xref: /freebsd/sys/netinet6/raw_ip6.c (revision fe267a559009cbf34f9341666fe4d88a92c02d5e)
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 
70 #include <sys/param.h>
71 #include <sys/errno.h>
72 #include <sys/jail.h>
73 #include <sys/kernel.h>
74 #include <sys/lock.h>
75 #include <sys/malloc.h>
76 #include <sys/mbuf.h>
77 #include <sys/priv.h>
78 #include <sys/proc.h>
79 #include <sys/protosw.h>
80 #include <sys/signalvar.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/sx.h>
84 #include <sys/syslog.h>
85 
86 #include <net/if.h>
87 #include <net/if_var.h>
88 #include <net/if_types.h>
89 #include <net/route.h>
90 #include <net/vnet.h>
91 
92 #include <netinet/in.h>
93 #include <netinet/in_var.h>
94 #include <netinet/in_systm.h>
95 #include <netinet/in_pcb.h>
96 
97 #include <netinet/icmp6.h>
98 #include <netinet/ip6.h>
99 #include <netinet/ip_var.h>
100 #include <netinet6/ip6protosw.h>
101 #include <netinet6/ip6_mroute.h>
102 #include <netinet6/in6_pcb.h>
103 #include <netinet6/ip6_var.h>
104 #include <netinet6/nd6.h>
105 #include <netinet6/raw_ip6.h>
106 #include <netinet6/scope6_var.h>
107 #include <netinet6/send.h>
108 
109 #include <netipsec/ipsec_support.h>
110 
111 #include <machine/stdarg.h>
112 
113 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
114 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
115 
116 /*
117  * Raw interface to IP6 protocol.
118  */
119 
120 VNET_DECLARE(struct inpcbhead, ripcb);
121 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
122 #define	V_ripcb				VNET(ripcb)
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 /*
155  * Setup generic address and protocol structures for raw_input routine, then
156  * pass them along with mbuf chain.
157  */
158 int
159 rip6_input(struct mbuf **mp, int *offp, int proto)
160 {
161 	struct ifnet *ifp;
162 	struct mbuf *m = *mp;
163 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
164 	struct inpcb *in6p;
165 	struct inpcb *last = NULL;
166 	struct mbuf *opts = NULL;
167 	struct sockaddr_in6 fromsa;
168 
169 	RIP6STAT_INC(rip6s_ipackets);
170 
171 	init_sin6(&fromsa, m, 0); /* general init */
172 
173 	ifp = m->m_pkthdr.rcvif;
174 
175 	INP_INFO_RLOCK(&V_ripcbinfo);
176 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
177 		/* XXX inp locking */
178 		if ((in6p->inp_vflag & INP_IPV6) == 0)
179 			continue;
180 		if (in6p->inp_ip_p &&
181 		    in6p->inp_ip_p != proto)
182 			continue;
183 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
184 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
185 			continue;
186 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
187 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
188 			continue;
189 		if (jailed_without_vnet(in6p->inp_cred)) {
190 			/*
191 			 * Allow raw socket in jail to receive multicast;
192 			 * assume process had PRIV_NETINET_RAW at attach,
193 			 * and fall through into normal filter path if so.
194 			 */
195 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
196 			    prison_check_ip6(in6p->inp_cred,
197 			    &ip6->ip6_dst) != 0)
198 				continue;
199 		}
200 		INP_RLOCK(in6p);
201 		if (in6p->in6p_cksum != -1) {
202 			RIP6STAT_INC(rip6s_isum);
203 			if (in6_cksum(m, proto, *offp,
204 			    m->m_pkthdr.len - *offp)) {
205 				INP_RUNLOCK(in6p);
206 				RIP6STAT_INC(rip6s_badsum);
207 				continue;
208 			}
209 		}
210 		/*
211 		 * If this raw socket has multicast state, and we
212 		 * have received a multicast, check if this socket
213 		 * should receive it, as multicast filtering is now
214 		 * the responsibility of the transport layer.
215 		 */
216 		if (in6p->in6p_moptions &&
217 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
218 			/*
219 			 * If the incoming datagram is for MLD, allow it
220 			 * through unconditionally to the raw socket.
221 			 *
222 			 * Use the M_RTALERT_MLD flag to check for MLD
223 			 * traffic without having to inspect the mbuf chain
224 			 * more deeply, as all MLDv1/v2 host messages MUST
225 			 * contain the Router Alert option.
226 			 *
227 			 * In the case of MLDv1, we may not have explicitly
228 			 * joined the group, and may have set IFF_ALLMULTI
229 			 * on the interface. im6o_mc_filter() may discard
230 			 * control traffic we actually need to see.
231 			 *
232 			 * Userland multicast routing daemons should continue
233 			 * filter the control traffic appropriately.
234 			 */
235 			int blocked;
236 
237 			blocked = MCAST_PASS;
238 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
239 				struct sockaddr_in6 mcaddr;
240 
241 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
242 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
243 				mcaddr.sin6_family = AF_INET6;
244 				mcaddr.sin6_addr = ip6->ip6_dst;
245 
246 				blocked = im6o_mc_filter(in6p->in6p_moptions,
247 				    ifp,
248 				    (struct sockaddr *)&mcaddr,
249 				    (struct sockaddr *)&fromsa);
250 			}
251 			if (blocked != MCAST_PASS) {
252 				IP6STAT_INC(ip6s_notmember);
253 				INP_RUNLOCK(in6p);
254 				continue;
255 			}
256 		}
257 		if (last != NULL) {
258 			struct mbuf *n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
259 
260 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
261 			/*
262 			 * Check AH/ESP integrity.
263 			 */
264 			if (IPSEC_ENABLED(ipv6)) {
265 				if (n != NULL &&
266 				    IPSEC_CHECK_POLICY(ipv6, n, last) != 0) {
267 					m_freem(n);
268 					/* Do not inject data into pcb. */
269 					n = NULL;
270 				}
271 			}
272 #endif /* IPSEC */
273 			if (n) {
274 				if (last->inp_flags & INP_CONTROLOPTS ||
275 				    last->inp_socket->so_options & SO_TIMESTAMP)
276 					ip6_savecontrol(last, n, &opts);
277 				/* strip intermediate headers */
278 				m_adj(n, *offp);
279 				if (sbappendaddr(&last->inp_socket->so_rcv,
280 						(struct sockaddr *)&fromsa,
281 						 n, opts) == 0) {
282 					m_freem(n);
283 					if (opts)
284 						m_freem(opts);
285 					RIP6STAT_INC(rip6s_fullsock);
286 				} else
287 					sorwakeup(last->inp_socket);
288 				opts = NULL;
289 			}
290 			INP_RUNLOCK(last);
291 		}
292 		last = in6p;
293 	}
294 	INP_INFO_RUNLOCK(&V_ripcbinfo);
295 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
296 	/*
297 	 * Check AH/ESP integrity.
298 	 */
299 	if (IPSEC_ENABLED(ipv6) && last != NULL &&
300 	    IPSEC_CHECK_POLICY(ipv6, m, last) != 0) {
301 		m_freem(m);
302 		IP6STAT_DEC(ip6s_delivered);
303 		/* Do not inject data into pcb. */
304 		INP_RUNLOCK(last);
305 	} else
306 #endif /* IPSEC */
307 	if (last != NULL) {
308 		if (last->inp_flags & INP_CONTROLOPTS ||
309 		    last->inp_socket->so_options & SO_TIMESTAMP)
310 			ip6_savecontrol(last, m, &opts);
311 		/* Strip intermediate headers. */
312 		m_adj(m, *offp);
313 		if (sbappendaddr(&last->inp_socket->so_rcv,
314 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
315 			m_freem(m);
316 			if (opts)
317 				m_freem(opts);
318 			RIP6STAT_INC(rip6s_fullsock);
319 		} else
320 			sorwakeup(last->inp_socket);
321 		INP_RUNLOCK(last);
322 	} else {
323 		RIP6STAT_INC(rip6s_nosock);
324 		if (m->m_flags & M_MCAST)
325 			RIP6STAT_INC(rip6s_nosockmcast);
326 		if (proto == IPPROTO_NONE)
327 			m_freem(m);
328 		else {
329 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
330 			icmp6_error(m, ICMP6_PARAM_PROB,
331 			    ICMP6_PARAMPROB_NEXTHEADER,
332 			    prvnxtp - mtod(m, char *));
333 		}
334 		IP6STAT_DEC(ip6s_delivered);
335 	}
336 	return (IPPROTO_DONE);
337 }
338 
339 void
340 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
341 {
342 	struct ip6_hdr *ip6;
343 	struct mbuf *m;
344 	int off = 0;
345 	struct ip6ctlparam *ip6cp = NULL;
346 	const struct sockaddr_in6 *sa6_src = NULL;
347 	void *cmdarg;
348 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
349 
350 	if (sa->sa_family != AF_INET6 ||
351 	    sa->sa_len != sizeof(struct sockaddr_in6))
352 		return;
353 
354 	if ((unsigned)cmd >= PRC_NCMDS)
355 		return;
356 	if (PRC_IS_REDIRECT(cmd))
357 		notify = in6_rtchange, d = NULL;
358 	else if (cmd == PRC_HOSTDEAD)
359 		d = NULL;
360 	else if (inet6ctlerrmap[cmd] == 0)
361 		return;
362 
363 	/*
364 	 * If the parameter is from icmp6, decode it.
365 	 */
366 	if (d != NULL) {
367 		ip6cp = (struct ip6ctlparam *)d;
368 		m = ip6cp->ip6c_m;
369 		ip6 = ip6cp->ip6c_ip6;
370 		off = ip6cp->ip6c_off;
371 		cmdarg = ip6cp->ip6c_cmdarg;
372 		sa6_src = ip6cp->ip6c_src;
373 	} else {
374 		m = NULL;
375 		ip6 = NULL;
376 		cmdarg = NULL;
377 		sa6_src = &sa6_any;
378 	}
379 
380 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
381 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
382 }
383 
384 /*
385  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
386  * may have setup with control call.
387  */
388 int
389 rip6_output(struct mbuf *m, struct socket *so, ...)
390 {
391 	struct mbuf *control;
392 	struct m_tag *mtag;
393 	struct sockaddr_in6 *dstsock;
394 	struct in6_addr *dst;
395 	struct ip6_hdr *ip6;
396 	struct inpcb *in6p;
397 	u_int	plen = m->m_pkthdr.len;
398 	int error = 0;
399 	struct ip6_pktopts opt, *optp;
400 	struct ifnet *oifp = NULL;
401 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
402 	int scope_ambiguous = 0;
403 	int use_defzone = 0;
404 	int hlim = 0;
405 	struct in6_addr in6a;
406 	va_list ap;
407 
408 	va_start(ap, so);
409 	dstsock = va_arg(ap, struct sockaddr_in6 *);
410 	control = va_arg(ap, struct mbuf *);
411 	va_end(ap);
412 
413 	in6p = sotoinpcb(so);
414 	INP_WLOCK(in6p);
415 
416 	dst = &dstsock->sin6_addr;
417 	if (control != NULL) {
418 		if ((error = ip6_setpktopts(control, &opt,
419 		    in6p->in6p_outputopts, so->so_cred,
420 		    so->so_proto->pr_protocol)) != 0) {
421 			goto bad;
422 		}
423 		optp = &opt;
424 	} else
425 		optp = in6p->in6p_outputopts;
426 
427 	/*
428 	 * Check and convert scope zone ID into internal form.
429 	 *
430 	 * XXX: we may still need to determine the zone later.
431 	 */
432 	if (!(so->so_state & SS_ISCONNECTED)) {
433 		if (!optp || !optp->ip6po_pktinfo ||
434 		    !optp->ip6po_pktinfo->ipi6_ifindex)
435 			use_defzone = V_ip6_use_defzone;
436 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
437 			scope_ambiguous = 1;
438 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
439 			goto bad;
440 	}
441 
442 	/*
443 	 * For an ICMPv6 packet, we should know its type and code to update
444 	 * statistics.
445 	 */
446 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
447 		struct icmp6_hdr *icmp6;
448 		if (m->m_len < sizeof(struct icmp6_hdr) &&
449 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
450 			error = ENOBUFS;
451 			goto bad;
452 		}
453 		icmp6 = mtod(m, struct icmp6_hdr *);
454 		type = icmp6->icmp6_type;
455 		code = icmp6->icmp6_code;
456 	}
457 
458 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
459 	if (m == NULL) {
460 		error = ENOBUFS;
461 		goto bad;
462 	}
463 	ip6 = mtod(m, struct ip6_hdr *);
464 
465 	/*
466 	 * Source address selection.
467 	 */
468 	error = in6_selectsrc_socket(dstsock, optp, in6p, so->so_cred,
469 	    scope_ambiguous, &in6a, &hlim);
470 
471 	if (error)
472 		goto bad;
473 	error = prison_check_ip6(in6p->inp_cred, &in6a);
474 	if (error != 0)
475 		goto bad;
476 	ip6->ip6_src = in6a;
477 
478 	ip6->ip6_dst = dstsock->sin6_addr;
479 
480 	/*
481 	 * Fill in the rest of the IPv6 header fields.
482 	 */
483 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
484 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
485 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
486 	    (IPV6_VERSION & IPV6_VERSION_MASK);
487 
488 	/*
489 	 * ip6_plen will be filled in ip6_output, so not fill it here.
490 	 */
491 	ip6->ip6_nxt = in6p->inp_ip_p;
492 	ip6->ip6_hlim = hlim;
493 
494 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
495 	    in6p->in6p_cksum != -1) {
496 		struct mbuf *n;
497 		int off;
498 		u_int16_t *p;
499 
500 		/* Compute checksum. */
501 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
502 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
503 		else
504 			off = in6p->in6p_cksum;
505 		if (plen < off + 1) {
506 			error = EINVAL;
507 			goto bad;
508 		}
509 		off += sizeof(struct ip6_hdr);
510 
511 		n = m;
512 		while (n && n->m_len <= off) {
513 			off -= n->m_len;
514 			n = n->m_next;
515 		}
516 		if (!n)
517 			goto bad;
518 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
519 		*p = 0;
520 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
521 	}
522 
523 	/*
524 	 * Send RA/RS messages to user land for protection, before sending
525 	 * them to rtadvd/rtsol.
526 	 */
527 	if ((send_sendso_input_hook != NULL) &&
528 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
529 		switch (type) {
530 		case ND_ROUTER_ADVERT:
531 		case ND_ROUTER_SOLICIT:
532 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
533 				sizeof(unsigned short), M_NOWAIT);
534 			if (mtag == NULL)
535 				goto bad;
536 			m_tag_prepend(m, mtag);
537 		}
538 	}
539 
540 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
541 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
542 		if (oifp)
543 			icmp6_ifoutstat_inc(oifp, type, code);
544 		ICMP6STAT_INC(icp6s_outhist[type]);
545 	} else
546 		RIP6STAT_INC(rip6s_opackets);
547 
548 	goto freectl;
549 
550  bad:
551 	if (m)
552 		m_freem(m);
553 
554  freectl:
555 	if (control != NULL) {
556 		ip6_clearpktopts(&opt, -1);
557 		m_freem(control);
558 	}
559 	INP_WUNLOCK(in6p);
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;
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 = sotoinpcb(so);
582 			INP_WLOCK(inp);
583 			inp->inp_inc.inc_fibnum = so->so_fibnum;
584 			INP_WUNLOCK(inp);
585 			return (0);
586 		}
587 		return (EINVAL);
588 	}
589 
590 	error = 0;
591 
592 	switch (sopt->sopt_dir) {
593 	case SOPT_GET:
594 		switch (sopt->sopt_name) {
595 		case MRT6_INIT:
596 		case MRT6_DONE:
597 		case MRT6_ADD_MIF:
598 		case MRT6_DEL_MIF:
599 		case MRT6_ADD_MFC:
600 		case MRT6_DEL_MFC:
601 		case MRT6_PIM:
602 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
603 			    EOPNOTSUPP;
604 			break;
605 		case IPV6_CHECKSUM:
606 			error = ip6_raw_ctloutput(so, sopt);
607 			break;
608 		default:
609 			error = ip6_ctloutput(so, sopt);
610 			break;
611 		}
612 		break;
613 
614 	case SOPT_SET:
615 		switch (sopt->sopt_name) {
616 		case MRT6_INIT:
617 		case MRT6_DONE:
618 		case MRT6_ADD_MIF:
619 		case MRT6_DEL_MIF:
620 		case MRT6_ADD_MFC:
621 		case MRT6_DEL_MFC:
622 		case MRT6_PIM:
623 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
624 			    EOPNOTSUPP;
625 			break;
626 		case IPV6_CHECKSUM:
627 			error = ip6_raw_ctloutput(so, sopt);
628 			break;
629 		default:
630 			error = ip6_ctloutput(so, sopt);
631 			break;
632 		}
633 		break;
634 	}
635 
636 	return (error);
637 }
638 
639 static int
640 rip6_attach(struct socket *so, int proto, struct thread *td)
641 {
642 	struct inpcb *inp;
643 	struct icmp6_filter *filter;
644 	int error;
645 
646 	inp = sotoinpcb(so);
647 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
648 
649 	error = priv_check(td, PRIV_NETINET_RAW);
650 	if (error)
651 		return (error);
652 	error = soreserve(so, rip_sendspace, rip_recvspace);
653 	if (error)
654 		return (error);
655 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
656 	if (filter == NULL)
657 		return (ENOMEM);
658 	INP_INFO_WLOCK(&V_ripcbinfo);
659 	error = in_pcballoc(so, &V_ripcbinfo);
660 	if (error) {
661 		INP_INFO_WUNLOCK(&V_ripcbinfo);
662 		free(filter, M_PCB);
663 		return (error);
664 	}
665 	inp = (struct inpcb *)so->so_pcb;
666 	INP_INFO_WUNLOCK(&V_ripcbinfo);
667 	inp->inp_vflag |= INP_IPV6;
668 	inp->inp_ip_p = (long)proto;
669 	inp->in6p_hops = -1;	/* use kernel default */
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_INFO_WLOCK(&V_ripcbinfo);
689 	INP_WLOCK(inp);
690 	free(inp->in6p_icmp6filt, M_PCB);
691 	in_pcbdetach(inp);
692 	in_pcbfree(inp);
693 	INP_INFO_WUNLOCK(&V_ripcbinfo);
694 }
695 
696 /* XXXRW: This can't ever be called. */
697 static void
698 rip6_abort(struct socket *so)
699 {
700 	struct inpcb *inp;
701 
702 	inp = sotoinpcb(so);
703 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
704 
705 	soisdisconnected(so);
706 }
707 
708 static void
709 rip6_close(struct socket *so)
710 {
711 	struct inpcb *inp;
712 
713 	inp = sotoinpcb(so);
714 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
715 
716 	soisdisconnected(so);
717 }
718 
719 static int
720 rip6_disconnect(struct socket *so)
721 {
722 	struct inpcb *inp;
723 
724 	inp = sotoinpcb(so);
725 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
726 
727 	if ((so->so_state & SS_ISCONNECTED) == 0)
728 		return (ENOTCONN);
729 	inp->in6p_faddr = in6addr_any;
730 	rip6_abort(so);
731 	return (0);
732 }
733 
734 static int
735 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
736 {
737 	struct inpcb *inp;
738 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
739 	struct ifaddr *ifa = NULL;
740 	int error = 0;
741 
742 	inp = sotoinpcb(so);
743 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
744 
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 (TAILQ_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 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
755 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
756 		return (EADDRNOTAVAIL);
757 	if (ifa != NULL &&
758 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
759 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
760 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
761 		ifa_free(ifa);
762 		return (EADDRNOTAVAIL);
763 	}
764 	if (ifa != NULL)
765 		ifa_free(ifa);
766 	INP_INFO_WLOCK(&V_ripcbinfo);
767 	INP_WLOCK(inp);
768 	inp->in6p_laddr = addr->sin6_addr;
769 	INP_WUNLOCK(inp);
770 	INP_INFO_WUNLOCK(&V_ripcbinfo);
771 	return (0);
772 }
773 
774 static int
775 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
776 {
777 	struct inpcb *inp;
778 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
779 	struct in6_addr in6a;
780 	int error = 0, scope_ambiguous = 0;
781 
782 	inp = sotoinpcb(so);
783 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
784 
785 	if (nam->sa_len != sizeof(*addr))
786 		return (EINVAL);
787 	if (TAILQ_EMPTY(&V_ifnet))
788 		return (EADDRNOTAVAIL);
789 	if (addr->sin6_family != AF_INET6)
790 		return (EAFNOSUPPORT);
791 
792 	/*
793 	 * Application should provide a proper zone ID or the use of default
794 	 * zone IDs should be enabled.  Unfortunately, some applications do
795 	 * not behave as it should, so we need a workaround.  Even if an
796 	 * appropriate ID is not determined, we'll see if we can determine
797 	 * the outgoing interface.  If we can, determine the zone ID based on
798 	 * the interface below.
799 	 */
800 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
801 		scope_ambiguous = 1;
802 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
803 		return (error);
804 
805 	INP_INFO_WLOCK(&V_ripcbinfo);
806 	INP_WLOCK(inp);
807 	/* Source address selection. XXX: need pcblookup? */
808 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
809 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
810 	if (error) {
811 		INP_WUNLOCK(inp);
812 		INP_INFO_WUNLOCK(&V_ripcbinfo);
813 		return (error);
814 	}
815 
816 	inp->in6p_faddr = addr->sin6_addr;
817 	inp->in6p_laddr = in6a;
818 	soisconnected(so);
819 	INP_WUNLOCK(inp);
820 	INP_INFO_WUNLOCK(&V_ripcbinfo);
821 	return (0);
822 }
823 
824 static int
825 rip6_shutdown(struct socket *so)
826 {
827 	struct inpcb *inp;
828 
829 	inp = sotoinpcb(so);
830 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
831 
832 	INP_WLOCK(inp);
833 	socantsendmore(so);
834 	INP_WUNLOCK(inp);
835 	return (0);
836 }
837 
838 static int
839 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
840     struct mbuf *control, struct thread *td)
841 {
842 	struct inpcb *inp;
843 	struct sockaddr_in6 tmp;
844 	struct sockaddr_in6 *dst;
845 	int ret;
846 
847 	inp = sotoinpcb(so);
848 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
849 
850 	/* Always copy sockaddr to avoid overwrites. */
851 	/* Unlocked read. */
852 	if (so->so_state & SS_ISCONNECTED) {
853 		if (nam) {
854 			m_freem(m);
855 			return (EISCONN);
856 		}
857 		/* XXX */
858 		bzero(&tmp, sizeof(tmp));
859 		tmp.sin6_family = AF_INET6;
860 		tmp.sin6_len = sizeof(struct sockaddr_in6);
861 		INP_RLOCK(inp);
862 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
863 		    sizeof(struct in6_addr));
864 		INP_RUNLOCK(inp);
865 		dst = &tmp;
866 	} else {
867 		if (nam == NULL) {
868 			m_freem(m);
869 			return (ENOTCONN);
870 		}
871 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
872 			m_freem(m);
873 			return (EINVAL);
874 		}
875 		tmp = *(struct sockaddr_in6 *)nam;
876 		dst = &tmp;
877 
878 		if (dst->sin6_family == AF_UNSPEC) {
879 			/*
880 			 * XXX: we allow this case for backward
881 			 * compatibility to buggy applications that
882 			 * rely on old (and wrong) kernel behavior.
883 			 */
884 			log(LOG_INFO, "rip6 SEND: address family is "
885 			    "unspec. Assume AF_INET6\n");
886 			dst->sin6_family = AF_INET6;
887 		} else if (dst->sin6_family != AF_INET6) {
888 			m_freem(m);
889 			return(EAFNOSUPPORT);
890 		}
891 	}
892 	ret = rip6_output(m, so, dst, control);
893 	return (ret);
894 }
895 
896 struct pr_usrreqs rip6_usrreqs = {
897 	.pru_abort =		rip6_abort,
898 	.pru_attach =		rip6_attach,
899 	.pru_bind =		rip6_bind,
900 	.pru_connect =		rip6_connect,
901 	.pru_control =		in6_control,
902 	.pru_detach =		rip6_detach,
903 	.pru_disconnect =	rip6_disconnect,
904 	.pru_peeraddr =		in6_getpeeraddr,
905 	.pru_send =		rip6_send,
906 	.pru_shutdown =		rip6_shutdown,
907 	.pru_sockaddr =		in6_getsockaddr,
908 	.pru_close =		rip6_close,
909 };
910