xref: /freebsd/sys/netinet6/raw_ip6.c (revision b0d29bc47dba79f6f38e67eabadfb4b32ffd9390)
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 *inp;
165 	struct inpcb *last = NULL;
166 	struct mbuf *opts = NULL;
167 	struct sockaddr_in6 fromsa;
168 
169 	NET_EPOCH_ASSERT();
170 
171 	RIP6STAT_INC(rip6s_ipackets);
172 
173 	init_sin6(&fromsa, m, 0); /* general init */
174 
175 	ifp = m->m_pkthdr.rcvif;
176 
177 	CK_LIST_FOREACH(inp, &V_ripcb, inp_list) {
178 		/* XXX inp locking */
179 		if ((inp->inp_vflag & INP_IPV6) == 0)
180 			continue;
181 		if (inp->inp_ip_p &&
182 		    inp->inp_ip_p != proto)
183 			continue;
184 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
185 		    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
186 			continue;
187 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
188 		    !IN6_ARE_ADDR_EQUAL(&inp->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(inp);
227 		if (__predict_false(inp->inp_flags2 & INP_FREED))
228 			goto skip_2;
229 		if (jailed_without_vnet(inp->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(inp->inp_cred,
237 			    &ip6->ip6_dst) != 0)
238 				goto skip_2;
239 		}
240 		if (inp->in6p_cksum != -1) {
241 			RIP6STAT_INC(rip6s_isum);
242 			if (m->m_pkthdr.len - (*offp + inp->in6p_cksum) < 2 ||
243 			    in6_cksum(m, proto, *offp,
244 			    m->m_pkthdr.len - *offp)) {
245 				RIP6STAT_INC(rip6s_badsum);
246 				/*
247 				 * Drop the received message, don't send an
248 				 * ICMP6 message. Set proto to IPPROTO_NONE
249 				 * to achieve that.
250 				 */
251 				proto = IPPROTO_NONE;
252 				goto skip_2;
253 			}
254 		}
255 		/*
256 		 * If this raw socket has multicast state, and we
257 		 * have received a multicast, check if this socket
258 		 * should receive it, as multicast filtering is now
259 		 * the responsibility of the transport layer.
260 		 */
261 		if (inp->in6p_moptions &&
262 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
263 			/*
264 			 * If the incoming datagram is for MLD, allow it
265 			 * through unconditionally to the raw socket.
266 			 *
267 			 * Use the M_RTALERT_MLD flag to check for MLD
268 			 * traffic without having to inspect the mbuf chain
269 			 * more deeply, as all MLDv1/v2 host messages MUST
270 			 * contain the Router Alert option.
271 			 *
272 			 * In the case of MLDv1, we may not have explicitly
273 			 * joined the group, and may have set IFF_ALLMULTI
274 			 * on the interface. im6o_mc_filter() may discard
275 			 * control traffic we actually need to see.
276 			 *
277 			 * Userland multicast routing daemons should continue
278 			 * filter the control traffic appropriately.
279 			 */
280 			int blocked;
281 
282 			blocked = MCAST_PASS;
283 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
284 				struct sockaddr_in6 mcaddr;
285 
286 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
287 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
288 				mcaddr.sin6_family = AF_INET6;
289 				mcaddr.sin6_addr = ip6->ip6_dst;
290 
291 				blocked = im6o_mc_filter(inp->in6p_moptions,
292 				    ifp,
293 				    (struct sockaddr *)&mcaddr,
294 				    (struct sockaddr *)&fromsa);
295 			}
296 			if (blocked != MCAST_PASS) {
297 				IP6STAT_INC(ip6s_notmember);
298 				goto skip_2;
299 			}
300 		}
301 		last = inp;
302 		continue;
303 skip_2:
304 		INP_RUNLOCK(inp);
305 	}
306 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
307 	/*
308 	 * Check AH/ESP integrity.
309 	 */
310 	if (IPSEC_ENABLED(ipv6) && last != NULL &&
311 	    IPSEC_CHECK_POLICY(ipv6, m, last) != 0) {
312 		m_freem(m);
313 		IP6STAT_DEC(ip6s_delivered);
314 		/* Do not inject data into pcb. */
315 		INP_RUNLOCK(last);
316 	} else
317 #endif /* IPSEC */
318 	if (last != NULL) {
319 		if (last->inp_flags & INP_CONTROLOPTS ||
320 		    last->inp_socket->so_options & SO_TIMESTAMP)
321 			ip6_savecontrol(last, m, &opts);
322 		/* Strip intermediate headers. */
323 		m_adj(m, *offp);
324 		if (sbappendaddr(&last->inp_socket->so_rcv,
325 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
326 			m_freem(m);
327 			if (opts)
328 				m_freem(opts);
329 			RIP6STAT_INC(rip6s_fullsock);
330 		} else
331 			sorwakeup(last->inp_socket);
332 		INP_RUNLOCK(last);
333 	} else {
334 		RIP6STAT_INC(rip6s_nosock);
335 		if (m->m_flags & M_MCAST)
336 			RIP6STAT_INC(rip6s_nosockmcast);
337 		if (proto == IPPROTO_NONE)
338 			m_freem(m);
339 		else
340 			icmp6_error(m, ICMP6_PARAM_PROB,
341 			    ICMP6_PARAMPROB_NEXTHEADER,
342 			    ip6_get_prevhdr(m, *offp));
343 		IP6STAT_DEC(ip6s_delivered);
344 	}
345 	return (IPPROTO_DONE);
346 }
347 
348 void
349 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
350 {
351 	struct ip6ctlparam *ip6cp = NULL;
352 	const struct sockaddr_in6 *sa6_src = NULL;
353 	void *cmdarg;
354 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
355 
356 	if (sa->sa_family != AF_INET6 ||
357 	    sa->sa_len != sizeof(struct sockaddr_in6))
358 		return;
359 
360 	if ((unsigned)cmd >= PRC_NCMDS)
361 		return;
362 	if (PRC_IS_REDIRECT(cmd))
363 		notify = in6_rtchange, d = NULL;
364 	else if (cmd == PRC_HOSTDEAD)
365 		d = NULL;
366 	else if (inet6ctlerrmap[cmd] == 0)
367 		return;
368 
369 	/*
370 	 * If the parameter is from icmp6, decode it.
371 	 */
372 	if (d != NULL) {
373 		ip6cp = (struct ip6ctlparam *)d;
374 		cmdarg = ip6cp->ip6c_cmdarg;
375 		sa6_src = ip6cp->ip6c_src;
376 	} else {
377 		cmdarg = NULL;
378 		sa6_src = &sa6_any;
379 	}
380 
381 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
382 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
383 }
384 
385 /*
386  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
387  * may have setup with control call.
388  */
389 int
390 rip6_output(struct mbuf *m, struct socket *so, ...)
391 {
392 	struct epoch_tracker et;
393 	struct mbuf *control;
394 	struct m_tag *mtag;
395 	struct sockaddr_in6 *dstsock;
396 	struct ip6_hdr *ip6;
397 	struct inpcb *inp;
398 	u_int	plen = m->m_pkthdr.len;
399 	int error = 0;
400 	struct ip6_pktopts opt, *optp;
401 	struct ifnet *oifp = NULL;
402 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
403 	int scope_ambiguous = 0;
404 	int use_defzone = 0;
405 	int hlim = 0;
406 	struct in6_addr in6a;
407 	va_list ap;
408 
409 	va_start(ap, so);
410 	dstsock = va_arg(ap, struct sockaddr_in6 *);
411 	control = va_arg(ap, struct mbuf *);
412 	va_end(ap);
413 
414 	inp = sotoinpcb(so);
415 	INP_WLOCK(inp);
416 
417 	if (control != NULL) {
418 		if ((error = ip6_setpktopts(control, &opt,
419 		    inp->in6p_outputopts, so->so_cred,
420 		    so->so_proto->pr_protocol)) != 0) {
421 			goto bad;
422 		}
423 		optp = &opt;
424 	} else
425 		optp = inp->in6p_outputopts;
426 
427 	/*
428 	 * Check and convert scope zone ID into internal form.
429 	 *
430 	 * XXX: we may still need to determine the zone later.
431 	 */
432 	if (!(so->so_state & SS_ISCONNECTED)) {
433 		if (!optp || !optp->ip6po_pktinfo ||
434 		    !optp->ip6po_pktinfo->ipi6_ifindex)
435 			use_defzone = V_ip6_use_defzone;
436 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
437 			scope_ambiguous = 1;
438 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
439 			goto bad;
440 	}
441 
442 	/*
443 	 * For an ICMPv6 packet, we should know its type and code to update
444 	 * statistics.
445 	 */
446 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
447 		struct icmp6_hdr *icmp6;
448 		if (m->m_len < sizeof(struct icmp6_hdr) &&
449 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
450 			error = ENOBUFS;
451 			goto bad;
452 		}
453 		icmp6 = mtod(m, struct icmp6_hdr *);
454 		type = icmp6->icmp6_type;
455 		code = icmp6->icmp6_code;
456 	}
457 
458 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
459 	if (m == NULL) {
460 		error = ENOBUFS;
461 		goto bad;
462 	}
463 	ip6 = mtod(m, struct ip6_hdr *);
464 
465 	/*
466 	 * Source address selection.
467 	 */
468 	error = in6_selectsrc_socket(dstsock, optp, inp, so->so_cred,
469 	    scope_ambiguous, &in6a, &hlim);
470 
471 	if (error)
472 		goto bad;
473 	error = prison_check_ip6(inp->inp_cred, &in6a);
474 	if (error != 0)
475 		goto bad;
476 	ip6->ip6_src = in6a;
477 
478 	ip6->ip6_dst = dstsock->sin6_addr;
479 
480 	/*
481 	 * Fill in the rest of the IPv6 header fields.
482 	 */
483 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
484 	    (inp->inp_flow & IPV6_FLOWINFO_MASK);
485 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
486 	    (IPV6_VERSION & IPV6_VERSION_MASK);
487 
488 	/*
489 	 * ip6_plen will be filled in ip6_output, so not fill it here.
490 	 */
491 	ip6->ip6_nxt = inp->inp_ip_p;
492 	ip6->ip6_hlim = hlim;
493 
494 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
495 	    inp->in6p_cksum != -1) {
496 		struct mbuf *n;
497 		int off;
498 		u_int16_t *p;
499 
500 		/* Compute checksum. */
501 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
502 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
503 		else
504 			off = inp->in6p_cksum;
505 		if (plen < off + 2) {
506 			error = EINVAL;
507 			goto bad;
508 		}
509 		off += sizeof(struct ip6_hdr);
510 
511 		n = m;
512 		while (n && n->m_len <= off) {
513 			off -= n->m_len;
514 			n = n->m_next;
515 		}
516 		if (!n)
517 			goto bad;
518 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
519 		*p = 0;
520 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
521 	}
522 
523 	/*
524 	 * Send RA/RS messages to user land for protection, before sending
525 	 * them to rtadvd/rtsol.
526 	 */
527 	if ((send_sendso_input_hook != NULL) &&
528 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
529 		switch (type) {
530 		case ND_ROUTER_ADVERT:
531 		case ND_ROUTER_SOLICIT:
532 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
533 				sizeof(unsigned short), M_NOWAIT);
534 			if (mtag == NULL)
535 				goto bad;
536 			m_tag_prepend(m, mtag);
537 		}
538 	}
539 
540 	NET_EPOCH_ENTER(et);
541 	error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
542 	NET_EPOCH_EXIT(et);
543 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
544 		if (oifp)
545 			icmp6_ifoutstat_inc(oifp, type, code);
546 		ICMP6STAT_INC(icp6s_outhist[type]);
547 	} else
548 		RIP6STAT_INC(rip6s_opackets);
549 
550 	goto freectl;
551 
552  bad:
553 	if (m)
554 		m_freem(m);
555 
556  freectl:
557 	if (control != NULL) {
558 		ip6_clearpktopts(&opt, -1);
559 		m_freem(control);
560 	}
561 	INP_WUNLOCK(inp);
562 	return (error);
563 }
564 
565 /*
566  * Raw IPv6 socket option processing.
567  */
568 int
569 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
570 {
571 	struct inpcb *inp;
572 	int error;
573 
574 	if (sopt->sopt_level == IPPROTO_ICMPV6)
575 		/*
576 		 * XXX: is it better to call icmp6_ctloutput() directly
577 		 * from protosw?
578 		 */
579 		return (icmp6_ctloutput(so, sopt));
580 	else if (sopt->sopt_level != IPPROTO_IPV6) {
581 		if (sopt->sopt_level == SOL_SOCKET &&
582 		    sopt->sopt_name == SO_SETFIB) {
583 			inp = sotoinpcb(so);
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 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
605 			    EOPNOTSUPP;
606 			break;
607 		case IPV6_CHECKSUM:
608 			error = ip6_raw_ctloutput(so, sopt);
609 			break;
610 		default:
611 			error = ip6_ctloutput(so, sopt);
612 			break;
613 		}
614 		break;
615 
616 	case SOPT_SET:
617 		switch (sopt->sopt_name) {
618 		case MRT6_INIT:
619 		case MRT6_DONE:
620 		case MRT6_ADD_MIF:
621 		case MRT6_DEL_MIF:
622 		case MRT6_ADD_MFC:
623 		case MRT6_DEL_MFC:
624 		case MRT6_PIM:
625 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
626 			    EOPNOTSUPP;
627 			break;
628 		case IPV6_CHECKSUM:
629 			error = ip6_raw_ctloutput(so, sopt);
630 			break;
631 		default:
632 			error = ip6_ctloutput(so, sopt);
633 			break;
634 		}
635 		break;
636 	}
637 
638 	return (error);
639 }
640 
641 static int
642 rip6_attach(struct socket *so, int proto, struct thread *td)
643 {
644 	struct inpcb *inp;
645 	struct icmp6_filter *filter;
646 	int error;
647 
648 	inp = sotoinpcb(so);
649 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
650 
651 	error = priv_check(td, PRIV_NETINET_RAW);
652 	if (error)
653 		return (error);
654 	error = soreserve(so, rip_sendspace, rip_recvspace);
655 	if (error)
656 		return (error);
657 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
658 	if (filter == NULL)
659 		return (ENOMEM);
660 	INP_INFO_WLOCK(&V_ripcbinfo);
661 	error = in_pcballoc(so, &V_ripcbinfo);
662 	if (error) {
663 		INP_INFO_WUNLOCK(&V_ripcbinfo);
664 		free(filter, M_PCB);
665 		return (error);
666 	}
667 	inp = (struct inpcb *)so->so_pcb;
668 	INP_INFO_WUNLOCK(&V_ripcbinfo);
669 	inp->inp_vflag |= INP_IPV6;
670 	inp->inp_ip_p = (long)proto;
671 	inp->in6p_hops = -1;	/* use kernel default */
672 	inp->in6p_cksum = -1;
673 	inp->in6p_icmp6filt = filter;
674 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
675 	INP_WUNLOCK(inp);
676 	return (0);
677 }
678 
679 static void
680 rip6_detach(struct socket *so)
681 {
682 	struct inpcb *inp;
683 
684 	inp = sotoinpcb(so);
685 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
686 
687 	if (so == V_ip6_mrouter && ip6_mrouter_done)
688 		ip6_mrouter_done();
689 	/* xxx: RSVP */
690 	INP_INFO_WLOCK(&V_ripcbinfo);
691 	INP_WLOCK(inp);
692 	free(inp->in6p_icmp6filt, M_PCB);
693 	in_pcbdetach(inp);
694 	in_pcbfree(inp);
695 	INP_INFO_WUNLOCK(&V_ripcbinfo);
696 }
697 
698 /* XXXRW: This can't ever be called. */
699 static void
700 rip6_abort(struct socket *so)
701 {
702 	struct inpcb *inp;
703 
704 	inp = sotoinpcb(so);
705 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
706 
707 	soisdisconnected(so);
708 }
709 
710 static void
711 rip6_close(struct socket *so)
712 {
713 	struct inpcb *inp;
714 
715 	inp = sotoinpcb(so);
716 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
717 
718 	soisdisconnected(so);
719 }
720 
721 static int
722 rip6_disconnect(struct socket *so)
723 {
724 	struct inpcb *inp;
725 
726 	inp = sotoinpcb(so);
727 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
728 
729 	if ((so->so_state & SS_ISCONNECTED) == 0)
730 		return (ENOTCONN);
731 	inp->in6p_faddr = in6addr_any;
732 	rip6_abort(so);
733 	return (0);
734 }
735 
736 static int
737 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
738 {
739 	struct epoch_tracker et;
740 	struct inpcb *inp;
741 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
742 	struct ifaddr *ifa = NULL;
743 	int error = 0;
744 
745 	inp = sotoinpcb(so);
746 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
747 
748 	if (nam->sa_len != sizeof(*addr))
749 		return (EINVAL);
750 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
751 		return (error);
752 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
753 		return (EADDRNOTAVAIL);
754 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
755 		return (error);
756 
757 	NET_EPOCH_ENTER(et);
758 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
759 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
760 		NET_EPOCH_EXIT(et);
761 		return (EADDRNOTAVAIL);
762 	}
763 	if (ifa != NULL &&
764 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
765 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
766 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
767 		NET_EPOCH_EXIT(et);
768 		return (EADDRNOTAVAIL);
769 	}
770 	NET_EPOCH_EXIT(et);
771 	INP_INFO_WLOCK(&V_ripcbinfo);
772 	INP_WLOCK(inp);
773 	inp->in6p_laddr = addr->sin6_addr;
774 	INP_WUNLOCK(inp);
775 	INP_INFO_WUNLOCK(&V_ripcbinfo);
776 	return (0);
777 }
778 
779 static int
780 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
781 {
782 	struct inpcb *inp;
783 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
784 	struct in6_addr in6a;
785 	int error = 0, scope_ambiguous = 0;
786 
787 	inp = sotoinpcb(so);
788 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
789 
790 	if (nam->sa_len != sizeof(*addr))
791 		return (EINVAL);
792 	if (CK_STAILQ_EMPTY(&V_ifnet))
793 		return (EADDRNOTAVAIL);
794 	if (addr->sin6_family != AF_INET6)
795 		return (EAFNOSUPPORT);
796 
797 	/*
798 	 * Application should provide a proper zone ID or the use of default
799 	 * zone IDs should be enabled.  Unfortunately, some applications do
800 	 * not behave as it should, so we need a workaround.  Even if an
801 	 * appropriate ID is not determined, we'll see if we can determine
802 	 * the outgoing interface.  If we can, determine the zone ID based on
803 	 * the interface below.
804 	 */
805 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
806 		scope_ambiguous = 1;
807 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
808 		return (error);
809 
810 	INP_INFO_WLOCK(&V_ripcbinfo);
811 	INP_WLOCK(inp);
812 	/* Source address selection. XXX: need pcblookup? */
813 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
814 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
815 	if (error) {
816 		INP_WUNLOCK(inp);
817 		INP_INFO_WUNLOCK(&V_ripcbinfo);
818 		return (error);
819 	}
820 
821 	inp->in6p_faddr = addr->sin6_addr;
822 	inp->in6p_laddr = in6a;
823 	soisconnected(so);
824 	INP_WUNLOCK(inp);
825 	INP_INFO_WUNLOCK(&V_ripcbinfo);
826 	return (0);
827 }
828 
829 static int
830 rip6_shutdown(struct socket *so)
831 {
832 	struct inpcb *inp;
833 
834 	inp = sotoinpcb(so);
835 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
836 
837 	INP_WLOCK(inp);
838 	socantsendmore(so);
839 	INP_WUNLOCK(inp);
840 	return (0);
841 }
842 
843 static int
844 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
845     struct mbuf *control, struct thread *td)
846 {
847 	struct inpcb *inp;
848 	struct sockaddr_in6 tmp;
849 	struct sockaddr_in6 *dst;
850 	int ret;
851 
852 	inp = sotoinpcb(so);
853 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
854 
855 	/* Always copy sockaddr to avoid overwrites. */
856 	/* Unlocked read. */
857 	if (so->so_state & SS_ISCONNECTED) {
858 		if (nam) {
859 			m_freem(m);
860 			return (EISCONN);
861 		}
862 		/* XXX */
863 		bzero(&tmp, sizeof(tmp));
864 		tmp.sin6_family = AF_INET6;
865 		tmp.sin6_len = sizeof(struct sockaddr_in6);
866 		INP_RLOCK(inp);
867 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
868 		    sizeof(struct in6_addr));
869 		INP_RUNLOCK(inp);
870 		dst = &tmp;
871 	} else {
872 		if (nam == NULL) {
873 			m_freem(m);
874 			return (ENOTCONN);
875 		}
876 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
877 			m_freem(m);
878 			return (EINVAL);
879 		}
880 		tmp = *(struct sockaddr_in6 *)nam;
881 		dst = &tmp;
882 
883 		if (dst->sin6_family == AF_UNSPEC) {
884 			/*
885 			 * XXX: we allow this case for backward
886 			 * compatibility to buggy applications that
887 			 * rely on old (and wrong) kernel behavior.
888 			 */
889 			log(LOG_INFO, "rip6 SEND: address family is "
890 			    "unspec. Assume AF_INET6\n");
891 			dst->sin6_family = AF_INET6;
892 		} else if (dst->sin6_family != AF_INET6) {
893 			m_freem(m);
894 			return(EAFNOSUPPORT);
895 		}
896 	}
897 	ret = rip6_output(m, so, dst, control);
898 	return (ret);
899 }
900 
901 struct pr_usrreqs rip6_usrreqs = {
902 	.pru_abort =		rip6_abort,
903 	.pru_attach =		rip6_attach,
904 	.pru_bind =		rip6_bind,
905 	.pru_connect =		rip6_connect,
906 	.pru_control =		in6_control,
907 	.pru_detach =		rip6_detach,
908 	.pru_disconnect =	rip6_disconnect,
909 	.pru_peeraddr =		in6_getpeeraddr,
910 	.pru_send =		rip6_send,
911 	.pru_shutdown =		rip6_shutdown,
912 	.pru_sockaddr =		in6_getsockaddr,
913 	.pru_close =		rip6_close,
914 };
915