in_pcb.c revision 1.80 1 /* $NetBSD: in_pcb.c,v 1.80 2002/10/22 02:31:16 simonb 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 1.80 2002/10/22 02:31:16 simonb 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", NULL);
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_pcbpolicy(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
227 if (TAILQ_FIRST(&in_ifaddr) == 0)
228 return (EADDRNOTAVAIL);
229 if (inp->inp_lport || !in_nullhost(inp->inp_laddr))
230 return (EINVAL);
231 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
232 wild = 1;
233 if (nam == 0)
234 goto noname;
235 sin = mtod(nam, struct sockaddr_in *);
236 if (nam->m_len != sizeof (*sin))
237 return (EINVAL);
238 #ifdef notdef
239 /*
240 * We should check the family, but old programs
241 * incorrectly fail to initialize it.
242 */
243 if (sin->sin_family != AF_INET)
244 return (EAFNOSUPPORT);
245 #endif
246 lport = sin->sin_port;
247 if (IN_MULTICAST(sin->sin_addr.s_addr)) {
248 /*
249 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
250 * allow complete duplication of binding if
251 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
252 * and a multicast address is bound on both
253 * new and duplicated sockets.
254 */
255 if (so->so_options & SO_REUSEADDR)
256 reuseport = SO_REUSEADDR|SO_REUSEPORT;
257 } else if (!in_nullhost(sin->sin_addr)) {
258 sin->sin_port = 0; /* yech... */
259 if (ifa_ifwithaddr(sintosa(sin)) == 0)
260 return (EADDRNOTAVAIL);
261 }
262 if (lport) {
263 struct inpcb *t;
264 #ifndef IPNOPRIVPORTS
265 /* GROSS */
266 if (ntohs(lport) < IPPORT_RESERVED &&
267 (p == 0 || suser(p->p_ucred, &p->p_acflag)))
268 return (EACCES);
269 #endif
270 if (so->so_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) {
271 t = in_pcblookup_port(table, sin->sin_addr, lport, 1);
272 /*
273 * XXX: investigate ramifications of loosening this
274 * restriction so that as long as both ports have
275 * SO_REUSEPORT allow the bind
276 */
277 if (t &&
278 (!in_nullhost(sin->sin_addr) ||
279 !in_nullhost(t->inp_laddr) ||
280 (t->inp_socket->so_options & SO_REUSEPORT) == 0)
281 && (so->so_uid != t->inp_socket->so_uid)) {
282 return (EADDRINUSE);
283 }
284 }
285 t = in_pcblookup_port(table, sin->sin_addr, lport, wild);
286 if (t && (reuseport & t->inp_socket->so_options) == 0)
287 return (EADDRINUSE);
288 }
289 inp->inp_laddr = sin->sin_addr;
290
291 noname:
292 if (lport == 0) {
293 int cnt;
294 u_int16_t min, max;
295 u_int16_t *lastport;
296
297 if (inp->inp_flags & INP_LOWPORT) {
298 #ifndef IPNOPRIVPORTS
299 if (p == 0 || suser(p->p_ucred, &p->p_acflag))
300 return (EACCES);
301 #endif
302 min = lowportmin;
303 max = lowportmax;
304 lastport = &table->inpt_lastlow;
305 } else {
306 min = anonportmin;
307 max = anonportmax;
308 lastport = &table->inpt_lastport;
309 }
310 if (min > max) { /* sanity check */
311 u_int16_t swp;
312
313 swp = min;
314 min = max;
315 max = swp;
316 }
317
318 lport = *lastport - 1;
319 for (cnt = max - min + 1; cnt; cnt--, lport--) {
320 if (lport < min || lport > max)
321 lport = max;
322 if (!in_pcblookup_port(table, inp->inp_laddr,
323 htons(lport), 1))
324 goto found;
325 }
326 if (!in_nullhost(inp->inp_laddr))
327 inp->inp_laddr.s_addr = INADDR_ANY;
328 return (EAGAIN);
329 found:
330 inp->inp_flags |= INP_ANONPORT;
331 *lastport = lport;
332 lport = htons(lport);
333 }
334 inp->inp_lport = lport;
335 in_pcbstate(inp, INP_BOUND);
336 return (0);
337 }
338
339 /*
340 * Connect from a socket to a specified address.
341 * Both address and port must be specified in argument sin.
342 * If don't have a local address for this socket yet,
343 * then pick one.
344 */
345 int
346 in_pcbconnect(v, nam)
347 void *v;
348 struct mbuf *nam;
349 {
350 struct inpcb *inp = v;
351 struct in_ifaddr *ia;
352 struct sockaddr_in *ifaddr = NULL;
353 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *);
354 int error;
355
356 if (nam->m_len != sizeof (*sin))
357 return (EINVAL);
358 if (sin->sin_family != AF_INET)
359 return (EAFNOSUPPORT);
360 if (sin->sin_port == 0)
361 return (EADDRNOTAVAIL);
362 if (TAILQ_FIRST(&in_ifaddr) != 0) {
363 /*
364 * If the destination address is INADDR_ANY,
365 * use any local address (likely loopback).
366 * If the supplied address is INADDR_BROADCAST,
367 * use the broadcast address of an interface
368 * which supports broadcast. (loopback does not)
369 */
370
371 if (in_nullhost(sin->sin_addr)) {
372 sin->sin_addr =
373 TAILQ_FIRST(&in_ifaddr)->ia_addr.sin_addr;
374 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) {
375 TAILQ_FOREACH(ia, &in_ifaddr, ia_list) {
376 if (ia->ia_ifp->if_flags & IFF_BROADCAST) {
377 sin->sin_addr =
378 ia->ia_broadaddr.sin_addr;
379 break;
380 }
381 }
382 }
383 }
384 /*
385 * If we haven't bound which network number to use as ours,
386 * we will use the number of the outgoing interface.
387 * This depends on having done a routing lookup, which
388 * we will probably have to do anyway, so we might
389 * as well do it now. On the other hand if we are
390 * sending to multiple destinations we may have already
391 * done the lookup, so see if we can use the route
392 * from before. In any case, we only
393 * chose a port number once, even if sending to multiple
394 * destinations.
395 */
396 if (in_nullhost(inp->inp_laddr)) {
397 int error;
398 ifaddr = in_selectsrc(sin, &inp->inp_route,
399 inp->inp_socket->so_options, inp->inp_moptions, &error);
400 if (ifaddr == NULL) {
401 if (error == 0)
402 error = EADDRNOTAVAIL;
403 return error;
404 }
405 }
406 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port,
407 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr,
408 inp->inp_lport) != 0)
409 return (EADDRINUSE);
410 if (in_nullhost(inp->inp_laddr)) {
411 if (inp->inp_lport == 0) {
412 error = in_pcbbind(inp, (struct mbuf *)0,
413 (struct proc *)0);
414 /*
415 * This used to ignore the return value
416 * completely, but we need to check for
417 * ephemeral port shortage.
418 * XXX Should we check for other errors, too?
419 */
420 if (error == EAGAIN)
421 return (error);
422 }
423 inp->inp_laddr = ifaddr->sin_addr;
424 }
425 inp->inp_faddr = sin->sin_addr;
426 inp->inp_fport = sin->sin_port;
427 in_pcbstate(inp, INP_CONNECTED);
428 #ifdef IPSEC
429 if (inp->inp_socket->so_type == SOCK_STREAM)
430 ipsec_pcbconn(inp->inp_sp);
431 #endif
432 return (0);
433 }
434
435 void
436 in_pcbdisconnect(v)
437 void *v;
438 {
439 struct inpcb *inp = v;
440
441 inp->inp_faddr = zeroin_addr;
442 inp->inp_fport = 0;
443 in_pcbstate(inp, INP_BOUND);
444 if (inp->inp_socket->so_state & SS_NOFDREF)
445 in_pcbdetach(inp);
446 #ifdef IPSEC
447 ipsec_pcbdisconn(inp->inp_sp);
448 #endif
449 }
450
451 void
452 in_pcbdetach(v)
453 void *v;
454 {
455 struct inpcb *inp = v;
456 struct socket *so = inp->inp_socket;
457 int s;
458
459 #ifdef IPSEC
460 ipsec4_delete_pcbpolicy(inp);
461 #endif /*IPSEC*/
462 so->so_pcb = 0;
463 sofree(so);
464 if (inp->inp_options)
465 (void)m_free(inp->inp_options);
466 if (inp->inp_route.ro_rt)
467 rtfree(inp->inp_route.ro_rt);
468 ip_freemoptions(inp->inp_moptions);
469 s = splnet();
470 in_pcbstate(inp, INP_ATTACHED);
471 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, inp, inp_queue);
472 splx(s);
473 pool_put(&inpcb_pool, inp);
474 }
475
476 void
477 in_setsockaddr(inp, nam)
478 struct inpcb *inp;
479 struct mbuf *nam;
480 {
481 struct sockaddr_in *sin;
482
483 nam->m_len = sizeof (*sin);
484 sin = mtod(nam, struct sockaddr_in *);
485 bzero((caddr_t)sin, sizeof (*sin));
486 sin->sin_family = AF_INET;
487 sin->sin_len = sizeof(*sin);
488 sin->sin_port = inp->inp_lport;
489 sin->sin_addr = inp->inp_laddr;
490 }
491
492 void
493 in_setpeeraddr(inp, nam)
494 struct inpcb *inp;
495 struct mbuf *nam;
496 {
497 struct sockaddr_in *sin;
498
499 nam->m_len = sizeof (*sin);
500 sin = mtod(nam, struct sockaddr_in *);
501 bzero((caddr_t)sin, sizeof (*sin));
502 sin->sin_family = AF_INET;
503 sin->sin_len = sizeof(*sin);
504 sin->sin_port = inp->inp_fport;
505 sin->sin_addr = inp->inp_faddr;
506 }
507
508 /*
509 * Pass some notification to all connections of a protocol
510 * associated with address dst. The local address and/or port numbers
511 * may be specified to limit the search. The "usual action" will be
512 * taken, depending on the ctlinput cmd. The caller must filter any
513 * cmds that are uninteresting (e.g., no error in the map).
514 * Call the protocol specific routine (if any) to report
515 * any errors for each matching socket.
516 *
517 * Must be called at splsoftnet.
518 */
519 int
520 in_pcbnotify(table, faddr, fport_arg, laddr, lport_arg, errno, notify)
521 struct inpcbtable *table;
522 struct in_addr faddr, laddr;
523 u_int fport_arg, lport_arg;
524 int errno;
525 void (*notify) __P((struct inpcb *, int));
526 {
527 struct inpcbhead *head;
528 struct inpcb *inp, *ninp;
529 u_int16_t fport = fport_arg, lport = lport_arg;
530 int nmatch;
531
532 if (in_nullhost(faddr) || notify == 0)
533 return (0);
534
535 nmatch = 0;
536 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
537 for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) {
538 ninp = LIST_NEXT(inp, inp_hash);
539 if (in_hosteq(inp->inp_faddr, faddr) &&
540 inp->inp_fport == fport &&
541 inp->inp_lport == lport &&
542 in_hosteq(inp->inp_laddr, laddr)) {
543 (*notify)(inp, errno);
544 nmatch++;
545 }
546 }
547 return (nmatch);
548 }
549
550 void
551 in_pcbnotifyall(table, faddr, errno, notify)
552 struct inpcbtable *table;
553 struct in_addr faddr;
554 int errno;
555 void (*notify) __P((struct inpcb *, int));
556 {
557 struct inpcb *inp, *ninp;
558
559 if (in_nullhost(faddr) || notify == 0)
560 return;
561
562 for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
563 inp != (void *)&table->inpt_queue;
564 inp = ninp) {
565 ninp = CIRCLEQ_NEXT(inp, inp_queue);
566 if (in_hosteq(inp->inp_faddr, faddr))
567 (*notify)(inp, errno);
568 }
569 }
570
571 void
572 in_pcbpurgeif0(table, ifp)
573 struct inpcbtable *table;
574 struct ifnet *ifp;
575 {
576 struct inpcb *inp, *ninp;
577 struct ip_moptions *imo;
578 int i, gap;
579
580 for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
581 inp != (void *)&table->inpt_queue;
582 inp = ninp) {
583 ninp = CIRCLEQ_NEXT(inp, inp_queue);
584 imo = inp->inp_moptions;
585 if (imo != NULL) {
586 /*
587 * Unselect the outgoing interface if it is being
588 * detached.
589 */
590 if (imo->imo_multicast_ifp == ifp)
591 imo->imo_multicast_ifp = NULL;
592
593 /*
594 * Drop multicast group membership if we joined
595 * through the interface being detached.
596 */
597 for (i = 0, gap = 0; i < imo->imo_num_memberships;
598 i++) {
599 if (imo->imo_membership[i]->inm_ifp == ifp) {
600 in_delmulti(imo->imo_membership[i]);
601 gap++;
602 } else if (gap != 0)
603 imo->imo_membership[i - gap] =
604 imo->imo_membership[i];
605 }
606 imo->imo_num_memberships -= gap;
607 }
608 }
609 }
610
611 void
612 in_pcbpurgeif(table, ifp)
613 struct inpcbtable *table;
614 struct ifnet *ifp;
615 {
616 struct inpcb *inp, *ninp;
617
618 for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
619 inp != (void *)&table->inpt_queue;
620 inp = ninp) {
621 ninp = CIRCLEQ_NEXT(inp, inp_queue);
622 if (inp->inp_route.ro_rt != NULL &&
623 inp->inp_route.ro_rt->rt_ifp == ifp)
624 in_rtchange(inp, 0);
625 }
626 }
627
628 /*
629 * Check for alternatives when higher level complains
630 * about service problems. For now, invalidate cached
631 * routing information. If the route was created dynamically
632 * (by a redirect), time to try a default gateway again.
633 */
634 void
635 in_losing(inp)
636 struct inpcb *inp;
637 {
638 struct rtentry *rt;
639 struct rt_addrinfo info;
640
641 if ((rt = inp->inp_route.ro_rt)) {
642 inp->inp_route.ro_rt = 0;
643 bzero((caddr_t)&info, sizeof(info));
644 info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst;
645 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
646 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
647 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
648 if (rt->rt_flags & RTF_DYNAMIC)
649 (void) rtrequest(RTM_DELETE, rt_key(rt),
650 rt->rt_gateway, rt_mask(rt), rt->rt_flags,
651 (struct rtentry **)0);
652 else
653 /*
654 * A new route can be allocated
655 * the next time output is attempted.
656 */
657 rtfree(rt);
658 }
659 }
660
661 /*
662 * After a routing change, flush old routing
663 * and allocate a (hopefully) better one.
664 */
665 void
666 in_rtchange(inp, errno)
667 struct inpcb *inp;
668 int errno;
669 {
670
671 if (inp->inp_route.ro_rt) {
672 rtfree(inp->inp_route.ro_rt);
673 inp->inp_route.ro_rt = 0;
674 /*
675 * A new route can be allocated the next time
676 * output is attempted.
677 */
678 }
679 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */
680 }
681
682 struct inpcb *
683 in_pcblookup_port(table, laddr, lport_arg, lookup_wildcard)
684 struct inpcbtable *table;
685 struct in_addr laddr;
686 u_int lport_arg;
687 int lookup_wildcard;
688 {
689 struct inpcb *inp, *match = 0;
690 int matchwild = 3, wildcard;
691 u_int16_t lport = lport_arg;
692
693 CIRCLEQ_FOREACH(inp, &table->inpt_queue, inp_queue) {
694 if (inp->inp_lport != lport)
695 continue;
696 wildcard = 0;
697 if (!in_nullhost(inp->inp_faddr))
698 wildcard++;
699 if (in_nullhost(inp->inp_laddr)) {
700 if (!in_nullhost(laddr))
701 wildcard++;
702 } else {
703 if (in_nullhost(laddr))
704 wildcard++;
705 else {
706 if (!in_hosteq(inp->inp_laddr, laddr))
707 continue;
708 }
709 }
710 if (wildcard && !lookup_wildcard)
711 continue;
712 if (wildcard < matchwild) {
713 match = inp;
714 matchwild = wildcard;
715 if (matchwild == 0)
716 break;
717 }
718 }
719 return (match);
720 }
721
722 #ifdef DIAGNOSTIC
723 int in_pcbnotifymiss = 0;
724 #endif
725
726 struct inpcb *
727 in_pcblookup_connect(table, faddr, fport_arg, laddr, lport_arg)
728 struct inpcbtable *table;
729 struct in_addr faddr, laddr;
730 u_int fport_arg, lport_arg;
731 {
732 struct inpcbhead *head;
733 struct inpcb *inp;
734 u_int16_t fport = fport_arg, lport = lport_arg;
735
736 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
737 LIST_FOREACH(inp, head, inp_hash) {
738 if (in_hosteq(inp->inp_faddr, faddr) &&
739 inp->inp_fport == fport &&
740 inp->inp_lport == lport &&
741 in_hosteq(inp->inp_laddr, laddr))
742 goto out;
743 }
744 #ifdef DIAGNOSTIC
745 if (in_pcbnotifymiss) {
746 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n",
747 ntohl(faddr.s_addr), ntohs(fport),
748 ntohl(laddr.s_addr), ntohs(lport));
749 }
750 #endif
751 return (0);
752
753 out:
754 /* Move this PCB to the head of hash chain. */
755 if (inp != LIST_FIRST(head)) {
756 LIST_REMOVE(inp, inp_hash);
757 LIST_INSERT_HEAD(head, inp, inp_hash);
758 }
759 return (inp);
760 }
761
762 struct inpcb *
763 in_pcblookup_bind(table, laddr, lport_arg)
764 struct inpcbtable *table;
765 struct in_addr laddr;
766 u_int lport_arg;
767 {
768 struct inpcbhead *head;
769 struct inpcb *inp;
770 u_int16_t lport = lport_arg;
771
772 head = INPCBHASH_BIND(table, laddr, lport);
773 LIST_FOREACH(inp, head, inp_hash) {
774 if (inp->inp_lport == lport &&
775 in_hosteq(inp->inp_laddr, laddr))
776 goto out;
777 }
778 head = INPCBHASH_BIND(table, zeroin_addr, lport);
779 LIST_FOREACH(inp, head, inp_hash) {
780 if (inp->inp_lport == lport &&
781 in_hosteq(inp->inp_laddr, zeroin_addr))
782 goto out;
783 }
784 #ifdef DIAGNOSTIC
785 if (in_pcbnotifymiss) {
786 printf("in_pcblookup_bind: laddr=%08x lport=%d\n",
787 ntohl(laddr.s_addr), ntohs(lport));
788 }
789 #endif
790 return (0);
791
792 out:
793 /* Move this PCB to the head of hash chain. */
794 if (inp != LIST_FIRST(head)) {
795 LIST_REMOVE(inp, inp_hash);
796 LIST_INSERT_HEAD(head, inp, inp_hash);
797 }
798 return (inp);
799 }
800
801 void
802 in_pcbstate(inp, state)
803 struct inpcb *inp;
804 int state;
805 {
806
807 if (inp->inp_state > INP_ATTACHED)
808 LIST_REMOVE(inp, inp_hash);
809
810 switch (state) {
811 case INP_BOUND:
812 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table,
813 inp->inp_laddr, inp->inp_lport), inp, inp_hash);
814 break;
815 case INP_CONNECTED:
816 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table,
817 inp->inp_faddr, inp->inp_fport,
818 inp->inp_laddr, inp->inp_lport), inp, inp_hash);
819 break;
820 }
821
822 inp->inp_state = state;
823 }
824
825 struct rtentry *
826 in_pcbrtentry(inp)
827 struct inpcb *inp;
828 {
829 struct route *ro;
830
831 ro = &inp->inp_route;
832
833 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
834 !in_hosteq(satosin(&ro->ro_dst)->sin_addr, inp->inp_faddr))) {
835 RTFREE(ro->ro_rt);
836 ro->ro_rt = (struct rtentry *)NULL;
837 }
838 if (ro->ro_rt == (struct rtentry *)NULL &&
839 !in_nullhost(inp->inp_faddr)) {
840 bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
841 ro->ro_dst.sa_family = AF_INET;
842 ro->ro_dst.sa_len = sizeof(ro->ro_dst);
843 satosin(&ro->ro_dst)->sin_addr = inp->inp_faddr;
844 rtalloc(ro);
845 }
846 return (ro->ro_rt);
847 }
848
849 struct sockaddr_in *
850 in_selectsrc(sin, ro, soopts, mopts, errorp)
851 struct sockaddr_in *sin;
852 struct route *ro;
853 int soopts;
854 struct ip_moptions *mopts;
855 int *errorp;
856 {
857 struct in_ifaddr *ia;
858
859 ia = (struct in_ifaddr *)0;
860 /*
861 * If route is known or can be allocated now,
862 * our src addr is taken from the i/f, else punt.
863 * Note that we should check the address family of the cached
864 * destination, in case of sharing the cache with IPv6.
865 */
866 if (ro->ro_rt &&
867 (ro->ro_dst.sa_family != AF_INET ||
868 !in_hosteq(satosin(&ro->ro_dst)->sin_addr, sin->sin_addr) ||
869 soopts & SO_DONTROUTE)) {
870 RTFREE(ro->ro_rt);
871 ro->ro_rt = (struct rtentry *)0;
872 }
873 if ((soopts & SO_DONTROUTE) == 0 && /*XXX*/
874 (ro->ro_rt == (struct rtentry *)0 ||
875 ro->ro_rt->rt_ifp == (struct ifnet *)0)) {
876 /* No route yet, so try to acquire one */
877 bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
878 ro->ro_dst.sa_family = AF_INET;
879 ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
880 satosin(&ro->ro_dst)->sin_addr = sin->sin_addr;
881 rtalloc(ro);
882 }
883 /*
884 * If we found a route, use the address
885 * corresponding to the outgoing interface
886 * unless it is the loopback (in case a route
887 * to our address on another net goes to loopback).
888 *
889 * XXX Is this still true? Do we care?
890 */
891 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
892 ia = ifatoia(ro->ro_rt->rt_ifa);
893 if (ia == NULL) {
894 u_int16_t fport = sin->sin_port;
895
896 sin->sin_port = 0;
897 ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
898 sin->sin_port = fport;
899 if (ia == 0) {
900 /* Find 1st non-loopback AF_INET address */
901 TAILQ_FOREACH(ia, &in_ifaddr, ia_list) {
902 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK))
903 break;
904 }
905 }
906 if (ia == NULL) {
907 *errorp = EADDRNOTAVAIL;
908 return NULL;
909 }
910 }
911 /*
912 * If the destination address is multicast and an outgoing
913 * interface has been set as a multicast option, use the
914 * address of that interface as our source address.
915 */
916 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) {
917 struct ip_moptions *imo;
918 struct ifnet *ifp;
919
920 imo = mopts;
921 if (imo->imo_multicast_ifp != NULL) {
922 ifp = imo->imo_multicast_ifp;
923 IFP_TO_IA(ifp, ia); /* XXX */
924 if (ia == 0) {
925 *errorp = EADDRNOTAVAIL;
926 return NULL;
927 }
928 }
929 }
930 return satosin(&ia->ia_addr);
931 }
932