Home | History | Annotate | Line # | Download | only in netinet6
raw_ip6.c revision 1.157.2.6
      1 /*	$NetBSD: raw_ip6.c,v 1.157.2.6 2023/03/23 12:08:39 martin 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 1.157.2.6 2023/03/23 12:08:39 martin 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/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 			/*
    196 			 * Although in6_cksum() does not need the position of
    197 			 * the checksum field for verification, enforce that it
    198 			 * is located within the packet.  Userland has given
    199 			 * a checksum offset, a packet too short for that is
    200 			 * invalid.  Avoid overflow with user supplied offset.
    201 			 */
    202 			if (m->m_pkthdr.len < *offp + 2 ||
    203 			    m->m_pkthdr.len - *offp - 2 < in6p->in6p_cksum ||
    204 			    in6_cksum(m, proto, *offp,
    205 			    m->m_pkthdr.len - *offp)) {
    206 				RIP6_STATINC(RIP6_STAT_BADSUM);
    207 				continue;
    208 			}
    209 		}
    210 		if (last) {
    211 			struct	mbuf *n;
    212 
    213 #ifdef IPSEC
    214 			/*
    215 			 * Check AH/ESP integrity
    216 			 */
    217 			if (!ipsec_used ||
    218 			    (ipsec_used && !ipsec6_in_reject(m, last)))
    219 #endif /* IPSEC */
    220 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
    221 				if (last->in6p_flags & IN6P_CONTROLOPTS)
    222 					ip6_savecontrol(last, &opts, ip6, n);
    223 				/* strip intermediate headers */
    224 				m_adj(n, *offp);
    225 				if (sbappendaddr(&last->in6p_socket->so_rcv,
    226 				    sin6tosa(&rip6src), n, opts) == 0) {
    227 					soroverflow(last->in6p_socket);
    228 					m_freem(n);
    229 					if (opts)
    230 						m_freem(opts);
    231 					RIP6_STATINC(RIP6_STAT_FULLSOCK);
    232 				} else
    233 					sorwakeup(last->in6p_socket);
    234 				opts = NULL;
    235 			}
    236 		}
    237 		last = in6p;
    238 	}
    239 #ifdef IPSEC
    240 	if (ipsec_used && last && ipsec6_in_reject(m, last)) {
    241 		m_freem(m);
    242 		/*
    243 		 * XXX ipsec6_in_reject update stat if there is an error
    244 		 * so we just need to update stats by hand in the case of last is
    245 		 * NULL
    246 		 */
    247 		if (!last)
    248 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
    249 			IP6_STATDEC(IP6_STAT_DELIVERED);
    250 			/* do not inject data into pcb */
    251 		} else
    252 #endif /* IPSEC */
    253 	if (last) {
    254 		if (last->in6p_flags & IN6P_CONTROLOPTS)
    255 			ip6_savecontrol(last, &opts, ip6, m);
    256 		/* strip intermediate headers */
    257 		m_adj(m, *offp);
    258 		if (sbappendaddr(&last->in6p_socket->so_rcv,
    259 		    sin6tosa(&rip6src), m, opts) == 0) {
    260 			soroverflow(last->in6p_socket);
    261 			m_freem(m);
    262 			if (opts)
    263 				m_freem(opts);
    264 			RIP6_STATINC(RIP6_STAT_FULLSOCK);
    265 		} else
    266 			sorwakeup(last->in6p_socket);
    267 	} else {
    268 		RIP6_STATINC(RIP6_STAT_NOSOCK);
    269 		if (m->m_flags & M_MCAST)
    270 			RIP6_STATINC(RIP6_STAT_NOSOCKMCAST);
    271 		if (proto == IPPROTO_NONE)
    272 			m_freem(m);
    273 		else {
    274 			int s;
    275 			struct ifnet *rcvif = m_get_rcvif(m, &s);
    276 			const int prvnxt = ip6_get_prevhdr(m, *offp);
    277 			in6_ifstat_inc(rcvif, ifs6_in_protounknown);
    278 			m_put_rcvif(rcvif, &s);
    279 			icmp6_error(m, ICMP6_PARAM_PROB,
    280 			    ICMP6_PARAMPROB_NEXTHEADER,
    281 			    prvnxt);
    282 		}
    283 		IP6_STATDEC(IP6_STAT_DELIVERED);
    284 	}
    285 	return IPPROTO_DONE;
    286 }
    287 
    288 void *
    289 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
    290 {
    291 	struct ip6_hdr *ip6;
    292 	struct ip6ctlparam *ip6cp = NULL;
    293 	const struct sockaddr_in6 *sa6_src = NULL;
    294 	void *cmdarg;
    295 	void (*notify)(struct in6pcb *, int) = in6_rtchange;
    296 	int nxt;
    297 
    298 	if (sa->sa_family != AF_INET6 ||
    299 	    sa->sa_len != sizeof(struct sockaddr_in6))
    300 		return NULL;
    301 
    302 	if ((unsigned)cmd >= PRC_NCMDS)
    303 		return NULL;
    304 	if (PRC_IS_REDIRECT(cmd))
    305 		notify = in6_rtchange, d = NULL;
    306 	else if (cmd == PRC_HOSTDEAD)
    307 		d = NULL;
    308 	else if (cmd == PRC_MSGSIZE)
    309 		; /* special code is present, see below */
    310 	else if (inet6ctlerrmap[cmd] == 0)
    311 		return NULL;
    312 
    313 	/* if the parameter is from icmp6, decode it. */
    314 	if (d != NULL) {
    315 		ip6cp = (struct ip6ctlparam *)d;
    316 		ip6 = ip6cp->ip6c_ip6;
    317 		cmdarg = ip6cp->ip6c_cmdarg;
    318 		sa6_src = ip6cp->ip6c_src;
    319 		nxt = ip6cp->ip6c_nxt;
    320 	} else {
    321 		ip6 = NULL;
    322 		cmdarg = NULL;
    323 		sa6_src = &sa6_any;
    324 		nxt = -1;
    325 	}
    326 
    327 	if (ip6 && cmd == PRC_MSGSIZE) {
    328 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
    329 		int valid = 0;
    330 		struct in6pcb *in6p;
    331 
    332 		/*
    333 		 * Check to see if we have a valid raw IPv6 socket
    334 		 * corresponding to the address in the ICMPv6 message
    335 		 * payload, and the protocol (ip6_nxt) meets the socket.
    336 		 * XXX chase extension headers, or pass final nxt value
    337 		 * from icmp6_notify_error()
    338 		 */
    339 		in6p = NULL;
    340 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
    341 					     (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0, 0);
    342 #if 0
    343 		if (!in6p) {
    344 			/*
    345 			 * As the use of sendto(2) is fairly popular,
    346 			 * we may want to allow non-connected pcb too.
    347 			 * But it could be too weak against attacks...
    348 			 * We should at least check if the local
    349 			 * address (= s) is really ours.
    350 			 */
    351 			in6p = in6_pcblookup_bind(&raw6cbtable,
    352 			    &sa6->sin6_addr, 0, 0);
    353 		}
    354 #endif
    355 
    356 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
    357 		    in6p->in6p_ip6.ip6_nxt == nxt)
    358 			valid++;
    359 
    360 		/*
    361 		 * Depending on the value of "valid" and routing table
    362 		 * size (mtudisc_{hi,lo}wat), we will:
    363 		 * - recalculate the new MTU and create the
    364 		 *   corresponding routing entry, or
    365 		 * - ignore the MTU change notification.
    366 		 */
    367 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
    368 
    369 		/*
    370 		 * regardless of if we called icmp6_mtudisc_update(),
    371 		 * we need to call in6_pcbnotify(), to notify path MTU
    372 		 * change to the userland (RFC3542), because some
    373 		 * unconnected sockets may share the same destination
    374 		 * and want to know the path MTU.
    375 		 */
    376 	}
    377 
    378 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
    379 	    sin6tocsa(sa6_src), 0, cmd, cmdarg, notify);
    380 	return NULL;
    381 }
    382 
    383 /*
    384  * Generate IPv6 header and pass packet to ip6_output.
    385  * Tack on options user may have setup with control call.
    386  */
    387 int
    388 rip6_output(struct mbuf *m, struct socket * const so,
    389     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
    390 {
    391 	struct in6_addr *dst;
    392 	struct ip6_hdr *ip6;
    393 	struct in6pcb *in6p;
    394 	u_int	plen = m->m_pkthdr.len;
    395 	int error = 0;
    396 	struct ip6_pktopts opt, *optp = NULL;
    397 	struct ifnet *oifp = NULL;
    398 	int type, code;		/* for ICMPv6 output statistics only */
    399 	int scope_ambiguous = 0;
    400 	int bound = curlwp_bind();
    401 	struct psref psref;
    402 
    403 	in6p = sotoin6pcb(so);
    404 
    405 	dst = &dstsock->sin6_addr;
    406 	if (control) {
    407 		if ((error = ip6_setpktopts(control, &opt,
    408 		    in6p->in6p_outputopts,
    409 		    kauth_cred_get(), so->so_proto->pr_protocol)) != 0) {
    410 			goto bad;
    411 		}
    412 		optp = &opt;
    413 	} else
    414 		optp = in6p->in6p_outputopts;
    415 
    416 	/*
    417 	 * Check and convert scope zone ID into internal form.
    418 	 * XXX: we may still need to determine the zone later.
    419 	 */
    420 	if (!(so->so_state & SS_ISCONNECTED)) {
    421 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
    422 			scope_ambiguous = 1;
    423 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
    424 			goto bad;
    425 	}
    426 
    427 	/*
    428 	 * For an ICMPv6 packet, we should know its type and code
    429 	 * to update statistics.
    430 	 */
    431 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    432 		struct icmp6_hdr *icmp6;
    433 		if (m->m_len < sizeof(struct icmp6_hdr) &&
    434 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
    435 			error = ENOBUFS;
    436 			goto bad;
    437 		}
    438 		icmp6 = mtod(m, struct icmp6_hdr *);
    439 		type = icmp6->icmp6_type;
    440 		code = icmp6->icmp6_code;
    441 	} else {
    442 		type = 0;
    443 		code = 0;
    444 	}
    445 
    446 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
    447 	if (!m) {
    448 		error = ENOBUFS;
    449 		goto bad;
    450 	}
    451 	ip6 = mtod(m, struct ip6_hdr *);
    452 
    453 	/*
    454 	 * Next header might not be ICMP6 but use its pseudo header anyway.
    455 	 */
    456 	ip6->ip6_dst = *dst;
    457 
    458 	/*
    459 	 * Source address selection.
    460 	 */
    461 	error = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
    462 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp, &psref, &ip6->ip6_src);
    463 	if (error != 0)
    464 		goto bad;
    465 
    466 	if (oifp && scope_ambiguous) {
    467 		/*
    468 		 * Application should provide a proper zone ID or the use of
    469 		 * default zone IDs should be enabled.  Unfortunately, some
    470 		 * applications do not behave as it should, so we need a
    471 		 * workaround.  Even if an appropriate ID is not determined
    472 		 * (when it's required), if we can determine the outgoing
    473 		 * interface. determine the zone ID based on the interface.
    474 		 */
    475 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
    476 		if (error != 0)
    477 			goto bad;
    478 	}
    479 	ip6->ip6_dst = dstsock->sin6_addr;
    480 
    481 	/* fill in the rest of the IPv6 header fields */
    482 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
    483 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
    484 	ip6->ip6_vfc  |= IPV6_VERSION;
    485 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
    486 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
    487 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
    488 
    489 	if_put(oifp, &psref);
    490 	oifp = NULL;
    491 
    492 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
    493 	    in6p->in6p_cksum != -1) {
    494 		const uint8_t nxt = ip6->ip6_nxt;
    495 		int off;
    496 		u_int16_t sum;
    497 
    498 		/* compute checksum */
    499 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
    500 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
    501 		else
    502 			off = in6p->in6p_cksum;
    503 		if (plen < 2 || plen - 2 < off) {
    504 			error = EINVAL;
    505 			goto bad;
    506 		}
    507 		off += sizeof(struct ip6_hdr);
    508 
    509 		sum = 0;
    510 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    511 		    M_DONTWAIT);
    512 		if (m == NULL) {
    513 			error = ENOBUFS;
    514 			goto bad;
    515 		}
    516 		sum = in6_cksum(m, nxt, sizeof(*ip6), plen);
    517 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    518 		    M_DONTWAIT);
    519 		if (m == NULL) {
    520 			error = ENOBUFS;
    521 			goto bad;
    522 		}
    523 	}
    524 
    525 	{
    526 		struct ifnet *ret_oifp = NULL;
    527 
    528 		error = ip6_output(m, optp, &in6p->in6p_route, 0,
    529 		    in6p->in6p_moptions, in6p, &ret_oifp);
    530 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    531 			if (ret_oifp)
    532 				icmp6_ifoutstat_inc(ret_oifp, type, code);
    533 			ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
    534 		} else
    535 			RIP6_STATINC(RIP6_STAT_OPACKETS);
    536 	}
    537 
    538 	goto freectl;
    539 
    540  bad:
    541 	if (m)
    542 		m_freem(m);
    543 
    544  freectl:
    545 	if (control) {
    546 		ip6_clearpktopts(&opt, -1);
    547 		m_freem(control);
    548 	}
    549 	if_put(oifp, &psref);
    550 	curlwp_bindx(bound);
    551 	return error;
    552 }
    553 
    554 /*
    555  * Raw IPv6 socket option processing.
    556  */
    557 int
    558 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    559 {
    560 	int error = 0;
    561 
    562 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    563 		int optval;
    564 
    565 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
    566 		if (op == PRCO_GETOPT) {
    567 			optval = 1;
    568 			error = sockopt_set(sopt, &optval, sizeof(optval));
    569 		} else if (op == PRCO_SETOPT) {
    570 			error = sockopt_getint(sopt, &optval);
    571 			if (error)
    572 				goto out;
    573 			if (optval == 0)
    574 				error = EINVAL;
    575 		}
    576 
    577 		goto out;
    578 	} else if (sopt->sopt_level != IPPROTO_IPV6)
    579 		return ip6_ctloutput(op, so, sopt);
    580 
    581 	switch (sopt->sopt_name) {
    582 	case MRT6_INIT:
    583 	case MRT6_DONE:
    584 	case MRT6_ADD_MIF:
    585 	case MRT6_DEL_MIF:
    586 	case MRT6_ADD_MFC:
    587 	case MRT6_DEL_MFC:
    588 	case MRT6_PIM:
    589 		if (op == PRCO_SETOPT)
    590 			error = ip6_mrouter_set(so, sopt);
    591 		else if (op == PRCO_GETOPT)
    592 			error = ip6_mrouter_get(so, sopt);
    593 		else
    594 			error = EINVAL;
    595 		break;
    596 	case IPV6_CHECKSUM:
    597 		return ip6_raw_ctloutput(op, so, sopt);
    598 	default:
    599 		return ip6_ctloutput(op, so, sopt);
    600 	}
    601  out:
    602 	return error;
    603 }
    604 
    605 extern	u_long rip6_sendspace;
    606 extern	u_long rip6_recvspace;
    607 
    608 int
    609 rip6_attach(struct socket *so, int proto)
    610 {
    611 	struct in6pcb *in6p;
    612 	int s, error;
    613 
    614 	KASSERT(sotoin6pcb(so) == NULL);
    615 	sosetlock(so);
    616 
    617 	error = kauth_authorize_network(curlwp->l_cred,
    618 	    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
    619 	    KAUTH_ARG(AF_INET6),
    620 	    KAUTH_ARG(SOCK_RAW),
    621 	    KAUTH_ARG(so->so_proto->pr_protocol));
    622 	if (error) {
    623 		return error;
    624 	}
    625 	s = splsoftnet();
    626 	error = soreserve(so, rip6_sendspace, rip6_recvspace);
    627 	if (error) {
    628 		splx(s);
    629 		return error;
    630 	}
    631 	if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
    632 		splx(s);
    633 		return error;
    634 	}
    635 	splx(s);
    636 	in6p = sotoin6pcb(so);
    637 	in6p->in6p_ip6.ip6_nxt = proto;
    638 	in6p->in6p_cksum = -1;
    639 
    640 	in6p->in6p_icmp6filt = kmem_alloc(sizeof(struct icmp6_filter), KM_SLEEP);
    641 	ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
    642 	KASSERT(solocked(so));
    643 	return error;
    644 }
    645 
    646 static void
    647 rip6_detach(struct socket *so)
    648 {
    649 	struct in6pcb *in6p = sotoin6pcb(so);
    650 
    651 	KASSERT(solocked(so));
    652 	KASSERT(in6p != NULL);
    653 
    654 	if (so == ip6_mrouter) {
    655 		ip6_mrouter_done();
    656 	}
    657 	/* xxx: RSVP */
    658 	if (in6p->in6p_icmp6filt != NULL) {
    659 		kmem_free(in6p->in6p_icmp6filt, sizeof(struct icmp6_filter));
    660 		in6p->in6p_icmp6filt = NULL;
    661 	}
    662 	in6_pcbdetach(in6p);
    663 }
    664 
    665 static int
    666 rip6_accept(struct socket *so, struct sockaddr *nam)
    667 {
    668 	KASSERT(solocked(so));
    669 
    670 	return EOPNOTSUPP;
    671 }
    672 
    673 static int
    674 rip6_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    675 {
    676 	struct in6pcb *in6p = sotoin6pcb(so);
    677 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    678 	struct ifaddr *ifa = NULL;
    679 	int error = 0;
    680 	int s;
    681 
    682 	KASSERT(solocked(so));
    683 	KASSERT(in6p != NULL);
    684 	KASSERT(nam != NULL);
    685 
    686 	if (addr->sin6_len != sizeof(*addr))
    687 		return EINVAL;
    688 	if (IFNET_READER_EMPTY() || addr->sin6_family != AF_INET6)
    689 		return EADDRNOTAVAIL;
    690 
    691 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    692 		return error;
    693 
    694 	/*
    695 	 * we don't support mapped address here, it would confuse
    696 	 * users so reject it
    697 	 */
    698 	if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr))
    699 		return EADDRNOTAVAIL;
    700 	s = pserialize_read_enter();
    701 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
    702 	    (ifa = ifa_ifwithaddr(sin6tosa(addr))) == NULL) {
    703 		error = EADDRNOTAVAIL;
    704 		goto out;
    705 	}
    706 	if (ifa && (ifatoia6(ifa))->ia6_flags &
    707 	    (IN6_IFF_ANYCAST | IN6_IFF_DUPLICATED)) {
    708 		error = EADDRNOTAVAIL;
    709 		goto out;
    710 	}
    711 
    712 	in6p->in6p_laddr = addr->sin6_addr;
    713 	error = 0;
    714 out:
    715 	pserialize_read_exit(s);
    716 	return error;
    717 }
    718 
    719 static int
    720 rip6_listen(struct socket *so, struct lwp *l)
    721 {
    722 	KASSERT(solocked(so));
    723 
    724 	return EOPNOTSUPP;
    725 }
    726 
    727 static int
    728 rip6_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    729 {
    730 	struct in6pcb *in6p = sotoin6pcb(so);
    731 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    732 	struct in6_addr in6a;
    733 	struct ifnet *ifp = NULL;
    734 	int scope_ambiguous = 0;
    735 	int error = 0;
    736 	struct psref psref;
    737 	int bound;
    738 
    739 	KASSERT(solocked(so));
    740 	KASSERT(in6p != NULL);
    741 	KASSERT(nam != NULL);
    742 
    743 	if (IFNET_READER_EMPTY())
    744 		return EADDRNOTAVAIL;
    745 	if (addr->sin6_family != AF_INET6)
    746 		return EAFNOSUPPORT;
    747 
    748 	/*
    749 	 * Application should provide a proper zone ID or the use of
    750 	 * default zone IDs should be enabled.  Unfortunately, some
    751 	 * applications do not behave as it should, so we need a
    752 	 * workaround.  Even if an appropriate ID is not determined,
    753 	 * we'll see if we can determine the outgoing interface.  If we
    754 	 * can, determine the zone ID based on the interface below.
    755 	 */
    756 	if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
    757 		scope_ambiguous = 1;
    758 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    759 		return error;
    760 
    761 	bound = curlwp_bind();
    762 	/* Source address selection. XXX: need pcblookup? */
    763 	error = in6_selectsrc(addr, in6p->in6p_outputopts,
    764 	    in6p->in6p_moptions, &in6p->in6p_route,
    765 	    &in6p->in6p_laddr, &ifp, &psref, &in6a);
    766 	if (error != 0)
    767 		goto out;
    768 	/* XXX: see above */
    769 	if (ifp && scope_ambiguous &&
    770 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
    771 		goto out;
    772 	}
    773 	in6p->in6p_laddr = in6a;
    774 	in6p->in6p_faddr = addr->sin6_addr;
    775 	soisconnected(so);
    776 out:
    777 	if_put(ifp, &psref);
    778 	curlwp_bindx(bound);
    779 	return error;
    780 }
    781 
    782 static int
    783 rip6_connect2(struct socket *so, struct socket *so2)
    784 {
    785 	KASSERT(solocked(so));
    786 
    787 	return EOPNOTSUPP;
    788 }
    789 
    790 static int
    791 rip6_disconnect(struct socket *so)
    792 {
    793 	struct in6pcb *in6p = sotoin6pcb(so);
    794 
    795 	KASSERT(solocked(so));
    796 	KASSERT(in6p != NULL);
    797 
    798 	if ((so->so_state & SS_ISCONNECTED) == 0)
    799 		return ENOTCONN;
    800 
    801 	in6p->in6p_faddr = in6addr_any;
    802 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
    803 	return 0;
    804 }
    805 
    806 static int
    807 rip6_shutdown(struct socket *so)
    808 {
    809 	KASSERT(solocked(so));
    810 
    811 	/*
    812 	 * Mark the connection as being incapable of futther input.
    813 	 */
    814 	socantsendmore(so);
    815 	return 0;
    816 }
    817 
    818 static int
    819 rip6_abort(struct socket *so)
    820 {
    821 	KASSERT(solocked(so));
    822 
    823 	soisdisconnected(so);
    824 	rip6_detach(so);
    825 	return 0;
    826 }
    827 
    828 static int
    829 rip6_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    830 {
    831 	return in6_control(so, cmd, nam, ifp);
    832 }
    833 
    834 static int
    835 rip6_stat(struct socket *so, struct stat *ub)
    836 {
    837 	KASSERT(solocked(so));
    838 
    839 	/* stat: don't bother with a blocksize */
    840 	return 0;
    841 }
    842 
    843 static int
    844 rip6_peeraddr(struct socket *so, struct sockaddr *nam)
    845 {
    846 	KASSERT(solocked(so));
    847 	KASSERT(sotoin6pcb(so) != NULL);
    848 	KASSERT(nam != NULL);
    849 
    850 	in6_setpeeraddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    851 	return 0;
    852 }
    853 
    854 static int
    855 rip6_sockaddr(struct socket *so, struct sockaddr *nam)
    856 {
    857 	KASSERT(solocked(so));
    858 	KASSERT(sotoin6pcb(so) != NULL);
    859 	KASSERT(nam != NULL);
    860 
    861 	in6_setsockaddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    862 	return 0;
    863 }
    864 
    865 static int
    866 rip6_rcvd(struct socket *so, int flags, struct lwp *l)
    867 {
    868 	KASSERT(solocked(so));
    869 
    870 	return EOPNOTSUPP;
    871 }
    872 
    873 static int
    874 rip6_recvoob(struct socket *so, struct mbuf *m, int flags)
    875 {
    876 	KASSERT(solocked(so));
    877 
    878 	return EOPNOTSUPP;
    879 }
    880 
    881 static int
    882 rip6_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    883     struct mbuf *control, struct lwp *l)
    884 {
    885 	struct in6pcb *in6p = sotoin6pcb(so);
    886 	struct sockaddr_in6 tmp;
    887 	struct sockaddr_in6 *dst;
    888 	int error = 0;
    889 
    890 	KASSERT(solocked(so));
    891 	KASSERT(in6p != NULL);
    892 	KASSERT(m != NULL);
    893 
    894 	/*
    895 	 * Ship a packet out. The appropriate raw output
    896 	 * routine handles any messaging necessary.
    897 	 */
    898 
    899 	/* always copy sockaddr to avoid overwrites */
    900 	if (so->so_state & SS_ISCONNECTED) {
    901 		if (nam) {
    902 			error = EISCONN;
    903 			goto release;
    904 		}
    905 		/* XXX */
    906 		sockaddr_in6_init(&tmp, &in6p->in6p_faddr, 0, 0, 0);
    907 		dst = &tmp;
    908 	} else {
    909 		if (nam == NULL) {
    910 			error = ENOTCONN;
    911 			goto release;
    912 		}
    913 		tmp = *(struct sockaddr_in6 *)nam;
    914 		dst = &tmp;
    915 
    916 		if (dst->sin6_family != AF_INET6) {
    917 			error = EAFNOSUPPORT;
    918 			goto release;
    919 		}
    920 	}
    921 	error = rip6_output(m, so, dst, control);
    922 	m = NULL;
    923 
    924 release:
    925 	if (m)
    926 		m_freem(m);
    927 
    928 	return error;
    929 }
    930 
    931 static int
    932 rip6_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    933 {
    934 	KASSERT(solocked(so));
    935 
    936 	m_freem(m);
    937 	m_freem(control);
    938 
    939 	return EOPNOTSUPP;
    940 }
    941 
    942 static int
    943 rip6_purgeif(struct socket *so, struct ifnet *ifp)
    944 {
    945 
    946 	mutex_enter(softnet_lock);
    947 	in6_pcbpurgeif0(&raw6cbtable, ifp);
    948 #ifdef NET_MPSAFE
    949 	mutex_exit(softnet_lock);
    950 #endif
    951 	in6_purgeif(ifp);
    952 #ifdef NET_MPSAFE
    953 	mutex_enter(softnet_lock);
    954 #endif
    955 	in6_pcbpurgeif(&raw6cbtable, ifp);
    956 	mutex_exit(softnet_lock);
    957 
    958 	return 0;
    959 }
    960 
    961 static int
    962 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
    963 {
    964 
    965 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
    966 }
    967 
    968 static void
    969 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
    970 {
    971 
    972 	sysctl_createv(clog, 0, NULL, NULL,
    973 		       CTLFLAG_PERMANENT,
    974 		       CTLTYPE_NODE, "inet6", NULL,
    975 		       NULL, 0, NULL, 0,
    976 		       CTL_NET, PF_INET6, CTL_EOL);
    977 	sysctl_createv(clog, 0, NULL, NULL,
    978 		       CTLFLAG_PERMANENT,
    979 		       CTLTYPE_NODE, "raw6",
    980 		       SYSCTL_DESCR("Raw IPv6 settings"),
    981 		       NULL, 0, NULL, 0,
    982 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
    983 
    984 	sysctl_createv(clog, 0, NULL, NULL,
    985 		       CTLFLAG_PERMANENT,
    986 		       CTLTYPE_STRUCT, "pcblist",
    987 		       SYSCTL_DESCR("Raw IPv6 control block list"),
    988 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
    989 		       CTL_NET, PF_INET6, IPPROTO_RAW,
    990 		       CTL_CREATE, CTL_EOL);
    991 	sysctl_createv(clog, 0, NULL, NULL,
    992 		       CTLFLAG_PERMANENT,
    993 		       CTLTYPE_STRUCT, "stats",
    994 		       SYSCTL_DESCR("Raw IPv6 statistics"),
    995 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
    996 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
    997 		       CTL_EOL);
    998 }
    999 
   1000 PR_WRAP_USRREQS(rip6)
   1001 #define	rip6_attach		rip6_attach_wrapper
   1002 #define	rip6_detach		rip6_detach_wrapper
   1003 #define	rip6_accept		rip6_accept_wrapper
   1004 #define	rip6_bind		rip6_bind_wrapper
   1005 #define	rip6_listen		rip6_listen_wrapper
   1006 #define	rip6_connect		rip6_connect_wrapper
   1007 #define	rip6_connect2		rip6_connect2_wrapper
   1008 #define	rip6_disconnect		rip6_disconnect_wrapper
   1009 #define	rip6_shutdown		rip6_shutdown_wrapper
   1010 #define	rip6_abort		rip6_abort_wrapper
   1011 #define	rip6_ioctl		rip6_ioctl_wrapper
   1012 #define	rip6_stat		rip6_stat_wrapper
   1013 #define	rip6_peeraddr		rip6_peeraddr_wrapper
   1014 #define	rip6_sockaddr		rip6_sockaddr_wrapper
   1015 #define	rip6_rcvd		rip6_rcvd_wrapper
   1016 #define	rip6_recvoob		rip6_recvoob_wrapper
   1017 #define	rip6_send		rip6_send_wrapper
   1018 #define	rip6_sendoob		rip6_sendoob_wrapper
   1019 #define	rip6_purgeif		rip6_purgeif_wrapper
   1020 
   1021 const struct pr_usrreqs rip6_usrreqs = {
   1022 	.pr_attach	= rip6_attach,
   1023 	.pr_detach	= rip6_detach,
   1024 	.pr_accept	= rip6_accept,
   1025 	.pr_bind	= rip6_bind,
   1026 	.pr_listen	= rip6_listen,
   1027 	.pr_connect	= rip6_connect,
   1028 	.pr_connect2	= rip6_connect2,
   1029 	.pr_disconnect	= rip6_disconnect,
   1030 	.pr_shutdown	= rip6_shutdown,
   1031 	.pr_abort	= rip6_abort,
   1032 	.pr_ioctl	= rip6_ioctl,
   1033 	.pr_stat	= rip6_stat,
   1034 	.pr_peeraddr	= rip6_peeraddr,
   1035 	.pr_sockaddr	= rip6_sockaddr,
   1036 	.pr_rcvd	= rip6_rcvd,
   1037 	.pr_recvoob	= rip6_recvoob,
   1038 	.pr_send	= rip6_send,
   1039 	.pr_sendoob	= rip6_sendoob,
   1040 	.pr_purgeif	= rip6_purgeif,
   1041 };
   1042