xref: /freebsd/sys/netinet6/raw_ip6.c (revision 3231251399f9e60fc7175b6352eb5291e9296fe8)
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 
70 #include <sys/param.h>
71 #include <sys/errno.h>
72 #include <sys/jail.h>
73 #include <sys/kernel.h>
74 #include <sys/lock.h>
75 #include <sys/malloc.h>
76 #include <sys/mbuf.h>
77 #include <sys/priv.h>
78 #include <sys/proc.h>
79 #include <sys/protosw.h>
80 #include <sys/signalvar.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/sx.h>
84 #include <sys/syslog.h>
85 
86 #include <net/if.h>
87 #include <net/if_var.h>
88 #include <net/if_types.h>
89 #include <net/route.h>
90 #include <net/vnet.h>
91 
92 #include <netinet/in.h>
93 #include <netinet/in_var.h>
94 #include <netinet/in_systm.h>
95 #include <netinet/in_pcb.h>
96 
97 #include <netinet/icmp6.h>
98 #include <netinet/ip6.h>
99 #include <netinet/ip_var.h>
100 #include <netinet6/ip6protosw.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/scope6_var.h>
107 #include <netinet6/send.h>
108 
109 #include <netipsec/ipsec_support.h>
110 
111 #include <machine/stdarg.h>
112 
113 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
114 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
115 
116 /*
117  * Raw interface to IP6 protocol.
118  */
119 
120 VNET_DECLARE(struct inpcbhead, ripcb);
121 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
122 #define	V_ripcb				VNET(ripcb)
123 #define	V_ripcbinfo			VNET(ripcbinfo)
124 
125 extern u_long	rip_sendspace;
126 extern u_long	rip_recvspace;
127 
128 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
129 VNET_PCPUSTAT_SYSINIT(rip6stat);
130 
131 #ifdef VIMAGE
132 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
133 #endif /* VIMAGE */
134 
135 /*
136  * Hooks for multicast routing. They all default to NULL, so leave them not
137  * initialized and rely on BSS being set to 0.
138  */
139 
140 /*
141  * The socket used to communicate with the multicast routing daemon.
142  */
143 VNET_DEFINE(struct socket *, ip6_mrouter);
144 
145 /*
146  * The various mrouter functions.
147  */
148 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
149 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
150 int (*ip6_mrouter_done)(void);
151 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
152 int (*mrt6_ioctl)(u_long, caddr_t);
153 
154 /*
155  * Setup generic address and protocol structures for raw_input routine, then
156  * pass them along with mbuf chain.
157  */
158 int
159 rip6_input(struct mbuf **mp, int *offp, int proto)
160 {
161 	struct ifnet *ifp;
162 	struct mbuf *m = *mp;
163 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
164 	struct inpcb *in6p;
165 	struct inpcb *last = NULL;
166 	struct mbuf *opts = NULL;
167 	struct sockaddr_in6 fromsa;
168 	struct epoch_tracker et;
169 
170 	RIP6STAT_INC(rip6s_ipackets);
171 
172 	init_sin6(&fromsa, m, 0); /* general init */
173 
174 	ifp = m->m_pkthdr.rcvif;
175 
176 	INP_INFO_RLOCK_ET(&V_ripcbinfo, et);
177 	CK_LIST_FOREACH(in6p, &V_ripcb, inp_list) {
178 		/* XXX inp locking */
179 		if ((in6p->inp_vflag & INP_IPV6) == 0)
180 			continue;
181 		if (in6p->inp_ip_p &&
182 		    in6p->inp_ip_p != proto)
183 			continue;
184 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
185 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
186 			continue;
187 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
188 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
189 			continue;
190 		if (last != NULL) {
191 			struct mbuf *n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
192 
193 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
194 			/*
195 			 * Check AH/ESP integrity.
196 			 */
197 			if (IPSEC_ENABLED(ipv6)) {
198 				if (n != NULL &&
199 				    IPSEC_CHECK_POLICY(ipv6, n, last) != 0) {
200 					m_freem(n);
201 					/* Do not inject data into pcb. */
202 					n = NULL;
203 				}
204 			}
205 #endif /* IPSEC */
206 			if (n) {
207 				if (last->inp_flags & INP_CONTROLOPTS ||
208 				    last->inp_socket->so_options & SO_TIMESTAMP)
209 					ip6_savecontrol(last, n, &opts);
210 				/* strip intermediate headers */
211 				m_adj(n, *offp);
212 				if (sbappendaddr(&last->inp_socket->so_rcv,
213 						(struct sockaddr *)&fromsa,
214 						 n, opts) == 0) {
215 					m_freem(n);
216 					if (opts)
217 						m_freem(opts);
218 					RIP6STAT_INC(rip6s_fullsock);
219 				} else
220 					sorwakeup(last->inp_socket);
221 				opts = NULL;
222 			}
223 			INP_RUNLOCK(last);
224 			last = NULL;
225 		}
226 		INP_RLOCK(in6p);
227 		if (__predict_false(in6p->inp_flags2 & INP_FREED))
228 			goto skip_2;
229 		if (jailed_without_vnet(in6p->inp_cred)) {
230 			/*
231 			 * Allow raw socket in jail to receive multicast;
232 			 * assume process had PRIV_NETINET_RAW at attach,
233 			 * and fall through into normal filter path if so.
234 			 */
235 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
236 			    prison_check_ip6(in6p->inp_cred,
237 			    &ip6->ip6_dst) != 0)
238 				goto skip_2;
239 		}
240 		if (in6p->in6p_cksum != -1) {
241 			RIP6STAT_INC(rip6s_isum);
242 			if (in6_cksum(m, proto, *offp,
243 			    m->m_pkthdr.len - *offp)) {
244 				RIP6STAT_INC(rip6s_badsum);
245 				goto skip_2;
246 			}
247 		}
248 		/*
249 		 * If this raw socket has multicast state, and we
250 		 * have received a multicast, check if this socket
251 		 * should receive it, as multicast filtering is now
252 		 * the responsibility of the transport layer.
253 		 */
254 		if (in6p->in6p_moptions &&
255 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
256 			/*
257 			 * If the incoming datagram is for MLD, allow it
258 			 * through unconditionally to the raw socket.
259 			 *
260 			 * Use the M_RTALERT_MLD flag to check for MLD
261 			 * traffic without having to inspect the mbuf chain
262 			 * more deeply, as all MLDv1/v2 host messages MUST
263 			 * contain the Router Alert option.
264 			 *
265 			 * In the case of MLDv1, we may not have explicitly
266 			 * joined the group, and may have set IFF_ALLMULTI
267 			 * on the interface. im6o_mc_filter() may discard
268 			 * control traffic we actually need to see.
269 			 *
270 			 * Userland multicast routing daemons should continue
271 			 * filter the control traffic appropriately.
272 			 */
273 			int blocked;
274 
275 			blocked = MCAST_PASS;
276 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
277 				struct sockaddr_in6 mcaddr;
278 
279 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
280 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
281 				mcaddr.sin6_family = AF_INET6;
282 				mcaddr.sin6_addr = ip6->ip6_dst;
283 
284 				blocked = im6o_mc_filter(in6p->in6p_moptions,
285 				    ifp,
286 				    (struct sockaddr *)&mcaddr,
287 				    (struct sockaddr *)&fromsa);
288 			}
289 			if (blocked != MCAST_PASS) {
290 				IP6STAT_INC(ip6s_notmember);
291 				goto skip_2;
292 			}
293 		}
294 		last = in6p;
295 		continue;
296 skip_2:
297 		INP_RUNLOCK(in6p);
298 	}
299 	INP_INFO_RUNLOCK_ET(&V_ripcbinfo, et);
300 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
301 	/*
302 	 * Check AH/ESP integrity.
303 	 */
304 	if (IPSEC_ENABLED(ipv6) && last != NULL &&
305 	    IPSEC_CHECK_POLICY(ipv6, m, last) != 0) {
306 		m_freem(m);
307 		IP6STAT_DEC(ip6s_delivered);
308 		/* Do not inject data into pcb. */
309 		INP_RUNLOCK(last);
310 	} else
311 #endif /* IPSEC */
312 	if (last != NULL) {
313 		if (last->inp_flags & INP_CONTROLOPTS ||
314 		    last->inp_socket->so_options & SO_TIMESTAMP)
315 			ip6_savecontrol(last, m, &opts);
316 		/* Strip intermediate headers. */
317 		m_adj(m, *offp);
318 		if (sbappendaddr(&last->inp_socket->so_rcv,
319 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
320 			m_freem(m);
321 			if (opts)
322 				m_freem(opts);
323 			RIP6STAT_INC(rip6s_fullsock);
324 		} else
325 			sorwakeup(last->inp_socket);
326 		INP_RUNLOCK(last);
327 	} else {
328 		RIP6STAT_INC(rip6s_nosock);
329 		if (m->m_flags & M_MCAST)
330 			RIP6STAT_INC(rip6s_nosockmcast);
331 		if (proto == IPPROTO_NONE)
332 			m_freem(m);
333 		else
334 			icmp6_error(m, ICMP6_PARAM_PROB,
335 			    ICMP6_PARAMPROB_NEXTHEADER,
336 			    ip6_get_prevhdr(m, *offp));
337 		IP6STAT_DEC(ip6s_delivered);
338 	}
339 	return (IPPROTO_DONE);
340 }
341 
342 void
343 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
344 {
345 	struct ip6ctlparam *ip6cp = NULL;
346 	const struct sockaddr_in6 *sa6_src = NULL;
347 	void *cmdarg;
348 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
349 
350 	if (sa->sa_family != AF_INET6 ||
351 	    sa->sa_len != sizeof(struct sockaddr_in6))
352 		return;
353 
354 	if ((unsigned)cmd >= PRC_NCMDS)
355 		return;
356 	if (PRC_IS_REDIRECT(cmd))
357 		notify = in6_rtchange, d = NULL;
358 	else if (cmd == PRC_HOSTDEAD)
359 		d = NULL;
360 	else if (inet6ctlerrmap[cmd] == 0)
361 		return;
362 
363 	/*
364 	 * If the parameter is from icmp6, decode it.
365 	 */
366 	if (d != NULL) {
367 		ip6cp = (struct ip6ctlparam *)d;
368 		cmdarg = ip6cp->ip6c_cmdarg;
369 		sa6_src = ip6cp->ip6c_src;
370 	} else {
371 		cmdarg = NULL;
372 		sa6_src = &sa6_any;
373 	}
374 
375 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
376 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
377 }
378 
379 /*
380  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
381  * may have setup with control call.
382  */
383 int
384 rip6_output(struct mbuf *m, struct socket *so, ...)
385 {
386 	struct mbuf *control;
387 	struct m_tag *mtag;
388 	struct sockaddr_in6 *dstsock;
389 	struct ip6_hdr *ip6;
390 	struct inpcb *in6p;
391 	u_int	plen = m->m_pkthdr.len;
392 	int error = 0;
393 	struct ip6_pktopts opt, *optp;
394 	struct ifnet *oifp = NULL;
395 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
396 	int scope_ambiguous = 0;
397 	int use_defzone = 0;
398 	int hlim = 0;
399 	struct in6_addr in6a;
400 	va_list ap;
401 
402 	va_start(ap, so);
403 	dstsock = va_arg(ap, struct sockaddr_in6 *);
404 	control = va_arg(ap, struct mbuf *);
405 	va_end(ap);
406 
407 	in6p = sotoinpcb(so);
408 	INP_WLOCK(in6p);
409 
410 	if (control != NULL) {
411 		if ((error = ip6_setpktopts(control, &opt,
412 		    in6p->in6p_outputopts, so->so_cred,
413 		    so->so_proto->pr_protocol)) != 0) {
414 			goto bad;
415 		}
416 		optp = &opt;
417 	} else
418 		optp = in6p->in6p_outputopts;
419 
420 	/*
421 	 * Check and convert scope zone ID into internal form.
422 	 *
423 	 * XXX: we may still need to determine the zone later.
424 	 */
425 	if (!(so->so_state & SS_ISCONNECTED)) {
426 		if (!optp || !optp->ip6po_pktinfo ||
427 		    !optp->ip6po_pktinfo->ipi6_ifindex)
428 			use_defzone = V_ip6_use_defzone;
429 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
430 			scope_ambiguous = 1;
431 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
432 			goto bad;
433 	}
434 
435 	/*
436 	 * For an ICMPv6 packet, we should know its type and code to update
437 	 * statistics.
438 	 */
439 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
440 		struct icmp6_hdr *icmp6;
441 		if (m->m_len < sizeof(struct icmp6_hdr) &&
442 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
443 			error = ENOBUFS;
444 			goto bad;
445 		}
446 		icmp6 = mtod(m, struct icmp6_hdr *);
447 		type = icmp6->icmp6_type;
448 		code = icmp6->icmp6_code;
449 	}
450 
451 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
452 	if (m == NULL) {
453 		error = ENOBUFS;
454 		goto bad;
455 	}
456 	ip6 = mtod(m, struct ip6_hdr *);
457 
458 	/*
459 	 * Source address selection.
460 	 */
461 	error = in6_selectsrc_socket(dstsock, optp, in6p, so->so_cred,
462 	    scope_ambiguous, &in6a, &hlim);
463 
464 	if (error)
465 		goto bad;
466 	error = prison_check_ip6(in6p->inp_cred, &in6a);
467 	if (error != 0)
468 		goto bad;
469 	ip6->ip6_src = in6a;
470 
471 	ip6->ip6_dst = dstsock->sin6_addr;
472 
473 	/*
474 	 * Fill in the rest of the IPv6 header fields.
475 	 */
476 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
477 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
478 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
479 	    (IPV6_VERSION & IPV6_VERSION_MASK);
480 
481 	/*
482 	 * ip6_plen will be filled in ip6_output, so not fill it here.
483 	 */
484 	ip6->ip6_nxt = in6p->inp_ip_p;
485 	ip6->ip6_hlim = hlim;
486 
487 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
488 	    in6p->in6p_cksum != -1) {
489 		struct mbuf *n;
490 		int off;
491 		u_int16_t *p;
492 
493 		/* Compute checksum. */
494 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
495 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
496 		else
497 			off = in6p->in6p_cksum;
498 		if (plen < off + 1) {
499 			error = EINVAL;
500 			goto bad;
501 		}
502 		off += sizeof(struct ip6_hdr);
503 
504 		n = m;
505 		while (n && n->m_len <= off) {
506 			off -= n->m_len;
507 			n = n->m_next;
508 		}
509 		if (!n)
510 			goto bad;
511 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
512 		*p = 0;
513 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
514 	}
515 
516 	/*
517 	 * Send RA/RS messages to user land for protection, before sending
518 	 * them to rtadvd/rtsol.
519 	 */
520 	if ((send_sendso_input_hook != NULL) &&
521 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
522 		switch (type) {
523 		case ND_ROUTER_ADVERT:
524 		case ND_ROUTER_SOLICIT:
525 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
526 				sizeof(unsigned short), M_NOWAIT);
527 			if (mtag == NULL)
528 				goto bad;
529 			m_tag_prepend(m, mtag);
530 		}
531 	}
532 
533 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
534 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
535 		if (oifp)
536 			icmp6_ifoutstat_inc(oifp, type, code);
537 		ICMP6STAT_INC(icp6s_outhist[type]);
538 	} else
539 		RIP6STAT_INC(rip6s_opackets);
540 
541 	goto freectl;
542 
543  bad:
544 	if (m)
545 		m_freem(m);
546 
547  freectl:
548 	if (control != NULL) {
549 		ip6_clearpktopts(&opt, -1);
550 		m_freem(control);
551 	}
552 	INP_WUNLOCK(in6p);
553 	return (error);
554 }
555 
556 /*
557  * Raw IPv6 socket option processing.
558  */
559 int
560 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
561 {
562 	struct inpcb *inp;
563 	int error;
564 
565 	if (sopt->sopt_level == IPPROTO_ICMPV6)
566 		/*
567 		 * XXX: is it better to call icmp6_ctloutput() directly
568 		 * from protosw?
569 		 */
570 		return (icmp6_ctloutput(so, sopt));
571 	else if (sopt->sopt_level != IPPROTO_IPV6) {
572 		if (sopt->sopt_level == SOL_SOCKET &&
573 		    sopt->sopt_name == SO_SETFIB) {
574 			inp = sotoinpcb(so);
575 			INP_WLOCK(inp);
576 			inp->inp_inc.inc_fibnum = so->so_fibnum;
577 			INP_WUNLOCK(inp);
578 			return (0);
579 		}
580 		return (EINVAL);
581 	}
582 
583 	error = 0;
584 
585 	switch (sopt->sopt_dir) {
586 	case SOPT_GET:
587 		switch (sopt->sopt_name) {
588 		case MRT6_INIT:
589 		case MRT6_DONE:
590 		case MRT6_ADD_MIF:
591 		case MRT6_DEL_MIF:
592 		case MRT6_ADD_MFC:
593 		case MRT6_DEL_MFC:
594 		case MRT6_PIM:
595 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
596 			    EOPNOTSUPP;
597 			break;
598 		case IPV6_CHECKSUM:
599 			error = ip6_raw_ctloutput(so, sopt);
600 			break;
601 		default:
602 			error = ip6_ctloutput(so, sopt);
603 			break;
604 		}
605 		break;
606 
607 	case SOPT_SET:
608 		switch (sopt->sopt_name) {
609 		case MRT6_INIT:
610 		case MRT6_DONE:
611 		case MRT6_ADD_MIF:
612 		case MRT6_DEL_MIF:
613 		case MRT6_ADD_MFC:
614 		case MRT6_DEL_MFC:
615 		case MRT6_PIM:
616 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
617 			    EOPNOTSUPP;
618 			break;
619 		case IPV6_CHECKSUM:
620 			error = ip6_raw_ctloutput(so, sopt);
621 			break;
622 		default:
623 			error = ip6_ctloutput(so, sopt);
624 			break;
625 		}
626 		break;
627 	}
628 
629 	return (error);
630 }
631 
632 static int
633 rip6_attach(struct socket *so, int proto, struct thread *td)
634 {
635 	struct inpcb *inp;
636 	struct icmp6_filter *filter;
637 	int error;
638 
639 	inp = sotoinpcb(so);
640 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
641 
642 	error = priv_check(td, PRIV_NETINET_RAW);
643 	if (error)
644 		return (error);
645 	error = soreserve(so, rip_sendspace, rip_recvspace);
646 	if (error)
647 		return (error);
648 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
649 	if (filter == NULL)
650 		return (ENOMEM);
651 	INP_INFO_WLOCK(&V_ripcbinfo);
652 	error = in_pcballoc(so, &V_ripcbinfo);
653 	if (error) {
654 		INP_INFO_WUNLOCK(&V_ripcbinfo);
655 		free(filter, M_PCB);
656 		return (error);
657 	}
658 	inp = (struct inpcb *)so->so_pcb;
659 	INP_INFO_WUNLOCK(&V_ripcbinfo);
660 	inp->inp_vflag |= INP_IPV6;
661 	inp->inp_ip_p = (long)proto;
662 	inp->in6p_hops = -1;	/* use kernel default */
663 	inp->in6p_cksum = -1;
664 	inp->in6p_icmp6filt = filter;
665 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
666 	INP_WUNLOCK(inp);
667 	return (0);
668 }
669 
670 static void
671 rip6_detach(struct socket *so)
672 {
673 	struct inpcb *inp;
674 
675 	inp = sotoinpcb(so);
676 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
677 
678 	if (so == V_ip6_mrouter && ip6_mrouter_done)
679 		ip6_mrouter_done();
680 	/* xxx: RSVP */
681 	INP_INFO_WLOCK(&V_ripcbinfo);
682 	INP_WLOCK(inp);
683 	free(inp->in6p_icmp6filt, M_PCB);
684 	in_pcbdetach(inp);
685 	in_pcbfree(inp);
686 	INP_INFO_WUNLOCK(&V_ripcbinfo);
687 }
688 
689 /* XXXRW: This can't ever be called. */
690 static void
691 rip6_abort(struct socket *so)
692 {
693 	struct inpcb *inp;
694 
695 	inp = sotoinpcb(so);
696 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
697 
698 	soisdisconnected(so);
699 }
700 
701 static void
702 rip6_close(struct socket *so)
703 {
704 	struct inpcb *inp;
705 
706 	inp = sotoinpcb(so);
707 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
708 
709 	soisdisconnected(so);
710 }
711 
712 static int
713 rip6_disconnect(struct socket *so)
714 {
715 	struct inpcb *inp;
716 
717 	inp = sotoinpcb(so);
718 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
719 
720 	if ((so->so_state & SS_ISCONNECTED) == 0)
721 		return (ENOTCONN);
722 	inp->in6p_faddr = in6addr_any;
723 	rip6_abort(so);
724 	return (0);
725 }
726 
727 static int
728 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
729 {
730 	struct epoch_tracker et;
731 	struct inpcb *inp;
732 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
733 	struct ifaddr *ifa = NULL;
734 	int error = 0;
735 
736 	inp = sotoinpcb(so);
737 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
738 
739 	if (nam->sa_len != sizeof(*addr))
740 		return (EINVAL);
741 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
742 		return (error);
743 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
744 		return (EADDRNOTAVAIL);
745 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
746 		return (error);
747 
748 	NET_EPOCH_ENTER(et);
749 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
750 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
751 		NET_EPOCH_EXIT(et);
752 		return (EADDRNOTAVAIL);
753 	}
754 	if (ifa != NULL &&
755 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
756 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
757 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
758 		NET_EPOCH_EXIT(et);
759 		return (EADDRNOTAVAIL);
760 	}
761 	NET_EPOCH_EXIT(et);
762 	INP_INFO_WLOCK(&V_ripcbinfo);
763 	INP_WLOCK(inp);
764 	inp->in6p_laddr = addr->sin6_addr;
765 	INP_WUNLOCK(inp);
766 	INP_INFO_WUNLOCK(&V_ripcbinfo);
767 	return (0);
768 }
769 
770 static int
771 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
772 {
773 	struct inpcb *inp;
774 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
775 	struct in6_addr in6a;
776 	int error = 0, scope_ambiguous = 0;
777 
778 	inp = sotoinpcb(so);
779 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
780 
781 	if (nam->sa_len != sizeof(*addr))
782 		return (EINVAL);
783 	if (CK_STAILQ_EMPTY(&V_ifnet))
784 		return (EADDRNOTAVAIL);
785 	if (addr->sin6_family != AF_INET6)
786 		return (EAFNOSUPPORT);
787 
788 	/*
789 	 * Application should provide a proper zone ID or the use of default
790 	 * zone IDs should be enabled.  Unfortunately, some applications do
791 	 * not behave as it should, so we need a workaround.  Even if an
792 	 * appropriate ID is not determined, we'll see if we can determine
793 	 * the outgoing interface.  If we can, determine the zone ID based on
794 	 * the interface below.
795 	 */
796 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
797 		scope_ambiguous = 1;
798 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
799 		return (error);
800 
801 	INP_INFO_WLOCK(&V_ripcbinfo);
802 	INP_WLOCK(inp);
803 	/* Source address selection. XXX: need pcblookup? */
804 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
805 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
806 	if (error) {
807 		INP_WUNLOCK(inp);
808 		INP_INFO_WUNLOCK(&V_ripcbinfo);
809 		return (error);
810 	}
811 
812 	inp->in6p_faddr = addr->sin6_addr;
813 	inp->in6p_laddr = in6a;
814 	soisconnected(so);
815 	INP_WUNLOCK(inp);
816 	INP_INFO_WUNLOCK(&V_ripcbinfo);
817 	return (0);
818 }
819 
820 static int
821 rip6_shutdown(struct socket *so)
822 {
823 	struct inpcb *inp;
824 
825 	inp = sotoinpcb(so);
826 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
827 
828 	INP_WLOCK(inp);
829 	socantsendmore(so);
830 	INP_WUNLOCK(inp);
831 	return (0);
832 }
833 
834 static int
835 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
836     struct mbuf *control, struct thread *td)
837 {
838 	struct inpcb *inp;
839 	struct sockaddr_in6 tmp;
840 	struct sockaddr_in6 *dst;
841 	int ret;
842 
843 	inp = sotoinpcb(so);
844 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
845 
846 	/* Always copy sockaddr to avoid overwrites. */
847 	/* Unlocked read. */
848 	if (so->so_state & SS_ISCONNECTED) {
849 		if (nam) {
850 			m_freem(m);
851 			return (EISCONN);
852 		}
853 		/* XXX */
854 		bzero(&tmp, sizeof(tmp));
855 		tmp.sin6_family = AF_INET6;
856 		tmp.sin6_len = sizeof(struct sockaddr_in6);
857 		INP_RLOCK(inp);
858 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
859 		    sizeof(struct in6_addr));
860 		INP_RUNLOCK(inp);
861 		dst = &tmp;
862 	} else {
863 		if (nam == NULL) {
864 			m_freem(m);
865 			return (ENOTCONN);
866 		}
867 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
868 			m_freem(m);
869 			return (EINVAL);
870 		}
871 		tmp = *(struct sockaddr_in6 *)nam;
872 		dst = &tmp;
873 
874 		if (dst->sin6_family == AF_UNSPEC) {
875 			/*
876 			 * XXX: we allow this case for backward
877 			 * compatibility to buggy applications that
878 			 * rely on old (and wrong) kernel behavior.
879 			 */
880 			log(LOG_INFO, "rip6 SEND: address family is "
881 			    "unspec. Assume AF_INET6\n");
882 			dst->sin6_family = AF_INET6;
883 		} else if (dst->sin6_family != AF_INET6) {
884 			m_freem(m);
885 			return(EAFNOSUPPORT);
886 		}
887 	}
888 	ret = rip6_output(m, so, dst, control);
889 	return (ret);
890 }
891 
892 struct pr_usrreqs rip6_usrreqs = {
893 	.pru_abort =		rip6_abort,
894 	.pru_attach =		rip6_attach,
895 	.pru_bind =		rip6_bind,
896 	.pru_connect =		rip6_connect,
897 	.pru_control =		in6_control,
898 	.pru_detach =		rip6_detach,
899 	.pru_disconnect =	rip6_disconnect,
900 	.pru_peeraddr =		in6_getpeeraddr,
901 	.pru_send =		rip6_send,
902 	.pru_shutdown =		rip6_shutdown,
903 	.pru_sockaddr =		in6_getsockaddr,
904 	.pru_close =		rip6_close,
905 };
906