Home | History | Annotate | Line # | Download | only in netinet6
raw_ip6.c revision 1.148
      1 /*	$NetBSD: raw_ip6.c,v 1.148 2016/07/15 07:40:09 ozaki-r 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 1.148 2016/07/15 07:40:09 ozaki-r 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/if_types.h>
     84 #include <net/net_stats.h>
     85 
     86 #include <netinet/in.h>
     87 #include <netinet/in_var.h>
     88 #include <netinet/ip6.h>
     89 #include <netinet6/ip6_var.h>
     90 #include <netinet6/ip6_private.h>
     91 #include <netinet6/ip6_mroute.h>
     92 #include <netinet/icmp6.h>
     93 #include <netinet6/icmp6_private.h>
     94 #include <netinet6/in6_pcb.h>
     95 #include <netinet6/nd6.h>
     96 #include <netinet6/ip6protosw.h>
     97 #include <netinet6/scope6_var.h>
     98 #include <netinet6/raw_ip6.h>
     99 
    100 #ifdef IPSEC
    101 #include <netipsec/ipsec.h>
    102 #include <netipsec/ipsec_var.h>
    103 #include <netipsec/ipsec_private.h>
    104 #include <netipsec/ipsec6.h>
    105 #endif
    106 
    107 #include "faith.h"
    108 #if defined(NFAITH) && 0 < NFAITH
    109 #include <net/if_faith.h>
    110 #endif
    111 
    112 extern struct inpcbtable rawcbtable;
    113 struct	inpcbtable raw6cbtable;
    114 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
    115 
    116 /*
    117  * Raw interface to IP6 protocol.
    118  */
    119 
    120 static percpu_t *rip6stat_percpu;
    121 
    122 #define	RIP6_STATINC(x)		_NET_STATINC(rip6stat_percpu, x)
    123 
    124 static void sysctl_net_inet6_raw6_setup(struct sysctllog **);
    125 
    126 /*
    127  * Initialize raw connection block queue.
    128  */
    129 void
    130 rip6_init(void)
    131 {
    132 
    133 	sysctl_net_inet6_raw6_setup(NULL);
    134 	in6_pcbinit(&raw6cbtable, 1, 1);
    135 
    136 	rip6stat_percpu = percpu_alloc(sizeof(uint64_t) * RIP6_NSTATS);
    137 }
    138 
    139 /*
    140  * Setup generic address and protocol structures
    141  * for raw_input routine, then pass them along with
    142  * mbuf chain.
    143  */
    144 int
    145 rip6_input(struct mbuf **mp, int *offp, int proto)
    146 {
    147 	struct mbuf *m = *mp;
    148 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
    149 	struct inpcb_hdr *inph;
    150 	struct in6pcb *in6p;
    151 	struct in6pcb *last = NULL;
    152 	struct sockaddr_in6 rip6src;
    153 	struct mbuf *opts = NULL;
    154 
    155 	RIP6_STATINC(RIP6_STAT_IPACKETS);
    156 
    157 #if defined(NFAITH) && 0 < NFAITH
    158 	if (faithprefix(&ip6->ip6_dst)) {
    159 		/* send icmp6 host unreach? */
    160 		m_freem(m);
    161 		return IPPROTO_DONE;
    162 	}
    163 #endif
    164 
    165 	/* Be proactive about malicious use of IPv4 mapped address */
    166 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    167 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    168 		/* XXX stat */
    169 		m_freem(m);
    170 		return IPPROTO_DONE;
    171 	}
    172 
    173 	sockaddr_in6_init(&rip6src, &ip6->ip6_src, 0, 0, 0);
    174 	if (sa6_recoverscope(&rip6src) != 0) {
    175 		/* XXX: should be impossible. */
    176 		m_freem(m);
    177 		return IPPROTO_DONE;
    178 	}
    179 
    180 	TAILQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
    181 		in6p = (struct in6pcb *)inph;
    182 		if (in6p->in6p_af != AF_INET6)
    183 			continue;
    184 		if (in6p->in6p_ip6.ip6_nxt &&
    185 		    in6p->in6p_ip6.ip6_nxt != proto)
    186 			continue;
    187 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
    188 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
    189 			continue;
    190 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
    191 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
    192 			continue;
    193 		if (in6p->in6p_cksum != -1) {
    194 			RIP6_STATINC(RIP6_STAT_ISUM);
    195 			if (in6_cksum(m, proto, *offp,
    196 			    m->m_pkthdr.len - *offp)) {
    197 				RIP6_STATINC(RIP6_STAT_BADSUM);
    198 				continue;
    199 			}
    200 		}
    201 		if (last) {
    202 			struct	mbuf *n;
    203 
    204 #ifdef IPSEC
    205 			/*
    206 			 * Check AH/ESP integrity
    207 			 */
    208 			if (ipsec_used && !ipsec6_in_reject(m, last))
    209 #endif /* IPSEC */
    210 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
    211 				if (last->in6p_flags & IN6P_CONTROLOPTS)
    212 					ip6_savecontrol(last, &opts, ip6, n);
    213 				/* strip intermediate headers */
    214 				m_adj(n, *offp);
    215 				if (sbappendaddr(&last->in6p_socket->so_rcv,
    216 				    sin6tosa(&rip6src), n, opts) == 0) {
    217 					/* should notify about lost packet */
    218 					m_freem(n);
    219 					if (opts)
    220 						m_freem(opts);
    221 					RIP6_STATINC(RIP6_STAT_FULLSOCK);
    222 				} else
    223 					sorwakeup(last->in6p_socket);
    224 				opts = NULL;
    225 			}
    226 		}
    227 		last = in6p;
    228 	}
    229 #ifdef IPSEC
    230 	if (ipsec_used && last && ipsec6_in_reject(m, last)) {
    231 		m_freem(m);
    232 		/*
    233 		 * XXX ipsec6_in_reject update stat if there is an error
    234 		 * so we just need to update stats by hand in the case of last is
    235 		 * NULL
    236 		 */
    237 		if (!last)
    238 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
    239 			IP6_STATDEC(IP6_STAT_DELIVERED);
    240 			/* do not inject data into pcb */
    241 		} else
    242 #endif /* IPSEC */
    243 	if (last) {
    244 		if (last->in6p_flags & IN6P_CONTROLOPTS)
    245 			ip6_savecontrol(last, &opts, ip6, m);
    246 		/* strip intermediate headers */
    247 		m_adj(m, *offp);
    248 		if (sbappendaddr(&last->in6p_socket->so_rcv,
    249 		    sin6tosa(&rip6src), m, opts) == 0) {
    250 			m_freem(m);
    251 			if (opts)
    252 				m_freem(opts);
    253 			RIP6_STATINC(RIP6_STAT_FULLSOCK);
    254 		} else
    255 			sorwakeup(last->in6p_socket);
    256 	} else {
    257 		RIP6_STATINC(RIP6_STAT_NOSOCK);
    258 		if (m->m_flags & M_MCAST)
    259 			RIP6_STATINC(RIP6_STAT_NOSOCKMCAST);
    260 		if (proto == IPPROTO_NONE)
    261 			m_freem(m);
    262 		else {
    263 			int s;
    264 			struct ifnet *rcvif = m_get_rcvif(m, &s);
    265 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
    266 			in6_ifstat_inc(rcvif, ifs6_in_protounknown);
    267 			m_put_rcvif(rcvif, &s);
    268 			icmp6_error(m, ICMP6_PARAM_PROB,
    269 			    ICMP6_PARAMPROB_NEXTHEADER,
    270 			    prvnxtp - mtod(m, u_int8_t *));
    271 		}
    272 		IP6_STATDEC(IP6_STAT_DELIVERED);
    273 	}
    274 	return IPPROTO_DONE;
    275 }
    276 
    277 void *
    278 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
    279 {
    280 	struct ip6_hdr *ip6;
    281 	struct ip6ctlparam *ip6cp = NULL;
    282 	const struct sockaddr_in6 *sa6_src = NULL;
    283 	void *cmdarg;
    284 	void (*notify)(struct in6pcb *, int) = in6_rtchange;
    285 	int nxt;
    286 
    287 	if (sa->sa_family != AF_INET6 ||
    288 	    sa->sa_len != sizeof(struct sockaddr_in6))
    289 		return NULL;
    290 
    291 	if ((unsigned)cmd >= PRC_NCMDS)
    292 		return NULL;
    293 	if (PRC_IS_REDIRECT(cmd))
    294 		notify = in6_rtchange, d = NULL;
    295 	else if (cmd == PRC_HOSTDEAD)
    296 		d = NULL;
    297 	else if (cmd == PRC_MSGSIZE)
    298 		; /* special code is present, see below */
    299 	else if (inet6ctlerrmap[cmd] == 0)
    300 		return NULL;
    301 
    302 	/* if the parameter is from icmp6, decode it. */
    303 	if (d != NULL) {
    304 		ip6cp = (struct ip6ctlparam *)d;
    305 		ip6 = ip6cp->ip6c_ip6;
    306 		cmdarg = ip6cp->ip6c_cmdarg;
    307 		sa6_src = ip6cp->ip6c_src;
    308 		nxt = ip6cp->ip6c_nxt;
    309 	} else {
    310 		ip6 = NULL;
    311 		cmdarg = NULL;
    312 		sa6_src = &sa6_any;
    313 		nxt = -1;
    314 	}
    315 
    316 	if (ip6 && cmd == PRC_MSGSIZE) {
    317 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
    318 		int valid = 0;
    319 		struct in6pcb *in6p;
    320 
    321 		/*
    322 		 * Check to see if we have a valid raw IPv6 socket
    323 		 * corresponding to the address in the ICMPv6 message
    324 		 * payload, and the protocol (ip6_nxt) meets the socket.
    325 		 * XXX chase extension headers, or pass final nxt value
    326 		 * from icmp6_notify_error()
    327 		 */
    328 		in6p = NULL;
    329 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
    330 					     (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0, 0);
    331 #if 0
    332 		if (!in6p) {
    333 			/*
    334 			 * As the use of sendto(2) is fairly popular,
    335 			 * we may want to allow non-connected pcb too.
    336 			 * But it could be too weak against attacks...
    337 			 * We should at least check if the local
    338 			 * address (= s) is really ours.
    339 			 */
    340 			in6p = in6_pcblookup_bind(&raw6cbtable,
    341 			    &sa6->sin6_addr, 0, 0);
    342 		}
    343 #endif
    344 
    345 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
    346 		    in6p->in6p_ip6.ip6_nxt == nxt)
    347 			valid++;
    348 
    349 		/*
    350 		 * Depending on the value of "valid" and routing table
    351 		 * size (mtudisc_{hi,lo}wat), we will:
    352 		 * - recalculate the new MTU and create the
    353 		 *   corresponding routing entry, or
    354 		 * - ignore the MTU change notification.
    355 		 */
    356 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
    357 
    358 		/*
    359 		 * regardless of if we called icmp6_mtudisc_update(),
    360 		 * we need to call in6_pcbnotify(), to notify path MTU
    361 		 * change to the userland (RFC3542), because some
    362 		 * unconnected sockets may share the same destination
    363 		 * and want to know the path MTU.
    364 		 */
    365 	}
    366 
    367 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
    368 	    sin6tocsa(sa6_src), 0, cmd, cmdarg, notify);
    369 	return NULL;
    370 }
    371 
    372 /*
    373  * Generate IPv6 header and pass packet to ip6_output.
    374  * Tack on options user may have setup with control call.
    375  */
    376 int
    377 rip6_output(struct mbuf *m, struct socket * const so,
    378     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
    379 {
    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 scope_ambiguous = 0;
    389 	struct in6_addr *in6a;
    390 	int bound = curlwp_bind();
    391 	struct psref psref;
    392 
    393 	in6p = sotoin6pcb(so);
    394 
    395 	dst = &dstsock->sin6_addr;
    396 	if (control) {
    397 		if ((error = ip6_setpktopts(control, &opt,
    398 		    in6p->in6p_outputopts,
    399 		    kauth_cred_get(), so->so_proto->pr_protocol)) != 0) {
    400 			goto bad;
    401 		}
    402 		optp = &opt;
    403 	} else
    404 		optp = in6p->in6p_outputopts;
    405 
    406 	/*
    407 	 * Check and convert scope zone ID into internal form.
    408 	 * XXX: we may still need to determine the zone later.
    409 	 */
    410 	if (!(so->so_state & SS_ISCONNECTED)) {
    411 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
    412 			scope_ambiguous = 1;
    413 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
    414 			goto bad;
    415 	}
    416 
    417 	/*
    418 	 * For an ICMPv6 packet, we should know its type and code
    419 	 * to update statistics.
    420 	 */
    421 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    422 		struct icmp6_hdr *icmp6;
    423 		if (m->m_len < sizeof(struct icmp6_hdr) &&
    424 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
    425 			error = ENOBUFS;
    426 			goto bad;
    427 		}
    428 		icmp6 = mtod(m, struct icmp6_hdr *);
    429 		type = icmp6->icmp6_type;
    430 		code = icmp6->icmp6_code;
    431 	} else {
    432 		type = 0;
    433 		code = 0;
    434 	}
    435 
    436 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
    437 	if (!m) {
    438 		error = ENOBUFS;
    439 		goto bad;
    440 	}
    441 	ip6 = mtod(m, struct ip6_hdr *);
    442 
    443 	/*
    444 	 * Next header might not be ICMP6 but use its pseudo header anyway.
    445 	 */
    446 	ip6->ip6_dst = *dst;
    447 
    448 	/*
    449 	 * Source address selection.
    450 	 */
    451 	if ((in6a = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
    452 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp, &psref, &error)) == 0) {
    453 		if (error == 0)
    454 			error = EADDRNOTAVAIL;
    455 		goto bad;
    456 	}
    457 	ip6->ip6_src = *in6a;
    458 
    459 	if (oifp && scope_ambiguous) {
    460 		/*
    461 		 * Application should provide a proper zone ID or the use of
    462 		 * default zone IDs should be enabled.  Unfortunately, some
    463 		 * applications do not behave as it should, so we need a
    464 		 * workaround.  Even if an appropriate ID is not determined
    465 		 * (when it's required), if we can determine the outgoing
    466 		 * interface. determine the zone ID based on the interface.
    467 		 */
    468 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
    469 		if (error != 0)
    470 			goto bad;
    471 	}
    472 	ip6->ip6_dst = dstsock->sin6_addr;
    473 
    474 	/* fill in the rest of the IPv6 header fields */
    475 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
    476 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
    477 	ip6->ip6_vfc  |= IPV6_VERSION;
    478 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
    479 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
    480 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
    481 
    482 	if_put(oifp, &psref);
    483 	oifp = NULL;
    484 
    485 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
    486 	    in6p->in6p_cksum != -1) {
    487 		int off;
    488 		u_int16_t sum;
    489 
    490 		/* compute checksum */
    491 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
    492 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
    493 		else
    494 			off = in6p->in6p_cksum;
    495 		if (plen < off + 1) {
    496 			error = EINVAL;
    497 			goto bad;
    498 		}
    499 		off += sizeof(struct ip6_hdr);
    500 
    501 		sum = 0;
    502 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    503 		    M_DONTWAIT);
    504 		if (m == NULL) {
    505 			error = ENOBUFS;
    506 			goto bad;
    507 		}
    508 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
    509 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    510 		    M_DONTWAIT);
    511 		if (m == NULL) {
    512 			error = ENOBUFS;
    513 			goto bad;
    514 		}
    515 	}
    516 
    517 	{
    518 		struct ifnet *ret_oifp = NULL;
    519 
    520 		error = ip6_output(m, optp, &in6p->in6p_route, 0,
    521 		    in6p->in6p_moptions, so, &ret_oifp);
    522 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    523 			if (ret_oifp)
    524 				icmp6_ifoutstat_inc(ret_oifp, type, code);
    525 			ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
    526 		} else
    527 			RIP6_STATINC(RIP6_STAT_OPACKETS);
    528 	}
    529 
    530 	goto freectl;
    531 
    532  bad:
    533 	if (m)
    534 		m_freem(m);
    535 
    536  freectl:
    537 	if (control) {
    538 		ip6_clearpktopts(&opt, -1);
    539 		m_freem(control);
    540 	}
    541 	if_put(oifp, &psref);
    542 	curlwp_bindx(bound);
    543 	return error;
    544 }
    545 
    546 /*
    547  * Raw IPv6 socket option processing.
    548  */
    549 int
    550 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    551 {
    552 	int error = 0;
    553 
    554 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    555 		int optval;
    556 
    557 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
    558 		if (op == PRCO_GETOPT) {
    559 			optval = 1;
    560 			error = sockopt_set(sopt, &optval, sizeof(optval));
    561 		} else if (op == PRCO_SETOPT) {
    562 			error = sockopt_getint(sopt, &optval);
    563 			if (error)
    564 				goto out;
    565 			if (optval == 0)
    566 				error = EINVAL;
    567 		}
    568 
    569 		goto out;
    570 	} else if (sopt->sopt_level != IPPROTO_IPV6)
    571 		return ip6_ctloutput(op, so, sopt);
    572 
    573 	switch (sopt->sopt_name) {
    574 	case MRT6_INIT:
    575 	case MRT6_DONE:
    576 	case MRT6_ADD_MIF:
    577 	case MRT6_DEL_MIF:
    578 	case MRT6_ADD_MFC:
    579 	case MRT6_DEL_MFC:
    580 	case MRT6_PIM:
    581 		if (op == PRCO_SETOPT)
    582 			error = ip6_mrouter_set(so, sopt);
    583 		else if (op == PRCO_GETOPT)
    584 			error = ip6_mrouter_get(so, sopt);
    585 		else
    586 			error = EINVAL;
    587 		break;
    588 	case IPV6_CHECKSUM:
    589 		return ip6_raw_ctloutput(op, so, sopt);
    590 	default:
    591 		return ip6_ctloutput(op, so, sopt);
    592 	}
    593  out:
    594 	return error;
    595 }
    596 
    597 extern	u_long rip6_sendspace;
    598 extern	u_long rip6_recvspace;
    599 
    600 int
    601 rip6_attach(struct socket *so, int proto)
    602 {
    603 	struct in6pcb *in6p;
    604 	int s, error;
    605 
    606 	KASSERT(sotoin6pcb(so) == NULL);
    607 	sosetlock(so);
    608 
    609 	error = kauth_authorize_network(curlwp->l_cred,
    610 	    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
    611 	    KAUTH_ARG(AF_INET6),
    612 	    KAUTH_ARG(SOCK_RAW),
    613 	    KAUTH_ARG(so->so_proto->pr_protocol));
    614 	if (error) {
    615 		return error;
    616 	}
    617 	s = splsoftnet();
    618 	error = soreserve(so, rip6_sendspace, rip6_recvspace);
    619 	if (error) {
    620 		splx(s);
    621 		return error;
    622 	}
    623 	if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
    624 		splx(s);
    625 		return error;
    626 	}
    627 	splx(s);
    628 	in6p = sotoin6pcb(so);
    629 	in6p->in6p_ip6.ip6_nxt = proto;
    630 	in6p->in6p_cksum = -1;
    631 
    632 	in6p->in6p_icmp6filt = kmem_alloc(sizeof(struct icmp6_filter), KM_SLEEP);
    633 	if (in6p->in6p_icmp6filt == NULL) {
    634 		in6_pcbdetach(in6p);
    635 		return ENOMEM;
    636 	}
    637 	ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
    638 	KASSERT(solocked(so));
    639 	return error;
    640 }
    641 
    642 static void
    643 rip6_detach(struct socket *so)
    644 {
    645 	struct in6pcb *in6p = sotoin6pcb(so);
    646 
    647 	KASSERT(solocked(so));
    648 	KASSERT(in6p != NULL);
    649 
    650 	if (so == ip6_mrouter) {
    651 		ip6_mrouter_done();
    652 	}
    653 	/* xxx: RSVP */
    654 	if (in6p->in6p_icmp6filt != NULL) {
    655 		kmem_free(in6p->in6p_icmp6filt, sizeof(struct icmp6_filter));
    656 		in6p->in6p_icmp6filt = NULL;
    657 	}
    658 	in6_pcbdetach(in6p);
    659 }
    660 
    661 static int
    662 rip6_accept(struct socket *so, struct sockaddr *nam)
    663 {
    664 	KASSERT(solocked(so));
    665 
    666 	return EOPNOTSUPP;
    667 }
    668 
    669 static int
    670 rip6_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    671 {
    672 	struct in6pcb *in6p = sotoin6pcb(so);
    673 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    674 	struct ifaddr *ia = NULL;
    675 	int error = 0;
    676 
    677 	KASSERT(solocked(so));
    678 	KASSERT(in6p != NULL);
    679 	KASSERT(nam != NULL);
    680 
    681 	if (addr->sin6_len != sizeof(*addr))
    682 		return EINVAL;
    683 	if (IFNET_READER_EMPTY() || addr->sin6_family != AF_INET6)
    684 		return EADDRNOTAVAIL;
    685 
    686 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    687 		return error;
    688 
    689 	/*
    690 	 * we don't support mapped address here, it would confuse
    691 	 * users so reject it
    692 	 */
    693 	if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr))
    694 		return EADDRNOTAVAIL;
    695 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
    696 	    (ia = ifa_ifwithaddr(sin6tosa(addr))) == 0)
    697 		return EADDRNOTAVAIL;
    698 	if (ia && ifatoia6(ia)->ia6_flags &
    699 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
    700 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED))
    701 		return EADDRNOTAVAIL;
    702 	in6p->in6p_laddr = addr->sin6_addr;
    703 	return 0;
    704 }
    705 
    706 static int
    707 rip6_listen(struct socket *so, struct lwp *l)
    708 {
    709 	KASSERT(solocked(so));
    710 
    711 	return EOPNOTSUPP;
    712 }
    713 
    714 static int
    715 rip6_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    716 {
    717 	struct in6pcb *in6p = sotoin6pcb(so);
    718 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    719 	struct in6_addr *in6a = NULL;
    720 	struct ifnet *ifp = NULL;
    721 	int scope_ambiguous = 0;
    722 	int error = 0;
    723 	struct psref psref;
    724 	int bound;
    725 
    726 	KASSERT(solocked(so));
    727 	KASSERT(in6p != NULL);
    728 	KASSERT(nam != NULL);
    729 
    730 	if (IFNET_READER_EMPTY())
    731 		return EADDRNOTAVAIL;
    732 	if (addr->sin6_family != AF_INET6)
    733 		return EAFNOSUPPORT;
    734 
    735 	/*
    736 	 * Application should provide a proper zone ID or the use of
    737 	 * default zone IDs should be enabled.  Unfortunately, some
    738 	 * applications do not behave as it should, so we need a
    739 	 * workaround.  Even if an appropriate ID is not determined,
    740 	 * we'll see if we can determine the outgoing interface.  If we
    741 	 * can, determine the zone ID based on the interface below.
    742 	 */
    743 	if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
    744 		scope_ambiguous = 1;
    745 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    746 		return error;
    747 
    748 	bound = curlwp_bind();
    749 	/* Source address selection. XXX: need pcblookup? */
    750 	in6a = in6_selectsrc(addr, in6p->in6p_outputopts,
    751 	    in6p->in6p_moptions, &in6p->in6p_route,
    752 	    &in6p->in6p_laddr, &ifp, &psref, &error);
    753 	if (in6a == NULL) {
    754 		if (error == 0)
    755 			error = EADDRNOTAVAIL;
    756 		goto out;
    757 	}
    758 	/* XXX: see above */
    759 	if (ifp && scope_ambiguous &&
    760 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
    761 		goto out;
    762 	}
    763 	in6p->in6p_laddr = *in6a;
    764 	in6p->in6p_faddr = addr->sin6_addr;
    765 	soisconnected(so);
    766 out:
    767 	if_put(ifp, &psref);
    768 	curlwp_bindx(bound);
    769 	return error;
    770 }
    771 
    772 static int
    773 rip6_connect2(struct socket *so, struct socket *so2)
    774 {
    775 	KASSERT(solocked(so));
    776 
    777 	return EOPNOTSUPP;
    778 }
    779 
    780 static int
    781 rip6_disconnect(struct socket *so)
    782 {
    783 	struct in6pcb *in6p = sotoin6pcb(so);
    784 
    785 	KASSERT(solocked(so));
    786 	KASSERT(in6p != NULL);
    787 
    788 	if ((so->so_state & SS_ISCONNECTED) == 0)
    789 		return ENOTCONN;
    790 
    791 	in6p->in6p_faddr = in6addr_any;
    792 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
    793 	return 0;
    794 }
    795 
    796 static int
    797 rip6_shutdown(struct socket *so)
    798 {
    799 	KASSERT(solocked(so));
    800 
    801 	/*
    802 	 * Mark the connection as being incapable of futther input.
    803 	 */
    804 	socantsendmore(so);
    805 	return 0;
    806 }
    807 
    808 static int
    809 rip6_abort(struct socket *so)
    810 {
    811 	KASSERT(solocked(so));
    812 
    813 	soisdisconnected(so);
    814 	rip6_detach(so);
    815 	return 0;
    816 }
    817 
    818 static int
    819 rip6_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    820 {
    821 	return in6_control(so, cmd, nam, ifp);
    822 }
    823 
    824 static int
    825 rip6_stat(struct socket *so, struct stat *ub)
    826 {
    827 	KASSERT(solocked(so));
    828 
    829 	/* stat: don't bother with a blocksize */
    830 	return 0;
    831 }
    832 
    833 static int
    834 rip6_peeraddr(struct socket *so, struct sockaddr *nam)
    835 {
    836 	KASSERT(solocked(so));
    837 	KASSERT(sotoin6pcb(so) != NULL);
    838 	KASSERT(nam != NULL);
    839 
    840 	in6_setpeeraddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    841 	return 0;
    842 }
    843 
    844 static int
    845 rip6_sockaddr(struct socket *so, struct sockaddr *nam)
    846 {
    847 	KASSERT(solocked(so));
    848 	KASSERT(sotoin6pcb(so) != NULL);
    849 	KASSERT(nam != NULL);
    850 
    851 	in6_setsockaddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    852 	return 0;
    853 }
    854 
    855 static int
    856 rip6_rcvd(struct socket *so, int flags, struct lwp *l)
    857 {
    858 	KASSERT(solocked(so));
    859 
    860 	return EOPNOTSUPP;
    861 }
    862 
    863 static int
    864 rip6_recvoob(struct socket *so, struct mbuf *m, int flags)
    865 {
    866 	KASSERT(solocked(so));
    867 
    868 	return EOPNOTSUPP;
    869 }
    870 
    871 static int
    872 rip6_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    873     struct mbuf *control, struct lwp *l)
    874 {
    875 	struct in6pcb *in6p = sotoin6pcb(so);
    876 	struct sockaddr_in6 tmp;
    877 	struct sockaddr_in6 *dst;
    878 	int error = 0;
    879 
    880 	KASSERT(solocked(so));
    881 	KASSERT(in6p != NULL);
    882 	KASSERT(m != NULL);
    883 
    884 	/*
    885 	 * Ship a packet out. The appropriate raw output
    886 	 * routine handles any messaging necessary.
    887 	 */
    888 
    889 	/* always copy sockaddr to avoid overwrites */
    890 	if (so->so_state & SS_ISCONNECTED) {
    891 		if (nam) {
    892 			error = EISCONN;
    893 			goto release;
    894 		}
    895 		/* XXX */
    896 		sockaddr_in6_init(&tmp, &in6p->in6p_faddr, 0, 0, 0);
    897 		dst = &tmp;
    898 	} else {
    899 		if (nam == NULL) {
    900 			error = ENOTCONN;
    901 			goto release;
    902 		}
    903 		tmp = *(struct sockaddr_in6 *)nam;
    904 		dst = &tmp;
    905 
    906 		if (dst->sin6_family != AF_INET6) {
    907 			error = EAFNOSUPPORT;
    908 			goto release;
    909 		}
    910 	}
    911 	error = rip6_output(m, so, dst, control);
    912 	m = NULL;
    913 
    914 release:
    915 	if (m)
    916 		m_freem(m);
    917 
    918 	return error;
    919 }
    920 
    921 static int
    922 rip6_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    923 {
    924 	KASSERT(solocked(so));
    925 
    926 	if (m)
    927 	 	m_freem(m);
    928 
    929 	return EOPNOTSUPP;
    930 }
    931 
    932 static int
    933 rip6_purgeif(struct socket *so, struct ifnet *ifp)
    934 {
    935 
    936 	mutex_enter(softnet_lock);
    937 	in6_pcbpurgeif0(&raw6cbtable, ifp);
    938 	in6_purgeif(ifp);
    939 	in6_pcbpurgeif(&raw6cbtable, ifp);
    940 	mutex_exit(softnet_lock);
    941 
    942 	return 0;
    943 }
    944 
    945 static int
    946 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
    947 {
    948 
    949 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
    950 }
    951 
    952 static void
    953 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
    954 {
    955 
    956 	sysctl_createv(clog, 0, NULL, NULL,
    957 		       CTLFLAG_PERMANENT,
    958 		       CTLTYPE_NODE, "inet6", NULL,
    959 		       NULL, 0, NULL, 0,
    960 		       CTL_NET, PF_INET6, CTL_EOL);
    961 	sysctl_createv(clog, 0, NULL, NULL,
    962 		       CTLFLAG_PERMANENT,
    963 		       CTLTYPE_NODE, "raw6",
    964 		       SYSCTL_DESCR("Raw IPv6 settings"),
    965 		       NULL, 0, NULL, 0,
    966 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
    967 
    968 	sysctl_createv(clog, 0, NULL, NULL,
    969 		       CTLFLAG_PERMANENT,
    970 		       CTLTYPE_STRUCT, "pcblist",
    971 		       SYSCTL_DESCR("Raw IPv6 control block list"),
    972 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
    973 		       CTL_NET, PF_INET6, IPPROTO_RAW,
    974 		       CTL_CREATE, CTL_EOL);
    975 	sysctl_createv(clog, 0, NULL, NULL,
    976 		       CTLFLAG_PERMANENT,
    977 		       CTLTYPE_STRUCT, "stats",
    978 		       SYSCTL_DESCR("Raw IPv6 statistics"),
    979 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
    980 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
    981 		       CTL_EOL);
    982 }
    983 
    984 PR_WRAP_USRREQS(rip6)
    985 #define	rip6_attach		rip6_attach_wrapper
    986 #define	rip6_detach		rip6_detach_wrapper
    987 #define	rip6_accept		rip6_accept_wrapper
    988 #define	rip6_bind		rip6_bind_wrapper
    989 #define	rip6_listen		rip6_listen_wrapper
    990 #define	rip6_connect		rip6_connect_wrapper
    991 #define	rip6_connect2		rip6_connect2_wrapper
    992 #define	rip6_disconnect		rip6_disconnect_wrapper
    993 #define	rip6_shutdown		rip6_shutdown_wrapper
    994 #define	rip6_abort		rip6_abort_wrapper
    995 #define	rip6_ioctl		rip6_ioctl_wrapper
    996 #define	rip6_stat		rip6_stat_wrapper
    997 #define	rip6_peeraddr		rip6_peeraddr_wrapper
    998 #define	rip6_sockaddr		rip6_sockaddr_wrapper
    999 #define	rip6_rcvd		rip6_rcvd_wrapper
   1000 #define	rip6_recvoob		rip6_recvoob_wrapper
   1001 #define	rip6_send		rip6_send_wrapper
   1002 #define	rip6_sendoob		rip6_sendoob_wrapper
   1003 #define	rip6_purgeif		rip6_purgeif_wrapper
   1004 
   1005 const struct pr_usrreqs rip6_usrreqs = {
   1006 	.pr_attach	= rip6_attach,
   1007 	.pr_detach	= rip6_detach,
   1008 	.pr_accept	= rip6_accept,
   1009 	.pr_bind	= rip6_bind,
   1010 	.pr_listen	= rip6_listen,
   1011 	.pr_connect	= rip6_connect,
   1012 	.pr_connect2	= rip6_connect2,
   1013 	.pr_disconnect	= rip6_disconnect,
   1014 	.pr_shutdown	= rip6_shutdown,
   1015 	.pr_abort	= rip6_abort,
   1016 	.pr_ioctl	= rip6_ioctl,
   1017 	.pr_stat	= rip6_stat,
   1018 	.pr_peeraddr	= rip6_peeraddr,
   1019 	.pr_sockaddr	= rip6_sockaddr,
   1020 	.pr_rcvd	= rip6_rcvd,
   1021 	.pr_recvoob	= rip6_recvoob,
   1022 	.pr_send	= rip6_send,
   1023 	.pr_sendoob	= rip6_sendoob,
   1024 	.pr_purgeif	= rip6_purgeif,
   1025 };
   1026