Home | History | Annotate | Line # | Download | only in netinet6
raw_ip6.c revision 1.175.4.1
      1 /*	$NetBSD: raw_ip6.c,v 1.175.4.1 2023/03/23 12:06:49 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 1.175.4.1 2023/03/23 12:06:49 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/ipsec6.h>
    103 #endif
    104 
    105 #include "faith.h"
    106 #if defined(NFAITH) && 0 < NFAITH
    107 #include <net/if_faith.h>
    108 #endif
    109 
    110 extern struct inpcbtable rawcbtable;
    111 struct	inpcbtable raw6cbtable;
    112 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
    113 
    114 /*
    115  * Raw interface to IP6 protocol.
    116  */
    117 
    118 static percpu_t *rip6stat_percpu;
    119 
    120 #define	RIP6_STATINC(x)		_NET_STATINC(rip6stat_percpu, x)
    121 
    122 static void sysctl_net_inet6_raw6_setup(struct sysctllog **);
    123 
    124 /*
    125  * Initialize raw connection block queue.
    126  */
    127 void
    128 rip6_init(void)
    129 {
    130 
    131 	sysctl_net_inet6_raw6_setup(NULL);
    132 	in6_pcbinit(&raw6cbtable, 1, 1);
    133 
    134 	rip6stat_percpu = percpu_alloc(sizeof(uint64_t) * RIP6_NSTATS);
    135 }
    136 
    137 static void
    138 rip6_sbappendaddr(struct in6pcb *last, struct ip6_hdr *ip6,
    139     const struct sockaddr *sa, int hlen, struct mbuf *n)
    140 {
    141 	struct mbuf *opts = NULL;
    142 
    143 	if (last->in6p_flags & IN6P_CONTROLOPTS)
    144 		ip6_savecontrol(last, &opts, ip6, n);
    145 
    146 	m_adj(n, hlen);
    147 
    148 	if (sbappendaddr(&last->in6p_socket->so_rcv, sa, n, opts) == 0) {
    149 		soroverflow(last->in6p_socket);
    150 		m_freem(n);
    151 		if (opts)
    152 			m_freem(opts);
    153 		RIP6_STATINC(RIP6_STAT_FULLSOCK);
    154 	} else {
    155 		sorwakeup(last->in6p_socket);
    156 	}
    157 }
    158 
    159 /*
    160  * Setup generic address and protocol structures
    161  * for raw_input routine, then pass them along with
    162  * mbuf chain.
    163  */
    164 int
    165 rip6_input(struct mbuf **mp, int *offp, int proto)
    166 {
    167 	struct mbuf *m = *mp;
    168 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
    169 	struct inpcb_hdr *inph;
    170 	struct in6pcb *in6p;
    171 	struct in6pcb *last = NULL;
    172 	struct sockaddr_in6 rip6src;
    173 	struct mbuf *n;
    174 
    175 	RIP6_STATINC(RIP6_STAT_IPACKETS);
    176 
    177 #if defined(NFAITH) && 0 < NFAITH
    178 	if (faithprefix(&ip6->ip6_dst)) {
    179 		/* send icmp6 host unreach? */
    180 		m_freem(m);
    181 		return IPPROTO_DONE;
    182 	}
    183 #endif
    184 
    185 	sockaddr_in6_init(&rip6src, &ip6->ip6_src, 0, 0, 0);
    186 	if (sa6_recoverscope(&rip6src) != 0) {
    187 		/* XXX: should be impossible. */
    188 		m_freem(m);
    189 		return IPPROTO_DONE;
    190 	}
    191 
    192 	TAILQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
    193 		in6p = (struct in6pcb *)inph;
    194 		if (in6p->in6p_af != AF_INET6)
    195 			continue;
    196 		if (in6p->in6p_ip6.ip6_nxt &&
    197 		    in6p->in6p_ip6.ip6_nxt != proto)
    198 			continue;
    199 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
    200 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
    201 			continue;
    202 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
    203 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
    204 			continue;
    205 		if (in6p->in6p_cksum != -1) {
    206 			RIP6_STATINC(RIP6_STAT_ISUM);
    207 			/*
    208 			 * Although in6_cksum() does not need the position of
    209 			 * the checksum field for verification, enforce that it
    210 			 * is located within the packet.  Userland has given
    211 			 * a checksum offset, a packet too short for that is
    212 			 * invalid.  Avoid overflow with user supplied offset.
    213 			 */
    214 			if (m->m_pkthdr.len < *offp + 2 ||
    215 			    m->m_pkthdr.len - *offp - 2 < in6p->in6p_cksum ||
    216 			    in6_cksum(m, proto, *offp,
    217 			    m->m_pkthdr.len - *offp)) {
    218 				RIP6_STATINC(RIP6_STAT_BADSUM);
    219 				continue;
    220 			}
    221 		}
    222 
    223 		if (last == NULL) {
    224 			;
    225 		}
    226 #ifdef IPSEC
    227 		else if (ipsec_used && ipsec_in_reject(m, last)) {
    228 			/* do not inject data into pcb */
    229 		}
    230 #endif
    231 		else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) {
    232 			rip6_sbappendaddr(last, ip6, sin6tosa(&rip6src),
    233 			    *offp, n);
    234 		}
    235 
    236 		last = in6p;
    237 	}
    238 
    239 #ifdef IPSEC
    240 	if (ipsec_used && last && ipsec_in_reject(m, last)) {
    241 		m_freem(m);
    242 		IP6_STATDEC(IP6_STAT_DELIVERED);
    243 		/* do not inject data into pcb */
    244 	} else
    245 #endif
    246 	if (last != NULL) {
    247 		rip6_sbappendaddr(last, ip6, sin6tosa(&rip6src), *offp, m);
    248 	} else {
    249 		RIP6_STATINC(RIP6_STAT_NOSOCK);
    250 		if (m->m_flags & M_MCAST)
    251 			RIP6_STATINC(RIP6_STAT_NOSOCKMCAST);
    252 		if (proto == IPPROTO_NONE)
    253 			m_freem(m);
    254 		else {
    255 			int s;
    256 			struct ifnet *rcvif = m_get_rcvif(m, &s);
    257 			const int prvnxt = ip6_get_prevhdr(m, *offp);
    258 			in6_ifstat_inc(rcvif, ifs6_in_protounknown);
    259 			m_put_rcvif(rcvif, &s);
    260 			icmp6_error(m, ICMP6_PARAM_PROB,
    261 			    ICMP6_PARAMPROB_NEXTHEADER,
    262 			    prvnxt);
    263 		}
    264 		IP6_STATDEC(IP6_STAT_DELIVERED);
    265 	}
    266 	return IPPROTO_DONE;
    267 }
    268 
    269 void *
    270 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
    271 {
    272 	struct ip6_hdr *ip6;
    273 	struct ip6ctlparam *ip6cp = NULL;
    274 	const struct sockaddr_in6 *sa6_src = NULL;
    275 	void *cmdarg;
    276 	void (*notify)(struct in6pcb *, int) = in6_rtchange;
    277 	int nxt;
    278 
    279 	if (sa->sa_family != AF_INET6 ||
    280 	    sa->sa_len != sizeof(struct sockaddr_in6))
    281 		return NULL;
    282 
    283 	if ((unsigned)cmd >= PRC_NCMDS)
    284 		return NULL;
    285 	if (PRC_IS_REDIRECT(cmd))
    286 		notify = in6_rtchange, d = NULL;
    287 	else if (cmd == PRC_HOSTDEAD)
    288 		d = NULL;
    289 	else if (cmd == PRC_MSGSIZE)
    290 		; /* special code is present, see below */
    291 	else if (inet6ctlerrmap[cmd] == 0)
    292 		return NULL;
    293 
    294 	/* if the parameter is from icmp6, decode it. */
    295 	if (d != NULL) {
    296 		ip6cp = (struct ip6ctlparam *)d;
    297 		ip6 = ip6cp->ip6c_ip6;
    298 		cmdarg = ip6cp->ip6c_cmdarg;
    299 		sa6_src = ip6cp->ip6c_src;
    300 		nxt = ip6cp->ip6c_nxt;
    301 	} else {
    302 		ip6 = NULL;
    303 		cmdarg = NULL;
    304 		sa6_src = &sa6_any;
    305 		nxt = -1;
    306 	}
    307 
    308 	if (ip6 && cmd == PRC_MSGSIZE) {
    309 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
    310 		int valid = 0;
    311 		struct in6pcb *in6p;
    312 
    313 		/*
    314 		 * Check to see if we have a valid raw IPv6 socket
    315 		 * corresponding to the address in the ICMPv6 message
    316 		 * payload, and the protocol (ip6_nxt) meets the socket.
    317 		 * XXX chase extension headers, or pass final nxt value
    318 		 * from icmp6_notify_error()
    319 		 */
    320 		in6p = NULL;
    321 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
    322 					     (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0, 0);
    323 #if 0
    324 		if (!in6p) {
    325 			/*
    326 			 * As the use of sendto(2) is fairly popular,
    327 			 * we may want to allow non-connected pcb too.
    328 			 * But it could be too weak against attacks...
    329 			 * We should at least check if the local
    330 			 * address (= s) is really ours.
    331 			 */
    332 			in6p = in6_pcblookup_bind(&raw6cbtable,
    333 			    &sa6->sin6_addr, 0, 0);
    334 		}
    335 #endif
    336 
    337 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
    338 		    in6p->in6p_ip6.ip6_nxt == nxt)
    339 			valid++;
    340 
    341 		/*
    342 		 * Depending on the value of "valid" and routing table
    343 		 * size (mtudisc_{hi,lo}wat), we will:
    344 		 * - recalculate the new MTU and create the
    345 		 *   corresponding routing entry, or
    346 		 * - ignore the MTU change notification.
    347 		 */
    348 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
    349 
    350 		/*
    351 		 * regardless of if we called icmp6_mtudisc_update(),
    352 		 * we need to call in6_pcbnotify(), to notify path MTU
    353 		 * change to the userland (RFC3542), because some
    354 		 * unconnected sockets may share the same destination
    355 		 * and want to know the path MTU.
    356 		 */
    357 	}
    358 
    359 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
    360 	    sin6tocsa(sa6_src), 0, cmd, cmdarg, notify);
    361 	return NULL;
    362 }
    363 
    364 /*
    365  * Generate IPv6 header and pass packet to ip6_output.
    366  * Tack on options user may have setup with control call.
    367  */
    368 int
    369 rip6_output(struct mbuf *m, struct socket * const so,
    370     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
    371 {
    372 	struct in6_addr *dst;
    373 	struct ip6_hdr *ip6;
    374 	struct in6pcb *in6p;
    375 	u_int	plen = m->m_pkthdr.len;
    376 	int error = 0;
    377 	struct ip6_pktopts opt, *optp = NULL;
    378 	struct ifnet *oifp = NULL;
    379 	int type, code;		/* for ICMPv6 output statistics only */
    380 	int scope_ambiguous = 0;
    381 	int bound = curlwp_bind();
    382 	struct psref psref;
    383 
    384 	in6p = sotoin6pcb(so);
    385 
    386 	dst = &dstsock->sin6_addr;
    387 	if (control) {
    388 		if ((error = ip6_setpktopts(control, &opt,
    389 		    in6p->in6p_outputopts,
    390 		    kauth_cred_get(), so->so_proto->pr_protocol)) != 0) {
    391 			goto bad;
    392 		}
    393 		optp = &opt;
    394 	} else
    395 		optp = in6p->in6p_outputopts;
    396 
    397 	/*
    398 	 * Check and convert scope zone ID into internal form.
    399 	 * XXX: we may still need to determine the zone later.
    400 	 */
    401 	if (!(so->so_state & SS_ISCONNECTED)) {
    402 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
    403 			scope_ambiguous = 1;
    404 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
    405 			goto bad;
    406 	}
    407 
    408 	/*
    409 	 * For an ICMPv6 packet, we should know its type and code
    410 	 * to update statistics.
    411 	 */
    412 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    413 		struct icmp6_hdr *icmp6;
    414 		if (m->m_len < sizeof(struct icmp6_hdr) &&
    415 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
    416 			error = ENOBUFS;
    417 			goto bad;
    418 		}
    419 		icmp6 = mtod(m, struct icmp6_hdr *);
    420 		type = icmp6->icmp6_type;
    421 		code = icmp6->icmp6_code;
    422 	} else {
    423 		type = 0;
    424 		code = 0;
    425 	}
    426 
    427 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
    428 	if (!m) {
    429 		error = ENOBUFS;
    430 		goto bad;
    431 	}
    432 	ip6 = mtod(m, struct ip6_hdr *);
    433 
    434 	/*
    435 	 * Next header might not be ICMP6 but use its pseudo header anyway.
    436 	 */
    437 	ip6->ip6_dst = *dst;
    438 
    439 	/*
    440 	 * Source address selection.
    441 	 */
    442 	error = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
    443 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp, &psref, &ip6->ip6_src);
    444 	if (error != 0)
    445 		goto bad;
    446 
    447 	if (oifp && scope_ambiguous) {
    448 		/*
    449 		 * Application should provide a proper zone ID or the use of
    450 		 * default zone IDs should be enabled.  Unfortunately, some
    451 		 * applications do not behave as it should, so we need a
    452 		 * workaround.  Even if an appropriate ID is not determined
    453 		 * (when it's required), if we can determine the outgoing
    454 		 * interface. determine the zone ID based on the interface.
    455 		 */
    456 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
    457 		if (error != 0)
    458 			goto bad;
    459 	}
    460 	ip6->ip6_dst = dstsock->sin6_addr;
    461 
    462 	/* fill in the rest of the IPv6 header fields */
    463 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
    464 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
    465 	ip6->ip6_vfc  |= IPV6_VERSION;
    466 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
    467 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
    468 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
    469 
    470 	if_put(oifp, &psref);
    471 	oifp = NULL;
    472 
    473 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
    474 	    in6p->in6p_cksum != -1) {
    475 		const uint8_t nxt = ip6->ip6_nxt;
    476 		int off;
    477 		u_int16_t sum;
    478 
    479 		/* compute checksum */
    480 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
    481 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
    482 		else
    483 			off = in6p->in6p_cksum;
    484 		if (plen < 2 || plen - 2 < off) {
    485 			error = EINVAL;
    486 			goto bad;
    487 		}
    488 		off += sizeof(struct ip6_hdr);
    489 
    490 		sum = 0;
    491 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    492 		    M_DONTWAIT);
    493 		if (m == NULL) {
    494 			error = ENOBUFS;
    495 			goto bad;
    496 		}
    497 		sum = in6_cksum(m, nxt, sizeof(*ip6), plen);
    498 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
    499 		    M_DONTWAIT);
    500 		if (m == NULL) {
    501 			error = ENOBUFS;
    502 			goto bad;
    503 		}
    504 	}
    505 
    506 	{
    507 		struct ifnet *ret_oifp = NULL;
    508 
    509 		error = ip6_output(m, optp, &in6p->in6p_route, 0,
    510 		    in6p->in6p_moptions, in6p, &ret_oifp);
    511 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
    512 			if (ret_oifp)
    513 				icmp6_ifoutstat_inc(ret_oifp, type, code);
    514 			ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
    515 		} else
    516 			RIP6_STATINC(RIP6_STAT_OPACKETS);
    517 	}
    518 
    519 	goto freectl;
    520 
    521  bad:
    522 	if (m)
    523 		m_freem(m);
    524 
    525  freectl:
    526 	if (control) {
    527 		ip6_clearpktopts(&opt, -1);
    528 		m_freem(control);
    529 	}
    530 	if_put(oifp, &psref);
    531 	curlwp_bindx(bound);
    532 	return error;
    533 }
    534 
    535 /*
    536  * Raw IPv6 socket option processing.
    537  */
    538 int
    539 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    540 {
    541 	int error = 0;
    542 
    543 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    544 		int optval;
    545 
    546 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
    547 		if (op == PRCO_GETOPT) {
    548 			optval = 1;
    549 			error = sockopt_set(sopt, &optval, sizeof(optval));
    550 		} else if (op == PRCO_SETOPT) {
    551 			error = sockopt_getint(sopt, &optval);
    552 			if (error)
    553 				goto out;
    554 			if (optval == 0)
    555 				error = EINVAL;
    556 		}
    557 
    558 		goto out;
    559 	} else if (sopt->sopt_level != IPPROTO_IPV6)
    560 		return ip6_ctloutput(op, so, sopt);
    561 
    562 	switch (sopt->sopt_name) {
    563 	case MRT6_INIT:
    564 	case MRT6_DONE:
    565 	case MRT6_ADD_MIF:
    566 	case MRT6_DEL_MIF:
    567 	case MRT6_ADD_MFC:
    568 	case MRT6_DEL_MFC:
    569 	case MRT6_PIM:
    570 		if (op == PRCO_SETOPT)
    571 			error = ip6_mrouter_set(so, sopt);
    572 		else if (op == PRCO_GETOPT)
    573 			error = ip6_mrouter_get(so, sopt);
    574 		else
    575 			error = EINVAL;
    576 		break;
    577 	case IPV6_CHECKSUM:
    578 		return ip6_raw_ctloutput(op, so, sopt);
    579 	default:
    580 		return ip6_ctloutput(op, so, sopt);
    581 	}
    582  out:
    583 	return error;
    584 }
    585 
    586 extern	u_long rip6_sendspace;
    587 extern	u_long rip6_recvspace;
    588 
    589 int
    590 rip6_attach(struct socket *so, int proto)
    591 {
    592 	struct in6pcb *in6p;
    593 	int s, error;
    594 
    595 	KASSERT(sotoin6pcb(so) == NULL);
    596 	sosetlock(so);
    597 
    598 	error = kauth_authorize_network(curlwp->l_cred,
    599 	    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
    600 	    KAUTH_ARG(AF_INET6),
    601 	    KAUTH_ARG(SOCK_RAW),
    602 	    KAUTH_ARG(so->so_proto->pr_protocol));
    603 	if (error) {
    604 		return error;
    605 	}
    606 	s = splsoftnet();
    607 	error = soreserve(so, rip6_sendspace, rip6_recvspace);
    608 	if (error) {
    609 		splx(s);
    610 		return error;
    611 	}
    612 	if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
    613 		splx(s);
    614 		return error;
    615 	}
    616 	splx(s);
    617 	in6p = sotoin6pcb(so);
    618 	in6p->in6p_ip6.ip6_nxt = proto;
    619 	in6p->in6p_cksum = -1;
    620 
    621 	in6p->in6p_icmp6filt = kmem_alloc(sizeof(struct icmp6_filter), KM_SLEEP);
    622 	ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
    623 	KASSERT(solocked(so));
    624 	return error;
    625 }
    626 
    627 static void
    628 rip6_detach(struct socket *so)
    629 {
    630 	struct in6pcb *in6p = sotoin6pcb(so);
    631 
    632 	KASSERT(solocked(so));
    633 	KASSERT(in6p != NULL);
    634 
    635 	if (so == ip6_mrouter) {
    636 		ip6_mrouter_done();
    637 	}
    638 	/* xxx: RSVP */
    639 	if (in6p->in6p_icmp6filt != NULL) {
    640 		kmem_free(in6p->in6p_icmp6filt, sizeof(struct icmp6_filter));
    641 		in6p->in6p_icmp6filt = NULL;
    642 	}
    643 	in6_pcbdetach(in6p);
    644 }
    645 
    646 static int
    647 rip6_accept(struct socket *so, struct sockaddr *nam)
    648 {
    649 	KASSERT(solocked(so));
    650 
    651 	return EOPNOTSUPP;
    652 }
    653 
    654 static int
    655 rip6_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    656 {
    657 	struct in6pcb *in6p = sotoin6pcb(so);
    658 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    659 	struct ifaddr *ifa = NULL;
    660 	int error = 0;
    661 	int s;
    662 
    663 	KASSERT(solocked(so));
    664 	KASSERT(in6p != NULL);
    665 	KASSERT(nam != NULL);
    666 
    667 	if (addr->sin6_len != sizeof(*addr))
    668 		return EINVAL;
    669 	if (IFNET_READER_EMPTY() || addr->sin6_family != AF_INET6)
    670 		return EADDRNOTAVAIL;
    671 
    672 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    673 		return error;
    674 
    675 	/*
    676 	 * we don't support mapped address here, it would confuse
    677 	 * users so reject it
    678 	 */
    679 	if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr))
    680 		return EADDRNOTAVAIL;
    681 	s = pserialize_read_enter();
    682 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
    683 	    (ifa = ifa_ifwithaddr(sin6tosa(addr))) == NULL) {
    684 		error = EADDRNOTAVAIL;
    685 		goto out;
    686 	}
    687 	if (ifa && (ifatoia6(ifa))->ia6_flags &
    688 	    (IN6_IFF_ANYCAST | IN6_IFF_DUPLICATED)) {
    689 		error = EADDRNOTAVAIL;
    690 		goto out;
    691 	}
    692 
    693 	in6p->in6p_laddr = addr->sin6_addr;
    694 	error = 0;
    695 out:
    696 	pserialize_read_exit(s);
    697 	return error;
    698 }
    699 
    700 static int
    701 rip6_listen(struct socket *so, struct lwp *l)
    702 {
    703 	KASSERT(solocked(so));
    704 
    705 	return EOPNOTSUPP;
    706 }
    707 
    708 static int
    709 rip6_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    710 {
    711 	struct in6pcb *in6p = sotoin6pcb(so);
    712 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
    713 	struct in6_addr in6a;
    714 	struct ifnet *ifp = NULL;
    715 	int scope_ambiguous = 0;
    716 	int error = 0;
    717 	struct psref psref;
    718 	int bound;
    719 
    720 	KASSERT(solocked(so));
    721 	KASSERT(in6p != NULL);
    722 	KASSERT(nam != NULL);
    723 
    724 	if (IFNET_READER_EMPTY())
    725 		return EADDRNOTAVAIL;
    726 	if (addr->sin6_family != AF_INET6)
    727 		return EAFNOSUPPORT;
    728 	if (addr->sin6_len != sizeof(*addr))
    729 		return EINVAL;
    730 
    731 	/*
    732 	 * Application should provide a proper zone ID or the use of
    733 	 * default zone IDs should be enabled.  Unfortunately, some
    734 	 * applications do not behave as it should, so we need a
    735 	 * workaround.  Even if an appropriate ID is not determined,
    736 	 * we'll see if we can determine the outgoing interface.  If we
    737 	 * can, determine the zone ID based on the interface below.
    738 	 */
    739 	if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
    740 		scope_ambiguous = 1;
    741 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
    742 		return error;
    743 
    744 	bound = curlwp_bind();
    745 	/* Source address selection. XXX: need pcblookup? */
    746 	error = in6_selectsrc(addr, in6p->in6p_outputopts,
    747 	    in6p->in6p_moptions, &in6p->in6p_route,
    748 	    &in6p->in6p_laddr, &ifp, &psref, &in6a);
    749 	if (error != 0)
    750 		goto out;
    751 	/* XXX: see above */
    752 	if (ifp && scope_ambiguous &&
    753 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
    754 		goto out;
    755 	}
    756 	in6p->in6p_laddr = in6a;
    757 	in6p->in6p_faddr = addr->sin6_addr;
    758 	soisconnected(so);
    759 out:
    760 	if_put(ifp, &psref);
    761 	curlwp_bindx(bound);
    762 	return error;
    763 }
    764 
    765 static int
    766 rip6_connect2(struct socket *so, struct socket *so2)
    767 {
    768 	KASSERT(solocked(so));
    769 
    770 	return EOPNOTSUPP;
    771 }
    772 
    773 static int
    774 rip6_disconnect(struct socket *so)
    775 {
    776 	struct in6pcb *in6p = sotoin6pcb(so);
    777 
    778 	KASSERT(solocked(so));
    779 	KASSERT(in6p != NULL);
    780 
    781 	if ((so->so_state & SS_ISCONNECTED) == 0)
    782 		return ENOTCONN;
    783 
    784 	in6p->in6p_faddr = in6addr_any;
    785 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
    786 	return 0;
    787 }
    788 
    789 static int
    790 rip6_shutdown(struct socket *so)
    791 {
    792 	KASSERT(solocked(so));
    793 
    794 	/*
    795 	 * Mark the connection as being incapable of futther input.
    796 	 */
    797 	socantsendmore(so);
    798 	return 0;
    799 }
    800 
    801 static int
    802 rip6_abort(struct socket *so)
    803 {
    804 	KASSERT(solocked(so));
    805 
    806 	soisdisconnected(so);
    807 	rip6_detach(so);
    808 	return 0;
    809 }
    810 
    811 static int
    812 rip6_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    813 {
    814 	return in6_control(so, cmd, nam, ifp);
    815 }
    816 
    817 static int
    818 rip6_stat(struct socket *so, struct stat *ub)
    819 {
    820 	KASSERT(solocked(so));
    821 
    822 	/* stat: don't bother with a blocksize */
    823 	return 0;
    824 }
    825 
    826 static int
    827 rip6_peeraddr(struct socket *so, struct sockaddr *nam)
    828 {
    829 	KASSERT(solocked(so));
    830 	KASSERT(sotoin6pcb(so) != NULL);
    831 	KASSERT(nam != NULL);
    832 
    833 	in6_setpeeraddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    834 	return 0;
    835 }
    836 
    837 static int
    838 rip6_sockaddr(struct socket *so, struct sockaddr *nam)
    839 {
    840 	KASSERT(solocked(so));
    841 	KASSERT(sotoin6pcb(so) != NULL);
    842 	KASSERT(nam != NULL);
    843 
    844 	in6_setsockaddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
    845 	return 0;
    846 }
    847 
    848 static int
    849 rip6_rcvd(struct socket *so, int flags, struct lwp *l)
    850 {
    851 	KASSERT(solocked(so));
    852 
    853 	return EOPNOTSUPP;
    854 }
    855 
    856 static int
    857 rip6_recvoob(struct socket *so, struct mbuf *m, int flags)
    858 {
    859 	KASSERT(solocked(so));
    860 
    861 	return EOPNOTSUPP;
    862 }
    863 
    864 static int
    865 rip6_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    866     struct mbuf *control, struct lwp *l)
    867 {
    868 	struct in6pcb *in6p = sotoin6pcb(so);
    869 	struct sockaddr_in6 tmp;
    870 	struct sockaddr_in6 *dst;
    871 	int error = 0;
    872 
    873 	KASSERT(solocked(so));
    874 	KASSERT(in6p != NULL);
    875 	KASSERT(m != NULL);
    876 
    877 	/*
    878 	 * Ship a packet out. The appropriate raw output
    879 	 * routine handles any messaging necessary.
    880 	 */
    881 
    882 	/* always copy sockaddr to avoid overwrites */
    883 	if (so->so_state & SS_ISCONNECTED) {
    884 		if (nam) {
    885 			error = EISCONN;
    886 			goto release;
    887 		}
    888 		/* XXX */
    889 		sockaddr_in6_init(&tmp, &in6p->in6p_faddr, 0, 0, 0);
    890 		dst = &tmp;
    891 	} else {
    892 		if (nam == NULL) {
    893 			error = ENOTCONN;
    894 			goto release;
    895 		}
    896 		tmp = *(struct sockaddr_in6 *)nam;
    897 		dst = &tmp;
    898 
    899 		if (dst->sin6_family != AF_INET6) {
    900 			error = EAFNOSUPPORT;
    901 			goto release;
    902 		}
    903 		if (dst->sin6_len != sizeof(*dst)) {
    904 			error = EINVAL;
    905 			goto release;
    906 		}
    907 	}
    908 	error = rip6_output(m, so, dst, control);
    909 	m = NULL;
    910 
    911 release:
    912 	if (m)
    913 		m_freem(m);
    914 
    915 	return error;
    916 }
    917 
    918 static int
    919 rip6_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    920 {
    921 	KASSERT(solocked(so));
    922 
    923 	m_freem(m);
    924 	m_freem(control);
    925 
    926 	return EOPNOTSUPP;
    927 }
    928 
    929 static int
    930 rip6_purgeif(struct socket *so, struct ifnet *ifp)
    931 {
    932 
    933 	mutex_enter(softnet_lock);
    934 	in6_pcbpurgeif0(&raw6cbtable, ifp);
    935 #ifdef NET_MPSAFE
    936 	mutex_exit(softnet_lock);
    937 #endif
    938 	in6_purgeif(ifp);
    939 #ifdef NET_MPSAFE
    940 	mutex_enter(softnet_lock);
    941 #endif
    942 	in6_pcbpurgeif(&raw6cbtable, ifp);
    943 	mutex_exit(softnet_lock);
    944 
    945 	return 0;
    946 }
    947 
    948 static int
    949 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
    950 {
    951 
    952 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
    953 }
    954 
    955 static void
    956 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
    957 {
    958 
    959 	sysctl_createv(clog, 0, NULL, NULL,
    960 		       CTLFLAG_PERMANENT,
    961 		       CTLTYPE_NODE, "inet6", NULL,
    962 		       NULL, 0, NULL, 0,
    963 		       CTL_NET, PF_INET6, CTL_EOL);
    964 	sysctl_createv(clog, 0, NULL, NULL,
    965 		       CTLFLAG_PERMANENT,
    966 		       CTLTYPE_NODE, "raw6",
    967 		       SYSCTL_DESCR("Raw IPv6 settings"),
    968 		       NULL, 0, NULL, 0,
    969 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
    970 
    971 	sysctl_createv(clog, 0, NULL, NULL,
    972 		       CTLFLAG_PERMANENT,
    973 		       CTLTYPE_STRUCT, "pcblist",
    974 		       SYSCTL_DESCR("Raw IPv6 control block list"),
    975 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
    976 		       CTL_NET, PF_INET6, IPPROTO_RAW,
    977 		       CTL_CREATE, CTL_EOL);
    978 	sysctl_createv(clog, 0, NULL, NULL,
    979 		       CTLFLAG_PERMANENT,
    980 		       CTLTYPE_STRUCT, "stats",
    981 		       SYSCTL_DESCR("Raw IPv6 statistics"),
    982 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
    983 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
    984 		       CTL_EOL);
    985 }
    986 
    987 PR_WRAP_USRREQS(rip6)
    988 #define	rip6_attach		rip6_attach_wrapper
    989 #define	rip6_detach		rip6_detach_wrapper
    990 #define	rip6_accept		rip6_accept_wrapper
    991 #define	rip6_bind		rip6_bind_wrapper
    992 #define	rip6_listen		rip6_listen_wrapper
    993 #define	rip6_connect		rip6_connect_wrapper
    994 #define	rip6_connect2		rip6_connect2_wrapper
    995 #define	rip6_disconnect		rip6_disconnect_wrapper
    996 #define	rip6_shutdown		rip6_shutdown_wrapper
    997 #define	rip6_abort		rip6_abort_wrapper
    998 #define	rip6_ioctl		rip6_ioctl_wrapper
    999 #define	rip6_stat		rip6_stat_wrapper
   1000 #define	rip6_peeraddr		rip6_peeraddr_wrapper
   1001 #define	rip6_sockaddr		rip6_sockaddr_wrapper
   1002 #define	rip6_rcvd		rip6_rcvd_wrapper
   1003 #define	rip6_recvoob		rip6_recvoob_wrapper
   1004 #define	rip6_send		rip6_send_wrapper
   1005 #define	rip6_sendoob		rip6_sendoob_wrapper
   1006 #define	rip6_purgeif		rip6_purgeif_wrapper
   1007 
   1008 const struct pr_usrreqs rip6_usrreqs = {
   1009 	.pr_attach	= rip6_attach,
   1010 	.pr_detach	= rip6_detach,
   1011 	.pr_accept	= rip6_accept,
   1012 	.pr_bind	= rip6_bind,
   1013 	.pr_listen	= rip6_listen,
   1014 	.pr_connect	= rip6_connect,
   1015 	.pr_connect2	= rip6_connect2,
   1016 	.pr_disconnect	= rip6_disconnect,
   1017 	.pr_shutdown	= rip6_shutdown,
   1018 	.pr_abort	= rip6_abort,
   1019 	.pr_ioctl	= rip6_ioctl,
   1020 	.pr_stat	= rip6_stat,
   1021 	.pr_peeraddr	= rip6_peeraddr,
   1022 	.pr_sockaddr	= rip6_sockaddr,
   1023 	.pr_rcvd	= rip6_rcvd,
   1024 	.pr_recvoob	= rip6_recvoob,
   1025 	.pr_send	= rip6_send,
   1026 	.pr_sendoob	= rip6_sendoob,
   1027 	.pr_purgeif	= rip6_purgeif,
   1028 };
   1029