xref: /netbsd-src/sys/netinet6/raw_ip6.c (revision 5b84b3983f71fd20a534cfa5d1556623a8aaa717)
1 /*	$NetBSD: raw_ip6.c,v 1.73 2005/08/28 21:04:09 rpaulo 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 1.73 2005/08/28 21:04:09 rpaulo 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 
80 #include <net/if.h>
81 #include <net/route.h>
82 #include <net/if_types.h>
83 
84 #include <netinet/in.h>
85 #include <netinet/in_var.h>
86 #include <netinet/ip6.h>
87 #include <netinet6/ip6_var.h>
88 #include <netinet6/ip6_mroute.h>
89 #include <netinet/icmp6.h>
90 #include <netinet6/in6_pcb.h>
91 #include <netinet6/nd6.h>
92 #include <netinet6/ip6protosw.h>
93 #ifdef ENABLE_DEFAULT_SCOPE
94 #include <netinet6/scope6_var.h>
95 #endif
96 #include <netinet6/raw_ip6.h>
97 
98 #ifdef IPSEC
99 #include <netinet6/ipsec.h>
100 #endif /* IPSEC */
101 
102 #include <machine/stdarg.h>
103 
104 #include "faith.h"
105 #if defined(NFAITH) && 0 < NFAITH
106 #include <net/if_faith.h>
107 #endif
108 
109 extern struct inpcbtable rawcbtable;
110 struct	inpcbtable raw6cbtable;
111 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
112 
113 /*
114  * Raw interface to IP6 protocol.
115  */
116 
117 struct rip6stat rip6stat;
118 
119 /*
120  * Initialize raw connection block queue.
121  */
122 void
123 rip6_init()
124 {
125 
126 	in6_pcbinit(&raw6cbtable, 1, 1);
127 }
128 
129 /*
130  * Setup generic address and protocol structures
131  * for raw_input routine, then pass them along with
132  * mbuf chain.
133  */
134 int
135 rip6_input(mp, offp, proto)
136 	struct	mbuf **mp;
137 	int	*offp, proto;
138 {
139 	struct mbuf *m = *mp;
140 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
141 	struct inpcb_hdr *inph;
142 	struct in6pcb *in6p;
143 	struct in6pcb *last = NULL;
144 	struct sockaddr_in6 rip6src;
145 	struct mbuf *opts = NULL;
146 
147 	rip6stat.rip6s_ipackets++;
148 
149 #if defined(NFAITH) && 0 < NFAITH
150 	if (faithprefix(&ip6->ip6_dst)) {
151 		/* send icmp6 host unreach? */
152 		m_freem(m);
153 		return IPPROTO_DONE;
154 	}
155 #endif
156 
157 	/* Be proactive about malicious use of IPv4 mapped address */
158 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
159 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
160 		/* XXX stat */
161 		m_freem(m);
162 		return IPPROTO_DONE;
163 	}
164 
165 	bzero(&rip6src, sizeof(rip6src));
166 	rip6src.sin6_len = sizeof(struct sockaddr_in6);
167 	rip6src.sin6_family = AF_INET6;
168 	/* KAME hack: recover scopeid */
169 	(void)in6_recoverscope(&rip6src, &ip6->ip6_src, m->m_pkthdr.rcvif);
170 
171 	CIRCLEQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
172 		in6p = (struct in6pcb *)inph;
173 		if (in6p->in6p_af != AF_INET6)
174 			continue;
175 		if (in6p->in6p_ip6.ip6_nxt &&
176 		    in6p->in6p_ip6.ip6_nxt != proto)
177 			continue;
178 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
179 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
180 			continue;
181 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
182 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
183 			continue;
184 		if (in6p->in6p_cksum != -1) {
185 			rip6stat.rip6s_isum++;
186 			if (in6_cksum(m, proto, *offp,
187 			    m->m_pkthdr.len - *offp)) {
188 				rip6stat.rip6s_badsum++;
189 				continue;
190 			}
191 		}
192 		if (last) {
193 			struct	mbuf *n;
194 
195 #ifdef IPSEC
196 			/*
197 			 * Check AH/ESP integrity.
198 			 */
199 			if (ipsec6_in_reject(m, last)) {
200 				ipsec6stat.in_polvio++;
201 				/* do not inject data into pcb */
202 			} else
203 #endif /* IPSEC */
204 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
205 				if (last->in6p_flags & IN6P_CONTROLOPTS)
206 					ip6_savecontrol(last, &opts, ip6, n);
207 				/* strip intermediate headers */
208 				m_adj(n, *offp);
209 				if (sbappendaddr(&last->in6p_socket->so_rcv,
210 				    (struct sockaddr *)&rip6src, n, opts) == 0) {
211 					/* should notify about lost packet */
212 					m_freem(n);
213 					if (opts)
214 						m_freem(opts);
215 					rip6stat.rip6s_fullsock++;
216 				} else
217 					sorwakeup(last->in6p_socket);
218 				opts = NULL;
219 			}
220 		}
221 		last = in6p;
222 	}
223 #ifdef IPSEC
224 	/*
225 	 * Check AH/ESP integrity.
226 	 */
227 	if (last && ipsec6_in_reject(m, last)) {
228 		m_freem(m);
229 		ipsec6stat.in_polvio++;
230 		ip6stat.ip6s_delivered--;
231 		/* do not inject data into pcb */
232 	} else
233 #endif /* IPSEC */
234 	if (last) {
235 		if (last->in6p_flags & IN6P_CONTROLOPTS)
236 			ip6_savecontrol(last, &opts, ip6, m);
237 		/* strip intermediate headers */
238 		m_adj(m, *offp);
239 		if (sbappendaddr(&last->in6p_socket->so_rcv,
240 		    (struct sockaddr *)&rip6src, m, opts) == 0) {
241 			m_freem(m);
242 			if (opts)
243 				m_freem(opts);
244 			rip6stat.rip6s_fullsock++;
245 		} else
246 			sorwakeup(last->in6p_socket);
247 	} else {
248 		rip6stat.rip6s_nosock++;
249 		if (m->m_flags & M_MCAST)
250 			rip6stat.rip6s_nosockmcast++;
251 		if (proto == IPPROTO_NONE)
252 			m_freem(m);
253 		else {
254 			u_int8_t *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
255 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_protounknown);
256 			icmp6_error(m, ICMP6_PARAM_PROB,
257 			    ICMP6_PARAMPROB_NEXTHEADER,
258 			    prvnxtp - mtod(m, u_int8_t *));
259 		}
260 		ip6stat.ip6s_delivered--;
261 	}
262 	return IPPROTO_DONE;
263 }
264 
265 void
266 rip6_ctlinput(cmd, sa, d)
267 	int cmd;
268 	struct sockaddr *sa;
269 	void *d;
270 {
271 	struct ip6_hdr *ip6;
272 	struct ip6ctlparam *ip6cp = NULL;
273 	const struct sockaddr_in6 *sa6_src = NULL;
274 	void *cmdarg;
275 	void (*notify) __P((struct in6pcb *, int)) = in6_rtchange;
276 	int nxt;
277 
278 	if (sa->sa_family != AF_INET6 ||
279 	    sa->sa_len != sizeof(struct sockaddr_in6))
280 		return;
281 
282 	if ((unsigned)cmd >= PRC_NCMDS)
283 		return;
284 	if (PRC_IS_REDIRECT(cmd))
285 		notify = in6_rtchange, d = NULL;
286 	else if (cmd == PRC_HOSTDEAD)
287 		d = NULL;
288 	else if (cmd == PRC_MSGSIZE)
289 		; /* special code is present, see below */
290 	else if (inet6ctlerrmap[cmd] == 0)
291 		return;
292 
293 	/* if the parameter is from icmp6, decode it. */
294 	if (d != NULL) {
295 		ip6cp = (struct ip6ctlparam *)d;
296 		ip6 = ip6cp->ip6c_ip6;
297 		cmdarg = ip6cp->ip6c_cmdarg;
298 		sa6_src = ip6cp->ip6c_src;
299 		nxt = ip6cp->ip6c_nxt;
300 	} else {
301 		ip6 = NULL;
302 		cmdarg = NULL;
303 		sa6_src = &sa6_any;
304 		nxt = -1;
305 	}
306 
307 	if (ip6 && cmd == PRC_MSGSIZE) {
308 		struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)sa;
309 		int valid = 0;
310 		struct in6pcb *in6p;
311 
312 		/*
313 		 * Check to see if we have a valid raw IPv6 socket
314 		 * corresponding to the address in the ICMPv6 message
315 		 * payload, and the protocol (ip6_nxt) meets the socket.
316 		 * XXX chase extension headers, or pass final nxt value
317 		 * from icmp6_notify_error()
318 		 */
319 		in6p = NULL;
320 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
321 		    (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0);
322 #if 0
323 		if (!in6p) {
324 			/*
325 			 * As the use of sendto(2) is fairly popular,
326 			 * we may want to allow non-connected pcb too.
327 			 * But it could be too weak against attacks...
328 			 * We should at least check if the local
329 			 * address (= s) is really ours.
330 			 */
331 			in6p = in6_pcblookup_bind(&raw6cbtable,
332 			    &sa6->sin6_addr, 0, 0))
333 		}
334 #endif
335 
336 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
337 		    in6p->in6p_ip6.ip6_nxt == nxt)
338 			valid++;
339 
340 		/*
341 		 * Depending on the value of "valid" and routing table
342 		 * size (mtudisc_{hi,lo}wat), we will:
343 		 * - recalculate the new MTU and create the
344 		 *   corresponding routing entry, or
345 		 * - ignore the MTU change notification.
346 		 */
347 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
348 
349 		/*
350 		 * regardless of if we called icmp6_mtudisc_update(),
351 		 * we need to call in6_pcbnotify(), to notify path
352 		 * MTU change to the userland (2292bis-02), because
353 		 * some unconnected sockets may share the same
354 		 * destination and want to know the path MTU.
355 		 */
356 	}
357 
358 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
359 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
360 }
361 
362 /*
363  * Generate IPv6 header and pass packet to ip6_output.
364  * Tack on options user may have setup with control call.
365  */
366 int
367 #if __STDC__
368 rip6_output(struct mbuf *m, ...)
369 #else
370 rip6_output(m, va_alist)
371 	struct mbuf *m;
372 	va_dcl
373 #endif
374 {
375 	struct socket *so;
376 	struct sockaddr_in6 *dstsock;
377 	struct mbuf *control;
378 	struct in6_addr *dst;
379 	struct ip6_hdr *ip6;
380 	struct in6pcb *in6p;
381 	u_int	plen = m->m_pkthdr.len;
382 	int error = 0;
383 	struct ip6_pktopts opt, *optp = NULL, *origoptp;
384 	struct ifnet *oifp = NULL;
385 	int type, code;		/* for ICMPv6 output statistics only */
386 	int priv = 0;
387 	va_list ap;
388 	int flags;
389 
390 	va_start(ap, m);
391 	so = va_arg(ap, struct socket *);
392 	dstsock = va_arg(ap, struct sockaddr_in6 *);
393 	control = va_arg(ap, struct mbuf *);
394 	va_end(ap);
395 
396 	in6p = sotoin6pcb(so);
397 
398 	priv = 0;
399     {
400 	struct proc *p = curproc;	/* XXX */
401 
402 	if (p && !suser(p->p_ucred, &p->p_acflag))
403 		priv = 1;
404     }
405 	dst = &dstsock->sin6_addr;
406 	if (control) {
407 		if ((error = ip6_setpktoptions(control, &opt, priv)) != 0)
408 			goto bad;
409 		optp = &opt;
410 	} else
411 		optp = in6p->in6p_outputopts;
412 
413 	/*
414 	 * For an ICMPv6 packet, we should know its type and code
415 	 * to update statistics.
416 	 */
417 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
418 		struct icmp6_hdr *icmp6;
419 		if (m->m_len < sizeof(struct icmp6_hdr) &&
420 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
421 			error = ENOBUFS;
422 			goto bad;
423 		}
424 		icmp6 = mtod(m, struct icmp6_hdr *);
425 		type = icmp6->icmp6_type;
426 		code = icmp6->icmp6_code;
427 	} else {
428 		type = 0;
429 		code = 0;
430 	}
431 
432 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
433 	if (!m) {
434 		error = ENOBUFS;
435 		goto bad;
436 	}
437 	ip6 = mtod(m, struct ip6_hdr *);
438 
439 	/*
440 	 * Next header might not be ICMP6 but use its pseudo header anyway.
441 	 */
442 	ip6->ip6_dst = *dst;
443 
444 	/* KAME hack: embed scopeid */
445 	origoptp = in6p->in6p_outputopts;
446 	in6p->in6p_outputopts = optp;
447 	if (in6_embedscope(&ip6->ip6_dst, dstsock, in6p, &oifp) != 0) {
448 		error = EINVAL;
449 		goto bad;
450 	}
451 	in6p->in6p_outputopts = origoptp;
452 
453 	/*
454 	 * Source address selection.
455 	 */
456 	{
457 		struct in6_addr *in6a;
458 
459 		if ((in6a = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
460 		    &in6p->in6p_route, &in6p->in6p_laddr, &error)) == 0) {
461 			if (error == 0)
462 				error = EADDRNOTAVAIL;
463 			goto bad;
464 		}
465 		ip6->ip6_src = *in6a;
466 		if (in6p->in6p_route.ro_rt) {
467 			/* what if oifp contradicts ? */
468 			oifp = ifindex2ifnet[in6p->in6p_route.ro_rt->rt_ifp->if_index];
469 		}
470 	}
471 
472 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
473 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
474 	ip6->ip6_vfc  |= IPV6_VERSION;
475 #if 0				/* ip6_plen will be filled in ip6_output. */
476 	ip6->ip6_plen  = htons((u_int16_t)plen);
477 #endif
478 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
479 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
480 
481 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
482 	    in6p->in6p_cksum != -1) {
483 		int off;
484 		u_int16_t sum;
485 
486 		/* compute checksum */
487 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
488 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
489 		else
490 			off = in6p->in6p_cksum;
491 		if (plen < off + 1) {
492 			error = EINVAL;
493 			goto bad;
494 		}
495 		off += sizeof(struct ip6_hdr);
496 
497 		sum = 0;
498 		m = m_copyback_cow(m, off, sizeof(sum), (caddr_t)&sum,
499 		    M_DONTWAIT);
500 		if (m == NULL) {
501 			error = ENOBUFS;
502 			goto bad;
503 		}
504 		sum = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
505 		m = m_copyback_cow(m, off, sizeof(sum), (caddr_t)&sum,
506 		    M_DONTWAIT);
507 		if (m == NULL) {
508 			error = ENOBUFS;
509 			goto bad;
510 		}
511 	}
512 
513 	flags = 0;
514 	if (in6p->in6p_flags & IN6P_MINMTU)
515 		flags |= IPV6_MINMTU;
516 
517 	error = ip6_output(m, optp, &in6p->in6p_route, flags,
518 	    in6p->in6p_moptions, so, &oifp);
519 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
520 		if (oifp)
521 			icmp6_ifoutstat_inc(oifp, type, code);
522 		icmp6stat.icp6s_outhist[type]++;
523 	} else
524 		rip6stat.rip6s_opackets++;
525 
526 	goto freectl;
527 
528  bad:
529 	if (m)
530 		m_freem(m);
531 
532  freectl:
533 	if (optp == &opt && optp->ip6po_rthdr && optp->ip6po_route.ro_rt)
534 		RTFREE(optp->ip6po_route.ro_rt);
535 	if (control)
536 		m_freem(control);
537 	return (error);
538 }
539 
540 /*
541  * Raw IPv6 socket option processing.
542  */
543 int
544 rip6_ctloutput(op, so, level, optname, mp)
545 	int op;
546 	struct socket *so;
547 	int level, optname;
548 	struct mbuf **mp;
549 {
550 	int error = 0;
551 
552 	switch (level) {
553 	case IPPROTO_IPV6:
554 		switch (optname) {
555 		case MRT6_INIT:
556 		case MRT6_DONE:
557 		case MRT6_ADD_MIF:
558 		case MRT6_DEL_MIF:
559 		case MRT6_ADD_MFC:
560 		case MRT6_DEL_MFC:
561 		case MRT6_PIM:
562 			if (op == PRCO_SETOPT) {
563 				error = ip6_mrouter_set(optname, so, *mp);
564 				if (*mp)
565 					(void)m_free(*mp);
566 			} else if (op == PRCO_GETOPT)
567 				error = ip6_mrouter_get(optname, so, mp);
568 			else
569 				error = EINVAL;
570 			return (error);
571 		case IPV6_CHECKSUM:
572 			return (ip6_raw_ctloutput(op, so, level, optname, mp));
573 		default:
574 			return (ip6_ctloutput(op, so, level, optname, mp));
575 		}
576 
577 	case IPPROTO_ICMPV6:
578 		/*
579 		 * XXX: is it better to call icmp6_ctloutput() directly
580 		 * from protosw?
581 		 */
582 		return (icmp6_ctloutput(op, so, level, optname, mp));
583 
584 	default:
585 		if (op == PRCO_SETOPT && *mp)
586 			m_free(*mp);
587 		return EINVAL;
588 	}
589 }
590 
591 extern	u_long rip6_sendspace;
592 extern	u_long rip6_recvspace;
593 
594 int
595 rip6_usrreq(so, req, m, nam, control, p)
596 	struct socket *so;
597 	int req;
598 	struct mbuf *m, *nam, *control;
599 	struct proc *p;
600 {
601 	struct in6pcb *in6p = sotoin6pcb(so);
602 	int s;
603 	int error = 0;
604 	int priv;
605 
606 	priv = 0;
607 	if (p && !suser(p->p_ucred, &p->p_acflag))
608 		priv++;
609 
610 	if (req == PRU_CONTROL)
611 		return (in6_control(so, (u_long)m, (caddr_t)nam,
612 		    (struct ifnet *)control, p));
613 
614 	if (req == PRU_PURGEIF) {
615 		in6_pcbpurgeif0(&raw6cbtable, (struct ifnet *)control);
616 		in6_purgeif((struct ifnet *)control);
617 		in6_pcbpurgeif(&raw6cbtable, (struct ifnet *)control);
618 		return (0);
619 	}
620 
621 	switch (req) {
622 	case PRU_ATTACH:
623 		if (in6p)
624 			panic("rip6_attach");
625 		if (!priv) {
626 			error = EACCES;
627 			break;
628 		}
629 		s = splsoftnet();
630 		if ((error = soreserve(so, rip6_sendspace, rip6_recvspace)) != 0) {
631 			splx(s);
632 			break;
633 		}
634 		if ((error = in6_pcballoc(so, &raw6cbtable)) != 0)
635 		{
636 			splx(s);
637 			break;
638 		}
639 		splx(s);
640 		in6p = sotoin6pcb(so);
641 		in6p->in6p_ip6.ip6_nxt = (long)nam;
642 		in6p->in6p_cksum = -1;
643 
644 		MALLOC(in6p->in6p_icmp6filt, struct icmp6_filter *,
645 		    sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
646 		if (in6p->in6p_icmp6filt == NULL) {
647 			in6_pcbdetach(in6p);
648 			error = ENOMEM;
649 			break;
650 		}
651 		ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
652 		break;
653 
654 	case PRU_DISCONNECT:
655 		if ((so->so_state & SS_ISCONNECTED) == 0) {
656 			error = ENOTCONN;
657 			break;
658 		}
659 		in6p->in6p_faddr = in6addr_any;
660 		so->so_state &= ~SS_ISCONNECTED;	/* XXX */
661 		break;
662 
663 	case PRU_ABORT:
664 		soisdisconnected(so);
665 		/* Fallthrough */
666 	case PRU_DETACH:
667 		if (in6p == 0)
668 			panic("rip6_detach");
669 		if (so == ip6_mrouter)
670 			ip6_mrouter_done();
671 		/* xxx: RSVP */
672 		if (in6p->in6p_icmp6filt) {
673 			FREE(in6p->in6p_icmp6filt, M_PCB);
674 			in6p->in6p_icmp6filt = NULL;
675 		}
676 		in6_pcbdetach(in6p);
677 		break;
678 
679 	case PRU_BIND:
680 	    {
681 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
682 		struct ifaddr *ia = NULL;
683 
684 		if (nam->m_len != sizeof(*addr)) {
685 			error = EINVAL;
686 			break;
687 		}
688 		if ((ifnet.tqh_first == 0) || (addr->sin6_family != AF_INET6)) {
689 			error = EADDRNOTAVAIL;
690 			break;
691 		}
692 #ifdef ENABLE_DEFAULT_SCOPE
693 		if (addr->sin6_scope_id == 0)	/* not change if specified  */
694 			addr->sin6_scope_id =
695 			    scope6_addr2default(&addr->sin6_addr);
696 #endif
697 		/* KAME hack: embed scopeid */
698 		if (in6_embedscope(&addr->sin6_addr, addr, in6p, NULL) != 0)
699 			return EINVAL;
700 #ifndef SCOPEDROUTING
701 		addr->sin6_scope_id = 0; /* for ifa_ifwithaddr */
702 #endif
703 
704 		/*
705 		 * we don't support mapped address here, it would confuse
706 		 * users so reject it
707 		 */
708 		if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr)) {
709 			error = EADDRNOTAVAIL;
710 			break;
711 		}
712 		if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
713 		    (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0) {
714 			error = EADDRNOTAVAIL;
715 			break;
716 		}
717 		if (ia && ((struct in6_ifaddr *)ia)->ia6_flags &
718 		    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
719 		     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
720 			error = EADDRNOTAVAIL;
721 			break;
722 		}
723 		in6p->in6p_laddr = addr->sin6_addr;
724 		break;
725 	    }
726 
727 	case PRU_CONNECT:
728 	{
729 		struct sockaddr_in6 *addr = mtod(nam, struct sockaddr_in6 *);
730 		struct in6_addr *in6a = NULL;
731 #ifdef ENABLE_DEFAULT_SCOPE
732 		struct sockaddr_in6 sin6;
733 #endif
734 
735 		if (nam->m_len != sizeof(*addr)) {
736 			error = EINVAL;
737 			break;
738 		}
739 		if (ifnet.tqh_first == 0)
740 		{
741 			error = EADDRNOTAVAIL;
742 			break;
743 		}
744 		if (addr->sin6_family != AF_INET6) {
745 			error = EAFNOSUPPORT;
746 			break;
747 		}
748 
749 #ifdef ENABLE_DEFAULT_SCOPE
750 		if (addr->sin6_scope_id == 0) {
751 			/* protect *addr */
752 			sin6 = *addr;
753 			addr = &sin6;
754 			addr->sin6_scope_id =
755 			    scope6_addr2default(&addr->sin6_addr);
756 		}
757 #endif
758 
759 		/* Source address selection. XXX: need pcblookup? */
760 		in6a = in6_selectsrc(addr, in6p->in6p_outputopts,
761 		    in6p->in6p_moptions, &in6p->in6p_route,
762 		    &in6p->in6p_laddr, &error);
763 		if (in6a == NULL) {
764 			if (error == 0)
765 				error = EADDRNOTAVAIL;
766 			break;
767 		}
768 		in6p->in6p_laddr = *in6a;
769 		in6p->in6p_faddr = addr->sin6_addr;
770 		soisconnected(so);
771 		break;
772 	}
773 
774 	case PRU_CONNECT2:
775 		error = EOPNOTSUPP;
776 		break;
777 
778 	/*
779 	 * Mark the connection as being incapable of futther input.
780 	 */
781 	case PRU_SHUTDOWN:
782 		socantsendmore(so);
783 		break;
784 	/*
785 	 * Ship a packet out. The appropriate raw output
786 	 * routine handles any messaging necessary.
787 	 */
788 	case PRU_SEND:
789 	{
790 		struct sockaddr_in6 tmp;
791 		struct sockaddr_in6 *dst;
792 
793 		/* always copy sockaddr to avoid overwrites */
794 		if (so->so_state & SS_ISCONNECTED) {
795 			if (nam) {
796 				error = EISCONN;
797 				break;
798 			}
799 			/* XXX */
800 			bzero(&tmp, sizeof(tmp));
801 			tmp.sin6_family = AF_INET6;
802 			tmp.sin6_len = sizeof(struct sockaddr_in6);
803 			bcopy(&in6p->in6p_faddr, &tmp.sin6_addr,
804 			    sizeof(struct in6_addr));
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 #ifdef ENABLE_DEFAULT_SCOPE
825 		if (dst->sin6_scope_id == 0) {
826 			dst->sin6_scope_id =
827 			    scope6_addr2default(&dst->sin6_addr);
828 		}
829 #endif
830 		error = rip6_output(m, so, dst, control);
831 		m = NULL;
832 		break;
833 	}
834 
835 	case PRU_SENSE:
836 		/*
837 		 * stat: don't bother with a blocksize
838 		 */
839 		return (0);
840 	/*
841 	 * Not supported.
842 	 */
843 	case PRU_RCVOOB:
844 	case PRU_RCVD:
845 	case PRU_LISTEN:
846 	case PRU_ACCEPT:
847 	case PRU_SENDOOB:
848 		error = EOPNOTSUPP;
849 		break;
850 
851 	case PRU_SOCKADDR:
852 		in6_setsockaddr(in6p, nam);
853 		break;
854 
855 	case PRU_PEERADDR:
856 		in6_setpeeraddr(in6p, nam);
857 		break;
858 
859 	default:
860 		panic("rip6_usrreq");
861 	}
862 	if (m != NULL)
863 		m_freem(m);
864 	return (error);
865 }
866 
867 SYSCTL_SETUP(sysctl_net_inet6_raw6_setup, "sysctl net.inet6.raw6 subtree setup")
868 {
869 
870 	sysctl_createv(clog, 0, NULL, NULL,
871 		       CTLFLAG_PERMANENT,
872 		       CTLTYPE_NODE, "net", NULL,
873 		       NULL, 0, NULL, 0,
874 		       CTL_NET, CTL_EOL);
875 	sysctl_createv(clog, 0, NULL, NULL,
876 		       CTLFLAG_PERMANENT,
877 		       CTLTYPE_NODE, "inet6", NULL,
878 		       NULL, 0, NULL, 0,
879 		       CTL_NET, PF_INET6, CTL_EOL);
880 	sysctl_createv(clog, 0, NULL, NULL,
881 		       CTLFLAG_PERMANENT,
882 		       CTLTYPE_NODE, "raw6",
883 		       SYSCTL_DESCR("Raw IPv6 settings"),
884 		       NULL, 0, NULL, 0,
885 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
886 
887 	sysctl_createv(clog, 0, NULL, NULL,
888 		       CTLFLAG_PERMANENT,
889 		       CTLTYPE_STRUCT, "pcblist",
890 		       SYSCTL_DESCR("Raw IPv6 control block list"),
891 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
892 		       CTL_NET, PF_INET6, IPPROTO_RAW,
893 		       CTL_CREATE, CTL_EOL);
894 	sysctl_createv(clog, 0, NULL, NULL,
895 		       CTLFLAG_PERMANENT,
896 		       CTLTYPE_STRUCT, "stats",
897 		       SYSCTL_DESCR("Raw IPv6 statistics"),
898 		       NULL, 0, &rip6stat, sizeof(rip6stat),
899 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
900 		       CTL_EOL);
901 }
902