1 /* $NetBSD: in_pcb.c,v 1.127 2008/08/04 07:01:05 spz 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 1.127 2008/08/04 07:01:05 spz 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/pool.h> 110 #include <sys/proc.h> 111 #include <sys/kauth.h> 112 113 #include <net/if.h> 114 #include <net/route.h> 115 116 #include <netinet/in.h> 117 #include <netinet/in_systm.h> 118 #include <netinet/ip.h> 119 #include <netinet/in_pcb.h> 120 #include <netinet/in_var.h> 121 #include <netinet/ip_var.h> 122 123 #ifdef INET6 124 #include <netinet/ip6.h> 125 #include <netinet6/ip6_var.h> 126 #include <netinet6/in6_pcb.h> 127 #endif 128 129 #ifdef IPSEC 130 #include <netinet6/ipsec.h> 131 #include <netkey/key.h> 132 #elif FAST_IPSEC 133 #include <netipsec/ipsec.h> 134 #include <netipsec/key.h> 135 #endif /* IPSEC */ 136 137 struct in_addr zeroin_addr; 138 139 #define INPCBHASH_PORT(table, lport) \ 140 &(table)->inpt_porthashtbl[ntohs(lport) & (table)->inpt_porthash] 141 #define INPCBHASH_BIND(table, laddr, lport) \ 142 &(table)->inpt_bindhashtbl[ \ 143 ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash] 144 #define INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \ 145 &(table)->inpt_connecthashtbl[ \ 146 ((ntohl((faddr).s_addr) + ntohs(fport)) + \ 147 (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash] 148 149 int anonportmin = IPPORT_ANONMIN; 150 int anonportmax = IPPORT_ANONMAX; 151 int lowportmin = IPPORT_RESERVEDMIN; 152 int lowportmax = IPPORT_RESERVEDMAX; 153 154 POOL_INIT(inpcb_pool, sizeof(struct inpcb), 0, 0, 0, "inpcbpl", NULL, 155 IPL_NET); 156 157 void 158 in_pcbinit(struct inpcbtable *table, int bindhashsize, int connecthashsize) 159 { 160 161 CIRCLEQ_INIT(&table->inpt_queue); 162 table->inpt_porthashtbl = hashinit(bindhashsize, HASH_LIST, true, 163 &table->inpt_porthash); 164 table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, true, 165 &table->inpt_bindhash); 166 table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST, true, 167 &table->inpt_connecthash); 168 table->inpt_lastlow = IPPORT_RESERVEDMAX; 169 table->inpt_lastport = (u_int16_t)anonportmax; 170 } 171 172 int 173 in_pcballoc(struct socket *so, void *v) 174 { 175 struct inpcbtable *table = v; 176 struct inpcb *inp; 177 int s; 178 #if defined(IPSEC) || defined(FAST_IPSEC) 179 int error; 180 #endif 181 182 s = splnet(); 183 inp = pool_get(&inpcb_pool, PR_NOWAIT); 184 splx(s); 185 if (inp == NULL) 186 return (ENOBUFS); 187 bzero((void *)inp, sizeof(*inp)); 188 inp->inp_af = AF_INET; 189 inp->inp_table = table; 190 inp->inp_socket = so; 191 inp->inp_errormtu = -1; 192 #if defined(IPSEC) || defined(FAST_IPSEC) 193 error = ipsec_init_pcbpolicy(so, &inp->inp_sp); 194 if (error != 0) { 195 s = splnet(); 196 pool_put(&inpcb_pool, inp); 197 splx(s); 198 return error; 199 } 200 #endif 201 so->so_pcb = inp; 202 s = splnet(); 203 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, &inp->inp_head, 204 inph_queue); 205 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 206 inph_lhash); 207 in_pcbstate(inp, INP_ATTACHED); 208 splx(s); 209 return (0); 210 } 211 212 int 213 in_pcbbind(void *v, struct mbuf *nam, struct lwp *l) 214 { 215 struct in_ifaddr *ia = NULL; 216 struct inpcb *inp = v; 217 struct socket *so = inp->inp_socket; 218 struct inpcbtable *table = inp->inp_table; 219 struct sockaddr_in *sin = NULL; /* XXXGCC */ 220 u_int16_t lport = 0; 221 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 222 kauth_cred_t cred = l->l_cred; 223 224 if (inp->inp_af != AF_INET) 225 return (EINVAL); 226 227 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 228 return (EADDRNOTAVAIL); 229 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 230 return (EINVAL); 231 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 232 wild = 1; 233 if (nam == 0) 234 goto noname; 235 sin = mtod(nam, struct sockaddr_in *); 236 if (nam->m_len != sizeof (*sin)) 237 return (EINVAL); 238 if (sin->sin_family != AF_INET) 239 return (EAFNOSUPPORT); 240 lport = sin->sin_port; 241 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 242 /* 243 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 244 * allow complete duplication of binding if 245 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 246 * and a multicast address is bound on both 247 * new and duplicated sockets. 248 */ 249 if (so->so_options & SO_REUSEADDR) 250 reuseport = SO_REUSEADDR|SO_REUSEPORT; 251 } else if (!in_nullhost(sin->sin_addr)) { 252 sin->sin_port = 0; /* yech... */ 253 INADDR_TO_IA(sin->sin_addr, ia); 254 /* check for broadcast addresses */ 255 if (ia == NULL) 256 ia = ifatoia(ifa_ifwithaddr(sintosa(sin))); 257 if (ia == NULL) 258 return (EADDRNOTAVAIL); 259 } 260 if (lport) { 261 struct inpcb *t; 262 #ifdef INET6 263 struct in6pcb *t6; 264 struct in6_addr mapped; 265 #endif 266 #ifndef IPNOPRIVPORTS 267 /* GROSS */ 268 if (ntohs(lport) < IPPORT_RESERVED && 269 kauth_authorize_network(cred, 270 KAUTH_NETWORK_BIND, 271 KAUTH_REQ_NETWORK_BIND_PRIVPORT, so, sin, 272 NULL)) 273 return (EACCES); 274 #endif 275 #ifdef INET6 276 memset(&mapped, 0, sizeof(mapped)); 277 mapped.s6_addr16[5] = 0xffff; 278 memcpy(&mapped.s6_addr32[3], &sin->sin_addr, 279 sizeof(mapped.s6_addr32[3])); 280 t6 = in6_pcblookup_port(table, &mapped, lport, wild); 281 if (t6 && (reuseport & t6->in6p_socket->so_options) == 0) 282 return (EADDRINUSE); 283 #endif 284 if (so->so_uidinfo->ui_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) { 285 t = in_pcblookup_port(table, sin->sin_addr, lport, 1); 286 /* 287 * XXX: investigate ramifications of loosening this 288 * restriction so that as long as both ports have 289 * SO_REUSEPORT allow the bind 290 */ 291 if (t && 292 (!in_nullhost(sin->sin_addr) || 293 !in_nullhost(t->inp_laddr) || 294 (t->inp_socket->so_options & SO_REUSEPORT) == 0) 295 && (so->so_uidinfo->ui_uid != t->inp_socket->so_uidinfo->ui_uid)) { 296 return (EADDRINUSE); 297 } 298 } 299 t = in_pcblookup_port(table, sin->sin_addr, lport, wild); 300 if (t && (reuseport & t->inp_socket->so_options) == 0) 301 return (EADDRINUSE); 302 } 303 inp->inp_laddr = sin->sin_addr; 304 305 noname: 306 if (lport == 0) { 307 int cnt; 308 u_int16_t mymin, mymax; 309 u_int16_t *lastport; 310 311 if (inp->inp_flags & INP_LOWPORT) { 312 #ifndef IPNOPRIVPORTS 313 if (kauth_authorize_network(cred, 314 KAUTH_NETWORK_BIND, 315 KAUTH_REQ_NETWORK_BIND_PRIVPORT, so, 316 sin, NULL)) 317 return (EACCES); 318 #endif 319 mymin = lowportmin; 320 mymax = lowportmax; 321 lastport = &table->inpt_lastlow; 322 } else { 323 mymin = anonportmin; 324 mymax = anonportmax; 325 lastport = &table->inpt_lastport; 326 } 327 if (mymin > mymax) { /* sanity check */ 328 u_int16_t swp; 329 330 swp = mymin; 331 mymin = mymax; 332 mymax = swp; 333 } 334 335 lport = *lastport - 1; 336 for (cnt = mymax - mymin + 1; cnt; cnt--, lport--) { 337 if (lport < mymin || lport > mymax) 338 lport = mymax; 339 if (!in_pcblookup_port(table, inp->inp_laddr, 340 htons(lport), 1)) 341 goto found; 342 } 343 if (!in_nullhost(inp->inp_laddr)) 344 inp->inp_laddr.s_addr = INADDR_ANY; 345 return (EAGAIN); 346 found: 347 inp->inp_flags |= INP_ANONPORT; 348 *lastport = lport; 349 lport = htons(lport); 350 } 351 inp->inp_lport = lport; 352 LIST_REMOVE(&inp->inp_head, inph_lhash); 353 LIST_INSERT_HEAD(INPCBHASH_PORT(table, inp->inp_lport), &inp->inp_head, 354 inph_lhash); 355 in_pcbstate(inp, INP_BOUND); 356 return (0); 357 } 358 359 /* 360 * Connect from a socket to a specified address. 361 * Both address and port must be specified in argument sin. 362 * If don't have a local address for this socket yet, 363 * then pick one. 364 */ 365 int 366 in_pcbconnect(void *v, struct mbuf *nam, struct lwp *l) 367 { 368 struct inpcb *inp = v; 369 struct in_ifaddr *ia = NULL; 370 struct sockaddr_in *ifaddr = NULL; 371 struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); 372 int error; 373 374 if (inp->inp_af != AF_INET) 375 return (EINVAL); 376 377 if (nam->m_len != sizeof (*sin)) 378 return (EINVAL); 379 if (sin->sin_family != AF_INET) 380 return (EAFNOSUPPORT); 381 if (sin->sin_port == 0) 382 return (EADDRNOTAVAIL); 383 if (TAILQ_FIRST(&in_ifaddrhead) != 0) { 384 /* 385 * If the destination address is INADDR_ANY, 386 * use any local address (likely loopback). 387 * If the supplied address is INADDR_BROADCAST, 388 * use the broadcast address of an interface 389 * which supports broadcast. (loopback does not) 390 */ 391 392 if (in_nullhost(sin->sin_addr)) { 393 sin->sin_addr = 394 TAILQ_FIRST(&in_ifaddrhead)->ia_addr.sin_addr; 395 } else if (sin->sin_addr.s_addr == INADDR_BROADCAST) { 396 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 397 if (ia->ia_ifp->if_flags & IFF_BROADCAST) { 398 sin->sin_addr = 399 ia->ia_broadaddr.sin_addr; 400 break; 401 } 402 } 403 } 404 } 405 /* 406 * If we haven't bound which network number to use as ours, 407 * we will use the number of the outgoing interface. 408 * This depends on having done a routing lookup, which 409 * we will probably have to do anyway, so we might 410 * as well do it now. On the other hand if we are 411 * sending to multiple destinations we may have already 412 * done the lookup, so see if we can use the route 413 * from before. In any case, we only 414 * chose a port number once, even if sending to multiple 415 * destinations. 416 */ 417 if (in_nullhost(inp->inp_laddr)) { 418 int xerror; 419 ifaddr = in_selectsrc(sin, &inp->inp_route, 420 inp->inp_socket->so_options, inp->inp_moptions, &xerror); 421 if (ifaddr == NULL) { 422 if (xerror == 0) 423 xerror = EADDRNOTAVAIL; 424 return xerror; 425 } 426 INADDR_TO_IA(ifaddr->sin_addr, ia); 427 if (ia == NULL) 428 return (EADDRNOTAVAIL); 429 } 430 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port, 431 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr, 432 inp->inp_lport) != 0) 433 return (EADDRINUSE); 434 if (in_nullhost(inp->inp_laddr)) { 435 if (inp->inp_lport == 0) { 436 error = in_pcbbind(inp, NULL, l); 437 /* 438 * This used to ignore the return value 439 * completely, but we need to check for 440 * ephemeral port shortage. 441 * And attempts to request low ports if not root. 442 */ 443 if (error != 0) 444 return (error); 445 } 446 inp->inp_laddr = ifaddr->sin_addr; 447 } 448 inp->inp_faddr = sin->sin_addr; 449 inp->inp_fport = sin->sin_port; 450 in_pcbstate(inp, INP_CONNECTED); 451 #if defined(IPSEC) || defined(FAST_IPSEC) 452 if (inp->inp_socket->so_type == SOCK_STREAM) 453 ipsec_pcbconn(inp->inp_sp); 454 #endif 455 return (0); 456 } 457 458 void 459 in_pcbdisconnect(void *v) 460 { 461 struct inpcb *inp = v; 462 463 if (inp->inp_af != AF_INET) 464 return; 465 466 inp->inp_faddr = zeroin_addr; 467 inp->inp_fport = 0; 468 in_pcbstate(inp, INP_BOUND); 469 #if defined(IPSEC) || defined(FAST_IPSEC) 470 ipsec_pcbdisconn(inp->inp_sp); 471 #endif 472 if (inp->inp_socket->so_state & SS_NOFDREF) 473 in_pcbdetach(inp); 474 } 475 476 void 477 in_pcbdetach(void *v) 478 { 479 struct inpcb *inp = v; 480 struct socket *so = inp->inp_socket; 481 int s; 482 483 if (inp->inp_af != AF_INET) 484 return; 485 486 #if defined(IPSEC) || defined(FAST_IPSEC) 487 ipsec4_delete_pcbpolicy(inp); 488 #endif /*IPSEC*/ 489 so->so_pcb = 0; 490 if (inp->inp_options) 491 (void)m_free(inp->inp_options); 492 rtcache_free(&inp->inp_route); 493 ip_freemoptions(inp->inp_moptions); 494 s = splnet(); 495 in_pcbstate(inp, INP_ATTACHED); 496 LIST_REMOVE(&inp->inp_head, inph_lhash); 497 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, &inp->inp_head, 498 inph_queue); 499 pool_put(&inpcb_pool, inp); 500 splx(s); 501 sofree(so); /* drops the socket's lock */ 502 mutex_enter(softnet_lock); /* reacquire the softnet_lock */ 503 } 504 505 void 506 in_setsockaddr(struct inpcb *inp, struct mbuf *nam) 507 { 508 struct sockaddr_in *sin; 509 510 if (inp->inp_af != AF_INET) 511 return; 512 513 sin = mtod(nam, struct sockaddr_in *); 514 sockaddr_in_init(sin, &inp->inp_laddr, inp->inp_lport); 515 nam->m_len = sin->sin_len; 516 } 517 518 void 519 in_setpeeraddr(struct inpcb *inp, struct mbuf *nam) 520 { 521 struct sockaddr_in *sin; 522 523 if (inp->inp_af != AF_INET) 524 return; 525 526 sin = mtod(nam, struct sockaddr_in *); 527 sockaddr_in_init(sin, &inp->inp_faddr, inp->inp_fport); 528 nam->m_len = sin->sin_len; 529 } 530 531 /* 532 * Pass some notification to all connections of a protocol 533 * associated with address dst. The local address and/or port numbers 534 * may be specified to limit the search. The "usual action" will be 535 * taken, depending on the ctlinput cmd. The caller must filter any 536 * cmds that are uninteresting (e.g., no error in the map). 537 * Call the protocol specific routine (if any) to report 538 * any errors for each matching socket. 539 * 540 * Must be called at splsoftnet. 541 */ 542 int 543 in_pcbnotify(struct inpcbtable *table, struct in_addr faddr, u_int fport_arg, 544 struct in_addr laddr, u_int lport_arg, int errno, 545 void (*notify)(struct inpcb *, int)) 546 { 547 struct inpcbhead *head; 548 struct inpcb *inp, *ninp; 549 u_int16_t fport = fport_arg, lport = lport_arg; 550 int nmatch; 551 552 if (in_nullhost(faddr) || notify == 0) 553 return (0); 554 555 nmatch = 0; 556 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 557 for (inp = (struct inpcb *)LIST_FIRST(head); inp != NULL; inp = ninp) { 558 ninp = (struct inpcb *)LIST_NEXT(inp, inp_hash); 559 if (inp->inp_af != AF_INET) 560 continue; 561 if (in_hosteq(inp->inp_faddr, faddr) && 562 inp->inp_fport == fport && 563 inp->inp_lport == lport && 564 in_hosteq(inp->inp_laddr, laddr)) { 565 (*notify)(inp, errno); 566 nmatch++; 567 } 568 } 569 return (nmatch); 570 } 571 572 void 573 in_pcbnotifyall(struct inpcbtable *table, struct in_addr faddr, int errno, 574 void (*notify)(struct inpcb *, int)) 575 { 576 struct inpcb *inp, *ninp; 577 578 if (in_nullhost(faddr) || notify == 0) 579 return; 580 581 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 582 inp != (void *)&table->inpt_queue; 583 inp = ninp) { 584 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 585 if (inp->inp_af != AF_INET) 586 continue; 587 if (in_hosteq(inp->inp_faddr, faddr)) 588 (*notify)(inp, errno); 589 } 590 } 591 592 void 593 in_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp) 594 { 595 struct inpcb *inp, *ninp; 596 struct ip_moptions *imo; 597 int i, gap; 598 599 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 600 inp != (void *)&table->inpt_queue; 601 inp = ninp) { 602 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 603 if (inp->inp_af != AF_INET) 604 continue; 605 imo = inp->inp_moptions; 606 if (imo != NULL) { 607 /* 608 * Unselect the outgoing interface if it is being 609 * detached. 610 */ 611 if (imo->imo_multicast_ifp == ifp) 612 imo->imo_multicast_ifp = NULL; 613 614 /* 615 * Drop multicast group membership if we joined 616 * through the interface being detached. 617 */ 618 for (i = 0, gap = 0; i < imo->imo_num_memberships; 619 i++) { 620 if (imo->imo_membership[i]->inm_ifp == ifp) { 621 in_delmulti(imo->imo_membership[i]); 622 gap++; 623 } else if (gap != 0) 624 imo->imo_membership[i - gap] = 625 imo->imo_membership[i]; 626 } 627 imo->imo_num_memberships -= gap; 628 } 629 } 630 } 631 632 void 633 in_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp) 634 { 635 struct rtentry *rt; 636 struct inpcb *inp, *ninp; 637 638 for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); 639 inp != (void *)&table->inpt_queue; 640 inp = ninp) { 641 ninp = (struct inpcb *)CIRCLEQ_NEXT(inp, inp_queue); 642 if (inp->inp_af != AF_INET) 643 continue; 644 if ((rt = rtcache_validate(&inp->inp_route)) != NULL && 645 rt->rt_ifp == ifp) 646 in_rtchange(inp, 0); 647 } 648 } 649 650 /* 651 * Check for alternatives when higher level complains 652 * about service problems. For now, invalidate cached 653 * routing information. If the route was created dynamically 654 * (by a redirect), time to try a default gateway again. 655 */ 656 void 657 in_losing(struct inpcb *inp) 658 { 659 struct rtentry *rt; 660 struct rt_addrinfo info; 661 662 if (inp->inp_af != AF_INET) 663 return; 664 665 if ((rt = rtcache_validate(&inp->inp_route)) == NULL) 666 return; 667 668 memset(&info, 0, sizeof(info)); 669 info.rti_info[RTAX_DST] = rtcache_getdst(&inp->inp_route); 670 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 671 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 672 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 673 if (rt->rt_flags & RTF_DYNAMIC) 674 (void) rtrequest(RTM_DELETE, rt_getkey(rt), 675 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 676 NULL); 677 /* 678 * A new route can be allocated 679 * the next time output is attempted. 680 */ 681 rtcache_free(&inp->inp_route); 682 } 683 684 /* 685 * After a routing change, flush old routing. A new route can be 686 * allocated the next time output is attempted. 687 */ 688 void 689 in_rtchange(struct inpcb *inp, int errno) 690 { 691 692 if (inp->inp_af != AF_INET) 693 return; 694 695 rtcache_free(&inp->inp_route); 696 697 /* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 698 } 699 700 struct inpcb * 701 in_pcblookup_port(struct inpcbtable *table, struct in_addr laddr, 702 u_int lport_arg, int lookup_wildcard) 703 { 704 struct inpcbhead *head; 705 struct inpcb_hdr *inph; 706 struct inpcb *inp, *match = 0; 707 int matchwild = 3, wildcard; 708 u_int16_t lport = lport_arg; 709 710 head = INPCBHASH_PORT(table, lport); 711 LIST_FOREACH(inph, head, inph_lhash) { 712 inp = (struct inpcb *)inph; 713 if (inp->inp_af != AF_INET) 714 continue; 715 716 if (inp->inp_lport != lport) 717 continue; 718 wildcard = 0; 719 if (!in_nullhost(inp->inp_faddr)) 720 wildcard++; 721 if (in_nullhost(inp->inp_laddr)) { 722 if (!in_nullhost(laddr)) 723 wildcard++; 724 } else { 725 if (in_nullhost(laddr)) 726 wildcard++; 727 else { 728 if (!in_hosteq(inp->inp_laddr, laddr)) 729 continue; 730 } 731 } 732 if (wildcard && !lookup_wildcard) 733 continue; 734 if (wildcard < matchwild) { 735 match = inp; 736 matchwild = wildcard; 737 if (matchwild == 0) 738 break; 739 } 740 } 741 return (match); 742 } 743 744 #ifdef DIAGNOSTIC 745 int in_pcbnotifymiss = 0; 746 #endif 747 748 struct inpcb * 749 in_pcblookup_connect(struct inpcbtable *table, 750 struct in_addr faddr, u_int fport_arg, 751 struct in_addr laddr, u_int lport_arg) 752 { 753 struct inpcbhead *head; 754 struct inpcb_hdr *inph; 755 struct inpcb *inp; 756 u_int16_t fport = fport_arg, lport = lport_arg; 757 758 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 759 LIST_FOREACH(inph, head, inph_hash) { 760 inp = (struct inpcb *)inph; 761 if (inp->inp_af != AF_INET) 762 continue; 763 764 if (in_hosteq(inp->inp_faddr, faddr) && 765 inp->inp_fport == fport && 766 inp->inp_lport == lport && 767 in_hosteq(inp->inp_laddr, laddr)) 768 goto out; 769 } 770 #ifdef DIAGNOSTIC 771 if (in_pcbnotifymiss) { 772 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 773 ntohl(faddr.s_addr), ntohs(fport), 774 ntohl(laddr.s_addr), ntohs(lport)); 775 } 776 #endif 777 return (0); 778 779 out: 780 /* Move this PCB to the head of hash chain. */ 781 inph = &inp->inp_head; 782 if (inph != LIST_FIRST(head)) { 783 LIST_REMOVE(inph, inph_hash); 784 LIST_INSERT_HEAD(head, inph, inph_hash); 785 } 786 return (inp); 787 } 788 789 struct inpcb * 790 in_pcblookup_bind(struct inpcbtable *table, 791 struct in_addr laddr, u_int lport_arg) 792 { 793 struct inpcbhead *head; 794 struct inpcb_hdr *inph; 795 struct inpcb *inp; 796 u_int16_t lport = lport_arg; 797 798 head = INPCBHASH_BIND(table, laddr, lport); 799 LIST_FOREACH(inph, head, inph_hash) { 800 inp = (struct inpcb *)inph; 801 if (inp->inp_af != AF_INET) 802 continue; 803 804 if (inp->inp_lport == lport && 805 in_hosteq(inp->inp_laddr, laddr)) 806 goto out; 807 } 808 head = INPCBHASH_BIND(table, zeroin_addr, lport); 809 LIST_FOREACH(inph, head, inph_hash) { 810 inp = (struct inpcb *)inph; 811 if (inp->inp_af != AF_INET) 812 continue; 813 814 if (inp->inp_lport == lport && 815 in_hosteq(inp->inp_laddr, zeroin_addr)) 816 goto out; 817 } 818 #ifdef DIAGNOSTIC 819 if (in_pcbnotifymiss) { 820 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 821 ntohl(laddr.s_addr), ntohs(lport)); 822 } 823 #endif 824 return (0); 825 826 out: 827 /* Move this PCB to the head of hash chain. */ 828 inph = &inp->inp_head; 829 if (inph != LIST_FIRST(head)) { 830 LIST_REMOVE(inph, inph_hash); 831 LIST_INSERT_HEAD(head, inph, inph_hash); 832 } 833 return (inp); 834 } 835 836 void 837 in_pcbstate(struct inpcb *inp, int state) 838 { 839 840 if (inp->inp_af != AF_INET) 841 return; 842 843 if (inp->inp_state > INP_ATTACHED) 844 LIST_REMOVE(&inp->inp_head, inph_hash); 845 846 switch (state) { 847 case INP_BOUND: 848 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 849 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 850 inph_hash); 851 break; 852 case INP_CONNECTED: 853 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 854 inp->inp_faddr, inp->inp_fport, 855 inp->inp_laddr, inp->inp_lport), &inp->inp_head, 856 inph_hash); 857 break; 858 } 859 860 inp->inp_state = state; 861 } 862 863 struct rtentry * 864 in_pcbrtentry(struct inpcb *inp) 865 { 866 struct route *ro; 867 union { 868 struct sockaddr dst; 869 struct sockaddr_in dst4; 870 } u; 871 872 if (inp->inp_af != AF_INET) 873 return (NULL); 874 875 ro = &inp->inp_route; 876 877 sockaddr_in_init(&u.dst4, &inp->inp_faddr, 0); 878 return rtcache_lookup(ro, &u.dst); 879 } 880 881 struct sockaddr_in * 882 in_selectsrc(struct sockaddr_in *sin, struct route *ro, 883 int soopts, struct ip_moptions *mopts, int *errorp) 884 { 885 struct rtentry *rt = NULL; 886 struct in_ifaddr *ia = NULL; 887 888 /* 889 * If route is known or can be allocated now, take the 890 * source address from the interface. Otherwise, punt. 891 */ 892 if ((soopts & SO_DONTROUTE) != 0) 893 rtcache_free(ro); 894 else { 895 union { 896 struct sockaddr dst; 897 struct sockaddr_in dst4; 898 } u; 899 900 sockaddr_in_init(&u.dst4, &sin->sin_addr, 0); 901 rt = rtcache_lookup(ro, &u.dst); 902 } 903 /* 904 * If we found a route, use the address 905 * corresponding to the outgoing interface 906 * unless it is the loopback (in case a route 907 * to our address on another net goes to loopback). 908 * 909 * XXX Is this still true? Do we care? 910 */ 911 if (rt != NULL && (rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) 912 ia = ifatoia(rt->rt_ifa); 913 if (ia == NULL) { 914 u_int16_t fport = sin->sin_port; 915 916 sin->sin_port = 0; 917 ia = ifatoia(ifa_ifwithladdr(sintosa(sin))); 918 sin->sin_port = fport; 919 if (ia == NULL) { 920 /* Find 1st non-loopback AF_INET address */ 921 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 922 if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK)) 923 break; 924 } 925 } 926 if (ia == NULL) { 927 *errorp = EADDRNOTAVAIL; 928 return NULL; 929 } 930 } 931 /* 932 * If the destination address is multicast and an outgoing 933 * interface has been set as a multicast option, use the 934 * address of that interface as our source address. 935 */ 936 if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) { 937 struct ip_moptions *imo; 938 struct ifnet *ifp; 939 940 imo = mopts; 941 if (imo->imo_multicast_ifp != NULL) { 942 ifp = imo->imo_multicast_ifp; 943 IFP_TO_IA(ifp, ia); /* XXX */ 944 if (ia == 0) { 945 *errorp = EADDRNOTAVAIL; 946 return NULL; 947 } 948 } 949 } 950 if (ia->ia_ifa.ifa_getifa != NULL) { 951 ia = ifatoia((*ia->ia_ifa.ifa_getifa)(&ia->ia_ifa, 952 sintosa(sin))); 953 } 954 #ifdef GETIFA_DEBUG 955 else 956 printf("%s: missing ifa_getifa\n", __func__); 957 #endif 958 return satosin(&ia->ia_addr); 959 } 960