xref: /freebsd/sys/netinet6/raw_ip6.c (revision 3416500aef140042c64bc149cb1ec6620483bc44)
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  * 3. 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 #include <netipsec/ipsec_support.h>
108 
109 #include <machine/stdarg.h>
110 
111 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
112 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
113 
114 /*
115  * Raw interface to IP6 protocol.
116  */
117 
118 VNET_DECLARE(struct inpcbhead, ripcb);
119 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
120 #define	V_ripcb				VNET(ripcb)
121 #define	V_ripcbinfo			VNET(ripcbinfo)
122 
123 extern u_long	rip_sendspace;
124 extern u_long	rip_recvspace;
125 
126 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
127 VNET_PCPUSTAT_SYSINIT(rip6stat);
128 
129 #ifdef VIMAGE
130 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
131 #endif /* VIMAGE */
132 
133 /*
134  * Hooks for multicast routing. They all default to NULL, so leave them not
135  * initialized and rely on BSS being set to 0.
136  */
137 
138 /*
139  * The socket used to communicate with the multicast routing daemon.
140  */
141 VNET_DEFINE(struct socket *, ip6_mrouter);
142 
143 /*
144  * The various mrouter functions.
145  */
146 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
147 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
148 int (*ip6_mrouter_done)(void);
149 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
150 int (*mrt6_ioctl)(u_long, caddr_t);
151 
152 /*
153  * Setup generic address and protocol structures for raw_input routine, then
154  * pass them along with mbuf chain.
155  */
156 int
157 rip6_input(struct mbuf **mp, int *offp, int proto)
158 {
159 	struct ifnet *ifp;
160 	struct mbuf *m = *mp;
161 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
162 	struct inpcb *in6p;
163 	struct inpcb *last = NULL;
164 	struct mbuf *opts = NULL;
165 	struct sockaddr_in6 fromsa;
166 
167 	RIP6STAT_INC(rip6s_ipackets);
168 
169 	init_sin6(&fromsa, m, 0); /* 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_without_vnet(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 		INP_RLOCK(in6p);
199 		if (in6p->in6p_cksum != -1) {
200 			RIP6STAT_INC(rip6s_isum);
201 			if (in6_cksum(m, proto, *offp,
202 			    m->m_pkthdr.len - *offp)) {
203 				INP_RUNLOCK(in6p);
204 				RIP6STAT_INC(rip6s_badsum);
205 				continue;
206 			}
207 		}
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 			/*
217 			 * If the incoming datagram is for MLD, allow it
218 			 * through unconditionally to the raw socket.
219 			 *
220 			 * Use the M_RTALERT_MLD flag to check for MLD
221 			 * traffic without having to inspect the mbuf chain
222 			 * more deeply, as all MLDv1/v2 host messages MUST
223 			 * contain the Router Alert option.
224 			 *
225 			 * In the case of MLDv1, we may not have explicitly
226 			 * joined the group, and may have set IFF_ALLMULTI
227 			 * on the interface. im6o_mc_filter() may discard
228 			 * control traffic we actually need to see.
229 			 *
230 			 * Userland multicast routing daemons should continue
231 			 * filter the control traffic appropriately.
232 			 */
233 			int blocked;
234 
235 			blocked = MCAST_PASS;
236 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
237 				struct sockaddr_in6 mcaddr;
238 
239 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
240 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
241 				mcaddr.sin6_family = AF_INET6;
242 				mcaddr.sin6_addr = ip6->ip6_dst;
243 
244 				blocked = im6o_mc_filter(in6p->in6p_moptions,
245 				    ifp,
246 				    (struct sockaddr *)&mcaddr,
247 				    (struct sockaddr *)&fromsa);
248 			}
249 			if (blocked != MCAST_PASS) {
250 				IP6STAT_INC(ip6s_notmember);
251 				INP_RUNLOCK(in6p);
252 				continue;
253 			}
254 		}
255 		if (last != NULL) {
256 			struct mbuf *n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
257 
258 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
259 			/*
260 			 * Check AH/ESP integrity.
261 			 */
262 			if (IPSEC_ENABLED(ipv6)) {
263 				if (n != NULL &&
264 				    IPSEC_CHECK_POLICY(ipv6, n, last) != 0) {
265 					m_freem(n);
266 					/* Do not inject data into pcb. */
267 					n = NULL;
268 				}
269 			}
270 #endif /* IPSEC */
271 			if (n) {
272 				if (last->inp_flags & INP_CONTROLOPTS ||
273 				    last->inp_socket->so_options & SO_TIMESTAMP)
274 					ip6_savecontrol(last, n, &opts);
275 				/* strip intermediate headers */
276 				m_adj(n, *offp);
277 				if (sbappendaddr(&last->inp_socket->so_rcv,
278 						(struct sockaddr *)&fromsa,
279 						 n, opts) == 0) {
280 					m_freem(n);
281 					if (opts)
282 						m_freem(opts);
283 					RIP6STAT_INC(rip6s_fullsock);
284 				} else
285 					sorwakeup(last->inp_socket);
286 				opts = NULL;
287 			}
288 			INP_RUNLOCK(last);
289 		}
290 		last = in6p;
291 	}
292 	INP_INFO_RUNLOCK(&V_ripcbinfo);
293 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
294 	/*
295 	 * Check AH/ESP integrity.
296 	 */
297 	if (IPSEC_ENABLED(ipv6) && last != NULL &&
298 	    IPSEC_CHECK_POLICY(ipv6, m, last) != 0) {
299 		m_freem(m);
300 		IP6STAT_DEC(ip6s_delivered);
301 		/* Do not inject data into pcb. */
302 		INP_RUNLOCK(last);
303 	} else
304 #endif /* IPSEC */
305 	if (last != NULL) {
306 		if (last->inp_flags & INP_CONTROLOPTS ||
307 		    last->inp_socket->so_options & SO_TIMESTAMP)
308 			ip6_savecontrol(last, m, &opts);
309 		/* Strip intermediate headers. */
310 		m_adj(m, *offp);
311 		if (sbappendaddr(&last->inp_socket->so_rcv,
312 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
313 			m_freem(m);
314 			if (opts)
315 				m_freem(opts);
316 			RIP6STAT_INC(rip6s_fullsock);
317 		} else
318 			sorwakeup(last->inp_socket);
319 		INP_RUNLOCK(last);
320 	} else {
321 		RIP6STAT_INC(rip6s_nosock);
322 		if (m->m_flags & M_MCAST)
323 			RIP6STAT_INC(rip6s_nosockmcast);
324 		if (proto == IPPROTO_NONE)
325 			m_freem(m);
326 		else {
327 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
328 			icmp6_error(m, ICMP6_PARAM_PROB,
329 			    ICMP6_PARAMPROB_NEXTHEADER,
330 			    prvnxtp - mtod(m, char *));
331 		}
332 		IP6STAT_DEC(ip6s_delivered);
333 	}
334 	return (IPPROTO_DONE);
335 }
336 
337 void
338 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
339 {
340 	struct ip6_hdr *ip6;
341 	struct mbuf *m;
342 	int off = 0;
343 	struct ip6ctlparam *ip6cp = NULL;
344 	const struct sockaddr_in6 *sa6_src = NULL;
345 	void *cmdarg;
346 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
347 
348 	if (sa->sa_family != AF_INET6 ||
349 	    sa->sa_len != sizeof(struct sockaddr_in6))
350 		return;
351 
352 	if ((unsigned)cmd >= PRC_NCMDS)
353 		return;
354 	if (PRC_IS_REDIRECT(cmd))
355 		notify = in6_rtchange, d = NULL;
356 	else if (cmd == PRC_HOSTDEAD)
357 		d = NULL;
358 	else if (inet6ctlerrmap[cmd] == 0)
359 		return;
360 
361 	/*
362 	 * If the parameter is from icmp6, decode it.
363 	 */
364 	if (d != NULL) {
365 		ip6cp = (struct ip6ctlparam *)d;
366 		m = ip6cp->ip6c_m;
367 		ip6 = ip6cp->ip6c_ip6;
368 		off = ip6cp->ip6c_off;
369 		cmdarg = ip6cp->ip6c_cmdarg;
370 		sa6_src = ip6cp->ip6c_src;
371 	} else {
372 		m = NULL;
373 		ip6 = NULL;
374 		cmdarg = NULL;
375 		sa6_src = &sa6_any;
376 	}
377 
378 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
379 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
380 }
381 
382 /*
383  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
384  * may have setup with control call.
385  */
386 int
387 rip6_output(struct mbuf *m, struct socket *so, ...)
388 {
389 	struct mbuf *control;
390 	struct m_tag *mtag;
391 	struct sockaddr_in6 *dstsock;
392 	struct in6_addr *dst;
393 	struct ip6_hdr *ip6;
394 	struct inpcb *in6p;
395 	u_int	plen = m->m_pkthdr.len;
396 	int error = 0;
397 	struct ip6_pktopts opt, *optp;
398 	struct ifnet *oifp = NULL;
399 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
400 	int scope_ambiguous = 0;
401 	int use_defzone = 0;
402 	int hlim = 0;
403 	struct in6_addr in6a;
404 	va_list ap;
405 
406 	va_start(ap, so);
407 	dstsock = va_arg(ap, struct sockaddr_in6 *);
408 	control = va_arg(ap, struct mbuf *);
409 	va_end(ap);
410 
411 	in6p = sotoinpcb(so);
412 	INP_WLOCK(in6p);
413 
414 	dst = &dstsock->sin6_addr;
415 	if (control != NULL) {
416 		if ((error = ip6_setpktopts(control, &opt,
417 		    in6p->in6p_outputopts, so->so_cred,
418 		    so->so_proto->pr_protocol)) != 0) {
419 			goto bad;
420 		}
421 		optp = &opt;
422 	} else
423 		optp = in6p->in6p_outputopts;
424 
425 	/*
426 	 * Check and convert scope zone ID into internal form.
427 	 *
428 	 * XXX: we may still need to determine the zone later.
429 	 */
430 	if (!(so->so_state & SS_ISCONNECTED)) {
431 		if (!optp || !optp->ip6po_pktinfo ||
432 		    !optp->ip6po_pktinfo->ipi6_ifindex)
433 			use_defzone = V_ip6_use_defzone;
434 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
435 			scope_ambiguous = 1;
436 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
437 			goto bad;
438 	}
439 
440 	/*
441 	 * For an ICMPv6 packet, we should know its type and code to update
442 	 * statistics.
443 	 */
444 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
445 		struct icmp6_hdr *icmp6;
446 		if (m->m_len < sizeof(struct icmp6_hdr) &&
447 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
448 			error = ENOBUFS;
449 			goto bad;
450 		}
451 		icmp6 = mtod(m, struct icmp6_hdr *);
452 		type = icmp6->icmp6_type;
453 		code = icmp6->icmp6_code;
454 	}
455 
456 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
457 	if (m == NULL) {
458 		error = ENOBUFS;
459 		goto bad;
460 	}
461 	ip6 = mtod(m, struct ip6_hdr *);
462 
463 	/*
464 	 * Source address selection.
465 	 */
466 	error = in6_selectsrc_socket(dstsock, optp, in6p, so->so_cred,
467 	    scope_ambiguous, &in6a, &hlim);
468 
469 	if (error)
470 		goto bad;
471 	error = prison_check_ip6(in6p->inp_cred, &in6a);
472 	if (error != 0)
473 		goto bad;
474 	ip6->ip6_src = in6a;
475 
476 	ip6->ip6_dst = dstsock->sin6_addr;
477 
478 	/*
479 	 * Fill in the rest of the IPv6 header fields.
480 	 */
481 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
482 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
483 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
484 	    (IPV6_VERSION & IPV6_VERSION_MASK);
485 
486 	/*
487 	 * ip6_plen will be filled in ip6_output, so not fill it here.
488 	 */
489 	ip6->ip6_nxt = in6p->inp_ip_p;
490 	ip6->ip6_hlim = hlim;
491 
492 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
493 	    in6p->in6p_cksum != -1) {
494 		struct mbuf *n;
495 		int off;
496 		u_int16_t *p;
497 
498 		/* Compute checksum. */
499 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
500 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
501 		else
502 			off = in6p->in6p_cksum;
503 		if (plen < off + 1) {
504 			error = EINVAL;
505 			goto bad;
506 		}
507 		off += sizeof(struct ip6_hdr);
508 
509 		n = m;
510 		while (n && n->m_len <= off) {
511 			off -= n->m_len;
512 			n = n->m_next;
513 		}
514 		if (!n)
515 			goto bad;
516 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
517 		*p = 0;
518 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
519 	}
520 
521 	/*
522 	 * Send RA/RS messages to user land for protection, before sending
523 	 * them to rtadvd/rtsol.
524 	 */
525 	if ((send_sendso_input_hook != NULL) &&
526 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
527 		switch (type) {
528 		case ND_ROUTER_ADVERT:
529 		case ND_ROUTER_SOLICIT:
530 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
531 				sizeof(unsigned short), M_NOWAIT);
532 			if (mtag == NULL)
533 				goto bad;
534 			m_tag_prepend(m, mtag);
535 		}
536 	}
537 
538 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
539 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
540 		if (oifp)
541 			icmp6_ifoutstat_inc(oifp, type, code);
542 		ICMP6STAT_INC(icp6s_outhist[type]);
543 	} else
544 		RIP6STAT_INC(rip6s_opackets);
545 
546 	goto freectl;
547 
548  bad:
549 	if (m)
550 		m_freem(m);
551 
552  freectl:
553 	if (control != NULL) {
554 		ip6_clearpktopts(&opt, -1);
555 		m_freem(control);
556 	}
557 	INP_WUNLOCK(in6p);
558 	return (error);
559 }
560 
561 /*
562  * Raw IPv6 socket option processing.
563  */
564 int
565 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
566 {
567 	struct inpcb *inp;
568 	int error;
569 
570 	if (sopt->sopt_level == IPPROTO_ICMPV6)
571 		/*
572 		 * XXX: is it better to call icmp6_ctloutput() directly
573 		 * from protosw?
574 		 */
575 		return (icmp6_ctloutput(so, sopt));
576 	else if (sopt->sopt_level != IPPROTO_IPV6) {
577 		if (sopt->sopt_level == SOL_SOCKET &&
578 		    sopt->sopt_name == SO_SETFIB) {
579 			inp = sotoinpcb(so);
580 			INP_WLOCK(inp);
581 			inp->inp_inc.inc_fibnum = so->so_fibnum;
582 			INP_WUNLOCK(inp);
583 			return (0);
584 		}
585 		return (EINVAL);
586 	}
587 
588 	error = 0;
589 
590 	switch (sopt->sopt_dir) {
591 	case SOPT_GET:
592 		switch (sopt->sopt_name) {
593 		case MRT6_INIT:
594 		case MRT6_DONE:
595 		case MRT6_ADD_MIF:
596 		case MRT6_DEL_MIF:
597 		case MRT6_ADD_MFC:
598 		case MRT6_DEL_MFC:
599 		case MRT6_PIM:
600 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
601 			    EOPNOTSUPP;
602 			break;
603 		case IPV6_CHECKSUM:
604 			error = ip6_raw_ctloutput(so, sopt);
605 			break;
606 		default:
607 			error = ip6_ctloutput(so, sopt);
608 			break;
609 		}
610 		break;
611 
612 	case SOPT_SET:
613 		switch (sopt->sopt_name) {
614 		case MRT6_INIT:
615 		case MRT6_DONE:
616 		case MRT6_ADD_MIF:
617 		case MRT6_DEL_MIF:
618 		case MRT6_ADD_MFC:
619 		case MRT6_DEL_MFC:
620 		case MRT6_PIM:
621 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
622 			    EOPNOTSUPP;
623 			break;
624 		case IPV6_CHECKSUM:
625 			error = ip6_raw_ctloutput(so, sopt);
626 			break;
627 		default:
628 			error = ip6_ctloutput(so, sopt);
629 			break;
630 		}
631 		break;
632 	}
633 
634 	return (error);
635 }
636 
637 static int
638 rip6_attach(struct socket *so, int proto, struct thread *td)
639 {
640 	struct inpcb *inp;
641 	struct icmp6_filter *filter;
642 	int error;
643 
644 	inp = sotoinpcb(so);
645 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
646 
647 	error = priv_check(td, PRIV_NETINET_RAW);
648 	if (error)
649 		return (error);
650 	error = soreserve(so, rip_sendspace, rip_recvspace);
651 	if (error)
652 		return (error);
653 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
654 	if (filter == NULL)
655 		return (ENOMEM);
656 	INP_INFO_WLOCK(&V_ripcbinfo);
657 	error = in_pcballoc(so, &V_ripcbinfo);
658 	if (error) {
659 		INP_INFO_WUNLOCK(&V_ripcbinfo);
660 		free(filter, M_PCB);
661 		return (error);
662 	}
663 	inp = (struct inpcb *)so->so_pcb;
664 	INP_INFO_WUNLOCK(&V_ripcbinfo);
665 	inp->inp_vflag |= INP_IPV6;
666 	inp->inp_ip_p = (long)proto;
667 	inp->in6p_hops = -1;	/* use kernel default */
668 	inp->in6p_cksum = -1;
669 	inp->in6p_icmp6filt = filter;
670 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
671 	INP_WUNLOCK(inp);
672 	return (0);
673 }
674 
675 static void
676 rip6_detach(struct socket *so)
677 {
678 	struct inpcb *inp;
679 
680 	inp = sotoinpcb(so);
681 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
682 
683 	if (so == V_ip6_mrouter && ip6_mrouter_done)
684 		ip6_mrouter_done();
685 	/* xxx: RSVP */
686 	INP_INFO_WLOCK(&V_ripcbinfo);
687 	INP_WLOCK(inp);
688 	free(inp->in6p_icmp6filt, M_PCB);
689 	in_pcbdetach(inp);
690 	in_pcbfree(inp);
691 	INP_INFO_WUNLOCK(&V_ripcbinfo);
692 }
693 
694 /* XXXRW: This can't ever be called. */
695 static void
696 rip6_abort(struct socket *so)
697 {
698 	struct inpcb *inp;
699 
700 	inp = sotoinpcb(so);
701 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
702 
703 	soisdisconnected(so);
704 }
705 
706 static void
707 rip6_close(struct socket *so)
708 {
709 	struct inpcb *inp;
710 
711 	inp = sotoinpcb(so);
712 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
713 
714 	soisdisconnected(so);
715 }
716 
717 static int
718 rip6_disconnect(struct socket *so)
719 {
720 	struct inpcb *inp;
721 
722 	inp = sotoinpcb(so);
723 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
724 
725 	if ((so->so_state & SS_ISCONNECTED) == 0)
726 		return (ENOTCONN);
727 	inp->in6p_faddr = in6addr_any;
728 	rip6_abort(so);
729 	return (0);
730 }
731 
732 static int
733 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
734 {
735 	struct inpcb *inp;
736 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
737 	struct ifaddr *ifa = NULL;
738 	int error = 0;
739 
740 	inp = sotoinpcb(so);
741 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
742 
743 	if (nam->sa_len != sizeof(*addr))
744 		return (EINVAL);
745 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
746 		return (error);
747 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
748 		return (EADDRNOTAVAIL);
749 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
750 		return (error);
751 
752 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
753 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
754 		return (EADDRNOTAVAIL);
755 	if (ifa != NULL &&
756 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
757 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
758 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
759 		ifa_free(ifa);
760 		return (EADDRNOTAVAIL);
761 	}
762 	if (ifa != NULL)
763 		ifa_free(ifa);
764 	INP_INFO_WLOCK(&V_ripcbinfo);
765 	INP_WLOCK(inp);
766 	inp->in6p_laddr = addr->sin6_addr;
767 	INP_WUNLOCK(inp);
768 	INP_INFO_WUNLOCK(&V_ripcbinfo);
769 	return (0);
770 }
771 
772 static int
773 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
774 {
775 	struct inpcb *inp;
776 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
777 	struct in6_addr in6a;
778 	int error = 0, scope_ambiguous = 0;
779 
780 	inp = sotoinpcb(so);
781 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
782 
783 	if (nam->sa_len != sizeof(*addr))
784 		return (EINVAL);
785 	if (TAILQ_EMPTY(&V_ifnet))
786 		return (EADDRNOTAVAIL);
787 	if (addr->sin6_family != AF_INET6)
788 		return (EAFNOSUPPORT);
789 
790 	/*
791 	 * Application should provide a proper zone ID or the use of default
792 	 * zone IDs should be enabled.  Unfortunately, some applications do
793 	 * not behave as it should, so we need a workaround.  Even if an
794 	 * appropriate ID is not determined, we'll see if we can determine
795 	 * the outgoing interface.  If we can, determine the zone ID based on
796 	 * the interface below.
797 	 */
798 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
799 		scope_ambiguous = 1;
800 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
801 		return (error);
802 
803 	INP_INFO_WLOCK(&V_ripcbinfo);
804 	INP_WLOCK(inp);
805 	/* Source address selection. XXX: need pcblookup? */
806 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
807 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
808 	if (error) {
809 		INP_WUNLOCK(inp);
810 		INP_INFO_WUNLOCK(&V_ripcbinfo);
811 		return (error);
812 	}
813 
814 	inp->in6p_faddr = addr->sin6_addr;
815 	inp->in6p_laddr = in6a;
816 	soisconnected(so);
817 	INP_WUNLOCK(inp);
818 	INP_INFO_WUNLOCK(&V_ripcbinfo);
819 	return (0);
820 }
821 
822 static int
823 rip6_shutdown(struct socket *so)
824 {
825 	struct inpcb *inp;
826 
827 	inp = sotoinpcb(so);
828 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
829 
830 	INP_WLOCK(inp);
831 	socantsendmore(so);
832 	INP_WUNLOCK(inp);
833 	return (0);
834 }
835 
836 static int
837 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
838     struct mbuf *control, struct thread *td)
839 {
840 	struct inpcb *inp;
841 	struct sockaddr_in6 tmp;
842 	struct sockaddr_in6 *dst;
843 	int ret;
844 
845 	inp = sotoinpcb(so);
846 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
847 
848 	/* Always copy sockaddr to avoid overwrites. */
849 	/* Unlocked read. */
850 	if (so->so_state & SS_ISCONNECTED) {
851 		if (nam) {
852 			m_freem(m);
853 			return (EISCONN);
854 		}
855 		/* XXX */
856 		bzero(&tmp, sizeof(tmp));
857 		tmp.sin6_family = AF_INET6;
858 		tmp.sin6_len = sizeof(struct sockaddr_in6);
859 		INP_RLOCK(inp);
860 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
861 		    sizeof(struct in6_addr));
862 		INP_RUNLOCK(inp);
863 		dst = &tmp;
864 	} else {
865 		if (nam == NULL) {
866 			m_freem(m);
867 			return (ENOTCONN);
868 		}
869 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
870 			m_freem(m);
871 			return (EINVAL);
872 		}
873 		tmp = *(struct sockaddr_in6 *)nam;
874 		dst = &tmp;
875 
876 		if (dst->sin6_family == AF_UNSPEC) {
877 			/*
878 			 * XXX: we allow this case for backward
879 			 * compatibility to buggy applications that
880 			 * rely on old (and wrong) kernel behavior.
881 			 */
882 			log(LOG_INFO, "rip6 SEND: address family is "
883 			    "unspec. Assume AF_INET6\n");
884 			dst->sin6_family = AF_INET6;
885 		} else if (dst->sin6_family != AF_INET6) {
886 			m_freem(m);
887 			return(EAFNOSUPPORT);
888 		}
889 	}
890 	ret = rip6_output(m, so, dst, control);
891 	return (ret);
892 }
893 
894 struct pr_usrreqs rip6_usrreqs = {
895 	.pru_abort =		rip6_abort,
896 	.pru_attach =		rip6_attach,
897 	.pru_bind =		rip6_bind,
898 	.pru_connect =		rip6_connect,
899 	.pru_control =		in6_control,
900 	.pru_detach =		rip6_detach,
901 	.pru_disconnect =	rip6_disconnect,
902 	.pru_peeraddr =		in6_getpeeraddr,
903 	.pru_send =		rip6_send,
904 	.pru_shutdown =		rip6_shutdown,
905 	.pru_sockaddr =		in6_getsockaddr,
906 	.pru_close =		rip6_close,
907 };
908