in_pcb.c revision 1.86
1 /* $NetBSD: in_pcb.c,v 1.86 2003/08/07 16:33:10 agc 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 1.86 2003/08/07 16:33:10 agc Exp $"); 102 103 #include "opt_ipsec.h" 104 105 #include <sys/param.h> 106 #include <sys/systm.h> 107 #include <sys/malloc.h> 108 #include <sys/mbuf.h> 109 #include <sys/protosw.h> 110 #include <sys/socket.h> 111 #include <sys/socketvar.h> 112 #include <sys/ioctl.h> 113 #include <sys/errno.h> 114 #include <sys/time.h> 115 #include <sys/pool.h> 116 #include <sys/proc.h> 117 118 #include <net/if.h> 119 #include <net/route.h> 120 121 #include <netinet/in.h> 122 #include <netinet/in_systm.h> 123 #include <netinet/ip.h> 124 #include <netinet/in_pcb.h> 125 #include <netinet/in_var.h> 126 #include <netinet/ip_var.h> 127 128 #ifdef IPSEC 129 #include <netinet6/ipsec.h> 130 #include <netkey/key.h> 131 #endif /* IPSEC */ 132 133 struct in_addr zeroin_addr; 134 135 #define INPCBHASH_BIND(table, laddr, lport) \ 136 &(table)->inpt_bindhashtbl[ \ 137 ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash] 138 #define INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \ 139 &(table)->inpt_connecthashtbl[ \ 140 ((ntohl((faddr).s_addr) + ntohs(fport)) + \ 141 (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash] 142 143 struct inpcb * 144 in_pcblookup_port __P((struct inpcbtable *, 145 struct in_addr, u_int, int)); 146 147 int anonportmin = IPPORT_ANONMIN; 148 int anonportmax = IPPORT_ANONMAX; 149 int lowportmin = IPPORT_RESERVEDMIN; 150 int lowportmax = IPPORT_RESERVEDMAX; 151 152 struct pool inpcb_pool; 153 154 void 155 in_pcbinit(table, bindhashsize, connecthashsize) 156 struct inpcbtable *table; 157 int bindhashsize, connecthashsize; 158 { 159 static int inpcb_pool_initialized; 160 161 if (inpcb_pool_initialized == 0) { 162 pool_init(&inpcb_pool, sizeof(struct inpcb), 0, 0, 0, 163 "inpcbpl", NULL); 164 inpcb_pool_initialized = 1; 165 } 166 167 CIRCLEQ_INIT(&table->inpt_queue); 168 table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, M_PCB, 169 M_WAITOK, &table->inpt_bindhash); 170 table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST, 171 M_PCB, M_WAITOK, &table->inpt_connecthash); 172 table->inpt_lastlow = IPPORT_RESERVEDMAX; 173 table->inpt_lastport = (u_int16_t)anonportmax; 174 } 175 176 int 177 in_pcballoc(so, v) 178 struct socket *so; 179 void *v; 180 { 181 struct inpcbtable *table = v; 182 struct inpcb *inp; 183 int s; 184 #ifdef 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_table = table; 193 inp->inp_socket = so; 194 inp->inp_errormtu = -1; 195 #ifdef IPSEC 196 error = ipsec_init_pcbpolicy(so, &inp->inp_sp); 197 if (error != 0) { 198 pool_put(&inpcb_pool, inp); 199 return error; 200 } 201 #endif 202 so->so_pcb = inp; 203 s = splnet(); 204 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, inp, inp_queue); 205 in_pcbstate(inp, INP_ATTACHED); 206 splx(s); 207 return (0); 208 } 209 210 int 211 in_pcbbind(v, nam, p) 212 void *v; 213 struct mbuf *nam; 214 struct proc *p; 215 { 216 struct in_ifaddr *ia = NULL; 217 struct inpcb *inp = v; 218 struct socket *so = inp->inp_socket; 219 struct inpcbtable *table = inp->inp_table; 220 struct sockaddr_in *sin; 221 u_int16_t lport = 0; 222 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 223 224 if (TAILQ_FIRST(&in_ifaddr) == 0) 225 return (EADDRNOTAVAIL); 226 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 227 return (EINVAL); 228 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 229 wild = 1; 230 if (nam == 0) 231 goto noname; 232 sin = mtod(nam, struct sockaddr_in *); 233 if (nam->m_len != sizeof (*sin)) 234 return (EINVAL); 235 if (sin->sin_family != AF_INET) 236 return (EAFNOSUPPORT); 237 lport = sin->sin_port; 238 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 239 /* 240 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 241 * allow complete duplication of binding if 242 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 243 * and a multicast address is bound on both 244 * new and duplicated sockets. 245 */ 246 if (so->so_options & SO_REUSEADDR) 247 reuseport = SO_REUSEADDR|SO_REUSEPORT; 248 } else if (!in_nullhost(sin->sin_addr)) { 249 sin->sin_port = 0; /* yech... */ 250 INADDR_TO_IA(sin->sin_addr, ia); 251 /* check for broadcast addresses */ 252 if (ia == NULL) 253 ia = ifatoia(ifa_ifwithaddr(sintosa(sin))); 254 if (ia == NULL) 255 return (EADDRNOTAVAIL); 256 } 257 if (lport) { 258 struct inpcb *t; 259 #ifndef IPNOPRIVPORTS 260 /* GROSS */ 261 if (ntohs(lport) < IPPORT_RESERVED && 262 (p == 0 || suser(p->p_ucred, &p->p_acflag))) 263 return (EACCES); 264 #endif 265 if (so->so_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) { 266 t = in_pcblookup_port(table, sin->sin_addr, lport, 1); 267 /* 268 * XXX: investigate ramifications of loosening this 269 * restriction so that as long as both ports have 270 * SO_REUSEPORT allow the bind 271 */ 272 if (t && 273 (!in_nullhost(sin->sin_addr) || 274 !in_nullhost(t->inp_laddr) || 275 (t->inp_socket->so_options & SO_REUSEPORT) == 0) 276 && (so->so_uid != t->inp_socket->so_uid)) { 277 return (EADDRINUSE); 278 } 279 } 280 t = in_pcblookup_port(table, sin->sin_addr, lport, wild); 281 if (t && (reuseport & t->inp_socket->so_options) == 0) 282 return (EADDRINUSE); 283 } 284 if (!in_nullhost(inp->inp_laddr)) { 285 KASSERT(inp->inp_ia != NULL); 286 LIST_REMOVE(inp, inp_ialink); 287 IFAFREE(&inp->inp_ia->ia_ifa); 288 inp->inp_ia = NULL; 289 } 290 inp->inp_laddr = sin->sin_addr; 291 if (ia != NULL) { 292 inp->inp_ia = ia; 293 LIST_INSERT_HEAD(&ia->ia_inpcbs, inp, inp_ialink); 294 IFAREF(&ia->ia_ifa); 295 } 296 297 298 noname: 299 if (lport == 0) { 300 int cnt; 301 u_int16_t min, max; 302 u_int16_t *lastport; 303 304 if (inp->inp_flags & INP_LOWPORT) { 305 #ifndef IPNOPRIVPORTS 306 if (p == 0 || suser(p->p_ucred, &p->p_acflag)) 307 return (EACCES); 308 #endif 309 min = lowportmin; 310 max = lowportmax; 311 lastport = &table->inpt_lastlow; 312 } else { 313 min = anonportmin; 314 max = anonportmax; 315 lastport = &table->inpt_lastport; 316 } 317 if (min > max) { /* sanity check */ 318 u_int16_t swp; 319 320 swp = min; 321 min = max; 322 max = swp; 323 } 324 325 lport = *lastport - 1; 326 for (cnt = max - min + 1; cnt; cnt--, lport--) { 327 if (lport < min || lport > max) 328 lport = max; 329 if (!in_pcblookup_port(table, inp->inp_laddr, 330 htons(lport), 1)) 331 goto found; 332 } 333 if (!in_nullhost(inp->inp_laddr)) { 334 if (inp->inp_ia != NULL) { 335 LIST_REMOVE(inp, inp_ialink); 336 IFAFREE(&inp->inp_ia->ia_ifa); 337 inp->inp_ia = NULL; 338 } 339 inp->inp_laddr.s_addr = INADDR_ANY; 340 } 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 in_pcbstate(inp, INP_BOUND); 349 return (0); 350 } 351 352 /* 353 * Connect from a socket to a specified address. 354 * Both address and port must be specified in argument sin. 355 * If don't have a local address for this socket yet, 356 * then pick one. 357 */ 358 int 359 in_pcbconnect(v, nam) 360 void *v; 361 struct mbuf *nam; 362 { 363 struct inpcb *inp = v; 364 struct in_ifaddr *ia = NULL; 365 struct sockaddr_in *ifaddr = NULL; 366 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); 367 int error; 368 369 if (nam->m_len != sizeof (*sin)) 370 return (EINVAL); 371 if (sin->sin_family != AF_INET) 372 return (EAFNOSUPPORT); 373 if (sin->sin_port == 0) 374 return (EADDRNOTAVAIL); 375 if (TAILQ_FIRST(&in_ifaddr) != 0) { 376 /* 377 * If the destination address is INADDR_ANY, 378 * use any local address (likely loopback). 379 * If the supplied address is INADDR_BROADCAST, 380 * use the broadcast address of an interface 381 * which supports broadcast. (loopback does not) 382 */ 383 384 if (in_nullhost(sin->sin_addr)) { 385 sin->sin_addr = 386 TAILQ_FIRST(&in_ifaddr)->ia_addr.sin_addr; 387 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) { 388 TAILQ_FOREACH(ia, &in_ifaddr, ia_list) { 389 if (ia->ia_ifp->if_flags & IFF_BROADCAST) { 390 sin->sin_addr = 391 ia->ia_broadaddr.sin_addr; 392 break; 393 } 394 } 395 } 396 } 397 /* 398 * If we haven't bound which network number to use as ours, 399 * we will use the number of the outgoing interface. 400 * This depends on having done a routing lookup, which 401 * we will probably have to do anyway, so we might 402 * as well do it now. On the other hand if we are 403 * sending to multiple destinations we may have already 404 * done the lookup, so see if we can use the route 405 * from before. In any case, we only 406 * chose a port number once, even if sending to multiple 407 * destinations. 408 */ 409 if (in_nullhost(inp->inp_laddr)) { 410 int error; 411 ifaddr = in_selectsrc(sin, &inp->inp_route, 412 inp->inp_socket->so_options, inp->inp_moptions, &error); 413 if (ifaddr == NULL) { 414 if (error == 0) 415 error = EADDRNOTAVAIL; 416 return error; 417 } 418 INADDR_TO_IA(ifaddr->sin_addr, ia); 419 if (ia == NULL) 420 return (EADDRNOTAVAIL); 421 } 422 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port, 423 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr, 424 inp->inp_lport) != 0) 425 return (EADDRINUSE); 426 if (in_nullhost(inp->inp_laddr)) { 427 if (inp->inp_lport == 0) { 428 error = in_pcbbind(inp, (struct mbuf *)0, 429 (struct proc *)0); 430 /* 431 * This used to ignore the return value 432 * completely, but we need to check for 433 * ephemeral port shortage. 434 * XXX Should we check for other errors, too? 435 */ 436 if (error == EAGAIN) 437 return (error); 438 } 439 KASSERT(inp->inp_ia == NULL); 440 inp->inp_laddr = ia->ia_addr.sin_addr; 441 inp->inp_ia = ia; 442 LIST_INSERT_HEAD(&ia->ia_inpcbs, inp, inp_ialink); 443 IFAREF(&ia->ia_ifa); 444 inp->inp_laddr = ifaddr->sin_addr; 445 } 446 inp->inp_faddr = sin->sin_addr; 447 inp->inp_fport = sin->sin_port; 448 in_pcbstate(inp, INP_CONNECTED); 449 #ifdef IPSEC 450 if (inp->inp_socket->so_type == SOCK_STREAM) 451 ipsec_pcbconn(inp->inp_sp); 452 #endif 453 return (0); 454 } 455 456 void 457 in_pcbdisconnect(v) 458 void *v; 459 { 460 struct inpcb *inp = v; 461 462 inp->inp_faddr = zeroin_addr; 463 inp->inp_fport = 0; 464 in_pcbstate(inp, INP_BOUND); 465 if (inp->inp_socket->so_state & SS_NOFDREF) 466 in_pcbdetach(inp); 467 #ifdef IPSEC 468 ipsec_pcbdisconn(inp->inp_sp); 469 #endif 470 } 471 472 void 473 in_pcbdetach(v) 474 void *v; 475 { 476 struct inpcb *inp = v; 477 struct socket *so = inp->inp_socket; 478 int s; 479 480 #ifdef IPSEC 481 ipsec4_delete_pcbpolicy(inp); 482 #endif /*IPSEC*/ 483 so->so_pcb = 0; 484 sofree(so); 485 if (inp->inp_options) 486 (void)m_free(inp->inp_options); 487 if (inp->inp_route.ro_rt) 488 rtfree(inp->inp_route.ro_rt); 489 ip_freemoptions(inp->inp_moptions); 490 if (inp->inp_ia != NULL) { 491 LIST_REMOVE(inp, inp_ialink); 492 IFAFREE(&inp->inp_ia->ia_ifa); 493 inp->inp_ia = NULL; 494 } 495 s = splnet(); 496 in_pcbstate(inp, INP_ATTACHED); 497 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, inp, inp_queue); 498 splx(s); 499 pool_put(&inpcb_pool, inp); 500 } 501 502 void 503 in_setsockaddr(inp, nam) 504 struct inpcb *inp; 505 struct mbuf *nam; 506 { 507 struct sockaddr_in *sin; 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(inp, nam) 520 struct inpcb *inp; 521 struct mbuf *nam; 522 { 523 struct sockaddr_in *sin; 524 525 nam->m_len = sizeof (*sin); 526 sin = mtod(nam, struct sockaddr_in *); 527 bzero((caddr_t)sin, sizeof (*sin)); 528 sin->sin_family = AF_INET; 529 sin->sin_len = sizeof(*sin); 530 sin->sin_port = inp->inp_fport; 531 sin->sin_addr = inp->inp_faddr; 532 } 533 534 /* 535 * Pass some notification to all connections of a protocol 536 * associated with address dst. The local address and/or port numbers 537 * may be specified to limit the search. The "usual action" will be 538 * taken, depending on the ctlinput cmd. The caller must filter any 539 * cmds that are uninteresting (e.g., no error in the map). 540 * Call the protocol specific routine (if any) to report 541 * any errors for each matching socket. 542 * 543 * Must be called at splsoftnet. 544 */ 545 int 546 in_pcbnotify(table, faddr, fport_arg, laddr, lport_arg, errno, notify) 547 struct inpcbtable *table; 548 struct in_addr faddr, laddr; 549 u_int fport_arg, lport_arg; 550 int errno; 551 void (*notify) __P((struct inpcb *, int)); 552 { 553 struct inpcbhead *head; 554 struct inpcb *inp, *ninp; 555 u_int16_t fport = fport_arg, lport = lport_arg; 556 int nmatch; 557 558 if (in_nullhost(faddr) || notify == 0) 559 return (0); 560 561 nmatch = 0; 562 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 563 for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) { 564 ninp = LIST_NEXT(inp, inp_hash); 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(table, faddr, errno, notify) 578 struct inpcbtable *table; 579 struct in_addr faddr; 580 int errno; 581 void (*notify) __P((struct inpcb *, int)); 582 { 583 struct inpcb *inp, *ninp; 584 585 if (in_nullhost(faddr) || notify == 0) 586 return; 587 588 for (inp = CIRCLEQ_FIRST(&table->inpt_queue); 589 inp != (void *)&table->inpt_queue; 590 inp = ninp) { 591 ninp = CIRCLEQ_NEXT(inp, inp_queue); 592 if (in_hosteq(inp->inp_faddr, faddr)) 593 (*notify)(inp, errno); 594 } 595 } 596 597 void 598 in_pcbpurgeif0(table, ifp) 599 struct inpcbtable *table; 600 struct ifnet *ifp; 601 { 602 struct inpcb *inp, *ninp; 603 struct ip_moptions *imo; 604 int i, gap; 605 606 for (inp = CIRCLEQ_FIRST(&table->inpt_queue); 607 inp != (void *)&table->inpt_queue; 608 inp = ninp) { 609 ninp = CIRCLEQ_NEXT(inp, inp_queue); 610 imo = inp->inp_moptions; 611 if (imo != NULL) { 612 /* 613 * Unselect the outgoing interface if it is being 614 * detached. 615 */ 616 if (imo->imo_multicast_ifp == ifp) 617 imo->imo_multicast_ifp = NULL; 618 619 /* 620 * Drop multicast group membership if we joined 621 * through the interface being detached. 622 */ 623 for (i = 0, gap = 0; i < imo->imo_num_memberships; 624 i++) { 625 if (imo->imo_membership[i]->inm_ifp == ifp) { 626 in_delmulti(imo->imo_membership[i]); 627 gap++; 628 } else if (gap != 0) 629 imo->imo_membership[i - gap] = 630 imo->imo_membership[i]; 631 } 632 imo->imo_num_memberships -= gap; 633 } 634 } 635 } 636 637 void 638 in_pcbpurgeif(table, ifp) 639 struct inpcbtable *table; 640 struct ifnet *ifp; 641 { 642 struct inpcb *inp, *ninp; 643 644 for (inp = CIRCLEQ_FIRST(&table->inpt_queue); 645 inp != (void *)&table->inpt_queue; 646 inp = ninp) { 647 ninp = CIRCLEQ_NEXT(inp, inp_queue); 648 if (inp->inp_route.ro_rt != NULL && 649 inp->inp_route.ro_rt->rt_ifp == ifp) 650 in_rtchange(inp, 0); 651 } 652 } 653 654 /* 655 * Check for alternatives when higher level complains 656 * about service problems. For now, invalidate cached 657 * routing information. If the route was created dynamically 658 * (by a redirect), time to try a default gateway again. 659 */ 660 void 661 in_losing(inp) 662 struct inpcb *inp; 663 { 664 struct rtentry *rt; 665 struct rt_addrinfo info; 666 667 if ((rt = inp->inp_route.ro_rt)) { 668 inp->inp_route.ro_rt = 0; 669 bzero((caddr_t)&info, sizeof(info)); 670 info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst; 671 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 672 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 673 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 674 if (rt->rt_flags & RTF_DYNAMIC) 675 (void) rtrequest(RTM_DELETE, rt_key(rt), 676 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 677 (struct rtentry **)0); 678 else 679 /* 680 * A new route can be allocated 681 * the next time output is attempted. 682 */ 683 rtfree(rt); 684 } 685 } 686 687 /* 688 * After a routing change, flush old routing 689 * and allocate a (hopefully) better one. 690 */ 691 void 692 in_rtchange(inp, errno) 693 struct inpcb *inp; 694 int errno; 695 { 696 697 if (inp->inp_route.ro_rt) { 698 rtfree(inp->inp_route.ro_rt); 699 inp->inp_route.ro_rt = 0; 700 /* 701 * A new route can be allocated the next time 702 * output is attempted. 703 */ 704 } 705 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 706 } 707 708 struct inpcb * 709 in_pcblookup_port(table, laddr, lport_arg, lookup_wildcard) 710 struct inpcbtable *table; 711 struct in_addr laddr; 712 u_int lport_arg; 713 int lookup_wildcard; 714 { 715 struct inpcb *inp, *match = 0; 716 int matchwild = 3, wildcard; 717 u_int16_t lport = lport_arg; 718 719 CIRCLEQ_FOREACH(inp, &table->inpt_queue, inp_queue) { 720 if (inp->inp_lport != lport) 721 continue; 722 wildcard = 0; 723 if (!in_nullhost(inp->inp_faddr)) 724 wildcard++; 725 if (in_nullhost(inp->inp_laddr)) { 726 if (!in_nullhost(laddr)) 727 wildcard++; 728 } else { 729 if (in_nullhost(laddr)) 730 wildcard++; 731 else { 732 if (!in_hosteq(inp->inp_laddr, laddr)) 733 continue; 734 } 735 } 736 if (wildcard && !lookup_wildcard) 737 continue; 738 if (wildcard < matchwild) { 739 match = inp; 740 matchwild = wildcard; 741 if (matchwild == 0) 742 break; 743 } 744 } 745 return (match); 746 } 747 748 #ifdef DIAGNOSTIC 749 int in_pcbnotifymiss = 0; 750 #endif 751 752 struct inpcb * 753 in_pcblookup_connect(table, faddr, fport_arg, laddr, lport_arg) 754 struct inpcbtable *table; 755 struct in_addr faddr, laddr; 756 u_int fport_arg, lport_arg; 757 { 758 struct inpcbhead *head; 759 struct inpcb *inp; 760 u_int16_t fport = fport_arg, lport = lport_arg; 761 762 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 763 LIST_FOREACH(inp, head, inp_hash) { 764 if (in_hosteq(inp->inp_faddr, faddr) && 765 inp->inp_fport == fport && 766 inp->inp_lport == lport && 767 in_hosteq(inp->inp_laddr, laddr)) 768 goto out; 769 } 770 #ifdef DIAGNOSTIC 771 if (in_pcbnotifymiss) { 772 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 773 ntohl(faddr.s_addr), ntohs(fport), 774 ntohl(laddr.s_addr), ntohs(lport)); 775 } 776 #endif 777 return (0); 778 779 out: 780 /* Move this PCB to the head of hash chain. */ 781 if (inp != LIST_FIRST(head)) { 782 LIST_REMOVE(inp, inp_hash); 783 LIST_INSERT_HEAD(head, inp, inp_hash); 784 } 785 return (inp); 786 } 787 788 struct inpcb * 789 in_pcblookup_bind(table, laddr, lport_arg) 790 struct inpcbtable *table; 791 struct in_addr laddr; 792 u_int lport_arg; 793 { 794 struct inpcbhead *head; 795 struct inpcb *inp; 796 u_int16_t lport = lport_arg; 797 798 head = INPCBHASH_BIND(table, laddr, lport); 799 LIST_FOREACH(inp, head, inp_hash) { 800 if (inp->inp_lport == lport && 801 in_hosteq(inp->inp_laddr, laddr)) 802 goto out; 803 } 804 head = INPCBHASH_BIND(table, zeroin_addr, lport); 805 LIST_FOREACH(inp, head, inp_hash) { 806 if (inp->inp_lport == lport && 807 in_hosteq(inp->inp_laddr, zeroin_addr)) 808 goto out; 809 } 810 #ifdef DIAGNOSTIC 811 if (in_pcbnotifymiss) { 812 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 813 ntohl(laddr.s_addr), ntohs(lport)); 814 } 815 #endif 816 return (0); 817 818 out: 819 /* Move this PCB to the head of hash chain. */ 820 if (inp != LIST_FIRST(head)) { 821 LIST_REMOVE(inp, inp_hash); 822 LIST_INSERT_HEAD(head, inp, inp_hash); 823 } 824 return (inp); 825 } 826 827 void 828 in_pcbstate(inp, state) 829 struct inpcb *inp; 830 int state; 831 { 832 833 if (inp->inp_state > INP_ATTACHED) 834 LIST_REMOVE(inp, inp_hash); 835 836 switch (state) { 837 case INP_BOUND: 838 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 839 inp->inp_laddr, inp->inp_lport), inp, inp_hash); 840 break; 841 case INP_CONNECTED: 842 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 843 inp->inp_faddr, inp->inp_fport, 844 inp->inp_laddr, inp->inp_lport), inp, inp_hash); 845 break; 846 } 847 848 inp->inp_state = state; 849 } 850 851 struct rtentry * 852 in_pcbrtentry(inp) 853 struct inpcb *inp; 854 { 855 struct route *ro; 856 857 ro = &inp->inp_route; 858 859 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 860 !in_hosteq(satosin(&ro->ro_dst)->sin_addr, inp->inp_faddr))) { 861 RTFREE(ro->ro_rt); 862 ro->ro_rt = (struct rtentry *)NULL; 863 } 864 if (ro->ro_rt == (struct rtentry *)NULL && 865 !in_nullhost(inp->inp_faddr)) { 866 bzero(&ro->ro_dst, sizeof(struct sockaddr_in)); 867 ro->ro_dst.sa_family = AF_INET; 868 ro->ro_dst.sa_len = sizeof(ro->ro_dst); 869 satosin(&ro->ro_dst)->sin_addr = inp->inp_faddr; 870 rtalloc(ro); 871 } 872 return (ro->ro_rt); 873 } 874 875 struct sockaddr_in * 876 in_selectsrc(sin, ro, soopts, mopts, errorp) 877 struct sockaddr_in *sin; 878 struct route *ro; 879 int soopts; 880 struct ip_moptions *mopts; 881 int *errorp; 882 { 883 struct in_ifaddr *ia; 884 885 ia = (struct in_ifaddr *)0; 886 /* 887 * If route is known or can be allocated now, 888 * our src addr is taken from the i/f, else punt. 889 * Note that we should check the address family of the cached 890 * destination, in case of sharing the cache with IPv6. 891 */ 892 if (ro->ro_rt && 893 (ro->ro_dst.sa_family != AF_INET || 894 !in_hosteq(satosin(&ro->ro_dst)->sin_addr, sin->sin_addr) || 895 soopts & SO_DONTROUTE)) { 896 RTFREE(ro->ro_rt); 897 ro->ro_rt = (struct rtentry *)0; 898 } 899 if ((soopts & SO_DONTROUTE) == 0 && /*XXX*/ 900 (ro->ro_rt == (struct rtentry *)0 || 901 ro->ro_rt->rt_ifp == (struct ifnet *)0)) { 902 /* No route yet, so try to acquire one */ 903 bzero(&ro->ro_dst, sizeof(struct sockaddr_in)); 904 ro->ro_dst.sa_family = AF_INET; 905 ro->ro_dst.sa_len = sizeof(struct sockaddr_in); 906 satosin(&ro->ro_dst)->sin_addr = sin->sin_addr; 907 rtalloc(ro); 908 } 909 /* 910 * If we found a route, use the address 911 * corresponding to the outgoing interface 912 * unless it is the loopback (in case a route 913 * to our address on another net goes to loopback). 914 * 915 * XXX Is this still true? Do we care? 916 */ 917 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) 918 ia = ifatoia(ro->ro_rt->rt_ifa); 919 if (ia == NULL) { 920 u_int16_t fport = sin->sin_port; 921 922 sin->sin_port = 0; 923 ia = ifatoia(ifa_ifwithladdr(sintosa(sin))); 924 sin->sin_port = fport; 925 if (ia == 0) { 926 /* Find 1st non-loopback AF_INET address */ 927 TAILQ_FOREACH(ia, &in_ifaddr, ia_list) { 928 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK)) 929 break; 930 } 931 } 932 if (ia == NULL) { 933 *errorp = EADDRNOTAVAIL; 934 return NULL; 935 } 936 } 937 /* 938 * If the destination address is multicast and an outgoing 939 * interface has been set as a multicast option, use the 940 * address of that interface as our source address. 941 */ 942 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) { 943 struct ip_moptions *imo; 944 struct ifnet *ifp; 945 946 imo = mopts; 947 if (imo->imo_multicast_ifp != NULL) { 948 ifp = imo->imo_multicast_ifp; 949 IFP_TO_IA(ifp, ia); /* XXX */ 950 if (ia == 0) { 951 *errorp = EADDRNOTAVAIL; 952 return NULL; 953 } 954 } 955 } 956 return satosin(&ia->ia_addr); 957 } 958
Indexes created Mon Sep 22 05:09:51 GMT 2025