in_pcb.c revision 1.123.2.2
1 /* $NetBSD: in_pcb.c,v 1.123.2.2 2009/05/04 08:14:17 yamt 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 1.123.2.2 2009/05/04 08:14:17 yamt 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 (error); 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_nullhost(sin->sin_addr)) { 307 struct in_ifaddr *ia = NULL; 308 309 INADDR_TO_IA(sin->sin_addr, ia); 310 /* check for broadcast addresses */ 311 if (ia == NULL) 312 ia = ifatoia(ifa_ifwithaddr(sintosa(sin))); 313 if (ia == NULL) 314 return (EADDRNOTAVAIL); 315 } 316 317 inp->inp_laddr = sin->sin_addr; 318 319 return (0); 320 } 321 322 static int 323 in_pcbbind_port(struct inpcb *inp, struct sockaddr_in *sin, kauth_cred_t cred) 324 { 325 struct inpcbtable *table = inp->inp_table; 326 struct socket *so = inp->inp_socket; 327 int reuseport = (so->so_options & SO_REUSEPORT); 328 int wild = 0, error; 329 330 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 331 /* 332 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 333 * allow complete duplication of binding if 334 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 335 * and a multicast address is bound on both 336 * new and duplicated sockets. 337 */ 338 if (so->so_options & SO_REUSEADDR) 339 reuseport = SO_REUSEADDR|SO_REUSEPORT; 340 } 341 342 if (sin->sin_port == 0) { 343 error = in_pcbsetport(sin, inp, cred); 344 if (error) 345 return (error); 346 } else { 347 struct inpcb *t; 348 #ifdef INET6 349 struct in6pcb *t6; 350 struct in6_addr mapped; 351 #endif 352 enum kauth_network_req req; 353 354 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 355 wild = 1; 356 357 #ifndef IPNOPRIVPORTS 358 if (ntohs(sin->sin_port) < IPPORT_RESERVED) 359 req = KAUTH_REQ_NETWORK_BIND_PRIVPORT; 360 else 361 #endif /* !IPNOPRIVPORTS */ 362 req = KAUTH_REQ_NETWORK_BIND_PORT; 363 364 error = kauth_authorize_network(cred, KAUTH_NETWORK_BIND, req, 365 so, sin, NULL); 366 if (error) 367 return (error); 368 369 #ifdef INET6 370 memset(&mapped, 0, sizeof(mapped)); 371 mapped.s6_addr16[5] = 0xffff; 372 memcpy(&mapped.s6_addr32[3], &sin->sin_addr, 373 sizeof(mapped.s6_addr32[3])); 374 t6 = in6_pcblookup_port(table, &mapped, sin->sin_port, wild); 375 if (t6 && (reuseport & t6->in6p_socket->so_options) == 0) 376 return (EADDRINUSE); 377 #endif 378 379 /* XXX-kauth */ 380 if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) { 381 t = in_pcblookup_port(table, sin->sin_addr, sin->sin_port, 1); 382 /* 383 * XXX: investigate ramifications of loosening this 384 * restriction so that as long as both ports have 385 * SO_REUSEPORT allow the bind 386 */ 387 if (t && 388 (!in_nullhost(sin->sin_addr) || 389 !in_nullhost(t->inp_laddr) || 390 (t->inp_socket->so_options & SO_REUSEPORT) == 0) 391 && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) { 392 return (EADDRINUSE); 393 } 394 } 395 t = in_pcblookup_port(table, sin->sin_addr, sin->sin_port, wild); 396 if (t && (reuseport & t->inp_socket->so_options) == 0) 397 return (EADDRINUSE); 398 399 inp->inp_lport = sin->sin_port; 400 in_pcbstate(inp, INP_BOUND); 401 } 402 403 LIST_REMOVE(&inp->inp_head, inph_lhash); 404 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 405 inph_lhash); 406 407 return (0); 408 } 409 410 int 411 in_pcbbind(void *v, struct mbuf *nam, struct lwp *l) 412 { 413 struct inpcb *inp = v; 414 struct sockaddr_in *sin = NULL; /* XXXGCC */ 415 struct sockaddr_in lsin; 416 int error; 417 418 if (inp->inp_af != AF_INET) 419 return (EINVAL); 420 421 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 422 return (EADDRNOTAVAIL); 423 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 424 return (EINVAL); 425 426 if (nam != NULL) { 427 sin = mtod(nam, struct sockaddr_in *); 428 if (nam->m_len != sizeof (*sin)) 429 return (EINVAL); 430 } else { 431 lsin = *((const struct sockaddr_in *) 432 inp->inp_socket->so_proto->pr_domain->dom_sa_any); 433 sin = &lsin; 434 } 435 436 /* Bind address. */ 437 error = in_pcbbind_addr(inp, sin, l->l_cred); 438 if (error) 439 return (error); 440 441 /* Bind port. */ 442 error = in_pcbbind_port(inp, sin, l->l_cred); 443 if (error) { 444 inp->inp_laddr.s_addr = INADDR_ANY; 445 446 return (error); 447 } 448 449 return (0); 450 } 451 452 /* 453 * Connect from a socket to a specified address. 454 * Both address and port must be specified in argument sin. 455 * If don't have a local address for this socket yet, 456 * then pick one. 457 */ 458 int 459 in_pcbconnect(void *v, struct mbuf *nam, struct lwp *l) 460 { 461 struct inpcb *inp = v; 462 struct in_ifaddr *ia = NULL; 463 struct sockaddr_in *ifaddr = NULL; 464 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); 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) != 0) 526 return (EADDRINUSE); 527 if (in_nullhost(inp->inp_laddr)) { 528 if (inp->inp_lport == 0) { 529 error = in_pcbbind(inp, NULL, l); 530 /* 531 * This used to ignore the return value 532 * completely, but we need to check for 533 * ephemeral port shortage. 534 * And attempts to request low ports if not root. 535 */ 536 if (error != 0) 537 return (error); 538 } 539 inp->inp_laddr = ifaddr->sin_addr; 540 } 541 inp->inp_faddr = sin->sin_addr; 542 inp->inp_fport = sin->sin_port; 543 in_pcbstate(inp, INP_CONNECTED); 544 #if defined(IPSEC) || defined(FAST_IPSEC) 545 if (inp->inp_socket->so_type == SOCK_STREAM) 546 ipsec_pcbconn(inp->inp_sp); 547 #endif 548 return (0); 549 } 550 551 void 552 in_pcbdisconnect(void *v) 553 { 554 struct inpcb *inp = v; 555 556 if (inp->inp_af != AF_INET) 557 return; 558 559 inp->inp_faddr = zeroin_addr; 560 inp->inp_fport = 0; 561 in_pcbstate(inp, INP_BOUND); 562 #if defined(IPSEC) || defined(FAST_IPSEC) 563 ipsec_pcbdisconn(inp->inp_sp); 564 #endif 565 if (inp->inp_socket->so_state & SS_NOFDREF) 566 in_pcbdetach(inp); 567 } 568 569 void 570 in_pcbdetach(void *v) 571 { 572 struct inpcb *inp = v; 573 struct socket *so = inp->inp_socket; 574 int s; 575 576 if (inp->inp_af != AF_INET) 577 return; 578 579 #if defined(IPSEC) || defined(FAST_IPSEC) 580 ipsec4_delete_pcbpolicy(inp); 581 #endif /*IPSEC*/ 582 so->so_pcb = 0; 583 if (inp->inp_options) 584 (void)m_free(inp->inp_options); 585 rtcache_free(&inp->inp_route); 586 ip_freemoptions(inp->inp_moptions); 587 s = splnet(); 588 in_pcbstate(inp, INP_ATTACHED); 589 LIST_REMOVE(&inp->inp_head, inph_lhash); 590 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head, 591 inph_queue); 592 pool_put(&inpcb_pool, inp); 593 splx(s); 594 sofree(so); /* drops the socket's lock */ 595 mutex_enter(softnet_lock); /* reacquire the softnet_lock */ 596 } 597 598 void 599 in_setsockaddr(struct inpcb *inp, struct mbuf *nam) 600 { 601 struct sockaddr_in *sin; 602 603 if (inp->inp_af != AF_INET) 604 return; 605 606 sin = mtod(nam, struct sockaddr_in *); 607 sockaddr_in_init(sin, &inp->inp_laddr, inp->inp_lport); 608 nam->m_len = sin->sin_len; 609 } 610 611 void 612 in_setpeeraddr(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_faddr, inp->inp_fport); 621 nam->m_len = sin->sin_len; 622 } 623 624 /* 625 * Pass some notification to all connections of a protocol 626 * associated with address dst. The local address and/or port numbers 627 * may be specified to limit the search. The "usual action" will be 628 * taken, depending on the ctlinput cmd. The caller must filter any 629 * cmds that are uninteresting (e.g., no error in the map). 630 * Call the protocol specific routine (if any) to report 631 * any errors for each matching socket. 632 * 633 * Must be called at splsoftnet. 634 */ 635 int 636 in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg, 637 struct in_addr laddr, u_int lport_arg, int errno, 638 void (*notify)(struct inpcb *, int)) 639 { 640 struct inpcbhead *head; 641 struct inpcb *inp, *ninp; 642 u_int16_t fport = fport_arg, lport = lport_arg; 643 int nmatch; 644 645 if (in_nullhost(faddr) || notify == 0) 646 return (0); 647 648 nmatch = 0; 649 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 650 for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) { 651 ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash); 652 if (inp->inp_af != AF_INET) 653 continue; 654 if (in_hosteq(inp->inp_faddr, faddr) && 655 inp->inp_fport == fport && 656 inp->inp_lport == lport && 657 in_hosteq(inp->inp_laddr, laddr)) { 658 (*notify)(inp, errno); 659 nmatch++; 660 } 661 } 662 return (nmatch); 663 } 664 665 void 666 in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno, 667 void (*notify)(struct inpcb *, int)) 668 { 669 struct inpcb *inp, *ninp; 670 671 if (in_nullhost(faddr) || notify == 0) 672 return; 673 674 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 675 inp != (void *)&table->inpt_queue; 676 inp = ninp) { 677 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 678 if (inp->inp_af != AF_INET) 679 continue; 680 if (in_hosteq(inp->inp_faddr, faddr)) 681 (*notify)(inp, errno); 682 } 683 } 684 685 void 686 in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp) 687 { 688 struct inpcb *inp, *ninp; 689 struct ip_moptions *imo; 690 int i, gap; 691 692 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 693 inp != (void *)&table->inpt_queue; 694 inp = ninp) { 695 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 696 if (inp->inp_af != AF_INET) 697 continue; 698 imo = inp->inp_moptions; 699 if (imo != NULL) { 700 /* 701 * Unselect the outgoing interface if it is being 702 * detached. 703 */ 704 if (imo->imo_multicast_ifp == ifp) 705 imo->imo_multicast_ifp = NULL; 706 707 /* 708 * Drop multicast group membership if we joined 709 * through the interface being detached. 710 */ 711 for (i = 0, gap = 0; i < imo->imo_num_memberships; 712 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] = 718 imo->imo_membership[i]; 719 } 720 imo->imo_num_memberships -= gap; 721 } 722 } 723 } 724 725 void 726 in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp) 727 { 728 struct rtentry *rt; 729 struct inpcb *inp, *ninp; 730 731 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 732 inp != (void *)&table->inpt_queue; 733 inp = ninp) { 734 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 735 if (inp->inp_af != AF_INET) 736 continue; 737 if ((rt = rtcache_validate(&inp->inp_route)) != NULL && 738 rt->rt_ifp == ifp) 739 in_rtchange(inp, 0); 740 } 741 } 742 743 /* 744 * Check for alternatives when higher level complains 745 * about service problems. For now, invalidate cached 746 * routing information. If the route was created dynamically 747 * (by a redirect), time to try a default gateway again. 748 */ 749 void 750 in_losing(struct inpcb *inp) 751 { 752 struct rtentry *rt; 753 struct rt_addrinfo info; 754 755 if (inp->inp_af != AF_INET) 756 return; 757 758 if ((rt = rtcache_validate(&inp->inp_route)) == NULL) 759 return; 760 761 memset(&info, 0, sizeof(info)); 762 info.rti_info[RTAX_DST] = rtcache_getdst(&inp->inp_route); 763 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 764 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 765 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 766 if (rt->rt_flags & RTF_DYNAMIC) 767 (void) rtrequest(RTM_DELETE, rt_getkey(rt), 768 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 769 NULL); 770 /* 771 * A new route can be allocated 772 * the next time output is attempted. 773 */ 774 rtcache_free(&inp->inp_route); 775 } 776 777 /* 778 * After a routing change, flush old routing. A new route can be 779 * allocated the next time output is attempted. 780 */ 781 void 782 in_rtchange(struct inpcb *inp, int errno) 783 { 784 785 if (inp->inp_af != AF_INET) 786 return; 787 788 rtcache_free(&inp->inp_route); 789 790 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 791 } 792 793 struct inpcb * 794 in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr, 795 u_int lport_arg, int lookup_wildcard) 796 { 797 struct inpcbhead *head; 798 struct inpcb_hdr *inph; 799 struct inpcb *inp, *match = 0; 800 int matchwild = 3, wildcard; 801 u_int16_t lport = lport_arg; 802 803 head = INPCBHASH_PORT(table, lport); 804 LIST_FOREACH(inph, head, inph_lhash) { 805 inp = (struct inpcb *)inph; 806 if (inp->inp_af != AF_INET) 807 continue; 808 809 if (inp->inp_lport != lport) 810 continue; 811 wildcard = 0; 812 if (!in_nullhost(inp->inp_faddr)) 813 wildcard++; 814 if (in_nullhost(inp->inp_laddr)) { 815 if (!in_nullhost(laddr)) 816 wildcard++; 817 } else { 818 if (in_nullhost(laddr)) 819 wildcard++; 820 else { 821 if (!in_hosteq(inp->inp_laddr, laddr)) 822 continue; 823 } 824 } 825 if (wildcard && !lookup_wildcard) 826 continue; 827 if (wildcard < matchwild) { 828 match = inp; 829 matchwild = wildcard; 830 if (matchwild == 0) 831 break; 832 } 833 } 834 return (match); 835 } 836 837 #ifdef DIAGNOSTIC 838 int in_pcbnotifymiss = 0; 839 #endif 840 841 struct inpcb * 842 in_pcblookup_connect(struct inpcbtable *table, 843 struct in_addr faddr, u_int fport_arg, 844 struct in_addr laddr, u_int lport_arg) 845 { 846 struct inpcbhead *head; 847 struct inpcb_hdr *inph; 848 struct inpcb *inp; 849 u_int16_t fport = fport_arg, lport = lport_arg; 850 851 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 852 LIST_FOREACH(inph, head, inph_hash) { 853 inp = (struct inpcb *)inph; 854 if (inp->inp_af != AF_INET) 855 continue; 856 857 if (in_hosteq(inp->inp_faddr, faddr) && 858 inp->inp_fport == fport && 859 inp->inp_lport == lport && 860 in_hosteq(inp->inp_laddr, laddr)) 861 goto out; 862 } 863 #ifdef DIAGNOSTIC 864 if (in_pcbnotifymiss) { 865 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 866 ntohl(faddr.s_addr), ntohs(fport), 867 ntohl(laddr.s_addr), ntohs(lport)); 868 } 869 #endif 870 return (0); 871 872 out: 873 /* Move this PCB to the head of hash chain. */ 874 inph = &inp->inp_head; 875 if (inph != LIST_FIRST(head)) { 876 LIST_REMOVE(inph, inph_hash); 877 LIST_INSERT_HEAD(head, inph, inph_hash); 878 } 879 return (inp); 880 } 881 882 struct inpcb * 883 in_pcblookup_bind(struct inpcbtable *table, 884 struct in_addr laddr, u_int lport_arg) 885 { 886 struct inpcbhead *head; 887 struct inpcb_hdr *inph; 888 struct inpcb *inp; 889 u_int16_t lport = lport_arg; 890 891 head = INPCBHASH_BIND(table, laddr, lport); 892 LIST_FOREACH(inph, head, inph_hash) { 893 inp = (struct inpcb *)inph; 894 if (inp->inp_af != AF_INET) 895 continue; 896 897 if (inp->inp_lport == lport && 898 in_hosteq(inp->inp_laddr, laddr)) 899 goto out; 900 } 901 head = INPCBHASH_BIND(table, zeroin_addr, lport); 902 LIST_FOREACH(inph, head, inph_hash) { 903 inp = (struct inpcb *)inph; 904 if (inp->inp_af != AF_INET) 905 continue; 906 907 if (inp->inp_lport == lport && 908 in_hosteq(inp->inp_laddr, zeroin_addr)) 909 goto out; 910 } 911 #ifdef DIAGNOSTIC 912 if (in_pcbnotifymiss) { 913 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 914 ntohl(laddr.s_addr), ntohs(lport)); 915 } 916 #endif 917 return (0); 918 919 out: 920 /* Move this PCB to the head of hash chain. */ 921 inph = &inp->inp_head; 922 if (inph != LIST_FIRST(head)) { 923 LIST_REMOVE(inph, inph_hash); 924 LIST_INSERT_HEAD(head, inph, inph_hash); 925 } 926 return (inp); 927 } 928 929 void 930 in_pcbstate(struct inpcb *inp, int state) 931 { 932 933 if (inp->inp_af != AF_INET) 934 return; 935 936 if (inp->inp_state > INP_ATTACHED) 937 LIST_REMOVE(&inp->inp_head, inph_hash); 938 939 switch (state) { 940 case INP_BOUND: 941 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 942 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 943 inph_hash); 944 break; 945 case INP_CONNECTED: 946 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 947 inp->inp_faddr, inp->inp_fport, 948 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 949 inph_hash); 950 break; 951 } 952 953 inp->inp_state = state; 954 } 955 956 struct rtentry * 957 in_pcbrtentry(struct inpcb *inp) 958 { 959 struct route *ro; 960 union { 961 struct sockaddr dst; 962 struct sockaddr_in dst4; 963 } u; 964 965 if (inp->inp_af != AF_INET) 966 return (NULL); 967 968 ro = &inp->inp_route; 969 970 sockaddr_in_init(&u.dst4, &inp->inp_faddr, 0); 971 return rtcache_lookup(ro, &u.dst); 972 } 973 974 struct sockaddr_in * 975 in_selectsrc(struct sockaddr_in *sin, struct route *ro, 976 int soopts, struct ip_moptions *mopts, int *errorp) 977 { 978 struct rtentry *rt = NULL; 979 struct in_ifaddr *ia = NULL; 980 981 /* 982 * If route is known or can be allocated now, take the 983 * source address from the interface. Otherwise, punt. 984 */ 985 if ((soopts & SO_DONTROUTE) != 0) 986 rtcache_free(ro); 987 else { 988 union { 989 struct sockaddr dst; 990 struct sockaddr_in dst4; 991 } u; 992 993 sockaddr_in_init(&u.dst4, &sin->sin_addr, 0); 994 rt = rtcache_lookup(ro, &u.dst); 995 } 996 /* 997 * If we found a route, use the address 998 * corresponding to the outgoing interface 999 * unless it is the loopback (in case a route 1000 * to our address on another net goes to loopback). 1001 * 1002 * XXX Is this still true? Do we care? 1003 */ 1004 if (rt != NULL && (rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) 1005 ia = ifatoia(rt->rt_ifa); 1006 if (ia == NULL) { 1007 u_int16_t fport = sin->sin_port; 1008 1009 sin->sin_port = 0; 1010 ia = ifatoia(ifa_ifwithladdr(sintosa(sin))); 1011 sin->sin_port = fport; 1012 if (ia == NULL) { 1013 /* Find 1st non-loopback AF_INET address */ 1014 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 1015 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK)) 1016 break; 1017 } 1018 } 1019 if (ia == NULL) { 1020 *errorp = EADDRNOTAVAIL; 1021 return NULL; 1022 } 1023 } 1024 /* 1025 * If the destination address is multicast and an outgoing 1026 * interface has been set as a multicast option, use the 1027 * address of that interface as our source address. 1028 */ 1029 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) { 1030 struct ip_moptions *imo; 1031 struct ifnet *ifp; 1032 1033 imo = mopts; 1034 if (imo->imo_multicast_ifp != NULL) { 1035 ifp = imo->imo_multicast_ifp; 1036 IFP_TO_IA(ifp, ia); /* XXX */ 1037 if (ia == 0) { 1038 *errorp = EADDRNOTAVAIL; 1039 return NULL; 1040 } 1041 } 1042 } 1043 if (ia->ia_ifa.ifa_getifa != NULL) { 1044 ia = ifatoia((*ia->ia_ifa.ifa_getifa)(&ia->ia_ifa, 1045 sintosa(sin))); 1046 } 1047 #ifdef GETIFA_DEBUG 1048 else 1049 printf("%s: missing ifa_getifa\n", __func__); 1050 #endif 1051 return satosin(&ia->ia_addr); 1052 } 1053
Indexes created Fri Oct 17 03:10:13 GMT 2025