xref: /netbsd-src/sys/netinet/in_pcb.c (revision 1ad9454efb13a65cd7535ccf867508cb14d9d30e)
1 /*	$NetBSD: in_pcb.c,v 1.105 2006/09/19 21:42:30 elad Exp $	*/
2 
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) 1998 The NetBSD Foundation, Inc.
34  * All rights reserved.
35  *
36  * This code is derived from software contributed to The NetBSD Foundation
37  * by Public Access Networks Corporation ("Panix").  It was developed under
38  * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
39  *
40  * Redistribution and use in source and binary forms, with or without
41  * modification, are permitted provided that the following conditions
42  * are met:
43  * 1. Redistributions of source code must retain the above copyright
44  *    notice, this list of conditions and the following disclaimer.
45  * 2. Redistributions in binary form must reproduce the above copyright
46  *    notice, this list of conditions and the following disclaimer in the
47  *    documentation and/or other materials provided with the distribution.
48  * 3. All advertising materials mentioning features or use of this software
49  *    must display the following acknowledgement:
50  *	This product includes software developed by the NetBSD
51  *	Foundation, Inc. and its contributors.
52  * 4. Neither the name of The NetBSD Foundation nor the names of its
53  *    contributors may be used to endorse or promote products derived
54  *    from this software without specific prior written permission.
55  *
56  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
57  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
58  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
59  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
60  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
61  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
62  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
63  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
64  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
65  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
66  * POSSIBILITY OF SUCH DAMAGE.
67  */
68 
69 /*
70  * Copyright (c) 1982, 1986, 1991, 1993, 1995
71  *	The Regents of the University of California.  All rights reserved.
72  *
73  * Redistribution and use in source and binary forms, with or without
74  * modification, are permitted provided that the following conditions
75  * are met:
76  * 1. Redistributions of source code must retain the above copyright
77  *    notice, this list of conditions and the following disclaimer.
78  * 2. Redistributions in binary form must reproduce the above copyright
79  *    notice, this list of conditions and the following disclaimer in the
80  *    documentation and/or other materials provided with the distribution.
81  * 3. Neither the name of the University nor the names of its contributors
82  *    may be used to endorse or promote products derived from this software
83  *    without specific prior written permission.
84  *
85  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
86  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
87  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
88  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
89  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
90  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
91  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
92  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
93  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
94  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
95  * SUCH DAMAGE.
96  *
97  *	@(#)in_pcb.c	8.4 (Berkeley) 5/24/95
98  */
99 
100 #include <sys/cdefs.h>
101 __KERNEL_RCSID(0, "$NetBSD: in_pcb.c,v 1.105 2006/09/19 21:42:30 elad Exp $");
102 
103 #include "opt_inet.h"
104 #include "opt_ipsec.h"
105 
106 #include <sys/param.h>
107 #include <sys/systm.h>
108 #include <sys/malloc.h>
109 #include <sys/mbuf.h>
110 #include <sys/protosw.h>
111 #include <sys/socket.h>
112 #include <sys/socketvar.h>
113 #include <sys/ioctl.h>
114 #include <sys/errno.h>
115 #include <sys/time.h>
116 #include <sys/pool.h>
117 #include <sys/proc.h>
118 #include <sys/kauth.h>
119 
120 #include <net/if.h>
121 #include <net/route.h>
122 
123 #include <netinet/in.h>
124 #include <netinet/in_systm.h>
125 #include <netinet/ip.h>
126 #include <netinet/in_pcb.h>
127 #include <netinet/in_var.h>
128 #include <netinet/ip_var.h>
129 
130 #ifdef INET6
131 #include <netinet/ip6.h>
132 #include <netinet6/ip6_var.h>
133 #include <netinet6/in6_pcb.h>
134 #endif
135 
136 #ifdef IPSEC
137 #include <netinet6/ipsec.h>
138 #include <netkey/key.h>
139 #elif FAST_IPSEC
140 #include <netipsec/ipsec.h>
141 #include <netipsec/key.h>
142 #endif /* IPSEC */
143 
144 struct	in_addr zeroin_addr;
145 
146 #define	INPCBHASH_PORT(table, lport) \
147 	&(table)->inpt_porthashtbl[ntohs(lport) & (table)->inpt_porthash]
148 #define	INPCBHASH_BIND(table, laddr, lport) \
149 	&(table)->inpt_bindhashtbl[ \
150 	    ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash]
151 #define	INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \
152 	&(table)->inpt_connecthashtbl[ \
153 	    ((ntohl((faddr).s_addr) + ntohs(fport)) + \
154 	     (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash]
155 
156 int	anonportmin = IPPORT_ANONMIN;
157 int	anonportmax = IPPORT_ANONMAX;
158 int	lowportmin  = IPPORT_RESERVEDMIN;
159 int	lowportmax  = IPPORT_RESERVEDMAX;
160 
161 POOL_INIT(inpcb_pool, sizeof(struct inpcb), 0, 0, 0, "inpcbpl", NULL);
162 
163 void
164 in_pcbinit(struct inpcbtable *table, int bindhashsize, int connecthashsize)
165 {
166 
167 	CIRCLEQ_INIT(&table->inpt_queue);
168 	table->inpt_porthashtbl = hashinit(bindhashsize, HASH_LIST, M_PCB,
169 	    M_WAITOK, &table->inpt_porthash);
170 	table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, M_PCB,
171 	    M_WAITOK, &table->inpt_bindhash);
172 	table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST,
173 	    M_PCB, M_WAITOK, &table->inpt_connecthash);
174 	table->inpt_lastlow = IPPORT_RESERVEDMAX;
175 	table->inpt_lastport = (u_int16_t)anonportmax;
176 }
177 
178 int
179 in_pcballoc(struct socket *so, void *v)
180 {
181 	struct inpcbtable *table = v;
182 	struct inpcb *inp;
183 	int s;
184 #if defined(IPSEC) || defined(FAST_IPSEC)
185 	int error;
186 #endif
187 
188 	inp = pool_get(&inpcb_pool, PR_NOWAIT);
189 	if (inp == NULL)
190 		return (ENOBUFS);
191 	bzero((caddr_t)inp, sizeof(*inp));
192 	inp->inp_af = AF_INET;
193 	inp->inp_table = table;
194 	inp->inp_socket = so;
195 	inp->inp_errormtu = -1;
196 #if defined(IPSEC) || defined(FAST_IPSEC)
197 	error = ipsec_init_pcbpolicy(so, &inp->inp_sp);
198 	if (error != 0) {
199 		pool_put(&inpcb_pool, inp);
200 		return error;
201 	}
202 #endif
203 	so->so_pcb = inp;
204 	s = splnet();
205 	CIRCLEQ_INSERT_HEAD(&table->inpt_queue, &inp->inp_head,
206 	    inph_queue);
207 	LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head,
208 	    inph_lhash);
209 	in_pcbstate(inp, INP_ATTACHED);
210 	splx(s);
211 	return (0);
212 }
213 
214 int
215 in_pcbbind(void *v, struct mbuf *nam, struct lwp *l)
216 {
217 	struct in_ifaddr *ia = NULL;
218 	struct inpcb *inp = v;
219 	struct socket *so = inp->inp_socket;
220 	struct inpcbtable *table = inp->inp_table;
221 	struct sockaddr_in *sin = NULL; /* XXXGCC */
222 	u_int16_t lport = 0;
223 	int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
224 
225 	if (inp->inp_af != AF_INET)
226 		return (EINVAL);
227 
228 	if (TAILQ_FIRST(&in_ifaddrhead) == 0)
229 		return (EADDRNOTAVAIL);
230 	if (inp->inp_lport || !in_nullhost(inp->inp_laddr))
231 		return (EINVAL);
232 	if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
233 		wild = 1;
234 	if (nam == 0)
235 		goto noname;
236 	sin = mtod(nam, struct sockaddr_in *);
237 	if (nam->m_len != sizeof (*sin))
238 		return (EINVAL);
239 	if (sin->sin_family != AF_INET)
240 		return (EAFNOSUPPORT);
241 	lport = sin->sin_port;
242 	if (IN_MULTICAST(sin->sin_addr.s_addr)) {
243 		/*
244 		 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
245 		 * allow complete duplication of binding if
246 		 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
247 		 * and a multicast address is bound on both
248 		 * new and duplicated sockets.
249 		 */
250 		if (so->so_options & SO_REUSEADDR)
251 			reuseport = SO_REUSEADDR|SO_REUSEPORT;
252 	} else if (!in_nullhost(sin->sin_addr)) {
253 		sin->sin_port = 0;		/* yech... */
254 		INADDR_TO_IA(sin->sin_addr, ia);
255 		/* check for broadcast addresses */
256 		if (ia == NULL)
257 			ia = ifatoia(ifa_ifwithaddr(sintosa(sin)));
258 		if (ia == NULL)
259 			return (EADDRNOTAVAIL);
260 	}
261 	if (lport) {
262 		struct inpcb *t;
263 #ifdef INET6
264 		struct in6pcb *t6;
265 		struct in6_addr mapped;
266 #endif
267 #ifndef IPNOPRIVPORTS
268 		/* GROSS */
269 		if (ntohs(lport) < IPPORT_RESERVED &&
270 		    (l == 0 || kauth_authorize_network(l->l_cred,
271 		    KAUTH_NETWORK_BIND,
272 		    KAUTH_REQ_NETWORK_BIND_PRIVPORT, so, sin,
273 		    NULL)))
274 			return (EACCES);
275 #endif
276 #ifdef INET6
277 		memset(&mapped, 0, sizeof(mapped));
278 		mapped.s6_addr16[5] = 0xffff;
279 		memcpy(&mapped.s6_addr32[3], &sin->sin_addr,
280 		    sizeof(mapped.s6_addr32[3]));
281 		t6 = in6_pcblookup_port(table, &mapped, lport, wild);
282 		if (t6 && (reuseport & t6->in6p_socket->so_options) == 0)
283 			return (EADDRINUSE);
284 #endif
285 		if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) {
286 			t = in_pcblookup_port(table, sin->sin_addr, lport, 1);
287 		/*
288 		 * XXX:	investigate ramifications of loosening this
289 		 *	restriction so that as long as both ports have
290 		 *	SO_REUSEPORT allow the bind
291 		 */
292 			if (t &&
293 			    (!in_nullhost(sin->sin_addr) ||
294 			     !in_nullhost(t->inp_laddr) ||
295 			     (t->inp_socket->so_options & SO_REUSEPORT) == 0)
296 			    && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) {
297 				return (EADDRINUSE);
298 			}
299 		}
300 		t = in_pcblookup_port(table, sin->sin_addr, lport, wild);
301 		if (t && (reuseport & t->inp_socket->so_options) == 0)
302 			return (EADDRINUSE);
303 	}
304 	inp->inp_laddr = sin->sin_addr;
305 
306 noname:
307 	if (lport == 0) {
308 		int	   cnt;
309 		u_int16_t  mymin, mymax;
310 		u_int16_t *lastport;
311 
312 		if (inp->inp_flags & INP_LOWPORT) {
313 #ifndef IPNOPRIVPORTS
314 			if (l == 0 || kauth_authorize_network(l->l_cred,
315 			    KAUTH_NETWORK_BIND,
316 			    KAUTH_REQ_NETWORK_BIND_PRIVPORT, so,
317 			    sin, NULL))
318 				return (EACCES);
319 #endif
320 			mymin = lowportmin;
321 			mymax = lowportmax;
322 			lastport = &table->inpt_lastlow;
323 		} else {
324 			mymin = anonportmin;
325 			mymax = anonportmax;
326 			lastport = &table->inpt_lastport;
327 		}
328 		if (mymin > mymax) {	/* sanity check */
329 			u_int16_t swp;
330 
331 			swp = mymin;
332 			mymin = mymax;
333 			mymax = swp;
334 		}
335 
336 		lport = *lastport - 1;
337 		for (cnt = mymax - mymin + 1; cnt; cnt--, lport--) {
338 			if (lport < mymin || lport > mymax)
339 				lport = mymax;
340 			if (!in_pcblookup_port(table, inp->inp_laddr,
341 			    htons(lport), 1))
342 				goto found;
343 		}
344 		if (!in_nullhost(inp->inp_laddr))
345 			inp->inp_laddr.s_addr = INADDR_ANY;
346 		return (EAGAIN);
347 	found:
348 		inp->inp_flags |= INP_ANONPORT;
349 		*lastport = lport;
350 		lport = htons(lport);
351 	}
352 	inp->inp_lport = lport;
353 	LIST_REMOVE(&inp->inp_head, inph_lhash);
354 	LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head,
355 	    inph_lhash);
356 	in_pcbstate(inp, INP_BOUND);
357 	return (0);
358 }
359 
360 /*
361  * Connect from a socket to a specified address.
362  * Both address and port must be specified in argument sin.
363  * If don't have a local address for this socket yet,
364  * then pick one.
365  */
366 int
367 in_pcbconnect(void *v, struct mbuf *nam, struct lwp *l)
368 {
369 	struct inpcb *inp = v;
370 	struct in_ifaddr *ia = NULL;
371 	struct sockaddr_in *ifaddr = NULL;
372 	struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *);
373 	int error;
374 
375 	if (inp->inp_af != AF_INET)
376 		return (EINVAL);
377 
378 	if (nam->m_len != sizeof (*sin))
379 		return (EINVAL);
380 	if (sin->sin_family != AF_INET)
381 		return (EAFNOSUPPORT);
382 	if (sin->sin_port == 0)
383 		return (EADDRNOTAVAIL);
384 	if (TAILQ_FIRST(&in_ifaddrhead) != 0) {
385 		/*
386 		 * If the destination address is INADDR_ANY,
387 		 * use any local address (likely loopback).
388 		 * If the supplied address is INADDR_BROADCAST,
389 		 * use the broadcast address of an interface
390 		 * which supports broadcast. (loopback does not)
391 		 */
392 
393 		if (in_nullhost(sin->sin_addr)) {
394 			sin->sin_addr =
395 			    TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr;
396 		} else if (sin->sin_addr.s_addr == INADDR_BROADCAST) {
397 			TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
398 				if (ia->ia_ifp->if_flags & IFF_BROADCAST) {
399 					sin->sin_addr =
400 					    ia->ia_broadaddr.sin_addr;
401 					break;
402 				}
403 			}
404 		}
405 	}
406 	/*
407 	 * If we haven't bound which network number to use as ours,
408 	 * we will use the number of the outgoing interface.
409 	 * This depends on having done a routing lookup, which
410 	 * we will probably have to do anyway, so we might
411 	 * as well do it now.  On the other hand if we are
412 	 * sending to multiple destinations we may have already
413 	 * done the lookup, so see if we can use the route
414 	 * from before.  In any case, we only
415 	 * chose a port number once, even if sending to multiple
416 	 * destinations.
417 	 */
418 	if (in_nullhost(inp->inp_laddr)) {
419 		int xerror;
420 		ifaddr = in_selectsrc(sin, &inp->inp_route,
421 		    inp->inp_socket->so_options, inp->inp_moptions, &xerror);
422 		if (ifaddr == NULL) {
423 			if (xerror == 0)
424 				xerror = EADDRNOTAVAIL;
425 			return xerror;
426 		}
427 		INADDR_TO_IA(ifaddr->sin_addr, ia);
428 		if (ia == NULL)
429 			return (EADDRNOTAVAIL);
430 	}
431 	if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port,
432 	    !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr,
433 	    inp->inp_lport) != 0)
434 		return (EADDRINUSE);
435 	if (in_nullhost(inp->inp_laddr)) {
436 		if (inp->inp_lport == 0) {
437 			error = in_pcbbind(inp, NULL, l);
438 			/*
439 			 * This used to ignore the return value
440 			 * completely, but we need to check for
441 			 * ephemeral port shortage.
442 			 * And attempts to request low ports if not root.
443 			 */
444 			if (error != 0)
445 				return (error);
446 		}
447 		inp->inp_laddr = ifaddr->sin_addr;
448 	}
449 	inp->inp_faddr = sin->sin_addr;
450 	inp->inp_fport = sin->sin_port;
451 	in_pcbstate(inp, INP_CONNECTED);
452 #if defined(IPSEC) || defined(FAST_IPSEC)
453 	if (inp->inp_socket->so_type == SOCK_STREAM)
454 		ipsec_pcbconn(inp->inp_sp);
455 #endif
456 	return (0);
457 }
458 
459 void
460 in_pcbdisconnect(void *v)
461 {
462 	struct inpcb *inp = v;
463 
464 	if (inp->inp_af != AF_INET)
465 		return;
466 
467 	inp->inp_faddr = zeroin_addr;
468 	inp->inp_fport = 0;
469 	in_pcbstate(inp, INP_BOUND);
470 #if defined(IPSEC) || defined(FAST_IPSEC)
471 	ipsec_pcbdisconn(inp->inp_sp);
472 #endif
473 	if (inp->inp_socket->so_state & SS_NOFDREF)
474 		in_pcbdetach(inp);
475 }
476 
477 void
478 in_pcbdetach(void *v)
479 {
480 	struct inpcb *inp = v;
481 	struct socket *so = inp->inp_socket;
482 	int s;
483 
484 	if (inp->inp_af != AF_INET)
485 		return;
486 
487 #if defined(IPSEC) || defined(FAST_IPSEC)
488 	ipsec4_delete_pcbpolicy(inp);
489 #endif /*IPSEC*/
490 	so->so_pcb = 0;
491 	sofree(so);
492 	if (inp->inp_options)
493 		(void)m_free(inp->inp_options);
494 	if (inp->inp_route.ro_rt)
495 		rtfree(inp->inp_route.ro_rt);
496 	ip_freemoptions(inp->inp_moptions);
497 	s = splnet();
498 	in_pcbstate(inp, INP_ATTACHED);
499 	LIST_REMOVE(&inp->inp_head, inph_lhash);
500 	CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head,
501 	    inph_queue);
502 	splx(s);
503 	pool_put(&inpcb_pool, inp);
504 }
505 
506 void
507 in_setsockaddr(struct inpcb *inp, struct mbuf *nam)
508 {
509 	struct sockaddr_in *sin;
510 
511 	if (inp->inp_af != AF_INET)
512 		return;
513 
514 	nam->m_len = sizeof (*sin);
515 	sin = mtod(nam, struct sockaddr_in *);
516 	bzero((caddr_t)sin, sizeof (*sin));
517 	sin->sin_family = AF_INET;
518 	sin->sin_len = sizeof(*sin);
519 	sin->sin_port = inp->inp_lport;
520 	sin->sin_addr = inp->inp_laddr;
521 }
522 
523 void
524 in_setpeeraddr(struct inpcb *inp, struct mbuf *nam)
525 {
526 	struct sockaddr_in *sin;
527 
528 	if (inp->inp_af != AF_INET)
529 		return;
530 
531 	nam->m_len = sizeof (*sin);
532 	sin = mtod(nam, struct sockaddr_in *);
533 	bzero((caddr_t)sin, sizeof (*sin));
534 	sin->sin_family = AF_INET;
535 	sin->sin_len = sizeof(*sin);
536 	sin->sin_port = inp->inp_fport;
537 	sin->sin_addr = inp->inp_faddr;
538 }
539 
540 /*
541  * Pass some notification to all connections of a protocol
542  * associated with address dst.  The local address and/or port numbers
543  * may be specified to limit the search.  The "usual action" will be
544  * taken, depending on the ctlinput cmd.  The caller must filter any
545  * cmds that are uninteresting (e.g., no error in the map).
546  * Call the protocol specific routine (if any) to report
547  * any errors for each matching socket.
548  *
549  * Must be called at splsoftnet.
550  */
551 int
552 in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg,
553     struct in_addr laddr, u_int lport_arg, int errno,
554     void (*notify)(struct inpcb *, int))
555 {
556 	struct inpcbhead *head;
557 	struct inpcb *inp, *ninp;
558 	u_int16_t fport = fport_arg, lport = lport_arg;
559 	int nmatch;
560 
561 	if (in_nullhost(faddr) || notify == 0)
562 		return (0);
563 
564 	nmatch = 0;
565 	head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
566 	for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) {
567 		ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash);
568 		if (inp->inp_af != AF_INET)
569 			continue;
570 		if (in_hosteq(inp->inp_faddr, faddr) &&
571 		    inp->inp_fport == fport &&
572 		    inp->inp_lport == lport &&
573 		    in_hosteq(inp->inp_laddr, laddr)) {
574 			(*notify)(inp, errno);
575 			nmatch++;
576 		}
577 	}
578 	return (nmatch);
579 }
580 
581 void
582 in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno,
583     void (*notify)(struct inpcb *, int))
584 {
585 	struct inpcb *inp, *ninp;
586 
587 	if (in_nullhost(faddr) || notify == 0)
588 		return;
589 
590 	for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
591 	    inp != (void *)&table->inpt_queue;
592 	    inp = ninp) {
593 		ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
594 		if (inp->inp_af != AF_INET)
595 			continue;
596 		if (in_hosteq(inp->inp_faddr, faddr))
597 			(*notify)(inp, errno);
598 	}
599 }
600 
601 void
602 in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp)
603 {
604 	struct inpcb *inp, *ninp;
605 	struct ip_moptions *imo;
606 	int i, gap;
607 
608 	for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
609 	    inp != (void *)&table->inpt_queue;
610 	    inp = ninp) {
611 		ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
612 		if (inp->inp_af != AF_INET)
613 			continue;
614 		imo = inp->inp_moptions;
615 		if (imo != NULL) {
616 			/*
617 			 * Unselect the outgoing interface if it is being
618 			 * detached.
619 			 */
620 			if (imo->imo_multicast_ifp == ifp)
621 				imo->imo_multicast_ifp = NULL;
622 
623 			/*
624 			 * Drop multicast group membership if we joined
625 			 * through the interface being detached.
626 			 */
627 			for (i = 0, gap = 0; i < imo->imo_num_memberships;
628 			    i++) {
629 				if (imo->imo_membership[i]->inm_ifp == ifp) {
630 					in_delmulti(imo->imo_membership[i]);
631 					gap++;
632 				} else if (gap != 0)
633 					imo->imo_membership[i - gap] =
634 					    imo->imo_membership[i];
635 			}
636 			imo->imo_num_memberships -= gap;
637 		}
638 	}
639 }
640 
641 void
642 in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp)
643 {
644 	struct inpcb *inp, *ninp;
645 
646 	for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
647 	    inp != (void *)&table->inpt_queue;
648 	    inp = ninp) {
649 		ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
650 		if (inp->inp_af != AF_INET)
651 			continue;
652 		if (inp->inp_route.ro_rt != NULL &&
653 		    inp->inp_route.ro_rt->rt_ifp == ifp)
654 			in_rtchange(inp, 0);
655 	}
656 }
657 
658 /*
659  * Check for alternatives when higher level complains
660  * about service problems.  For now, invalidate cached
661  * routing information.  If the route was created dynamically
662  * (by a redirect), time to try a default gateway again.
663  */
664 void
665 in_losing(struct inpcb *inp)
666 {
667 	struct rtentry *rt;
668 	struct rt_addrinfo info;
669 
670 	if (inp->inp_af != AF_INET)
671 		return;
672 
673 	if ((rt = inp->inp_route.ro_rt)) {
674 		inp->inp_route.ro_rt = 0;
675 		bzero((caddr_t)&info, sizeof(info));
676 		info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst;
677 		info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
678 		info.rti_info[RTAX_NETMASK] = rt_mask(rt);
679 		rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
680 		if (rt->rt_flags & RTF_DYNAMIC)
681 			(void) rtrequest(RTM_DELETE, rt_key(rt),
682 				rt->rt_gateway, rt_mask(rt), rt->rt_flags,
683 				(struct rtentry **)0);
684 		else
685 		/*
686 		 * A new route can be allocated
687 		 * the next time output is attempted.
688 		 */
689 			rtfree(rt);
690 	}
691 }
692 
693 /*
694  * After a routing change, flush old routing
695  * and allocate a (hopefully) better one.
696  */
697 void
698 in_rtchange(struct inpcb *inp, int errno)
699 {
700 
701 	if (inp->inp_af != AF_INET)
702 		return;
703 
704 	if (inp->inp_route.ro_rt) {
705 		rtfree(inp->inp_route.ro_rt);
706 		inp->inp_route.ro_rt = 0;
707 		/*
708 		 * A new route can be allocated the next time
709 		 * output is attempted.
710 		 */
711 	}
712 	/* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */
713 }
714 
715 struct inpcb *
716 in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr,
717     u_int lport_arg, int lookup_wildcard)
718 {
719 	struct inpcbhead *head;
720 	struct inpcb_hdr *inph;
721 	struct inpcb *inp, *match = 0;
722 	int matchwild = 3, wildcard;
723 	u_int16_t lport = lport_arg;
724 
725 	head = INPCBHASH_PORT(table, lport);
726 	LIST_FOREACH(inph, head, inph_lhash) {
727 		inp = (struct inpcb *)inph;
728 		if (inp->inp_af != AF_INET)
729 			continue;
730 
731 		if (inp->inp_lport != lport)
732 			continue;
733 		wildcard = 0;
734 		if (!in_nullhost(inp->inp_faddr))
735 			wildcard++;
736 		if (in_nullhost(inp->inp_laddr)) {
737 			if (!in_nullhost(laddr))
738 				wildcard++;
739 		} else {
740 			if (in_nullhost(laddr))
741 				wildcard++;
742 			else {
743 				if (!in_hosteq(inp->inp_laddr, laddr))
744 					continue;
745 			}
746 		}
747 		if (wildcard && !lookup_wildcard)
748 			continue;
749 		if (wildcard < matchwild) {
750 			match = inp;
751 			matchwild = wildcard;
752 			if (matchwild == 0)
753 				break;
754 		}
755 	}
756 	return (match);
757 }
758 
759 #ifdef DIAGNOSTIC
760 int	in_pcbnotifymiss = 0;
761 #endif
762 
763 struct inpcb *
764 in_pcblookup_connect(struct inpcbtable *table,
765     struct in_addr faddr, u_int fport_arg,
766     struct in_addr laddr, u_int lport_arg)
767 {
768 	struct inpcbhead *head;
769 	struct inpcb_hdr *inph;
770 	struct inpcb *inp;
771 	u_int16_t fport = fport_arg, lport = lport_arg;
772 
773 	head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
774 	LIST_FOREACH(inph, head, inph_hash) {
775 		inp = (struct inpcb *)inph;
776 		if (inp->inp_af != AF_INET)
777 			continue;
778 
779 		if (in_hosteq(inp->inp_faddr, faddr) &&
780 		    inp->inp_fport == fport &&
781 		    inp->inp_lport == lport &&
782 		    in_hosteq(inp->inp_laddr, laddr))
783 			goto out;
784 	}
785 #ifdef DIAGNOSTIC
786 	if (in_pcbnotifymiss) {
787 		printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n",
788 		    ntohl(faddr.s_addr), ntohs(fport),
789 		    ntohl(laddr.s_addr), ntohs(lport));
790 	}
791 #endif
792 	return (0);
793 
794 out:
795 	/* Move this PCB to the head of hash chain. */
796 	inph = &inp->inp_head;
797 	if (inph != LIST_FIRST(head)) {
798 		LIST_REMOVE(inph, inph_hash);
799 		LIST_INSERT_HEAD(head, inph, inph_hash);
800 	}
801 	return (inp);
802 }
803 
804 struct inpcb *
805 in_pcblookup_bind(struct inpcbtable *table,
806     struct in_addr laddr, u_int lport_arg)
807 {
808 	struct inpcbhead *head;
809 	struct inpcb_hdr *inph;
810 	struct inpcb *inp;
811 	u_int16_t lport = lport_arg;
812 
813 	head = INPCBHASH_BIND(table, laddr, lport);
814 	LIST_FOREACH(inph, head, inph_hash) {
815 		inp = (struct inpcb *)inph;
816 		if (inp->inp_af != AF_INET)
817 			continue;
818 
819 		if (inp->inp_lport == lport &&
820 		    in_hosteq(inp->inp_laddr, laddr))
821 			goto out;
822 	}
823 	head = INPCBHASH_BIND(table, zeroin_addr, lport);
824 	LIST_FOREACH(inph, head, inph_hash) {
825 		inp = (struct inpcb *)inph;
826 		if (inp->inp_af != AF_INET)
827 			continue;
828 
829 		if (inp->inp_lport == lport &&
830 		    in_hosteq(inp->inp_laddr, zeroin_addr))
831 			goto out;
832 	}
833 #ifdef DIAGNOSTIC
834 	if (in_pcbnotifymiss) {
835 		printf("in_pcblookup_bind: laddr=%08x lport=%d\n",
836 		    ntohl(laddr.s_addr), ntohs(lport));
837 	}
838 #endif
839 	return (0);
840 
841 out:
842 	/* Move this PCB to the head of hash chain. */
843 	inph = &inp->inp_head;
844 	if (inph != LIST_FIRST(head)) {
845 		LIST_REMOVE(inph, inph_hash);
846 		LIST_INSERT_HEAD(head, inph, inph_hash);
847 	}
848 	return (inp);
849 }
850 
851 void
852 in_pcbstate(struct inpcb *inp, int state)
853 {
854 
855 	if (inp->inp_af != AF_INET)
856 		return;
857 
858 	if (inp->inp_state > INP_ATTACHED)
859 		LIST_REMOVE(&inp->inp_head, inph_hash);
860 
861 	switch (state) {
862 	case INP_BOUND:
863 		LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table,
864 		    inp->inp_laddr, inp->inp_lport), &inp->inp_head,
865 		    inph_hash);
866 		break;
867 	case INP_CONNECTED:
868 		LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table,
869 		    inp->inp_faddr, inp->inp_fport,
870 		    inp->inp_laddr, inp->inp_lport), &inp->inp_head,
871 		    inph_hash);
872 		break;
873 	}
874 
875 	inp->inp_state = state;
876 }
877 
878 struct rtentry *
879 in_pcbrtentry(struct inpcb *inp)
880 {
881 	struct route *ro;
882 
883 	if (inp->inp_af != AF_INET)
884 		return (NULL);
885 
886 	ro = &inp->inp_route;
887 
888 	if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
889 	    !in_hosteq(satosin(&ro->ro_dst)->sin_addr, inp->inp_faddr))) {
890 		RTFREE(ro->ro_rt);
891 		ro->ro_rt = (struct rtentry *)NULL;
892 	}
893 	if (ro->ro_rt == (struct rtentry *)NULL &&
894 	    !in_nullhost(inp->inp_faddr)) {
895 		bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
896 		ro->ro_dst.sa_family = AF_INET;
897 		ro->ro_dst.sa_len = sizeof(ro->ro_dst);
898 		satosin(&ro->ro_dst)->sin_addr = inp->inp_faddr;
899 		rtalloc(ro);
900 	}
901 	return (ro->ro_rt);
902 }
903 
904 struct sockaddr_in *
905 in_selectsrc(struct sockaddr_in *sin, struct route *ro,
906     int soopts, struct ip_moptions *mopts, int *errorp)
907 {
908 	struct in_ifaddr *ia;
909 
910 	ia = (struct in_ifaddr *)0;
911 	/*
912 	 * If route is known or can be allocated now,
913 	 * our src addr is taken from the i/f, else punt.
914 	 * Note that we should check the address family of the cached
915 	 * destination, in case of sharing the cache with IPv6.
916 	 */
917 	if (ro->ro_rt &&
918 	    (ro->ro_dst.sa_family != AF_INET ||
919 	    !in_hosteq(satosin(&ro->ro_dst)->sin_addr, sin->sin_addr) ||
920 	    soopts & SO_DONTROUTE)) {
921 		RTFREE(ro->ro_rt);
922 		ro->ro_rt = (struct rtentry *)0;
923 	}
924 	if ((soopts & SO_DONTROUTE) == 0 && /*XXX*/
925 	    (ro->ro_rt == (struct rtentry *)0 ||
926 	     ro->ro_rt->rt_ifp == (struct ifnet *)0)) {
927 		/* No route yet, so try to acquire one */
928 		bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
929 		ro->ro_dst.sa_family = AF_INET;
930 		ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
931 		satosin(&ro->ro_dst)->sin_addr = sin->sin_addr;
932 		rtalloc(ro);
933 	}
934 	/*
935 	 * If we found a route, use the address
936 	 * corresponding to the outgoing interface
937 	 * unless it is the loopback (in case a route
938 	 * to our address on another net goes to loopback).
939 	 *
940 	 * XXX Is this still true?  Do we care?
941 	 */
942 	if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
943 		ia = ifatoia(ro->ro_rt->rt_ifa);
944 	if (ia == NULL) {
945 		u_int16_t fport = sin->sin_port;
946 
947 		sin->sin_port = 0;
948 		ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
949 		sin->sin_port = fport;
950 		if (ia == 0) {
951 			/* Find 1st non-loopback AF_INET address */
952 			TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
953 				if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK))
954 					break;
955 			}
956 		}
957 		if (ia == NULL) {
958 			*errorp = EADDRNOTAVAIL;
959 			return NULL;
960 		}
961 	}
962 	/*
963 	 * If the destination address is multicast and an outgoing
964 	 * interface has been set as a multicast option, use the
965 	 * address of that interface as our source address.
966 	 */
967 	if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) {
968 		struct ip_moptions *imo;
969 		struct ifnet *ifp;
970 
971 		imo = mopts;
972 		if (imo->imo_multicast_ifp != NULL) {
973 			ifp = imo->imo_multicast_ifp;
974 			IFP_TO_IA(ifp, ia);		/* XXX */
975 			if (ia == 0) {
976 				*errorp = EADDRNOTAVAIL;
977 				return NULL;
978 			}
979 		}
980 	}
981 	return satosin(&ia->ia_addr);
982 }
983