1 /* 2 * Copyright (c) 2004 Jeffrey M. Hsu. All rights reserved. 3 * Copyright (c) 2004 The DragonFly Project. All rights reserved. 4 * 5 * This code is derived from software contributed to The DragonFly Project 6 * by Jeffrey M. Hsu. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of The DragonFly Project nor the names of its 17 * contributors may be used to endorse or promote products derived 18 * from this software without specific, prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 22 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS 23 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 24 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 25 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING, 26 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 27 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 28 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 29 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 30 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 34 /* 35 * Copyright (c) 1982, 1986, 1991, 1993, 1995 36 * The Regents of the University of California. All rights reserved. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. All advertising materials mentioning features or use of this software 47 * must display the following acknowledgement: 48 * This product includes software developed by the University of 49 * California, Berkeley and its contributors. 50 * 4. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)in_pcb.c 8.4 (Berkeley) 5/24/95 67 * $FreeBSD: src/sys/netinet/in_pcb.c,v 1.59.2.27 2004/01/02 04:06:42 ambrisko Exp $ 68 * $DragonFly: src/sys/netinet/in_pcb.c,v 1.48 2008/11/08 03:38:23 sephe Exp $ 69 */ 70 71 #include "opt_ipsec.h" 72 #include "opt_inet6.h" 73 74 #include <sys/param.h> 75 #include <sys/systm.h> 76 #include <sys/malloc.h> 77 #include <sys/mbuf.h> 78 #include <sys/domain.h> 79 #include <sys/protosw.h> 80 #include <sys/socket.h> 81 #include <sys/socketvar.h> 82 #include <sys/proc.h> 83 #include <sys/priv.h> 84 #include <sys/jail.h> 85 #include <sys/kernel.h> 86 #include <sys/sysctl.h> 87 88 #include <sys/thread2.h> 89 #include <sys/socketvar2.h> 90 #include <sys/msgport2.h> 91 92 #include <machine/limits.h> 93 94 #include <net/if.h> 95 #include <net/if_types.h> 96 #include <net/route.h> 97 98 #include <netinet/in.h> 99 #include <netinet/in_pcb.h> 100 #include <netinet/in_var.h> 101 #include <netinet/ip_var.h> 102 #ifdef INET6 103 #include <netinet/ip6.h> 104 #include <netinet6/ip6_var.h> 105 #endif /* INET6 */ 106 107 #ifdef IPSEC 108 #include <netinet6/ipsec.h> 109 #include <netproto/key/key.h> 110 #endif 111 112 #ifdef FAST_IPSEC 113 #if defined(IPSEC) || defined(IPSEC_ESP) 114 #error "Bad idea: don't compile with both IPSEC and FAST_IPSEC!" 115 #endif 116 117 #include <netproto/ipsec/ipsec.h> 118 #include <netproto/ipsec/key.h> 119 #define IPSEC 120 #endif /* FAST_IPSEC */ 121 122 struct in_addr zeroin_addr; 123 124 /* 125 * These configure the range of local port addresses assigned to 126 * "unspecified" outgoing connections/packets/whatever. 127 */ 128 int ipport_lowfirstauto = IPPORT_RESERVED - 1; /* 1023 */ 129 int ipport_lowlastauto = IPPORT_RESERVEDSTART; /* 600 */ 130 131 int ipport_firstauto = IPPORT_RESERVED; /* 1024 */ 132 int ipport_lastauto = IPPORT_USERRESERVED; /* 5000 */ 133 134 int ipport_hifirstauto = IPPORT_HIFIRSTAUTO; /* 49152 */ 135 int ipport_hilastauto = IPPORT_HILASTAUTO; /* 65535 */ 136 137 #define RANGECHK(var, min, max) \ 138 if ((var) < (min)) { (var) = (min); } \ 139 else if ((var) > (max)) { (var) = (max); } 140 141 static int 142 sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS) 143 { 144 int error; 145 146 error = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req); 147 if (!error) { 148 RANGECHK(ipport_lowfirstauto, 1, IPPORT_RESERVED - 1); 149 RANGECHK(ipport_lowlastauto, 1, IPPORT_RESERVED - 1); 150 151 RANGECHK(ipport_firstauto, IPPORT_RESERVED, USHRT_MAX); 152 RANGECHK(ipport_lastauto, IPPORT_RESERVED, USHRT_MAX); 153 154 RANGECHK(ipport_hifirstauto, IPPORT_RESERVED, USHRT_MAX); 155 RANGECHK(ipport_hilastauto, IPPORT_RESERVED, USHRT_MAX); 156 } 157 return (error); 158 } 159 160 #undef RANGECHK 161 162 SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW, 0, "IP Ports"); 163 164 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW, 165 &ipport_lowfirstauto, 0, &sysctl_net_ipport_check, "I", ""); 166 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW, 167 &ipport_lowlastauto, 0, &sysctl_net_ipport_check, "I", ""); 168 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW, 169 &ipport_firstauto, 0, &sysctl_net_ipport_check, "I", ""); 170 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW, 171 &ipport_lastauto, 0, &sysctl_net_ipport_check, "I", ""); 172 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW, 173 &ipport_hifirstauto, 0, &sysctl_net_ipport_check, "I", ""); 174 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW, 175 &ipport_hilastauto, 0, &sysctl_net_ipport_check, "I", ""); 176 177 /* 178 * in_pcb.c: manage the Protocol Control Blocks. 179 * 180 * NOTE: It is assumed that most of these functions will be called from 181 * a critical section. XXX - There are, unfortunately, a few exceptions 182 * to this rule that should be fixed. 183 * 184 * NOTE: The caller should initialize the cpu field to the cpu running the 185 * protocol stack associated with this inpcbinfo. 186 */ 187 188 void 189 in_pcbinfo_init(struct inpcbinfo *pcbinfo) 190 { 191 LIST_INIT(&pcbinfo->pcblisthead); 192 pcbinfo->cpu = -1; 193 pcbinfo->portsave = kmalloc(sizeof(*pcbinfo->portsave), M_PCB, 194 M_WAITOK | M_ZERO); 195 } 196 197 /* 198 * Allocate a PCB and associate it with the socket. 199 */ 200 int 201 in_pcballoc(struct socket *so, struct inpcbinfo *pcbinfo) 202 { 203 struct inpcb *inp; 204 #ifdef IPSEC 205 int error; 206 #endif 207 208 inp = kmalloc(pcbinfo->ipi_size, M_PCB, M_WAITOK|M_ZERO); 209 inp->inp_gencnt = ++pcbinfo->ipi_gencnt; 210 inp->inp_pcbinfo = inp->inp_cpcbinfo = pcbinfo; 211 inp->inp_socket = so; 212 #ifdef IPSEC 213 error = ipsec_init_policy(so, &inp->inp_sp); 214 if (error != 0) { 215 kfree(inp, M_PCB); 216 return (error); 217 } 218 #endif 219 #ifdef INET6 220 if (INP_SOCKAF(so) == AF_INET6 && ip6_v6only) 221 inp->inp_flags |= IN6P_IPV6_V6ONLY; 222 if (ip6_auto_flowlabel) 223 inp->inp_flags |= IN6P_AUTOFLOWLABEL; 224 #endif 225 soreference(so); 226 so->so_pcb = inp; 227 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, inp, inp_list); 228 pcbinfo->ipi_count++; 229 return (0); 230 } 231 232 /* 233 * Unlink a pcb with the intention of moving it to another cpu with a 234 * different pcbinfo. While unlinked nothing should attempt to dereference 235 * inp_pcbinfo, NULL it out so we assert if it does. 236 */ 237 void 238 in_pcbunlink(struct inpcb *inp, struct inpcbinfo *pcbinfo) 239 { 240 KKASSERT(inp->inp_pcbinfo == pcbinfo); 241 242 LIST_REMOVE(inp, inp_list); 243 pcbinfo->ipi_count--; 244 inp->inp_pcbinfo = NULL; 245 } 246 247 /* 248 * Relink a pcb into a new pcbinfo. 249 */ 250 void 251 in_pcblink(struct inpcb *inp, struct inpcbinfo *pcbinfo) 252 { 253 KKASSERT(inp->inp_pcbinfo == NULL); 254 inp->inp_pcbinfo = pcbinfo; 255 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, inp, inp_list); 256 pcbinfo->ipi_count++; 257 } 258 259 int 260 in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 261 { 262 struct socket *so = inp->inp_socket; 263 struct proc *p = td->td_proc; 264 unsigned short *lastport; 265 struct sockaddr_in *sin; 266 struct sockaddr_in jsin; 267 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 268 struct ucred *cred = NULL; 269 u_short lport = 0; 270 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 271 int error; 272 273 KKASSERT(p); 274 275 if (TAILQ_EMPTY(&in_ifaddrheads[mycpuid])) /* XXX broken! */ 276 return (EADDRNOTAVAIL); 277 if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY) 278 return (EINVAL); /* already bound */ 279 280 if (!(so->so_options & (SO_REUSEADDR|SO_REUSEPORT))) 281 wild = 1; /* neither SO_REUSEADDR nor SO_REUSEPORT is set */ 282 if (p) 283 cred = p->p_ucred; 284 285 /* 286 * This has to be atomic. If the porthash is shared across multiple 287 * protocol threads (aka tcp) then the token will be non-NULL. 288 */ 289 if (pcbinfo->porttoken) 290 lwkt_gettoken(pcbinfo->porttoken); 291 292 if (nam != NULL) { 293 sin = (struct sockaddr_in *)nam; 294 if (nam->sa_len != sizeof *sin) { 295 error = EINVAL; 296 goto done; 297 } 298 #ifdef notdef 299 /* 300 * We should check the family, but old programs 301 * incorrectly fail to initialize it. 302 */ 303 if (sin->sin_family != AF_INET) { 304 error = EAFNOSUPPORT; 305 goto done; 306 } 307 #endif 308 if (!prison_replace_wildcards(td, nam)) { 309 error = EINVAL; 310 goto done; 311 } 312 lport = sin->sin_port; 313 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { 314 /* 315 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 316 * allow complete duplication of binding if 317 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 318 * and a multicast address is bound on both 319 * new and duplicated sockets. 320 */ 321 if (so->so_options & SO_REUSEADDR) 322 reuseport = SO_REUSEADDR | SO_REUSEPORT; 323 } else if (sin->sin_addr.s_addr != INADDR_ANY) { 324 sin->sin_port = 0; /* yech... */ 325 bzero(&sin->sin_zero, sizeof sin->sin_zero); 326 if (ifa_ifwithaddr((struct sockaddr *)sin) == NULL) { 327 error = EADDRNOTAVAIL; 328 goto done; 329 } 330 } 331 if (lport != 0) { 332 struct inpcb *t; 333 334 /* GROSS */ 335 if (ntohs(lport) < IPPORT_RESERVED && 336 cred && 337 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 0)) { 338 error = EACCES; 339 goto done; 340 } 341 if (so->so_cred->cr_uid != 0 && 342 !IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { 343 t = in_pcblookup_local(pcbinfo, 344 sin->sin_addr, 345 lport, 346 INPLOOKUP_WILDCARD, 347 cred); 348 if (t && 349 (!in_nullhost(sin->sin_addr) || 350 !in_nullhost(t->inp_laddr) || 351 (t->inp_socket->so_options & 352 SO_REUSEPORT) == 0) && 353 (so->so_cred->cr_uid != 354 t->inp_socket->so_cred->cr_uid)) { 355 #ifdef INET6 356 if (!in_nullhost(sin->sin_addr) || 357 !in_nullhost(t->inp_laddr) || 358 INP_SOCKAF(so) == 359 INP_SOCKAF(t->inp_socket)) 360 #endif 361 { 362 error = EADDRINUSE; 363 goto done; 364 } 365 } 366 } 367 if (cred && !prison_replace_wildcards(td, nam)) { 368 error = EADDRNOTAVAIL; 369 goto done; 370 } 371 t = in_pcblookup_local(pcbinfo, sin->sin_addr, lport, 372 wild, cred); 373 if (t && !(reuseport & t->inp_socket->so_options)) { 374 #ifdef INET6 375 if (!in_nullhost(sin->sin_addr) || 376 !in_nullhost(t->inp_laddr) || 377 INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket)) 378 #endif 379 { 380 error = EADDRINUSE; 381 goto done; 382 } 383 } 384 } 385 inp->inp_laddr = sin->sin_addr; 386 } 387 if (lport == 0) { 388 ushort first, last; 389 int count; 390 391 jsin.sin_family = AF_INET; 392 jsin.sin_addr.s_addr = inp->inp_laddr.s_addr; 393 if (!prison_replace_wildcards(td, (struct sockaddr *)&jsin)) { 394 inp->inp_laddr.s_addr = INADDR_ANY; 395 error = EINVAL; 396 goto done; 397 } 398 inp->inp_laddr.s_addr = jsin.sin_addr.s_addr; 399 400 inp->inp_flags |= INP_ANONPORT; 401 402 if (inp->inp_flags & INP_HIGHPORT) { 403 first = ipport_hifirstauto; /* sysctl */ 404 last = ipport_hilastauto; 405 lastport = &pcbinfo->lasthi; 406 } else if (inp->inp_flags & INP_LOWPORT) { 407 if (cred && 408 (error = priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 0))) { 409 inp->inp_laddr.s_addr = INADDR_ANY; 410 goto done; 411 } 412 first = ipport_lowfirstauto; /* 1023 */ 413 last = ipport_lowlastauto; /* 600 */ 414 lastport = &pcbinfo->lastlow; 415 } else { 416 first = ipport_firstauto; /* sysctl */ 417 last = ipport_lastauto; 418 lastport = &pcbinfo->lastport; 419 } 420 /* 421 * Simple check to ensure all ports are not used up causing 422 * a deadlock here. 423 * 424 * We split the two cases (up and down) so that the direction 425 * is not being tested on each round of the loop. 426 */ 427 if (first > last) { 428 /* 429 * counting down 430 */ 431 count = first - last; 432 433 do { 434 if (count-- < 0) { /* completely used? */ 435 inp->inp_laddr.s_addr = INADDR_ANY; 436 error = EADDRNOTAVAIL; 437 goto done; 438 } 439 --*lastport; 440 if (*lastport > first || *lastport < last) 441 *lastport = first; 442 lport = htons(*lastport); 443 } while (in_pcblookup_local(pcbinfo, inp->inp_laddr, 444 lport, wild, cred)); 445 } else { 446 /* 447 * counting up 448 */ 449 count = last - first; 450 451 do { 452 if (count-- < 0) { /* completely used? */ 453 inp->inp_laddr.s_addr = INADDR_ANY; 454 error = EADDRNOTAVAIL; 455 goto done; 456 } 457 ++*lastport; 458 if (*lastport < first || *lastport > last) 459 *lastport = first; 460 lport = htons(*lastport); 461 } while (in_pcblookup_local(pcbinfo, inp->inp_laddr, 462 lport, wild, cred)); 463 } 464 } 465 inp->inp_lport = lport; 466 467 jsin.sin_family = AF_INET; 468 jsin.sin_addr.s_addr = inp->inp_laddr.s_addr; 469 if (!prison_replace_wildcards(td, (struct sockaddr*)&jsin)) { 470 inp->inp_laddr.s_addr = INADDR_ANY; 471 inp->inp_lport = 0; 472 error = EINVAL; 473 goto done; 474 } 475 inp->inp_laddr.s_addr = jsin.sin_addr.s_addr; 476 477 if (in_pcbinsporthash(inp) != 0) { 478 inp->inp_laddr.s_addr = INADDR_ANY; 479 inp->inp_lport = 0; 480 error = EAGAIN; 481 goto done; 482 } 483 error = 0; 484 done: 485 if (pcbinfo->porttoken) 486 lwkt_reltoken(pcbinfo->porttoken); 487 return error; 488 } 489 490 /* 491 * Transform old in_pcbconnect() into an inner subroutine for new 492 * in_pcbconnect(): Do some validity-checking on the remote 493 * address (in mbuf 'nam') and then determine local host address 494 * (i.e., which interface) to use to access that remote host. 495 * 496 * This preserves definition of in_pcbconnect(), while supporting a 497 * slightly different version for T/TCP. (This is more than 498 * a bit of a kludge, but cleaning up the internal interfaces would 499 * have forced minor changes in every protocol). 500 */ 501 int 502 in_pcbladdr(struct inpcb *inp, struct sockaddr *nam, 503 struct sockaddr_in **plocal_sin, struct thread *td) 504 { 505 struct in_ifaddr *ia; 506 struct ucred *cred = NULL; 507 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 508 struct sockaddr *jsin; 509 int jailed = 0, alloc_route = 0; 510 511 if (nam->sa_len != sizeof *sin) 512 return (EINVAL); 513 if (sin->sin_family != AF_INET) 514 return (EAFNOSUPPORT); 515 if (sin->sin_port == 0) 516 return (EADDRNOTAVAIL); 517 if (td && td->td_proc && td->td_proc->p_ucred) 518 cred = td->td_proc->p_ucred; 519 if (cred && cred->cr_prison) 520 jailed = 1; 521 if (!TAILQ_EMPTY(&in_ifaddrheads[mycpuid])) { 522 ia = TAILQ_FIRST(&in_ifaddrheads[mycpuid])->ia; 523 /* 524 * If the destination address is INADDR_ANY, 525 * use the primary local address. 526 * If the supplied address is INADDR_BROADCAST, 527 * and the primary interface supports broadcast, 528 * choose the broadcast address for that interface. 529 */ 530 if (sin->sin_addr.s_addr == INADDR_ANY) 531 sin->sin_addr = IA_SIN(ia)->sin_addr; 532 else if (sin->sin_addr.s_addr == (u_long)INADDR_BROADCAST && 533 (ia->ia_ifp->if_flags & IFF_BROADCAST)) 534 sin->sin_addr = satosin(&ia->ia_broadaddr)->sin_addr; 535 } 536 if (inp->inp_laddr.s_addr == INADDR_ANY) { 537 struct route *ro; 538 539 ia = NULL; 540 /* 541 * If route is known or can be allocated now, 542 * our src addr is taken from the i/f, else punt. 543 * Note that we should check the address family of the cached 544 * destination, in case of sharing the cache with IPv6. 545 */ 546 ro = &inp->inp_route; 547 if (ro->ro_rt && 548 (!(ro->ro_rt->rt_flags & RTF_UP) || 549 ro->ro_dst.sa_family != AF_INET || 550 satosin(&ro->ro_dst)->sin_addr.s_addr != 551 sin->sin_addr.s_addr || 552 inp->inp_socket->so_options & SO_DONTROUTE)) { 553 RTFREE(ro->ro_rt); 554 ro->ro_rt = NULL; 555 } 556 if (!(inp->inp_socket->so_options & SO_DONTROUTE) && /*XXX*/ 557 (ro->ro_rt == NULL || 558 ro->ro_rt->rt_ifp == NULL)) { 559 /* No route yet, so try to acquire one */ 560 bzero(&ro->ro_dst, sizeof(struct sockaddr_in)); 561 ro->ro_dst.sa_family = AF_INET; 562 ro->ro_dst.sa_len = sizeof(struct sockaddr_in); 563 ((struct sockaddr_in *) &ro->ro_dst)->sin_addr = 564 sin->sin_addr; 565 rtalloc(ro); 566 alloc_route = 1; 567 } 568 /* 569 * If we found a route, use the address 570 * corresponding to the outgoing interface 571 * unless it is the loopback (in case a route 572 * to our address on another net goes to loopback). 573 */ 574 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) { 575 if (jailed) { 576 if (jailed_ip(cred->cr_prison, 577 ro->ro_rt->rt_ifa->ifa_addr)) { 578 ia = ifatoia(ro->ro_rt->rt_ifa); 579 } 580 } else { 581 ia = ifatoia(ro->ro_rt->rt_ifa); 582 } 583 } 584 if (ia == NULL) { 585 u_short fport = sin->sin_port; 586 587 sin->sin_port = 0; 588 ia = ifatoia(ifa_ifwithdstaddr(sintosa(sin))); 589 if (ia && jailed && !jailed_ip(cred->cr_prison, 590 sintosa(&ia->ia_addr))) 591 ia = NULL; 592 if (ia == NULL) 593 ia = ifatoia(ifa_ifwithnet(sintosa(sin))); 594 if (ia && jailed && !jailed_ip(cred->cr_prison, 595 sintosa(&ia->ia_addr))) 596 ia = NULL; 597 sin->sin_port = fport; 598 if (ia == NULL && 599 !TAILQ_EMPTY(&in_ifaddrheads[mycpuid])) 600 ia = TAILQ_FIRST(&in_ifaddrheads[mycpuid])->ia; 601 if (ia && jailed && !jailed_ip(cred->cr_prison, 602 sintosa(&ia->ia_addr))) 603 ia = NULL; 604 605 if (!jailed && ia == NULL) 606 goto fail; 607 } 608 /* 609 * If the destination address is multicast and an outgoing 610 * interface has been set as a multicast option, use the 611 * address of that interface as our source address. 612 */ 613 if (!jailed && IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) && 614 inp->inp_moptions != NULL) { 615 struct ip_moptions *imo; 616 struct ifnet *ifp; 617 618 imo = inp->inp_moptions; 619 if (imo->imo_multicast_ifp != NULL) { 620 struct in_ifaddr_container *iac; 621 622 ifp = imo->imo_multicast_ifp; 623 ia = NULL; 624 TAILQ_FOREACH(iac, 625 &in_ifaddrheads[mycpuid], ia_link) { 626 if (iac->ia->ia_ifp == ifp) { 627 ia = iac->ia; 628 break; 629 } 630 } 631 if (ia == NULL) 632 goto fail; 633 } 634 } 635 /* 636 * Don't do pcblookup call here; return interface in plocal_sin 637 * and exit to caller, that will do the lookup. 638 */ 639 if (ia == NULL && jailed) { 640 if ((jsin = prison_get_nonlocal(cred->cr_prison, AF_INET, NULL)) != NULL || 641 (jsin = prison_get_local(cred->cr_prison, AF_INET, NULL)) != NULL) { 642 *plocal_sin = satosin(jsin); 643 } else { 644 /* IPv6 only Jail */ 645 goto fail; 646 } 647 } else { 648 *plocal_sin = &ia->ia_addr; 649 } 650 } 651 return (0); 652 fail: 653 if (alloc_route) { 654 struct route *ro = &inp->inp_route; 655 656 if (ro->ro_rt != NULL) 657 RTFREE(ro->ro_rt); 658 bzero(ro, sizeof(*ro)); 659 } 660 return (EADDRNOTAVAIL); 661 } 662 663 /* 664 * Outer subroutine: 665 * Connect from a socket to a specified address. 666 * Both address and port must be specified in argument sin. 667 * If don't have a local address for this socket yet, 668 * then pick one. 669 */ 670 int 671 in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 672 { 673 struct sockaddr_in *if_sin; 674 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 675 int error; 676 677 /* Call inner routine to assign local interface address. */ 678 if ((error = in_pcbladdr(inp, nam, &if_sin, td)) != 0) 679 return (error); 680 681 if (in_pcblookup_hash(inp->inp_cpcbinfo, sin->sin_addr, sin->sin_port, 682 inp->inp_laddr.s_addr ? 683 inp->inp_laddr : if_sin->sin_addr, 684 inp->inp_lport, FALSE, NULL) != NULL) { 685 return (EADDRINUSE); 686 } 687 if (inp->inp_laddr.s_addr == INADDR_ANY) { 688 if (inp->inp_lport == 0) { 689 error = in_pcbbind(inp, NULL, td); 690 if (error) 691 return (error); 692 } 693 inp->inp_laddr = if_sin->sin_addr; 694 } 695 inp->inp_faddr = sin->sin_addr; 696 inp->inp_fport = sin->sin_port; 697 in_pcbinsconnhash(inp); 698 return (0); 699 } 700 701 void 702 in_pcbdisconnect(struct inpcb *inp) 703 { 704 705 inp->inp_faddr.s_addr = INADDR_ANY; 706 inp->inp_fport = 0; 707 in_pcbremconnhash(inp); 708 if (inp->inp_socket->so_state & SS_NOFDREF) 709 in_pcbdetach(inp); 710 } 711 712 void 713 in_pcbdetach(struct inpcb *inp) 714 { 715 struct socket *so = inp->inp_socket; 716 struct inpcbinfo *ipi = inp->inp_pcbinfo; 717 718 #ifdef IPSEC 719 ipsec4_delete_pcbpolicy(inp); 720 #endif /*IPSEC*/ 721 inp->inp_gencnt = ++ipi->ipi_gencnt; 722 KKASSERT((so->so_state & SS_ASSERTINPROG) == 0); 723 in_pcbremlists(inp); 724 so->so_pcb = NULL; 725 sofree(so); /* remove pcb ref */ 726 if (inp->inp_options) 727 m_free(inp->inp_options); 728 if (inp->inp_route.ro_rt) 729 rtfree(inp->inp_route.ro_rt); 730 ip_freemoptions(inp->inp_moptions); 731 inp->inp_vflag = 0; 732 kfree(inp, M_PCB); 733 } 734 735 /* 736 * The calling convention of in_setsockaddr() and in_setpeeraddr() was 737 * modified to match the pru_sockaddr() and pru_peeraddr() entry points 738 * in struct pr_usrreqs, so that protocols can just reference then directly 739 * without the need for a wrapper function. The socket must have a valid 740 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one 741 * except through a kernel programming error, so it is acceptable to panic 742 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap 743 * because there actually /is/ a programming error somewhere... XXX) 744 */ 745 int 746 in_setsockaddr(struct socket *so, struct sockaddr **nam) 747 { 748 struct inpcb *inp; 749 struct sockaddr_in *sin; 750 751 /* 752 * Do the malloc first in case it blocks. 753 */ 754 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, 755 M_WAITOK | M_ZERO); 756 sin->sin_family = AF_INET; 757 sin->sin_len = sizeof *sin; 758 759 crit_enter(); 760 inp = so->so_pcb; 761 if (!inp) { 762 crit_exit(); 763 kfree(sin, M_SONAME); 764 return (ECONNRESET); 765 } 766 sin->sin_port = inp->inp_lport; 767 sin->sin_addr = inp->inp_laddr; 768 crit_exit(); 769 770 *nam = (struct sockaddr *)sin; 771 return (0); 772 } 773 774 void 775 in_setsockaddr_dispatch(netmsg_t msg) 776 { 777 int error; 778 779 error = in_setsockaddr(msg->base.nm_so, msg->peeraddr.nm_nam); 780 lwkt_replymsg(&msg->lmsg, error); 781 } 782 783 int 784 in_setpeeraddr(struct socket *so, struct sockaddr **nam) 785 { 786 struct inpcb *inp; 787 struct sockaddr_in *sin; 788 789 /* 790 * Do the malloc first in case it blocks. 791 */ 792 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, 793 M_WAITOK | M_ZERO); 794 sin->sin_family = AF_INET; 795 sin->sin_len = sizeof *sin; 796 797 crit_enter(); 798 inp = so->so_pcb; 799 if (!inp) { 800 crit_exit(); 801 kfree(sin, M_SONAME); 802 return (ECONNRESET); 803 } 804 sin->sin_port = inp->inp_fport; 805 sin->sin_addr = inp->inp_faddr; 806 crit_exit(); 807 808 *nam = (struct sockaddr *)sin; 809 return (0); 810 } 811 812 void 813 in_setpeeraddr_dispatch(netmsg_t msg) 814 { 815 int error; 816 817 error = in_setpeeraddr(msg->base.nm_so, msg->peeraddr.nm_nam); 818 lwkt_replymsg(&msg->lmsg, error); 819 } 820 821 void 822 in_pcbnotifyall(struct inpcbhead *head, struct in_addr faddr, int err, 823 void (*notify)(struct inpcb *, int)) 824 { 825 struct inpcb *inp, *ninp; 826 827 /* 828 * note: if INP_PLACEMARKER is set we must ignore the rest of 829 * the structure and skip it. 830 */ 831 crit_enter(); 832 LIST_FOREACH_MUTABLE(inp, head, inp_list, ninp) { 833 if (inp->inp_flags & INP_PLACEMARKER) 834 continue; 835 #ifdef INET6 836 if (!(inp->inp_vflag & INP_IPV4)) 837 continue; 838 #endif 839 if (inp->inp_faddr.s_addr != faddr.s_addr || 840 inp->inp_socket == NULL) 841 continue; 842 (*notify)(inp, err); /* can remove inp from list! */ 843 } 844 crit_exit(); 845 } 846 847 void 848 in_pcbpurgeif0(struct inpcb *head, struct ifnet *ifp) 849 { 850 struct inpcb *inp; 851 struct ip_moptions *imo; 852 int i, gap; 853 854 for (inp = head; inp != NULL; inp = LIST_NEXT(inp, inp_list)) { 855 if (inp->inp_flags & INP_PLACEMARKER) 856 continue; 857 imo = inp->inp_moptions; 858 if ((inp->inp_vflag & INP_IPV4) && imo != NULL) { 859 /* 860 * Unselect the outgoing interface if it is being 861 * detached. 862 */ 863 if (imo->imo_multicast_ifp == ifp) 864 imo->imo_multicast_ifp = NULL; 865 866 /* 867 * Drop multicast group membership if we joined 868 * through the interface being detached. 869 */ 870 for (i = 0, gap = 0; i < imo->imo_num_memberships; 871 i++) { 872 if (imo->imo_membership[i]->inm_ifp == ifp) { 873 in_delmulti(imo->imo_membership[i]); 874 gap++; 875 } else if (gap != 0) 876 imo->imo_membership[i - gap] = 877 imo->imo_membership[i]; 878 } 879 imo->imo_num_memberships -= gap; 880 } 881 } 882 } 883 884 /* 885 * Check for alternatives when higher level complains 886 * about service problems. For now, invalidate cached 887 * routing information. If the route was created dynamically 888 * (by a redirect), time to try a default gateway again. 889 */ 890 void 891 in_losing(struct inpcb *inp) 892 { 893 struct rtentry *rt; 894 struct rt_addrinfo rtinfo; 895 896 if ((rt = inp->inp_route.ro_rt)) { 897 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 898 rtinfo.rti_info[RTAX_DST] = rt_key(rt); 899 rtinfo.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 900 rtinfo.rti_info[RTAX_NETMASK] = rt_mask(rt); 901 rtinfo.rti_flags = rt->rt_flags; 902 rt_missmsg(RTM_LOSING, &rtinfo, rt->rt_flags, 0); 903 if (rt->rt_flags & RTF_DYNAMIC) 904 rtrequest1_global(RTM_DELETE, &rtinfo, NULL, NULL); 905 inp->inp_route.ro_rt = NULL; 906 rtfree(rt); 907 /* 908 * A new route can be allocated 909 * the next time output is attempted. 910 */ 911 } 912 } 913 914 /* 915 * After a routing change, flush old routing 916 * and allocate a (hopefully) better one. 917 */ 918 void 919 in_rtchange(struct inpcb *inp, int err) 920 { 921 if (inp->inp_route.ro_rt) { 922 rtfree(inp->inp_route.ro_rt); 923 inp->inp_route.ro_rt = NULL; 924 /* 925 * A new route can be allocated the next time 926 * output is attempted. 927 */ 928 } 929 } 930 931 /* 932 * Lookup a PCB based on the local address and port. 933 */ 934 struct inpcb * 935 in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr, 936 u_int lport_arg, int wild_okay, struct ucred *cred) 937 { 938 struct inpcb *inp; 939 int matchwild = 3, wildcard; 940 u_short lport = lport_arg; 941 struct inpcbporthead *porthash; 942 struct inpcbport *phd; 943 struct inpcb *match = NULL; 944 945 /* 946 * If the porthashbase is shared across several cpus we need 947 * to lock. 948 */ 949 if (pcbinfo->porttoken) 950 lwkt_gettoken(pcbinfo->porttoken); 951 952 /* 953 * Best fit PCB lookup. 954 * 955 * First see if this local port is in use by looking on the 956 * port hash list. 957 */ 958 porthash = &pcbinfo->porthashbase[ 959 INP_PCBPORTHASH(lport, pcbinfo->porthashmask)]; 960 LIST_FOREACH(phd, porthash, phd_hash) { 961 if (phd->phd_port == lport) 962 break; 963 } 964 if (phd != NULL) { 965 /* 966 * Port is in use by one or more PCBs. Look for best 967 * fit. 968 */ 969 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 970 wildcard = 0; 971 #ifdef INET6 972 if ((inp->inp_vflag & INP_IPV4) == 0) 973 continue; 974 #endif 975 if (inp->inp_faddr.s_addr != INADDR_ANY) 976 wildcard++; 977 if (inp->inp_laddr.s_addr != INADDR_ANY) { 978 if (laddr.s_addr == INADDR_ANY) 979 wildcard++; 980 else if (inp->inp_laddr.s_addr != laddr.s_addr) 981 continue; 982 } else { 983 if (laddr.s_addr != INADDR_ANY) 984 wildcard++; 985 } 986 if (wildcard && !wild_okay) 987 continue; 988 if (wildcard < matchwild && 989 (cred == NULL || 990 cred->cr_prison == 991 inp->inp_socket->so_cred->cr_prison)) { 992 match = inp; 993 matchwild = wildcard; 994 if (matchwild == 0) { 995 break; 996 } 997 } 998 } 999 } 1000 if (pcbinfo->porttoken) 1001 lwkt_reltoken(pcbinfo->porttoken); 1002 return (match); 1003 } 1004 1005 /* 1006 * Lookup PCB in hash list. 1007 */ 1008 struct inpcb * 1009 in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr, 1010 u_int fport_arg, struct in_addr laddr, u_int lport_arg, 1011 boolean_t wildcard, struct ifnet *ifp) 1012 { 1013 struct inpcbhead *head; 1014 struct inpcb *inp, *jinp=NULL; 1015 u_short fport = fport_arg, lport = lport_arg; 1016 1017 /* 1018 * First look for an exact match. 1019 */ 1020 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr.s_addr, fport, 1021 laddr.s_addr, lport, pcbinfo->hashmask)]; 1022 LIST_FOREACH(inp, head, inp_hash) { 1023 #ifdef INET6 1024 if (!(inp->inp_vflag & INP_IPV4)) 1025 continue; 1026 #endif 1027 if (in_hosteq(inp->inp_faddr, faddr) && 1028 in_hosteq(inp->inp_laddr, laddr) && 1029 inp->inp_fport == fport && inp->inp_lport == lport) { 1030 /* found */ 1031 if (inp->inp_socket == NULL || 1032 inp->inp_socket->so_cred->cr_prison == NULL) { 1033 return (inp); 1034 } else { 1035 if (jinp == NULL) 1036 jinp = inp; 1037 } 1038 } 1039 } 1040 if (jinp != NULL) 1041 return (jinp); 1042 if (wildcard) { 1043 struct inpcb *local_wild = NULL; 1044 struct inpcb *jinp_wild = NULL; 1045 #ifdef INET6 1046 struct inpcb *local_wild_mapped = NULL; 1047 #endif 1048 struct inpcontainer *ic; 1049 struct inpcontainerhead *chead; 1050 struct sockaddr_in jsin; 1051 struct ucred *cred; 1052 1053 /* 1054 * Order of socket selection: 1055 * 1. non-jailed, non-wild. 1056 * 2. non-jailed, wild. 1057 * 3. jailed, non-wild. 1058 * 4. jailed, wild. 1059 */ 1060 jsin.sin_family = AF_INET; 1061 chead = &pcbinfo->wildcardhashbase[ 1062 INP_PCBWILDCARDHASH(lport, pcbinfo->wildcardhashmask)]; 1063 LIST_FOREACH(ic, chead, ic_list) { 1064 inp = ic->ic_inp; 1065 jsin.sin_addr.s_addr = laddr.s_addr; 1066 #ifdef INET6 1067 if (!(inp->inp_vflag & INP_IPV4)) 1068 continue; 1069 #endif 1070 if (inp->inp_socket != NULL) 1071 cred = inp->inp_socket->so_cred; 1072 else 1073 cred = NULL; 1074 if (cred != NULL && jailed(cred)) { 1075 if (jinp != NULL) 1076 continue; 1077 else 1078 if (!jailed_ip(cred->cr_prison, 1079 (struct sockaddr *)&jsin)) 1080 continue; 1081 } 1082 if (inp->inp_lport == lport) { 1083 if (ifp && ifp->if_type == IFT_FAITH && 1084 !(inp->inp_flags & INP_FAITH)) 1085 continue; 1086 if (inp->inp_laddr.s_addr == laddr.s_addr) { 1087 if (cred != NULL && jailed(cred)) 1088 jinp = inp; 1089 else 1090 return (inp); 1091 } 1092 if (inp->inp_laddr.s_addr == INADDR_ANY) { 1093 #ifdef INET6 1094 if (INP_CHECK_SOCKAF(inp->inp_socket, 1095 AF_INET6)) 1096 local_wild_mapped = inp; 1097 else 1098 #endif 1099 if (cred != NULL && 1100 jailed(cred)) 1101 jinp_wild = inp; 1102 else 1103 local_wild = inp; 1104 } 1105 } 1106 } 1107 if (local_wild != NULL) 1108 return (local_wild); 1109 #ifdef INET6 1110 if (local_wild_mapped != NULL) 1111 return (local_wild_mapped); 1112 #endif 1113 if (jinp != NULL) 1114 return (jinp); 1115 return (jinp_wild); 1116 } 1117 1118 /* 1119 * Not found. 1120 */ 1121 return (NULL); 1122 } 1123 1124 /* 1125 * Insert PCB into connection hash table. 1126 */ 1127 void 1128 in_pcbinsconnhash(struct inpcb *inp) 1129 { 1130 struct inpcbinfo *pcbinfo = inp->inp_cpcbinfo; 1131 struct inpcbhead *bucket; 1132 u_int32_t hashkey_faddr, hashkey_laddr; 1133 1134 #ifdef INET6 1135 if (inp->inp_vflag & INP_IPV6) { 1136 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX JH */; 1137 hashkey_laddr = inp->in6p_laddr.s6_addr32[3] /* XXX JH */; 1138 } else { 1139 #endif 1140 hashkey_faddr = inp->inp_faddr.s_addr; 1141 hashkey_laddr = inp->inp_laddr.s_addr; 1142 #ifdef INET6 1143 } 1144 #endif 1145 1146 KASSERT(!(inp->inp_flags & INP_CONNECTED), ("already on hash list")); 1147 inp->inp_flags |= INP_CONNECTED; 1148 1149 /* 1150 * Insert into the connection hash table. 1151 */ 1152 bucket = &pcbinfo->hashbase[INP_PCBCONNHASH(hashkey_faddr, 1153 inp->inp_fport, hashkey_laddr, inp->inp_lport, pcbinfo->hashmask)]; 1154 LIST_INSERT_HEAD(bucket, inp, inp_hash); 1155 } 1156 1157 /* 1158 * Remove PCB from connection hash table. 1159 */ 1160 void 1161 in_pcbremconnhash(struct inpcb *inp) 1162 { 1163 KASSERT(inp->inp_flags & INP_CONNECTED, ("inp not connected")); 1164 LIST_REMOVE(inp, inp_hash); 1165 inp->inp_flags &= ~INP_CONNECTED; 1166 } 1167 1168 /* 1169 * Insert PCB into port hash table. 1170 */ 1171 int 1172 in_pcbinsporthash(struct inpcb *inp) 1173 { 1174 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1175 struct inpcbporthead *pcbporthash; 1176 struct inpcbport *phd; 1177 1178 /* 1179 * If the porthashbase is shared across several cpus we need 1180 * to lock. 1181 */ 1182 if (pcbinfo->porttoken) 1183 lwkt_gettoken(pcbinfo->porttoken); 1184 1185 /* 1186 * Insert into the port hash table. 1187 */ 1188 pcbporthash = &pcbinfo->porthashbase[ 1189 INP_PCBPORTHASH(inp->inp_lport, pcbinfo->porthashmask)]; 1190 1191 /* Go through port list and look for a head for this lport. */ 1192 LIST_FOREACH(phd, pcbporthash, phd_hash) { 1193 if (phd->phd_port == inp->inp_lport) 1194 break; 1195 } 1196 1197 /* If none exists, malloc one and tack it on. */ 1198 if (phd == NULL) { 1199 KKASSERT(pcbinfo->portsave != NULL); 1200 phd = pcbinfo->portsave; 1201 pcbinfo->portsave = NULL; 1202 phd->phd_port = inp->inp_lport; 1203 LIST_INIT(&phd->phd_pcblist); 1204 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash); 1205 } 1206 1207 inp->inp_phd = phd; 1208 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist); 1209 1210 if (pcbinfo->porttoken) 1211 lwkt_reltoken(pcbinfo->porttoken); 1212 if (pcbinfo->portsave == NULL) { 1213 pcbinfo->portsave = kmalloc(sizeof(*pcbinfo->portsave), 1214 M_PCB, M_INTWAIT | M_ZERO); 1215 } 1216 return (0); 1217 } 1218 1219 void 1220 in_pcbinswildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo) 1221 { 1222 struct inpcontainer *ic; 1223 struct inpcontainerhead *bucket; 1224 1225 bucket = &pcbinfo->wildcardhashbase[ 1226 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)]; 1227 1228 ic = kmalloc(sizeof(struct inpcontainer), M_TEMP, M_INTWAIT); 1229 ic->ic_inp = inp; 1230 LIST_INSERT_HEAD(bucket, ic, ic_list); 1231 } 1232 1233 /* 1234 * Insert PCB into wildcard hash table. 1235 */ 1236 void 1237 in_pcbinswildcardhash(struct inpcb *inp) 1238 { 1239 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1240 1241 KKASSERT(pcbinfo != NULL); 1242 1243 in_pcbinswildcardhash_oncpu(inp, pcbinfo); 1244 inp->inp_flags |= INP_WILDCARD; 1245 } 1246 1247 void 1248 in_pcbremwildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo) 1249 { 1250 struct inpcontainer *ic; 1251 struct inpcontainerhead *head; 1252 1253 /* find bucket */ 1254 head = &pcbinfo->wildcardhashbase[ 1255 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)]; 1256 1257 LIST_FOREACH(ic, head, ic_list) { 1258 if (ic->ic_inp == inp) 1259 goto found; 1260 } 1261 return; /* not found! */ 1262 1263 found: 1264 LIST_REMOVE(ic, ic_list); /* remove container from bucket chain */ 1265 kfree(ic, M_TEMP); /* deallocate container */ 1266 } 1267 1268 /* 1269 * Remove PCB from wildcard hash table. 1270 */ 1271 void 1272 in_pcbremwildcardhash(struct inpcb *inp) 1273 { 1274 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1275 1276 KASSERT(inp->inp_flags & INP_WILDCARD, ("inp not wildcard")); 1277 in_pcbremwildcardhash_oncpu(inp, pcbinfo); 1278 inp->inp_flags &= ~INP_WILDCARD; 1279 } 1280 1281 /* 1282 * Remove PCB from various lists. 1283 */ 1284 void 1285 in_pcbremlists(struct inpcb *inp) 1286 { 1287 struct inpcbinfo *pcbinfo; 1288 1289 if (inp->inp_lport) { 1290 struct inpcbport *phd; 1291 1292 pcbinfo = inp->inp_pcbinfo; 1293 if (pcbinfo->porttoken) 1294 lwkt_gettoken(pcbinfo->porttoken); 1295 1296 phd = inp->inp_phd; 1297 LIST_REMOVE(inp, inp_portlist); 1298 if (LIST_FIRST(&phd->phd_pcblist) == NULL) { 1299 LIST_REMOVE(phd, phd_hash); 1300 kfree(phd, M_PCB); 1301 } 1302 if (pcbinfo->porttoken) 1303 lwkt_reltoken(pcbinfo->porttoken); 1304 } 1305 if (inp->inp_flags & INP_WILDCARD) { 1306 in_pcbremwildcardhash(inp); 1307 } else if (inp->inp_flags & INP_CONNECTED) { 1308 in_pcbremconnhash(inp); 1309 } 1310 LIST_REMOVE(inp, inp_list); 1311 inp->inp_pcbinfo->ipi_count--; 1312 } 1313 1314 int 1315 prison_xinpcb(struct thread *td, struct inpcb *inp) 1316 { 1317 struct ucred *cr; 1318 1319 if (td->td_proc == NULL) 1320 return (0); 1321 cr = td->td_proc->p_ucred; 1322 if (cr->cr_prison == NULL) 1323 return (0); 1324 if (inp->inp_socket && inp->inp_socket->so_cred && 1325 inp->inp_socket->so_cred->cr_prison && 1326 cr->cr_prison == inp->inp_socket->so_cred->cr_prison) 1327 return (0); 1328 return (1); 1329 } 1330 1331 int 1332 in_pcblist_global(SYSCTL_HANDLER_ARGS) 1333 { 1334 struct inpcbinfo *pcbinfo = arg1; 1335 struct inpcb *inp, *marker; 1336 struct xinpcb xi; 1337 int error, i, n; 1338 1339 /* 1340 * The process of preparing the TCB list is too time-consuming and 1341 * resource-intensive to repeat twice on every request. 1342 */ 1343 if (req->oldptr == NULL) { 1344 n = pcbinfo->ipi_count; 1345 req->oldidx = (n + n/8 + 10) * sizeof(struct xinpcb); 1346 return 0; 1347 } 1348 1349 if (req->newptr != NULL) 1350 return EPERM; 1351 1352 /* 1353 * OK, now we're committed to doing something. Re-fetch ipi_count 1354 * after obtaining the generation count. 1355 */ 1356 n = pcbinfo->ipi_count; 1357 1358 marker = kmalloc(sizeof(struct inpcb), M_TEMP, M_WAITOK|M_ZERO); 1359 marker->inp_flags |= INP_PLACEMARKER; 1360 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list); 1361 1362 i = 0; 1363 error = 0; 1364 1365 while ((inp = LIST_NEXT(marker, inp_list)) != NULL && i < n) { 1366 LIST_REMOVE(marker, inp_list); 1367 LIST_INSERT_AFTER(inp, marker, inp_list); 1368 1369 if (inp->inp_flags & INP_PLACEMARKER) 1370 continue; 1371 if (prison_xinpcb(req->td, inp)) 1372 continue; 1373 bzero(&xi, sizeof xi); 1374 xi.xi_len = sizeof xi; 1375 bcopy(inp, &xi.xi_inp, sizeof *inp); 1376 if (inp->inp_socket) 1377 sotoxsocket(inp->inp_socket, &xi.xi_socket); 1378 if ((error = SYSCTL_OUT(req, &xi, sizeof xi)) != 0) 1379 break; 1380 ++i; 1381 } 1382 LIST_REMOVE(marker, inp_list); 1383 if (error == 0 && i < n) { 1384 bzero(&xi, sizeof xi); 1385 xi.xi_len = sizeof xi; 1386 while (i < n) { 1387 error = SYSCTL_OUT(req, &xi, sizeof xi); 1388 ++i; 1389 } 1390 } 1391 kfree(marker, M_TEMP); 1392 return(error); 1393 } 1394