in_pcb.c revision 1.101.12.1 1 /* $NetBSD: in_pcb.c,v 1.101.12.1 2006/05/24 15:50:44 tron 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 1.101.12.1 2006/05/24 15:50:44 tron 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 proc *p)
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;
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 (p == 0 || kauth_authorize_generic(p->p_cred, KAUTH_GENERIC_ISSUSER, &p->p_acflag)))
271 return (EACCES);
272 #endif
273 #ifdef INET6
274 memset(&mapped, 0, sizeof(mapped));
275 mapped.s6_addr16[5] = 0xffff;
276 memcpy(&mapped.s6_addr32[3], &sin->sin_addr,
277 sizeof(mapped.s6_addr32[3]));
278 t6 = in6_pcblookup_port(table, &mapped, lport, wild);
279 if (t6 && (reuseport & t6->in6p_socket->so_options) == 0)
280 return (EADDRINUSE);
281 #endif
282 if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) {
283 t = in_pcblookup_port(table, sin->sin_addr, lport, 1);
284 /*
285 * XXX: investigate ramifications of loosening this
286 * restriction so that as long as both ports have
287 * SO_REUSEPORT allow the bind
288 */
289 if (t &&
290 (!in_nullhost(sin->sin_addr) ||
291 !in_nullhost(t->inp_laddr) ||
292 (t->inp_socket->so_options & SO_REUSEPORT) == 0)
293 && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) {
294 return (EADDRINUSE);
295 }
296 }
297 t = in_pcblookup_port(table, sin->sin_addr, lport, wild);
298 if (t && (reuseport & t->inp_socket->so_options) == 0)
299 return (EADDRINUSE);
300 }
301 inp->inp_laddr = sin->sin_addr;
302
303 noname:
304 if (lport == 0) {
305 int cnt;
306 u_int16_t mymin, mymax;
307 u_int16_t *lastport;
308
309 if (inp->inp_flags & INP_LOWPORT) {
310 #ifndef IPNOPRIVPORTS
311 if (p == 0 || kauth_authorize_generic(p->p_cred, KAUTH_GENERIC_ISSUSER,
312 &p->p_acflag))
313 return (EACCES);
314 #endif
315 mymin = lowportmin;
316 mymax = lowportmax;
317 lastport = &table->inpt_lastlow;
318 } else {
319 mymin = anonportmin;
320 mymax = anonportmax;
321 lastport = &table->inpt_lastport;
322 }
323 if (mymin > mymax) { /* sanity check */
324 u_int16_t swp;
325
326 swp = mymin;
327 mymin = mymax;
328 mymax = swp;
329 }
330
331 lport = *lastport - 1;
332 for (cnt = mymax - mymin + 1; cnt; cnt--, lport--) {
333 if (lport < mymin || lport > mymax)
334 lport = mymax;
335 if (!in_pcblookup_port(table, inp->inp_laddr,
336 htons(lport), 1))
337 goto found;
338 }
339 if (!in_nullhost(inp->inp_laddr))
340 inp->inp_laddr.s_addr = INADDR_ANY;
341 return (EAGAIN);
342 found:
343 inp->inp_flags |= INP_ANONPORT;
344 *lastport = lport;
345 lport = htons(lport);
346 }
347 inp->inp_lport = lport;
348 LIST_REMOVE(&inp->inp_head, inph_lhash);
349 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head,
350 inph_lhash);
351 in_pcbstate(inp, INP_BOUND);
352 return (0);
353 }
354
355 /*
356 * Connect from a socket to a specified address.
357 * Both address and port must be specified in argument sin.
358 * If don't have a local address for this socket yet,
359 * then pick one.
360 */
361 int
362 in_pcbconnect(void *v, struct mbuf *nam, struct proc *p)
363 {
364 struct inpcb *inp = v;
365 struct in_ifaddr *ia = NULL;
366 struct sockaddr_in *ifaddr = NULL;
367 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *);
368 int error;
369
370 if (inp->inp_af != AF_INET)
371 return (EINVAL);
372
373 if (nam->m_len != sizeof (*sin))
374 return (EINVAL);
375 if (sin->sin_family != AF_INET)
376 return (EAFNOSUPPORT);
377 if (sin->sin_port == 0)
378 return (EADDRNOTAVAIL);
379 if (TAILQ_FIRST(&in_ifaddrhead) != 0) {
380 /*
381 * If the destination address is INADDR_ANY,
382 * use any local address (likely loopback).
383 * If the supplied address is INADDR_BROADCAST,
384 * use the broadcast address of an interface
385 * which supports broadcast. (loopback does not)
386 */
387
388 if (in_nullhost(sin->sin_addr)) {
389 sin->sin_addr =
390 TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr;
391 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) {
392 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
393 if (ia->ia_ifp->if_flags & IFF_BROADCAST) {
394 sin->sin_addr =
395 ia->ia_broadaddr.sin_addr;
396 break;
397 }
398 }
399 }
400 }
401 /*
402 * If we haven't bound which network number to use as ours,
403 * we will use the number of the outgoing interface.
404 * This depends on having done a routing lookup, which
405 * we will probably have to do anyway, so we might
406 * as well do it now. On the other hand if we are
407 * sending to multiple destinations we may have already
408 * done the lookup, so see if we can use the route
409 * from before. In any case, we only
410 * chose a port number once, even if sending to multiple
411 * destinations.
412 */
413 if (in_nullhost(inp->inp_laddr)) {
414 int xerror;
415 ifaddr = in_selectsrc(sin, &inp->inp_route,
416 inp->inp_socket->so_options, inp->inp_moptions, &xerror);
417 if (ifaddr == NULL) {
418 if (xerror == 0)
419 xerror = EADDRNOTAVAIL;
420 return xerror;
421 }
422 INADDR_TO_IA(ifaddr->sin_addr, ia);
423 if (ia == NULL)
424 return (EADDRNOTAVAIL);
425 }
426 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port,
427 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr,
428 inp->inp_lport) != 0)
429 return (EADDRINUSE);
430 if (in_nullhost(inp->inp_laddr)) {
431 if (inp->inp_lport == 0) {
432 error = in_pcbbind(inp, NULL, p);
433 /*
434 * This used to ignore the return value
435 * completely, but we need to check for
436 * ephemeral port shortage.
437 * And attempts to request low ports if not root.
438 */
439 if (error != 0)
440 return (error);
441 }
442 inp->inp_laddr = ifaddr->sin_addr;
443 }
444 inp->inp_faddr = sin->sin_addr;
445 inp->inp_fport = sin->sin_port;
446 in_pcbstate(inp, INP_CONNECTED);
447 #if defined(IPSEC) || defined(FAST_IPSEC)
448 if (inp->inp_socket->so_type == SOCK_STREAM)
449 ipsec_pcbconn(inp->inp_sp);
450 #endif
451 return (0);
452 }
453
454 void
455 in_pcbdisconnect(void *v)
456 {
457 struct inpcb *inp = v;
458
459 if (inp->inp_af != AF_INET)
460 return;
461
462 inp->inp_faddr = zeroin_addr;
463 inp->inp_fport = 0;
464 in_pcbstate(inp, INP_BOUND);
465 #if defined(IPSEC) || defined(FAST_IPSEC)
466 ipsec_pcbdisconn(inp->inp_sp);
467 #endif
468 if (inp->inp_socket->so_state & SS_NOFDREF)
469 in_pcbdetach(inp);
470 }
471
472 void
473 in_pcbdetach(void *v)
474 {
475 struct inpcb *inp = v;
476 struct socket *so = inp->inp_socket;
477 int s;
478
479 if (inp->inp_af != AF_INET)
480 return;
481
482 #if defined(IPSEC) || defined(FAST_IPSEC)
483 ipsec4_delete_pcbpolicy(inp);
484 #endif /*IPSEC*/
485 so->so_pcb = 0;
486 sofree(so);
487 if (inp->inp_options)
488 (void)m_free(inp->inp_options);
489 if (inp->inp_route.ro_rt)
490 rtfree(inp->inp_route.ro_rt);
491 ip_freemoptions(inp->inp_moptions);
492 s = splnet();
493 in_pcbstate(inp, INP_ATTACHED);
494 LIST_REMOVE(&inp->inp_head, inph_lhash);
495 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head,
496 inph_queue);
497 splx(s);
498 pool_put(&inpcb_pool, inp);
499 }
500
501 void
502 in_setsockaddr(struct inpcb *inp, struct mbuf *nam)
503 {
504 struct sockaddr_in *sin;
505
506 if (inp->inp_af != AF_INET)
507 return;
508
509 nam->m_len = sizeof (*sin);
510 sin = mtod(nam, struct sockaddr_in *);
511 bzero((caddr_t)sin, sizeof (*sin));
512 sin->sin_family = AF_INET;
513 sin->sin_len = sizeof(*sin);
514 sin->sin_port = inp->inp_lport;
515 sin->sin_addr = inp->inp_laddr;
516 }
517
518 void
519 in_setpeeraddr(struct inpcb *inp, struct mbuf *nam)
520 {
521 struct sockaddr_in *sin;
522
523 if (inp->inp_af != AF_INET)
524 return;
525
526 nam->m_len = sizeof (*sin);
527 sin = mtod(nam, struct sockaddr_in *);
528 bzero((caddr_t)sin, sizeof (*sin));
529 sin->sin_family = AF_INET;
530 sin->sin_len = sizeof(*sin);
531 sin->sin_port = inp->inp_fport;
532 sin->sin_addr = inp->inp_faddr;
533 }
534
535 /*
536 * Pass some notification to all connections of a protocol
537 * associated with address dst. The local address and/or port numbers
538 * may be specified to limit the search. The "usual action" will be
539 * taken, depending on the ctlinput cmd. The caller must filter any
540 * cmds that are uninteresting (e.g., no error in the map).
541 * Call the protocol specific routine (if any) to report
542 * any errors for each matching socket.
543 *
544 * Must be called at splsoftnet.
545 */
546 int
547 in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg,
548 struct in_addr laddr, u_int lport_arg, int errno,
549 void (*notify)(struct inpcb *, int))
550 {
551 struct inpcbhead *head;
552 struct inpcb *inp, *ninp;
553 u_int16_t fport = fport_arg, lport = lport_arg;
554 int nmatch;
555
556 if (in_nullhost(faddr) || notify == 0)
557 return (0);
558
559 nmatch = 0;
560 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
561 for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) {
562 ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash);
563 if (inp->inp_af != AF_INET)
564 continue;
565 if (in_hosteq(inp->inp_faddr, faddr) &&
566 inp->inp_fport == fport &&
567 inp->inp_lport == lport &&
568 in_hosteq(inp->inp_laddr, laddr)) {
569 (*notify)(inp, errno);
570 nmatch++;
571 }
572 }
573 return (nmatch);
574 }
575
576 void
577 in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno,
578 void (*notify)(struct inpcb *, int))
579 {
580 struct inpcb *inp, *ninp;
581
582 if (in_nullhost(faddr) || notify == 0)
583 return;
584
585 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
586 inp != (void *)&table->inpt_queue;
587 inp = ninp) {
588 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
589 if (inp->inp_af != AF_INET)
590 continue;
591 if (in_hosteq(inp->inp_faddr, faddr))
592 (*notify)(inp, errno);
593 }
594 }
595
596 void
597 in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp)
598 {
599 struct inpcb *inp, *ninp;
600 struct ip_moptions *imo;
601 int i, gap;
602
603 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
604 inp != (void *)&table->inpt_queue;
605 inp = ninp) {
606 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
607 if (inp->inp_af != AF_INET)
608 continue;
609 imo = inp->inp_moptions;
610 if (imo != NULL) {
611 /*
612 * Unselect the outgoing interface if it is being
613 * detached.
614 */
615 if (imo->imo_multicast_ifp == ifp)
616 imo->imo_multicast_ifp = NULL;
617
618 /*
619 * Drop multicast group membership if we joined
620 * through the interface being detached.
621 */
622 for (i = 0, gap = 0; i < imo->imo_num_memberships;
623 i++) {
624 if (imo->imo_membership[i]->inm_ifp == ifp) {
625 in_delmulti(imo->imo_membership[i]);
626 gap++;
627 } else if (gap != 0)
628 imo->imo_membership[i - gap] =
629 imo->imo_membership[i];
630 }
631 imo->imo_num_memberships -= gap;
632 }
633 }
634 }
635
636 void
637 in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp)
638 {
639 struct inpcb *inp, *ninp;
640
641 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue);
642 inp != (void *)&table->inpt_queue;
643 inp = ninp) {
644 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue);
645 if (inp->inp_af != AF_INET)
646 continue;
647 if (inp->inp_route.ro_rt != NULL &&
648 inp->inp_route.ro_rt->rt_ifp == ifp)
649 in_rtchange(inp, 0);
650 }
651 }
652
653 /*
654 * Check for alternatives when higher level complains
655 * about service problems. For now, invalidate cached
656 * routing information. If the route was created dynamically
657 * (by a redirect), time to try a default gateway again.
658 */
659 void
660 in_losing(struct inpcb *inp)
661 {
662 struct rtentry *rt;
663 struct rt_addrinfo info;
664
665 if (inp->inp_af != AF_INET)
666 return;
667
668 if ((rt = inp->inp_route.ro_rt)) {
669 inp->inp_route.ro_rt = 0;
670 bzero((caddr_t)&info, sizeof(info));
671 info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst;
672 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
673 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
674 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
675 if (rt->rt_flags & RTF_DYNAMIC)
676 (void) rtrequest(RTM_DELETE, rt_key(rt),
677 rt->rt_gateway, rt_mask(rt), rt->rt_flags,
678 (struct rtentry **)0);
679 else
680 /*
681 * A new route can be allocated
682 * the next time output is attempted.
683 */
684 rtfree(rt);
685 }
686 }
687
688 /*
689 * After a routing change, flush old routing
690 * and allocate a (hopefully) better one.
691 */
692 void
693 in_rtchange(struct inpcb *inp, int errno)
694 {
695
696 if (inp->inp_af != AF_INET)
697 return;
698
699 if (inp->inp_route.ro_rt) {
700 rtfree(inp->inp_route.ro_rt);
701 inp->inp_route.ro_rt = 0;
702 /*
703 * A new route can be allocated the next time
704 * output is attempted.
705 */
706 }
707 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */
708 }
709
710 struct inpcb *
711 in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr,
712 u_int lport_arg, int lookup_wildcard)
713 {
714 struct inpcbhead *head;
715 struct inpcb_hdr *inph;
716 struct inpcb *inp, *match = 0;
717 int matchwild = 3, wildcard;
718 u_int16_t lport = lport_arg;
719
720 head = INPCBHASH_PORT(table, lport);
721 LIST_FOREACH(inph, head, inph_lhash) {
722 inp = (struct inpcb *)inph;
723 if (inp->inp_af != AF_INET)
724 continue;
725
726 if (inp->inp_lport != lport)
727 continue;
728 wildcard = 0;
729 if (!in_nullhost(inp->inp_faddr))
730 wildcard++;
731 if (in_nullhost(inp->inp_laddr)) {
732 if (!in_nullhost(laddr))
733 wildcard++;
734 } else {
735 if (in_nullhost(laddr))
736 wildcard++;
737 else {
738 if (!in_hosteq(inp->inp_laddr, laddr))
739 continue;
740 }
741 }
742 if (wildcard && !lookup_wildcard)
743 continue;
744 if (wildcard < matchwild) {
745 match = inp;
746 matchwild = wildcard;
747 if (matchwild == 0)
748 break;
749 }
750 }
751 return (match);
752 }
753
754 #ifdef DIAGNOSTIC
755 int in_pcbnotifymiss = 0;
756 #endif
757
758 struct inpcb *
759 in_pcblookup_connect(struct inpcbtable *table,
760 struct in_addr faddr, u_int fport_arg,
761 struct in_addr laddr, u_int lport_arg)
762 {
763 struct inpcbhead *head;
764 struct inpcb_hdr *inph;
765 struct inpcb *inp;
766 u_int16_t fport = fport_arg, lport = lport_arg;
767
768 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
769 LIST_FOREACH(inph, head, inph_hash) {
770 inp = (struct inpcb *)inph;
771 if (inp->inp_af != AF_INET)
772 continue;
773
774 if (in_hosteq(inp->inp_faddr, faddr) &&
775 inp->inp_fport == fport &&
776 inp->inp_lport == lport &&
777 in_hosteq(inp->inp_laddr, laddr))
778 goto out;
779 }
780 #ifdef DIAGNOSTIC
781 if (in_pcbnotifymiss) {
782 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n",
783 ntohl(faddr.s_addr), ntohs(fport),
784 ntohl(laddr.s_addr), ntohs(lport));
785 }
786 #endif
787 return (0);
788
789 out:
790 /* Move this PCB to the head of hash chain. */
791 inph = &inp->inp_head;
792 if (inph != LIST_FIRST(head)) {
793 LIST_REMOVE(inph, inph_hash);
794 LIST_INSERT_HEAD(head, inph, inph_hash);
795 }
796 return (inp);
797 }
798
799 struct inpcb *
800 in_pcblookup_bind(struct inpcbtable *table,
801 struct in_addr laddr, u_int lport_arg)
802 {
803 struct inpcbhead *head;
804 struct inpcb_hdr *inph;
805 struct inpcb *inp;
806 u_int16_t lport = lport_arg;
807
808 head = INPCBHASH_BIND(table, laddr, lport);
809 LIST_FOREACH(inph, head, inph_hash) {
810 inp = (struct inpcb *)inph;
811 if (inp->inp_af != AF_INET)
812 continue;
813
814 if (inp->inp_lport == lport &&
815 in_hosteq(inp->inp_laddr, laddr))
816 goto out;
817 }
818 head = INPCBHASH_BIND(table, zeroin_addr, lport);
819 LIST_FOREACH(inph, head, inph_hash) {
820 inp = (struct inpcb *)inph;
821 if (inp->inp_af != AF_INET)
822 continue;
823
824 if (inp->inp_lport == lport &&
825 in_hosteq(inp->inp_laddr, zeroin_addr))
826 goto out;
827 }
828 #ifdef DIAGNOSTIC
829 if (in_pcbnotifymiss) {
830 printf("in_pcblookup_bind: laddr=%08x lport=%d\n",
831 ntohl(laddr.s_addr), ntohs(lport));
832 }
833 #endif
834 return (0);
835
836 out:
837 /* Move this PCB to the head of hash chain. */
838 inph = &inp->inp_head;
839 if (inph != LIST_FIRST(head)) {
840 LIST_REMOVE(inph, inph_hash);
841 LIST_INSERT_HEAD(head, inph, inph_hash);
842 }
843 return (inp);
844 }
845
846 void
847 in_pcbstate(struct inpcb *inp, int state)
848 {
849
850 if (inp->inp_af != AF_INET)
851 return;
852
853 if (inp->inp_state > INP_ATTACHED)
854 LIST_REMOVE(&inp->inp_head, inph_hash);
855
856 switch (state) {
857 case INP_BOUND:
858 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table,
859 inp->inp_laddr, inp->inp_lport), &inp->inp_head,
860 inph_hash);
861 break;
862 case INP_CONNECTED:
863 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table,
864 inp->inp_faddr, inp->inp_fport,
865 inp->inp_laddr, inp->inp_lport), &inp->inp_head,
866 inph_hash);
867 break;
868 }
869
870 inp->inp_state = state;
871 }
872
873 struct rtentry *
874 in_pcbrtentry(struct inpcb *inp)
875 {
876 struct route *ro;
877
878 if (inp->inp_af != AF_INET)
879 return (NULL);
880
881 ro = &inp->inp_route;
882
883 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
884 !in_hosteq(satosin(&ro->ro_dst)->sin_addr, inp->inp_faddr))) {
885 RTFREE(ro->ro_rt);
886 ro->ro_rt = (struct rtentry *)NULL;
887 }
888 if (ro->ro_rt == (struct rtentry *)NULL &&
889 !in_nullhost(inp->inp_faddr)) {
890 bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
891 ro->ro_dst.sa_family = AF_INET;
892 ro->ro_dst.sa_len = sizeof(ro->ro_dst);
893 satosin(&ro->ro_dst)->sin_addr = inp->inp_faddr;
894 rtalloc(ro);
895 }
896 return (ro->ro_rt);
897 }
898
899 struct sockaddr_in *
900 in_selectsrc(struct sockaddr_in *sin, struct route *ro,
901 int soopts, struct ip_moptions *mopts, int *errorp)
902 {
903 struct in_ifaddr *ia;
904
905 ia = (struct in_ifaddr *)0;
906 /*
907 * If route is known or can be allocated now,
908 * our src addr is taken from the i/f, else punt.
909 * Note that we should check the address family of the cached
910 * destination, in case of sharing the cache with IPv6.
911 */
912 if (ro->ro_rt &&
913 (ro->ro_dst.sa_family != AF_INET ||
914 !in_hosteq(satosin(&ro->ro_dst)->sin_addr, sin->sin_addr) ||
915 soopts & SO_DONTROUTE)) {
916 RTFREE(ro->ro_rt);
917 ro->ro_rt = (struct rtentry *)0;
918 }
919 if ((soopts & SO_DONTROUTE) == 0 && /*XXX*/
920 (ro->ro_rt == (struct rtentry *)0 ||
921 ro->ro_rt->rt_ifp == (struct ifnet *)0)) {
922 /* No route yet, so try to acquire one */
923 bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
924 ro->ro_dst.sa_family = AF_INET;
925 ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
926 satosin(&ro->ro_dst)->sin_addr = sin->sin_addr;
927 rtalloc(ro);
928 }
929 /*
930 * If we found a route, use the address
931 * corresponding to the outgoing interface
932 * unless it is the loopback (in case a route
933 * to our address on another net goes to loopback).
934 *
935 * XXX Is this still true? Do we care?
936 */
937 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
938 ia = ifatoia(ro->ro_rt->rt_ifa);
939 if (ia == NULL) {
940 u_int16_t fport = sin->sin_port;
941
942 sin->sin_port = 0;
943 ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
944 sin->sin_port = fport;
945 if (ia == 0) {
946 /* Find 1st non-loopback AF_INET address */
947 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
948 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK))
949 break;
950 }
951 }
952 if (ia == NULL) {
953 *errorp = EADDRNOTAVAIL;
954 return NULL;
955 }
956 }
957 /*
958 * If the destination address is multicast and an outgoing
959 * interface has been set as a multicast option, use the
960 * address of that interface as our source address.
961 */
962 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) {
963 struct ip_moptions *imo;
964 struct ifnet *ifp;
965
966 imo = mopts;
967 if (imo->imo_multicast_ifp != NULL) {
968 ifp = imo->imo_multicast_ifp;
969 IFP_TO_IA(ifp, ia); /* XXX */
970 if (ia == 0) {
971 *errorp = EADDRNOTAVAIL;
972 return NULL;
973 }
974 }
975 }
976 return satosin(&ia->ia_addr);
977 }
978