in_pcb.c revision 1.161
1 /* $NetBSD: in_pcb.c,v 1.161 2015/05/24 15:43:45 rtr 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, 2011 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Coyote Point Systems, Inc. 38 * This code is derived from software contributed to The NetBSD Foundation 39 * by Public Access Networks Corporation ("Panix"). It was developed under 40 * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 51 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 52 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 53 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 54 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 55 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 56 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 57 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 58 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 59 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 60 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 61 * POSSIBILITY OF SUCH DAMAGE. 62 */ 63 64 /* 65 * Copyright (c) 1982, 1986, 1991, 1993, 1995 66 * The Regents of the University of California. All rights reserved. 67 * 68 * Redistribution and use in source and binary forms, with or without 69 * modification, are permitted provided that the following conditions 70 * are met: 71 * 1. Redistributions of source code must retain the above copyright 72 * notice, this list of conditions and the following disclaimer. 73 * 2. Redistributions in binary form must reproduce the above copyright 74 * notice, this list of conditions and the following disclaimer in the 75 * documentation and/or other materials provided with the distribution. 76 * 3. Neither the name of the University nor the names of its contributors 77 * may be used to endorse or promote products derived from this software 78 * without specific prior written permission. 79 * 80 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 81 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 82 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 83 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 84 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 85 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 86 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 87 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 88 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 89 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 90 * SUCH DAMAGE. 91 * 92 * @(#)in_pcb.c 8.4 (Berkeley) 5/24/95 93 */ 94 95 #include <sys/cdefs.h> 96 __KERNEL_RCSID(0, "$NetBSD: in_pcb.c,v 1.161 2015/05/24 15:43:45 rtr Exp $"); 97 98 #include "opt_inet.h" 99 #include "opt_ipsec.h" 100 101 #include <sys/param.h> 102 #include <sys/systm.h> 103 #include <sys/mbuf.h> 104 #include <sys/protosw.h> 105 #include <sys/socket.h> 106 #include <sys/socketvar.h> 107 #include <sys/ioctl.h> 108 #include <sys/errno.h> 109 #include <sys/time.h> 110 #include <sys/once.h> 111 #include <sys/pool.h> 112 #include <sys/proc.h> 113 #include <sys/kauth.h> 114 #include <sys/uidinfo.h> 115 #include <sys/domain.h> 116 117 #include <net/if.h> 118 #include <net/route.h> 119 120 #include <netinet/in.h> 121 #include <netinet/in_systm.h> 122 #include <netinet/ip.h> 123 #include <netinet/in_pcb.h> 124 #include <netinet/in_var.h> 125 #include <netinet/ip_var.h> 126 #include <netinet/portalgo.h> 127 128 #ifdef INET6 129 #include <netinet/ip6.h> 130 #include <netinet6/ip6_var.h> 131 #include <netinet6/in6_pcb.h> 132 #endif 133 134 #ifdef IPSEC 135 #include <netipsec/ipsec.h> 136 #include <netipsec/key.h> 137 #endif /* IPSEC */ 138 139 #include <netinet/tcp_vtw.h> 140 141 struct in_addr zeroin_addr; 142 143 #define INPCBHASH_PORT(table, lport) \ 144 &(table)->inpt_porthashtbl[ntohs(lport) & (table)->inpt_porthash] 145 #define INPCBHASH_BIND(table, laddr, lport) \ 146 &(table)->inpt_bindhashtbl[ \ 147 ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash] 148 #define INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \ 149 &(table)->inpt_connecthashtbl[ \ 150 ((ntohl((faddr).s_addr) + ntohs(fport)) + \ 151 (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash] 152 153 int anonportmin = IPPORT_ANONMIN; 154 int anonportmax = IPPORT_ANONMAX; 155 int lowportmin = IPPORT_RESERVEDMIN; 156 int lowportmax = IPPORT_RESERVEDMAX; 157 158 static struct pool inpcb_pool; 159 160 static int 161 inpcb_poolinit(void) 162 { 163 164 pool_init(&inpcb_pool, sizeof(struct inpcb), 0, 0, 0, "inpcbpl", NULL, 165 IPL_NET); 166 return 0; 167 } 168 169 void 170 in_pcbinit(struct inpcbtable *table, int bindhashsize, int connecthashsize) 171 { 172 static ONCE_DECL(control); 173 174 TAILQ_INIT(&table->inpt_queue); 175 table->inpt_porthashtbl = hashinit(bindhashsize, HASH_LIST, true, 176 &table->inpt_porthash); 177 table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, true, 178 &table->inpt_bindhash); 179 table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST, true, 180 &table->inpt_connecthash); 181 table->inpt_lastlow = IPPORT_RESERVEDMAX; 182 table->inpt_lastport = (u_int16_t)anonportmax; 183 184 RUN_ONCE(&control, inpcb_poolinit); 185 } 186 187 int 188 in_pcballoc(struct socket *so, void *v) 189 { 190 struct inpcbtable *table = v; 191 struct inpcb *inp; 192 int s; 193 194 s = splnet(); 195 inp = pool_get(&inpcb_pool, PR_NOWAIT); 196 splx(s); 197 if (inp == NULL) 198 return (ENOBUFS); 199 memset(inp, 0, sizeof(*inp)); 200 inp->inp_af = AF_INET; 201 inp->inp_table = table; 202 inp->inp_socket = so; 203 inp->inp_errormtu = -1; 204 inp->inp_portalgo = PORTALGO_DEFAULT; 205 inp->inp_bindportonsend = false; 206 #if defined(IPSEC) 207 if (ipsec_enabled) { 208 int error = ipsec_init_pcbpolicy(so, &inp->inp_sp); 209 if (error != 0) { 210 s = splnet(); 211 pool_put(&inpcb_pool, inp); 212 splx(s); 213 return error; 214 } 215 } 216 #endif 217 so->so_pcb = inp; 218 s = splnet(); 219 TAILQ_INSERT_HEAD(&table->inpt_queue, &inp->inp_head, inph_queue); 220 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 221 inph_lhash); 222 in_pcbstate(inp, INP_ATTACHED); 223 splx(s); 224 return (0); 225 } 226 227 static int 228 in_pcbsetport(struct sockaddr_in *sin, struct inpcb *inp, kauth_cred_t cred) 229 { 230 struct inpcbtable *table = inp->inp_table; 231 struct socket *so = inp->inp_socket; 232 u_int16_t *lastport; 233 u_int16_t lport = 0; 234 enum kauth_network_req req; 235 int error; 236 237 if (inp->inp_flags & INP_LOWPORT) { 238 #ifndef IPNOPRIVPORTS 239 req = KAUTH_REQ_NETWORK_BIND_PRIVPORT; 240 #else 241 req = KAUTH_REQ_NETWORK_BIND_PORT; 242 #endif 243 244 lastport = &table->inpt_lastlow; 245 } else { 246 req = KAUTH_REQ_NETWORK_BIND_PORT; 247 248 lastport = &table->inpt_lastport; 249 } 250 251 /* XXX-kauth: KAUTH_REQ_NETWORK_BIND_AUTOASSIGN_{,PRIV}PORT */ 252 error = kauth_authorize_network(cred, KAUTH_NETWORK_BIND, req, so, sin, 253 NULL); 254 if (error) 255 return (EACCES); 256 257 /* 258 * Use RFC6056 randomized port selection 259 */ 260 error = portalgo_randport(&lport, &inp->inp_head, cred); 261 if (error) 262 return error; 263 264 inp->inp_flags |= INP_ANONPORT; 265 *lastport = lport; 266 lport = htons(lport); 267 inp->inp_lport = lport; 268 in_pcbstate(inp, INP_BOUND); 269 270 return (0); 271 } 272 273 static int 274 in_pcbbind_addr(struct inpcb *inp, struct sockaddr_in *sin, kauth_cred_t cred) 275 { 276 if (sin->sin_family != AF_INET) 277 return (EAFNOSUPPORT); 278 279 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 280 /* Always succeed; port reuse handled in in_pcbbind_port(). */ 281 } else if (!in_nullhost(sin->sin_addr)) { 282 struct in_ifaddr *ia = NULL; 283 284 INADDR_TO_IA(sin->sin_addr, ia); 285 /* check for broadcast addresses */ 286 if (ia == NULL) 287 ia = ifatoia(ifa_ifwithaddr(sintosa(sin))); 288 if (ia == NULL) 289 return (EADDRNOTAVAIL); 290 if (ia->ia4_flags & (IN_IFF_NOTREADY | IN_IFF_DETACHED)) 291 return (EADDRNOTAVAIL); 292 } 293 294 inp->inp_laddr = sin->sin_addr; 295 296 return (0); 297 } 298 299 static int 300 in_pcbbind_port(struct inpcb *inp, struct sockaddr_in *sin, kauth_cred_t cred) 301 { 302 struct inpcbtable *table = inp->inp_table; 303 struct socket *so = inp->inp_socket; 304 int reuseport = (so->so_options & SO_REUSEPORT); 305 int wild = 0, error; 306 307 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 308 /* 309 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 310 * allow complete duplication of binding if 311 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 312 * and a multicast address is bound on both 313 * new and duplicated sockets. 314 */ 315 if (so->so_options & (SO_REUSEADDR | SO_REUSEPORT)) 316 reuseport = SO_REUSEADDR|SO_REUSEPORT; 317 } 318 319 if (sin->sin_port == 0) { 320 error = in_pcbsetport(sin, inp, cred); 321 if (error) 322 return (error); 323 } else { 324 struct inpcb *t; 325 vestigial_inpcb_t vestige; 326 #ifdef INET6 327 struct in6pcb *t6; 328 struct in6_addr mapped; 329 #endif 330 enum kauth_network_req req; 331 332 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 333 wild = 1; 334 335 #ifndef IPNOPRIVPORTS 336 if (ntohs(sin->sin_port) < IPPORT_RESERVED) 337 req = KAUTH_REQ_NETWORK_BIND_PRIVPORT; 338 else 339 #endif /* !IPNOPRIVPORTS */ 340 req = KAUTH_REQ_NETWORK_BIND_PORT; 341 342 error = kauth_authorize_network(cred, KAUTH_NETWORK_BIND, req, 343 so, sin, NULL); 344 if (error) 345 return (EACCES); 346 347 #ifdef INET6 348 memset(&mapped, 0, sizeof(mapped)); 349 mapped.s6_addr16[5] = 0xffff; 350 memcpy(&mapped.s6_addr32[3], &sin->sin_addr, 351 sizeof(mapped.s6_addr32[3])); 352 t6 = in6_pcblookup_port(table, &mapped, sin->sin_port, wild, &vestige); 353 if (t6 && (reuseport & t6->in6p_socket->so_options) == 0) 354 return (EADDRINUSE); 355 if (!t6 && vestige.valid) { 356 if (!!reuseport != !!vestige.reuse_port) { 357 return EADDRINUSE; 358 } 359 } 360 #endif 361 362 /* XXX-kauth */ 363 if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) { 364 t = in_pcblookup_port(table, sin->sin_addr, sin->sin_port, 1, &vestige); 365 /* 366 * XXX: investigate ramifications of loosening this 367 * restriction so that as long as both ports have 368 * SO_REUSEPORT allow the bind 369 */ 370 if (t && 371 (!in_nullhost(sin->sin_addr) || 372 !in_nullhost(t->inp_laddr) || 373 (t->inp_socket->so_options & SO_REUSEPORT) == 0) 374 && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) { 375 return (EADDRINUSE); 376 } 377 if (!t && vestige.valid) { 378 if ((!in_nullhost(sin->sin_addr) 379 || !in_nullhost(vestige.laddr.v4) 380 || !vestige.reuse_port) 381 && so->so_uidinfo->ui_uid != vestige.uid) { 382 return EADDRINUSE; 383 } 384 } 385 } 386 t = in_pcblookup_port(table, sin->sin_addr, sin->sin_port, wild, &vestige); 387 if (t && (reuseport & t->inp_socket->so_options) == 0) 388 return (EADDRINUSE); 389 if (!t 390 && vestige.valid 391 && !(reuseport && vestige.reuse_port)) 392 return EADDRINUSE; 393 394 inp->inp_lport = sin->sin_port; 395 in_pcbstate(inp, INP_BOUND); 396 } 397 398 LIST_REMOVE(&inp->inp_head, inph_lhash); 399 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 400 inph_lhash); 401 402 return (0); 403 } 404 405 int 406 in_pcbbind(void *v, struct sockaddr_in *sin, struct lwp *l) 407 { 408 struct inpcb *inp = v; 409 struct sockaddr_in lsin; 410 int error; 411 412 if (inp->inp_af != AF_INET) 413 return (EINVAL); 414 415 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 416 return (EADDRNOTAVAIL); 417 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 418 return (EINVAL); 419 420 if (NULL != sin) { 421 if (sin->sin_len != sizeof(*sin)) 422 return (EINVAL); 423 } else { 424 lsin = *((const struct sockaddr_in *) 425 inp->inp_socket->so_proto->pr_domain->dom_sa_any); 426 sin = &lsin; 427 } 428 429 /* Bind address. */ 430 error = in_pcbbind_addr(inp, sin, l->l_cred); 431 if (error) 432 return (error); 433 434 /* Bind port. */ 435 error = in_pcbbind_port(inp, sin, l->l_cred); 436 if (error) { 437 inp->inp_laddr.s_addr = INADDR_ANY; 438 439 return (error); 440 } 441 442 return (0); 443 } 444 445 /* 446 * Connect from a socket to a specified address. 447 * Both address and port must be specified in argument sin. 448 * If don't have a local address for this socket yet, 449 * then pick one. 450 */ 451 int 452 in_pcbconnect(void *v, struct sockaddr_in *sin, struct lwp *l) 453 { 454 struct inpcb *inp = v; 455 struct in_ifaddr *ia = NULL; 456 struct sockaddr_in *ifaddr = NULL; 457 vestigial_inpcb_t vestige; 458 int error; 459 460 if (inp->inp_af != AF_INET) 461 return (EINVAL); 462 463 if (sin->sin_len != sizeof (*sin)) 464 return (EINVAL); 465 if (sin->sin_family != AF_INET) 466 return (EAFNOSUPPORT); 467 if (sin->sin_port == 0) 468 return (EADDRNOTAVAIL); 469 470 if (IN_MULTICAST(sin->sin_addr.s_addr) && 471 inp->inp_socket->so_type == SOCK_STREAM) 472 return EADDRNOTAVAIL; 473 474 if (TAILQ_FIRST(&in_ifaddrhead) != 0) { 475 /* 476 * If the destination address is INADDR_ANY, 477 * use any local address (likely loopback). 478 * If the supplied address is INADDR_BROADCAST, 479 * use the broadcast address of an interface 480 * which supports broadcast. (loopback does not) 481 */ 482 483 if (in_nullhost(sin->sin_addr)) { 484 sin->sin_addr = 485 TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr; 486 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) { 487 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 488 if (ia->ia_ifp->if_flags & IFF_BROADCAST) { 489 sin->sin_addr = 490 ia->ia_broadaddr.sin_addr; 491 break; 492 } 493 } 494 } 495 } 496 /* 497 * If we haven't bound which network number to use as ours, 498 * we will use the number of the outgoing interface. 499 * This depends on having done a routing lookup, which 500 * we will probably have to do anyway, so we might 501 * as well do it now. On the other hand if we are 502 * sending to multiple destinations we may have already 503 * done the lookup, so see if we can use the route 504 * from before. In any case, we only 505 * chose a port number once, even if sending to multiple 506 * destinations. 507 */ 508 if (in_nullhost(inp->inp_laddr)) { 509 int xerror; 510 ifaddr = in_selectsrc(sin, &inp->inp_route, 511 inp->inp_socket->so_options, inp->inp_moptions, &xerror); 512 if (ifaddr == NULL) { 513 if (xerror == 0) 514 xerror = EADDRNOTAVAIL; 515 return xerror; 516 } 517 INADDR_TO_IA(ifaddr->sin_addr, ia); 518 if (ia == NULL) 519 return (EADDRNOTAVAIL); 520 } 521 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port, 522 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr, 523 inp->inp_lport, &vestige) != 0 524 || vestige.valid) 525 return (EADDRINUSE); 526 if (in_nullhost(inp->inp_laddr)) { 527 if (inp->inp_lport == 0) { 528 error = in_pcbbind(inp, NULL, l); 529 /* 530 * This used to ignore the return value 531 * completely, but we need to check for 532 * ephemeral port shortage. 533 * And attempts to request low ports if not root. 534 */ 535 if (error != 0) 536 return (error); 537 } 538 inp->inp_laddr = ifaddr->sin_addr; 539 } 540 inp->inp_faddr = sin->sin_addr; 541 inp->inp_fport = sin->sin_port; 542 543 /* Late bind, if needed */ 544 if (inp->inp_bindportonsend) { 545 struct sockaddr_in lsin = *((const struct sockaddr_in *) 546 inp->inp_socket->so_proto->pr_domain->dom_sa_any); 547 lsin.sin_addr = inp->inp_laddr; 548 lsin.sin_port = 0; 549 550 if ((error = in_pcbbind_port(inp, &lsin, l->l_cred)) != 0) 551 return error; 552 } 553 554 in_pcbstate(inp, INP_CONNECTED); 555 #if defined(IPSEC) 556 if (ipsec_enabled && inp->inp_socket->so_type == SOCK_STREAM) 557 ipsec_pcbconn(inp->inp_sp); 558 #endif 559 return (0); 560 } 561 562 void 563 in_pcbdisconnect(void *v) 564 { 565 struct inpcb *inp = v; 566 567 if (inp->inp_af != AF_INET) 568 return; 569 570 inp->inp_faddr = zeroin_addr; 571 inp->inp_fport = 0; 572 in_pcbstate(inp, INP_BOUND); 573 #if defined(IPSEC) 574 if (ipsec_enabled) 575 ipsec_pcbdisconn(inp->inp_sp); 576 #endif 577 if (inp->inp_socket->so_state & SS_NOFDREF) 578 in_pcbdetach(inp); 579 } 580 581 void 582 in_pcbdetach(void *v) 583 { 584 struct inpcb *inp = v; 585 struct socket *so = inp->inp_socket; 586 int s; 587 588 if (inp->inp_af != AF_INET) 589 return; 590 591 #if defined(IPSEC) 592 if (ipsec_enabled) 593 ipsec4_delete_pcbpolicy(inp); 594 #endif 595 so->so_pcb = NULL; 596 597 s = splnet(); 598 in_pcbstate(inp, INP_ATTACHED); 599 LIST_REMOVE(&inp->inp_head, inph_lhash); 600 TAILQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head, inph_queue); 601 splx(s); 602 603 if (inp->inp_options) { 604 m_free(inp->inp_options); 605 } 606 rtcache_free(&inp->inp_route); 607 ip_freemoptions(inp->inp_moptions); 608 sofree(so); /* drops the socket's lock */ 609 610 pool_put(&inpcb_pool, inp); 611 mutex_enter(softnet_lock); /* reacquire the softnet_lock */ 612 } 613 614 void 615 in_setsockaddr(struct inpcb *inp, struct sockaddr_in *sin) 616 { 617 618 if (inp->inp_af != AF_INET) 619 return; 620 621 sockaddr_in_init(sin, &inp->inp_laddr, inp->inp_lport); 622 } 623 624 void 625 in_setpeeraddr(struct inpcb *inp, struct sockaddr_in *sin) 626 { 627 628 if (inp->inp_af != AF_INET) 629 return; 630 631 sockaddr_in_init(sin, &inp->inp_faddr, inp->inp_fport); 632 } 633 634 /* 635 * Pass some notification to all connections of a protocol 636 * associated with address dst. The local address and/or port numbers 637 * may be specified to limit the search. The "usual action" will be 638 * taken, depending on the ctlinput cmd. The caller must filter any 639 * cmds that are uninteresting (e.g., no error in the map). 640 * Call the protocol specific routine (if any) to report 641 * any errors for each matching socket. 642 * 643 * Must be called at splsoftnet. 644 */ 645 int 646 in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg, 647 struct in_addr laddr, u_int lport_arg, int errno, 648 void (*notify)(struct inpcb *, int)) 649 { 650 struct inpcbhead *head; 651 struct inpcb *inp, *ninp; 652 u_int16_t fport = fport_arg, lport = lport_arg; 653 int nmatch; 654 655 if (in_nullhost(faddr) || notify == 0) 656 return (0); 657 658 nmatch = 0; 659 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 660 for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) { 661 ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash); 662 if (inp->inp_af != AF_INET) 663 continue; 664 if (in_hosteq(inp->inp_faddr, faddr) && 665 inp->inp_fport == fport && 666 inp->inp_lport == lport && 667 in_hosteq(inp->inp_laddr, laddr)) { 668 (*notify)(inp, errno); 669 nmatch++; 670 } 671 } 672 return (nmatch); 673 } 674 675 void 676 in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno, 677 void (*notify)(struct inpcb *, int)) 678 { 679 struct inpcb_hdr *inph, *ninph; 680 681 if (in_nullhost(faddr) || notify == 0) 682 return; 683 684 TAILQ_FOREACH_SAFE(inph, &table->inpt_queue, inph_queue, ninph) { 685 struct inpcb *inp = (struct inpcb *)inph; 686 if (inp->inp_af != AF_INET) 687 continue; 688 if (in_hosteq(inp->inp_faddr, faddr)) 689 (*notify)(inp, errno); 690 } 691 } 692 693 void 694 in_purgeifmcast(struct ip_moptions *imo, struct ifnet *ifp) 695 { 696 int i, gap; 697 698 if (imo == NULL) 699 return; 700 701 /* 702 * Unselect the outgoing interface if it is being 703 * detached. 704 */ 705 if (imo->imo_multicast_ifp == ifp) 706 imo->imo_multicast_ifp = NULL; 707 708 /* 709 * Drop multicast group membership if we joined 710 * through the interface being detached. 711 */ 712 for (i = 0, gap = 0; i < imo->imo_num_memberships; i++) { 713 if (imo->imo_membership[i]->inm_ifp == ifp) { 714 in_delmulti(imo->imo_membership[i]); 715 gap++; 716 } else if (gap != 0) 717 imo->imo_membership[i - gap] = imo->imo_membership[i]; 718 } 719 imo->imo_num_memberships -= gap; 720 } 721 722 void 723 in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp) 724 { 725 struct inpcb_hdr *inph, *ninph; 726 727 TAILQ_FOREACH_SAFE(inph, &table->inpt_queue, inph_queue, ninph) { 728 struct inpcb *inp = (struct inpcb *)inph; 729 if (inp->inp_af != AF_INET) 730 continue; 731 in_purgeifmcast(inp->inp_moptions, ifp); 732 } 733 } 734 735 void 736 in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp) 737 { 738 struct rtentry *rt; 739 struct inpcb_hdr *inph, *ninph; 740 741 TAILQ_FOREACH_SAFE(inph, &table->inpt_queue, inph_queue, ninph) { 742 struct inpcb *inp = (struct inpcb *)inph; 743 if (inp->inp_af != AF_INET) 744 continue; 745 if ((rt = rtcache_validate(&inp->inp_route)) != NULL && 746 rt->rt_ifp == ifp) 747 in_rtchange(inp, 0); 748 } 749 } 750 751 /* 752 * Check for alternatives when higher level complains 753 * about service problems. For now, invalidate cached 754 * routing information. If the route was created dynamically 755 * (by a redirect), time to try a default gateway again. 756 */ 757 void 758 in_losing(struct inpcb *inp) 759 { 760 struct rtentry *rt; 761 struct rt_addrinfo info; 762 763 if (inp->inp_af != AF_INET) 764 return; 765 766 if ((rt = rtcache_validate(&inp->inp_route)) == NULL) 767 return; 768 769 memset(&info, 0, sizeof(info)); 770 info.rti_info[RTAX_DST] = rtcache_getdst(&inp->inp_route); 771 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 772 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 773 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 774 if (rt->rt_flags & RTF_DYNAMIC) 775 (void) rtrequest(RTM_DELETE, rt_getkey(rt), 776 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 777 NULL); 778 /* 779 * A new route can be allocated 780 * the next time output is attempted. 781 */ 782 rtcache_free(&inp->inp_route); 783 } 784 785 /* 786 * After a routing change, flush old routing. A new route can be 787 * allocated the next time output is attempted. 788 */ 789 void 790 in_rtchange(struct inpcb *inp, int errno) 791 { 792 793 if (inp->inp_af != AF_INET) 794 return; 795 796 rtcache_free(&inp->inp_route); 797 798 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 799 } 800 801 struct inpcb * 802 in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr, 803 u_int lport_arg, int lookup_wildcard, vestigial_inpcb_t *vp) 804 { 805 struct inpcbhead *head; 806 struct inpcb_hdr *inph; 807 struct inpcb *match = NULL; 808 int matchwild = 3; 809 int wildcard; 810 u_int16_t lport = lport_arg; 811 812 if (vp) 813 vp->valid = 0; 814 815 head = INPCBHASH_PORT(table, lport); 816 LIST_FOREACH(inph, head, inph_lhash) { 817 struct inpcb * const inp = (struct inpcb *)inph; 818 819 if (inp->inp_af != AF_INET) 820 continue; 821 if (inp->inp_lport != lport) 822 continue; 823 /* 824 * check if inp's faddr and laddr match with ours. 825 * our faddr is considered null. 826 * count the number of wildcard matches. (0 - 2) 827 * 828 * null null match 829 * A null wildcard match 830 * null B wildcard match 831 * A B non match 832 * A A match 833 */ 834 wildcard = 0; 835 if (!in_nullhost(inp->inp_faddr)) 836 wildcard++; 837 if (in_nullhost(inp->inp_laddr)) { 838 if (!in_nullhost(laddr)) 839 wildcard++; 840 } else { 841 if (in_nullhost(laddr)) 842 wildcard++; 843 else { 844 if (!in_hosteq(inp->inp_laddr, laddr)) 845 continue; 846 } 847 } 848 if (wildcard && !lookup_wildcard) 849 continue; 850 /* 851 * prefer an address with less wildcards. 852 */ 853 if (wildcard < matchwild) { 854 match = inp; 855 matchwild = wildcard; 856 if (matchwild == 0) 857 break; 858 } 859 } 860 if (match && matchwild == 0) 861 return match; 862 863 if (vp && table->vestige) { 864 void *state = (*table->vestige->init_ports4)(laddr, lport_arg, lookup_wildcard); 865 vestigial_inpcb_t better; 866 867 while (table->vestige 868 && (*table->vestige->next_port4)(state, vp)) { 869 870 if (vp->lport != lport) 871 continue; 872 wildcard = 0; 873 if (!in_nullhost(vp->faddr.v4)) 874 wildcard++; 875 if (in_nullhost(vp->laddr.v4)) { 876 if (!in_nullhost(laddr)) 877 wildcard++; 878 } else { 879 if (in_nullhost(laddr)) 880 wildcard++; 881 else { 882 if (!in_hosteq(vp->laddr.v4, laddr)) 883 continue; 884 } 885 } 886 if (wildcard && !lookup_wildcard) 887 continue; 888 if (wildcard < matchwild) { 889 better = *vp; 890 match = (void*)&better; 891 892 matchwild = wildcard; 893 if (matchwild == 0) 894 break; 895 } 896 } 897 898 if (match) { 899 if (match != (void*)&better) 900 return match; 901 else { 902 *vp = better; 903 return 0; 904 } 905 } 906 } 907 908 return (match); 909 } 910 911 #ifdef DIAGNOSTIC 912 int in_pcbnotifymiss = 0; 913 #endif 914 915 struct inpcb * 916 in_pcblookup_connect(struct inpcbtable *table, 917 struct in_addr faddr, u_int fport_arg, 918 struct in_addr laddr, u_int lport_arg, 919 vestigial_inpcb_t *vp) 920 { 921 struct inpcbhead *head; 922 struct inpcb_hdr *inph; 923 struct inpcb *inp; 924 u_int16_t fport = fport_arg, lport = lport_arg; 925 926 if (vp) 927 vp->valid = 0; 928 929 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 930 LIST_FOREACH(inph, head, inph_hash) { 931 inp = (struct inpcb *)inph; 932 if (inp->inp_af != AF_INET) 933 continue; 934 935 if (in_hosteq(inp->inp_faddr, faddr) && 936 inp->inp_fport == fport && 937 inp->inp_lport == lport && 938 in_hosteq(inp->inp_laddr, laddr)) 939 goto out; 940 } 941 if (vp && table->vestige) { 942 if ((*table->vestige->lookup4)(faddr, fport_arg, 943 laddr, lport_arg, vp)) 944 return 0; 945 } 946 947 #ifdef DIAGNOSTIC 948 if (in_pcbnotifymiss) { 949 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 950 ntohl(faddr.s_addr), ntohs(fport), 951 ntohl(laddr.s_addr), ntohs(lport)); 952 } 953 #endif 954 return (0); 955 956 out: 957 /* Move this PCB to the head of hash chain. */ 958 inph = &inp->inp_head; 959 if (inph != LIST_FIRST(head)) { 960 LIST_REMOVE(inph, inph_hash); 961 LIST_INSERT_HEAD(head, inph, inph_hash); 962 } 963 return (inp); 964 } 965 966 struct inpcb * 967 in_pcblookup_bind(struct inpcbtable *table, 968 struct in_addr laddr, u_int lport_arg) 969 { 970 struct inpcbhead *head; 971 struct inpcb_hdr *inph; 972 struct inpcb *inp; 973 u_int16_t lport = lport_arg; 974 975 head = INPCBHASH_BIND(table, laddr, lport); 976 LIST_FOREACH(inph, head, inph_hash) { 977 inp = (struct inpcb *)inph; 978 if (inp->inp_af != AF_INET) 979 continue; 980 981 if (inp->inp_lport == lport && 982 in_hosteq(inp->inp_laddr, laddr)) 983 goto out; 984 } 985 head = INPCBHASH_BIND(table, zeroin_addr, lport); 986 LIST_FOREACH(inph, head, inph_hash) { 987 inp = (struct inpcb *)inph; 988 if (inp->inp_af != AF_INET) 989 continue; 990 991 if (inp->inp_lport == lport && 992 in_hosteq(inp->inp_laddr, zeroin_addr)) 993 goto out; 994 } 995 #ifdef DIAGNOSTIC 996 if (in_pcbnotifymiss) { 997 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 998 ntohl(laddr.s_addr), ntohs(lport)); 999 } 1000 #endif 1001 return (0); 1002 1003 out: 1004 /* Move this PCB to the head of hash chain. */ 1005 inph = &inp->inp_head; 1006 if (inph != LIST_FIRST(head)) { 1007 LIST_REMOVE(inph, inph_hash); 1008 LIST_INSERT_HEAD(head, inph, inph_hash); 1009 } 1010 return (inp); 1011 } 1012 1013 void 1014 in_pcbstate(struct inpcb *inp, int state) 1015 { 1016 1017 if (inp->inp_af != AF_INET) 1018 return; 1019 1020 if (inp->inp_state > INP_ATTACHED) 1021 LIST_REMOVE(&inp->inp_head, inph_hash); 1022 1023 switch (state) { 1024 case INP_BOUND: 1025 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 1026 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 1027 inph_hash); 1028 break; 1029 case INP_CONNECTED: 1030 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 1031 inp->inp_faddr, inp->inp_fport, 1032 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 1033 inph_hash); 1034 break; 1035 } 1036 1037 inp->inp_state = state; 1038 } 1039 1040 struct rtentry * 1041 in_pcbrtentry(struct inpcb *inp) 1042 { 1043 struct route *ro; 1044 union { 1045 struct sockaddr dst; 1046 struct sockaddr_in dst4; 1047 } u; 1048 1049 if (inp->inp_af != AF_INET) 1050 return (NULL); 1051 1052 ro = &inp->inp_route; 1053 1054 sockaddr_in_init(&u.dst4, &inp->inp_faddr, 0); 1055 return rtcache_lookup(ro, &u.dst); 1056 } 1057
Indexes created Wed Oct 15 16:09:53 GMT 2025