Home | History | Annotate | Line # | Download | only in netinet6
raw_ip6.c revision 1.77
      1 /*	$NetBSD: raw_ip6.c,v 1.77 2006/05/14 21:19:34 elad 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 1.77 2006/05/14 21:19:34 elad 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 #include <machine/stdarg.h>
    102 
    103 #include "faith.h"
    104 #if defined(NFAITH) && 0 < NFAITH
    105 #include <net/if_faith.h>
    106 #endif
    107 
    108 extern struct inpcbtable rawcbtable;
    109 struct	inpcbtable raw6cbtable;
    110 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
    111 
    112 /*
    113  * Raw interface to IP6 protocol.
    114  */
    115 
    116 struct rip6stat rip6stat;
    117 
    118 /*
    119  * Initialize raw connection block queue.
    120  */
    121 void
    122 rip6_init()
    123 {
    124 
    125 	in6_pcbinit(&raw6cbtable, 1, 1);
    126 }
    127 
    128 /*
    129  * Setup generic address and protocol structures
    130  * for raw_input routine, then pass them along with
    131  * mbuf chain.
    132  */
    133 int
    134 rip6_input(mp, offp, proto)
    135 	struct	mbuf **mp;
    136 	int	*offp, proto;
    137 {
    138 	struct mbuf *m = *mp;
    139 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
    140 	struct inpcb_hdr *inph;
    141 	struct in6pcb *in6p;
    142 	struct in6pcb *last = NULL;
    143 	struct sockaddr_in6 rip6src;
    144 	struct mbuf *opts = NULL;
    145 
    146 	rip6stat.rip6s_ipackets++;
    147 
    148 #if defined(NFAITH) && 0 < NFAITH
    149 	if (faithprefix(&ip6->ip6_dst)) {
    150 		/* send icmp6 host unreach? */
    151 		m_freem(m);
    152 		return IPPROTO_DONE;
    153 	}
    154 #endif
    155 
    156 	/* Be proactive about malicious use of IPv4 mapped address */
    157 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    158 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    159 		/* XXX stat */
    160 		m_freem(m);
    161 		return IPPROTO_DONE;
    162 	}
    163 
    164 	bzero(&rip6src, sizeof(rip6src));
    165 	rip6src.sin6_len = sizeof(struct sockaddr_in6);
    166 	rip6src.sin6_family = AF_INET6;
    167 	rip6src.sin6_addr = ip6->ip6_src;
    168 	if (sa6_recoverscope(&rip6src) != 0) {
    169 		/* XXX: should be impossible. */
    170 		m_freem(m);
    171 		return IPPROTO_DONE;
    172 	}
    173 
    174 	CIRCLEQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
    175 		in6p = (struct in6pcb *)inph;
    176 		if (in6p->in6p_af != AF_INET6)
    177 			continue;
    178 		if (in6p->in6p_ip6.ip6_nxt &&
    179 		    in6p->in6p_ip6.ip6_nxt != proto)
    180 			continue;
    181 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
    182 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
    183 			continue;
    184 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
    185 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
    186 			continue;
    187 		if (in6p->in6p_cksum != -1) {
    188 			rip6stat.rip6s_isum++;
    189 			if (in6_cksum(m, proto, *offp,
    190 			    m->m_pkthdr.len - *offp)) {
    191 				rip6stat.rip6s_badsum++;
    192 				continue;
    193 			}
    194 		}
    195 		if (last) {
    196 			struct	mbuf *n;
    197 
    198 #ifdef IPSEC
    199 			/*
    200 			 * Check AH/ESP integrity.
    201 			 */
    202 			if (ipsec6_in_reject(m, last)) {
    203 				ipsec6stat.in_polvio++;
    204 				/* do not inject data into pcb */
    205 			} else
    206 #endif /* IPSEC */
    207 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
    208 				if (last->in6p_flags & IN6P_CONTROLOPTS)
    209 					ip6_savecontrol(last, &opts, ip6, n);
    210 				/* strip intermediate headers */
    211 				m_adj(n, *offp);
    212 				if (sbappendaddr(&last->in6p_socket->so_rcv,
    213 				    (struct sockaddr *)&rip6src, n, opts) == 0) {
    214 					/* should notify about lost packet */
    215 					m_freem(n);
    216 					if (opts)
    217 						m_freem(opts);
    218 					rip6stat.rip6s_fullsock++;
    219 				} else
    220 					sorwakeup(last->in6p_socket);
    221 				opts = NULL;
    222 			}
    223 		}
    224 		last = in6p;
    225 	}
    226 #ifdef IPSEC
    227 	/*
    228 	 * Check AH/ESP integrity.
    229 	 */
    230 	if (last && ipsec6_in_reject(m, last)) {
    231 		m_freem(m);
    232 		ipsec6stat.in_polvio++;
    233 		ip6stat.ip6s_delivered--;
    234 		/* do not inject data into pcb */
    235 	} else
    236 #endif /* IPSEC */
    237 	if (last) {
    238 		if (last->in6p_flags & IN6P_CONTROLOPTS)
    239 			ip6_savecontrol(last, &opts, ip6, m);
    240 		/* strip intermediate headers */
    241 		m_adj(m, *offp);
    242 		if (sbappendaddr(&last->in6p_socket->so_rcv,
    243 		    (struct sockaddr *)&rip6src, m, opts) == 0) {
    244 			m_freem(m);
    245 			if (opts)
    246 				m_freem(opts);
    247 			rip6stat.rip6s_fullsock++;
    248 		} else
    249 			sorwakeup(last->in6p_socket);
    250 	} else {
    251 		rip6stat.rip6s_nosock++;
    252 		if (m->m_flags & M_MCAST)
    253 			rip6stat.rip6s_nosockmcast++;
    254 		if (proto == IPPROTO_NONE)
    255 			m_freem(m);
    256 		else {
    257 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
    258 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_protounknown);
    259 			icmp6_error(m, ICMP6_PARAM_PROB,
    260 			    ICMP6_PARAMPROB_NEXTHEADER,
    261 			    prvnxtp - mtod(m, u_int8_t *));
    262 		}
    263 		ip6stat.ip6s_delivered--;
    264 	}
    265 	return IPPROTO_DONE;
    266 }
    267 
    268 void
    269 rip6_ctlinput(cmd, sa, d)
    270 	int cmd;
    271 	struct sockaddr *sa;
    272 	void *d;
    273 {
    274 	struct ip6_hdr *ip6;
    275 	struct ip6ctlparam *ip6cp = NULL;
    276 	const struct sockaddr_in6 *sa6_src = NULL;
    277 	void *cmdarg;
    278 	void (*notify) __P((struct in6pcb *, int)) = in6_rtchange;
    279 	int nxt;
    280 
    281 	if (sa->sa_family != AF_INET6 ||
    282 	    sa->sa_len != sizeof(struct sockaddr_in6))
    283 		return;
    284 
    285 	if ((unsigned)cmd >= PRC_NCMDS)
    286 		return;
    287 	if (PRC_IS_REDIRECT(cmd))
    288 		notify = in6_rtchange, d = NULL;
    289 	else if (cmd == PRC_HOSTDEAD)
    290 		d = NULL;
    291 	else if (cmd == PRC_MSGSIZE)
    292 		; /* special code is present, see below */
    293 	else if (inet6ctlerrmap[cmd] == 0)
    294 		return;
    295 
    296 	/* if the parameter is from icmp6, decode it. */
    297 	if (d != NULL) {
    298 		ip6cp = (struct ip6ctlparam *)d;
    299 		ip6 = ip6cp->ip6c_ip6;
    300 		cmdarg = ip6cp->ip6c_cmdarg;
    301 		sa6_src = ip6cp->ip6c_src;
    302 		nxt = ip6cp->ip6c_nxt;
    303 	} else {
    304 		ip6 = NULL;
    305 		cmdarg = NULL;
    306 		sa6_src = &sa6_any;
    307 		nxt = -1;
    308 	}
    309 
    310 	if (ip6 && cmd == PRC_MSGSIZE) {
    311 		struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)sa;
    312 		int valid = 0;
    313 		struct in6pcb *in6p;
    314 
    315 		/*
    316 		 * Check to see if we have a valid raw IPv6 socket
    317 		 * corresponding to the address in the ICMPv6 message
    318 		 * payload, and the protocol (ip6_nxt) meets the socket.
    319 		 * XXX chase extension headers, or pass final nxt value
    320 		 * from icmp6_notify_error()
    321 		 */
    322 		in6p = NULL;
    323 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
    324 		    (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0);
    325 #if 0
    326 		if (!in6p) {
    327 			/*
    328 			 * As the use of sendto(2) is fairly popular,
    329 			 * we may want to allow non-connected pcb too.
    330 			 * But it could be too weak against attacks...
    331 			 * We should at least check if the local
    332 			 * address (= s) is really ours.
    333 			 */
    334 			in6p = in6_pcblookup_bind(&raw6cbtable,
    335 			    &sa6->sin6_addr, 0, 0);
    336 		}
    337 #endif
    338 
    339 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
    340 		    in6p->in6p_ip6.ip6_nxt == nxt)
    341 			valid++;
    342 
    343 		/*
    344 		 * Depending on the value of "valid" and routing table
    345 		 * size (mtudisc_{hi,lo}wat), we will:
    346 		 * - recalculate the new MTU and create the
    347 		 *   corresponding routing entry, or
    348 		 * - ignore the MTU change notification.
    349 		 */
    350 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
    351 
    352 		/*
    353 		 * regardless of if we called icmp6_mtudisc_update(),
    354 		 * we need to call in6_pcbnotify(), to notify path MTU
    355 		 * change to the userland (RFC3542), because some
    356 		 * unconnected sockets may share the same destination
    357 		 * and want to know the path MTU.
    358 		 */
    359 	}
    360 
    361 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
    362 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
    363 }
    364 
    365 /*
    366  * Generate IPv6 header and pass packet to ip6_output.
    367  * Tack on options user may have setup with control call.
    368  */
    369 int
    370 #if __STDC__
    371 rip6_output(struct mbuf *m, ...)
    372 #else
    373 rip6_output(m, va_alist)
    374 	struct mbuf *m;
    375 	va_dcl
    376 #endif
    377 {
    378 	struct socket *so;
    379 	struct sockaddr_in6 *dstsock;
    380 	struct mbuf *control;
    381 	struct in6_addr *dst;
    382 	struct ip6_hdr *ip6;
    383 	struct in6pcb *in6p;
    384 	u_int	plen = m->m_pkthdr.len;
    385 	int error = 0;
    386 	struct ip6_pktopts opt, *optp = NULL;
    387 	struct ifnet *oifp = NULL;
    388 	int type, code;		/* for ICMPv6 output statistics only */
    389 	int priv = 0;
    390 	int scope_ambiguous = 0;
    391 	struct in6_addr *in6a;
    392 	va_list ap;
    393 
    394 	va_start(ap, m);
    395 	so = va_arg(ap, struct socket *);
    396 	dstsock = va_arg(ap, struct sockaddr_in6 *);
    397 	control = va_arg(ap, struct mbuf *);
    398 	va_end(ap);
    399 
    400 	in6p = sotoin6pcb(so);
    401 
    402 	priv = 0;
    403 	if (curproc && !kauth_authorize_generic(curproc->p_cred,
    404 					  KAUTH_GENERIC_ISSUSER,
    405 					  &curproc->p_acflag))
    406 		priv = 1;
    407 
    408 	dst = &dstsock->sin6_addr;
    409 	if (control) {
    410 		if ((error = ip6_setpktopts(control, &opt,
    411 		    in6p->in6p_outputopts,
    412 		    priv, so->so_proto->pr_protocol)) != 0) {
    413 			goto bad;
    414 		}
    415 		optp = &opt;
    416 	} else
    417 		optp = in6p->in6p_outputopts;
    418 
    419 	/*
    420 	 * Check and convert scope zone ID into internal form.
    421 	 * XXX: we may still need to determine the zone later.
    422 	 */
    423 	if (!(so->so_state & SS_ISCONNECTED)) {
    424 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
    425 			scope_ambiguous = 1;
    426 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
    427 			goto bad;
    428 	}
    429 
    430 	/*
    431 	 * For an ICMPv6 packet, we should know its type and code
    432 	 * to update statistics.
    433 	 */
    434 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    435 		struct icmp6_hdr *icmp6;
    436 		if (m->m_len < sizeof(struct icmp6_hdr) &&
    437 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
    438 			error = ENOBUFS;
    439 			goto bad;
    440 		}
    441 		icmp6 = mtod(m, struct icmp6_hdr *);
    442 		type = icmp6->icmp6_type;
    443 		code = icmp6->icmp6_code;
    444 	} else {
    445 		type = 0;
    446 		code = 0;
    447 	}
    448 
    449 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
    450 	if (!m) {
    451 		error = ENOBUFS;
    452 		goto bad;
    453 	}
    454 	ip6 = mtod(m, struct ip6_hdr *);
    455 
    456 	/*
    457 	 * Next header might not be ICMP6 but use its pseudo header anyway.
    458 	 */
    459 	ip6->ip6_dst = *dst;
    460 
    461 	/*
    462 	 * Source address selection.
    463 	 */
    464 	if ((in6a = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
    465 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp, &error)) == 0) {
    466 		if (error == 0)
    467 			error = EADDRNOTAVAIL;
    468 		goto bad;
    469 	}
    470 	ip6->ip6_src = *in6a;
    471 
    472 	if (oifp && scope_ambiguous) {
    473 		/*
    474 		 * Application should provide a proper zone ID or the use of
    475 		 * default zone IDs should be enabled.  Unfortunately, some
    476 		 * applications do not behave as it should, so we need a
    477 		 * workaround.  Even if an appropriate ID is not determined
    478 		 * (when it's required), if we can determine the outgoing
    479 		 * interface. determine the zone ID based on the interface.
    480 		 */
    481 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
    482 		if (error != 0)
    483 			goto bad;
    484 	}
    485 	ip6->ip6_dst = dstsock->sin6_addr;
    486 
    487 	/* fill in the rest of the IPv6 header fields */
    488 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
    489 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
    490 	ip6->ip6_vfc  |= IPV6_VERSION;
    491 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
    492 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
    493 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
    494 
    495 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
    496 	    in6p->in6p_cksum != -1) {
    497 		int off;
    498 		u_int16_t sum;
    499 
    500 		/* compute checksum */
    501 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
    502 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
    503 		else
    504 			off = in6p->in6p_cksum;
    505 		if (plen < off + 1) {
    506 			error = EINVAL;
    507 			goto bad;
    508 		}
    509 		off += sizeof(struct ip6_hdr);
    510 
    511 		sum = 0;
    512 		m = m_copyback_cow(m, off, sizeof(sum), (caddr_t)&sum,
    513 		    M_DONTWAIT);
    514 		if (m == NULL) {
    515 			error = ENOBUFS;
    516 			goto bad;
    517 		}
    518 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
    519 		m = m_copyback_cow(m, off, sizeof(sum), (caddr_t)&sum,
    520 		    M_DONTWAIT);
    521 		if (m == NULL) {
    522 			error = ENOBUFS;
    523 			goto bad;
    524 		}
    525 	}
    526 
    527 	error = ip6_output(m, optp, &in6p->in6p_route, 0,
    528 	    in6p->in6p_moptions, so, &oifp);
    529 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    530 		if (oifp)
    531 			icmp6_ifoutstat_inc(oifp, type, code);
    532 		icmp6stat.icp6s_outhist[type]++;
    533 	} else
    534 		rip6stat.rip6s_opackets++;
    535 
    536 	goto freectl;
    537 
    538  bad:
    539 	if (m)
    540 		m_freem(m);
    541 
    542  freectl:
    543 	if (control) {
    544 		ip6_clearpktopts(&opt, -1);
    545 		m_freem(control);
    546 	}
    547 	return (error);
    548 }
    549 
    550 /*
    551  * Raw IPv6 socket option processing.
    552  */
    553 int
    554 rip6_ctloutput(op, so, level, optname, mp)
    555 	int op;
    556 	struct socket *so;
    557 	int level, optname;
    558 	struct mbuf **mp;
    559 {
    560 	int error = 0;
    561 
    562 	switch (level) {
    563 	case IPPROTO_IPV6:
    564 		switch (optname) {
    565 		case MRT6_INIT:
    566 		case MRT6_DONE:
    567 		case MRT6_ADD_MIF:
    568 		case MRT6_DEL_MIF:
    569 		case MRT6_ADD_MFC:
    570 		case MRT6_DEL_MFC:
    571 		case MRT6_PIM:
    572 			if (op == PRCO_SETOPT) {
    573 				error = ip6_mrouter_set(optname, so, *mp);
    574 				if (*mp)
    575 					(void)m_free(*mp);
    576 			} else if (op == PRCO_GETOPT)
    577 				error = ip6_mrouter_get(optname, so, mp);
    578 			else
    579 				error = EINVAL;
    580 			return (error);
    581 		case IPV6_CHECKSUM:
    582 			return (ip6_raw_ctloutput(op, so, level, optname, mp));
    583 		default:
    584 			return (ip6_ctloutput(op, so, level, optname, mp));
    585 		}
    586 
    587 	case IPPROTO_ICMPV6:
    588 		/*
    589 		 * XXX: is it better to call icmp6_ctloutput() directly
    590 		 * from protosw?
    591 		 */
    592 		return (icmp6_ctloutput(op, so, level, optname, mp));
    593 
    594 	default:
    595 		if (op == PRCO_SETOPT && *mp)
    596 			m_free(*mp);
    597 		return EINVAL;
    598 	}
    599 }
    600 
    601 extern	u_long rip6_sendspace;
    602 extern	u_long rip6_recvspace;
    603 
    604 int
    605 rip6_usrreq(so, req, m, nam, control, l)
    606 	struct socket *so;
    607 	int req;
    608 	struct mbuf *m, *nam, *control;
    609 	struct lwp *l;
    610 {
    611 	struct in6pcb *in6p = sotoin6pcb(so);
    612 	struct proc *p;
    613 	int s;
    614 	int error = 0;
    615 	int priv;
    616 
    617 	priv = 0;
    618 	p = l ? l->l_proc : NULL;
    619 	if (p && !kauth_authorize_generic(p->p_cred, KAUTH_GENERIC_ISSUSER, &p->p_acflag))
    620 		priv++;
    621 
    622 	if (req == PRU_CONTROL)
    623 		return (in6_control(so, (u_long)m, (caddr_t)nam,
    624 		    (struct ifnet *)control, p));
    625 
    626 	if (req == PRU_PURGEIF) {
    627 		in6_pcbpurgeif0(&raw6cbtable, (struct ifnet *)control);
    628 		in6_purgeif((struct ifnet *)control);
    629 		in6_pcbpurgeif(&raw6cbtable, (struct ifnet *)control);
    630 		return (0);
    631 	}
    632 
    633 	switch (req) {
    634 	case PRU_ATTACH:
    635 		if (in6p)
    636 			panic("rip6_attach");
    637 		if (!priv) {
    638 			error = EACCES;
    639 			break;
    640 		}
    641 		s = splsoftnet();
    642 		if ((error = soreserve(so, rip6_sendspace, rip6_recvspace)) != 0) {
    643 			splx(s);
    644 			break;
    645 		}
    646 		if ((error = in6_pcballoc(so, &raw6cbtable)) != 0)
    647 		{
    648 			splx(s);
    649 			break;
    650 		}
    651 		splx(s);
    652 		in6p = sotoin6pcb(so);
    653 		in6p->in6p_ip6.ip6_nxt = (long)nam;
    654 		in6p->in6p_cksum = -1;
    655 
    656 		MALLOC(in6p->in6p_icmp6filt, struct icmp6_filter *,
    657 		    sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
    658 		if (in6p->in6p_icmp6filt == NULL) {
    659 			in6_pcbdetach(in6p);
    660 			error = ENOMEM;
    661 			break;
    662 		}
    663 		ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
    664 		break;
    665 
    666 	case PRU_DISCONNECT:
    667 		if ((so->so_state & SS_ISCONNECTED) == 0) {
    668 			error = ENOTCONN;
    669 			break;
    670 		}
    671 		in6p->in6p_faddr = in6addr_any;
    672 		so->so_state &= ~SS_ISCONNECTED;	/* XXX */
    673 		break;
    674 
    675 	case PRU_ABORT:
    676 		soisdisconnected(so);
    677 		/* Fallthrough */
    678 	case PRU_DETACH:
    679 		if (in6p == 0)
    680 			panic("rip6_detach");
    681 		if (so == ip6_mrouter)
    682 			ip6_mrouter_done();
    683 		/* xxx: RSVP */
    684 		if (in6p->in6p_icmp6filt) {
    685 			FREE(in6p->in6p_icmp6filt, M_PCB);
    686 			in6p->in6p_icmp6filt = NULL;
    687 		}
    688 		in6_pcbdetach(in6p);
    689 		break;
    690 
    691 	case PRU_BIND:
    692 	    {
    693 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
    694 		struct ifaddr *ia = NULL;
    695 
    696 		if (nam->m_len != sizeof(*addr)) {
    697 			error = EINVAL;
    698 			break;
    699 		}
    700 		if ((ifnet.tqh_first == 0) || (addr->sin6_family != AF_INET6)) {
    701 			error = EADDRNOTAVAIL;
    702 			break;
    703 		}
    704 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    705 			break;
    706 
    707 		/*
    708 		 * we don't support mapped address here, it would confuse
    709 		 * users so reject it
    710 		 */
    711 		if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr)) {
    712 			error = EADDRNOTAVAIL;
    713 			break;
    714 		}
    715 		if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
    716 		    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0) {
    717 			error = EADDRNOTAVAIL;
    718 			break;
    719 		}
    720 		if (ia && ((struct in6_ifaddr *)ia)->ia6_flags &
    721 		    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
    722 		     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
    723 			error = EADDRNOTAVAIL;
    724 			break;
    725 		}
    726 		in6p->in6p_laddr = addr->sin6_addr;
    727 		break;
    728 	    }
    729 
    730 	case PRU_CONNECT:
    731 	{
    732 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
    733 		struct in6_addr *in6a = NULL;
    734 		struct ifnet *ifp = NULL;
    735 		int scope_ambiguous = 0;
    736 
    737 		if (nam->m_len != sizeof(*addr)) {
    738 			error = EINVAL;
    739 			break;
    740 		}
    741 		if (ifnet.tqh_first == 0)
    742 		{
    743 			error = EADDRNOTAVAIL;
    744 			break;
    745 		}
    746 		if (addr->sin6_family != AF_INET6) {
    747 			error = EAFNOSUPPORT;
    748 			break;
    749 		}
    750 
    751 		/*
    752 		 * Application should provide a proper zone ID or the use of
    753 		 * default zone IDs should be enabled.  Unfortunately, some
    754 		 * applications do not behave as it should, so we need a
    755 		 * workaround.  Even if an appropriate ID is not determined,
    756 		 * we'll see if we can determine the outgoing interface.  If we
    757 		 * can, determine the zone ID based on the interface below.
    758 		 */
    759 		if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
    760 			scope_ambiguous = 1;
    761 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    762 			return(error);
    763 
    764 		/* Source address selection. XXX: need pcblookup? */
    765 		in6a = in6_selectsrc(addr, in6p->in6p_outputopts,
    766 		    in6p->in6p_moptions, &in6p->in6p_route,
    767 		    &in6p->in6p_laddr, &ifp, &error);
    768 		if (in6a == NULL) {
    769 			if (error == 0)
    770 				error = EADDRNOTAVAIL;
    771 			break;
    772 		}
    773 		/* XXX: see above */
    774 		if (ifp && scope_ambiguous &&
    775 		    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
    776 			break;
    777 		}
    778 		in6p->in6p_laddr = *in6a;
    779 		in6p->in6p_faddr = addr->sin6_addr;
    780 		soisconnected(so);
    781 		break;
    782 	}
    783 
    784 	case PRU_CONNECT2:
    785 		error = EOPNOTSUPP;
    786 		break;
    787 
    788 	/*
    789 	 * Mark the connection as being incapable of futther input.
    790 	 */
    791 	case PRU_SHUTDOWN:
    792 		socantsendmore(so);
    793 		break;
    794 	/*
    795 	 * Ship a packet out. The appropriate raw output
    796 	 * routine handles any messaging necessary.
    797 	 */
    798 	case PRU_SEND:
    799 	{
    800 		struct sockaddr_in6 tmp;
    801 		struct sockaddr_in6 *dst;
    802 
    803 		/* always copy sockaddr to avoid overwrites */
    804 		if (so->so_state & SS_ISCONNECTED) {
    805 			if (nam) {
    806 				error = EISCONN;
    807 				break;
    808 			}
    809 			/* XXX */
    810 			bzero(&tmp, sizeof(tmp));
    811 			tmp.sin6_family = AF_INET6;
    812 			tmp.sin6_len = sizeof(struct sockaddr_in6);
    813 			bcopy(&in6p->in6p_faddr, &tmp.sin6_addr,
    814 			    sizeof(struct in6_addr));
    815 			dst = &tmp;
    816 		} else {
    817 			if (nam == NULL) {
    818 				error = ENOTCONN;
    819 				break;
    820 			}
    821 			if (nam->m_len != sizeof(tmp)) {
    822 				error = EINVAL;
    823 				break;
    824 			}
    825 
    826 			tmp = *mtod(nam, struct sockaddr_in6 *);
    827 			dst = &tmp;
    828 
    829 			if (dst->sin6_family != AF_INET6) {
    830 				error = EAFNOSUPPORT;
    831 				break;
    832 			}
    833 		}
    834 		error = rip6_output(m, so, dst, control);
    835 		m = NULL;
    836 		break;
    837 	}
    838 
    839 	case PRU_SENSE:
    840 		/*
    841 		 * stat: don't bother with a blocksize
    842 		 */
    843 		return (0);
    844 	/*
    845 	 * Not supported.
    846 	 */
    847 	case PRU_RCVOOB:
    848 	case PRU_RCVD:
    849 	case PRU_LISTEN:
    850 	case PRU_ACCEPT:
    851 	case PRU_SENDOOB:
    852 		error = EOPNOTSUPP;
    853 		break;
    854 
    855 	case PRU_SOCKADDR:
    856 		in6_setsockaddr(in6p, nam);
    857 		break;
    858 
    859 	case PRU_PEERADDR:
    860 		in6_setpeeraddr(in6p, nam);
    861 		break;
    862 
    863 	default:
    864 		panic("rip6_usrreq");
    865 	}
    866 	if (m != NULL)
    867 		m_freem(m);
    868 	return (error);
    869 }
    870 
    871 SYSCTL_SETUP(sysctl_net_inet6_raw6_setup, "sysctl net.inet6.raw6 subtree setup")
    872 {
    873 
    874 	sysctl_createv(clog, 0, NULL, NULL,
    875 		       CTLFLAG_PERMANENT,
    876 		       CTLTYPE_NODE, "net", NULL,
    877 		       NULL, 0, NULL, 0,
    878 		       CTL_NET, CTL_EOL);
    879 	sysctl_createv(clog, 0, NULL, NULL,
    880 		       CTLFLAG_PERMANENT,
    881 		       CTLTYPE_NODE, "inet6", NULL,
    882 		       NULL, 0, NULL, 0,
    883 		       CTL_NET, PF_INET6, CTL_EOL);
    884 	sysctl_createv(clog, 0, NULL, NULL,
    885 		       CTLFLAG_PERMANENT,
    886 		       CTLTYPE_NODE, "raw6",
    887 		       SYSCTL_DESCR("Raw IPv6 settings"),
    888 		       NULL, 0, NULL, 0,
    889 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
    890 
    891 	sysctl_createv(clog, 0, NULL, NULL,
    892 		       CTLFLAG_PERMANENT,
    893 		       CTLTYPE_STRUCT, "pcblist",
    894 		       SYSCTL_DESCR("Raw IPv6 control block list"),
    895 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
    896 		       CTL_NET, PF_INET6, IPPROTO_RAW,
    897 		       CTL_CREATE, CTL_EOL);
    898 	sysctl_createv(clog, 0, NULL, NULL,
    899 		       CTLFLAG_PERMANENT,
    900 		       CTLTYPE_STRUCT, "stats",
    901 		       SYSCTL_DESCR("Raw IPv6 statistics"),
    902 		       NULL, 0, &rip6stat, sizeof(rip6stat),
    903 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
    904 		       CTL_EOL);
    905 }
    906