xref: /freebsd/sys/netinet6/raw_ip6.c (revision 7ef62cebc2f965b0f640263e179276928885e33d)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /*-
33  * Copyright (c) 1982, 1986, 1988, 1993
34  *	The Regents of the University of California.
35  * 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 __FBSDID("$FreeBSD$");
66 
67 #include "opt_ipsec.h"
68 #include "opt_inet6.h"
69 #include "opt_route.h"
70 
71 #include <sys/param.h>
72 #include <sys/errno.h>
73 #include <sys/jail.h>
74 #include <sys/kernel.h>
75 #include <sys/lock.h>
76 #include <sys/malloc.h>
77 #include <sys/mbuf.h>
78 #include <sys/priv.h>
79 #include <sys/proc.h>
80 #include <sys/protosw.h>
81 #include <sys/signalvar.h>
82 #include <sys/socket.h>
83 #include <sys/socketvar.h>
84 #include <sys/sx.h>
85 #include <sys/syslog.h>
86 
87 #include <net/if.h>
88 #include <net/if_var.h>
89 #include <net/if_private.h>
90 #include <net/if_types.h>
91 #include <net/route.h>
92 #include <net/vnet.h>
93 
94 #include <netinet/in.h>
95 #include <netinet/in_var.h>
96 #include <netinet/in_systm.h>
97 #include <netinet/in_pcb.h>
98 
99 #include <netinet/icmp6.h>
100 #include <netinet/ip6.h>
101 #include <netinet/ip_var.h>
102 #include <netinet6/ip6_mroute.h>
103 #include <netinet6/in6_pcb.h>
104 #include <netinet6/ip6_var.h>
105 #include <netinet6/nd6.h>
106 #include <netinet6/raw_ip6.h>
107 #include <netinet6/in6_fib.h>
108 #include <netinet6/scope6_var.h>
109 #include <netinet6/send.h>
110 
111 #include <netipsec/ipsec_support.h>
112 
113 #include <machine/stdarg.h>
114 
115 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
116 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
117 
118 /*
119  * Raw interface to IP6 protocol.
120  */
121 
122 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
123 #define	V_ripcbinfo			VNET(ripcbinfo)
124 
125 extern u_long	rip_sendspace;
126 extern u_long	rip_recvspace;
127 
128 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
129 VNET_PCPUSTAT_SYSINIT(rip6stat);
130 
131 #ifdef VIMAGE
132 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
133 #endif /* VIMAGE */
134 
135 /*
136  * Hooks for multicast routing. They all default to NULL, so leave them not
137  * initialized and rely on BSS being set to 0.
138  */
139 
140 /*
141  * The socket used to communicate with the multicast routing daemon.
142  */
143 VNET_DEFINE(struct socket *, ip6_mrouter);
144 
145 /*
146  * The various mrouter functions.
147  */
148 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
149 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
150 int (*ip6_mrouter_done)(void);
151 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
152 int (*mrt6_ioctl)(u_long, caddr_t);
153 
154 struct rip6_inp_match_ctx {
155 	struct ip6_hdr *ip6;
156 	int proto;
157 };
158 
159 static bool
160 rip6_inp_match(const struct inpcb *inp, void *v)
161 {
162 	struct rip6_inp_match_ctx *c = v;
163 	struct ip6_hdr *ip6 = c->ip6;
164 	int proto = c->proto;
165 
166 	/* XXX inp locking */
167 	if ((inp->inp_vflag & INP_IPV6) == 0)
168 		return (false);
169 	if (inp->inp_ip_p && inp->inp_ip_p != proto)
170 		return (false);
171 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
172 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
173 		return (false);
174 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
175 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src))
176 		return (false);
177 
178 	return (true);
179 }
180 
181 /*
182  * Setup generic address and protocol structures for raw_input routine, then
183  * pass them along with mbuf chain.
184  */
185 int
186 rip6_input(struct mbuf **mp, int *offp, int proto)
187 {
188 	struct ifnet *ifp;
189 	struct mbuf *n, *m = *mp;
190 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
191 	struct inpcb *inp;
192 	struct mbuf *opts = NULL;
193 	struct sockaddr_in6 fromsa;
194 	struct rip6_inp_match_ctx ctx = { .ip6 = ip6, .proto = proto };
195 	struct inpcb_iterator inpi = INP_ITERATOR(&V_ripcbinfo,
196 	    INPLOOKUP_RLOCKPCB, rip6_inp_match, &ctx);
197 	int delivered = 0;
198 
199 	NET_EPOCH_ASSERT();
200 
201 	RIP6STAT_INC(rip6s_ipackets);
202 
203 	init_sin6(&fromsa, m, 0); /* general init */
204 
205 	ifp = m->m_pkthdr.rcvif;
206 
207 	while ((inp = inp_next(&inpi)) != NULL) {
208 		INP_RLOCK_ASSERT(inp);
209 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
210 		/*
211 		 * Check AH/ESP integrity.
212 		 */
213 		if (IPSEC_ENABLED(ipv6) &&
214 		    IPSEC_CHECK_POLICY(ipv6, m, inp) != 0) {
215 			/* Do not inject data into pcb. */
216 			continue;
217 		}
218 #endif /* IPSEC */
219 		if (jailed_without_vnet(inp->inp_cred) &&
220 		    !IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
221 		    prison_check_ip6(inp->inp_cred, &ip6->ip6_dst) != 0)
222 			/*
223 			 * Allow raw socket in jail to receive multicast;
224 			 * assume process had PRIV_NETINET_RAW at attach,
225 			 * and fall through into normal filter path if so.
226 			 */
227 			continue;
228 		if (inp->in6p_cksum != -1) {
229 			RIP6STAT_INC(rip6s_isum);
230 			if (m->m_pkthdr.len - (*offp + inp->in6p_cksum) < 2 ||
231 			    in6_cksum(m, proto, *offp,
232 			    m->m_pkthdr.len - *offp)) {
233 				RIP6STAT_INC(rip6s_badsum);
234 				/*
235 				 * Drop the received message, don't send an
236 				 * ICMP6 message. Set proto to IPPROTO_NONE
237 				 * to achieve that.
238 				 */
239 				INP_RUNLOCK(inp);
240 				proto = IPPROTO_NONE;
241 				break;
242 			}
243 		}
244 		/*
245 		 * If this raw socket has multicast state, and we
246 		 * have received a multicast, check if this socket
247 		 * should receive it, as multicast filtering is now
248 		 * the responsibility of the transport layer.
249 		 */
250 		if (inp->in6p_moptions &&
251 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
252 			/*
253 			 * If the incoming datagram is for MLD, allow it
254 			 * through unconditionally to the raw socket.
255 			 *
256 			 * Use the M_RTALERT_MLD flag to check for MLD
257 			 * traffic without having to inspect the mbuf chain
258 			 * more deeply, as all MLDv1/v2 host messages MUST
259 			 * contain the Router Alert option.
260 			 *
261 			 * In the case of MLDv1, we may not have explicitly
262 			 * joined the group, and may have set IFF_ALLMULTI
263 			 * on the interface. im6o_mc_filter() may discard
264 			 * control traffic we actually need to see.
265 			 *
266 			 * Userland multicast routing daemons should continue
267 			 * filter the control traffic appropriately.
268 			 */
269 			int blocked;
270 
271 			blocked = MCAST_PASS;
272 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
273 				struct sockaddr_in6 mcaddr;
274 
275 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
276 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
277 				mcaddr.sin6_family = AF_INET6;
278 				mcaddr.sin6_addr = ip6->ip6_dst;
279 
280 				blocked = im6o_mc_filter(inp->in6p_moptions,
281 				    ifp,
282 				    (struct sockaddr *)&mcaddr,
283 				    (struct sockaddr *)&fromsa);
284 			}
285 			if (blocked != MCAST_PASS) {
286 				IP6STAT_INC(ip6s_notmember);
287 				continue;
288 			}
289 		}
290 		if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL)
291 			continue;
292 		if (inp->inp_flags & INP_CONTROLOPTS ||
293 		    inp->inp_socket->so_options & SO_TIMESTAMP)
294 			ip6_savecontrol(inp, n, &opts);
295 		/* strip intermediate headers */
296 		m_adj(n, *offp);
297 		if (sbappendaddr(&inp->inp_socket->so_rcv,
298 		    (struct sockaddr *)&fromsa, n, opts) == 0) {
299 			soroverflow(inp->inp_socket);
300 			m_freem(n);
301 			if (opts)
302 				m_freem(opts);
303 			RIP6STAT_INC(rip6s_fullsock);
304 		} else {
305 			sorwakeup(inp->inp_socket);
306 			delivered++;
307 		}
308 		opts = NULL;
309 	}
310 	if (delivered == 0) {
311 		RIP6STAT_INC(rip6s_nosock);
312 		if (m->m_flags & M_MCAST)
313 			RIP6STAT_INC(rip6s_nosockmcast);
314 		if (proto == IPPROTO_NONE)
315 			m_freem(m);
316 		else
317 			icmp6_error(m, ICMP6_PARAM_PROB,
318 			    ICMP6_PARAMPROB_NEXTHEADER,
319 			    ip6_get_prevhdr(m, *offp));
320 		IP6STAT_DEC(ip6s_delivered);
321 	} else
322 		m_freem(m);
323 	return (IPPROTO_DONE);
324 }
325 
326 void
327 rip6_ctlinput(struct ip6ctlparam *ip6cp)
328 {
329 	int errno;
330 
331 	if ((errno = icmp6_errmap(ip6cp->ip6c_icmp6)) != 0)
332 		in6_pcbnotify(&V_ripcbinfo, ip6cp->ip6c_finaldst, 0,
333 		    ip6cp->ip6c_src, 0, errno, ip6cp->ip6c_cmdarg,
334 		    in6_rtchange);
335 }
336 
337 /*
338  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
339  * may have setup with control call.
340  */
341 static int
342 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
343     struct mbuf *control, struct thread *td)
344 {
345 	struct epoch_tracker et;
346 	struct inpcb *inp;
347 	struct sockaddr_in6 tmp, *dstsock;
348 	struct m_tag *mtag;
349 	struct ip6_hdr *ip6;
350 	u_int	plen = m->m_pkthdr.len;
351 	struct ip6_pktopts opt, *optp;
352 	struct ifnet *oifp = NULL;
353 	int error;
354 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
355 	int scope_ambiguous = 0;
356 	int use_defzone = 0;
357 	int hlim = 0;
358 	struct in6_addr in6a;
359 
360 	inp = sotoinpcb(so);
361 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
362 
363 	/* Always copy sockaddr to avoid overwrites. */
364 	/* Unlocked read. */
365 	if (so->so_state & SS_ISCONNECTED) {
366 		if (nam) {
367 			error = EISCONN;
368 			goto release;
369 		}
370 		tmp = (struct sockaddr_in6 ){
371 			.sin6_family = AF_INET6,
372 			.sin6_len = sizeof(struct sockaddr_in6),
373 		};
374 		INP_RLOCK(inp);
375 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
376 		    sizeof(struct in6_addr));
377 		INP_RUNLOCK(inp);
378 		dstsock = &tmp;
379 	} else {
380 		if (nam == NULL)
381 			error = ENOTCONN;
382 		else if (nam->sa_family != AF_INET6)
383 			error = EAFNOSUPPORT;
384 		else if (nam->sa_len != sizeof(struct sockaddr_in6))
385 			error = EINVAL;
386 		else
387 			error = 0;
388 		if (error != 0)
389 			goto release;
390 		dstsock = (struct sockaddr_in6 *)nam;
391 		if (dstsock->sin6_family != AF_INET6) {
392 			error = EAFNOSUPPORT;
393 			goto release;
394 		}
395 	}
396 
397 	INP_WLOCK(inp);
398 
399 	if (control != NULL) {
400 		NET_EPOCH_ENTER(et);
401 		error = ip6_setpktopts(control, &opt, inp->in6p_outputopts,
402 		    so->so_cred, inp->inp_ip_p);
403 		NET_EPOCH_EXIT(et);
404 
405 		if (error != 0) {
406 			goto bad;
407 		}
408 		optp = &opt;
409 	} else
410 		optp = inp->in6p_outputopts;
411 
412 	/*
413 	 * Check and convert scope zone ID into internal form.
414 	 *
415 	 * XXX: we may still need to determine the zone later.
416 	 */
417 	if (!(so->so_state & SS_ISCONNECTED)) {
418 		if (!optp || !optp->ip6po_pktinfo ||
419 		    !optp->ip6po_pktinfo->ipi6_ifindex)
420 			use_defzone = V_ip6_use_defzone;
421 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
422 			scope_ambiguous = 1;
423 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
424 			goto bad;
425 	}
426 
427 	/*
428 	 * For an ICMPv6 packet, we should know its type and code to update
429 	 * statistics.
430 	 */
431 	if (inp->inp_ip_p == 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 	}
442 
443 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
444 	if (m == NULL) {
445 		error = ENOBUFS;
446 		goto bad;
447 	}
448 	ip6 = mtod(m, struct ip6_hdr *);
449 
450 #ifdef ROUTE_MPATH
451 	if (CALC_FLOWID_OUTBOUND) {
452 		uint32_t hash_type, hash_val;
453 
454 		hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
455 		    &dstsock->sin6_addr, 0, 0, inp->inp_ip_p, &hash_type);
456 		inp->inp_flowid = hash_val;
457 		inp->inp_flowtype = hash_type;
458 	}
459 #endif
460 	/*
461 	 * Source address selection.
462 	 */
463 	NET_EPOCH_ENTER(et);
464 	error = in6_selectsrc_socket(dstsock, optp, inp, so->so_cred,
465 	    scope_ambiguous, &in6a, &hlim);
466 	NET_EPOCH_EXIT(et);
467 
468 	if (error)
469 		goto bad;
470 	error = prison_check_ip6(inp->inp_cred, &in6a);
471 	if (error != 0)
472 		goto bad;
473 	ip6->ip6_src = in6a;
474 
475 	ip6->ip6_dst = dstsock->sin6_addr;
476 
477 	/*
478 	 * Fill in the rest of the IPv6 header fields.
479 	 */
480 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
481 	    (inp->inp_flow & IPV6_FLOWINFO_MASK);
482 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
483 	    (IPV6_VERSION & IPV6_VERSION_MASK);
484 
485 	/*
486 	 * ip6_plen will be filled in ip6_output, so not fill it here.
487 	 */
488 	ip6->ip6_nxt = inp->inp_ip_p;
489 	ip6->ip6_hlim = hlim;
490 
491 	if (inp->inp_ip_p == IPPROTO_ICMPV6 || inp->in6p_cksum != -1) {
492 		struct mbuf *n;
493 		int off;
494 		u_int16_t *p;
495 
496 		/* Compute checksum. */
497 		if (inp->inp_ip_p == IPPROTO_ICMPV6)
498 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
499 		else
500 			off = inp->in6p_cksum;
501 		if (plen < off + 2) {
502 			error = EINVAL;
503 			goto bad;
504 		}
505 		off += sizeof(struct ip6_hdr);
506 
507 		n = m;
508 		while (n && n->m_len <= off) {
509 			off -= n->m_len;
510 			n = n->m_next;
511 		}
512 		if (!n)
513 			goto bad;
514 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
515 		*p = 0;
516 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
517 	}
518 
519 	/*
520 	 * Send RA/RS messages to user land for protection, before sending
521 	 * them to rtadvd/rtsol.
522 	 */
523 	if ((send_sendso_input_hook != NULL) &&
524 	    inp->inp_ip_p == IPPROTO_ICMPV6) {
525 		switch (type) {
526 		case ND_ROUTER_ADVERT:
527 		case ND_ROUTER_SOLICIT:
528 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
529 				sizeof(unsigned short), M_NOWAIT);
530 			if (mtag == NULL)
531 				goto bad;
532 			m_tag_prepend(m, mtag);
533 		}
534 	}
535 
536 	NET_EPOCH_ENTER(et);
537 	error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
538 	NET_EPOCH_EXIT(et);
539 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
540 		if (oifp)
541 			icmp6_ifoutstat_inc(oifp, type, code);
542 		ICMP6STAT_INC(icp6s_outhist[type]);
543 	} else
544 		RIP6STAT_INC(rip6s_opackets);
545 
546 	goto freectl;
547 
548  bad:
549 	if (m)
550 		m_freem(m);
551 
552  freectl:
553 	if (control != NULL) {
554 		ip6_clearpktopts(&opt, -1);
555 		m_freem(control);
556 	}
557 	INP_WUNLOCK(inp);
558 	return (error);
559 
560 release:
561 	if (control != NULL)
562 		m_freem(control);
563 	m_freem(m);
564 	return (error);
565 }
566 
567 /*
568  * Raw IPv6 socket option processing.
569  */
570 int
571 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
572 {
573 	struct inpcb *inp = sotoinpcb(so);
574 	int error;
575 
576 	if (sopt->sopt_level == IPPROTO_ICMPV6)
577 		/*
578 		 * XXX: is it better to call icmp6_ctloutput() directly
579 		 * from protosw?
580 		 */
581 		return (icmp6_ctloutput(so, sopt));
582 	else if (sopt->sopt_level != IPPROTO_IPV6) {
583 		if (sopt->sopt_level == SOL_SOCKET &&
584 		    sopt->sopt_name == SO_SETFIB) {
585 			INP_WLOCK(inp);
586 			inp->inp_inc.inc_fibnum = so->so_fibnum;
587 			INP_WUNLOCK(inp);
588 			return (0);
589 		}
590 		return (EINVAL);
591 	}
592 
593 	error = 0;
594 
595 	switch (sopt->sopt_dir) {
596 	case SOPT_GET:
597 		switch (sopt->sopt_name) {
598 		case MRT6_INIT:
599 		case MRT6_DONE:
600 		case MRT6_ADD_MIF:
601 		case MRT6_DEL_MIF:
602 		case MRT6_ADD_MFC:
603 		case MRT6_DEL_MFC:
604 		case MRT6_PIM:
605 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
606 				return (EOPNOTSUPP);
607 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
608 			    EOPNOTSUPP;
609 			break;
610 		case IPV6_CHECKSUM:
611 			error = ip6_raw_ctloutput(so, sopt);
612 			break;
613 		default:
614 			error = ip6_ctloutput(so, sopt);
615 			break;
616 		}
617 		break;
618 
619 	case SOPT_SET:
620 		switch (sopt->sopt_name) {
621 		case MRT6_INIT:
622 		case MRT6_DONE:
623 		case MRT6_ADD_MIF:
624 		case MRT6_DEL_MIF:
625 		case MRT6_ADD_MFC:
626 		case MRT6_DEL_MFC:
627 		case MRT6_PIM:
628 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
629 				return (EOPNOTSUPP);
630 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
631 			    EOPNOTSUPP;
632 			break;
633 		case IPV6_CHECKSUM:
634 			error = ip6_raw_ctloutput(so, sopt);
635 			break;
636 		default:
637 			error = ip6_ctloutput(so, sopt);
638 			break;
639 		}
640 		break;
641 	}
642 
643 	return (error);
644 }
645 
646 static int
647 rip6_attach(struct socket *so, int proto, struct thread *td)
648 {
649 	struct inpcb *inp;
650 	struct icmp6_filter *filter;
651 	int error;
652 
653 	inp = sotoinpcb(so);
654 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
655 
656 	error = priv_check(td, PRIV_NETINET_RAW);
657 	if (error)
658 		return (error);
659 	if (proto >= IPPROTO_MAX || proto < 0)
660 		return (EPROTONOSUPPORT);
661 	error = soreserve(so, rip_sendspace, rip_recvspace);
662 	if (error)
663 		return (error);
664 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
665 	if (filter == NULL)
666 		return (ENOMEM);
667 	error = in_pcballoc(so, &V_ripcbinfo);
668 	if (error) {
669 		free(filter, M_PCB);
670 		return (error);
671 	}
672 	inp = (struct inpcb *)so->so_pcb;
673 	inp->inp_ip_p = proto;
674 	inp->in6p_cksum = -1;
675 	inp->in6p_icmp6filt = filter;
676 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
677 	INP_WUNLOCK(inp);
678 	return (0);
679 }
680 
681 static void
682 rip6_detach(struct socket *so)
683 {
684 	struct inpcb *inp;
685 
686 	inp = sotoinpcb(so);
687 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
688 
689 	if (so == V_ip6_mrouter && ip6_mrouter_done)
690 		ip6_mrouter_done();
691 	/* xxx: RSVP */
692 	INP_WLOCK(inp);
693 	free(inp->in6p_icmp6filt, M_PCB);
694 	in_pcbdetach(inp);
695 	in_pcbfree(inp);
696 }
697 
698 /* XXXRW: This can't ever be called. */
699 static void
700 rip6_abort(struct socket *so)
701 {
702 	struct inpcb *inp __diagused;
703 
704 	inp = sotoinpcb(so);
705 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
706 
707 	soisdisconnected(so);
708 }
709 
710 static void
711 rip6_close(struct socket *so)
712 {
713 	struct inpcb *inp __diagused;
714 
715 	inp = sotoinpcb(so);
716 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
717 
718 	soisdisconnected(so);
719 }
720 
721 static int
722 rip6_disconnect(struct socket *so)
723 {
724 	struct inpcb *inp;
725 
726 	inp = sotoinpcb(so);
727 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
728 
729 	if ((so->so_state & SS_ISCONNECTED) == 0)
730 		return (ENOTCONN);
731 	inp->in6p_faddr = in6addr_any;
732 	rip6_abort(so);
733 	return (0);
734 }
735 
736 static int
737 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
738 {
739 	struct epoch_tracker et;
740 	struct inpcb *inp;
741 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
742 	struct ifaddr *ifa = NULL;
743 	int error = 0;
744 
745 	inp = sotoinpcb(so);
746 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
747 
748 	if (nam->sa_family != AF_INET6)
749 		return (EAFNOSUPPORT);
750 	if (nam->sa_len != sizeof(*addr))
751 		return (EINVAL);
752 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
753 		return (error);
754 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
755 		return (EADDRNOTAVAIL);
756 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
757 		return (error);
758 
759 	NET_EPOCH_ENTER(et);
760 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
761 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
762 		NET_EPOCH_EXIT(et);
763 		return (EADDRNOTAVAIL);
764 	}
765 	if (ifa != NULL &&
766 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
767 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
768 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
769 		NET_EPOCH_EXIT(et);
770 		return (EADDRNOTAVAIL);
771 	}
772 	NET_EPOCH_EXIT(et);
773 	INP_WLOCK(inp);
774 	INP_INFO_WLOCK(&V_ripcbinfo);
775 	inp->in6p_laddr = addr->sin6_addr;
776 	INP_INFO_WUNLOCK(&V_ripcbinfo);
777 	INP_WUNLOCK(inp);
778 	return (0);
779 }
780 
781 static int
782 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
783 {
784 	struct inpcb *inp;
785 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
786 	struct in6_addr in6a;
787 	struct epoch_tracker et;
788 	int error = 0, scope_ambiguous = 0;
789 
790 	inp = sotoinpcb(so);
791 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
792 
793 	if (nam->sa_len != sizeof(*addr))
794 		return (EINVAL);
795 	if (CK_STAILQ_EMPTY(&V_ifnet))
796 		return (EADDRNOTAVAIL);
797 	if (addr->sin6_family != AF_INET6)
798 		return (EAFNOSUPPORT);
799 
800 	/*
801 	 * Application should provide a proper zone ID or the use of default
802 	 * zone IDs should be enabled.  Unfortunately, some applications do
803 	 * not behave as it should, so we need a workaround.  Even if an
804 	 * appropriate ID is not determined, we'll see if we can determine
805 	 * the outgoing interface.  If we can, determine the zone ID based on
806 	 * the interface below.
807 	 */
808 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
809 		scope_ambiguous = 1;
810 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
811 		return (error);
812 
813 	INP_WLOCK(inp);
814 	INP_INFO_WLOCK(&V_ripcbinfo);
815 	/* Source address selection. XXX: need pcblookup? */
816 	NET_EPOCH_ENTER(et);
817 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
818 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
819 	NET_EPOCH_EXIT(et);
820 	if (error) {
821 		INP_INFO_WUNLOCK(&V_ripcbinfo);
822 		INP_WUNLOCK(inp);
823 		return (error);
824 	}
825 
826 	inp->in6p_faddr = addr->sin6_addr;
827 	inp->in6p_laddr = in6a;
828 	soisconnected(so);
829 	INP_INFO_WUNLOCK(&V_ripcbinfo);
830 	INP_WUNLOCK(inp);
831 	return (0);
832 }
833 
834 static int
835 rip6_shutdown(struct socket *so)
836 {
837 	struct inpcb *inp;
838 
839 	inp = sotoinpcb(so);
840 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
841 
842 	INP_WLOCK(inp);
843 	socantsendmore(so);
844 	INP_WUNLOCK(inp);
845 	return (0);
846 }
847 
848 struct protosw rip6_protosw = {
849 	.pr_type =		SOCK_RAW,
850 	.pr_flags =		PR_ATOMIC|PR_ADDR,
851 	.pr_ctloutput =		rip6_ctloutput,
852 	.pr_abort =		rip6_abort,
853 	.pr_attach =		rip6_attach,
854 	.pr_bind =		rip6_bind,
855 	.pr_connect =		rip6_connect,
856 	.pr_control =		in6_control,
857 	.pr_detach =		rip6_detach,
858 	.pr_disconnect =	rip6_disconnect,
859 	.pr_peeraddr =		in6_getpeeraddr,
860 	.pr_send =		rip6_send,
861 	.pr_shutdown =		rip6_shutdown,
862 	.pr_sockaddr =		in6_getsockaddr,
863 	.pr_close =		rip6_close
864 };
865