Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)
[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]
FreeBSD/Linux Kernel Cross Reference
sys/netinet/in_pcb.c
Version:
- FREEBSD - FREEBSD-13-STABLE - FREEBSD-13-0 - FREEBSD-12-STABLE - FREEBSD-12-0 - FREEBSD-11-STABLE - FREEBSD-11-0 - FREEBSD-10-STABLE - FREEBSD-10-0 - FREEBSD-9-STABLE - FREEBSD-9-0 - FREEBSD-8-STABLE - FREEBSD-8-0 - FREEBSD-7-STABLE - FREEBSD-7-0 - FREEBSD-6-STABLE - FREEBSD-6-0 - FREEBSD-5-STABLE - FREEBSD-5-0 - FREEBSD-4-STABLE - FREEBSD-3-STABLE - FREEBSD22 - l41 - OPENBSD - linux-2.6 - MK84 - PLAN9 - xnu-8792
SearchContext: - none - 3 - 10
SearchContext: - none - 3 - 10
1 /* 2 * Copyright (c) 1982, 1986, 1991, 1993, 1995 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * @(#)in_pcb.c 8.4 (Berkeley) 5/24/95 34 * $FreeBSD: releng/5.0/sys/netinet/in_pcb.c 106681 2002-11-08 23:50:32Z sam $ 35 */ 36 37 #include "opt_ipsec.h" 38 #include "opt_inet6.h" 39 40 #include <sys/param.h> 41 #include <sys/systm.h> 42 #include <sys/malloc.h> 43 #include <sys/mbuf.h> 44 #include <sys/domain.h> 45 #include <sys/protosw.h> 46 #include <sys/socket.h> 47 #include <sys/socketvar.h> 48 #include <sys/proc.h> 49 #include <sys/jail.h> 50 #include <sys/kernel.h> 51 #include <sys/sysctl.h> 52 53 #include <machine/limits.h> 54 55 #include <vm/uma.h> 56 57 #include <net/if.h> 58 #include <net/if_types.h> 59 #include <net/route.h> 60 61 #include <netinet/in.h> 62 #include <netinet/in_pcb.h> 63 #include <netinet/in_var.h> 64 #include <netinet/ip_var.h> 65 #ifdef INET6 66 #include <netinet/ip6.h> 67 #include <netinet6/ip6_var.h> 68 #endif /* INET6 */ 69 70 #ifdef IPSEC 71 #include <netinet6/ipsec.h> 72 #include <netkey/key.h> 73 #endif /* IPSEC */ 74 75 #ifdef FAST_IPSEC 76 #if defined(IPSEC) || defined(IPSEC_ESP) 77 #error "Bad idea: don't compile with both IPSEC and FAST_IPSEC!" 78 #endif 79 #if defined(INET6) 80 #error "Bad idea: don't use IPv6 with FAST_IPSEC (for the moment)!" 81 #endif 82 83 #include <netipsec/ipsec.h> 84 #include <netipsec/key.h> 85 #define IPSEC 86 #endif /* FAST_IPSEC */ 87 88 struct in_addr zeroin_addr; 89 90 /* 91 * These configure the range of local port addresses assigned to 92 * "unspecified" outgoing connections/packets/whatever. 93 */ 94 int ipport_lowfirstauto = IPPORT_RESERVED - 1; /* 1023 */ 95 int ipport_lowlastauto = IPPORT_RESERVEDSTART; /* 600 */ 96 int ipport_firstauto = IPPORT_HIFIRSTAUTO; /* 49152 */ 97 int ipport_lastauto = IPPORT_HILASTAUTO; /* 65535 */ 98 int ipport_hifirstauto = IPPORT_HIFIRSTAUTO; /* 49152 */ 99 int ipport_hilastauto = IPPORT_HILASTAUTO; /* 65535 */ 100 101 #define RANGECHK(var, min, max) \ 102 if ((var) < (min)) { (var) = (min); } \ 103 else if ((var) > (max)) { (var) = (max); } 104 105 static int 106 sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS) 107 { 108 int error = sysctl_handle_int(oidp, 109 oidp->oid_arg1, oidp->oid_arg2, req); 110 if (!error) { 111 RANGECHK(ipport_lowfirstauto, 1, IPPORT_RESERVED - 1); 112 RANGECHK(ipport_lowlastauto, 1, IPPORT_RESERVED - 1); 113 RANGECHK(ipport_firstauto, IPPORT_RESERVED, USHRT_MAX); 114 RANGECHK(ipport_lastauto, IPPORT_RESERVED, USHRT_MAX); 115 RANGECHK(ipport_hifirstauto, IPPORT_RESERVED, USHRT_MAX); 116 RANGECHK(ipport_hilastauto, IPPORT_RESERVED, USHRT_MAX); 117 } 118 return error; 119 } 120 121 #undef RANGECHK 122 123 SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW, 0, "IP Ports"); 124 125 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW, 126 &ipport_lowfirstauto, 0, &sysctl_net_ipport_check, "I", ""); 127 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW, 128 &ipport_lowlastauto, 0, &sysctl_net_ipport_check, "I", ""); 129 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW, 130 &ipport_firstauto, 0, &sysctl_net_ipport_check, "I", ""); 131 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW, 132 &ipport_lastauto, 0, &sysctl_net_ipport_check, "I", ""); 133 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW, 134 &ipport_hifirstauto, 0, &sysctl_net_ipport_check, "I", ""); 135 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW, 136 &ipport_hilastauto, 0, &sysctl_net_ipport_check, "I", ""); 137 138 /* 139 * in_pcb.c: manage the Protocol Control Blocks. 140 * 141 * NOTE: It is assumed that most of these functions will be called at 142 * splnet(). XXX - There are, unfortunately, a few exceptions to this 143 * rule that should be fixed. 144 */ 145 146 /* 147 * Allocate a PCB and associate it with the socket. 148 */ 149 int 150 in_pcballoc(so, pcbinfo, td) 151 struct socket *so; 152 struct inpcbinfo *pcbinfo; 153 struct thread *td; 154 { 155 register struct inpcb *inp; 156 #ifdef IPSEC 157 int error; 158 #endif 159 160 inp = uma_zalloc(pcbinfo->ipi_zone, M_NOWAIT); 161 if (inp == NULL) 162 return (ENOBUFS); 163 bzero((caddr_t)inp, sizeof(*inp)); 164 inp->inp_gencnt = ++pcbinfo->ipi_gencnt; 165 inp->inp_pcbinfo = pcbinfo; 166 inp->inp_socket = so; 167 #ifdef IPSEC 168 error = ipsec_init_policy(so, &inp->inp_sp); 169 if (error != 0) { 170 uma_zfree(pcbinfo->ipi_zone, inp); 171 return error; 172 } 173 #endif /*IPSEC*/ 174 #if defined(INET6) 175 if (INP_SOCKAF(so) == AF_INET6 && ip6_v6only) 176 inp->inp_flags |= IN6P_IPV6_V6ONLY; 177 #endif 178 LIST_INSERT_HEAD(pcbinfo->listhead, inp, inp_list); 179 pcbinfo->ipi_count++; 180 so->so_pcb = (caddr_t)inp; 181 INP_LOCK_INIT(inp, "inp"); 182 #ifdef INET6 183 if (ip6_auto_flowlabel) 184 inp->inp_flags |= IN6P_AUTOFLOWLABEL; 185 #endif 186 return (0); 187 } 188 189 int 190 in_pcbbind(inp, nam, td) 191 register struct inpcb *inp; 192 struct sockaddr *nam; 193 struct thread *td; 194 { 195 int anonport, error; 196 197 if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY) 198 return (EINVAL); 199 anonport = inp->inp_lport == 0 && (nam == NULL || 200 ((struct sockaddr_in *)nam)->sin_port == 0); 201 error = in_pcbbind_setup(inp, nam, &inp->inp_laddr.s_addr, 202 &inp->inp_lport, td); 203 if (error) 204 return (error); 205 if (in_pcbinshash(inp) != 0) { 206 inp->inp_laddr.s_addr = INADDR_ANY; 207 inp->inp_lport = 0; 208 return (EAGAIN); 209 } 210 if (anonport) 211 inp->inp_flags |= INP_ANONPORT; 212 return (0); 213 } 214 215 /* 216 * Set up a bind operation on a PCB, performing port allocation 217 * as required, but do not actually modify the PCB. Callers can 218 * either complete the bind by setting inp_laddr/inp_lport and 219 * calling in_pcbinshash(), or they can just use the resulting 220 * port and address to authorise the sending of a once-off packet. 221 * 222 * On error, the values of *laddrp and *lportp are not changed. 223 */ 224 int 225 in_pcbbind_setup(inp, nam, laddrp, lportp, td) 226 struct inpcb *inp; 227 struct sockaddr *nam; 228 in_addr_t *laddrp; 229 u_short *lportp; 230 struct thread *td; 231 { 232 struct socket *so = inp->inp_socket; 233 unsigned short *lastport; 234 struct sockaddr_in *sin; 235 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 236 struct in_addr laddr; 237 u_short lport = 0; 238 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 239 int error, prison = 0; 240 241 if (TAILQ_EMPTY(&in_ifaddrhead)) /* XXX broken! */ 242 return (EADDRNOTAVAIL); 243 laddr.s_addr = *laddrp; 244 if (nam != NULL && laddr.s_addr != INADDR_ANY) 245 return (EINVAL); 246 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 247 wild = 1; 248 if (nam) { 249 sin = (struct sockaddr_in *)nam; 250 if (nam->sa_len != sizeof (*sin)) 251 return (EINVAL); 252 #ifdef notdef 253 /* 254 * We should check the family, but old programs 255 * incorrectly fail to initialize it. 256 */ 257 if (sin->sin_family != AF_INET) 258 return (EAFNOSUPPORT); 259 #endif 260 if (sin->sin_addr.s_addr != INADDR_ANY) 261 if (prison_ip(td->td_ucred, 0, &sin->sin_addr.s_addr)) 262 return(EINVAL); 263 if (sin->sin_port != *lportp) { 264 /* Don't allow the port to change. */ 265 if (*lportp != 0) 266 return (EINVAL); 267 lport = sin->sin_port; 268 } 269 /* NB: lport is left as 0 if the port isn't being changed. */ 270 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { 271 /* 272 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 273 * allow complete duplication of binding if 274 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 275 * and a multicast address is bound on both 276 * new and duplicated sockets. 277 */ 278 if (so->so_options & SO_REUSEADDR) 279 reuseport = SO_REUSEADDR|SO_REUSEPORT; 280 } else if (sin->sin_addr.s_addr != INADDR_ANY) { 281 sin->sin_port = 0; /* yech... */ 282 bzero(&sin->sin_zero, sizeof(sin->sin_zero)); 283 if (ifa_ifwithaddr((struct sockaddr *)sin) == 0) 284 return (EADDRNOTAVAIL); 285 } 286 laddr = sin->sin_addr; 287 if (lport) { 288 struct inpcb *t; 289 /* GROSS */ 290 if (ntohs(lport) < IPPORT_RESERVED && td && 291 suser_cred(td->td_ucred, PRISON_ROOT)) 292 return (EACCES); 293 if (td && jailed(td->td_ucred)) 294 prison = 1; 295 if (so->so_cred->cr_uid != 0 && 296 !IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { 297 t = in_pcblookup_local(inp->inp_pcbinfo, 298 sin->sin_addr, lport, 299 prison ? 0 : INPLOOKUP_WILDCARD); 300 if (t && 301 (ntohl(sin->sin_addr.s_addr) != INADDR_ANY || 302 ntohl(t->inp_laddr.s_addr) != INADDR_ANY || 303 (t->inp_socket->so_options & 304 SO_REUSEPORT) == 0) && 305 (so->so_cred->cr_uid != 306 t->inp_socket->so_cred->cr_uid)) { 307 #if defined(INET6) 308 if (ntohl(sin->sin_addr.s_addr) != 309 INADDR_ANY || 310 ntohl(t->inp_laddr.s_addr) != 311 INADDR_ANY || 312 INP_SOCKAF(so) == 313 INP_SOCKAF(t->inp_socket)) 314 #endif /* defined(INET6) */ 315 return (EADDRINUSE); 316 } 317 } 318 if (prison && 319 prison_ip(td->td_ucred, 0, &sin->sin_addr.s_addr)) 320 return (EADDRNOTAVAIL); 321 t = in_pcblookup_local(pcbinfo, sin->sin_addr, 322 lport, prison ? 0 : wild); 323 if (t && 324 (reuseport & t->inp_socket->so_options) == 0) { 325 #if defined(INET6) 326 if (ntohl(sin->sin_addr.s_addr) != 327 INADDR_ANY || 328 ntohl(t->inp_laddr.s_addr) != 329 INADDR_ANY || 330 INP_SOCKAF(so) == 331 INP_SOCKAF(t->inp_socket)) 332 #endif /* defined(INET6) */ 333 return (EADDRINUSE); 334 } 335 } 336 } 337 if (*lportp != 0) 338 lport = *lportp; 339 if (lport == 0) { 340 ushort first, last; 341 int count; 342 343 if (laddr.s_addr != INADDR_ANY) 344 if (prison_ip(td->td_ucred, 0, &laddr.s_addr)) 345 return (EINVAL); 346 347 if (inp->inp_flags & INP_HIGHPORT) { 348 first = ipport_hifirstauto; /* sysctl */ 349 last = ipport_hilastauto; 350 lastport = &pcbinfo->lasthi; 351 } else if (inp->inp_flags & INP_LOWPORT) { 352 if (td && (error = suser_cred(td->td_ucred, 353 PRISON_ROOT)) != 0) 354 return error; 355 first = ipport_lowfirstauto; /* 1023 */ 356 last = ipport_lowlastauto; /* 600 */ 357 lastport = &pcbinfo->lastlow; 358 } else { 359 first = ipport_firstauto; /* sysctl */ 360 last = ipport_lastauto; 361 lastport = &pcbinfo->lastport; 362 } 363 /* 364 * Simple check to ensure all ports are not used up causing 365 * a deadlock here. 366 * 367 * We split the two cases (up and down) so that the direction 368 * is not being tested on each round of the loop. 369 */ 370 if (first > last) { 371 /* 372 * counting down 373 */ 374 count = first - last; 375 376 do { 377 if (count-- < 0) /* completely used? */ 378 return (EADDRNOTAVAIL); 379 --*lastport; 380 if (*lastport > first || *lastport < last) 381 *lastport = first; 382 lport = htons(*lastport); 383 } while (in_pcblookup_local(pcbinfo, laddr, lport, 384 wild)); 385 } else { 386 /* 387 * counting up 388 */ 389 count = last - first; 390 391 do { 392 if (count-- < 0) /* completely used? */ 393 return (EADDRNOTAVAIL); 394 ++*lastport; 395 if (*lastport < first || *lastport > last) 396 *lastport = first; 397 lport = htons(*lastport); 398 } while (in_pcblookup_local(pcbinfo, laddr, lport, 399 wild)); 400 } 401 } 402 if (prison_ip(td->td_ucred, 0, &laddr.s_addr)) 403 return (EINVAL); 404 *laddrp = laddr.s_addr; 405 *lportp = lport; 406 return (0); 407 } 408 409 /* 410 * Connect from a socket to a specified address. 411 * Both address and port must be specified in argument sin. 412 * If don't have a local address for this socket yet, 413 * then pick one. 414 */ 415 int 416 in_pcbconnect(inp, nam, td) 417 register struct inpcb *inp; 418 struct sockaddr *nam; 419 struct thread *td; 420 { 421 u_short lport, fport; 422 in_addr_t laddr, faddr; 423 int anonport, error; 424 425 lport = inp->inp_lport; 426 laddr = inp->inp_laddr.s_addr; 427 anonport = (lport == 0); 428 error = in_pcbconnect_setup(inp, nam, &laddr, &lport, &faddr, &fport, 429 NULL, td); 430 if (error) 431 return (error); 432 433 /* Do the initial binding of the local address if required. */ 434 if (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0) { 435 inp->inp_lport = lport; 436 inp->inp_laddr.s_addr = laddr; 437 if (in_pcbinshash(inp) != 0) { 438 inp->inp_laddr.s_addr = INADDR_ANY; 439 inp->inp_lport = 0; 440 return (EAGAIN); 441 } 442 } 443 444 /* Commit the remaining changes. */ 445 inp->inp_lport = lport; 446 inp->inp_laddr.s_addr = laddr; 447 inp->inp_faddr.s_addr = faddr; 448 inp->inp_fport = fport; 449 in_pcbrehash(inp); 450 if (anonport) 451 inp->inp_flags |= INP_ANONPORT; 452 return (0); 453 } 454 455 /* 456 * Set up for a connect from a socket to the specified address. 457 * On entry, *laddrp and *lportp should contain the current local 458 * address and port for the PCB; these are updated to the values 459 * that should be placed in inp_laddr and inp_lport to complete 460 * the connect. 461 * 462 * On success, *faddrp and *fportp will be set to the remote address 463 * and port. These are not updated in the error case. 464 * 465 * If the operation fails because the connection already exists, 466 * *oinpp will be set to the PCB of that connection so that the 467 * caller can decide to override it. In all other cases, *oinpp 468 * is set to NULL. 469 */ 470 int 471 in_pcbconnect_setup(inp, nam, laddrp, lportp, faddrp, fportp, oinpp, td) 472 register struct inpcb *inp; 473 struct sockaddr *nam; 474 in_addr_t *laddrp; 475 u_short *lportp; 476 in_addr_t *faddrp; 477 u_short *fportp; 478 struct inpcb **oinpp; 479 struct thread *td; 480 { 481 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 482 struct in_ifaddr *ia; 483 struct sockaddr_in sa; 484 struct ucred *cred; 485 struct inpcb *oinp; 486 struct in_addr laddr, faddr; 487 u_short lport, fport; 488 int error; 489 490 if (oinpp != NULL) 491 *oinpp = NULL; 492 if (nam->sa_len != sizeof (*sin)) 493 return (EINVAL); 494 if (sin->sin_family != AF_INET) 495 return (EAFNOSUPPORT); 496 if (sin->sin_port == 0) 497 return (EADDRNOTAVAIL); 498 laddr.s_addr = *laddrp; 499 lport = *lportp; 500 faddr = sin->sin_addr; 501 fport = sin->sin_port; 502 cred = inp->inp_socket->so_cred; 503 if (laddr.s_addr == INADDR_ANY && jailed(cred)) { 504 bzero(&sa, sizeof(sa)); 505 sa.sin_addr.s_addr = htonl(prison_getip(cred)); 506 sa.sin_len = sizeof(sa); 507 sa.sin_family = AF_INET; 508 error = in_pcbbind_setup(inp, (struct sockaddr *)&sa, 509 &laddr.s_addr, &lport, td); 510 if (error) 511 return (error); 512 } 513 514 if (!TAILQ_EMPTY(&in_ifaddrhead)) { 515 /* 516 * If the destination address is INADDR_ANY, 517 * use the primary local address. 518 * If the supplied address is INADDR_BROADCAST, 519 * and the primary interface supports broadcast, 520 * choose the broadcast address for that interface. 521 */ 522 if (faddr.s_addr == INADDR_ANY) 523 faddr = IA_SIN(TAILQ_FIRST(&in_ifaddrhead))->sin_addr; 524 else if (faddr.s_addr == (u_long)INADDR_BROADCAST && 525 (TAILQ_FIRST(&in_ifaddrhead)->ia_ifp->if_flags & 526 IFF_BROADCAST)) 527 faddr = satosin(&TAILQ_FIRST( 528 &in_ifaddrhead)->ia_broadaddr)->sin_addr; 529 } 530 if (laddr.s_addr == INADDR_ANY) { 531 register struct route *ro; 532 533 ia = (struct in_ifaddr *)0; 534 /* 535 * If route is known or can be allocated now, 536 * our src addr is taken from the i/f, else punt. 537 * Note that we should check the address family of the cached 538 * destination, in case of sharing the cache with IPv6. 539 */ 540 ro = &inp->inp_route; 541 if (ro->ro_rt && 542 (ro->ro_dst.sa_family != AF_INET || 543 satosin(&ro->ro_dst)->sin_addr.s_addr != faddr.s_addr || 544 inp->inp_socket->so_options & SO_DONTROUTE)) { 545 RTFREE(ro->ro_rt); 546 ro->ro_rt = (struct rtentry *)0; 547 } 548 if ((inp->inp_socket->so_options & SO_DONTROUTE) == 0 && /*XXX*/ 549 (ro->ro_rt == (struct rtentry *)0 || 550 ro->ro_rt->rt_ifp == (struct ifnet *)0)) { 551 /* No route yet, so try to acquire one */ 552 bzero(&ro->ro_dst, sizeof(struct sockaddr_in)); 553 ro->ro_dst.sa_family = AF_INET; 554 ro->ro_dst.sa_len = sizeof(struct sockaddr_in); 555 ((struct sockaddr_in *)&ro->ro_dst)->sin_addr = faddr; 556 rtalloc(ro); 557 } 558 /* 559 * If we found a route, use the address 560 * corresponding to the outgoing interface 561 * unless it is the loopback (in case a route 562 * to our address on another net goes to loopback). 563 */ 564 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) 565 ia = ifatoia(ro->ro_rt->rt_ifa); 566 if (ia == 0) { 567 bzero(&sa, sizeof(sa)); 568 sa.sin_addr = faddr; 569 sa.sin_len = sizeof(sa); 570 sa.sin_family = AF_INET; 571 572 ia = ifatoia(ifa_ifwithdstaddr(sintosa(&sa))); 573 if (ia == 0) 574 ia = ifatoia(ifa_ifwithnet(sintosa(&sa))); 575 if (ia == 0) 576 ia = TAILQ_FIRST(&in_ifaddrhead); 577 if (ia == 0) 578 return (EADDRNOTAVAIL); 579 } 580 /* 581 * If the destination address is multicast and an outgoing 582 * interface has been set as a multicast option, use the 583 * address of that interface as our source address. 584 */ 585 if (IN_MULTICAST(ntohl(faddr.s_addr)) && 586 inp->inp_moptions != NULL) { 587 struct ip_moptions *imo; 588 struct ifnet *ifp; 589 590 imo = inp->inp_moptions; 591 if (imo->imo_multicast_ifp != NULL) { 592 ifp = imo->imo_multicast_ifp; 593 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) 594 if (ia->ia_ifp == ifp) 595 break; 596 if (ia == 0) 597 return (EADDRNOTAVAIL); 598 } 599 } 600 laddr = ia->ia_addr.sin_addr; 601 } 602 603 oinp = in_pcblookup_hash(inp->inp_pcbinfo, faddr, fport, laddr, lport, 604 0, NULL); 605 if (oinp != NULL) { 606 if (oinpp != NULL) 607 *oinpp = oinp; 608 return (EADDRINUSE); 609 } 610 if (lport == 0) { 611 error = in_pcbbind_setup(inp, NULL, &laddr.s_addr, &lport, td); 612 if (error) 613 return (error); 614 } 615 *laddrp = laddr.s_addr; 616 *lportp = lport; 617 *faddrp = faddr.s_addr; 618 *fportp = fport; 619 return (0); 620 } 621 622 void 623 in_pcbdisconnect(inp) 624 struct inpcb *inp; 625 { 626 627 inp->inp_faddr.s_addr = INADDR_ANY; 628 inp->inp_fport = 0; 629 in_pcbrehash(inp); 630 if (inp->inp_socket->so_state & SS_NOFDREF) 631 in_pcbdetach(inp); 632 } 633 634 void 635 in_pcbdetach(inp) 636 struct inpcb *inp; 637 { 638 struct socket *so = inp->inp_socket; 639 struct inpcbinfo *ipi = inp->inp_pcbinfo; 640 641 #ifdef IPSEC 642 ipsec4_delete_pcbpolicy(inp); 643 #endif /*IPSEC*/ 644 inp->inp_gencnt = ++ipi->ipi_gencnt; 645 in_pcbremlists(inp); 646 so->so_pcb = 0; 647 sotryfree(so); 648 if (inp->inp_options) 649 (void)m_free(inp->inp_options); 650 if (inp->inp_route.ro_rt) 651 rtfree(inp->inp_route.ro_rt); 652 ip_freemoptions(inp->inp_moptions); 653 inp->inp_vflag = 0; 654 INP_LOCK_DESTROY(inp); 655 uma_zfree(ipi->ipi_zone, inp); 656 } 657 658 struct sockaddr * 659 in_sockaddr(port, addr_p) 660 in_port_t port; 661 struct in_addr *addr_p; 662 { 663 struct sockaddr_in *sin; 664 665 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, 666 M_WAITOK | M_ZERO); 667 sin->sin_family = AF_INET; 668 sin->sin_len = sizeof(*sin); 669 sin->sin_addr = *addr_p; 670 sin->sin_port = port; 671 672 return (struct sockaddr *)sin; 673 } 674 675 /* 676 * The wrapper function will pass down the pcbinfo for this function to lock. 677 * The socket must have a valid 678 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one 679 * except through a kernel programming error, so it is acceptable to panic 680 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap 681 * because there actually /is/ a programming error somewhere... XXX) 682 */ 683 int 684 in_setsockaddr(so, nam, pcbinfo) 685 struct socket *so; 686 struct sockaddr **nam; 687 struct inpcbinfo *pcbinfo; 688 { 689 int s; 690 register struct inpcb *inp; 691 struct in_addr addr; 692 in_port_t port; 693 694 s = splnet(); 695 INP_INFO_RLOCK(pcbinfo); 696 inp = sotoinpcb(so); 697 if (!inp) { 698 INP_INFO_RUNLOCK(pcbinfo); 699 splx(s); 700 return ECONNRESET; 701 } 702 INP_LOCK(inp); 703 port = inp->inp_lport; 704 addr = inp->inp_laddr; 705 INP_UNLOCK(inp); 706 INP_INFO_RUNLOCK(pcbinfo); 707 splx(s); 708 709 *nam = in_sockaddr(port, &addr); 710 return 0; 711 } 712 713 /* 714 * The wrapper function will pass down the pcbinfo for this function to lock. 715 */ 716 int 717 in_setpeeraddr(so, nam, pcbinfo) 718 struct socket *so; 719 struct sockaddr **nam; 720 struct inpcbinfo *pcbinfo; 721 { 722 int s; 723 register struct inpcb *inp; 724 struct in_addr addr; 725 in_port_t port; 726 727 s = splnet(); 728 INP_INFO_RLOCK(pcbinfo); 729 inp = sotoinpcb(so); 730 if (!inp) { 731 INP_INFO_RUNLOCK(pcbinfo); 732 splx(s); 733 return ECONNRESET; 734 } 735 INP_LOCK(inp); 736 port = inp->inp_fport; 737 addr = inp->inp_faddr; 738 INP_UNLOCK(inp); 739 INP_INFO_RUNLOCK(pcbinfo); 740 splx(s); 741 742 *nam = in_sockaddr(port, &addr); 743 return 0; 744 } 745 746 void 747 in_pcbnotifyall(pcbinfo, faddr, errno, notify) 748 struct inpcbinfo *pcbinfo; 749 struct in_addr faddr; 750 int errno; 751 struct inpcb *(*notify)(struct inpcb *, int); 752 { 753 struct inpcb *inp, *ninp; 754 struct inpcbhead *head; 755 int s; 756 757 s = splnet(); 758 INP_INFO_RLOCK(pcbinfo); 759 head = pcbinfo->listhead; 760 for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) { 761 INP_LOCK(inp); 762 ninp = LIST_NEXT(inp, inp_list); 763 #ifdef INET6 764 if ((inp->inp_vflag & INP_IPV4) == 0) { 765 INP_UNLOCK(inp); 766 continue; 767 } 768 #endif 769 if (inp->inp_faddr.s_addr != faddr.s_addr || 770 inp->inp_socket == NULL) { 771 INP_UNLOCK(inp); 772 continue; 773 } 774 (*notify)(inp, errno); 775 INP_UNLOCK(inp); 776 } 777 INP_INFO_RUNLOCK(pcbinfo); 778 splx(s); 779 } 780 781 void 782 in_pcbpurgeif0(pcbinfo, ifp) 783 struct inpcbinfo *pcbinfo; 784 struct ifnet *ifp; 785 { 786 struct inpcb *inp; 787 struct ip_moptions *imo; 788 int i, gap; 789 790 /* why no splnet here? XXX */ 791 INP_INFO_RLOCK(pcbinfo); 792 LIST_FOREACH(inp, pcbinfo->listhead, inp_list) { 793 INP_LOCK(inp); 794 imo = inp->inp_moptions; 795 if ((inp->inp_vflag & INP_IPV4) && 796 imo != NULL) { 797 /* 798 * Unselect the outgoing interface if it is being 799 * detached. 800 */ 801 if (imo->imo_multicast_ifp == ifp) 802 imo->imo_multicast_ifp = NULL; 803 804 /* 805 * Drop multicast group membership if we joined 806 * through the interface being detached. 807 */ 808 for (i = 0, gap = 0; i < imo->imo_num_memberships; 809 i++) { 810 if (imo->imo_membership[i]->inm_ifp == ifp) { 811 in_delmulti(imo->imo_membership[i]); 812 gap++; 813 } else if (gap != 0) 814 imo->imo_membership[i - gap] = 815 imo->imo_membership[i]; 816 } 817 imo->imo_num_memberships -= gap; 818 } 819 INP_UNLOCK(inp); 820 } 821 INP_INFO_RUNLOCK(pcbinfo); 822 } 823 824 /* 825 * Check for alternatives when higher level complains 826 * about service problems. For now, invalidate cached 827 * routing information. If the route was created dynamically 828 * (by a redirect), time to try a default gateway again. 829 */ 830 void 831 in_losing(inp) 832 struct inpcb *inp; 833 { 834 register struct rtentry *rt; 835 struct rt_addrinfo info; 836 837 if ((rt = inp->inp_route.ro_rt)) { 838 bzero((caddr_t)&info, sizeof(info)); 839 info.rti_flags = rt->rt_flags; 840 info.rti_info[RTAX_DST] = rt_key(rt); 841 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 842 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 843 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 844 if (rt->rt_flags & RTF_DYNAMIC) 845 (void) rtrequest1(RTM_DELETE, &info, NULL); 846 inp->inp_route.ro_rt = NULL; 847 rtfree(rt); 848 /* 849 * A new route can be allocated 850 * the next time output is attempted. 851 */ 852 } 853 } 854 855 /* 856 * After a routing change, flush old routing 857 * and allocate a (hopefully) better one. 858 */ 859 struct inpcb * 860 in_rtchange(inp, errno) 861 register struct inpcb *inp; 862 int errno; 863 { 864 if (inp->inp_route.ro_rt) { 865 rtfree(inp->inp_route.ro_rt); 866 inp->inp_route.ro_rt = 0; 867 /* 868 * A new route can be allocated the next time 869 * output is attempted. 870 */ 871 } 872 return inp; 873 } 874 875 /* 876 * Lookup a PCB based on the local address and port. 877 */ 878 struct inpcb * 879 in_pcblookup_local(pcbinfo, laddr, lport_arg, wild_okay) 880 struct inpcbinfo *pcbinfo; 881 struct in_addr laddr; 882 u_int lport_arg; 883 int wild_okay; 884 { 885 register struct inpcb *inp; 886 int matchwild = 3, wildcard; 887 u_short lport = lport_arg; 888 889 if (!wild_okay) { 890 struct inpcbhead *head; 891 /* 892 * Look for an unconnected (wildcard foreign addr) PCB that 893 * matches the local address and port we're looking for. 894 */ 895 head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0, pcbinfo->hashmask)]; 896 LIST_FOREACH(inp, head, inp_hash) { 897 #ifdef INET6 898 if ((inp->inp_vflag & INP_IPV4) == 0) 899 continue; 900 #endif 901 if (inp->inp_faddr.s_addr == INADDR_ANY && 902 inp->inp_laddr.s_addr == laddr.s_addr && 903 inp->inp_lport == lport) { 904 /* 905 * Found. 906 */ 907 return (inp); 908 } 909 } 910 /* 911 * Not found. 912 */ 913 return (NULL); 914 } else { 915 struct inpcbporthead *porthash; 916 struct inpcbport *phd; 917 struct inpcb *match = NULL; 918 /* 919 * Best fit PCB lookup. 920 * 921 * First see if this local port is in use by looking on the 922 * port hash list. 923 */ 924 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport, 925 pcbinfo->porthashmask)]; 926 LIST_FOREACH(phd, porthash, phd_hash) { 927 if (phd->phd_port == lport) 928 break; 929 } 930 if (phd != NULL) { 931 /* 932 * Port is in use by one or more PCBs. Look for best 933 * fit. 934 */ 935 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 936 wildcard = 0; 937 #ifdef INET6 938 if ((inp->inp_vflag & INP_IPV4) == 0) 939 continue; 940 #endif 941 if (inp->inp_faddr.s_addr != INADDR_ANY) 942 wildcard++; 943 if (inp->inp_laddr.s_addr != INADDR_ANY) { 944 if (laddr.s_addr == INADDR_ANY) 945 wildcard++; 946 else if (inp->inp_laddr.s_addr != laddr.s_addr) 947 continue; 948 } else { 949 if (laddr.s_addr != INADDR_ANY) 950 wildcard++; 951 } 952 if (wildcard < matchwild) { 953 match = inp; 954 matchwild = wildcard; 955 if (matchwild == 0) { 956 break; 957 } 958 } 959 } 960 } 961 return (match); 962 } 963 } 964 965 /* 966 * Lookup PCB in hash list. 967 */ 968 struct inpcb * 969 in_pcblookup_hash(pcbinfo, faddr, fport_arg, laddr, lport_arg, wildcard, 970 ifp) 971 struct inpcbinfo *pcbinfo; 972 struct in_addr faddr, laddr; 973 u_int fport_arg, lport_arg; 974 int wildcard; 975 struct ifnet *ifp; 976 { 977 struct inpcbhead *head; 978 register struct inpcb *inp; 979 u_short fport = fport_arg, lport = lport_arg; 980 981 /* 982 * First look for an exact match. 983 */ 984 head = &pcbinfo->hashbase[INP_PCBHASH(faddr.s_addr, lport, fport, pcbinfo->hashmask)]; 985 LIST_FOREACH(inp, head, inp_hash) { 986 #ifdef INET6 987 if ((inp->inp_vflag & INP_IPV4) == 0) 988 continue; 989 #endif 990 if (inp->inp_faddr.s_addr == faddr.s_addr && 991 inp->inp_laddr.s_addr == laddr.s_addr && 992 inp->inp_fport == fport && 993 inp->inp_lport == lport) { 994 /* 995 * Found. 996 */ 997 return (inp); 998 } 999 } 1000 if (wildcard) { 1001 struct inpcb *local_wild = NULL; 1002 #if defined(INET6) 1003 struct inpcb *local_wild_mapped = NULL; 1004 #endif /* defined(INET6) */ 1005 1006 head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0, pcbinfo->hashmask)]; 1007 LIST_FOREACH(inp, head, inp_hash) { 1008 #ifdef INET6 1009 if ((inp->inp_vflag & INP_IPV4) == 0) 1010 continue; 1011 #endif 1012 if (inp->inp_faddr.s_addr == INADDR_ANY && 1013 inp->inp_lport == lport) { 1014 if (ifp && ifp->if_type == IFT_FAITH && 1015 (inp->inp_flags & INP_FAITH) == 0) 1016 continue; 1017 if (inp->inp_laddr.s_addr == laddr.s_addr) 1018 return (inp); 1019 else if (inp->inp_laddr.s_addr == INADDR_ANY) { 1020 #if defined(INET6) 1021 if (INP_CHECK_SOCKAF(inp->inp_socket, 1022 AF_INET6)) 1023 local_wild_mapped = inp; 1024 else 1025 #endif /* defined(INET6) */ 1026 local_wild = inp; 1027 } 1028 } 1029 } 1030 #if defined(INET6) 1031 if (local_wild == NULL) 1032 return (local_wild_mapped); 1033 #endif /* defined(INET6) */ 1034 return (local_wild); 1035 } 1036 1037 /* 1038 * Not found. 1039 */ 1040 return (NULL); 1041 } 1042 1043 /* 1044 * Insert PCB onto various hash lists. 1045 */ 1046 int 1047 in_pcbinshash(inp) 1048 struct inpcb *inp; 1049 { 1050 struct inpcbhead *pcbhash; 1051 struct inpcbporthead *pcbporthash; 1052 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1053 struct inpcbport *phd; 1054 u_int32_t hashkey_faddr; 1055 1056 #ifdef INET6 1057 if (inp->inp_vflag & INP_IPV6) 1058 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX */; 1059 else 1060 #endif /* INET6 */ 1061 hashkey_faddr = inp->inp_faddr.s_addr; 1062 1063 pcbhash = &pcbinfo->hashbase[INP_PCBHASH(hashkey_faddr, 1064 inp->inp_lport, inp->inp_fport, pcbinfo->hashmask)]; 1065 1066 pcbporthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(inp->inp_lport, 1067 pcbinfo->porthashmask)]; 1068 1069 /* 1070 * Go through port list and look for a head for this lport. 1071 */ 1072 LIST_FOREACH(phd, pcbporthash, phd_hash) { 1073 if (phd->phd_port == inp->inp_lport) 1074 break; 1075 } 1076 /* 1077 * If none exists, malloc one and tack it on. 1078 */ 1079 if (phd == NULL) { 1080 MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport), M_PCB, M_NOWAIT); 1081 if (phd == NULL) { 1082 return (ENOBUFS); /* XXX */ 1083 } 1084 phd->phd_port = inp->inp_lport; 1085 LIST_INIT(&phd->phd_pcblist); 1086 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash); 1087 } 1088 inp->inp_phd = phd; 1089 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist); 1090 LIST_INSERT_HEAD(pcbhash, inp, inp_hash); 1091 return (0); 1092 } 1093 1094 /* 1095 * Move PCB to the proper hash bucket when { faddr, fport } have been 1096 * changed. NOTE: This does not handle the case of the lport changing (the 1097 * hashed port list would have to be updated as well), so the lport must 1098 * not change after in_pcbinshash() has been called. 1099 */ 1100 void 1101 in_pcbrehash(inp) 1102 struct inpcb *inp; 1103 { 1104 struct inpcbhead *head; 1105 u_int32_t hashkey_faddr; 1106 1107 #ifdef INET6 1108 if (inp->inp_vflag & INP_IPV6) 1109 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX */; 1110 else 1111 #endif /* INET6 */ 1112 hashkey_faddr = inp->inp_faddr.s_addr; 1113 1114 head = &inp->inp_pcbinfo->hashbase[INP_PCBHASH(hashkey_faddr, 1115 inp->inp_lport, inp->inp_fport, inp->inp_pcbinfo->hashmask)]; 1116 1117 LIST_REMOVE(inp, inp_hash); 1118 LIST_INSERT_HEAD(head, inp, inp_hash); 1119 } 1120 1121 /* 1122 * Remove PCB from various lists. 1123 */ 1124 void 1125 in_pcbremlists(inp) 1126 struct inpcb *inp; 1127 { 1128 inp->inp_gencnt = ++inp->inp_pcbinfo->ipi_gencnt; 1129 if (inp->inp_lport) { 1130 struct inpcbport *phd = inp->inp_phd; 1131 1132 LIST_REMOVE(inp, inp_hash); 1133 LIST_REMOVE(inp, inp_portlist); 1134 if (LIST_FIRST(&phd->phd_pcblist) == NULL) { 1135 LIST_REMOVE(phd, phd_hash); 1136 free(phd, M_PCB); 1137 } 1138 } 1139 LIST_REMOVE(inp, inp_list); 1140 inp->inp_pcbinfo->ipi_count--; 1141 } 1142 1143 int 1144 prison_xinpcb(struct thread *td, struct inpcb *inp) 1145 { 1146 if (!jailed(td->td_ucred)) 1147 return (0); 1148 if (ntohl(inp->inp_laddr.s_addr) == prison_getip(td->td_ucred)) 1149 return (0); 1150 return (1); 1151 }
Cache object: 39070d0a284121ee49de33a0b4fc7a4c
[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]