xref: /freebsd/sys/netinet6/raw_ip6.c (revision f7c32ed617858bcd22f8d1b03199099d50125721)
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/ip6protosw.h>
102 #include <netinet6/ip6_mroute.h>
103 #include <netinet6/in6_pcb.h>
104 #include <netinet6/ip6_var.h>
105 #include <netinet6/nd6.h>
106 #include <netinet6/raw_ip6.h>
107 #include <netinet6/in6_fib.h>
108 #include <netinet6/scope6_var.h>
109 #include <netinet6/send.h>
110 
111 #include <netipsec/ipsec_support.h>
112 
113 #include <machine/stdarg.h>
114 
115 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
116 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
117 
118 /*
119  * Raw interface to IP6 protocol.
120  */
121 
122 VNET_DECLARE(struct inpcbhead, ripcb);
123 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
124 #define	V_ripcb				VNET(ripcb)
125 #define	V_ripcbinfo			VNET(ripcbinfo)
126 
127 extern u_long	rip_sendspace;
128 extern u_long	rip_recvspace;
129 
130 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
131 VNET_PCPUSTAT_SYSINIT(rip6stat);
132 
133 #ifdef VIMAGE
134 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
135 #endif /* VIMAGE */
136 
137 /*
138  * Hooks for multicast routing. They all default to NULL, so leave them not
139  * initialized and rely on BSS being set to 0.
140  */
141 
142 /*
143  * The socket used to communicate with the multicast routing daemon.
144  */
145 VNET_DEFINE(struct socket *, ip6_mrouter);
146 
147 /*
148  * The various mrouter functions.
149  */
150 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
151 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
152 int (*ip6_mrouter_done)(void);
153 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
154 int (*mrt6_ioctl)(u_long, caddr_t);
155 
156 /*
157  * Setup generic address and protocol structures for raw_input routine, then
158  * pass them along with mbuf chain.
159  */
160 int
161 rip6_input(struct mbuf **mp, int *offp, int proto)
162 {
163 	struct ifnet *ifp;
164 	struct mbuf *m = *mp;
165 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
166 	struct inpcb *inp;
167 	struct inpcb *last = NULL;
168 	struct mbuf *opts = NULL;
169 	struct sockaddr_in6 fromsa;
170 
171 	NET_EPOCH_ASSERT();
172 
173 	RIP6STAT_INC(rip6s_ipackets);
174 
175 	init_sin6(&fromsa, m, 0); /* general init */
176 
177 	ifp = m->m_pkthdr.rcvif;
178 
179 	CK_LIST_FOREACH(inp, &V_ripcb, inp_list) {
180 		/* XXX inp locking */
181 		if ((inp->inp_vflag & INP_IPV6) == 0)
182 			continue;
183 		if (inp->inp_ip_p &&
184 		    inp->inp_ip_p != proto)
185 			continue;
186 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
187 		    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
188 			continue;
189 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
190 		    !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src))
191 			continue;
192 		if (last != NULL) {
193 			struct mbuf *n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
194 
195 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
196 			/*
197 			 * Check AH/ESP integrity.
198 			 */
199 			if (IPSEC_ENABLED(ipv6)) {
200 				if (n != NULL &&
201 				    IPSEC_CHECK_POLICY(ipv6, n, last) != 0) {
202 					m_freem(n);
203 					/* Do not inject data into pcb. */
204 					n = NULL;
205 				}
206 			}
207 #endif /* IPSEC */
208 			if (n) {
209 				if (last->inp_flags & INP_CONTROLOPTS ||
210 				    last->inp_socket->so_options & SO_TIMESTAMP)
211 					ip6_savecontrol(last, n, &opts);
212 				/* strip intermediate headers */
213 				m_adj(n, *offp);
214 				if (sbappendaddr(&last->inp_socket->so_rcv,
215 						(struct sockaddr *)&fromsa,
216 						 n, opts) == 0) {
217 					soroverflow(last->inp_socket);
218 					m_freem(n);
219 					if (opts)
220 						m_freem(opts);
221 					RIP6STAT_INC(rip6s_fullsock);
222 				} else
223 					sorwakeup(last->inp_socket);
224 				opts = NULL;
225 			}
226 			INP_RUNLOCK(last);
227 			last = NULL;
228 		}
229 		INP_RLOCK(inp);
230 		if (__predict_false(inp->inp_flags2 & INP_FREED))
231 			goto skip_2;
232 		if (jailed_without_vnet(inp->inp_cred)) {
233 			/*
234 			 * Allow raw socket in jail to receive multicast;
235 			 * assume process had PRIV_NETINET_RAW at attach,
236 			 * and fall through into normal filter path if so.
237 			 */
238 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
239 			    prison_check_ip6(inp->inp_cred,
240 			    &ip6->ip6_dst) != 0)
241 				goto skip_2;
242 		}
243 		if (inp->in6p_cksum != -1) {
244 			RIP6STAT_INC(rip6s_isum);
245 			if (m->m_pkthdr.len - (*offp + inp->in6p_cksum) < 2 ||
246 			    in6_cksum(m, proto, *offp,
247 			    m->m_pkthdr.len - *offp)) {
248 				RIP6STAT_INC(rip6s_badsum);
249 				/*
250 				 * Drop the received message, don't send an
251 				 * ICMP6 message. Set proto to IPPROTO_NONE
252 				 * to achieve that.
253 				 */
254 				proto = IPPROTO_NONE;
255 				goto skip_2;
256 			}
257 		}
258 		/*
259 		 * If this raw socket has multicast state, and we
260 		 * have received a multicast, check if this socket
261 		 * should receive it, as multicast filtering is now
262 		 * the responsibility of the transport layer.
263 		 */
264 		if (inp->in6p_moptions &&
265 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
266 			/*
267 			 * If the incoming datagram is for MLD, allow it
268 			 * through unconditionally to the raw socket.
269 			 *
270 			 * Use the M_RTALERT_MLD flag to check for MLD
271 			 * traffic without having to inspect the mbuf chain
272 			 * more deeply, as all MLDv1/v2 host messages MUST
273 			 * contain the Router Alert option.
274 			 *
275 			 * In the case of MLDv1, we may not have explicitly
276 			 * joined the group, and may have set IFF_ALLMULTI
277 			 * on the interface. im6o_mc_filter() may discard
278 			 * control traffic we actually need to see.
279 			 *
280 			 * Userland multicast routing daemons should continue
281 			 * filter the control traffic appropriately.
282 			 */
283 			int blocked;
284 
285 			blocked = MCAST_PASS;
286 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
287 				struct sockaddr_in6 mcaddr;
288 
289 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
290 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
291 				mcaddr.sin6_family = AF_INET6;
292 				mcaddr.sin6_addr = ip6->ip6_dst;
293 
294 				blocked = im6o_mc_filter(inp->in6p_moptions,
295 				    ifp,
296 				    (struct sockaddr *)&mcaddr,
297 				    (struct sockaddr *)&fromsa);
298 			}
299 			if (blocked != MCAST_PASS) {
300 				IP6STAT_INC(ip6s_notmember);
301 				goto skip_2;
302 			}
303 		}
304 		last = inp;
305 		continue;
306 skip_2:
307 		INP_RUNLOCK(inp);
308 	}
309 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
310 	/*
311 	 * Check AH/ESP integrity.
312 	 */
313 	if (IPSEC_ENABLED(ipv6) && last != NULL &&
314 	    IPSEC_CHECK_POLICY(ipv6, m, last) != 0) {
315 		m_freem(m);
316 		IP6STAT_DEC(ip6s_delivered);
317 		/* Do not inject data into pcb. */
318 		INP_RUNLOCK(last);
319 	} else
320 #endif /* IPSEC */
321 	if (last != NULL) {
322 		if (last->inp_flags & INP_CONTROLOPTS ||
323 		    last->inp_socket->so_options & SO_TIMESTAMP)
324 			ip6_savecontrol(last, m, &opts);
325 		/* Strip intermediate headers. */
326 		m_adj(m, *offp);
327 		if (sbappendaddr(&last->inp_socket->so_rcv,
328 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
329 			soroverflow(last->inp_socket);
330 			m_freem(m);
331 			if (opts)
332 				m_freem(opts);
333 			RIP6STAT_INC(rip6s_fullsock);
334 		} else
335 			sorwakeup(last->inp_socket);
336 		INP_RUNLOCK(last);
337 	} else {
338 		RIP6STAT_INC(rip6s_nosock);
339 		if (m->m_flags & M_MCAST)
340 			RIP6STAT_INC(rip6s_nosockmcast);
341 		if (proto == IPPROTO_NONE)
342 			m_freem(m);
343 		else
344 			icmp6_error(m, ICMP6_PARAM_PROB,
345 			    ICMP6_PARAMPROB_NEXTHEADER,
346 			    ip6_get_prevhdr(m, *offp));
347 		IP6STAT_DEC(ip6s_delivered);
348 	}
349 	return (IPPROTO_DONE);
350 }
351 
352 void
353 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
354 {
355 	struct ip6ctlparam *ip6cp = NULL;
356 	const struct sockaddr_in6 *sa6_src = NULL;
357 	void *cmdarg;
358 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
359 
360 	if (sa->sa_family != AF_INET6 ||
361 	    sa->sa_len != sizeof(struct sockaddr_in6))
362 		return;
363 
364 	if ((unsigned)cmd >= PRC_NCMDS)
365 		return;
366 	if (PRC_IS_REDIRECT(cmd))
367 		notify = in6_rtchange, d = NULL;
368 	else if (cmd == PRC_HOSTDEAD)
369 		d = NULL;
370 	else if (inet6ctlerrmap[cmd] == 0)
371 		return;
372 
373 	/*
374 	 * If the parameter is from icmp6, decode it.
375 	 */
376 	if (d != NULL) {
377 		ip6cp = (struct ip6ctlparam *)d;
378 		cmdarg = ip6cp->ip6c_cmdarg;
379 		sa6_src = ip6cp->ip6c_src;
380 	} else {
381 		cmdarg = NULL;
382 		sa6_src = &sa6_any;
383 	}
384 
385 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
386 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
387 }
388 
389 /*
390  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
391  * may have setup with control call.
392  */
393 int
394 rip6_output(struct mbuf *m, struct socket *so, ...)
395 {
396 	struct epoch_tracker et;
397 	struct mbuf *control;
398 	struct m_tag *mtag;
399 	struct sockaddr_in6 *dstsock;
400 	struct ip6_hdr *ip6;
401 	struct inpcb *inp;
402 	u_int	plen = m->m_pkthdr.len;
403 	int error = 0;
404 	struct ip6_pktopts opt, *optp;
405 	struct ifnet *oifp = NULL;
406 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
407 	int scope_ambiguous = 0;
408 	int use_defzone = 0;
409 	int hlim = 0;
410 	struct in6_addr in6a;
411 	va_list ap;
412 
413 	va_start(ap, so);
414 	dstsock = va_arg(ap, struct sockaddr_in6 *);
415 	control = va_arg(ap, struct mbuf *);
416 	va_end(ap);
417 
418 	inp = sotoinpcb(so);
419 	INP_WLOCK(inp);
420 
421 	if (control != NULL) {
422 		NET_EPOCH_ENTER(et);
423 		error = ip6_setpktopts(control, &opt,
424 		    inp->in6p_outputopts, so->so_cred,
425 		    so->so_proto->pr_protocol);
426 		NET_EPOCH_EXIT(et);
427 
428 		if (error != 0) {
429 			goto bad;
430 		}
431 		optp = &opt;
432 	} else
433 		optp = inp->in6p_outputopts;
434 
435 	/*
436 	 * Check and convert scope zone ID into internal form.
437 	 *
438 	 * XXX: we may still need to determine the zone later.
439 	 */
440 	if (!(so->so_state & SS_ISCONNECTED)) {
441 		if (!optp || !optp->ip6po_pktinfo ||
442 		    !optp->ip6po_pktinfo->ipi6_ifindex)
443 			use_defzone = V_ip6_use_defzone;
444 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
445 			scope_ambiguous = 1;
446 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
447 			goto bad;
448 	}
449 
450 	/*
451 	 * For an ICMPv6 packet, we should know its type and code to update
452 	 * statistics.
453 	 */
454 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
455 		struct icmp6_hdr *icmp6;
456 		if (m->m_len < sizeof(struct icmp6_hdr) &&
457 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
458 			error = ENOBUFS;
459 			goto bad;
460 		}
461 		icmp6 = mtod(m, struct icmp6_hdr *);
462 		type = icmp6->icmp6_type;
463 		code = icmp6->icmp6_code;
464 	}
465 
466 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
467 	if (m == NULL) {
468 		error = ENOBUFS;
469 		goto bad;
470 	}
471 	ip6 = mtod(m, struct ip6_hdr *);
472 
473 #ifdef ROUTE_MPATH
474 	if (CALC_FLOWID_OUTBOUND) {
475 		uint32_t hash_type, hash_val;
476 
477 		hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
478 		    &dstsock->sin6_addr, 0, 0, so->so_proto->pr_protocol,
479 		    &hash_type);
480 		inp->inp_flowid = hash_val;
481 		inp->inp_flowtype = hash_type;
482 	}
483 #endif
484 	/*
485 	 * Source address selection.
486 	 */
487 	NET_EPOCH_ENTER(et);
488 	error = in6_selectsrc_socket(dstsock, optp, inp, so->so_cred,
489 	    scope_ambiguous, &in6a, &hlim);
490 	NET_EPOCH_EXIT(et);
491 
492 	if (error)
493 		goto bad;
494 	error = prison_check_ip6(inp->inp_cred, &in6a);
495 	if (error != 0)
496 		goto bad;
497 	ip6->ip6_src = in6a;
498 
499 	ip6->ip6_dst = dstsock->sin6_addr;
500 
501 	/*
502 	 * Fill in the rest of the IPv6 header fields.
503 	 */
504 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
505 	    (inp->inp_flow & IPV6_FLOWINFO_MASK);
506 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
507 	    (IPV6_VERSION & IPV6_VERSION_MASK);
508 
509 	/*
510 	 * ip6_plen will be filled in ip6_output, so not fill it here.
511 	 */
512 	ip6->ip6_nxt = inp->inp_ip_p;
513 	ip6->ip6_hlim = hlim;
514 
515 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
516 	    inp->in6p_cksum != -1) {
517 		struct mbuf *n;
518 		int off;
519 		u_int16_t *p;
520 
521 		/* Compute checksum. */
522 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
523 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
524 		else
525 			off = inp->in6p_cksum;
526 		if (plen < off + 2) {
527 			error = EINVAL;
528 			goto bad;
529 		}
530 		off += sizeof(struct ip6_hdr);
531 
532 		n = m;
533 		while (n && n->m_len <= off) {
534 			off -= n->m_len;
535 			n = n->m_next;
536 		}
537 		if (!n)
538 			goto bad;
539 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
540 		*p = 0;
541 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
542 	}
543 
544 	/*
545 	 * Send RA/RS messages to user land for protection, before sending
546 	 * them to rtadvd/rtsol.
547 	 */
548 	if ((send_sendso_input_hook != NULL) &&
549 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
550 		switch (type) {
551 		case ND_ROUTER_ADVERT:
552 		case ND_ROUTER_SOLICIT:
553 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
554 				sizeof(unsigned short), M_NOWAIT);
555 			if (mtag == NULL)
556 				goto bad;
557 			m_tag_prepend(m, mtag);
558 		}
559 	}
560 
561 	NET_EPOCH_ENTER(et);
562 	error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
563 	NET_EPOCH_EXIT(et);
564 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
565 		if (oifp)
566 			icmp6_ifoutstat_inc(oifp, type, code);
567 		ICMP6STAT_INC(icp6s_outhist[type]);
568 	} else
569 		RIP6STAT_INC(rip6s_opackets);
570 
571 	goto freectl;
572 
573  bad:
574 	if (m)
575 		m_freem(m);
576 
577  freectl:
578 	if (control != NULL) {
579 		ip6_clearpktopts(&opt, -1);
580 		m_freem(control);
581 	}
582 	INP_WUNLOCK(inp);
583 	return (error);
584 }
585 
586 /*
587  * Raw IPv6 socket option processing.
588  */
589 int
590 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
591 {
592 	struct inpcb *inp;
593 	int error;
594 
595 	if (sopt->sopt_level == IPPROTO_ICMPV6)
596 		/*
597 		 * XXX: is it better to call icmp6_ctloutput() directly
598 		 * from protosw?
599 		 */
600 		return (icmp6_ctloutput(so, sopt));
601 	else if (sopt->sopt_level != IPPROTO_IPV6) {
602 		if (sopt->sopt_level == SOL_SOCKET &&
603 		    sopt->sopt_name == SO_SETFIB) {
604 			inp = sotoinpcb(so);
605 			INP_WLOCK(inp);
606 			inp->inp_inc.inc_fibnum = so->so_fibnum;
607 			INP_WUNLOCK(inp);
608 			return (0);
609 		}
610 		return (EINVAL);
611 	}
612 
613 	error = 0;
614 
615 	switch (sopt->sopt_dir) {
616 	case SOPT_GET:
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_get ?  ip6_mrouter_get(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 	case SOPT_SET:
638 		switch (sopt->sopt_name) {
639 		case MRT6_INIT:
640 		case MRT6_DONE:
641 		case MRT6_ADD_MIF:
642 		case MRT6_DEL_MIF:
643 		case MRT6_ADD_MFC:
644 		case MRT6_DEL_MFC:
645 		case MRT6_PIM:
646 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
647 			    EOPNOTSUPP;
648 			break;
649 		case IPV6_CHECKSUM:
650 			error = ip6_raw_ctloutput(so, sopt);
651 			break;
652 		default:
653 			error = ip6_ctloutput(so, sopt);
654 			break;
655 		}
656 		break;
657 	}
658 
659 	return (error);
660 }
661 
662 static int
663 rip6_attach(struct socket *so, int proto, struct thread *td)
664 {
665 	struct inpcb *inp;
666 	struct icmp6_filter *filter;
667 	int error;
668 
669 	inp = sotoinpcb(so);
670 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
671 
672 	error = priv_check(td, PRIV_NETINET_RAW);
673 	if (error)
674 		return (error);
675 	error = soreserve(so, rip_sendspace, rip_recvspace);
676 	if (error)
677 		return (error);
678 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
679 	if (filter == NULL)
680 		return (ENOMEM);
681 	INP_INFO_WLOCK(&V_ripcbinfo);
682 	error = in_pcballoc(so, &V_ripcbinfo);
683 	if (error) {
684 		INP_INFO_WUNLOCK(&V_ripcbinfo);
685 		free(filter, M_PCB);
686 		return (error);
687 	}
688 	inp = (struct inpcb *)so->so_pcb;
689 	INP_INFO_WUNLOCK(&V_ripcbinfo);
690 	inp->inp_vflag |= INP_IPV6;
691 	inp->inp_ip_p = (long)proto;
692 	inp->in6p_hops = -1;	/* use kernel default */
693 	inp->in6p_cksum = -1;
694 	inp->in6p_icmp6filt = filter;
695 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
696 	INP_WUNLOCK(inp);
697 	return (0);
698 }
699 
700 static void
701 rip6_detach(struct socket *so)
702 {
703 	struct inpcb *inp;
704 
705 	inp = sotoinpcb(so);
706 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
707 
708 	if (so == V_ip6_mrouter && ip6_mrouter_done)
709 		ip6_mrouter_done();
710 	/* xxx: RSVP */
711 	INP_INFO_WLOCK(&V_ripcbinfo);
712 	INP_WLOCK(inp);
713 	free(inp->in6p_icmp6filt, M_PCB);
714 	in_pcbdetach(inp);
715 	in_pcbfree(inp);
716 	INP_INFO_WUNLOCK(&V_ripcbinfo);
717 }
718 
719 /* XXXRW: This can't ever be called. */
720 static void
721 rip6_abort(struct socket *so)
722 {
723 	struct inpcb *inp;
724 
725 	inp = sotoinpcb(so);
726 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
727 
728 	soisdisconnected(so);
729 }
730 
731 static void
732 rip6_close(struct socket *so)
733 {
734 	struct inpcb *inp;
735 
736 	inp = sotoinpcb(so);
737 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
738 
739 	soisdisconnected(so);
740 }
741 
742 static int
743 rip6_disconnect(struct socket *so)
744 {
745 	struct inpcb *inp;
746 
747 	inp = sotoinpcb(so);
748 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
749 
750 	if ((so->so_state & SS_ISCONNECTED) == 0)
751 		return (ENOTCONN);
752 	inp->in6p_faddr = in6addr_any;
753 	rip6_abort(so);
754 	return (0);
755 }
756 
757 static int
758 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
759 {
760 	struct epoch_tracker et;
761 	struct inpcb *inp;
762 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
763 	struct ifaddr *ifa = NULL;
764 	int error = 0;
765 
766 	inp = sotoinpcb(so);
767 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
768 
769 	if (nam->sa_family != AF_INET6)
770 		return (EAFNOSUPPORT);
771 	if (nam->sa_len != sizeof(*addr))
772 		return (EINVAL);
773 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
774 		return (error);
775 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
776 		return (EADDRNOTAVAIL);
777 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
778 		return (error);
779 
780 	NET_EPOCH_ENTER(et);
781 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
782 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
783 		NET_EPOCH_EXIT(et);
784 		return (EADDRNOTAVAIL);
785 	}
786 	if (ifa != NULL &&
787 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
788 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
789 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
790 		NET_EPOCH_EXIT(et);
791 		return (EADDRNOTAVAIL);
792 	}
793 	NET_EPOCH_EXIT(et);
794 	INP_INFO_WLOCK(&V_ripcbinfo);
795 	INP_WLOCK(inp);
796 	inp->in6p_laddr = addr->sin6_addr;
797 	INP_WUNLOCK(inp);
798 	INP_INFO_WUNLOCK(&V_ripcbinfo);
799 	return (0);
800 }
801 
802 static int
803 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
804 {
805 	struct inpcb *inp;
806 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
807 	struct in6_addr in6a;
808 	struct epoch_tracker et;
809 	int error = 0, scope_ambiguous = 0;
810 
811 	inp = sotoinpcb(so);
812 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
813 
814 	if (nam->sa_len != sizeof(*addr))
815 		return (EINVAL);
816 	if (CK_STAILQ_EMPTY(&V_ifnet))
817 		return (EADDRNOTAVAIL);
818 	if (addr->sin6_family != AF_INET6)
819 		return (EAFNOSUPPORT);
820 
821 	/*
822 	 * Application should provide a proper zone ID or the use of default
823 	 * zone IDs should be enabled.  Unfortunately, some applications do
824 	 * not behave as it should, so we need a workaround.  Even if an
825 	 * appropriate ID is not determined, we'll see if we can determine
826 	 * the outgoing interface.  If we can, determine the zone ID based on
827 	 * the interface below.
828 	 */
829 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
830 		scope_ambiguous = 1;
831 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
832 		return (error);
833 
834 	INP_INFO_WLOCK(&V_ripcbinfo);
835 	INP_WLOCK(inp);
836 	/* Source address selection. XXX: need pcblookup? */
837 	NET_EPOCH_ENTER(et);
838 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
839 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
840 	NET_EPOCH_EXIT(et);
841 	if (error) {
842 		INP_WUNLOCK(inp);
843 		INP_INFO_WUNLOCK(&V_ripcbinfo);
844 		return (error);
845 	}
846 
847 	inp->in6p_faddr = addr->sin6_addr;
848 	inp->in6p_laddr = in6a;
849 	soisconnected(so);
850 	INP_WUNLOCK(inp);
851 	INP_INFO_WUNLOCK(&V_ripcbinfo);
852 	return (0);
853 }
854 
855 static int
856 rip6_shutdown(struct socket *so)
857 {
858 	struct inpcb *inp;
859 
860 	inp = sotoinpcb(so);
861 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
862 
863 	INP_WLOCK(inp);
864 	socantsendmore(so);
865 	INP_WUNLOCK(inp);
866 	return (0);
867 }
868 
869 static int
870 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
871     struct mbuf *control, struct thread *td)
872 {
873 	struct inpcb *inp;
874 	struct sockaddr_in6 tmp;
875 	struct sockaddr_in6 *dst;
876 	int error;
877 
878 	inp = sotoinpcb(so);
879 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
880 
881 	/* Always copy sockaddr to avoid overwrites. */
882 	/* Unlocked read. */
883 	if (so->so_state & SS_ISCONNECTED) {
884 		if (nam) {
885 			error = EISCONN;
886 			goto release;
887 		}
888 		/* XXX */
889 		bzero(&tmp, sizeof(tmp));
890 		tmp.sin6_family = AF_INET6;
891 		tmp.sin6_len = sizeof(struct sockaddr_in6);
892 		INP_RLOCK(inp);
893 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
894 		    sizeof(struct in6_addr));
895 		INP_RUNLOCK(inp);
896 		dst = &tmp;
897 	} else {
898 		error = 0;
899 		if (nam == NULL)
900 			error = ENOTCONN;
901 		else if (nam->sa_family != AF_INET6)
902 			error = EAFNOSUPPORT;
903 		else if (nam->sa_len != sizeof(struct sockaddr_in6))
904 			error = EINVAL;
905 		if (error != 0)
906 			goto release;
907 		tmp = *(struct sockaddr_in6 *)nam;
908 		dst = &tmp;
909 
910 		if (dst->sin6_family == AF_UNSPEC) {
911 			/*
912 			 * XXX: we allow this case for backward
913 			 * compatibility to buggy applications that
914 			 * rely on old (and wrong) kernel behavior.
915 			 */
916 			log(LOG_INFO, "rip6 SEND: address family is "
917 			    "unspec. Assume AF_INET6\n");
918 			dst->sin6_family = AF_INET6;
919 		} else if (dst->sin6_family != AF_INET6) {
920 			error = EAFNOSUPPORT;
921 			goto release;
922 		}
923 	}
924 	return (rip6_output(m, so, dst, control));
925 
926 release:
927 	if (control != NULL)
928 		m_freem(control);
929 	m_freem(m);
930 	return (error);
931 }
932 
933 struct pr_usrreqs rip6_usrreqs = {
934 	.pru_abort =		rip6_abort,
935 	.pru_attach =		rip6_attach,
936 	.pru_bind =		rip6_bind,
937 	.pru_connect =		rip6_connect,
938 	.pru_control =		in6_control,
939 	.pru_detach =		rip6_detach,
940 	.pru_disconnect =	rip6_disconnect,
941 	.pru_peeraddr =		in6_getpeeraddr,
942 	.pru_send =		rip6_send,
943 	.pru_shutdown =		rip6_shutdown,
944 	.pru_sockaddr =		in6_getsockaddr,
945 	.pru_close =		rip6_close,
946 };
947