xref: /netbsd-src/sys/netinet6/raw_ip6.c (revision 53b02e147d4ed531c0d2a5ca9b3e8026ba3e99b5)
1 /*	$NetBSD: raw_ip6.c,v 1.176 2021/09/21 15:08:44 christos 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 1.176 2021/09/21 15:08:44 christos Exp $");
66 
67 #ifdef _KERNEL_OPT
68 #include "opt_ipsec.h"
69 #include "opt_net_mpsafe.h"
70 #endif
71 
72 #include <sys/param.h>
73 #include <sys/sysctl.h>
74 #include <sys/mbuf.h>
75 #include <sys/socket.h>
76 #include <sys/protosw.h>
77 #include <sys/socketvar.h>
78 #include <sys/systm.h>
79 #include <sys/proc.h>
80 #include <sys/kauth.h>
81 #include <sys/kmem.h>
82 
83 #include <net/if.h>
84 #include <net/if_types.h>
85 #include <net/net_stats.h>
86 
87 #include <netinet/in.h>
88 #include <netinet/in_var.h>
89 #include <netinet/ip6.h>
90 #include <netinet6/ip6_var.h>
91 #include <netinet6/ip6_private.h>
92 #include <netinet6/ip6_mroute.h>
93 #include <netinet/icmp6.h>
94 #include <netinet6/icmp6_private.h>
95 #include <netinet6/in6_pcb.h>
96 #include <netinet6/ip6protosw.h>
97 #include <netinet6/scope6_var.h>
98 #include <netinet6/raw_ip6.h>
99 
100 #ifdef IPSEC
101 #include <netipsec/ipsec.h>
102 #include <netipsec/ipsec6.h>
103 #endif
104 
105 #include "faith.h"
106 #if defined(NFAITH) && 0 < NFAITH
107 #include <net/if_faith.h>
108 #endif
109 
110 extern struct inpcbtable rawcbtable;
111 struct	inpcbtable raw6cbtable;
112 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
113 
114 /*
115  * Raw interface to IP6 protocol.
116  */
117 
118 static percpu_t *rip6stat_percpu;
119 
120 #define	RIP6_STATINC(x)		_NET_STATINC(rip6stat_percpu, x)
121 
122 static void sysctl_net_inet6_raw6_setup(struct sysctllog **);
123 
124 /*
125  * Initialize raw connection block queue.
126  */
127 void
128 rip6_init(void)
129 {
130 
131 	sysctl_net_inet6_raw6_setup(NULL);
132 	in6_pcbinit(&raw6cbtable, 1, 1);
133 
134 	rip6stat_percpu = percpu_alloc(sizeof(uint64_t) * RIP6_NSTATS);
135 }
136 
137 static void
138 rip6_sbappendaddr(struct in6pcb *last, struct ip6_hdr *ip6,
139     const struct sockaddr *sa, int hlen, struct mbuf *n)
140 {
141 	struct mbuf *opts = NULL;
142 
143 	if (last->in6p_flags & IN6P_CONTROLOPTS)
144 		ip6_savecontrol(last, &opts, ip6, n);
145 
146 	m_adj(n, hlen);
147 
148 	if (sbappendaddr(&last->in6p_socket->so_rcv, sa, n, opts) == 0) {
149 		soroverflow(last->in6p_socket);
150 		m_freem(n);
151 		if (opts)
152 			m_freem(opts);
153 		RIP6_STATINC(RIP6_STAT_FULLSOCK);
154 	} else {
155 		sorwakeup(last->in6p_socket);
156 	}
157 }
158 
159 /*
160  * Setup generic address and protocol structures
161  * for raw_input routine, then pass them along with
162  * mbuf chain.
163  */
164 int
165 rip6_input(struct mbuf **mp, int *offp, int proto)
166 {
167 	struct mbuf *m = *mp;
168 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
169 	struct inpcb_hdr *inph;
170 	struct in6pcb *in6p;
171 	struct in6pcb *last = NULL;
172 	struct sockaddr_in6 rip6src;
173 	struct mbuf *n;
174 
175 	RIP6_STATINC(RIP6_STAT_IPACKETS);
176 
177 #if defined(NFAITH) && 0 < NFAITH
178 	if (faithprefix(&ip6->ip6_dst)) {
179 		/* send icmp6 host unreach? */
180 		m_freem(m);
181 		return IPPROTO_DONE;
182 	}
183 #endif
184 
185 	sockaddr_in6_init(&rip6src, &ip6->ip6_src, 0, 0, 0);
186 	if (sa6_recoverscope(&rip6src) != 0) {
187 		/* XXX: should be impossible. */
188 		m_freem(m);
189 		return IPPROTO_DONE;
190 	}
191 
192 	TAILQ_FOREACH(inph, &raw6cbtable.inpt_queue, inph_queue) {
193 		in6p = (struct in6pcb *)inph;
194 		if (in6p->in6p_af != AF_INET6)
195 			continue;
196 		if (in6p->in6p_ip6.ip6_nxt &&
197 		    in6p->in6p_ip6.ip6_nxt != proto)
198 			continue;
199 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
200 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
201 			continue;
202 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
203 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
204 			continue;
205 		if (in6p->in6p_cksum != -1) {
206 			RIP6_STATINC(RIP6_STAT_ISUM);
207 			if (in6_cksum(m, proto, *offp,
208 			    m->m_pkthdr.len - *offp)) {
209 				RIP6_STATINC(RIP6_STAT_BADSUM);
210 				continue;
211 			}
212 		}
213 
214 		if (last == NULL) {
215 			;
216 		}
217 #ifdef IPSEC
218 		else if (ipsec_used && ipsec_in_reject(m, last)) {
219 			/* do not inject data into pcb */
220 		}
221 #endif
222 		else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) {
223 			rip6_sbappendaddr(last, ip6, sin6tosa(&rip6src),
224 			    *offp, n);
225 		}
226 
227 		last = in6p;
228 	}
229 
230 #ifdef IPSEC
231 	if (ipsec_used && last && ipsec_in_reject(m, last)) {
232 		m_freem(m);
233 		IP6_STATDEC(IP6_STAT_DELIVERED);
234 		/* do not inject data into pcb */
235 	} else
236 #endif
237 	if (last != NULL) {
238 		rip6_sbappendaddr(last, ip6, sin6tosa(&rip6src), *offp, m);
239 	} else {
240 		RIP6_STATINC(RIP6_STAT_NOSOCK);
241 		if (m->m_flags & M_MCAST)
242 			RIP6_STATINC(RIP6_STAT_NOSOCKMCAST);
243 		if (proto == IPPROTO_NONE)
244 			m_freem(m);
245 		else {
246 			int s;
247 			struct ifnet *rcvif = m_get_rcvif(m, &s);
248 			const int prvnxt = ip6_get_prevhdr(m, *offp);
249 			in6_ifstat_inc(rcvif, ifs6_in_protounknown);
250 			m_put_rcvif(rcvif, &s);
251 			icmp6_error(m, ICMP6_PARAM_PROB,
252 			    ICMP6_PARAMPROB_NEXTHEADER,
253 			    prvnxt);
254 		}
255 		IP6_STATDEC(IP6_STAT_DELIVERED);
256 	}
257 	return IPPROTO_DONE;
258 }
259 
260 void *
261 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
262 {
263 	struct ip6_hdr *ip6;
264 	struct ip6ctlparam *ip6cp = NULL;
265 	const struct sockaddr_in6 *sa6_src = NULL;
266 	void *cmdarg;
267 	void (*notify)(struct in6pcb *, int) = in6_rtchange;
268 	int nxt;
269 
270 	if (sa->sa_family != AF_INET6 ||
271 	    sa->sa_len != sizeof(struct sockaddr_in6))
272 		return NULL;
273 
274 	if ((unsigned)cmd >= PRC_NCMDS)
275 		return NULL;
276 	if (PRC_IS_REDIRECT(cmd))
277 		notify = in6_rtchange, d = NULL;
278 	else if (cmd == PRC_HOSTDEAD)
279 		d = NULL;
280 	else if (cmd == PRC_MSGSIZE)
281 		; /* special code is present, see below */
282 	else if (inet6ctlerrmap[cmd] == 0)
283 		return NULL;
284 
285 	/* if the parameter is from icmp6, decode it. */
286 	if (d != NULL) {
287 		ip6cp = (struct ip6ctlparam *)d;
288 		ip6 = ip6cp->ip6c_ip6;
289 		cmdarg = ip6cp->ip6c_cmdarg;
290 		sa6_src = ip6cp->ip6c_src;
291 		nxt = ip6cp->ip6c_nxt;
292 	} else {
293 		ip6 = NULL;
294 		cmdarg = NULL;
295 		sa6_src = &sa6_any;
296 		nxt = -1;
297 	}
298 
299 	if (ip6 && cmd == PRC_MSGSIZE) {
300 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
301 		int valid = 0;
302 		struct in6pcb *in6p;
303 
304 		/*
305 		 * Check to see if we have a valid raw IPv6 socket
306 		 * corresponding to the address in the ICMPv6 message
307 		 * payload, and the protocol (ip6_nxt) meets the socket.
308 		 * XXX chase extension headers, or pass final nxt value
309 		 * from icmp6_notify_error()
310 		 */
311 		in6p = NULL;
312 		in6p = in6_pcblookup_connect(&raw6cbtable, &sa6->sin6_addr, 0,
313 					     (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0, 0);
314 #if 0
315 		if (!in6p) {
316 			/*
317 			 * As the use of sendto(2) is fairly popular,
318 			 * we may want to allow non-connected pcb too.
319 			 * But it could be too weak against attacks...
320 			 * We should at least check if the local
321 			 * address (= s) is really ours.
322 			 */
323 			in6p = in6_pcblookup_bind(&raw6cbtable,
324 			    &sa6->sin6_addr, 0, 0);
325 		}
326 #endif
327 
328 		if (in6p && in6p->in6p_ip6.ip6_nxt &&
329 		    in6p->in6p_ip6.ip6_nxt == nxt)
330 			valid++;
331 
332 		/*
333 		 * Depending on the value of "valid" and routing table
334 		 * size (mtudisc_{hi,lo}wat), we will:
335 		 * - recalculate the new MTU and create the
336 		 *   corresponding routing entry, or
337 		 * - ignore the MTU change notification.
338 		 */
339 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
340 
341 		/*
342 		 * regardless of if we called icmp6_mtudisc_update(),
343 		 * we need to call in6_pcbnotify(), to notify path MTU
344 		 * change to the userland (RFC3542), because some
345 		 * unconnected sockets may share the same destination
346 		 * and want to know the path MTU.
347 		 */
348 	}
349 
350 	(void) in6_pcbnotify(&raw6cbtable, sa, 0,
351 	    sin6tocsa(sa6_src), 0, cmd, cmdarg, notify);
352 	return NULL;
353 }
354 
355 /*
356  * Generate IPv6 header and pass packet to ip6_output.
357  * Tack on options user may have setup with control call.
358  */
359 int
360 rip6_output(struct mbuf *m, struct socket * const so,
361     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
362 {
363 	struct in6_addr *dst;
364 	struct ip6_hdr *ip6;
365 	struct in6pcb *in6p;
366 	u_int	plen = m->m_pkthdr.len;
367 	int error = 0;
368 	struct ip6_pktopts opt, *optp = NULL;
369 	struct ifnet *oifp = NULL;
370 	int type, code;		/* for ICMPv6 output statistics only */
371 	int scope_ambiguous = 0;
372 	int bound = curlwp_bind();
373 	struct psref psref;
374 
375 	in6p = sotoin6pcb(so);
376 
377 	dst = &dstsock->sin6_addr;
378 	if (control) {
379 		if ((error = ip6_setpktopts(control, &opt,
380 		    in6p->in6p_outputopts,
381 		    kauth_cred_get(), so->so_proto->pr_protocol)) != 0) {
382 			goto bad;
383 		}
384 		optp = &opt;
385 	} else
386 		optp = in6p->in6p_outputopts;
387 
388 	/*
389 	 * Check and convert scope zone ID into internal form.
390 	 * XXX: we may still need to determine the zone later.
391 	 */
392 	if (!(so->so_state & SS_ISCONNECTED)) {
393 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
394 			scope_ambiguous = 1;
395 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
396 			goto bad;
397 	}
398 
399 	/*
400 	 * For an ICMPv6 packet, we should know its type and code
401 	 * to update statistics.
402 	 */
403 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
404 		struct icmp6_hdr *icmp6;
405 		if (m->m_len < sizeof(struct icmp6_hdr) &&
406 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
407 			error = ENOBUFS;
408 			goto bad;
409 		}
410 		icmp6 = mtod(m, struct icmp6_hdr *);
411 		type = icmp6->icmp6_type;
412 		code = icmp6->icmp6_code;
413 	} else {
414 		type = 0;
415 		code = 0;
416 	}
417 
418 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
419 	if (!m) {
420 		error = ENOBUFS;
421 		goto bad;
422 	}
423 	ip6 = mtod(m, struct ip6_hdr *);
424 
425 	/*
426 	 * Next header might not be ICMP6 but use its pseudo header anyway.
427 	 */
428 	ip6->ip6_dst = *dst;
429 
430 	/*
431 	 * Source address selection.
432 	 */
433 	error = in6_selectsrc(dstsock, optp, in6p->in6p_moptions,
434 	    &in6p->in6p_route, &in6p->in6p_laddr, &oifp, &psref, &ip6->ip6_src);
435 	if (error != 0)
436 		goto bad;
437 
438 	if (oifp && scope_ambiguous) {
439 		/*
440 		 * Application should provide a proper zone ID or the use of
441 		 * default zone IDs should be enabled.  Unfortunately, some
442 		 * applications do not behave as it should, so we need a
443 		 * workaround.  Even if an appropriate ID is not determined
444 		 * (when it's required), if we can determine the outgoing
445 		 * interface. determine the zone ID based on the interface.
446 		 */
447 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
448 		if (error != 0)
449 			goto bad;
450 	}
451 	ip6->ip6_dst = dstsock->sin6_addr;
452 
453 	/* fill in the rest of the IPv6 header fields */
454 	ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
455 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
456 	ip6->ip6_vfc  |= IPV6_VERSION;
457 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
458 	ip6->ip6_nxt   = in6p->in6p_ip6.ip6_nxt;
459 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
460 
461 	if_put(oifp, &psref);
462 	oifp = NULL;
463 
464 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
465 	    in6p->in6p_cksum != -1) {
466 		const uint8_t nxt = ip6->ip6_nxt;
467 		int off;
468 		u_int16_t sum;
469 
470 		/* compute checksum */
471 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
472 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
473 		else
474 			off = in6p->in6p_cksum;
475 		if (plen < off + 1) {
476 			error = EINVAL;
477 			goto bad;
478 		}
479 		off += sizeof(struct ip6_hdr);
480 
481 		sum = 0;
482 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
483 		    M_DONTWAIT);
484 		if (m == NULL) {
485 			error = ENOBUFS;
486 			goto bad;
487 		}
488 		sum = in6_cksum(m, nxt, sizeof(*ip6), plen);
489 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
490 		    M_DONTWAIT);
491 		if (m == NULL) {
492 			error = ENOBUFS;
493 			goto bad;
494 		}
495 	}
496 
497 	{
498 		struct ifnet *ret_oifp = NULL;
499 
500 		error = ip6_output(m, optp, &in6p->in6p_route, 0,
501 		    in6p->in6p_moptions, in6p, &ret_oifp);
502 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
503 			if (ret_oifp)
504 				icmp6_ifoutstat_inc(ret_oifp, type, code);
505 			ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
506 		} else
507 			RIP6_STATINC(RIP6_STAT_OPACKETS);
508 	}
509 
510 	goto freectl;
511 
512  bad:
513 	if (m)
514 		m_freem(m);
515 
516  freectl:
517 	if (control) {
518 		ip6_clearpktopts(&opt, -1);
519 		m_freem(control);
520 	}
521 	if_put(oifp, &psref);
522 	curlwp_bindx(bound);
523 	return error;
524 }
525 
526 /*
527  * Raw IPv6 socket option processing.
528  */
529 int
530 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
531 {
532 	int error = 0;
533 
534 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
535 		int optval;
536 
537 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
538 		if (op == PRCO_GETOPT) {
539 			optval = 1;
540 			error = sockopt_set(sopt, &optval, sizeof(optval));
541 		} else if (op == PRCO_SETOPT) {
542 			error = sockopt_getint(sopt, &optval);
543 			if (error)
544 				goto out;
545 			if (optval == 0)
546 				error = EINVAL;
547 		}
548 
549 		goto out;
550 	} else if (sopt->sopt_level != IPPROTO_IPV6)
551 		return ip6_ctloutput(op, so, sopt);
552 
553 	switch (sopt->sopt_name) {
554 	case MRT6_INIT:
555 	case MRT6_DONE:
556 	case MRT6_ADD_MIF:
557 	case MRT6_DEL_MIF:
558 	case MRT6_ADD_MFC:
559 	case MRT6_DEL_MFC:
560 	case MRT6_PIM:
561 		if (op == PRCO_SETOPT)
562 			error = ip6_mrouter_set(so, sopt);
563 		else if (op == PRCO_GETOPT)
564 			error = ip6_mrouter_get(so, sopt);
565 		else
566 			error = EINVAL;
567 		break;
568 	case IPV6_CHECKSUM:
569 		return ip6_raw_ctloutput(op, so, sopt);
570 	default:
571 		return ip6_ctloutput(op, so, sopt);
572 	}
573  out:
574 	return error;
575 }
576 
577 extern	u_long rip6_sendspace;
578 extern	u_long rip6_recvspace;
579 
580 int
581 rip6_attach(struct socket *so, int proto)
582 {
583 	struct in6pcb *in6p;
584 	int s, error;
585 
586 	KASSERT(sotoin6pcb(so) == NULL);
587 	sosetlock(so);
588 
589 	error = kauth_authorize_network(kauth_cred_get(),
590 	    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
591 	    KAUTH_ARG(AF_INET6),
592 	    KAUTH_ARG(SOCK_RAW),
593 	    KAUTH_ARG(so->so_proto->pr_protocol));
594 	if (error) {
595 		return error;
596 	}
597 	s = splsoftnet();
598 	error = soreserve(so, rip6_sendspace, rip6_recvspace);
599 	if (error) {
600 		splx(s);
601 		return error;
602 	}
603 	if ((error = in6_pcballoc(so, &raw6cbtable)) != 0) {
604 		splx(s);
605 		return error;
606 	}
607 	splx(s);
608 	in6p = sotoin6pcb(so);
609 	in6p->in6p_ip6.ip6_nxt = proto;
610 	in6p->in6p_cksum = -1;
611 
612 	in6p->in6p_icmp6filt = kmem_alloc(sizeof(struct icmp6_filter), KM_SLEEP);
613 	ICMP6_FILTER_SETPASSALL(in6p->in6p_icmp6filt);
614 	KASSERT(solocked(so));
615 	return error;
616 }
617 
618 static void
619 rip6_detach(struct socket *so)
620 {
621 	struct in6pcb *in6p = sotoin6pcb(so);
622 
623 	KASSERT(solocked(so));
624 	KASSERT(in6p != NULL);
625 
626 	if (so == ip6_mrouter) {
627 		ip6_mrouter_done();
628 	}
629 	/* xxx: RSVP */
630 	if (in6p->in6p_icmp6filt != NULL) {
631 		kmem_free(in6p->in6p_icmp6filt, sizeof(struct icmp6_filter));
632 		in6p->in6p_icmp6filt = NULL;
633 	}
634 	in6_pcbdetach(in6p);
635 }
636 
637 static int
638 rip6_accept(struct socket *so, struct sockaddr *nam)
639 {
640 	KASSERT(solocked(so));
641 
642 	return EOPNOTSUPP;
643 }
644 
645 static int
646 rip6_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
647 {
648 	struct in6pcb *in6p = sotoin6pcb(so);
649 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
650 	struct ifaddr *ifa = NULL;
651 	int error = 0;
652 	int s;
653 
654 	KASSERT(solocked(so));
655 	KASSERT(in6p != NULL);
656 	KASSERT(nam != NULL);
657 
658 	if (addr->sin6_len != sizeof(*addr))
659 		return EINVAL;
660 	if (IFNET_READER_EMPTY() || addr->sin6_family != AF_INET6)
661 		return EADDRNOTAVAIL;
662 
663 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
664 		return error;
665 
666 	/*
667 	 * we don't support mapped address here, it would confuse
668 	 * users so reject it
669 	 */
670 	if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr))
671 		return EADDRNOTAVAIL;
672 	s = pserialize_read_enter();
673 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
674 	    (ifa = ifa_ifwithaddr(sin6tosa(addr))) == NULL) {
675 		error = EADDRNOTAVAIL;
676 		goto out;
677 	}
678 	if (ifa && (ifatoia6(ifa))->ia6_flags &
679 	    (IN6_IFF_ANYCAST | IN6_IFF_DUPLICATED)) {
680 		error = EADDRNOTAVAIL;
681 		goto out;
682 	}
683 
684 	in6p->in6p_laddr = addr->sin6_addr;
685 	error = 0;
686 out:
687 	pserialize_read_exit(s);
688 	return error;
689 }
690 
691 static int
692 rip6_listen(struct socket *so, struct lwp *l)
693 {
694 	KASSERT(solocked(so));
695 
696 	return EOPNOTSUPP;
697 }
698 
699 static int
700 rip6_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
701 {
702 	struct in6pcb *in6p = sotoin6pcb(so);
703 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
704 	struct in6_addr in6a;
705 	struct ifnet *ifp = NULL;
706 	int scope_ambiguous = 0;
707 	int error = 0;
708 	struct psref psref;
709 	int bound;
710 
711 	KASSERT(solocked(so));
712 	KASSERT(in6p != NULL);
713 	KASSERT(nam != NULL);
714 
715 	if (IFNET_READER_EMPTY())
716 		return EADDRNOTAVAIL;
717 	if (addr->sin6_family != AF_INET6)
718 		return EAFNOSUPPORT;
719 	if (addr->sin6_len != sizeof(*addr))
720 		return EINVAL;
721 
722 	/*
723 	 * Application should provide a proper zone ID or the use of
724 	 * default zone IDs should be enabled.  Unfortunately, some
725 	 * applications do not behave as it should, so we need a
726 	 * workaround.  Even if an appropriate ID is not determined,
727 	 * we'll see if we can determine the outgoing interface.  If we
728 	 * can, determine the zone ID based on the interface below.
729 	 */
730 	if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
731 		scope_ambiguous = 1;
732 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
733 		return error;
734 
735 	bound = curlwp_bind();
736 	/* Source address selection. XXX: need pcblookup? */
737 	error = in6_selectsrc(addr, in6p->in6p_outputopts,
738 	    in6p->in6p_moptions, &in6p->in6p_route,
739 	    &in6p->in6p_laddr, &ifp, &psref, &in6a);
740 	if (error != 0)
741 		goto out;
742 	/* XXX: see above */
743 	if (ifp && scope_ambiguous &&
744 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
745 		goto out;
746 	}
747 	in6p->in6p_laddr = in6a;
748 	in6p->in6p_faddr = addr->sin6_addr;
749 	soisconnected(so);
750 out:
751 	if_put(ifp, &psref);
752 	curlwp_bindx(bound);
753 	return error;
754 }
755 
756 static int
757 rip6_connect2(struct socket *so, struct socket *so2)
758 {
759 	KASSERT(solocked(so));
760 
761 	return EOPNOTSUPP;
762 }
763 
764 static int
765 rip6_disconnect(struct socket *so)
766 {
767 	struct in6pcb *in6p = sotoin6pcb(so);
768 
769 	KASSERT(solocked(so));
770 	KASSERT(in6p != NULL);
771 
772 	if ((so->so_state & SS_ISCONNECTED) == 0)
773 		return ENOTCONN;
774 
775 	in6p->in6p_faddr = in6addr_any;
776 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
777 	return 0;
778 }
779 
780 static int
781 rip6_shutdown(struct socket *so)
782 {
783 	KASSERT(solocked(so));
784 
785 	/*
786 	 * Mark the connection as being incapable of futther input.
787 	 */
788 	socantsendmore(so);
789 	return 0;
790 }
791 
792 static int
793 rip6_abort(struct socket *so)
794 {
795 	KASSERT(solocked(so));
796 
797 	soisdisconnected(so);
798 	rip6_detach(so);
799 	return 0;
800 }
801 
802 static int
803 rip6_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
804 {
805 	return in6_control(so, cmd, nam, ifp);
806 }
807 
808 static int
809 rip6_stat(struct socket *so, struct stat *ub)
810 {
811 	KASSERT(solocked(so));
812 
813 	/* stat: don't bother with a blocksize */
814 	return 0;
815 }
816 
817 static int
818 rip6_peeraddr(struct socket *so, struct sockaddr *nam)
819 {
820 	KASSERT(solocked(so));
821 	KASSERT(sotoin6pcb(so) != NULL);
822 	KASSERT(nam != NULL);
823 
824 	in6_setpeeraddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
825 	return 0;
826 }
827 
828 static int
829 rip6_sockaddr(struct socket *so, struct sockaddr *nam)
830 {
831 	KASSERT(solocked(so));
832 	KASSERT(sotoin6pcb(so) != NULL);
833 	KASSERT(nam != NULL);
834 
835 	in6_setsockaddr(sotoin6pcb(so), (struct sockaddr_in6 *)nam);
836 	return 0;
837 }
838 
839 static int
840 rip6_rcvd(struct socket *so, int flags, struct lwp *l)
841 {
842 	KASSERT(solocked(so));
843 
844 	return EOPNOTSUPP;
845 }
846 
847 static int
848 rip6_recvoob(struct socket *so, struct mbuf *m, int flags)
849 {
850 	KASSERT(solocked(so));
851 
852 	return EOPNOTSUPP;
853 }
854 
855 static int
856 rip6_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
857     struct mbuf *control, struct lwp *l)
858 {
859 	struct in6pcb *in6p = sotoin6pcb(so);
860 	struct sockaddr_in6 tmp;
861 	struct sockaddr_in6 *dst;
862 	int error = 0;
863 
864 	KASSERT(solocked(so));
865 	KASSERT(in6p != NULL);
866 	KASSERT(m != NULL);
867 
868 	/*
869 	 * Ship a packet out. The appropriate raw output
870 	 * routine handles any messaging necessary.
871 	 */
872 
873 	/* always copy sockaddr to avoid overwrites */
874 	if (so->so_state & SS_ISCONNECTED) {
875 		if (nam) {
876 			error = EISCONN;
877 			goto release;
878 		}
879 		/* XXX */
880 		sockaddr_in6_init(&tmp, &in6p->in6p_faddr, 0, 0, 0);
881 		dst = &tmp;
882 	} else {
883 		if (nam == NULL) {
884 			error = ENOTCONN;
885 			goto release;
886 		}
887 		tmp = *(struct sockaddr_in6 *)nam;
888 		dst = &tmp;
889 
890 		if (dst->sin6_family != AF_INET6) {
891 			error = EAFNOSUPPORT;
892 			goto release;
893 		}
894 		if (dst->sin6_len != sizeof(*dst)) {
895 			error = EINVAL;
896 			goto release;
897 		}
898 	}
899 	error = rip6_output(m, so, dst, control);
900 	m = NULL;
901 
902 release:
903 	if (m)
904 		m_freem(m);
905 
906 	return error;
907 }
908 
909 static int
910 rip6_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
911 {
912 	KASSERT(solocked(so));
913 
914 	m_freem(m);
915 	m_freem(control);
916 
917 	return EOPNOTSUPP;
918 }
919 
920 static int
921 rip6_purgeif(struct socket *so, struct ifnet *ifp)
922 {
923 
924 	mutex_enter(softnet_lock);
925 	in6_pcbpurgeif0(&raw6cbtable, ifp);
926 #ifdef NET_MPSAFE
927 	mutex_exit(softnet_lock);
928 #endif
929 	in6_purgeif(ifp);
930 #ifdef NET_MPSAFE
931 	mutex_enter(softnet_lock);
932 #endif
933 	in6_pcbpurgeif(&raw6cbtable, ifp);
934 	mutex_exit(softnet_lock);
935 
936 	return 0;
937 }
938 
939 static int
940 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
941 {
942 
943 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
944 }
945 
946 static void
947 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
948 {
949 
950 	sysctl_createv(clog, 0, NULL, NULL,
951 		       CTLFLAG_PERMANENT,
952 		       CTLTYPE_NODE, "inet6", NULL,
953 		       NULL, 0, NULL, 0,
954 		       CTL_NET, PF_INET6, CTL_EOL);
955 	sysctl_createv(clog, 0, NULL, NULL,
956 		       CTLFLAG_PERMANENT,
957 		       CTLTYPE_NODE, "raw6",
958 		       SYSCTL_DESCR("Raw IPv6 settings"),
959 		       NULL, 0, NULL, 0,
960 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
961 
962 	sysctl_createv(clog, 0, NULL, NULL,
963 		       CTLFLAG_PERMANENT,
964 		       CTLTYPE_STRUCT, "pcblist",
965 		       SYSCTL_DESCR("Raw IPv6 control block list"),
966 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
967 		       CTL_NET, PF_INET6, IPPROTO_RAW,
968 		       CTL_CREATE, CTL_EOL);
969 	sysctl_createv(clog, 0, NULL, NULL,
970 		       CTLFLAG_PERMANENT,
971 		       CTLTYPE_STRUCT, "stats",
972 		       SYSCTL_DESCR("Raw IPv6 statistics"),
973 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
974 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
975 		       CTL_EOL);
976 }
977 
978 PR_WRAP_USRREQS(rip6)
979 #define	rip6_attach		rip6_attach_wrapper
980 #define	rip6_detach		rip6_detach_wrapper
981 #define	rip6_accept		rip6_accept_wrapper
982 #define	rip6_bind		rip6_bind_wrapper
983 #define	rip6_listen		rip6_listen_wrapper
984 #define	rip6_connect		rip6_connect_wrapper
985 #define	rip6_connect2		rip6_connect2_wrapper
986 #define	rip6_disconnect		rip6_disconnect_wrapper
987 #define	rip6_shutdown		rip6_shutdown_wrapper
988 #define	rip6_abort		rip6_abort_wrapper
989 #define	rip6_ioctl		rip6_ioctl_wrapper
990 #define	rip6_stat		rip6_stat_wrapper
991 #define	rip6_peeraddr		rip6_peeraddr_wrapper
992 #define	rip6_sockaddr		rip6_sockaddr_wrapper
993 #define	rip6_rcvd		rip6_rcvd_wrapper
994 #define	rip6_recvoob		rip6_recvoob_wrapper
995 #define	rip6_send		rip6_send_wrapper
996 #define	rip6_sendoob		rip6_sendoob_wrapper
997 #define	rip6_purgeif		rip6_purgeif_wrapper
998 
999 const struct pr_usrreqs rip6_usrreqs = {
1000 	.pr_attach	= rip6_attach,
1001 	.pr_detach	= rip6_detach,
1002 	.pr_accept	= rip6_accept,
1003 	.pr_bind	= rip6_bind,
1004 	.pr_listen	= rip6_listen,
1005 	.pr_connect	= rip6_connect,
1006 	.pr_connect2	= rip6_connect2,
1007 	.pr_disconnect	= rip6_disconnect,
1008 	.pr_shutdown	= rip6_shutdown,
1009 	.pr_abort	= rip6_abort,
1010 	.pr_ioctl	= rip6_ioctl,
1011 	.pr_stat	= rip6_stat,
1012 	.pr_peeraddr	= rip6_peeraddr,
1013 	.pr_sockaddr	= rip6_sockaddr,
1014 	.pr_rcvd	= rip6_rcvd,
1015 	.pr_recvoob	= rip6_recvoob,
1016 	.pr_send	= rip6_send,
1017 	.pr_sendoob	= rip6_sendoob,
1018 	.pr_purgeif	= rip6_purgeif,
1019 };
1020