xref: /freebsd/sys/netinet6/raw_ip6.c (revision 9517e866259191fcd39434a97ad849a9b59b9b9f)
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 VNET_DEFINE(struct rip6stat, rip6stat);
123 
124 extern u_long	rip_sendspace;
125 extern u_long	rip_recvspace;
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(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 			struct sockaddr_in6 mcaddr;
217 			int blocked;
218 
219 			bzero(&mcaddr, sizeof(struct sockaddr_in6));
220 			mcaddr.sin6_len = sizeof(struct sockaddr_in6);
221 			mcaddr.sin6_family = AF_INET6;
222 			mcaddr.sin6_addr = ip6->ip6_dst;
223 
224 			blocked = im6o_mc_filter(in6p->in6p_moptions, ifp,
225 			    (struct sockaddr *)&mcaddr,
226 			    (struct sockaddr *)&fromsa);
227 			if (blocked != MCAST_PASS) {
228 				IP6STAT_INC(ip6s_notmember);
229 				continue;
230 			}
231 		}
232 		if (last != NULL) {
233 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
234 
235 #ifdef IPSEC
236 			/*
237 			 * Check AH/ESP integrity.
238 			 */
239 			if (n && ipsec6_in_reject(n, last)) {
240 				m_freem(n);
241 				V_ipsec6stat.in_polvio++;
242 				/* Do not inject data into pcb. */
243 			} else
244 #endif /* IPSEC */
245 			if (n) {
246 				if (last->inp_flags & INP_CONTROLOPTS ||
247 				    last->inp_socket->so_options & SO_TIMESTAMP)
248 					ip6_savecontrol(last, n, &opts);
249 				/* strip intermediate headers */
250 				m_adj(n, *offp);
251 				if (sbappendaddr(&last->inp_socket->so_rcv,
252 						(struct sockaddr *)&fromsa,
253 						 n, opts) == 0) {
254 					m_freem(n);
255 					if (opts)
256 						m_freem(opts);
257 					V_rip6stat.rip6s_fullsock++;
258 				} else
259 					sorwakeup(last->inp_socket);
260 				opts = NULL;
261 			}
262 			INP_RUNLOCK(last);
263 		}
264 		last = in6p;
265 	}
266 	INP_INFO_RUNLOCK(&V_ripcbinfo);
267 #ifdef IPSEC
268 	/*
269 	 * Check AH/ESP integrity.
270 	 */
271 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
272 		m_freem(m);
273 		V_ipsec6stat.in_polvio++;
274 		V_ip6stat.ip6s_delivered--;
275 		/* Do not inject data into pcb. */
276 		INP_RUNLOCK(last);
277 	} else
278 #endif /* IPSEC */
279 	if (last != NULL) {
280 		if (last->inp_flags & INP_CONTROLOPTS ||
281 		    last->inp_socket->so_options & SO_TIMESTAMP)
282 			ip6_savecontrol(last, m, &opts);
283 		/* Strip intermediate headers. */
284 		m_adj(m, *offp);
285 		if (sbappendaddr(&last->inp_socket->so_rcv,
286 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
287 			m_freem(m);
288 			if (opts)
289 				m_freem(opts);
290 			V_rip6stat.rip6s_fullsock++;
291 		} else
292 			sorwakeup(last->inp_socket);
293 		INP_RUNLOCK(last);
294 	} else {
295 		V_rip6stat.rip6s_nosock++;
296 		if (m->m_flags & M_MCAST)
297 			V_rip6stat.rip6s_nosockmcast++;
298 		if (proto == IPPROTO_NONE)
299 			m_freem(m);
300 		else {
301 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
302 			icmp6_error(m, ICMP6_PARAM_PROB,
303 			    ICMP6_PARAMPROB_NEXTHEADER,
304 			    prvnxtp - mtod(m, char *));
305 		}
306 		V_ip6stat.ip6s_delivered--;
307 	}
308 	return (IPPROTO_DONE);
309 }
310 
311 void
312 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
313 {
314 	struct ip6_hdr *ip6;
315 	struct mbuf *m;
316 	int off = 0;
317 	struct ip6ctlparam *ip6cp = NULL;
318 	const struct sockaddr_in6 *sa6_src = NULL;
319 	void *cmdarg;
320 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
321 
322 	if (sa->sa_family != AF_INET6 ||
323 	    sa->sa_len != sizeof(struct sockaddr_in6))
324 		return;
325 
326 	if ((unsigned)cmd >= PRC_NCMDS)
327 		return;
328 	if (PRC_IS_REDIRECT(cmd))
329 		notify = in6_rtchange, d = NULL;
330 	else if (cmd == PRC_HOSTDEAD)
331 		d = NULL;
332 	else if (inet6ctlerrmap[cmd] == 0)
333 		return;
334 
335 	/*
336 	 * If the parameter is from icmp6, decode it.
337 	 */
338 	if (d != NULL) {
339 		ip6cp = (struct ip6ctlparam *)d;
340 		m = ip6cp->ip6c_m;
341 		ip6 = ip6cp->ip6c_ip6;
342 		off = ip6cp->ip6c_off;
343 		cmdarg = ip6cp->ip6c_cmdarg;
344 		sa6_src = ip6cp->ip6c_src;
345 	} else {
346 		m = NULL;
347 		ip6 = NULL;
348 		cmdarg = NULL;
349 		sa6_src = &sa6_any;
350 	}
351 
352 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
353 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
354 }
355 
356 /*
357  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
358  * may have setup with control call.
359  */
360 int
361 #if __STDC__
362 rip6_output(struct mbuf *m, ...)
363 #else
364 rip6_output(m, va_alist)
365 	struct mbuf *m;
366 	va_dcl
367 #endif
368 {
369 	struct mbuf *control;
370 	struct socket *so;
371 	struct sockaddr_in6 *dstsock;
372 	struct in6_addr *dst;
373 	struct ip6_hdr *ip6;
374 	struct inpcb *in6p;
375 	u_int	plen = m->m_pkthdr.len;
376 	int error = 0;
377 	struct ip6_pktopts opt, *optp;
378 	struct ifnet *oifp = NULL;
379 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
380 	int scope_ambiguous = 0;
381 	struct in6_addr in6a;
382 	va_list ap;
383 
384 	va_start(ap, m);
385 	so = va_arg(ap, struct socket *);
386 	dstsock = va_arg(ap, struct sockaddr_in6 *);
387 	control = va_arg(ap, struct mbuf *);
388 	va_end(ap);
389 
390 	in6p = sotoinpcb(so);
391 	INP_WLOCK(in6p);
392 
393 	dst = &dstsock->sin6_addr;
394 	if (control != NULL) {
395 		if ((error = ip6_setpktopts(control, &opt,
396 		    in6p->in6p_outputopts, so->so_cred,
397 		    so->so_proto->pr_protocol)) != 0) {
398 			goto bad;
399 		}
400 		optp = &opt;
401 	} else
402 		optp = in6p->in6p_outputopts;
403 
404 	/*
405 	 * Check and convert scope zone ID into internal form.
406 	 *
407 	 * XXX: we may still need to determine the zone later.
408 	 */
409 	if (!(so->so_state & SS_ISCONNECTED)) {
410 		if (dstsock->sin6_scope_id == 0 && !V_ip6_use_defzone)
411 			scope_ambiguous = 1;
412 		if ((error = sa6_embedscope(dstsock, V_ip6_use_defzone)) != 0)
413 			goto bad;
414 	}
415 
416 	/*
417 	 * For an ICMPv6 packet, we should know its type and code to update
418 	 * statistics.
419 	 */
420 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
421 		struct icmp6_hdr *icmp6;
422 		if (m->m_len < sizeof(struct icmp6_hdr) &&
423 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
424 			error = ENOBUFS;
425 			goto bad;
426 		}
427 		icmp6 = mtod(m, struct icmp6_hdr *);
428 		type = icmp6->icmp6_type;
429 		code = icmp6->icmp6_code;
430 	}
431 
432 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
433 	if (m == NULL) {
434 		error = ENOBUFS;
435 		goto bad;
436 	}
437 	ip6 = mtod(m, struct ip6_hdr *);
438 
439 	/*
440 	 * Source address selection.
441 	 */
442 	error = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
443 	    &oifp, &in6a);
444 	if (error)
445 		goto bad;
446 	error = prison_get_ip6(in6p->inp_cred, &in6a);
447 	if (error != 0)
448 		goto bad;
449 	ip6->ip6_src = in6a;
450 
451 	if (oifp && scope_ambiguous) {
452 		/*
453 		 * Application should provide a proper zone ID or the use of
454 		 * default zone IDs should be enabled.  Unfortunately, some
455 		 * applications do not behave as it should, so we need a
456 		 * workaround.  Even if an appropriate ID is not determined
457 		 * (when it's required), if we can determine the outgoing
458 		 * interface. determine the zone ID based on the interface.
459 		 */
460 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
461 		if (error != 0)
462 			goto bad;
463 	}
464 	ip6->ip6_dst = dstsock->sin6_addr;
465 
466 	/*
467 	 * Fill in the rest of the IPv6 header fields.
468 	 */
469 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
470 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
471 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
472 	    (IPV6_VERSION & IPV6_VERSION_MASK);
473 
474 	/*
475 	 * ip6_plen will be filled in ip6_output, so not fill it here.
476 	 */
477 	ip6->ip6_nxt = in6p->inp_ip_p;
478 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
479 
480 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
481 	    in6p->in6p_cksum != -1) {
482 		struct mbuf *n;
483 		int off;
484 		u_int16_t *p;
485 
486 		/* Compute checksum. */
487 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
488 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
489 		else
490 			off = in6p->in6p_cksum;
491 		if (plen < off + 1) {
492 			error = EINVAL;
493 			goto bad;
494 		}
495 		off += sizeof(struct ip6_hdr);
496 
497 		n = m;
498 		while (n && n->m_len <= off) {
499 			off -= n->m_len;
500 			n = n->m_next;
501 		}
502 		if (!n)
503 			goto bad;
504 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
505 		*p = 0;
506 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
507 	}
508 
509 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
510 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
511 		if (oifp)
512 			icmp6_ifoutstat_inc(oifp, type, code);
513 		ICMP6STAT_INC(icp6s_outhist[type]);
514 	} else
515 		V_rip6stat.rip6s_opackets++;
516 
517 	goto freectl;
518 
519  bad:
520 	if (m)
521 		m_freem(m);
522 
523  freectl:
524 	if (control != NULL) {
525 		ip6_clearpktopts(&opt, -1);
526 		m_freem(control);
527 	}
528 	INP_WUNLOCK(in6p);
529 	return (error);
530 }
531 
532 /*
533  * Raw IPv6 socket option processing.
534  */
535 int
536 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
537 {
538 	int error;
539 
540 	if (sopt->sopt_level == IPPROTO_ICMPV6)
541 		/*
542 		 * XXX: is it better to call icmp6_ctloutput() directly
543 		 * from protosw?
544 		 */
545 		return (icmp6_ctloutput(so, sopt));
546 	else if (sopt->sopt_level != IPPROTO_IPV6)
547 		return (EINVAL);
548 
549 	error = 0;
550 
551 	switch (sopt->sopt_dir) {
552 	case SOPT_GET:
553 		switch (sopt->sopt_name) {
554 		case MRT6_INIT:
555 		case MRT6_DONE:
556 		case MRT6_ADD_MIF:
557 		case MRT6_DEL_MIF:
558 		case MRT6_ADD_MFC:
559 		case MRT6_DEL_MFC:
560 		case MRT6_PIM:
561 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
562 			    EOPNOTSUPP;
563 			break;
564 		case IPV6_CHECKSUM:
565 			error = ip6_raw_ctloutput(so, sopt);
566 			break;
567 		default:
568 			error = ip6_ctloutput(so, sopt);
569 			break;
570 		}
571 		break;
572 
573 	case SOPT_SET:
574 		switch (sopt->sopt_name) {
575 		case MRT6_INIT:
576 		case MRT6_DONE:
577 		case MRT6_ADD_MIF:
578 		case MRT6_DEL_MIF:
579 		case MRT6_ADD_MFC:
580 		case MRT6_DEL_MFC:
581 		case MRT6_PIM:
582 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
583 			    EOPNOTSUPP;
584 			break;
585 		case IPV6_CHECKSUM:
586 			error = ip6_raw_ctloutput(so, sopt);
587 			break;
588 		default:
589 			error = ip6_ctloutput(so, sopt);
590 			break;
591 		}
592 		break;
593 	}
594 
595 	return (error);
596 }
597 
598 static int
599 rip6_attach(struct socket *so, int proto, struct thread *td)
600 {
601 	struct inpcb *inp;
602 	struct icmp6_filter *filter;
603 	int error;
604 
605 	inp = sotoinpcb(so);
606 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
607 
608 	error = priv_check(td, PRIV_NETINET_RAW);
609 	if (error)
610 		return (error);
611 	error = soreserve(so, rip_sendspace, rip_recvspace);
612 	if (error)
613 		return (error);
614 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
615 	if (filter == NULL)
616 		return (ENOMEM);
617 	INP_INFO_WLOCK(&V_ripcbinfo);
618 	error = in_pcballoc(so, &V_ripcbinfo);
619 	if (error) {
620 		INP_INFO_WUNLOCK(&V_ripcbinfo);
621 		free(filter, M_PCB);
622 		return (error);
623 	}
624 	inp = (struct inpcb *)so->so_pcb;
625 	INP_INFO_WUNLOCK(&V_ripcbinfo);
626 	inp->inp_vflag |= INP_IPV6;
627 	inp->inp_ip_p = (long)proto;
628 	inp->in6p_hops = -1;	/* use kernel default */
629 	inp->in6p_cksum = -1;
630 	inp->in6p_icmp6filt = filter;
631 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
632 	INP_WUNLOCK(inp);
633 	return (0);
634 }
635 
636 static void
637 rip6_detach(struct socket *so)
638 {
639 	struct inpcb *inp;
640 
641 	inp = sotoinpcb(so);
642 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
643 
644 	if (so == V_ip6_mrouter && ip6_mrouter_done)
645 		ip6_mrouter_done();
646 	/* xxx: RSVP */
647 	INP_INFO_WLOCK(&V_ripcbinfo);
648 	INP_WLOCK(inp);
649 	free(inp->in6p_icmp6filt, M_PCB);
650 	in_pcbdetach(inp);
651 	in_pcbfree(inp);
652 	INP_INFO_WUNLOCK(&V_ripcbinfo);
653 }
654 
655 /* XXXRW: This can't ever be called. */
656 static void
657 rip6_abort(struct socket *so)
658 {
659 	struct inpcb *inp;
660 
661 	inp = sotoinpcb(so);
662 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
663 
664 	soisdisconnected(so);
665 }
666 
667 static void
668 rip6_close(struct socket *so)
669 {
670 	struct inpcb *inp;
671 
672 	inp = sotoinpcb(so);
673 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
674 
675 	soisdisconnected(so);
676 }
677 
678 static int
679 rip6_disconnect(struct socket *so)
680 {
681 	struct inpcb *inp;
682 
683 	inp = sotoinpcb(so);
684 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
685 
686 	if ((so->so_state & SS_ISCONNECTED) == 0)
687 		return (ENOTCONN);
688 	inp->in6p_faddr = in6addr_any;
689 	rip6_abort(so);
690 	return (0);
691 }
692 
693 static int
694 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
695 {
696 	struct inpcb *inp;
697 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
698 	struct ifaddr *ifa = NULL;
699 	int error = 0;
700 
701 	inp = sotoinpcb(so);
702 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
703 
704 	if (nam->sa_len != sizeof(*addr))
705 		return (EINVAL);
706 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
707 		return (error);
708 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
709 		return (EADDRNOTAVAIL);
710 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
711 		return (error);
712 
713 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
714 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
715 		return (EADDRNOTAVAIL);
716 	if (ifa != NULL &&
717 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
718 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
719 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
720 		ifa_free(ifa);
721 		return (EADDRNOTAVAIL);
722 	}
723 	if (ifa != NULL)
724 		ifa_free(ifa);
725 	INP_INFO_WLOCK(&V_ripcbinfo);
726 	INP_WLOCK(inp);
727 	inp->in6p_laddr = addr->sin6_addr;
728 	INP_WUNLOCK(inp);
729 	INP_INFO_WUNLOCK(&V_ripcbinfo);
730 	return (0);
731 }
732 
733 static int
734 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
735 {
736 	struct inpcb *inp;
737 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
738 	struct in6_addr in6a;
739 	struct ifnet *ifp = NULL;
740 	int error = 0, scope_ambiguous = 0;
741 
742 	inp = sotoinpcb(so);
743 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
744 
745 	if (nam->sa_len != sizeof(*addr))
746 		return (EINVAL);
747 	if (TAILQ_EMPTY(&V_ifnet))
748 		return (EADDRNOTAVAIL);
749 	if (addr->sin6_family != AF_INET6)
750 		return (EAFNOSUPPORT);
751 
752 	/*
753 	 * Application should provide a proper zone ID or the use of default
754 	 * zone IDs should be enabled.  Unfortunately, some applications do
755 	 * not behave as it should, so we need a workaround.  Even if an
756 	 * appropriate ID is not determined, we'll see if we can determine
757 	 * the outgoing interface.  If we can, determine the zone ID based on
758 	 * the interface below.
759 	 */
760 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
761 		scope_ambiguous = 1;
762 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
763 		return (error);
764 
765 	INP_INFO_WLOCK(&V_ripcbinfo);
766 	INP_WLOCK(inp);
767 	/* Source address selection. XXX: need pcblookup? */
768 	error = in6_selectsrc(addr, inp->in6p_outputopts,
769 	    inp, NULL, so->so_cred, &ifp, &in6a);
770 	if (error) {
771 		INP_WUNLOCK(inp);
772 		INP_INFO_WUNLOCK(&V_ripcbinfo);
773 		return (error);
774 	}
775 
776 	/* XXX: see above */
777 	if (ifp && scope_ambiguous &&
778 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
779 		INP_WUNLOCK(inp);
780 		INP_INFO_WUNLOCK(&V_ripcbinfo);
781 		return (error);
782 	}
783 	inp->in6p_faddr = addr->sin6_addr;
784 	inp->in6p_laddr = in6a;
785 	soisconnected(so);
786 	INP_WUNLOCK(inp);
787 	INP_INFO_WUNLOCK(&V_ripcbinfo);
788 	return (0);
789 }
790 
791 static int
792 rip6_shutdown(struct socket *so)
793 {
794 	struct inpcb *inp;
795 
796 	inp = sotoinpcb(so);
797 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
798 
799 	INP_WLOCK(inp);
800 	socantsendmore(so);
801 	INP_WUNLOCK(inp);
802 	return (0);
803 }
804 
805 static int
806 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
807     struct mbuf *control, struct thread *td)
808 {
809 	struct inpcb *inp;
810 	struct sockaddr_in6 tmp;
811 	struct sockaddr_in6 *dst;
812 	int ret;
813 
814 	inp = sotoinpcb(so);
815 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
816 
817 	/* Always copy sockaddr to avoid overwrites. */
818 	/* Unlocked read. */
819 	if (so->so_state & SS_ISCONNECTED) {
820 		if (nam) {
821 			m_freem(m);
822 			return (EISCONN);
823 		}
824 		/* XXX */
825 		bzero(&tmp, sizeof(tmp));
826 		tmp.sin6_family = AF_INET6;
827 		tmp.sin6_len = sizeof(struct sockaddr_in6);
828 		INP_RLOCK(inp);
829 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
830 		    sizeof(struct in6_addr));
831 		INP_RUNLOCK(inp);
832 		dst = &tmp;
833 	} else {
834 		if (nam == NULL) {
835 			m_freem(m);
836 			return (ENOTCONN);
837 		}
838 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
839 			m_freem(m);
840 			return (EINVAL);
841 		}
842 		tmp = *(struct sockaddr_in6 *)nam;
843 		dst = &tmp;
844 
845 		if (dst->sin6_family == AF_UNSPEC) {
846 			/*
847 			 * XXX: we allow this case for backward
848 			 * compatibility to buggy applications that
849 			 * rely on old (and wrong) kernel behavior.
850 			 */
851 			log(LOG_INFO, "rip6 SEND: address family is "
852 			    "unspec. Assume AF_INET6\n");
853 			dst->sin6_family = AF_INET6;
854 		} else if (dst->sin6_family != AF_INET6) {
855 			m_freem(m);
856 			return(EAFNOSUPPORT);
857 		}
858 	}
859 	ret = rip6_output(m, so, dst, control);
860 	return (ret);
861 }
862 
863 struct pr_usrreqs rip6_usrreqs = {
864 	.pru_abort =		rip6_abort,
865 	.pru_attach =		rip6_attach,
866 	.pru_bind =		rip6_bind,
867 	.pru_connect =		rip6_connect,
868 	.pru_control =		in6_control,
869 	.pru_detach =		rip6_detach,
870 	.pru_disconnect =	rip6_disconnect,
871 	.pru_peeraddr =		in6_getpeeraddr,
872 	.pru_send =		rip6_send,
873 	.pru_shutdown =		rip6_shutdown,
874 	.pru_sockaddr =		in6_getsockaddr,
875 	.pru_close =		rip6_close,
876 };
877