1/* $NetBSD: in_pcb.c,v 1.139 2011/09/24 17:18:17 christos 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 1.139 2011/09/24 17:18:17 christos 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/malloc.h> 104#include <sys/mbuf.h> 105#include <sys/protosw.h> 106#include <sys/socket.h> 107#include <sys/socketvar.h> 108#include <sys/ioctl.h> 109#include <sys/errno.h> 110#include <sys/time.h> 111#include <sys/once.h> 112#include <sys/pool.h> 113#include <sys/proc.h> 114#include <sys/kauth.h> 115#include <sys/uidinfo.h> 116#include <sys/domain.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#include <netinet/rfc6056.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 KAME_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#include <netinet/tcp_vtw.h> 144 145struct in_addr zeroin_addr; 146 147#define INPCBHASH_PORT(table, lport) \ 148 &(table)->inpt_porthashtbl[ntohs(lport) & (table)->inpt_porthash] 149#define INPCBHASH_BIND(table, laddr, lport) \ 150 &(table)->inpt_bindhashtbl[ \ 151 ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash] 152#define INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \ 153 &(table)->inpt_connecthashtbl[ \ 154 ((ntohl((faddr).s_addr) + ntohs(fport)) + \ 155 (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash] 156 157int anonportmin = IPPORT_ANONMIN; 158int anonportmax = IPPORT_ANONMAX; 159int lowportmin = IPPORT_RESERVEDMIN; 160int lowportmax = IPPORT_RESERVEDMAX; 161 162static struct pool inpcb_pool; 163 164static int 165inpcb_poolinit(void) 166{ 167 168 pool_init(&inpcb_pool, sizeof(struct inpcb), 0, 0, 0, "inpcbpl", NULL, 169 IPL_NET); 170 return 0; 171} 172 173void 174in_pcbinit(struct inpcbtable *table, int bindhashsize, int connecthashsize) 175{ 176 static ONCE_DECL(control); 177 178 CIRCLEQ_INIT(&table->inpt_queue); 179 table->inpt_porthashtbl = hashinit(bindhashsize, HASH_LIST, true, 180 &table->inpt_porthash); 181 table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, true, 182 &table->inpt_bindhash); 183 table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST, true, 184 &table->inpt_connecthash); 185 table->inpt_lastlow = IPPORT_RESERVEDMAX; 186 table->inpt_lastport = (u_int16_t)anonportmax; 187 188 RUN_ONCE(&control, inpcb_poolinit); 189} 190 191int 192in_pcballoc(struct socket *so, void *v) 193{ 194 struct inpcbtable *table = v; 195 struct inpcb *inp; 196 int s; 197#if defined(KAME_IPSEC) || defined(FAST_IPSEC) 198 int error; 199#endif 200 201 s = splnet(); 202 inp = pool_get(&inpcb_pool, PR_NOWAIT); 203 splx(s); 204 if (inp == NULL) 205 return (ENOBUFS); 206 memset(inp, 0, sizeof(*inp)); 207 inp->inp_af = AF_INET; 208 inp->inp_table = table; 209 inp->inp_socket = so; 210 inp->inp_errormtu = -1; 211 inp->inp_rfc6056algo = RFC6056_ALGO_DEFAULT; 212 inp->inp_bindportonsend = false; 213#if defined(KAME_IPSEC) || defined(FAST_IPSEC) 214 error = ipsec_init_pcbpolicy(so, &inp->inp_sp); 215 if (error != 0) { 216 s = splnet(); 217 pool_put(&inpcb_pool, inp); 218 splx(s); 219 return error; 220 } 221#endif 222 so->so_pcb = inp; 223 s = splnet(); 224 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, &inp->inp_head, 225 inph_queue); 226 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 227 inph_lhash); 228 in_pcbstate(inp, INP_ATTACHED); 229 splx(s); 230 return (0); 231} 232 233static int 234in_pcbsetport(struct sockaddr_in *sin, struct inpcb *inp, kauth_cred_t cred) 235{ 236 struct inpcbtable *table = inp->inp_table; 237 struct socket *so = inp->inp_socket; 238 u_int16_t *lastport; 239 u_int16_t lport = 0; 240 enum kauth_network_req req; 241 int error; 242 243 if (inp->inp_flags & INP_LOWPORT) { 244#ifndef IPNOPRIVPORTS 245 req = KAUTH_REQ_NETWORK_BIND_PRIVPORT; 246#else 247 req = KAUTH_REQ_NETWORK_BIND_PORT; 248#endif 249 250 lastport = &table->inpt_lastlow; 251 } else { 252 req = KAUTH_REQ_NETWORK_BIND_PORT; 253 254 lastport = &table->inpt_lastport; 255 } 256 257 /* XXX-kauth: KAUTH_REQ_NETWORK_BIND_AUTOASSIGN_{,PRIV}PORT */ 258 error = kauth_authorize_network(cred, KAUTH_NETWORK_BIND, req, so, sin, 259 NULL); 260 if (error) 261 return (EACCES); 262 263 /* 264 * Use RFC6056 randomized port selection 265 */ 266 error = rfc6056_randport(&lport, &inp->inp_head, cred); 267 if (error) 268 return error; 269 270 inp->inp_flags |= INP_ANONPORT; 271 *lastport = lport; 272 lport = htons(lport); 273 inp->inp_lport = lport; 274 in_pcbstate(inp, INP_BOUND); 275 276 return (0); 277} 278 279static int 280in_pcbbind_addr(struct inpcb *inp, struct sockaddr_in *sin, kauth_cred_t cred) 281{ 282 if (sin->sin_family != AF_INET) 283 return (EAFNOSUPPORT); 284 285 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 286 /* Always succeed; port reuse handled in in_pcbbind_port(). */ 287 } else if (!in_nullhost(sin->sin_addr)) { 288 struct in_ifaddr *ia = NULL; 289 290 INADDR_TO_IA(sin->sin_addr, ia); 291 /* check for broadcast addresses */ 292 if (ia == NULL) 293 ia = ifatoia(ifa_ifwithaddr(sintosa(sin))); 294 if (ia == NULL) 295 return (EADDRNOTAVAIL); 296 } 297 298 inp->inp_laddr = sin->sin_addr; 299 300 return (0); 301} 302 303static int 304in_pcbbind_port(struct inpcb *inp, struct sockaddr_in *sin, kauth_cred_t cred) 305{ 306 struct inpcbtable *table = inp->inp_table; 307 struct socket *so = inp->inp_socket; 308 int reuseport = (so->so_options & SO_REUSEPORT); 309 int wild = 0, error; 310 311 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 312 /* 313 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 314 * allow complete duplication of binding if 315 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 316 * and a multicast address is bound on both 317 * new and duplicated sockets. 318 */ 319 if (so->so_options & SO_REUSEADDR) 320 reuseport = SO_REUSEADDR|SO_REUSEPORT; 321 } 322 323 if (sin->sin_port == 0) { 324 error = in_pcbsetport(sin, inp, cred); 325 if (error) 326 return (error); 327 } else { 328 struct inpcb *t; 329 vestigial_inpcb_t vestige; 330#ifdef INET6 331 struct in6pcb *t6; 332 struct in6_addr mapped; 333#endif 334 enum kauth_network_req req; 335 336 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 337 wild = 1; 338 339#ifndef IPNOPRIVPORTS 340 if (ntohs(sin->sin_port) < IPPORT_RESERVED) 341 req = KAUTH_REQ_NETWORK_BIND_PRIVPORT; 342 else 343#endif /* !IPNOPRIVPORTS */ 344 req = KAUTH_REQ_NETWORK_BIND_PORT; 345 346 error = kauth_authorize_network(cred, KAUTH_NETWORK_BIND, req, 347 so, sin, NULL); 348 if (error) 349 return (EACCES); 350 351#ifdef INET6 352 memset(&mapped, 0, sizeof(mapped)); 353 mapped.s6_addr16[5] = 0xffff; 354 memcpy(&mapped.s6_addr32[3], &sin->sin_addr, 355 sizeof(mapped.s6_addr32[3])); 356 t6 = in6_pcblookup_port(table, &mapped, sin->sin_port, wild, &vestige); 357 if (t6 && (reuseport & t6->in6p_socket->so_options) == 0) 358 return (EADDRINUSE); 359 if (!t6 && vestige.valid) { 360 if (!!reuseport != !!vestige.reuse_port) { 361 return EADDRINUSE; 362 } 363 } 364#endif 365 366 /* XXX-kauth */ 367 if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) { 368 t = in_pcblookup_port(table, sin->sin_addr, sin->sin_port, 1, &vestige); 369 /* 370 * XXX: investigate ramifications of loosening this 371 * restriction so that as long as both ports have 372 * SO_REUSEPORT allow the bind 373 */ 374 if (t && 375 (!in_nullhost(sin->sin_addr) || 376 !in_nullhost(t->inp_laddr) || 377 (t->inp_socket->so_options & SO_REUSEPORT) == 0) 378 && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) { 379 return (EADDRINUSE); 380 } 381 if (!t && vestige.valid) { 382 if ((!in_nullhost(sin->sin_addr) 383 || !in_nullhost(vestige.laddr.v4) 384 || !vestige.reuse_port) 385 && so->so_uidinfo->ui_uid != vestige.uid) { 386 return EADDRINUSE; 387 } 388 } 389 } 390 t = in_pcblookup_port(table, sin->sin_addr, sin->sin_port, wild, &vestige); 391 if (t && (reuseport & t->inp_socket->so_options) == 0) 392 return (EADDRINUSE); 393 if (!t 394 && vestige.valid 395 && !(reuseport && vestige.reuse_port)) 396 return EADDRINUSE; 397 398 inp->inp_lport = sin->sin_port; 399 in_pcbstate(inp, INP_BOUND); 400 } 401 402 LIST_REMOVE(&inp->inp_head, inph_lhash); 403 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 404 inph_lhash); 405 406 return (0); 407} 408 409int 410in_pcbbind(void *v, struct mbuf *nam, struct lwp *l) 411{ 412 struct inpcb *inp = v; 413 struct sockaddr_in *sin = NULL; /* XXXGCC */ 414 struct sockaddr_in lsin; 415 int error; 416 417 if (inp->inp_af != AF_INET) 418 return (EINVAL); 419 420 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 421 return (EADDRNOTAVAIL); 422 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 423 return (EINVAL); 424 425 if (nam != NULL) { 426 sin = mtod(nam, struct sockaddr_in *); 427 if (nam->m_len != sizeof (*sin)) 428 return (EINVAL); 429 } else { 430 lsin = *((const struct sockaddr_in *) 431 inp->inp_socket->so_proto->pr_domain->dom_sa_any); 432 sin = &lsin; 433 } 434 435 /* Bind address. */ 436 error = in_pcbbind_addr(inp, sin, l->l_cred); 437 if (error) 438 return (error); 439 440 /* Bind port. */ 441 error = in_pcbbind_port(inp, sin, l->l_cred); 442 if (error) { 443 inp->inp_laddr.s_addr = INADDR_ANY; 444 445 return (error); 446 } 447 448 return (0); 449} 450 451/* 452 * Connect from a socket to a specified address. 453 * Both address and port must be specified in argument sin. 454 * If don't have a local address for this socket yet, 455 * then pick one. 456 */ 457int 458in_pcbconnect(void *v, struct mbuf *nam, struct lwp *l) 459{ 460 struct inpcb *inp = v; 461 struct in_ifaddr *ia = NULL; 462 struct sockaddr_in *ifaddr = NULL; 463 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); 464 vestigial_inpcb_t vestige; 465 int error; 466 467 if (inp->inp_af != AF_INET) 468 return (EINVAL); 469 470 if (nam->m_len != sizeof (*sin)) 471 return (EINVAL); 472 if (sin->sin_family != AF_INET) 473 return (EAFNOSUPPORT); 474 if (sin->sin_port == 0) 475 return (EADDRNOTAVAIL); 476 if (TAILQ_FIRST(&in_ifaddrhead) != 0) { 477 /* 478 * If the destination address is INADDR_ANY, 479 * use any local address (likely loopback). 480 * If the supplied address is INADDR_BROADCAST, 481 * use the broadcast address of an interface 482 * which supports broadcast. (loopback does not) 483 */ 484 485 if (in_nullhost(sin->sin_addr)) { 486 sin->sin_addr = 487 TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr; 488 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) { 489 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 490 if (ia->ia_ifp->if_flags & IFF_BROADCAST) { 491 sin->sin_addr = 492 ia->ia_broadaddr.sin_addr; 493 break; 494 } 495 } 496 } 497 } 498 /* 499 * If we haven't bound which network number to use as ours, 500 * we will use the number of the outgoing interface. 501 * This depends on having done a routing lookup, which 502 * we will probably have to do anyway, so we might 503 * as well do it now. On the other hand if we are 504 * sending to multiple destinations we may have already 505 * done the lookup, so see if we can use the route 506 * from before. In any case, we only 507 * chose a port number once, even if sending to multiple 508 * destinations. 509 */ 510 if (in_nullhost(inp->inp_laddr)) { 511 int xerror; 512 ifaddr = in_selectsrc(sin, &inp->inp_route, 513 inp->inp_socket->so_options, inp->inp_moptions, &xerror); 514 if (ifaddr == NULL) { 515 if (xerror == 0) 516 xerror = EADDRNOTAVAIL; 517 return xerror; 518 } 519 INADDR_TO_IA(ifaddr->sin_addr, ia); 520 if (ia == NULL) 521 return (EADDRNOTAVAIL); 522 } 523 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port, 524 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr, 525 inp->inp_lport, &vestige) != 0 526 || vestige.valid) 527 return (EADDRINUSE); 528 if (in_nullhost(inp->inp_laddr)) { 529 if (inp->inp_lport == 0) { 530 error = in_pcbbind(inp, NULL, l); 531 /* 532 * This used to ignore the return value 533 * completely, but we need to check for 534 * ephemeral port shortage. 535 * And attempts to request low ports if not root. 536 */ 537 if (error != 0) 538 return (error); 539 } 540 inp->inp_laddr = ifaddr->sin_addr; 541 } 542 inp->inp_faddr = sin->sin_addr; 543 inp->inp_fport = sin->sin_port; 544 545 /* Late bind, if needed */ 546 if (inp->inp_bindportonsend) { 547 struct sockaddr_in lsin = *((const struct sockaddr_in *) 548 inp->inp_socket->so_proto->pr_domain->dom_sa_any); 549 lsin.sin_addr = inp->inp_laddr; 550 lsin.sin_port = 0; 551 552 if ((error = in_pcbbind_port(inp, &lsin, l->l_cred)) != 0) 553 return error; 554 } 555 556 in_pcbstate(inp, INP_CONNECTED); 557#if defined(KAME_IPSEC) || defined(FAST_IPSEC) 558 if (inp->inp_socket->so_type == SOCK_STREAM) 559 ipsec_pcbconn(inp->inp_sp); 560#endif 561 return (0); 562} 563 564void 565in_pcbdisconnect(void *v) 566{ 567 struct inpcb *inp = v; 568 569 if (inp->inp_af != AF_INET) 570 return; 571 572 inp->inp_faddr = zeroin_addr; 573 inp->inp_fport = 0; 574 in_pcbstate(inp, INP_BOUND); 575#if defined(KAME_IPSEC) || defined(FAST_IPSEC) 576 ipsec_pcbdisconn(inp->inp_sp); 577#endif 578 if (inp->inp_socket->so_state & SS_NOFDREF) 579 in_pcbdetach(inp); 580} 581 582void 583in_pcbdetach(void *v) 584{ 585 struct inpcb *inp = v; 586 struct socket *so = inp->inp_socket; 587 int s; 588 589 if (inp->inp_af != AF_INET) 590 return; 591 592#if defined(KAME_IPSEC) || defined(FAST_IPSEC) 593 ipsec4_delete_pcbpolicy(inp); 594#endif /*IPSEC*/ 595 so->so_pcb = 0; 596 if (inp->inp_options) 597 (void)m_free(inp->inp_options); 598 rtcache_free(&inp->inp_route); 599 ip_freemoptions(inp->inp_moptions); 600 s = splnet(); 601 in_pcbstate(inp, INP_ATTACHED); 602 LIST_REMOVE(&inp->inp_head, inph_lhash); 603 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head, 604 inph_queue); 605 pool_put(&inpcb_pool, inp); 606 splx(s); 607 sofree(so); /* drops the socket's lock */ 608 mutex_enter(softnet_lock); /* reacquire the softnet_lock */ 609} 610 611void 612in_setsockaddr(struct inpcb *inp, struct mbuf *nam) 613{ 614 struct sockaddr_in *sin; 615 616 if (inp->inp_af != AF_INET) 617 return; 618 619 sin = mtod(nam, struct sockaddr_in *); 620 sockaddr_in_init(sin, &inp->inp_laddr, inp->inp_lport); 621 nam->m_len = sin->sin_len; 622} 623 624void 625in_setpeeraddr(struct inpcb *inp, struct mbuf *nam) 626{ 627 struct sockaddr_in *sin; 628 629 if (inp->inp_af != AF_INET) 630 return; 631 632 sin = mtod(nam, struct sockaddr_in *); 633 sockaddr_in_init(sin, &inp->inp_faddr, inp->inp_fport); 634 nam->m_len = sin->sin_len; 635} 636 637/* 638 * Pass some notification to all connections of a protocol 639 * associated with address dst. The local address and/or port numbers 640 * may be specified to limit the search. The "usual action" will be 641 * taken, depending on the ctlinput cmd. The caller must filter any 642 * cmds that are uninteresting (e.g., no error in the map). 643 * Call the protocol specific routine (if any) to report 644 * any errors for each matching socket. 645 * 646 * Must be called at splsoftnet. 647 */ 648int 649in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg, 650 struct in_addr laddr, u_int lport_arg, int errno, 651 void (*notify)(struct inpcb *, int)) 652{ 653 struct inpcbhead *head; 654 struct inpcb *inp, *ninp; 655 u_int16_t fport = fport_arg, lport = lport_arg; 656 int nmatch; 657 658 if (in_nullhost(faddr) || notify == 0) 659 return (0); 660 661 nmatch = 0; 662 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 663 for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) { 664 ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash); 665 if (inp->inp_af != AF_INET) 666 continue; 667 if (in_hosteq(inp->inp_faddr, faddr) && 668 inp->inp_fport == fport && 669 inp->inp_lport == lport && 670 in_hosteq(inp->inp_laddr, laddr)) { 671 (*notify)(inp, errno); 672 nmatch++; 673 } 674 } 675 return (nmatch); 676} 677 678void 679in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno, 680 void (*notify)(struct inpcb *, int)) 681{ 682 struct inpcb *inp, *ninp; 683 684 if (in_nullhost(faddr) || notify == 0) 685 return; 686 687 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 688 inp != (void *)&table->inpt_queue; 689 inp = ninp) { 690 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 691 if (inp->inp_af != AF_INET) 692 continue; 693 if (in_hosteq(inp->inp_faddr, faddr)) 694 (*notify)(inp, errno); 695 } 696} 697 698void 699in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp) 700{ 701 struct inpcb *inp, *ninp; 702 struct ip_moptions *imo; 703 int i, gap; 704 705 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 706 inp != (void *)&table->inpt_queue; 707 inp = ninp) { 708 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 709 if (inp->inp_af != AF_INET) 710 continue; 711 imo = inp->inp_moptions; 712 if (imo != NULL) { 713 /* 714 * Unselect the outgoing interface if it is being 715 * detached. 716 */ 717 if (imo->imo_multicast_ifp == ifp) 718 imo->imo_multicast_ifp = NULL; 719 720 /* 721 * Drop multicast group membership if we joined 722 * through the interface being detached. 723 */ 724 for (i = 0, gap = 0; i < imo->imo_num_memberships; 725 i++) { 726 if (imo->imo_membership[i]->inm_ifp == ifp) { 727 in_delmulti(imo->imo_membership[i]); 728 gap++; 729 } else if (gap != 0) 730 imo->imo_membership[i - gap] = 731 imo->imo_membership[i]; 732 } 733 imo->imo_num_memberships -= gap; 734 } 735 } 736} 737 738void 739in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp) 740{ 741 struct rtentry *rt; 742 struct inpcb *inp, *ninp; 743 744 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 745 inp != (void *)&table->inpt_queue; 746 inp = ninp) { 747 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 748 if (inp->inp_af != AF_INET) 749 continue; 750 if ((rt = rtcache_validate(&inp->inp_route)) != NULL && 751 rt->rt_ifp == ifp) 752 in_rtchange(inp, 0); 753 } 754} 755 756/* 757 * Check for alternatives when higher level complains 758 * about service problems. For now, invalidate cached 759 * routing information. If the route was created dynamically 760 * (by a redirect), time to try a default gateway again. 761 */ 762void 763in_losing(struct inpcb *inp) 764{ 765 struct rtentry *rt; 766 struct rt_addrinfo info; 767 768 if (inp->inp_af != AF_INET) 769 return; 770 771 if ((rt = rtcache_validate(&inp->inp_route)) == NULL) 772 return; 773 774 memset(&info, 0, sizeof(info)); 775 info.rti_info[RTAX_DST] = rtcache_getdst(&inp->inp_route); 776 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 777 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 778 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 779 if (rt->rt_flags & RTF_DYNAMIC) 780 (void) rtrequest(RTM_DELETE, rt_getkey(rt), 781 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 782 NULL); 783 /* 784 * A new route can be allocated 785 * the next time output is attempted. 786 */ 787 rtcache_free(&inp->inp_route); 788} 789 790/* 791 * After a routing change, flush old routing. A new route can be 792 * allocated the next time output is attempted. 793 */ 794void 795in_rtchange(struct inpcb *inp, int errno) 796{ 797 798 if (inp->inp_af != AF_INET) 799 return; 800 801 rtcache_free(&inp->inp_route); 802 803 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 804} 805 806struct inpcb * 807in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr, 808 u_int lport_arg, int lookup_wildcard, vestigial_inpcb_t *vp) 809{ 810 struct inpcbhead *head; 811 struct inpcb_hdr *inph; 812 struct inpcb *inp, *match = 0; 813 int matchwild = 3, wildcard; 814 u_int16_t lport = lport_arg; 815 816 if (vp) 817 vp->valid = 0; 818 819 head = INPCBHASH_PORT(table, lport); 820 LIST_FOREACH(inph, head, inph_lhash) { 821 inp = (struct inpcb *)inph; 822 if (inp->inp_af != AF_INET) 823 continue; 824 825 if (inp->inp_lport != lport) 826 continue; 827 wildcard = 0; 828 if (!in_nullhost(inp->inp_faddr)) 829 wildcard++; 830 if (in_nullhost(inp->inp_laddr)) { 831 if (!in_nullhost(laddr)) 832 wildcard++; 833 } else { 834 if (in_nullhost(laddr)) 835 wildcard++; 836 else { 837 if (!in_hosteq(inp->inp_laddr, laddr)) 838 continue; 839 } 840 } 841 if (wildcard && !lookup_wildcard) 842 continue; 843 if (wildcard < matchwild) { 844 match = inp; 845 matchwild = wildcard; 846 if (matchwild == 0) 847 break; 848 } 849 } 850 if (match && matchwild == 0) 851 return match; 852 853 if (vp && table->vestige) { 854 void *state = (*table->vestige->init_ports4)(laddr, lport_arg, lookup_wildcard); 855 vestigial_inpcb_t better; 856 857 while (table->vestige 858 && (*table->vestige->next_port4)(state, vp)) { 859 860 if (vp->lport != lport) 861 continue; 862 wildcard = 0; 863 if (!in_nullhost(vp->faddr.v4)) 864 wildcard++; 865 if (in_nullhost(vp->laddr.v4)) { 866 if (!in_nullhost(laddr)) 867 wildcard++; 868 } else { 869 if (in_nullhost(laddr)) 870 wildcard++; 871 else { 872 if (!in_hosteq(vp->laddr.v4, laddr)) 873 continue; 874 } 875 } 876 if (wildcard && !lookup_wildcard) 877 continue; 878 if (wildcard < matchwild) { 879 better = *vp; 880 match = (void*)&better; 881 882 matchwild = wildcard; 883 if (matchwild == 0) 884 break; 885 } 886 } 887 888 if (match) { 889 if (match != (void*)&better) 890 return match; 891 else { 892 *vp = better; 893 return 0; 894 } 895 } 896 } 897 898 return (match); 899} 900 901#ifdef DIAGNOSTIC 902int in_pcbnotifymiss = 0; 903#endif 904 905struct inpcb * 906in_pcblookup_connect(struct inpcbtable *table, 907 struct in_addr faddr, u_int fport_arg, 908 struct in_addr laddr, u_int lport_arg, 909 vestigial_inpcb_t *vp) 910{ 911 struct inpcbhead *head; 912 struct inpcb_hdr *inph; 913 struct inpcb *inp; 914 u_int16_t fport = fport_arg, lport = lport_arg; 915 916 if (vp) 917 vp->valid = 0; 918 919 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 920 LIST_FOREACH(inph, head, inph_hash) { 921 inp = (struct inpcb *)inph; 922 if (inp->inp_af != AF_INET) 923 continue; 924 925 if (in_hosteq(inp->inp_faddr, faddr) && 926 inp->inp_fport == fport && 927 inp->inp_lport == lport && 928 in_hosteq(inp->inp_laddr, laddr)) 929 goto out; 930 } 931 if (vp && table->vestige) { 932 if ((*table->vestige->lookup4)(faddr, fport_arg, 933 laddr, lport_arg, vp)) 934 return 0; 935 } 936 937#ifdef DIAGNOSTIC 938 if (in_pcbnotifymiss) { 939 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 940 ntohl(faddr.s_addr), ntohs(fport), 941 ntohl(laddr.s_addr), ntohs(lport)); 942 } 943#endif 944 return (0); 945 946out: 947 /* Move this PCB to the head of hash chain. */ 948 inph = &inp->inp_head; 949 if (inph != LIST_FIRST(head)) { 950 LIST_REMOVE(inph, inph_hash); 951 LIST_INSERT_HEAD(head, inph, inph_hash); 952 } 953 return (inp); 954} 955 956struct inpcb * 957in_pcblookup_bind(struct inpcbtable *table, 958 struct in_addr laddr, u_int lport_arg) 959{ 960 struct inpcbhead *head; 961 struct inpcb_hdr *inph; 962 struct inpcb *inp; 963 u_int16_t lport = lport_arg; 964 965 head = INPCBHASH_BIND(table, laddr, lport); 966 LIST_FOREACH(inph, head, inph_hash) { 967 inp = (struct inpcb *)inph; 968 if (inp->inp_af != AF_INET) 969 continue; 970 971 if (inp->inp_lport == lport && 972 in_hosteq(inp->inp_laddr, laddr)) 973 goto out; 974 } 975 head = INPCBHASH_BIND(table, zeroin_addr, 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, zeroin_addr)) 983 goto out; 984 } 985#ifdef DIAGNOSTIC 986 if (in_pcbnotifymiss) { 987 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 988 ntohl(laddr.s_addr), ntohs(lport)); 989 } 990#endif 991 return (0); 992 993out: 994 /* Move this PCB to the head of hash chain. */ 995 inph = &inp->inp_head; 996 if (inph != LIST_FIRST(head)) { 997 LIST_REMOVE(inph, inph_hash); 998 LIST_INSERT_HEAD(head, inph, inph_hash); 999 } 1000 return (inp); 1001} 1002 1003void 1004in_pcbstate(struct inpcb *inp, int state) 1005{ 1006 1007 if (inp->inp_af != AF_INET) 1008 return; 1009 1010 if (inp->inp_state > INP_ATTACHED) 1011 LIST_REMOVE(&inp->inp_head, inph_hash); 1012 1013 switch (state) { 1014 case INP_BOUND: 1015 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 1016 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 1017 inph_hash); 1018 break; 1019 case INP_CONNECTED: 1020 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 1021 inp->inp_faddr, inp->inp_fport, 1022 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 1023 inph_hash); 1024 break; 1025 } 1026 1027 inp->inp_state = state; 1028} 1029 1030struct rtentry * 1031in_pcbrtentry(struct inpcb *inp) 1032{ 1033 struct route *ro; 1034 union { 1035 struct sockaddr dst; 1036 struct sockaddr_in dst4; 1037 } u; 1038 1039 if (inp->inp_af != AF_INET) 1040 return (NULL); 1041 1042 ro = &inp->inp_route; 1043 1044 sockaddr_in_init(&u.dst4, &inp->inp_faddr, 0); 1045 return rtcache_lookup(ro, &u.dst); 1046} 1047 1048struct sockaddr_in * 1049in_selectsrc(struct sockaddr_in *sin, struct route *ro, 1050 int soopts, struct ip_moptions *mopts, int *errorp) 1051{ 1052 struct rtentry *rt = NULL; 1053 struct in_ifaddr *ia = NULL; 1054 1055 /* 1056 * If route is known or can be allocated now, take the 1057 * source address from the interface. Otherwise, punt. 1058 */ 1059 if ((soopts & SO_DONTROUTE) != 0) 1060 rtcache_free(ro); 1061 else { 1062 union { 1063 struct sockaddr dst; 1064 struct sockaddr_in dst4; 1065 } u; 1066 1067 sockaddr_in_init(&u.dst4, &sin->sin_addr, 0); 1068 rt = rtcache_lookup(ro, &u.dst); 1069 } 1070 /* 1071 * If we found a route, use the address 1072 * corresponding to the outgoing interface 1073 * unless it is the loopback (in case a route 1074 * to our address on another net goes to loopback). 1075 * 1076 * XXX Is this still true? Do we care? 1077 */ 1078 if (rt != NULL && (rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) 1079 ia = ifatoia(rt->rt_ifa); 1080 if (ia == NULL) { 1081 u_int16_t fport = sin->sin_port; 1082 1083 sin->sin_port = 0; 1084 ia = ifatoia(ifa_ifwithladdr(sintosa(sin))); 1085 sin->sin_port = fport; 1086 if (ia == NULL) { 1087 /* Find 1st non-loopback AF_INET address */ 1088 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 1089 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK)) 1090 break; 1091 } 1092 } 1093 if (ia == NULL) { 1094 *errorp = EADDRNOTAVAIL; 1095 return NULL; 1096 } 1097 } 1098 /* 1099 * If the destination address is multicast and an outgoing 1100 * interface has been set as a multicast option, use the 1101 * address of that interface as our source address. 1102 */ 1103 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) { 1104 struct ip_moptions *imo; 1105 struct ifnet *ifp; 1106 1107 imo = mopts; 1108 if (imo->imo_multicast_ifp != NULL) { 1109 ifp = imo->imo_multicast_ifp; 1110 IFP_TO_IA(ifp, ia); /* XXX */ 1111 if (ia == 0) { 1112 *errorp = EADDRNOTAVAIL; 1113 return NULL; 1114 } 1115 } 1116 } 1117 if (ia->ia_ifa.ifa_getifa != NULL) { 1118 ia = ifatoia((*ia->ia_ifa.ifa_getifa)(&ia->ia_ifa, 1119 sintosa(sin))); 1120 } 1121#ifdef GETIFA_DEBUG 1122 else 1123 printf("%s: missing ifa_getifa\n", __func__); 1124#endif 1125 return satosin(&ia->ia_addr); 1126} 1127