1 /*- 2 * Copyright (c) 1982, 1986, 1991, 1993, 1995 3 * The Regents of the University of California. 4 * Copyright (c) 2007 Robert N. M. Watson 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 * 4. Neither the name of the University 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 REGENTS 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 REGENTS 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 * @(#)in_pcb.c 8.4 (Berkeley) 5/24/95 32 */ 33 34 #include <sys/cdefs.h> 35 __FBSDID("$FreeBSD$"); 36 37 #include "opt_ddb.h" 38 #include "opt_ipsec.h" 39 #include "opt_inet6.h" 40 #include "opt_mac.h" 41 42 #include <sys/param.h> 43 #include <sys/systm.h> 44 #include <sys/malloc.h> 45 #include <sys/mbuf.h> 46 #include <sys/domain.h> 47 #include <sys/protosw.h> 48 #include <sys/socket.h> 49 #include <sys/socketvar.h> 50 #include <sys/priv.h> 51 #include <sys/proc.h> 52 #include <sys/jail.h> 53 #include <sys/kernel.h> 54 #include <sys/sysctl.h> 55 56 #ifdef DDB 57 #include <ddb/ddb.h> 58 #endif 59 60 #include <vm/uma.h> 61 62 #include <net/if.h> 63 #include <net/if_types.h> 64 #include <net/route.h> 65 66 #include <netinet/in.h> 67 #include <netinet/in_pcb.h> 68 #include <netinet/in_var.h> 69 #include <netinet/ip_var.h> 70 #include <netinet/tcp_var.h> 71 #include <netinet/udp.h> 72 #include <netinet/udp_var.h> 73 #ifdef INET6 74 #include <netinet/ip6.h> 75 #include <netinet6/ip6_var.h> 76 #endif /* INET6 */ 77 78 79 #ifdef IPSEC 80 #include <netipsec/ipsec.h> 81 #include <netipsec/key.h> 82 #endif /* IPSEC */ 83 84 #include <security/mac/mac_framework.h> 85 86 /* 87 * These configure the range of local port addresses assigned to 88 * "unspecified" outgoing connections/packets/whatever. 89 */ 90 int ipport_lowfirstauto = IPPORT_RESERVED - 1; /* 1023 */ 91 int ipport_lowlastauto = IPPORT_RESERVEDSTART; /* 600 */ 92 int ipport_firstauto = IPPORT_EPHEMERALFIRST; /* 10000 */ 93 int ipport_lastauto = IPPORT_EPHEMERALLAST; /* 65535 */ 94 int ipport_hifirstauto = IPPORT_HIFIRSTAUTO; /* 49152 */ 95 int ipport_hilastauto = IPPORT_HILASTAUTO; /* 65535 */ 96 97 /* 98 * Reserved ports accessible only to root. There are significant 99 * security considerations that must be accounted for when changing these, 100 * but the security benefits can be great. Please be careful. 101 */ 102 int ipport_reservedhigh = IPPORT_RESERVED - 1; /* 1023 */ 103 int ipport_reservedlow = 0; 104 105 /* Variables dealing with random ephemeral port allocation. */ 106 int ipport_randomized = 1; /* user controlled via sysctl */ 107 int ipport_randomcps = 10; /* user controlled via sysctl */ 108 int ipport_randomtime = 45; /* user controlled via sysctl */ 109 int ipport_stoprandom = 0; /* toggled by ipport_tick */ 110 int ipport_tcpallocs; 111 int ipport_tcplastcount; 112 113 #define RANGECHK(var, min, max) \ 114 if ((var) < (min)) { (var) = (min); } \ 115 else if ((var) > (max)) { (var) = (max); } 116 117 static int 118 sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS) 119 { 120 int error; 121 122 error = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req); 123 if (error == 0) { 124 RANGECHK(ipport_lowfirstauto, 1, IPPORT_RESERVED - 1); 125 RANGECHK(ipport_lowlastauto, 1, IPPORT_RESERVED - 1); 126 RANGECHK(ipport_firstauto, IPPORT_RESERVED, IPPORT_MAX); 127 RANGECHK(ipport_lastauto, IPPORT_RESERVED, IPPORT_MAX); 128 RANGECHK(ipport_hifirstauto, IPPORT_RESERVED, IPPORT_MAX); 129 RANGECHK(ipport_hilastauto, IPPORT_RESERVED, IPPORT_MAX); 130 } 131 return (error); 132 } 133 134 #undef RANGECHK 135 136 SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW, 0, "IP Ports"); 137 138 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW, 139 &ipport_lowfirstauto, 0, &sysctl_net_ipport_check, "I", ""); 140 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW, 141 &ipport_lowlastauto, 0, &sysctl_net_ipport_check, "I", ""); 142 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW, 143 &ipport_firstauto, 0, &sysctl_net_ipport_check, "I", ""); 144 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW, 145 &ipport_lastauto, 0, &sysctl_net_ipport_check, "I", ""); 146 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW, 147 &ipport_hifirstauto, 0, &sysctl_net_ipport_check, "I", ""); 148 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW, 149 &ipport_hilastauto, 0, &sysctl_net_ipport_check, "I", ""); 150 SYSCTL_INT(_net_inet_ip_portrange, OID_AUTO, reservedhigh, 151 CTLFLAG_RW|CTLFLAG_SECURE, &ipport_reservedhigh, 0, ""); 152 SYSCTL_INT(_net_inet_ip_portrange, OID_AUTO, reservedlow, 153 CTLFLAG_RW|CTLFLAG_SECURE, &ipport_reservedlow, 0, ""); 154 SYSCTL_INT(_net_inet_ip_portrange, OID_AUTO, randomized, CTLFLAG_RW, 155 &ipport_randomized, 0, "Enable random port allocation"); 156 SYSCTL_INT(_net_inet_ip_portrange, OID_AUTO, randomcps, CTLFLAG_RW, 157 &ipport_randomcps, 0, "Maximum number of random port " 158 "allocations before switching to a sequental one"); 159 SYSCTL_INT(_net_inet_ip_portrange, OID_AUTO, randomtime, CTLFLAG_RW, 160 &ipport_randomtime, 0, "Minimum time to keep sequental port " 161 "allocation before switching to a random one"); 162 163 /* 164 * in_pcb.c: manage the Protocol Control Blocks. 165 * 166 * NOTE: It is assumed that most of these functions will be called with 167 * the pcbinfo lock held, and often, the inpcb lock held, as these utility 168 * functions often modify hash chains or addresses in pcbs. 169 */ 170 171 /* 172 * Allocate a PCB and associate it with the socket. 173 * On success return with the PCB locked. 174 */ 175 int 176 in_pcballoc(struct socket *so, struct inpcbinfo *pcbinfo) 177 { 178 struct inpcb *inp; 179 int error; 180 181 INP_INFO_WLOCK_ASSERT(pcbinfo); 182 error = 0; 183 inp = uma_zalloc(pcbinfo->ipi_zone, M_NOWAIT); 184 if (inp == NULL) 185 return (ENOBUFS); 186 bzero(inp, inp_zero_size); 187 inp->inp_pcbinfo = pcbinfo; 188 inp->inp_socket = so; 189 inp->inp_inc.inc_fibnum = so->so_fibnum; 190 #ifdef MAC 191 error = mac_inpcb_init(inp, M_NOWAIT); 192 if (error != 0) 193 goto out; 194 SOCK_LOCK(so); 195 mac_inpcb_create(so, inp); 196 SOCK_UNLOCK(so); 197 #endif 198 199 #ifdef IPSEC 200 error = ipsec_init_policy(so, &inp->inp_sp); 201 if (error != 0) { 202 #ifdef MAC 203 mac_inpcb_destroy(inp); 204 #endif 205 goto out; 206 } 207 #endif /*IPSEC*/ 208 #ifdef INET6 209 if (INP_SOCKAF(so) == AF_INET6) { 210 inp->inp_vflag |= INP_IPV6PROTO; 211 if (ip6_v6only) 212 inp->inp_flags |= IN6P_IPV6_V6ONLY; 213 } 214 #endif 215 LIST_INSERT_HEAD(pcbinfo->ipi_listhead, inp, inp_list); 216 pcbinfo->ipi_count++; 217 so->so_pcb = (caddr_t)inp; 218 #ifdef INET6 219 if (ip6_auto_flowlabel) 220 inp->inp_flags |= IN6P_AUTOFLOWLABEL; 221 #endif 222 INP_WLOCK(inp); 223 inp->inp_gencnt = ++pcbinfo->ipi_gencnt; 224 225 #if defined(IPSEC) || defined(MAC) 226 out: 227 if (error != 0) 228 uma_zfree(pcbinfo->ipi_zone, inp); 229 #endif 230 return (error); 231 } 232 233 int 234 in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred) 235 { 236 int anonport, error; 237 238 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 239 INP_WLOCK_ASSERT(inp); 240 241 if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY) 242 return (EINVAL); 243 anonport = inp->inp_lport == 0 && (nam == NULL || 244 ((struct sockaddr_in *)nam)->sin_port == 0); 245 error = in_pcbbind_setup(inp, nam, &inp->inp_laddr.s_addr, 246 &inp->inp_lport, cred); 247 if (error) 248 return (error); 249 if (in_pcbinshash(inp) != 0) { 250 inp->inp_laddr.s_addr = INADDR_ANY; 251 inp->inp_lport = 0; 252 return (EAGAIN); 253 } 254 if (anonport) 255 inp->inp_flags |= INP_ANONPORT; 256 return (0); 257 } 258 259 /* 260 * Set up a bind operation on a PCB, performing port allocation 261 * as required, but do not actually modify the PCB. Callers can 262 * either complete the bind by setting inp_laddr/inp_lport and 263 * calling in_pcbinshash(), or they can just use the resulting 264 * port and address to authorise the sending of a once-off packet. 265 * 266 * On error, the values of *laddrp and *lportp are not changed. 267 */ 268 int 269 in_pcbbind_setup(struct inpcb *inp, struct sockaddr *nam, in_addr_t *laddrp, 270 u_short *lportp, struct ucred *cred) 271 { 272 struct socket *so = inp->inp_socket; 273 unsigned short *lastport; 274 struct sockaddr_in *sin; 275 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 276 struct in_addr laddr; 277 u_short lport = 0; 278 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 279 int error, prison = 0; 280 int dorandom; 281 282 /* 283 * Because no actual state changes occur here, a global write lock on 284 * the pcbinfo isn't required. 285 */ 286 INP_INFO_LOCK_ASSERT(pcbinfo); 287 INP_LOCK_ASSERT(inp); 288 289 if (TAILQ_EMPTY(&in_ifaddrhead)) /* XXX broken! */ 290 return (EADDRNOTAVAIL); 291 laddr.s_addr = *laddrp; 292 if (nam != NULL && laddr.s_addr != INADDR_ANY) 293 return (EINVAL); 294 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 295 wild = INPLOOKUP_WILDCARD; 296 if (nam) { 297 sin = (struct sockaddr_in *)nam; 298 if (nam->sa_len != sizeof (*sin)) 299 return (EINVAL); 300 #ifdef notdef 301 /* 302 * We should check the family, but old programs 303 * incorrectly fail to initialize it. 304 */ 305 if (sin->sin_family != AF_INET) 306 return (EAFNOSUPPORT); 307 #endif 308 if (sin->sin_addr.s_addr != INADDR_ANY) 309 if (prison_ip(cred, 0, &sin->sin_addr.s_addr)) 310 return(EINVAL); 311 if (sin->sin_port != *lportp) { 312 /* Don't allow the port to change. */ 313 if (*lportp != 0) 314 return (EINVAL); 315 lport = sin->sin_port; 316 } 317 /* NB: lport is left as 0 if the port isn't being changed. */ 318 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { 319 /* 320 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 321 * allow complete duplication of binding if 322 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 323 * and a multicast address is bound on both 324 * new and duplicated sockets. 325 */ 326 if (so->so_options & SO_REUSEADDR) 327 reuseport = SO_REUSEADDR|SO_REUSEPORT; 328 } else if (sin->sin_addr.s_addr != INADDR_ANY) { 329 sin->sin_port = 0; /* yech... */ 330 bzero(&sin->sin_zero, sizeof(sin->sin_zero)); 331 if (ifa_ifwithaddr((struct sockaddr *)sin) == 0) 332 return (EADDRNOTAVAIL); 333 } 334 laddr = sin->sin_addr; 335 if (lport) { 336 struct inpcb *t; 337 struct tcptw *tw; 338 339 /* GROSS */ 340 if (ntohs(lport) <= ipport_reservedhigh && 341 ntohs(lport) >= ipport_reservedlow && 342 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 343 0)) 344 return (EACCES); 345 if (jailed(cred)) 346 prison = 1; 347 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) && 348 priv_check_cred(so->so_cred, 349 PRIV_NETINET_REUSEPORT, 0) != 0) { 350 t = in_pcblookup_local(pcbinfo, sin->sin_addr, 351 lport, prison ? 0 : INPLOOKUP_WILDCARD, 352 cred); 353 /* 354 * XXX 355 * This entire block sorely needs a rewrite. 356 */ 357 if (t && 358 ((t->inp_vflag & INP_TIMEWAIT) == 0) && 359 (so->so_type != SOCK_STREAM || 360 ntohl(t->inp_faddr.s_addr) == INADDR_ANY) && 361 (ntohl(sin->sin_addr.s_addr) != INADDR_ANY || 362 ntohl(t->inp_laddr.s_addr) != INADDR_ANY || 363 (t->inp_socket->so_options & 364 SO_REUSEPORT) == 0) && 365 (so->so_cred->cr_uid != 366 t->inp_socket->so_cred->cr_uid)) 367 return (EADDRINUSE); 368 } 369 if (prison && prison_ip(cred, 0, &sin->sin_addr.s_addr)) 370 return (EADDRNOTAVAIL); 371 t = in_pcblookup_local(pcbinfo, sin->sin_addr, 372 lport, prison ? 0 : wild, cred); 373 if (t && (t->inp_vflag & INP_TIMEWAIT)) { 374 /* 375 * XXXRW: If an incpb has had its timewait 376 * state recycled, we treat the address as 377 * being in use (for now). This is better 378 * than a panic, but not desirable. 379 */ 380 tw = intotw(inp); 381 if (tw == NULL || 382 (reuseport & tw->tw_so_options) == 0) 383 return (EADDRINUSE); 384 } else if (t && 385 (reuseport & t->inp_socket->so_options) == 0) { 386 #ifdef INET6 387 if (ntohl(sin->sin_addr.s_addr) != 388 INADDR_ANY || 389 ntohl(t->inp_laddr.s_addr) != 390 INADDR_ANY || 391 INP_SOCKAF(so) == 392 INP_SOCKAF(t->inp_socket)) 393 #endif 394 return (EADDRINUSE); 395 } 396 } 397 } 398 if (*lportp != 0) 399 lport = *lportp; 400 if (lport == 0) { 401 u_short first, last, aux; 402 int count; 403 404 if (laddr.s_addr != INADDR_ANY) 405 if (prison_ip(cred, 0, &laddr.s_addr)) 406 return (EINVAL); 407 408 if (inp->inp_flags & INP_HIGHPORT) { 409 first = ipport_hifirstauto; /* sysctl */ 410 last = ipport_hilastauto; 411 lastport = &pcbinfo->ipi_lasthi; 412 } else if (inp->inp_flags & INP_LOWPORT) { 413 error = priv_check_cred(cred, 414 PRIV_NETINET_RESERVEDPORT, 0); 415 if (error) 416 return error; 417 first = ipport_lowfirstauto; /* 1023 */ 418 last = ipport_lowlastauto; /* 600 */ 419 lastport = &pcbinfo->ipi_lastlow; 420 } else { 421 first = ipport_firstauto; /* sysctl */ 422 last = ipport_lastauto; 423 lastport = &pcbinfo->ipi_lastport; 424 } 425 /* 426 * For UDP, use random port allocation as long as the user 427 * allows it. For TCP (and as of yet unknown) connections, 428 * use random port allocation only if the user allows it AND 429 * ipport_tick() allows it. 430 */ 431 if (ipport_randomized && 432 (!ipport_stoprandom || pcbinfo == &udbinfo)) 433 dorandom = 1; 434 else 435 dorandom = 0; 436 /* 437 * It makes no sense to do random port allocation if 438 * we have the only port available. 439 */ 440 if (first == last) 441 dorandom = 0; 442 /* Make sure to not include UDP packets in the count. */ 443 if (pcbinfo != &udbinfo) 444 ipport_tcpallocs++; 445 /* 446 * Simple check to ensure all ports are not used up causing 447 * a deadlock here. 448 */ 449 if (first > last) { 450 aux = first; 451 first = last; 452 last = aux; 453 } 454 455 if (dorandom) 456 *lastport = first + 457 (arc4random() % (last - first)); 458 459 count = last - first; 460 461 do { 462 if (count-- < 0) /* completely used? */ 463 return (EADDRNOTAVAIL); 464 ++*lastport; 465 if (*lastport < first || *lastport > last) 466 *lastport = first; 467 lport = htons(*lastport); 468 } while (in_pcblookup_local(pcbinfo, laddr, 469 lport, wild, cred)); 470 } 471 if (prison_ip(cred, 0, &laddr.s_addr)) 472 return (EINVAL); 473 *laddrp = laddr.s_addr; 474 *lportp = lport; 475 return (0); 476 } 477 478 /* 479 * Connect from a socket to a specified address. 480 * Both address and port must be specified in argument sin. 481 * If don't have a local address for this socket yet, 482 * then pick one. 483 */ 484 int 485 in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred) 486 { 487 u_short lport, fport; 488 in_addr_t laddr, faddr; 489 int anonport, error; 490 491 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 492 INP_WLOCK_ASSERT(inp); 493 494 lport = inp->inp_lport; 495 laddr = inp->inp_laddr.s_addr; 496 anonport = (lport == 0); 497 error = in_pcbconnect_setup(inp, nam, &laddr, &lport, &faddr, &fport, 498 NULL, cred); 499 if (error) 500 return (error); 501 502 /* Do the initial binding of the local address if required. */ 503 if (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0) { 504 inp->inp_lport = lport; 505 inp->inp_laddr.s_addr = laddr; 506 if (in_pcbinshash(inp) != 0) { 507 inp->inp_laddr.s_addr = INADDR_ANY; 508 inp->inp_lport = 0; 509 return (EAGAIN); 510 } 511 } 512 513 /* Commit the remaining changes. */ 514 inp->inp_lport = lport; 515 inp->inp_laddr.s_addr = laddr; 516 inp->inp_faddr.s_addr = faddr; 517 inp->inp_fport = fport; 518 in_pcbrehash(inp); 519 520 if (anonport) 521 inp->inp_flags |= INP_ANONPORT; 522 return (0); 523 } 524 525 /* 526 * Set up for a connect from a socket to the specified address. 527 * On entry, *laddrp and *lportp should contain the current local 528 * address and port for the PCB; these are updated to the values 529 * that should be placed in inp_laddr and inp_lport to complete 530 * the connect. 531 * 532 * On success, *faddrp and *fportp will be set to the remote address 533 * and port. These are not updated in the error case. 534 * 535 * If the operation fails because the connection already exists, 536 * *oinpp will be set to the PCB of that connection so that the 537 * caller can decide to override it. In all other cases, *oinpp 538 * is set to NULL. 539 */ 540 int 541 in_pcbconnect_setup(struct inpcb *inp, struct sockaddr *nam, 542 in_addr_t *laddrp, u_short *lportp, in_addr_t *faddrp, u_short *fportp, 543 struct inpcb **oinpp, struct ucred *cred) 544 { 545 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 546 struct in_ifaddr *ia; 547 struct sockaddr_in sa; 548 struct ucred *socred; 549 struct inpcb *oinp; 550 struct in_addr laddr, faddr; 551 u_short lport, fport; 552 int error; 553 554 /* 555 * Because a global state change doesn't actually occur here, a read 556 * lock is sufficient. 557 */ 558 INP_INFO_LOCK_ASSERT(inp->inp_pcbinfo); 559 INP_LOCK_ASSERT(inp); 560 561 if (oinpp != NULL) 562 *oinpp = NULL; 563 if (nam->sa_len != sizeof (*sin)) 564 return (EINVAL); 565 if (sin->sin_family != AF_INET) 566 return (EAFNOSUPPORT); 567 if (sin->sin_port == 0) 568 return (EADDRNOTAVAIL); 569 laddr.s_addr = *laddrp; 570 lport = *lportp; 571 faddr = sin->sin_addr; 572 fport = sin->sin_port; 573 socred = inp->inp_socket->so_cred; 574 if (laddr.s_addr == INADDR_ANY && jailed(socred)) { 575 bzero(&sa, sizeof(sa)); 576 sa.sin_addr.s_addr = htonl(prison_getip(socred)); 577 sa.sin_len = sizeof(sa); 578 sa.sin_family = AF_INET; 579 error = in_pcbbind_setup(inp, (struct sockaddr *)&sa, 580 &laddr.s_addr, &lport, cred); 581 if (error) 582 return (error); 583 } 584 if (!TAILQ_EMPTY(&in_ifaddrhead)) { 585 /* 586 * If the destination address is INADDR_ANY, 587 * use the primary local address. 588 * If the supplied address is INADDR_BROADCAST, 589 * and the primary interface supports broadcast, 590 * choose the broadcast address for that interface. 591 */ 592 if (faddr.s_addr == INADDR_ANY) 593 faddr = IA_SIN(TAILQ_FIRST(&in_ifaddrhead))->sin_addr; 594 else if (faddr.s_addr == (u_long)INADDR_BROADCAST && 595 (TAILQ_FIRST(&in_ifaddrhead)->ia_ifp->if_flags & 596 IFF_BROADCAST)) 597 faddr = satosin(&TAILQ_FIRST( 598 &in_ifaddrhead)->ia_broadaddr)->sin_addr; 599 } 600 if (laddr.s_addr == INADDR_ANY) { 601 ia = NULL; 602 /* 603 * If route is known our src addr is taken from the i/f, 604 * else punt. 605 * 606 * Find out route to destination 607 */ 608 if ((inp->inp_socket->so_options & SO_DONTROUTE) == 0) 609 ia = ip_rtaddr(faddr, inp->inp_inc.inc_fibnum); 610 /* 611 * If we found a route, use the address corresponding to 612 * the outgoing interface. 613 * 614 * Otherwise assume faddr is reachable on a directly connected 615 * network and try to find a corresponding interface to take 616 * the source address from. 617 */ 618 if (ia == NULL) { 619 bzero(&sa, sizeof(sa)); 620 sa.sin_addr = faddr; 621 sa.sin_len = sizeof(sa); 622 sa.sin_family = AF_INET; 623 624 ia = ifatoia(ifa_ifwithdstaddr(sintosa(&sa))); 625 if (ia == NULL) 626 ia = ifatoia(ifa_ifwithnet(sintosa(&sa))); 627 if (ia == NULL) 628 return (ENETUNREACH); 629 } 630 /* 631 * If the destination address is multicast and an outgoing 632 * interface has been set as a multicast option, use the 633 * address of that interface as our source address. 634 */ 635 if (IN_MULTICAST(ntohl(faddr.s_addr)) && 636 inp->inp_moptions != NULL) { 637 struct ip_moptions *imo; 638 struct ifnet *ifp; 639 640 imo = inp->inp_moptions; 641 if (imo->imo_multicast_ifp != NULL) { 642 ifp = imo->imo_multicast_ifp; 643 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) 644 if (ia->ia_ifp == ifp) 645 break; 646 if (ia == NULL) 647 return (EADDRNOTAVAIL); 648 } 649 } 650 laddr = ia->ia_addr.sin_addr; 651 } 652 653 oinp = in_pcblookup_hash(inp->inp_pcbinfo, faddr, fport, laddr, lport, 654 0, NULL); 655 if (oinp != NULL) { 656 if (oinpp != NULL) 657 *oinpp = oinp; 658 return (EADDRINUSE); 659 } 660 if (lport == 0) { 661 error = in_pcbbind_setup(inp, NULL, &laddr.s_addr, &lport, 662 cred); 663 if (error) 664 return (error); 665 } 666 *laddrp = laddr.s_addr; 667 *lportp = lport; 668 *faddrp = faddr.s_addr; 669 *fportp = fport; 670 return (0); 671 } 672 673 void 674 in_pcbdisconnect(struct inpcb *inp) 675 { 676 677 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 678 INP_WLOCK_ASSERT(inp); 679 680 inp->inp_faddr.s_addr = INADDR_ANY; 681 inp->inp_fport = 0; 682 in_pcbrehash(inp); 683 } 684 685 /* 686 * In the old world order, in_pcbdetach() served two functions: to detach the 687 * pcb from the socket/potentially free the socket, and to free the pcb 688 * itself. In the new world order, the protocol code is responsible for 689 * managing the relationship with the socket, and this code simply frees the 690 * pcb. 691 */ 692 void 693 in_pcbdetach(struct inpcb *inp) 694 { 695 696 KASSERT(inp->inp_socket != NULL, ("in_pcbdetach: inp_socket == NULL")); 697 inp->inp_socket->so_pcb = NULL; 698 inp->inp_socket = NULL; 699 } 700 701 void 702 in_pcbfree(struct inpcb *inp) 703 { 704 struct inpcbinfo *ipi = inp->inp_pcbinfo; 705 706 KASSERT(inp->inp_socket == NULL, ("in_pcbfree: inp_socket != NULL")); 707 708 INP_INFO_WLOCK_ASSERT(ipi); 709 INP_WLOCK_ASSERT(inp); 710 711 #ifdef IPSEC 712 ipsec4_delete_pcbpolicy(inp); 713 #endif /*IPSEC*/ 714 inp->inp_gencnt = ++ipi->ipi_gencnt; 715 in_pcbremlists(inp); 716 if (inp->inp_options) 717 (void)m_free(inp->inp_options); 718 if (inp->inp_moptions != NULL) 719 inp_freemoptions(inp->inp_moptions); 720 inp->inp_vflag = 0; 721 722 #ifdef MAC 723 mac_inpcb_destroy(inp); 724 #endif 725 INP_WUNLOCK(inp); 726 uma_zfree(ipi->ipi_zone, inp); 727 } 728 729 /* 730 * TCP needs to maintain its inpcb structure after the TCP connection has 731 * been torn down. However, it must be disconnected from the inpcb hashes as 732 * it must not prevent binding of future connections to the same port/ip 733 * combination by other inpcbs. 734 */ 735 void 736 in_pcbdrop(struct inpcb *inp) 737 { 738 739 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 740 INP_WLOCK_ASSERT(inp); 741 742 inp->inp_vflag |= INP_DROPPED; 743 if (inp->inp_lport) { 744 struct inpcbport *phd = inp->inp_phd; 745 746 LIST_REMOVE(inp, inp_hash); 747 LIST_REMOVE(inp, inp_portlist); 748 if (LIST_FIRST(&phd->phd_pcblist) == NULL) { 749 LIST_REMOVE(phd, phd_hash); 750 free(phd, M_PCB); 751 } 752 inp->inp_lport = 0; 753 } 754 } 755 756 /* 757 * Common routines to return the socket addresses associated with inpcbs. 758 */ 759 struct sockaddr * 760 in_sockaddr(in_port_t port, struct in_addr *addr_p) 761 { 762 struct sockaddr_in *sin; 763 764 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, 765 M_WAITOK | M_ZERO); 766 sin->sin_family = AF_INET; 767 sin->sin_len = sizeof(*sin); 768 sin->sin_addr = *addr_p; 769 sin->sin_port = port; 770 771 return (struct sockaddr *)sin; 772 } 773 774 int 775 in_getsockaddr(struct socket *so, struct sockaddr **nam) 776 { 777 struct inpcb *inp; 778 struct in_addr addr; 779 in_port_t port; 780 781 inp = sotoinpcb(so); 782 KASSERT(inp != NULL, ("in_getsockaddr: inp == NULL")); 783 784 INP_RLOCK(inp); 785 port = inp->inp_lport; 786 addr = inp->inp_laddr; 787 INP_RUNLOCK(inp); 788 789 *nam = in_sockaddr(port, &addr); 790 return 0; 791 } 792 793 int 794 in_getpeeraddr(struct socket *so, struct sockaddr **nam) 795 { 796 struct inpcb *inp; 797 struct in_addr addr; 798 in_port_t port; 799 800 inp = sotoinpcb(so); 801 KASSERT(inp != NULL, ("in_getpeeraddr: inp == NULL")); 802 803 INP_RLOCK(inp); 804 port = inp->inp_fport; 805 addr = inp->inp_faddr; 806 INP_RUNLOCK(inp); 807 808 *nam = in_sockaddr(port, &addr); 809 return 0; 810 } 811 812 void 813 in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr faddr, int errno, 814 struct inpcb *(*notify)(struct inpcb *, int)) 815 { 816 struct inpcb *inp, *inp_temp; 817 818 INP_INFO_WLOCK(pcbinfo); 819 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) { 820 INP_WLOCK(inp); 821 #ifdef INET6 822 if ((inp->inp_vflag & INP_IPV4) == 0) { 823 INP_WUNLOCK(inp); 824 continue; 825 } 826 #endif 827 if (inp->inp_faddr.s_addr != faddr.s_addr || 828 inp->inp_socket == NULL) { 829 INP_WUNLOCK(inp); 830 continue; 831 } 832 if ((*notify)(inp, errno)) 833 INP_WUNLOCK(inp); 834 } 835 INP_INFO_WUNLOCK(pcbinfo); 836 } 837 838 void 839 in_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) 840 { 841 struct inpcb *inp; 842 struct ip_moptions *imo; 843 int i, gap; 844 845 INP_INFO_RLOCK(pcbinfo); 846 LIST_FOREACH(inp, pcbinfo->ipi_listhead, inp_list) { 847 INP_WLOCK(inp); 848 imo = inp->inp_moptions; 849 if ((inp->inp_vflag & INP_IPV4) && 850 imo != NULL) { 851 /* 852 * Unselect the outgoing interface if it is being 853 * detached. 854 */ 855 if (imo->imo_multicast_ifp == ifp) 856 imo->imo_multicast_ifp = NULL; 857 858 /* 859 * Drop multicast group membership if we joined 860 * through the interface being detached. 861 */ 862 for (i = 0, gap = 0; i < imo->imo_num_memberships; 863 i++) { 864 if (imo->imo_membership[i]->inm_ifp == ifp) { 865 in_delmulti(imo->imo_membership[i]); 866 gap++; 867 } else if (gap != 0) 868 imo->imo_membership[i - gap] = 869 imo->imo_membership[i]; 870 } 871 imo->imo_num_memberships -= gap; 872 } 873 INP_WUNLOCK(inp); 874 } 875 INP_INFO_RUNLOCK(pcbinfo); 876 } 877 878 /* 879 * Lookup a PCB based on the local address and port. 880 */ 881 #define INP_LOOKUP_MAPPED_PCB_COST 3 882 struct inpcb * 883 in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr, 884 u_short lport, int wild_okay, struct ucred *cred) 885 { 886 struct inpcb *inp; 887 #ifdef INET6 888 int matchwild = 3 + INP_LOOKUP_MAPPED_PCB_COST; 889 #else 890 int matchwild = 3; 891 #endif 892 int wildcard; 893 894 INP_INFO_LOCK_ASSERT(pcbinfo); 895 896 if (!wild_okay) { 897 struct inpcbhead *head; 898 /* 899 * Look for an unconnected (wildcard foreign addr) PCB that 900 * matches the local address and port we're looking for. 901 */ 902 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 903 0, pcbinfo->ipi_hashmask)]; 904 LIST_FOREACH(inp, head, inp_hash) { 905 #ifdef INET6 906 if ((inp->inp_vflag & INP_IPV4) == 0) 907 continue; 908 #endif 909 if (inp->inp_faddr.s_addr == INADDR_ANY && 910 inp->inp_laddr.s_addr == laddr.s_addr && 911 inp->inp_lport == lport) { 912 /* 913 * Found. 914 */ 915 return (inp); 916 } 917 } 918 /* 919 * Not found. 920 */ 921 return (NULL); 922 } else { 923 struct inpcbporthead *porthash; 924 struct inpcbport *phd; 925 struct inpcb *match = NULL; 926 /* 927 * Best fit PCB lookup. 928 * 929 * First see if this local port is in use by looking on the 930 * port hash list. 931 */ 932 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport, 933 pcbinfo->ipi_porthashmask)]; 934 LIST_FOREACH(phd, porthash, phd_hash) { 935 if (phd->phd_port == lport) 936 break; 937 } 938 if (phd != NULL) { 939 /* 940 * Port is in use by one or more PCBs. Look for best 941 * fit. 942 */ 943 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 944 wildcard = 0; 945 #ifdef INET6 946 if ((inp->inp_vflag & INP_IPV4) == 0) 947 continue; 948 /* 949 * We never select the PCB that has 950 * INP_IPV6 flag and is bound to :: if 951 * we have another PCB which is bound 952 * to 0.0.0.0. If a PCB has the 953 * INP_IPV6 flag, then we set its cost 954 * higher than IPv4 only PCBs. 955 * 956 * Note that the case only happens 957 * when a socket is bound to ::, under 958 * the condition that the use of the 959 * mapped address is allowed. 960 */ 961 if ((inp->inp_vflag & INP_IPV6) != 0) 962 wildcard += INP_LOOKUP_MAPPED_PCB_COST; 963 #endif 964 if (inp->inp_faddr.s_addr != INADDR_ANY) 965 wildcard++; 966 if (inp->inp_laddr.s_addr != INADDR_ANY) { 967 if (laddr.s_addr == INADDR_ANY) 968 wildcard++; 969 else if (inp->inp_laddr.s_addr != laddr.s_addr) 970 continue; 971 } else { 972 if (laddr.s_addr != INADDR_ANY) 973 wildcard++; 974 } 975 if (wildcard < matchwild) { 976 match = inp; 977 matchwild = wildcard; 978 if (matchwild == 0) { 979 break; 980 } 981 } 982 } 983 } 984 return (match); 985 } 986 } 987 #undef INP_LOOKUP_MAPPED_PCB_COST 988 989 /* 990 * Lookup PCB in hash list. 991 */ 992 struct inpcb * 993 in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr, 994 u_int fport_arg, struct in_addr laddr, u_int lport_arg, int wildcard, 995 struct ifnet *ifp) 996 { 997 struct inpcbhead *head; 998 struct inpcb *inp; 999 u_short fport = fport_arg, lport = lport_arg; 1000 1001 INP_INFO_LOCK_ASSERT(pcbinfo); 1002 1003 /* 1004 * First look for an exact match. 1005 */ 1006 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(faddr.s_addr, lport, fport, 1007 pcbinfo->ipi_hashmask)]; 1008 LIST_FOREACH(inp, head, inp_hash) { 1009 #ifdef INET6 1010 if ((inp->inp_vflag & INP_IPV4) == 0) 1011 continue; 1012 #endif 1013 if (inp->inp_faddr.s_addr == faddr.s_addr && 1014 inp->inp_laddr.s_addr == laddr.s_addr && 1015 inp->inp_fport == fport && 1016 inp->inp_lport == lport) 1017 return (inp); 1018 } 1019 1020 /* 1021 * Then look for a wildcard match, if requested. 1022 */ 1023 if (wildcard) { 1024 struct inpcb *local_wild = NULL; 1025 #ifdef INET6 1026 struct inpcb *local_wild_mapped = NULL; 1027 #endif 1028 1029 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport, 1030 0, pcbinfo->ipi_hashmask)]; 1031 LIST_FOREACH(inp, head, inp_hash) { 1032 #ifdef INET6 1033 if ((inp->inp_vflag & INP_IPV4) == 0) 1034 continue; 1035 #endif 1036 if (inp->inp_faddr.s_addr == INADDR_ANY && 1037 inp->inp_lport == lport) { 1038 if (ifp && ifp->if_type == IFT_FAITH && 1039 (inp->inp_flags & INP_FAITH) == 0) 1040 continue; 1041 if (inp->inp_laddr.s_addr == laddr.s_addr) 1042 return (inp); 1043 else if (inp->inp_laddr.s_addr == INADDR_ANY) { 1044 #ifdef INET6 1045 if (INP_CHECK_SOCKAF(inp->inp_socket, 1046 AF_INET6)) 1047 local_wild_mapped = inp; 1048 else 1049 #endif 1050 local_wild = inp; 1051 } 1052 } 1053 } 1054 #ifdef INET6 1055 if (local_wild == NULL) 1056 return (local_wild_mapped); 1057 #endif 1058 return (local_wild); 1059 } 1060 return (NULL); 1061 } 1062 1063 /* 1064 * Insert PCB onto various hash lists. 1065 */ 1066 int 1067 in_pcbinshash(struct inpcb *inp) 1068 { 1069 struct inpcbhead *pcbhash; 1070 struct inpcbporthead *pcbporthash; 1071 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1072 struct inpcbport *phd; 1073 u_int32_t hashkey_faddr; 1074 1075 INP_INFO_WLOCK_ASSERT(pcbinfo); 1076 INP_WLOCK_ASSERT(inp); 1077 1078 #ifdef INET6 1079 if (inp->inp_vflag & INP_IPV6) 1080 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX */; 1081 else 1082 #endif /* INET6 */ 1083 hashkey_faddr = inp->inp_faddr.s_addr; 1084 1085 pcbhash = &pcbinfo->ipi_hashbase[INP_PCBHASH(hashkey_faddr, 1086 inp->inp_lport, inp->inp_fport, pcbinfo->ipi_hashmask)]; 1087 1088 pcbporthash = &pcbinfo->ipi_porthashbase[ 1089 INP_PCBPORTHASH(inp->inp_lport, pcbinfo->ipi_porthashmask)]; 1090 1091 /* 1092 * Go through port list and look for a head for this lport. 1093 */ 1094 LIST_FOREACH(phd, pcbporthash, phd_hash) { 1095 if (phd->phd_port == inp->inp_lport) 1096 break; 1097 } 1098 /* 1099 * If none exists, malloc one and tack it on. 1100 */ 1101 if (phd == NULL) { 1102 MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport), M_PCB, M_NOWAIT); 1103 if (phd == NULL) { 1104 return (ENOBUFS); /* XXX */ 1105 } 1106 phd->phd_port = inp->inp_lport; 1107 LIST_INIT(&phd->phd_pcblist); 1108 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash); 1109 } 1110 inp->inp_phd = phd; 1111 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist); 1112 LIST_INSERT_HEAD(pcbhash, inp, inp_hash); 1113 return (0); 1114 } 1115 1116 /* 1117 * Move PCB to the proper hash bucket when { faddr, fport } have been 1118 * changed. NOTE: This does not handle the case of the lport changing (the 1119 * hashed port list would have to be updated as well), so the lport must 1120 * not change after in_pcbinshash() has been called. 1121 */ 1122 void 1123 in_pcbrehash(struct inpcb *inp) 1124 { 1125 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1126 struct inpcbhead *head; 1127 u_int32_t hashkey_faddr; 1128 1129 INP_INFO_WLOCK_ASSERT(pcbinfo); 1130 INP_WLOCK_ASSERT(inp); 1131 1132 #ifdef INET6 1133 if (inp->inp_vflag & INP_IPV6) 1134 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX */; 1135 else 1136 #endif /* INET6 */ 1137 hashkey_faddr = inp->inp_faddr.s_addr; 1138 1139 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(hashkey_faddr, 1140 inp->inp_lport, inp->inp_fport, pcbinfo->ipi_hashmask)]; 1141 1142 LIST_REMOVE(inp, inp_hash); 1143 LIST_INSERT_HEAD(head, inp, inp_hash); 1144 } 1145 1146 /* 1147 * Remove PCB from various lists. 1148 */ 1149 void 1150 in_pcbremlists(struct inpcb *inp) 1151 { 1152 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1153 1154 INP_INFO_WLOCK_ASSERT(pcbinfo); 1155 INP_WLOCK_ASSERT(inp); 1156 1157 inp->inp_gencnt = ++pcbinfo->ipi_gencnt; 1158 if (inp->inp_lport) { 1159 struct inpcbport *phd = inp->inp_phd; 1160 1161 LIST_REMOVE(inp, inp_hash); 1162 LIST_REMOVE(inp, inp_portlist); 1163 if (LIST_FIRST(&phd->phd_pcblist) == NULL) { 1164 LIST_REMOVE(phd, phd_hash); 1165 free(phd, M_PCB); 1166 } 1167 } 1168 LIST_REMOVE(inp, inp_list); 1169 pcbinfo->ipi_count--; 1170 } 1171 1172 /* 1173 * A set label operation has occurred at the socket layer, propagate the 1174 * label change into the in_pcb for the socket. 1175 */ 1176 void 1177 in_pcbsosetlabel(struct socket *so) 1178 { 1179 #ifdef MAC 1180 struct inpcb *inp; 1181 1182 inp = sotoinpcb(so); 1183 KASSERT(inp != NULL, ("in_pcbsosetlabel: so->so_pcb == NULL")); 1184 1185 INP_WLOCK(inp); 1186 SOCK_LOCK(so); 1187 mac_inpcb_sosetlabel(so, inp); 1188 SOCK_UNLOCK(so); 1189 INP_WUNLOCK(inp); 1190 #endif 1191 } 1192 1193 /* 1194 * ipport_tick runs once per second, determining if random port allocation 1195 * should be continued. If more than ipport_randomcps ports have been 1196 * allocated in the last second, then we return to sequential port 1197 * allocation. We return to random allocation only once we drop below 1198 * ipport_randomcps for at least ipport_randomtime seconds. 1199 */ 1200 void 1201 ipport_tick(void *xtp) 1202 { 1203 1204 if (ipport_tcpallocs <= ipport_tcplastcount + ipport_randomcps) { 1205 if (ipport_stoprandom > 0) 1206 ipport_stoprandom--; 1207 } else 1208 ipport_stoprandom = ipport_randomtime; 1209 ipport_tcplastcount = ipport_tcpallocs; 1210 callout_reset(&ipport_tick_callout, hz, ipport_tick, NULL); 1211 } 1212 1213 void 1214 inp_wlock(struct inpcb *inp) 1215 { 1216 1217 INP_WLOCK(inp); 1218 } 1219 1220 void 1221 inp_wunlock(struct inpcb *inp) 1222 { 1223 1224 INP_WUNLOCK(inp); 1225 } 1226 1227 void 1228 inp_rlock(struct inpcb *inp) 1229 { 1230 1231 INP_RLOCK(inp); 1232 } 1233 1234 void 1235 inp_runlock(struct inpcb *inp) 1236 { 1237 1238 INP_RUNLOCK(inp); 1239 } 1240 1241 #ifdef INVARIANTS 1242 void 1243 inp_lock_assert(struct inpcb *inp) 1244 { 1245 1246 INP_WLOCK_ASSERT(inp); 1247 } 1248 1249 void 1250 inp_unlock_assert(struct inpcb *inp) 1251 { 1252 1253 INP_UNLOCK_ASSERT(inp); 1254 } 1255 #endif 1256 1257 void 1258 inp_apply_all(void (*func)(struct inpcb *, void *), void *arg) 1259 { 1260 struct inpcb *inp; 1261 1262 INP_INFO_RLOCK(&tcbinfo); 1263 LIST_FOREACH(inp, tcbinfo.ipi_listhead, inp_list) { 1264 INP_WLOCK(inp); 1265 func(inp, arg); 1266 INP_WUNLOCK(inp); 1267 } 1268 INP_INFO_RUNLOCK(&tcbinfo); 1269 } 1270 1271 struct socket * 1272 inp_inpcbtosocket(struct inpcb *inp) 1273 { 1274 1275 INP_WLOCK_ASSERT(inp); 1276 return (inp->inp_socket); 1277 } 1278 1279 struct tcpcb * 1280 inp_inpcbtotcpcb(struct inpcb *inp) 1281 { 1282 1283 INP_WLOCK_ASSERT(inp); 1284 return ((struct tcpcb *)inp->inp_ppcb); 1285 } 1286 1287 int 1288 inp_ip_tos_get(const struct inpcb *inp) 1289 { 1290 1291 return (inp->inp_ip_tos); 1292 } 1293 1294 void 1295 inp_ip_tos_set(struct inpcb *inp, int val) 1296 { 1297 1298 inp->inp_ip_tos = val; 1299 } 1300 1301 void 1302 inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp, 1303 uint32_t *faddr, uint16_t *fp) 1304 { 1305 1306 INP_LOCK_ASSERT(inp); 1307 *laddr = inp->inp_laddr.s_addr; 1308 *faddr = inp->inp_faddr.s_addr; 1309 *lp = inp->inp_lport; 1310 *fp = inp->inp_fport; 1311 } 1312 1313 struct inpcb * 1314 so_sotoinpcb(struct socket *so) 1315 { 1316 1317 return (sotoinpcb(so)); 1318 } 1319 1320 struct tcpcb * 1321 so_sototcpcb(struct socket *so) 1322 { 1323 1324 return (sototcpcb(so)); 1325 } 1326 1327 #ifdef DDB 1328 static void 1329 db_print_indent(int indent) 1330 { 1331 int i; 1332 1333 for (i = 0; i < indent; i++) 1334 db_printf(" "); 1335 } 1336 1337 static void 1338 db_print_inconninfo(struct in_conninfo *inc, const char *name, int indent) 1339 { 1340 char faddr_str[48], laddr_str[48]; 1341 1342 db_print_indent(indent); 1343 db_printf("%s at %p\n", name, inc); 1344 1345 indent += 2; 1346 1347 #ifdef INET6 1348 if (inc->inc_flags == 1) { 1349 /* IPv6. */ 1350 ip6_sprintf(laddr_str, &inc->inc6_laddr); 1351 ip6_sprintf(faddr_str, &inc->inc6_faddr); 1352 } else { 1353 #endif 1354 /* IPv4. */ 1355 inet_ntoa_r(inc->inc_laddr, laddr_str); 1356 inet_ntoa_r(inc->inc_faddr, faddr_str); 1357 #ifdef INET6 1358 } 1359 #endif 1360 db_print_indent(indent); 1361 db_printf("inc_laddr %s inc_lport %u\n", laddr_str, 1362 ntohs(inc->inc_lport)); 1363 db_print_indent(indent); 1364 db_printf("inc_faddr %s inc_fport %u\n", faddr_str, 1365 ntohs(inc->inc_fport)); 1366 } 1367 1368 static void 1369 db_print_inpflags(int inp_flags) 1370 { 1371 int comma; 1372 1373 comma = 0; 1374 if (inp_flags & INP_RECVOPTS) { 1375 db_printf("%sINP_RECVOPTS", comma ? ", " : ""); 1376 comma = 1; 1377 } 1378 if (inp_flags & INP_RECVRETOPTS) { 1379 db_printf("%sINP_RECVRETOPTS", comma ? ", " : ""); 1380 comma = 1; 1381 } 1382 if (inp_flags & INP_RECVDSTADDR) { 1383 db_printf("%sINP_RECVDSTADDR", comma ? ", " : ""); 1384 comma = 1; 1385 } 1386 if (inp_flags & INP_HDRINCL) { 1387 db_printf("%sINP_HDRINCL", comma ? ", " : ""); 1388 comma = 1; 1389 } 1390 if (inp_flags & INP_HIGHPORT) { 1391 db_printf("%sINP_HIGHPORT", comma ? ", " : ""); 1392 comma = 1; 1393 } 1394 if (inp_flags & INP_LOWPORT) { 1395 db_printf("%sINP_LOWPORT", comma ? ", " : ""); 1396 comma = 1; 1397 } 1398 if (inp_flags & INP_ANONPORT) { 1399 db_printf("%sINP_ANONPORT", comma ? ", " : ""); 1400 comma = 1; 1401 } 1402 if (inp_flags & INP_RECVIF) { 1403 db_printf("%sINP_RECVIF", comma ? ", " : ""); 1404 comma = 1; 1405 } 1406 if (inp_flags & INP_MTUDISC) { 1407 db_printf("%sINP_MTUDISC", comma ? ", " : ""); 1408 comma = 1; 1409 } 1410 if (inp_flags & INP_FAITH) { 1411 db_printf("%sINP_FAITH", comma ? ", " : ""); 1412 comma = 1; 1413 } 1414 if (inp_flags & INP_RECVTTL) { 1415 db_printf("%sINP_RECVTTL", comma ? ", " : ""); 1416 comma = 1; 1417 } 1418 if (inp_flags & INP_DONTFRAG) { 1419 db_printf("%sINP_DONTFRAG", comma ? ", " : ""); 1420 comma = 1; 1421 } 1422 if (inp_flags & IN6P_IPV6_V6ONLY) { 1423 db_printf("%sIN6P_IPV6_V6ONLY", comma ? ", " : ""); 1424 comma = 1; 1425 } 1426 if (inp_flags & IN6P_PKTINFO) { 1427 db_printf("%sIN6P_PKTINFO", comma ? ", " : ""); 1428 comma = 1; 1429 } 1430 if (inp_flags & IN6P_HOPLIMIT) { 1431 db_printf("%sIN6P_HOPLIMIT", comma ? ", " : ""); 1432 comma = 1; 1433 } 1434 if (inp_flags & IN6P_HOPOPTS) { 1435 db_printf("%sIN6P_HOPOPTS", comma ? ", " : ""); 1436 comma = 1; 1437 } 1438 if (inp_flags & IN6P_DSTOPTS) { 1439 db_printf("%sIN6P_DSTOPTS", comma ? ", " : ""); 1440 comma = 1; 1441 } 1442 if (inp_flags & IN6P_RTHDR) { 1443 db_printf("%sIN6P_RTHDR", comma ? ", " : ""); 1444 comma = 1; 1445 } 1446 if (inp_flags & IN6P_RTHDRDSTOPTS) { 1447 db_printf("%sIN6P_RTHDRDSTOPTS", comma ? ", " : ""); 1448 comma = 1; 1449 } 1450 if (inp_flags & IN6P_TCLASS) { 1451 db_printf("%sIN6P_TCLASS", comma ? ", " : ""); 1452 comma = 1; 1453 } 1454 if (inp_flags & IN6P_AUTOFLOWLABEL) { 1455 db_printf("%sIN6P_AUTOFLOWLABEL", comma ? ", " : ""); 1456 comma = 1; 1457 } 1458 if (inp_flags & IN6P_RFC2292) { 1459 db_printf("%sIN6P_RFC2292", comma ? ", " : ""); 1460 comma = 1; 1461 } 1462 if (inp_flags & IN6P_MTU) { 1463 db_printf("IN6P_MTU%s", comma ? ", " : ""); 1464 comma = 1; 1465 } 1466 } 1467 1468 static void 1469 db_print_inpvflag(u_char inp_vflag) 1470 { 1471 int comma; 1472 1473 comma = 0; 1474 if (inp_vflag & INP_IPV4) { 1475 db_printf("%sINP_IPV4", comma ? ", " : ""); 1476 comma = 1; 1477 } 1478 if (inp_vflag & INP_IPV6) { 1479 db_printf("%sINP_IPV6", comma ? ", " : ""); 1480 comma = 1; 1481 } 1482 if (inp_vflag & INP_IPV6PROTO) { 1483 db_printf("%sINP_IPV6PROTO", comma ? ", " : ""); 1484 comma = 1; 1485 } 1486 if (inp_vflag & INP_TIMEWAIT) { 1487 db_printf("%sINP_TIMEWAIT", comma ? ", " : ""); 1488 comma = 1; 1489 } 1490 if (inp_vflag & INP_ONESBCAST) { 1491 db_printf("%sINP_ONESBCAST", comma ? ", " : ""); 1492 comma = 1; 1493 } 1494 if (inp_vflag & INP_DROPPED) { 1495 db_printf("%sINP_DROPPED", comma ? ", " : ""); 1496 comma = 1; 1497 } 1498 if (inp_vflag & INP_SOCKREF) { 1499 db_printf("%sINP_SOCKREF", comma ? ", " : ""); 1500 comma = 1; 1501 } 1502 } 1503 1504 void 1505 db_print_inpcb(struct inpcb *inp, const char *name, int indent) 1506 { 1507 1508 db_print_indent(indent); 1509 db_printf("%s at %p\n", name, inp); 1510 1511 indent += 2; 1512 1513 db_print_indent(indent); 1514 db_printf("inp_flow: 0x%x\n", inp->inp_flow); 1515 1516 db_print_inconninfo(&inp->inp_inc, "inp_conninfo", indent); 1517 1518 db_print_indent(indent); 1519 db_printf("inp_ppcb: %p inp_pcbinfo: %p inp_socket: %p\n", 1520 inp->inp_ppcb, inp->inp_pcbinfo, inp->inp_socket); 1521 1522 db_print_indent(indent); 1523 db_printf("inp_label: %p inp_flags: 0x%x (", 1524 inp->inp_label, inp->inp_flags); 1525 db_print_inpflags(inp->inp_flags); 1526 db_printf(")\n"); 1527 1528 db_print_indent(indent); 1529 db_printf("inp_sp: %p inp_vflag: 0x%x (", inp->inp_sp, 1530 inp->inp_vflag); 1531 db_print_inpvflag(inp->inp_vflag); 1532 db_printf(")\n"); 1533 1534 db_print_indent(indent); 1535 db_printf("inp_ip_ttl: %d inp_ip_p: %d inp_ip_minttl: %d\n", 1536 inp->inp_ip_ttl, inp->inp_ip_p, inp->inp_ip_minttl); 1537 1538 db_print_indent(indent); 1539 #ifdef INET6 1540 if (inp->inp_vflag & INP_IPV6) { 1541 db_printf("in6p_options: %p in6p_outputopts: %p " 1542 "in6p_moptions: %p\n", inp->in6p_options, 1543 inp->in6p_outputopts, inp->in6p_moptions); 1544 db_printf("in6p_icmp6filt: %p in6p_cksum %d " 1545 "in6p_hops %u\n", inp->in6p_icmp6filt, inp->in6p_cksum, 1546 inp->in6p_hops); 1547 } else 1548 #endif 1549 { 1550 db_printf("inp_ip_tos: %d inp_ip_options: %p " 1551 "inp_ip_moptions: %p\n", inp->inp_ip_tos, 1552 inp->inp_options, inp->inp_moptions); 1553 } 1554 1555 db_print_indent(indent); 1556 db_printf("inp_phd: %p inp_gencnt: %ju\n", inp->inp_phd, 1557 (uintmax_t)inp->inp_gencnt); 1558 } 1559 1560 DB_SHOW_COMMAND(inpcb, db_show_inpcb) 1561 { 1562 struct inpcb *inp; 1563 1564 if (!have_addr) { 1565 db_printf("usage: show inpcb <addr>\n"); 1566 return; 1567 } 1568 inp = (struct inpcb *)addr; 1569 1570 db_print_inpcb(inp, "inpcb", 0); 1571 } 1572 #endif 1573