xref: /freebsd/sys/netinet6/raw_ip6.c (revision a812392203d7c4c3f0db9d8a0f3391374c49c71f)
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/kernel.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 
84 #include <net/if.h>
85 #include <net/if_var.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 
95 #include <netinet/icmp6.h>
96 #include <netinet/ip6.h>
97 #include <netinet/ip_var.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/send.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 VNET_DECLARE(struct inpcbhead, ripcb);
122 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
123 #define	V_ripcb				VNET(ripcb)
124 #define	V_ripcbinfo			VNET(ripcbinfo)
125 
126 extern u_long	rip_sendspace;
127 extern u_long	rip_recvspace;
128 
129 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
130 VNET_PCPUSTAT_SYSINIT(rip6stat);
131 
132 #ifdef VIMAGE
133 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
134 #endif /* VIMAGE */
135 
136 /*
137  * Hooks for multicast routing. They all default to NULL, so leave them not
138  * initialized and rely on BSS being set to 0.
139  */
140 
141 /*
142  * The socket used to communicate with the multicast routing daemon.
143  */
144 VNET_DEFINE(struct socket *, ip6_mrouter);
145 
146 /*
147  * The various mrouter functions.
148  */
149 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
150 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
151 int (*ip6_mrouter_done)(void);
152 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
153 int (*mrt6_ioctl)(u_long, caddr_t);
154 
155 /*
156  * Setup generic address and protocol structures for raw_input routine, then
157  * pass them along with mbuf chain.
158  */
159 int
160 rip6_input(struct mbuf **mp, int *offp, int proto)
161 {
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 	RIP6STAT_INC(rip6s_ipackets);
171 
172 	init_sin6(&fromsa, m); /* general init */
173 
174 	ifp = m->m_pkthdr.rcvif;
175 
176 	INP_INFO_RLOCK(&V_ripcbinfo);
177 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
178 		/* XXX inp locking */
179 		if ((in6p->inp_vflag & INP_IPV6) == 0)
180 			continue;
181 		if (in6p->inp_ip_p &&
182 		    in6p->inp_ip_p != proto)
183 			continue;
184 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
185 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
186 			continue;
187 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
188 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
189 			continue;
190 		if (jailed_without_vnet(in6p->inp_cred)) {
191 			/*
192 			 * Allow raw socket in jail to receive multicast;
193 			 * assume process had PRIV_NETINET_RAW at attach,
194 			 * and fall through into normal filter path if so.
195 			 */
196 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
197 			    prison_check_ip6(in6p->inp_cred,
198 			    &ip6->ip6_dst) != 0)
199 				continue;
200 		}
201 		INP_RLOCK(in6p);
202 		if (in6p->in6p_cksum != -1) {
203 			RIP6STAT_INC(rip6s_isum);
204 			if (in6_cksum(m, proto, *offp,
205 			    m->m_pkthdr.len - *offp)) {
206 				INP_RUNLOCK(in6p);
207 				RIP6STAT_INC(rip6s_badsum);
208 				continue;
209 			}
210 		}
211 		/*
212 		 * If this raw socket has multicast state, and we
213 		 * have received a multicast, check if this socket
214 		 * should receive it, as multicast filtering is now
215 		 * the responsibility of the transport layer.
216 		 */
217 		if (in6p->in6p_moptions &&
218 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
219 			/*
220 			 * If the incoming datagram is for MLD, allow it
221 			 * through unconditionally to the raw socket.
222 			 *
223 			 * Use the M_RTALERT_MLD flag to check for MLD
224 			 * traffic without having to inspect the mbuf chain
225 			 * more deeply, as all MLDv1/v2 host messages MUST
226 			 * contain the Router Alert option.
227 			 *
228 			 * In the case of MLDv1, we may not have explicitly
229 			 * joined the group, and may have set IFF_ALLMULTI
230 			 * on the interface. im6o_mc_filter() may discard
231 			 * control traffic we actually need to see.
232 			 *
233 			 * Userland multicast routing daemons should continue
234 			 * filter the control traffic appropriately.
235 			 */
236 			int blocked;
237 
238 			blocked = MCAST_PASS;
239 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
240 				struct sockaddr_in6 mcaddr;
241 
242 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
243 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
244 				mcaddr.sin6_family = AF_INET6;
245 				mcaddr.sin6_addr = ip6->ip6_dst;
246 
247 				blocked = im6o_mc_filter(in6p->in6p_moptions,
248 				    ifp,
249 				    (struct sockaddr *)&mcaddr,
250 				    (struct sockaddr *)&fromsa);
251 			}
252 			if (blocked != MCAST_PASS) {
253 				IP6STAT_INC(ip6s_notmember);
254 				INP_RUNLOCK(in6p);
255 				continue;
256 			}
257 		}
258 		if (last != NULL) {
259 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
260 
261 #ifdef IPSEC
262 			/*
263 			 * Check AH/ESP integrity.
264 			 */
265 			if (n && ipsec6_in_reject(n, last)) {
266 				m_freem(n);
267 				/* Do not inject data into pcb. */
268 			} else
269 #endif /* IPSEC */
270 			if (n) {
271 				if (last->inp_flags & INP_CONTROLOPTS ||
272 				    last->inp_socket->so_options & SO_TIMESTAMP)
273 					ip6_savecontrol(last, n, &opts);
274 				/* strip intermediate headers */
275 				m_adj(n, *offp);
276 				if (sbappendaddr(&last->inp_socket->so_rcv,
277 						(struct sockaddr *)&fromsa,
278 						 n, opts) == 0) {
279 					m_freem(n);
280 					if (opts)
281 						m_freem(opts);
282 					RIP6STAT_INC(rip6s_fullsock);
283 				} else
284 					sorwakeup(last->inp_socket);
285 				opts = NULL;
286 			}
287 			INP_RUNLOCK(last);
288 		}
289 		last = in6p;
290 	}
291 	INP_INFO_RUNLOCK(&V_ripcbinfo);
292 #ifdef IPSEC
293 	/*
294 	 * Check AH/ESP integrity.
295 	 */
296 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
297 		m_freem(m);
298 		IP6STAT_DEC(ip6s_delivered);
299 		/* Do not inject data into pcb. */
300 		INP_RUNLOCK(last);
301 	} else
302 #endif /* IPSEC */
303 	if (last != NULL) {
304 		if (last->inp_flags & INP_CONTROLOPTS ||
305 		    last->inp_socket->so_options & SO_TIMESTAMP)
306 			ip6_savecontrol(last, m, &opts);
307 		/* Strip intermediate headers. */
308 		m_adj(m, *offp);
309 		if (sbappendaddr(&last->inp_socket->so_rcv,
310 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
311 			m_freem(m);
312 			if (opts)
313 				m_freem(opts);
314 			RIP6STAT_INC(rip6s_fullsock);
315 		} else
316 			sorwakeup(last->inp_socket);
317 		INP_RUNLOCK(last);
318 	} else {
319 		RIP6STAT_INC(rip6s_nosock);
320 		if (m->m_flags & M_MCAST)
321 			RIP6STAT_INC(rip6s_nosockmcast);
322 		if (proto == IPPROTO_NONE)
323 			m_freem(m);
324 		else {
325 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
326 			icmp6_error(m, ICMP6_PARAM_PROB,
327 			    ICMP6_PARAMPROB_NEXTHEADER,
328 			    prvnxtp - mtod(m, char *));
329 		}
330 		IP6STAT_DEC(ip6s_delivered);
331 	}
332 	return (IPPROTO_DONE);
333 }
334 
335 void
336 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
337 {
338 	struct ip6_hdr *ip6;
339 	struct mbuf *m;
340 	int off = 0;
341 	struct ip6ctlparam *ip6cp = NULL;
342 	const struct sockaddr_in6 *sa6_src = NULL;
343 	void *cmdarg;
344 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
345 
346 	if (sa->sa_family != AF_INET6 ||
347 	    sa->sa_len != sizeof(struct sockaddr_in6))
348 		return;
349 
350 	if ((unsigned)cmd >= PRC_NCMDS)
351 		return;
352 	if (PRC_IS_REDIRECT(cmd))
353 		notify = in6_rtchange, d = NULL;
354 	else if (cmd == PRC_HOSTDEAD)
355 		d = NULL;
356 	else if (inet6ctlerrmap[cmd] == 0)
357 		return;
358 
359 	/*
360 	 * If the parameter is from icmp6, decode it.
361 	 */
362 	if (d != NULL) {
363 		ip6cp = (struct ip6ctlparam *)d;
364 		m = ip6cp->ip6c_m;
365 		ip6 = ip6cp->ip6c_ip6;
366 		off = ip6cp->ip6c_off;
367 		cmdarg = ip6cp->ip6c_cmdarg;
368 		sa6_src = ip6cp->ip6c_src;
369 	} else {
370 		m = NULL;
371 		ip6 = NULL;
372 		cmdarg = NULL;
373 		sa6_src = &sa6_any;
374 	}
375 
376 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
377 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
378 }
379 
380 /*
381  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
382  * may have setup with control call.
383  */
384 int
385 rip6_output(struct mbuf *m, struct socket *so, ...)
386 {
387 	struct mbuf *control;
388 	struct m_tag *mtag;
389 	struct sockaddr_in6 *dstsock;
390 	struct in6_addr *dst;
391 	struct ip6_hdr *ip6;
392 	struct inpcb *in6p;
393 	u_int	plen = m->m_pkthdr.len;
394 	int error = 0;
395 	struct ip6_pktopts opt, *optp;
396 	struct ifnet *oifp = NULL;
397 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
398 	int scope_ambiguous = 0;
399 	int use_defzone = 0;
400 	struct in6_addr in6a;
401 	va_list ap;
402 
403 	va_start(ap, so);
404 	dstsock = va_arg(ap, struct sockaddr_in6 *);
405 	control = va_arg(ap, struct mbuf *);
406 	va_end(ap);
407 
408 	in6p = sotoinpcb(so);
409 	INP_WLOCK(in6p);
410 
411 	dst = &dstsock->sin6_addr;
412 	if (control != NULL) {
413 		if ((error = ip6_setpktopts(control, &opt,
414 		    in6p->in6p_outputopts, so->so_cred,
415 		    so->so_proto->pr_protocol)) != 0) {
416 			goto bad;
417 		}
418 		optp = &opt;
419 	} else
420 		optp = in6p->in6p_outputopts;
421 
422 	/*
423 	 * Check and convert scope zone ID into internal form.
424 	 *
425 	 * XXX: we may still need to determine the zone later.
426 	 */
427 	if (!(so->so_state & SS_ISCONNECTED)) {
428 		if (!optp || !optp->ip6po_pktinfo ||
429 		    !optp->ip6po_pktinfo->ipi6_ifindex)
430 			use_defzone = V_ip6_use_defzone;
431 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
432 			scope_ambiguous = 1;
433 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
434 			goto bad;
435 	}
436 
437 	/*
438 	 * For an ICMPv6 packet, we should know its type and code to update
439 	 * statistics.
440 	 */
441 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
442 		struct icmp6_hdr *icmp6;
443 		if (m->m_len < sizeof(struct icmp6_hdr) &&
444 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
445 			error = ENOBUFS;
446 			goto bad;
447 		}
448 		icmp6 = mtod(m, struct icmp6_hdr *);
449 		type = icmp6->icmp6_type;
450 		code = icmp6->icmp6_code;
451 	}
452 
453 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
454 	if (m == NULL) {
455 		error = ENOBUFS;
456 		goto bad;
457 	}
458 	ip6 = mtod(m, struct ip6_hdr *);
459 
460 	/*
461 	 * Source address selection.
462 	 */
463 	error = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
464 	    &oifp, &in6a);
465 	if (error)
466 		goto bad;
467 	error = prison_check_ip6(in6p->inp_cred, &in6a);
468 	if (error != 0)
469 		goto bad;
470 	ip6->ip6_src = in6a;
471 
472 	if (oifp && scope_ambiguous) {
473 		/*
474 		 * Application should provide a proper zone ID or the use of
475 		 * default zone IDs should be enabled.  Unfortunately, some
476 		 * applications do not behave as it should, so we need a
477 		 * workaround.  Even if an appropriate ID is not determined
478 		 * (when it's required), if we can determine the outgoing
479 		 * interface. determine the zone ID based on the interface.
480 		 */
481 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
482 		if (error != 0)
483 			goto bad;
484 	}
485 	ip6->ip6_dst = dstsock->sin6_addr;
486 
487 	/*
488 	 * Fill in the rest of the IPv6 header fields.
489 	 */
490 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
491 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
492 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
493 	    (IPV6_VERSION & IPV6_VERSION_MASK);
494 
495 	/*
496 	 * ip6_plen will be filled in ip6_output, so not fill it here.
497 	 */
498 	ip6->ip6_nxt = in6p->inp_ip_p;
499 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
500 
501 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
502 	    in6p->in6p_cksum != -1) {
503 		struct mbuf *n;
504 		int off;
505 		u_int16_t *p;
506 
507 		/* Compute checksum. */
508 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
509 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
510 		else
511 			off = in6p->in6p_cksum;
512 		if (plen < off + 1) {
513 			error = EINVAL;
514 			goto bad;
515 		}
516 		off += sizeof(struct ip6_hdr);
517 
518 		n = m;
519 		while (n && n->m_len <= off) {
520 			off -= n->m_len;
521 			n = n->m_next;
522 		}
523 		if (!n)
524 			goto bad;
525 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
526 		*p = 0;
527 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
528 	}
529 
530 	/*
531 	 * Send RA/RS messages to user land for protection, before sending
532 	 * them to rtadvd/rtsol.
533 	 */
534 	if ((send_sendso_input_hook != NULL) &&
535 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
536 		switch (type) {
537 		case ND_ROUTER_ADVERT:
538 		case ND_ROUTER_SOLICIT:
539 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
540 				sizeof(unsigned short), M_NOWAIT);
541 			if (mtag == NULL)
542 				goto bad;
543 			m_tag_prepend(m, mtag);
544 		}
545 	}
546 
547 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
548 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
549 		if (oifp)
550 			icmp6_ifoutstat_inc(oifp, type, code);
551 		ICMP6STAT_INC(icp6s_outhist[type]);
552 	} else
553 		RIP6STAT_INC(rip6s_opackets);
554 
555 	goto freectl;
556 
557  bad:
558 	if (m)
559 		m_freem(m);
560 
561  freectl:
562 	if (control != NULL) {
563 		ip6_clearpktopts(&opt, -1);
564 		m_freem(control);
565 	}
566 	INP_WUNLOCK(in6p);
567 	return (error);
568 }
569 
570 /*
571  * Raw IPv6 socket option processing.
572  */
573 int
574 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
575 {
576 	struct inpcb *inp;
577 	int error;
578 
579 	if (sopt->sopt_level == IPPROTO_ICMPV6)
580 		/*
581 		 * XXX: is it better to call icmp6_ctloutput() directly
582 		 * from protosw?
583 		 */
584 		return (icmp6_ctloutput(so, sopt));
585 	else if (sopt->sopt_level != IPPROTO_IPV6) {
586 		if (sopt->sopt_level == SOL_SOCKET &&
587 		    sopt->sopt_name == SO_SETFIB) {
588 			inp = sotoinpcb(so);
589 			INP_WLOCK(inp);
590 			inp->inp_inc.inc_fibnum = so->so_fibnum;
591 			INP_WUNLOCK(inp);
592 			return (0);
593 		}
594 		return (EINVAL);
595 	}
596 
597 	error = 0;
598 
599 	switch (sopt->sopt_dir) {
600 	case SOPT_GET:
601 		switch (sopt->sopt_name) {
602 		case MRT6_INIT:
603 		case MRT6_DONE:
604 		case MRT6_ADD_MIF:
605 		case MRT6_DEL_MIF:
606 		case MRT6_ADD_MFC:
607 		case MRT6_DEL_MFC:
608 		case MRT6_PIM:
609 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
610 			    EOPNOTSUPP;
611 			break;
612 		case IPV6_CHECKSUM:
613 			error = ip6_raw_ctloutput(so, sopt);
614 			break;
615 		default:
616 			error = ip6_ctloutput(so, sopt);
617 			break;
618 		}
619 		break;
620 
621 	case SOPT_SET:
622 		switch (sopt->sopt_name) {
623 		case MRT6_INIT:
624 		case MRT6_DONE:
625 		case MRT6_ADD_MIF:
626 		case MRT6_DEL_MIF:
627 		case MRT6_ADD_MFC:
628 		case MRT6_DEL_MFC:
629 		case MRT6_PIM:
630 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
631 			    EOPNOTSUPP;
632 			break;
633 		case IPV6_CHECKSUM:
634 			error = ip6_raw_ctloutput(so, sopt);
635 			break;
636 		default:
637 			error = ip6_ctloutput(so, sopt);
638 			break;
639 		}
640 		break;
641 	}
642 
643 	return (error);
644 }
645 
646 static int
647 rip6_attach(struct socket *so, int proto, struct thread *td)
648 {
649 	struct inpcb *inp;
650 	struct icmp6_filter *filter;
651 	int error;
652 
653 	inp = sotoinpcb(so);
654 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
655 
656 	error = priv_check(td, PRIV_NETINET_RAW);
657 	if (error)
658 		return (error);
659 	error = soreserve(so, rip_sendspace, rip_recvspace);
660 	if (error)
661 		return (error);
662 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
663 	if (filter == NULL)
664 		return (ENOMEM);
665 	INP_INFO_WLOCK(&V_ripcbinfo);
666 	error = in_pcballoc(so, &V_ripcbinfo);
667 	if (error) {
668 		INP_INFO_WUNLOCK(&V_ripcbinfo);
669 		free(filter, M_PCB);
670 		return (error);
671 	}
672 	inp = (struct inpcb *)so->so_pcb;
673 	INP_INFO_WUNLOCK(&V_ripcbinfo);
674 	inp->inp_vflag |= INP_IPV6;
675 	inp->inp_ip_p = (long)proto;
676 	inp->in6p_hops = -1;	/* use kernel default */
677 	inp->in6p_cksum = -1;
678 	inp->in6p_icmp6filt = filter;
679 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
680 	INP_WUNLOCK(inp);
681 	return (0);
682 }
683 
684 static void
685 rip6_detach(struct socket *so)
686 {
687 	struct inpcb *inp;
688 
689 	inp = sotoinpcb(so);
690 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
691 
692 	if (so == V_ip6_mrouter && ip6_mrouter_done)
693 		ip6_mrouter_done();
694 	/* xxx: RSVP */
695 	INP_INFO_WLOCK(&V_ripcbinfo);
696 	INP_WLOCK(inp);
697 	free(inp->in6p_icmp6filt, M_PCB);
698 	in_pcbdetach(inp);
699 	in_pcbfree(inp);
700 	INP_INFO_WUNLOCK(&V_ripcbinfo);
701 }
702 
703 /* XXXRW: This can't ever be called. */
704 static void
705 rip6_abort(struct socket *so)
706 {
707 	struct inpcb *inp;
708 
709 	inp = sotoinpcb(so);
710 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
711 
712 	soisdisconnected(so);
713 }
714 
715 static void
716 rip6_close(struct socket *so)
717 {
718 	struct inpcb *inp;
719 
720 	inp = sotoinpcb(so);
721 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
722 
723 	soisdisconnected(so);
724 }
725 
726 static int
727 rip6_disconnect(struct socket *so)
728 {
729 	struct inpcb *inp;
730 
731 	inp = sotoinpcb(so);
732 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
733 
734 	if ((so->so_state & SS_ISCONNECTED) == 0)
735 		return (ENOTCONN);
736 	inp->in6p_faddr = in6addr_any;
737 	rip6_abort(so);
738 	return (0);
739 }
740 
741 static int
742 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
743 {
744 	struct inpcb *inp;
745 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
746 	struct ifaddr *ifa = NULL;
747 	int error = 0;
748 
749 	inp = sotoinpcb(so);
750 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
751 
752 	if (nam->sa_len != sizeof(*addr))
753 		return (EINVAL);
754 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
755 		return (error);
756 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
757 		return (EADDRNOTAVAIL);
758 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
759 		return (error);
760 
761 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
762 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
763 		return (EADDRNOTAVAIL);
764 	if (ifa != NULL &&
765 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
766 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
767 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
768 		ifa_free(ifa);
769 		return (EADDRNOTAVAIL);
770 	}
771 	if (ifa != NULL)
772 		ifa_free(ifa);
773 	INP_INFO_WLOCK(&V_ripcbinfo);
774 	INP_WLOCK(inp);
775 	inp->in6p_laddr = addr->sin6_addr;
776 	INP_WUNLOCK(inp);
777 	INP_INFO_WUNLOCK(&V_ripcbinfo);
778 	return (0);
779 }
780 
781 static int
782 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
783 {
784 	struct inpcb *inp;
785 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
786 	struct in6_addr in6a;
787 	struct ifnet *ifp = NULL;
788 	int error = 0, scope_ambiguous = 0;
789 
790 	inp = sotoinpcb(so);
791 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
792 
793 	if (nam->sa_len != sizeof(*addr))
794 		return (EINVAL);
795 	if (TAILQ_EMPTY(&V_ifnet))
796 		return (EADDRNOTAVAIL);
797 	if (addr->sin6_family != AF_INET6)
798 		return (EAFNOSUPPORT);
799 
800 	/*
801 	 * Application should provide a proper zone ID or the use of default
802 	 * zone IDs should be enabled.  Unfortunately, some applications do
803 	 * not behave as it should, so we need a workaround.  Even if an
804 	 * appropriate ID is not determined, we'll see if we can determine
805 	 * the outgoing interface.  If we can, determine the zone ID based on
806 	 * the interface below.
807 	 */
808 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
809 		scope_ambiguous = 1;
810 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
811 		return (error);
812 
813 	INP_INFO_WLOCK(&V_ripcbinfo);
814 	INP_WLOCK(inp);
815 	/* Source address selection. XXX: need pcblookup? */
816 	error = in6_selectsrc(addr, inp->in6p_outputopts,
817 	    inp, NULL, so->so_cred, &ifp, &in6a);
818 	if (error) {
819 		INP_WUNLOCK(inp);
820 		INP_INFO_WUNLOCK(&V_ripcbinfo);
821 		return (error);
822 	}
823 
824 	/* XXX: see above */
825 	if (ifp && scope_ambiguous &&
826 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
827 		INP_WUNLOCK(inp);
828 		INP_INFO_WUNLOCK(&V_ripcbinfo);
829 		return (error);
830 	}
831 	inp->in6p_faddr = addr->sin6_addr;
832 	inp->in6p_laddr = in6a;
833 	soisconnected(so);
834 	INP_WUNLOCK(inp);
835 	INP_INFO_WUNLOCK(&V_ripcbinfo);
836 	return (0);
837 }
838 
839 static int
840 rip6_shutdown(struct socket *so)
841 {
842 	struct inpcb *inp;
843 
844 	inp = sotoinpcb(so);
845 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
846 
847 	INP_WLOCK(inp);
848 	socantsendmore(so);
849 	INP_WUNLOCK(inp);
850 	return (0);
851 }
852 
853 static int
854 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
855     struct mbuf *control, struct thread *td)
856 {
857 	struct inpcb *inp;
858 	struct sockaddr_in6 tmp;
859 	struct sockaddr_in6 *dst;
860 	int ret;
861 
862 	inp = sotoinpcb(so);
863 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
864 
865 	/* Always copy sockaddr to avoid overwrites. */
866 	/* Unlocked read. */
867 	if (so->so_state & SS_ISCONNECTED) {
868 		if (nam) {
869 			m_freem(m);
870 			return (EISCONN);
871 		}
872 		/* XXX */
873 		bzero(&tmp, sizeof(tmp));
874 		tmp.sin6_family = AF_INET6;
875 		tmp.sin6_len = sizeof(struct sockaddr_in6);
876 		INP_RLOCK(inp);
877 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
878 		    sizeof(struct in6_addr));
879 		INP_RUNLOCK(inp);
880 		dst = &tmp;
881 	} else {
882 		if (nam == NULL) {
883 			m_freem(m);
884 			return (ENOTCONN);
885 		}
886 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
887 			m_freem(m);
888 			return (EINVAL);
889 		}
890 		tmp = *(struct sockaddr_in6 *)nam;
891 		dst = &tmp;
892 
893 		if (dst->sin6_family == AF_UNSPEC) {
894 			/*
895 			 * XXX: we allow this case for backward
896 			 * compatibility to buggy applications that
897 			 * rely on old (and wrong) kernel behavior.
898 			 */
899 			log(LOG_INFO, "rip6 SEND: address family is "
900 			    "unspec. Assume AF_INET6\n");
901 			dst->sin6_family = AF_INET6;
902 		} else if (dst->sin6_family != AF_INET6) {
903 			m_freem(m);
904 			return(EAFNOSUPPORT);
905 		}
906 	}
907 	ret = rip6_output(m, so, dst, control);
908 	return (ret);
909 }
910 
911 struct pr_usrreqs rip6_usrreqs = {
912 	.pru_abort =		rip6_abort,
913 	.pru_attach =		rip6_attach,
914 	.pru_bind =		rip6_bind,
915 	.pru_connect =		rip6_connect,
916 	.pru_control =		in6_control,
917 	.pru_detach =		rip6_detach,
918 	.pru_disconnect =	rip6_disconnect,
919 	.pru_peeraddr =		in6_getpeeraddr,
920 	.pru_send =		rip6_send,
921 	.pru_shutdown =		rip6_shutdown,
922 	.pru_sockaddr =		in6_getsockaddr,
923 	.pru_close =		rip6_close,
924 };
925