xref: /openbsd-src/sys/netinet6/raw_ip6.c (revision 3374c67d44f9b75b98444cbf63020f777792342e)
1 /*	$OpenBSD: raw_ip6.c,v 1.171 2022/10/17 14:49:02 mvs Exp $	*/
2 /*	$KAME: raw_ip6.c,v 1.69 2001/03/04 15:55:44 itojun 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 "pf.h"
65 
66 #include <sys/param.h>
67 #include <sys/malloc.h>
68 #include <sys/mbuf.h>
69 #include <sys/socket.h>
70 #include <sys/protosw.h>
71 #include <sys/socketvar.h>
72 #include <sys/errno.h>
73 #include <sys/systm.h>
74 #include <sys/sysctl.h>
75 
76 #include <net/if.h>
77 #include <net/if_var.h>
78 #include <net/route.h>
79 
80 #include <netinet/in.h>
81 #include <netinet6/in6_var.h>
82 #include <netinet/ip6.h>
83 #include <netinet6/ip6_var.h>
84 #ifdef MROUTING
85 #include <netinet6/ip6_mroute.h>
86 #endif
87 #include <netinet/icmp6.h>
88 #include <netinet/ip.h>
89 #include <netinet/in_pcb.h>
90 #include <netinet6/nd6.h>
91 #include <netinet6/ip6protosw.h>
92 #include <netinet6/raw_ip6.h>
93 
94 #if NPF > 0
95 #include <net/pfvar.h>
96 #endif
97 
98 #include <sys/stdarg.h>
99 
100 /*
101  * Raw interface to IP6 protocol.
102  */
103 
104 struct	inpcbtable rawin6pcbtable;
105 
106 struct cpumem *rip6counters;
107 
108 const struct pr_usrreqs rip6_usrreqs = {
109 	.pru_attach	= rip6_attach,
110 	.pru_detach	= rip6_detach,
111 	.pru_lock	= rip6_lock,
112 	.pru_unlock	= rip6_unlock,
113 	.pru_bind	= rip6_bind,
114 	.pru_connect	= rip6_connect,
115 	.pru_disconnect	= rip6_disconnect,
116 	.pru_shutdown	= rip6_shutdown,
117 	.pru_send	= rip6_send,
118 	.pru_control	= in6_control,
119 	.pru_sockaddr	= in6_sockaddr,
120 	.pru_peeraddr	= in6_peeraddr,
121 };
122 
123 /*
124  * Initialize raw connection block queue.
125  */
126 void
127 rip6_init(void)
128 {
129 	in_pcbinit(&rawin6pcbtable, 1);
130 	rip6counters = counters_alloc(rip6s_ncounters);
131 }
132 
133 int
134 rip6_input(struct mbuf **mp, int *offp, int proto, int af)
135 {
136 	struct mbuf *m = *mp;
137 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
138 	struct inpcb *in6p;
139 	SIMPLEQ_HEAD(, inpcb) inpcblist;
140 	struct in6_addr *key;
141 	struct sockaddr_in6 rip6src;
142 	uint8_t type;
143 
144 	KASSERT(af == AF_INET6);
145 
146 	if (proto == IPPROTO_ICMPV6) {
147 		struct icmp6_hdr *icmp6;
148 
149 		IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, *offp,
150 		    sizeof(*icmp6));
151 		if (icmp6 == NULL)
152 			return IPPROTO_DONE;
153 		type = icmp6->icmp6_type;
154 	} else
155 		rip6stat_inc(rip6s_ipackets);
156 
157 	bzero(&rip6src, sizeof(rip6src));
158 	rip6src.sin6_len = sizeof(struct sockaddr_in6);
159 	rip6src.sin6_family = AF_INET6;
160 	/* KAME hack: recover scopeid */
161 	in6_recoverscope(&rip6src, &ip6->ip6_src);
162 
163 	key = &ip6->ip6_dst;
164 #if NPF > 0
165 	if (m->m_pkthdr.pf.flags & PF_TAG_DIVERTED) {
166 		struct pf_divert *divert;
167 
168 		divert = pf_find_divert(m);
169 		KASSERT(divert != NULL);
170 		switch (divert->type) {
171 		case PF_DIVERT_TO:
172 			key = &divert->addr.v6;
173 			break;
174 		case PF_DIVERT_REPLY:
175 			break;
176 		default:
177 			panic("%s: unknown divert type %d, mbuf %p, divert %p",
178 			    __func__, divert->type, m, divert);
179 		}
180 	}
181 #endif
182 	SIMPLEQ_INIT(&inpcblist);
183 	rw_enter_write(&rawin6pcbtable.inpt_notify);
184 	mtx_enter(&rawin6pcbtable.inpt_mtx);
185 	TAILQ_FOREACH(in6p, &rawin6pcbtable.inpt_queue, inp_queue) {
186 		if (in6p->inp_socket->so_state & SS_CANTRCVMORE)
187 			continue;
188 		if (rtable_l2(in6p->inp_rtableid) !=
189 		    rtable_l2(m->m_pkthdr.ph_rtableid))
190 			continue;
191 
192 		if (!(in6p->inp_flags & INP_IPV6))
193 			continue;
194 		if ((in6p->inp_ipv6.ip6_nxt || proto == IPPROTO_ICMPV6) &&
195 		    in6p->inp_ipv6.ip6_nxt != proto)
196 			continue;
197 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->inp_laddr6) &&
198 		    !IN6_ARE_ADDR_EQUAL(&in6p->inp_laddr6, key))
199 			continue;
200 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->inp_faddr6) &&
201 		    !IN6_ARE_ADDR_EQUAL(&in6p->inp_faddr6, &ip6->ip6_src))
202 			continue;
203 		if (proto == IPPROTO_ICMPV6 && in6p->inp_icmp6filt) {
204 			if (ICMP6_FILTER_WILLBLOCK(type, in6p->inp_icmp6filt))
205 				continue;
206 		}
207 		if (proto != IPPROTO_ICMPV6 && in6p->inp_cksum6 != -1) {
208 			rip6stat_inc(rip6s_isum);
209 			/*
210 			 * Although in6_cksum() does not need the position of
211 			 * the checksum field for verification, enforce that it
212 			 * is located within the packet.  Userland has given
213 			 * a checksum offset, a packet too short for that is
214 			 * invalid.  Avoid overflow with user supplied offset.
215 			 */
216 			if (m->m_pkthdr.len < *offp + 2 ||
217 			    m->m_pkthdr.len - *offp - 2 < in6p->inp_cksum6 ||
218 			    in6_cksum(m, proto, *offp,
219 			    m->m_pkthdr.len - *offp)) {
220 				rip6stat_inc(rip6s_badsum);
221 				continue;
222 			}
223 		}
224 
225 		in_pcbref(in6p);
226 		SIMPLEQ_INSERT_TAIL(&inpcblist, in6p, inp_notify);
227 	}
228 	mtx_leave(&rawin6pcbtable.inpt_mtx);
229 
230 	if (SIMPLEQ_EMPTY(&inpcblist)) {
231 		struct counters_ref ref;
232 		uint64_t *counters;
233 
234 		rw_exit_write(&rawin6pcbtable.inpt_notify);
235 
236 		if (proto != IPPROTO_ICMPV6) {
237 			rip6stat_inc(rip6s_nosock);
238 			if (m->m_flags & M_MCAST)
239 				rip6stat_inc(rip6s_nosockmcast);
240 		}
241 		if (proto == IPPROTO_NONE || proto == IPPROTO_ICMPV6) {
242 			m_freem(m);
243 		} else {
244 			int prvnxt = ip6_get_prevhdr(m, *offp);
245 
246 			icmp6_error(m, ICMP6_PARAM_PROB,
247 			    ICMP6_PARAMPROB_NEXTHEADER, prvnxt);
248 		}
249 		counters = counters_enter(&ref, ip6counters);
250 		counters[ip6s_delivered]--;
251 		counters_leave(&ref, ip6counters);
252 
253 		return IPPROTO_DONE;
254 	}
255 
256 	while ((in6p = SIMPLEQ_FIRST(&inpcblist)) != NULL) {
257 		struct mbuf *n, *opts = NULL;
258 
259 		SIMPLEQ_REMOVE_HEAD(&inpcblist, inp_notify);
260 		if (SIMPLEQ_EMPTY(&inpcblist))
261 			n = m;
262 		else
263 			n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
264 		if (n != NULL) {
265 			int ret;
266 
267 			if (in6p->inp_flags & IN6P_CONTROLOPTS)
268 				ip6_savecontrol(in6p, n, &opts);
269 			/* strip intermediate headers */
270 			m_adj(n, *offp);
271 
272 			mtx_enter(&in6p->inp_mtx);
273 			ret = sbappendaddr(in6p->inp_socket,
274 			    &in6p->inp_socket->so_rcv,
275 			    sin6tosa(&rip6src), n, opts);
276 			mtx_leave(&in6p->inp_mtx);
277 
278 			if (ret == 0) {
279 				/* should notify about lost packet */
280 				m_freem(n);
281 				m_freem(opts);
282 				rip6stat_inc(rip6s_fullsock);
283 			} else
284 				sorwakeup(in6p->inp_socket);
285 		}
286 		in_pcbunref(in6p);
287 	}
288 	rw_exit_write(&rawin6pcbtable.inpt_notify);
289 
290 	return IPPROTO_DONE;
291 }
292 
293 void
294 rip6_ctlinput(int cmd, struct sockaddr *sa, u_int rdomain, void *d)
295 {
296 	struct ip6_hdr *ip6;
297 	struct ip6ctlparam *ip6cp = NULL;
298 	struct sockaddr_in6 *sa6 = satosin6(sa);
299 	const struct sockaddr_in6 *sa6_src = NULL;
300 	void *cmdarg;
301 	void (*notify)(struct inpcb *, int) = in_rtchange;
302 	int nxt;
303 
304 	if (sa->sa_family != AF_INET6 ||
305 	    sa->sa_len != sizeof(struct sockaddr_in6))
306 		return;
307 
308 	if ((unsigned)cmd >= PRC_NCMDS)
309 		return;
310 	if (PRC_IS_REDIRECT(cmd))
311 		notify = in_rtchange, d = NULL;
312 	else if (cmd == PRC_HOSTDEAD)
313 		d = NULL;
314 	else if (cmd == PRC_MSGSIZE)
315 		; /* special code is present, see below */
316 	else if (inet6ctlerrmap[cmd] == 0)
317 		return;
318 
319 	/* if the parameter is from icmp6, decode it. */
320 	if (d != NULL) {
321 		ip6cp = (struct ip6ctlparam *)d;
322 		ip6 = ip6cp->ip6c_ip6;
323 		cmdarg = ip6cp->ip6c_cmdarg;
324 		sa6_src = ip6cp->ip6c_src;
325 		nxt = ip6cp->ip6c_nxt;
326 	} else {
327 		ip6 = NULL;
328 		cmdarg = NULL;
329 		sa6_src = &sa6_any;
330 		nxt = -1;
331 	}
332 
333 	if (ip6 && cmd == PRC_MSGSIZE) {
334 		int valid = 0;
335 		struct inpcb *in6p;
336 
337 		/*
338 		 * Check to see if we have a valid raw IPv6 socket
339 		 * corresponding to the address in the ICMPv6 message
340 		 * payload, and the protocol (ip6_nxt) meets the socket.
341 		 * XXX chase extension headers, or pass final nxt value
342 		 * from icmp6_notify_error()
343 		 */
344 		in6p = in6_pcblookup(&rawin6pcbtable, &sa6->sin6_addr, 0,
345 		    &sa6_src->sin6_addr, 0, rdomain);
346 
347 		if (in6p && in6p->inp_ipv6.ip6_nxt &&
348 		    in6p->inp_ipv6.ip6_nxt == nxt)
349 			valid = 1;
350 
351 		/*
352 		 * Depending on the value of "valid" and routing table
353 		 * size (mtudisc_{hi,lo}wat), we will:
354 		 * - recalculate the new MTU and create the
355 		 *   corresponding routing entry, or
356 		 * - ignore the MTU change notification.
357 		 */
358 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
359 		in_pcbunref(in6p);
360 
361 		/*
362 		 * regardless of if we called icmp6_mtudisc_update(),
363 		 * we need to call in6_pcbnotify(), to notify path
364 		 * MTU change to the userland (2292bis-02), because
365 		 * some unconnected sockets may share the same
366 		 * destination and want to know the path MTU.
367 		 */
368 	}
369 
370 	in6_pcbnotify(&rawin6pcbtable, sa6, 0,
371 	    sa6_src, 0, rdomain, cmd, cmdarg, notify);
372 }
373 
374 /*
375  * Generate IPv6 header and pass packet to ip6_output.
376  * Tack on options user may have setup with control call.
377  */
378 int
379 rip6_output(struct mbuf *m, struct socket *so, struct sockaddr *dstaddr,
380     struct mbuf *control)
381 {
382 	struct in6_addr *dst;
383 	struct ip6_hdr *ip6;
384 	struct inpcb *in6p;
385 	u_int	plen = m->m_pkthdr.len;
386 	int error = 0;
387 	struct ip6_pktopts opt, *optp = NULL, *origoptp;
388 	int type;		/* for ICMPv6 output statistics only */
389 	int priv = 0;
390 	int flags;
391 
392 	in6p = sotoinpcb(so);
393 
394 	priv = 0;
395 	if ((so->so_state & SS_PRIV) != 0)
396 		priv = 1;
397 	if (control) {
398 		if ((error = ip6_setpktopts(control, &opt,
399 		    in6p->inp_outputopts6,
400 		    priv, so->so_proto->pr_protocol)) != 0)
401 			goto bad;
402 		optp = &opt;
403 	} else
404 		optp = in6p->inp_outputopts6;
405 
406 	if (dstaddr->sa_family != AF_INET6) {
407 		error = EAFNOSUPPORT;
408 		goto bad;
409 	}
410 	dst = &satosin6(dstaddr)->sin6_addr;
411 	if (IN6_IS_ADDR_V4MAPPED(dst)) {
412 		error = EADDRNOTAVAIL;
413 		goto bad;
414 	}
415 
416 	/*
417 	 * For an ICMPv6 packet, we should know its type and code
418 	 * to update statistics.
419 	 */
420 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
421 		struct icmp6_hdr *icmp6;
422 		if (m->m_len < sizeof(struct icmp6_hdr) &&
423 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
424 			error = ENOBUFS;
425 			goto bad;
426 		}
427 		icmp6 = mtod(m, struct icmp6_hdr *);
428 		type = icmp6->icmp6_type;
429 	}
430 
431 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
432 	if (!m) {
433 		error = ENOBUFS;
434 		goto bad;
435 	}
436 	ip6 = mtod(m, struct ip6_hdr *);
437 
438 	/*
439 	 * Next header might not be ICMP6 but use its pseudo header anyway.
440 	 */
441 	ip6->ip6_dst = *dst;
442 
443 	/* KAME hack: embed scopeid */
444 	origoptp = in6p->inp_outputopts6;
445 	in6p->inp_outputopts6 = optp;
446 	if (in6_embedscope(&ip6->ip6_dst, satosin6(dstaddr), in6p) != 0) {
447 		error = EINVAL;
448 		goto bad;
449 	}
450 	in6p->inp_outputopts6 = origoptp;
451 
452 	/*
453 	 * Source address selection.
454 	 */
455 	{
456 		struct in6_addr *in6a;
457 
458 		error = in6_pcbselsrc(&in6a, satosin6(dstaddr), in6p, optp);
459 		if (error)
460 			goto bad;
461 
462 		ip6->ip6_src = *in6a;
463 	}
464 
465 	ip6->ip6_flow = in6p->inp_flowinfo & IPV6_FLOWINFO_MASK;
466 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
467 	ip6->ip6_vfc  |= IPV6_VERSION;
468 #if 0				/* ip6_plen will be filled in ip6_output. */
469 	ip6->ip6_plen  = htons((u_short)plen);
470 #endif
471 	ip6->ip6_nxt   = in6p->inp_ipv6.ip6_nxt;
472 	ip6->ip6_hlim = in6_selecthlim(in6p);
473 
474 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
475 	    in6p->inp_cksum6 != -1) {
476 		struct mbuf *n;
477 		int off;
478 		u_int16_t *sump;
479 		int sumoff;
480 
481 		/* compute checksum */
482 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
483 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
484 		else
485 			off = in6p->inp_cksum6;
486 		if (plen < 2 || plen - 2 < off) {
487 			error = EINVAL;
488 			goto bad;
489 		}
490 		off += sizeof(struct ip6_hdr);
491 
492 		n = m_pulldown(m, off, sizeof(*sump), &sumoff);
493 		if (n == NULL) {
494 			m = NULL;
495 			error = ENOBUFS;
496 			goto bad;
497 		}
498 		sump = (u_int16_t *)(mtod(n, caddr_t) + sumoff);
499 		*sump = 0;
500 		*sump = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
501 	}
502 
503 	flags = 0;
504 	if (in6p->inp_flags & IN6P_MINMTU)
505 		flags |= IPV6_MINMTU;
506 
507 	/* force routing table */
508 	m->m_pkthdr.ph_rtableid = in6p->inp_rtableid;
509 
510 #if NPF > 0
511 	if (in6p->inp_socket->so_state & SS_ISCONNECTED &&
512 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
513 		pf_mbuf_link_inpcb(m, in6p);
514 #endif
515 
516 	error = ip6_output(m, optp, &in6p->inp_route6, flags,
517 	    in6p->inp_moptions6, in6p);
518 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
519 		icmp6stat_inc(icp6s_outhist + type);
520 	} else
521 		rip6stat_inc(rip6s_opackets);
522 
523 	goto freectl;
524 
525  bad:
526 	m_freem(m);
527 
528  freectl:
529 	if (control) {
530 		ip6_clearpktopts(&opt, -1);
531 		m_freem(control);
532 	}
533 	return (error);
534 }
535 
536 /*
537  * Raw IPv6 socket option processing.
538  */
539 int
540 rip6_ctloutput(int op, struct socket *so, int level, int optname,
541     struct mbuf *m)
542 {
543 #ifdef MROUTING
544 	int error;
545 #endif
546 
547 	switch (level) {
548 	case IPPROTO_IPV6:
549 		switch (optname) {
550 #ifdef MROUTING
551 		case MRT6_INIT:
552 		case MRT6_DONE:
553 		case MRT6_ADD_MIF:
554 		case MRT6_DEL_MIF:
555 		case MRT6_ADD_MFC:
556 		case MRT6_DEL_MFC:
557 			if (op == PRCO_SETOPT) {
558 				error = ip6_mrouter_set(optname, so, m);
559 			} else if (op == PRCO_GETOPT)
560 				error = ip6_mrouter_get(optname, so, m);
561 			else
562 				error = EINVAL;
563 			return (error);
564 #endif
565 		case IPV6_CHECKSUM:
566 			return (ip6_raw_ctloutput(op, so, level, optname, m));
567 		default:
568 			return (ip6_ctloutput(op, so, level, optname, m));
569 		}
570 
571 	case IPPROTO_ICMPV6:
572 		/*
573 		 * XXX: is it better to call icmp6_ctloutput() directly
574 		 * from protosw?
575 		 */
576 		return (icmp6_ctloutput(op, so, level, optname, m));
577 
578 	default:
579 		return EINVAL;
580 	}
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, int wait)
588 {
589 	struct inpcb *in6p;
590 	int error;
591 
592 	if (so->so_pcb)
593 		panic("%s", __func__);
594 	if ((so->so_state & SS_PRIV) == 0)
595 		return (EACCES);
596 	if (proto < 0 || proto >= IPPROTO_MAX)
597 		return EPROTONOSUPPORT;
598 
599 	if ((error = soreserve(so, rip6_sendspace, rip6_recvspace)))
600 		return error;
601 	NET_ASSERT_LOCKED();
602 	if ((error = in_pcballoc(so, &rawin6pcbtable, wait)))
603 		return error;
604 
605 	in6p = sotoinpcb(so);
606 	in6p->inp_ipv6.ip6_nxt = proto;
607 	in6p->inp_cksum6 = -1;
608 
609 	in6p->inp_icmp6filt = malloc(sizeof(struct icmp6_filter), M_PCB,
610 	    wait == M_WAIT ? M_WAITOK : M_NOWAIT);
611 	if (in6p->inp_icmp6filt == NULL) {
612 		in_pcbdetach(in6p);
613 		return ENOMEM;
614 	}
615 	ICMP6_FILTER_SETPASSALL(in6p->inp_icmp6filt);
616 	return 0;
617 }
618 
619 int
620 rip6_detach(struct socket *so)
621 {
622 	struct inpcb *in6p = sotoinpcb(so);
623 
624 	soassertlocked(so);
625 
626 	if (in6p == NULL)
627 		panic("%s", __func__);
628 #ifdef MROUTING
629 	if (so == ip6_mrouter[in6p->inp_rtableid])
630 		ip6_mrouter_done(so);
631 #endif
632 	free(in6p->inp_icmp6filt, M_PCB, sizeof(struct icmp6_filter));
633 	in6p->inp_icmp6filt = NULL;
634 
635 	in_pcbdetach(in6p);
636 
637 	return (0);
638 }
639 
640 void
641 rip6_lock(struct socket *so)
642 {
643 	struct inpcb *in6p = sotoinpcb(so);
644 
645 	NET_ASSERT_LOCKED();
646 	mtx_enter(&in6p->inp_mtx);
647 }
648 
649 void
650 rip6_unlock(struct socket *so)
651 {
652 	struct inpcb *in6p = sotoinpcb(so);
653 
654 	NET_ASSERT_LOCKED();
655 	mtx_leave(&in6p->inp_mtx);
656 }
657 
658 int
659 rip6_bind(struct socket *so, struct mbuf *nam, struct proc *p)
660 {
661 	struct inpcb *in6p = sotoinpcb(so);
662 	struct sockaddr_in6 *addr;
663 	int error;
664 
665 	soassertlocked(so);
666 
667 	if ((error = in6_nam2sin6(nam, &addr)))
668 		return (error);
669 
670 	/*
671 	 * Make sure to not enter in_pcblookup_local(), local ports
672 	 * are non-sensical for raw sockets.
673 	 */
674 	addr->sin6_port = 0;
675 
676 	if ((error = in6_pcbaddrisavail(in6p, addr, 0, p)))
677 		return (error);
678 
679 	in6p->inp_laddr6 = addr->sin6_addr;
680 	return (0);
681 }
682 
683 int
684 rip6_connect(struct socket *so, struct mbuf *nam)
685 {
686 	struct inpcb *in6p = sotoinpcb(so);
687 	struct sockaddr_in6 *addr;
688 	struct in6_addr *in6a = NULL;
689 	int error;
690 
691 	soassertlocked(so);
692 
693 	if ((error = in6_nam2sin6(nam, &addr)))
694 		return (error);
695 
696 	/* Source address selection. XXX: need pcblookup? */
697 	error = in6_pcbselsrc(&in6a, addr, in6p, in6p->inp_outputopts6);
698 	if (error)
699 		return (error);
700 
701 	in6p->inp_laddr6 = *in6a;
702 	in6p->inp_faddr6 = addr->sin6_addr;
703 	soisconnected(so);
704 	return (0);
705 }
706 
707 int
708 rip6_disconnect(struct socket *so)
709 {
710 	struct inpcb *in6p = sotoinpcb(so);
711 
712 	soassertlocked(so);
713 
714 	if ((so->so_state & SS_ISCONNECTED) == 0)
715 		return (ENOTCONN);
716 
717 	in6p->inp_faddr6 = in6addr_any;
718 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
719 	return (0);
720 }
721 
722 int
723 rip6_shutdown(struct socket *so)
724 {
725 	/*
726 	 * Mark the connection as being incapable of further input.
727 	 */
728 	soassertlocked(so);
729 	socantsendmore(so);
730 	return (0);
731 }
732 
733 int
734 rip6_send(struct socket *so, struct mbuf *m, struct mbuf *nam,
735 	struct mbuf *control)
736 {
737 	struct inpcb *in6p = sotoinpcb(so);
738 	struct sockaddr_in6 dst;
739 	int error;
740 
741 	soassertlocked(so);
742 
743 	/*
744 	 * Ship a packet out. The appropriate raw output
745 	 * routine handles any messaging necessary.
746 	 */
747 
748 	/* always copy sockaddr to avoid overwrites */
749 	memset(&dst, 0, sizeof(dst));
750 	dst.sin6_family = AF_INET6;
751 	dst.sin6_len = sizeof(dst);
752 	if (so->so_state & SS_ISCONNECTED) {
753 		if (nam) {
754 			error = EISCONN;
755 			goto out;
756 		}
757 		dst.sin6_addr = in6p->inp_faddr6;
758 	} else {
759 		struct sockaddr_in6 *addr6;
760 
761 		if (nam == NULL) {
762 			error = ENOTCONN;
763 			goto out;
764 		}
765 		if ((error = in6_nam2sin6(nam, &addr6)))
766 			goto out;
767 		dst.sin6_addr = addr6->sin6_addr;
768 		dst.sin6_scope_id = addr6->sin6_scope_id;
769 	}
770 	error = rip6_output(m, so, sin6tosa(&dst), control);
771 	control = NULL;
772 	m = NULL;
773 
774 out:
775 	m_freem(control);
776 	m_freem(m);
777 
778 	return (error);
779 }
780 
781 int
782 rip6_sysctl_rip6stat(void *oldp, size_t *oldplen, void *newp)
783 {
784 	struct rip6stat rip6stat;
785 
786 	CTASSERT(sizeof(rip6stat) == rip6s_ncounters * sizeof(uint64_t));
787 	counters_read(rip6counters, (uint64_t *)&rip6stat, rip6s_ncounters);
788 
789 	return (sysctl_rdstruct(oldp, oldplen, newp,
790 	    &rip6stat, sizeof(rip6stat)));
791 }
792 
793 int
794 rip6_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
795     void *newp, size_t newlen)
796 {
797 	/* All sysctl names at this level are terminal. */
798 	if (namelen != 1)
799 		return ENOTDIR;
800 
801 	switch (name[0]) {
802 	case RIPV6CTL_STATS:
803 		return (rip6_sysctl_rip6stat(oldp, oldlenp, newp));
804 	default:
805 		return (EOPNOTSUPP);
806 	}
807 	/* NOTREACHED */
808 }
809