xref: /freebsd/sys/netinet6/raw_ip6.c (revision 7aa383846770374466b1dcb2cefd71bde9acf463)
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/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_types.h>
85 #include <net/route.h>
86 #include <net/vnet.h>
87 
88 #include <netinet/in.h>
89 #include <netinet/in_var.h>
90 #include <netinet/in_systm.h>
91 #include <netinet/in_pcb.h>
92 
93 #include <netinet/icmp6.h>
94 #include <netinet/ip6.h>
95 #include <netinet6/ip6protosw.h>
96 #include <netinet6/ip6_mroute.h>
97 #include <netinet6/in6_pcb.h>
98 #include <netinet6/ip6_var.h>
99 #include <netinet6/nd6.h>
100 #include <netinet6/raw_ip6.h>
101 #include <netinet6/scope6_var.h>
102 
103 #ifdef IPSEC
104 #include <netipsec/ipsec.h>
105 #include <netipsec/ipsec6.h>
106 #endif /* IPSEC */
107 
108 #include <machine/stdarg.h>
109 
110 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
111 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
112 
113 /*
114  * Raw interface to IP6 protocol.
115  */
116 
117 VNET_DECLARE(struct inpcbhead, ripcb);
118 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
119 #define	V_ripcb				VNET(ripcb)
120 #define	V_ripcbinfo			VNET(ripcbinfo)
121 
122 extern u_long	rip_sendspace;
123 extern u_long	rip_recvspace;
124 
125 VNET_DEFINE(struct rip6stat, rip6stat);
126 
127 /*
128  * Hooks for multicast routing. They all default to NULL, so leave them not
129  * initialized and rely on BSS being set to 0.
130  */
131 
132 /*
133  * The socket used to communicate with the multicast routing daemon.
134  */
135 VNET_DEFINE(struct socket *, ip6_mrouter);
136 
137 /*
138  * The various mrouter functions.
139  */
140 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
141 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
142 int (*ip6_mrouter_done)(void);
143 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
144 int (*mrt6_ioctl)(u_long, caddr_t);
145 
146 /*
147  * Setup generic address and protocol structures for raw_input routine, then
148  * pass them along with mbuf chain.
149  */
150 int
151 rip6_input(struct mbuf **mp, int *offp, int proto)
152 {
153 	struct ifnet *ifp;
154 	struct mbuf *m = *mp;
155 	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
156 	register struct inpcb *in6p;
157 	struct inpcb *last = 0;
158 	struct mbuf *opts = NULL;
159 	struct sockaddr_in6 fromsa;
160 
161 	V_rip6stat.rip6s_ipackets++;
162 
163 	if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
164 		/* XXX Send icmp6 host/port unreach? */
165 		m_freem(m);
166 		return (IPPROTO_DONE);
167 	}
168 
169 	init_sin6(&fromsa, m); /* general init */
170 
171 	ifp = m->m_pkthdr.rcvif;
172 
173 	INP_INFO_RLOCK(&V_ripcbinfo);
174 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
175 		/* XXX inp locking */
176 		if ((in6p->inp_vflag & INP_IPV6) == 0)
177 			continue;
178 		if (in6p->inp_ip_p &&
179 		    in6p->inp_ip_p != proto)
180 			continue;
181 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
182 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
183 			continue;
184 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
185 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
186 			continue;
187 		if (jailed_without_vnet(in6p->inp_cred)) {
188 			/*
189 			 * Allow raw socket in jail to receive multicast;
190 			 * assume process had PRIV_NETINET_RAW at attach,
191 			 * and fall through into normal filter path if so.
192 			 */
193 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
194 			    prison_check_ip6(in6p->inp_cred,
195 			    &ip6->ip6_dst) != 0)
196 				continue;
197 		}
198 		if (in6p->in6p_cksum != -1) {
199 			V_rip6stat.rip6s_isum++;
200 			if (in6_cksum(m, proto, *offp,
201 			    m->m_pkthdr.len - *offp)) {
202 				INP_RUNLOCK(in6p);
203 				V_rip6stat.rip6s_badsum++;
204 				continue;
205 			}
206 		}
207 		INP_RLOCK(in6p);
208 		/*
209 		 * If this raw socket has multicast state, and we
210 		 * have received a multicast, check if this socket
211 		 * should receive it, as multicast filtering is now
212 		 * the responsibility of the transport layer.
213 		 */
214 		if (in6p->in6p_moptions &&
215 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
216 			/*
217 			 * If the incoming datagram is for MLD, allow it
218 			 * through unconditionally to the raw socket.
219 			 *
220 			 * Use the M_RTALERT_MLD flag to check for MLD
221 			 * traffic without having to inspect the mbuf chain
222 			 * more deeply, as all MLDv1/v2 host messages MUST
223 			 * contain the Router Alert option.
224 			 *
225 			 * In the case of MLDv1, we may not have explicitly
226 			 * joined the group, and may have set IFF_ALLMULTI
227 			 * on the interface. im6o_mc_filter() may discard
228 			 * control traffic we actually need to see.
229 			 *
230 			 * Userland multicast routing daemons should continue
231 			 * filter the control traffic appropriately.
232 			 */
233 			int blocked;
234 
235 			blocked = MCAST_PASS;
236 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
237 				struct sockaddr_in6 mcaddr;
238 
239 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
240 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
241 				mcaddr.sin6_family = AF_INET6;
242 				mcaddr.sin6_addr = ip6->ip6_dst;
243 
244 				blocked = im6o_mc_filter(in6p->in6p_moptions,
245 				    ifp,
246 				    (struct sockaddr *)&mcaddr,
247 				    (struct sockaddr *)&fromsa);
248 			}
249 			if (blocked != MCAST_PASS) {
250 				IP6STAT_INC(ip6s_notmember);
251 				continue;
252 			}
253 		}
254 		if (last != NULL) {
255 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
256 
257 #ifdef IPSEC
258 			/*
259 			 * Check AH/ESP integrity.
260 			 */
261 			if (n && ipsec6_in_reject(n, last)) {
262 				m_freem(n);
263 				V_ipsec6stat.in_polvio++;
264 				/* Do not inject data into pcb. */
265 			} else
266 #endif /* IPSEC */
267 			if (n) {
268 				if (last->inp_flags & INP_CONTROLOPTS ||
269 				    last->inp_socket->so_options & SO_TIMESTAMP)
270 					ip6_savecontrol(last, n, &opts);
271 				/* strip intermediate headers */
272 				m_adj(n, *offp);
273 				if (sbappendaddr(&last->inp_socket->so_rcv,
274 						(struct sockaddr *)&fromsa,
275 						 n, opts) == 0) {
276 					m_freem(n);
277 					if (opts)
278 						m_freem(opts);
279 					V_rip6stat.rip6s_fullsock++;
280 				} else
281 					sorwakeup(last->inp_socket);
282 				opts = NULL;
283 			}
284 			INP_RUNLOCK(last);
285 		}
286 		last = in6p;
287 	}
288 	INP_INFO_RUNLOCK(&V_ripcbinfo);
289 #ifdef IPSEC
290 	/*
291 	 * Check AH/ESP integrity.
292 	 */
293 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
294 		m_freem(m);
295 		V_ipsec6stat.in_polvio++;
296 		V_ip6stat.ip6s_delivered--;
297 		/* Do not inject data into pcb. */
298 		INP_RUNLOCK(last);
299 	} else
300 #endif /* IPSEC */
301 	if (last != NULL) {
302 		if (last->inp_flags & INP_CONTROLOPTS ||
303 		    last->inp_socket->so_options & SO_TIMESTAMP)
304 			ip6_savecontrol(last, m, &opts);
305 		/* Strip intermediate headers. */
306 		m_adj(m, *offp);
307 		if (sbappendaddr(&last->inp_socket->so_rcv,
308 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
309 			m_freem(m);
310 			if (opts)
311 				m_freem(opts);
312 			V_rip6stat.rip6s_fullsock++;
313 		} else
314 			sorwakeup(last->inp_socket);
315 		INP_RUNLOCK(last);
316 	} else {
317 		V_rip6stat.rip6s_nosock++;
318 		if (m->m_flags & M_MCAST)
319 			V_rip6stat.rip6s_nosockmcast++;
320 		if (proto == IPPROTO_NONE)
321 			m_freem(m);
322 		else {
323 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
324 			icmp6_error(m, ICMP6_PARAM_PROB,
325 			    ICMP6_PARAMPROB_NEXTHEADER,
326 			    prvnxtp - mtod(m, char *));
327 		}
328 		V_ip6stat.ip6s_delivered--;
329 	}
330 	return (IPPROTO_DONE);
331 }
332 
333 void
334 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
335 {
336 	struct ip6_hdr *ip6;
337 	struct mbuf *m;
338 	int off = 0;
339 	struct ip6ctlparam *ip6cp = NULL;
340 	const struct sockaddr_in6 *sa6_src = NULL;
341 	void *cmdarg;
342 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
343 
344 	if (sa->sa_family != AF_INET6 ||
345 	    sa->sa_len != sizeof(struct sockaddr_in6))
346 		return;
347 
348 	if ((unsigned)cmd >= PRC_NCMDS)
349 		return;
350 	if (PRC_IS_REDIRECT(cmd))
351 		notify = in6_rtchange, d = NULL;
352 	else if (cmd == PRC_HOSTDEAD)
353 		d = NULL;
354 	else if (inet6ctlerrmap[cmd] == 0)
355 		return;
356 
357 	/*
358 	 * If the parameter is from icmp6, decode it.
359 	 */
360 	if (d != NULL) {
361 		ip6cp = (struct ip6ctlparam *)d;
362 		m = ip6cp->ip6c_m;
363 		ip6 = ip6cp->ip6c_ip6;
364 		off = ip6cp->ip6c_off;
365 		cmdarg = ip6cp->ip6c_cmdarg;
366 		sa6_src = ip6cp->ip6c_src;
367 	} else {
368 		m = NULL;
369 		ip6 = NULL;
370 		cmdarg = NULL;
371 		sa6_src = &sa6_any;
372 	}
373 
374 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
375 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
376 }
377 
378 /*
379  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
380  * may have setup with control call.
381  */
382 int
383 #if __STDC__
384 rip6_output(struct mbuf *m, ...)
385 #else
386 rip6_output(m, va_alist)
387 	struct mbuf *m;
388 	va_dcl
389 #endif
390 {
391 	struct mbuf *control;
392 	struct socket *so;
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 	struct in6_addr in6a;
404 	va_list ap;
405 
406 	va_start(ap, m);
407 	so = va_arg(ap, struct socket *);
408 	dstsock = va_arg(ap, struct sockaddr_in6 *);
409 	control = va_arg(ap, struct mbuf *);
410 	va_end(ap);
411 
412 	in6p = sotoinpcb(so);
413 	INP_WLOCK(in6p);
414 
415 	dst = &dstsock->sin6_addr;
416 	if (control != NULL) {
417 		if ((error = ip6_setpktopts(control, &opt,
418 		    in6p->in6p_outputopts, so->so_cred,
419 		    so->so_proto->pr_protocol)) != 0) {
420 			goto bad;
421 		}
422 		optp = &opt;
423 	} else
424 		optp = in6p->in6p_outputopts;
425 
426 	/*
427 	 * Check and convert scope zone ID into internal form.
428 	 *
429 	 * XXX: we may still need to determine the zone later.
430 	 */
431 	if (!(so->so_state & SS_ISCONNECTED)) {
432 		if (dstsock->sin6_scope_id == 0 && !V_ip6_use_defzone)
433 			scope_ambiguous = 1;
434 		if ((error = sa6_embedscope(dstsock, V_ip6_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_DONTWAIT);
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(dstsock, optp, in6p, NULL, so->so_cred,
465 	    &oifp, &in6a);
466 	if (error)
467 		goto bad;
468 	error = prison_check_ip6(in6p->inp_cred, &in6a);
469 	if (error != 0)
470 		goto bad;
471 	ip6->ip6_src = in6a;
472 
473 	if (oifp && scope_ambiguous) {
474 		/*
475 		 * Application should provide a proper zone ID or the use of
476 		 * default zone IDs should be enabled.  Unfortunately, some
477 		 * applications do not behave as it should, so we need a
478 		 * workaround.  Even if an appropriate ID is not determined
479 		 * (when it's required), if we can determine the outgoing
480 		 * interface. determine the zone ID based on the interface.
481 		 */
482 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
483 		if (error != 0)
484 			goto bad;
485 	}
486 	ip6->ip6_dst = dstsock->sin6_addr;
487 
488 	/*
489 	 * Fill in the rest of the IPv6 header fields.
490 	 */
491 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
492 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
493 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
494 	    (IPV6_VERSION & IPV6_VERSION_MASK);
495 
496 	/*
497 	 * ip6_plen will be filled in ip6_output, so not fill it here.
498 	 */
499 	ip6->ip6_nxt = in6p->inp_ip_p;
500 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
501 
502 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
503 	    in6p->in6p_cksum != -1) {
504 		struct mbuf *n;
505 		int off;
506 		u_int16_t *p;
507 
508 		/* Compute checksum. */
509 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
510 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
511 		else
512 			off = in6p->in6p_cksum;
513 		if (plen < off + 1) {
514 			error = EINVAL;
515 			goto bad;
516 		}
517 		off += sizeof(struct ip6_hdr);
518 
519 		n = m;
520 		while (n && n->m_len <= off) {
521 			off -= n->m_len;
522 			n = n->m_next;
523 		}
524 		if (!n)
525 			goto bad;
526 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
527 		*p = 0;
528 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
529 	}
530 
531 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
532 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
533 		if (oifp)
534 			icmp6_ifoutstat_inc(oifp, type, code);
535 		ICMP6STAT_INC(icp6s_outhist[type]);
536 	} else
537 		V_rip6stat.rip6s_opackets++;
538 
539 	goto freectl;
540 
541  bad:
542 	if (m)
543 		m_freem(m);
544 
545  freectl:
546 	if (control != NULL) {
547 		ip6_clearpktopts(&opt, -1);
548 		m_freem(control);
549 	}
550 	INP_WUNLOCK(in6p);
551 	return (error);
552 }
553 
554 /*
555  * Raw IPv6 socket option processing.
556  */
557 int
558 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
559 {
560 	int error;
561 
562 	if (sopt->sopt_level == IPPROTO_ICMPV6)
563 		/*
564 		 * XXX: is it better to call icmp6_ctloutput() directly
565 		 * from protosw?
566 		 */
567 		return (icmp6_ctloutput(so, sopt));
568 	else if (sopt->sopt_level != IPPROTO_IPV6)
569 		return (EINVAL);
570 
571 	error = 0;
572 
573 	switch (sopt->sopt_dir) {
574 	case SOPT_GET:
575 		switch (sopt->sopt_name) {
576 		case MRT6_INIT:
577 		case MRT6_DONE:
578 		case MRT6_ADD_MIF:
579 		case MRT6_DEL_MIF:
580 		case MRT6_ADD_MFC:
581 		case MRT6_DEL_MFC:
582 		case MRT6_PIM:
583 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
584 			    EOPNOTSUPP;
585 			break;
586 		case IPV6_CHECKSUM:
587 			error = ip6_raw_ctloutput(so, sopt);
588 			break;
589 		default:
590 			error = ip6_ctloutput(so, sopt);
591 			break;
592 		}
593 		break;
594 
595 	case SOPT_SET:
596 		switch (sopt->sopt_name) {
597 		case MRT6_INIT:
598 		case MRT6_DONE:
599 		case MRT6_ADD_MIF:
600 		case MRT6_DEL_MIF:
601 		case MRT6_ADD_MFC:
602 		case MRT6_DEL_MFC:
603 		case MRT6_PIM:
604 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
605 			    EOPNOTSUPP;
606 			break;
607 		case IPV6_CHECKSUM:
608 			error = ip6_raw_ctloutput(so, sopt);
609 			break;
610 		default:
611 			error = ip6_ctloutput(so, sopt);
612 			break;
613 		}
614 		break;
615 	}
616 
617 	return (error);
618 }
619 
620 static int
621 rip6_attach(struct socket *so, int proto, struct thread *td)
622 {
623 	struct inpcb *inp;
624 	struct icmp6_filter *filter;
625 	int error;
626 
627 	inp = sotoinpcb(so);
628 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
629 
630 	error = priv_check(td, PRIV_NETINET_RAW);
631 	if (error)
632 		return (error);
633 	error = soreserve(so, rip_sendspace, rip_recvspace);
634 	if (error)
635 		return (error);
636 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
637 	if (filter == NULL)
638 		return (ENOMEM);
639 	INP_INFO_WLOCK(&V_ripcbinfo);
640 	error = in_pcballoc(so, &V_ripcbinfo);
641 	if (error) {
642 		INP_INFO_WUNLOCK(&V_ripcbinfo);
643 		free(filter, M_PCB);
644 		return (error);
645 	}
646 	inp = (struct inpcb *)so->so_pcb;
647 	INP_INFO_WUNLOCK(&V_ripcbinfo);
648 	inp->inp_vflag |= INP_IPV6;
649 	inp->inp_ip_p = (long)proto;
650 	inp->in6p_hops = -1;	/* use kernel default */
651 	inp->in6p_cksum = -1;
652 	inp->in6p_icmp6filt = filter;
653 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
654 	INP_WUNLOCK(inp);
655 	return (0);
656 }
657 
658 static void
659 rip6_detach(struct socket *so)
660 {
661 	struct inpcb *inp;
662 
663 	inp = sotoinpcb(so);
664 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
665 
666 	if (so == V_ip6_mrouter && ip6_mrouter_done)
667 		ip6_mrouter_done();
668 	/* xxx: RSVP */
669 	INP_INFO_WLOCK(&V_ripcbinfo);
670 	INP_WLOCK(inp);
671 	free(inp->in6p_icmp6filt, M_PCB);
672 	in_pcbdetach(inp);
673 	in_pcbfree(inp);
674 	INP_INFO_WUNLOCK(&V_ripcbinfo);
675 }
676 
677 /* XXXRW: This can't ever be called. */
678 static void
679 rip6_abort(struct socket *so)
680 {
681 	struct inpcb *inp;
682 
683 	inp = sotoinpcb(so);
684 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
685 
686 	soisdisconnected(so);
687 }
688 
689 static void
690 rip6_close(struct socket *so)
691 {
692 	struct inpcb *inp;
693 
694 	inp = sotoinpcb(so);
695 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
696 
697 	soisdisconnected(so);
698 }
699 
700 static int
701 rip6_disconnect(struct socket *so)
702 {
703 	struct inpcb *inp;
704 
705 	inp = sotoinpcb(so);
706 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
707 
708 	if ((so->so_state & SS_ISCONNECTED) == 0)
709 		return (ENOTCONN);
710 	inp->in6p_faddr = in6addr_any;
711 	rip6_abort(so);
712 	return (0);
713 }
714 
715 static int
716 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
717 {
718 	struct inpcb *inp;
719 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
720 	struct ifaddr *ifa = NULL;
721 	int error = 0;
722 
723 	inp = sotoinpcb(so);
724 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
725 
726 	if (nam->sa_len != sizeof(*addr))
727 		return (EINVAL);
728 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
729 		return (error);
730 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
731 		return (EADDRNOTAVAIL);
732 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
733 		return (error);
734 
735 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
736 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
737 		return (EADDRNOTAVAIL);
738 	if (ifa != NULL &&
739 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
740 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
741 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
742 		ifa_free(ifa);
743 		return (EADDRNOTAVAIL);
744 	}
745 	if (ifa != NULL)
746 		ifa_free(ifa);
747 	INP_INFO_WLOCK(&V_ripcbinfo);
748 	INP_WLOCK(inp);
749 	inp->in6p_laddr = addr->sin6_addr;
750 	INP_WUNLOCK(inp);
751 	INP_INFO_WUNLOCK(&V_ripcbinfo);
752 	return (0);
753 }
754 
755 static int
756 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
757 {
758 	struct inpcb *inp;
759 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
760 	struct in6_addr in6a;
761 	struct ifnet *ifp = NULL;
762 	int error = 0, scope_ambiguous = 0;
763 
764 	inp = sotoinpcb(so);
765 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
766 
767 	if (nam->sa_len != sizeof(*addr))
768 		return (EINVAL);
769 	if (TAILQ_EMPTY(&V_ifnet))
770 		return (EADDRNOTAVAIL);
771 	if (addr->sin6_family != AF_INET6)
772 		return (EAFNOSUPPORT);
773 
774 	/*
775 	 * Application should provide a proper zone ID or the use of default
776 	 * zone IDs should be enabled.  Unfortunately, some applications do
777 	 * not behave as it should, so we need a workaround.  Even if an
778 	 * appropriate ID is not determined, we'll see if we can determine
779 	 * the outgoing interface.  If we can, determine the zone ID based on
780 	 * the interface below.
781 	 */
782 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
783 		scope_ambiguous = 1;
784 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
785 		return (error);
786 
787 	INP_INFO_WLOCK(&V_ripcbinfo);
788 	INP_WLOCK(inp);
789 	/* Source address selection. XXX: need pcblookup? */
790 	error = in6_selectsrc(addr, inp->in6p_outputopts,
791 	    inp, NULL, so->so_cred, &ifp, &in6a);
792 	if (error) {
793 		INP_WUNLOCK(inp);
794 		INP_INFO_WUNLOCK(&V_ripcbinfo);
795 		return (error);
796 	}
797 
798 	/* XXX: see above */
799 	if (ifp && scope_ambiguous &&
800 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
801 		INP_WUNLOCK(inp);
802 		INP_INFO_WUNLOCK(&V_ripcbinfo);
803 		return (error);
804 	}
805 	inp->in6p_faddr = addr->sin6_addr;
806 	inp->in6p_laddr = in6a;
807 	soisconnected(so);
808 	INP_WUNLOCK(inp);
809 	INP_INFO_WUNLOCK(&V_ripcbinfo);
810 	return (0);
811 }
812 
813 static int
814 rip6_shutdown(struct socket *so)
815 {
816 	struct inpcb *inp;
817 
818 	inp = sotoinpcb(so);
819 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
820 
821 	INP_WLOCK(inp);
822 	socantsendmore(so);
823 	INP_WUNLOCK(inp);
824 	return (0);
825 }
826 
827 static int
828 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
829     struct mbuf *control, struct thread *td)
830 {
831 	struct inpcb *inp;
832 	struct sockaddr_in6 tmp;
833 	struct sockaddr_in6 *dst;
834 	int ret;
835 
836 	inp = sotoinpcb(so);
837 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
838 
839 	/* Always copy sockaddr to avoid overwrites. */
840 	/* Unlocked read. */
841 	if (so->so_state & SS_ISCONNECTED) {
842 		if (nam) {
843 			m_freem(m);
844 			return (EISCONN);
845 		}
846 		/* XXX */
847 		bzero(&tmp, sizeof(tmp));
848 		tmp.sin6_family = AF_INET6;
849 		tmp.sin6_len = sizeof(struct sockaddr_in6);
850 		INP_RLOCK(inp);
851 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
852 		    sizeof(struct in6_addr));
853 		INP_RUNLOCK(inp);
854 		dst = &tmp;
855 	} else {
856 		if (nam == NULL) {
857 			m_freem(m);
858 			return (ENOTCONN);
859 		}
860 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
861 			m_freem(m);
862 			return (EINVAL);
863 		}
864 		tmp = *(struct sockaddr_in6 *)nam;
865 		dst = &tmp;
866 
867 		if (dst->sin6_family == AF_UNSPEC) {
868 			/*
869 			 * XXX: we allow this case for backward
870 			 * compatibility to buggy applications that
871 			 * rely on old (and wrong) kernel behavior.
872 			 */
873 			log(LOG_INFO, "rip6 SEND: address family is "
874 			    "unspec. Assume AF_INET6\n");
875 			dst->sin6_family = AF_INET6;
876 		} else if (dst->sin6_family != AF_INET6) {
877 			m_freem(m);
878 			return(EAFNOSUPPORT);
879 		}
880 	}
881 	ret = rip6_output(m, so, dst, control);
882 	return (ret);
883 }
884 
885 struct pr_usrreqs rip6_usrreqs = {
886 	.pru_abort =		rip6_abort,
887 	.pru_attach =		rip6_attach,
888 	.pru_bind =		rip6_bind,
889 	.pru_connect =		rip6_connect,
890 	.pru_control =		in6_control,
891 	.pru_detach =		rip6_detach,
892 	.pru_disconnect =	rip6_disconnect,
893 	.pru_peeraddr =		in6_getpeeraddr,
894 	.pru_send =		rip6_send,
895 	.pru_shutdown =		rip6_shutdown,
896 	.pru_sockaddr =		in6_getsockaddr,
897 	.pru_close =		rip6_close,
898 };
899