Home | History | Annotate | Line # | Download | only in netinet6
raw_ip6.c revision 1.141
      1 /*	$NetBSD: raw_ip6.c,v 1.141 2015/08/24 22:21:27 pooka 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 1.141 2015/08/24 22:21:27 pooka Exp $");
     66 
     67 #ifdef _KERNEL_OPT
     68 #include "opt_ipsec.h"
     69 #endif
     70 
     71 #include <sys/param.h>
     72 #include <sys/sysctl.h>
     73 #include <sys/mbuf.h>
     74 #include <sys/socket.h>
     75 #include <sys/protosw.h>
     76 #include <sys/socketvar.h>
     77 #include <sys/systm.h>
     78 #include <sys/proc.h>
     79 #include <sys/kauth.h>
     80 #include <sys/kmem.h>
     81 
     82 #include <net/if.h>
     83 #include <net/route.h>
     84 #include <net/if_types.h>
     85 #include <net/net_stats.h>
     86 
     87 #include <netinet/in.h>
     88 #include <netinet/in_var.h>
     89 #include <netinet/ip6.h>
     90 #include <netinet6/ip6_var.h>
     91 #include <netinet6/ip6_private.h>
     92 #include <netinet6/ip6_mroute.h>
     93 #include <netinet/icmp6.h>
     94 #include <netinet6/icmp6_private.h>
     95 #include <netinet6/in6_pcb.h>
     96 #include <netinet6/nd6.h>
     97 #include <netinet6/ip6protosw.h>
     98 #include <netinet6/scope6_var.h>
     99 #include <netinet6/raw_ip6.h>
    100 
    101 #ifdef IPSEC
    102 #include <netipsec/ipsec.h>
    103 #include <netipsec/ipsec_var.h>
    104 #include <netipsec/ipsec_private.h>
    105 #include <netipsec/ipsec6.h>
    106 #endif
    107 
    108 #include "faith.h"
    109 #if defined(NFAITH) && 0 < NFAITH
    110 #include <net/if_faith.h>
    111 #endif
    112 
    113 extern struct inpcbtable rawcbtable;
    114 struct	inpcbtable raw6cbtable;
    115 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
    116 
    117 /*
    118  * Raw interface to IP6 protocol.
    119  */
    120 
    121 static percpu_t *rip6stat_percpu;
    122 
    123 #define	RIP6_STATINC(x)		_NET_STATINC(rip6stat_percpu, x)
    124 
    125 static void sysctl_net_inet6_raw6_setup(struct sysctllog **);
    126 
    127 /*
    128  * Initialize raw connection block queue.
    129  */
    130 void
    131 rip6_init(void)
    132 {
    133 
    134 	sysctl_net_inet6_raw6_setup(NULL);
    135 	in6_pcbinit(&raw6cbtable, 1, 1);
    136 
    137 	rip6stat_percpu = percpu_alloc(sizeof(uint64_t) * RIP6_NSTATS);
    138 }
    139 
    140 /*
    141  * Setup generic address and protocol structures
    142  * for raw_input routine, then pass them along with
    143  * mbuf chain.
    144  */
    145 int
    146 rip6_input(struct mbuf **mp, int *offp, int proto)
    147 {
    148 	struct mbuf *m = *mp;
    149 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
    150 	struct inpcb_hdr *inph;
    151 	struct in6pcb *in6p;
    152 	struct in6pcb *last = NULL;
    153 	struct sockaddr_in6 rip6src;
    154 	struct mbuf *opts = NULL;
    155 
    156 	RIP6_STATINC(RIP6_STAT_IPACKETS);
    157 
    158 #if defined(NFAITH) && 0 < NFAITH
    159 	if (faithprefix(&ip6->ip6_dst)) {
    160 		/* send icmp6 host unreach? */
    161 		m_freem(m);
    162 		return IPPROTO_DONE;
    163 	}
    164 #endif
    165 
    166 	/* Be proactive about malicious use of IPv4 mapped address */
    167 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    168 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    169 		/* XXX stat */
    170 		m_freem(m);
    171 		return IPPROTO_DONE;
    172 	}
    173 
    174 	sockaddr_in6_init(&rip6src, &ip6->ip6_src, 0, 0, 0);
    175 	if (sa6_recoverscope(&rip6src) != 0) {
    176 		/* XXX: should be impossible. */
    177 		m_freem(m);
    178 		return IPPROTO_DONE;
    179 	}
    180 
    181 	TAILQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
    182 		in6p = (struct in6pcb *)inph;
    183 		if (in6p->in6p_af != AF_INET6)
    184 			continue;
    185 		if (in6p->in6p_ip6.ip6_nxt &&
    186 		    in6p->in6p_ip6.ip6_nxt != proto)
    187 			continue;
    188 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
    189 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
    190 			continue;
    191 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
    192 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
    193 			continue;
    194 		if (in6p->in6p_cksum != -1) {
    195 			RIP6_STATINC(RIP6_STAT_ISUM);
    196 			if (in6_cksum(m, proto, *offp,
    197 			    m->m_pkthdr.len - *offp)) {
    198 				RIP6_STATINC(RIP6_STAT_BADSUM);
    199 				continue;
    200 			}
    201 		}
    202 		if (last) {
    203 			struct	mbuf *n;
    204 
    205 #ifdef IPSEC
    206 			/*
    207 			 * Check AH/ESP integrity
    208 			 */
    209 			if (ipsec_used && !ipsec6_in_reject(m, last))
    210 #endif /* IPSEC */
    211 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
    212 				if (last->in6p_flags & IN6P_CONTROLOPTS)
    213 					ip6_savecontrol(last, &opts, ip6, n);
    214 				/* strip intermediate headers */
    215 				m_adj(n, *offp);
    216 				if (sbappendaddr(&last->in6p_socket->so_rcv,
    217 				    (struct sockaddr *)&rip6src, n, opts) == 0) {
    218 					/* should notify about lost packet */
    219 					m_freem(n);
    220 					if (opts)
    221 						m_freem(opts);
    222 					RIP6_STATINC(RIP6_STAT_FULLSOCK);
    223 				} else
    224 					sorwakeup(last->in6p_socket);
    225 				opts = NULL;
    226 			}
    227 		}
    228 		last = in6p;
    229 	}
    230 #ifdef IPSEC
    231 	if (ipsec_used && last && ipsec6_in_reject(m, last)) {
    232 		m_freem(m);
    233 		/*
    234 		 * XXX ipsec6_in_reject update stat if there is an error
    235 		 * so we just need to update stats by hand in the case of last is
    236 		 * NULL
    237 		 */
    238 		if (!last)
    239 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
    240 			IP6_STATDEC(IP6_STAT_DELIVERED);
    241 			/* do not inject data into pcb */
    242 		} else
    243 #endif /* IPSEC */
    244 	if (last) {
    245 		if (last->in6p_flags & IN6P_CONTROLOPTS)
    246 			ip6_savecontrol(last, &opts, ip6, m);
    247 		/* strip intermediate headers */
    248 		m_adj(m, *offp);
    249 		if (sbappendaddr(&last->in6p_socket->so_rcv,
    250 		    (struct sockaddr *)&rip6src, m, opts) == 0) {
    251 			m_freem(m);
    252 			if (opts)
    253 				m_freem(opts);
    254 			RIP6_STATINC(RIP6_STAT_FULLSOCK);
    255 		} else
    256 			sorwakeup(last->in6p_socket);
    257 	} else {
    258 		RIP6_STATINC(RIP6_STAT_NOSOCK);
    259 		if (m->m_flags & M_MCAST)
    260 			RIP6_STATINC(RIP6_STAT_NOSOCKMCAST);
    261 		if (proto == IPPROTO_NONE)
    262 			m_freem(m);
    263 		else {
    264 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
    265 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_protounknown);
    266 			icmp6_error(m, ICMP6_PARAM_PROB,
    267 			    ICMP6_PARAMPROB_NEXTHEADER,
    268 			    prvnxtp - mtod(m, u_int8_t *));
    269 		}
    270 		IP6_STATDEC(IP6_STAT_DELIVERED);
    271 	}
    272 	return IPPROTO_DONE;
    273 }
    274 
    275 void *
    276 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
    277 {
    278 	struct ip6_hdr *ip6;
    279 	struct ip6ctlparam *ip6cp = NULL;
    280 	const struct sockaddr_in6 *sa6_src = NULL;
    281 	void *cmdarg;
    282 	void (*notify)(struct in6pcb *, int) = in6_rtchange;
    283 	int nxt;
    284 
    285 	if (sa->sa_family != AF_INET6 ||
    286 	    sa->sa_len != sizeof(struct sockaddr_in6))
    287 		return NULL;
    288 
    289 	if ((unsigned)cmd >= PRC_NCMDS)
    290 		return NULL;
    291 	if (PRC_IS_REDIRECT(cmd))
    292 		notify = in6_rtchange, d = NULL;
    293 	else if (cmd == PRC_HOSTDEAD)
    294 		d = NULL;
    295 	else if (cmd == PRC_MSGSIZE)
    296 		; /* special code is present, see below */
    297 	else if (inet6ctlerrmap[cmd] == 0)
    298 		return NULL;
    299 
    300 	/* if the parameter is from icmp6, decode it. */
    301 	if (d != NULL) {
    302 		ip6cp = (struct ip6ctlparam *)d;
    303 		ip6 = ip6cp->ip6c_ip6;
    304 		cmdarg = ip6cp->ip6c_cmdarg;
    305 		sa6_src = ip6cp->ip6c_src;
    306 		nxt = ip6cp->ip6c_nxt;
    307 	} else {
    308 		ip6 = NULL;
    309 		cmdarg = NULL;
    310 		sa6_src = &sa6_any;
    311 		nxt = -1;
    312 	}
    313 
    314 	if (ip6 && cmd == PRC_MSGSIZE) {
    315 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
    316 		int valid = 0;
    317 		struct in6pcb *in6p;
    318 
    319 		/*
    320 		 * Check to see if we have a valid raw IPv6 socket
    321 		 * corresponding to the address in the ICMPv6 message
    322 		 * payload, and the protocol (ip6_nxt) meets the socket.
    323 		 * XXX chase extension headers, or pass final nxt value
    324 		 * from icmp6_notify_error()
    325 		 */
    326 		in6p = NULL;
    327 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
    328 					     (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0, 0);
    329 #if 0
    330 		if (!in6p) {
    331 			/*
    332 			 * As the use of sendto(2) is fairly popular,
    333 			 * we may want to allow non-connected pcb too.
    334 			 * But it could be too weak against attacks...
    335 			 * We should at least check if the local
    336 			 * address (= s) is really ours.
    337 			 */
    338 			in6p = in6_pcblookup_bind(&raw6cbtable,
    339 			    &sa6->sin6_addr, 0, 0);
    340 		}
    341 #endif
    342 
    343 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
    344 		    in6p->in6p_ip6.ip6_nxt == nxt)
    345 			valid++;
    346 
    347 		/*
    348 		 * Depending on the value of "valid" and routing table
    349 		 * size (mtudisc_{hi,lo}wat), we will:
    350 		 * - recalculate the new MTU and create the
    351 		 *   corresponding routing entry, or
    352 		 * - ignore the MTU change notification.
    353 		 */
    354 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
    355 
    356 		/*
    357 		 * regardless of if we called icmp6_mtudisc_update(),
    358 		 * we need to call in6_pcbnotify(), to notify path MTU
    359 		 * change to the userland (RFC3542), because some
    360 		 * unconnected sockets may share the same destination
    361 		 * and want to know the path MTU.
    362 		 */
    363 	}
    364 
    365 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
    366 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
    367 	return NULL;
    368 }
    369 
    370 /*
    371  * Generate IPv6 header and pass packet to ip6_output.
    372  * Tack on options user may have setup with control call.
    373  */
    374 int
    375 rip6_output(struct mbuf *m, struct socket * const so,
    376     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
    377 {
    378 	struct in6_addr *dst;
    379 	struct ip6_hdr *ip6;
    380 	struct in6pcb *in6p;
    381 	u_int	plen = m->m_pkthdr.len;
    382 	int error = 0;
    383 	struct ip6_pktopts opt, *optp = NULL;
    384 	struct ifnet *oifp = NULL;
    385 	int type, code;		/* for ICMPv6 output statistics only */
    386 	int scope_ambiguous = 0;
    387 	struct in6_addr *in6a;
    388 
    389 	in6p = sotoin6pcb(so);
    390 
    391 	dst = &dstsock->sin6_addr;
    392 	if (control) {
    393 		if ((error = ip6_setpktopts(control, &opt,
    394 		    in6p->in6p_outputopts,
    395 		    kauth_cred_get(), so->so_proto->pr_protocol)) != 0) {
    396 			goto bad;
    397 		}
    398 		optp = &opt;
    399 	} else
    400 		optp = in6p->in6p_outputopts;
    401 
    402 	/*
    403 	 * Check and convert scope zone ID into internal form.
    404 	 * XXX: we may still need to determine the zone later.
    405 	 */
    406 	if (!(so->so_state & SS_ISCONNECTED)) {
    407 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
    408 			scope_ambiguous = 1;
    409 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
    410 			goto bad;
    411 	}
    412 
    413 	/*
    414 	 * For an ICMPv6 packet, we should know its type and code
    415 	 * to update statistics.
    416 	 */
    417 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    418 		struct icmp6_hdr *icmp6;
    419 		if (m->m_len < sizeof(struct icmp6_hdr) &&
    420 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
    421 			error = ENOBUFS;
    422 			goto bad;
    423 		}
    424 		icmp6 = mtod(m, struct icmp6_hdr *);
    425 		type = icmp6->icmp6_type;
    426 		code = icmp6->icmp6_code;
    427 	} else {
    428 		type = 0;
    429 		code = 0;
    430 	}
    431 
    432 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
    433 	if (!m) {
    434 		error = ENOBUFS;
    435 		goto bad;
    436 	}
    437 	ip6 = mtod(m, struct ip6_hdr *);
    438 
    439 	/*
    440 	 * Next header might not be ICMP6 but use its pseudo header anyway.
    441 	 */
    442 	ip6->ip6_dst = *dst;
    443 
    444 	/*
    445 	 * Source address selection.
    446 	 */
    447 	if ((in6a = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
    448 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp,
    449 	    &error)) == 0) {
    450 		if (error == 0)
    451 			error = EADDRNOTAVAIL;
    452 		goto bad;
    453 	}
    454 	ip6->ip6_src = *in6a;
    455 
    456 	if (oifp && scope_ambiguous) {
    457 		/*
    458 		 * Application should provide a proper zone ID or the use of
    459 		 * default zone IDs should be enabled.  Unfortunately, some
    460 		 * applications do not behave as it should, so we need a
    461 		 * workaround.  Even if an appropriate ID is not determined
    462 		 * (when it's required), if we can determine the outgoing
    463 		 * interface. determine the zone ID based on the interface.
    464 		 */
    465 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
    466 		if (error != 0)
    467 			goto bad;
    468 	}
    469 	ip6->ip6_dst = dstsock->sin6_addr;
    470 
    471 	/* fill in the rest of the IPv6 header fields */
    472 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
    473 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
    474 	ip6->ip6_vfc  |= IPV6_VERSION;
    475 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
    476 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
    477 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
    478 
    479 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
    480 	    in6p->in6p_cksum != -1) {
    481 		int off;
    482 		u_int16_t sum;
    483 
    484 		/* compute checksum */
    485 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
    486 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
    487 		else
    488 			off = in6p->in6p_cksum;
    489 		if (plen < off + 1) {
    490 			error = EINVAL;
    491 			goto bad;
    492 		}
    493 		off += sizeof(struct ip6_hdr);
    494 
    495 		sum = 0;
    496 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    497 		    M_DONTWAIT);
    498 		if (m == NULL) {
    499 			error = ENOBUFS;
    500 			goto bad;
    501 		}
    502 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
    503 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    504 		    M_DONTWAIT);
    505 		if (m == NULL) {
    506 			error = ENOBUFS;
    507 			goto bad;
    508 		}
    509 	}
    510 
    511 	error = ip6_output(m, optp, &in6p->in6p_route, 0,
    512 	    in6p->in6p_moptions, so, &oifp);
    513 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    514 		if (oifp)
    515 			icmp6_ifoutstat_inc(oifp, type, code);
    516 		ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
    517 	} else
    518 		RIP6_STATINC(RIP6_STAT_OPACKETS);
    519 
    520 	goto freectl;
    521 
    522  bad:
    523 	if (m)
    524 		m_freem(m);
    525 
    526  freectl:
    527 	if (control) {
    528 		ip6_clearpktopts(&opt, -1);
    529 		m_freem(control);
    530 	}
    531 	return error;
    532 }
    533 
    534 /*
    535  * Raw IPv6 socket option processing.
    536  */
    537 int
    538 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    539 {
    540 	int error = 0;
    541 
    542 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    543 		int optval;
    544 
    545 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
    546 		if (op == PRCO_GETOPT) {
    547 			optval = 1;
    548 			error = sockopt_set(sopt, &optval, sizeof(optval));
    549 		} else if (op == PRCO_SETOPT) {
    550 			error = sockopt_getint(sopt, &optval);
    551 			if (error)
    552 				goto out;
    553 			if (optval == 0)
    554 				error = EINVAL;
    555 		}
    556 
    557 		goto out;
    558 	} else if (sopt->sopt_level != IPPROTO_IPV6)
    559 		return ip6_ctloutput(op, so, sopt);
    560 
    561 	switch (sopt->sopt_name) {
    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(so, sopt);
    571 		else if (op == PRCO_GETOPT)
    572 			error = ip6_mrouter_get(so, sopt);
    573 		else
    574 			error = EINVAL;
    575 		break;
    576 	case IPV6_CHECKSUM:
    577 		return ip6_raw_ctloutput(op, so, sopt);
    578 	default:
    579 		return ip6_ctloutput(op, so, sopt);
    580 	}
    581  out:
    582 	return error;
    583 }
    584 
    585 extern	u_long rip6_sendspace;
    586 extern	u_long rip6_recvspace;
    587 
    588 int
    589 rip6_attach(struct socket *so, int proto)
    590 {
    591 	struct in6pcb *in6p;
    592 	int s, error;
    593 
    594 	KASSERT(sotoin6pcb(so) == NULL);
    595 	sosetlock(so);
    596 
    597 	error = kauth_authorize_network(curlwp->l_cred,
    598 	    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
    599 	    KAUTH_ARG(AF_INET6),
    600 	    KAUTH_ARG(SOCK_RAW),
    601 	    KAUTH_ARG(so->so_proto->pr_protocol));
    602 	if (error) {
    603 		return error;
    604 	}
    605 	s = splsoftnet();
    606 	error = soreserve(so, rip6_sendspace, rip6_recvspace);
    607 	if (error) {
    608 		splx(s);
    609 		return error;
    610 	}
    611 	if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
    612 		splx(s);
    613 		return error;
    614 	}
    615 	splx(s);
    616 	in6p = sotoin6pcb(so);
    617 	in6p->in6p_ip6.ip6_nxt = proto;
    618 	in6p->in6p_cksum = -1;
    619 
    620 	in6p->in6p_icmp6filt = kmem_alloc(sizeof(struct icmp6_filter), KM_SLEEP);
    621 	if (in6p->in6p_icmp6filt == NULL) {
    622 		in6_pcbdetach(in6p);
    623 		return ENOMEM;
    624 	}
    625 	ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
    626 	KASSERT(solocked(so));
    627 	return error;
    628 }
    629 
    630 static void
    631 rip6_detach(struct socket *so)
    632 {
    633 	struct in6pcb *in6p = sotoin6pcb(so);
    634 
    635 	KASSERT(solocked(so));
    636 	KASSERT(in6p != NULL);
    637 
    638 	if (so == ip6_mrouter) {
    639 		ip6_mrouter_done();
    640 	}
    641 	/* xxx: RSVP */
    642 	if (in6p->in6p_icmp6filt != NULL) {
    643 		kmem_free(in6p->in6p_icmp6filt, sizeof(struct icmp6_filter));
    644 		in6p->in6p_icmp6filt = NULL;
    645 	}
    646 	in6_pcbdetach(in6p);
    647 }
    648 
    649 static int
    650 rip6_accept(struct socket *so, struct sockaddr *nam)
    651 {
    652 	KASSERT(solocked(so));
    653 
    654 	return EOPNOTSUPP;
    655 }
    656 
    657 static int
    658 rip6_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    659 {
    660 	struct in6pcb *in6p = sotoin6pcb(so);
    661 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    662 	struct ifaddr *ia = NULL;
    663 	int error = 0;
    664 
    665 	KASSERT(solocked(so));
    666 	KASSERT(in6p != NULL);
    667 	KASSERT(nam != NULL);
    668 
    669 	if (addr->sin6_len != sizeof(*addr))
    670 		return EINVAL;
    671 	if (IFNET_EMPTY() || addr->sin6_family != AF_INET6)
    672 		return EADDRNOTAVAIL;
    673 
    674 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    675 		return error;
    676 
    677 	/*
    678 	 * we don't support mapped address here, it would confuse
    679 	 * users so reject it
    680 	 */
    681 	if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr))
    682 		return EADDRNOTAVAIL;
    683 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
    684 	    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0)
    685 		return EADDRNOTAVAIL;
    686 	if (ia && ((struct in6_ifaddr *)ia)->ia6_flags &
    687 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
    688 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED))
    689 		return EADDRNOTAVAIL;
    690 	in6p->in6p_laddr = addr->sin6_addr;
    691 	return 0;
    692 }
    693 
    694 static int
    695 rip6_listen(struct socket *so, struct lwp *l)
    696 {
    697 	KASSERT(solocked(so));
    698 
    699 	return EOPNOTSUPP;
    700 }
    701 
    702 static int
    703 rip6_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    704 {
    705 	struct in6pcb *in6p = sotoin6pcb(so);
    706 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    707 	struct in6_addr *in6a = NULL;
    708 	struct ifnet *ifp = NULL;
    709 	int scope_ambiguous = 0;
    710 	int error = 0;
    711 
    712 	KASSERT(solocked(so));
    713 	KASSERT(in6p != NULL);
    714 	KASSERT(nam != NULL);
    715 
    716 	if (IFNET_EMPTY())
    717 		return EADDRNOTAVAIL;
    718 	if (addr->sin6_family != AF_INET6)
    719 		return EAFNOSUPPORT;
    720 
    721 	/*
    722 	 * Application should provide a proper zone ID or the use of
    723 	 * default zone IDs should be enabled.  Unfortunately, some
    724 	 * applications do not behave as it should, so we need a
    725 	 * workaround.  Even if an appropriate ID is not determined,
    726 	 * we'll see if we can determine the outgoing interface.  If we
    727 	 * can, determine the zone ID based on the interface below.
    728 	 */
    729 	if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
    730 		scope_ambiguous = 1;
    731 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    732 		return error;
    733 
    734 	/* Source address selection. XXX: need pcblookup? */
    735 	in6a = in6_selectsrc(addr, in6p->in6p_outputopts,
    736 	    in6p->in6p_moptions, &in6p->in6p_route,
    737 	    &in6p->in6p_laddr, &ifp, &error);
    738 	if (in6a == NULL) {
    739 		if (error == 0)
    740 			return EADDRNOTAVAIL;
    741 		return error;
    742 	}
    743 	/* XXX: see above */
    744 	if (ifp && scope_ambiguous &&
    745 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
    746 		return error;
    747 	}
    748 	in6p->in6p_laddr = *in6a;
    749 	in6p->in6p_faddr = addr->sin6_addr;
    750 	soisconnected(so);
    751 	return error;
    752 }
    753 
    754 static int
    755 rip6_connect2(struct socket *so, struct socket *so2)
    756 {
    757 	KASSERT(solocked(so));
    758 
    759 	return EOPNOTSUPP;
    760 }
    761 
    762 static int
    763 rip6_disconnect(struct socket *so)
    764 {
    765 	struct in6pcb *in6p = sotoin6pcb(so);
    766 
    767 	KASSERT(solocked(so));
    768 	KASSERT(in6p != NULL);
    769 
    770 	if ((so->so_state & SS_ISCONNECTED) == 0)
    771 		return ENOTCONN;
    772 
    773 	in6p->in6p_faddr = in6addr_any;
    774 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
    775 	return 0;
    776 }
    777 
    778 static int
    779 rip6_shutdown(struct socket *so)
    780 {
    781 	KASSERT(solocked(so));
    782 
    783 	/*
    784 	 * Mark the connection as being incapable of futther input.
    785 	 */
    786 	socantsendmore(so);
    787 	return 0;
    788 }
    789 
    790 static int
    791 rip6_abort(struct socket *so)
    792 {
    793 	KASSERT(solocked(so));
    794 
    795 	soisdisconnected(so);
    796 	rip6_detach(so);
    797 	return 0;
    798 }
    799 
    800 static int
    801 rip6_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    802 {
    803 	return in6_control(so, cmd, nam, ifp);
    804 }
    805 
    806 static int
    807 rip6_stat(struct socket *so, struct stat *ub)
    808 {
    809 	KASSERT(solocked(so));
    810 
    811 	/* stat: don't bother with a blocksize */
    812 	return 0;
    813 }
    814 
    815 static int
    816 rip6_peeraddr(struct socket *so, struct sockaddr *nam)
    817 {
    818 	KASSERT(solocked(so));
    819 	KASSERT(sotoin6pcb(so) != NULL);
    820 	KASSERT(nam != NULL);
    821 
    822 	in6_setpeeraddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    823 	return 0;
    824 }
    825 
    826 static int
    827 rip6_sockaddr(struct socket *so, struct sockaddr *nam)
    828 {
    829 	KASSERT(solocked(so));
    830 	KASSERT(sotoin6pcb(so) != NULL);
    831 	KASSERT(nam != NULL);
    832 
    833 	in6_setsockaddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    834 	return 0;
    835 }
    836 
    837 static int
    838 rip6_rcvd(struct socket *so, int flags, struct lwp *l)
    839 {
    840 	KASSERT(solocked(so));
    841 
    842 	return EOPNOTSUPP;
    843 }
    844 
    845 static int
    846 rip6_recvoob(struct socket *so, struct mbuf *m, int flags)
    847 {
    848 	KASSERT(solocked(so));
    849 
    850 	return EOPNOTSUPP;
    851 }
    852 
    853 static int
    854 rip6_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    855     struct mbuf *control, struct lwp *l)
    856 {
    857 	struct in6pcb *in6p = sotoin6pcb(so);
    858 	struct sockaddr_in6 tmp;
    859 	struct sockaddr_in6 *dst;
    860 	int error = 0;
    861 
    862 	KASSERT(solocked(so));
    863 	KASSERT(in6p != NULL);
    864 	KASSERT(m != NULL);
    865 
    866 	/*
    867 	 * Ship a packet out. The appropriate raw output
    868 	 * routine handles any messaging necessary.
    869 	 */
    870 
    871 	/* always copy sockaddr to avoid overwrites */
    872 	if (so->so_state & SS_ISCONNECTED) {
    873 		if (nam) {
    874 			error = EISCONN;
    875 			goto release;
    876 		}
    877 		/* XXX */
    878 		sockaddr_in6_init(&tmp, &in6p->in6p_faddr, 0, 0, 0);
    879 		dst = &tmp;
    880 	} else {
    881 		if (nam == NULL) {
    882 			error = ENOTCONN;
    883 			goto release;
    884 		}
    885 		tmp = *(struct sockaddr_in6 *)nam;
    886 		dst = &tmp;
    887 
    888 		if (dst->sin6_family != AF_INET6) {
    889 			error = EAFNOSUPPORT;
    890 			goto release;
    891 		}
    892 	}
    893 	error = rip6_output(m, so, dst, control);
    894 	m = NULL;
    895 
    896 release:
    897 	if (m)
    898 		m_freem(m);
    899 
    900 	return error;
    901 }
    902 
    903 static int
    904 rip6_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    905 {
    906 	KASSERT(solocked(so));
    907 
    908 	if (m)
    909 	 	m_freem(m);
    910 
    911 	return EOPNOTSUPP;
    912 }
    913 
    914 static int
    915 rip6_purgeif(struct socket *so, struct ifnet *ifp)
    916 {
    917 
    918 	mutex_enter(softnet_lock);
    919 	in6_pcbpurgeif0(&raw6cbtable, ifp);
    920 	in6_purgeif(ifp);
    921 	in6_pcbpurgeif(&raw6cbtable, ifp);
    922 	mutex_exit(softnet_lock);
    923 
    924 	return 0;
    925 }
    926 
    927 static int
    928 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
    929 {
    930 
    931 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
    932 }
    933 
    934 static void
    935 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
    936 {
    937 
    938 	sysctl_createv(clog, 0, NULL, NULL,
    939 		       CTLFLAG_PERMANENT,
    940 		       CTLTYPE_NODE, "inet6", NULL,
    941 		       NULL, 0, NULL, 0,
    942 		       CTL_NET, PF_INET6, CTL_EOL);
    943 	sysctl_createv(clog, 0, NULL, NULL,
    944 		       CTLFLAG_PERMANENT,
    945 		       CTLTYPE_NODE, "raw6",
    946 		       SYSCTL_DESCR("Raw IPv6 settings"),
    947 		       NULL, 0, NULL, 0,
    948 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
    949 
    950 	sysctl_createv(clog, 0, NULL, NULL,
    951 		       CTLFLAG_PERMANENT,
    952 		       CTLTYPE_STRUCT, "pcblist",
    953 		       SYSCTL_DESCR("Raw IPv6 control block list"),
    954 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
    955 		       CTL_NET, PF_INET6, IPPROTO_RAW,
    956 		       CTL_CREATE, CTL_EOL);
    957 	sysctl_createv(clog, 0, NULL, NULL,
    958 		       CTLFLAG_PERMANENT,
    959 		       CTLTYPE_STRUCT, "stats",
    960 		       SYSCTL_DESCR("Raw IPv6 statistics"),
    961 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
    962 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
    963 		       CTL_EOL);
    964 }
    965 
    966 PR_WRAP_USRREQS(rip6)
    967 #define	rip6_attach		rip6_attach_wrapper
    968 #define	rip6_detach		rip6_detach_wrapper
    969 #define	rip6_accept		rip6_accept_wrapper
    970 #define	rip6_bind		rip6_bind_wrapper
    971 #define	rip6_listen		rip6_listen_wrapper
    972 #define	rip6_connect		rip6_connect_wrapper
    973 #define	rip6_connect2		rip6_connect2_wrapper
    974 #define	rip6_disconnect		rip6_disconnect_wrapper
    975 #define	rip6_shutdown		rip6_shutdown_wrapper
    976 #define	rip6_abort		rip6_abort_wrapper
    977 #define	rip6_ioctl		rip6_ioctl_wrapper
    978 #define	rip6_stat		rip6_stat_wrapper
    979 #define	rip6_peeraddr		rip6_peeraddr_wrapper
    980 #define	rip6_sockaddr		rip6_sockaddr_wrapper
    981 #define	rip6_rcvd		rip6_rcvd_wrapper
    982 #define	rip6_recvoob		rip6_recvoob_wrapper
    983 #define	rip6_send		rip6_send_wrapper
    984 #define	rip6_sendoob		rip6_sendoob_wrapper
    985 #define	rip6_purgeif		rip6_purgeif_wrapper
    986 
    987 const struct pr_usrreqs rip6_usrreqs = {
    988 	.pr_attach	= rip6_attach,
    989 	.pr_detach	= rip6_detach,
    990 	.pr_accept	= rip6_accept,
    991 	.pr_bind	= rip6_bind,
    992 	.pr_listen	= rip6_listen,
    993 	.pr_connect	= rip6_connect,
    994 	.pr_connect2	= rip6_connect2,
    995 	.pr_disconnect	= rip6_disconnect,
    996 	.pr_shutdown	= rip6_shutdown,
    997 	.pr_abort	= rip6_abort,
    998 	.pr_ioctl	= rip6_ioctl,
    999 	.pr_stat	= rip6_stat,
   1000 	.pr_peeraddr	= rip6_peeraddr,
   1001 	.pr_sockaddr	= rip6_sockaddr,
   1002 	.pr_rcvd	= rip6_rcvd,
   1003 	.pr_recvoob	= rip6_recvoob,
   1004 	.pr_send	= rip6_send,
   1005 	.pr_sendoob	= rip6_sendoob,
   1006 	.pr_purgeif	= rip6_purgeif,
   1007 };
   1008