xref: /freebsd/sys/netinet6/raw_ip6.c (revision 595e514d0df2bac5b813d35f83e32875dbf16a83)
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 <netinet/ip_var.h>
96 #include <netinet6/ip6protosw.h>
97 #include <netinet6/ip6_mroute.h>
98 #include <netinet6/in6_pcb.h>
99 #include <netinet6/ip6_var.h>
100 #include <netinet6/nd6.h>
101 #include <netinet6/raw_ip6.h>
102 #include <netinet6/scope6_var.h>
103 #include <netinet6/send.h>
104 
105 #ifdef IPSEC
106 #include <netipsec/ipsec.h>
107 #include <netipsec/ipsec6.h>
108 #endif /* IPSEC */
109 
110 #include <machine/stdarg.h>
111 
112 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
113 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
114 
115 /*
116  * Raw interface to IP6 protocol.
117  */
118 
119 VNET_DECLARE(struct inpcbhead, ripcb);
120 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
121 #define	V_ripcb				VNET(ripcb)
122 #define	V_ripcbinfo			VNET(ripcbinfo)
123 
124 extern u_long	rip_sendspace;
125 extern u_long	rip_recvspace;
126 
127 VNET_DEFINE(struct rip6stat, rip6stat);
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 VNET_DEFINE(struct socket *, ip6_mrouter);
138 
139 /*
140  * The various mrouter functions.
141  */
142 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
143 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
144 int (*ip6_mrouter_done)(void);
145 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
146 int (*mrt6_ioctl)(u_long, caddr_t);
147 
148 /*
149  * Setup generic address and protocol structures for raw_input routine, then
150  * pass them along with mbuf chain.
151  */
152 int
153 rip6_input(struct mbuf **mp, int *offp, int proto)
154 {
155 	struct ifnet *ifp;
156 	struct mbuf *m = *mp;
157 	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
158 	register struct inpcb *in6p;
159 	struct inpcb *last = 0;
160 	struct mbuf *opts = NULL;
161 	struct sockaddr_in6 fromsa;
162 
163 	V_rip6stat.rip6s_ipackets++;
164 
165 	if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
166 		/* XXX Send icmp6 host/port unreach? */
167 		m_freem(m);
168 		return (IPPROTO_DONE);
169 	}
170 
171 	init_sin6(&fromsa, m); /* general init */
172 
173 	ifp = m->m_pkthdr.rcvif;
174 
175 	INP_INFO_RLOCK(&V_ripcbinfo);
176 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
177 		/* XXX inp locking */
178 		if ((in6p->inp_vflag & INP_IPV6) == 0)
179 			continue;
180 		if (in6p->inp_ip_p &&
181 		    in6p->inp_ip_p != proto)
182 			continue;
183 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
184 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
185 			continue;
186 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
187 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
188 			continue;
189 		if (jailed_without_vnet(in6p->inp_cred)) {
190 			/*
191 			 * Allow raw socket in jail to receive multicast;
192 			 * assume process had PRIV_NETINET_RAW at attach,
193 			 * and fall through into normal filter path if so.
194 			 */
195 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
196 			    prison_check_ip6(in6p->inp_cred,
197 			    &ip6->ip6_dst) != 0)
198 				continue;
199 		}
200 		INP_RLOCK(in6p);
201 		if (in6p->in6p_cksum != -1) {
202 			V_rip6stat.rip6s_isum++;
203 			if (in6_cksum(m, proto, *offp,
204 			    m->m_pkthdr.len - *offp)) {
205 				INP_RUNLOCK(in6p);
206 				V_rip6stat.rip6s_badsum++;
207 				continue;
208 			}
209 		}
210 		/*
211 		 * If this raw socket has multicast state, and we
212 		 * have received a multicast, check if this socket
213 		 * should receive it, as multicast filtering is now
214 		 * the responsibility of the transport layer.
215 		 */
216 		if (in6p->in6p_moptions &&
217 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
218 			/*
219 			 * If the incoming datagram is for MLD, allow it
220 			 * through unconditionally to the raw socket.
221 			 *
222 			 * Use the M_RTALERT_MLD flag to check for MLD
223 			 * traffic without having to inspect the mbuf chain
224 			 * more deeply, as all MLDv1/v2 host messages MUST
225 			 * contain the Router Alert option.
226 			 *
227 			 * In the case of MLDv1, we may not have explicitly
228 			 * joined the group, and may have set IFF_ALLMULTI
229 			 * on the interface. im6o_mc_filter() may discard
230 			 * control traffic we actually need to see.
231 			 *
232 			 * Userland multicast routing daemons should continue
233 			 * filter the control traffic appropriately.
234 			 */
235 			int blocked;
236 
237 			blocked = MCAST_PASS;
238 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
239 				struct sockaddr_in6 mcaddr;
240 
241 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
242 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
243 				mcaddr.sin6_family = AF_INET6;
244 				mcaddr.sin6_addr = ip6->ip6_dst;
245 
246 				blocked = im6o_mc_filter(in6p->in6p_moptions,
247 				    ifp,
248 				    (struct sockaddr *)&mcaddr,
249 				    (struct sockaddr *)&fromsa);
250 			}
251 			if (blocked != MCAST_PASS) {
252 				IP6STAT_INC(ip6s_notmember);
253 				INP_RUNLOCK(in6p);
254 				continue;
255 			}
256 		}
257 		if (last != NULL) {
258 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
259 
260 #ifdef IPSEC
261 			/*
262 			 * Check AH/ESP integrity.
263 			 */
264 			if (n && ipsec6_in_reject(n, last)) {
265 				m_freem(n);
266 				V_ipsec6stat.in_polvio++;
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 					V_rip6stat.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 		V_ipsec6stat.in_polvio++;
299 		IP6STAT_DEC(ip6s_delivered);
300 		/* Do not inject data into pcb. */
301 		INP_RUNLOCK(last);
302 	} else
303 #endif /* IPSEC */
304 	if (last != NULL) {
305 		if (last->inp_flags & INP_CONTROLOPTS ||
306 		    last->inp_socket->so_options & SO_TIMESTAMP)
307 			ip6_savecontrol(last, m, &opts);
308 		/* Strip intermediate headers. */
309 		m_adj(m, *offp);
310 		if (sbappendaddr(&last->inp_socket->so_rcv,
311 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
312 			m_freem(m);
313 			if (opts)
314 				m_freem(opts);
315 			V_rip6stat.rip6s_fullsock++;
316 		} else
317 			sorwakeup(last->inp_socket);
318 		INP_RUNLOCK(last);
319 	} else {
320 		V_rip6stat.rip6s_nosock++;
321 		if (m->m_flags & M_MCAST)
322 			V_rip6stat.rip6s_nosockmcast++;
323 		if (proto == IPPROTO_NONE)
324 			m_freem(m);
325 		else {
326 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
327 			icmp6_error(m, ICMP6_PARAM_PROB,
328 			    ICMP6_PARAMPROB_NEXTHEADER,
329 			    prvnxtp - mtod(m, char *));
330 		}
331 		IP6STAT_DEC(ip6s_delivered);
332 	}
333 	return (IPPROTO_DONE);
334 }
335 
336 void
337 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
338 {
339 	struct ip6_hdr *ip6;
340 	struct mbuf *m;
341 	int off = 0;
342 	struct ip6ctlparam *ip6cp = NULL;
343 	const struct sockaddr_in6 *sa6_src = NULL;
344 	void *cmdarg;
345 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
346 
347 	if (sa->sa_family != AF_INET6 ||
348 	    sa->sa_len != sizeof(struct sockaddr_in6))
349 		return;
350 
351 	if ((unsigned)cmd >= PRC_NCMDS)
352 		return;
353 	if (PRC_IS_REDIRECT(cmd))
354 		notify = in6_rtchange, d = NULL;
355 	else if (cmd == PRC_HOSTDEAD)
356 		d = NULL;
357 	else if (inet6ctlerrmap[cmd] == 0)
358 		return;
359 
360 	/*
361 	 * If the parameter is from icmp6, decode it.
362 	 */
363 	if (d != NULL) {
364 		ip6cp = (struct ip6ctlparam *)d;
365 		m = ip6cp->ip6c_m;
366 		ip6 = ip6cp->ip6c_ip6;
367 		off = ip6cp->ip6c_off;
368 		cmdarg = ip6cp->ip6c_cmdarg;
369 		sa6_src = ip6cp->ip6c_src;
370 	} else {
371 		m = NULL;
372 		ip6 = NULL;
373 		cmdarg = NULL;
374 		sa6_src = &sa6_any;
375 	}
376 
377 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
378 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
379 }
380 
381 /*
382  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
383  * may have setup with control call.
384  */
385 int
386 #if __STDC__
387 rip6_output(struct mbuf *m, ...)
388 #else
389 rip6_output(m, va_alist)
390 	struct mbuf *m;
391 	va_dcl
392 #endif
393 {
394 	struct mbuf *control;
395 	struct m_tag *mtag;
396 	struct socket *so;
397 	struct sockaddr_in6 *dstsock;
398 	struct in6_addr *dst;
399 	struct ip6_hdr *ip6;
400 	struct inpcb *in6p;
401 	u_int	plen = m->m_pkthdr.len;
402 	int error = 0;
403 	struct ip6_pktopts opt, *optp;
404 	struct ifnet *oifp = NULL;
405 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
406 	int scope_ambiguous = 0;
407 	int use_defzone = 0;
408 	struct in6_addr in6a;
409 	va_list ap;
410 
411 	va_start(ap, m);
412 	so = va_arg(ap, struct socket *);
413 	dstsock = va_arg(ap, struct sockaddr_in6 *);
414 	control = va_arg(ap, struct mbuf *);
415 	va_end(ap);
416 
417 	in6p = sotoinpcb(so);
418 	INP_WLOCK(in6p);
419 
420 	dst = &dstsock->sin6_addr;
421 	if (control != NULL) {
422 		if ((error = ip6_setpktopts(control, &opt,
423 		    in6p->in6p_outputopts, so->so_cred,
424 		    so->so_proto->pr_protocol)) != 0) {
425 			goto bad;
426 		}
427 		optp = &opt;
428 	} else
429 		optp = in6p->in6p_outputopts;
430 
431 	/*
432 	 * Check and convert scope zone ID into internal form.
433 	 *
434 	 * XXX: we may still need to determine the zone later.
435 	 */
436 	if (!(so->so_state & SS_ISCONNECTED)) {
437 		if (!optp || !optp->ip6po_pktinfo ||
438 		    !optp->ip6po_pktinfo->ipi6_ifindex)
439 			use_defzone = V_ip6_use_defzone;
440 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
441 			scope_ambiguous = 1;
442 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
443 			goto bad;
444 	}
445 
446 	/*
447 	 * For an ICMPv6 packet, we should know its type and code to update
448 	 * statistics.
449 	 */
450 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
451 		struct icmp6_hdr *icmp6;
452 		if (m->m_len < sizeof(struct icmp6_hdr) &&
453 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
454 			error = ENOBUFS;
455 			goto bad;
456 		}
457 		icmp6 = mtod(m, struct icmp6_hdr *);
458 		type = icmp6->icmp6_type;
459 		code = icmp6->icmp6_code;
460 	}
461 
462 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
463 	if (m == NULL) {
464 		error = ENOBUFS;
465 		goto bad;
466 	}
467 	ip6 = mtod(m, struct ip6_hdr *);
468 
469 	/*
470 	 * Source address selection.
471 	 */
472 	error = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
473 	    &oifp, &in6a);
474 	if (error)
475 		goto bad;
476 	error = prison_check_ip6(in6p->inp_cred, &in6a);
477 	if (error != 0)
478 		goto bad;
479 	ip6->ip6_src = in6a;
480 
481 	if (oifp && scope_ambiguous) {
482 		/*
483 		 * Application should provide a proper zone ID or the use of
484 		 * default zone IDs should be enabled.  Unfortunately, some
485 		 * applications do not behave as it should, so we need a
486 		 * workaround.  Even if an appropriate ID is not determined
487 		 * (when it's required), if we can determine the outgoing
488 		 * interface. determine the zone ID based on the interface.
489 		 */
490 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
491 		if (error != 0)
492 			goto bad;
493 	}
494 	ip6->ip6_dst = dstsock->sin6_addr;
495 
496 	/*
497 	 * Fill in the rest of the IPv6 header fields.
498 	 */
499 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
500 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
501 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
502 	    (IPV6_VERSION & IPV6_VERSION_MASK);
503 
504 	/*
505 	 * ip6_plen will be filled in ip6_output, so not fill it here.
506 	 */
507 	ip6->ip6_nxt = in6p->inp_ip_p;
508 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
509 
510 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
511 	    in6p->in6p_cksum != -1) {
512 		struct mbuf *n;
513 		int off;
514 		u_int16_t *p;
515 
516 		/* Compute checksum. */
517 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
518 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
519 		else
520 			off = in6p->in6p_cksum;
521 		if (plen < off + 1) {
522 			error = EINVAL;
523 			goto bad;
524 		}
525 		off += sizeof(struct ip6_hdr);
526 
527 		n = m;
528 		while (n && n->m_len <= off) {
529 			off -= n->m_len;
530 			n = n->m_next;
531 		}
532 		if (!n)
533 			goto bad;
534 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
535 		*p = 0;
536 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
537 	}
538 
539 	/*
540 	 * Send RA/RS messages to user land for protection, before sending
541 	 * them to rtadvd/rtsol.
542 	 */
543 	if ((send_sendso_input_hook != NULL) &&
544 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
545 		switch (type) {
546 		case ND_ROUTER_ADVERT:
547 		case ND_ROUTER_SOLICIT:
548 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
549 				sizeof(unsigned short), M_NOWAIT);
550 			if (mtag == NULL)
551 				goto bad;
552 			m_tag_prepend(m, mtag);
553 		}
554 	}
555 
556 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
557 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
558 		if (oifp)
559 			icmp6_ifoutstat_inc(oifp, type, code);
560 		ICMP6STAT_INC(icp6s_outhist[type]);
561 	} else
562 		V_rip6stat.rip6s_opackets++;
563 
564 	goto freectl;
565 
566  bad:
567 	if (m)
568 		m_freem(m);
569 
570  freectl:
571 	if (control != NULL) {
572 		ip6_clearpktopts(&opt, -1);
573 		m_freem(control);
574 	}
575 	INP_WUNLOCK(in6p);
576 	return (error);
577 }
578 
579 /*
580  * Raw IPv6 socket option processing.
581  */
582 int
583 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
584 {
585 	struct inpcb *inp;
586 	int error;
587 
588 	if (sopt->sopt_level == IPPROTO_ICMPV6)
589 		/*
590 		 * XXX: is it better to call icmp6_ctloutput() directly
591 		 * from protosw?
592 		 */
593 		return (icmp6_ctloutput(so, sopt));
594 	else if (sopt->sopt_level != IPPROTO_IPV6) {
595 		if (sopt->sopt_level == SOL_SOCKET &&
596 		    sopt->sopt_name == SO_SETFIB) {
597 			inp = sotoinpcb(so);
598 			INP_WLOCK(inp);
599 			inp->inp_inc.inc_fibnum = so->so_fibnum;
600 			INP_WUNLOCK(inp);
601 			return (0);
602 		}
603 		return (EINVAL);
604 	}
605 
606 	error = 0;
607 
608 	switch (sopt->sopt_dir) {
609 	case SOPT_GET:
610 		switch (sopt->sopt_name) {
611 		case MRT6_INIT:
612 		case MRT6_DONE:
613 		case MRT6_ADD_MIF:
614 		case MRT6_DEL_MIF:
615 		case MRT6_ADD_MFC:
616 		case MRT6_DEL_MFC:
617 		case MRT6_PIM:
618 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
619 			    EOPNOTSUPP;
620 			break;
621 		case IPV6_CHECKSUM:
622 			error = ip6_raw_ctloutput(so, sopt);
623 			break;
624 		default:
625 			error = ip6_ctloutput(so, sopt);
626 			break;
627 		}
628 		break;
629 
630 	case SOPT_SET:
631 		switch (sopt->sopt_name) {
632 		case MRT6_INIT:
633 		case MRT6_DONE:
634 		case MRT6_ADD_MIF:
635 		case MRT6_DEL_MIF:
636 		case MRT6_ADD_MFC:
637 		case MRT6_DEL_MFC:
638 		case MRT6_PIM:
639 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
640 			    EOPNOTSUPP;
641 			break;
642 		case IPV6_CHECKSUM:
643 			error = ip6_raw_ctloutput(so, sopt);
644 			break;
645 		default:
646 			error = ip6_ctloutput(so, sopt);
647 			break;
648 		}
649 		break;
650 	}
651 
652 	return (error);
653 }
654 
655 static int
656 rip6_attach(struct socket *so, int proto, struct thread *td)
657 {
658 	struct inpcb *inp;
659 	struct icmp6_filter *filter;
660 	int error;
661 
662 	inp = sotoinpcb(so);
663 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
664 
665 	error = priv_check(td, PRIV_NETINET_RAW);
666 	if (error)
667 		return (error);
668 	error = soreserve(so, rip_sendspace, rip_recvspace);
669 	if (error)
670 		return (error);
671 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
672 	if (filter == NULL)
673 		return (ENOMEM);
674 	INP_INFO_WLOCK(&V_ripcbinfo);
675 	error = in_pcballoc(so, &V_ripcbinfo);
676 	if (error) {
677 		INP_INFO_WUNLOCK(&V_ripcbinfo);
678 		free(filter, M_PCB);
679 		return (error);
680 	}
681 	inp = (struct inpcb *)so->so_pcb;
682 	INP_INFO_WUNLOCK(&V_ripcbinfo);
683 	inp->inp_vflag |= INP_IPV6;
684 	inp->inp_ip_p = (long)proto;
685 	inp->in6p_hops = -1;	/* use kernel default */
686 	inp->in6p_cksum = -1;
687 	inp->in6p_icmp6filt = filter;
688 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
689 	INP_WUNLOCK(inp);
690 	return (0);
691 }
692 
693 static void
694 rip6_detach(struct socket *so)
695 {
696 	struct inpcb *inp;
697 
698 	inp = sotoinpcb(so);
699 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
700 
701 	if (so == V_ip6_mrouter && ip6_mrouter_done)
702 		ip6_mrouter_done();
703 	/* xxx: RSVP */
704 	INP_INFO_WLOCK(&V_ripcbinfo);
705 	INP_WLOCK(inp);
706 	free(inp->in6p_icmp6filt, M_PCB);
707 	in_pcbdetach(inp);
708 	in_pcbfree(inp);
709 	INP_INFO_WUNLOCK(&V_ripcbinfo);
710 }
711 
712 /* XXXRW: This can't ever be called. */
713 static void
714 rip6_abort(struct socket *so)
715 {
716 	struct inpcb *inp;
717 
718 	inp = sotoinpcb(so);
719 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
720 
721 	soisdisconnected(so);
722 }
723 
724 static void
725 rip6_close(struct socket *so)
726 {
727 	struct inpcb *inp;
728 
729 	inp = sotoinpcb(so);
730 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
731 
732 	soisdisconnected(so);
733 }
734 
735 static int
736 rip6_disconnect(struct socket *so)
737 {
738 	struct inpcb *inp;
739 
740 	inp = sotoinpcb(so);
741 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
742 
743 	if ((so->so_state & SS_ISCONNECTED) == 0)
744 		return (ENOTCONN);
745 	inp->in6p_faddr = in6addr_any;
746 	rip6_abort(so);
747 	return (0);
748 }
749 
750 static int
751 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
752 {
753 	struct inpcb *inp;
754 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
755 	struct ifaddr *ifa = NULL;
756 	int error = 0;
757 
758 	inp = sotoinpcb(so);
759 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
760 
761 	if (nam->sa_len != sizeof(*addr))
762 		return (EINVAL);
763 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
764 		return (error);
765 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
766 		return (EADDRNOTAVAIL);
767 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
768 		return (error);
769 
770 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
771 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
772 		return (EADDRNOTAVAIL);
773 	if (ifa != NULL &&
774 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
775 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
776 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
777 		ifa_free(ifa);
778 		return (EADDRNOTAVAIL);
779 	}
780 	if (ifa != NULL)
781 		ifa_free(ifa);
782 	INP_INFO_WLOCK(&V_ripcbinfo);
783 	INP_WLOCK(inp);
784 	inp->in6p_laddr = addr->sin6_addr;
785 	INP_WUNLOCK(inp);
786 	INP_INFO_WUNLOCK(&V_ripcbinfo);
787 	return (0);
788 }
789 
790 static int
791 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
792 {
793 	struct inpcb *inp;
794 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
795 	struct in6_addr in6a;
796 	struct ifnet *ifp = NULL;
797 	int error = 0, scope_ambiguous = 0;
798 
799 	inp = sotoinpcb(so);
800 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
801 
802 	if (nam->sa_len != sizeof(*addr))
803 		return (EINVAL);
804 	if (TAILQ_EMPTY(&V_ifnet))
805 		return (EADDRNOTAVAIL);
806 	if (addr->sin6_family != AF_INET6)
807 		return (EAFNOSUPPORT);
808 
809 	/*
810 	 * Application should provide a proper zone ID or the use of default
811 	 * zone IDs should be enabled.  Unfortunately, some applications do
812 	 * not behave as it should, so we need a workaround.  Even if an
813 	 * appropriate ID is not determined, we'll see if we can determine
814 	 * the outgoing interface.  If we can, determine the zone ID based on
815 	 * the interface below.
816 	 */
817 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
818 		scope_ambiguous = 1;
819 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
820 		return (error);
821 
822 	INP_INFO_WLOCK(&V_ripcbinfo);
823 	INP_WLOCK(inp);
824 	/* Source address selection. XXX: need pcblookup? */
825 	error = in6_selectsrc(addr, inp->in6p_outputopts,
826 	    inp, NULL, so->so_cred, &ifp, &in6a);
827 	if (error) {
828 		INP_WUNLOCK(inp);
829 		INP_INFO_WUNLOCK(&V_ripcbinfo);
830 		return (error);
831 	}
832 
833 	/* XXX: see above */
834 	if (ifp && scope_ambiguous &&
835 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
836 		INP_WUNLOCK(inp);
837 		INP_INFO_WUNLOCK(&V_ripcbinfo);
838 		return (error);
839 	}
840 	inp->in6p_faddr = addr->sin6_addr;
841 	inp->in6p_laddr = in6a;
842 	soisconnected(so);
843 	INP_WUNLOCK(inp);
844 	INP_INFO_WUNLOCK(&V_ripcbinfo);
845 	return (0);
846 }
847 
848 static int
849 rip6_shutdown(struct socket *so)
850 {
851 	struct inpcb *inp;
852 
853 	inp = sotoinpcb(so);
854 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
855 
856 	INP_WLOCK(inp);
857 	socantsendmore(so);
858 	INP_WUNLOCK(inp);
859 	return (0);
860 }
861 
862 static int
863 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
864     struct mbuf *control, struct thread *td)
865 {
866 	struct inpcb *inp;
867 	struct sockaddr_in6 tmp;
868 	struct sockaddr_in6 *dst;
869 	int ret;
870 
871 	inp = sotoinpcb(so);
872 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
873 
874 	/* Always copy sockaddr to avoid overwrites. */
875 	/* Unlocked read. */
876 	if (so->so_state & SS_ISCONNECTED) {
877 		if (nam) {
878 			m_freem(m);
879 			return (EISCONN);
880 		}
881 		/* XXX */
882 		bzero(&tmp, sizeof(tmp));
883 		tmp.sin6_family = AF_INET6;
884 		tmp.sin6_len = sizeof(struct sockaddr_in6);
885 		INP_RLOCK(inp);
886 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
887 		    sizeof(struct in6_addr));
888 		INP_RUNLOCK(inp);
889 		dst = &tmp;
890 	} else {
891 		if (nam == NULL) {
892 			m_freem(m);
893 			return (ENOTCONN);
894 		}
895 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
896 			m_freem(m);
897 			return (EINVAL);
898 		}
899 		tmp = *(struct sockaddr_in6 *)nam;
900 		dst = &tmp;
901 
902 		if (dst->sin6_family == AF_UNSPEC) {
903 			/*
904 			 * XXX: we allow this case for backward
905 			 * compatibility to buggy applications that
906 			 * rely on old (and wrong) kernel behavior.
907 			 */
908 			log(LOG_INFO, "rip6 SEND: address family is "
909 			    "unspec. Assume AF_INET6\n");
910 			dst->sin6_family = AF_INET6;
911 		} else if (dst->sin6_family != AF_INET6) {
912 			m_freem(m);
913 			return(EAFNOSUPPORT);
914 		}
915 	}
916 	ret = rip6_output(m, so, dst, control);
917 	return (ret);
918 }
919 
920 struct pr_usrreqs rip6_usrreqs = {
921 	.pru_abort =		rip6_abort,
922 	.pru_attach =		rip6_attach,
923 	.pru_bind =		rip6_bind,
924 	.pru_connect =		rip6_connect,
925 	.pru_control =		in6_control,
926 	.pru_detach =		rip6_detach,
927 	.pru_disconnect =	rip6_disconnect,
928 	.pru_peeraddr =		in6_getpeeraddr,
929 	.pru_send =		rip6_send,
930 	.pru_shutdown =		rip6_shutdown,
931 	.pru_sockaddr =		in6_getsockaddr,
932 	.pru_close =		rip6_close,
933 };
934