xref: /freebsd/sys/netinet6/raw_ip6.c (revision ff0ba87247820afbdfdc1b307c803f7923d0e4d3)
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 = 0;
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 				IPSEC6STAT_INC(ips_in_polvio);
268 				/* Do not inject data into pcb. */
269 			} else
270 #endif /* IPSEC */
271 			if (n) {
272 				if (last->inp_flags & INP_CONTROLOPTS ||
273 				    last->inp_socket->so_options & SO_TIMESTAMP)
274 					ip6_savecontrol(last, n, &opts);
275 				/* strip intermediate headers */
276 				m_adj(n, *offp);
277 				if (sbappendaddr(&last->inp_socket->so_rcv,
278 						(struct sockaddr *)&fromsa,
279 						 n, opts) == 0) {
280 					m_freem(n);
281 					if (opts)
282 						m_freem(opts);
283 					RIP6STAT_INC(rip6s_fullsock);
284 				} else
285 					sorwakeup(last->inp_socket);
286 				opts = NULL;
287 			}
288 			INP_RUNLOCK(last);
289 		}
290 		last = in6p;
291 	}
292 	INP_INFO_RUNLOCK(&V_ripcbinfo);
293 #ifdef IPSEC
294 	/*
295 	 * Check AH/ESP integrity.
296 	 */
297 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
298 		m_freem(m);
299 		IPSEC6STAT_INC(ips_in_polvio);
300 		IP6STAT_DEC(ip6s_delivered);
301 		/* Do not inject data into pcb. */
302 		INP_RUNLOCK(last);
303 	} else
304 #endif /* IPSEC */
305 	if (last != NULL) {
306 		if (last->inp_flags & INP_CONTROLOPTS ||
307 		    last->inp_socket->so_options & SO_TIMESTAMP)
308 			ip6_savecontrol(last, m, &opts);
309 		/* Strip intermediate headers. */
310 		m_adj(m, *offp);
311 		if (sbappendaddr(&last->inp_socket->so_rcv,
312 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
313 			m_freem(m);
314 			if (opts)
315 				m_freem(opts);
316 			RIP6STAT_INC(rip6s_fullsock);
317 		} else
318 			sorwakeup(last->inp_socket);
319 		INP_RUNLOCK(last);
320 	} else {
321 		RIP6STAT_INC(rip6s_nosock);
322 		if (m->m_flags & M_MCAST)
323 			RIP6STAT_INC(rip6s_nosockmcast);
324 		if (proto == IPPROTO_NONE)
325 			m_freem(m);
326 		else {
327 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
328 			icmp6_error(m, ICMP6_PARAM_PROB,
329 			    ICMP6_PARAMPROB_NEXTHEADER,
330 			    prvnxtp - mtod(m, char *));
331 		}
332 		IP6STAT_DEC(ip6s_delivered);
333 	}
334 	return (IPPROTO_DONE);
335 }
336 
337 void
338 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
339 {
340 	struct ip6_hdr *ip6;
341 	struct mbuf *m;
342 	int off = 0;
343 	struct ip6ctlparam *ip6cp = NULL;
344 	const struct sockaddr_in6 *sa6_src = NULL;
345 	void *cmdarg;
346 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
347 
348 	if (sa->sa_family != AF_INET6 ||
349 	    sa->sa_len != sizeof(struct sockaddr_in6))
350 		return;
351 
352 	if ((unsigned)cmd >= PRC_NCMDS)
353 		return;
354 	if (PRC_IS_REDIRECT(cmd))
355 		notify = in6_rtchange, d = NULL;
356 	else if (cmd == PRC_HOSTDEAD)
357 		d = NULL;
358 	else if (inet6ctlerrmap[cmd] == 0)
359 		return;
360 
361 	/*
362 	 * If the parameter is from icmp6, decode it.
363 	 */
364 	if (d != NULL) {
365 		ip6cp = (struct ip6ctlparam *)d;
366 		m = ip6cp->ip6c_m;
367 		ip6 = ip6cp->ip6c_ip6;
368 		off = ip6cp->ip6c_off;
369 		cmdarg = ip6cp->ip6c_cmdarg;
370 		sa6_src = ip6cp->ip6c_src;
371 	} else {
372 		m = NULL;
373 		ip6 = NULL;
374 		cmdarg = NULL;
375 		sa6_src = &sa6_any;
376 	}
377 
378 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
379 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
380 }
381 
382 /*
383  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
384  * may have setup with control call.
385  */
386 int
387 rip6_output(struct mbuf *m, struct socket *so, ...)
388 {
389 	struct mbuf *control;
390 	struct m_tag *mtag;
391 	struct sockaddr_in6 *dstsock;
392 	struct in6_addr *dst;
393 	struct ip6_hdr *ip6;
394 	struct inpcb *in6p;
395 	u_int	plen = m->m_pkthdr.len;
396 	int error = 0;
397 	struct ip6_pktopts opt, *optp;
398 	struct ifnet *oifp = NULL;
399 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
400 	int scope_ambiguous = 0;
401 	int use_defzone = 0;
402 	struct in6_addr in6a;
403 	va_list ap;
404 
405 	va_start(ap, so);
406 	dstsock = va_arg(ap, struct sockaddr_in6 *);
407 	control = va_arg(ap, struct mbuf *);
408 	va_end(ap);
409 
410 	in6p = sotoinpcb(so);
411 	INP_WLOCK(in6p);
412 
413 	dst = &dstsock->sin6_addr;
414 	if (control != NULL) {
415 		if ((error = ip6_setpktopts(control, &opt,
416 		    in6p->in6p_outputopts, so->so_cred,
417 		    so->so_proto->pr_protocol)) != 0) {
418 			goto bad;
419 		}
420 		optp = &opt;
421 	} else
422 		optp = in6p->in6p_outputopts;
423 
424 	/*
425 	 * Check and convert scope zone ID into internal form.
426 	 *
427 	 * XXX: we may still need to determine the zone later.
428 	 */
429 	if (!(so->so_state & SS_ISCONNECTED)) {
430 		if (!optp || !optp->ip6po_pktinfo ||
431 		    !optp->ip6po_pktinfo->ipi6_ifindex)
432 			use_defzone = V_ip6_use_defzone;
433 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
434 			scope_ambiguous = 1;
435 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
436 			goto bad;
437 	}
438 
439 	/*
440 	 * For an ICMPv6 packet, we should know its type and code to update
441 	 * statistics.
442 	 */
443 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
444 		struct icmp6_hdr *icmp6;
445 		if (m->m_len < sizeof(struct icmp6_hdr) &&
446 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
447 			error = ENOBUFS;
448 			goto bad;
449 		}
450 		icmp6 = mtod(m, struct icmp6_hdr *);
451 		type = icmp6->icmp6_type;
452 		code = icmp6->icmp6_code;
453 	}
454 
455 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
456 	if (m == NULL) {
457 		error = ENOBUFS;
458 		goto bad;
459 	}
460 	ip6 = mtod(m, struct ip6_hdr *);
461 
462 	/*
463 	 * Source address selection.
464 	 */
465 	error = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
466 	    &oifp, &in6a);
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 	if (oifp && scope_ambiguous) {
475 		/*
476 		 * Application should provide a proper zone ID or the use of
477 		 * default zone IDs should be enabled.  Unfortunately, some
478 		 * applications do not behave as it should, so we need a
479 		 * workaround.  Even if an appropriate ID is not determined
480 		 * (when it's required), if we can determine the outgoing
481 		 * interface. determine the zone ID based on the interface.
482 		 */
483 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
484 		if (error != 0)
485 			goto bad;
486 	}
487 	ip6->ip6_dst = dstsock->sin6_addr;
488 
489 	/*
490 	 * Fill in the rest of the IPv6 header fields.
491 	 */
492 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
493 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
494 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
495 	    (IPV6_VERSION & IPV6_VERSION_MASK);
496 
497 	/*
498 	 * ip6_plen will be filled in ip6_output, so not fill it here.
499 	 */
500 	ip6->ip6_nxt = in6p->inp_ip_p;
501 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
502 
503 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
504 	    in6p->in6p_cksum != -1) {
505 		struct mbuf *n;
506 		int off;
507 		u_int16_t *p;
508 
509 		/* Compute checksum. */
510 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
511 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
512 		else
513 			off = in6p->in6p_cksum;
514 		if (plen < off + 1) {
515 			error = EINVAL;
516 			goto bad;
517 		}
518 		off += sizeof(struct ip6_hdr);
519 
520 		n = m;
521 		while (n && n->m_len <= off) {
522 			off -= n->m_len;
523 			n = n->m_next;
524 		}
525 		if (!n)
526 			goto bad;
527 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
528 		*p = 0;
529 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
530 	}
531 
532 	/*
533 	 * Send RA/RS messages to user land for protection, before sending
534 	 * them to rtadvd/rtsol.
535 	 */
536 	if ((send_sendso_input_hook != NULL) &&
537 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
538 		switch (type) {
539 		case ND_ROUTER_ADVERT:
540 		case ND_ROUTER_SOLICIT:
541 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
542 				sizeof(unsigned short), M_NOWAIT);
543 			if (mtag == NULL)
544 				goto bad;
545 			m_tag_prepend(m, mtag);
546 		}
547 	}
548 
549 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
550 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
551 		if (oifp)
552 			icmp6_ifoutstat_inc(oifp, type, code);
553 		ICMP6STAT_INC(icp6s_outhist[type]);
554 	} else
555 		RIP6STAT_INC(rip6s_opackets);
556 
557 	goto freectl;
558 
559  bad:
560 	if (m)
561 		m_freem(m);
562 
563  freectl:
564 	if (control != NULL) {
565 		ip6_clearpktopts(&opt, -1);
566 		m_freem(control);
567 	}
568 	INP_WUNLOCK(in6p);
569 	return (error);
570 }
571 
572 /*
573  * Raw IPv6 socket option processing.
574  */
575 int
576 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
577 {
578 	struct inpcb *inp;
579 	int error;
580 
581 	if (sopt->sopt_level == IPPROTO_ICMPV6)
582 		/*
583 		 * XXX: is it better to call icmp6_ctloutput() directly
584 		 * from protosw?
585 		 */
586 		return (icmp6_ctloutput(so, sopt));
587 	else if (sopt->sopt_level != IPPROTO_IPV6) {
588 		if (sopt->sopt_level == SOL_SOCKET &&
589 		    sopt->sopt_name == SO_SETFIB) {
590 			inp = sotoinpcb(so);
591 			INP_WLOCK(inp);
592 			inp->inp_inc.inc_fibnum = so->so_fibnum;
593 			INP_WUNLOCK(inp);
594 			return (0);
595 		}
596 		return (EINVAL);
597 	}
598 
599 	error = 0;
600 
601 	switch (sopt->sopt_dir) {
602 	case SOPT_GET:
603 		switch (sopt->sopt_name) {
604 		case MRT6_INIT:
605 		case MRT6_DONE:
606 		case MRT6_ADD_MIF:
607 		case MRT6_DEL_MIF:
608 		case MRT6_ADD_MFC:
609 		case MRT6_DEL_MFC:
610 		case MRT6_PIM:
611 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
612 			    EOPNOTSUPP;
613 			break;
614 		case IPV6_CHECKSUM:
615 			error = ip6_raw_ctloutput(so, sopt);
616 			break;
617 		default:
618 			error = ip6_ctloutput(so, sopt);
619 			break;
620 		}
621 		break;
622 
623 	case SOPT_SET:
624 		switch (sopt->sopt_name) {
625 		case MRT6_INIT:
626 		case MRT6_DONE:
627 		case MRT6_ADD_MIF:
628 		case MRT6_DEL_MIF:
629 		case MRT6_ADD_MFC:
630 		case MRT6_DEL_MFC:
631 		case MRT6_PIM:
632 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
633 			    EOPNOTSUPP;
634 			break;
635 		case IPV6_CHECKSUM:
636 			error = ip6_raw_ctloutput(so, sopt);
637 			break;
638 		default:
639 			error = ip6_ctloutput(so, sopt);
640 			break;
641 		}
642 		break;
643 	}
644 
645 	return (error);
646 }
647 
648 static int
649 rip6_attach(struct socket *so, int proto, struct thread *td)
650 {
651 	struct inpcb *inp;
652 	struct icmp6_filter *filter;
653 	int error;
654 
655 	inp = sotoinpcb(so);
656 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
657 
658 	error = priv_check(td, PRIV_NETINET_RAW);
659 	if (error)
660 		return (error);
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 	INP_INFO_WLOCK(&V_ripcbinfo);
668 	error = in_pcballoc(so, &V_ripcbinfo);
669 	if (error) {
670 		INP_INFO_WUNLOCK(&V_ripcbinfo);
671 		free(filter, M_PCB);
672 		return (error);
673 	}
674 	inp = (struct inpcb *)so->so_pcb;
675 	INP_INFO_WUNLOCK(&V_ripcbinfo);
676 	inp->inp_vflag |= INP_IPV6;
677 	inp->inp_ip_p = (long)proto;
678 	inp->in6p_hops = -1;	/* use kernel default */
679 	inp->in6p_cksum = -1;
680 	inp->in6p_icmp6filt = filter;
681 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
682 	INP_WUNLOCK(inp);
683 	return (0);
684 }
685 
686 static void
687 rip6_detach(struct socket *so)
688 {
689 	struct inpcb *inp;
690 
691 	inp = sotoinpcb(so);
692 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
693 
694 	if (so == V_ip6_mrouter && ip6_mrouter_done)
695 		ip6_mrouter_done();
696 	/* xxx: RSVP */
697 	INP_INFO_WLOCK(&V_ripcbinfo);
698 	INP_WLOCK(inp);
699 	free(inp->in6p_icmp6filt, M_PCB);
700 	in_pcbdetach(inp);
701 	in_pcbfree(inp);
702 	INP_INFO_WUNLOCK(&V_ripcbinfo);
703 }
704 
705 /* XXXRW: This can't ever be called. */
706 static void
707 rip6_abort(struct socket *so)
708 {
709 	struct inpcb *inp;
710 
711 	inp = sotoinpcb(so);
712 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
713 
714 	soisdisconnected(so);
715 }
716 
717 static void
718 rip6_close(struct socket *so)
719 {
720 	struct inpcb *inp;
721 
722 	inp = sotoinpcb(so);
723 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
724 
725 	soisdisconnected(so);
726 }
727 
728 static int
729 rip6_disconnect(struct socket *so)
730 {
731 	struct inpcb *inp;
732 
733 	inp = sotoinpcb(so);
734 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
735 
736 	if ((so->so_state & SS_ISCONNECTED) == 0)
737 		return (ENOTCONN);
738 	inp->in6p_faddr = in6addr_any;
739 	rip6_abort(so);
740 	return (0);
741 }
742 
743 static int
744 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
745 {
746 	struct inpcb *inp;
747 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
748 	struct ifaddr *ifa = NULL;
749 	int error = 0;
750 
751 	inp = sotoinpcb(so);
752 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
753 
754 	if (nam->sa_len != sizeof(*addr))
755 		return (EINVAL);
756 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
757 		return (error);
758 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
759 		return (EADDRNOTAVAIL);
760 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
761 		return (error);
762 
763 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
764 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
765 		return (EADDRNOTAVAIL);
766 	if (ifa != NULL &&
767 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
768 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
769 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
770 		ifa_free(ifa);
771 		return (EADDRNOTAVAIL);
772 	}
773 	if (ifa != NULL)
774 		ifa_free(ifa);
775 	INP_INFO_WLOCK(&V_ripcbinfo);
776 	INP_WLOCK(inp);
777 	inp->in6p_laddr = addr->sin6_addr;
778 	INP_WUNLOCK(inp);
779 	INP_INFO_WUNLOCK(&V_ripcbinfo);
780 	return (0);
781 }
782 
783 static int
784 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
785 {
786 	struct inpcb *inp;
787 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
788 	struct in6_addr in6a;
789 	struct ifnet *ifp = NULL;
790 	int error = 0, scope_ambiguous = 0;
791 
792 	inp = sotoinpcb(so);
793 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
794 
795 	if (nam->sa_len != sizeof(*addr))
796 		return (EINVAL);
797 	if (TAILQ_EMPTY(&V_ifnet))
798 		return (EADDRNOTAVAIL);
799 	if (addr->sin6_family != AF_INET6)
800 		return (EAFNOSUPPORT);
801 
802 	/*
803 	 * Application should provide a proper zone ID or the use of default
804 	 * zone IDs should be enabled.  Unfortunately, some applications do
805 	 * not behave as it should, so we need a workaround.  Even if an
806 	 * appropriate ID is not determined, we'll see if we can determine
807 	 * the outgoing interface.  If we can, determine the zone ID based on
808 	 * the interface below.
809 	 */
810 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
811 		scope_ambiguous = 1;
812 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
813 		return (error);
814 
815 	INP_INFO_WLOCK(&V_ripcbinfo);
816 	INP_WLOCK(inp);
817 	/* Source address selection. XXX: need pcblookup? */
818 	error = in6_selectsrc(addr, inp->in6p_outputopts,
819 	    inp, NULL, so->so_cred, &ifp, &in6a);
820 	if (error) {
821 		INP_WUNLOCK(inp);
822 		INP_INFO_WUNLOCK(&V_ripcbinfo);
823 		return (error);
824 	}
825 
826 	/* XXX: see above */
827 	if (ifp && scope_ambiguous &&
828 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
829 		INP_WUNLOCK(inp);
830 		INP_INFO_WUNLOCK(&V_ripcbinfo);
831 		return (error);
832 	}
833 	inp->in6p_faddr = addr->sin6_addr;
834 	inp->in6p_laddr = in6a;
835 	soisconnected(so);
836 	INP_WUNLOCK(inp);
837 	INP_INFO_WUNLOCK(&V_ripcbinfo);
838 	return (0);
839 }
840 
841 static int
842 rip6_shutdown(struct socket *so)
843 {
844 	struct inpcb *inp;
845 
846 	inp = sotoinpcb(so);
847 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
848 
849 	INP_WLOCK(inp);
850 	socantsendmore(so);
851 	INP_WUNLOCK(inp);
852 	return (0);
853 }
854 
855 static int
856 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
857     struct mbuf *control, struct thread *td)
858 {
859 	struct inpcb *inp;
860 	struct sockaddr_in6 tmp;
861 	struct sockaddr_in6 *dst;
862 	int ret;
863 
864 	inp = sotoinpcb(so);
865 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
866 
867 	/* Always copy sockaddr to avoid overwrites. */
868 	/* Unlocked read. */
869 	if (so->so_state & SS_ISCONNECTED) {
870 		if (nam) {
871 			m_freem(m);
872 			return (EISCONN);
873 		}
874 		/* XXX */
875 		bzero(&tmp, sizeof(tmp));
876 		tmp.sin6_family = AF_INET6;
877 		tmp.sin6_len = sizeof(struct sockaddr_in6);
878 		INP_RLOCK(inp);
879 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
880 		    sizeof(struct in6_addr));
881 		INP_RUNLOCK(inp);
882 		dst = &tmp;
883 	} else {
884 		if (nam == NULL) {
885 			m_freem(m);
886 			return (ENOTCONN);
887 		}
888 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
889 			m_freem(m);
890 			return (EINVAL);
891 		}
892 		tmp = *(struct sockaddr_in6 *)nam;
893 		dst = &tmp;
894 
895 		if (dst->sin6_family == AF_UNSPEC) {
896 			/*
897 			 * XXX: we allow this case for backward
898 			 * compatibility to buggy applications that
899 			 * rely on old (and wrong) kernel behavior.
900 			 */
901 			log(LOG_INFO, "rip6 SEND: address family is "
902 			    "unspec. Assume AF_INET6\n");
903 			dst->sin6_family = AF_INET6;
904 		} else if (dst->sin6_family != AF_INET6) {
905 			m_freem(m);
906 			return(EAFNOSUPPORT);
907 		}
908 	}
909 	ret = rip6_output(m, so, dst, control);
910 	return (ret);
911 }
912 
913 struct pr_usrreqs rip6_usrreqs = {
914 	.pru_abort =		rip6_abort,
915 	.pru_attach =		rip6_attach,
916 	.pru_bind =		rip6_bind,
917 	.pru_connect =		rip6_connect,
918 	.pru_control =		in6_control,
919 	.pru_detach =		rip6_detach,
920 	.pru_disconnect =	rip6_disconnect,
921 	.pru_peeraddr =		in6_getpeeraddr,
922 	.pru_send =		rip6_send,
923 	.pru_shutdown =		rip6_shutdown,
924 	.pru_sockaddr =		in6_getsockaddr,
925 	.pru_close =		rip6_close,
926 };
927