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) 2004 Jeffrey M. Hsu. All rights reserved. 36 * 37 * License terms: all terms for the DragonFly license above plus the following: 38 * 39 * 4. All advertising materials mentioning features or use of this software 40 * must display the following acknowledgement: 41 * 42 * This product includes software developed by Jeffrey M. Hsu 43 * for the DragonFly Project. 44 * 45 * This requirement may be waived with permission from Jeffrey Hsu. 46 * This requirement will sunset and may be removed on July 8 2005, 47 * after which the standard DragonFly license (as shown above) will 48 * apply. 49 */ 50 51 /* 52 * Copyright (c) 1982, 1986, 1991, 1993, 1995 53 * The Regents of the University of California. All rights reserved. 54 * 55 * Redistribution and use in source and binary forms, with or without 56 * modification, are permitted provided that the following conditions 57 * are met: 58 * 1. Redistributions of source code must retain the above copyright 59 * notice, this list of conditions and the following disclaimer. 60 * 2. Redistributions in binary form must reproduce the above copyright 61 * notice, this list of conditions and the following disclaimer in the 62 * documentation and/or other materials provided with the distribution. 63 * 3. All advertising materials mentioning features or use of this software 64 * must display the following acknowledgement: 65 * This product includes software developed by the University of 66 * California, Berkeley and its contributors. 67 * 4. Neither the name of the University nor the names of its contributors 68 * may be used to endorse or promote products derived from this software 69 * without specific prior written permission. 70 * 71 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 72 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 73 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 74 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 75 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 76 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 77 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 78 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 79 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 80 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 81 * SUCH DAMAGE. 82 * 83 * @(#)in_pcb.c 8.4 (Berkeley) 5/24/95 84 * $FreeBSD: src/sys/netinet/in_pcb.c,v 1.59.2.27 2004/01/02 04:06:42 ambrisko Exp $ 85 * $DragonFly: src/sys/netinet/in_pcb.c,v 1.36 2005/06/02 23:52:42 dillon Exp $ 86 */ 87 88 #include "opt_ipsec.h" 89 #include "opt_inet6.h" 90 91 #include <sys/param.h> 92 #include <sys/systm.h> 93 #include <sys/malloc.h> 94 #include <sys/mbuf.h> 95 #include <sys/domain.h> 96 #include <sys/protosw.h> 97 #include <sys/socket.h> 98 #include <sys/socketvar.h> 99 #include <sys/proc.h> 100 #include <sys/jail.h> 101 #include <sys/kernel.h> 102 #include <sys/sysctl.h> 103 #include <sys/thread2.h> 104 105 #include <machine/limits.h> 106 107 #include <vm/vm_zone.h> 108 109 #include <net/if.h> 110 #include <net/if_types.h> 111 #include <net/route.h> 112 113 #include <netinet/in.h> 114 #include <netinet/in_pcb.h> 115 #include <netinet/in_var.h> 116 #include <netinet/ip_var.h> 117 #ifdef INET6 118 #include <netinet/ip6.h> 119 #include <netinet6/ip6_var.h> 120 #endif /* INET6 */ 121 122 #ifdef IPSEC 123 #include <netinet6/ipsec.h> 124 #include <netproto/key/key.h> 125 #endif 126 127 #ifdef FAST_IPSEC 128 #if defined(IPSEC) || defined(IPSEC_ESP) 129 #error "Bad idea: don't compile with both IPSEC and FAST_IPSEC!" 130 #endif 131 132 #include <netproto/ipsec/ipsec.h> 133 #include <netproto/ipsec/key.h> 134 #define IPSEC 135 #endif /* FAST_IPSEC */ 136 137 struct in_addr zeroin_addr; 138 139 /* 140 * These configure the range of local port addresses assigned to 141 * "unspecified" outgoing connections/packets/whatever. 142 */ 143 int ipport_lowfirstauto = IPPORT_RESERVED - 1; /* 1023 */ 144 int ipport_lowlastauto = IPPORT_RESERVEDSTART; /* 600 */ 145 146 int ipport_firstauto = IPPORT_RESERVED; /* 1024 */ 147 int ipport_lastauto = IPPORT_USERRESERVED; /* 5000 */ 148 149 int ipport_hifirstauto = IPPORT_HIFIRSTAUTO; /* 49152 */ 150 int ipport_hilastauto = IPPORT_HILASTAUTO; /* 65535 */ 151 152 static __inline void 153 RANGECHK(int var, int min, int max) 154 { 155 if (var < min) 156 var = min; 157 else if (var > max) 158 var = max; 159 } 160 161 static int 162 sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS) 163 { 164 int error; 165 166 error = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req); 167 if (!error) { 168 RANGECHK(ipport_lowfirstauto, 1, IPPORT_RESERVED - 1); 169 RANGECHK(ipport_lowlastauto, 1, IPPORT_RESERVED - 1); 170 171 RANGECHK(ipport_firstauto, IPPORT_RESERVED, USHRT_MAX); 172 RANGECHK(ipport_lastauto, IPPORT_RESERVED, USHRT_MAX); 173 174 RANGECHK(ipport_hifirstauto, IPPORT_RESERVED, USHRT_MAX); 175 RANGECHK(ipport_hilastauto, IPPORT_RESERVED, USHRT_MAX); 176 } 177 return (error); 178 } 179 180 SYSCTL_NODE(_net_inet_ip, IPPROTO_IP, portrange, CTLFLAG_RW, 0, "IP Ports"); 181 182 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowfirst, CTLTYPE_INT|CTLFLAG_RW, 183 &ipport_lowfirstauto, 0, &sysctl_net_ipport_check, "I", ""); 184 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, lowlast, CTLTYPE_INT|CTLFLAG_RW, 185 &ipport_lowlastauto, 0, &sysctl_net_ipport_check, "I", ""); 186 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, first, CTLTYPE_INT|CTLFLAG_RW, 187 &ipport_firstauto, 0, &sysctl_net_ipport_check, "I", ""); 188 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, last, CTLTYPE_INT|CTLFLAG_RW, 189 &ipport_lastauto, 0, &sysctl_net_ipport_check, "I", ""); 190 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hifirst, CTLTYPE_INT|CTLFLAG_RW, 191 &ipport_hifirstauto, 0, &sysctl_net_ipport_check, "I", ""); 192 SYSCTL_PROC(_net_inet_ip_portrange, OID_AUTO, hilast, CTLTYPE_INT|CTLFLAG_RW, 193 &ipport_hilastauto, 0, &sysctl_net_ipport_check, "I", ""); 194 195 /* 196 * in_pcb.c: manage the Protocol Control Blocks. 197 * 198 * NOTE: It is assumed that most of these functions will be called from 199 * a critical section. XXX - There are, unfortunately, a few exceptions 200 * to this rule that should be fixed. 201 * 202 * NOTE: The caller should initialize the cpu field to the cpu running the 203 * protocol stack associated with this inpcbinfo. 204 */ 205 206 void 207 in_pcbinfo_init(struct inpcbinfo *pcbinfo) 208 { 209 LIST_INIT(&pcbinfo->pcblisthead); 210 pcbinfo->cpu = -1; 211 } 212 213 /* 214 * Allocate a PCB and associate it with the socket. 215 */ 216 int 217 in_pcballoc(struct socket *so, struct inpcbinfo *pcbinfo) 218 { 219 struct inpcb *inp; 220 #ifdef IPSEC 221 int error; 222 #endif 223 224 inp = zalloc(pcbinfo->ipi_zone); 225 if (inp == NULL) 226 return (ENOBUFS); 227 bzero(inp, sizeof *inp); 228 inp->inp_gencnt = ++pcbinfo->ipi_gencnt; 229 inp->inp_pcbinfo = inp->inp_cpcbinfo = pcbinfo; 230 inp->inp_socket = so; 231 #ifdef IPSEC 232 error = ipsec_init_policy(so, &inp->inp_sp); 233 if (error != 0) { 234 zfree(pcbinfo->ipi_zone, inp); 235 return (error); 236 } 237 #endif 238 #ifdef INET6 239 if (INP_SOCKAF(so) == AF_INET6 && ip6_v6only) 240 inp->inp_flags |= IN6P_IPV6_V6ONLY; 241 if (ip6_auto_flowlabel) 242 inp->inp_flags |= IN6P_AUTOFLOWLABEL; 243 #endif 244 so->so_pcb = inp; 245 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, inp, inp_list); 246 pcbinfo->ipi_count++; 247 return (0); 248 } 249 250 int 251 in_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 252 { 253 struct socket *so = inp->inp_socket; 254 struct proc *p = td->td_proc; 255 unsigned short *lastport; 256 struct sockaddr_in *sin; 257 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 258 u_short lport = 0; 259 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 260 int error, prison = 0; 261 262 KKASSERT(p); 263 264 if (TAILQ_EMPTY(&in_ifaddrhead)) /* XXX broken! */ 265 return (EADDRNOTAVAIL); 266 if (inp->inp_lport != 0 || inp->inp_laddr.s_addr != INADDR_ANY) 267 return (EINVAL); /* already bound */ 268 if (!(so->so_options & (SO_REUSEADDR|SO_REUSEPORT))) 269 wild = 1; /* neither SO_REUSEADDR nor SO_REUSEPORT is set */ 270 if (nam != NULL) { 271 sin = (struct sockaddr_in *)nam; 272 if (nam->sa_len != sizeof *sin) 273 return (EINVAL); 274 #ifdef notdef 275 /* 276 * We should check the family, but old programs 277 * incorrectly fail to initialize it. 278 */ 279 if (sin->sin_family != AF_INET) 280 return (EAFNOSUPPORT); 281 #endif 282 if (sin->sin_addr.s_addr != INADDR_ANY && 283 prison_ip(td, 0, &sin->sin_addr.s_addr)) 284 return (EINVAL); 285 lport = sin->sin_port; 286 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { 287 /* 288 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 289 * allow complete duplication of binding if 290 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 291 * and a multicast address is bound on both 292 * new and duplicated sockets. 293 */ 294 if (so->so_options & SO_REUSEADDR) 295 reuseport = SO_REUSEADDR | SO_REUSEPORT; 296 } else if (sin->sin_addr.s_addr != INADDR_ANY) { 297 sin->sin_port = 0; /* yech... */ 298 bzero(&sin->sin_zero, sizeof sin->sin_zero); 299 if (ifa_ifwithaddr((struct sockaddr *)sin) == NULL) 300 return (EADDRNOTAVAIL); 301 } 302 if (lport != 0) { 303 struct inpcb *t; 304 305 /* GROSS */ 306 if (ntohs(lport) < IPPORT_RESERVED && 307 p && suser_cred(p->p_ucred, PRISON_ROOT)) 308 return (EACCES); 309 if (p && p->p_ucred->cr_prison) 310 prison = 1; 311 if (so->so_cred->cr_uid != 0 && 312 !IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { 313 t = in_pcblookup_local(inp->inp_pcbinfo, 314 sin->sin_addr, lport, 315 prison ? 0 : INPLOOKUP_WILDCARD); 316 if (t && 317 (!in_nullhost(sin->sin_addr) || 318 !in_nullhost(t->inp_laddr) || 319 (t->inp_socket->so_options & 320 SO_REUSEPORT) == 0) && 321 (so->so_cred->cr_uid != 322 t->inp_socket->so_cred->cr_uid)) { 323 #ifdef INET6 324 if (!in_nullhost(sin->sin_addr) || 325 !in_nullhost(t->inp_laddr) || 326 INP_SOCKAF(so) == 327 INP_SOCKAF(t->inp_socket)) 328 #endif 329 return (EADDRINUSE); 330 } 331 } 332 if (prison && prison_ip(td, 0, &sin->sin_addr.s_addr)) 333 return (EADDRNOTAVAIL); 334 t = in_pcblookup_local(pcbinfo, sin->sin_addr, 335 lport, prison ? 0 : wild); 336 if (t && !(reuseport & t->inp_socket->so_options)) { 337 #ifdef INET6 338 if (!in_nullhost(sin->sin_addr) || 339 !in_nullhost(t->inp_laddr) || 340 INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket)) 341 #endif 342 return (EADDRINUSE); 343 } 344 } 345 inp->inp_laddr = sin->sin_addr; 346 } 347 if (lport == 0) { 348 ushort first, last; 349 int count; 350 351 if (inp->inp_laddr.s_addr != INADDR_ANY && 352 prison_ip(td, 0, &inp->inp_laddr.s_addr )) { 353 inp->inp_laddr.s_addr = INADDR_ANY; 354 return (EINVAL); 355 } 356 inp->inp_flags |= INP_ANONPORT; 357 358 if (inp->inp_flags & INP_HIGHPORT) { 359 first = ipport_hifirstauto; /* sysctl */ 360 last = ipport_hilastauto; 361 lastport = &pcbinfo->lasthi; 362 } else if (inp->inp_flags & INP_LOWPORT) { 363 if (p && 364 (error = suser_cred(p->p_ucred, PRISON_ROOT))) { 365 inp->inp_laddr.s_addr = INADDR_ANY; 366 return (error); 367 } 368 first = ipport_lowfirstauto; /* 1023 */ 369 last = ipport_lowlastauto; /* 600 */ 370 lastport = &pcbinfo->lastlow; 371 } else { 372 first = ipport_firstauto; /* sysctl */ 373 last = ipport_lastauto; 374 lastport = &pcbinfo->lastport; 375 } 376 /* 377 * Simple check to ensure all ports are not used up causing 378 * a deadlock here. 379 * 380 * We split the two cases (up and down) so that the direction 381 * is not being tested on each round of the loop. 382 */ 383 if (first > last) { 384 /* 385 * counting down 386 */ 387 count = first - last; 388 389 do { 390 if (count-- < 0) { /* completely used? */ 391 inp->inp_laddr.s_addr = INADDR_ANY; 392 return (EADDRNOTAVAIL); 393 } 394 --*lastport; 395 if (*lastport > first || *lastport < last) 396 *lastport = first; 397 lport = htons(*lastport); 398 } while (in_pcblookup_local(pcbinfo, 399 inp->inp_laddr, lport, wild)); 400 } else { 401 /* 402 * counting up 403 */ 404 count = last - first; 405 406 do { 407 if (count-- < 0) { /* completely used? */ 408 inp->inp_laddr.s_addr = INADDR_ANY; 409 return (EADDRNOTAVAIL); 410 } 411 ++*lastport; 412 if (*lastport < first || *lastport > last) 413 *lastport = first; 414 lport = htons(*lastport); 415 } while (in_pcblookup_local(pcbinfo, 416 inp->inp_laddr, lport, wild)); 417 } 418 } 419 inp->inp_lport = lport; 420 if (prison_ip(td, 0, &inp->inp_laddr.s_addr)) { 421 inp->inp_laddr.s_addr = INADDR_ANY; 422 inp->inp_lport = 0; 423 return (EINVAL); 424 } 425 if (in_pcbinsporthash(inp) != 0) { 426 inp->inp_laddr.s_addr = INADDR_ANY; 427 inp->inp_lport = 0; 428 return (EAGAIN); 429 } 430 return (0); 431 } 432 433 /* 434 * Transform old in_pcbconnect() into an inner subroutine for new 435 * in_pcbconnect(): Do some validity-checking on the remote 436 * address (in mbuf 'nam') and then determine local host address 437 * (i.e., which interface) to use to access that remote host. 438 * 439 * This preserves definition of in_pcbconnect(), while supporting a 440 * slightly different version for T/TCP. (This is more than 441 * a bit of a kludge, but cleaning up the internal interfaces would 442 * have forced minor changes in every protocol). 443 */ 444 int 445 in_pcbladdr(inp, nam, plocal_sin) 446 struct inpcb *inp; 447 struct sockaddr *nam; 448 struct sockaddr_in **plocal_sin; 449 { 450 struct in_ifaddr *ia; 451 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 452 453 if (nam->sa_len != sizeof *sin) 454 return (EINVAL); 455 if (sin->sin_family != AF_INET) 456 return (EAFNOSUPPORT); 457 if (sin->sin_port == 0) 458 return (EADDRNOTAVAIL); 459 if (!TAILQ_EMPTY(&in_ifaddrhead)) { 460 ia = TAILQ_FIRST(&in_ifaddrhead); 461 /* 462 * If the destination address is INADDR_ANY, 463 * use the primary local address. 464 * If the supplied address is INADDR_BROADCAST, 465 * and the primary interface supports broadcast, 466 * choose the broadcast address for that interface. 467 */ 468 if (sin->sin_addr.s_addr == INADDR_ANY) 469 sin->sin_addr = IA_SIN(ia)->sin_addr; 470 else if (sin->sin_addr.s_addr == (u_long)INADDR_BROADCAST && 471 (ia->ia_ifp->if_flags & IFF_BROADCAST)) 472 sin->sin_addr = satosin(&ia->ia_broadaddr)->sin_addr; 473 } 474 if (inp->inp_laddr.s_addr == INADDR_ANY) { 475 struct route *ro; 476 477 ia = (struct in_ifaddr *)NULL; 478 /* 479 * If route is known or can be allocated now, 480 * our src addr is taken from the i/f, else punt. 481 * Note that we should check the address family of the cached 482 * destination, in case of sharing the cache with IPv6. 483 */ 484 ro = &inp->inp_route; 485 if (ro->ro_rt && 486 (!(ro->ro_rt->rt_flags & RTF_UP) || 487 ro->ro_dst.sa_family != AF_INET || 488 satosin(&ro->ro_dst)->sin_addr.s_addr != 489 sin->sin_addr.s_addr || 490 inp->inp_socket->so_options & SO_DONTROUTE)) { 491 RTFREE(ro->ro_rt); 492 ro->ro_rt = (struct rtentry *)NULL; 493 } 494 if (!(inp->inp_socket->so_options & SO_DONTROUTE) && /*XXX*/ 495 (ro->ro_rt == (struct rtentry *)NULL || 496 ro->ro_rt->rt_ifp == (struct ifnet *)NULL)) { 497 /* No route yet, so try to acquire one */ 498 bzero(&ro->ro_dst, sizeof(struct sockaddr_in)); 499 ro->ro_dst.sa_family = AF_INET; 500 ro->ro_dst.sa_len = sizeof(struct sockaddr_in); 501 ((struct sockaddr_in *) &ro->ro_dst)->sin_addr = 502 sin->sin_addr; 503 rtalloc(ro); 504 } 505 /* 506 * If we found a route, use the address 507 * corresponding to the outgoing interface 508 * unless it is the loopback (in case a route 509 * to our address on another net goes to loopback). 510 */ 511 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) 512 ia = ifatoia(ro->ro_rt->rt_ifa); 513 if (ia == NULL) { 514 u_short fport = sin->sin_port; 515 516 sin->sin_port = 0; 517 ia = ifatoia(ifa_ifwithdstaddr(sintosa(sin))); 518 if (ia == NULL) 519 ia = ifatoia(ifa_ifwithnet(sintosa(sin))); 520 sin->sin_port = fport; 521 if (ia == NULL) 522 ia = TAILQ_FIRST(&in_ifaddrhead); 523 if (ia == NULL) 524 return (EADDRNOTAVAIL); 525 } 526 /* 527 * If the destination address is multicast and an outgoing 528 * interface has been set as a multicast option, use the 529 * address of that interface as our source address. 530 */ 531 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) && 532 inp->inp_moptions != NULL) { 533 struct ip_moptions *imo; 534 struct ifnet *ifp; 535 536 imo = inp->inp_moptions; 537 if (imo->imo_multicast_ifp != NULL) { 538 ifp = imo->imo_multicast_ifp; 539 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) 540 if (ia->ia_ifp == ifp) 541 break; 542 if (ia == NULL) 543 return (EADDRNOTAVAIL); 544 } 545 } 546 /* 547 * Don't do pcblookup call here; return interface in plocal_sin 548 * and exit to caller, that will do the lookup. 549 */ 550 *plocal_sin = &ia->ia_addr; 551 552 } 553 return (0); 554 } 555 556 /* 557 * Outer subroutine: 558 * Connect from a socket to a specified address. 559 * Both address and port must be specified in argument sin. 560 * If don't have a local address for this socket yet, 561 * then pick one. 562 */ 563 int 564 in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 565 { 566 struct sockaddr_in *if_sin; 567 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 568 struct sockaddr_in sa; 569 struct ucred *cr = td->td_proc ? td->td_proc->p_ucred : NULL; 570 int error; 571 572 if (cr && cr->cr_prison != NULL && in_nullhost(inp->inp_laddr)) { 573 bzero(&sa, sizeof sa); 574 sa.sin_addr.s_addr = htonl(cr->cr_prison->pr_ip); 575 sa.sin_len = sizeof sa; 576 sa.sin_family = AF_INET; 577 error = in_pcbbind(inp, (struct sockaddr *)&sa, td); 578 if (error) 579 return (error); 580 } 581 582 /* Call inner routine to assign local interface address. */ 583 if ((error = in_pcbladdr(inp, nam, &if_sin)) != 0) 584 return (error); 585 586 if (in_pcblookup_hash(inp->inp_cpcbinfo, sin->sin_addr, sin->sin_port, 587 inp->inp_laddr.s_addr ? inp->inp_laddr : if_sin->sin_addr, 588 inp->inp_lport, FALSE, NULL) != NULL) { 589 return (EADDRINUSE); 590 } 591 if (inp->inp_laddr.s_addr == INADDR_ANY) { 592 if (inp->inp_lport == 0) { 593 error = in_pcbbind(inp, (struct sockaddr *)NULL, td); 594 if (error) 595 return (error); 596 } 597 inp->inp_laddr = if_sin->sin_addr; 598 } 599 inp->inp_faddr = sin->sin_addr; 600 inp->inp_fport = sin->sin_port; 601 in_pcbinsconnhash(inp); 602 return (0); 603 } 604 605 void 606 in_pcbdisconnect(inp) 607 struct inpcb *inp; 608 { 609 610 inp->inp_faddr.s_addr = INADDR_ANY; 611 inp->inp_fport = 0; 612 in_pcbremconnhash(inp); 613 if (inp->inp_socket->so_state & SS_NOFDREF) 614 in_pcbdetach(inp); 615 } 616 617 void 618 in_pcbdetach(inp) 619 struct inpcb *inp; 620 { 621 struct socket *so = inp->inp_socket; 622 struct inpcbinfo *ipi = inp->inp_pcbinfo; 623 624 #ifdef IPSEC 625 ipsec4_delete_pcbpolicy(inp); 626 #endif /*IPSEC*/ 627 inp->inp_gencnt = ++ipi->ipi_gencnt; 628 in_pcbremlists(inp); 629 so->so_pcb = 0; 630 sofree(so); 631 if (inp->inp_options) 632 m_free(inp->inp_options); 633 if (inp->inp_route.ro_rt) 634 rtfree(inp->inp_route.ro_rt); 635 ip_freemoptions(inp->inp_moptions); 636 inp->inp_vflag = 0; 637 zfree(ipi->ipi_zone, inp); 638 } 639 640 /* 641 * The calling convention of in_setsockaddr() and in_setpeeraddr() was 642 * modified to match the pru_sockaddr() and pru_peeraddr() entry points 643 * in struct pr_usrreqs, so that protocols can just reference then directly 644 * without the need for a wrapper function. The socket must have a valid 645 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one 646 * except through a kernel programming error, so it is acceptable to panic 647 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap 648 * because there actually /is/ a programming error somewhere... XXX) 649 */ 650 int 651 in_setsockaddr(so, nam) 652 struct socket *so; 653 struct sockaddr **nam; 654 { 655 struct inpcb *inp; 656 struct sockaddr_in *sin; 657 658 /* 659 * Do the malloc first in case it blocks. 660 */ 661 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, 662 M_WAITOK | M_ZERO); 663 sin->sin_family = AF_INET; 664 sin->sin_len = sizeof *sin; 665 666 crit_enter(); 667 inp = so->so_pcb; 668 if (!inp) { 669 crit_exit(); 670 free(sin, M_SONAME); 671 return (ECONNRESET); 672 } 673 sin->sin_port = inp->inp_lport; 674 sin->sin_addr = inp->inp_laddr; 675 crit_exit(); 676 677 *nam = (struct sockaddr *)sin; 678 return (0); 679 } 680 681 int 682 in_setpeeraddr(so, nam) 683 struct socket *so; 684 struct sockaddr **nam; 685 { 686 struct inpcb *inp; 687 struct sockaddr_in *sin; 688 689 /* 690 * Do the malloc first in case it blocks. 691 */ 692 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, 693 M_WAITOK | M_ZERO); 694 sin->sin_family = AF_INET; 695 sin->sin_len = sizeof *sin; 696 697 crit_enter(); 698 inp = so->so_pcb; 699 if (!inp) { 700 crit_exit(); 701 free(sin, M_SONAME); 702 return (ECONNRESET); 703 } 704 sin->sin_port = inp->inp_fport; 705 sin->sin_addr = inp->inp_faddr; 706 crit_exit(); 707 708 *nam = (struct sockaddr *)sin; 709 return (0); 710 } 711 712 void 713 in_pcbnotifyall(head, faddr, errno, notify) 714 struct inpcbhead *head; 715 struct in_addr faddr; 716 void (*notify) (struct inpcb *, int); 717 { 718 struct inpcb *inp, *ninp; 719 720 /* 721 * note: if INP_PLACEMARKER is set we must ignore the rest of 722 * the structure and skip it. 723 */ 724 crit_enter(); 725 LIST_FOREACH_MUTABLE(inp, head, inp_list, ninp) { 726 if (inp->inp_flags & INP_PLACEMARKER) 727 continue; 728 #ifdef INET6 729 if (!(inp->inp_vflag & INP_IPV4)) 730 continue; 731 #endif 732 if (inp->inp_faddr.s_addr != faddr.s_addr || 733 inp->inp_socket == NULL) 734 continue; 735 (*notify)(inp, errno); /* can remove inp from list! */ 736 } 737 crit_exit(); 738 } 739 740 void 741 in_pcbpurgeif0(head, ifp) 742 struct inpcb *head; 743 struct ifnet *ifp; 744 { 745 struct inpcb *inp; 746 struct ip_moptions *imo; 747 int i, gap; 748 749 for (inp = head; inp != NULL; inp = LIST_NEXT(inp, inp_list)) { 750 if (inp->inp_flags & INP_PLACEMARKER) 751 continue; 752 imo = inp->inp_moptions; 753 if ((inp->inp_vflag & INP_IPV4) && imo != NULL) { 754 /* 755 * Unselect the outgoing interface if it is being 756 * detached. 757 */ 758 if (imo->imo_multicast_ifp == ifp) 759 imo->imo_multicast_ifp = NULL; 760 761 /* 762 * Drop multicast group membership if we joined 763 * through the interface being detached. 764 */ 765 for (i = 0, gap = 0; i < imo->imo_num_memberships; 766 i++) { 767 if (imo->imo_membership[i]->inm_ifp == ifp) { 768 in_delmulti(imo->imo_membership[i]); 769 gap++; 770 } else if (gap != 0) 771 imo->imo_membership[i - gap] = 772 imo->imo_membership[i]; 773 } 774 imo->imo_num_memberships -= gap; 775 } 776 } 777 } 778 779 /* 780 * Check for alternatives when higher level complains 781 * about service problems. For now, invalidate cached 782 * routing information. If the route was created dynamically 783 * (by a redirect), time to try a default gateway again. 784 */ 785 void 786 in_losing(struct inpcb *inp) 787 { 788 struct rtentry *rt; 789 struct rt_addrinfo rtinfo; 790 791 if ((rt = inp->inp_route.ro_rt)) { 792 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 793 rtinfo.rti_info[RTAX_DST] = rt_key(rt); 794 rtinfo.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 795 rtinfo.rti_info[RTAX_NETMASK] = rt_mask(rt); 796 rtinfo.rti_flags = rt->rt_flags; 797 rt_missmsg(RTM_LOSING, &rtinfo, rt->rt_flags, 0); 798 if (rt->rt_flags & RTF_DYNAMIC) 799 rtrequest1(RTM_DELETE, &rtinfo, NULL); 800 inp->inp_route.ro_rt = NULL; 801 rtfree(rt); 802 /* 803 * A new route can be allocated 804 * the next time output is attempted. 805 */ 806 } 807 } 808 809 /* 810 * After a routing change, flush old routing 811 * and allocate a (hopefully) better one. 812 */ 813 void 814 in_rtchange(inp, errno) 815 struct inpcb *inp; 816 int errno; 817 { 818 if (inp->inp_route.ro_rt) { 819 rtfree(inp->inp_route.ro_rt); 820 inp->inp_route.ro_rt = NULL; 821 /* 822 * A new route can be allocated the next time 823 * output is attempted. 824 */ 825 } 826 } 827 828 /* 829 * Lookup a PCB based on the local address and port. 830 */ 831 struct inpcb * 832 in_pcblookup_local(pcbinfo, laddr, lport_arg, wild_okay) 833 struct inpcbinfo *pcbinfo; 834 struct in_addr laddr; 835 u_int lport_arg; 836 int wild_okay; 837 { 838 struct inpcb *inp; 839 int matchwild = 3, wildcard; 840 u_short lport = lport_arg; 841 842 struct inpcbporthead *porthash; 843 struct inpcbport *phd; 844 struct inpcb *match = NULL; 845 846 /* 847 * Best fit PCB lookup. 848 * 849 * First see if this local port is in use by looking on the 850 * port hash list. 851 */ 852 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport, 853 pcbinfo->porthashmask)]; 854 LIST_FOREACH(phd, porthash, phd_hash) { 855 if (phd->phd_port == lport) 856 break; 857 } 858 if (phd != NULL) { 859 /* 860 * Port is in use by one or more PCBs. Look for best 861 * fit. 862 */ 863 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 864 wildcard = 0; 865 #ifdef INET6 866 if ((inp->inp_vflag & INP_IPV4) == 0) 867 continue; 868 #endif 869 if (inp->inp_faddr.s_addr != INADDR_ANY) 870 wildcard++; 871 if (inp->inp_laddr.s_addr != INADDR_ANY) { 872 if (laddr.s_addr == INADDR_ANY) 873 wildcard++; 874 else if (inp->inp_laddr.s_addr != laddr.s_addr) 875 continue; 876 } else { 877 if (laddr.s_addr != INADDR_ANY) 878 wildcard++; 879 } 880 if (wildcard && !wild_okay) 881 continue; 882 if (wildcard < matchwild) { 883 match = inp; 884 matchwild = wildcard; 885 if (matchwild == 0) { 886 break; 887 } 888 } 889 } 890 } 891 return (match); 892 } 893 894 /* 895 * Lookup PCB in hash list. 896 */ 897 struct inpcb * 898 in_pcblookup_hash(pcbinfo, faddr, fport_arg, laddr, lport_arg, wildcard, ifp) 899 struct inpcbinfo *pcbinfo; 900 struct in_addr faddr, laddr; 901 u_int fport_arg, lport_arg; 902 boolean_t wildcard; 903 struct ifnet *ifp; 904 { 905 struct inpcbhead *head; 906 struct inpcb *inp; 907 u_short fport = fport_arg, lport = lport_arg; 908 909 /* 910 * First look for an exact match. 911 */ 912 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr.s_addr, fport, 913 laddr.s_addr, lport, pcbinfo->hashmask)]; 914 LIST_FOREACH(inp, head, inp_hash) { 915 #ifdef INET6 916 if (!(inp->inp_vflag & INP_IPV4)) 917 continue; 918 #endif 919 if (in_hosteq(inp->inp_faddr, faddr) && 920 in_hosteq(inp->inp_laddr, laddr) && 921 inp->inp_fport == fport && inp->inp_lport == lport) { 922 /* found */ 923 return (inp); 924 } 925 } 926 927 if (wildcard) { 928 struct inpcb *local_wild = NULL; 929 #ifdef INET6 930 struct inpcb *local_wild_mapped = NULL; 931 #endif 932 struct inpcontainer *ic; 933 struct inpcontainerhead *chead; 934 935 chead = &pcbinfo->wildcardhashbase[ 936 INP_PCBWILDCARDHASH(lport, pcbinfo->wildcardhashmask)]; 937 LIST_FOREACH(ic, chead, ic_list) { 938 inp = ic->ic_inp; 939 #ifdef INET6 940 if (!(inp->inp_vflag & INP_IPV4)) 941 continue; 942 #endif 943 if (inp->inp_lport == lport) { 944 if (ifp && ifp->if_type == IFT_FAITH && 945 !(inp->inp_flags & INP_FAITH)) 946 continue; 947 if (inp->inp_laddr.s_addr == laddr.s_addr) 948 return (inp); 949 if (inp->inp_laddr.s_addr == INADDR_ANY) { 950 #ifdef INET6 951 if (INP_CHECK_SOCKAF(inp->inp_socket, 952 AF_INET6)) 953 local_wild_mapped = inp; 954 else 955 #endif 956 local_wild = inp; 957 } 958 } 959 } 960 #ifdef INET6 961 if (local_wild == NULL) 962 return (local_wild_mapped); 963 #endif 964 return (local_wild); 965 } 966 967 /* 968 * Not found. 969 */ 970 return (NULL); 971 } 972 973 /* 974 * Insert PCB into connection hash table. 975 */ 976 void 977 in_pcbinsconnhash(struct inpcb *inp) 978 { 979 struct inpcbinfo *pcbinfo = inp->inp_cpcbinfo; 980 struct inpcbhead *bucket; 981 u_int32_t hashkey_faddr, hashkey_laddr; 982 983 #ifdef INET6 984 if (inp->inp_vflag & INP_IPV6) { 985 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX JH */; 986 hashkey_laddr = inp->in6p_laddr.s6_addr32[3] /* XXX JH */; 987 } else { 988 #endif 989 hashkey_faddr = inp->inp_faddr.s_addr; 990 hashkey_laddr = inp->inp_laddr.s_addr; 991 #ifdef INET6 992 } 993 #endif 994 995 KASSERT(!(inp->inp_flags & INP_CONNECTED), ("already on hash list")); 996 inp->inp_flags |= INP_CONNECTED; 997 998 /* 999 * Insert into the connection hash table. 1000 */ 1001 bucket = &pcbinfo->hashbase[INP_PCBCONNHASH(hashkey_faddr, 1002 inp->inp_fport, hashkey_laddr, inp->inp_lport, pcbinfo->hashmask)]; 1003 LIST_INSERT_HEAD(bucket, inp, inp_hash); 1004 } 1005 1006 /* 1007 * Remove PCB from connection hash table. 1008 */ 1009 void 1010 in_pcbremconnhash(struct inpcb *inp) 1011 { 1012 KASSERT(inp->inp_flags & INP_CONNECTED, ("inp not connected")); 1013 LIST_REMOVE(inp, inp_hash); 1014 inp->inp_flags &= ~INP_CONNECTED; 1015 } 1016 1017 /* 1018 * Insert PCB into port hash table. 1019 */ 1020 int 1021 in_pcbinsporthash(struct inpcb *inp) 1022 { 1023 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1024 struct inpcbporthead *pcbporthash; 1025 struct inpcbport *phd; 1026 1027 /* 1028 * Insert into the port hash table. 1029 */ 1030 pcbporthash = &pcbinfo->porthashbase[ 1031 INP_PCBPORTHASH(inp->inp_lport, pcbinfo->porthashmask)]; 1032 1033 /* Go through port list and look for a head for this lport. */ 1034 LIST_FOREACH(phd, pcbporthash, phd_hash) 1035 if (phd->phd_port == inp->inp_lport) 1036 break; 1037 1038 /* If none exists, malloc one and tack it on. */ 1039 if (phd == NULL) { 1040 MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport), 1041 M_PCB, M_INTWAIT | M_NULLOK); 1042 if (phd == NULL) 1043 return (ENOBUFS); /* XXX */ 1044 phd->phd_port = inp->inp_lport; 1045 LIST_INIT(&phd->phd_pcblist); 1046 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash); 1047 } 1048 1049 inp->inp_phd = phd; 1050 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist); 1051 1052 return (0); 1053 } 1054 1055 void 1056 in_pcbinswildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo) 1057 { 1058 struct inpcontainer *ic; 1059 struct inpcontainerhead *bucket; 1060 1061 bucket = &pcbinfo->wildcardhashbase[ 1062 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)]; 1063 1064 ic = malloc(sizeof(struct inpcontainer), M_TEMP, M_INTWAIT); 1065 ic->ic_inp = inp; 1066 LIST_INSERT_HEAD(bucket, ic, ic_list); 1067 } 1068 1069 /* 1070 * Insert PCB into wildcard hash table. 1071 */ 1072 void 1073 in_pcbinswildcardhash(struct inpcb *inp) 1074 { 1075 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1076 1077 KKASSERT(pcbinfo != NULL); 1078 1079 in_pcbinswildcardhash_oncpu(inp, pcbinfo); 1080 inp->inp_flags |= INP_WILDCARD; 1081 } 1082 1083 void 1084 in_pcbremwildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo) 1085 { 1086 struct inpcontainer *ic; 1087 struct inpcontainerhead *head; 1088 1089 /* find bucket */ 1090 head = &pcbinfo->wildcardhashbase[ 1091 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)]; 1092 1093 LIST_FOREACH(ic, head, ic_list) { 1094 if (ic->ic_inp == inp) 1095 goto found; 1096 } 1097 return; /* not found! */ 1098 1099 found: 1100 LIST_REMOVE(ic, ic_list); /* remove container from bucket chain */ 1101 free(ic, M_TEMP); /* deallocate container */ 1102 } 1103 1104 /* 1105 * Remove PCB from wildcard hash table. 1106 */ 1107 void 1108 in_pcbremwildcardhash(struct inpcb *inp) 1109 { 1110 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1111 1112 KASSERT(inp->inp_flags & INP_WILDCARD, ("inp not wildcard")); 1113 in_pcbremwildcardhash_oncpu(inp, pcbinfo); 1114 inp->inp_flags &= ~INP_WILDCARD; 1115 } 1116 1117 /* 1118 * Remove PCB from various lists. 1119 */ 1120 void 1121 in_pcbremlists(inp) 1122 struct inpcb *inp; 1123 { 1124 if (inp->inp_lport) { 1125 struct inpcbport *phd = inp->inp_phd; 1126 1127 LIST_REMOVE(inp, inp_portlist); 1128 if (LIST_FIRST(&phd->phd_pcblist) == NULL) { 1129 LIST_REMOVE(phd, phd_hash); 1130 free(phd, M_PCB); 1131 } 1132 } 1133 if (inp->inp_flags & INP_WILDCARD) { 1134 in_pcbremwildcardhash(inp); 1135 } else if (inp->inp_flags & INP_CONNECTED) { 1136 in_pcbremconnhash(inp); 1137 } 1138 LIST_REMOVE(inp, inp_list); 1139 inp->inp_pcbinfo->ipi_count--; 1140 } 1141 1142 int 1143 prison_xinpcb(struct thread *td, struct inpcb *inp) 1144 { 1145 struct ucred *cr; 1146 1147 if (td->td_proc == NULL) 1148 return (0); 1149 cr = td->td_proc->p_ucred; 1150 if (cr->cr_prison == NULL) 1151 return (0); 1152 if (ntohl(inp->inp_laddr.s_addr) == cr->cr_prison->pr_ip) 1153 return (0); 1154 return (1); 1155 } 1156 1157 int 1158 in_pcblist_global(SYSCTL_HANDLER_ARGS) 1159 { 1160 struct inpcbinfo *pcbinfo = arg1; 1161 struct inpcb *inp, *marker; 1162 struct xinpcb xi; 1163 int error, i, n; 1164 inp_gen_t gencnt; 1165 1166 /* 1167 * The process of preparing the TCB list is too time-consuming and 1168 * resource-intensive to repeat twice on every request. 1169 */ 1170 if (req->oldptr == NULL) { 1171 n = pcbinfo->ipi_count; 1172 req->oldidx = (n + n/8 + 10) * sizeof(struct xinpcb); 1173 return 0; 1174 } 1175 1176 if (req->newptr != NULL) 1177 return EPERM; 1178 1179 /* 1180 * OK, now we're committed to doing something. Re-fetch ipi_count 1181 * after obtaining the generation count. 1182 */ 1183 gencnt = pcbinfo->ipi_gencnt; 1184 n = pcbinfo->ipi_count; 1185 1186 marker = malloc(sizeof(struct inpcb), M_TEMP, M_WAITOK|M_ZERO); 1187 marker->inp_flags |= INP_PLACEMARKER; 1188 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list); 1189 1190 i = 0; 1191 error = 0; 1192 1193 while ((inp = LIST_NEXT(marker, inp_list)) != NULL && i < n) { 1194 LIST_REMOVE(marker, inp_list); 1195 LIST_INSERT_AFTER(inp, marker, inp_list); 1196 1197 if (inp->inp_flags & INP_PLACEMARKER) 1198 continue; 1199 if (inp->inp_gencnt > gencnt) 1200 continue; 1201 if (prison_xinpcb(req->td, inp)) 1202 continue; 1203 bzero(&xi, sizeof xi); 1204 xi.xi_len = sizeof xi; 1205 bcopy(inp, &xi.xi_inp, sizeof *inp); 1206 if (inp->inp_socket) 1207 sotoxsocket(inp->inp_socket, &xi.xi_socket); 1208 if ((error = SYSCTL_OUT(req, &xi, sizeof xi)) != 0) 1209 break; 1210 ++i; 1211 } 1212 LIST_REMOVE(marker, inp_list); 1213 if (error == 0 && i < n) { 1214 bzero(&xi, sizeof xi); 1215 xi.xi_len = sizeof xi; 1216 while (i < n) { 1217 error = SYSCTL_OUT(req, &xi, sizeof xi); 1218 ++i; 1219 } 1220 } 1221 free(marker, M_TEMP); 1222 return(error); 1223 } 1224