xref: /freebsd/sys/netinet6/raw_ip6.c (revision 6137b5f7b8c183ee8806d79b3f1d8e5e3ddb3df3)
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 
62 #include <sys/cdefs.h>
63 #include "opt_ipsec.h"
64 #include "opt_inet6.h"
65 #include "opt_route.h"
66 
67 #include <sys/param.h>
68 #include <sys/errno.h>
69 #include <sys/jail.h>
70 #include <sys/kernel.h>
71 #include <sys/lock.h>
72 #include <sys/malloc.h>
73 #include <sys/mbuf.h>
74 #include <sys/priv.h>
75 #include <sys/proc.h>
76 #include <sys/protosw.h>
77 #include <sys/signalvar.h>
78 #include <sys/socket.h>
79 #include <sys/socketvar.h>
80 #include <sys/sx.h>
81 #include <sys/syslog.h>
82 
83 #include <net/if.h>
84 #include <net/if_var.h>
85 #include <net/if_private.h>
86 #include <net/if_types.h>
87 #include <net/route.h>
88 #include <net/vnet.h>
89 
90 #include <netinet/in.h>
91 #include <netinet/in_var.h>
92 #include <netinet/in_systm.h>
93 #include <netinet/in_pcb.h>
94 
95 #include <netinet/icmp6.h>
96 #include <netinet/ip6.h>
97 #include <netinet/ip_var.h>
98 #include <netinet6/ip6_mroute.h>
99 #include <netinet6/in6_pcb.h>
100 #include <netinet6/ip6_var.h>
101 #include <netinet6/nd6.h>
102 #include <netinet6/raw_ip6.h>
103 #include <netinet6/in6_fib.h>
104 #include <netinet6/scope6_var.h>
105 #include <netinet6/send.h>
106 
107 #include <netipsec/ipsec_support.h>
108 
109 #include <machine/stdarg.h>
110 
111 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
112 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
113 
114 /*
115  * Raw interface to IP6 protocol.
116  */
117 
118 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
119 #define	V_ripcbinfo			VNET(ripcbinfo)
120 
121 extern u_long	rip_sendspace;
122 extern u_long	rip_recvspace;
123 
124 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
125 VNET_PCPUSTAT_SYSINIT(rip6stat);
126 
127 #ifdef VIMAGE
128 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
129 #endif /* VIMAGE */
130 
131 /*
132  * Hooks for multicast routing. They all default to NULL, so leave them not
133  * initialized and rely on BSS being set to 0.
134  */
135 
136 /*
137  * The socket used to communicate with the multicast routing daemon.
138  */
139 VNET_DEFINE(struct socket *, ip6_mrouter);
140 
141 /*
142  * The various mrouter functions.
143  */
144 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
145 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
146 int (*ip6_mrouter_done)(void);
147 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
148 int (*mrt6_ioctl)(u_long, caddr_t);
149 
150 struct rip6_inp_match_ctx {
151 	struct ip6_hdr *ip6;
152 	int proto;
153 };
154 
155 static bool
156 rip6_inp_match(const struct inpcb *inp, void *v)
157 {
158 	struct rip6_inp_match_ctx *c = v;
159 	struct ip6_hdr *ip6 = c->ip6;
160 	int proto = c->proto;
161 
162 	/* XXX inp locking */
163 	if ((inp->inp_vflag & INP_IPV6) == 0)
164 		return (false);
165 	if (inp->inp_ip_p && inp->inp_ip_p != proto)
166 		return (false);
167 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
168 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
169 		return (false);
170 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
171 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src))
172 		return (false);
173 
174 	return (true);
175 }
176 
177 /*
178  * Setup generic address and protocol structures for raw_input routine, then
179  * pass them along with mbuf chain.
180  */
181 int
182 rip6_input(struct mbuf **mp, int *offp, int proto)
183 {
184 	struct ifnet *ifp;
185 	struct mbuf *n, *m = *mp;
186 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
187 	struct inpcb *inp;
188 	struct mbuf *opts = NULL;
189 	struct sockaddr_in6 fromsa;
190 	struct rip6_inp_match_ctx ctx = { .ip6 = ip6, .proto = proto };
191 	struct inpcb_iterator inpi = INP_ITERATOR(&V_ripcbinfo,
192 	    INPLOOKUP_RLOCKPCB, rip6_inp_match, &ctx);
193 	int delivered = 0;
194 
195 	NET_EPOCH_ASSERT();
196 
197 	RIP6STAT_INC(rip6s_ipackets);
198 
199 	init_sin6(&fromsa, m, 0); /* general init */
200 
201 	ifp = m->m_pkthdr.rcvif;
202 
203 	while ((inp = inp_next(&inpi)) != NULL) {
204 		INP_RLOCK_ASSERT(inp);
205 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
206 		/*
207 		 * Check AH/ESP integrity.
208 		 */
209 		if (IPSEC_ENABLED(ipv6) &&
210 		    IPSEC_CHECK_POLICY(ipv6, m, inp) != 0) {
211 			/* Do not inject data into pcb. */
212 			continue;
213 		}
214 #endif /* IPSEC */
215 		if (jailed_without_vnet(inp->inp_cred) &&
216 		    !IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
217 		    prison_check_ip6(inp->inp_cred, &ip6->ip6_dst) != 0)
218 			/*
219 			 * Allow raw socket in jail to receive multicast;
220 			 * assume process had PRIV_NETINET_RAW at attach,
221 			 * and fall through into normal filter path if so.
222 			 */
223 			continue;
224 		if (inp->in6p_cksum != -1) {
225 			RIP6STAT_INC(rip6s_isum);
226 			if (m->m_pkthdr.len - (*offp + inp->in6p_cksum) < 2 ||
227 			    in6_cksum(m, proto, *offp,
228 			    m->m_pkthdr.len - *offp)) {
229 				RIP6STAT_INC(rip6s_badsum);
230 				/*
231 				 * Drop the received message, don't send an
232 				 * ICMP6 message. Set proto to IPPROTO_NONE
233 				 * to achieve that.
234 				 */
235 				INP_RUNLOCK(inp);
236 				proto = IPPROTO_NONE;
237 				break;
238 			}
239 		}
240 		/*
241 		 * If this raw socket has multicast state, and we
242 		 * have received a multicast, check if this socket
243 		 * should receive it, as multicast filtering is now
244 		 * the responsibility of the transport layer.
245 		 */
246 		if (inp->in6p_moptions &&
247 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
248 			/*
249 			 * If the incoming datagram is for MLD, allow it
250 			 * through unconditionally to the raw socket.
251 			 *
252 			 * Use the M_RTALERT_MLD flag to check for MLD
253 			 * traffic without having to inspect the mbuf chain
254 			 * more deeply, as all MLDv1/v2 host messages MUST
255 			 * contain the Router Alert option.
256 			 *
257 			 * In the case of MLDv1, we may not have explicitly
258 			 * joined the group, and may have set IFF_ALLMULTI
259 			 * on the interface. im6o_mc_filter() may discard
260 			 * control traffic we actually need to see.
261 			 *
262 			 * Userland multicast routing daemons should continue
263 			 * filter the control traffic appropriately.
264 			 */
265 			int blocked;
266 
267 			blocked = MCAST_PASS;
268 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
269 				struct sockaddr_in6 mcaddr;
270 
271 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
272 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
273 				mcaddr.sin6_family = AF_INET6;
274 				mcaddr.sin6_addr = ip6->ip6_dst;
275 
276 				blocked = im6o_mc_filter(inp->in6p_moptions,
277 				    ifp,
278 				    (struct sockaddr *)&mcaddr,
279 				    (struct sockaddr *)&fromsa);
280 			}
281 			if (blocked != MCAST_PASS) {
282 				IP6STAT_INC(ip6s_notmember);
283 				continue;
284 			}
285 		}
286 		if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL)
287 			continue;
288 		if (inp->inp_flags & INP_CONTROLOPTS ||
289 		    inp->inp_socket->so_options & SO_TIMESTAMP)
290 			ip6_savecontrol(inp, n, &opts);
291 		/* strip intermediate headers */
292 		m_adj(n, *offp);
293 		if (sbappendaddr(&inp->inp_socket->so_rcv,
294 		    (struct sockaddr *)&fromsa, n, opts) == 0) {
295 			soroverflow(inp->inp_socket);
296 			m_freem(n);
297 			if (opts)
298 				m_freem(opts);
299 			RIP6STAT_INC(rip6s_fullsock);
300 		} else {
301 			sorwakeup(inp->inp_socket);
302 			delivered++;
303 		}
304 		opts = NULL;
305 	}
306 	if (delivered == 0) {
307 		RIP6STAT_INC(rip6s_nosock);
308 		if (m->m_flags & M_MCAST)
309 			RIP6STAT_INC(rip6s_nosockmcast);
310 		if (proto == IPPROTO_NONE)
311 			m_freem(m);
312 		else
313 			icmp6_error(m, ICMP6_PARAM_PROB,
314 			    ICMP6_PARAMPROB_NEXTHEADER,
315 			    ip6_get_prevhdr(m, *offp));
316 		IP6STAT_DEC(ip6s_delivered);
317 	} else
318 		m_freem(m);
319 	return (IPPROTO_DONE);
320 }
321 
322 void
323 rip6_ctlinput(struct ip6ctlparam *ip6cp)
324 {
325 	int errno;
326 
327 	if ((errno = icmp6_errmap(ip6cp->ip6c_icmp6)) != 0)
328 		in6_pcbnotify(&V_ripcbinfo, ip6cp->ip6c_finaldst, 0,
329 		    ip6cp->ip6c_src, 0, errno, ip6cp->ip6c_cmdarg,
330 		    in6_rtchange);
331 }
332 
333 /*
334  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
335  * may have setup with control call.
336  */
337 static int
338 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
339     struct mbuf *control, struct thread *td)
340 {
341 	struct epoch_tracker et;
342 	struct inpcb *inp;
343 	struct sockaddr_in6 tmp, *dstsock;
344 	struct m_tag *mtag;
345 	struct ip6_hdr *ip6;
346 	u_int	plen = m->m_pkthdr.len;
347 	struct ip6_pktopts opt, *optp;
348 	struct ifnet *oifp = NULL;
349 	int error;
350 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
351 	int scope_ambiguous = 0;
352 	int use_defzone = 0;
353 	int hlim = 0;
354 	struct in6_addr in6a;
355 
356 	inp = sotoinpcb(so);
357 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
358 
359 	/* Always copy sockaddr to avoid overwrites. */
360 	/* Unlocked read. */
361 	if (so->so_state & SS_ISCONNECTED) {
362 		if (nam) {
363 			error = EISCONN;
364 			goto release;
365 		}
366 		tmp = (struct sockaddr_in6 ){
367 			.sin6_family = AF_INET6,
368 			.sin6_len = sizeof(struct sockaddr_in6),
369 		};
370 		INP_RLOCK(inp);
371 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
372 		    sizeof(struct in6_addr));
373 		INP_RUNLOCK(inp);
374 		dstsock = &tmp;
375 	} else {
376 		if (nam == NULL)
377 			error = ENOTCONN;
378 		else if (nam->sa_family != AF_INET6)
379 			error = EAFNOSUPPORT;
380 		else if (nam->sa_len != sizeof(struct sockaddr_in6))
381 			error = EINVAL;
382 		else
383 			error = 0;
384 		if (error != 0)
385 			goto release;
386 		dstsock = (struct sockaddr_in6 *)nam;
387 		if (dstsock->sin6_family != AF_INET6) {
388 			error = EAFNOSUPPORT;
389 			goto release;
390 		}
391 	}
392 
393 	INP_WLOCK(inp);
394 
395 	if (control != NULL) {
396 		NET_EPOCH_ENTER(et);
397 		error = ip6_setpktopts(control, &opt, inp->in6p_outputopts,
398 		    so->so_cred, inp->inp_ip_p);
399 		NET_EPOCH_EXIT(et);
400 
401 		if (error != 0) {
402 			goto bad;
403 		}
404 		optp = &opt;
405 	} else
406 		optp = inp->in6p_outputopts;
407 
408 	/*
409 	 * Check and convert scope zone ID into internal form.
410 	 *
411 	 * XXX: we may still need to determine the zone later.
412 	 */
413 	if (!(so->so_state & SS_ISCONNECTED)) {
414 		if (!optp || !optp->ip6po_pktinfo ||
415 		    !optp->ip6po_pktinfo->ipi6_ifindex)
416 			use_defzone = V_ip6_use_defzone;
417 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
418 			scope_ambiguous = 1;
419 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
420 			goto bad;
421 	}
422 
423 	/*
424 	 * For an ICMPv6 packet, we should know its type and code to update
425 	 * statistics.
426 	 */
427 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
428 		struct icmp6_hdr *icmp6;
429 		if (m->m_len < sizeof(struct icmp6_hdr) &&
430 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
431 			error = ENOBUFS;
432 			goto bad;
433 		}
434 		icmp6 = mtod(m, struct icmp6_hdr *);
435 		type = icmp6->icmp6_type;
436 		code = icmp6->icmp6_code;
437 	}
438 
439 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
440 	if (m == NULL) {
441 		error = ENOBUFS;
442 		goto bad;
443 	}
444 	ip6 = mtod(m, struct ip6_hdr *);
445 
446 #ifdef ROUTE_MPATH
447 	if (CALC_FLOWID_OUTBOUND) {
448 		uint32_t hash_type, hash_val;
449 
450 		hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
451 		    &dstsock->sin6_addr, 0, 0, inp->inp_ip_p, &hash_type);
452 		inp->inp_flowid = hash_val;
453 		inp->inp_flowtype = hash_type;
454 	}
455 #endif
456 	/*
457 	 * Source address selection.
458 	 */
459 	NET_EPOCH_ENTER(et);
460 	error = in6_selectsrc_socket(dstsock, optp, inp, so->so_cred,
461 	    scope_ambiguous, &in6a, &hlim);
462 	NET_EPOCH_EXIT(et);
463 
464 	if (error)
465 		goto bad;
466 	error = prison_check_ip6(inp->inp_cred, &in6a);
467 	if (error != 0)
468 		goto bad;
469 	ip6->ip6_src = in6a;
470 
471 	ip6->ip6_dst = dstsock->sin6_addr;
472 
473 	/*
474 	 * Fill in the rest of the IPv6 header fields.
475 	 */
476 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
477 	    (inp->inp_flow & IPV6_FLOWINFO_MASK);
478 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
479 	    (IPV6_VERSION & IPV6_VERSION_MASK);
480 
481 	/*
482 	 * ip6_plen will be filled in ip6_output, so not fill it here.
483 	 */
484 	ip6->ip6_nxt = inp->inp_ip_p;
485 	ip6->ip6_hlim = hlim;
486 
487 	if (inp->inp_ip_p == IPPROTO_ICMPV6 || inp->in6p_cksum != -1) {
488 		struct mbuf *n;
489 		int off;
490 		u_int16_t *p;
491 
492 		/* Compute checksum. */
493 		if (inp->inp_ip_p == IPPROTO_ICMPV6)
494 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
495 		else
496 			off = inp->in6p_cksum;
497 		if (plen < off + 2) {
498 			error = EINVAL;
499 			goto bad;
500 		}
501 		off += sizeof(struct ip6_hdr);
502 
503 		n = m;
504 		while (n && n->m_len <= off) {
505 			off -= n->m_len;
506 			n = n->m_next;
507 		}
508 		if (!n)
509 			goto bad;
510 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
511 		*p = 0;
512 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
513 	}
514 
515 	/*
516 	 * Send RA/RS messages to user land for protection, before sending
517 	 * them to rtadvd/rtsol.
518 	 */
519 	if ((send_sendso_input_hook != NULL) &&
520 	    inp->inp_ip_p == IPPROTO_ICMPV6) {
521 		switch (type) {
522 		case ND_ROUTER_ADVERT:
523 		case ND_ROUTER_SOLICIT:
524 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
525 				sizeof(unsigned short), M_NOWAIT);
526 			if (mtag == NULL)
527 				goto bad;
528 			m_tag_prepend(m, mtag);
529 		}
530 	}
531 
532 	NET_EPOCH_ENTER(et);
533 	error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
534 	NET_EPOCH_EXIT(et);
535 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
536 		if (oifp)
537 			icmp6_ifoutstat_inc(oifp, type, code);
538 		ICMP6STAT_INC(icp6s_outhist[type]);
539 	} else
540 		RIP6STAT_INC(rip6s_opackets);
541 
542 	goto freectl;
543 
544  bad:
545 	if (m)
546 		m_freem(m);
547 
548  freectl:
549 	if (control != NULL) {
550 		ip6_clearpktopts(&opt, -1);
551 		m_freem(control);
552 	}
553 	INP_WUNLOCK(inp);
554 	return (error);
555 
556 release:
557 	if (control != NULL)
558 		m_freem(control);
559 	m_freem(m);
560 	return (error);
561 }
562 
563 /*
564  * Raw IPv6 socket option processing.
565  */
566 int
567 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
568 {
569 	struct inpcb *inp = sotoinpcb(so);
570 	int error;
571 
572 	if (sopt->sopt_level == IPPROTO_ICMPV6)
573 		/*
574 		 * XXX: is it better to call icmp6_ctloutput() directly
575 		 * from protosw?
576 		 */
577 		return (icmp6_ctloutput(so, sopt));
578 	else if (sopt->sopt_level != IPPROTO_IPV6) {
579 		if (sopt->sopt_level == SOL_SOCKET &&
580 		    sopt->sopt_name == SO_SETFIB) {
581 			INP_WLOCK(inp);
582 			inp->inp_inc.inc_fibnum = so->so_fibnum;
583 			INP_WUNLOCK(inp);
584 			return (0);
585 		}
586 		return (EINVAL);
587 	}
588 
589 	error = 0;
590 
591 	switch (sopt->sopt_dir) {
592 	case SOPT_GET:
593 		switch (sopt->sopt_name) {
594 		case MRT6_INIT:
595 		case MRT6_DONE:
596 		case MRT6_ADD_MIF:
597 		case MRT6_DEL_MIF:
598 		case MRT6_ADD_MFC:
599 		case MRT6_DEL_MFC:
600 		case MRT6_PIM:
601 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
602 				return (EOPNOTSUPP);
603 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
604 			    EOPNOTSUPP;
605 			break;
606 		case IPV6_CHECKSUM:
607 			error = ip6_raw_ctloutput(so, sopt);
608 			break;
609 		default:
610 			error = ip6_ctloutput(so, sopt);
611 			break;
612 		}
613 		break;
614 
615 	case SOPT_SET:
616 		switch (sopt->sopt_name) {
617 		case MRT6_INIT:
618 		case MRT6_DONE:
619 		case MRT6_ADD_MIF:
620 		case MRT6_DEL_MIF:
621 		case MRT6_ADD_MFC:
622 		case MRT6_DEL_MFC:
623 		case MRT6_PIM:
624 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
625 				return (EOPNOTSUPP);
626 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
627 			    EOPNOTSUPP;
628 			break;
629 		case IPV6_CHECKSUM:
630 			error = ip6_raw_ctloutput(so, sopt);
631 			break;
632 		default:
633 			error = ip6_ctloutput(so, sopt);
634 			break;
635 		}
636 		break;
637 	}
638 
639 	return (error);
640 }
641 
642 static int
643 rip6_attach(struct socket *so, int proto, struct thread *td)
644 {
645 	struct inpcb *inp;
646 	struct icmp6_filter *filter;
647 	int error;
648 
649 	inp = sotoinpcb(so);
650 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
651 
652 	error = priv_check(td, PRIV_NETINET_RAW);
653 	if (error)
654 		return (error);
655 	if (proto >= IPPROTO_MAX || proto < 0)
656 		return (EPROTONOSUPPORT);
657 	error = soreserve(so, rip_sendspace, rip_recvspace);
658 	if (error)
659 		return (error);
660 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
661 	if (filter == NULL)
662 		return (ENOMEM);
663 	error = in_pcballoc(so, &V_ripcbinfo);
664 	if (error) {
665 		free(filter, M_PCB);
666 		return (error);
667 	}
668 	inp = (struct inpcb *)so->so_pcb;
669 	inp->inp_ip_p = proto;
670 	inp->in6p_cksum = -1;
671 	inp->in6p_icmp6filt = filter;
672 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
673 	INP_WUNLOCK(inp);
674 	return (0);
675 }
676 
677 static void
678 rip6_detach(struct socket *so)
679 {
680 	struct inpcb *inp;
681 
682 	inp = sotoinpcb(so);
683 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
684 
685 	if (so == V_ip6_mrouter && ip6_mrouter_done)
686 		ip6_mrouter_done();
687 	/* xxx: RSVP */
688 	INP_WLOCK(inp);
689 	free(inp->in6p_icmp6filt, M_PCB);
690 	in_pcbfree(inp);
691 }
692 
693 /* XXXRW: This can't ever be called. */
694 static void
695 rip6_abort(struct socket *so)
696 {
697 	struct inpcb *inp __diagused;
698 
699 	inp = sotoinpcb(so);
700 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
701 
702 	soisdisconnected(so);
703 }
704 
705 static void
706 rip6_close(struct socket *so)
707 {
708 	struct inpcb *inp __diagused;
709 
710 	inp = sotoinpcb(so);
711 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
712 
713 	soisdisconnected(so);
714 }
715 
716 static int
717 rip6_disconnect(struct socket *so)
718 {
719 	struct inpcb *inp;
720 
721 	inp = sotoinpcb(so);
722 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
723 
724 	if ((so->so_state & SS_ISCONNECTED) == 0)
725 		return (ENOTCONN);
726 	inp->in6p_faddr = in6addr_any;
727 	rip6_abort(so);
728 	return (0);
729 }
730 
731 static int
732 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
733 {
734 	struct epoch_tracker et;
735 	struct inpcb *inp;
736 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
737 	struct ifaddr *ifa = NULL;
738 	int error = 0;
739 
740 	inp = sotoinpcb(so);
741 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
742 
743 	if (nam->sa_family != AF_INET6)
744 		return (EAFNOSUPPORT);
745 	if (nam->sa_len != sizeof(*addr))
746 		return (EINVAL);
747 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
748 		return (error);
749 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
750 		return (EADDRNOTAVAIL);
751 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
752 		return (error);
753 
754 	NET_EPOCH_ENTER(et);
755 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
756 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
757 		NET_EPOCH_EXIT(et);
758 		return (EADDRNOTAVAIL);
759 	}
760 	if (ifa != NULL &&
761 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
762 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
763 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
764 		NET_EPOCH_EXIT(et);
765 		return (EADDRNOTAVAIL);
766 	}
767 	NET_EPOCH_EXIT(et);
768 	INP_WLOCK(inp);
769 	INP_INFO_WLOCK(&V_ripcbinfo);
770 	inp->in6p_laddr = addr->sin6_addr;
771 	INP_INFO_WUNLOCK(&V_ripcbinfo);
772 	INP_WUNLOCK(inp);
773 	return (0);
774 }
775 
776 static int
777 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
778 {
779 	struct inpcb *inp;
780 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
781 	struct in6_addr in6a;
782 	struct epoch_tracker et;
783 	int error = 0, scope_ambiguous = 0;
784 
785 	inp = sotoinpcb(so);
786 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
787 
788 	if (nam->sa_len != sizeof(*addr))
789 		return (EINVAL);
790 	if (CK_STAILQ_EMPTY(&V_ifnet))
791 		return (EADDRNOTAVAIL);
792 	if (addr->sin6_family != AF_INET6)
793 		return (EAFNOSUPPORT);
794 
795 	/*
796 	 * Application should provide a proper zone ID or the use of default
797 	 * zone IDs should be enabled.  Unfortunately, some applications do
798 	 * not behave as it should, so we need a workaround.  Even if an
799 	 * appropriate ID is not determined, we'll see if we can determine
800 	 * the outgoing interface.  If we can, determine the zone ID based on
801 	 * the interface below.
802 	 */
803 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
804 		scope_ambiguous = 1;
805 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
806 		return (error);
807 
808 	INP_WLOCK(inp);
809 	INP_INFO_WLOCK(&V_ripcbinfo);
810 	/* Source address selection. XXX: need pcblookup? */
811 	NET_EPOCH_ENTER(et);
812 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
813 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
814 	NET_EPOCH_EXIT(et);
815 	if (error) {
816 		INP_INFO_WUNLOCK(&V_ripcbinfo);
817 		INP_WUNLOCK(inp);
818 		return (error);
819 	}
820 
821 	inp->in6p_faddr = addr->sin6_addr;
822 	inp->in6p_laddr = in6a;
823 	soisconnected(so);
824 	INP_INFO_WUNLOCK(&V_ripcbinfo);
825 	INP_WUNLOCK(inp);
826 	return (0);
827 }
828 
829 static int
830 rip6_shutdown(struct socket *so)
831 {
832 	struct inpcb *inp;
833 
834 	inp = sotoinpcb(so);
835 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
836 
837 	INP_WLOCK(inp);
838 	socantsendmore(so);
839 	INP_WUNLOCK(inp);
840 	return (0);
841 }
842 
843 struct protosw rip6_protosw = {
844 	.pr_type =		SOCK_RAW,
845 	.pr_flags =		PR_ATOMIC|PR_ADDR,
846 	.pr_ctloutput =		rip6_ctloutput,
847 	.pr_abort =		rip6_abort,
848 	.pr_attach =		rip6_attach,
849 	.pr_bind =		rip6_bind,
850 	.pr_connect =		rip6_connect,
851 	.pr_control =		in6_control,
852 	.pr_detach =		rip6_detach,
853 	.pr_disconnect =	rip6_disconnect,
854 	.pr_peeraddr =		in6_getpeeraddr,
855 	.pr_send =		rip6_send,
856 	.pr_shutdown =		rip6_shutdown,
857 	.pr_sockaddr =		in6_getsockaddr,
858 	.pr_close =		rip6_close
859 };
860