in_pcb.c revision 1.117
1 /* $NetBSD: in_pcb.c,v 1.117 2007/05/02 20:40:24 dyoung 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 1.117 2007/05/02 20:40:24 dyoung 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 IPL_NET); 163 164 void 165 in_pcbinit(struct inpcbtable *table, int bindhashsize, int 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(struct socket *so, void *v) 181 { 182 struct inpcbtable *table = v; 183 struct inpcb *inp; 184 int s; 185 #if defined(IPSEC) || defined(FAST_IPSEC) 186 int error; 187 #endif 188 189 s = splnet(); 190 inp = pool_get(&inpcb_pool, PR_NOWAIT); 191 splx(s); 192 if (inp == NULL) 193 return (ENOBUFS); 194 bzero((void *)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 s = splnet(); 203 pool_put(&inpcb_pool, inp); 204 splx(s); 205 return error; 206 } 207 #endif 208 so->so_pcb = inp; 209 s = splnet(); 210 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, &inp->inp_head, 211 inph_queue); 212 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 213 inph_lhash); 214 in_pcbstate(inp, INP_ATTACHED); 215 splx(s); 216 return (0); 217 } 218 219 int 220 in_pcbbind(void *v, struct mbuf *nam, struct lwp *l) 221 { 222 struct in_ifaddr *ia = NULL; 223 struct inpcb *inp = v; 224 struct socket *so = inp->inp_socket; 225 struct inpcbtable *table = inp->inp_table; 226 struct sockaddr_in *sin = NULL; /* XXXGCC */ 227 u_int16_t lport = 0; 228 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 229 230 if (inp->inp_af != AF_INET) 231 return (EINVAL); 232 233 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 234 return (EADDRNOTAVAIL); 235 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 236 return (EINVAL); 237 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 238 wild = 1; 239 if (nam == 0) 240 goto noname; 241 sin = mtod(nam, struct sockaddr_in *); 242 if (nam->m_len != sizeof (*sin)) 243 return (EINVAL); 244 if (sin->sin_family != AF_INET) 245 return (EAFNOSUPPORT); 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 INADDR_TO_IA(sin->sin_addr, ia); 260 /* check for broadcast addresses */ 261 if (ia == NULL) 262 ia = ifatoia(ifa_ifwithaddr(sintosa(sin))); 263 if (ia == NULL) 264 return (EADDRNOTAVAIL); 265 } 266 if (lport) { 267 struct inpcb *t; 268 #ifdef INET6 269 struct in6pcb *t6; 270 struct in6_addr mapped; 271 #endif 272 #ifndef IPNOPRIVPORTS 273 /* GROSS */ 274 if (ntohs(lport) < IPPORT_RESERVED && 275 (l == 0 || kauth_authorize_network(l->l_cred, 276 KAUTH_NETWORK_BIND, 277 KAUTH_REQ_NETWORK_BIND_PRIVPORT, so, sin, 278 NULL))) 279 return (EACCES); 280 #endif 281 #ifdef INET6 282 memset(&mapped, 0, sizeof(mapped)); 283 mapped.s6_addr16[5] = 0xffff; 284 memcpy(&mapped.s6_addr32[3], &sin->sin_addr, 285 sizeof(mapped.s6_addr32[3])); 286 t6 = in6_pcblookup_port(table, &mapped, lport, wild); 287 if (t6 && (reuseport & t6->in6p_socket->so_options) == 0) 288 return (EADDRINUSE); 289 #endif 290 if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) { 291 t = in_pcblookup_port(table, sin->sin_addr, lport, 1); 292 /* 293 * XXX: investigate ramifications of loosening this 294 * restriction so that as long as both ports have 295 * SO_REUSEPORT allow the bind 296 */ 297 if (t && 298 (!in_nullhost(sin->sin_addr) || 299 !in_nullhost(t->inp_laddr) || 300 (t->inp_socket->so_options & SO_REUSEPORT) == 0) 301 && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) { 302 return (EADDRINUSE); 303 } 304 } 305 t = in_pcblookup_port(table, sin->sin_addr, lport, wild); 306 if (t && (reuseport & t->inp_socket->so_options) == 0) 307 return (EADDRINUSE); 308 } 309 inp->inp_laddr = sin->sin_addr; 310 311 noname: 312 if (lport == 0) { 313 int cnt; 314 u_int16_t mymin, mymax; 315 u_int16_t *lastport; 316 317 if (inp->inp_flags & INP_LOWPORT) { 318 #ifndef IPNOPRIVPORTS 319 if (l == 0 || kauth_authorize_network(l->l_cred, 320 KAUTH_NETWORK_BIND, 321 KAUTH_REQ_NETWORK_BIND_PRIVPORT, so, 322 sin, NULL)) 323 return (EACCES); 324 #endif 325 mymin = lowportmin; 326 mymax = lowportmax; 327 lastport = &table->inpt_lastlow; 328 } else { 329 mymin = anonportmin; 330 mymax = anonportmax; 331 lastport = &table->inpt_lastport; 332 } 333 if (mymin > mymax) { /* sanity check */ 334 u_int16_t swp; 335 336 swp = mymin; 337 mymin = mymax; 338 mymax = swp; 339 } 340 341 lport = *lastport - 1; 342 for (cnt = mymax - mymin + 1; cnt; cnt--, lport--) { 343 if (lport < mymin || lport > mymax) 344 lport = mymax; 345 if (!in_pcblookup_port(table, inp->inp_laddr, 346 htons(lport), 1)) 347 goto found; 348 } 349 if (!in_nullhost(inp->inp_laddr)) 350 inp->inp_laddr.s_addr = INADDR_ANY; 351 return (EAGAIN); 352 found: 353 inp->inp_flags |= INP_ANONPORT; 354 *lastport = lport; 355 lport = htons(lport); 356 } 357 inp->inp_lport = lport; 358 LIST_REMOVE(&inp->inp_head, inph_lhash); 359 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 360 inph_lhash); 361 in_pcbstate(inp, INP_BOUND); 362 return (0); 363 } 364 365 /* 366 * Connect from a socket to a specified address. 367 * Both address and port must be specified in argument sin. 368 * If don't have a local address for this socket yet, 369 * then pick one. 370 */ 371 int 372 in_pcbconnect(void *v, struct mbuf *nam, struct lwp *l) 373 { 374 struct inpcb *inp = v; 375 struct in_ifaddr *ia = NULL; 376 struct sockaddr_in *ifaddr = NULL; 377 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); 378 int error; 379 380 if (inp->inp_af != AF_INET) 381 return (EINVAL); 382 383 if (nam->m_len != sizeof (*sin)) 384 return (EINVAL); 385 if (sin->sin_family != AF_INET) 386 return (EAFNOSUPPORT); 387 if (sin->sin_port == 0) 388 return (EADDRNOTAVAIL); 389 if (TAILQ_FIRST(&in_ifaddrhead) != 0) { 390 /* 391 * If the destination address is INADDR_ANY, 392 * use any local address (likely loopback). 393 * If the supplied address is INADDR_BROADCAST, 394 * use the broadcast address of an interface 395 * which supports broadcast. (loopback does not) 396 */ 397 398 if (in_nullhost(sin->sin_addr)) { 399 sin->sin_addr = 400 TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr; 401 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) { 402 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 403 if (ia->ia_ifp->if_flags & IFF_BROADCAST) { 404 sin->sin_addr = 405 ia->ia_broadaddr.sin_addr; 406 break; 407 } 408 } 409 } 410 } 411 /* 412 * If we haven't bound which network number to use as ours, 413 * we will use the number of the outgoing interface. 414 * This depends on having done a routing lookup, which 415 * we will probably have to do anyway, so we might 416 * as well do it now. On the other hand if we are 417 * sending to multiple destinations we may have already 418 * done the lookup, so see if we can use the route 419 * from before. In any case, we only 420 * chose a port number once, even if sending to multiple 421 * destinations. 422 */ 423 if (in_nullhost(inp->inp_laddr)) { 424 int xerror; 425 ifaddr = in_selectsrc(sin, &inp->inp_route, 426 inp->inp_socket->so_options, inp->inp_moptions, &xerror); 427 if (ifaddr == NULL) { 428 if (xerror == 0) 429 xerror = EADDRNOTAVAIL; 430 return xerror; 431 } 432 INADDR_TO_IA(ifaddr->sin_addr, ia); 433 if (ia == NULL) 434 return (EADDRNOTAVAIL); 435 } 436 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port, 437 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr, 438 inp->inp_lport) != 0) 439 return (EADDRINUSE); 440 if (in_nullhost(inp->inp_laddr)) { 441 if (inp->inp_lport == 0) { 442 error = in_pcbbind(inp, NULL, l); 443 /* 444 * This used to ignore the return value 445 * completely, but we need to check for 446 * ephemeral port shortage. 447 * And attempts to request low ports if not root. 448 */ 449 if (error != 0) 450 return (error); 451 } 452 inp->inp_laddr = ifaddr->sin_addr; 453 } 454 inp->inp_faddr = sin->sin_addr; 455 inp->inp_fport = sin->sin_port; 456 in_pcbstate(inp, INP_CONNECTED); 457 #if defined(IPSEC) || defined(FAST_IPSEC) 458 if (inp->inp_socket->so_type == SOCK_STREAM) 459 ipsec_pcbconn(inp->inp_sp); 460 #endif 461 return (0); 462 } 463 464 void 465 in_pcbdisconnect(void *v) 466 { 467 struct inpcb *inp = v; 468 469 if (inp->inp_af != AF_INET) 470 return; 471 472 inp->inp_faddr = zeroin_addr; 473 inp->inp_fport = 0; 474 in_pcbstate(inp, INP_BOUND); 475 #if defined(IPSEC) || defined(FAST_IPSEC) 476 ipsec_pcbdisconn(inp->inp_sp); 477 #endif 478 if (inp->inp_socket->so_state & SS_NOFDREF) 479 in_pcbdetach(inp); 480 } 481 482 void 483 in_pcbdetach(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 rtcache_free(&inp->inp_route); 500 ip_freemoptions(inp->inp_moptions); 501 s = splnet(); 502 in_pcbstate(inp, INP_ATTACHED); 503 LIST_REMOVE(&inp->inp_head, inph_lhash); 504 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head, 505 inph_queue); 506 pool_put(&inpcb_pool, inp); 507 splx(s); 508 } 509 510 void 511 in_setsockaddr(struct inpcb *inp, struct mbuf *nam) 512 { 513 struct sockaddr_in *sin; 514 515 if (inp->inp_af != AF_INET) 516 return; 517 518 nam->m_len = sizeof (*sin); 519 sin = mtod(nam, struct sockaddr_in *); 520 bzero((void *)sin, sizeof (*sin)); 521 sin->sin_family = AF_INET; 522 sin->sin_len = sizeof(*sin); 523 sin->sin_port = inp->inp_lport; 524 sin->sin_addr = inp->inp_laddr; 525 } 526 527 void 528 in_setpeeraddr(struct inpcb *inp, struct mbuf *nam) 529 { 530 struct sockaddr_in *sin; 531 532 if (inp->inp_af != AF_INET) 533 return; 534 535 nam->m_len = sizeof (*sin); 536 sin = mtod(nam, struct sockaddr_in *); 537 bzero((void *)sin, sizeof (*sin)); 538 sin->sin_family = AF_INET; 539 sin->sin_len = sizeof(*sin); 540 sin->sin_port = inp->inp_fport; 541 sin->sin_addr = inp->inp_faddr; 542 } 543 544 /* 545 * Pass some notification to all connections of a protocol 546 * associated with address dst. The local address and/or port numbers 547 * may be specified to limit the search. The "usual action" will be 548 * taken, depending on the ctlinput cmd. The caller must filter any 549 * cmds that are uninteresting (e.g., no error in the map). 550 * Call the protocol specific routine (if any) to report 551 * any errors for each matching socket. 552 * 553 * Must be called at splsoftnet. 554 */ 555 int 556 in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg, 557 struct in_addr laddr, u_int lport_arg, int errno, 558 void (*notify)(struct inpcb *, int)) 559 { 560 struct inpcbhead *head; 561 struct inpcb *inp, *ninp; 562 u_int16_t fport = fport_arg, lport = lport_arg; 563 int nmatch; 564 565 if (in_nullhost(faddr) || notify == 0) 566 return (0); 567 568 nmatch = 0; 569 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 570 for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) { 571 ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash); 572 if (inp->inp_af != AF_INET) 573 continue; 574 if (in_hosteq(inp->inp_faddr, faddr) && 575 inp->inp_fport == fport && 576 inp->inp_lport == lport && 577 in_hosteq(inp->inp_laddr, laddr)) { 578 (*notify)(inp, errno); 579 nmatch++; 580 } 581 } 582 return (nmatch); 583 } 584 585 void 586 in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno, 587 void (*notify)(struct inpcb *, int)) 588 { 589 struct inpcb *inp, *ninp; 590 591 if (in_nullhost(faddr) || notify == 0) 592 return; 593 594 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 595 inp != (void *)&table->inpt_queue; 596 inp = ninp) { 597 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 598 if (inp->inp_af != AF_INET) 599 continue; 600 if (in_hosteq(inp->inp_faddr, faddr)) 601 (*notify)(inp, errno); 602 } 603 } 604 605 void 606 in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp) 607 { 608 struct inpcb *inp, *ninp; 609 struct ip_moptions *imo; 610 int i, gap; 611 612 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 613 inp != (void *)&table->inpt_queue; 614 inp = ninp) { 615 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 616 if (inp->inp_af != AF_INET) 617 continue; 618 imo = inp->inp_moptions; 619 if (imo != NULL) { 620 /* 621 * Unselect the outgoing interface if it is being 622 * detached. 623 */ 624 if (imo->imo_multicast_ifp == ifp) 625 imo->imo_multicast_ifp = NULL; 626 627 /* 628 * Drop multicast group membership if we joined 629 * through the interface being detached. 630 */ 631 for (i = 0, gap = 0; i < imo->imo_num_memberships; 632 i++) { 633 if (imo->imo_membership[i]->inm_ifp == ifp) { 634 in_delmulti(imo->imo_membership[i]); 635 gap++; 636 } else if (gap != 0) 637 imo->imo_membership[i - gap] = 638 imo->imo_membership[i]; 639 } 640 imo->imo_num_memberships -= gap; 641 } 642 } 643 } 644 645 void 646 in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp) 647 { 648 struct inpcb *inp, *ninp; 649 650 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 651 inp != (void *)&table->inpt_queue; 652 inp = ninp) { 653 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 654 if (inp->inp_af != AF_INET) 655 continue; 656 if (inp->inp_route.ro_rt != NULL && 657 inp->inp_route.ro_rt->rt_ifp == ifp) 658 in_rtchange(inp, 0); 659 } 660 } 661 662 /* 663 * Check for alternatives when higher level complains 664 * about service problems. For now, invalidate cached 665 * routing information. If the route was created dynamically 666 * (by a redirect), time to try a default gateway again. 667 */ 668 void 669 in_losing(struct inpcb *inp) 670 { 671 struct rtentry *rt; 672 struct rt_addrinfo info; 673 674 if (inp->inp_af != AF_INET) 675 return; 676 677 if ((rt = inp->inp_route.ro_rt) != NULL) { 678 memset(&info, 0, sizeof(info)); 679 info.rti_info[RTAX_DST] = rtcache_getdst(&inp->inp_route); 680 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 681 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 682 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 683 if (rt->rt_flags & RTF_DYNAMIC) 684 (void) rtrequest(RTM_DELETE, rt_key(rt), 685 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 686 NULL); 687 /* 688 * A new route can be allocated 689 * the next time output is attempted. 690 */ 691 rtcache_free(&inp->inp_route); 692 } 693 } 694 695 /* 696 * After a routing change, flush old routing. A new route can be 697 * allocated the next time output is attempted. 698 */ 699 void 700 in_rtchange(struct inpcb *inp, int errno) 701 { 702 703 if (inp->inp_af != AF_INET) 704 return; 705 706 rtcache_free(&inp->inp_route); 707 708 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 709 } 710 711 struct inpcb * 712 in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr, 713 u_int lport_arg, int lookup_wildcard) 714 { 715 struct inpcbhead *head; 716 struct inpcb_hdr *inph; 717 struct inpcb *inp, *match = 0; 718 int matchwild = 3, wildcard; 719 u_int16_t lport = lport_arg; 720 721 head = INPCBHASH_PORT(table, lport); 722 LIST_FOREACH(inph, head, inph_lhash) { 723 inp = (struct inpcb *)inph; 724 if (inp->inp_af != AF_INET) 725 continue; 726 727 if (inp->inp_lport != lport) 728 continue; 729 wildcard = 0; 730 if (!in_nullhost(inp->inp_faddr)) 731 wildcard++; 732 if (in_nullhost(inp->inp_laddr)) { 733 if (!in_nullhost(laddr)) 734 wildcard++; 735 } else { 736 if (in_nullhost(laddr)) 737 wildcard++; 738 else { 739 if (!in_hosteq(inp->inp_laddr, laddr)) 740 continue; 741 } 742 } 743 if (wildcard && !lookup_wildcard) 744 continue; 745 if (wildcard < matchwild) { 746 match = inp; 747 matchwild = wildcard; 748 if (matchwild == 0) 749 break; 750 } 751 } 752 return (match); 753 } 754 755 #ifdef DIAGNOSTIC 756 int in_pcbnotifymiss = 0; 757 #endif 758 759 struct inpcb * 760 in_pcblookup_connect(struct inpcbtable *table, 761 struct in_addr faddr, u_int fport_arg, 762 struct in_addr laddr, u_int lport_arg) 763 { 764 struct inpcbhead *head; 765 struct inpcb_hdr *inph; 766 struct inpcb *inp; 767 u_int16_t fport = fport_arg, lport = lport_arg; 768 769 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 770 LIST_FOREACH(inph, head, inph_hash) { 771 inp = (struct inpcb *)inph; 772 if (inp->inp_af != AF_INET) 773 continue; 774 775 if (in_hosteq(inp->inp_faddr, faddr) && 776 inp->inp_fport == fport && 777 inp->inp_lport == lport && 778 in_hosteq(inp->inp_laddr, laddr)) 779 goto out; 780 } 781 #ifdef DIAGNOSTIC 782 if (in_pcbnotifymiss) { 783 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 784 ntohl(faddr.s_addr), ntohs(fport), 785 ntohl(laddr.s_addr), ntohs(lport)); 786 } 787 #endif 788 return (0); 789 790 out: 791 /* Move this PCB to the head of hash chain. */ 792 inph = &inp->inp_head; 793 if (inph != LIST_FIRST(head)) { 794 LIST_REMOVE(inph, inph_hash); 795 LIST_INSERT_HEAD(head, inph, inph_hash); 796 } 797 return (inp); 798 } 799 800 struct inpcb * 801 in_pcblookup_bind(struct inpcbtable *table, 802 struct in_addr laddr, u_int lport_arg) 803 { 804 struct inpcbhead *head; 805 struct inpcb_hdr *inph; 806 struct inpcb *inp; 807 u_int16_t lport = lport_arg; 808 809 head = INPCBHASH_BIND(table, laddr, lport); 810 LIST_FOREACH(inph, head, inph_hash) { 811 inp = (struct inpcb *)inph; 812 if (inp->inp_af != AF_INET) 813 continue; 814 815 if (inp->inp_lport == lport && 816 in_hosteq(inp->inp_laddr, laddr)) 817 goto out; 818 } 819 head = INPCBHASH_BIND(table, zeroin_addr, lport); 820 LIST_FOREACH(inph, head, inph_hash) { 821 inp = (struct inpcb *)inph; 822 if (inp->inp_af != AF_INET) 823 continue; 824 825 if (inp->inp_lport == lport && 826 in_hosteq(inp->inp_laddr, zeroin_addr)) 827 goto out; 828 } 829 #ifdef DIAGNOSTIC 830 if (in_pcbnotifymiss) { 831 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 832 ntohl(laddr.s_addr), ntohs(lport)); 833 } 834 #endif 835 return (0); 836 837 out: 838 /* Move this PCB to the head of hash chain. */ 839 inph = &inp->inp_head; 840 if (inph != LIST_FIRST(head)) { 841 LIST_REMOVE(inph, inph_hash); 842 LIST_INSERT_HEAD(head, inph, inph_hash); 843 } 844 return (inp); 845 } 846 847 void 848 in_pcbstate(struct inpcb *inp, int state) 849 { 850 851 if (inp->inp_af != AF_INET) 852 return; 853 854 if (inp->inp_state > INP_ATTACHED) 855 LIST_REMOVE(&inp->inp_head, inph_hash); 856 857 switch (state) { 858 case INP_BOUND: 859 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 860 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 861 inph_hash); 862 break; 863 case INP_CONNECTED: 864 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 865 inp->inp_faddr, inp->inp_fport, 866 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 867 inph_hash); 868 break; 869 } 870 871 inp->inp_state = state; 872 } 873 874 struct rtentry * 875 in_pcbrtentry(struct inpcb *inp) 876 { 877 struct route *ro; 878 union { 879 struct sockaddr dst; 880 struct sockaddr_in dst4; 881 } u; 882 883 if (inp->inp_af != AF_INET) 884 return (NULL); 885 886 ro = &inp->inp_route; 887 888 sockaddr_in_init(&u.dst4, &inp->inp_faddr, 0); 889 return rtcache_lookup(ro, &u.dst); 890 } 891 892 struct sockaddr_in * 893 in_selectsrc(struct sockaddr_in *sin, struct route *ro, 894 int soopts, struct ip_moptions *mopts, int *errorp) 895 { 896 struct in_ifaddr *ia = NULL; 897 898 /* 899 * If route is known or can be allocated now, take the 900 * source address from the interface. Otherwise, punt. 901 */ 902 if ((soopts & SO_DONTROUTE) != 0) 903 rtcache_free(ro); 904 else { 905 union { 906 struct sockaddr dst; 907 struct sockaddr_in dst4; 908 } u; 909 910 sockaddr_in_init(&u.dst4, &sin->sin_addr, 0); 911 (void)rtcache_lookup(ro, &u.dst); 912 } 913 /* 914 * If we found a route, use the address 915 * corresponding to the outgoing interface 916 * unless it is the loopback (in case a route 917 * to our address on another net goes to loopback). 918 * 919 * XXX Is this still true? Do we care? 920 */ 921 if (ro->ro_rt != NULL && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) 922 ia = ifatoia(ro->ro_rt->rt_ifa); 923 if (ia == NULL) { 924 u_int16_t fport = sin->sin_port; 925 926 sin->sin_port = 0; 927 ia = ifatoia(ifa_ifwithladdr(sintosa(sin))); 928 sin->sin_port = fport; 929 if (ia == NULL) { 930 /* Find 1st non-loopback AF_INET address */ 931 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 932 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK)) 933 break; 934 } 935 } 936 if (ia == NULL) { 937 *errorp = EADDRNOTAVAIL; 938 return NULL; 939 } 940 } 941 /* 942 * If the destination address is multicast and an outgoing 943 * interface has been set as a multicast option, use the 944 * address of that interface as our source address. 945 */ 946 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) { 947 struct ip_moptions *imo; 948 struct ifnet *ifp; 949 950 imo = mopts; 951 if (imo->imo_multicast_ifp != NULL) { 952 ifp = imo->imo_multicast_ifp; 953 IFP_TO_IA(ifp, ia); /* XXX */ 954 if (ia == 0) { 955 *errorp = EADDRNOTAVAIL; 956 return NULL; 957 } 958 } 959 } 960 if (ia->ia_ifa.ifa_getifa != NULL) { 961 ia = ifatoia((*ia->ia_ifa.ifa_getifa)(&ia->ia_ifa, 962 sintosa(sin))); 963 } 964 #ifdef GETIFA_DEBUG 965 else 966 printf("%s: missing ifa_getifa\n", __func__); 967 #endif 968 return satosin(&ia->ia_addr); 969 } 970
Indexes created Mon Sep 22 13:09:51 GMT 2025