xref: /freebsd/sys/netinet6/raw_ip6.c (revision d5b0e70f7e04d971691517ce1304d86a1e367e2e)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /*-
33  * Copyright (c) 1982, 1986, 1988, 1993
34  *	The Regents of the University of California.
35  * 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 __FBSDID("$FreeBSD$");
66 
67 #include "opt_ipsec.h"
68 #include "opt_inet6.h"
69 #include "opt_route.h"
70 
71 #include <sys/param.h>
72 #include <sys/errno.h>
73 #include <sys/jail.h>
74 #include <sys/kernel.h>
75 #include <sys/lock.h>
76 #include <sys/malloc.h>
77 #include <sys/mbuf.h>
78 #include <sys/priv.h>
79 #include <sys/proc.h>
80 #include <sys/protosw.h>
81 #include <sys/signalvar.h>
82 #include <sys/socket.h>
83 #include <sys/socketvar.h>
84 #include <sys/sx.h>
85 #include <sys/syslog.h>
86 
87 #include <net/if.h>
88 #include <net/if_var.h>
89 #include <net/if_types.h>
90 #include <net/route.h>
91 #include <net/vnet.h>
92 
93 #include <netinet/in.h>
94 #include <netinet/in_var.h>
95 #include <netinet/in_systm.h>
96 #include <netinet/in_pcb.h>
97 
98 #include <netinet/icmp6.h>
99 #include <netinet/ip6.h>
100 #include <netinet/ip_var.h>
101 #include <netinet6/ip6_mroute.h>
102 #include <netinet6/in6_pcb.h>
103 #include <netinet6/ip6_var.h>
104 #include <netinet6/nd6.h>
105 #include <netinet6/raw_ip6.h>
106 #include <netinet6/in6_fib.h>
107 #include <netinet6/scope6_var.h>
108 #include <netinet6/send.h>
109 
110 #include <netipsec/ipsec_support.h>
111 
112 #include <machine/stdarg.h>
113 
114 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
115 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
116 
117 /*
118  * Raw interface to IP6 protocol.
119  */
120 
121 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
122 #define	V_ripcbinfo			VNET(ripcbinfo)
123 
124 extern u_long	rip_sendspace;
125 extern u_long	rip_recvspace;
126 
127 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
128 VNET_PCPUSTAT_SYSINIT(rip6stat);
129 
130 #ifdef VIMAGE
131 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
132 #endif /* VIMAGE */
133 
134 /*
135  * Hooks for multicast routing. They all default to NULL, so leave them not
136  * initialized and rely on BSS being set to 0.
137  */
138 
139 /*
140  * The socket used to communicate with the multicast routing daemon.
141  */
142 VNET_DEFINE(struct socket *, ip6_mrouter);
143 
144 /*
145  * The various mrouter functions.
146  */
147 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
148 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
149 int (*ip6_mrouter_done)(void);
150 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
151 int (*mrt6_ioctl)(u_long, caddr_t);
152 
153 struct rip6_inp_match_ctx {
154 	struct ip6_hdr *ip6;
155 	int proto;
156 };
157 
158 static bool
159 rip6_inp_match(const struct inpcb *inp, void *v)
160 {
161 	struct rip6_inp_match_ctx *c = v;
162 	struct ip6_hdr *ip6 = c->ip6;
163 	int proto = c->proto;
164 
165 	/* XXX inp locking */
166 	if ((inp->inp_vflag & INP_IPV6) == 0)
167 		return (false);
168 	if (inp->inp_ip_p && inp->inp_ip_p != proto)
169 		return (false);
170 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
171 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
172 		return (false);
173 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
174 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src))
175 		return (false);
176 
177 	return (true);
178 }
179 
180 /*
181  * Setup generic address and protocol structures for raw_input routine, then
182  * pass them along with mbuf chain.
183  */
184 int
185 rip6_input(struct mbuf **mp, int *offp, int proto)
186 {
187 	struct ifnet *ifp;
188 	struct mbuf *n, *m = *mp;
189 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
190 	struct inpcb *inp;
191 	struct mbuf *opts = NULL;
192 	struct sockaddr_in6 fromsa;
193 	struct rip6_inp_match_ctx ctx = { .ip6 = ip6, .proto = proto };
194 	struct inpcb_iterator inpi = INP_ITERATOR(&V_ripcbinfo,
195 	    INPLOOKUP_RLOCKPCB, rip6_inp_match, &ctx);
196 	int delivered = 0;
197 
198 	NET_EPOCH_ASSERT();
199 
200 	RIP6STAT_INC(rip6s_ipackets);
201 
202 	init_sin6(&fromsa, m, 0); /* general init */
203 
204 	ifp = m->m_pkthdr.rcvif;
205 
206 	while ((inp = inp_next(&inpi)) != NULL) {
207 		INP_RLOCK_ASSERT(inp);
208 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
209 		/*
210 		 * Check AH/ESP integrity.
211 		 */
212 		if (IPSEC_ENABLED(ipv6) &&
213 		    IPSEC_CHECK_POLICY(ipv6, m, inp) != 0) {
214 			/* Do not inject data into pcb. */
215 			continue;
216 		}
217 #endif /* IPSEC */
218 		if (jailed_without_vnet(inp->inp_cred) &&
219 		    !IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
220 		    prison_check_ip6(inp->inp_cred, &ip6->ip6_dst) != 0)
221 			/*
222 			 * Allow raw socket in jail to receive multicast;
223 			 * assume process had PRIV_NETINET_RAW at attach,
224 			 * and fall through into normal filter path if so.
225 			 */
226 			continue;
227 		if (inp->in6p_cksum != -1) {
228 			RIP6STAT_INC(rip6s_isum);
229 			if (m->m_pkthdr.len - (*offp + inp->in6p_cksum) < 2 ||
230 			    in6_cksum(m, proto, *offp,
231 			    m->m_pkthdr.len - *offp)) {
232 				RIP6STAT_INC(rip6s_badsum);
233 				/*
234 				 * Drop the received message, don't send an
235 				 * ICMP6 message. Set proto to IPPROTO_NONE
236 				 * to achieve that.
237 				 */
238 				INP_RUNLOCK(inp);
239 				proto = IPPROTO_NONE;
240 				break;
241 			}
242 		}
243 		/*
244 		 * If this raw socket has multicast state, and we
245 		 * have received a multicast, check if this socket
246 		 * should receive it, as multicast filtering is now
247 		 * the responsibility of the transport layer.
248 		 */
249 		if (inp->in6p_moptions &&
250 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
251 			/*
252 			 * If the incoming datagram is for MLD, allow it
253 			 * through unconditionally to the raw socket.
254 			 *
255 			 * Use the M_RTALERT_MLD flag to check for MLD
256 			 * traffic without having to inspect the mbuf chain
257 			 * more deeply, as all MLDv1/v2 host messages MUST
258 			 * contain the Router Alert option.
259 			 *
260 			 * In the case of MLDv1, we may not have explicitly
261 			 * joined the group, and may have set IFF_ALLMULTI
262 			 * on the interface. im6o_mc_filter() may discard
263 			 * control traffic we actually need to see.
264 			 *
265 			 * Userland multicast routing daemons should continue
266 			 * filter the control traffic appropriately.
267 			 */
268 			int blocked;
269 
270 			blocked = MCAST_PASS;
271 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
272 				struct sockaddr_in6 mcaddr;
273 
274 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
275 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
276 				mcaddr.sin6_family = AF_INET6;
277 				mcaddr.sin6_addr = ip6->ip6_dst;
278 
279 				blocked = im6o_mc_filter(inp->in6p_moptions,
280 				    ifp,
281 				    (struct sockaddr *)&mcaddr,
282 				    (struct sockaddr *)&fromsa);
283 			}
284 			if (blocked != MCAST_PASS) {
285 				IP6STAT_INC(ip6s_notmember);
286 				continue;
287 			}
288 		}
289 		if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL)
290 			continue;
291 		if (inp->inp_flags & INP_CONTROLOPTS ||
292 		    inp->inp_socket->so_options & SO_TIMESTAMP)
293 			ip6_savecontrol(inp, n, &opts);
294 		/* strip intermediate headers */
295 		m_adj(n, *offp);
296 		if (sbappendaddr(&inp->inp_socket->so_rcv,
297 		    (struct sockaddr *)&fromsa, n, opts) == 0) {
298 			soroverflow(inp->inp_socket);
299 			m_freem(n);
300 			if (opts)
301 				m_freem(opts);
302 			RIP6STAT_INC(rip6s_fullsock);
303 		} else {
304 			sorwakeup(inp->inp_socket);
305 			delivered++;
306 		}
307 		opts = NULL;
308 	}
309 	if (delivered == 0) {
310 		RIP6STAT_INC(rip6s_nosock);
311 		if (m->m_flags & M_MCAST)
312 			RIP6STAT_INC(rip6s_nosockmcast);
313 		if (proto == IPPROTO_NONE)
314 			m_freem(m);
315 		else
316 			icmp6_error(m, ICMP6_PARAM_PROB,
317 			    ICMP6_PARAMPROB_NEXTHEADER,
318 			    ip6_get_prevhdr(m, *offp));
319 		IP6STAT_DEC(ip6s_delivered);
320 	} else
321 		m_freem(m);
322 	return (IPPROTO_DONE);
323 }
324 
325 void
326 rip6_ctlinput(struct ip6ctlparam *ip6cp)
327 {
328 	int errno;
329 
330 	if ((errno = icmp6_errmap(ip6cp->ip6c_icmp6)) != 0)
331 		in6_pcbnotify(&V_ripcbinfo, ip6cp->ip6c_finaldst, 0,
332 		    ip6cp->ip6c_src, 0, errno, ip6cp->ip6c_cmdarg,
333 		    in6_rtchange);
334 }
335 
336 /*
337  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
338  * may have setup with control call.
339  */
340 static int
341 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
342     struct mbuf *control, struct thread *td)
343 {
344 	struct epoch_tracker et;
345 	struct inpcb *inp;
346 	struct sockaddr_in6 tmp, *dstsock;
347 	struct m_tag *mtag;
348 	struct ip6_hdr *ip6;
349 	u_int	plen = m->m_pkthdr.len;
350 	struct ip6_pktopts opt, *optp;
351 	struct ifnet *oifp = NULL;
352 	int error;
353 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
354 	int scope_ambiguous = 0;
355 	int use_defzone = 0;
356 	int hlim = 0;
357 	struct in6_addr in6a;
358 
359 	inp = sotoinpcb(so);
360 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
361 
362 	/* Always copy sockaddr to avoid overwrites. */
363 	/* Unlocked read. */
364 	if (so->so_state & SS_ISCONNECTED) {
365 		if (nam) {
366 			error = EISCONN;
367 			goto release;
368 		}
369 		tmp = (struct sockaddr_in6 ){
370 			.sin6_family = AF_INET6,
371 			.sin6_len = sizeof(struct sockaddr_in6),
372 		};
373 		INP_RLOCK(inp);
374 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
375 		    sizeof(struct in6_addr));
376 		INP_RUNLOCK(inp);
377 		dstsock = &tmp;
378 	} else {
379 		if (nam == NULL)
380 			error = ENOTCONN;
381 		else if (nam->sa_family != AF_INET6)
382 			error = EAFNOSUPPORT;
383 		else if (nam->sa_len != sizeof(struct sockaddr_in6))
384 			error = EINVAL;
385 		else
386 			error = 0;
387 		if (error != 0)
388 			goto release;
389 		dstsock = (struct sockaddr_in6 *)nam;
390 		if (dstsock->sin6_family != AF_INET6) {
391 			error = EAFNOSUPPORT;
392 			goto release;
393 		}
394 	}
395 
396 	INP_WLOCK(inp);
397 
398 	if (control != NULL) {
399 		NET_EPOCH_ENTER(et);
400 		error = ip6_setpktopts(control, &opt, inp->in6p_outputopts,
401 		    so->so_cred, inp->inp_ip_p);
402 		NET_EPOCH_EXIT(et);
403 
404 		if (error != 0) {
405 			goto bad;
406 		}
407 		optp = &opt;
408 	} else
409 		optp = inp->in6p_outputopts;
410 
411 	/*
412 	 * Check and convert scope zone ID into internal form.
413 	 *
414 	 * XXX: we may still need to determine the zone later.
415 	 */
416 	if (!(so->so_state & SS_ISCONNECTED)) {
417 		if (!optp || !optp->ip6po_pktinfo ||
418 		    !optp->ip6po_pktinfo->ipi6_ifindex)
419 			use_defzone = V_ip6_use_defzone;
420 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
421 			scope_ambiguous = 1;
422 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
423 			goto bad;
424 	}
425 
426 	/*
427 	 * For an ICMPv6 packet, we should know its type and code to update
428 	 * statistics.
429 	 */
430 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
431 		struct icmp6_hdr *icmp6;
432 		if (m->m_len < sizeof(struct icmp6_hdr) &&
433 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
434 			error = ENOBUFS;
435 			goto bad;
436 		}
437 		icmp6 = mtod(m, struct icmp6_hdr *);
438 		type = icmp6->icmp6_type;
439 		code = icmp6->icmp6_code;
440 	}
441 
442 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
443 	if (m == NULL) {
444 		error = ENOBUFS;
445 		goto bad;
446 	}
447 	ip6 = mtod(m, struct ip6_hdr *);
448 
449 #ifdef ROUTE_MPATH
450 	if (CALC_FLOWID_OUTBOUND) {
451 		uint32_t hash_type, hash_val;
452 
453 		hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
454 		    &dstsock->sin6_addr, 0, 0, inp->inp_ip_p, &hash_type);
455 		inp->inp_flowid = hash_val;
456 		inp->inp_flowtype = hash_type;
457 	}
458 #endif
459 	/*
460 	 * Source address selection.
461 	 */
462 	NET_EPOCH_ENTER(et);
463 	error = in6_selectsrc_socket(dstsock, optp, inp, so->so_cred,
464 	    scope_ambiguous, &in6a, &hlim);
465 	NET_EPOCH_EXIT(et);
466 
467 	if (error)
468 		goto bad;
469 	error = prison_check_ip6(inp->inp_cred, &in6a);
470 	if (error != 0)
471 		goto bad;
472 	ip6->ip6_src = in6a;
473 
474 	ip6->ip6_dst = dstsock->sin6_addr;
475 
476 	/*
477 	 * Fill in the rest of the IPv6 header fields.
478 	 */
479 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
480 	    (inp->inp_flow & IPV6_FLOWINFO_MASK);
481 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
482 	    (IPV6_VERSION & IPV6_VERSION_MASK);
483 
484 	/*
485 	 * ip6_plen will be filled in ip6_output, so not fill it here.
486 	 */
487 	ip6->ip6_nxt = inp->inp_ip_p;
488 	ip6->ip6_hlim = hlim;
489 
490 	if (inp->inp_ip_p == IPPROTO_ICMPV6 || inp->in6p_cksum != -1) {
491 		struct mbuf *n;
492 		int off;
493 		u_int16_t *p;
494 
495 		/* Compute checksum. */
496 		if (inp->inp_ip_p == IPPROTO_ICMPV6)
497 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
498 		else
499 			off = inp->in6p_cksum;
500 		if (plen < off + 2) {
501 			error = EINVAL;
502 			goto bad;
503 		}
504 		off += sizeof(struct ip6_hdr);
505 
506 		n = m;
507 		while (n && n->m_len <= off) {
508 			off -= n->m_len;
509 			n = n->m_next;
510 		}
511 		if (!n)
512 			goto bad;
513 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
514 		*p = 0;
515 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
516 	}
517 
518 	/*
519 	 * Send RA/RS messages to user land for protection, before sending
520 	 * them to rtadvd/rtsol.
521 	 */
522 	if ((send_sendso_input_hook != NULL) &&
523 	    inp->inp_ip_p == IPPROTO_ICMPV6) {
524 		switch (type) {
525 		case ND_ROUTER_ADVERT:
526 		case ND_ROUTER_SOLICIT:
527 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
528 				sizeof(unsigned short), M_NOWAIT);
529 			if (mtag == NULL)
530 				goto bad;
531 			m_tag_prepend(m, mtag);
532 		}
533 	}
534 
535 	NET_EPOCH_ENTER(et);
536 	error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
537 	NET_EPOCH_EXIT(et);
538 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
539 		if (oifp)
540 			icmp6_ifoutstat_inc(oifp, type, code);
541 		ICMP6STAT_INC(icp6s_outhist[type]);
542 	} else
543 		RIP6STAT_INC(rip6s_opackets);
544 
545 	goto freectl;
546 
547  bad:
548 	if (m)
549 		m_freem(m);
550 
551  freectl:
552 	if (control != NULL) {
553 		ip6_clearpktopts(&opt, -1);
554 		m_freem(control);
555 	}
556 	INP_WUNLOCK(inp);
557 	return (error);
558 
559 release:
560 	if (control != NULL)
561 		m_freem(control);
562 	m_freem(m);
563 	return (error);
564 }
565 
566 /*
567  * Raw IPv6 socket option processing.
568  */
569 int
570 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
571 {
572 	struct inpcb *inp = sotoinpcb(so);
573 	int error;
574 
575 	if (sopt->sopt_level == IPPROTO_ICMPV6)
576 		/*
577 		 * XXX: is it better to call icmp6_ctloutput() directly
578 		 * from protosw?
579 		 */
580 		return (icmp6_ctloutput(so, sopt));
581 	else if (sopt->sopt_level != IPPROTO_IPV6) {
582 		if (sopt->sopt_level == SOL_SOCKET &&
583 		    sopt->sopt_name == SO_SETFIB) {
584 			INP_WLOCK(inp);
585 			inp->inp_inc.inc_fibnum = so->so_fibnum;
586 			INP_WUNLOCK(inp);
587 			return (0);
588 		}
589 		return (EINVAL);
590 	}
591 
592 	error = 0;
593 
594 	switch (sopt->sopt_dir) {
595 	case SOPT_GET:
596 		switch (sopt->sopt_name) {
597 		case MRT6_INIT:
598 		case MRT6_DONE:
599 		case MRT6_ADD_MIF:
600 		case MRT6_DEL_MIF:
601 		case MRT6_ADD_MFC:
602 		case MRT6_DEL_MFC:
603 		case MRT6_PIM:
604 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
605 				return (EOPNOTSUPP);
606 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
607 			    EOPNOTSUPP;
608 			break;
609 		case IPV6_CHECKSUM:
610 			error = ip6_raw_ctloutput(so, sopt);
611 			break;
612 		default:
613 			error = ip6_ctloutput(so, sopt);
614 			break;
615 		}
616 		break;
617 
618 	case SOPT_SET:
619 		switch (sopt->sopt_name) {
620 		case MRT6_INIT:
621 		case MRT6_DONE:
622 		case MRT6_ADD_MIF:
623 		case MRT6_DEL_MIF:
624 		case MRT6_ADD_MFC:
625 		case MRT6_DEL_MFC:
626 		case MRT6_PIM:
627 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
628 				return (EOPNOTSUPP);
629 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
630 			    EOPNOTSUPP;
631 			break;
632 		case IPV6_CHECKSUM:
633 			error = ip6_raw_ctloutput(so, sopt);
634 			break;
635 		default:
636 			error = ip6_ctloutput(so, sopt);
637 			break;
638 		}
639 		break;
640 	}
641 
642 	return (error);
643 }
644 
645 static int
646 rip6_attach(struct socket *so, int proto, struct thread *td)
647 {
648 	struct inpcb *inp;
649 	struct icmp6_filter *filter;
650 	int error;
651 
652 	inp = sotoinpcb(so);
653 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
654 
655 	error = priv_check(td, PRIV_NETINET_RAW);
656 	if (error)
657 		return (error);
658 	if (proto >= IPPROTO_MAX || proto < 0)
659 		return (EPROTONOSUPPORT);
660 	error = soreserve(so, rip_sendspace, rip_recvspace);
661 	if (error)
662 		return (error);
663 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
664 	if (filter == NULL)
665 		return (ENOMEM);
666 	error = in_pcballoc(so, &V_ripcbinfo);
667 	if (error) {
668 		free(filter, M_PCB);
669 		return (error);
670 	}
671 	inp = (struct inpcb *)so->so_pcb;
672 	inp->inp_ip_p = proto;
673 	inp->in6p_cksum = -1;
674 	inp->in6p_icmp6filt = filter;
675 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
676 	INP_WUNLOCK(inp);
677 	return (0);
678 }
679 
680 static void
681 rip6_detach(struct socket *so)
682 {
683 	struct inpcb *inp;
684 
685 	inp = sotoinpcb(so);
686 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
687 
688 	if (so == V_ip6_mrouter && ip6_mrouter_done)
689 		ip6_mrouter_done();
690 	/* xxx: RSVP */
691 	INP_WLOCK(inp);
692 	free(inp->in6p_icmp6filt, M_PCB);
693 	in_pcbdetach(inp);
694 	in_pcbfree(inp);
695 }
696 
697 /* XXXRW: This can't ever be called. */
698 static void
699 rip6_abort(struct socket *so)
700 {
701 	struct inpcb *inp __diagused;
702 
703 	inp = sotoinpcb(so);
704 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
705 
706 	soisdisconnected(so);
707 }
708 
709 static void
710 rip6_close(struct socket *so)
711 {
712 	struct inpcb *inp __diagused;
713 
714 	inp = sotoinpcb(so);
715 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
716 
717 	soisdisconnected(so);
718 }
719 
720 static int
721 rip6_disconnect(struct socket *so)
722 {
723 	struct inpcb *inp;
724 
725 	inp = sotoinpcb(so);
726 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
727 
728 	if ((so->so_state & SS_ISCONNECTED) == 0)
729 		return (ENOTCONN);
730 	inp->in6p_faddr = in6addr_any;
731 	rip6_abort(so);
732 	return (0);
733 }
734 
735 static int
736 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
737 {
738 	struct epoch_tracker et;
739 	struct inpcb *inp;
740 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
741 	struct ifaddr *ifa = NULL;
742 	int error = 0;
743 
744 	inp = sotoinpcb(so);
745 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
746 
747 	if (nam->sa_family != AF_INET6)
748 		return (EAFNOSUPPORT);
749 	if (nam->sa_len != sizeof(*addr))
750 		return (EINVAL);
751 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
752 		return (error);
753 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
754 		return (EADDRNOTAVAIL);
755 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
756 		return (error);
757 
758 	NET_EPOCH_ENTER(et);
759 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
760 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
761 		NET_EPOCH_EXIT(et);
762 		return (EADDRNOTAVAIL);
763 	}
764 	if (ifa != NULL &&
765 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
766 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
767 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
768 		NET_EPOCH_EXIT(et);
769 		return (EADDRNOTAVAIL);
770 	}
771 	NET_EPOCH_EXIT(et);
772 	INP_WLOCK(inp);
773 	INP_INFO_WLOCK(&V_ripcbinfo);
774 	inp->in6p_laddr = addr->sin6_addr;
775 	INP_INFO_WUNLOCK(&V_ripcbinfo);
776 	INP_WUNLOCK(inp);
777 	return (0);
778 }
779 
780 static int
781 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
782 {
783 	struct inpcb *inp;
784 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
785 	struct in6_addr in6a;
786 	struct epoch_tracker et;
787 	int error = 0, scope_ambiguous = 0;
788 
789 	inp = sotoinpcb(so);
790 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
791 
792 	if (nam->sa_len != sizeof(*addr))
793 		return (EINVAL);
794 	if (CK_STAILQ_EMPTY(&V_ifnet))
795 		return (EADDRNOTAVAIL);
796 	if (addr->sin6_family != AF_INET6)
797 		return (EAFNOSUPPORT);
798 
799 	/*
800 	 * Application should provide a proper zone ID or the use of default
801 	 * zone IDs should be enabled.  Unfortunately, some applications do
802 	 * not behave as it should, so we need a workaround.  Even if an
803 	 * appropriate ID is not determined, we'll see if we can determine
804 	 * the outgoing interface.  If we can, determine the zone ID based on
805 	 * the interface below.
806 	 */
807 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
808 		scope_ambiguous = 1;
809 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
810 		return (error);
811 
812 	INP_WLOCK(inp);
813 	INP_INFO_WLOCK(&V_ripcbinfo);
814 	/* Source address selection. XXX: need pcblookup? */
815 	NET_EPOCH_ENTER(et);
816 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
817 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
818 	NET_EPOCH_EXIT(et);
819 	if (error) {
820 		INP_INFO_WUNLOCK(&V_ripcbinfo);
821 		INP_WUNLOCK(inp);
822 		return (error);
823 	}
824 
825 	inp->in6p_faddr = addr->sin6_addr;
826 	inp->in6p_laddr = in6a;
827 	soisconnected(so);
828 	INP_INFO_WUNLOCK(&V_ripcbinfo);
829 	INP_WUNLOCK(inp);
830 	return (0);
831 }
832 
833 static int
834 rip6_shutdown(struct socket *so)
835 {
836 	struct inpcb *inp;
837 
838 	inp = sotoinpcb(so);
839 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
840 
841 	INP_WLOCK(inp);
842 	socantsendmore(so);
843 	INP_WUNLOCK(inp);
844 	return (0);
845 }
846 
847 struct protosw rip6_protosw = {
848 	.pr_type =		SOCK_RAW,
849 	.pr_flags =		PR_ATOMIC|PR_ADDR,
850 	.pr_ctloutput =		rip6_ctloutput,
851 	.pr_abort =		rip6_abort,
852 	.pr_attach =		rip6_attach,
853 	.pr_bind =		rip6_bind,
854 	.pr_connect =		rip6_connect,
855 	.pr_control =		in6_control,
856 	.pr_detach =		rip6_detach,
857 	.pr_disconnect =	rip6_disconnect,
858 	.pr_peeraddr =		in6_getpeeraddr,
859 	.pr_send =		rip6_send,
860 	.pr_shutdown =		rip6_shutdown,
861 	.pr_sockaddr =		in6_getsockaddr,
862 	.pr_close =		rip6_close
863 };
864