Home | History | Annotate | Line # | Download | only in netinet6
raw_ip6.c revision 1.136.4.7
      1 /*	$NetBSD: raw_ip6.c,v 1.136.4.7 2016/12/05 10:55:28 skrll 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 1.136.4.7 2016/12/05 10:55:28 skrll Exp $");
     66 
     67 #ifdef _KERNEL_OPT
     68 #include "opt_ipsec.h"
     69 #include "opt_net_mpsafe.h"
     70 #endif
     71 
     72 #include <sys/param.h>
     73 #include <sys/sysctl.h>
     74 #include <sys/mbuf.h>
     75 #include <sys/socket.h>
     76 #include <sys/protosw.h>
     77 #include <sys/socketvar.h>
     78 #include <sys/systm.h>
     79 #include <sys/proc.h>
     80 #include <sys/kauth.h>
     81 #include <sys/kmem.h>
     82 
     83 #include <net/if.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 				    sin6tosa(&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 		    sin6tosa(&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 			int s;
    265 			struct ifnet *rcvif = m_get_rcvif(m, &s);
    266 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
    267 			in6_ifstat_inc(rcvif, ifs6_in_protounknown);
    268 			m_put_rcvif(rcvif, &s);
    269 			icmp6_error(m, ICMP6_PARAM_PROB,
    270 			    ICMP6_PARAMPROB_NEXTHEADER,
    271 			    prvnxtp - mtod(m, u_int8_t *));
    272 		}
    273 		IP6_STATDEC(IP6_STAT_DELIVERED);
    274 	}
    275 	return IPPROTO_DONE;
    276 }
    277 
    278 void *
    279 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
    280 {
    281 	struct ip6_hdr *ip6;
    282 	struct ip6ctlparam *ip6cp = NULL;
    283 	const struct sockaddr_in6 *sa6_src = NULL;
    284 	void *cmdarg;
    285 	void (*notify)(struct in6pcb *, int) = in6_rtchange;
    286 	int nxt;
    287 
    288 	if (sa->sa_family != AF_INET6 ||
    289 	    sa->sa_len != sizeof(struct sockaddr_in6))
    290 		return NULL;
    291 
    292 	if ((unsigned)cmd >= PRC_NCMDS)
    293 		return NULL;
    294 	if (PRC_IS_REDIRECT(cmd))
    295 		notify = in6_rtchange, d = NULL;
    296 	else if (cmd == PRC_HOSTDEAD)
    297 		d = NULL;
    298 	else if (cmd == PRC_MSGSIZE)
    299 		; /* special code is present, see below */
    300 	else if (inet6ctlerrmap[cmd] == 0)
    301 		return NULL;
    302 
    303 	/* if the parameter is from icmp6, decode it. */
    304 	if (d != NULL) {
    305 		ip6cp = (struct ip6ctlparam *)d;
    306 		ip6 = ip6cp->ip6c_ip6;
    307 		cmdarg = ip6cp->ip6c_cmdarg;
    308 		sa6_src = ip6cp->ip6c_src;
    309 		nxt = ip6cp->ip6c_nxt;
    310 	} else {
    311 		ip6 = NULL;
    312 		cmdarg = NULL;
    313 		sa6_src = &sa6_any;
    314 		nxt = -1;
    315 	}
    316 
    317 	if (ip6 && cmd == PRC_MSGSIZE) {
    318 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
    319 		int valid = 0;
    320 		struct in6pcb *in6p;
    321 
    322 		/*
    323 		 * Check to see if we have a valid raw IPv6 socket
    324 		 * corresponding to the address in the ICMPv6 message
    325 		 * payload, and the protocol (ip6_nxt) meets the socket.
    326 		 * XXX chase extension headers, or pass final nxt value
    327 		 * from icmp6_notify_error()
    328 		 */
    329 		in6p = NULL;
    330 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
    331 					     (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0, 0);
    332 #if 0
    333 		if (!in6p) {
    334 			/*
    335 			 * As the use of sendto(2) is fairly popular,
    336 			 * we may want to allow non-connected pcb too.
    337 			 * But it could be too weak against attacks...
    338 			 * We should at least check if the local
    339 			 * address (= s) is really ours.
    340 			 */
    341 			in6p = in6_pcblookup_bind(&raw6cbtable,
    342 			    &sa6->sin6_addr, 0, 0);
    343 		}
    344 #endif
    345 
    346 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
    347 		    in6p->in6p_ip6.ip6_nxt == nxt)
    348 			valid++;
    349 
    350 		/*
    351 		 * Depending on the value of "valid" and routing table
    352 		 * size (mtudisc_{hi,lo}wat), we will:
    353 		 * - recalculate the new MTU and create the
    354 		 *   corresponding routing entry, or
    355 		 * - ignore the MTU change notification.
    356 		 */
    357 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
    358 
    359 		/*
    360 		 * regardless of if we called icmp6_mtudisc_update(),
    361 		 * we need to call in6_pcbnotify(), to notify path MTU
    362 		 * change to the userland (RFC3542), because some
    363 		 * unconnected sockets may share the same destination
    364 		 * and want to know the path MTU.
    365 		 */
    366 	}
    367 
    368 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
    369 	    sin6tocsa(sa6_src), 0, cmd, cmdarg, notify);
    370 	return NULL;
    371 }
    372 
    373 /*
    374  * Generate IPv6 header and pass packet to ip6_output.
    375  * Tack on options user may have setup with control call.
    376  */
    377 int
    378 rip6_output(struct mbuf *m, struct socket * const so,
    379     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
    380 {
    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 scope_ambiguous = 0;
    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 	error = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
    452 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp, &psref, &ip6->ip6_src);
    453 	if (error != 0)
    454 		goto bad;
    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_put(oifp, &psref);
    480 	oifp = NULL;
    481 
    482 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
    483 	    in6p->in6p_cksum != -1) {
    484 		int off;
    485 		u_int16_t sum;
    486 
    487 		/* compute checksum */
    488 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
    489 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
    490 		else
    491 			off = in6p->in6p_cksum;
    492 		if (plen < off + 1) {
    493 			error = EINVAL;
    494 			goto bad;
    495 		}
    496 		off += sizeof(struct ip6_hdr);
    497 
    498 		sum = 0;
    499 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    500 		    M_DONTWAIT);
    501 		if (m == NULL) {
    502 			error = ENOBUFS;
    503 			goto bad;
    504 		}
    505 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
    506 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    507 		    M_DONTWAIT);
    508 		if (m == NULL) {
    509 			error = ENOBUFS;
    510 			goto bad;
    511 		}
    512 	}
    513 
    514 	{
    515 		struct ifnet *ret_oifp = NULL;
    516 
    517 		error = ip6_output(m, optp, &in6p->in6p_route, 0,
    518 		    in6p->in6p_moptions, so, &ret_oifp);
    519 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    520 			if (ret_oifp)
    521 				icmp6_ifoutstat_inc(ret_oifp, type, code);
    522 			ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
    523 		} else
    524 			RIP6_STATINC(RIP6_STAT_OPACKETS);
    525 	}
    526 
    527 	goto freectl;
    528 
    529  bad:
    530 	if (m)
    531 		m_freem(m);
    532 
    533  freectl:
    534 	if (control) {
    535 		ip6_clearpktopts(&opt, -1);
    536 		m_freem(control);
    537 	}
    538 	if_put(oifp, &psref);
    539 	curlwp_bindx(bound);
    540 	return error;
    541 }
    542 
    543 /*
    544  * Raw IPv6 socket option processing.
    545  */
    546 int
    547 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    548 {
    549 	int error = 0;
    550 
    551 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    552 		int optval;
    553 
    554 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
    555 		if (op == PRCO_GETOPT) {
    556 			optval = 1;
    557 			error = sockopt_set(sopt, &optval, sizeof(optval));
    558 		} else if (op == PRCO_SETOPT) {
    559 			error = sockopt_getint(sopt, &optval);
    560 			if (error)
    561 				goto out;
    562 			if (optval == 0)
    563 				error = EINVAL;
    564 		}
    565 
    566 		goto out;
    567 	} else if (sopt->sopt_level != IPPROTO_IPV6)
    568 		return ip6_ctloutput(op, so, sopt);
    569 
    570 	switch (sopt->sopt_name) {
    571 	case MRT6_INIT:
    572 	case MRT6_DONE:
    573 	case MRT6_ADD_MIF:
    574 	case MRT6_DEL_MIF:
    575 	case MRT6_ADD_MFC:
    576 	case MRT6_DEL_MFC:
    577 	case MRT6_PIM:
    578 		if (op == PRCO_SETOPT)
    579 			error = ip6_mrouter_set(so, sopt);
    580 		else if (op == PRCO_GETOPT)
    581 			error = ip6_mrouter_get(so, sopt);
    582 		else
    583 			error = EINVAL;
    584 		break;
    585 	case IPV6_CHECKSUM:
    586 		return ip6_raw_ctloutput(op, so, sopt);
    587 	default:
    588 		return ip6_ctloutput(op, so, sopt);
    589 	}
    590  out:
    591 	return error;
    592 }
    593 
    594 extern	u_long rip6_sendspace;
    595 extern	u_long rip6_recvspace;
    596 
    597 int
    598 rip6_attach(struct socket *so, int proto)
    599 {
    600 	struct in6pcb *in6p;
    601 	int s, error;
    602 
    603 	KASSERT(sotoin6pcb(so) == NULL);
    604 	sosetlock(so);
    605 
    606 	error = kauth_authorize_network(curlwp->l_cred,
    607 	    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
    608 	    KAUTH_ARG(AF_INET6),
    609 	    KAUTH_ARG(SOCK_RAW),
    610 	    KAUTH_ARG(so->so_proto->pr_protocol));
    611 	if (error) {
    612 		return error;
    613 	}
    614 	s = splsoftnet();
    615 	error = soreserve(so, rip6_sendspace, rip6_recvspace);
    616 	if (error) {
    617 		splx(s);
    618 		return error;
    619 	}
    620 	if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
    621 		splx(s);
    622 		return error;
    623 	}
    624 	splx(s);
    625 	in6p = sotoin6pcb(so);
    626 	in6p->in6p_ip6.ip6_nxt = proto;
    627 	in6p->in6p_cksum = -1;
    628 
    629 	in6p->in6p_icmp6filt = kmem_alloc(sizeof(struct icmp6_filter), KM_SLEEP);
    630 	if (in6p->in6p_icmp6filt == NULL) {
    631 		in6_pcbdetach(in6p);
    632 		return ENOMEM;
    633 	}
    634 	ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
    635 	KASSERT(solocked(so));
    636 	return error;
    637 }
    638 
    639 static void
    640 rip6_detach(struct socket *so)
    641 {
    642 	struct in6pcb *in6p = sotoin6pcb(so);
    643 
    644 	KASSERT(solocked(so));
    645 	KASSERT(in6p != NULL);
    646 
    647 	if (so == ip6_mrouter) {
    648 		ip6_mrouter_done();
    649 	}
    650 	/* xxx: RSVP */
    651 	if (in6p->in6p_icmp6filt != NULL) {
    652 		kmem_free(in6p->in6p_icmp6filt, sizeof(struct icmp6_filter));
    653 		in6p->in6p_icmp6filt = NULL;
    654 	}
    655 	in6_pcbdetach(in6p);
    656 }
    657 
    658 static int
    659 rip6_accept(struct socket *so, struct sockaddr *nam)
    660 {
    661 	KASSERT(solocked(so));
    662 
    663 	return EOPNOTSUPP;
    664 }
    665 
    666 static int
    667 rip6_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    668 {
    669 	struct in6pcb *in6p = sotoin6pcb(so);
    670 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    671 	struct ifaddr *ifa = NULL;
    672 	int error = 0;
    673 	int s;
    674 
    675 	KASSERT(solocked(so));
    676 	KASSERT(in6p != NULL);
    677 	KASSERT(nam != NULL);
    678 
    679 	if (addr->sin6_len != sizeof(*addr))
    680 		return EINVAL;
    681 	if (IFNET_READER_EMPTY() || addr->sin6_family != AF_INET6)
    682 		return EADDRNOTAVAIL;
    683 
    684 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    685 		return error;
    686 
    687 	/*
    688 	 * we don't support mapped address here, it would confuse
    689 	 * users so reject it
    690 	 */
    691 	if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr))
    692 		return EADDRNOTAVAIL;
    693 	s = pserialize_read_enter();
    694 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
    695 	    (ifa = ifa_ifwithaddr(sin6tosa(addr))) == NULL) {
    696 		error = EADDRNOTAVAIL;
    697 		goto out;
    698 	}
    699 	if (ifa && (ifatoia6(ifa))->ia6_flags &
    700 	    (IN6_IFF_ANYCAST | IN6_IFF_DUPLICATED)) {
    701 		error = EADDRNOTAVAIL;
    702 		goto out;
    703 	}
    704 
    705 	in6p->in6p_laddr = addr->sin6_addr;
    706 	error = 0;
    707 out:
    708 	pserialize_read_exit(s);
    709 	return error;
    710 }
    711 
    712 static int
    713 rip6_listen(struct socket *so, struct lwp *l)
    714 {
    715 	KASSERT(solocked(so));
    716 
    717 	return EOPNOTSUPP;
    718 }
    719 
    720 static int
    721 rip6_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    722 {
    723 	struct in6pcb *in6p = sotoin6pcb(so);
    724 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    725 	struct in6_addr in6a;
    726 	struct ifnet *ifp = NULL;
    727 	int scope_ambiguous = 0;
    728 	int error = 0;
    729 	struct psref psref;
    730 	int bound;
    731 
    732 	KASSERT(solocked(so));
    733 	KASSERT(in6p != NULL);
    734 	KASSERT(nam != NULL);
    735 
    736 	if (IFNET_READER_EMPTY())
    737 		return EADDRNOTAVAIL;
    738 	if (addr->sin6_family != AF_INET6)
    739 		return EAFNOSUPPORT;
    740 
    741 	/*
    742 	 * Application should provide a proper zone ID or the use of
    743 	 * default zone IDs should be enabled.  Unfortunately, some
    744 	 * applications do not behave as it should, so we need a
    745 	 * workaround.  Even if an appropriate ID is not determined,
    746 	 * we'll see if we can determine the outgoing interface.  If we
    747 	 * can, determine the zone ID based on the interface below.
    748 	 */
    749 	if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
    750 		scope_ambiguous = 1;
    751 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    752 		return error;
    753 
    754 	bound = curlwp_bind();
    755 	/* Source address selection. XXX: need pcblookup? */
    756 	error = in6_selectsrc(addr, in6p->in6p_outputopts,
    757 	    in6p->in6p_moptions, &in6p->in6p_route,
    758 	    &in6p->in6p_laddr, &ifp, &psref, &in6a);
    759 	if (error != 0)
    760 		goto out;
    761 	/* XXX: see above */
    762 	if (ifp && scope_ambiguous &&
    763 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
    764 		goto out;
    765 	}
    766 	in6p->in6p_laddr = in6a;
    767 	in6p->in6p_faddr = addr->sin6_addr;
    768 	soisconnected(so);
    769 out:
    770 	if_put(ifp, &psref);
    771 	curlwp_bindx(bound);
    772 	return error;
    773 }
    774 
    775 static int
    776 rip6_connect2(struct socket *so, struct socket *so2)
    777 {
    778 	KASSERT(solocked(so));
    779 
    780 	return EOPNOTSUPP;
    781 }
    782 
    783 static int
    784 rip6_disconnect(struct socket *so)
    785 {
    786 	struct in6pcb *in6p = sotoin6pcb(so);
    787 
    788 	KASSERT(solocked(so));
    789 	KASSERT(in6p != NULL);
    790 
    791 	if ((so->so_state & SS_ISCONNECTED) == 0)
    792 		return ENOTCONN;
    793 
    794 	in6p->in6p_faddr = in6addr_any;
    795 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
    796 	return 0;
    797 }
    798 
    799 static int
    800 rip6_shutdown(struct socket *so)
    801 {
    802 	KASSERT(solocked(so));
    803 
    804 	/*
    805 	 * Mark the connection as being incapable of futther input.
    806 	 */
    807 	socantsendmore(so);
    808 	return 0;
    809 }
    810 
    811 static int
    812 rip6_abort(struct socket *so)
    813 {
    814 	KASSERT(solocked(so));
    815 
    816 	soisdisconnected(so);
    817 	rip6_detach(so);
    818 	return 0;
    819 }
    820 
    821 static int
    822 rip6_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    823 {
    824 	return in6_control(so, cmd, nam, ifp);
    825 }
    826 
    827 static int
    828 rip6_stat(struct socket *so, struct stat *ub)
    829 {
    830 	KASSERT(solocked(so));
    831 
    832 	/* stat: don't bother with a blocksize */
    833 	return 0;
    834 }
    835 
    836 static int
    837 rip6_peeraddr(struct socket *so, struct sockaddr *nam)
    838 {
    839 	KASSERT(solocked(so));
    840 	KASSERT(sotoin6pcb(so) != NULL);
    841 	KASSERT(nam != NULL);
    842 
    843 	in6_setpeeraddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    844 	return 0;
    845 }
    846 
    847 static int
    848 rip6_sockaddr(struct socket *so, struct sockaddr *nam)
    849 {
    850 	KASSERT(solocked(so));
    851 	KASSERT(sotoin6pcb(so) != NULL);
    852 	KASSERT(nam != NULL);
    853 
    854 	in6_setsockaddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    855 	return 0;
    856 }
    857 
    858 static int
    859 rip6_rcvd(struct socket *so, int flags, struct lwp *l)
    860 {
    861 	KASSERT(solocked(so));
    862 
    863 	return EOPNOTSUPP;
    864 }
    865 
    866 static int
    867 rip6_recvoob(struct socket *so, struct mbuf *m, int flags)
    868 {
    869 	KASSERT(solocked(so));
    870 
    871 	return EOPNOTSUPP;
    872 }
    873 
    874 static int
    875 rip6_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    876     struct mbuf *control, struct lwp *l)
    877 {
    878 	struct in6pcb *in6p = sotoin6pcb(so);
    879 	struct sockaddr_in6 tmp;
    880 	struct sockaddr_in6 *dst;
    881 	int error = 0;
    882 
    883 	KASSERT(solocked(so));
    884 	KASSERT(in6p != NULL);
    885 	KASSERT(m != NULL);
    886 
    887 	/*
    888 	 * Ship a packet out. The appropriate raw output
    889 	 * routine handles any messaging necessary.
    890 	 */
    891 
    892 	/* always copy sockaddr to avoid overwrites */
    893 	if (so->so_state & SS_ISCONNECTED) {
    894 		if (nam) {
    895 			error = EISCONN;
    896 			goto release;
    897 		}
    898 		/* XXX */
    899 		sockaddr_in6_init(&tmp, &in6p->in6p_faddr, 0, 0, 0);
    900 		dst = &tmp;
    901 	} else {
    902 		if (nam == NULL) {
    903 			error = ENOTCONN;
    904 			goto release;
    905 		}
    906 		tmp = *(struct sockaddr_in6 *)nam;
    907 		dst = &tmp;
    908 
    909 		if (dst->sin6_family != AF_INET6) {
    910 			error = EAFNOSUPPORT;
    911 			goto release;
    912 		}
    913 	}
    914 	error = rip6_output(m, so, dst, control);
    915 	m = NULL;
    916 
    917 release:
    918 	if (m)
    919 		m_freem(m);
    920 
    921 	return error;
    922 }
    923 
    924 static int
    925 rip6_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    926 {
    927 	KASSERT(solocked(so));
    928 
    929 	if (m)
    930 	 	m_freem(m);
    931 
    932 	return EOPNOTSUPP;
    933 }
    934 
    935 static int
    936 rip6_purgeif(struct socket *so, struct ifnet *ifp)
    937 {
    938 
    939 #ifndef NET_MPSAFE
    940 	mutex_enter(softnet_lock);
    941 #endif
    942 	in6_pcbpurgeif0(&raw6cbtable, ifp);
    943 	in6_purgeif(ifp);
    944 	in6_pcbpurgeif(&raw6cbtable, ifp);
    945 #ifndef NET_MPSAFE
    946 	mutex_exit(softnet_lock);
    947 #endif
    948 
    949 	return 0;
    950 }
    951 
    952 static int
    953 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
    954 {
    955 
    956 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
    957 }
    958 
    959 static void
    960 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
    961 {
    962 
    963 	sysctl_createv(clog, 0, NULL, NULL,
    964 		       CTLFLAG_PERMANENT,
    965 		       CTLTYPE_NODE, "inet6", NULL,
    966 		       NULL, 0, NULL, 0,
    967 		       CTL_NET, PF_INET6, CTL_EOL);
    968 	sysctl_createv(clog, 0, NULL, NULL,
    969 		       CTLFLAG_PERMANENT,
    970 		       CTLTYPE_NODE, "raw6",
    971 		       SYSCTL_DESCR("Raw IPv6 settings"),
    972 		       NULL, 0, NULL, 0,
    973 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
    974 
    975 	sysctl_createv(clog, 0, NULL, NULL,
    976 		       CTLFLAG_PERMANENT,
    977 		       CTLTYPE_STRUCT, "pcblist",
    978 		       SYSCTL_DESCR("Raw IPv6 control block list"),
    979 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
    980 		       CTL_NET, PF_INET6, IPPROTO_RAW,
    981 		       CTL_CREATE, CTL_EOL);
    982 	sysctl_createv(clog, 0, NULL, NULL,
    983 		       CTLFLAG_PERMANENT,
    984 		       CTLTYPE_STRUCT, "stats",
    985 		       SYSCTL_DESCR("Raw IPv6 statistics"),
    986 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
    987 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
    988 		       CTL_EOL);
    989 }
    990 
    991 PR_WRAP_USRREQS(rip6)
    992 #define	rip6_attach		rip6_attach_wrapper
    993 #define	rip6_detach		rip6_detach_wrapper
    994 #define	rip6_accept		rip6_accept_wrapper
    995 #define	rip6_bind		rip6_bind_wrapper
    996 #define	rip6_listen		rip6_listen_wrapper
    997 #define	rip6_connect		rip6_connect_wrapper
    998 #define	rip6_connect2		rip6_connect2_wrapper
    999 #define	rip6_disconnect		rip6_disconnect_wrapper
   1000 #define	rip6_shutdown		rip6_shutdown_wrapper
   1001 #define	rip6_abort		rip6_abort_wrapper
   1002 #define	rip6_ioctl		rip6_ioctl_wrapper
   1003 #define	rip6_stat		rip6_stat_wrapper
   1004 #define	rip6_peeraddr		rip6_peeraddr_wrapper
   1005 #define	rip6_sockaddr		rip6_sockaddr_wrapper
   1006 #define	rip6_rcvd		rip6_rcvd_wrapper
   1007 #define	rip6_recvoob		rip6_recvoob_wrapper
   1008 #define	rip6_send		rip6_send_wrapper
   1009 #define	rip6_sendoob		rip6_sendoob_wrapper
   1010 #define	rip6_purgeif		rip6_purgeif_wrapper
   1011 
   1012 const struct pr_usrreqs rip6_usrreqs = {
   1013 	.pr_attach	= rip6_attach,
   1014 	.pr_detach	= rip6_detach,
   1015 	.pr_accept	= rip6_accept,
   1016 	.pr_bind	= rip6_bind,
   1017 	.pr_listen	= rip6_listen,
   1018 	.pr_connect	= rip6_connect,
   1019 	.pr_connect2	= rip6_connect2,
   1020 	.pr_disconnect	= rip6_disconnect,
   1021 	.pr_shutdown	= rip6_shutdown,
   1022 	.pr_abort	= rip6_abort,
   1023 	.pr_ioctl	= rip6_ioctl,
   1024 	.pr_stat	= rip6_stat,
   1025 	.pr_peeraddr	= rip6_peeraddr,
   1026 	.pr_sockaddr	= rip6_sockaddr,
   1027 	.pr_rcvd	= rip6_rcvd,
   1028 	.pr_recvoob	= rip6_recvoob,
   1029 	.pr_send	= rip6_send,
   1030 	.pr_sendoob	= rip6_sendoob,
   1031 	.pr_purgeif	= rip6_purgeif,
   1032 };
   1033