xref: /freebsd/sys/netinet6/raw_ip6.c (revision 49dae58b287906be26f56ba3e3dc693c3ba8cf37)
1 /*-
2  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. Neither the name of the project nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  */
29 
30 /*-
31  * Copyright (c) 1982, 1986, 1988, 1993
32  *	The Regents of the University of California.
33  * All rights reserved.
34  *
35  * Redistribution and use in source and binary forms, with or without
36  * modification, are permitted provided that the following conditions
37  * are met:
38  * 1. Redistributions of source code must retain the above copyright
39  *    notice, this list of conditions and the following disclaimer.
40  * 2. Redistributions in binary form must reproduce the above copyright
41  *    notice, this list of conditions and the following disclaimer in the
42  *    documentation and/or other materials provided with the distribution.
43  * 4. Neither the name of the University nor the names of its contributors
44  *    may be used to endorse or promote products derived from this software
45  *    without specific prior written permission.
46  *
47  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
48  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
49  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
50  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
51  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
52  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
53  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
54  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
55  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
56  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
57  * SUCH DAMAGE.
58  *
59  *	@(#)raw_ip.c	8.2 (Berkeley) 1/4/94
60  */
61 
62 #include <sys/cdefs.h>
63 __FBSDID("$FreeBSD$");
64 
65 #include "opt_ipsec.h"
66 #include "opt_inet6.h"
67 
68 #include <sys/param.h>
69 #include <sys/errno.h>
70 #include <sys/jail.h>
71 #include <sys/kernel.h>
72 #include <sys/lock.h>
73 #include <sys/malloc.h>
74 #include <sys/mbuf.h>
75 #include <sys/priv.h>
76 #include <sys/proc.h>
77 #include <sys/protosw.h>
78 #include <sys/signalvar.h>
79 #include <sys/socket.h>
80 #include <sys/socketvar.h>
81 #include <sys/sx.h>
82 #include <sys/syslog.h>
83 
84 #include <net/if.h>
85 #include <net/if_var.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/ip6protosw.h>
99 #include <netinet6/ip6_mroute.h>
100 #include <netinet6/in6_pcb.h>
101 #include <netinet6/ip6_var.h>
102 #include <netinet6/nd6.h>
103 #include <netinet6/raw_ip6.h>
104 #include <netinet6/scope6_var.h>
105 #include <netinet6/send.h>
106 
107 #ifdef IPSEC
108 #include <netipsec/ipsec.h>
109 #include <netipsec/ipsec6.h>
110 #endif /* IPSEC */
111 
112 #include <machine/stdarg.h>
113 
114 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
115 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
116 
117 /*
118  * Raw interface to IP6 protocol.
119  */
120 
121 VNET_DECLARE(struct inpcbhead, ripcb);
122 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
123 #define	V_ripcb				VNET(ripcb)
124 #define	V_ripcbinfo			VNET(ripcbinfo)
125 
126 extern u_long	rip_sendspace;
127 extern u_long	rip_recvspace;
128 
129 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
130 VNET_PCPUSTAT_SYSINIT(rip6stat);
131 
132 #ifdef VIMAGE
133 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
134 #endif /* VIMAGE */
135 
136 /*
137  * Hooks for multicast routing. They all default to NULL, so leave them not
138  * initialized and rely on BSS being set to 0.
139  */
140 
141 /*
142  * The socket used to communicate with the multicast routing daemon.
143  */
144 VNET_DEFINE(struct socket *, ip6_mrouter);
145 
146 /*
147  * The various mrouter functions.
148  */
149 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
150 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
151 int (*ip6_mrouter_done)(void);
152 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
153 int (*mrt6_ioctl)(u_long, caddr_t);
154 
155 /*
156  * Setup generic address and protocol structures for raw_input routine, then
157  * pass them along with mbuf chain.
158  */
159 int
160 rip6_input(struct mbuf **mp, int *offp, int proto)
161 {
162 	struct ifnet *ifp;
163 	struct mbuf *m = *mp;
164 	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
165 	register struct inpcb *in6p;
166 	struct inpcb *last = NULL;
167 	struct mbuf *opts = NULL;
168 	struct sockaddr_in6 fromsa;
169 
170 	RIP6STAT_INC(rip6s_ipackets);
171 
172 	init_sin6(&fromsa, m); /* general init */
173 
174 	ifp = m->m_pkthdr.rcvif;
175 
176 	INP_INFO_RLOCK(&V_ripcbinfo);
177 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
178 		/* XXX inp locking */
179 		if ((in6p->inp_vflag & INP_IPV6) == 0)
180 			continue;
181 		if (in6p->inp_ip_p &&
182 		    in6p->inp_ip_p != proto)
183 			continue;
184 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
185 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
186 			continue;
187 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
188 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
189 			continue;
190 		if (jailed_without_vnet(in6p->inp_cred)) {
191 			/*
192 			 * Allow raw socket in jail to receive multicast;
193 			 * assume process had PRIV_NETINET_RAW at attach,
194 			 * and fall through into normal filter path if so.
195 			 */
196 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
197 			    prison_check_ip6(in6p->inp_cred,
198 			    &ip6->ip6_dst) != 0)
199 				continue;
200 		}
201 		INP_RLOCK(in6p);
202 		if (in6p->in6p_cksum != -1) {
203 			RIP6STAT_INC(rip6s_isum);
204 			if (in6_cksum(m, proto, *offp,
205 			    m->m_pkthdr.len - *offp)) {
206 				INP_RUNLOCK(in6p);
207 				RIP6STAT_INC(rip6s_badsum);
208 				continue;
209 			}
210 		}
211 		/*
212 		 * If this raw socket has multicast state, and we
213 		 * have received a multicast, check if this socket
214 		 * should receive it, as multicast filtering is now
215 		 * the responsibility of the transport layer.
216 		 */
217 		if (in6p->in6p_moptions &&
218 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
219 			/*
220 			 * If the incoming datagram is for MLD, allow it
221 			 * through unconditionally to the raw socket.
222 			 *
223 			 * Use the M_RTALERT_MLD flag to check for MLD
224 			 * traffic without having to inspect the mbuf chain
225 			 * more deeply, as all MLDv1/v2 host messages MUST
226 			 * contain the Router Alert option.
227 			 *
228 			 * In the case of MLDv1, we may not have explicitly
229 			 * joined the group, and may have set IFF_ALLMULTI
230 			 * on the interface. im6o_mc_filter() may discard
231 			 * control traffic we actually need to see.
232 			 *
233 			 * Userland multicast routing daemons should continue
234 			 * filter the control traffic appropriately.
235 			 */
236 			int blocked;
237 
238 			blocked = MCAST_PASS;
239 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
240 				struct sockaddr_in6 mcaddr;
241 
242 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
243 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
244 				mcaddr.sin6_family = AF_INET6;
245 				mcaddr.sin6_addr = ip6->ip6_dst;
246 
247 				blocked = im6o_mc_filter(in6p->in6p_moptions,
248 				    ifp,
249 				    (struct sockaddr *)&mcaddr,
250 				    (struct sockaddr *)&fromsa);
251 			}
252 			if (blocked != MCAST_PASS) {
253 				IP6STAT_INC(ip6s_notmember);
254 				INP_RUNLOCK(in6p);
255 				continue;
256 			}
257 		}
258 		if (last != NULL) {
259 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
260 
261 #ifdef IPSEC
262 			/*
263 			 * Check AH/ESP integrity.
264 			 */
265 			if (n && ipsec6_in_reject(n, last)) {
266 				m_freem(n);
267 				/* Do not inject data into pcb. */
268 			} else
269 #endif /* IPSEC */
270 			if (n) {
271 				if (last->inp_flags & INP_CONTROLOPTS ||
272 				    last->inp_socket->so_options & SO_TIMESTAMP)
273 					ip6_savecontrol(last, n, &opts);
274 				/* strip intermediate headers */
275 				m_adj(n, *offp);
276 				if (sbappendaddr(&last->inp_socket->so_rcv,
277 						(struct sockaddr *)&fromsa,
278 						 n, opts) == 0) {
279 					m_freem(n);
280 					if (opts)
281 						m_freem(opts);
282 					RIP6STAT_INC(rip6s_fullsock);
283 				} else
284 					sorwakeup(last->inp_socket);
285 				opts = NULL;
286 			}
287 			INP_RUNLOCK(last);
288 		}
289 		last = in6p;
290 	}
291 	INP_INFO_RUNLOCK(&V_ripcbinfo);
292 #ifdef IPSEC
293 	/*
294 	 * Check AH/ESP integrity.
295 	 */
296 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
297 		m_freem(m);
298 		IP6STAT_DEC(ip6s_delivered);
299 		/* Do not inject data into pcb. */
300 		INP_RUNLOCK(last);
301 	} else
302 #endif /* IPSEC */
303 	if (last != NULL) {
304 		if (last->inp_flags & INP_CONTROLOPTS ||
305 		    last->inp_socket->so_options & SO_TIMESTAMP)
306 			ip6_savecontrol(last, m, &opts);
307 		/* Strip intermediate headers. */
308 		m_adj(m, *offp);
309 		if (sbappendaddr(&last->inp_socket->so_rcv,
310 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
311 			m_freem(m);
312 			if (opts)
313 				m_freem(opts);
314 			RIP6STAT_INC(rip6s_fullsock);
315 		} else
316 			sorwakeup(last->inp_socket);
317 		INP_RUNLOCK(last);
318 	} else {
319 		RIP6STAT_INC(rip6s_nosock);
320 		if (m->m_flags & M_MCAST)
321 			RIP6STAT_INC(rip6s_nosockmcast);
322 		if (proto == IPPROTO_NONE)
323 			m_freem(m);
324 		else {
325 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
326 			icmp6_error(m, ICMP6_PARAM_PROB,
327 			    ICMP6_PARAMPROB_NEXTHEADER,
328 			    prvnxtp - mtod(m, char *));
329 		}
330 		IP6STAT_DEC(ip6s_delivered);
331 	}
332 	return (IPPROTO_DONE);
333 }
334 
335 void
336 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
337 {
338 	struct ip6_hdr *ip6;
339 	struct mbuf *m;
340 	int off = 0;
341 	struct ip6ctlparam *ip6cp = NULL;
342 	const struct sockaddr_in6 *sa6_src = NULL;
343 	void *cmdarg;
344 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
345 
346 	if (sa->sa_family != AF_INET6 ||
347 	    sa->sa_len != sizeof(struct sockaddr_in6))
348 		return;
349 
350 	if ((unsigned)cmd >= PRC_NCMDS)
351 		return;
352 	if (PRC_IS_REDIRECT(cmd))
353 		notify = in6_rtchange, d = NULL;
354 	else if (cmd == PRC_HOSTDEAD)
355 		d = NULL;
356 	else if (inet6ctlerrmap[cmd] == 0)
357 		return;
358 
359 	/*
360 	 * If the parameter is from icmp6, decode it.
361 	 */
362 	if (d != NULL) {
363 		ip6cp = (struct ip6ctlparam *)d;
364 		m = ip6cp->ip6c_m;
365 		ip6 = ip6cp->ip6c_ip6;
366 		off = ip6cp->ip6c_off;
367 		cmdarg = ip6cp->ip6c_cmdarg;
368 		sa6_src = ip6cp->ip6c_src;
369 	} else {
370 		m = NULL;
371 		ip6 = NULL;
372 		cmdarg = NULL;
373 		sa6_src = &sa6_any;
374 	}
375 
376 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
377 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
378 }
379 
380 /*
381  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
382  * may have setup with control call.
383  */
384 int
385 rip6_output(struct mbuf *m, struct socket *so, ...)
386 {
387 	struct mbuf *control;
388 	struct m_tag *mtag;
389 	struct sockaddr_in6 *dstsock;
390 	struct in6_addr *dst;
391 	struct ip6_hdr *ip6;
392 	struct inpcb *in6p;
393 	u_int	plen = m->m_pkthdr.len;
394 	int error = 0;
395 	struct ip6_pktopts opt, *optp;
396 	struct ifnet *oifp = NULL;
397 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
398 	int scope_ambiguous = 0;
399 	int use_defzone = 0;
400 	int hlim = 0;
401 	struct in6_addr in6a;
402 	va_list ap;
403 
404 	va_start(ap, so);
405 	dstsock = va_arg(ap, struct sockaddr_in6 *);
406 	control = va_arg(ap, struct mbuf *);
407 	va_end(ap);
408 
409 	in6p = sotoinpcb(so);
410 	INP_WLOCK(in6p);
411 
412 	dst = &dstsock->sin6_addr;
413 	if (control != NULL) {
414 		if ((error = ip6_setpktopts(control, &opt,
415 		    in6p->in6p_outputopts, so->so_cred,
416 		    so->so_proto->pr_protocol)) != 0) {
417 			goto bad;
418 		}
419 		optp = &opt;
420 	} else
421 		optp = in6p->in6p_outputopts;
422 
423 	/*
424 	 * Check and convert scope zone ID into internal form.
425 	 *
426 	 * XXX: we may still need to determine the zone later.
427 	 */
428 	if (!(so->so_state & SS_ISCONNECTED)) {
429 		if (!optp || !optp->ip6po_pktinfo ||
430 		    !optp->ip6po_pktinfo->ipi6_ifindex)
431 			use_defzone = V_ip6_use_defzone;
432 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
433 			scope_ambiguous = 1;
434 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
435 			goto bad;
436 	}
437 
438 	/*
439 	 * For an ICMPv6 packet, we should know its type and code to update
440 	 * statistics.
441 	 */
442 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
443 		struct icmp6_hdr *icmp6;
444 		if (m->m_len < sizeof(struct icmp6_hdr) &&
445 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
446 			error = ENOBUFS;
447 			goto bad;
448 		}
449 		icmp6 = mtod(m, struct icmp6_hdr *);
450 		type = icmp6->icmp6_type;
451 		code = icmp6->icmp6_code;
452 	}
453 
454 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
455 	if (m == NULL) {
456 		error = ENOBUFS;
457 		goto bad;
458 	}
459 	ip6 = mtod(m, struct ip6_hdr *);
460 
461 	/*
462 	 * Source address selection.
463 	 */
464 	error = in6_selectsrc_socket(dstsock, optp, in6p, so->so_cred,
465 	    scope_ambiguous, &in6a, &hlim);
466 
467 	if (error)
468 		goto bad;
469 	error = prison_check_ip6(in6p->inp_cred, &in6a);
470 	if (error != 0)
471 		goto bad;
472 	ip6->ip6_src = in6a;
473 
474 	ip6->ip6_dst = dstsock->sin6_addr;
475 
476 	/*
477 	 * Fill in the rest of the IPv6 header fields.
478 	 */
479 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
480 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
481 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
482 	    (IPV6_VERSION & IPV6_VERSION_MASK);
483 
484 	/*
485 	 * ip6_plen will be filled in ip6_output, so not fill it here.
486 	 */
487 	ip6->ip6_nxt = in6p->inp_ip_p;
488 	ip6->ip6_hlim = hlim;
489 
490 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
491 	    in6p->in6p_cksum != -1) {
492 		struct mbuf *n;
493 		int off;
494 		u_int16_t *p;
495 
496 		/* Compute checksum. */
497 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
498 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
499 		else
500 			off = in6p->in6p_cksum;
501 		if (plen < off + 1) {
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 	    so->so_proto->pr_protocol == 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 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
537 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
538 		if (oifp)
539 			icmp6_ifoutstat_inc(oifp, type, code);
540 		ICMP6STAT_INC(icp6s_outhist[type]);
541 	} else
542 		RIP6STAT_INC(rip6s_opackets);
543 
544 	goto freectl;
545 
546  bad:
547 	if (m)
548 		m_freem(m);
549 
550  freectl:
551 	if (control != NULL) {
552 		ip6_clearpktopts(&opt, -1);
553 		m_freem(control);
554 	}
555 	INP_WUNLOCK(in6p);
556 	return (error);
557 }
558 
559 /*
560  * Raw IPv6 socket option processing.
561  */
562 int
563 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
564 {
565 	struct inpcb *inp;
566 	int error;
567 
568 	if (sopt->sopt_level == IPPROTO_ICMPV6)
569 		/*
570 		 * XXX: is it better to call icmp6_ctloutput() directly
571 		 * from protosw?
572 		 */
573 		return (icmp6_ctloutput(so, sopt));
574 	else if (sopt->sopt_level != IPPROTO_IPV6) {
575 		if (sopt->sopt_level == SOL_SOCKET &&
576 		    sopt->sopt_name == SO_SETFIB) {
577 			inp = sotoinpcb(so);
578 			INP_WLOCK(inp);
579 			inp->inp_inc.inc_fibnum = so->so_fibnum;
580 			INP_WUNLOCK(inp);
581 			return (0);
582 		}
583 		return (EINVAL);
584 	}
585 
586 	error = 0;
587 
588 	switch (sopt->sopt_dir) {
589 	case SOPT_GET:
590 		switch (sopt->sopt_name) {
591 		case MRT6_INIT:
592 		case MRT6_DONE:
593 		case MRT6_ADD_MIF:
594 		case MRT6_DEL_MIF:
595 		case MRT6_ADD_MFC:
596 		case MRT6_DEL_MFC:
597 		case MRT6_PIM:
598 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
599 			    EOPNOTSUPP;
600 			break;
601 		case IPV6_CHECKSUM:
602 			error = ip6_raw_ctloutput(so, sopt);
603 			break;
604 		default:
605 			error = ip6_ctloutput(so, sopt);
606 			break;
607 		}
608 		break;
609 
610 	case SOPT_SET:
611 		switch (sopt->sopt_name) {
612 		case MRT6_INIT:
613 		case MRT6_DONE:
614 		case MRT6_ADD_MIF:
615 		case MRT6_DEL_MIF:
616 		case MRT6_ADD_MFC:
617 		case MRT6_DEL_MFC:
618 		case MRT6_PIM:
619 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
620 			    EOPNOTSUPP;
621 			break;
622 		case IPV6_CHECKSUM:
623 			error = ip6_raw_ctloutput(so, sopt);
624 			break;
625 		default:
626 			error = ip6_ctloutput(so, sopt);
627 			break;
628 		}
629 		break;
630 	}
631 
632 	return (error);
633 }
634 
635 static int
636 rip6_attach(struct socket *so, int proto, struct thread *td)
637 {
638 	struct inpcb *inp;
639 	struct icmp6_filter *filter;
640 	int error;
641 
642 	inp = sotoinpcb(so);
643 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
644 
645 	error = priv_check(td, PRIV_NETINET_RAW);
646 	if (error)
647 		return (error);
648 	error = soreserve(so, rip_sendspace, rip_recvspace);
649 	if (error)
650 		return (error);
651 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
652 	if (filter == NULL)
653 		return (ENOMEM);
654 	INP_INFO_WLOCK(&V_ripcbinfo);
655 	error = in_pcballoc(so, &V_ripcbinfo);
656 	if (error) {
657 		INP_INFO_WUNLOCK(&V_ripcbinfo);
658 		free(filter, M_PCB);
659 		return (error);
660 	}
661 	inp = (struct inpcb *)so->so_pcb;
662 	INP_INFO_WUNLOCK(&V_ripcbinfo);
663 	inp->inp_vflag |= INP_IPV6;
664 	inp->inp_ip_p = (long)proto;
665 	inp->in6p_hops = -1;	/* use kernel default */
666 	inp->in6p_cksum = -1;
667 	inp->in6p_icmp6filt = filter;
668 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
669 	INP_WUNLOCK(inp);
670 	return (0);
671 }
672 
673 static void
674 rip6_detach(struct socket *so)
675 {
676 	struct inpcb *inp;
677 
678 	inp = sotoinpcb(so);
679 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
680 
681 	if (so == V_ip6_mrouter && ip6_mrouter_done)
682 		ip6_mrouter_done();
683 	/* xxx: RSVP */
684 	INP_INFO_WLOCK(&V_ripcbinfo);
685 	INP_WLOCK(inp);
686 	free(inp->in6p_icmp6filt, M_PCB);
687 	in_pcbdetach(inp);
688 	in_pcbfree(inp);
689 	INP_INFO_WUNLOCK(&V_ripcbinfo);
690 }
691 
692 /* XXXRW: This can't ever be called. */
693 static void
694 rip6_abort(struct socket *so)
695 {
696 	struct inpcb *inp;
697 
698 	inp = sotoinpcb(so);
699 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
700 
701 	soisdisconnected(so);
702 }
703 
704 static void
705 rip6_close(struct socket *so)
706 {
707 	struct inpcb *inp;
708 
709 	inp = sotoinpcb(so);
710 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
711 
712 	soisdisconnected(so);
713 }
714 
715 static int
716 rip6_disconnect(struct socket *so)
717 {
718 	struct inpcb *inp;
719 
720 	inp = sotoinpcb(so);
721 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
722 
723 	if ((so->so_state & SS_ISCONNECTED) == 0)
724 		return (ENOTCONN);
725 	inp->in6p_faddr = in6addr_any;
726 	rip6_abort(so);
727 	return (0);
728 }
729 
730 static int
731 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
732 {
733 	struct inpcb *inp;
734 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
735 	struct ifaddr *ifa = NULL;
736 	int error = 0;
737 
738 	inp = sotoinpcb(so);
739 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
740 
741 	if (nam->sa_len != sizeof(*addr))
742 		return (EINVAL);
743 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
744 		return (error);
745 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
746 		return (EADDRNOTAVAIL);
747 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
748 		return (error);
749 
750 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
751 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
752 		return (EADDRNOTAVAIL);
753 	if (ifa != NULL &&
754 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
755 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
756 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
757 		ifa_free(ifa);
758 		return (EADDRNOTAVAIL);
759 	}
760 	if (ifa != NULL)
761 		ifa_free(ifa);
762 	INP_INFO_WLOCK(&V_ripcbinfo);
763 	INP_WLOCK(inp);
764 	inp->in6p_laddr = addr->sin6_addr;
765 	INP_WUNLOCK(inp);
766 	INP_INFO_WUNLOCK(&V_ripcbinfo);
767 	return (0);
768 }
769 
770 static int
771 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
772 {
773 	struct inpcb *inp;
774 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
775 	struct in6_addr in6a;
776 	int error = 0, scope_ambiguous = 0;
777 
778 	inp = sotoinpcb(so);
779 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
780 
781 	if (nam->sa_len != sizeof(*addr))
782 		return (EINVAL);
783 	if (TAILQ_EMPTY(&V_ifnet))
784 		return (EADDRNOTAVAIL);
785 	if (addr->sin6_family != AF_INET6)
786 		return (EAFNOSUPPORT);
787 
788 	/*
789 	 * Application should provide a proper zone ID or the use of default
790 	 * zone IDs should be enabled.  Unfortunately, some applications do
791 	 * not behave as it should, so we need a workaround.  Even if an
792 	 * appropriate ID is not determined, we'll see if we can determine
793 	 * the outgoing interface.  If we can, determine the zone ID based on
794 	 * the interface below.
795 	 */
796 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
797 		scope_ambiguous = 1;
798 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
799 		return (error);
800 
801 	INP_INFO_WLOCK(&V_ripcbinfo);
802 	INP_WLOCK(inp);
803 	/* Source address selection. XXX: need pcblookup? */
804 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
805 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
806 	if (error) {
807 		INP_WUNLOCK(inp);
808 		INP_INFO_WUNLOCK(&V_ripcbinfo);
809 		return (error);
810 	}
811 
812 	inp->in6p_faddr = addr->sin6_addr;
813 	inp->in6p_laddr = in6a;
814 	soisconnected(so);
815 	INP_WUNLOCK(inp);
816 	INP_INFO_WUNLOCK(&V_ripcbinfo);
817 	return (0);
818 }
819 
820 static int
821 rip6_shutdown(struct socket *so)
822 {
823 	struct inpcb *inp;
824 
825 	inp = sotoinpcb(so);
826 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
827 
828 	INP_WLOCK(inp);
829 	socantsendmore(so);
830 	INP_WUNLOCK(inp);
831 	return (0);
832 }
833 
834 static int
835 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
836     struct mbuf *control, struct thread *td)
837 {
838 	struct inpcb *inp;
839 	struct sockaddr_in6 tmp;
840 	struct sockaddr_in6 *dst;
841 	int ret;
842 
843 	inp = sotoinpcb(so);
844 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
845 
846 	/* Always copy sockaddr to avoid overwrites. */
847 	/* Unlocked read. */
848 	if (so->so_state & SS_ISCONNECTED) {
849 		if (nam) {
850 			m_freem(m);
851 			return (EISCONN);
852 		}
853 		/* XXX */
854 		bzero(&tmp, sizeof(tmp));
855 		tmp.sin6_family = AF_INET6;
856 		tmp.sin6_len = sizeof(struct sockaddr_in6);
857 		INP_RLOCK(inp);
858 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
859 		    sizeof(struct in6_addr));
860 		INP_RUNLOCK(inp);
861 		dst = &tmp;
862 	} else {
863 		if (nam == NULL) {
864 			m_freem(m);
865 			return (ENOTCONN);
866 		}
867 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
868 			m_freem(m);
869 			return (EINVAL);
870 		}
871 		tmp = *(struct sockaddr_in6 *)nam;
872 		dst = &tmp;
873 
874 		if (dst->sin6_family == AF_UNSPEC) {
875 			/*
876 			 * XXX: we allow this case for backward
877 			 * compatibility to buggy applications that
878 			 * rely on old (and wrong) kernel behavior.
879 			 */
880 			log(LOG_INFO, "rip6 SEND: address family is "
881 			    "unspec. Assume AF_INET6\n");
882 			dst->sin6_family = AF_INET6;
883 		} else if (dst->sin6_family != AF_INET6) {
884 			m_freem(m);
885 			return(EAFNOSUPPORT);
886 		}
887 	}
888 	ret = rip6_output(m, so, dst, control);
889 	return (ret);
890 }
891 
892 struct pr_usrreqs rip6_usrreqs = {
893 	.pru_abort =		rip6_abort,
894 	.pru_attach =		rip6_attach,
895 	.pru_bind =		rip6_bind,
896 	.pru_connect =		rip6_connect,
897 	.pru_control =		in6_control,
898 	.pru_detach =		rip6_detach,
899 	.pru_disconnect =	rip6_disconnect,
900 	.pru_peeraddr =		in6_getpeeraddr,
901 	.pru_send =		rip6_send,
902 	.pru_shutdown =		rip6_shutdown,
903 	.pru_sockaddr =		in6_getsockaddr,
904 	.pru_close =		rip6_close,
905 };
906