xref: /netbsd-src/sys/netinet6/raw_ip6.c (revision b7b7574d3bf8eeb51a1fa3977b59142ec6434a55)
1 /*	$NetBSD: raw_ip6.c,v 1.121 2014/07/01 05:49:19 rtr 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 1.121 2014/07/01 05:49:19 rtr Exp $");
66 
67 #include "opt_ipsec.h"
68 
69 #include <sys/param.h>
70 #include <sys/sysctl.h>
71 #include <sys/mbuf.h>
72 #include <sys/socket.h>
73 #include <sys/protosw.h>
74 #include <sys/socketvar.h>
75 #include <sys/systm.h>
76 #include <sys/proc.h>
77 #include <sys/kauth.h>
78 #include <sys/kmem.h>
79 
80 #include <net/if.h>
81 #include <net/route.h>
82 #include <net/if_types.h>
83 #include <net/net_stats.h>
84 
85 #include <netinet/in.h>
86 #include <netinet/in_var.h>
87 #include <netinet/ip6.h>
88 #include <netinet6/ip6_var.h>
89 #include <netinet6/ip6_private.h>
90 #include <netinet6/ip6_mroute.h>
91 #include <netinet/icmp6.h>
92 #include <netinet6/icmp6_private.h>
93 #include <netinet6/in6_pcb.h>
94 #include <netinet6/nd6.h>
95 #include <netinet6/ip6protosw.h>
96 #include <netinet6/scope6_var.h>
97 #include <netinet6/raw_ip6.h>
98 
99 #ifdef IPSEC
100 #include <netipsec/ipsec.h>
101 #include <netipsec/ipsec_var.h>
102 #include <netipsec/ipsec_private.h>
103 #include <netipsec/ipsec6.h>
104 #endif
105 
106 #include "faith.h"
107 #if defined(NFAITH) && 0 < NFAITH
108 #include <net/if_faith.h>
109 #endif
110 
111 extern struct inpcbtable rawcbtable;
112 struct	inpcbtable raw6cbtable;
113 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
114 
115 /*
116  * Raw interface to IP6 protocol.
117  */
118 
119 static percpu_t *rip6stat_percpu;
120 
121 #define	RIP6_STATINC(x)		_NET_STATINC(rip6stat_percpu, x)
122 
123 static void sysctl_net_inet6_raw6_setup(struct sysctllog **);
124 
125 /*
126  * Initialize raw connection block queue.
127  */
128 void
129 rip6_init(void)
130 {
131 
132 	sysctl_net_inet6_raw6_setup(NULL);
133 	in6_pcbinit(&raw6cbtable, 1, 1);
134 
135 	rip6stat_percpu = percpu_alloc(sizeof(uint64_t) * RIP6_NSTATS);
136 }
137 
138 /*
139  * Setup generic address and protocol structures
140  * for raw_input routine, then pass them along with
141  * mbuf chain.
142  */
143 int
144 rip6_input(struct mbuf **mp, int *offp, int proto)
145 {
146 	struct mbuf *m = *mp;
147 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
148 	struct inpcb_hdr *inph;
149 	struct in6pcb *in6p;
150 	struct in6pcb *last = NULL;
151 	struct sockaddr_in6 rip6src;
152 	struct mbuf *opts = NULL;
153 
154 	RIP6_STATINC(RIP6_STAT_IPACKETS);
155 
156 #if defined(NFAITH) && 0 < NFAITH
157 	if (faithprefix(&ip6->ip6_dst)) {
158 		/* send icmp6 host unreach? */
159 		m_freem(m);
160 		return IPPROTO_DONE;
161 	}
162 #endif
163 
164 	/* Be proactive about malicious use of IPv4 mapped address */
165 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
166 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
167 		/* XXX stat */
168 		m_freem(m);
169 		return IPPROTO_DONE;
170 	}
171 
172 	sockaddr_in6_init(&rip6src, &ip6->ip6_src, 0, 0, 0);
173 	if (sa6_recoverscope(&rip6src) != 0) {
174 		/* XXX: should be impossible. */
175 		m_freem(m);
176 		return IPPROTO_DONE;
177 	}
178 
179 	TAILQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
180 		in6p = (struct in6pcb *)inph;
181 		if (in6p->in6p_af != AF_INET6)
182 			continue;
183 		if (in6p->in6p_ip6.ip6_nxt &&
184 		    in6p->in6p_ip6.ip6_nxt != proto)
185 			continue;
186 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
187 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
188 			continue;
189 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
190 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
191 			continue;
192 		if (in6p->in6p_cksum != -1) {
193 			RIP6_STATINC(RIP6_STAT_ISUM);
194 			if (in6_cksum(m, proto, *offp,
195 			    m->m_pkthdr.len - *offp)) {
196 				RIP6_STATINC(RIP6_STAT_BADSUM);
197 				continue;
198 			}
199 		}
200 		if (last) {
201 			struct	mbuf *n;
202 
203 #ifdef IPSEC
204 			/*
205 			 * Check AH/ESP integrity
206 			 */
207 			if (ipsec_used && !ipsec6_in_reject(m, last))
208 #endif /* IPSEC */
209 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
210 				if (last->in6p_flags & IN6P_CONTROLOPTS)
211 					ip6_savecontrol(last, &opts, ip6, n);
212 				/* strip intermediate headers */
213 				m_adj(n, *offp);
214 				if (sbappendaddr(&last->in6p_socket->so_rcv,
215 				    (struct sockaddr *)&rip6src, n, opts) == 0) {
216 					/* should notify about lost packet */
217 					m_freem(n);
218 					if (opts)
219 						m_freem(opts);
220 					RIP6_STATINC(RIP6_STAT_FULLSOCK);
221 				} else
222 					sorwakeup(last->in6p_socket);
223 				opts = NULL;
224 			}
225 		}
226 		last = in6p;
227 	}
228 #ifdef IPSEC
229 	if (ipsec_used && last && ipsec6_in_reject(m, last)) {
230 		m_freem(m);
231 		/*
232 		 * XXX ipsec6_in_reject update stat if there is an error
233 		 * so we just need to update stats by hand in the case of last is
234 		 * NULL
235 		 */
236 		if (!last)
237 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
238 			IP6_STATDEC(IP6_STAT_DELIVERED);
239 			/* do not inject data into pcb */
240 		} else
241 #endif /* IPSEC */
242 	if (last) {
243 		if (last->in6p_flags & IN6P_CONTROLOPTS)
244 			ip6_savecontrol(last, &opts, ip6, m);
245 		/* strip intermediate headers */
246 		m_adj(m, *offp);
247 		if (sbappendaddr(&last->in6p_socket->so_rcv,
248 		    (struct sockaddr *)&rip6src, m, opts) == 0) {
249 			m_freem(m);
250 			if (opts)
251 				m_freem(opts);
252 			RIP6_STATINC(RIP6_STAT_FULLSOCK);
253 		} else
254 			sorwakeup(last->in6p_socket);
255 	} else {
256 		RIP6_STATINC(RIP6_STAT_NOSOCK);
257 		if (m->m_flags & M_MCAST)
258 			RIP6_STATINC(RIP6_STAT_NOSOCKMCAST);
259 		if (proto == IPPROTO_NONE)
260 			m_freem(m);
261 		else {
262 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
263 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_protounknown);
264 			icmp6_error(m, ICMP6_PARAM_PROB,
265 			    ICMP6_PARAMPROB_NEXTHEADER,
266 			    prvnxtp - mtod(m, u_int8_t *));
267 		}
268 		IP6_STATDEC(IP6_STAT_DELIVERED);
269 	}
270 	return IPPROTO_DONE;
271 }
272 
273 void *
274 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
275 {
276 	struct ip6_hdr *ip6;
277 	struct ip6ctlparam *ip6cp = NULL;
278 	const struct sockaddr_in6 *sa6_src = NULL;
279 	void *cmdarg;
280 	void (*notify)(struct in6pcb *, int) = in6_rtchange;
281 	int nxt;
282 
283 	if (sa->sa_family != AF_INET6 ||
284 	    sa->sa_len != sizeof(struct sockaddr_in6))
285 		return NULL;
286 
287 	if ((unsigned)cmd >= PRC_NCMDS)
288 		return NULL;
289 	if (PRC_IS_REDIRECT(cmd))
290 		notify = in6_rtchange, d = NULL;
291 	else if (cmd == PRC_HOSTDEAD)
292 		d = NULL;
293 	else if (cmd == PRC_MSGSIZE)
294 		; /* special code is present, see below */
295 	else if (inet6ctlerrmap[cmd] == 0)
296 		return NULL;
297 
298 	/* if the parameter is from icmp6, decode it. */
299 	if (d != NULL) {
300 		ip6cp = (struct ip6ctlparam *)d;
301 		ip6 = ip6cp->ip6c_ip6;
302 		cmdarg = ip6cp->ip6c_cmdarg;
303 		sa6_src = ip6cp->ip6c_src;
304 		nxt = ip6cp->ip6c_nxt;
305 	} else {
306 		ip6 = NULL;
307 		cmdarg = NULL;
308 		sa6_src = &sa6_any;
309 		nxt = -1;
310 	}
311 
312 	if (ip6 && cmd == PRC_MSGSIZE) {
313 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
314 		int valid = 0;
315 		struct in6pcb *in6p;
316 
317 		/*
318 		 * Check to see if we have a valid raw IPv6 socket
319 		 * corresponding to the address in the ICMPv6 message
320 		 * payload, and the protocol (ip6_nxt) meets the socket.
321 		 * XXX chase extension headers, or pass final nxt value
322 		 * from icmp6_notify_error()
323 		 */
324 		in6p = NULL;
325 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
326 					     (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0, 0);
327 #if 0
328 		if (!in6p) {
329 			/*
330 			 * As the use of sendto(2) is fairly popular,
331 			 * we may want to allow non-connected pcb too.
332 			 * But it could be too weak against attacks...
333 			 * We should at least check if the local
334 			 * address (= s) is really ours.
335 			 */
336 			in6p = in6_pcblookup_bind(&raw6cbtable,
337 			    &sa6->sin6_addr, 0, 0);
338 		}
339 #endif
340 
341 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
342 		    in6p->in6p_ip6.ip6_nxt == nxt)
343 			valid++;
344 
345 		/*
346 		 * Depending on the value of "valid" and routing table
347 		 * size (mtudisc_{hi,lo}wat), we will:
348 		 * - recalculate the new MTU and create the
349 		 *   corresponding routing entry, or
350 		 * - ignore the MTU change notification.
351 		 */
352 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
353 
354 		/*
355 		 * regardless of if we called icmp6_mtudisc_update(),
356 		 * we need to call in6_pcbnotify(), to notify path MTU
357 		 * change to the userland (RFC3542), because some
358 		 * unconnected sockets may share the same destination
359 		 * and want to know the path MTU.
360 		 */
361 	}
362 
363 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
364 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
365 	return NULL;
366 }
367 
368 /*
369  * Generate IPv6 header and pass packet to ip6_output.
370  * Tack on options user may have setup with control call.
371  */
372 int
373 rip6_output(struct mbuf *m, struct socket * const so,
374     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
375 {
376 	struct in6_addr *dst;
377 	struct ip6_hdr *ip6;
378 	struct in6pcb *in6p;
379 	u_int	plen = m->m_pkthdr.len;
380 	int error = 0;
381 	struct ip6_pktopts opt, *optp = NULL;
382 	struct ifnet *oifp = NULL;
383 	int type, code;		/* for ICMPv6 output statistics only */
384 	int scope_ambiguous = 0;
385 	struct in6_addr *in6a;
386 
387 	in6p = sotoin6pcb(so);
388 
389 	dst = &dstsock->sin6_addr;
390 	if (control) {
391 		if ((error = ip6_setpktopts(control, &opt,
392 		    in6p->in6p_outputopts,
393 		    kauth_cred_get(), so->so_proto->pr_protocol)) != 0) {
394 			goto bad;
395 		}
396 		optp = &opt;
397 	} else
398 		optp = in6p->in6p_outputopts;
399 
400 	/*
401 	 * Check and convert scope zone ID into internal form.
402 	 * XXX: we may still need to determine the zone later.
403 	 */
404 	if (!(so->so_state & SS_ISCONNECTED)) {
405 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
406 			scope_ambiguous = 1;
407 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
408 			goto bad;
409 	}
410 
411 	/*
412 	 * For an ICMPv6 packet, we should know its type and code
413 	 * to update statistics.
414 	 */
415 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
416 		struct icmp6_hdr *icmp6;
417 		if (m->m_len < sizeof(struct icmp6_hdr) &&
418 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
419 			error = ENOBUFS;
420 			goto bad;
421 		}
422 		icmp6 = mtod(m, struct icmp6_hdr *);
423 		type = icmp6->icmp6_type;
424 		code = icmp6->icmp6_code;
425 	} else {
426 		type = 0;
427 		code = 0;
428 	}
429 
430 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
431 	if (!m) {
432 		error = ENOBUFS;
433 		goto bad;
434 	}
435 	ip6 = mtod(m, struct ip6_hdr *);
436 
437 	/*
438 	 * Next header might not be ICMP6 but use its pseudo header anyway.
439 	 */
440 	ip6->ip6_dst = *dst;
441 
442 	/*
443 	 * Source address selection.
444 	 */
445 	if ((in6a = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
446 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp,
447 	    &error)) == 0) {
448 		if (error == 0)
449 			error = EADDRNOTAVAIL;
450 		goto bad;
451 	}
452 	ip6->ip6_src = *in6a;
453 
454 	if (oifp && scope_ambiguous) {
455 		/*
456 		 * Application should provide a proper zone ID or the use of
457 		 * default zone IDs should be enabled.  Unfortunately, some
458 		 * applications do not behave as it should, so we need a
459 		 * workaround.  Even if an appropriate ID is not determined
460 		 * (when it's required), if we can determine the outgoing
461 		 * interface. determine the zone ID based on the interface.
462 		 */
463 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
464 		if (error != 0)
465 			goto bad;
466 	}
467 	ip6->ip6_dst = dstsock->sin6_addr;
468 
469 	/* fill in the rest of the IPv6 header fields */
470 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
471 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
472 	ip6->ip6_vfc  |= IPV6_VERSION;
473 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
474 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
475 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
476 
477 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
478 	    in6p->in6p_cksum != -1) {
479 		int off;
480 		u_int16_t sum;
481 
482 		/* compute checksum */
483 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
484 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
485 		else
486 			off = in6p->in6p_cksum;
487 		if (plen < off + 1) {
488 			error = EINVAL;
489 			goto bad;
490 		}
491 		off += sizeof(struct ip6_hdr);
492 
493 		sum = 0;
494 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
495 		    M_DONTWAIT);
496 		if (m == NULL) {
497 			error = ENOBUFS;
498 			goto bad;
499 		}
500 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
501 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
502 		    M_DONTWAIT);
503 		if (m == NULL) {
504 			error = ENOBUFS;
505 			goto bad;
506 		}
507 	}
508 
509 	error = ip6_output(m, optp, &in6p->in6p_route, 0,
510 	    in6p->in6p_moptions, so, &oifp);
511 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
512 		if (oifp)
513 			icmp6_ifoutstat_inc(oifp, type, code);
514 		ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
515 	} else
516 		RIP6_STATINC(RIP6_STAT_OPACKETS);
517 
518 	goto freectl;
519 
520  bad:
521 	if (m)
522 		m_freem(m);
523 
524  freectl:
525 	if (control) {
526 		ip6_clearpktopts(&opt, -1);
527 		m_freem(control);
528 	}
529 	return error;
530 }
531 
532 /*
533  * Raw IPv6 socket option processing.
534  */
535 int
536 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
537 {
538 	int error = 0;
539 
540 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
541 		int optval;
542 
543 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
544 		if (op == PRCO_GETOPT) {
545 			optval = 1;
546 			error = sockopt_set(sopt, &optval, sizeof(optval));
547 		} else if (op == PRCO_SETOPT) {
548 			error = sockopt_getint(sopt, &optval);
549 			if (error)
550 				goto out;
551 			if (optval == 0)
552 				error = EINVAL;
553 		}
554 
555 		goto out;
556 	} else if (sopt->sopt_level != IPPROTO_IPV6)
557 		return ip6_ctloutput(op, so, sopt);
558 
559 	switch (sopt->sopt_name) {
560 	case MRT6_INIT:
561 	case MRT6_DONE:
562 	case MRT6_ADD_MIF:
563 	case MRT6_DEL_MIF:
564 	case MRT6_ADD_MFC:
565 	case MRT6_DEL_MFC:
566 	case MRT6_PIM:
567 		if (op == PRCO_SETOPT)
568 			error = ip6_mrouter_set(so, sopt);
569 		else if (op == PRCO_GETOPT)
570 			error = ip6_mrouter_get(so, sopt);
571 		else
572 			error = EINVAL;
573 		break;
574 	case IPV6_CHECKSUM:
575 		return ip6_raw_ctloutput(op, so, sopt);
576 	default:
577 		return ip6_ctloutput(op, so, sopt);
578 	}
579  out:
580 	return error;
581 }
582 
583 extern	u_long rip6_sendspace;
584 extern	u_long rip6_recvspace;
585 
586 int
587 rip6_attach(struct socket *so, int proto)
588 {
589 	struct in6pcb *in6p;
590 	int s, error;
591 
592 	KASSERT(sotoin6pcb(so) == NULL);
593 	sosetlock(so);
594 
595 	error = kauth_authorize_network(curlwp->l_cred,
596 	    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
597 	    KAUTH_ARG(AF_INET6),
598 	    KAUTH_ARG(SOCK_RAW),
599 	    KAUTH_ARG(so->so_proto->pr_protocol));
600 	if (error) {
601 		return error;
602 	}
603 	s = splsoftnet();
604 	error = soreserve(so, rip6_sendspace, rip6_recvspace);
605 	if (error) {
606 		splx(s);
607 		return error;
608 	}
609 	if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
610 		splx(s);
611 		return error;
612 	}
613 	splx(s);
614 	in6p = sotoin6pcb(so);
615 	in6p->in6p_ip6.ip6_nxt = proto;
616 	in6p->in6p_cksum = -1;
617 
618 	in6p->in6p_icmp6filt = kmem_alloc(sizeof(struct icmp6_filter), KM_SLEEP);
619 	if (in6p->in6p_icmp6filt == NULL) {
620 		in6_pcbdetach(in6p);
621 		return ENOMEM;
622 	}
623 	ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
624 	KASSERT(solocked(so));
625 	return error;
626 }
627 
628 static void
629 rip6_detach(struct socket *so)
630 {
631 	struct in6pcb *in6p = sotoin6pcb(so);
632 
633 	KASSERT(solocked(so));
634 	KASSERT(in6p != NULL);
635 
636 	if (so == ip6_mrouter) {
637 		ip6_mrouter_done();
638 	}
639 	/* xxx: RSVP */
640 	if (in6p->in6p_icmp6filt != NULL) {
641 		kmem_free(in6p->in6p_icmp6filt, sizeof(struct icmp6_filter));
642 		in6p->in6p_icmp6filt = NULL;
643 	}
644 	in6_pcbdetach(in6p);
645 }
646 
647 static int
648 rip6_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
649 {
650 	return in6_control(so, cmd, nam, ifp);
651 }
652 
653 int
654 rip6_usrreq(struct socket *so, int req, struct mbuf *m,
655 	struct mbuf *nam, struct mbuf *control, struct lwp *l)
656 {
657 	struct in6pcb *in6p = sotoin6pcb(so);
658 	int error = 0;
659 
660 	KASSERT(req != PRU_CONTROL);
661 
662 	if (req == PRU_PURGEIF) {
663 		mutex_enter(softnet_lock);
664 		in6_pcbpurgeif0(&raw6cbtable, (struct ifnet *)control);
665 		in6_purgeif((struct ifnet *)control);
666 		in6_pcbpurgeif(&raw6cbtable, (struct ifnet *)control);
667 		mutex_exit(softnet_lock);
668 		return 0;
669 	}
670 
671 	switch (req) {
672 	case PRU_DISCONNECT:
673 		if ((so->so_state & SS_ISCONNECTED) == 0) {
674 			error = ENOTCONN;
675 			break;
676 		}
677 		in6p->in6p_faddr = in6addr_any;
678 		so->so_state &= ~SS_ISCONNECTED;	/* XXX */
679 		break;
680 
681 	case PRU_ABORT:
682 		soisdisconnected(so);
683 		rip6_detach(so);
684 		break;
685 
686 	case PRU_BIND:
687 	    {
688 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
689 		struct ifaddr *ia = NULL;
690 
691 		if (nam->m_len != sizeof(*addr)) {
692 			error = EINVAL;
693 			break;
694 		}
695 		if (!IFNET_FIRST() || addr->sin6_family != AF_INET6) {
696 			error = EADDRNOTAVAIL;
697 			break;
698 		}
699 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
700 			break;
701 
702 		/*
703 		 * we don't support mapped address here, it would confuse
704 		 * users so reject it
705 		 */
706 		if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr)) {
707 			error = EADDRNOTAVAIL;
708 			break;
709 		}
710 		if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
711 		    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0) {
712 			error = EADDRNOTAVAIL;
713 			break;
714 		}
715 		if (ia && ((struct in6_ifaddr *)ia)->ia6_flags &
716 		    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
717 		     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
718 			error = EADDRNOTAVAIL;
719 			break;
720 		}
721 		in6p->in6p_laddr = addr->sin6_addr;
722 		break;
723 	    }
724 
725 	case PRU_CONNECT:
726 	{
727 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
728 		struct in6_addr *in6a = NULL;
729 		struct ifnet *ifp = NULL;
730 		int scope_ambiguous = 0;
731 
732 		if (nam->m_len != sizeof(*addr)) {
733 			error = EINVAL;
734 			break;
735 		}
736 		if (!IFNET_FIRST()) {
737 			error = EADDRNOTAVAIL;
738 			break;
739 		}
740 		if (addr->sin6_family != AF_INET6) {
741 			error = EAFNOSUPPORT;
742 			break;
743 		}
744 
745 		/*
746 		 * Application should provide a proper zone ID or the use of
747 		 * default zone IDs should be enabled.  Unfortunately, some
748 		 * applications do not behave as it should, so we need a
749 		 * workaround.  Even if an appropriate ID is not determined,
750 		 * we'll see if we can determine the outgoing interface.  If we
751 		 * can, determine the zone ID based on the interface below.
752 		 */
753 		if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
754 			scope_ambiguous = 1;
755 		if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
756 			return error;
757 
758 		/* Source address selection. XXX: need pcblookup? */
759 		in6a = in6_selectsrc(addr, in6p->in6p_outputopts,
760 		    in6p->in6p_moptions, &in6p->in6p_route,
761 		    &in6p->in6p_laddr, &ifp, &error);
762 		if (in6a == NULL) {
763 			if (error == 0)
764 				error = EADDRNOTAVAIL;
765 			break;
766 		}
767 		/* XXX: see above */
768 		if (ifp && scope_ambiguous &&
769 		    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
770 			break;
771 		}
772 		in6p->in6p_laddr = *in6a;
773 		in6p->in6p_faddr = addr->sin6_addr;
774 		soisconnected(so);
775 		break;
776 	}
777 
778 	case PRU_CONNECT2:
779 		error = EOPNOTSUPP;
780 		break;
781 
782 	/*
783 	 * Mark the connection as being incapable of futther input.
784 	 */
785 	case PRU_SHUTDOWN:
786 		socantsendmore(so);
787 		break;
788 	/*
789 	 * Ship a packet out. The appropriate raw output
790 	 * routine handles any messaging necessary.
791 	 */
792 	case PRU_SEND:
793 	{
794 		struct sockaddr_in6 tmp;
795 		struct sockaddr_in6 *dst;
796 
797 		/* always copy sockaddr to avoid overwrites */
798 		if (so->so_state & SS_ISCONNECTED) {
799 			if (nam) {
800 				error = EISCONN;
801 				break;
802 			}
803 			/* XXX */
804 			sockaddr_in6_init(&tmp, &in6p->in6p_faddr, 0, 0, 0);
805 			dst = &tmp;
806 		} else {
807 			if (nam == NULL) {
808 				error = ENOTCONN;
809 				break;
810 			}
811 			if (nam->m_len != sizeof(tmp)) {
812 				error = EINVAL;
813 				break;
814 			}
815 
816 			tmp = *mtod(nam, struct sockaddr_in6 *);
817 			dst = &tmp;
818 
819 			if (dst->sin6_family != AF_INET6) {
820 				error = EAFNOSUPPORT;
821 				break;
822 			}
823 		}
824 		error = rip6_output(m, so, dst, control);
825 		m = NULL;
826 		break;
827 	}
828 
829 	case PRU_SENSE:
830 		/*
831 		 * stat: don't bother with a blocksize
832 		 */
833 		return 0;
834 	/*
835 	 * Not supported.
836 	 */
837 	case PRU_RCVOOB:
838 	case PRU_RCVD:
839 	case PRU_LISTEN:
840 	case PRU_ACCEPT:
841 	case PRU_SENDOOB:
842 		error = EOPNOTSUPP;
843 		break;
844 
845 	case PRU_SOCKADDR:
846 		in6_setsockaddr(in6p, nam);
847 		break;
848 
849 	case PRU_PEERADDR:
850 		in6_setpeeraddr(in6p, nam);
851 		break;
852 
853 	default:
854 		panic("rip6_usrreq");
855 	}
856 	if (m != NULL)
857 		m_freem(m);
858 	return error;
859 }
860 
861 static int
862 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
863 {
864 
865 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
866 }
867 
868 static void
869 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
870 {
871 
872 	sysctl_createv(clog, 0, NULL, NULL,
873 		       CTLFLAG_PERMANENT,
874 		       CTLTYPE_NODE, "inet6", NULL,
875 		       NULL, 0, NULL, 0,
876 		       CTL_NET, PF_INET6, CTL_EOL);
877 	sysctl_createv(clog, 0, NULL, NULL,
878 		       CTLFLAG_PERMANENT,
879 		       CTLTYPE_NODE, "raw6",
880 		       SYSCTL_DESCR("Raw IPv6 settings"),
881 		       NULL, 0, NULL, 0,
882 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
883 
884 	sysctl_createv(clog, 0, NULL, NULL,
885 		       CTLFLAG_PERMANENT,
886 		       CTLTYPE_STRUCT, "pcblist",
887 		       SYSCTL_DESCR("Raw IPv6 control block list"),
888 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
889 		       CTL_NET, PF_INET6, IPPROTO_RAW,
890 		       CTL_CREATE, CTL_EOL);
891 	sysctl_createv(clog, 0, NULL, NULL,
892 		       CTLFLAG_PERMANENT,
893 		       CTLTYPE_STRUCT, "stats",
894 		       SYSCTL_DESCR("Raw IPv6 statistics"),
895 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
896 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
897 		       CTL_EOL);
898 }
899 
900 PR_WRAP_USRREQS(rip6)
901 #define	rip6_attach		rip6_attach_wrapper
902 #define	rip6_detach		rip6_detach_wrapper
903 #define	rip6_ioctl		rip6_ioctl_wrapper
904 #define	rip6_usrreq		rip6_usrreq_wrapper
905 
906 const struct pr_usrreqs rip6_usrreqs = {
907 	.pr_attach	= rip6_attach,
908 	.pr_detach	= rip6_detach,
909 	.pr_ioctl	= rip6_ioctl,
910 	.pr_generic	= rip6_usrreq,
911 };
912