xref: /freebsd/sys/netinet6/raw_ip6.c (revision 5861f9665471e98e544f6fa3ce73c4912229ff82)
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 #include <sys/vimage.h>
83 
84 #include <net/if.h>
85 #include <net/if_types.h>
86 #include <net/route.h>
87 #include <net/vnet.h>
88 
89 #include <netinet/in.h>
90 #include <netinet/in_var.h>
91 #include <netinet/in_systm.h>
92 #include <netinet/in_pcb.h>
93 #include <netinet/vinet.h>
94 
95 #include <netinet/icmp6.h>
96 #include <netinet/ip6.h>
97 #include <netinet6/ip6protosw.h>
98 #include <netinet6/ip6_mroute.h>
99 #include <netinet6/in6_pcb.h>
100 #include <netinet6/ip6_var.h>
101 #include <netinet6/nd6.h>
102 #include <netinet6/raw_ip6.h>
103 #include <netinet6/scope6_var.h>
104 #include <netinet6/vinet6.h>
105 
106 #ifdef IPSEC
107 #include <netipsec/ipsec.h>
108 #include <netipsec/ipsec6.h>
109 #endif /* IPSEC */
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 #ifdef VIMAGE_GLOBALS
121 extern struct	inpcbhead ripcb;
122 extern struct	inpcbinfo ripcbinfo;
123 struct rip6stat rip6stat;
124 #endif
125 
126 extern u_long	rip_sendspace;
127 extern u_long	rip_recvspace;
128 
129 /*
130  * Hooks for multicast routing. They all default to NULL, so leave them not
131  * initialized and rely on BSS being set to 0.
132  */
133 
134 /*
135  * The socket used to communicate with the multicast routing daemon.
136  */
137 #ifdef VIMAGE_GLOBALS
138 struct socket *ip6_mrouter;
139 #endif
140 
141 /*
142  * The various mrouter functions.
143  */
144 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
145 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
146 int (*ip6_mrouter_done)(void);
147 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
148 int (*mrt6_ioctl)(u_long, caddr_t);
149 
150 /*
151  * Setup generic address and protocol structures for raw_input routine, then
152  * pass them along with mbuf chain.
153  */
154 int
155 rip6_input(struct mbuf **mp, int *offp, int proto)
156 {
157 	INIT_VNET_INET(curvnet);
158 	INIT_VNET_INET6(curvnet);
159 #ifdef IPSEC
160 	INIT_VNET_IPSEC(curvnet);
161 #endif
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 	V_rip6stat.rip6s_ipackets++;
171 
172 	if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
173 		/* XXX Send icmp6 host/port unreach? */
174 		m_freem(m);
175 		return (IPPROTO_DONE);
176 	}
177 
178 	init_sin6(&fromsa, m); /* general init */
179 
180 	ifp = m->m_pkthdr.rcvif;
181 
182 	INP_INFO_RLOCK(&V_ripcbinfo);
183 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
184 		/* XXX inp locking */
185 		if ((in6p->inp_vflag & INP_IPV6) == 0)
186 			continue;
187 		if (in6p->inp_ip_p &&
188 		    in6p->inp_ip_p != proto)
189 			continue;
190 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
191 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
192 			continue;
193 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
194 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
195 			continue;
196 		if (jailed(in6p->inp_cred)) {
197 			/*
198 			 * Allow raw socket in jail to receive multicast;
199 			 * assume process had PRIV_NETINET_RAW at attach,
200 			 * and fall through into normal filter path if so.
201 			 */
202 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
203 			    prison_check_ip6(in6p->inp_cred,
204 			    &ip6->ip6_dst) != 0)
205 				continue;
206 		}
207 		if (in6p->in6p_cksum != -1) {
208 			V_rip6stat.rip6s_isum++;
209 			if (in6_cksum(m, proto, *offp,
210 			    m->m_pkthdr.len - *offp)) {
211 				INP_RUNLOCK(in6p);
212 				V_rip6stat.rip6s_badsum++;
213 				continue;
214 			}
215 		}
216 		INP_RLOCK(in6p);
217 		/*
218 		 * If this raw socket has multicast state, and we
219 		 * have received a multicast, check if this socket
220 		 * should receive it, as multicast filtering is now
221 		 * the responsibility of the transport layer.
222 		 */
223 		if (in6p->in6p_moptions &&
224 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
225 			struct sockaddr_in6 mcaddr;
226 			int blocked;
227 
228 			bzero(&mcaddr, sizeof(struct sockaddr_in6));
229 			mcaddr.sin6_len = sizeof(struct sockaddr_in6);
230 			mcaddr.sin6_family = AF_INET6;
231 			mcaddr.sin6_addr = ip6->ip6_dst;
232 
233 			blocked = im6o_mc_filter(in6p->in6p_moptions, ifp,
234 			    (struct sockaddr *)&mcaddr,
235 			    (struct sockaddr *)&fromsa);
236 			if (blocked != MCAST_PASS) {
237 				IP6STAT_INC(ip6s_notmember);
238 				continue;
239 			}
240 		}
241 		if (last != NULL) {
242 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
243 
244 #ifdef IPSEC
245 			/*
246 			 * Check AH/ESP integrity.
247 			 */
248 			if (n && ipsec6_in_reject(n, last)) {
249 				m_freem(n);
250 				V_ipsec6stat.in_polvio++;
251 				/* Do not inject data into pcb. */
252 			} else
253 #endif /* IPSEC */
254 			if (n) {
255 				if (last->inp_flags & INP_CONTROLOPTS ||
256 				    last->inp_socket->so_options & SO_TIMESTAMP)
257 					ip6_savecontrol(last, n, &opts);
258 				/* strip intermediate headers */
259 				m_adj(n, *offp);
260 				if (sbappendaddr(&last->inp_socket->so_rcv,
261 						(struct sockaddr *)&fromsa,
262 						 n, opts) == 0) {
263 					m_freem(n);
264 					if (opts)
265 						m_freem(opts);
266 					V_rip6stat.rip6s_fullsock++;
267 				} else
268 					sorwakeup(last->inp_socket);
269 				opts = NULL;
270 			}
271 			INP_RUNLOCK(last);
272 		}
273 		last = in6p;
274 	}
275 	INP_INFO_RUNLOCK(&V_ripcbinfo);
276 #ifdef IPSEC
277 	/*
278 	 * Check AH/ESP integrity.
279 	 */
280 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
281 		m_freem(m);
282 		V_ipsec6stat.in_polvio++;
283 		V_ip6stat.ip6s_delivered--;
284 		/* Do not inject data into pcb. */
285 		INP_RUNLOCK(last);
286 	} else
287 #endif /* IPSEC */
288 	if (last != NULL) {
289 		if (last->inp_flags & INP_CONTROLOPTS ||
290 		    last->inp_socket->so_options & SO_TIMESTAMP)
291 			ip6_savecontrol(last, m, &opts);
292 		/* Strip intermediate headers. */
293 		m_adj(m, *offp);
294 		if (sbappendaddr(&last->inp_socket->so_rcv,
295 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
296 			m_freem(m);
297 			if (opts)
298 				m_freem(opts);
299 			V_rip6stat.rip6s_fullsock++;
300 		} else
301 			sorwakeup(last->inp_socket);
302 		INP_RUNLOCK(last);
303 	} else {
304 		V_rip6stat.rip6s_nosock++;
305 		if (m->m_flags & M_MCAST)
306 			V_rip6stat.rip6s_nosockmcast++;
307 		if (proto == IPPROTO_NONE)
308 			m_freem(m);
309 		else {
310 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
311 			icmp6_error(m, ICMP6_PARAM_PROB,
312 			    ICMP6_PARAMPROB_NEXTHEADER,
313 			    prvnxtp - mtod(m, char *));
314 		}
315 		V_ip6stat.ip6s_delivered--;
316 	}
317 	return (IPPROTO_DONE);
318 }
319 
320 void
321 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
322 {
323 	INIT_VNET_INET(curvnet);
324 	struct ip6_hdr *ip6;
325 	struct mbuf *m;
326 	int off = 0;
327 	struct ip6ctlparam *ip6cp = NULL;
328 	const struct sockaddr_in6 *sa6_src = NULL;
329 	void *cmdarg;
330 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
331 
332 	if (sa->sa_family != AF_INET6 ||
333 	    sa->sa_len != sizeof(struct sockaddr_in6))
334 		return;
335 
336 	if ((unsigned)cmd >= PRC_NCMDS)
337 		return;
338 	if (PRC_IS_REDIRECT(cmd))
339 		notify = in6_rtchange, d = NULL;
340 	else if (cmd == PRC_HOSTDEAD)
341 		d = NULL;
342 	else if (inet6ctlerrmap[cmd] == 0)
343 		return;
344 
345 	/*
346 	 * If the parameter is from icmp6, decode it.
347 	 */
348 	if (d != NULL) {
349 		ip6cp = (struct ip6ctlparam *)d;
350 		m = ip6cp->ip6c_m;
351 		ip6 = ip6cp->ip6c_ip6;
352 		off = ip6cp->ip6c_off;
353 		cmdarg = ip6cp->ip6c_cmdarg;
354 		sa6_src = ip6cp->ip6c_src;
355 	} else {
356 		m = NULL;
357 		ip6 = NULL;
358 		cmdarg = NULL;
359 		sa6_src = &sa6_any;
360 	}
361 
362 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
363 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
364 }
365 
366 /*
367  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
368  * may have setup with control call.
369  */
370 int
371 #if __STDC__
372 rip6_output(struct mbuf *m, ...)
373 #else
374 rip6_output(m, va_alist)
375 	struct mbuf *m;
376 	va_dcl
377 #endif
378 {
379 	INIT_VNET_INET6(curvnet);
380 	struct mbuf *control;
381 	struct socket *so;
382 	struct sockaddr_in6 *dstsock;
383 	struct in6_addr *dst;
384 	struct ip6_hdr *ip6;
385 	struct inpcb *in6p;
386 	u_int	plen = m->m_pkthdr.len;
387 	int error = 0;
388 	struct ip6_pktopts opt, *optp;
389 	struct ifnet *oifp = NULL;
390 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
391 	int scope_ambiguous = 0;
392 	struct in6_addr in6a;
393 	va_list ap;
394 
395 	va_start(ap, m);
396 	so = va_arg(ap, struct socket *);
397 	dstsock = va_arg(ap, struct sockaddr_in6 *);
398 	control = va_arg(ap, struct mbuf *);
399 	va_end(ap);
400 
401 	in6p = sotoinpcb(so);
402 	INP_WLOCK(in6p);
403 
404 	dst = &dstsock->sin6_addr;
405 	if (control != NULL) {
406 		if ((error = ip6_setpktopts(control, &opt,
407 		    in6p->in6p_outputopts, so->so_cred,
408 		    so->so_proto->pr_protocol)) != 0) {
409 			goto bad;
410 		}
411 		optp = &opt;
412 	} else
413 		optp = in6p->in6p_outputopts;
414 
415 	/*
416 	 * Check and convert scope zone ID into internal form.
417 	 *
418 	 * XXX: we may still need to determine the zone later.
419 	 */
420 	if (!(so->so_state & SS_ISCONNECTED)) {
421 		if (dstsock->sin6_scope_id == 0 && !V_ip6_use_defzone)
422 			scope_ambiguous = 1;
423 		if ((error = sa6_embedscope(dstsock, V_ip6_use_defzone)) != 0)
424 			goto bad;
425 	}
426 
427 	/*
428 	 * For an ICMPv6 packet, we should know its type and code to update
429 	 * statistics.
430 	 */
431 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
432 		struct icmp6_hdr *icmp6;
433 		if (m->m_len < sizeof(struct icmp6_hdr) &&
434 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
435 			error = ENOBUFS;
436 			goto bad;
437 		}
438 		icmp6 = mtod(m, struct icmp6_hdr *);
439 		type = icmp6->icmp6_type;
440 		code = icmp6->icmp6_code;
441 	}
442 
443 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
444 	if (m == NULL) {
445 		error = ENOBUFS;
446 		goto bad;
447 	}
448 	ip6 = mtod(m, struct ip6_hdr *);
449 
450 	/*
451 	 * Source address selection.
452 	 */
453 	error = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
454 	    &oifp, &in6a);
455 	if (error)
456 		goto bad;
457 	error = prison_get_ip6(in6p->inp_cred, &in6a);
458 	if (error != 0)
459 		goto bad;
460 	ip6->ip6_src = in6a;
461 
462 	if (oifp && scope_ambiguous) {
463 		/*
464 		 * Application should provide a proper zone ID or the use of
465 		 * default zone IDs should be enabled.  Unfortunately, some
466 		 * applications do not behave as it should, so we need a
467 		 * workaround.  Even if an appropriate ID is not determined
468 		 * (when it's required), if we can determine the outgoing
469 		 * interface. determine the zone ID based on the interface.
470 		 */
471 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
472 		if (error != 0)
473 			goto bad;
474 	}
475 	ip6->ip6_dst = dstsock->sin6_addr;
476 
477 	/*
478 	 * Fill in the rest of the IPv6 header fields.
479 	 */
480 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
481 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
482 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
483 	    (IPV6_VERSION & IPV6_VERSION_MASK);
484 
485 	/*
486 	 * ip6_plen will be filled in ip6_output, so not fill it here.
487 	 */
488 	ip6->ip6_nxt = in6p->inp_ip_p;
489 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
490 
491 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
492 	    in6p->in6p_cksum != -1) {
493 		struct mbuf *n;
494 		int off;
495 		u_int16_t *p;
496 
497 		/* Compute checksum. */
498 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
499 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
500 		else
501 			off = in6p->in6p_cksum;
502 		if (plen < off + 1) {
503 			error = EINVAL;
504 			goto bad;
505 		}
506 		off += sizeof(struct ip6_hdr);
507 
508 		n = m;
509 		while (n && n->m_len <= off) {
510 			off -= n->m_len;
511 			n = n->m_next;
512 		}
513 		if (!n)
514 			goto bad;
515 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
516 		*p = 0;
517 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
518 	}
519 
520 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
521 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
522 		if (oifp)
523 			icmp6_ifoutstat_inc(oifp, type, code);
524 		ICMP6STAT_INC(icp6s_outhist[type]);
525 	} else
526 		V_rip6stat.rip6s_opackets++;
527 
528 	goto freectl;
529 
530  bad:
531 	if (m)
532 		m_freem(m);
533 
534  freectl:
535 	if (control != NULL) {
536 		ip6_clearpktopts(&opt, -1);
537 		m_freem(control);
538 	}
539 	INP_WUNLOCK(in6p);
540 	return (error);
541 }
542 
543 /*
544  * Raw IPv6 socket option processing.
545  */
546 int
547 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
548 {
549 	int error;
550 
551 	if (sopt->sopt_level == IPPROTO_ICMPV6)
552 		/*
553 		 * XXX: is it better to call icmp6_ctloutput() directly
554 		 * from protosw?
555 		 */
556 		return (icmp6_ctloutput(so, sopt));
557 	else if (sopt->sopt_level != IPPROTO_IPV6)
558 		return (EINVAL);
559 
560 	error = 0;
561 
562 	switch (sopt->sopt_dir) {
563 	case SOPT_GET:
564 		switch (sopt->sopt_name) {
565 		case MRT6_INIT:
566 		case MRT6_DONE:
567 		case MRT6_ADD_MIF:
568 		case MRT6_DEL_MIF:
569 		case MRT6_ADD_MFC:
570 		case MRT6_DEL_MFC:
571 		case MRT6_PIM:
572 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
573 			    EOPNOTSUPP;
574 			break;
575 		case IPV6_CHECKSUM:
576 			error = ip6_raw_ctloutput(so, sopt);
577 			break;
578 		default:
579 			error = ip6_ctloutput(so, sopt);
580 			break;
581 		}
582 		break;
583 
584 	case SOPT_SET:
585 		switch (sopt->sopt_name) {
586 		case MRT6_INIT:
587 		case MRT6_DONE:
588 		case MRT6_ADD_MIF:
589 		case MRT6_DEL_MIF:
590 		case MRT6_ADD_MFC:
591 		case MRT6_DEL_MFC:
592 		case MRT6_PIM:
593 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
594 			    EOPNOTSUPP;
595 			break;
596 		case IPV6_CHECKSUM:
597 			error = ip6_raw_ctloutput(so, sopt);
598 			break;
599 		default:
600 			error = ip6_ctloutput(so, sopt);
601 			break;
602 		}
603 		break;
604 	}
605 
606 	return (error);
607 }
608 
609 static int
610 rip6_attach(struct socket *so, int proto, struct thread *td)
611 {
612 	INIT_VNET_INET(so->so_vnet);
613 	struct inpcb *inp;
614 	struct icmp6_filter *filter;
615 	int error;
616 
617 	inp = sotoinpcb(so);
618 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
619 
620 	error = priv_check(td, PRIV_NETINET_RAW);
621 	if (error)
622 		return (error);
623 	error = soreserve(so, rip_sendspace, rip_recvspace);
624 	if (error)
625 		return (error);
626 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
627 	if (filter == NULL)
628 		return (ENOMEM);
629 	INP_INFO_WLOCK(&V_ripcbinfo);
630 	error = in_pcballoc(so, &V_ripcbinfo);
631 	if (error) {
632 		INP_INFO_WUNLOCK(&V_ripcbinfo);
633 		free(filter, M_PCB);
634 		return (error);
635 	}
636 	inp = (struct inpcb *)so->so_pcb;
637 	INP_INFO_WUNLOCK(&V_ripcbinfo);
638 	inp->inp_vflag |= INP_IPV6;
639 	inp->inp_ip_p = (long)proto;
640 	inp->in6p_hops = -1;	/* use kernel default */
641 	inp->in6p_cksum = -1;
642 	inp->in6p_icmp6filt = filter;
643 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
644 	INP_WUNLOCK(inp);
645 	return (0);
646 }
647 
648 static void
649 rip6_detach(struct socket *so)
650 {
651 	INIT_VNET_INET(so->so_vnet);
652 	INIT_VNET_INET6(so->so_vnet);
653 	struct inpcb *inp;
654 
655 	inp = sotoinpcb(so);
656 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
657 
658 	if (so == V_ip6_mrouter && ip6_mrouter_done)
659 		ip6_mrouter_done();
660 	/* xxx: RSVP */
661 	INP_INFO_WLOCK(&V_ripcbinfo);
662 	INP_WLOCK(inp);
663 	free(inp->in6p_icmp6filt, M_PCB);
664 	in_pcbdetach(inp);
665 	in_pcbfree(inp);
666 	INP_INFO_WUNLOCK(&V_ripcbinfo);
667 }
668 
669 /* XXXRW: This can't ever be called. */
670 static void
671 rip6_abort(struct socket *so)
672 {
673 	struct inpcb *inp;
674 
675 	inp = sotoinpcb(so);
676 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
677 
678 	soisdisconnected(so);
679 }
680 
681 static void
682 rip6_close(struct socket *so)
683 {
684 	struct inpcb *inp;
685 
686 	inp = sotoinpcb(so);
687 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
688 
689 	soisdisconnected(so);
690 }
691 
692 static int
693 rip6_disconnect(struct socket *so)
694 {
695 	struct inpcb *inp;
696 
697 	inp = sotoinpcb(so);
698 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
699 
700 	if ((so->so_state & SS_ISCONNECTED) == 0)
701 		return (ENOTCONN);
702 	inp->in6p_faddr = in6addr_any;
703 	rip6_abort(so);
704 	return (0);
705 }
706 
707 static int
708 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
709 {
710 	INIT_VNET_NET(so->so_vnet);
711 	INIT_VNET_INET(so->so_vnet);
712 	INIT_VNET_INET6(so->so_vnet);
713 	struct inpcb *inp;
714 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
715 	struct ifaddr *ifa = NULL;
716 	int error = 0;
717 
718 	inp = sotoinpcb(so);
719 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
720 
721 	if (nam->sa_len != sizeof(*addr))
722 		return (EINVAL);
723 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
724 		return (error);
725 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
726 		return (EADDRNOTAVAIL);
727 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
728 		return (error);
729 
730 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
731 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
732 		return (EADDRNOTAVAIL);
733 	if (ifa != NULL &&
734 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
735 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
736 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
737 		ifa_free(ifa);
738 		return (EADDRNOTAVAIL);
739 	}
740 	if (ifa != NULL)
741 		ifa_free(ifa);
742 	INP_INFO_WLOCK(&V_ripcbinfo);
743 	INP_WLOCK(inp);
744 	inp->in6p_laddr = addr->sin6_addr;
745 	INP_WUNLOCK(inp);
746 	INP_INFO_WUNLOCK(&V_ripcbinfo);
747 	return (0);
748 }
749 
750 static int
751 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
752 {
753 	INIT_VNET_NET(so->so_vnet);
754 	INIT_VNET_INET(so->so_vnet);
755 	INIT_VNET_INET6(so->so_vnet);
756 	struct inpcb *inp;
757 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
758 	struct in6_addr in6a;
759 	struct ifnet *ifp = NULL;
760 	int error = 0, scope_ambiguous = 0;
761 
762 	inp = sotoinpcb(so);
763 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
764 
765 	if (nam->sa_len != sizeof(*addr))
766 		return (EINVAL);
767 	if (TAILQ_EMPTY(&V_ifnet))
768 		return (EADDRNOTAVAIL);
769 	if (addr->sin6_family != AF_INET6)
770 		return (EAFNOSUPPORT);
771 
772 	/*
773 	 * Application should provide a proper zone ID or the use of default
774 	 * zone IDs should be enabled.  Unfortunately, some applications do
775 	 * not behave as it should, so we need a workaround.  Even if an
776 	 * appropriate ID is not determined, we'll see if we can determine
777 	 * the outgoing interface.  If we can, determine the zone ID based on
778 	 * the interface below.
779 	 */
780 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
781 		scope_ambiguous = 1;
782 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
783 		return (error);
784 
785 	INP_INFO_WLOCK(&V_ripcbinfo);
786 	INP_WLOCK(inp);
787 	/* Source address selection. XXX: need pcblookup? */
788 	error = in6_selectsrc(addr, inp->in6p_outputopts,
789 	    inp, NULL, so->so_cred, &ifp, &in6a);
790 	if (error) {
791 		INP_WUNLOCK(inp);
792 		INP_INFO_WUNLOCK(&V_ripcbinfo);
793 		return (error);
794 	}
795 
796 	/* XXX: see above */
797 	if (ifp && scope_ambiguous &&
798 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
799 		INP_WUNLOCK(inp);
800 		INP_INFO_WUNLOCK(&V_ripcbinfo);
801 		return (error);
802 	}
803 	inp->in6p_faddr = addr->sin6_addr;
804 	inp->in6p_laddr = in6a;
805 	soisconnected(so);
806 	INP_WUNLOCK(inp);
807 	INP_INFO_WUNLOCK(&V_ripcbinfo);
808 	return (0);
809 }
810 
811 static int
812 rip6_shutdown(struct socket *so)
813 {
814 	struct inpcb *inp;
815 
816 	inp = sotoinpcb(so);
817 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
818 
819 	INP_WLOCK(inp);
820 	socantsendmore(so);
821 	INP_WUNLOCK(inp);
822 	return (0);
823 }
824 
825 static int
826 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
827     struct mbuf *control, struct thread *td)
828 {
829 	struct inpcb *inp;
830 	struct sockaddr_in6 tmp;
831 	struct sockaddr_in6 *dst;
832 	int ret;
833 
834 	inp = sotoinpcb(so);
835 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
836 
837 	/* Always copy sockaddr to avoid overwrites. */
838 	/* Unlocked read. */
839 	if (so->so_state & SS_ISCONNECTED) {
840 		if (nam) {
841 			m_freem(m);
842 			return (EISCONN);
843 		}
844 		/* XXX */
845 		bzero(&tmp, sizeof(tmp));
846 		tmp.sin6_family = AF_INET6;
847 		tmp.sin6_len = sizeof(struct sockaddr_in6);
848 		INP_RLOCK(inp);
849 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
850 		    sizeof(struct in6_addr));
851 		INP_RUNLOCK(inp);
852 		dst = &tmp;
853 	} else {
854 		if (nam == NULL) {
855 			m_freem(m);
856 			return (ENOTCONN);
857 		}
858 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
859 			m_freem(m);
860 			return (EINVAL);
861 		}
862 		tmp = *(struct sockaddr_in6 *)nam;
863 		dst = &tmp;
864 
865 		if (dst->sin6_family == AF_UNSPEC) {
866 			/*
867 			 * XXX: we allow this case for backward
868 			 * compatibility to buggy applications that
869 			 * rely on old (and wrong) kernel behavior.
870 			 */
871 			log(LOG_INFO, "rip6 SEND: address family is "
872 			    "unspec. Assume AF_INET6\n");
873 			dst->sin6_family = AF_INET6;
874 		} else if (dst->sin6_family != AF_INET6) {
875 			m_freem(m);
876 			return(EAFNOSUPPORT);
877 		}
878 	}
879 	ret = rip6_output(m, so, dst, control);
880 	return (ret);
881 }
882 
883 struct pr_usrreqs rip6_usrreqs = {
884 	.pru_abort =		rip6_abort,
885 	.pru_attach =		rip6_attach,
886 	.pru_bind =		rip6_bind,
887 	.pru_connect =		rip6_connect,
888 	.pru_control =		in6_control,
889 	.pru_detach =		rip6_detach,
890 	.pru_disconnect =	rip6_disconnect,
891 	.pru_peeraddr =		in6_getpeeraddr,
892 	.pru_send =		rip6_send,
893 	.pru_shutdown =		rip6_shutdown,
894 	.pru_sockaddr =		in6_getsockaddr,
895 	.pru_close =		rip6_close,
896 };
897