xref: /freebsd/sys/netinet6/raw_ip6.c (revision f677a9e2672665f4eb3dd4111c07ee8f1f954262)
1 /*-
2  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. Neither the name of the project nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  */
29 
30 /*-
31  * Copyright (c) 1982, 1986, 1988, 1993
32  *	The Regents of the University of California.
33  * All rights reserved.
34  *
35  * Redistribution and use in source and binary forms, with or without
36  * modification, are permitted provided that the following conditions
37  * are met:
38  * 1. Redistributions of source code must retain the above copyright
39  *    notice, this list of conditions and the following disclaimer.
40  * 2. Redistributions in binary form must reproduce the above copyright
41  *    notice, this list of conditions and the following disclaimer in the
42  *    documentation and/or other materials provided with the distribution.
43  * 4. Neither the name of the University nor the names of its contributors
44  *    may be used to endorse or promote products derived from this software
45  *    without specific prior written permission.
46  *
47  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
48  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
49  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
50  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
51  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
52  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
53  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
54  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
55  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
56  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
57  * SUCH DAMAGE.
58  *
59  *	@(#)raw_ip.c	8.2 (Berkeley) 1/4/94
60  */
61 
62 #include <sys/cdefs.h>
63 __FBSDID("$FreeBSD$");
64 
65 #include "opt_ipsec.h"
66 #include "opt_inet6.h"
67 
68 #include <sys/param.h>
69 #include <sys/errno.h>
70 #include <sys/jail.h>
71 #include <sys/kernel.h>
72 #include <sys/lock.h>
73 #include <sys/malloc.h>
74 #include <sys/mbuf.h>
75 #include <sys/priv.h>
76 #include <sys/proc.h>
77 #include <sys/protosw.h>
78 #include <sys/signalvar.h>
79 #include <sys/socket.h>
80 #include <sys/socketvar.h>
81 #include <sys/sx.h>
82 #include <sys/syslog.h>
83 
84 #include <net/if.h>
85 #include <net/if_types.h>
86 #include <net/route.h>
87 #include <net/vnet.h>
88 
89 #include <netinet/in.h>
90 #include <netinet/in_var.h>
91 #include <netinet/in_systm.h>
92 #include <netinet/in_pcb.h>
93 
94 #include <netinet/icmp6.h>
95 #include <netinet/ip6.h>
96 #include <netinet/ip_var.h>
97 #include <netinet6/ip6protosw.h>
98 #include <netinet6/ip6_mroute.h>
99 #include <netinet6/in6_pcb.h>
100 #include <netinet6/ip6_var.h>
101 #include <netinet6/nd6.h>
102 #include <netinet6/raw_ip6.h>
103 #include <netinet6/scope6_var.h>
104 #include <netinet6/send.h>
105 
106 #ifdef IPSEC
107 #include <netipsec/ipsec.h>
108 #include <netipsec/ipsec6.h>
109 #endif /* IPSEC */
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 	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
164 	register struct inpcb *in6p;
165 	struct inpcb *last = 0;
166 	struct mbuf *opts = NULL;
167 	struct sockaddr_in6 fromsa;
168 
169 	RIP6STAT_INC(rip6s_ipackets);
170 
171 	if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
172 		/* XXX Send icmp6 host/port unreach? */
173 		m_freem(m);
174 		return (IPPROTO_DONE);
175 	}
176 
177 	init_sin6(&fromsa, m); /* general init */
178 
179 	ifp = m->m_pkthdr.rcvif;
180 
181 	INP_INFO_RLOCK(&V_ripcbinfo);
182 	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
183 		/* XXX inp locking */
184 		if ((in6p->inp_vflag & INP_IPV6) == 0)
185 			continue;
186 		if (in6p->inp_ip_p &&
187 		    in6p->inp_ip_p != proto)
188 			continue;
189 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
190 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
191 			continue;
192 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
193 		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
194 			continue;
195 		if (jailed_without_vnet(in6p->inp_cred)) {
196 			/*
197 			 * Allow raw socket in jail to receive multicast;
198 			 * assume process had PRIV_NETINET_RAW at attach,
199 			 * and fall through into normal filter path if so.
200 			 */
201 			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
202 			    prison_check_ip6(in6p->inp_cred,
203 			    &ip6->ip6_dst) != 0)
204 				continue;
205 		}
206 		INP_RLOCK(in6p);
207 		if (in6p->in6p_cksum != -1) {
208 			RIP6STAT_INC(rip6s_isum);
209 			if (in6_cksum(m, proto, *offp,
210 			    m->m_pkthdr.len - *offp)) {
211 				INP_RUNLOCK(in6p);
212 				RIP6STAT_INC(rip6s_badsum);
213 				continue;
214 			}
215 		}
216 		/*
217 		 * If this raw socket has multicast state, and we
218 		 * have received a multicast, check if this socket
219 		 * should receive it, as multicast filtering is now
220 		 * the responsibility of the transport layer.
221 		 */
222 		if (in6p->in6p_moptions &&
223 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
224 			/*
225 			 * If the incoming datagram is for MLD, allow it
226 			 * through unconditionally to the raw socket.
227 			 *
228 			 * Use the M_RTALERT_MLD flag to check for MLD
229 			 * traffic without having to inspect the mbuf chain
230 			 * more deeply, as all MLDv1/v2 host messages MUST
231 			 * contain the Router Alert option.
232 			 *
233 			 * In the case of MLDv1, we may not have explicitly
234 			 * joined the group, and may have set IFF_ALLMULTI
235 			 * on the interface. im6o_mc_filter() may discard
236 			 * control traffic we actually need to see.
237 			 *
238 			 * Userland multicast routing daemons should continue
239 			 * filter the control traffic appropriately.
240 			 */
241 			int blocked;
242 
243 			blocked = MCAST_PASS;
244 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
245 				struct sockaddr_in6 mcaddr;
246 
247 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
248 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
249 				mcaddr.sin6_family = AF_INET6;
250 				mcaddr.sin6_addr = ip6->ip6_dst;
251 
252 				blocked = im6o_mc_filter(in6p->in6p_moptions,
253 				    ifp,
254 				    (struct sockaddr *)&mcaddr,
255 				    (struct sockaddr *)&fromsa);
256 			}
257 			if (blocked != MCAST_PASS) {
258 				IP6STAT_INC(ip6s_notmember);
259 				INP_RUNLOCK(in6p);
260 				continue;
261 			}
262 		}
263 		if (last != NULL) {
264 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
265 
266 #ifdef IPSEC
267 			/*
268 			 * Check AH/ESP integrity.
269 			 */
270 			if (n && ipsec6_in_reject(n, last)) {
271 				m_freem(n);
272 				IPSEC6STAT_INC(ips_in_polvio);
273 				/* Do not inject data into pcb. */
274 			} else
275 #endif /* IPSEC */
276 			if (n) {
277 				if (last->inp_flags & INP_CONTROLOPTS ||
278 				    last->inp_socket->so_options & SO_TIMESTAMP)
279 					ip6_savecontrol(last, n, &opts);
280 				/* strip intermediate headers */
281 				m_adj(n, *offp);
282 				if (sbappendaddr(&last->inp_socket->so_rcv,
283 						(struct sockaddr *)&fromsa,
284 						 n, opts) == 0) {
285 					m_freem(n);
286 					if (opts)
287 						m_freem(opts);
288 					RIP6STAT_INC(rip6s_fullsock);
289 				} else
290 					sorwakeup(last->inp_socket);
291 				opts = NULL;
292 			}
293 			INP_RUNLOCK(last);
294 		}
295 		last = in6p;
296 	}
297 	INP_INFO_RUNLOCK(&V_ripcbinfo);
298 #ifdef IPSEC
299 	/*
300 	 * Check AH/ESP integrity.
301 	 */
302 	if ((last != NULL) && ipsec6_in_reject(m, last)) {
303 		m_freem(m);
304 		IPSEC6STAT_INC(ips_in_polvio);
305 		IP6STAT_DEC(ip6s_delivered);
306 		/* Do not inject data into pcb. */
307 		INP_RUNLOCK(last);
308 	} else
309 #endif /* IPSEC */
310 	if (last != NULL) {
311 		if (last->inp_flags & INP_CONTROLOPTS ||
312 		    last->inp_socket->so_options & SO_TIMESTAMP)
313 			ip6_savecontrol(last, m, &opts);
314 		/* Strip intermediate headers. */
315 		m_adj(m, *offp);
316 		if (sbappendaddr(&last->inp_socket->so_rcv,
317 		    (struct sockaddr *)&fromsa, m, opts) == 0) {
318 			m_freem(m);
319 			if (opts)
320 				m_freem(opts);
321 			RIP6STAT_INC(rip6s_fullsock);
322 		} else
323 			sorwakeup(last->inp_socket);
324 		INP_RUNLOCK(last);
325 	} else {
326 		RIP6STAT_INC(rip6s_nosock);
327 		if (m->m_flags & M_MCAST)
328 			RIP6STAT_INC(rip6s_nosockmcast);
329 		if (proto == IPPROTO_NONE)
330 			m_freem(m);
331 		else {
332 			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
333 			icmp6_error(m, ICMP6_PARAM_PROB,
334 			    ICMP6_PARAMPROB_NEXTHEADER,
335 			    prvnxtp - mtod(m, char *));
336 		}
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 ip6_hdr *ip6;
346 	struct mbuf *m;
347 	int off = 0;
348 	struct ip6ctlparam *ip6cp = NULL;
349 	const struct sockaddr_in6 *sa6_src = NULL;
350 	void *cmdarg;
351 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
352 
353 	if (sa->sa_family != AF_INET6 ||
354 	    sa->sa_len != sizeof(struct sockaddr_in6))
355 		return;
356 
357 	if ((unsigned)cmd >= PRC_NCMDS)
358 		return;
359 	if (PRC_IS_REDIRECT(cmd))
360 		notify = in6_rtchange, d = NULL;
361 	else if (cmd == PRC_HOSTDEAD)
362 		d = NULL;
363 	else if (inet6ctlerrmap[cmd] == 0)
364 		return;
365 
366 	/*
367 	 * If the parameter is from icmp6, decode it.
368 	 */
369 	if (d != NULL) {
370 		ip6cp = (struct ip6ctlparam *)d;
371 		m = ip6cp->ip6c_m;
372 		ip6 = ip6cp->ip6c_ip6;
373 		off = ip6cp->ip6c_off;
374 		cmdarg = ip6cp->ip6c_cmdarg;
375 		sa6_src = ip6cp->ip6c_src;
376 	} else {
377 		m = NULL;
378 		ip6 = NULL;
379 		cmdarg = NULL;
380 		sa6_src = &sa6_any;
381 	}
382 
383 	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
384 	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
385 }
386 
387 /*
388  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
389  * may have setup with control call.
390  */
391 int
392 #if __STDC__
393 rip6_output(struct mbuf *m, ...)
394 #else
395 rip6_output(m, va_alist)
396 	struct mbuf *m;
397 	va_dcl
398 #endif
399 {
400 	struct mbuf *control;
401 	struct m_tag *mtag;
402 	struct socket *so;
403 	struct sockaddr_in6 *dstsock;
404 	struct in6_addr *dst;
405 	struct ip6_hdr *ip6;
406 	struct inpcb *in6p;
407 	u_int	plen = m->m_pkthdr.len;
408 	int error = 0;
409 	struct ip6_pktopts opt, *optp;
410 	struct ifnet *oifp = NULL;
411 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
412 	int scope_ambiguous = 0;
413 	int use_defzone = 0;
414 	struct in6_addr in6a;
415 	va_list ap;
416 
417 	va_start(ap, m);
418 	so = va_arg(ap, struct socket *);
419 	dstsock = va_arg(ap, struct sockaddr_in6 *);
420 	control = va_arg(ap, struct mbuf *);
421 	va_end(ap);
422 
423 	in6p = sotoinpcb(so);
424 	INP_WLOCK(in6p);
425 
426 	dst = &dstsock->sin6_addr;
427 	if (control != NULL) {
428 		if ((error = ip6_setpktopts(control, &opt,
429 		    in6p->in6p_outputopts, so->so_cred,
430 		    so->so_proto->pr_protocol)) != 0) {
431 			goto bad;
432 		}
433 		optp = &opt;
434 	} else
435 		optp = in6p->in6p_outputopts;
436 
437 	/*
438 	 * Check and convert scope zone ID into internal form.
439 	 *
440 	 * XXX: we may still need to determine the zone later.
441 	 */
442 	if (!(so->so_state & SS_ISCONNECTED)) {
443 		if (!optp || !optp->ip6po_pktinfo ||
444 		    !optp->ip6po_pktinfo->ipi6_ifindex)
445 			use_defzone = V_ip6_use_defzone;
446 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
447 			scope_ambiguous = 1;
448 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
449 			goto bad;
450 	}
451 
452 	/*
453 	 * For an ICMPv6 packet, we should know its type and code to update
454 	 * statistics.
455 	 */
456 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
457 		struct icmp6_hdr *icmp6;
458 		if (m->m_len < sizeof(struct icmp6_hdr) &&
459 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
460 			error = ENOBUFS;
461 			goto bad;
462 		}
463 		icmp6 = mtod(m, struct icmp6_hdr *);
464 		type = icmp6->icmp6_type;
465 		code = icmp6->icmp6_code;
466 	}
467 
468 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
469 	if (m == NULL) {
470 		error = ENOBUFS;
471 		goto bad;
472 	}
473 	ip6 = mtod(m, struct ip6_hdr *);
474 
475 	/*
476 	 * Source address selection.
477 	 */
478 	error = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
479 	    &oifp, &in6a);
480 	if (error)
481 		goto bad;
482 	error = prison_check_ip6(in6p->inp_cred, &in6a);
483 	if (error != 0)
484 		goto bad;
485 	ip6->ip6_src = in6a;
486 
487 	if (oifp && scope_ambiguous) {
488 		/*
489 		 * Application should provide a proper zone ID or the use of
490 		 * default zone IDs should be enabled.  Unfortunately, some
491 		 * applications do not behave as it should, so we need a
492 		 * workaround.  Even if an appropriate ID is not determined
493 		 * (when it's required), if we can determine the outgoing
494 		 * interface. determine the zone ID based on the interface.
495 		 */
496 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
497 		if (error != 0)
498 			goto bad;
499 	}
500 	ip6->ip6_dst = dstsock->sin6_addr;
501 
502 	/*
503 	 * Fill in the rest of the IPv6 header fields.
504 	 */
505 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
506 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
507 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
508 	    (IPV6_VERSION & IPV6_VERSION_MASK);
509 
510 	/*
511 	 * ip6_plen will be filled in ip6_output, so not fill it here.
512 	 */
513 	ip6->ip6_nxt = in6p->inp_ip_p;
514 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
515 
516 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
517 	    in6p->in6p_cksum != -1) {
518 		struct mbuf *n;
519 		int off;
520 		u_int16_t *p;
521 
522 		/* Compute checksum. */
523 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
524 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
525 		else
526 			off = in6p->in6p_cksum;
527 		if (plen < off + 1) {
528 			error = EINVAL;
529 			goto bad;
530 		}
531 		off += sizeof(struct ip6_hdr);
532 
533 		n = m;
534 		while (n && n->m_len <= off) {
535 			off -= n->m_len;
536 			n = n->m_next;
537 		}
538 		if (!n)
539 			goto bad;
540 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
541 		*p = 0;
542 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
543 	}
544 
545 	/*
546 	 * Send RA/RS messages to user land for protection, before sending
547 	 * them to rtadvd/rtsol.
548 	 */
549 	if ((send_sendso_input_hook != NULL) &&
550 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
551 		switch (type) {
552 		case ND_ROUTER_ADVERT:
553 		case ND_ROUTER_SOLICIT:
554 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
555 				sizeof(unsigned short), M_NOWAIT);
556 			if (mtag == NULL)
557 				goto bad;
558 			m_tag_prepend(m, mtag);
559 		}
560 	}
561 
562 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
563 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
564 		if (oifp)
565 			icmp6_ifoutstat_inc(oifp, type, code);
566 		ICMP6STAT_INC(icp6s_outhist[type]);
567 	} else
568 		RIP6STAT_INC(rip6s_opackets);
569 
570 	goto freectl;
571 
572  bad:
573 	if (m)
574 		m_freem(m);
575 
576  freectl:
577 	if (control != NULL) {
578 		ip6_clearpktopts(&opt, -1);
579 		m_freem(control);
580 	}
581 	INP_WUNLOCK(in6p);
582 	return (error);
583 }
584 
585 /*
586  * Raw IPv6 socket option processing.
587  */
588 int
589 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
590 {
591 	struct inpcb *inp;
592 	int error;
593 
594 	if (sopt->sopt_level == IPPROTO_ICMPV6)
595 		/*
596 		 * XXX: is it better to call icmp6_ctloutput() directly
597 		 * from protosw?
598 		 */
599 		return (icmp6_ctloutput(so, sopt));
600 	else if (sopt->sopt_level != IPPROTO_IPV6) {
601 		if (sopt->sopt_level == SOL_SOCKET &&
602 		    sopt->sopt_name == SO_SETFIB) {
603 			inp = sotoinpcb(so);
604 			INP_WLOCK(inp);
605 			inp->inp_inc.inc_fibnum = so->so_fibnum;
606 			INP_WUNLOCK(inp);
607 			return (0);
608 		}
609 		return (EINVAL);
610 	}
611 
612 	error = 0;
613 
614 	switch (sopt->sopt_dir) {
615 	case SOPT_GET:
616 		switch (sopt->sopt_name) {
617 		case MRT6_INIT:
618 		case MRT6_DONE:
619 		case MRT6_ADD_MIF:
620 		case MRT6_DEL_MIF:
621 		case MRT6_ADD_MFC:
622 		case MRT6_DEL_MFC:
623 		case MRT6_PIM:
624 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
625 			    EOPNOTSUPP;
626 			break;
627 		case IPV6_CHECKSUM:
628 			error = ip6_raw_ctloutput(so, sopt);
629 			break;
630 		default:
631 			error = ip6_ctloutput(so, sopt);
632 			break;
633 		}
634 		break;
635 
636 	case SOPT_SET:
637 		switch (sopt->sopt_name) {
638 		case MRT6_INIT:
639 		case MRT6_DONE:
640 		case MRT6_ADD_MIF:
641 		case MRT6_DEL_MIF:
642 		case MRT6_ADD_MFC:
643 		case MRT6_DEL_MFC:
644 		case MRT6_PIM:
645 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
646 			    EOPNOTSUPP;
647 			break;
648 		case IPV6_CHECKSUM:
649 			error = ip6_raw_ctloutput(so, sopt);
650 			break;
651 		default:
652 			error = ip6_ctloutput(so, sopt);
653 			break;
654 		}
655 		break;
656 	}
657 
658 	return (error);
659 }
660 
661 static int
662 rip6_attach(struct socket *so, int proto, struct thread *td)
663 {
664 	struct inpcb *inp;
665 	struct icmp6_filter *filter;
666 	int error;
667 
668 	inp = sotoinpcb(so);
669 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
670 
671 	error = priv_check(td, PRIV_NETINET_RAW);
672 	if (error)
673 		return (error);
674 	error = soreserve(so, rip_sendspace, rip_recvspace);
675 	if (error)
676 		return (error);
677 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
678 	if (filter == NULL)
679 		return (ENOMEM);
680 	INP_INFO_WLOCK(&V_ripcbinfo);
681 	error = in_pcballoc(so, &V_ripcbinfo);
682 	if (error) {
683 		INP_INFO_WUNLOCK(&V_ripcbinfo);
684 		free(filter, M_PCB);
685 		return (error);
686 	}
687 	inp = (struct inpcb *)so->so_pcb;
688 	INP_INFO_WUNLOCK(&V_ripcbinfo);
689 	inp->inp_vflag |= INP_IPV6;
690 	inp->inp_ip_p = (long)proto;
691 	inp->in6p_hops = -1;	/* use kernel default */
692 	inp->in6p_cksum = -1;
693 	inp->in6p_icmp6filt = filter;
694 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
695 	INP_WUNLOCK(inp);
696 	return (0);
697 }
698 
699 static void
700 rip6_detach(struct socket *so)
701 {
702 	struct inpcb *inp;
703 
704 	inp = sotoinpcb(so);
705 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
706 
707 	if (so == V_ip6_mrouter && ip6_mrouter_done)
708 		ip6_mrouter_done();
709 	/* xxx: RSVP */
710 	INP_INFO_WLOCK(&V_ripcbinfo);
711 	INP_WLOCK(inp);
712 	free(inp->in6p_icmp6filt, M_PCB);
713 	in_pcbdetach(inp);
714 	in_pcbfree(inp);
715 	INP_INFO_WUNLOCK(&V_ripcbinfo);
716 }
717 
718 /* XXXRW: This can't ever be called. */
719 static void
720 rip6_abort(struct socket *so)
721 {
722 	struct inpcb *inp;
723 
724 	inp = sotoinpcb(so);
725 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
726 
727 	soisdisconnected(so);
728 }
729 
730 static void
731 rip6_close(struct socket *so)
732 {
733 	struct inpcb *inp;
734 
735 	inp = sotoinpcb(so);
736 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
737 
738 	soisdisconnected(so);
739 }
740 
741 static int
742 rip6_disconnect(struct socket *so)
743 {
744 	struct inpcb *inp;
745 
746 	inp = sotoinpcb(so);
747 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
748 
749 	if ((so->so_state & SS_ISCONNECTED) == 0)
750 		return (ENOTCONN);
751 	inp->in6p_faddr = in6addr_any;
752 	rip6_abort(so);
753 	return (0);
754 }
755 
756 static int
757 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
758 {
759 	struct inpcb *inp;
760 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
761 	struct ifaddr *ifa = NULL;
762 	int error = 0;
763 
764 	inp = sotoinpcb(so);
765 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
766 
767 	if (nam->sa_len != sizeof(*addr))
768 		return (EINVAL);
769 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
770 		return (error);
771 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
772 		return (EADDRNOTAVAIL);
773 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
774 		return (error);
775 
776 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
777 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
778 		return (EADDRNOTAVAIL);
779 	if (ifa != NULL &&
780 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
781 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
782 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
783 		ifa_free(ifa);
784 		return (EADDRNOTAVAIL);
785 	}
786 	if (ifa != NULL)
787 		ifa_free(ifa);
788 	INP_INFO_WLOCK(&V_ripcbinfo);
789 	INP_WLOCK(inp);
790 	inp->in6p_laddr = addr->sin6_addr;
791 	INP_WUNLOCK(inp);
792 	INP_INFO_WUNLOCK(&V_ripcbinfo);
793 	return (0);
794 }
795 
796 static int
797 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
798 {
799 	struct inpcb *inp;
800 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
801 	struct in6_addr in6a;
802 	struct ifnet *ifp = NULL;
803 	int error = 0, scope_ambiguous = 0;
804 
805 	inp = sotoinpcb(so);
806 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
807 
808 	if (nam->sa_len != sizeof(*addr))
809 		return (EINVAL);
810 	if (TAILQ_EMPTY(&V_ifnet))
811 		return (EADDRNOTAVAIL);
812 	if (addr->sin6_family != AF_INET6)
813 		return (EAFNOSUPPORT);
814 
815 	/*
816 	 * Application should provide a proper zone ID or the use of default
817 	 * zone IDs should be enabled.  Unfortunately, some applications do
818 	 * not behave as it should, so we need a workaround.  Even if an
819 	 * appropriate ID is not determined, we'll see if we can determine
820 	 * the outgoing interface.  If we can, determine the zone ID based on
821 	 * the interface below.
822 	 */
823 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
824 		scope_ambiguous = 1;
825 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
826 		return (error);
827 
828 	INP_INFO_WLOCK(&V_ripcbinfo);
829 	INP_WLOCK(inp);
830 	/* Source address selection. XXX: need pcblookup? */
831 	error = in6_selectsrc(addr, inp->in6p_outputopts,
832 	    inp, NULL, so->so_cred, &ifp, &in6a);
833 	if (error) {
834 		INP_WUNLOCK(inp);
835 		INP_INFO_WUNLOCK(&V_ripcbinfo);
836 		return (error);
837 	}
838 
839 	/* XXX: see above */
840 	if (ifp && scope_ambiguous &&
841 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
842 		INP_WUNLOCK(inp);
843 		INP_INFO_WUNLOCK(&V_ripcbinfo);
844 		return (error);
845 	}
846 	inp->in6p_faddr = addr->sin6_addr;
847 	inp->in6p_laddr = in6a;
848 	soisconnected(so);
849 	INP_WUNLOCK(inp);
850 	INP_INFO_WUNLOCK(&V_ripcbinfo);
851 	return (0);
852 }
853 
854 static int
855 rip6_shutdown(struct socket *so)
856 {
857 	struct inpcb *inp;
858 
859 	inp = sotoinpcb(so);
860 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
861 
862 	INP_WLOCK(inp);
863 	socantsendmore(so);
864 	INP_WUNLOCK(inp);
865 	return (0);
866 }
867 
868 static int
869 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
870     struct mbuf *control, struct thread *td)
871 {
872 	struct inpcb *inp;
873 	struct sockaddr_in6 tmp;
874 	struct sockaddr_in6 *dst;
875 	int ret;
876 
877 	inp = sotoinpcb(so);
878 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
879 
880 	/* Always copy sockaddr to avoid overwrites. */
881 	/* Unlocked read. */
882 	if (so->so_state & SS_ISCONNECTED) {
883 		if (nam) {
884 			m_freem(m);
885 			return (EISCONN);
886 		}
887 		/* XXX */
888 		bzero(&tmp, sizeof(tmp));
889 		tmp.sin6_family = AF_INET6;
890 		tmp.sin6_len = sizeof(struct sockaddr_in6);
891 		INP_RLOCK(inp);
892 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
893 		    sizeof(struct in6_addr));
894 		INP_RUNLOCK(inp);
895 		dst = &tmp;
896 	} else {
897 		if (nam == NULL) {
898 			m_freem(m);
899 			return (ENOTCONN);
900 		}
901 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
902 			m_freem(m);
903 			return (EINVAL);
904 		}
905 		tmp = *(struct sockaddr_in6 *)nam;
906 		dst = &tmp;
907 
908 		if (dst->sin6_family == AF_UNSPEC) {
909 			/*
910 			 * XXX: we allow this case for backward
911 			 * compatibility to buggy applications that
912 			 * rely on old (and wrong) kernel behavior.
913 			 */
914 			log(LOG_INFO, "rip6 SEND: address family is "
915 			    "unspec. Assume AF_INET6\n");
916 			dst->sin6_family = AF_INET6;
917 		} else if (dst->sin6_family != AF_INET6) {
918 			m_freem(m);
919 			return(EAFNOSUPPORT);
920 		}
921 	}
922 	ret = rip6_output(m, so, dst, control);
923 	return (ret);
924 }
925 
926 struct pr_usrreqs rip6_usrreqs = {
927 	.pru_abort =		rip6_abort,
928 	.pru_attach =		rip6_attach,
929 	.pru_bind =		rip6_bind,
930 	.pru_connect =		rip6_connect,
931 	.pru_control =		in6_control,
932 	.pru_detach =		rip6_detach,
933 	.pru_disconnect =	rip6_disconnect,
934 	.pru_peeraddr =		in6_getpeeraddr,
935 	.pru_send =		rip6_send,
936 	.pru_shutdown =		rip6_shutdown,
937 	.pru_sockaddr =		in6_getsockaddr,
938 	.pru_close =		rip6_close,
939 };
940