xref: /freebsd/sys/netinet6/raw_ip6.c (revision cacdd70cc751fb68dec4b86c5e5b8c969b6e26ef)
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/lock.h>
71 #include <sys/malloc.h>
72 #include <sys/mbuf.h>
73 #include <sys/priv.h>
74 #include <sys/proc.h>
75 #include <sys/protosw.h>
76 #include <sys/signalvar.h>
77 #include <sys/socket.h>
78 #include <sys/socketvar.h>
79 #include <sys/sx.h>
80 #include <sys/syslog.h>
81 
82 #include <net/if.h>
83 #include <net/if_types.h>
84 #include <net/route.h>
85 
86 #include <netinet/in.h>
87 #include <netinet/in_var.h>
88 #include <netinet/in_systm.h>
89 #include <netinet/icmp6.h>
90 #include <netinet/in_pcb.h>
91 #include <netinet/ip6.h>
92 #include <netinet6/ip6protosw.h>
93 #include <netinet6/ip6_mroute.h>
94 #include <netinet6/in6_pcb.h>
95 #include <netinet6/ip6_var.h>
96 #include <netinet6/nd6.h>
97 #include <netinet6/raw_ip6.h>
98 #include <netinet6/scope6_var.h>
99 
100 #ifdef IPSEC
101 #include <netipsec/ipsec.h>
102 #include <netipsec/ipsec6.h>
103 #endif /* IPSEC */
104 
105 #include <machine/stdarg.h>
106 
107 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
108 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
109 
110 /*
111  * Raw interface to IP6 protocol.
112  */
113 
114 extern struct	inpcbhead ripcb;
115 extern struct	inpcbinfo ripcbinfo;
116 extern u_long	rip_sendspace;
117 extern u_long	rip_recvspace;
118 
119 struct rip6stat rip6stat;
120 
121 /*
122  * Hooks for multicast forwarding.
123  */
124 struct socket *ip6_mrouter = NULL;
125 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
126 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
127 int (*ip6_mrouter_done)(void);
128 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
129 int (*mrt6_ioctl)(int, caddr_t);
130 
131 /*
132  * Setup generic address and protocol structures for raw_input routine, then
133  * pass them along with mbuf chain.
134  */
135 int
136 rip6_input(struct mbuf **mp, int *offp, int proto)
137 {
138 	struct mbuf *m = *mp;
139 	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
140 	register struct inpcb *in6p;
141 	struct inpcb *last = 0;
142 	struct mbuf *opts = NULL;
143 	struct sockaddr_in6 fromsa;
144 
145 	rip6stat.rip6s_ipackets++;
146 
147 	if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
148 		/* XXX Send icmp6 host/port unreach? */
149 		m_freem(m);
150 		return (IPPROTO_DONE);
151 	}
152 
153 	init_sin6(&fromsa, m); /* general init */
154 
155 	INP_INFO_RLOCK(&ripcbinfo);
156 	LIST_FOREACH(in6p, &ripcb, inp_list) {
157 		if ((in6p->in6p_vflag & INP_IPV6) == 0)
158 			continue;
159 		if (in6p->in6p_ip6_nxt &&
160 		    in6p->in6p_ip6_nxt != proto)
161 			continue;
162 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
163 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
164 			continue;
165 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
166 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
167 			continue;
168 		INP_RLOCK(in6p);
169 		if (in6p->in6p_cksum != -1) {
170 			rip6stat.rip6s_isum++;
171 			if (in6_cksum(m, proto, *offp,
172 			    m->m_pkthdr.len - *offp)) {
173 				INP_RUNLOCK(in6p);
174 				rip6stat.rip6s_badsum++;
175 				continue;
176 			}
177 		}
178 		if (last) {
179 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
180 
181 #ifdef IPSEC
182 			/*
183 			 * Check AH/ESP integrity.
184 			 */
185 			if (n && ipsec6_in_reject(n, last)) {
186 				m_freem(n);
187 				ipsec6stat.in_polvio++;
188 				/* Do not inject data into pcb. */
189 			} else
190 #endif /* IPSEC */
191 			if (n) {
192 				if (last->in6p_flags & IN6P_CONTROLOPTS ||
193 				    last->in6p_socket->so_options & SO_TIMESTAMP)
194 					ip6_savecontrol(last, n, &opts);
195 				/* strip intermediate headers */
196 				m_adj(n, *offp);
197 				if (sbappendaddr(&last->in6p_socket->so_rcv,
198 						(struct sockaddr *)&fromsa,
199 						 n, opts) == 0) {
200 					m_freem(n);
201 					if (opts)
202 						m_freem(opts);
203 					rip6stat.rip6s_fullsock++;
204 				} else
205 					sorwakeup(last->in6p_socket);
206 				opts = NULL;
207 			}
208 			INP_RUNLOCK(last);
209 		}
210 		last = in6p;
211 	}
212 	INP_INFO_RUNLOCK(&ripcbinfo);
213 #ifdef IPSEC
214 	/*
215 	 * Check AH/ESP integrity.
216 	 */
217 	if (last && ipsec6_in_reject(m, last)) {
218 		m_freem(m);
219 		ipsec6stat.in_polvio++;
220 		ip6stat.ip6s_delivered--;
221 		/* Do not inject data into pcb. */
222 		INP_RUNLOCK(last);
223 	} else
224 #endif /* IPSEC */
225 	if (last) {
226 		if (last->in6p_flags & IN6P_CONTROLOPTS ||
227 		    last->in6p_socket->so_options & SO_TIMESTAMP)
228 			ip6_savecontrol(last, m, &opts);
229 		/* Strip intermediate headers. */
230 		m_adj(m, *offp);
231 		if (sbappendaddr(&last->in6p_socket->so_rcv,
232 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
233 			m_freem(m);
234 			if (opts)
235 				m_freem(opts);
236 			rip6stat.rip6s_fullsock++;
237 		} else
238 			sorwakeup(last->in6p_socket);
239 		INP_RUNLOCK(last);
240 	} else {
241 		rip6stat.rip6s_nosock++;
242 		if (m->m_flags & M_MCAST)
243 			rip6stat.rip6s_nosockmcast++;
244 		if (proto == IPPROTO_NONE)
245 			m_freem(m);
246 		else {
247 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
248 			icmp6_error(m, ICMP6_PARAM_PROB,
249 			    ICMP6_PARAMPROB_NEXTHEADER,
250 			    prvnxtp - mtod(m, char *));
251 		}
252 		ip6stat.ip6s_delivered--;
253 	}
254 	return (IPPROTO_DONE);
255 }
256 
257 void
258 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
259 {
260 	struct ip6_hdr *ip6;
261 	struct mbuf *m;
262 	int off = 0;
263 	struct ip6ctlparam *ip6cp = NULL;
264 	const struct sockaddr_in6 *sa6_src = NULL;
265 	void *cmdarg;
266 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
267 
268 	if (sa->sa_family != AF_INET6 ||
269 	    sa->sa_len != sizeof(struct sockaddr_in6))
270 		return;
271 
272 	if ((unsigned)cmd >= PRC_NCMDS)
273 		return;
274 	if (PRC_IS_REDIRECT(cmd))
275 		notify = in6_rtchange, d = NULL;
276 	else if (cmd == PRC_HOSTDEAD)
277 		d = NULL;
278 	else if (inet6ctlerrmap[cmd] == 0)
279 		return;
280 
281 	/*
282 	 * If the parameter is from icmp6, decode it.
283 	 */
284 	if (d != NULL) {
285 		ip6cp = (struct ip6ctlparam *)d;
286 		m = ip6cp->ip6c_m;
287 		ip6 = ip6cp->ip6c_ip6;
288 		off = ip6cp->ip6c_off;
289 		cmdarg = ip6cp->ip6c_cmdarg;
290 		sa6_src = ip6cp->ip6c_src;
291 	} else {
292 		m = NULL;
293 		ip6 = NULL;
294 		cmdarg = NULL;
295 		sa6_src = &sa6_any;
296 	}
297 
298 	(void) in6_pcbnotify(&ripcbinfo, sa, 0,
299 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
300 }
301 
302 /*
303  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
304  * may have setup with control call.
305  */
306 int
307 #if __STDC__
308 rip6_output(struct mbuf *m, ...)
309 #else
310 rip6_output(m, va_alist)
311 	struct mbuf *m;
312 	va_dcl
313 #endif
314 {
315 	struct mbuf *control;
316 	struct socket *so;
317 	struct sockaddr_in6 *dstsock;
318 	struct in6_addr *dst;
319 	struct ip6_hdr *ip6;
320 	struct inpcb *in6p;
321 	u_int	plen = m->m_pkthdr.len;
322 	int error = 0;
323 	struct ip6_pktopts opt, *optp;
324 	struct ifnet *oifp = NULL;
325 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
326 	int scope_ambiguous = 0;
327 	struct in6_addr *in6a;
328 	va_list ap;
329 
330 	va_start(ap, m);
331 	so = va_arg(ap, struct socket *);
332 	dstsock = va_arg(ap, struct sockaddr_in6 *);
333 	control = va_arg(ap, struct mbuf *);
334 	va_end(ap);
335 
336 	in6p = sotoin6pcb(so);
337 	INP_WLOCK(in6p);
338 
339 	dst = &dstsock->sin6_addr;
340 	if (control) {
341 		if ((error = ip6_setpktopts(control, &opt,
342 		    in6p->in6p_outputopts, so->so_cred,
343 		    so->so_proto->pr_protocol)) != 0) {
344 			goto bad;
345 		}
346 		optp = &opt;
347 	} else
348 		optp = in6p->in6p_outputopts;
349 
350 	/*
351 	 * Check and convert scope zone ID into internal form.
352 	 *
353 	 * XXX: we may still need to determine the zone later.
354 	 */
355 	if (!(so->so_state & SS_ISCONNECTED)) {
356 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
357 			scope_ambiguous = 1;
358 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
359 			goto bad;
360 	}
361 
362 	/*
363 	 * For an ICMPv6 packet, we should know its type and code to update
364 	 * statistics.
365 	 */
366 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
367 		struct icmp6_hdr *icmp6;
368 		if (m->m_len < sizeof(struct icmp6_hdr) &&
369 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
370 			error = ENOBUFS;
371 			goto bad;
372 		}
373 		icmp6 = mtod(m, struct icmp6_hdr *);
374 		type = icmp6->icmp6_type;
375 		code = icmp6->icmp6_code;
376 	}
377 
378 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
379 	if (m == NULL) {
380 		error = ENOBUFS;
381 		goto bad;
382 	}
383 	ip6 = mtod(m, struct ip6_hdr *);
384 
385 	/*
386 	 * Source address selection.
387 	 */
388 	if ((in6a = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
389 	    &oifp, &error)) == NULL) {
390 		if (error == 0)
391 			error = EADDRNOTAVAIL;
392 		goto bad;
393 	}
394 	ip6->ip6_src = *in6a;
395 
396 	if (oifp && scope_ambiguous) {
397 		/*
398 		 * Application should provide a proper zone ID or the use of
399 		 * default zone IDs should be enabled.  Unfortunately, some
400 		 * applications do not behave as it should, so we need a
401 		 * workaround.  Even if an appropriate ID is not determined
402 		 * (when it's required), if we can determine the outgoing
403 		 * interface. determine the zone ID based on the interface.
404 		 */
405 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
406 		if (error != 0)
407 			goto bad;
408 	}
409 	ip6->ip6_dst = dstsock->sin6_addr;
410 
411 	/*
412 	 * Fill in the rest of the IPv6 header fields.
413 	 */
414 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
415 	    (in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK);
416 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
417 	    (IPV6_VERSION & IPV6_VERSION_MASK);
418 
419 	/*
420 	 * ip6_plen will be filled in ip6_output, so not fill it here.
421 	 */
422 	ip6->ip6_nxt = in6p->in6p_ip6_nxt;
423 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
424 
425 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
426 	    in6p->in6p_cksum != -1) {
427 		struct mbuf *n;
428 		int off;
429 		u_int16_t *p;
430 
431 		/* Compute checksum. */
432 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
433 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
434 		else
435 			off = in6p->in6p_cksum;
436 		if (plen < off + 1) {
437 			error = EINVAL;
438 			goto bad;
439 		}
440 		off += sizeof(struct ip6_hdr);
441 
442 		n = m;
443 		while (n && n->m_len <= off) {
444 			off -= n->m_len;
445 			n = n->m_next;
446 		}
447 		if (!n)
448 			goto bad;
449 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
450 		*p = 0;
451 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
452 	}
453 
454 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
455 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
456 		if (oifp)
457 			icmp6_ifoutstat_inc(oifp, type, code);
458 		icmp6stat.icp6s_outhist[type]++;
459 	} else
460 		rip6stat.rip6s_opackets++;
461 
462 	goto freectl;
463 
464  bad:
465 	if (m)
466 		m_freem(m);
467 
468  freectl:
469 	if (control) {
470 		ip6_clearpktopts(&opt, -1);
471 		m_freem(control);
472 	}
473 	INP_WUNLOCK(in6p);
474 	return (error);
475 }
476 
477 /*
478  * Raw IPv6 socket option processing.
479  */
480 int
481 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
482 {
483 	int error;
484 
485 	if (sopt->sopt_level == IPPROTO_ICMPV6)
486 		/*
487 		 * XXX: is it better to call icmp6_ctloutput() directly
488 		 * from protosw?
489 		 */
490 		return (icmp6_ctloutput(so, sopt));
491 	else if (sopt->sopt_level != IPPROTO_IPV6)
492 		return (EINVAL);
493 
494 	error = 0;
495 
496 	switch (sopt->sopt_dir) {
497 	case SOPT_GET:
498 		switch (sopt->sopt_name) {
499 		case MRT6_INIT:
500 		case MRT6_DONE:
501 		case MRT6_ADD_MIF:
502 		case MRT6_DEL_MIF:
503 		case MRT6_ADD_MFC:
504 		case MRT6_DEL_MFC:
505 		case MRT6_PIM:
506 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
507 			    EOPNOTSUPP;
508 			break;
509 		case IPV6_CHECKSUM:
510 			error = ip6_raw_ctloutput(so, sopt);
511 			break;
512 		default:
513 			error = ip6_ctloutput(so, sopt);
514 			break;
515 		}
516 		break;
517 
518 	case SOPT_SET:
519 		switch (sopt->sopt_name) {
520 		case MRT6_INIT:
521 		case MRT6_DONE:
522 		case MRT6_ADD_MIF:
523 		case MRT6_DEL_MIF:
524 		case MRT6_ADD_MFC:
525 		case MRT6_DEL_MFC:
526 		case MRT6_PIM:
527 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
528 			    EOPNOTSUPP;
529 			break;
530 		case IPV6_CHECKSUM:
531 			error = ip6_raw_ctloutput(so, sopt);
532 			break;
533 		default:
534 			error = ip6_ctloutput(so, sopt);
535 			break;
536 		}
537 		break;
538 	}
539 
540 	return (error);
541 }
542 
543 static int
544 rip6_attach(struct socket *so, int proto, struct thread *td)
545 {
546 	struct inpcb *inp;
547 	struct icmp6_filter *filter;
548 	int error;
549 
550 	inp = sotoinpcb(so);
551 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
552 
553 	error = priv_check(td, PRIV_NETINET_RAW);
554 	if (error)
555 		return (error);
556 	error = soreserve(so, rip_sendspace, rip_recvspace);
557 	if (error)
558 		return (error);
559 	MALLOC(filter, struct icmp6_filter *,
560 	       sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
561 	if (filter == NULL)
562 		return (ENOMEM);
563 	INP_INFO_WLOCK(&ripcbinfo);
564 	error = in_pcballoc(so, &ripcbinfo);
565 	if (error) {
566 		INP_INFO_WUNLOCK(&ripcbinfo);
567 		FREE(filter, M_PCB);
568 		return (error);
569 	}
570 	inp = (struct inpcb *)so->so_pcb;
571 	INP_INFO_WUNLOCK(&ripcbinfo);
572 	inp->inp_vflag |= INP_IPV6;
573 	inp->in6p_ip6_nxt = (long)proto;
574 	inp->in6p_hops = -1;	/* use kernel default */
575 	inp->in6p_cksum = -1;
576 	inp->in6p_icmp6filt = filter;
577 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
578 	INP_WUNLOCK(inp);
579 	return (0);
580 }
581 
582 static void
583 rip6_detach(struct socket *so)
584 {
585 	struct inpcb *inp;
586 
587 	inp = sotoinpcb(so);
588 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
589 
590 	if (so == ip6_mrouter && ip6_mrouter_done)
591 		ip6_mrouter_done();
592 	/* xxx: RSVP */
593 	INP_INFO_WLOCK(&ripcbinfo);
594 	INP_WLOCK(inp);
595 	FREE(inp->in6p_icmp6filt, M_PCB);
596 	in6_pcbdetach(inp);
597 	in6_pcbfree(inp);
598 	INP_INFO_WUNLOCK(&ripcbinfo);
599 }
600 
601 /* XXXRW: This can't ever be called. */
602 static void
603 rip6_abort(struct socket *so)
604 {
605 	struct inpcb *inp;
606 
607 	inp = sotoinpcb(so);
608 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
609 
610 	soisdisconnected(so);
611 }
612 
613 static void
614 rip6_close(struct socket *so)
615 {
616 	struct inpcb *inp;
617 
618 	inp = sotoinpcb(so);
619 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
620 
621 	soisdisconnected(so);
622 }
623 
624 static int
625 rip6_disconnect(struct socket *so)
626 {
627 	struct inpcb *inp;
628 
629 	inp = sotoinpcb(so);
630 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
631 
632 	if ((so->so_state & SS_ISCONNECTED) == 0)
633 		return (ENOTCONN);
634 	inp->in6p_faddr = in6addr_any;
635 	rip6_abort(so);
636 	return (0);
637 }
638 
639 static int
640 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
641 {
642 	struct inpcb *inp;
643 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
644 	struct ifaddr *ia = NULL;
645 	int error = 0;
646 
647 	inp = sotoinpcb(so);
648 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
649 
650 	if (nam->sa_len != sizeof(*addr))
651 		return (EINVAL);
652 	if (TAILQ_EMPTY(&ifnet) || addr->sin6_family != AF_INET6)
653 		return (EADDRNOTAVAIL);
654 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
655 		return (error);
656 
657 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
658 	    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0)
659 		return (EADDRNOTAVAIL);
660 	if (ia &&
661 	    ((struct in6_ifaddr *)ia)->ia6_flags &
662 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
663 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
664 		return (EADDRNOTAVAIL);
665 	}
666 	INP_INFO_WLOCK(&ripcbinfo);
667 	INP_WLOCK(inp);
668 	inp->in6p_laddr = addr->sin6_addr;
669 	INP_WUNLOCK(inp);
670 	INP_INFO_WUNLOCK(&ripcbinfo);
671 	return (0);
672 }
673 
674 static int
675 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
676 {
677 	struct inpcb *inp;
678 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
679 	struct in6_addr *in6a = NULL;
680 	struct ifnet *ifp = NULL;
681 	int error = 0, scope_ambiguous = 0;
682 
683 	inp = sotoinpcb(so);
684 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
685 
686 	if (nam->sa_len != sizeof(*addr))
687 		return (EINVAL);
688 	if (TAILQ_EMPTY(&ifnet))
689 		return (EADDRNOTAVAIL);
690 	if (addr->sin6_family != AF_INET6)
691 		return (EAFNOSUPPORT);
692 
693 	/*
694 	 * Application should provide a proper zone ID or the use of default
695 	 * zone IDs should be enabled.  Unfortunately, some applications do
696 	 * not behave as it should, so we need a workaround.  Even if an
697 	 * appropriate ID is not determined, we'll see if we can determine
698 	 * the outgoing interface.  If we can, determine the zone ID based on
699 	 * the interface below.
700 	 */
701 	if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
702 		scope_ambiguous = 1;
703 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
704 		return (error);
705 
706 	INP_INFO_WLOCK(&ripcbinfo);
707 	INP_WLOCK(inp);
708 	/* Source address selection. XXX: need pcblookup? */
709 	in6a = in6_selectsrc(addr, inp->in6p_outputopts,
710 			     inp, NULL, so->so_cred,
711 			     &ifp, &error);
712 	if (in6a == NULL) {
713 		INP_WUNLOCK(inp);
714 		INP_INFO_WUNLOCK(&ripcbinfo);
715 		return (error ? error : EADDRNOTAVAIL);
716 	}
717 
718 	/* XXX: see above */
719 	if (ifp && scope_ambiguous &&
720 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
721 		INP_WUNLOCK(inp);
722 		INP_INFO_WUNLOCK(&ripcbinfo);
723 		return (error);
724 	}
725 	inp->in6p_faddr = addr->sin6_addr;
726 	inp->in6p_laddr = *in6a;
727 	soisconnected(so);
728 	INP_WUNLOCK(inp);
729 	INP_INFO_WUNLOCK(&ripcbinfo);
730 	return (0);
731 }
732 
733 static int
734 rip6_shutdown(struct socket *so)
735 {
736 	struct inpcb *inp;
737 
738 	inp = sotoinpcb(so);
739 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
740 
741 	INP_WLOCK(inp);
742 	socantsendmore(so);
743 	INP_WUNLOCK(inp);
744 	return (0);
745 }
746 
747 static int
748 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
749     struct mbuf *control, struct thread *td)
750 {
751 	struct inpcb *inp;
752 	struct sockaddr_in6 tmp;
753 	struct sockaddr_in6 *dst;
754 	int ret;
755 
756 	inp = sotoinpcb(so);
757 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
758 
759 	/* Always copy sockaddr to avoid overwrites. */
760 	/* Unlocked read. */
761 	if (so->so_state & SS_ISCONNECTED) {
762 		if (nam) {
763 			m_freem(m);
764 			return (EISCONN);
765 		}
766 		/* XXX */
767 		bzero(&tmp, sizeof(tmp));
768 		tmp.sin6_family = AF_INET6;
769 		tmp.sin6_len = sizeof(struct sockaddr_in6);
770 		INP_RLOCK(inp);
771 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
772 		    sizeof(struct in6_addr));
773 		INP_RUNLOCK(inp);
774 		dst = &tmp;
775 	} else {
776 		if (nam == NULL) {
777 			m_freem(m);
778 			return (ENOTCONN);
779 		}
780 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
781 			m_freem(m);
782 			return (EINVAL);
783 		}
784 		tmp = *(struct sockaddr_in6 *)nam;
785 		dst = &tmp;
786 
787 		if (dst->sin6_family == AF_UNSPEC) {
788 			/*
789 			 * XXX: we allow this case for backward
790 			 * compatibility to buggy applications that
791 			 * rely on old (and wrong) kernel behavior.
792 			 */
793 			log(LOG_INFO, "rip6 SEND: address family is "
794 			    "unspec. Assume AF_INET6\n");
795 			dst->sin6_family = AF_INET6;
796 		} else if (dst->sin6_family != AF_INET6) {
797 			m_freem(m);
798 			return(EAFNOSUPPORT);
799 		}
800 	}
801 	ret = rip6_output(m, so, dst, control);
802 	return (ret);
803 }
804 
805 struct pr_usrreqs rip6_usrreqs = {
806 	.pru_abort =		rip6_abort,
807 	.pru_attach =		rip6_attach,
808 	.pru_bind =		rip6_bind,
809 	.pru_connect =		rip6_connect,
810 	.pru_control =		in6_control,
811 	.pru_detach =		rip6_detach,
812 	.pru_disconnect =	rip6_disconnect,
813 	.pru_peeraddr =		in6_getpeeraddr,
814 	.pru_send =		rip6_send,
815 	.pru_shutdown =		rip6_shutdown,
816 	.pru_sockaddr =		in6_getsockaddr,
817 	.pru_close =		rip6_close,
818 };
819