1 /* $NetBSD: in_pcb.c,v 1.137 2009/05/12 22:22:46 elad 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 * 49 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 50 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 51 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 52 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 53 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 54 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 55 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 56 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 57 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 58 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 59 * POSSIBILITY OF SUCH DAMAGE. 60 */ 61 62 /* 63 * Copyright (c) 1982, 1986, 1991, 1993, 1995 64 * The Regents of the University of California. All rights reserved. 65 * 66 * Redistribution and use in source and binary forms, with or without 67 * modification, are permitted provided that the following conditions 68 * are met: 69 * 1. Redistributions of source code must retain the above copyright 70 * notice, this list of conditions and the following disclaimer. 71 * 2. Redistributions in binary form must reproduce the above copyright 72 * notice, this list of conditions and the following disclaimer in the 73 * documentation and/or other materials provided with the distribution. 74 * 3. Neither the name of the University nor the names of its contributors 75 * may be used to endorse or promote products derived from this software 76 * without specific prior written permission. 77 * 78 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 79 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 80 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 81 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 82 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 83 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 84 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 85 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 86 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 87 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 88 * SUCH DAMAGE. 89 * 90 * @(#)in_pcb.c 8.4 (Berkeley) 5/24/95 91 */ 92 93 #include <sys/cdefs.h> 94 __KERNEL_RCSID(0, "$NetBSD: in_pcb.c,v 1.137 2009/05/12 22:22:46 elad Exp $"); 95 96 #include "opt_inet.h" 97 #include "opt_ipsec.h" 98 99 #include <sys/param.h> 100 #include <sys/systm.h> 101 #include <sys/malloc.h> 102 #include <sys/mbuf.h> 103 #include <sys/protosw.h> 104 #include <sys/socket.h> 105 #include <sys/socketvar.h> 106 #include <sys/ioctl.h> 107 #include <sys/errno.h> 108 #include <sys/time.h> 109 #include <sys/once.h> 110 #include <sys/pool.h> 111 #include <sys/proc.h> 112 #include <sys/kauth.h> 113 #include <sys/uidinfo.h> 114 #include <sys/domain.h> 115 116 #include <net/if.h> 117 #include <net/route.h> 118 119 #include <netinet/in.h> 120 #include <netinet/in_systm.h> 121 #include <netinet/ip.h> 122 #include <netinet/in_pcb.h> 123 #include <netinet/in_var.h> 124 #include <netinet/ip_var.h> 125 126 #ifdef INET6 127 #include <netinet/ip6.h> 128 #include <netinet6/ip6_var.h> 129 #include <netinet6/in6_pcb.h> 130 #endif 131 132 #ifdef IPSEC 133 #include <netinet6/ipsec.h> 134 #include <netkey/key.h> 135 #elif FAST_IPSEC 136 #include <netipsec/ipsec.h> 137 #include <netipsec/key.h> 138 #endif /* IPSEC */ 139 140 struct in_addr zeroin_addr; 141 142 #define INPCBHASH_PORT(table, lport) \ 143 &(table)->inpt_porthashtbl[ntohs(lport) & (table)->inpt_porthash] 144 #define INPCBHASH_BIND(table, laddr, lport) \ 145 &(table)->inpt_bindhashtbl[ \ 146 ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash] 147 #define INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \ 148 &(table)->inpt_connecthashtbl[ \ 149 ((ntohl((faddr).s_addr) + ntohs(fport)) + \ 150 (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash] 151 152 int anonportmin = IPPORT_ANONMIN; 153 int anonportmax = IPPORT_ANONMAX; 154 int lowportmin = IPPORT_RESERVEDMIN; 155 int lowportmax = IPPORT_RESERVEDMAX; 156 157 static struct pool inpcb_pool; 158 159 static int 160 inpcb_poolinit(void) 161 { 162 163 pool_init(&inpcb_pool, sizeof(struct inpcb), 0, 0, 0, "inpcbpl", NULL, 164 IPL_NET); 165 return 0; 166 } 167 168 void 169 in_pcbinit(struct inpcbtable *table, int bindhashsize, int connecthashsize) 170 { 171 static ONCE_DECL(control); 172 173 CIRCLEQ_INIT(&table->inpt_queue); 174 table->inpt_porthashtbl = hashinit(bindhashsize, HASH_LIST, true, 175 &table->inpt_porthash); 176 table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, true, 177 &table->inpt_bindhash); 178 table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST, true, 179 &table->inpt_connecthash); 180 table->inpt_lastlow = IPPORT_RESERVEDMAX; 181 table->inpt_lastport = (u_int16_t)anonportmax; 182 183 RUN_ONCE(&control, inpcb_poolinit); 184 } 185 186 int 187 in_pcballoc(struct socket *so, void *v) 188 { 189 struct inpcbtable *table = v; 190 struct inpcb *inp; 191 int s; 192 #if defined(IPSEC) || defined(FAST_IPSEC) 193 int error; 194 #endif 195 196 s = splnet(); 197 inp = pool_get(&inpcb_pool, PR_NOWAIT); 198 splx(s); 199 if (inp == NULL) 200 return (ENOBUFS); 201 memset((void *)inp, 0, sizeof(*inp)); 202 inp->inp_af = AF_INET; 203 inp->inp_table = table; 204 inp->inp_socket = so; 205 inp->inp_errormtu = -1; 206 #if defined(IPSEC) || defined(FAST_IPSEC) 207 error = ipsec_init_pcbpolicy(so, &inp->inp_sp); 208 if (error != 0) { 209 s = splnet(); 210 pool_put(&inpcb_pool, inp); 211 splx(s); 212 return error; 213 } 214 #endif 215 so->so_pcb = inp; 216 s = splnet(); 217 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, &inp->inp_head, 218 inph_queue); 219 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 220 inph_lhash); 221 in_pcbstate(inp, INP_ATTACHED); 222 splx(s); 223 return (0); 224 } 225 226 static int 227 in_pcbsetport(struct sockaddr_in *sin, struct inpcb *inp, kauth_cred_t cred) 228 { 229 struct inpcbtable *table = inp->inp_table; 230 struct socket *so = inp->inp_socket; 231 int cnt; 232 u_int16_t mymin, mymax; 233 u_int16_t *lastport; 234 u_int16_t lport = 0; 235 enum kauth_network_req req; 236 int error; 237 238 if (inp->inp_flags & INP_LOWPORT) { 239 #ifndef IPNOPRIVPORTS 240 req = KAUTH_REQ_NETWORK_BIND_PRIVPORT; 241 #else 242 req = KAUTH_REQ_NETWORK_BIND_PORT; 243 #endif 244 245 mymin = lowportmin; 246 mymax = lowportmax; 247 lastport = &table->inpt_lastlow; 248 } else { 249 req = KAUTH_REQ_NETWORK_BIND_PORT; 250 251 mymin = anonportmin; 252 mymax = anonportmax; 253 lastport = &table->inpt_lastport; 254 } 255 256 /* XXX-kauth: KAUTH_REQ_NETWORK_BIND_AUTOASSIGN_{,PRIV}PORT */ 257 error = kauth_authorize_network(cred, KAUTH_NETWORK_BIND, req, so, sin, 258 NULL); 259 if (error) 260 return (EACCES); 261 262 if (mymin > mymax) { /* sanity check */ 263 u_int16_t swp; 264 265 swp = mymin; 266 mymin = mymax; 267 mymax = swp; 268 } 269 270 lport = *lastport - 1; 271 for (cnt = mymax - mymin + 1; cnt; cnt--, lport--) { 272 if (lport < mymin || lport > mymax) 273 lport = mymax; 274 if (!in_pcblookup_port(table, sin->sin_addr, htons(lport), 1)) { 275 /* We have a free port, check with the secmodel(s). */ 276 sin->sin_port = lport; 277 error = kauth_authorize_network(cred, 278 KAUTH_NETWORK_BIND, req, so, sin, NULL); 279 if (error) { 280 /* Secmodel says no. Keep looking. */ 281 continue; 282 } 283 284 goto found; 285 } 286 } 287 288 return (EAGAIN); 289 290 found: 291 inp->inp_flags |= INP_ANONPORT; 292 *lastport = lport; 293 lport = htons(lport); 294 inp->inp_lport = lport; 295 in_pcbstate(inp, INP_BOUND); 296 297 return (0); 298 } 299 300 static int 301 in_pcbbind_addr(struct inpcb *inp, struct sockaddr_in *sin, kauth_cred_t cred) 302 { 303 if (sin->sin_family != AF_INET) 304 return (EAFNOSUPPORT); 305 306 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 307 /* Always succeed; port reuse handled in in_pcbbind_port(). */ 308 } else if (!in_nullhost(sin->sin_addr)) { 309 struct in_ifaddr *ia = NULL; 310 311 INADDR_TO_IA(sin->sin_addr, ia); 312 /* check for broadcast addresses */ 313 if (ia == NULL) 314 ia = ifatoia(ifa_ifwithaddr(sintosa(sin))); 315 if (ia == NULL) 316 return (EADDRNOTAVAIL); 317 } 318 319 inp->inp_laddr = sin->sin_addr; 320 321 return (0); 322 } 323 324 static int 325 in_pcbbind_port(struct inpcb *inp, struct sockaddr_in *sin, kauth_cred_t cred) 326 { 327 struct inpcbtable *table = inp->inp_table; 328 struct socket *so = inp->inp_socket; 329 int reuseport = (so->so_options & SO_REUSEPORT); 330 int wild = 0, error; 331 332 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 333 /* 334 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 335 * allow complete duplication of binding if 336 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 337 * and a multicast address is bound on both 338 * new and duplicated sockets. 339 */ 340 if (so->so_options & SO_REUSEADDR) 341 reuseport = SO_REUSEADDR|SO_REUSEPORT; 342 } 343 344 if (sin->sin_port == 0) { 345 error = in_pcbsetport(sin, inp, cred); 346 if (error) 347 return (error); 348 } else { 349 struct inpcb *t; 350 #ifdef INET6 351 struct in6pcb *t6; 352 struct in6_addr mapped; 353 #endif 354 enum kauth_network_req req; 355 356 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 357 wild = 1; 358 359 #ifndef IPNOPRIVPORTS 360 if (ntohs(sin->sin_port) < IPPORT_RESERVED) 361 req = KAUTH_REQ_NETWORK_BIND_PRIVPORT; 362 else 363 #endif /* !IPNOPRIVPORTS */ 364 req = KAUTH_REQ_NETWORK_BIND_PORT; 365 366 error = kauth_authorize_network(cred, KAUTH_NETWORK_BIND, req, 367 so, sin, NULL); 368 if (error) 369 return (EACCES); 370 371 #ifdef INET6 372 memset(&mapped, 0, sizeof(mapped)); 373 mapped.s6_addr16[5] = 0xffff; 374 memcpy(&mapped.s6_addr32[3], &sin->sin_addr, 375 sizeof(mapped.s6_addr32[3])); 376 t6 = in6_pcblookup_port(table, &mapped, sin->sin_port, wild); 377 if (t6 && (reuseport & t6->in6p_socket->so_options) == 0) 378 return (EADDRINUSE); 379 #endif 380 381 /* XXX-kauth */ 382 if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) { 383 t = in_pcblookup_port(table, sin->sin_addr, sin->sin_port, 1); 384 /* 385 * XXX: investigate ramifications of loosening this 386 * restriction so that as long as both ports have 387 * SO_REUSEPORT allow the bind 388 */ 389 if (t && 390 (!in_nullhost(sin->sin_addr) || 391 !in_nullhost(t->inp_laddr) || 392 (t->inp_socket->so_options & SO_REUSEPORT) == 0) 393 && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) { 394 return (EADDRINUSE); 395 } 396 } 397 t = in_pcblookup_port(table, sin->sin_addr, sin->sin_port, wild); 398 if (t && (reuseport & t->inp_socket->so_options) == 0) 399 return (EADDRINUSE); 400 401 inp->inp_lport = sin->sin_port; 402 in_pcbstate(inp, INP_BOUND); 403 } 404 405 LIST_REMOVE(&inp->inp_head, inph_lhash); 406 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 407 inph_lhash); 408 409 return (0); 410 } 411 412 int 413 in_pcbbind(void *v, struct mbuf *nam, struct lwp *l) 414 { 415 struct inpcb *inp = v; 416 struct sockaddr_in *sin = NULL; /* XXXGCC */ 417 struct sockaddr_in lsin; 418 int error; 419 420 if (inp->inp_af != AF_INET) 421 return (EINVAL); 422 423 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 424 return (EADDRNOTAVAIL); 425 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 426 return (EINVAL); 427 428 if (nam != NULL) { 429 sin = mtod(nam, struct sockaddr_in *); 430 if (nam->m_len != sizeof (*sin)) 431 return (EINVAL); 432 } else { 433 lsin = *((const struct sockaddr_in *) 434 inp->inp_socket->so_proto->pr_domain->dom_sa_any); 435 sin = &lsin; 436 } 437 438 /* Bind address. */ 439 error = in_pcbbind_addr(inp, sin, l->l_cred); 440 if (error) 441 return (error); 442 443 /* Bind port. */ 444 error = in_pcbbind_port(inp, sin, l->l_cred); 445 if (error) { 446 inp->inp_laddr.s_addr = INADDR_ANY; 447 448 return (error); 449 } 450 451 return (0); 452 } 453 454 /* 455 * Connect from a socket to a specified address. 456 * Both address and port must be specified in argument sin. 457 * If don't have a local address for this socket yet, 458 * then pick one. 459 */ 460 int 461 in_pcbconnect(void *v, struct mbuf *nam, struct lwp *l) 462 { 463 struct inpcb *inp = v; 464 struct in_ifaddr *ia = NULL; 465 struct sockaddr_in *ifaddr = NULL; 466 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); 467 int error; 468 469 if (inp->inp_af != AF_INET) 470 return (EINVAL); 471 472 if (nam->m_len != sizeof (*sin)) 473 return (EINVAL); 474 if (sin->sin_family != AF_INET) 475 return (EAFNOSUPPORT); 476 if (sin->sin_port == 0) 477 return (EADDRNOTAVAIL); 478 if (TAILQ_FIRST(&in_ifaddrhead) != 0) { 479 /* 480 * If the destination address is INADDR_ANY, 481 * use any local address (likely loopback). 482 * If the supplied address is INADDR_BROADCAST, 483 * use the broadcast address of an interface 484 * which supports broadcast. (loopback does not) 485 */ 486 487 if (in_nullhost(sin->sin_addr)) { 488 sin->sin_addr = 489 TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr; 490 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) { 491 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 492 if (ia->ia_ifp->if_flags & IFF_BROADCAST) { 493 sin->sin_addr = 494 ia->ia_broadaddr.sin_addr; 495 break; 496 } 497 } 498 } 499 } 500 /* 501 * If we haven't bound which network number to use as ours, 502 * we will use the number of the outgoing interface. 503 * This depends on having done a routing lookup, which 504 * we will probably have to do anyway, so we might 505 * as well do it now. On the other hand if we are 506 * sending to multiple destinations we may have already 507 * done the lookup, so see if we can use the route 508 * from before. In any case, we only 509 * chose a port number once, even if sending to multiple 510 * destinations. 511 */ 512 if (in_nullhost(inp->inp_laddr)) { 513 int xerror; 514 ifaddr = in_selectsrc(sin, &inp->inp_route, 515 inp->inp_socket->so_options, inp->inp_moptions, &xerror); 516 if (ifaddr == NULL) { 517 if (xerror == 0) 518 xerror = EADDRNOTAVAIL; 519 return xerror; 520 } 521 INADDR_TO_IA(ifaddr->sin_addr, ia); 522 if (ia == NULL) 523 return (EADDRNOTAVAIL); 524 } 525 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port, 526 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr, 527 inp->inp_lport) != 0) 528 return (EADDRINUSE); 529 if (in_nullhost(inp->inp_laddr)) { 530 if (inp->inp_lport == 0) { 531 error = in_pcbbind(inp, NULL, l); 532 /* 533 * This used to ignore the return value 534 * completely, but we need to check for 535 * ephemeral port shortage. 536 * And attempts to request low ports if not root. 537 */ 538 if (error != 0) 539 return (error); 540 } 541 inp->inp_laddr = ifaddr->sin_addr; 542 } 543 inp->inp_faddr = sin->sin_addr; 544 inp->inp_fport = sin->sin_port; 545 in_pcbstate(inp, INP_CONNECTED); 546 #if defined(IPSEC) || defined(FAST_IPSEC) 547 if (inp->inp_socket->so_type == SOCK_STREAM) 548 ipsec_pcbconn(inp->inp_sp); 549 #endif 550 return (0); 551 } 552 553 void 554 in_pcbdisconnect(void *v) 555 { 556 struct inpcb *inp = v; 557 558 if (inp->inp_af != AF_INET) 559 return; 560 561 inp->inp_faddr = zeroin_addr; 562 inp->inp_fport = 0; 563 in_pcbstate(inp, INP_BOUND); 564 #if defined(IPSEC) || defined(FAST_IPSEC) 565 ipsec_pcbdisconn(inp->inp_sp); 566 #endif 567 if (inp->inp_socket->so_state & SS_NOFDREF) 568 in_pcbdetach(inp); 569 } 570 571 void 572 in_pcbdetach(void *v) 573 { 574 struct inpcb *inp = v; 575 struct socket *so = inp->inp_socket; 576 int s; 577 578 if (inp->inp_af != AF_INET) 579 return; 580 581 #if defined(IPSEC) || defined(FAST_IPSEC) 582 ipsec4_delete_pcbpolicy(inp); 583 #endif /*IPSEC*/ 584 so->so_pcb = 0; 585 if (inp->inp_options) 586 (void)m_free(inp->inp_options); 587 rtcache_free(&inp->inp_route); 588 ip_freemoptions(inp->inp_moptions); 589 s = splnet(); 590 in_pcbstate(inp, INP_ATTACHED); 591 LIST_REMOVE(&inp->inp_head, inph_lhash); 592 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head, 593 inph_queue); 594 pool_put(&inpcb_pool, inp); 595 splx(s); 596 sofree(so); /* drops the socket's lock */ 597 mutex_enter(softnet_lock); /* reacquire the softnet_lock */ 598 } 599 600 void 601 in_setsockaddr(struct inpcb *inp, struct mbuf *nam) 602 { 603 struct sockaddr_in *sin; 604 605 if (inp->inp_af != AF_INET) 606 return; 607 608 sin = mtod(nam, struct sockaddr_in *); 609 sockaddr_in_init(sin, &inp->inp_laddr, inp->inp_lport); 610 nam->m_len = sin->sin_len; 611 } 612 613 void 614 in_setpeeraddr(struct inpcb *inp, struct mbuf *nam) 615 { 616 struct sockaddr_in *sin; 617 618 if (inp->inp_af != AF_INET) 619 return; 620 621 sin = mtod(nam, struct sockaddr_in *); 622 sockaddr_in_init(sin, &inp->inp_faddr, inp->inp_fport); 623 nam->m_len = sin->sin_len; 624 } 625 626 /* 627 * Pass some notification to all connections of a protocol 628 * associated with address dst. The local address and/or port numbers 629 * may be specified to limit the search. The "usual action" will be 630 * taken, depending on the ctlinput cmd. The caller must filter any 631 * cmds that are uninteresting (e.g., no error in the map). 632 * Call the protocol specific routine (if any) to report 633 * any errors for each matching socket. 634 * 635 * Must be called at splsoftnet. 636 */ 637 int 638 in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg, 639 struct in_addr laddr, u_int lport_arg, int errno, 640 void (*notify)(struct inpcb *, int)) 641 { 642 struct inpcbhead *head; 643 struct inpcb *inp, *ninp; 644 u_int16_t fport = fport_arg, lport = lport_arg; 645 int nmatch; 646 647 if (in_nullhost(faddr) || notify == 0) 648 return (0); 649 650 nmatch = 0; 651 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 652 for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) { 653 ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash); 654 if (inp->inp_af != AF_INET) 655 continue; 656 if (in_hosteq(inp->inp_faddr, faddr) && 657 inp->inp_fport == fport && 658 inp->inp_lport == lport && 659 in_hosteq(inp->inp_laddr, laddr)) { 660 (*notify)(inp, errno); 661 nmatch++; 662 } 663 } 664 return (nmatch); 665 } 666 667 void 668 in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno, 669 void (*notify)(struct inpcb *, int)) 670 { 671 struct inpcb *inp, *ninp; 672 673 if (in_nullhost(faddr) || notify == 0) 674 return; 675 676 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 677 inp != (void *)&table->inpt_queue; 678 inp = ninp) { 679 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 680 if (inp->inp_af != AF_INET) 681 continue; 682 if (in_hosteq(inp->inp_faddr, faddr)) 683 (*notify)(inp, errno); 684 } 685 } 686 687 void 688 in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp) 689 { 690 struct inpcb *inp, *ninp; 691 struct ip_moptions *imo; 692 int i, gap; 693 694 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 695 inp != (void *)&table->inpt_queue; 696 inp = ninp) { 697 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 698 if (inp->inp_af != AF_INET) 699 continue; 700 imo = inp->inp_moptions; 701 if (imo != NULL) { 702 /* 703 * Unselect the outgoing interface if it is being 704 * detached. 705 */ 706 if (imo->imo_multicast_ifp == ifp) 707 imo->imo_multicast_ifp = NULL; 708 709 /* 710 * Drop multicast group membership if we joined 711 * through the interface being detached. 712 */ 713 for (i = 0, gap = 0; i < imo->imo_num_memberships; 714 i++) { 715 if (imo->imo_membership[i]->inm_ifp == ifp) { 716 in_delmulti(imo->imo_membership[i]); 717 gap++; 718 } else if (gap != 0) 719 imo->imo_membership[i - gap] = 720 imo->imo_membership[i]; 721 } 722 imo->imo_num_memberships -= gap; 723 } 724 } 725 } 726 727 void 728 in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp) 729 { 730 struct rtentry *rt; 731 struct inpcb *inp, *ninp; 732 733 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 734 inp != (void *)&table->inpt_queue; 735 inp = ninp) { 736 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 737 if (inp->inp_af != AF_INET) 738 continue; 739 if ((rt = rtcache_validate(&inp->inp_route)) != NULL && 740 rt->rt_ifp == ifp) 741 in_rtchange(inp, 0); 742 } 743 } 744 745 /* 746 * Check for alternatives when higher level complains 747 * about service problems. For now, invalidate cached 748 * routing information. If the route was created dynamically 749 * (by a redirect), time to try a default gateway again. 750 */ 751 void 752 in_losing(struct inpcb *inp) 753 { 754 struct rtentry *rt; 755 struct rt_addrinfo info; 756 757 if (inp->inp_af != AF_INET) 758 return; 759 760 if ((rt = rtcache_validate(&inp->inp_route)) == NULL) 761 return; 762 763 memset(&info, 0, sizeof(info)); 764 info.rti_info[RTAX_DST] = rtcache_getdst(&inp->inp_route); 765 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 766 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 767 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 768 if (rt->rt_flags & RTF_DYNAMIC) 769 (void) rtrequest(RTM_DELETE, rt_getkey(rt), 770 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 771 NULL); 772 /* 773 * A new route can be allocated 774 * the next time output is attempted. 775 */ 776 rtcache_free(&inp->inp_route); 777 } 778 779 /* 780 * After a routing change, flush old routing. A new route can be 781 * allocated the next time output is attempted. 782 */ 783 void 784 in_rtchange(struct inpcb *inp, int errno) 785 { 786 787 if (inp->inp_af != AF_INET) 788 return; 789 790 rtcache_free(&inp->inp_route); 791 792 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 793 } 794 795 struct inpcb * 796 in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr, 797 u_int lport_arg, int lookup_wildcard) 798 { 799 struct inpcbhead *head; 800 struct inpcb_hdr *inph; 801 struct inpcb *inp, *match = 0; 802 int matchwild = 3, wildcard; 803 u_int16_t lport = lport_arg; 804 805 head = INPCBHASH_PORT(table, lport); 806 LIST_FOREACH(inph, head, inph_lhash) { 807 inp = (struct inpcb *)inph; 808 if (inp->inp_af != AF_INET) 809 continue; 810 811 if (inp->inp_lport != lport) 812 continue; 813 wildcard = 0; 814 if (!in_nullhost(inp->inp_faddr)) 815 wildcard++; 816 if (in_nullhost(inp->inp_laddr)) { 817 if (!in_nullhost(laddr)) 818 wildcard++; 819 } else { 820 if (in_nullhost(laddr)) 821 wildcard++; 822 else { 823 if (!in_hosteq(inp->inp_laddr, laddr)) 824 continue; 825 } 826 } 827 if (wildcard && !lookup_wildcard) 828 continue; 829 if (wildcard < matchwild) { 830 match = inp; 831 matchwild = wildcard; 832 if (matchwild == 0) 833 break; 834 } 835 } 836 return (match); 837 } 838 839 #ifdef DIAGNOSTIC 840 int in_pcbnotifymiss = 0; 841 #endif 842 843 struct inpcb * 844 in_pcblookup_connect(struct inpcbtable *table, 845 struct in_addr faddr, u_int fport_arg, 846 struct in_addr laddr, u_int lport_arg) 847 { 848 struct inpcbhead *head; 849 struct inpcb_hdr *inph; 850 struct inpcb *inp; 851 u_int16_t fport = fport_arg, lport = lport_arg; 852 853 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 854 LIST_FOREACH(inph, head, inph_hash) { 855 inp = (struct inpcb *)inph; 856 if (inp->inp_af != AF_INET) 857 continue; 858 859 if (in_hosteq(inp->inp_faddr, faddr) && 860 inp->inp_fport == fport && 861 inp->inp_lport == lport && 862 in_hosteq(inp->inp_laddr, laddr)) 863 goto out; 864 } 865 #ifdef DIAGNOSTIC 866 if (in_pcbnotifymiss) { 867 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 868 ntohl(faddr.s_addr), ntohs(fport), 869 ntohl(laddr.s_addr), ntohs(lport)); 870 } 871 #endif 872 return (0); 873 874 out: 875 /* Move this PCB to the head of hash chain. */ 876 inph = &inp->inp_head; 877 if (inph != LIST_FIRST(head)) { 878 LIST_REMOVE(inph, inph_hash); 879 LIST_INSERT_HEAD(head, inph, inph_hash); 880 } 881 return (inp); 882 } 883 884 struct inpcb * 885 in_pcblookup_bind(struct inpcbtable *table, 886 struct in_addr laddr, u_int lport_arg) 887 { 888 struct inpcbhead *head; 889 struct inpcb_hdr *inph; 890 struct inpcb *inp; 891 u_int16_t lport = lport_arg; 892 893 head = INPCBHASH_BIND(table, laddr, lport); 894 LIST_FOREACH(inph, head, inph_hash) { 895 inp = (struct inpcb *)inph; 896 if (inp->inp_af != AF_INET) 897 continue; 898 899 if (inp->inp_lport == lport && 900 in_hosteq(inp->inp_laddr, laddr)) 901 goto out; 902 } 903 head = INPCBHASH_BIND(table, zeroin_addr, lport); 904 LIST_FOREACH(inph, head, inph_hash) { 905 inp = (struct inpcb *)inph; 906 if (inp->inp_af != AF_INET) 907 continue; 908 909 if (inp->inp_lport == lport && 910 in_hosteq(inp->inp_laddr, zeroin_addr)) 911 goto out; 912 } 913 #ifdef DIAGNOSTIC 914 if (in_pcbnotifymiss) { 915 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 916 ntohl(laddr.s_addr), ntohs(lport)); 917 } 918 #endif 919 return (0); 920 921 out: 922 /* Move this PCB to the head of hash chain. */ 923 inph = &inp->inp_head; 924 if (inph != LIST_FIRST(head)) { 925 LIST_REMOVE(inph, inph_hash); 926 LIST_INSERT_HEAD(head, inph, inph_hash); 927 } 928 return (inp); 929 } 930 931 void 932 in_pcbstate(struct inpcb *inp, int state) 933 { 934 935 if (inp->inp_af != AF_INET) 936 return; 937 938 if (inp->inp_state > INP_ATTACHED) 939 LIST_REMOVE(&inp->inp_head, inph_hash); 940 941 switch (state) { 942 case INP_BOUND: 943 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 944 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 945 inph_hash); 946 break; 947 case INP_CONNECTED: 948 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 949 inp->inp_faddr, inp->inp_fport, 950 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 951 inph_hash); 952 break; 953 } 954 955 inp->inp_state = state; 956 } 957 958 struct rtentry * 959 in_pcbrtentry(struct inpcb *inp) 960 { 961 struct route *ro; 962 union { 963 struct sockaddr dst; 964 struct sockaddr_in dst4; 965 } u; 966 967 if (inp->inp_af != AF_INET) 968 return (NULL); 969 970 ro = &inp->inp_route; 971 972 sockaddr_in_init(&u.dst4, &inp->inp_faddr, 0); 973 return rtcache_lookup(ro, &u.dst); 974 } 975 976 struct sockaddr_in * 977 in_selectsrc(struct sockaddr_in *sin, struct route *ro, 978 int soopts, struct ip_moptions *mopts, int *errorp) 979 { 980 struct rtentry *rt = NULL; 981 struct in_ifaddr *ia = NULL; 982 983 /* 984 * If route is known or can be allocated now, take the 985 * source address from the interface. Otherwise, punt. 986 */ 987 if ((soopts & SO_DONTROUTE) != 0) 988 rtcache_free(ro); 989 else { 990 union { 991 struct sockaddr dst; 992 struct sockaddr_in dst4; 993 } u; 994 995 sockaddr_in_init(&u.dst4, &sin->sin_addr, 0); 996 rt = rtcache_lookup(ro, &u.dst); 997 } 998 /* 999 * If we found a route, use the address 1000 * corresponding to the outgoing interface 1001 * unless it is the loopback (in case a route 1002 * to our address on another net goes to loopback). 1003 * 1004 * XXX Is this still true? Do we care? 1005 */ 1006 if (rt != NULL && (rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) 1007 ia = ifatoia(rt->rt_ifa); 1008 if (ia == NULL) { 1009 u_int16_t fport = sin->sin_port; 1010 1011 sin->sin_port = 0; 1012 ia = ifatoia(ifa_ifwithladdr(sintosa(sin))); 1013 sin->sin_port = fport; 1014 if (ia == NULL) { 1015 /* Find 1st non-loopback AF_INET address */ 1016 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 1017 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK)) 1018 break; 1019 } 1020 } 1021 if (ia == NULL) { 1022 *errorp = EADDRNOTAVAIL; 1023 return NULL; 1024 } 1025 } 1026 /* 1027 * If the destination address is multicast and an outgoing 1028 * interface has been set as a multicast option, use the 1029 * address of that interface as our source address. 1030 */ 1031 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) { 1032 struct ip_moptions *imo; 1033 struct ifnet *ifp; 1034 1035 imo = mopts; 1036 if (imo->imo_multicast_ifp != NULL) { 1037 ifp = imo->imo_multicast_ifp; 1038 IFP_TO_IA(ifp, ia); /* XXX */ 1039 if (ia == 0) { 1040 *errorp = EADDRNOTAVAIL; 1041 return NULL; 1042 } 1043 } 1044 } 1045 if (ia->ia_ifa.ifa_getifa != NULL) { 1046 ia = ifatoia((*ia->ia_ifa.ifa_getifa)(&ia->ia_ifa, 1047 sintosa(sin))); 1048 } 1049 #ifdef GETIFA_DEBUG 1050 else 1051 printf("%s: missing ifa_getifa\n", __func__); 1052 #endif 1053 return satosin(&ia->ia_addr); 1054 } 1055