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: stable/10/sys/netinet6/raw_ip6.c 329158 2018-02-12 13:52:58Z ae $");
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
rip6_input(struct mbuf ** mp,int * offp,int proto)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 			icmp6_error(m, ICMP6_PARAM_PROB,
333 			    ICMP6_PARAMPROB_NEXTHEADER,
334 			    ip6_get_prevhdr(m, *offp));
335 		IP6STAT_DEC(ip6s_delivered);
336 	}
337 	return (IPPROTO_DONE);
338 }
339 
340 void
rip6_ctlinput(int cmd,struct sockaddr * sa,void * d)341 rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
342 {
343 	struct ip6_hdr *ip6;
344 	struct mbuf *m;
345 	int off = 0;
346 	struct ip6ctlparam *ip6cp = NULL;
347 	const struct sockaddr_in6 *sa6_src = NULL;
348 	void *cmdarg;
349 	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
350 
351 	if (sa->sa_family != AF_INET6 ||
352 	    sa->sa_len != sizeof(struct sockaddr_in6))
353 		return;
354 
355 	if ((unsigned)cmd >= PRC_NCMDS)
356 		return;
357 	if (PRC_IS_REDIRECT(cmd))
358 		notify = in6_rtchange, d = NULL;
359 	else if (cmd == PRC_HOSTDEAD)
360 		d = NULL;
361 	else if (inet6ctlerrmap[cmd] == 0)
362 		return;
363 
364 	/*
365 	 * If the parameter is from icmp6, decode it.
366 	 */
367 	if (d != NULL) {
368 		ip6cp = (struct ip6ctlparam *)d;
369 		m = ip6cp->ip6c_m;
370 		ip6 = ip6cp->ip6c_ip6;
371 		off = ip6cp->ip6c_off;
372 		cmdarg = ip6cp->ip6c_cmdarg;
373 		sa6_src = ip6cp->ip6c_src;
374 	} else {
375 		m = NULL;
376 		ip6 = NULL;
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 #if __STDC__
rip6_output(struct mbuf * m,...)391 rip6_output(struct mbuf *m, ...)
392 #else
393 rip6_output(m, va_alist)
394 	struct mbuf *m;
395 	va_dcl
396 #endif
397 {
398 	struct mbuf *control;
399 	struct m_tag *mtag;
400 	struct socket *so;
401 	struct sockaddr_in6 *dstsock;
402 	struct in6_addr *dst;
403 	struct ip6_hdr *ip6;
404 	struct inpcb *in6p;
405 	u_int	plen = m->m_pkthdr.len;
406 	int error = 0;
407 	struct ip6_pktopts opt, *optp;
408 	struct ifnet *oifp = NULL;
409 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
410 	int scope_ambiguous = 0;
411 	int use_defzone = 0;
412 	struct in6_addr in6a;
413 	va_list ap;
414 
415 	va_start(ap, m);
416 	so = va_arg(ap, struct socket *);
417 	dstsock = va_arg(ap, struct sockaddr_in6 *);
418 	control = va_arg(ap, struct mbuf *);
419 	va_end(ap);
420 
421 	in6p = sotoinpcb(so);
422 	INP_WLOCK(in6p);
423 
424 	dst = &dstsock->sin6_addr;
425 	if (control != NULL) {
426 		if ((error = ip6_setpktopts(control, &opt,
427 		    in6p->in6p_outputopts, so->so_cred,
428 		    so->so_proto->pr_protocol)) != 0) {
429 			goto bad;
430 		}
431 		optp = &opt;
432 	} else
433 		optp = in6p->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 	/*
474 	 * Source address selection.
475 	 */
476 	error = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
477 	    &oifp, &in6a);
478 	if (error)
479 		goto bad;
480 	error = prison_check_ip6(in6p->inp_cred, &in6a);
481 	if (error != 0)
482 		goto bad;
483 	ip6->ip6_src = in6a;
484 
485 	if (oifp && scope_ambiguous) {
486 		/*
487 		 * Application should provide a proper zone ID or the use of
488 		 * default zone IDs should be enabled.  Unfortunately, some
489 		 * applications do not behave as it should, so we need a
490 		 * workaround.  Even if an appropriate ID is not determined
491 		 * (when it's required), if we can determine the outgoing
492 		 * interface. determine the zone ID based on the interface.
493 		 */
494 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
495 		if (error != 0)
496 			goto bad;
497 	}
498 	ip6->ip6_dst = dstsock->sin6_addr;
499 
500 	/*
501 	 * Fill in the rest of the IPv6 header fields.
502 	 */
503 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
504 	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
505 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
506 	    (IPV6_VERSION & IPV6_VERSION_MASK);
507 
508 	/*
509 	 * ip6_plen will be filled in ip6_output, so not fill it here.
510 	 */
511 	ip6->ip6_nxt = in6p->inp_ip_p;
512 	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
513 
514 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
515 	    in6p->in6p_cksum != -1) {
516 		struct mbuf *n;
517 		int off;
518 		u_int16_t *p;
519 
520 		/* Compute checksum. */
521 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
522 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
523 		else
524 			off = in6p->in6p_cksum;
525 		if (plen < off + 1) {
526 			error = EINVAL;
527 			goto bad;
528 		}
529 		off += sizeof(struct ip6_hdr);
530 
531 		n = m;
532 		while (n && n->m_len <= off) {
533 			off -= n->m_len;
534 			n = n->m_next;
535 		}
536 		if (!n)
537 			goto bad;
538 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
539 		*p = 0;
540 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
541 	}
542 
543 	/*
544 	 * Send RA/RS messages to user land for protection, before sending
545 	 * them to rtadvd/rtsol.
546 	 */
547 	if ((send_sendso_input_hook != NULL) &&
548 	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
549 		switch (type) {
550 		case ND_ROUTER_ADVERT:
551 		case ND_ROUTER_SOLICIT:
552 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
553 				sizeof(unsigned short), M_NOWAIT);
554 			if (mtag == NULL)
555 				goto bad;
556 			m_tag_prepend(m, mtag);
557 		}
558 	}
559 
560 	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
561 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
562 		if (oifp)
563 			icmp6_ifoutstat_inc(oifp, type, code);
564 		ICMP6STAT_INC(icp6s_outhist[type]);
565 	} else
566 		RIP6STAT_INC(rip6s_opackets);
567 
568 	goto freectl;
569 
570  bad:
571 	if (m)
572 		m_freem(m);
573 
574  freectl:
575 	if (control != NULL) {
576 		ip6_clearpktopts(&opt, -1);
577 		m_freem(control);
578 	}
579 	INP_WUNLOCK(in6p);
580 	return (error);
581 }
582 
583 /*
584  * Raw IPv6 socket option processing.
585  */
586 int
rip6_ctloutput(struct socket * so,struct sockopt * sopt)587 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
588 {
589 	struct inpcb *inp;
590 	int error;
591 
592 	if (sopt->sopt_level == IPPROTO_ICMPV6)
593 		/*
594 		 * XXX: is it better to call icmp6_ctloutput() directly
595 		 * from protosw?
596 		 */
597 		return (icmp6_ctloutput(so, sopt));
598 	else if (sopt->sopt_level != IPPROTO_IPV6) {
599 		if (sopt->sopt_level == SOL_SOCKET &&
600 		    sopt->sopt_name == SO_SETFIB) {
601 			inp = sotoinpcb(so);
602 			INP_WLOCK(inp);
603 			inp->inp_inc.inc_fibnum = so->so_fibnum;
604 			INP_WUNLOCK(inp);
605 			return (0);
606 		}
607 		return (EINVAL);
608 	}
609 
610 	error = 0;
611 
612 	switch (sopt->sopt_dir) {
613 	case SOPT_GET:
614 		switch (sopt->sopt_name) {
615 		case MRT6_INIT:
616 		case MRT6_DONE:
617 		case MRT6_ADD_MIF:
618 		case MRT6_DEL_MIF:
619 		case MRT6_ADD_MFC:
620 		case MRT6_DEL_MFC:
621 		case MRT6_PIM:
622 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
623 			    EOPNOTSUPP;
624 			break;
625 		case IPV6_CHECKSUM:
626 			error = ip6_raw_ctloutput(so, sopt);
627 			break;
628 		default:
629 			error = ip6_ctloutput(so, sopt);
630 			break;
631 		}
632 		break;
633 
634 	case SOPT_SET:
635 		switch (sopt->sopt_name) {
636 		case MRT6_INIT:
637 		case MRT6_DONE:
638 		case MRT6_ADD_MIF:
639 		case MRT6_DEL_MIF:
640 		case MRT6_ADD_MFC:
641 		case MRT6_DEL_MFC:
642 		case MRT6_PIM:
643 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
644 			    EOPNOTSUPP;
645 			break;
646 		case IPV6_CHECKSUM:
647 			error = ip6_raw_ctloutput(so, sopt);
648 			break;
649 		default:
650 			error = ip6_ctloutput(so, sopt);
651 			break;
652 		}
653 		break;
654 	}
655 
656 	return (error);
657 }
658 
659 static int
rip6_attach(struct socket * so,int proto,struct thread * td)660 rip6_attach(struct socket *so, int proto, struct thread *td)
661 {
662 	struct inpcb *inp;
663 	struct icmp6_filter *filter;
664 	int error;
665 
666 	inp = sotoinpcb(so);
667 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
668 
669 	error = priv_check(td, PRIV_NETINET_RAW);
670 	if (error)
671 		return (error);
672 	error = soreserve(so, rip_sendspace, rip_recvspace);
673 	if (error)
674 		return (error);
675 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
676 	if (filter == NULL)
677 		return (ENOMEM);
678 	INP_INFO_WLOCK(&V_ripcbinfo);
679 	error = in_pcballoc(so, &V_ripcbinfo);
680 	if (error) {
681 		INP_INFO_WUNLOCK(&V_ripcbinfo);
682 		free(filter, M_PCB);
683 		return (error);
684 	}
685 	inp = (struct inpcb *)so->so_pcb;
686 	INP_INFO_WUNLOCK(&V_ripcbinfo);
687 	inp->inp_vflag |= INP_IPV6;
688 	inp->inp_ip_p = (long)proto;
689 	inp->in6p_hops = -1;	/* use kernel default */
690 	inp->in6p_cksum = -1;
691 	inp->in6p_icmp6filt = filter;
692 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
693 	INP_WUNLOCK(inp);
694 	return (0);
695 }
696 
697 static void
rip6_detach(struct socket * so)698 rip6_detach(struct socket *so)
699 {
700 	struct inpcb *inp;
701 
702 	inp = sotoinpcb(so);
703 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
704 
705 	if (so == V_ip6_mrouter && ip6_mrouter_done)
706 		ip6_mrouter_done();
707 	/* xxx: RSVP */
708 	INP_INFO_WLOCK(&V_ripcbinfo);
709 	INP_WLOCK(inp);
710 	free(inp->in6p_icmp6filt, M_PCB);
711 	in_pcbdetach(inp);
712 	in_pcbfree(inp);
713 	INP_INFO_WUNLOCK(&V_ripcbinfo);
714 }
715 
716 /* XXXRW: This can't ever be called. */
717 static void
rip6_abort(struct socket * so)718 rip6_abort(struct socket *so)
719 {
720 	struct inpcb *inp;
721 
722 	inp = sotoinpcb(so);
723 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
724 
725 	soisdisconnected(so);
726 }
727 
728 static void
rip6_close(struct socket * so)729 rip6_close(struct socket *so)
730 {
731 	struct inpcb *inp;
732 
733 	inp = sotoinpcb(so);
734 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
735 
736 	soisdisconnected(so);
737 }
738 
739 static int
rip6_disconnect(struct socket * so)740 rip6_disconnect(struct socket *so)
741 {
742 	struct inpcb *inp;
743 
744 	inp = sotoinpcb(so);
745 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
746 
747 	if ((so->so_state & SS_ISCONNECTED) == 0)
748 		return (ENOTCONN);
749 	inp->in6p_faddr = in6addr_any;
750 	rip6_abort(so);
751 	return (0);
752 }
753 
754 static int
rip6_bind(struct socket * so,struct sockaddr * nam,struct thread * td)755 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
756 {
757 	struct inpcb *inp;
758 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
759 	struct ifaddr *ifa = NULL;
760 	int error = 0;
761 
762 	inp = sotoinpcb(so);
763 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
764 
765 	if (nam->sa_len != sizeof(*addr))
766 		return (EINVAL);
767 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
768 		return (error);
769 	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
770 		return (EADDRNOTAVAIL);
771 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
772 		return (error);
773 
774 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
775 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
776 		return (EADDRNOTAVAIL);
777 	if (ifa != NULL &&
778 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
779 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
780 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
781 		ifa_free(ifa);
782 		return (EADDRNOTAVAIL);
783 	}
784 	if (ifa != NULL)
785 		ifa_free(ifa);
786 	INP_INFO_WLOCK(&V_ripcbinfo);
787 	INP_WLOCK(inp);
788 	inp->in6p_laddr = addr->sin6_addr;
789 	INP_WUNLOCK(inp);
790 	INP_INFO_WUNLOCK(&V_ripcbinfo);
791 	return (0);
792 }
793 
794 static int
rip6_connect(struct socket * so,struct sockaddr * nam,struct thread * td)795 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
796 {
797 	struct inpcb *inp;
798 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
799 	struct in6_addr in6a;
800 	struct ifnet *ifp = NULL;
801 	int error = 0, scope_ambiguous = 0;
802 
803 	inp = sotoinpcb(so);
804 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
805 
806 	if (nam->sa_len != sizeof(*addr))
807 		return (EINVAL);
808 	if (TAILQ_EMPTY(&V_ifnet))
809 		return (EADDRNOTAVAIL);
810 	if (addr->sin6_family != AF_INET6)
811 		return (EAFNOSUPPORT);
812 
813 	/*
814 	 * Application should provide a proper zone ID or the use of default
815 	 * zone IDs should be enabled.  Unfortunately, some applications do
816 	 * not behave as it should, so we need a workaround.  Even if an
817 	 * appropriate ID is not determined, we'll see if we can determine
818 	 * the outgoing interface.  If we can, determine the zone ID based on
819 	 * the interface below.
820 	 */
821 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
822 		scope_ambiguous = 1;
823 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
824 		return (error);
825 
826 	INP_INFO_WLOCK(&V_ripcbinfo);
827 	INP_WLOCK(inp);
828 	/* Source address selection. XXX: need pcblookup? */
829 	error = in6_selectsrc(addr, inp->in6p_outputopts,
830 	    inp, NULL, so->so_cred, &ifp, &in6a);
831 	if (error) {
832 		INP_WUNLOCK(inp);
833 		INP_INFO_WUNLOCK(&V_ripcbinfo);
834 		return (error);
835 	}
836 
837 	/* XXX: see above */
838 	if (ifp && scope_ambiguous &&
839 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
840 		INP_WUNLOCK(inp);
841 		INP_INFO_WUNLOCK(&V_ripcbinfo);
842 		return (error);
843 	}
844 	inp->in6p_faddr = addr->sin6_addr;
845 	inp->in6p_laddr = in6a;
846 	soisconnected(so);
847 	INP_WUNLOCK(inp);
848 	INP_INFO_WUNLOCK(&V_ripcbinfo);
849 	return (0);
850 }
851 
852 static int
rip6_shutdown(struct socket * so)853 rip6_shutdown(struct socket *so)
854 {
855 	struct inpcb *inp;
856 
857 	inp = sotoinpcb(so);
858 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
859 
860 	INP_WLOCK(inp);
861 	socantsendmore(so);
862 	INP_WUNLOCK(inp);
863 	return (0);
864 }
865 
866 static int
rip6_send(struct socket * so,int flags,struct mbuf * m,struct sockaddr * nam,struct mbuf * control,struct thread * td)867 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
868     struct mbuf *control, struct thread *td)
869 {
870 	struct inpcb *inp;
871 	struct sockaddr_in6 tmp;
872 	struct sockaddr_in6 *dst;
873 	int ret;
874 
875 	inp = sotoinpcb(so);
876 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
877 
878 	/* Always copy sockaddr to avoid overwrites. */
879 	/* Unlocked read. */
880 	if (so->so_state & SS_ISCONNECTED) {
881 		if (nam) {
882 			m_freem(m);
883 			return (EISCONN);
884 		}
885 		/* XXX */
886 		bzero(&tmp, sizeof(tmp));
887 		tmp.sin6_family = AF_INET6;
888 		tmp.sin6_len = sizeof(struct sockaddr_in6);
889 		INP_RLOCK(inp);
890 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
891 		    sizeof(struct in6_addr));
892 		INP_RUNLOCK(inp);
893 		dst = &tmp;
894 	} else {
895 		if (nam == NULL) {
896 			m_freem(m);
897 			return (ENOTCONN);
898 		}
899 		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
900 			m_freem(m);
901 			return (EINVAL);
902 		}
903 		tmp = *(struct sockaddr_in6 *)nam;
904 		dst = &tmp;
905 
906 		if (dst->sin6_family == AF_UNSPEC) {
907 			/*
908 			 * XXX: we allow this case for backward
909 			 * compatibility to buggy applications that
910 			 * rely on old (and wrong) kernel behavior.
911 			 */
912 			log(LOG_INFO, "rip6 SEND: address family is "
913 			    "unspec. Assume AF_INET6\n");
914 			dst->sin6_family = AF_INET6;
915 		} else if (dst->sin6_family != AF_INET6) {
916 			m_freem(m);
917 			return(EAFNOSUPPORT);
918 		}
919 	}
920 	ret = rip6_output(m, so, dst, control);
921 	return (ret);
922 }
923 
924 struct pr_usrreqs rip6_usrreqs = {
925 	.pru_abort =		rip6_abort,
926 	.pru_attach =		rip6_attach,
927 	.pru_bind =		rip6_bind,
928 	.pru_connect =		rip6_connect,
929 	.pru_control =		in6_control,
930 	.pru_detach =		rip6_detach,
931 	.pru_disconnect =	rip6_disconnect,
932 	.pru_peeraddr =		in6_getpeeraddr,
933 	.pru_send =		rip6_send,
934 	.pru_shutdown =		rip6_shutdown,
935 	.pru_sockaddr =		in6_getsockaddr,
936 	.pru_close =		rip6_close,
937 };
938