->in6p_outputopts,
    763 		    in6p->in6p_moptions, &in6p->in6p_route,
    764 		    &in6p->in6p_laddr, &ifp, &error);
    765 		if (in6a == NULL) {
    766 			if (error == 0)
    767 				error = EADDRNOTAVAIL;
    768 			break;
    769 		}
    770 		/* XXX: see above */
    771 		if (ifp && scope_ambiguous &&
    772 		    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
    773 			break;
    774 		}
    775 		in6p->in6p_laddr = *in6a;
    776 		in6p->in6p_faddr = addr->sin6_addr;
    777 		soisconnected(so);
    778 		break;
    779 	}
    780 
    781 	case PRU_CONNECT2:
    782 		error = EOPNOTSUPP;
    783 		break;
    784 
    785 	/*
    786 	 * Mark the connection as being incapable of futther input.
    787 	 */
    788 	case PRU_SHUTDOWN:
    789 		socantsendmore(so);
    790 		break;
    791 	/*
    792 	 * Ship a packet out. The appropriate raw output
    793 	 * routine handles any messaging necessary.
    794 	 */
    795 	case PRU_SEND:
    796 	{
    797 		struct sockaddr_in6 tmp;
    798 		struct sockaddr_in6 *dst;
    799 
    800 		/* always copy sockaddr to avoid overwrites */
    801 		if (so->so_state & SS_ISCONNECTED) {
    802 			if (nam) {
    803 				error = EISCONN;
    804 				break;
    805 			}
    806 			/* XXX */
    807 			bzero(&tmp, sizeof(tmp));
    808 			tmp.sin6_family = AF_INET6;
    809 			tmp.sin6_len = sizeof(struct sockaddr_in6);
    810 			bcopy(&in6p->in6p_faddr, &tmp.sin6_addr,
    811 			    sizeof(struct in6_addr));
    812 			dst = &tmp;
    813 		} else {
    814 			if (nam == NULL) {
    815 				error = ENOTCONN;
    816 				break;
    817 			}
    818 			if (nam->m_len != sizeof(tmp)) {
    819 				error = EINVAL;
    820 				break;
    821 			}
    822 
    823 			tmp = *mtod(nam, struct sockaddr_in6 *);
    824 			dst = &tmp;
    825 
    826 			if (dst->sin6_family != AF_INET6) {
    827 				error = EAFNOSUPPORT;
    828 				break;
    829 			}
    830 		}
    831 		error = rip6_output(m, so, dst, control);
    832 		m = NULL;
    833 		break;
    834 	}
    835 
    836 	case PRU_SENSE:
    837 		/*
    838 		 * stat: don't bother with a blocksize
    839 		 */
    840 		return (0);
    841 	/*
    842 	 * Not supported.
    843 	 */
    844 	case PRU_RCVOOB:
    845 	case PRU_RCVD:
    846 	case PRU_LISTEN:
    847 	case PRU_ACCEPT:
    848 	case PRU_SENDOOB:
    849 		error = EOPNOTSUPP;
    850 		break;
    851 
    852 	case PRU_SOCKADDR:
    853 		in6_setsockaddr(in6p, nam);
    854 		break;
    855 
    856 	case PRU_PEERADDR:
    857 		in6_setpeeraddr(in6p, nam);
    858 		break;
    859 
    860 	default:
    861 		panic("rip6_usrreq");
    862 	}
    863 	if (m != NULL)
    864 		m_freem(m);
    865 	return (error);
    866 }
    867 
    868 SYSCTL_SETUP(sysctl_net_inet6_raw6_setup, "sysctl net.inet6.raw6 subtree setup")
    869 {
    870 
    871 	sysctl_createv(clog, 0, NULL, NULL,
    872 		       CTLFLAG_PERMANENT,
    873 		       CTLTYPE_NODE, "net", NULL,
    874 		       NULL, 0, NULL, 0,
    875 		       CTL_NET, CTL_EOL);
    876 	sysctl_createv(clog, 0, NULL, NULL,
    877 		       CTLFLAG_PERMANENT,
    878 		       CTLTYPE_NODE, "inet6", NULL,
    879 		       NULL, 0, NULL, 0,
    880 		       CTL_NET, PF_INET6, CTL_EOL);
    881 	sysctl_createv(clog, 0, NULL, NULL,
    882 		       CTLFLAG_PERMANENT,
    883 		       CTLTYPE_NODE, "raw6",
    884 		       SYSCTL_DESCR("Raw IPv6 settings"),
    885 		       NULL, 0, NULL, 0,
    886 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
    887 
    888 	sysctl_createv(clog, 0, NULL, NULL,
    889 		       CTLFLAG_PERMANENT,
    890 		       CTLTYPE_STRUCT, "pcblist",
    891 		       SYSCTL_DESCR("Raw IPv6 control block list"),
    892 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
    893 		       CTL_NET, PF_INET6, IPPROTO_RAW,
    894 		       CTL_CREATE, CTL_EOL);
    895 	sysctl_createv(clog, 0, NULL, NULL,
    896 		       CTLFLAG_PERMANENT,
    897 		       CTLTYPE_STRUCT, "stats",
    898 		       SYSCTL_DESCR("Raw IPv6 statistics"),
    899 		       NULL, 0, &rip6stat, sizeof(rip6stat),
    900 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
    901 		       CTL_EOL);
    902 }
    903