xref: /netbsd-src/sys/netinet/in_pcb.c (revision ce2c90c7c172d95d2402a5b3d96d8f8e6d138a21)
1 /*	$NetBSD: in_pcb.c,v 1.107 2006/10/12 01:32:37 christos 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 1.107 2006/10/12 01:32:37 christos 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 	s = splnet();
189 	inp = pool_get(&inpcb_pool, PR_NOWAIT);
190 	splx(s);
191 	if (inp == NULL)
192 		return (ENOBUFS);
193 	bzero((caddr_t)inp, sizeof(*inp));
194 	inp->inp_af = AF_INET;
195 	inp->inp_table = table;
196 	inp->inp_socket = so;
197 	inp->inp_errormtu = -1;
198 #if defined(IPSEC) || defined(FAST_IPSEC)
199 	error = ipsec_init_pcbpolicy(so, &inp->inp_sp);
200 	if (error != 0) {
201 		s = splnet();
202 		pool_put(&inpcb_pool, inp);
203 		splx(s);
204 		return error;
205 	}
206 #endif
207 	so->so_pcb = inp;
208 	s = splnet();
209 	CIRCLEQ_INSERT_HEAD(&table->inpt_queue, &inp->inp_head,
210 	    inph_queue);
211 	LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head,
212 	    inph_lhash);
213 	in_pcbstate(inp, INP_ATTACHED);
214 	splx(s);
215 	return (0);
216 }
217 
218 int
219 in_pcbbind(void *v, struct mbuf *nam, struct lwp *l)
220 {
221 	struct in_ifaddr *ia = NULL;
222 	struct inpcb *inp = v;
223 	struct socket *so = inp->inp_socket;
224 	struct inpcbtable *table = inp->inp_table;
225 	struct sockaddr_in *sin = NULL; /* XXXGCC */
226 	u_int16_t lport = 0;
227 	int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
228 
229 	if (inp->inp_af != AF_INET)
230 		return (EINVAL);
231 
232 	if (TAILQ_FIRST(&in_ifaddrhead) == 0)
233 		return (EADDRNOTAVAIL);
234 	if (inp->inp_lport || !in_nullhost(inp->inp_laddr))
235 		return (EINVAL);
236 	if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
237 		wild = 1;
238 	if (nam == 0)
239 		goto noname;
240 	sin = mtod(nam, struct sockaddr_in *);
241 	if (nam->m_len != sizeof (*sin))
242 		return (EINVAL);
243 	if (sin->sin_family != AF_INET)
244 		return (EAFNOSUPPORT);
245 	lport = sin->sin_port;
246 	if (IN_MULTICAST(sin->sin_addr.s_addr)) {
247 		/*
248 		 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
249 		 * allow complete duplication of binding if
250 		 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
251 		 * and a multicast address is bound on both
252 		 * new and duplicated sockets.
253 		 */
254 		if (so->so_options & SO_REUSEADDR)
255 			reuseport = SO_REUSEADDR|SO_REUSEPORT;
256 	} else if (!in_nullhost(sin->sin_addr)) {
257 		sin->sin_port = 0;		/* yech... */
258 		INADDR_TO_IA(sin->sin_addr, ia);
259 		/* check for broadcast addresses */
260 		if (ia == NULL)
261 			ia = ifatoia(ifa_ifwithaddr(sintosa(sin)));
262 		if (ia == NULL)
263 			return (EADDRNOTAVAIL);
264 	}
265 	if (lport) {
266 		struct inpcb *t;
267 #ifdef INET6
268 		struct in6pcb *t6;
269 		struct in6_addr mapped;
270 #endif
271 #ifndef IPNOPRIVPORTS
272 		/* GROSS */
273 		if (ntohs(lport) < IPPORT_RESERVED &&
274 		    (l == 0 || kauth_authorize_network(l->l_cred,
275 		    KAUTH_NETWORK_BIND,
276 		    KAUTH_REQ_NETWORK_BIND_PRIVPORT, so, sin,
277 		    NULL)))
278 			return (EACCES);
279 #endif
280 #ifdef INET6
281 		memset(&mapped, 0, sizeof(mapped));
282 		mapped.s6_addr16[5] = 0xffff;
283 		memcpy(&mapped.s6_addr32[3], &sin->sin_addr,
284 		    sizeof(mapped.s6_addr32[3]));
285 		t6 = in6_pcblookup_port(table, &mapped, lport, wild);
286 		if (t6 && (reuseport & t6->in6p_socket->so_options) == 0)
287 			return (EADDRINUSE);
288 #endif
289 		if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) {
290 			t = in_pcblookup_port(table, sin->sin_addr, lport, 1);
291 		/*
292 		 * XXX:	investigate ramifications of loosening this
293 		 *	restriction so that as long as both ports have
294 		 *	SO_REUSEPORT allow the bind
295 		 */
296 			if (t &&
297 			    (!in_nullhost(sin->sin_addr) ||
298 			     !in_nullhost(t->inp_laddr) ||
299 			     (t->inp_socket->so_options & SO_REUSEPORT) == 0)
300 			    && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) {
301 				return (EADDRINUSE);
302 			}
303 		}
304 		t = in_pcblookup_port(table, sin->sin_addr, lport, wild);
305 		if (t && (reuseport & t->inp_socket->so_options) == 0)
306 			return (EADDRINUSE);
307 	}
308 	inp->inp_laddr = sin->sin_addr;
309 
310 noname:
311 	if (lport == 0) {
312 		int	   cnt;
313 		u_int16_t  mymin, mymax;
314 		u_int16_t *lastport;
315 
316 		if (inp->inp_flags & INP_LOWPORT) {
317 #ifndef IPNOPRIVPORTS
318 			if (l == 0 || kauth_authorize_network(l->l_cred,
319 			    KAUTH_NETWORK_BIND,
320 			    KAUTH_REQ_NETWORK_BIND_PRIVPORT, so,
321 			    sin, NULL))
322 				return (EACCES);
323 #endif
324 			mymin = lowportmin;
325 			mymax = lowportmax;
326 			lastport = &table->inpt_lastlow;
327 		} else {
328 			mymin = anonportmin;
329 			mymax = anonportmax;
330 			lastport = &table->inpt_lastport;
331 		}
332 		if (mymin > mymax) {	/* sanity check */
333 			u_int16_t swp;
334 
335 			swp = mymin;
336 			mymin = mymax;
337 			mymax = swp;
338 		}
339 
340 		lport = *lastport - 1;
341 		for (cnt = mymax - mymin + 1; cnt; cnt--, lport--) {
342 			if (lport < mymin || lport > mymax)
343 				lport = mymax;
344 			if (!in_pcblookup_port(table, inp->inp_laddr,
345 			    htons(lport), 1))
346 				goto found;
347 		}
348 		if (!in_nullhost(inp->inp_laddr))
349 			inp->inp_laddr.s_addr = INADDR_ANY;
350 		return (EAGAIN);
351 	found:
352 		inp->inp_flags |= INP_ANONPORT;
353 		*lastport = lport;
354 		lport = htons(lport);
355 	}
356 	inp->inp_lport = lport;
357 	LIST_REMOVE(&inp->inp_head, inph_lhash);
358 	LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head,
359 	    inph_lhash);
360 	in_pcbstate(inp, INP_BOUND);
361 	return (0);
362 }
363 
364 /*
365  * Connect from a socket to a specified address.
366  * Both address and port must be specified in argument sin.
367  * If don't have a local address for this socket yet,
368  * then pick one.
369  */
370 int
371 in_pcbconnect(void *v, struct mbuf *nam, struct lwp *l)
372 {
373 	struct inpcb *inp = v;
374 	struct in_ifaddr *ia = NULL;
375 	struct sockaddr_in *ifaddr = NULL;
376 	struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *);
377 	int error;
378 
379 	if (inp->inp_af != AF_INET)
380 		return (EINVAL);
381 
382 	if (nam->m_len != sizeof (*sin))
383 		return (EINVAL);
384 	if (sin->sin_family != AF_INET)
385 		return (EAFNOSUPPORT);
386 	if (sin->sin_port == 0)
387 		return (EADDRNOTAVAIL);
388 	if (TAILQ_FIRST(&in_ifaddrhead) != 0) {
389 		/*
390 		 * If the destination address is INADDR_ANY,
391 		 * use any local address (likely loopback).
392 		 * If the supplied address is INADDR_BROADCAST,
393 		 * use the broadcast address of an interface
394 		 * which supports broadcast. (loopback does not)
395 		 */
396 
397 		if (in_nullhost(sin->sin_addr)) {
398 			sin->sin_addr =
399 			    TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr;
400 		} else if (sin->sin_addr.s_addr == INADDR_BROADCAST) {
401 			TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
402 				if (ia->ia_ifp->if_flags & IFF_BROADCAST) {
403 					sin->sin_addr =
404 					    ia->ia_broadaddr.sin_addr;
405 					break;
406 				}
407 			}
408 		}
409 	}
410 	/*
411 	 * If we haven't bound which network number to use as ours,
412 	 * we will use the number of the outgoing interface.
413 	 * This depends on having done a routing lookup, which
414 	 * we will probably have to do anyway, so we might
415 	 * as well do it now.  On the other hand if we are
416 	 * sending to multiple destinations we may have already
417 	 * done the lookup, so see if we can use the route
418 	 * from before.  In any case, we only
419 	 * chose a port number once, even if sending to multiple
420 	 * destinations.
421 	 */
422 	if (in_nullhost(inp->inp_laddr)) {
423 		int xerror;
424 		ifaddr = in_selectsrc(sin, &inp->inp_route,
425 		    inp->inp_socket->so_options, inp->inp_moptions, &xerror);
426 		if (ifaddr == NULL) {
427 			if (xerror == 0)
428 				xerror = EADDRNOTAVAIL;
429 			return xerror;
430 		}
431 		INADDR_TO_IA(ifaddr->sin_addr, ia);
432 		if (ia == NULL)
433 			return (EADDRNOTAVAIL);
434 	}
435 	if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port,
436 	    !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr,
437 	    inp->inp_lport) != 0)
438 		return (EADDRINUSE);
439 	if (in_nullhost(inp->inp_laddr)) {
440 		if (inp->inp_lport == 0) {
441 			error = in_pcbbind(inp, NULL, l);
442 			/*
443 			 * This used to ignore the return value
444 			 * completely, but we need to check for
445 			 * ephemeral port shortage.
446 			 * And attempts to request low ports if not root.
447 			 */
448 			if (error != 0)
449 				return (error);
450 		}
451 		inp->inp_laddr = ifaddr->sin_addr;
452 	}
453 	inp->inp_faddr = sin->sin_addr;
454 	inp->inp_fport = sin->sin_port;
455 	in_pcbstate(inp, INP_CONNECTED);
456 #if defined(IPSEC) || defined(FAST_IPSEC)
457 	if (inp->inp_socket->so_type == SOCK_STREAM)
458 		ipsec_pcbconn(inp->inp_sp);
459 #endif
460 	return (0);
461 }
462 
463 void
464 in_pcbdisconnect(void *v)
465 {
466 	struct inpcb *inp = v;
467 
468 	if (inp->inp_af != AF_INET)
469 		return;
470 
471 	inp->inp_faddr = zeroin_addr;
472 	inp->inp_fport = 0;
473 	in_pcbstate(inp, INP_BOUND);
474 #if defined(IPSEC) || defined(FAST_IPSEC)
475 	ipsec_pcbdisconn(inp->inp_sp);
476 #endif
477 	if (inp->inp_socket->so_state & SS_NOFDREF)
478 		in_pcbdetach(inp);
479 }
480 
481 void
482 in_pcbdetach(void *v)
483 {
484 	struct inpcb *inp = v;
485 	struct socket *so = inp->inp_socket;
486 	int s;
487 
488 	if (inp->inp_af != AF_INET)
489 		return;
490 
491 #if defined(IPSEC) || defined(FAST_IPSEC)
492 	ipsec4_delete_pcbpolicy(inp);
493 #endif /*IPSEC*/
494 	so->so_pcb = 0;
495 	sofree(so);
496 	if (inp->inp_options)
497 		(void)m_free(inp->inp_options);
498 	if (inp->inp_route.ro_rt)
499 		rtfree(inp->inp_route.ro_rt);
500 	ip_freemoptions(inp->inp_moptions);
501 	s = splnet();
502 	in_pcbstate(inp, INP_ATTACHED);
503 	LIST_REMOVE(&inp->inp_head, inph_lhash);
504 	CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head,
505 	    inph_queue);
506 	pool_put(&inpcb_pool, inp);
507 	splx(s);
508 }
509 
510 void
511 in_setsockaddr(struct inpcb *inp, struct mbuf *nam)
512 {
513 	struct sockaddr_in *sin;
514 
515 	if (inp->inp_af != AF_INET)
516 		return;
517 
518 	nam->m_len = sizeof (*sin);
519 	sin = mtod(nam, struct sockaddr_in *);
520 	bzero((caddr_t)sin, sizeof (*sin));
521 	sin->sin_family = AF_INET;
522 	sin->sin_len = sizeof(*sin);
523 	sin->sin_port = inp->inp_lport;
524 	sin->sin_addr = inp->inp_laddr;
525 }
526 
527 void
528 in_setpeeraddr(struct inpcb *inp, struct mbuf *nam)
529 {
530 	struct sockaddr_in *sin;
531 
532 	if (inp->inp_af != AF_INET)
533 		return;
534 
535 	nam->m_len = sizeof (*sin);
536 	sin = mtod(nam, struct sockaddr_in *);
537 	bzero((caddr_t)sin, sizeof (*sin));
538 	sin->sin_family = AF_INET;
539 	sin->sin_len = sizeof(*sin);
540 	sin->sin_port = inp->inp_fport;
541 	sin->sin_addr = inp->inp_faddr;
542 }
543 
544 /*
545  * Pass some notification to all connections of a protocol
546  * associated with address dst.  The local address and/or port numbers
547  * may be specified to limit the search.  The "usual action" will be
548  * taken, depending on the ctlinput cmd.  The caller must filter any
549  * cmds that are uninteresting (e.g., no error in the map).
550  * Call the protocol specific routine (if any) to report
551  * any errors for each matching socket.
552  *
553  * Must be called at splsoftnet.
554  */
555 int
556 in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg,
557     struct in_addr laddr, u_int lport_arg, int errno,
558     void (*notify)(struct inpcb *, int))
559 {
560 	struct inpcbhead *head;
561 	struct inpcb *inp, *ninp;
562 	u_int16_t fport = fport_arg, lport = lport_arg;
563 	int nmatch;
564 
565 	if (in_nullhost(faddr) || notify == 0)
566 		return (0);
567 
568 	nmatch = 0;
569 	head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
570 	for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) {
571 		ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash);
572 		if (inp->inp_af != AF_INET)
573 			continue;
574 		if (in_hosteq(inp->inp_faddr, faddr) &&
575 		    inp->inp_fport == fport &&
576 		    inp->inp_lport == lport &&
577 		    in_hosteq(inp->inp_laddr, laddr)) {
578 			(*notify)(inp, errno);
579 			nmatch++;
580 		}
581 	}
582 	return (nmatch);
583 }
584 
585 void
586 in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno,
587     void (*notify)(struct inpcb *, int))
588 {
589 	struct inpcb *inp, *ninp;
590 
591 	if (in_nullhost(faddr) || notify == 0)
592 		return;
593 
594 	for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
595 	    inp != (void *)&table->inpt_queue;
596 	    inp = ninp) {
597 		ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
598 		if (inp->inp_af != AF_INET)
599 			continue;
600 		if (in_hosteq(inp->inp_faddr, faddr))
601 			(*notify)(inp, errno);
602 	}
603 }
604 
605 void
606 in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp)
607 {
608 	struct inpcb *inp, *ninp;
609 	struct ip_moptions *imo;
610 	int i, gap;
611 
612 	for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
613 	    inp != (void *)&table->inpt_queue;
614 	    inp = ninp) {
615 		ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
616 		if (inp->inp_af != AF_INET)
617 			continue;
618 		imo = inp->inp_moptions;
619 		if (imo != NULL) {
620 			/*
621 			 * Unselect the outgoing interface if it is being
622 			 * detached.
623 			 */
624 			if (imo->imo_multicast_ifp == ifp)
625 				imo->imo_multicast_ifp = NULL;
626 
627 			/*
628 			 * Drop multicast group membership if we joined
629 			 * through the interface being detached.
630 			 */
631 			for (i = 0, gap = 0; i < imo->imo_num_memberships;
632 			    i++) {
633 				if (imo->imo_membership[i]->inm_ifp == ifp) {
634 					in_delmulti(imo->imo_membership[i]);
635 					gap++;
636 				} else if (gap != 0)
637 					imo->imo_membership[i - gap] =
638 					    imo->imo_membership[i];
639 			}
640 			imo->imo_num_memberships -= gap;
641 		}
642 	}
643 }
644 
645 void
646 in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp)
647 {
648 	struct inpcb *inp, *ninp;
649 
650 	for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
651 	    inp != (void *)&table->inpt_queue;
652 	    inp = ninp) {
653 		ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
654 		if (inp->inp_af != AF_INET)
655 			continue;
656 		if (inp->inp_route.ro_rt != NULL &&
657 		    inp->inp_route.ro_rt->rt_ifp == ifp)
658 			in_rtchange(inp, 0);
659 	}
660 }
661 
662 /*
663  * Check for alternatives when higher level complains
664  * about service problems.  For now, invalidate cached
665  * routing information.  If the route was created dynamically
666  * (by a redirect), time to try a default gateway again.
667  */
668 void
669 in_losing(struct inpcb *inp)
670 {
671 	struct rtentry *rt;
672 	struct rt_addrinfo info;
673 
674 	if (inp->inp_af != AF_INET)
675 		return;
676 
677 	if ((rt = inp->inp_route.ro_rt)) {
678 		inp->inp_route.ro_rt = 0;
679 		bzero((caddr_t)&info, sizeof(info));
680 		info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst;
681 		info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
682 		info.rti_info[RTAX_NETMASK] = rt_mask(rt);
683 		rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
684 		if (rt->rt_flags & RTF_DYNAMIC)
685 			(void) rtrequest(RTM_DELETE, rt_key(rt),
686 				rt->rt_gateway, rt_mask(rt), rt->rt_flags,
687 				(struct rtentry **)0);
688 		else
689 		/*
690 		 * A new route can be allocated
691 		 * the next time output is attempted.
692 		 */
693 			rtfree(rt);
694 	}
695 }
696 
697 /*
698  * After a routing change, flush old routing
699  * and allocate a (hopefully) better one.
700  */
701 void
702 in_rtchange(struct inpcb *inp, int errno __unused)
703 {
704 
705 	if (inp->inp_af != AF_INET)
706 		return;
707 
708 	if (inp->inp_route.ro_rt) {
709 		rtfree(inp->inp_route.ro_rt);
710 		inp->inp_route.ro_rt = 0;
711 		/*
712 		 * A new route can be allocated the next time
713 		 * output is attempted.
714 		 */
715 	}
716 	/* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */
717 }
718 
719 struct inpcb *
720 in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr,
721     u_int lport_arg, int lookup_wildcard)
722 {
723 	struct inpcbhead *head;
724 	struct inpcb_hdr *inph;
725 	struct inpcb *inp, *match = 0;
726 	int matchwild = 3, wildcard;
727 	u_int16_t lport = lport_arg;
728 
729 	head = INPCBHASH_PORT(table, lport);
730 	LIST_FOREACH(inph, head, inph_lhash) {
731 		inp = (struct inpcb *)inph;
732 		if (inp->inp_af != AF_INET)
733 			continue;
734 
735 		if (inp->inp_lport != lport)
736 			continue;
737 		wildcard = 0;
738 		if (!in_nullhost(inp->inp_faddr))
739 			wildcard++;
740 		if (in_nullhost(inp->inp_laddr)) {
741 			if (!in_nullhost(laddr))
742 				wildcard++;
743 		} else {
744 			if (in_nullhost(laddr))
745 				wildcard++;
746 			else {
747 				if (!in_hosteq(inp->inp_laddr, laddr))
748 					continue;
749 			}
750 		}
751 		if (wildcard && !lookup_wildcard)
752 			continue;
753 		if (wildcard < matchwild) {
754 			match = inp;
755 			matchwild = wildcard;
756 			if (matchwild == 0)
757 				break;
758 		}
759 	}
760 	return (match);
761 }
762 
763 #ifdef DIAGNOSTIC
764 int	in_pcbnotifymiss = 0;
765 #endif
766 
767 struct inpcb *
768 in_pcblookup_connect(struct inpcbtable *table,
769     struct in_addr faddr, u_int fport_arg,
770     struct in_addr laddr, u_int lport_arg)
771 {
772 	struct inpcbhead *head;
773 	struct inpcb_hdr *inph;
774 	struct inpcb *inp;
775 	u_int16_t fport = fport_arg, lport = lport_arg;
776 
777 	head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
778 	LIST_FOREACH(inph, head, inph_hash) {
779 		inp = (struct inpcb *)inph;
780 		if (inp->inp_af != AF_INET)
781 			continue;
782 
783 		if (in_hosteq(inp->inp_faddr, faddr) &&
784 		    inp->inp_fport == fport &&
785 		    inp->inp_lport == lport &&
786 		    in_hosteq(inp->inp_laddr, laddr))
787 			goto out;
788 	}
789 #ifdef DIAGNOSTIC
790 	if (in_pcbnotifymiss) {
791 		printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n",
792 		    ntohl(faddr.s_addr), ntohs(fport),
793 		    ntohl(laddr.s_addr), ntohs(lport));
794 	}
795 #endif
796 	return (0);
797 
798 out:
799 	/* Move this PCB to the head of hash chain. */
800 	inph = &inp->inp_head;
801 	if (inph != LIST_FIRST(head)) {
802 		LIST_REMOVE(inph, inph_hash);
803 		LIST_INSERT_HEAD(head, inph, inph_hash);
804 	}
805 	return (inp);
806 }
807 
808 struct inpcb *
809 in_pcblookup_bind(struct inpcbtable *table,
810     struct in_addr laddr, u_int lport_arg)
811 {
812 	struct inpcbhead *head;
813 	struct inpcb_hdr *inph;
814 	struct inpcb *inp;
815 	u_int16_t lport = lport_arg;
816 
817 	head = INPCBHASH_BIND(table, laddr, lport);
818 	LIST_FOREACH(inph, head, inph_hash) {
819 		inp = (struct inpcb *)inph;
820 		if (inp->inp_af != AF_INET)
821 			continue;
822 
823 		if (inp->inp_lport == lport &&
824 		    in_hosteq(inp->inp_laddr, laddr))
825 			goto out;
826 	}
827 	head = INPCBHASH_BIND(table, zeroin_addr, lport);
828 	LIST_FOREACH(inph, head, inph_hash) {
829 		inp = (struct inpcb *)inph;
830 		if (inp->inp_af != AF_INET)
831 			continue;
832 
833 		if (inp->inp_lport == lport &&
834 		    in_hosteq(inp->inp_laddr, zeroin_addr))
835 			goto out;
836 	}
837 #ifdef DIAGNOSTIC
838 	if (in_pcbnotifymiss) {
839 		printf("in_pcblookup_bind: laddr=%08x lport=%d\n",
840 		    ntohl(laddr.s_addr), ntohs(lport));
841 	}
842 #endif
843 	return (0);
844 
845 out:
846 	/* Move this PCB to the head of hash chain. */
847 	inph = &inp->inp_head;
848 	if (inph != LIST_FIRST(head)) {
849 		LIST_REMOVE(inph, inph_hash);
850 		LIST_INSERT_HEAD(head, inph, inph_hash);
851 	}
852 	return (inp);
853 }
854 
855 void
856 in_pcbstate(struct inpcb *inp, int state)
857 {
858 
859 	if (inp->inp_af != AF_INET)
860 		return;
861 
862 	if (inp->inp_state > INP_ATTACHED)
863 		LIST_REMOVE(&inp->inp_head, inph_hash);
864 
865 	switch (state) {
866 	case INP_BOUND:
867 		LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table,
868 		    inp->inp_laddr, inp->inp_lport), &inp->inp_head,
869 		    inph_hash);
870 		break;
871 	case INP_CONNECTED:
872 		LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table,
873 		    inp->inp_faddr, inp->inp_fport,
874 		    inp->inp_laddr, inp->inp_lport), &inp->inp_head,
875 		    inph_hash);
876 		break;
877 	}
878 
879 	inp->inp_state = state;
880 }
881 
882 struct rtentry *
883 in_pcbrtentry(struct inpcb *inp)
884 {
885 	struct route *ro;
886 
887 	if (inp->inp_af != AF_INET)
888 		return (NULL);
889 
890 	ro = &inp->inp_route;
891 
892 	if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
893 	    !in_hosteq(satosin(&ro->ro_dst)->sin_addr, inp->inp_faddr))) {
894 		RTFREE(ro->ro_rt);
895 		ro->ro_rt = (struct rtentry *)NULL;
896 	}
897 	if (ro->ro_rt == (struct rtentry *)NULL &&
898 	    !in_nullhost(inp->inp_faddr)) {
899 		bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
900 		ro->ro_dst.sa_family = AF_INET;
901 		ro->ro_dst.sa_len = sizeof(ro->ro_dst);
902 		satosin(&ro->ro_dst)->sin_addr = inp->inp_faddr;
903 		rtalloc(ro);
904 	}
905 	return (ro->ro_rt);
906 }
907 
908 struct sockaddr_in *
909 in_selectsrc(struct sockaddr_in *sin, struct route *ro,
910     int soopts, struct ip_moptions *mopts, int *errorp)
911 {
912 	struct in_ifaddr *ia;
913 
914 	ia = (struct in_ifaddr *)0;
915 	/*
916 	 * If route is known or can be allocated now,
917 	 * our src addr is taken from the i/f, else punt.
918 	 * Note that we should check the address family of the cached
919 	 * destination, in case of sharing the cache with IPv6.
920 	 */
921 	if (ro->ro_rt &&
922 	    (ro->ro_dst.sa_family != AF_INET ||
923 	    !in_hosteq(satosin(&ro->ro_dst)->sin_addr, sin->sin_addr) ||
924 	    soopts & SO_DONTROUTE)) {
925 		RTFREE(ro->ro_rt);
926 		ro->ro_rt = (struct rtentry *)0;
927 	}
928 	if ((soopts & SO_DONTROUTE) == 0 && /*XXX*/
929 	    (ro->ro_rt == (struct rtentry *)0 ||
930 	     ro->ro_rt->rt_ifp == (struct ifnet *)0)) {
931 		/* No route yet, so try to acquire one */
932 		bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
933 		ro->ro_dst.sa_family = AF_INET;
934 		ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
935 		satosin(&ro->ro_dst)->sin_addr = sin->sin_addr;
936 		rtalloc(ro);
937 	}
938 	/*
939 	 * If we found a route, use the address
940 	 * corresponding to the outgoing interface
941 	 * unless it is the loopback (in case a route
942 	 * to our address on another net goes to loopback).
943 	 *
944 	 * XXX Is this still true?  Do we care?
945 	 */
946 	if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
947 		ia = ifatoia(ro->ro_rt->rt_ifa);
948 	if (ia == NULL) {
949 		u_int16_t fport = sin->sin_port;
950 
951 		sin->sin_port = 0;
952 		ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
953 		sin->sin_port = fport;
954 		if (ia == 0) {
955 			/* Find 1st non-loopback AF_INET address */
956 			TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
957 				if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK))
958 					break;
959 			}
960 		}
961 		if (ia == NULL) {
962 			*errorp = EADDRNOTAVAIL;
963 			return NULL;
964 		}
965 	}
966 	/*
967 	 * If the destination address is multicast and an outgoing
968 	 * interface has been set as a multicast option, use the
969 	 * address of that interface as our source address.
970 	 */
971 	if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) {
972 		struct ip_moptions *imo;
973 		struct ifnet *ifp;
974 
975 		imo = mopts;
976 		if (imo->imo_multicast_ifp != NULL) {
977 			ifp = imo->imo_multicast_ifp;
978 			IFP_TO_IA(ifp, ia);		/* XXX */
979 			if (ia == 0) {
980 				*errorp = EADDRNOTAVAIL;
981 				return NULL;
982 			}
983 		}
984 	}
985 	return satosin(&ia->ia_addr);
986 }
987