xref: /freebsd/sys/netinet6/in6_pcb.c (revision 35a04710d7286aa9538917fd7f8e417dbee95b82)
1 /*	$FreeBSD$	*/
2 /*	$KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $	*/
3 
4 /*-
5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the project nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  *
32  */
33 
34 /*-
35  * Copyright (c) 1982, 1986, 1991, 1993
36  *	The Regents of the University of California.  All rights reserved.
37  *
38  * Redistribution and use in source and binary forms, with or without
39  * modification, are permitted provided that the following conditions
40  * are met:
41  * 1. Redistributions of source code must retain the above copyright
42  *    notice, this list of conditions and the following disclaimer.
43  * 2. Redistributions in binary form must reproduce the above copyright
44  *    notice, this list of conditions and the following disclaimer in the
45  *    documentation and/or other materials provided with the distribution.
46  * 4. Neither the name of the University nor the names of its contributors
47  *    may be used to endorse or promote products derived from this software
48  *    without specific prior written permission.
49  *
50  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60  * SUCH DAMAGE.
61  *
62  *	@(#)in_pcb.c	8.2 (Berkeley) 1/4/94
63  */
64 
65 #include "opt_inet.h"
66 #include "opt_inet6.h"
67 #include "opt_ipsec.h"
68 
69 #include <sys/param.h>
70 #include <sys/systm.h>
71 #include <sys/malloc.h>
72 #include <sys/mbuf.h>
73 #include <sys/domain.h>
74 #include <sys/protosw.h>
75 #include <sys/socket.h>
76 #include <sys/socketvar.h>
77 #include <sys/sockio.h>
78 #include <sys/errno.h>
79 #include <sys/time.h>
80 #include <sys/priv.h>
81 #include <sys/proc.h>
82 #include <sys/jail.h>
83 
84 #include <vm/uma.h>
85 
86 #include <net/if.h>
87 #include <net/if_types.h>
88 #include <net/route.h>
89 
90 #include <netinet/in.h>
91 #include <netinet/in_var.h>
92 #include <netinet/in_systm.h>
93 #include <netinet/tcp_var.h>
94 #include <netinet/ip6.h>
95 #include <netinet/ip_var.h>
96 #include <netinet6/ip6_var.h>
97 #include <netinet6/nd6.h>
98 #include <netinet/in_pcb.h>
99 #include <netinet6/in6_pcb.h>
100 #include <netinet6/scope6_var.h>
101 
102 #ifdef IPSEC
103 #include <netipsec/ipsec.h>
104 #include <netipsec/ipsec6.h>
105 #include <netipsec/key.h>
106 #endif /* IPSEC */
107 
108 struct	in6_addr zeroin6_addr;
109 
110 int
111 in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam,
112     struct ucred *cred)
113 {
114 	struct socket *so = inp->inp_socket;
115 	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
116 	struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
117 	u_short	lport = 0;
118 	int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
119 
120 	INP_INFO_WLOCK_ASSERT(pcbinfo);
121 	INP_LOCK_ASSERT(inp);
122 
123 	if (!in6_ifaddr) /* XXX broken! */
124 		return (EADDRNOTAVAIL);
125 	if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
126 		return (EINVAL);
127 	if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
128 		wild = INPLOOKUP_WILDCARD;
129 	if (nam) {
130 		int error;
131 
132 		sin6 = (struct sockaddr_in6 *)nam;
133 		if (nam->sa_len != sizeof(*sin6))
134 			return (EINVAL);
135 		/*
136 		 * family check.
137 		 */
138 		if (nam->sa_family != AF_INET6)
139 			return (EAFNOSUPPORT);
140 
141 		if ((error = sa6_embedscope(sin6, ip6_use_defzone)) != 0)
142 			return(error);
143 
144 		lport = sin6->sin6_port;
145 		if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
146 			/*
147 			 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
148 			 * allow compepte duplication of binding if
149 			 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
150 			 * and a multicast address is bound on both
151 			 * new and duplicated sockets.
152 			 */
153 			if (so->so_options & SO_REUSEADDR)
154 				reuseport = SO_REUSEADDR|SO_REUSEPORT;
155 		} else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
156 			struct ifaddr *ia = NULL;
157 
158 			sin6->sin6_port = 0;		/* yech... */
159 			if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0)
160 				return (EADDRNOTAVAIL);
161 
162 			/*
163 			 * XXX: bind to an anycast address might accidentally
164 			 * cause sending a packet with anycast source address.
165 			 * We should allow to bind to a deprecated address, since
166 			 * the application dares to use it.
167 			 */
168 			if (ia &&
169 			    ((struct in6_ifaddr *)ia)->ia6_flags &
170 			    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) {
171 				return (EADDRNOTAVAIL);
172 			}
173 		}
174 		if (lport) {
175 			struct inpcb *t;
176 
177 			/* GROSS */
178 			if (ntohs(lport) <= ipport_reservedhigh &&
179 			    ntohs(lport) >= ipport_reservedlow &&
180 			    priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT,
181 			    0))
182 				return (EACCES);
183 			if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) &&
184 			    priv_check_cred(so->so_cred,
185 			    PRIV_NETINET_REUSEPORT, 0) != 0) {
186 				t = in6_pcblookup_local(pcbinfo,
187 				    &sin6->sin6_addr, lport,
188 				    INPLOOKUP_WILDCARD);
189 				if (t &&
190 				    ((t->inp_vflag & INP_TIMEWAIT) == 0) &&
191 				    (so->so_type != SOCK_STREAM ||
192 				     IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) &&
193 				    (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
194 				     !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
195 				     (t->inp_socket->so_options & SO_REUSEPORT)
196 				      == 0) && (so->so_cred->cr_uid !=
197 				     t->inp_socket->so_cred->cr_uid))
198 					return (EADDRINUSE);
199 				if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
200 				    IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
201 					struct sockaddr_in sin;
202 
203 					in6_sin6_2_sin(&sin, sin6);
204 					t = in_pcblookup_local(pcbinfo,
205 						sin.sin_addr, lport,
206 						INPLOOKUP_WILDCARD);
207 					if (t &&
208 					    ((t->inp_vflag &
209 					      INP_TIMEWAIT) == 0) &&
210 					    (so->so_type != SOCK_STREAM ||
211 					     ntohl(t->inp_faddr.s_addr) ==
212 					      INADDR_ANY) &&
213 					    (so->so_cred->cr_uid !=
214 					     t->inp_socket->so_cred->cr_uid))
215 						return (EADDRINUSE);
216 				}
217 			}
218 			t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
219 						lport, wild);
220 			if (t && (reuseport & ((t->inp_vflag & INP_TIMEWAIT) ?
221 			    intotw(t)->tw_so_options :
222 			    t->inp_socket->so_options)) == 0)
223 				return (EADDRINUSE);
224 			if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
225 			    IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
226 				struct sockaddr_in sin;
227 
228 				in6_sin6_2_sin(&sin, sin6);
229 				t = in_pcblookup_local(pcbinfo, sin.sin_addr,
230 						       lport, wild);
231 				if (t && t->inp_vflag & INP_TIMEWAIT) {
232 					if ((reuseport &
233 					    intotw(t)->tw_so_options) == 0 &&
234 					    (ntohl(t->inp_laddr.s_addr) !=
235 					     INADDR_ANY || ((inp->inp_vflag &
236 					     INP_IPV6PROTO) ==
237 					     (t->inp_vflag & INP_IPV6PROTO))))
238 						return (EADDRINUSE);
239 				}
240 				else if (t &&
241 				    (reuseport & t->inp_socket->so_options)
242 				    == 0 && (ntohl(t->inp_laddr.s_addr) !=
243 				    INADDR_ANY || INP_SOCKAF(so) ==
244 				     INP_SOCKAF(t->inp_socket)))
245 					return (EADDRINUSE);
246 			}
247 		}
248 		inp->in6p_laddr = sin6->sin6_addr;
249 	}
250 	if (lport == 0) {
251 		int e;
252 		if ((e = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0)
253 			return (e);
254 	}
255 	else {
256 		inp->inp_lport = lport;
257 		if (in_pcbinshash(inp) != 0) {
258 			inp->in6p_laddr = in6addr_any;
259 			inp->inp_lport = 0;
260 			return (EAGAIN);
261 		}
262 	}
263 	return (0);
264 }
265 
266 /*
267  *   Transform old in6_pcbconnect() into an inner subroutine for new
268  *   in6_pcbconnect(): Do some validity-checking on the remote
269  *   address (in mbuf 'nam') and then determine local host address
270  *   (i.e., which interface) to use to access that remote host.
271  *
272  *   This preserves definition of in6_pcbconnect(), while supporting a
273  *   slightly different version for T/TCP.  (This is more than
274  *   a bit of a kludge, but cleaning up the internal interfaces would
275  *   have forced minor changes in every protocol).
276  */
277 int
278 in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam,
279     struct in6_addr **plocal_addr6)
280 {
281 	register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
282 	int error = 0;
283 	struct ifnet *ifp = NULL;
284 	int scope_ambiguous = 0;
285 
286 	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
287 	INP_LOCK_ASSERT(inp);
288 
289 	if (nam->sa_len != sizeof (*sin6))
290 		return (EINVAL);
291 	if (sin6->sin6_family != AF_INET6)
292 		return (EAFNOSUPPORT);
293 	if (sin6->sin6_port == 0)
294 		return (EADDRNOTAVAIL);
295 
296 	if (sin6->sin6_scope_id == 0 && !ip6_use_defzone)
297 		scope_ambiguous = 1;
298 	if ((error = sa6_embedscope(sin6, ip6_use_defzone)) != 0)
299 		return(error);
300 
301 	if (in6_ifaddr) {
302 		/*
303 		 * If the destination address is UNSPECIFIED addr,
304 		 * use the loopback addr, e.g ::1.
305 		 */
306 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
307 			sin6->sin6_addr = in6addr_loopback;
308 	}
309 
310 	/*
311 	 * XXX: in6_selectsrc might replace the bound local address
312 	 * with the address specified by setsockopt(IPV6_PKTINFO).
313 	 * Is it the intended behavior?
314 	 */
315 	*plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts,
316 				      inp->in6p_moptions, NULL,
317 				      &inp->in6p_laddr, &ifp, &error);
318 	if (ifp && scope_ambiguous &&
319 	    (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) {
320 		return(error);
321 	}
322 
323 	if (*plocal_addr6 == 0) {
324 		if (error == 0)
325 			error = EADDRNOTAVAIL;
326 		return (error);
327 	}
328 	/*
329 	 * Don't do pcblookup call here; return interface in
330 	 * plocal_addr6
331 	 * and exit to caller, that will do the lookup.
332 	 */
333 
334 	return (0);
335 }
336 
337 /*
338  * Outer subroutine:
339  * Connect from a socket to a specified address.
340  * Both address and port must be specified in argument sin.
341  * If don't have a local address for this socket yet,
342  * then pick one.
343  */
344 int
345 in6_pcbconnect(register struct inpcb *inp, struct sockaddr *nam,
346     struct ucred *cred)
347 {
348 	struct in6_addr *addr6;
349 	register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
350 	int error;
351 
352 	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
353 	INP_LOCK_ASSERT(inp);
354 
355 	/*
356 	 * Call inner routine, to assign local interface address.
357 	 * in6_pcbladdr() may automatically fill in sin6_scope_id.
358 	 */
359 	if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0)
360 		return (error);
361 
362 	if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr,
363 			       sin6->sin6_port,
364 			      IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
365 			      ? addr6 : &inp->in6p_laddr,
366 			      inp->inp_lport, 0, NULL) != NULL) {
367 		return (EADDRINUSE);
368 	}
369 	if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
370 		if (inp->inp_lport == 0) {
371 			error = in6_pcbbind(inp, (struct sockaddr *)0, cred);
372 			if (error)
373 				return (error);
374 		}
375 		inp->in6p_laddr = *addr6;
376 	}
377 	inp->in6p_faddr = sin6->sin6_addr;
378 	inp->inp_fport = sin6->sin6_port;
379 	/* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
380 	inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
381 	if (inp->in6p_flags & IN6P_AUTOFLOWLABEL)
382 		inp->in6p_flowinfo |=
383 		    (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
384 
385 	in_pcbrehash(inp);
386 
387 	return (0);
388 }
389 
390 void
391 in6_pcbdisconnect(struct inpcb *inp)
392 {
393 
394 	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
395 	INP_LOCK_ASSERT(inp);
396 
397 	bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
398 	inp->inp_fport = 0;
399 	/* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
400 	inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
401 	in_pcbrehash(inp);
402 }
403 
404 void
405 in6_pcbdetach(struct inpcb *inp)
406 {
407 
408 	KASSERT(inp->inp_socket != NULL, ("in6_pcbdetach: inp_socket == NULL"));
409 	inp->inp_socket->so_pcb = NULL;
410 	inp->inp_socket = NULL;
411 }
412 
413 void
414 in6_pcbfree(struct inpcb *inp)
415 {
416 	struct inpcbinfo *ipi = inp->inp_pcbinfo;
417 
418 	KASSERT(inp->inp_socket == NULL, ("in6_pcbfree: inp_socket != NULL"));
419 	INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
420 	INP_LOCK_ASSERT(inp);
421 
422 #ifdef IPSEC
423 	if (inp->in6p_sp != NULL)
424 		ipsec6_delete_pcbpolicy(inp);
425 #endif /* IPSEC */
426 	inp->inp_gencnt = ++ipi->ipi_gencnt;
427 	in_pcbremlists(inp);
428 	ip6_freepcbopts(inp->in6p_outputopts);
429 	ip6_freemoptions(inp->in6p_moptions);
430 	/* Check and free IPv4 related resources in case of mapped addr */
431 	if (inp->inp_options)
432 		(void)m_free(inp->inp_options);
433 	if (inp->inp_moptions != NULL)
434 		inp_freemoptions(inp->inp_moptions);
435 	inp->inp_vflag = 0;
436 	INP_UNLOCK(inp);
437 	uma_zfree(ipi->ipi_zone, inp);
438 }
439 
440 struct sockaddr *
441 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
442 {
443 	struct sockaddr_in6 *sin6;
444 
445 	MALLOC(sin6, struct sockaddr_in6 *, sizeof *sin6, M_SONAME, M_WAITOK);
446 	bzero(sin6, sizeof *sin6);
447 	sin6->sin6_family = AF_INET6;
448 	sin6->sin6_len = sizeof(*sin6);
449 	sin6->sin6_port = port;
450 	sin6->sin6_addr = *addr_p;
451 	(void)sa6_recoverscope(sin6); /* XXX: should catch errors */
452 
453 	return (struct sockaddr *)sin6;
454 }
455 
456 struct sockaddr *
457 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p)
458 {
459 	struct sockaddr_in sin;
460 	struct sockaddr_in6 *sin6_p;
461 
462 	bzero(&sin, sizeof sin);
463 	sin.sin_family = AF_INET;
464 	sin.sin_len = sizeof(sin);
465 	sin.sin_port = port;
466 	sin.sin_addr = *addr_p;
467 
468 	MALLOC(sin6_p, struct sockaddr_in6 *, sizeof *sin6_p, M_SONAME,
469 		M_WAITOK);
470 	in6_sin_2_v4mapsin6(&sin, sin6_p);
471 
472 	return (struct sockaddr *)sin6_p;
473 }
474 
475 int
476 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
477 {
478 	register struct inpcb *inp;
479 	struct in6_addr addr;
480 	in_port_t port;
481 
482 	inp = sotoinpcb(so);
483 	KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL"));
484 
485 	INP_LOCK(inp);
486 	port = inp->inp_lport;
487 	addr = inp->in6p_laddr;
488 	INP_UNLOCK(inp);
489 
490 	*nam = in6_sockaddr(port, &addr);
491 	return 0;
492 }
493 
494 int
495 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
496 {
497 	struct inpcb *inp;
498 	struct in6_addr addr;
499 	in_port_t port;
500 
501 	inp = sotoinpcb(so);
502 	KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL"));
503 
504 	INP_LOCK(inp);
505 	port = inp->inp_fport;
506 	addr = inp->in6p_faddr;
507 	INP_UNLOCK(inp);
508 
509 	*nam = in6_sockaddr(port, &addr);
510 	return 0;
511 }
512 
513 int
514 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
515 {
516 	struct	inpcb *inp;
517 	int	error;
518 
519 	inp = sotoinpcb(so);
520 	KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL"));
521 
522 	if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
523 		error = in_getsockaddr(so, nam);
524 		if (error == 0)
525 			in6_sin_2_v4mapsin6_in_sock(nam);
526 	} else {
527 		/* scope issues will be handled in in6_getsockaddr(). */
528 		error = in6_getsockaddr(so, nam);
529 	}
530 
531 	return error;
532 }
533 
534 int
535 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
536 {
537 	struct	inpcb *inp;
538 	int	error;
539 
540 	inp = sotoinpcb(so);
541 	KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL"));
542 
543 	if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
544 		error = in_getpeeraddr(so, nam);
545 		if (error == 0)
546 			in6_sin_2_v4mapsin6_in_sock(nam);
547 	} else
548 	/* scope issues will be handled in in6_getpeeraddr(). */
549 	error = in6_getpeeraddr(so, nam);
550 
551 	return error;
552 }
553 
554 /*
555  * Pass some notification to all connections of a protocol
556  * associated with address dst.  The local address and/or port numbers
557  * may be specified to limit the search.  The "usual action" will be
558  * taken, depending on the ctlinput cmd.  The caller must filter any
559  * cmds that are uninteresting (e.g., no error in the map).
560  * Call the protocol specific routine (if any) to report
561  * any errors for each matching socket.
562  */
563 void
564 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst,
565     u_int fport_arg, const struct sockaddr *src, u_int lport_arg,
566     int cmd, void *cmdarg,
567     struct inpcb *(*notify) __P((struct inpcb *, int)))
568 {
569 	struct inpcbhead *head;
570 	struct inpcb *inp, *ninp;
571 	struct sockaddr_in6 sa6_src, *sa6_dst;
572 	u_short	fport = fport_arg, lport = lport_arg;
573 	u_int32_t flowinfo;
574 	int errno;
575 
576 	if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
577 		return;
578 
579 	sa6_dst = (struct sockaddr_in6 *)dst;
580 	if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
581 		return;
582 
583 	/*
584 	 * note that src can be NULL when we get notify by local fragmentation.
585 	 */
586 	sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
587 	flowinfo = sa6_src.sin6_flowinfo;
588 
589 	/*
590 	 * Redirects go to all references to the destination,
591 	 * and use in6_rtchange to invalidate the route cache.
592 	 * Dead host indications: also use in6_rtchange to invalidate
593 	 * the cache, and deliver the error to all the sockets.
594 	 * Otherwise, if we have knowledge of the local port and address,
595 	 * deliver only to that socket.
596 	 */
597 	if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
598 		fport = 0;
599 		lport = 0;
600 		bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
601 
602 		if (cmd != PRC_HOSTDEAD)
603 			notify = in6_rtchange;
604 	}
605 	errno = inet6ctlerrmap[cmd];
606 	head = pcbinfo->ipi_listhead;
607 	INP_INFO_WLOCK(pcbinfo);
608 	for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) {
609 		INP_LOCK(inp);
610 		ninp = LIST_NEXT(inp, inp_list);
611 
612 		if ((inp->inp_vflag & INP_IPV6) == 0) {
613 			INP_UNLOCK(inp);
614 			continue;
615 		}
616 
617 		/*
618 		 * If the error designates a new path MTU for a destination
619 		 * and the application (associated with this socket) wanted to
620 		 * know the value, notify. Note that we notify for all
621 		 * disconnected sockets if the corresponding application
622 		 * wanted. This is because some UDP applications keep sending
623 		 * sockets disconnected.
624 		 * XXX: should we avoid to notify the value to TCP sockets?
625 		 */
626 		if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 &&
627 		    (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
628 		     IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) {
629 			ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst,
630 					(u_int32_t *)cmdarg);
631 		}
632 
633 		/*
634 		 * Detect if we should notify the error. If no source and
635 		 * destination ports are specifed, but non-zero flowinfo and
636 		 * local address match, notify the error. This is the case
637 		 * when the error is delivered with an encrypted buffer
638 		 * by ESP. Otherwise, just compare addresses and ports
639 		 * as usual.
640 		 */
641 		if (lport == 0 && fport == 0 && flowinfo &&
642 		    inp->inp_socket != NULL &&
643 		    flowinfo == (inp->in6p_flowinfo & IPV6_FLOWLABEL_MASK) &&
644 		    IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
645 			goto do_notify;
646 		else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
647 					     &sa6_dst->sin6_addr) ||
648 			 inp->inp_socket == 0 ||
649 			 (lport && inp->inp_lport != lport) ||
650 			 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
651 			  !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
652 					      &sa6_src.sin6_addr)) ||
653 			 (fport && inp->inp_fport != fport)) {
654 			INP_UNLOCK(inp);
655 			continue;
656 		}
657 
658 	  do_notify:
659 		if (notify) {
660 			if ((*notify)(inp, errno))
661 				INP_UNLOCK(inp);
662 		} else
663 			INP_UNLOCK(inp);
664 	}
665 	INP_INFO_WUNLOCK(pcbinfo);
666 }
667 
668 /*
669  * Lookup a PCB based on the local address and port.
670  */
671 struct inpcb *
672 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
673     u_int lport_arg, int wild_okay)
674 {
675 	register struct inpcb *inp;
676 	int matchwild = 3, wildcard;
677 	u_short lport = lport_arg;
678 
679 	INP_INFO_WLOCK_ASSERT(pcbinfo);
680 
681 	if (!wild_okay) {
682 		struct inpcbhead *head;
683 		/*
684 		 * Look for an unconnected (wildcard foreign addr) PCB that
685 		 * matches the local address and port we're looking for.
686 		 */
687 		head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
688 		    0, pcbinfo->ipi_hashmask)];
689 		LIST_FOREACH(inp, head, inp_hash) {
690 			if ((inp->inp_vflag & INP_IPV6) == 0)
691 				continue;
692 			if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
693 			    IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
694 			    inp->inp_lport == lport) {
695 				/*
696 				 * Found.
697 				 */
698 				return (inp);
699 			}
700 		}
701 		/*
702 		 * Not found.
703 		 */
704 		return (NULL);
705 	} else {
706 		struct inpcbporthead *porthash;
707 		struct inpcbport *phd;
708 		struct inpcb *match = NULL;
709 		/*
710 		 * Best fit PCB lookup.
711 		 *
712 		 * First see if this local port is in use by looking on the
713 		 * port hash list.
714 		 */
715 		porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
716 		    pcbinfo->ipi_porthashmask)];
717 		LIST_FOREACH(phd, porthash, phd_hash) {
718 			if (phd->phd_port == lport)
719 				break;
720 		}
721 		if (phd != NULL) {
722 			/*
723 			 * Port is in use by one or more PCBs. Look for best
724 			 * fit.
725 			 */
726 			LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
727 				wildcard = 0;
728 				if ((inp->inp_vflag & INP_IPV6) == 0)
729 					continue;
730 				if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
731 					wildcard++;
732 				if (!IN6_IS_ADDR_UNSPECIFIED(
733 					&inp->in6p_laddr)) {
734 					if (IN6_IS_ADDR_UNSPECIFIED(laddr))
735 						wildcard++;
736 					else if (!IN6_ARE_ADDR_EQUAL(
737 						&inp->in6p_laddr, laddr))
738 						continue;
739 				} else {
740 					if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
741 						wildcard++;
742 				}
743 				if (wildcard < matchwild) {
744 					match = inp;
745 					matchwild = wildcard;
746 					if (matchwild == 0) {
747 						break;
748 					}
749 				}
750 			}
751 		}
752 		return (match);
753 	}
754 }
755 
756 void
757 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
758 {
759 	struct in6pcb *in6p;
760 	struct ip6_moptions *im6o;
761 	struct in6_multi_mship *imm, *nimm;
762 
763 	INP_INFO_RLOCK(pcbinfo);
764 	LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) {
765 		INP_LOCK(in6p);
766 		im6o = in6p->in6p_moptions;
767 		if ((in6p->inp_vflag & INP_IPV6) &&
768 		    im6o) {
769 			/*
770 			 * Unselect the outgoing interface if it is being
771 			 * detached.
772 			 */
773 			if (im6o->im6o_multicast_ifp == ifp)
774 				im6o->im6o_multicast_ifp = NULL;
775 
776 			/*
777 			 * Drop multicast group membership if we joined
778 			 * through the interface being detached.
779 			 * XXX controversial - is it really legal for kernel
780 			 * to force this?
781 			 */
782 			for (imm = im6o->im6o_memberships.lh_first;
783 			     imm != NULL; imm = nimm) {
784 				nimm = imm->i6mm_chain.le_next;
785 				if (imm->i6mm_maddr->in6m_ifp == ifp) {
786 					LIST_REMOVE(imm, i6mm_chain);
787 					in6_delmulti(imm->i6mm_maddr);
788 					free(imm, M_IP6MADDR);
789 				}
790 			}
791 		}
792 		INP_UNLOCK(in6p);
793 	}
794 	INP_INFO_RUNLOCK(pcbinfo);
795 }
796 
797 /*
798  * Check for alternatives when higher level complains
799  * about service problems.  For now, invalidate cached
800  * routing information.  If the route was created dynamically
801  * (by a redirect), time to try a default gateway again.
802  */
803 void
804 in6_losing(struct inpcb *in6p)
805 {
806 
807 	/*
808 	 * We don't store route pointers in the routing table anymore
809 	 */
810 	return;
811 }
812 
813 /*
814  * After a routing change, flush old routing
815  * and allocate a (hopefully) better one.
816  */
817 struct inpcb *
818 in6_rtchange(struct inpcb *inp, int errno)
819 {
820 	/*
821 	 * We don't store route pointers in the routing table anymore
822 	 */
823 	return inp;
824 }
825 
826 /*
827  * Lookup PCB in hash list.
828  */
829 struct inpcb *
830 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
831     u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
832     int wildcard, struct ifnet *ifp)
833 {
834 	struct inpcbhead *head;
835 	register struct inpcb *inp;
836 	u_short fport = fport_arg, lport = lport_arg;
837 	int faith;
838 
839 	INP_INFO_RLOCK_ASSERT(pcbinfo);
840 
841 	if (faithprefix_p != NULL)
842 		faith = (*faithprefix_p)(laddr);
843 	else
844 		faith = 0;
845 
846 	/*
847 	 * First look for an exact match.
848 	 */
849 	head = &pcbinfo->ipi_hashbase[
850 	    INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport,
851 	    pcbinfo->ipi_hashmask)];
852 	LIST_FOREACH(inp, head, inp_hash) {
853 		if ((inp->inp_vflag & INP_IPV6) == 0)
854 			continue;
855 		if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
856 		    IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
857 		    inp->inp_fport == fport &&
858 		    inp->inp_lport == lport) {
859 			/*
860 			 * Found.
861 			 */
862 			return (inp);
863 		}
864 	}
865 	if (wildcard) {
866 		struct inpcb *local_wild = NULL;
867 
868 		head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
869 		    0, pcbinfo->ipi_hashmask)];
870 		LIST_FOREACH(inp, head, inp_hash) {
871 			if ((inp->inp_vflag & INP_IPV6) == 0)
872 				continue;
873 			if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
874 			    inp->inp_lport == lport) {
875 				if (faith && (inp->inp_flags & INP_FAITH) == 0)
876 					continue;
877 				if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
878 						       laddr))
879 					return (inp);
880 				else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
881 					local_wild = inp;
882 			}
883 		}
884 		return (local_wild);
885 	}
886 
887 	/*
888 	 * Not found.
889 	 */
890 	return (NULL);
891 }
892 
893 void
894 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
895 {
896 	struct ip6_hdr *ip;
897 
898 	ip = mtod(m, struct ip6_hdr *);
899 	bzero(sin6, sizeof(*sin6));
900 	sin6->sin6_len = sizeof(*sin6);
901 	sin6->sin6_family = AF_INET6;
902 	sin6->sin6_addr = ip->ip6_src;
903 
904 	(void)sa6_recoverscope(sin6); /* XXX: should catch errors... */
905 
906 	return;
907 }
908