xref: /freebsd/sys/netinet6/raw_ip6.c (revision 9162f64b58d01ec01481d60b6cdc06ffd8e8c7fc)
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 #include <sys/vimage.h>
83 
84 #include <net/if.h>
85 #include <net/if_types.h>
86 #include <net/route.h>
87 #include <net/vnet.h>
88 
89 #include <netinet/in.h>
90 #include <netinet/in_var.h>
91 #include <netinet/in_systm.h>
92 #include <netinet/in_pcb.h>
93 #include <netinet/vinet.h>
94 
95 #include <netinet/icmp6.h>
96 #include <netinet/ip6.h>
97 #include <netinet6/ip6protosw.h>
98 #include <netinet6/ip6_mroute.h>
99 #include <netinet6/in6_pcb.h>
100 #include <netinet6/ip6_var.h>
101 #include <netinet6/nd6.h>
102 #include <netinet6/raw_ip6.h>
103 #include <netinet6/scope6_var.h>
104 #include <netinet6/vinet6.h>
105 
106 #ifdef IPSEC
107 #include <netipsec/ipsec.h>
108 #include <netipsec/ipsec6.h>
109 #endif /* IPSEC */
110 
111 #include <machine/stdarg.h>
112 
113 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
114 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
115 
116 /*
117  * Raw interface to IP6 protocol.
118  */
119 
120 #ifdef VIMAGE_GLOBALS
121 extern struct	inpcbhead ripcb;
122 extern struct	inpcbinfo ripcbinfo;
123 struct rip6stat rip6stat;
124 #endif
125 
126 extern u_long	rip_sendspace;
127 extern u_long	rip_recvspace;
128 
129 /*
130  * Hooks for multicast forwarding.
131  */
132 struct socket *ip6_mrouter = NULL;
133 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
134 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
135 int (*ip6_mrouter_done)(void);
136 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
137 int (*mrt6_ioctl)(int, caddr_t);
138 
139 /*
140  * Setup generic address and protocol structures for raw_input routine, then
141  * pass them along with mbuf chain.
142  */
143 int
144 rip6_input(struct mbuf **mp, int *offp, int proto)
145 {
146 	INIT_VNET_INET(curvnet);
147 	INIT_VNET_INET6(curvnet);
148 #ifdef IPSEC
149 	INIT_VNET_IPSEC(curvnet);
150 #endif
151 	struct mbuf *m = *mp;
152 	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
153 	register struct inpcb *in6p;
154 	struct inpcb *last = 0;
155 	struct mbuf *opts = NULL;
156 	struct sockaddr_in6 fromsa;
157 
158 	V_rip6stat.rip6s_ipackets++;
159 
160 	if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
161 		/* XXX Send icmp6 host/port unreach? */
162 		m_freem(m);
163 		return (IPPROTO_DONE);
164 	}
165 
166 	init_sin6(&fromsa, m); /* general init */
167 
168 	INP_INFO_RLOCK(&V_ripcbinfo);
169 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
170 		/* XXX inp locking */
171 		if ((in6p->inp_vflag & INP_IPV6) == 0)
172 			continue;
173 		if (in6p->inp_ip_p &&
174 		    in6p->inp_ip_p != proto)
175 			continue;
176 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
177 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
178 			continue;
179 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
180 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
181 			continue;
182 		if (jailed(in6p->inp_cred)) {
183 			if (!prison_check_ip6(in6p->inp_cred, &ip6->ip6_dst))
184 				continue;
185 		}
186 		INP_RLOCK(in6p);
187 		if (in6p->in6p_cksum != -1) {
188 			V_rip6stat.rip6s_isum++;
189 			if (in6_cksum(m, proto, *offp,
190 			    m->m_pkthdr.len - *offp)) {
191 				INP_RUNLOCK(in6p);
192 				V_rip6stat.rip6s_badsum++;
193 				continue;
194 			}
195 		}
196 		if (last != NULL) {
197 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
198 
199 #ifdef IPSEC
200 			/*
201 			 * Check AH/ESP integrity.
202 			 */
203 			if (n && ipsec6_in_reject(n, last)) {
204 				m_freem(n);
205 				V_ipsec6stat.in_polvio++;
206 				/* Do not inject data into pcb. */
207 			} else
208 #endif /* IPSEC */
209 			if (n) {
210 				if (last->inp_flags & INP_CONTROLOPTS ||
211 				    last->inp_socket->so_options & SO_TIMESTAMP)
212 					ip6_savecontrol(last, n, &opts);
213 				/* strip intermediate headers */
214 				m_adj(n, *offp);
215 				if (sbappendaddr(&last->inp_socket->so_rcv,
216 						(struct sockaddr *)&fromsa,
217 						 n, opts) == 0) {
218 					m_freem(n);
219 					if (opts)
220 						m_freem(opts);
221 					V_rip6stat.rip6s_fullsock++;
222 				} else
223 					sorwakeup(last->inp_socket);
224 				opts = NULL;
225 			}
226 			INP_RUNLOCK(last);
227 		}
228 		last = in6p;
229 	}
230 	INP_INFO_RUNLOCK(&V_ripcbinfo);
231 #ifdef IPSEC
232 	/*
233 	 * Check AH/ESP integrity.
234 	 */
235 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
236 		m_freem(m);
237 		V_ipsec6stat.in_polvio++;
238 		V_ip6stat.ip6s_delivered--;
239 		/* Do not inject data into pcb. */
240 		INP_RUNLOCK(last);
241 	} else
242 #endif /* IPSEC */
243 	if (last != NULL) {
244 		if (last->inp_flags & INP_CONTROLOPTS ||
245 		    last->inp_socket->so_options & SO_TIMESTAMP)
246 			ip6_savecontrol(last, m, &opts);
247 		/* Strip intermediate headers. */
248 		m_adj(m, *offp);
249 		if (sbappendaddr(&last->inp_socket->so_rcv,
250 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
251 			m_freem(m);
252 			if (opts)
253 				m_freem(opts);
254 			V_rip6stat.rip6s_fullsock++;
255 		} else
256 			sorwakeup(last->inp_socket);
257 		INP_RUNLOCK(last);
258 	} else {
259 		V_rip6stat.rip6s_nosock++;
260 		if (m->m_flags & M_MCAST)
261 			V_rip6stat.rip6s_nosockmcast++;
262 		if (proto == IPPROTO_NONE)
263 			m_freem(m);
264 		else {
265 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
266 			icmp6_error(m, ICMP6_PARAM_PROB,
267 			    ICMP6_PARAMPROB_NEXTHEADER,
268 			    prvnxtp - mtod(m, char *));
269 		}
270 		V_ip6stat.ip6s_delivered--;
271 	}
272 	return (IPPROTO_DONE);
273 }
274 
275 void
276 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
277 {
278 	INIT_VNET_INET(curvnet);
279 	struct ip6_hdr *ip6;
280 	struct mbuf *m;
281 	int off = 0;
282 	struct ip6ctlparam *ip6cp = NULL;
283 	const struct sockaddr_in6 *sa6_src = NULL;
284 	void *cmdarg;
285 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
286 
287 	if (sa->sa_family != AF_INET6 ||
288 	    sa->sa_len != sizeof(struct sockaddr_in6))
289 		return;
290 
291 	if ((unsigned)cmd >= PRC_NCMDS)
292 		return;
293 	if (PRC_IS_REDIRECT(cmd))
294 		notify = in6_rtchange, d = NULL;
295 	else if (cmd == PRC_HOSTDEAD)
296 		d = NULL;
297 	else if (inet6ctlerrmap[cmd] == 0)
298 		return;
299 
300 	/*
301 	 * If the parameter is from icmp6, decode it.
302 	 */
303 	if (d != NULL) {
304 		ip6cp = (struct ip6ctlparam *)d;
305 		m = ip6cp->ip6c_m;
306 		ip6 = ip6cp->ip6c_ip6;
307 		off = ip6cp->ip6c_off;
308 		cmdarg = ip6cp->ip6c_cmdarg;
309 		sa6_src = ip6cp->ip6c_src;
310 	} else {
311 		m = NULL;
312 		ip6 = NULL;
313 		cmdarg = NULL;
314 		sa6_src = &sa6_any;
315 	}
316 
317 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
318 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
319 }
320 
321 /*
322  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
323  * may have setup with control call.
324  */
325 int
326 #if __STDC__
327 rip6_output(struct mbuf *m, ...)
328 #else
329 rip6_output(m, va_alist)
330 	struct mbuf *m;
331 	va_dcl
332 #endif
333 {
334 	INIT_VNET_INET6(curvnet);
335 	struct mbuf *control;
336 	struct socket *so;
337 	struct sockaddr_in6 *dstsock;
338 	struct in6_addr *dst;
339 	struct ip6_hdr *ip6;
340 	struct inpcb *in6p;
341 	u_int	plen = m->m_pkthdr.len;
342 	int error = 0;
343 	struct ip6_pktopts opt, *optp;
344 	struct ifnet *oifp = NULL;
345 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
346 	int scope_ambiguous = 0;
347 	struct in6_addr *in6a;
348 	va_list ap;
349 
350 	va_start(ap, m);
351 	so = va_arg(ap, struct socket *);
352 	dstsock = va_arg(ap, struct sockaddr_in6 *);
353 	control = va_arg(ap, struct mbuf *);
354 	va_end(ap);
355 
356 	in6p = sotoinpcb(so);
357 	INP_WLOCK(in6p);
358 
359 	dst = &dstsock->sin6_addr;
360 	if (control != NULL) {
361 		if ((error = ip6_setpktopts(control, &opt,
362 		    in6p->in6p_outputopts, so->so_cred,
363 		    so->so_proto->pr_protocol)) != 0) {
364 			goto bad;
365 		}
366 		optp = &opt;
367 	} else
368 		optp = in6p->in6p_outputopts;
369 
370 	/*
371 	 * Check and convert scope zone ID into internal form.
372 	 *
373 	 * XXX: we may still need to determine the zone later.
374 	 */
375 	if (!(so->so_state & SS_ISCONNECTED)) {
376 		if (dstsock->sin6_scope_id == 0 && !V_ip6_use_defzone)
377 			scope_ambiguous = 1;
378 		if ((error = sa6_embedscope(dstsock, V_ip6_use_defzone)) != 0)
379 			goto bad;
380 	}
381 
382 	/*
383 	 * For an ICMPv6 packet, we should know its type and code to update
384 	 * statistics.
385 	 */
386 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
387 		struct icmp6_hdr *icmp6;
388 		if (m->m_len < sizeof(struct icmp6_hdr) &&
389 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
390 			error = ENOBUFS;
391 			goto bad;
392 		}
393 		icmp6 = mtod(m, struct icmp6_hdr *);
394 		type = icmp6->icmp6_type;
395 		code = icmp6->icmp6_code;
396 	}
397 
398 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
399 	if (m == NULL) {
400 		error = ENOBUFS;
401 		goto bad;
402 	}
403 	ip6 = mtod(m, struct ip6_hdr *);
404 
405 	/*
406 	 * Source address selection.
407 	 */
408 	if ((in6a = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
409 	    &oifp, &error)) == NULL) {
410 		if (error == 0)
411 			error = EADDRNOTAVAIL;
412 		goto bad;
413 	}
414 	if (jailed(in6p->inp_cred))
415 		if (prison_getip6(in6p->inp_cred, in6a) != 0) {
416 			error = EPERM;
417 			goto bad;
418 		}
419 	ip6->ip6_src = *in6a;
420 
421 	if (oifp && scope_ambiguous) {
422 		/*
423 		 * Application should provide a proper zone ID or the use of
424 		 * default zone IDs should be enabled.  Unfortunately, some
425 		 * applications do not behave as it should, so we need a
426 		 * workaround.  Even if an appropriate ID is not determined
427 		 * (when it's required), if we can determine the outgoing
428 		 * interface. determine the zone ID based on the interface.
429 		 */
430 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
431 		if (error != 0)
432 			goto bad;
433 	}
434 	ip6->ip6_dst = dstsock->sin6_addr;
435 
436 	/*
437 	 * Fill in the rest of the IPv6 header fields.
438 	 */
439 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
440 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
441 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
442 	    (IPV6_VERSION & IPV6_VERSION_MASK);
443 
444 	/*
445 	 * ip6_plen will be filled in ip6_output, so not fill it here.
446 	 */
447 	ip6->ip6_nxt = in6p->inp_ip_p;
448 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
449 
450 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
451 	    in6p->in6p_cksum != -1) {
452 		struct mbuf *n;
453 		int off;
454 		u_int16_t *p;
455 
456 		/* Compute checksum. */
457 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
458 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
459 		else
460 			off = in6p->in6p_cksum;
461 		if (plen < off + 1) {
462 			error = EINVAL;
463 			goto bad;
464 		}
465 		off += sizeof(struct ip6_hdr);
466 
467 		n = m;
468 		while (n && n->m_len <= off) {
469 			off -= n->m_len;
470 			n = n->m_next;
471 		}
472 		if (!n)
473 			goto bad;
474 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
475 		*p = 0;
476 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
477 	}
478 
479 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
480 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
481 		if (oifp)
482 			icmp6_ifoutstat_inc(oifp, type, code);
483 		V_icmp6stat.icp6s_outhist[type]++;
484 	} else
485 		V_rip6stat.rip6s_opackets++;
486 
487 	goto freectl;
488 
489  bad:
490 	if (m)
491 		m_freem(m);
492 
493  freectl:
494 	if (control != NULL) {
495 		ip6_clearpktopts(&opt, -1);
496 		m_freem(control);
497 	}
498 	INP_WUNLOCK(in6p);
499 	return (error);
500 }
501 
502 /*
503  * Raw IPv6 socket option processing.
504  */
505 int
506 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
507 {
508 	int error;
509 
510 	if (sopt->sopt_level == IPPROTO_ICMPV6)
511 		/*
512 		 * XXX: is it better to call icmp6_ctloutput() directly
513 		 * from protosw?
514 		 */
515 		return (icmp6_ctloutput(so, sopt));
516 	else if (sopt->sopt_level != IPPROTO_IPV6)
517 		return (EINVAL);
518 
519 	error = 0;
520 
521 	switch (sopt->sopt_dir) {
522 	case SOPT_GET:
523 		switch (sopt->sopt_name) {
524 		case MRT6_INIT:
525 		case MRT6_DONE:
526 		case MRT6_ADD_MIF:
527 		case MRT6_DEL_MIF:
528 		case MRT6_ADD_MFC:
529 		case MRT6_DEL_MFC:
530 		case MRT6_PIM:
531 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
532 			    EOPNOTSUPP;
533 			break;
534 		case IPV6_CHECKSUM:
535 			error = ip6_raw_ctloutput(so, sopt);
536 			break;
537 		default:
538 			error = ip6_ctloutput(so, sopt);
539 			break;
540 		}
541 		break;
542 
543 	case SOPT_SET:
544 		switch (sopt->sopt_name) {
545 		case MRT6_INIT:
546 		case MRT6_DONE:
547 		case MRT6_ADD_MIF:
548 		case MRT6_DEL_MIF:
549 		case MRT6_ADD_MFC:
550 		case MRT6_DEL_MFC:
551 		case MRT6_PIM:
552 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
553 			    EOPNOTSUPP;
554 			break;
555 		case IPV6_CHECKSUM:
556 			error = ip6_raw_ctloutput(so, sopt);
557 			break;
558 		default:
559 			error = ip6_ctloutput(so, sopt);
560 			break;
561 		}
562 		break;
563 	}
564 
565 	return (error);
566 }
567 
568 static int
569 rip6_attach(struct socket *so, int proto, struct thread *td)
570 {
571 	INIT_VNET_INET(so->so_vnet);
572 	struct inpcb *inp;
573 	struct icmp6_filter *filter;
574 	int error;
575 
576 	inp = sotoinpcb(so);
577 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
578 
579 	error = priv_check(td, PRIV_NETINET_RAW);
580 	if (error)
581 		return (error);
582 	error = soreserve(so, rip_sendspace, rip_recvspace);
583 	if (error)
584 		return (error);
585 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
586 	if (filter == NULL)
587 		return (ENOMEM);
588 	INP_INFO_WLOCK(&V_ripcbinfo);
589 	error = in_pcballoc(so, &V_ripcbinfo);
590 	if (error) {
591 		INP_INFO_WUNLOCK(&V_ripcbinfo);
592 		free(filter, M_PCB);
593 		return (error);
594 	}
595 	inp = (struct inpcb *)so->so_pcb;
596 	INP_INFO_WUNLOCK(&V_ripcbinfo);
597 	inp->inp_vflag |= INP_IPV6;
598 	inp->inp_ip_p = (long)proto;
599 	inp->in6p_hops = -1;	/* use kernel default */
600 	inp->in6p_cksum = -1;
601 	inp->in6p_icmp6filt = filter;
602 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
603 	INP_WUNLOCK(inp);
604 	return (0);
605 }
606 
607 static void
608 rip6_detach(struct socket *so)
609 {
610 	INIT_VNET_INET(so->so_vnet);
611 	struct inpcb *inp;
612 
613 	inp = sotoinpcb(so);
614 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
615 
616 	if (so == ip6_mrouter && ip6_mrouter_done)
617 		ip6_mrouter_done();
618 	/* xxx: RSVP */
619 	INP_INFO_WLOCK(&V_ripcbinfo);
620 	INP_WLOCK(inp);
621 	free(inp->in6p_icmp6filt, M_PCB);
622 	in_pcbdetach(inp);
623 	in_pcbfree(inp);
624 	INP_INFO_WUNLOCK(&V_ripcbinfo);
625 }
626 
627 /* XXXRW: This can't ever be called. */
628 static void
629 rip6_abort(struct socket *so)
630 {
631 	struct inpcb *inp;
632 
633 	inp = sotoinpcb(so);
634 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
635 
636 	soisdisconnected(so);
637 }
638 
639 static void
640 rip6_close(struct socket *so)
641 {
642 	struct inpcb *inp;
643 
644 	inp = sotoinpcb(so);
645 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
646 
647 	soisdisconnected(so);
648 }
649 
650 static int
651 rip6_disconnect(struct socket *so)
652 {
653 	struct inpcb *inp;
654 
655 	inp = sotoinpcb(so);
656 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
657 
658 	if ((so->so_state & SS_ISCONNECTED) == 0)
659 		return (ENOTCONN);
660 	inp->in6p_faddr = in6addr_any;
661 	rip6_abort(so);
662 	return (0);
663 }
664 
665 static int
666 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
667 {
668 	INIT_VNET_NET(so->so_vnet);
669 	INIT_VNET_INET(so->so_vnet);
670 	INIT_VNET_INET6(so->so_vnet);
671 	struct inpcb *inp;
672 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
673 	struct ifaddr *ia = NULL;
674 	int error = 0;
675 
676 	inp = sotoinpcb(so);
677 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
678 
679 	if (nam->sa_len != sizeof(*addr))
680 		return (EINVAL);
681 	if (!prison_check_ip6(td->td_ucred, &addr->sin6_addr))
682 		return (EADDRNOTAVAIL);
683 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
684 		return (EADDRNOTAVAIL);
685 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
686 		return (error);
687 
688 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
689 	    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0)
690 		return (EADDRNOTAVAIL);
691 	if (ia &&
692 	    ((struct in6_ifaddr *)ia)->ia6_flags &
693 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
694 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
695 		return (EADDRNOTAVAIL);
696 	}
697 	INP_INFO_WLOCK(&V_ripcbinfo);
698 	INP_WLOCK(inp);
699 	inp->in6p_laddr = addr->sin6_addr;
700 	INP_WUNLOCK(inp);
701 	INP_INFO_WUNLOCK(&V_ripcbinfo);
702 	return (0);
703 }
704 
705 static int
706 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
707 {
708 	INIT_VNET_NET(so->so_vnet);
709 	INIT_VNET_INET(so->so_vnet);
710 	INIT_VNET_INET6(so->so_vnet);
711 	struct inpcb *inp;
712 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
713 	struct in6_addr *in6a = NULL;
714 	struct ifnet *ifp = NULL;
715 	int error = 0, scope_ambiguous = 0;
716 
717 	inp = sotoinpcb(so);
718 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
719 
720 	if (nam->sa_len != sizeof(*addr))
721 		return (EINVAL);
722 	if (TAILQ_EMPTY(&V_ifnet))
723 		return (EADDRNOTAVAIL);
724 	if (addr->sin6_family != AF_INET6)
725 		return (EAFNOSUPPORT);
726 
727 	/*
728 	 * Application should provide a proper zone ID or the use of default
729 	 * zone IDs should be enabled.  Unfortunately, some applications do
730 	 * not behave as it should, so we need a workaround.  Even if an
731 	 * appropriate ID is not determined, we'll see if we can determine
732 	 * the outgoing interface.  If we can, determine the zone ID based on
733 	 * the interface below.
734 	 */
735 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
736 		scope_ambiguous = 1;
737 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
738 		return (error);
739 
740 	INP_INFO_WLOCK(&V_ripcbinfo);
741 	INP_WLOCK(inp);
742 	/* Source address selection. XXX: need pcblookup? */
743 	in6a = in6_selectsrc(addr, inp->in6p_outputopts,
744 			     inp, NULL, so->so_cred,
745 			     &ifp, &error);
746 	if (in6a == NULL) {
747 		INP_WUNLOCK(inp);
748 		INP_INFO_WUNLOCK(&V_ripcbinfo);
749 		return (error ? error : EADDRNOTAVAIL);
750 	}
751 
752 	/* XXX: see above */
753 	if (ifp && scope_ambiguous &&
754 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
755 		INP_WUNLOCK(inp);
756 		INP_INFO_WUNLOCK(&V_ripcbinfo);
757 		return (error);
758 	}
759 	inp->in6p_faddr = addr->sin6_addr;
760 	inp->in6p_laddr = *in6a;
761 	soisconnected(so);
762 	INP_WUNLOCK(inp);
763 	INP_INFO_WUNLOCK(&V_ripcbinfo);
764 	return (0);
765 }
766 
767 static int
768 rip6_shutdown(struct socket *so)
769 {
770 	struct inpcb *inp;
771 
772 	inp = sotoinpcb(so);
773 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
774 
775 	INP_WLOCK(inp);
776 	socantsendmore(so);
777 	INP_WUNLOCK(inp);
778 	return (0);
779 }
780 
781 static int
782 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
783     struct mbuf *control, struct thread *td)
784 {
785 	struct inpcb *inp;
786 	struct sockaddr_in6 tmp;
787 	struct sockaddr_in6 *dst;
788 	int ret;
789 
790 	inp = sotoinpcb(so);
791 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
792 
793 	/* Always copy sockaddr to avoid overwrites. */
794 	/* Unlocked read. */
795 	if (so->so_state & SS_ISCONNECTED) {
796 		if (nam) {
797 			m_freem(m);
798 			return (EISCONN);
799 		}
800 		/* XXX */
801 		bzero(&tmp, sizeof(tmp));
802 		tmp.sin6_family = AF_INET6;
803 		tmp.sin6_len = sizeof(struct sockaddr_in6);
804 		INP_RLOCK(inp);
805 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
806 		    sizeof(struct in6_addr));
807 		INP_RUNLOCK(inp);
808 		dst = &tmp;
809 	} else {
810 		if (nam == NULL) {
811 			m_freem(m);
812 			return (ENOTCONN);
813 		}
814 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
815 			m_freem(m);
816 			return (EINVAL);
817 		}
818 		tmp = *(struct sockaddr_in6 *)nam;
819 		dst = &tmp;
820 
821 		if (dst->sin6_family == AF_UNSPEC) {
822 			/*
823 			 * XXX: we allow this case for backward
824 			 * compatibility to buggy applications that
825 			 * rely on old (and wrong) kernel behavior.
826 			 */
827 			log(LOG_INFO, "rip6 SEND: address family is "
828 			    "unspec. Assume AF_INET6\n");
829 			dst->sin6_family = AF_INET6;
830 		} else if (dst->sin6_family != AF_INET6) {
831 			m_freem(m);
832 			return(EAFNOSUPPORT);
833 		}
834 	}
835 	ret = rip6_output(m, so, dst, control);
836 	return (ret);
837 }
838 
839 struct pr_usrreqs rip6_usrreqs = {
840 	.pru_abort =		rip6_abort,
841 	.pru_attach =		rip6_attach,
842 	.pru_bind =		rip6_bind,
843 	.pru_connect =		rip6_connect,
844 	.pru_control =		in6_control,
845 	.pru_detach =		rip6_detach,
846 	.pru_disconnect =	rip6_disconnect,
847 	.pru_peeraddr =		in6_getpeeraddr,
848 	.pru_send =		rip6_send,
849 	.pru_shutdown =		rip6_shutdown,
850 	.pru_sockaddr =		in6_getsockaddr,
851 	.pru_close =		rip6_close,
852 };
853