xref: /openbsd/sys/netinet6/raw_ip6.c (revision 9ea232b5)
1 /*	$OpenBSD: raw_ip6.c,v 1.180 2024/02/03 22:50:09 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_locked	= rip6_locked,
114 	.pru_bind	= rip6_bind,
115 	.pru_connect	= rip6_connect,
116 	.pru_disconnect	= rip6_disconnect,
117 	.pru_shutdown	= rip6_shutdown,
118 	.pru_send	= rip6_send,
119 	.pru_control	= in6_control,
120 	.pru_sockaddr	= in6_sockaddr,
121 	.pru_peeraddr	= in6_peeraddr,
122 };
123 
124 /*
125  * Initialize raw connection block queue.
126  */
127 void
128 rip6_init(void)
129 {
130 	in_pcbinit(&rawin6pcbtable, 1);
131 	rip6counters = counters_alloc(rip6s_ncounters);
132 }
133 
134 int
135 rip6_input(struct mbuf **mp, int *offp, int proto, int af)
136 {
137 	struct mbuf *m = *mp;
138 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
139 	struct inpcb *inp;
140 	SIMPLEQ_HEAD(, inpcb) inpcblist;
141 	struct in6_addr *key;
142 	struct sockaddr_in6 rip6src;
143 	uint8_t type;
144 
145 	KASSERT(af == AF_INET6);
146 
147 	if (proto == IPPROTO_ICMPV6) {
148 		struct icmp6_hdr *icmp6;
149 
150 		IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, *offp,
151 		    sizeof(*icmp6));
152 		if (icmp6 == NULL)
153 			return IPPROTO_DONE;
154 		type = icmp6->icmp6_type;
155 	} else
156 		rip6stat_inc(rip6s_ipackets);
157 
158 	bzero(&rip6src, sizeof(rip6src));
159 	rip6src.sin6_len = sizeof(struct sockaddr_in6);
160 	rip6src.sin6_family = AF_INET6;
161 	/* KAME hack: recover scopeid */
162 	in6_recoverscope(&rip6src, &ip6->ip6_src);
163 
164 	key = &ip6->ip6_dst;
165 #if NPF > 0
166 	if (m->m_pkthdr.pf.flags & PF_TAG_DIVERTED) {
167 		struct pf_divert *divert;
168 
169 		divert = pf_find_divert(m);
170 		KASSERT(divert != NULL);
171 		switch (divert->type) {
172 		case PF_DIVERT_TO:
173 			key = &divert->addr.v6;
174 			break;
175 		case PF_DIVERT_REPLY:
176 			break;
177 		default:
178 			panic("%s: unknown divert type %d, mbuf %p, divert %p",
179 			    __func__, divert->type, m, divert);
180 		}
181 	}
182 #endif
183 	SIMPLEQ_INIT(&inpcblist);
184 	rw_enter_write(&rawin6pcbtable.inpt_notify);
185 	mtx_enter(&rawin6pcbtable.inpt_mtx);
186 	TAILQ_FOREACH(inp, &rawin6pcbtable.inpt_queue, inp_queue) {
187 		KASSERT(ISSET(inp->inp_flags, INP_IPV6));
188 
189 		if (inp->inp_socket->so_rcv.sb_state & SS_CANTRCVMORE)
190 			continue;
191 		if (rtable_l2(inp->inp_rtableid) !=
192 		    rtable_l2(m->m_pkthdr.ph_rtableid))
193 			continue;
194 
195 		if ((inp->inp_ipv6.ip6_nxt || proto == IPPROTO_ICMPV6) &&
196 		    inp->inp_ipv6.ip6_nxt != proto)
197 			continue;
198 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->inp_laddr6) &&
199 		    !IN6_ARE_ADDR_EQUAL(&inp->inp_laddr6, key))
200 			continue;
201 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->inp_faddr6) &&
202 		    !IN6_ARE_ADDR_EQUAL(&inp->inp_faddr6, &ip6->ip6_src))
203 			continue;
204 		if (proto == IPPROTO_ICMPV6 && inp->inp_icmp6filt) {
205 			if (ICMP6_FILTER_WILLBLOCK(type, inp->inp_icmp6filt))
206 				continue;
207 		}
208 		if (proto != IPPROTO_ICMPV6 && inp->inp_cksum6 != -1) {
209 			rip6stat_inc(rip6s_isum);
210 			/*
211 			 * Although in6_cksum() does not need the position of
212 			 * the checksum field for verification, enforce that it
213 			 * is located within the packet.  Userland has given
214 			 * a checksum offset, a packet too short for that is
215 			 * invalid.  Avoid overflow with user supplied offset.
216 			 */
217 			if (m->m_pkthdr.len < *offp + 2 ||
218 			    m->m_pkthdr.len - *offp - 2 < inp->inp_cksum6 ||
219 			    in6_cksum(m, proto, *offp,
220 			    m->m_pkthdr.len - *offp)) {
221 				rip6stat_inc(rip6s_badsum);
222 				continue;
223 			}
224 		}
225 
226 		in_pcbref(inp);
227 		SIMPLEQ_INSERT_TAIL(&inpcblist, inp, inp_notify);
228 	}
229 	mtx_leave(&rawin6pcbtable.inpt_mtx);
230 
231 	if (SIMPLEQ_EMPTY(&inpcblist)) {
232 		struct counters_ref ref;
233 		uint64_t *counters;
234 
235 		rw_exit_write(&rawin6pcbtable.inpt_notify);
236 
237 		if (proto != IPPROTO_ICMPV6) {
238 			rip6stat_inc(rip6s_nosock);
239 			if (m->m_flags & M_MCAST)
240 				rip6stat_inc(rip6s_nosockmcast);
241 		}
242 		if (proto == IPPROTO_NONE || proto == IPPROTO_ICMPV6) {
243 			m_freem(m);
244 		} else {
245 			int prvnxt = ip6_get_prevhdr(m, *offp);
246 
247 			icmp6_error(m, ICMP6_PARAM_PROB,
248 			    ICMP6_PARAMPROB_NEXTHEADER, prvnxt);
249 		}
250 		counters = counters_enter(&ref, ip6counters);
251 		counters[ip6s_delivered]--;
252 		counters_leave(&ref, ip6counters);
253 
254 		return IPPROTO_DONE;
255 	}
256 
257 	while ((inp = SIMPLEQ_FIRST(&inpcblist)) != NULL) {
258 		struct mbuf *n, *opts = NULL;
259 
260 		SIMPLEQ_REMOVE_HEAD(&inpcblist, inp_notify);
261 		if (SIMPLEQ_EMPTY(&inpcblist))
262 			n = m;
263 		else
264 			n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
265 		if (n != NULL) {
266 			int ret;
267 
268 			if (inp->inp_flags & IN6P_CONTROLOPTS)
269 				ip6_savecontrol(inp, n, &opts);
270 			/* strip intermediate headers */
271 			m_adj(n, *offp);
272 
273 			mtx_enter(&inp->inp_mtx);
274 			ret = sbappendaddr(inp->inp_socket,
275 			    &inp->inp_socket->so_rcv,
276 			    sin6tosa(&rip6src), n, opts);
277 			mtx_leave(&inp->inp_mtx);
278 
279 			if (ret == 0) {
280 				/* should notify about lost packet */
281 				m_freem(n);
282 				m_freem(opts);
283 				rip6stat_inc(rip6s_fullsock);
284 			} else
285 				sorwakeup(inp->inp_socket);
286 		}
287 		in_pcbunref(inp);
288 	}
289 	rw_exit_write(&rawin6pcbtable.inpt_notify);
290 
291 	return IPPROTO_DONE;
292 }
293 
294 void
295 rip6_ctlinput(int cmd, struct sockaddr *sa, u_int rdomain, void *d)
296 {
297 	struct ip6_hdr *ip6;
298 	struct ip6ctlparam *ip6cp = NULL;
299 	struct sockaddr_in6 *sa6 = satosin6(sa);
300 	const struct sockaddr_in6 *sa6_src = NULL;
301 	void *cmdarg;
302 	void (*notify)(struct inpcb *, int) = in_rtchange;
303 	int nxt;
304 
305 	if (sa->sa_family != AF_INET6 ||
306 	    sa->sa_len != sizeof(struct sockaddr_in6))
307 		return;
308 
309 	if ((unsigned)cmd >= PRC_NCMDS)
310 		return;
311 	if (PRC_IS_REDIRECT(cmd))
312 		notify = in_rtchange, d = NULL;
313 	else if (cmd == PRC_HOSTDEAD)
314 		d = NULL;
315 	else if (cmd == PRC_MSGSIZE)
316 		; /* special code is present, see below */
317 	else if (inet6ctlerrmap[cmd] == 0)
318 		return;
319 
320 	/* if the parameter is from icmp6, decode it. */
321 	if (d != NULL) {
322 		ip6cp = (struct ip6ctlparam *)d;
323 		ip6 = ip6cp->ip6c_ip6;
324 		cmdarg = ip6cp->ip6c_cmdarg;
325 		sa6_src = ip6cp->ip6c_src;
326 		nxt = ip6cp->ip6c_nxt;
327 	} else {
328 		ip6 = NULL;
329 		cmdarg = NULL;
330 		sa6_src = &sa6_any;
331 		nxt = -1;
332 	}
333 
334 	if (ip6 && cmd == PRC_MSGSIZE) {
335 		int valid = 0;
336 		struct inpcb *inp;
337 
338 		/*
339 		 * Check to see if we have a valid raw IPv6 socket
340 		 * corresponding to the address in the ICMPv6 message
341 		 * payload, and the protocol (ip6_nxt) meets the socket.
342 		 * XXX chase extension headers, or pass final nxt value
343 		 * from icmp6_notify_error()
344 		 */
345 		inp = in6_pcblookup(&rawin6pcbtable, &sa6->sin6_addr, 0,
346 		    &sa6_src->sin6_addr, 0, rdomain);
347 
348 		if (inp && inp->inp_ipv6.ip6_nxt &&
349 		    inp->inp_ipv6.ip6_nxt == nxt)
350 			valid = 1;
351 
352 		/*
353 		 * Depending on the value of "valid" and routing table
354 		 * size (mtudisc_{hi,lo}wat), we will:
355 		 * - recalculate the new MTU and create the
356 		 *   corresponding routing entry, or
357 		 * - ignore the MTU change notification.
358 		 */
359 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
360 		in_pcbunref(inp);
361 
362 		/*
363 		 * regardless of if we called icmp6_mtudisc_update(),
364 		 * we need to call in6_pcbnotify(), to notify path
365 		 * MTU change to the userland (2292bis-02), because
366 		 * some unconnected sockets may share the same
367 		 * destination and want to know the path MTU.
368 		 */
369 	}
370 
371 	in6_pcbnotify(&rawin6pcbtable, sa6, 0,
372 	    sa6_src, 0, rdomain, cmd, cmdarg, notify);
373 }
374 
375 /*
376  * Generate IPv6 header and pass packet to ip6_output.
377  * Tack on options user may have setup with control call.
378  */
379 int
380 rip6_output(struct mbuf *m, struct socket *so, struct sockaddr *dstaddr,
381     struct mbuf *control)
382 {
383 	struct in6_addr *dst;
384 	struct ip6_hdr *ip6;
385 	struct inpcb *inp;
386 	u_int	plen = m->m_pkthdr.len;
387 	int error = 0;
388 	struct ip6_pktopts opt, *optp = NULL;
389 	int type;		/* for ICMPv6 output statistics only */
390 	int priv = 0;
391 	int flags;
392 
393 	inp = sotoinpcb(so);
394 
395 	priv = 0;
396 	if ((so->so_state & SS_PRIV) != 0)
397 		priv = 1;
398 	if (control) {
399 		if ((error = ip6_setpktopts(control, &opt,
400 		    inp->inp_outputopts6,
401 		    priv, so->so_proto->pr_protocol)) != 0)
402 			goto bad;
403 		optp = &opt;
404 	} else
405 		optp = inp->inp_outputopts6;
406 
407 	if (dstaddr->sa_family != AF_INET6) {
408 		error = EAFNOSUPPORT;
409 		goto bad;
410 	}
411 	dst = &satosin6(dstaddr)->sin6_addr;
412 	if (IN6_IS_ADDR_V4MAPPED(dst)) {
413 		error = EADDRNOTAVAIL;
414 		goto bad;
415 	}
416 
417 	/*
418 	 * For an ICMPv6 packet, we should know its type and code
419 	 * to update statistics.
420 	 */
421 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
422 		struct icmp6_hdr *icmp6;
423 		if (m->m_len < sizeof(struct icmp6_hdr) &&
424 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
425 			error = ENOBUFS;
426 			goto bad;
427 		}
428 		icmp6 = mtod(m, struct icmp6_hdr *);
429 		type = icmp6->icmp6_type;
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 	if (in6_embedscope(&ip6->ip6_dst, satosin6(dstaddr),
446 	    optp, inp->inp_moptions6) != 0) {
447 		error = EINVAL;
448 		goto bad;
449 	}
450 
451 	/*
452 	 * Source address selection.
453 	 */
454 	{
455 		const struct in6_addr *in6a;
456 
457 		error = in6_pcbselsrc(&in6a, satosin6(dstaddr), inp, optp);
458 		if (error)
459 			goto bad;
460 
461 		ip6->ip6_src = *in6a;
462 	}
463 
464 	ip6->ip6_flow = inp->inp_flowinfo & IPV6_FLOWINFO_MASK;
465 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
466 	ip6->ip6_vfc  |= IPV6_VERSION;
467 #if 0				/* ip6_plen will be filled in ip6_output. */
468 	ip6->ip6_plen  = htons((u_short)plen);
469 #endif
470 	ip6->ip6_nxt   = inp->inp_ipv6.ip6_nxt;
471 	ip6->ip6_hlim = in6_selecthlim(inp);
472 
473 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
474 	    inp->inp_cksum6 != -1) {
475 		struct mbuf *n;
476 		int off;
477 		u_int16_t *sump;
478 		int sumoff;
479 
480 		/* compute checksum */
481 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
482 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
483 		else
484 			off = inp->inp_cksum6;
485 		if (plen < 2 || plen - 2 < off) {
486 			error = EINVAL;
487 			goto bad;
488 		}
489 		off += sizeof(struct ip6_hdr);
490 
491 		n = m_pulldown(m, off, sizeof(*sump), &sumoff);
492 		if (n == NULL) {
493 			m = NULL;
494 			error = ENOBUFS;
495 			goto bad;
496 		}
497 		sump = (u_int16_t *)(mtod(n, caddr_t) + sumoff);
498 		*sump = 0;
499 		*sump = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
500 	}
501 
502 	flags = 0;
503 	if (inp->inp_flags & IN6P_MINMTU)
504 		flags |= IPV6_MINMTU;
505 
506 	/* force routing table */
507 	m->m_pkthdr.ph_rtableid = inp->inp_rtableid;
508 
509 #if NPF > 0
510 	if (inp->inp_socket->so_state & SS_ISCONNECTED &&
511 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
512 		pf_mbuf_link_inpcb(m, inp);
513 #endif
514 
515 	error = ip6_output(m, optp, &inp->inp_route6, flags,
516 	    inp->inp_moptions6, inp->inp_seclevel);
517 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
518 		icmp6stat_inc(icp6s_outhist + type);
519 	} else
520 		rip6stat_inc(rip6s_opackets);
521 
522 	goto freectl;
523 
524  bad:
525 	m_freem(m);
526 
527  freectl:
528 	if (control) {
529 		ip6_clearpktopts(&opt, -1);
530 		m_freem(control);
531 	}
532 	return (error);
533 }
534 
535 /*
536  * Raw IPv6 socket option processing.
537  */
538 int
539 rip6_ctloutput(int op, struct socket *so, int level, int optname,
540     struct mbuf *m)
541 {
542 #ifdef MROUTING
543 	int error;
544 #endif
545 
546 	switch (level) {
547 	case IPPROTO_IPV6:
548 		switch (optname) {
549 #ifdef MROUTING
550 		case MRT6_INIT:
551 		case MRT6_DONE:
552 		case MRT6_ADD_MIF:
553 		case MRT6_DEL_MIF:
554 		case MRT6_ADD_MFC:
555 		case MRT6_DEL_MFC:
556 			if (op == PRCO_SETOPT) {
557 				error = ip6_mrouter_set(optname, so, m);
558 			} else if (op == PRCO_GETOPT)
559 				error = ip6_mrouter_get(optname, so, m);
560 			else
561 				error = EINVAL;
562 			return (error);
563 #endif
564 		case IPV6_CHECKSUM:
565 			return (ip6_raw_ctloutput(op, so, level, optname, m));
566 		default:
567 			return (ip6_ctloutput(op, so, level, optname, m));
568 		}
569 
570 	case IPPROTO_ICMPV6:
571 		/*
572 		 * XXX: is it better to call icmp6_ctloutput() directly
573 		 * from protosw?
574 		 */
575 		return (icmp6_ctloutput(op, so, level, optname, m));
576 
577 	default:
578 		return EINVAL;
579 	}
580 }
581 
582 extern	u_long rip6_sendspace;
583 extern	u_long rip6_recvspace;
584 
585 int
586 rip6_attach(struct socket *so, int proto, int wait)
587 {
588 	struct inpcb *inp;
589 	int error;
590 
591 	if (so->so_pcb)
592 		panic("%s", __func__);
593 	if ((so->so_state & SS_PRIV) == 0)
594 		return (EACCES);
595 	if (proto < 0 || proto >= IPPROTO_MAX)
596 		return EPROTONOSUPPORT;
597 
598 	if ((error = soreserve(so, rip6_sendspace, rip6_recvspace)))
599 		return error;
600 	NET_ASSERT_LOCKED();
601 	if ((error = in_pcballoc(so, &rawin6pcbtable, wait)))
602 		return error;
603 
604 	inp = sotoinpcb(so);
605 	inp->inp_ipv6.ip6_nxt = proto;
606 	inp->inp_cksum6 = -1;
607 
608 	inp->inp_icmp6filt = malloc(sizeof(struct icmp6_filter), M_PCB,
609 	    wait == M_WAIT ? M_WAITOK : M_NOWAIT);
610 	if (inp->inp_icmp6filt == NULL) {
611 		in_pcbdetach(inp);
612 		return ENOMEM;
613 	}
614 	ICMP6_FILTER_SETPASSALL(inp->inp_icmp6filt);
615 	return 0;
616 }
617 
618 int
619 rip6_detach(struct socket *so)
620 {
621 	struct inpcb *inp = sotoinpcb(so);
622 
623 	soassertlocked(so);
624 
625 	if (inp == NULL)
626 		panic("%s", __func__);
627 #ifdef MROUTING
628 	if (so == ip6_mrouter[inp->inp_rtableid])
629 		ip6_mrouter_done(so);
630 #endif
631 	free(inp->inp_icmp6filt, M_PCB, sizeof(struct icmp6_filter));
632 	inp->inp_icmp6filt = NULL;
633 
634 	in_pcbdetach(inp);
635 
636 	return (0);
637 }
638 
639 void
640 rip6_lock(struct socket *so)
641 {
642 	struct inpcb *inp = sotoinpcb(so);
643 
644 	NET_ASSERT_LOCKED();
645 	mtx_enter(&inp->inp_mtx);
646 }
647 
648 void
649 rip6_unlock(struct socket *so)
650 {
651 	struct inpcb *inp = sotoinpcb(so);
652 
653 	NET_ASSERT_LOCKED();
654 	mtx_leave(&inp->inp_mtx);
655 }
656 
657 int
658 rip6_locked(struct socket *so)
659 {
660 	struct inpcb *inp = sotoinpcb(so);
661 
662 	return mtx_owned(&inp->inp_mtx);
663 }
664 
665 int
666 rip6_bind(struct socket *so, struct mbuf *nam, struct proc *p)
667 {
668 	struct inpcb *inp = sotoinpcb(so);
669 	struct sockaddr_in6 *addr;
670 	int error;
671 
672 	soassertlocked(so);
673 
674 	if ((error = in6_nam2sin6(nam, &addr)))
675 		return (error);
676 
677 	/*
678 	 * Make sure to not enter in_pcblookup_local(), local ports
679 	 * are non-sensical for raw sockets.
680 	 */
681 	addr->sin6_port = 0;
682 
683 	if ((error = in6_pcbaddrisavail(inp, addr, 0, p)))
684 		return (error);
685 
686 	mtx_enter(&rawin6pcbtable.inpt_mtx);
687 	inp->inp_laddr6 = addr->sin6_addr;
688 	mtx_leave(&rawin6pcbtable.inpt_mtx);
689 
690 	return (0);
691 }
692 
693 int
694 rip6_connect(struct socket *so, struct mbuf *nam)
695 {
696 	struct inpcb *inp = sotoinpcb(so);
697 	struct sockaddr_in6 *addr;
698 	const struct in6_addr *in6a;
699 	int error;
700 
701 	soassertlocked(so);
702 
703 	if ((error = in6_nam2sin6(nam, &addr)))
704 		return (error);
705 
706 	/* Source address selection. XXX: need pcblookup? */
707 	error = in6_pcbselsrc(&in6a, addr, inp, inp->inp_outputopts6);
708 	if (error)
709 		return (error);
710 
711 	mtx_enter(&rawin6pcbtable.inpt_mtx);
712 	inp->inp_laddr6 = *in6a;
713 	inp->inp_faddr6 = addr->sin6_addr;
714 	mtx_leave(&rawin6pcbtable.inpt_mtx);
715 	soisconnected(so);
716 
717 	return (0);
718 }
719 
720 int
721 rip6_disconnect(struct socket *so)
722 {
723 	struct inpcb *inp = sotoinpcb(so);
724 
725 	soassertlocked(so);
726 
727 	if ((so->so_state & SS_ISCONNECTED) == 0)
728 		return (ENOTCONN);
729 
730 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
731 	mtx_enter(&rawin6pcbtable.inpt_mtx);
732 	inp->inp_faddr6 = in6addr_any;
733 	mtx_leave(&rawin6pcbtable.inpt_mtx);
734 
735 	return (0);
736 }
737 
738 int
739 rip6_shutdown(struct socket *so)
740 {
741 	/*
742 	 * Mark the connection as being incapable of further input.
743 	 */
744 	soassertlocked(so);
745 	socantsendmore(so);
746 	return (0);
747 }
748 
749 int
750 rip6_send(struct socket *so, struct mbuf *m, struct mbuf *nam,
751 	struct mbuf *control)
752 {
753 	struct inpcb *inp = sotoinpcb(so);
754 	struct sockaddr_in6 dst;
755 	int error;
756 
757 	soassertlocked(so);
758 
759 	/*
760 	 * Ship a packet out. The appropriate raw output
761 	 * routine handles any messaging necessary.
762 	 */
763 
764 	/* always copy sockaddr to avoid overwrites */
765 	memset(&dst, 0, sizeof(dst));
766 	dst.sin6_family = AF_INET6;
767 	dst.sin6_len = sizeof(dst);
768 	if (so->so_state & SS_ISCONNECTED) {
769 		if (nam) {
770 			error = EISCONN;
771 			goto out;
772 		}
773 		dst.sin6_addr = inp->inp_faddr6;
774 	} else {
775 		struct sockaddr_in6 *addr6;
776 
777 		if (nam == NULL) {
778 			error = ENOTCONN;
779 			goto out;
780 		}
781 		if ((error = in6_nam2sin6(nam, &addr6)))
782 			goto out;
783 		dst.sin6_addr = addr6->sin6_addr;
784 		dst.sin6_scope_id = addr6->sin6_scope_id;
785 	}
786 	error = rip6_output(m, so, sin6tosa(&dst), control);
787 	control = NULL;
788 	m = NULL;
789 
790 out:
791 	m_freem(control);
792 	m_freem(m);
793 
794 	return (error);
795 }
796 
797 int
798 rip6_sysctl_rip6stat(void *oldp, size_t *oldplen, void *newp)
799 {
800 	struct rip6stat rip6stat;
801 
802 	CTASSERT(sizeof(rip6stat) == rip6s_ncounters * sizeof(uint64_t));
803 	counters_read(rip6counters, (uint64_t *)&rip6stat, rip6s_ncounters,
804 	    NULL);
805 
806 	return (sysctl_rdstruct(oldp, oldplen, newp,
807 	    &rip6stat, sizeof(rip6stat)));
808 }
809 
810 int
811 rip6_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
812     void *newp, size_t newlen)
813 {
814 	/* All sysctl names at this level are terminal. */
815 	if (namelen != 1)
816 		return ENOTDIR;
817 
818 	switch (name[0]) {
819 	case RIPV6CTL_STATS:
820 		return (rip6_sysctl_rip6stat(oldp, oldlenp, newp));
821 	default:
822 		return (EOPNOTSUPP);
823 	}
824 	/* NOTREACHED */
825 }
826