in_pcb.c revision 1.97
1 /* $NetBSD: in_pcb.c,v 1.97 2005/02/02 21:41:55 perry 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 1.97 2005/02/02 21:41:55 perry 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> 118 119 #include <net/if.h> 120 #include <net/route.h> 121 122 #include <netinet/in.h> 123 #include <netinet/in_systm.h> 124 #include <netinet/ip.h> 125 #include <netinet/in_pcb.h> 126 #include <netinet/in_var.h> 127 #include <netinet/ip_var.h> 128 129 #ifdef INET6 130 #include <netinet/ip6.h> 131 #include <netinet6/ip6_var.h> 132 #include <netinet6/in6_pcb.h> 133 #endif 134 135 #ifdef IPSEC 136 #include <netinet6/ipsec.h> 137 #include <netkey/key.h> 138 #elif FAST_IPSEC 139 #include <netipsec/ipsec.h> 140 #include <netipsec/key.h> 141 #endif /* IPSEC */ 142 143 struct in_addr zeroin_addr; 144 145 #define INPCBHASH_PORT(table, lport) \ 146 &(table)->inpt_porthashtbl[ntohs(lport) & (table)->inpt_porthash] 147 #define INPCBHASH_BIND(table, laddr, lport) \ 148 &(table)->inpt_bindhashtbl[ \ 149 ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash] 150 #define INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \ 151 &(table)->inpt_connecthashtbl[ \ 152 ((ntohl((faddr).s_addr) + ntohs(fport)) + \ 153 (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash] 154 155 int anonportmin = IPPORT_ANONMIN; 156 int anonportmax = IPPORT_ANONMAX; 157 int lowportmin = IPPORT_RESERVEDMIN; 158 int lowportmax = IPPORT_RESERVEDMAX; 159 160 POOL_INIT(inpcb_pool, sizeof(struct inpcb), 0, 0, 0, "inpcbpl", NULL); 161 162 void 163 in_pcbinit(table, bindhashsize, connecthashsize) 164 struct inpcbtable *table; 165 int bindhashsize, connecthashsize; 166 { 167 168 CIRCLEQ_INIT(&table->inpt_queue); 169 table->inpt_porthashtbl = hashinit(bindhashsize, HASH_LIST, M_PCB, 170 M_WAITOK, &table->inpt_porthash); 171 table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, M_PCB, 172 M_WAITOK, &table->inpt_bindhash); 173 table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST, 174 M_PCB, M_WAITOK, &table->inpt_connecthash); 175 table->inpt_lastlow = IPPORT_RESERVEDMAX; 176 table->inpt_lastport = (u_int16_t)anonportmax; 177 } 178 179 int 180 in_pcballoc(so, v) 181 struct socket *so; 182 void *v; 183 { 184 struct inpcbtable *table = v; 185 struct inpcb *inp; 186 int s; 187 #if defined(IPSEC) || defined(FAST_IPSEC) 188 int error; 189 #endif 190 191 inp = pool_get(&inpcb_pool, PR_NOWAIT); 192 if (inp == NULL) 193 return (ENOBUFS); 194 bzero((caddr_t)inp, sizeof(*inp)); 195 inp->inp_af = AF_INET; 196 inp->inp_table = table; 197 inp->inp_socket = so; 198 inp->inp_errormtu = -1; 199 #if defined(IPSEC) || defined(FAST_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_head, 209 inph_queue); 210 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 211 inph_lhash); 212 in_pcbstate(inp, INP_ATTACHED); 213 splx(s); 214 return (0); 215 } 216 217 int 218 in_pcbbind(v, nam, p) 219 void *v; 220 struct mbuf *nam; 221 struct proc *p; 222 { 223 struct in_ifaddr *ia = NULL; 224 struct inpcb *inp = v; 225 struct socket *so = inp->inp_socket; 226 struct inpcbtable *table = inp->inp_table; 227 struct sockaddr_in *sin; 228 u_int16_t lport = 0; 229 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 230 231 if (inp->inp_af != AF_INET) 232 return (EINVAL); 233 234 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 235 return (EADDRNOTAVAIL); 236 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 237 return (EINVAL); 238 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 239 wild = 1; 240 if (nam == 0) 241 goto noname; 242 sin = mtod(nam, struct sockaddr_in *); 243 if (nam->m_len != sizeof (*sin)) 244 return (EINVAL); 245 if (sin->sin_family != AF_INET) 246 return (EAFNOSUPPORT); 247 lport = sin->sin_port; 248 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 249 /* 250 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 251 * allow complete duplication of binding if 252 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 253 * and a multicast address is bound on both 254 * new and duplicated sockets. 255 */ 256 if (so->so_options & SO_REUSEADDR) 257 reuseport = SO_REUSEADDR|SO_REUSEPORT; 258 } else if (!in_nullhost(sin->sin_addr)) { 259 sin->sin_port = 0; /* yech... */ 260 INADDR_TO_IA(sin->sin_addr, ia); 261 /* check for broadcast addresses */ 262 if (ia == NULL) 263 ia = ifatoia(ifa_ifwithaddr(sintosa(sin))); 264 if (ia == NULL) 265 return (EADDRNOTAVAIL); 266 } 267 if (lport) { 268 struct inpcb *t; 269 #ifdef INET6 270 struct in6pcb *t6; 271 struct in6_addr mapped; 272 #endif 273 #ifndef IPNOPRIVPORTS 274 /* GROSS */ 275 if (ntohs(lport) < IPPORT_RESERVED && 276 (p == 0 || suser(p->p_ucred, &p->p_acflag))) 277 return (EACCES); 278 #endif 279 #ifdef INET6 280 memset(&mapped, 0, sizeof(mapped)); 281 mapped.s6_addr16[5] = 0xffff; 282 memcpy(&mapped.s6_addr32[3], &sin->sin_addr, 283 sizeof(mapped.s6_addr32[3])); 284 t6 = in6_pcblookup_port(table, &mapped, lport, wild); 285 if (t6 && (reuseport & t6->in6p_socket->so_options) == 0) 286 return (EADDRINUSE); 287 #endif 288 if (so->so_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) { 289 t = in_pcblookup_port(table, sin->sin_addr, lport, 1); 290 /* 291 * XXX: investigate ramifications of loosening this 292 * restriction so that as long as both ports have 293 * SO_REUSEPORT allow the bind 294 */ 295 if (t && 296 (!in_nullhost(sin->sin_addr) || 297 !in_nullhost(t->inp_laddr) || 298 (t->inp_socket->so_options & SO_REUSEPORT) == 0) 299 && (so->so_uid != t->inp_socket->so_uid)) { 300 return (EADDRINUSE); 301 } 302 } 303 t = in_pcblookup_port(table, sin->sin_addr, lport, wild); 304 if (t && (reuseport & t->inp_socket->so_options) == 0) 305 return (EADDRINUSE); 306 } 307 inp->inp_laddr = sin->sin_addr; 308 309 noname: 310 if (lport == 0) { 311 int cnt; 312 u_int16_t min, max; 313 u_int16_t *lastport; 314 315 if (inp->inp_flags & INP_LOWPORT) { 316 #ifndef IPNOPRIVPORTS 317 if (p == 0 || suser(p->p_ucred, &p->p_acflag)) 318 return (EACCES); 319 #endif 320 min = lowportmin; 321 max = lowportmax; 322 lastport = &table->inpt_lastlow; 323 } else { 324 min = anonportmin; 325 max = anonportmax; 326 lastport = &table->inpt_lastport; 327 } 328 if (min > max) { /* sanity check */ 329 u_int16_t swp; 330 331 swp = min; 332 min = max; 333 max = swp; 334 } 335 336 lport = *lastport - 1; 337 for (cnt = max - min + 1; cnt; cnt--, lport--) { 338 if (lport < min || lport > max) 339 lport = max; 340 if (!in_pcblookup_port(table, inp->inp_laddr, 341 htons(lport), 1)) 342 goto found; 343 } 344 if (!in_nullhost(inp->inp_laddr)) 345 inp->inp_laddr.s_addr = INADDR_ANY; 346 return (EAGAIN); 347 found: 348 inp->inp_flags |= INP_ANONPORT; 349 *lastport = lport; 350 lport = htons(lport); 351 } 352 inp->inp_lport = lport; 353 LIST_REMOVE(&inp->inp_head, inph_lhash); 354 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 355 inph_lhash); 356 in_pcbstate(inp, INP_BOUND); 357 return (0); 358 } 359 360 /* 361 * Connect from a socket to a specified address. 362 * Both address and port must be specified in argument sin. 363 * If don't have a local address for this socket yet, 364 * then pick one. 365 */ 366 int 367 in_pcbconnect(v, nam) 368 void *v; 369 struct mbuf *nam; 370 { 371 struct inpcb *inp = v; 372 struct in_ifaddr *ia = NULL; 373 struct sockaddr_in *ifaddr = NULL; 374 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); 375 int error; 376 377 if (inp->inp_af != AF_INET) 378 return (EINVAL); 379 380 if (nam->m_len != sizeof (*sin)) 381 return (EINVAL); 382 if (sin->sin_family != AF_INET) 383 return (EAFNOSUPPORT); 384 if (sin->sin_port == 0) 385 return (EADDRNOTAVAIL); 386 if (TAILQ_FIRST(&in_ifaddrhead) != 0) { 387 /* 388 * If the destination address is INADDR_ANY, 389 * use any local address (likely loopback). 390 * If the supplied address is INADDR_BROADCAST, 391 * use the broadcast address of an interface 392 * which supports broadcast. (loopback does not) 393 */ 394 395 if (in_nullhost(sin->sin_addr)) { 396 sin->sin_addr = 397 TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr; 398 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) { 399 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 400 if (ia->ia_ifp->if_flags & IFF_BROADCAST) { 401 sin->sin_addr = 402 ia->ia_broadaddr.sin_addr; 403 break; 404 } 405 } 406 } 407 } 408 /* 409 * If we haven't bound which network number to use as ours, 410 * we will use the number of the outgoing interface. 411 * This depends on having done a routing lookup, which 412 * we will probably have to do anyway, so we might 413 * as well do it now. On the other hand if we are 414 * sending to multiple destinations we may have already 415 * done the lookup, so see if we can use the route 416 * from before. In any case, we only 417 * chose a port number once, even if sending to multiple 418 * destinations. 419 */ 420 if (in_nullhost(inp->inp_laddr)) { 421 int error; 422 ifaddr = in_selectsrc(sin, &inp->inp_route, 423 inp->inp_socket->so_options, inp->inp_moptions, &error); 424 if (ifaddr == NULL) { 425 if (error == 0) 426 error = EADDRNOTAVAIL; 427 return error; 428 } 429 INADDR_TO_IA(ifaddr->sin_addr, ia); 430 if (ia == NULL) 431 return (EADDRNOTAVAIL); 432 } 433 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port, 434 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr, 435 inp->inp_lport) != 0) 436 return (EADDRINUSE); 437 if (in_nullhost(inp->inp_laddr)) { 438 if (inp->inp_lport == 0) { 439 error = in_pcbbind(inp, (struct mbuf *)0, 440 (struct proc *)0); 441 /* 442 * This used to ignore the return value 443 * completely, but we need to check for 444 * ephemeral port shortage. 445 * XXX Should we check for other errors, too? 446 */ 447 if (error == EAGAIN) 448 return (error); 449 } 450 inp->inp_laddr = ifaddr->sin_addr; 451 } 452 inp->inp_faddr = sin->sin_addr; 453 inp->inp_fport = sin->sin_port; 454 in_pcbstate(inp, INP_CONNECTED); 455 #if defined(IPSEC) || defined(FAST_IPSEC) 456 if (inp->inp_socket->so_type == SOCK_STREAM) 457 ipsec_pcbconn(inp->inp_sp); 458 #endif 459 return (0); 460 } 461 462 void 463 in_pcbdisconnect(v) 464 void *v; 465 { 466 struct inpcb *inp = v; 467 468 if (inp->inp_af != AF_INET) 469 return; 470 471 inp->inp_faddr = zeroin_addr; 472 inp->inp_fport = 0; 473 in_pcbstate(inp, INP_BOUND); 474 #if defined(IPSEC) || defined(FAST_IPSEC) 475 ipsec_pcbdisconn(inp->inp_sp); 476 #endif 477 if (inp->inp_socket->so_state & SS_NOFDREF) 478 in_pcbdetach(inp); 479 } 480 481 void 482 in_pcbdetach(v) 483 void *v; 484 { 485 struct inpcb *inp = v; 486 struct socket *so = inp->inp_socket; 487 int s; 488 489 if (inp->inp_af != AF_INET) 490 return; 491 492 #if defined(IPSEC) || defined(FAST_IPSEC) 493 ipsec4_delete_pcbpolicy(inp); 494 #endif /*IPSEC*/ 495 so->so_pcb = 0; 496 sofree(so); 497 if (inp->inp_options) 498 (void)m_free(inp->inp_options); 499 if (inp->inp_route.ro_rt) 500 rtfree(inp->inp_route.ro_rt); 501 ip_freemoptions(inp->inp_moptions); 502 s = splnet(); 503 in_pcbstate(inp, INP_ATTACHED); 504 LIST_REMOVE(&inp->inp_head, inph_lhash); 505 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head, 506 inph_queue); 507 splx(s); 508 pool_put(&inpcb_pool, inp); 509 } 510 511 void 512 in_setsockaddr(inp, nam) 513 struct inpcb *inp; 514 struct mbuf *nam; 515 { 516 struct sockaddr_in *sin; 517 518 if (inp->inp_af != AF_INET) 519 return; 520 521 nam->m_len = sizeof (*sin); 522 sin = mtod(nam, struct sockaddr_in *); 523 bzero((caddr_t)sin, sizeof (*sin)); 524 sin->sin_family = AF_INET; 525 sin->sin_len = sizeof(*sin); 526 sin->sin_port = inp->inp_lport; 527 sin->sin_addr = inp->inp_laddr; 528 } 529 530 void 531 in_setpeeraddr(inp, nam) 532 struct inpcb *inp; 533 struct mbuf *nam; 534 { 535 struct sockaddr_in *sin; 536 537 if (inp->inp_af != AF_INET) 538 return; 539 540 nam->m_len = sizeof (*sin); 541 sin = mtod(nam, struct sockaddr_in *); 542 bzero((caddr_t)sin, sizeof (*sin)); 543 sin->sin_family = AF_INET; 544 sin->sin_len = sizeof(*sin); 545 sin->sin_port = inp->inp_fport; 546 sin->sin_addr = inp->inp_faddr; 547 } 548 549 /* 550 * Pass some notification to all connections of a protocol 551 * associated with address dst. The local address and/or port numbers 552 * may be specified to limit the search. The "usual action" will be 553 * taken, depending on the ctlinput cmd. The caller must filter any 554 * cmds that are uninteresting (e.g., no error in the map). 555 * Call the protocol specific routine (if any) to report 556 * any errors for each matching socket. 557 * 558 * Must be called at splsoftnet. 559 */ 560 int 561 in_pcbnotify(table, faddr, fport_arg, laddr, lport_arg, errno, notify) 562 struct inpcbtable *table; 563 struct in_addr faddr, laddr; 564 u_int fport_arg, lport_arg; 565 int errno; 566 void (*notify)(struct inpcb *, int); 567 { 568 struct inpcbhead *head; 569 struct inpcb *inp, *ninp; 570 u_int16_t fport = fport_arg, lport = lport_arg; 571 int nmatch; 572 573 if (in_nullhost(faddr) || notify == 0) 574 return (0); 575 576 nmatch = 0; 577 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 578 for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) { 579 ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash); 580 if (inp->inp_af != AF_INET) 581 continue; 582 if (in_hosteq(inp->inp_faddr, faddr) && 583 inp->inp_fport == fport && 584 inp->inp_lport == lport && 585 in_hosteq(inp->inp_laddr, laddr)) { 586 (*notify)(inp, errno); 587 nmatch++; 588 } 589 } 590 return (nmatch); 591 } 592 593 void 594 in_pcbnotifyall(table, faddr, errno, notify) 595 struct inpcbtable *table; 596 struct in_addr faddr; 597 int errno; 598 void (*notify)(struct inpcb *, int); 599 { 600 struct inpcb *inp, *ninp; 601 602 if (in_nullhost(faddr) || notify == 0) 603 return; 604 605 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 606 inp != (void *)&table->inpt_queue; 607 inp = ninp) { 608 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 609 if (inp->inp_af != AF_INET) 610 continue; 611 if (in_hosteq(inp->inp_faddr, faddr)) 612 (*notify)(inp, errno); 613 } 614 } 615 616 void 617 in_pcbpurgeif0(table, ifp) 618 struct inpcbtable *table; 619 struct ifnet *ifp; 620 { 621 struct inpcb *inp, *ninp; 622 struct ip_moptions *imo; 623 int i, gap; 624 625 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 626 inp != (void *)&table->inpt_queue; 627 inp = ninp) { 628 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 629 if (inp->inp_af != AF_INET) 630 continue; 631 imo = inp->inp_moptions; 632 if (imo != NULL) { 633 /* 634 * Unselect the outgoing interface if it is being 635 * detached. 636 */ 637 if (imo->imo_multicast_ifp == ifp) 638 imo->imo_multicast_ifp = NULL; 639 640 /* 641 * Drop multicast group membership if we joined 642 * through the interface being detached. 643 */ 644 for (i = 0, gap = 0; i < imo->imo_num_memberships; 645 i++) { 646 if (imo->imo_membership[i]->inm_ifp == ifp) { 647 in_delmulti(imo->imo_membership[i]); 648 gap++; 649 } else if (gap != 0) 650 imo->imo_membership[i - gap] = 651 imo->imo_membership[i]; 652 } 653 imo->imo_num_memberships -= gap; 654 } 655 } 656 } 657 658 void 659 in_pcbpurgeif(table, ifp) 660 struct inpcbtable *table; 661 struct ifnet *ifp; 662 { 663 struct inpcb *inp, *ninp; 664 665 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 666 inp != (void *)&table->inpt_queue; 667 inp = ninp) { 668 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 669 if (inp->inp_af != AF_INET) 670 continue; 671 if (inp->inp_route.ro_rt != NULL && 672 inp->inp_route.ro_rt->rt_ifp == ifp) 673 in_rtchange(inp, 0); 674 } 675 } 676 677 /* 678 * Check for alternatives when higher level complains 679 * about service problems. For now, invalidate cached 680 * routing information. If the route was created dynamically 681 * (by a redirect), time to try a default gateway again. 682 */ 683 void 684 in_losing(inp) 685 struct inpcb *inp; 686 { 687 struct rtentry *rt; 688 struct rt_addrinfo info; 689 690 if (inp->inp_af != AF_INET) 691 return; 692 693 if ((rt = inp->inp_route.ro_rt)) { 694 inp->inp_route.ro_rt = 0; 695 bzero((caddr_t)&info, sizeof(info)); 696 info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst; 697 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 698 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 699 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 700 if (rt->rt_flags & RTF_DYNAMIC) 701 (void) rtrequest(RTM_DELETE, rt_key(rt), 702 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 703 (struct rtentry **)0); 704 else 705 /* 706 * A new route can be allocated 707 * the next time output is attempted. 708 */ 709 rtfree(rt); 710 } 711 } 712 713 /* 714 * After a routing change, flush old routing 715 * and allocate a (hopefully) better one. 716 */ 717 void 718 in_rtchange(inp, errno) 719 struct inpcb *inp; 720 int errno; 721 { 722 723 if (inp->inp_af != AF_INET) 724 return; 725 726 if (inp->inp_route.ro_rt) { 727 rtfree(inp->inp_route.ro_rt); 728 inp->inp_route.ro_rt = 0; 729 /* 730 * A new route can be allocated the next time 731 * output is attempted. 732 */ 733 } 734 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 735 } 736 737 struct inpcb * 738 in_pcblookup_port(table, laddr, lport_arg, lookup_wildcard) 739 struct inpcbtable *table; 740 struct in_addr laddr; 741 u_int lport_arg; 742 int lookup_wildcard; 743 { 744 struct inpcbhead *head; 745 struct inpcb_hdr *inph; 746 struct inpcb *inp, *match = 0; 747 int matchwild = 3, wildcard; 748 u_int16_t lport = lport_arg; 749 750 head = INPCBHASH_PORT(table, lport); 751 LIST_FOREACH(inph, head, inph_lhash) { 752 inp = (struct inpcb *)inph; 753 if (inp->inp_af != AF_INET) 754 continue; 755 756 if (inp->inp_lport != lport) 757 continue; 758 wildcard = 0; 759 if (!in_nullhost(inp->inp_faddr)) 760 wildcard++; 761 if (in_nullhost(inp->inp_laddr)) { 762 if (!in_nullhost(laddr)) 763 wildcard++; 764 } else { 765 if (in_nullhost(laddr)) 766 wildcard++; 767 else { 768 if (!in_hosteq(inp->inp_laddr, laddr)) 769 continue; 770 } 771 } 772 if (wildcard && !lookup_wildcard) 773 continue; 774 if (wildcard < matchwild) { 775 match = inp; 776 matchwild = wildcard; 777 if (matchwild == 0) 778 break; 779 } 780 } 781 return (match); 782 } 783 784 #ifdef DIAGNOSTIC 785 int in_pcbnotifymiss = 0; 786 #endif 787 788 struct inpcb * 789 in_pcblookup_connect(table, faddr, fport_arg, laddr, lport_arg) 790 struct inpcbtable *table; 791 struct in_addr faddr, laddr; 792 u_int fport_arg, lport_arg; 793 { 794 struct inpcbhead *head; 795 struct inpcb_hdr *inph; 796 struct inpcb *inp; 797 u_int16_t fport = fport_arg, lport = lport_arg; 798 799 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 800 LIST_FOREACH(inph, head, inph_hash) { 801 inp = (struct inpcb *)inph; 802 if (inp->inp_af != AF_INET) 803 continue; 804 805 if (in_hosteq(inp->inp_faddr, faddr) && 806 inp->inp_fport == fport && 807 inp->inp_lport == lport && 808 in_hosteq(inp->inp_laddr, laddr)) 809 goto out; 810 } 811 #ifdef DIAGNOSTIC 812 if (in_pcbnotifymiss) { 813 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 814 ntohl(faddr.s_addr), ntohs(fport), 815 ntohl(laddr.s_addr), ntohs(lport)); 816 } 817 #endif 818 return (0); 819 820 out: 821 /* Move this PCB to the head of hash chain. */ 822 inph = &inp->inp_head; 823 if (inph != LIST_FIRST(head)) { 824 LIST_REMOVE(inph, inph_hash); 825 LIST_INSERT_HEAD(head, inph, inph_hash); 826 } 827 return (inp); 828 } 829 830 struct inpcb * 831 in_pcblookup_bind(table, laddr, lport_arg) 832 struct inpcbtable *table; 833 struct in_addr laddr; 834 u_int lport_arg; 835 { 836 struct inpcbhead *head; 837 struct inpcb_hdr *inph; 838 struct inpcb *inp; 839 u_int16_t lport = lport_arg; 840 841 head = INPCBHASH_BIND(table, laddr, lport); 842 LIST_FOREACH(inph, head, inph_hash) { 843 inp = (struct inpcb *)inph; 844 if (inp->inp_af != AF_INET) 845 continue; 846 847 if (inp->inp_lport == lport && 848 in_hosteq(inp->inp_laddr, laddr)) 849 goto out; 850 } 851 head = INPCBHASH_BIND(table, zeroin_addr, lport); 852 LIST_FOREACH(inph, head, inph_hash) { 853 inp = (struct inpcb *)inph; 854 if (inp->inp_af != AF_INET) 855 continue; 856 857 if (inp->inp_lport == lport && 858 in_hosteq(inp->inp_laddr, zeroin_addr)) 859 goto out; 860 } 861 #ifdef DIAGNOSTIC 862 if (in_pcbnotifymiss) { 863 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 864 ntohl(laddr.s_addr), ntohs(lport)); 865 } 866 #endif 867 return (0); 868 869 out: 870 /* Move this PCB to the head of hash chain. */ 871 inph = &inp->inp_head; 872 if (inph != LIST_FIRST(head)) { 873 LIST_REMOVE(inph, inph_hash); 874 LIST_INSERT_HEAD(head, inph, inph_hash); 875 } 876 return (inp); 877 } 878 879 void 880 in_pcbstate(inp, state) 881 struct inpcb *inp; 882 int state; 883 { 884 885 if (inp->inp_af != AF_INET) 886 return; 887 888 if (inp->inp_state > INP_ATTACHED) 889 LIST_REMOVE(&inp->inp_head, inph_hash); 890 891 switch (state) { 892 case INP_BOUND: 893 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 894 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 895 inph_hash); 896 break; 897 case INP_CONNECTED: 898 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 899 inp->inp_faddr, inp->inp_fport, 900 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 901 inph_hash); 902 break; 903 } 904 905 inp->inp_state = state; 906 } 907 908 struct rtentry * 909 in_pcbrtentry(inp) 910 struct inpcb *inp; 911 { 912 struct route *ro; 913 914 if (inp->inp_af != AF_INET) 915 return (NULL); 916 917 ro = &inp->inp_route; 918 919 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 920 !in_hosteq(satosin(&ro->ro_dst)->sin_addr, inp->inp_faddr))) { 921 RTFREE(ro->ro_rt); 922 ro->ro_rt = (struct rtentry *)NULL; 923 } 924 if (ro->ro_rt == (struct rtentry *)NULL && 925 !in_nullhost(inp->inp_faddr)) { 926 bzero(&ro->ro_dst, sizeof(struct sockaddr_in)); 927 ro->ro_dst.sa_family = AF_INET; 928 ro->ro_dst.sa_len = sizeof(ro->ro_dst); 929 satosin(&ro->ro_dst)->sin_addr = inp->inp_faddr; 930 rtalloc(ro); 931 } 932 return (ro->ro_rt); 933 } 934 935 struct sockaddr_in * 936 in_selectsrc(sin, ro, soopts, mopts, errorp) 937 struct sockaddr_in *sin; 938 struct route *ro; 939 int soopts; 940 struct ip_moptions *mopts; 941 int *errorp; 942 { 943 struct in_ifaddr *ia; 944 945 ia = (struct in_ifaddr *)0; 946 /* 947 * If route is known or can be allocated now, 948 * our src addr is taken from the i/f, else punt. 949 * Note that we should check the address family of the cached 950 * destination, in case of sharing the cache with IPv6. 951 */ 952 if (ro->ro_rt && 953 (ro->ro_dst.sa_family != AF_INET || 954 !in_hosteq(satosin(&ro->ro_dst)->sin_addr, sin->sin_addr) || 955 soopts & SO_DONTROUTE)) { 956 RTFREE(ro->ro_rt); 957 ro->ro_rt = (struct rtentry *)0; 958 } 959 if ((soopts & SO_DONTROUTE) == 0 && /*XXX*/ 960 (ro->ro_rt == (struct rtentry *)0 || 961 ro->ro_rt->rt_ifp == (struct ifnet *)0)) { 962 /* No route yet, so try to acquire one */ 963 bzero(&ro->ro_dst, sizeof(struct sockaddr_in)); 964 ro->ro_dst.sa_family = AF_INET; 965 ro->ro_dst.sa_len = sizeof(struct sockaddr_in); 966 satosin(&ro->ro_dst)->sin_addr = sin->sin_addr; 967 rtalloc(ro); 968 } 969 /* 970 * If we found a route, use the address 971 * corresponding to the outgoing interface 972 * unless it is the loopback (in case a route 973 * to our address on another net goes to loopback). 974 * 975 * XXX Is this still true? Do we care? 976 */ 977 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) 978 ia = ifatoia(ro->ro_rt->rt_ifa); 979 if (ia == NULL) { 980 u_int16_t fport = sin->sin_port; 981 982 sin->sin_port = 0; 983 ia = ifatoia(ifa_ifwithladdr(sintosa(sin))); 984 sin->sin_port = fport; 985 if (ia == 0) { 986 /* Find 1st non-loopback AF_INET address */ 987 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 988 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK)) 989 break; 990 } 991 } 992 if (ia == NULL) { 993 *errorp = EADDRNOTAVAIL; 994 return NULL; 995 } 996 } 997 /* 998 * If the destination address is multicast and an outgoing 999 * interface has been set as a multicast option, use the 1000 * address of that interface as our source address. 1001 */ 1002 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) { 1003 struct ip_moptions *imo; 1004 struct ifnet *ifp; 1005 1006 imo = mopts; 1007 if (imo->imo_multicast_ifp != NULL) { 1008 ifp = imo->imo_multicast_ifp; 1009 IFP_TO_IA(ifp, ia); /* XXX */ 1010 if (ia == 0) { 1011 *errorp = EADDRNOTAVAIL; 1012 return NULL; 1013 } 1014 } 1015 } 1016 return satosin(&ia->ia_addr); 1017 } 1018
Indexes created Wed Oct 15 16:09:53 GMT 2025