xref: /freebsd/sys/netinet6/raw_ip6.c (revision c0020399a650364d0134f79f3fa319f84064372d)
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 #include "opt_route.h"
68 
69 #include <sys/param.h>
70 #include <sys/errno.h>
71 #include <sys/jail.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 #include <sys/vimage.h>
84 
85 #include <net/if.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 #include <netinet/vinet.h>
95 
96 #include <netinet/icmp6.h>
97 #include <netinet/ip6.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/vinet6.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 #ifdef VIMAGE_GLOBALS
122 extern struct	inpcbhead ripcb;
123 extern struct	inpcbinfo ripcbinfo;
124 struct rip6stat rip6stat;
125 #endif
126 
127 extern u_long	rip_sendspace;
128 extern u_long	rip_recvspace;
129 
130 /*
131  * Hooks for multicast forwarding.
132  */
133 struct socket *ip6_mrouter = NULL;
134 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
135 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
136 int (*ip6_mrouter_done)(void);
137 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
138 int (*mrt6_ioctl)(int, caddr_t);
139 
140 /*
141  * Setup generic address and protocol structures for raw_input routine, then
142  * pass them along with mbuf chain.
143  */
144 int
145 rip6_input(struct mbuf **mp, int *offp, int proto)
146 {
147 	INIT_VNET_INET(curvnet);
148 	INIT_VNET_INET6(curvnet);
149 #ifdef IPSEC
150 	INIT_VNET_IPSEC(curvnet);
151 #endif
152 	struct mbuf *m = *mp;
153 	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
154 	register struct inpcb *in6p;
155 	struct inpcb *last = 0;
156 	struct mbuf *opts = NULL;
157 	struct sockaddr_in6 fromsa;
158 
159 	V_rip6stat.rip6s_ipackets++;
160 
161 	if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
162 		/* XXX Send icmp6 host/port unreach? */
163 		m_freem(m);
164 		return (IPPROTO_DONE);
165 	}
166 
167 	init_sin6(&fromsa, m); /* general init */
168 
169 	INP_INFO_RLOCK(&V_ripcbinfo);
170 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
171 		/* XXX inp locking */
172 		if ((in6p->inp_vflag & INP_IPV6) == 0)
173 			continue;
174 		if (in6p->inp_ip_p &&
175 		    in6p->inp_ip_p != proto)
176 			continue;
177 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
178 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
179 			continue;
180 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
181 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
182 			continue;
183 		if (prison_check_ip6(in6p->inp_cred, &ip6->ip6_dst) != 0)
184 			continue;
185 		INP_RLOCK(in6p);
186 		if (in6p->in6p_cksum != -1) {
187 			V_rip6stat.rip6s_isum++;
188 			if (in6_cksum(m, proto, *offp,
189 			    m->m_pkthdr.len - *offp)) {
190 				INP_RUNLOCK(in6p);
191 				V_rip6stat.rip6s_badsum++;
192 				continue;
193 			}
194 		}
195 		if (last != NULL) {
196 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
197 
198 #ifdef IPSEC
199 			/*
200 			 * Check AH/ESP integrity.
201 			 */
202 			if (n && ipsec6_in_reject(n, last)) {
203 				m_freem(n);
204 				V_ipsec6stat.in_polvio++;
205 				/* Do not inject data into pcb. */
206 			} else
207 #endif /* IPSEC */
208 			if (n) {
209 				if (last->inp_flags & INP_CONTROLOPTS ||
210 				    last->inp_socket->so_options & SO_TIMESTAMP)
211 					ip6_savecontrol(last, n, &opts);
212 				/* strip intermediate headers */
213 				m_adj(n, *offp);
214 				if (sbappendaddr(&last->inp_socket->so_rcv,
215 						(struct sockaddr *)&fromsa,
216 						 n, opts) == 0) {
217 					m_freem(n);
218 					if (opts)
219 						m_freem(opts);
220 					V_rip6stat.rip6s_fullsock++;
221 				} else
222 					sorwakeup(last->inp_socket);
223 				opts = NULL;
224 			}
225 			INP_RUNLOCK(last);
226 		}
227 		last = in6p;
228 	}
229 	INP_INFO_RUNLOCK(&V_ripcbinfo);
230 #ifdef IPSEC
231 	/*
232 	 * Check AH/ESP integrity.
233 	 */
234 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
235 		m_freem(m);
236 		V_ipsec6stat.in_polvio++;
237 		V_ip6stat.ip6s_delivered--;
238 		/* Do not inject data into pcb. */
239 		INP_RUNLOCK(last);
240 	} else
241 #endif /* IPSEC */
242 	if (last != NULL) {
243 		if (last->inp_flags & INP_CONTROLOPTS ||
244 		    last->inp_socket->so_options & SO_TIMESTAMP)
245 			ip6_savecontrol(last, m, &opts);
246 		/* Strip intermediate headers. */
247 		m_adj(m, *offp);
248 		if (sbappendaddr(&last->inp_socket->so_rcv,
249 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
250 			m_freem(m);
251 			if (opts)
252 				m_freem(opts);
253 			V_rip6stat.rip6s_fullsock++;
254 		} else
255 			sorwakeup(last->inp_socket);
256 		INP_RUNLOCK(last);
257 	} else {
258 		V_rip6stat.rip6s_nosock++;
259 		if (m->m_flags & M_MCAST)
260 			V_rip6stat.rip6s_nosockmcast++;
261 		if (proto == IPPROTO_NONE)
262 			m_freem(m);
263 		else {
264 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
265 			icmp6_error(m, ICMP6_PARAM_PROB,
266 			    ICMP6_PARAMPROB_NEXTHEADER,
267 			    prvnxtp - mtod(m, char *));
268 		}
269 		V_ip6stat.ip6s_delivered--;
270 	}
271 	return (IPPROTO_DONE);
272 }
273 
274 void
275 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
276 {
277 	INIT_VNET_INET(curvnet);
278 	struct ip6_hdr *ip6;
279 	struct mbuf *m;
280 	int off = 0;
281 	struct ip6ctlparam *ip6cp = NULL;
282 	const struct sockaddr_in6 *sa6_src = NULL;
283 	void *cmdarg;
284 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
285 
286 	if (sa->sa_family != AF_INET6 ||
287 	    sa->sa_len != sizeof(struct sockaddr_in6))
288 		return;
289 
290 	if ((unsigned)cmd >= PRC_NCMDS)
291 		return;
292 	if (PRC_IS_REDIRECT(cmd))
293 		notify = in6_rtchange, d = NULL;
294 	else if (cmd == PRC_HOSTDEAD)
295 		d = NULL;
296 	else if (inet6ctlerrmap[cmd] == 0)
297 		return;
298 
299 	/*
300 	 * If the parameter is from icmp6, decode it.
301 	 */
302 	if (d != NULL) {
303 		ip6cp = (struct ip6ctlparam *)d;
304 		m = ip6cp->ip6c_m;
305 		ip6 = ip6cp->ip6c_ip6;
306 		off = ip6cp->ip6c_off;
307 		cmdarg = ip6cp->ip6c_cmdarg;
308 		sa6_src = ip6cp->ip6c_src;
309 	} else {
310 		m = NULL;
311 		ip6 = NULL;
312 		cmdarg = NULL;
313 		sa6_src = &sa6_any;
314 	}
315 
316 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
317 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
318 }
319 
320 /*
321  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
322  * may have setup with control call.
323  */
324 int
325 #if __STDC__
326 rip6_output(struct mbuf *m, ...)
327 #else
328 rip6_output(m, va_alist)
329 	struct mbuf *m;
330 	va_dcl
331 #endif
332 {
333 	INIT_VNET_INET6(curvnet);
334 	struct mbuf *control;
335 	struct socket *so;
336 	struct sockaddr_in6 *dstsock;
337 	struct in6_addr *dst;
338 	struct ip6_hdr *ip6;
339 	struct inpcb *in6p;
340 	u_int	plen = m->m_pkthdr.len;
341 	int error = 0;
342 	struct ip6_pktopts opt, *optp;
343 	struct ifnet *oifp = NULL;
344 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
345 	int scope_ambiguous = 0;
346 	struct in6_addr *in6a;
347 	va_list ap;
348 
349 	va_start(ap, m);
350 	so = va_arg(ap, struct socket *);
351 	dstsock = va_arg(ap, struct sockaddr_in6 *);
352 	control = va_arg(ap, struct mbuf *);
353 	va_end(ap);
354 
355 	in6p = sotoinpcb(so);
356 	INP_WLOCK(in6p);
357 
358 	dst = &dstsock->sin6_addr;
359 	if (control != NULL) {
360 		if ((error = ip6_setpktopts(control, &opt,
361 		    in6p->in6p_outputopts, so->so_cred,
362 		    so->so_proto->pr_protocol)) != 0) {
363 			goto bad;
364 		}
365 		optp = &opt;
366 	} else
367 		optp = in6p->in6p_outputopts;
368 
369 	/*
370 	 * Check and convert scope zone ID into internal form.
371 	 *
372 	 * XXX: we may still need to determine the zone later.
373 	 */
374 	if (!(so->so_state & SS_ISCONNECTED)) {
375 		if (dstsock->sin6_scope_id == 0 && !V_ip6_use_defzone)
376 			scope_ambiguous = 1;
377 		if ((error = sa6_embedscope(dstsock, V_ip6_use_defzone)) != 0)
378 			goto bad;
379 	}
380 
381 	/*
382 	 * For an ICMPv6 packet, we should know its type and code to update
383 	 * statistics.
384 	 */
385 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
386 		struct icmp6_hdr *icmp6;
387 		if (m->m_len < sizeof(struct icmp6_hdr) &&
388 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
389 			error = ENOBUFS;
390 			goto bad;
391 		}
392 		icmp6 = mtod(m, struct icmp6_hdr *);
393 		type = icmp6->icmp6_type;
394 		code = icmp6->icmp6_code;
395 	}
396 
397 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
398 	if (m == NULL) {
399 		error = ENOBUFS;
400 		goto bad;
401 	}
402 	ip6 = mtod(m, struct ip6_hdr *);
403 
404 	/*
405 	 * Source address selection.
406 	 */
407 	if ((in6a = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
408 	    &oifp, &error)) == NULL) {
409 		if (error == 0)
410 			error = EADDRNOTAVAIL;
411 		goto bad;
412 	}
413 	error = prison_get_ip6(in6p->inp_cred, in6a);
414 	if (error != 0)
415 		goto bad;
416 	ip6->ip6_src = *in6a;
417 
418 	if (oifp && scope_ambiguous) {
419 		/*
420 		 * Application should provide a proper zone ID or the use of
421 		 * default zone IDs should be enabled.  Unfortunately, some
422 		 * applications do not behave as it should, so we need a
423 		 * workaround.  Even if an appropriate ID is not determined
424 		 * (when it's required), if we can determine the outgoing
425 		 * interface. determine the zone ID based on the interface.
426 		 */
427 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
428 		if (error != 0)
429 			goto bad;
430 	}
431 	ip6->ip6_dst = dstsock->sin6_addr;
432 
433 	/*
434 	 * Fill in the rest of the IPv6 header fields.
435 	 */
436 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
437 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
438 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
439 	    (IPV6_VERSION & IPV6_VERSION_MASK);
440 
441 	/*
442 	 * ip6_plen will be filled in ip6_output, so not fill it here.
443 	 */
444 	ip6->ip6_nxt = in6p->inp_ip_p;
445 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
446 
447 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
448 	    in6p->in6p_cksum != -1) {
449 		struct mbuf *n;
450 		int off;
451 		u_int16_t *p;
452 
453 		/* Compute checksum. */
454 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
455 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
456 		else
457 			off = in6p->in6p_cksum;
458 		if (plen < off + 1) {
459 			error = EINVAL;
460 			goto bad;
461 		}
462 		off += sizeof(struct ip6_hdr);
463 
464 		n = m;
465 		while (n && n->m_len <= off) {
466 			off -= n->m_len;
467 			n = n->m_next;
468 		}
469 		if (!n)
470 			goto bad;
471 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
472 		*p = 0;
473 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
474 	}
475 
476 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
477 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
478 		if (oifp)
479 			icmp6_ifoutstat_inc(oifp, type, code);
480 		ICMP6STAT_INC(icp6s_outhist[type]);
481 	} else
482 		V_rip6stat.rip6s_opackets++;
483 
484 	goto freectl;
485 
486  bad:
487 	if (m)
488 		m_freem(m);
489 
490  freectl:
491 	if (control != NULL) {
492 		ip6_clearpktopts(&opt, -1);
493 		m_freem(control);
494 	}
495 	INP_WUNLOCK(in6p);
496 	return (error);
497 }
498 
499 /*
500  * Raw IPv6 socket option processing.
501  */
502 int
503 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
504 {
505 	int error;
506 
507 	if (sopt->sopt_level == IPPROTO_ICMPV6)
508 		/*
509 		 * XXX: is it better to call icmp6_ctloutput() directly
510 		 * from protosw?
511 		 */
512 		return (icmp6_ctloutput(so, sopt));
513 	else if (sopt->sopt_level != IPPROTO_IPV6)
514 		return (EINVAL);
515 
516 	error = 0;
517 
518 	switch (sopt->sopt_dir) {
519 	case SOPT_GET:
520 		switch (sopt->sopt_name) {
521 		case MRT6_INIT:
522 		case MRT6_DONE:
523 		case MRT6_ADD_MIF:
524 		case MRT6_DEL_MIF:
525 		case MRT6_ADD_MFC:
526 		case MRT6_DEL_MFC:
527 		case MRT6_PIM:
528 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
529 			    EOPNOTSUPP;
530 			break;
531 		case IPV6_CHECKSUM:
532 			error = ip6_raw_ctloutput(so, sopt);
533 			break;
534 		default:
535 			error = ip6_ctloutput(so, sopt);
536 			break;
537 		}
538 		break;
539 
540 	case SOPT_SET:
541 		switch (sopt->sopt_name) {
542 		case MRT6_INIT:
543 		case MRT6_DONE:
544 		case MRT6_ADD_MIF:
545 		case MRT6_DEL_MIF:
546 		case MRT6_ADD_MFC:
547 		case MRT6_DEL_MFC:
548 		case MRT6_PIM:
549 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
550 			    EOPNOTSUPP;
551 			break;
552 		case IPV6_CHECKSUM:
553 			error = ip6_raw_ctloutput(so, sopt);
554 			break;
555 		default:
556 			error = ip6_ctloutput(so, sopt);
557 			break;
558 		}
559 		break;
560 	}
561 
562 	return (error);
563 }
564 
565 static int
566 rip6_attach(struct socket *so, int proto, struct thread *td)
567 {
568 	INIT_VNET_INET(so->so_vnet);
569 	struct inpcb *inp;
570 	struct icmp6_filter *filter;
571 	int error;
572 
573 	inp = sotoinpcb(so);
574 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
575 
576 	error = priv_check(td, PRIV_NETINET_RAW);
577 	if (error)
578 		return (error);
579 	error = soreserve(so, rip_sendspace, rip_recvspace);
580 	if (error)
581 		return (error);
582 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
583 	if (filter == NULL)
584 		return (ENOMEM);
585 	INP_INFO_WLOCK(&V_ripcbinfo);
586 	error = in_pcballoc(so, &V_ripcbinfo);
587 	if (error) {
588 		INP_INFO_WUNLOCK(&V_ripcbinfo);
589 		free(filter, M_PCB);
590 		return (error);
591 	}
592 	inp = (struct inpcb *)so->so_pcb;
593 	INP_INFO_WUNLOCK(&V_ripcbinfo);
594 	inp->inp_vflag |= INP_IPV6;
595 	inp->inp_ip_p = (long)proto;
596 	inp->in6p_hops = -1;	/* use kernel default */
597 	inp->in6p_cksum = -1;
598 	inp->in6p_icmp6filt = filter;
599 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
600 	INP_WUNLOCK(inp);
601 	return (0);
602 }
603 
604 static void
605 rip6_detach(struct socket *so)
606 {
607 	INIT_VNET_INET(so->so_vnet);
608 	struct inpcb *inp;
609 
610 	inp = sotoinpcb(so);
611 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
612 
613 	if (so == ip6_mrouter && ip6_mrouter_done)
614 		ip6_mrouter_done();
615 	/* xxx: RSVP */
616 	INP_INFO_WLOCK(&V_ripcbinfo);
617 	INP_WLOCK(inp);
618 	free(inp->in6p_icmp6filt, M_PCB);
619 	in_pcbdetach(inp);
620 	in_pcbfree(inp);
621 	INP_INFO_WUNLOCK(&V_ripcbinfo);
622 }
623 
624 /* XXXRW: This can't ever be called. */
625 static void
626 rip6_abort(struct socket *so)
627 {
628 	struct inpcb *inp;
629 
630 	inp = sotoinpcb(so);
631 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
632 
633 	soisdisconnected(so);
634 }
635 
636 static void
637 rip6_close(struct socket *so)
638 {
639 	struct inpcb *inp;
640 
641 	inp = sotoinpcb(so);
642 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
643 
644 	soisdisconnected(so);
645 }
646 
647 static int
648 rip6_disconnect(struct socket *so)
649 {
650 	struct inpcb *inp;
651 
652 	inp = sotoinpcb(so);
653 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
654 
655 	if ((so->so_state & SS_ISCONNECTED) == 0)
656 		return (ENOTCONN);
657 	inp->in6p_faddr = in6addr_any;
658 	rip6_abort(so);
659 	return (0);
660 }
661 
662 static int
663 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
664 {
665 	INIT_VNET_NET(so->so_vnet);
666 	INIT_VNET_INET(so->so_vnet);
667 	INIT_VNET_INET6(so->so_vnet);
668 	struct inpcb *inp;
669 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
670 	struct ifaddr *ia = NULL;
671 	int error = 0;
672 
673 	inp = sotoinpcb(so);
674 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
675 
676 	if (nam->sa_len != sizeof(*addr))
677 		return (EINVAL);
678 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
679 		return (error);
680 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
681 		return (EADDRNOTAVAIL);
682 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
683 		return (error);
684 
685 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
686 	    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0)
687 		return (EADDRNOTAVAIL);
688 	if (ia &&
689 	    ((struct in6_ifaddr *)ia)->ia6_flags &
690 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
691 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
692 		return (EADDRNOTAVAIL);
693 	}
694 	INP_INFO_WLOCK(&V_ripcbinfo);
695 	INP_WLOCK(inp);
696 	inp->in6p_laddr = addr->sin6_addr;
697 	INP_WUNLOCK(inp);
698 	INP_INFO_WUNLOCK(&V_ripcbinfo);
699 	return (0);
700 }
701 
702 static int
703 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
704 {
705 	INIT_VNET_NET(so->so_vnet);
706 	INIT_VNET_INET(so->so_vnet);
707 	INIT_VNET_INET6(so->so_vnet);
708 	struct inpcb *inp;
709 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
710 	struct in6_addr *in6a = NULL;
711 	struct ifnet *ifp = NULL;
712 	int error = 0, scope_ambiguous = 0;
713 
714 	inp = sotoinpcb(so);
715 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
716 
717 	if (nam->sa_len != sizeof(*addr))
718 		return (EINVAL);
719 	if (TAILQ_EMPTY(&V_ifnet))
720 		return (EADDRNOTAVAIL);
721 	if (addr->sin6_family != AF_INET6)
722 		return (EAFNOSUPPORT);
723 
724 	/*
725 	 * Application should provide a proper zone ID or the use of default
726 	 * zone IDs should be enabled.  Unfortunately, some applications do
727 	 * not behave as it should, so we need a workaround.  Even if an
728 	 * appropriate ID is not determined, we'll see if we can determine
729 	 * the outgoing interface.  If we can, determine the zone ID based on
730 	 * the interface below.
731 	 */
732 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
733 		scope_ambiguous = 1;
734 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
735 		return (error);
736 
737 	INP_INFO_WLOCK(&V_ripcbinfo);
738 	INP_WLOCK(inp);
739 	/* Source address selection. XXX: need pcblookup? */
740 	in6a = in6_selectsrc(addr, inp->in6p_outputopts,
741 			     inp, NULL, so->so_cred,
742 			     &ifp, &error);
743 	if (in6a == NULL) {
744 		INP_WUNLOCK(inp);
745 		INP_INFO_WUNLOCK(&V_ripcbinfo);
746 		return (error ? error : EADDRNOTAVAIL);
747 	}
748 
749 	/* XXX: see above */
750 	if (ifp && scope_ambiguous &&
751 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
752 		INP_WUNLOCK(inp);
753 		INP_INFO_WUNLOCK(&V_ripcbinfo);
754 		return (error);
755 	}
756 	inp->in6p_faddr = addr->sin6_addr;
757 	inp->in6p_laddr = *in6a;
758 	soisconnected(so);
759 	INP_WUNLOCK(inp);
760 	INP_INFO_WUNLOCK(&V_ripcbinfo);
761 	return (0);
762 }
763 
764 static int
765 rip6_shutdown(struct socket *so)
766 {
767 	struct inpcb *inp;
768 
769 	inp = sotoinpcb(so);
770 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
771 
772 	INP_WLOCK(inp);
773 	socantsendmore(so);
774 	INP_WUNLOCK(inp);
775 	return (0);
776 }
777 
778 static int
779 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
780     struct mbuf *control, struct thread *td)
781 {
782 	struct inpcb *inp;
783 	struct sockaddr_in6 tmp;
784 	struct sockaddr_in6 *dst;
785 	int ret;
786 
787 	inp = sotoinpcb(so);
788 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
789 
790 	/* Always copy sockaddr to avoid overwrites. */
791 	/* Unlocked read. */
792 	if (so->so_state & SS_ISCONNECTED) {
793 		if (nam) {
794 			m_freem(m);
795 			return (EISCONN);
796 		}
797 		/* XXX */
798 		bzero(&tmp, sizeof(tmp));
799 		tmp.sin6_family = AF_INET6;
800 		tmp.sin6_len = sizeof(struct sockaddr_in6);
801 		INP_RLOCK(inp);
802 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
803 		    sizeof(struct in6_addr));
804 		INP_RUNLOCK(inp);
805 		dst = &tmp;
806 	} else {
807 		if (nam == NULL) {
808 			m_freem(m);
809 			return (ENOTCONN);
810 		}
811 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
812 			m_freem(m);
813 			return (EINVAL);
814 		}
815 		tmp = *(struct sockaddr_in6 *)nam;
816 		dst = &tmp;
817 
818 		if (dst->sin6_family == AF_UNSPEC) {
819 			/*
820 			 * XXX: we allow this case for backward
821 			 * compatibility to buggy applications that
822 			 * rely on old (and wrong) kernel behavior.
823 			 */
824 			log(LOG_INFO, "rip6 SEND: address family is "
825 			    "unspec. Assume AF_INET6\n");
826 			dst->sin6_family = AF_INET6;
827 		} else if (dst->sin6_family != AF_INET6) {
828 			m_freem(m);
829 			return(EAFNOSUPPORT);
830 		}
831 	}
832 	ret = rip6_output(m, so, dst, control);
833 	return (ret);
834 }
835 
836 struct pr_usrreqs rip6_usrreqs = {
837 	.pru_abort =		rip6_abort,
838 	.pru_attach =		rip6_attach,
839 	.pru_bind =		rip6_bind,
840 	.pru_connect =		rip6_connect,
841 	.pru_control =		in6_control,
842 	.pru_detach =		rip6_detach,
843 	.pru_disconnect =	rip6_disconnect,
844 	.pru_peeraddr =		in6_getpeeraddr,
845 	.pru_send =		rip6_send,
846 	.pru_shutdown =		rip6_shutdown,
847 	.pru_sockaddr =		in6_getsockaddr,
848 	.pru_close =		rip6_close,
849 };
850