xref: /netbsd-src/sys/netinet6/raw_ip6.c (revision b1c86f5f087524e68db12794ee9c3e3da1ab17a0)
1 /*	$NetBSD: raw_ip6.c,v 1.107 2010/07/08 01:22:28 dyoung Exp $	*/
2 /*	$KAME: raw_ip6.c,v 1.82 2001/07/23 18:57:56 jinmei Exp $	*/
3 
4 /*
5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the project nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 /*
34  * Copyright (c) 1982, 1986, 1988, 1993
35  *	The Regents of the University of California.  All rights reserved.
36  *
37  * Redistribution and use in source and binary forms, with or without
38  * modification, are permitted provided that the following conditions
39  * are met:
40  * 1. Redistributions of source code must retain the above copyright
41  *    notice, this list of conditions and the following disclaimer.
42  * 2. Redistributions in binary form must reproduce the above copyright
43  *    notice, this list of conditions and the following disclaimer in the
44  *    documentation and/or other materials provided with the distribution.
45  * 3. Neither the name of the University nor the names of its contributors
46  *    may be used to endorse or promote products derived from this software
47  *    without specific prior written permission.
48  *
49  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59  * SUCH DAMAGE.
60  *
61  *	@(#)raw_ip.c	8.2 (Berkeley) 1/4/94
62  */
63 
64 #include <sys/cdefs.h>
65 __KERNEL_RCSID(0, "$NetBSD: raw_ip6.c,v 1.107 2010/07/08 01:22:28 dyoung Exp $");
66 
67 #include "opt_ipsec.h"
68 
69 #include <sys/param.h>
70 #include <sys/sysctl.h>
71 #include <sys/malloc.h>
72 #include <sys/mbuf.h>
73 #include <sys/socket.h>
74 #include <sys/protosw.h>
75 #include <sys/socketvar.h>
76 #include <sys/errno.h>
77 #include <sys/systm.h>
78 #include <sys/proc.h>
79 #include <sys/kauth.h>
80 
81 #include <net/if.h>
82 #include <net/route.h>
83 #include <net/if_types.h>
84 #include <net/net_stats.h>
85 
86 #include <netinet/in.h>
87 #include <netinet/in_var.h>
88 #include <netinet/ip6.h>
89 #include <netinet6/ip6_var.h>
90 #include <netinet6/ip6_private.h>
91 #include <netinet6/ip6_mroute.h>
92 #include <netinet/icmp6.h>
93 #include <netinet6/icmp6_private.h>
94 #include <netinet6/in6_pcb.h>
95 #include <netinet6/nd6.h>
96 #include <netinet6/ip6protosw.h>
97 #include <netinet6/scope6_var.h>
98 #include <netinet6/raw_ip6.h>
99 
100 #ifdef IPSEC
101 #include <netinet6/ipsec.h>
102 #include <netinet6/ipsec_private.h>
103 #endif /* IPSEC */
104 
105 #ifdef FAST_IPSEC
106 #include <netipsec/ipsec.h>
107 #include <netipsec/ipsec_var.h>
108 #include <netipsec/ipsec_private.h>
109 #include <netipsec/ipsec6.h>
110 #endif
111 
112 #include "faith.h"
113 #if defined(NFAITH) && 0 < NFAITH
114 #include <net/if_faith.h>
115 #endif
116 
117 extern struct inpcbtable rawcbtable;
118 struct	inpcbtable raw6cbtable;
119 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
120 
121 /*
122  * Raw interface to IP6 protocol.
123  */
124 
125 static percpu_t *rip6stat_percpu;
126 
127 #define	RIP6_STATINC(x)		_NET_STATINC(rip6stat_percpu, x)
128 
129 static void sysctl_net_inet6_raw6_setup(struct sysctllog **);
130 
131 /*
132  * Initialize raw connection block queue.
133  */
134 void
135 rip6_init(void)
136 {
137 
138 	sysctl_net_inet6_raw6_setup(NULL);
139 	in6_pcbinit(&raw6cbtable, 1, 1);
140 
141 	rip6stat_percpu = percpu_alloc(sizeof(uint64_t) * RIP6_NSTATS);
142 }
143 
144 /*
145  * Setup generic address and protocol structures
146  * for raw_input routine, then pass them along with
147  * mbuf chain.
148  */
149 int
150 rip6_input(struct mbuf **mp, int *offp, int proto)
151 {
152 	struct mbuf *m = *mp;
153 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
154 	struct inpcb_hdr *inph;
155 	struct in6pcb *in6p;
156 	struct in6pcb *last = NULL;
157 	struct sockaddr_in6 rip6src;
158 	struct mbuf *opts = NULL;
159 
160 	RIP6_STATINC(RIP6_STAT_IPACKETS);
161 
162 #if defined(NFAITH) && 0 < NFAITH
163 	if (faithprefix(&ip6->ip6_dst)) {
164 		/* send icmp6 host unreach? */
165 		m_freem(m);
166 		return IPPROTO_DONE;
167 	}
168 #endif
169 
170 	/* Be proactive about malicious use of IPv4 mapped address */
171 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
172 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
173 		/* XXX stat */
174 		m_freem(m);
175 		return IPPROTO_DONE;
176 	}
177 
178 	sockaddr_in6_init(&rip6src, &ip6->ip6_src, 0, 0, 0);
179 	if (sa6_recoverscope(&rip6src) != 0) {
180 		/* XXX: should be impossible. */
181 		m_freem(m);
182 		return IPPROTO_DONE;
183 	}
184 
185 	CIRCLEQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
186 		in6p = (struct in6pcb *)inph;
187 		if (in6p->in6p_af != AF_INET6)
188 			continue;
189 		if (in6p->in6p_ip6.ip6_nxt &&
190 		    in6p->in6p_ip6.ip6_nxt != proto)
191 			continue;
192 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
193 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
194 			continue;
195 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
196 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
197 			continue;
198 		if (in6p->in6p_cksum != -1) {
199 			RIP6_STATINC(RIP6_STAT_ISUM);
200 			if (in6_cksum(m, proto, *offp,
201 			    m->m_pkthdr.len - *offp)) {
202 				RIP6_STATINC(RIP6_STAT_BADSUM);
203 				continue;
204 			}
205 		}
206 		if (last) {
207 			struct	mbuf *n;
208 
209 #ifdef IPSEC
210 			/*
211 			 * Check AH/ESP integrity.
212 			 */
213 			if (ipsec6_in_reject(m, last)) {
214 				IPSEC6_STATINC(IPSEC_STAT_IN_INVAL);
215 				/* do not inject data into pcb */
216 			} else
217 #endif /* IPSEC */
218 #ifdef FAST_IPSEC
219 			/*
220 			 * Check AH/ESP integrity
221 			 */
222 			if (!ipsec6_in_reject(m,last))
223 #endif /* FAST_IPSEC */
224 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
225 				if (last->in6p_flags & IN6P_CONTROLOPTS)
226 					ip6_savecontrol(last, &opts, ip6, n);
227 				/* strip intermediate headers */
228 				m_adj(n, *offp);
229 				if (sbappendaddr(&last->in6p_socket->so_rcv,
230 				    (struct sockaddr *)&rip6src, n, opts) == 0) {
231 					/* should notify about lost packet */
232 					m_freem(n);
233 					if (opts)
234 						m_freem(opts);
235 					RIP6_STATINC(RIP6_STAT_FULLSOCK);
236 				} else
237 					sorwakeup(last->in6p_socket);
238 				opts = NULL;
239 			}
240 		}
241 		last = in6p;
242 	}
243 #ifdef IPSEC
244 	/*
245 	 * Check AH/ESP integrity.
246 	 */
247 	if (last && ipsec6_in_reject(m, last)) {
248 		m_freem(m);
249 		IPSEC6_STATINC(IPSEC_STAT_IN_INVAL);
250 		IP6_STATDEC(IP6_STAT_DELIVERED);
251 		/* do not inject data into pcb */
252 	} else
253 #endif /* IPSEC */
254 #ifdef FAST_IPSEC
255 	if (last && ipsec6_in_reject(m, last)) {
256 		m_freem(m);
257 		/*
258 		 * XXX ipsec6_in_reject update stat if there is an error
259 		 * so we just need to update stats by hand in the case of last is
260 		 * NULL
261 		 */
262 		if (!last)
263 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
264 			IP6_STATDEC(IP6_STAT_DELIVERED);
265 			/* do not inject data into pcb */
266 		} else
267 #endif /* FAST_IPSEC */
268 	if (last) {
269 		if (last->in6p_flags & IN6P_CONTROLOPTS)
270 			ip6_savecontrol(last, &opts, ip6, m);
271 		/* strip intermediate headers */
272 		m_adj(m, *offp);
273 		if (sbappendaddr(&last->in6p_socket->so_rcv,
274 		    (struct sockaddr *)&rip6src, m, opts) == 0) {
275 			m_freem(m);
276 			if (opts)
277 				m_freem(opts);
278 			RIP6_STATINC(RIP6_STAT_FULLSOCK);
279 		} else
280 			sorwakeup(last->in6p_socket);
281 	} else {
282 		RIP6_STATINC(RIP6_STAT_NOSOCK);
283 		if (m->m_flags & M_MCAST)
284 			RIP6_STATINC(RIP6_STAT_NOSOCKMCAST);
285 		if (proto == IPPROTO_NONE)
286 			m_freem(m);
287 		else {
288 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
289 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_protounknown);
290 			icmp6_error(m, ICMP6_PARAM_PROB,
291 			    ICMP6_PARAMPROB_NEXTHEADER,
292 			    prvnxtp - mtod(m, u_int8_t *));
293 		}
294 		IP6_STATDEC(IP6_STAT_DELIVERED);
295 	}
296 	return IPPROTO_DONE;
297 }
298 
299 void *
300 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
301 {
302 	struct ip6_hdr *ip6;
303 	struct ip6ctlparam *ip6cp = NULL;
304 	const struct sockaddr_in6 *sa6_src = NULL;
305 	void *cmdarg;
306 	void (*notify)(struct in6pcb *, int) = in6_rtchange;
307 	int nxt;
308 
309 	if (sa->sa_family != AF_INET6 ||
310 	    sa->sa_len != sizeof(struct sockaddr_in6))
311 		return NULL;
312 
313 	if ((unsigned)cmd >= PRC_NCMDS)
314 		return NULL;
315 	if (PRC_IS_REDIRECT(cmd))
316 		notify = in6_rtchange, d = NULL;
317 	else if (cmd == PRC_HOSTDEAD)
318 		d = NULL;
319 	else if (cmd == PRC_MSGSIZE)
320 		; /* special code is present, see below */
321 	else if (inet6ctlerrmap[cmd] == 0)
322 		return NULL;
323 
324 	/* if the parameter is from icmp6, decode it. */
325 	if (d != NULL) {
326 		ip6cp = (struct ip6ctlparam *)d;
327 		ip6 = ip6cp->ip6c_ip6;
328 		cmdarg = ip6cp->ip6c_cmdarg;
329 		sa6_src = ip6cp->ip6c_src;
330 		nxt = ip6cp->ip6c_nxt;
331 	} else {
332 		ip6 = NULL;
333 		cmdarg = NULL;
334 		sa6_src = &sa6_any;
335 		nxt = -1;
336 	}
337 
338 	if (ip6 && cmd == PRC_MSGSIZE) {
339 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
340 		int valid = 0;
341 		struct in6pcb *in6p;
342 
343 		/*
344 		 * Check to see if we have a valid raw IPv6 socket
345 		 * corresponding to the address in the ICMPv6 message
346 		 * payload, and the protocol (ip6_nxt) meets the socket.
347 		 * XXX chase extension headers, or pass final nxt value
348 		 * from icmp6_notify_error()
349 		 */
350 		in6p = NULL;
351 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
352 		    (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0);
353 #if 0
354 		if (!in6p) {
355 			/*
356 			 * As the use of sendto(2) is fairly popular,
357 			 * we may want to allow non-connected pcb too.
358 			 * But it could be too weak against attacks...
359 			 * We should at least check if the local
360 			 * address (= s) is really ours.
361 			 */
362 			in6p = in6_pcblookup_bind(&raw6cbtable,
363 			    &sa6->sin6_addr, 0, 0);
364 		}
365 #endif
366 
367 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
368 		    in6p->in6p_ip6.ip6_nxt == nxt)
369 			valid++;
370 
371 		/*
372 		 * Depending on the value of "valid" and routing table
373 		 * size (mtudisc_{hi,lo}wat), we will:
374 		 * - recalculate the new MTU and create the
375 		 *   corresponding routing entry, or
376 		 * - ignore the MTU change notification.
377 		 */
378 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
379 
380 		/*
381 		 * regardless of if we called icmp6_mtudisc_update(),
382 		 * we need to call in6_pcbnotify(), to notify path MTU
383 		 * change to the userland (RFC3542), because some
384 		 * unconnected sockets may share the same destination
385 		 * and want to know the path MTU.
386 		 */
387 	}
388 
389 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
390 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
391 	return NULL;
392 }
393 
394 /*
395  * Generate IPv6 header and pass packet to ip6_output.
396  * Tack on options user may have setup with control call.
397  */
398 int
399 rip6_output(struct mbuf *m, struct socket * const so,
400     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
401 {
402 	struct in6_addr *dst;
403 	struct ip6_hdr *ip6;
404 	struct in6pcb *in6p;
405 	u_int	plen = m->m_pkthdr.len;
406 	int error = 0;
407 	struct ip6_pktopts opt, *optp = NULL;
408 	struct ifnet *oifp = NULL;
409 	int type, code;		/* for ICMPv6 output statistics only */
410 	int scope_ambiguous = 0;
411 	struct in6_addr *in6a;
412 
413 	in6p = sotoin6pcb(so);
414 
415 	dst = &dstsock->sin6_addr;
416 	if (control) {
417 		if ((error = ip6_setpktopts(control, &opt,
418 		    in6p->in6p_outputopts,
419 		    kauth_cred_get(), so->so_proto->pr_protocol)) != 0) {
420 			goto bad;
421 		}
422 		optp = &opt;
423 	} else
424 		optp = in6p->in6p_outputopts;
425 
426 	/*
427 	 * Check and convert scope zone ID into internal form.
428 	 * XXX: we may still need to determine the zone later.
429 	 */
430 	if (!(so->so_state & SS_ISCONNECTED)) {
431 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
432 			scope_ambiguous = 1;
433 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
434 			goto bad;
435 	}
436 
437 	/*
438 	 * For an ICMPv6 packet, we should know its type and code
439 	 * to update statistics.
440 	 */
441 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
442 		struct icmp6_hdr *icmp6;
443 		if (m->m_len < sizeof(struct icmp6_hdr) &&
444 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
445 			error = ENOBUFS;
446 			goto bad;
447 		}
448 		icmp6 = mtod(m, struct icmp6_hdr *);
449 		type = icmp6->icmp6_type;
450 		code = icmp6->icmp6_code;
451 	} else {
452 		type = 0;
453 		code = 0;
454 	}
455 
456 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
457 	if (!m) {
458 		error = ENOBUFS;
459 		goto bad;
460 	}
461 	ip6 = mtod(m, struct ip6_hdr *);
462 
463 	/*
464 	 * Next header might not be ICMP6 but use its pseudo header anyway.
465 	 */
466 	ip6->ip6_dst = *dst;
467 
468 	/*
469 	 * Source address selection.
470 	 */
471 	if ((in6a = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
472 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp,
473 	    &error)) == 0) {
474 		if (error == 0)
475 			error = EADDRNOTAVAIL;
476 		goto bad;
477 	}
478 	ip6->ip6_src = *in6a;
479 
480 	if (oifp && scope_ambiguous) {
481 		/*
482 		 * Application should provide a proper zone ID or the use of
483 		 * default zone IDs should be enabled.  Unfortunately, some
484 		 * applications do not behave as it should, so we need a
485 		 * workaround.  Even if an appropriate ID is not determined
486 		 * (when it's required), if we can determine the outgoing
487 		 * interface. determine the zone ID based on the interface.
488 		 */
489 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
490 		if (error != 0)
491 			goto bad;
492 	}
493 	ip6->ip6_dst = dstsock->sin6_addr;
494 
495 	/* fill in the rest of the IPv6 header fields */
496 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
497 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
498 	ip6->ip6_vfc  |= IPV6_VERSION;
499 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
500 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
501 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
502 
503 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
504 	    in6p->in6p_cksum != -1) {
505 		int off;
506 		u_int16_t sum;
507 
508 		/* compute checksum */
509 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
510 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
511 		else
512 			off = in6p->in6p_cksum;
513 		if (plen < off + 1) {
514 			error = EINVAL;
515 			goto bad;
516 		}
517 		off += sizeof(struct ip6_hdr);
518 
519 		sum = 0;
520 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
521 		    M_DONTWAIT);
522 		if (m == NULL) {
523 			error = ENOBUFS;
524 			goto bad;
525 		}
526 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
527 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
528 		    M_DONTWAIT);
529 		if (m == NULL) {
530 			error = ENOBUFS;
531 			goto bad;
532 		}
533 	}
534 
535 	error = ip6_output(m, optp, &in6p->in6p_route, 0,
536 	    in6p->in6p_moptions, so, &oifp);
537 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
538 		if (oifp)
539 			icmp6_ifoutstat_inc(oifp, type, code);
540 		ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
541 	} else
542 		RIP6_STATINC(RIP6_STAT_OPACKETS);
543 
544 	goto freectl;
545 
546  bad:
547 	if (m)
548 		m_freem(m);
549 
550  freectl:
551 	if (control) {
552 		ip6_clearpktopts(&opt, -1);
553 		m_freem(control);
554 	}
555 	return error;
556 }
557 
558 /*
559  * Raw IPv6 socket option processing.
560  */
561 int
562 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
563 {
564 	int error = 0;
565 
566 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
567 		int optval;
568 
569 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
570 		if (op == PRCO_GETOPT) {
571 			optval = 1;
572 			error = sockopt_set(sopt, &optval, sizeof(optval));
573 		} else if (op == PRCO_SETOPT) {
574 			error = sockopt_getint(sopt, &optval);
575 			if (error)
576 				goto out;
577 			if (optval == 0)
578 				error = EINVAL;
579 		}
580 
581 		goto out;
582 	} else if (sopt->sopt_level != IPPROTO_IPV6)
583 		return ip6_ctloutput(op, so, sopt);
584 
585 	switch (sopt->sopt_name) {
586 	case MRT6_INIT:
587 	case MRT6_DONE:
588 	case MRT6_ADD_MIF:
589 	case MRT6_DEL_MIF:
590 	case MRT6_ADD_MFC:
591 	case MRT6_DEL_MFC:
592 	case MRT6_PIM:
593 		if (op == PRCO_SETOPT)
594 			error = ip6_mrouter_set(so, sopt);
595 		else if (op == PRCO_GETOPT)
596 			error = ip6_mrouter_get(so, sopt);
597 		else
598 			error = EINVAL;
599 		break;
600 	case IPV6_CHECKSUM:
601 		return ip6_raw_ctloutput(op, so, sopt);
602 	default:
603 		return ip6_ctloutput(op, so, sopt);
604 	}
605  out:
606 	return error;
607 }
608 
609 extern	u_long rip6_sendspace;
610 extern	u_long rip6_recvspace;
611 
612 int
613 rip6_usrreq(struct socket *so, int req, struct mbuf *m,
614 	struct mbuf *nam, struct mbuf *control, struct lwp *l)
615 {
616 	struct in6pcb *in6p = sotoin6pcb(so);
617 	int s;
618 	int error = 0;
619 
620 	if (req == PRU_CONTROL)
621 		return in6_control(so, (u_long)m, (void *)nam,
622 		    (struct ifnet *)control, l);
623 
624 	if (req == PRU_PURGEIF) {
625 		mutex_enter(softnet_lock);
626 		in6_pcbpurgeif0(&raw6cbtable, (struct ifnet *)control);
627 		in6_purgeif((struct ifnet *)control);
628 		in6_pcbpurgeif(&raw6cbtable, (struct ifnet *)control);
629 		mutex_exit(softnet_lock);
630 		return 0;
631 	}
632 
633 	switch (req) {
634 	case PRU_ATTACH:
635 		error = kauth_authorize_network(l->l_cred,
636 		    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
637 		    NULL, NULL, NULL);
638 		sosetlock(so);
639 		if (in6p != NULL)
640 			panic("rip6_attach");
641 		if (error) {
642 			break;
643 		}
644 		s = splsoftnet();
645 		error = soreserve(so, rip6_sendspace, rip6_recvspace);
646 		if (error != 0) {
647 			splx(s);
648 			break;
649 		}
650 		if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
651 			splx(s);
652 			break;
653 		}
654 		splx(s);
655 		in6p = sotoin6pcb(so);
656 		in6p->in6p_ip6.ip6_nxt = (long)nam;
657 		in6p->in6p_cksum = -1;
658 
659 		in6p->in6p_icmp6filt = malloc(sizeof(struct icmp6_filter),
660 			M_PCB, M_NOWAIT);
661 		if (in6p->in6p_icmp6filt == NULL) {
662 			in6_pcbdetach(in6p);
663 			error = ENOMEM;
664 			break;
665 		}
666 		ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
667 		break;
668 
669 	case PRU_DISCONNECT:
670 		if ((so->so_state & SS_ISCONNECTED) == 0) {
671 			error = ENOTCONN;
672 			break;
673 		}
674 		in6p->in6p_faddr = in6addr_any;
675 		so->so_state &= ~SS_ISCONNECTED;	/* XXX */
676 		break;
677 
678 	case PRU_ABORT:
679 		soisdisconnected(so);
680 		/* Fallthrough */
681 	case PRU_DETACH:
682 		if (in6p == NULL)
683 			panic("rip6_detach");
684 		if (so == ip6_mrouter)
685 			ip6_mrouter_done();
686 		/* xxx: RSVP */
687 		if (in6p->in6p_icmp6filt != NULL) {
688 			free(in6p->in6p_icmp6filt, M_PCB);
689 			in6p->in6p_icmp6filt = NULL;
690 		}
691 		in6_pcbdetach(in6p);
692 		break;
693 
694 	case PRU_BIND:
695 	    {
696 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
697 		struct ifaddr *ia = NULL;
698 
699 		if (nam->m_len != sizeof(*addr)) {
700 			error = EINVAL;
701 			break;
702 		}
703 		if (TAILQ_EMPTY(&ifnet) || addr->sin6_family != AF_INET6) {
704 			error = EADDRNOTAVAIL;
705 			break;
706 		}
707 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
708 			break;
709 
710 		/*
711 		 * we don't support mapped address here, it would confuse
712 		 * users so reject it
713 		 */
714 		if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr)) {
715 			error = EADDRNOTAVAIL;
716 			break;
717 		}
718 		if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
719 		    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0) {
720 			error = EADDRNOTAVAIL;
721 			break;
722 		}
723 		if (ia && ((struct in6_ifaddr *)ia)->ia6_flags &
724 		    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
725 		     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
726 			error = EADDRNOTAVAIL;
727 			break;
728 		}
729 		in6p->in6p_laddr = addr->sin6_addr;
730 		break;
731 	    }
732 
733 	case PRU_CONNECT:
734 	{
735 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
736 		struct in6_addr *in6a = NULL;
737 		struct ifnet *ifp = NULL;
738 		int scope_ambiguous = 0;
739 
740 		if (nam->m_len != sizeof(*addr)) {
741 			error = EINVAL;
742 			break;
743 		}
744 		if (TAILQ_EMPTY(&ifnet)) {
745 			error = EADDRNOTAVAIL;
746 			break;
747 		}
748 		if (addr->sin6_family != AF_INET6) {
749 			error = EAFNOSUPPORT;
750 			break;
751 		}
752 
753 		/*
754 		 * Application should provide a proper zone ID or the use of
755 		 * default zone IDs should be enabled.  Unfortunately, some
756 		 * applications do not behave as it should, so we need a
757 		 * workaround.  Even if an appropriate ID is not determined,
758 		 * we'll see if we can determine the outgoing interface.  If we
759 		 * can, determine the zone ID based on the interface below.
760 		 */
761 		if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
762 			scope_ambiguous = 1;
763 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
764 			return error;
765 
766 		/* Source address selection. XXX: need pcblookup? */
767 		in6a = in6_selectsrc(addr, in6p->in6p_outputopts,
768 		    in6p->in6p_moptions, &in6p->in6p_route,
769 		    &in6p->in6p_laddr, &ifp, &error);
770 		if (in6a == NULL) {
771 			if (error == 0)
772 				error = EADDRNOTAVAIL;
773 			break;
774 		}
775 		/* XXX: see above */
776 		if (ifp && scope_ambiguous &&
777 		    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
778 			break;
779 		}
780 		in6p->in6p_laddr = *in6a;
781 		in6p->in6p_faddr = addr->sin6_addr;
782 		soisconnected(so);
783 		break;
784 	}
785 
786 	case PRU_CONNECT2:
787 		error = EOPNOTSUPP;
788 		break;
789 
790 	/*
791 	 * Mark the connection as being incapable of futther input.
792 	 */
793 	case PRU_SHUTDOWN:
794 		socantsendmore(so);
795 		break;
796 	/*
797 	 * Ship a packet out. The appropriate raw output
798 	 * routine handles any messaging necessary.
799 	 */
800 	case PRU_SEND:
801 	{
802 		struct sockaddr_in6 tmp;
803 		struct sockaddr_in6 *dst;
804 
805 		/* always copy sockaddr to avoid overwrites */
806 		if (so->so_state & SS_ISCONNECTED) {
807 			if (nam) {
808 				error = EISCONN;
809 				break;
810 			}
811 			/* XXX */
812 			sockaddr_in6_init(&tmp, &in6p->in6p_faddr, 0, 0, 0);
813 			dst = &tmp;
814 		} else {
815 			if (nam == NULL) {
816 				error = ENOTCONN;
817 				break;
818 			}
819 			if (nam->m_len != sizeof(tmp)) {
820 				error = EINVAL;
821 				break;
822 			}
823 
824 			tmp = *mtod(nam, struct sockaddr_in6 *);
825 			dst = &tmp;
826 
827 			if (dst->sin6_family != AF_INET6) {
828 				error = EAFNOSUPPORT;
829 				break;
830 			}
831 		}
832 		error = rip6_output(m, so, dst, control);
833 		m = NULL;
834 		break;
835 	}
836 
837 	case PRU_SENSE:
838 		/*
839 		 * stat: don't bother with a blocksize
840 		 */
841 		return 0;
842 	/*
843 	 * Not supported.
844 	 */
845 	case PRU_RCVOOB:
846 	case PRU_RCVD:
847 	case PRU_LISTEN:
848 	case PRU_ACCEPT:
849 	case PRU_SENDOOB:
850 		error = EOPNOTSUPP;
851 		break;
852 
853 	case PRU_SOCKADDR:
854 		in6_setsockaddr(in6p, nam);
855 		break;
856 
857 	case PRU_PEERADDR:
858 		in6_setpeeraddr(in6p, nam);
859 		break;
860 
861 	default:
862 		panic("rip6_usrreq");
863 	}
864 	if (m != NULL)
865 		m_freem(m);
866 	return error;
867 }
868 
869 static int
870 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
871 {
872 
873 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
874 }
875 
876 static void
877 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
878 {
879 
880 	sysctl_createv(clog, 0, NULL, NULL,
881 		       CTLFLAG_PERMANENT,
882 		       CTLTYPE_NODE, "net", NULL,
883 		       NULL, 0, NULL, 0,
884 		       CTL_NET, CTL_EOL);
885 	sysctl_createv(clog, 0, NULL, NULL,
886 		       CTLFLAG_PERMANENT,
887 		       CTLTYPE_NODE, "inet6", NULL,
888 		       NULL, 0, NULL, 0,
889 		       CTL_NET, PF_INET6, CTL_EOL);
890 	sysctl_createv(clog, 0, NULL, NULL,
891 		       CTLFLAG_PERMANENT,
892 		       CTLTYPE_NODE, "raw6",
893 		       SYSCTL_DESCR("Raw IPv6 settings"),
894 		       NULL, 0, NULL, 0,
895 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
896 
897 	sysctl_createv(clog, 0, NULL, NULL,
898 		       CTLFLAG_PERMANENT,
899 		       CTLTYPE_STRUCT, "pcblist",
900 		       SYSCTL_DESCR("Raw IPv6 control block list"),
901 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
902 		       CTL_NET, PF_INET6, IPPROTO_RAW,
903 		       CTL_CREATE, CTL_EOL);
904 	sysctl_createv(clog, 0, NULL, NULL,
905 		       CTLFLAG_PERMANENT,
906 		       CTLTYPE_STRUCT, "stats",
907 		       SYSCTL_DESCR("Raw IPv6 statistics"),
908 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
909 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
910 		       CTL_EOL);
911 }
912