1 /* $NetBSD: in_pcb.c,v 1.35 1996/10/13 02:03:04 christos Exp $ */ 2 3 /* 4 * Copyright (c) 1982, 1986, 1991, 1993 5 * The Regents of the University of California. 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. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed by the University of 18 * California, Berkeley and its contributors. 19 * 4. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 36 */ 37 38 #include <sys/param.h> 39 #include <sys/systm.h> 40 #include <sys/malloc.h> 41 #include <sys/mbuf.h> 42 #include <sys/protosw.h> 43 #include <sys/socket.h> 44 #include <sys/socketvar.h> 45 #include <sys/ioctl.h> 46 #include <sys/errno.h> 47 #include <sys/time.h> 48 #include <sys/proc.h> 49 50 #include <net/if.h> 51 #include <net/route.h> 52 53 #include <netinet/in.h> 54 #include <netinet/in_systm.h> 55 #include <netinet/ip.h> 56 #include <netinet/in_pcb.h> 57 #include <netinet/in_var.h> 58 #include <netinet/ip_var.h> 59 60 struct in_addr zeroin_addr; 61 62 #define INPCBHASH_BIND(table, laddr, lport) \ 63 &(table)->inpt_bindhashtbl[ \ 64 ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash] 65 #define INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \ 66 &(table)->inpt_connecthashtbl[ \ 67 ((ntohl((faddr).s_addr) + ntohs(fport)) + \ 68 (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash] 69 70 struct inpcb * 71 in_pcblookup_port __P((struct inpcbtable *, 72 struct in_addr, u_int, int)); 73 74 void 75 in_pcbinit(table, bindhashsize, connecthashsize) 76 struct inpcbtable *table; 77 int bindhashsize, connecthashsize; 78 { 79 80 CIRCLEQ_INIT(&table->inpt_queue); 81 table->inpt_bindhashtbl = 82 hashinit(bindhashsize, M_PCB, &table->inpt_bindhash); 83 table->inpt_connecthashtbl = 84 hashinit(connecthashsize, M_PCB, &table->inpt_connecthash); 85 table->inpt_lastport = 0; 86 } 87 88 int 89 in_pcballoc(so, v) 90 struct socket *so; 91 void *v; 92 { 93 struct inpcbtable *table = v; 94 register struct inpcb *inp; 95 int s; 96 97 MALLOC(inp, struct inpcb *, sizeof(*inp), M_PCB, M_WAITOK); 98 if (inp == NULL) 99 return (ENOBUFS); 100 bzero((caddr_t)inp, sizeof(*inp)); 101 inp->inp_table = table; 102 inp->inp_socket = so; 103 so->so_pcb = inp; 104 s = splnet(); 105 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, inp, inp_queue); 106 in_pcbstate(inp, INP_ATTACHED); 107 splx(s); 108 return (0); 109 } 110 111 int 112 in_pcbbind(v, nam, p) 113 void *v; 114 struct mbuf *nam; 115 struct proc *p; 116 { 117 register struct inpcb *inp = v; 118 register struct socket *so = inp->inp_socket; 119 register struct inpcbtable *table = inp->inp_table; 120 register struct sockaddr_in *sin; 121 u_int16_t lport = 0; 122 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 123 int error; 124 125 if (in_ifaddr.tqh_first == 0) 126 return (EADDRNOTAVAIL); 127 if (inp->inp_lport || !in_nullhost(inp->inp_laddr)) 128 return (EINVAL); 129 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 && 130 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 || 131 (so->so_options & SO_ACCEPTCONN) == 0)) 132 wild = INPLOOKUP_WILDCARD; 133 if (nam == 0) 134 goto noname; 135 sin = mtod(nam, struct sockaddr_in *); 136 if (nam->m_len != sizeof (*sin)) 137 return (EINVAL); 138 #ifdef notdef 139 /* 140 * We should check the family, but old programs 141 * incorrectly fail to initialize it. 142 */ 143 if (sin->sin_family != AF_INET) 144 return (EAFNOSUPPORT); 145 #endif 146 lport = sin->sin_port; 147 if (IN_MULTICAST(sin->sin_addr.s_addr)) { 148 /* 149 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 150 * allow complete duplication of binding if 151 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 152 * and a multicast address is bound on both 153 * new and duplicated sockets. 154 */ 155 if (so->so_options & SO_REUSEADDR) 156 reuseport = SO_REUSEADDR|SO_REUSEPORT; 157 } else if (!in_nullhost(sin->sin_addr)) { 158 sin->sin_port = 0; /* yech... */ 159 if (ifa_ifwithaddr(sintosa(sin)) == 0) 160 return (EADDRNOTAVAIL); 161 } 162 if (lport) { 163 struct inpcb *t; 164 #ifndef IPNOPRIVPORTS 165 /* GROSS */ 166 if (ntohs(lport) < IPPORT_RESERVED && 167 (p == 0 || (error = suser(p->p_ucred, &p->p_acflag)))) 168 return (EACCES); 169 #endif 170 t = in_pcblookup_port(table, sin->sin_addr, lport, wild); 171 if (t && (reuseport & t->inp_socket->so_options) == 0) 172 return (EADDRINUSE); 173 } 174 inp->inp_laddr = sin->sin_addr; 175 noname: 176 if (lport == 0) 177 do { 178 if (table->inpt_lastport++ < IPPORT_RESERVED || 179 table->inpt_lastport > IPPORT_USERRESERVED) 180 table->inpt_lastport = IPPORT_RESERVED; 181 lport = htons(table->inpt_lastport); 182 } while (in_pcblookup_port(table, inp->inp_laddr, lport, wild)); 183 inp->inp_lport = lport; 184 in_pcbstate(inp, INP_BOUND); 185 return (0); 186 } 187 188 /* 189 * Connect from a socket to a specified address. 190 * Both address and port must be specified in argument sin. 191 * If don't have a local address for this socket yet, 192 * then pick one. 193 */ 194 int 195 in_pcbconnect(v, nam) 196 register void *v; 197 struct mbuf *nam; 198 { 199 register struct inpcb *inp = v; 200 struct in_ifaddr *ia; 201 struct sockaddr_in *ifaddr = NULL; 202 register struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); 203 204 if (nam->m_len != sizeof (*sin)) 205 return (EINVAL); 206 if (sin->sin_family != AF_INET) 207 return (EAFNOSUPPORT); 208 if (sin->sin_port == 0) 209 return (EADDRNOTAVAIL); 210 if (in_ifaddr.tqh_first != 0) { 211 /* 212 * If the destination address is INADDR_ANY, 213 * use the primary local address. 214 * If the supplied address is INADDR_BROADCAST, 215 * and the primary interface supports broadcast, 216 * choose the broadcast address for that interface. 217 */ 218 if (in_nullhost(sin->sin_addr)) 219 sin->sin_addr = in_ifaddr.tqh_first->ia_addr.sin_addr; 220 else if (sin->sin_addr.s_addr == INADDR_BROADCAST && 221 (in_ifaddr.tqh_first->ia_ifp->if_flags & IFF_BROADCAST)) 222 sin->sin_addr = in_ifaddr.tqh_first->ia_broadaddr.sin_addr; 223 } 224 /* 225 * If we haven't bound which network number to use as ours, 226 * we will use the number of the outgoing interface. 227 * This depends on having done a routing lookup, which 228 * we will probably have to do anyway, so we might 229 * as well do it now. On the other hand if we are 230 * sending to multiple destinations we may have already 231 * done the lookup, so see if we can use the route 232 * from before. In any case, we only 233 * chose a port number once, even if sending to multiple 234 * destinations. 235 */ 236 if (in_nullhost(inp->inp_laddr)) { 237 register struct route *ro; 238 239 ia = (struct in_ifaddr *)0; 240 /* 241 * If route is known or can be allocated now, 242 * our src addr is taken from the i/f, else punt. 243 */ 244 ro = &inp->inp_route; 245 if (ro->ro_rt && 246 (!in_hosteq(satosin(&ro->ro_dst)->sin_addr, 247 sin->sin_addr) || 248 inp->inp_socket->so_options & SO_DONTROUTE)) { 249 RTFREE(ro->ro_rt); 250 ro->ro_rt = (struct rtentry *)0; 251 } 252 if ((inp->inp_socket->so_options & SO_DONTROUTE) == 0 && /*XXX*/ 253 (ro->ro_rt == (struct rtentry *)0 || 254 ro->ro_rt->rt_ifp == (struct ifnet *)0)) { 255 /* No route yet, so try to acquire one */ 256 ro->ro_dst.sa_family = AF_INET; 257 ro->ro_dst.sa_len = sizeof(struct sockaddr_in); 258 satosin(&ro->ro_dst)->sin_addr = sin->sin_addr; 259 rtalloc(ro); 260 } 261 /* 262 * If we found a route, use the address 263 * corresponding to the outgoing interface 264 * unless it is the loopback (in case a route 265 * to our address on another net goes to loopback). 266 */ 267 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) 268 ia = ifatoia(ro->ro_rt->rt_ifa); 269 if (ia == 0) { 270 u_int16_t fport = sin->sin_port; 271 272 sin->sin_port = 0; 273 ia = ifatoia(ifa_ifwithladdr(sintosa(sin))); 274 sin->sin_port = fport; 275 if (ia == 0) 276 ia = in_ifaddr.tqh_first; 277 if (ia == 0) 278 return (EADDRNOTAVAIL); 279 } 280 /* 281 * If the destination address is multicast and an outgoing 282 * interface has been set as a multicast option, use the 283 * address of that interface as our source address. 284 */ 285 if (IN_MULTICAST(sin->sin_addr.s_addr) && 286 inp->inp_moptions != NULL) { 287 struct ip_moptions *imo; 288 struct ifnet *ifp; 289 290 imo = inp->inp_moptions; 291 if (imo->imo_multicast_ifp != NULL) { 292 ifp = imo->imo_multicast_ifp; 293 for (ia = in_ifaddr.tqh_first; ia != 0; 294 ia = ia->ia_list.tqe_next) 295 if (ia->ia_ifp == ifp) 296 break; 297 if (ia == 0) 298 return (EADDRNOTAVAIL); 299 } 300 } 301 ifaddr = satosin(&ia->ia_addr); 302 } 303 if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port, 304 !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr, 305 inp->inp_lport) != 0) 306 return (EADDRINUSE); 307 if (in_nullhost(inp->inp_laddr)) { 308 if (inp->inp_lport == 0) 309 (void)in_pcbbind(inp, (struct mbuf *)0, 310 (struct proc *)0); 311 inp->inp_laddr = ifaddr->sin_addr; 312 } 313 inp->inp_faddr = sin->sin_addr; 314 inp->inp_fport = sin->sin_port; 315 in_pcbstate(inp, INP_CONNECTED); 316 return (0); 317 } 318 319 void 320 in_pcbdisconnect(v) 321 void *v; 322 { 323 struct inpcb *inp = v; 324 325 inp->inp_faddr = zeroin_addr; 326 inp->inp_fport = 0; 327 in_pcbstate(inp, INP_BOUND); 328 if (inp->inp_socket->so_state & SS_NOFDREF) 329 in_pcbdetach(inp); 330 } 331 332 void 333 in_pcbdetach(v) 334 void *v; 335 { 336 struct inpcb *inp = v; 337 struct socket *so = inp->inp_socket; 338 int s; 339 340 so->so_pcb = 0; 341 sofree(so); 342 if (inp->inp_options) 343 (void)m_free(inp->inp_options); 344 if (inp->inp_route.ro_rt) 345 rtfree(inp->inp_route.ro_rt); 346 ip_freemoptions(inp->inp_moptions); 347 s = splnet(); 348 in_pcbstate(inp, INP_ATTACHED); 349 CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, inp, inp_queue); 350 splx(s); 351 FREE(inp, M_PCB); 352 } 353 354 void 355 in_setsockaddr(inp, nam) 356 register struct inpcb *inp; 357 struct mbuf *nam; 358 { 359 register struct sockaddr_in *sin; 360 361 nam->m_len = sizeof (*sin); 362 sin = mtod(nam, struct sockaddr_in *); 363 bzero((caddr_t)sin, sizeof (*sin)); 364 sin->sin_family = AF_INET; 365 sin->sin_len = sizeof(*sin); 366 sin->sin_port = inp->inp_lport; 367 sin->sin_addr = inp->inp_laddr; 368 } 369 370 void 371 in_setpeeraddr(inp, nam) 372 struct inpcb *inp; 373 struct mbuf *nam; 374 { 375 register struct sockaddr_in *sin; 376 377 nam->m_len = sizeof (*sin); 378 sin = mtod(nam, struct sockaddr_in *); 379 bzero((caddr_t)sin, sizeof (*sin)); 380 sin->sin_family = AF_INET; 381 sin->sin_len = sizeof(*sin); 382 sin->sin_port = inp->inp_fport; 383 sin->sin_addr = inp->inp_faddr; 384 } 385 386 /* 387 * Pass some notification to all connections of a protocol 388 * associated with address dst. The local address and/or port numbers 389 * may be specified to limit the search. The "usual action" will be 390 * taken, depending on the ctlinput cmd. The caller must filter any 391 * cmds that are uninteresting (e.g., no error in the map). 392 * Call the protocol specific routine (if any) to report 393 * any errors for each matching socket. 394 * 395 * Must be called at splsoftnet. 396 */ 397 void 398 in_pcbnotify(table, faddr, fport_arg, laddr, lport_arg, errno, notify) 399 struct inpcbtable *table; 400 struct in_addr faddr, laddr; 401 u_int fport_arg, lport_arg; 402 int errno; 403 void (*notify) __P((struct inpcb *, int)); 404 { 405 struct inpcbhead *head; 406 register struct inpcb *inp, *ninp; 407 u_int16_t fport = fport_arg, lport = lport_arg; 408 409 if (in_nullhost(faddr) || notify == 0) 410 return; 411 412 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 413 for (inp = head->lh_first; inp != NULL; inp = ninp) { 414 ninp = inp->inp_hash.le_next; 415 if (in_hosteq(inp->inp_faddr, faddr) && 416 inp->inp_fport == fport && 417 inp->inp_lport == lport && 418 in_hosteq(inp->inp_laddr, laddr)) 419 (*notify)(inp, errno); 420 } 421 } 422 423 void 424 in_pcbnotifyall(table, faddr, errno, notify) 425 struct inpcbtable *table; 426 struct in_addr faddr; 427 int errno; 428 void (*notify) __P((struct inpcb *, int)); 429 { 430 register struct inpcb *inp, *ninp; 431 432 if (in_nullhost(faddr) || notify == 0) 433 return; 434 435 for (inp = table->inpt_queue.cqh_first; 436 inp != (struct inpcb *)&table->inpt_queue; 437 inp = ninp) { 438 ninp = inp->inp_queue.cqe_next; 439 if (in_hosteq(inp->inp_faddr, faddr)) 440 (*notify)(inp, errno); 441 } 442 } 443 444 /* 445 * Check for alternatives when higher level complains 446 * about service problems. For now, invalidate cached 447 * routing information. If the route was created dynamically 448 * (by a redirect), time to try a default gateway again. 449 */ 450 void 451 in_losing(inp) 452 struct inpcb *inp; 453 { 454 register struct rtentry *rt; 455 struct rt_addrinfo info; 456 457 if ((rt = inp->inp_route.ro_rt)) { 458 inp->inp_route.ro_rt = 0; 459 bzero((caddr_t)&info, sizeof(info)); 460 info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst; 461 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 462 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 463 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 464 if (rt->rt_flags & RTF_DYNAMIC) 465 (void) rtrequest(RTM_DELETE, rt_key(rt), 466 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 467 (struct rtentry **)0); 468 else 469 /* 470 * A new route can be allocated 471 * the next time output is attempted. 472 */ 473 rtfree(rt); 474 } 475 } 476 477 /* 478 * After a routing change, flush old routing 479 * and allocate a (hopefully) better one. 480 */ 481 void 482 in_rtchange(inp, errno) 483 register struct inpcb *inp; 484 int errno; 485 { 486 487 if (inp->inp_route.ro_rt) { 488 rtfree(inp->inp_route.ro_rt); 489 inp->inp_route.ro_rt = 0; 490 /* 491 * A new route can be allocated the next time 492 * output is attempted. 493 */ 494 } 495 /* SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */ 496 } 497 498 struct inpcb * 499 in_pcblookup_port(table, laddr, lport_arg, flags) 500 struct inpcbtable *table; 501 struct in_addr laddr; 502 u_int lport_arg; 503 int flags; 504 { 505 register struct inpcb *inp, *match = 0; 506 int matchwild = 3, wildcard; 507 u_int16_t lport = lport_arg; 508 509 for (inp = table->inpt_queue.cqh_first; 510 inp != (struct inpcb *)&table->inpt_queue; 511 inp = inp->inp_queue.cqe_next) { 512 if (inp->inp_lport != lport) 513 continue; 514 wildcard = 0; 515 if (!in_nullhost(inp->inp_faddr)) 516 wildcard++; 517 if (in_nullhost(inp->inp_laddr)) { 518 if (!in_nullhost(laddr)) 519 wildcard++; 520 } else { 521 if (in_nullhost(laddr)) 522 wildcard++; 523 else { 524 if (!in_hosteq(inp->inp_laddr, laddr)) 525 continue; 526 } 527 } 528 if (wildcard && (flags & INPLOOKUP_WILDCARD) == 0) 529 continue; 530 if (wildcard < matchwild) { 531 match = inp; 532 matchwild = wildcard; 533 if (matchwild == 0) 534 break; 535 } 536 } 537 return (match); 538 } 539 540 #ifdef DIAGNOSTIC 541 int in_pcbnotifymiss = 0; 542 #endif 543 544 struct inpcb * 545 in_pcblookup_connect(table, faddr, fport_arg, laddr, lport_arg) 546 struct inpcbtable *table; 547 struct in_addr faddr, laddr; 548 u_int fport_arg, lport_arg; 549 { 550 struct inpcbhead *head; 551 register struct inpcb *inp; 552 u_int16_t fport = fport_arg, lport = lport_arg; 553 554 head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport); 555 for (inp = head->lh_first; inp != NULL; inp = inp->inp_hash.le_next) { 556 if (in_hosteq(inp->inp_faddr, faddr) && 557 inp->inp_fport == fport && 558 inp->inp_lport == lport && 559 in_hosteq(inp->inp_laddr, laddr)) 560 goto out; 561 } 562 #ifdef DIAGNOSTIC 563 if (in_pcbnotifymiss) { 564 printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n", 565 ntohl(faddr.s_addr), ntohs(fport), 566 ntohl(laddr.s_addr), ntohs(lport)); 567 } 568 #endif 569 return (0); 570 571 out: 572 /* Move this PCB to the head of hash chain. */ 573 if (inp != head->lh_first) { 574 LIST_REMOVE(inp, inp_hash); 575 LIST_INSERT_HEAD(head, inp, inp_hash); 576 } 577 return (inp); 578 } 579 580 struct inpcb * 581 in_pcblookup_bind(table, laddr, lport_arg) 582 struct inpcbtable *table; 583 struct in_addr laddr; 584 u_int lport_arg; 585 { 586 struct inpcbhead *head; 587 register struct inpcb *inp; 588 u_int16_t lport = lport_arg; 589 590 head = INPCBHASH_BIND(table, laddr, lport); 591 for (inp = head->lh_first; inp != NULL; inp = inp->inp_hash.le_next) { 592 if (inp->inp_lport == lport && 593 in_hosteq(inp->inp_laddr, laddr)) 594 goto out; 595 } 596 head = INPCBHASH_BIND(table, zeroin_addr, lport); 597 for (inp = head->lh_first; inp != NULL; inp = inp->inp_hash.le_next) { 598 if (inp->inp_lport == lport && 599 in_hosteq(inp->inp_laddr, zeroin_addr)) 600 goto out; 601 } 602 #ifdef DIAGNOSTIC 603 if (in_pcbnotifymiss) { 604 printf("in_pcblookup_bind: laddr=%08x lport=%d\n", 605 ntohl(laddr.s_addr), ntohs(lport)); 606 } 607 #endif 608 return (0); 609 610 out: 611 /* Move this PCB to the head of hash chain. */ 612 if (inp != head->lh_first) { 613 LIST_REMOVE(inp, inp_hash); 614 LIST_INSERT_HEAD(head, inp, inp_hash); 615 } 616 return (inp); 617 } 618 619 void 620 in_pcbstate(inp, state) 621 struct inpcb *inp; 622 int state; 623 { 624 625 if (inp->inp_state > INP_ATTACHED) 626 LIST_REMOVE(inp, inp_hash); 627 628 switch (state) { 629 case INP_BOUND: 630 LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table, 631 inp->inp_laddr, inp->inp_lport), inp, inp_hash); 632 break; 633 case INP_CONNECTED: 634 LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table, 635 inp->inp_faddr, inp->inp_fport, 636 inp->inp_laddr, inp->inp_lport), inp, inp_hash); 637 break; 638 } 639 640 inp->inp_state = state; 641 } 642