xref: /netbsd-src/sys/netinet/in_pcb.c (revision 1ca5c1b28139779176bd5c13ad7c5f25c0bcd5f8)
1 /*	$NetBSD: in_pcb.c,v 1.73 2001/11/13 00:32:36 lukem 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. All advertising materials mentioning features or use of this software
82  *    must display the following acknowledgement:
83  *	This product includes software developed by the University of
84  *	California, Berkeley and its contributors.
85  * 4. Neither the name of the University nor the names of its contributors
86  *    may be used to endorse or promote products derived from this software
87  *    without specific prior written permission.
88  *
89  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
90  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
91  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
92  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
93  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
94  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
95  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
96  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
97  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
98  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
99  * SUCH DAMAGE.
100  *
101  *	@(#)in_pcb.c	8.4 (Berkeley) 5/24/95
102  */
103 
104 #include <sys/cdefs.h>
105 __KERNEL_RCSID(0, "$NetBSD: in_pcb.c,v 1.73 2001/11/13 00:32:36 lukem Exp $");
106 
107 #include "opt_ipsec.h"
108 
109 #include <sys/param.h>
110 #include <sys/systm.h>
111 #include <sys/malloc.h>
112 #include <sys/mbuf.h>
113 #include <sys/protosw.h>
114 #include <sys/socket.h>
115 #include <sys/socketvar.h>
116 #include <sys/ioctl.h>
117 #include <sys/errno.h>
118 #include <sys/time.h>
119 #include <sys/pool.h>
120 #include <sys/proc.h>
121 
122 #include <net/if.h>
123 #include <net/route.h>
124 
125 #include <netinet/in.h>
126 #include <netinet/in_systm.h>
127 #include <netinet/ip.h>
128 #include <netinet/in_pcb.h>
129 #include <netinet/in_var.h>
130 #include <netinet/ip_var.h>
131 
132 #ifdef IPSEC
133 #include <netinet6/ipsec.h>
134 #include <netkey/key.h>
135 #endif /* IPSEC */
136 
137 struct	in_addr zeroin_addr;
138 
139 #define	INPCBHASH_BIND(table, laddr, lport) \
140 	&(table)->inpt_bindhashtbl[ \
141 	    ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash]
142 #define	INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \
143 	&(table)->inpt_connecthashtbl[ \
144 	    ((ntohl((faddr).s_addr) + ntohs(fport)) + \
145 	     (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash]
146 
147 struct inpcb *
148 	in_pcblookup_port __P((struct inpcbtable *,
149 	    struct in_addr, u_int, int));
150 
151 int	anonportmin = IPPORT_ANONMIN;
152 int	anonportmax = IPPORT_ANONMAX;
153 int	lowportmin  = IPPORT_RESERVEDMIN;
154 int	lowportmax  = IPPORT_RESERVEDMAX;
155 
156 struct pool inpcb_pool;
157 
158 void
159 in_pcbinit(table, bindhashsize, connecthashsize)
160 	struct inpcbtable *table;
161 	int bindhashsize, connecthashsize;
162 {
163 	static int inpcb_pool_initialized;
164 
165 	if (inpcb_pool_initialized == 0) {
166 		pool_init(&inpcb_pool, sizeof(struct inpcb), 0, 0, 0,
167 		    "inpcbpl", 0, NULL, NULL, M_PCB);
168 		inpcb_pool_initialized = 1;
169 	}
170 
171 	CIRCLEQ_INIT(&table->inpt_queue);
172 	table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, M_PCB,
173 	    M_WAITOK, &table->inpt_bindhash);
174 	table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST,
175 	    M_PCB, M_WAITOK, &table->inpt_connecthash);
176 	table->inpt_lastlow = IPPORT_RESERVEDMAX;
177 	table->inpt_lastport = (u_int16_t)anonportmax;
178 }
179 
180 int
181 in_pcballoc(so, v)
182 	struct socket *so;
183 	void *v;
184 {
185 	struct inpcbtable *table = v;
186 	struct inpcb *inp;
187 	int s;
188 #ifdef IPSEC
189 	int error;
190 #endif
191 
192 	inp = pool_get(&inpcb_pool, PR_NOWAIT);
193 	if (inp == NULL)
194 		return (ENOBUFS);
195 	bzero((caddr_t)inp, sizeof(*inp));
196 	inp->inp_table = table;
197 	inp->inp_socket = so;
198 	inp->inp_errormtu = -1;
199 #ifdef IPSEC
200 	error = ipsec_init_policy(so, &inp->inp_sp);
201 	if (error != 0) {
202 		pool_put(&inpcb_pool, inp);
203 		return error;
204 	}
205 #endif
206 	so->so_pcb = inp;
207 	s = splnet();
208 	CIRCLEQ_INSERT_HEAD(&table->inpt_queue, inp, inp_queue);
209 	in_pcbstate(inp, INP_ATTACHED);
210 	splx(s);
211 	return (0);
212 }
213 
214 int
215 in_pcbbind(v, nam, p)
216 	void *v;
217 	struct mbuf *nam;
218 	struct proc *p;
219 {
220 	struct inpcb *inp = v;
221 	struct socket *so = inp->inp_socket;
222 	struct inpcbtable *table = inp->inp_table;
223 	struct sockaddr_in *sin;
224 	u_int16_t lport = 0;
225 	int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
226 #ifndef IPNOPRIVPORTS
227 	int error;
228 #endif
229 
230 	if (TAILQ_FIRST(&in_ifaddr) == 0)
231 		return (EADDRNOTAVAIL);
232 	if (inp->inp_lport || !in_nullhost(inp->inp_laddr))
233 		return (EINVAL);
234 	if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
235 		wild = 1;
236 	if (nam == 0)
237 		goto noname;
238 	sin = mtod(nam, struct sockaddr_in *);
239 	if (nam->m_len != sizeof (*sin))
240 		return (EINVAL);
241 #ifdef notdef
242 	/*
243 	 * We should check the family, but old programs
244 	 * incorrectly fail to initialize it.
245 	 */
246 	if (sin->sin_family != AF_INET)
247 		return (EAFNOSUPPORT);
248 #endif
249 	lport = sin->sin_port;
250 	if (IN_MULTICAST(sin->sin_addr.s_addr)) {
251 		/*
252 		 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
253 		 * allow complete duplication of binding if
254 		 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
255 		 * and a multicast address is bound on both
256 		 * new and duplicated sockets.
257 		 */
258 		if (so->so_options & SO_REUSEADDR)
259 			reuseport = SO_REUSEADDR|SO_REUSEPORT;
260 	} else if (!in_nullhost(sin->sin_addr)) {
261 		sin->sin_port = 0;		/* yech... */
262 		if (ifa_ifwithaddr(sintosa(sin)) == 0)
263 			return (EADDRNOTAVAIL);
264 	}
265 	if (lport) {
266 		struct inpcb *t;
267 #ifndef IPNOPRIVPORTS
268 		/* GROSS */
269 		if (ntohs(lport) < IPPORT_RESERVED &&
270 		    (p == 0 || (error = suser(p->p_ucred, &p->p_acflag))))
271 			return (EACCES);
272 #endif
273 		if (so->so_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) {
274 			t = in_pcblookup_port(table, sin->sin_addr, lport, 1);
275 		/*
276 		 * XXX:	investigate ramifications of loosening this
277 		 *	restriction so that as long as both ports have
278 		 *	SO_REUSEPORT allow the bind
279 		 */
280 			if (t &&
281 			    (!in_nullhost(sin->sin_addr) ||
282 			     !in_nullhost(t->inp_laddr) ||
283 			     (t->inp_socket->so_options & SO_REUSEPORT) == 0)
284 			    && (so->so_uid != t->inp_socket->so_uid)) {
285 				return (EADDRINUSE);
286 			}
287 		}
288 		t = in_pcblookup_port(table, sin->sin_addr, lport, wild);
289 		if (t && (reuseport & t->inp_socket->so_options) == 0)
290 			return (EADDRINUSE);
291 	}
292 	inp->inp_laddr = sin->sin_addr;
293 
294 noname:
295 	if (lport == 0) {
296 		int	   cnt;
297 		u_int16_t  min, max;
298 		u_int16_t *lastport;
299 
300 		if (inp->inp_flags & INP_LOWPORT) {
301 #ifndef IPNOPRIVPORTS
302 			if (p == 0 || (error = suser(p->p_ucred, &p->p_acflag)))
303 				return (EACCES);
304 #endif
305 			min = lowportmin;
306 			max = lowportmax;
307 			lastport = &table->inpt_lastlow;
308 		} else {
309 			min = anonportmin;
310 			max = anonportmax;
311 			lastport = &table->inpt_lastport;
312 		}
313 		if (min > max) {	/* sanity check */
314 			u_int16_t swp;
315 
316 			swp = min;
317 			min = max;
318 			max = swp;
319 		}
320 
321 		lport = *lastport - 1;
322 		for (cnt = max - min + 1; cnt; cnt--, lport--) {
323 			if (lport < min || lport > max)
324 				lport = max;
325 			if (!in_pcblookup_port(table, inp->inp_laddr,
326 			    htons(lport), 1))
327 				goto found;
328 		}
329 		if (!in_nullhost(inp->inp_laddr))
330 			inp->inp_laddr.s_addr = INADDR_ANY;
331 		return (EAGAIN);
332 	found:
333 		inp->inp_flags |= INP_ANONPORT;
334 		*lastport = lport;
335 		lport = htons(lport);
336 	}
337 	inp->inp_lport = lport;
338 	in_pcbstate(inp, INP_BOUND);
339 	return (0);
340 }
341 
342 /*
343  * Connect from a socket to a specified address.
344  * Both address and port must be specified in argument sin.
345  * If don't have a local address for this socket yet,
346  * then pick one.
347  */
348 int
349 in_pcbconnect(v, nam)
350 	void *v;
351 	struct mbuf *nam;
352 {
353 	struct inpcb *inp = v;
354 	struct in_ifaddr *ia;
355 	struct sockaddr_in *ifaddr = NULL;
356 	struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *);
357 	int error;
358 
359 	if (nam->m_len != sizeof (*sin))
360 		return (EINVAL);
361 	if (sin->sin_family != AF_INET)
362 		return (EAFNOSUPPORT);
363 	if (sin->sin_port == 0)
364 		return (EADDRNOTAVAIL);
365 	if (TAILQ_FIRST(&in_ifaddr) != 0) {
366 		/*
367 		 * If the destination address is INADDR_ANY,
368 		 * use any local address (likely loopback).
369 		 * If the supplied address is INADDR_BROADCAST,
370 		 * use the broadcast address of an interface
371 		 * which supports broadcast. (loopback does not)
372 		 */
373 
374 		if (in_nullhost(sin->sin_addr)) {
375 			sin->sin_addr =
376 			    TAILQ_FIRST(&in_ifaddr)->ia_addr.sin_addr;
377 		} else if (sin->sin_addr.s_addr == INADDR_BROADCAST) {
378 			TAILQ_FOREACH(ia, &in_ifaddr, ia_list) {
379 				if (ia->ia_ifp->if_flags & IFF_BROADCAST) {
380 					sin->sin_addr =
381 					    ia->ia_broadaddr.sin_addr;
382 					break;
383 				}
384 			}
385 		}
386 	}
387 	/*
388 	 * If we haven't bound which network number to use as ours,
389 	 * we will use the number of the outgoing interface.
390 	 * This depends on having done a routing lookup, which
391 	 * we will probably have to do anyway, so we might
392 	 * as well do it now.  On the other hand if we are
393 	 * sending to multiple destinations we may have already
394 	 * done the lookup, so see if we can use the route
395 	 * from before.  In any case, we only
396 	 * chose a port number once, even if sending to multiple
397 	 * destinations.
398 	 */
399 	if (in_nullhost(inp->inp_laddr)) {
400 #if 0
401 		struct route *ro;
402 
403 		ia = (struct in_ifaddr *)0;
404 		/*
405 		 * If route is known or can be allocated now,
406 		 * our src addr is taken from the i/f, else punt.
407 		 */
408 		ro = &inp->inp_route;
409 		if (ro->ro_rt &&
410 		    (!in_hosteq(satosin(&ro->ro_dst)->sin_addr,
411 			sin->sin_addr) ||
412 		    inp->inp_socket->so_options & SO_DONTROUTE)) {
413 			RTFREE(ro->ro_rt);
414 			ro->ro_rt = (struct rtentry *)0;
415 		}
416 		if ((inp->inp_socket->so_options & SO_DONTROUTE) == 0 && /*XXX*/
417 		    (ro->ro_rt == (struct rtentry *)0 ||
418 		    ro->ro_rt->rt_ifp == (struct ifnet *)0)) {
419 			/* No route yet, so try to acquire one */
420 			ro->ro_dst.sa_family = AF_INET;
421 			ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
422 			satosin(&ro->ro_dst)->sin_addr = sin->sin_addr;
423 			rtalloc(ro);
424 		}
425 		/*
426 		 * If we found a route, use the address
427 		 * corresponding to the outgoing interface
428 		 * unless it is the loopback (in case a route
429 		 * to our address on another net goes to loopback).
430 		 *
431 		 * XXX Is this still true?  Do we care?
432 		 */
433 		if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
434 			ia = ifatoia(ro->ro_rt->rt_ifa);
435 		if (ia == NULL) {
436 			u_int16_t fport = sin->sin_port;
437 
438 			sin->sin_port = 0;
439 			ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
440 			sin->sin_port = fport;
441 			if (ia == 0) {
442 				/* Find 1st non-loopback AF_INET address */
443 				TAILQ_FOREACH(ia, &in_ifaddr, ia_list) {
444 					if ((ia->ia_ifp->if_flags &
445 					     IFF_LOOPBACK) == 0)
446 						break;
447 				}
448 			}
449 			if (ia == NULL)
450 				return (EADDRNOTAVAIL);
451 		}
452 		/*
453 		 * If the destination address is multicast and an outgoing
454 		 * interface has been set as a multicast option, use the
455 		 * address of that interface as our source address.
456 		 */
457 		if (IN_MULTICAST(sin->sin_addr.s_addr) &&
458 		    inp->inp_moptions != NULL) {
459 			struct ip_moptions *imo;
460 			struct ifnet *ifp;
461 
462 			imo = inp->inp_moptions;
463 			if (imo->imo_multicast_ifp != NULL) {
464 				ifp = imo->imo_multicast_ifp;
465 				IFP_TO_IA(ifp, ia);		/* XXX */
466 				if (ia == 0)
467 					return (EADDRNOTAVAIL);
468 			}
469 		}
470 		ifaddr = satosin(&ia->ia_addr);
471 #else
472 		int error;
473 		ifaddr = in_selectsrc(sin, &inp->inp_route,
474 			inp->inp_socket->so_options, inp->inp_moptions, &error);
475 		if (ifaddr == NULL) {
476 			if (error == 0)
477 				error = EADDRNOTAVAIL;
478 			return error;
479 		}
480 #endif
481 	}
482 	if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port,
483 	    !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr,
484 	    inp->inp_lport) != 0)
485 		return (EADDRINUSE);
486 	if (in_nullhost(inp->inp_laddr)) {
487 		if (inp->inp_lport == 0) {
488 			error = in_pcbbind(inp, (struct mbuf *)0,
489 			    (struct proc *)0);
490 			/*
491 			 * This used to ignore the return value
492 			 * completely, but we need to check for
493 			 * ephemeral port shortage.
494 			 * XXX Should we check for other errors, too?
495 			 */
496 			if (error == EAGAIN)
497 				return (error);
498 		}
499 		inp->inp_laddr = ifaddr->sin_addr;
500 	}
501 	inp->inp_faddr = sin->sin_addr;
502 	inp->inp_fport = sin->sin_port;
503 	in_pcbstate(inp, INP_CONNECTED);
504 #ifdef IPSEC
505 	if (inp->inp_socket->so_type == SOCK_STREAM)
506 		ipsec_pcbconn(inp->inp_sp);
507 #endif
508 	return (0);
509 }
510 
511 void
512 in_pcbdisconnect(v)
513 	void *v;
514 {
515 	struct inpcb *inp = v;
516 
517 	inp->inp_faddr = zeroin_addr;
518 	inp->inp_fport = 0;
519 	in_pcbstate(inp, INP_BOUND);
520 	if (inp->inp_socket->so_state & SS_NOFDREF)
521 		in_pcbdetach(inp);
522 #ifdef IPSEC
523 	ipsec_pcbdisconn(inp->inp_sp);
524 #endif
525 }
526 
527 void
528 in_pcbdetach(v)
529 	void *v;
530 {
531 	struct inpcb *inp = v;
532 	struct socket *so = inp->inp_socket;
533 	int s;
534 
535 #ifdef IPSEC
536 	ipsec4_delete_pcbpolicy(inp);
537 #endif /*IPSEC*/
538 	so->so_pcb = 0;
539 	sofree(so);
540 	if (inp->inp_options)
541 		(void)m_free(inp->inp_options);
542 	if (inp->inp_route.ro_rt)
543 		rtfree(inp->inp_route.ro_rt);
544 	ip_freemoptions(inp->inp_moptions);
545 	s = splnet();
546 	in_pcbstate(inp, INP_ATTACHED);
547 	CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, inp, inp_queue);
548 	splx(s);
549 	pool_put(&inpcb_pool, inp);
550 }
551 
552 void
553 in_setsockaddr(inp, nam)
554 	struct inpcb *inp;
555 	struct mbuf *nam;
556 {
557 	struct sockaddr_in *sin;
558 
559 	nam->m_len = sizeof (*sin);
560 	sin = mtod(nam, struct sockaddr_in *);
561 	bzero((caddr_t)sin, sizeof (*sin));
562 	sin->sin_family = AF_INET;
563 	sin->sin_len = sizeof(*sin);
564 	sin->sin_port = inp->inp_lport;
565 	sin->sin_addr = inp->inp_laddr;
566 }
567 
568 void
569 in_setpeeraddr(inp, nam)
570 	struct inpcb *inp;
571 	struct mbuf *nam;
572 {
573 	struct sockaddr_in *sin;
574 
575 	nam->m_len = sizeof (*sin);
576 	sin = mtod(nam, struct sockaddr_in *);
577 	bzero((caddr_t)sin, sizeof (*sin));
578 	sin->sin_family = AF_INET;
579 	sin->sin_len = sizeof(*sin);
580 	sin->sin_port = inp->inp_fport;
581 	sin->sin_addr = inp->inp_faddr;
582 }
583 
584 /*
585  * Pass some notification to all connections of a protocol
586  * associated with address dst.  The local address and/or port numbers
587  * may be specified to limit the search.  The "usual action" will be
588  * taken, depending on the ctlinput cmd.  The caller must filter any
589  * cmds that are uninteresting (e.g., no error in the map).
590  * Call the protocol specific routine (if any) to report
591  * any errors for each matching socket.
592  *
593  * Must be called at splsoftnet.
594  */
595 int
596 in_pcbnotify(table, faddr, fport_arg, laddr, lport_arg, errno, notify)
597 	struct inpcbtable *table;
598 	struct in_addr faddr, laddr;
599 	u_int fport_arg, lport_arg;
600 	int errno;
601 	void (*notify) __P((struct inpcb *, int));
602 {
603 	struct inpcbhead *head;
604 	struct inpcb *inp, *ninp;
605 	u_int16_t fport = fport_arg, lport = lport_arg;
606 	int nmatch;
607 
608 	if (in_nullhost(faddr) || notify == 0)
609 		return (0);
610 
611 	nmatch = 0;
612 	head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
613 	for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) {
614 		ninp = LIST_NEXT(inp, inp_hash);
615 		if (in_hosteq(inp->inp_faddr, faddr) &&
616 		    inp->inp_fport == fport &&
617 		    inp->inp_lport == lport &&
618 		    in_hosteq(inp->inp_laddr, laddr)) {
619 			(*notify)(inp, errno);
620 			nmatch++;
621 		}
622 	}
623 	return (nmatch);
624 }
625 
626 void
627 in_pcbnotifyall(table, faddr, errno, notify)
628 	struct inpcbtable *table;
629 	struct in_addr faddr;
630 	int errno;
631 	void (*notify) __P((struct inpcb *, int));
632 {
633 	struct inpcb *inp, *ninp;
634 
635 	if (in_nullhost(faddr) || notify == 0)
636 		return;
637 
638 	for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
639 	    inp != (void *)&table->inpt_queue;
640 	    inp = ninp) {
641 		ninp = CIRCLEQ_NEXT(inp, inp_queue);
642 		if (in_hosteq(inp->inp_faddr, faddr))
643 			(*notify)(inp, errno);
644 	}
645 }
646 
647 void
648 in_pcbpurgeif0(table, ifp)
649 	struct inpcbtable *table;
650 	struct ifnet *ifp;
651 {
652 	struct inpcb *inp, *ninp;
653 	struct ip_moptions *imo;
654 	int i, gap;
655 
656 	for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
657 	    inp != (void *)&table->inpt_queue;
658 	    inp = ninp) {
659 		ninp = CIRCLEQ_NEXT(inp, inp_queue);
660 		imo = inp->inp_moptions;
661 		if (imo != NULL) {
662 			/*
663 			 * Unselect the outgoing interface if it is being
664 			 * detached.
665 			 */
666 			if (imo->imo_multicast_ifp == ifp)
667 				imo->imo_multicast_ifp = NULL;
668 
669 			/*
670 			 * Drop multicast group membership if we joined
671 			 * through the interface being detached.
672 			 */
673 			for (i = 0, gap = 0; i < imo->imo_num_memberships;
674 			    i++) {
675 				if (imo->imo_membership[i]->inm_ifp == ifp) {
676 					in_delmulti(imo->imo_membership[i]);
677 					gap++;
678 				} else if (gap != 0)
679 					imo->imo_membership[i - gap] =
680 					    imo->imo_membership[i];
681 			}
682 			imo->imo_num_memberships -= gap;
683 		}
684 	}
685 }
686 
687 void
688 in_pcbpurgeif(table, ifp)
689 	struct inpcbtable *table;
690 	struct ifnet *ifp;
691 {
692 	struct inpcb *inp, *ninp;
693 
694 	for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
695 	    inp != (void *)&table->inpt_queue;
696 	    inp = ninp) {
697 		ninp = CIRCLEQ_NEXT(inp, inp_queue);
698 		if (inp->inp_route.ro_rt != NULL &&
699 		    inp->inp_route.ro_rt->rt_ifp == ifp)
700 			in_rtchange(inp, 0);
701 	}
702 }
703 
704 /*
705  * Check for alternatives when higher level complains
706  * about service problems.  For now, invalidate cached
707  * routing information.  If the route was created dynamically
708  * (by a redirect), time to try a default gateway again.
709  */
710 void
711 in_losing(inp)
712 	struct inpcb *inp;
713 {
714 	struct rtentry *rt;
715 	struct rt_addrinfo info;
716 
717 	if ((rt = inp->inp_route.ro_rt)) {
718 		inp->inp_route.ro_rt = 0;
719 		bzero((caddr_t)&info, sizeof(info));
720 		info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst;
721 		info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
722 		info.rti_info[RTAX_NETMASK] = rt_mask(rt);
723 		rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
724 		if (rt->rt_flags & RTF_DYNAMIC)
725 			(void) rtrequest(RTM_DELETE, rt_key(rt),
726 				rt->rt_gateway, rt_mask(rt), rt->rt_flags,
727 				(struct rtentry **)0);
728 		else
729 		/*
730 		 * A new route can be allocated
731 		 * the next time output is attempted.
732 		 */
733 			rtfree(rt);
734 	}
735 }
736 
737 /*
738  * After a routing change, flush old routing
739  * and allocate a (hopefully) better one.
740  */
741 void
742 in_rtchange(inp, errno)
743 	struct inpcb *inp;
744 	int errno;
745 {
746 
747 	if (inp->inp_route.ro_rt) {
748 		rtfree(inp->inp_route.ro_rt);
749 		inp->inp_route.ro_rt = 0;
750 		/*
751 		 * A new route can be allocated the next time
752 		 * output is attempted.
753 		 */
754 	}
755 	/* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */
756 }
757 
758 struct inpcb *
759 in_pcblookup_port(table, laddr, lport_arg, lookup_wildcard)
760 	struct inpcbtable *table;
761 	struct in_addr laddr;
762 	u_int lport_arg;
763 	int lookup_wildcard;
764 {
765 	struct inpcb *inp, *match = 0;
766 	int matchwild = 3, wildcard;
767 	u_int16_t lport = lport_arg;
768 
769 	CIRCLEQ_FOREACH(inp, &table->inpt_queue, inp_queue) {
770 		if (inp->inp_lport != lport)
771 			continue;
772 		wildcard = 0;
773 		if (!in_nullhost(inp->inp_faddr))
774 			wildcard++;
775 		if (in_nullhost(inp->inp_laddr)) {
776 			if (!in_nullhost(laddr))
777 				wildcard++;
778 		} else {
779 			if (in_nullhost(laddr))
780 				wildcard++;
781 			else {
782 				if (!in_hosteq(inp->inp_laddr, laddr))
783 					continue;
784 			}
785 		}
786 		if (wildcard && !lookup_wildcard)
787 			continue;
788 		if (wildcard < matchwild) {
789 			match = inp;
790 			matchwild = wildcard;
791 			if (matchwild == 0)
792 				break;
793 		}
794 	}
795 	return (match);
796 }
797 
798 #ifdef DIAGNOSTIC
799 int	in_pcbnotifymiss = 0;
800 #endif
801 
802 struct inpcb *
803 in_pcblookup_connect(table, faddr, fport_arg, laddr, lport_arg)
804 	struct inpcbtable *table;
805 	struct in_addr faddr, laddr;
806 	u_int fport_arg, lport_arg;
807 {
808 	struct inpcbhead *head;
809 	struct inpcb *inp;
810 	u_int16_t fport = fport_arg, lport = lport_arg;
811 
812 	head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
813 	LIST_FOREACH(inp, head, inp_hash) {
814 		if (in_hosteq(inp->inp_faddr, faddr) &&
815 		    inp->inp_fport == fport &&
816 		    inp->inp_lport == lport &&
817 		    in_hosteq(inp->inp_laddr, laddr))
818 			goto out;
819 	}
820 #ifdef DIAGNOSTIC
821 	if (in_pcbnotifymiss) {
822 		printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n",
823 		    ntohl(faddr.s_addr), ntohs(fport),
824 		    ntohl(laddr.s_addr), ntohs(lport));
825 	}
826 #endif
827 	return (0);
828 
829 out:
830 	/* Move this PCB to the head of hash chain. */
831 	if (inp != LIST_FIRST(head)) {
832 		LIST_REMOVE(inp, inp_hash);
833 		LIST_INSERT_HEAD(head, inp, inp_hash);
834 	}
835 	return (inp);
836 }
837 
838 struct inpcb *
839 in_pcblookup_bind(table, laddr, lport_arg)
840 	struct inpcbtable *table;
841 	struct in_addr laddr;
842 	u_int lport_arg;
843 {
844 	struct inpcbhead *head;
845 	struct inpcb *inp;
846 	u_int16_t lport = lport_arg;
847 
848 	head = INPCBHASH_BIND(table, laddr, lport);
849 	LIST_FOREACH(inp, head, inp_hash) {
850 		if (inp->inp_lport == lport &&
851 		    in_hosteq(inp->inp_laddr, laddr))
852 			goto out;
853 	}
854 	head = INPCBHASH_BIND(table, zeroin_addr, lport);
855 	LIST_FOREACH(inp, head, inp_hash) {
856 		if (inp->inp_lport == lport &&
857 		    in_hosteq(inp->inp_laddr, zeroin_addr))
858 			goto out;
859 	}
860 #ifdef DIAGNOSTIC
861 	if (in_pcbnotifymiss) {
862 		printf("in_pcblookup_bind: laddr=%08x lport=%d\n",
863 		    ntohl(laddr.s_addr), ntohs(lport));
864 	}
865 #endif
866 	return (0);
867 
868 out:
869 	/* Move this PCB to the head of hash chain. */
870 	if (inp != LIST_FIRST(head)) {
871 		LIST_REMOVE(inp, inp_hash);
872 		LIST_INSERT_HEAD(head, inp, inp_hash);
873 	}
874 	return (inp);
875 }
876 
877 void
878 in_pcbstate(inp, state)
879 	struct inpcb *inp;
880 	int state;
881 {
882 
883 	if (inp->inp_state > INP_ATTACHED)
884 		LIST_REMOVE(inp, inp_hash);
885 
886 	switch (state) {
887 	case INP_BOUND:
888 		LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table,
889 		    inp->inp_laddr, inp->inp_lport), inp, inp_hash);
890 		break;
891 	case INP_CONNECTED:
892 		LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table,
893 		    inp->inp_faddr, inp->inp_fport,
894 		    inp->inp_laddr, inp->inp_lport), inp, inp_hash);
895 		break;
896 	}
897 
898 	inp->inp_state = state;
899 }
900 
901 struct rtentry *
902 in_pcbrtentry(inp)
903 	struct inpcb *inp;
904 {
905 	struct route *ro;
906 
907 	ro = &inp->inp_route;
908 
909 	if (ro->ro_rt == NULL) {
910 		/*
911 		 * No route yet, so try to acquire one.
912 		 */
913 		if (!in_nullhost(inp->inp_faddr)) {
914 			ro->ro_dst.sa_family = AF_INET;
915 			ro->ro_dst.sa_len = sizeof(ro->ro_dst);
916 			satosin(&ro->ro_dst)->sin_addr = inp->inp_faddr;
917 			rtalloc(ro);
918 		}
919 	}
920 	return (ro->ro_rt);
921 }
922 
923 struct sockaddr_in *
924 in_selectsrc(sin, ro, soopts, mopts, errorp)
925 	struct sockaddr_in *sin;
926 	struct route *ro;
927 	int soopts;
928 	struct ip_moptions *mopts;
929 	int *errorp;
930 {
931 	struct in_ifaddr *ia;
932 
933 	ia = (struct in_ifaddr *)0;
934 	/*
935 	 * If route is known or can be allocated now,
936 	 * our src addr is taken from the i/f, else punt.
937 	 */
938 	if (ro->ro_rt &&
939 	    (!in_hosteq(satosin(&ro->ro_dst)->sin_addr, sin->sin_addr) ||
940 	    soopts & SO_DONTROUTE)) {
941 		RTFREE(ro->ro_rt);
942 		ro->ro_rt = (struct rtentry *)0;
943 	}
944 	if ((soopts & SO_DONTROUTE) == 0 && /*XXX*/
945 	    (ro->ro_rt == (struct rtentry *)0 ||
946 	    ro->ro_rt->rt_ifp == (struct ifnet *)0)) {
947 		/* No route yet, so try to acquire one */
948 		ro->ro_dst.sa_family = AF_INET;
949 		ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
950 		satosin(&ro->ro_dst)->sin_addr = sin->sin_addr;
951 		rtalloc(ro);
952 	}
953 	/*
954 	 * If we found a route, use the address
955 	 * corresponding to the outgoing interface
956 	 * unless it is the loopback (in case a route
957 	 * to our address on another net goes to loopback).
958 	 *
959 	 * XXX Is this still true?  Do we care?
960 	 */
961 	if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
962 		ia = ifatoia(ro->ro_rt->rt_ifa);
963 	if (ia == NULL) {
964 		u_int16_t fport = sin->sin_port;
965 
966 		sin->sin_port = 0;
967 		ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
968 		sin->sin_port = fport;
969 		if (ia == 0) {
970 			/* Find 1st non-loopback AF_INET address */
971 			TAILQ_FOREACH(ia, &in_ifaddr, ia_list) {
972 				if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK))
973 					break;
974 			}
975 		}
976 		if (ia == NULL) {
977 			*errorp = EADDRNOTAVAIL;
978 			return NULL;
979 		}
980 	}
981 	/*
982 	 * If the destination address is multicast and an outgoing
983 	 * interface has been set as a multicast option, use the
984 	 * address of that interface as our source address.
985 	 */
986 	if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) {
987 		struct ip_moptions *imo;
988 		struct ifnet *ifp;
989 
990 		imo = mopts;
991 		if (imo->imo_multicast_ifp != NULL) {
992 			ifp = imo->imo_multicast_ifp;
993 			IFP_TO_IA(ifp, ia);		/* XXX */
994 			if (ia == 0) {
995 				*errorp = EADDRNOTAVAIL;
996 				return NULL;
997 			}
998 		}
999 	}
1000 	return satosin(&ia->ia_addr);
1001 }
1002