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 1.38 2006/01/31 19:05:40 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(struct inpcb *inp, struct sockaddr *nam, struct sockaddr_in **plocal_sin) 446 { 447 struct in_ifaddr *ia; 448 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 449 450 if (nam->sa_len != sizeof *sin) 451 return (EINVAL); 452 if (sin->sin_family != AF_INET) 453 return (EAFNOSUPPORT); 454 if (sin->sin_port == 0) 455 return (EADDRNOTAVAIL); 456 if (!TAILQ_EMPTY(&in_ifaddrhead)) { 457 ia = TAILQ_FIRST(&in_ifaddrhead); 458 /* 459 * If the destination address is INADDR_ANY, 460 * use the primary local address. 461 * If the supplied address is INADDR_BROADCAST, 462 * and the primary interface supports broadcast, 463 * choose the broadcast address for that interface. 464 */ 465 if (sin->sin_addr.s_addr == INADDR_ANY) 466 sin->sin_addr = IA_SIN(ia)->sin_addr; 467 else if (sin->sin_addr.s_addr == (u_long)INADDR_BROADCAST && 468 (ia->ia_ifp->if_flags & IFF_BROADCAST)) 469 sin->sin_addr = satosin(&ia->ia_broadaddr)->sin_addr; 470 } 471 if (inp->inp_laddr.s_addr == INADDR_ANY) { 472 struct route *ro; 473 474 ia = (struct in_ifaddr *)NULL; 475 /* 476 * If route is known or can be allocated now, 477 * our src addr is taken from the i/f, else punt. 478 * Note that we should check the address family of the cached 479 * destination, in case of sharing the cache with IPv6. 480 */ 481 ro = &inp->inp_route; 482 if (ro->ro_rt && 483 (!(ro->ro_rt->rt_flags & RTF_UP) || 484 ro->ro_dst.sa_family != AF_INET || 485 satosin(&ro->ro_dst)->sin_addr.s_addr != 486 sin->sin_addr.s_addr || 487 inp->inp_socket->so_options & SO_DONTROUTE)) { 488 RTFREE(ro->ro_rt); 489 ro->ro_rt = (struct rtentry *)NULL; 490 } 491 if (!(inp->inp_socket->so_options & SO_DONTROUTE) && /*XXX*/ 492 (ro->ro_rt == (struct rtentry *)NULL || 493 ro->ro_rt->rt_ifp == (struct ifnet *)NULL)) { 494 /* No route yet, so try to acquire one */ 495 bzero(&ro->ro_dst, sizeof(struct sockaddr_in)); 496 ro->ro_dst.sa_family = AF_INET; 497 ro->ro_dst.sa_len = sizeof(struct sockaddr_in); 498 ((struct sockaddr_in *) &ro->ro_dst)->sin_addr = 499 sin->sin_addr; 500 rtalloc(ro); 501 } 502 /* 503 * If we found a route, use the address 504 * corresponding to the outgoing interface 505 * unless it is the loopback (in case a route 506 * to our address on another net goes to loopback). 507 */ 508 if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK)) 509 ia = ifatoia(ro->ro_rt->rt_ifa); 510 if (ia == NULL) { 511 u_short fport = sin->sin_port; 512 513 sin->sin_port = 0; 514 ia = ifatoia(ifa_ifwithdstaddr(sintosa(sin))); 515 if (ia == NULL) 516 ia = ifatoia(ifa_ifwithnet(sintosa(sin))); 517 sin->sin_port = fport; 518 if (ia == NULL) 519 ia = TAILQ_FIRST(&in_ifaddrhead); 520 if (ia == NULL) 521 return (EADDRNOTAVAIL); 522 } 523 /* 524 * If the destination address is multicast and an outgoing 525 * interface has been set as a multicast option, use the 526 * address of that interface as our source address. 527 */ 528 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) && 529 inp->inp_moptions != NULL) { 530 struct ip_moptions *imo; 531 struct ifnet *ifp; 532 533 imo = inp->inp_moptions; 534 if (imo->imo_multicast_ifp != NULL) { 535 ifp = imo->imo_multicast_ifp; 536 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) 537 if (ia->ia_ifp == ifp) 538 break; 539 if (ia == NULL) 540 return (EADDRNOTAVAIL); 541 } 542 } 543 /* 544 * Don't do pcblookup call here; return interface in plocal_sin 545 * and exit to caller, that will do the lookup. 546 */ 547 *plocal_sin = &ia->ia_addr; 548 549 } 550 return (0); 551 } 552 553 /* 554 * Outer subroutine: 555 * Connect from a socket to a specified address. 556 * Both address and port must be specified in argument sin. 557 * If don't have a local address for this socket yet, 558 * then pick one. 559 */ 560 int 561 in_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td) 562 { 563 struct sockaddr_in *if_sin; 564 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 565 struct sockaddr_in sa; 566 struct ucred *cr = td->td_proc ? td->td_proc->p_ucred : NULL; 567 int error; 568 569 if (cr && cr->cr_prison != NULL && in_nullhost(inp->inp_laddr)) { 570 bzero(&sa, sizeof sa); 571 sa.sin_addr.s_addr = htonl(cr->cr_prison->pr_ip); 572 sa.sin_len = sizeof sa; 573 sa.sin_family = AF_INET; 574 error = in_pcbbind(inp, (struct sockaddr *)&sa, td); 575 if (error) 576 return (error); 577 } 578 579 /* Call inner routine to assign local interface address. */ 580 if ((error = in_pcbladdr(inp, nam, &if_sin)) != 0) 581 return (error); 582 583 if (in_pcblookup_hash(inp->inp_cpcbinfo, sin->sin_addr, sin->sin_port, 584 inp->inp_laddr.s_addr ? inp->inp_laddr : if_sin->sin_addr, 585 inp->inp_lport, FALSE, NULL) != NULL) { 586 return (EADDRINUSE); 587 } 588 if (inp->inp_laddr.s_addr == INADDR_ANY) { 589 if (inp->inp_lport == 0) { 590 error = in_pcbbind(inp, (struct sockaddr *)NULL, td); 591 if (error) 592 return (error); 593 } 594 inp->inp_laddr = if_sin->sin_addr; 595 } 596 inp->inp_faddr = sin->sin_addr; 597 inp->inp_fport = sin->sin_port; 598 in_pcbinsconnhash(inp); 599 return (0); 600 } 601 602 void 603 in_pcbdisconnect(struct inpcb *inp) 604 { 605 606 inp->inp_faddr.s_addr = INADDR_ANY; 607 inp->inp_fport = 0; 608 in_pcbremconnhash(inp); 609 if (inp->inp_socket->so_state & SS_NOFDREF) 610 in_pcbdetach(inp); 611 } 612 613 void 614 in_pcbdetach(struct inpcb *inp) 615 { 616 struct socket *so = inp->inp_socket; 617 struct inpcbinfo *ipi = inp->inp_pcbinfo; 618 619 #ifdef IPSEC 620 ipsec4_delete_pcbpolicy(inp); 621 #endif /*IPSEC*/ 622 inp->inp_gencnt = ++ipi->ipi_gencnt; 623 in_pcbremlists(inp); 624 so->so_pcb = 0; 625 sofree(so); 626 if (inp->inp_options) 627 m_free(inp->inp_options); 628 if (inp->inp_route.ro_rt) 629 rtfree(inp->inp_route.ro_rt); 630 ip_freemoptions(inp->inp_moptions); 631 inp->inp_vflag = 0; 632 zfree(ipi->ipi_zone, inp); 633 } 634 635 /* 636 * The calling convention of in_setsockaddr() and in_setpeeraddr() was 637 * modified to match the pru_sockaddr() and pru_peeraddr() entry points 638 * in struct pr_usrreqs, so that protocols can just reference then directly 639 * without the need for a wrapper function. The socket must have a valid 640 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one 641 * except through a kernel programming error, so it is acceptable to panic 642 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap 643 * because there actually /is/ a programming error somewhere... XXX) 644 */ 645 int 646 in_setsockaddr(struct socket *so, struct sockaddr **nam) 647 { 648 struct inpcb *inp; 649 struct sockaddr_in *sin; 650 651 /* 652 * Do the malloc first in case it blocks. 653 */ 654 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, 655 M_WAITOK | M_ZERO); 656 sin->sin_family = AF_INET; 657 sin->sin_len = sizeof *sin; 658 659 crit_enter(); 660 inp = so->so_pcb; 661 if (!inp) { 662 crit_exit(); 663 free(sin, M_SONAME); 664 return (ECONNRESET); 665 } 666 sin->sin_port = inp->inp_lport; 667 sin->sin_addr = inp->inp_laddr; 668 crit_exit(); 669 670 *nam = (struct sockaddr *)sin; 671 return (0); 672 } 673 674 int 675 in_setpeeraddr(struct socket *so, struct sockaddr **nam) 676 { 677 struct inpcb *inp; 678 struct sockaddr_in *sin; 679 680 /* 681 * Do the malloc first in case it blocks. 682 */ 683 MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME, 684 M_WAITOK | M_ZERO); 685 sin->sin_family = AF_INET; 686 sin->sin_len = sizeof *sin; 687 688 crit_enter(); 689 inp = so->so_pcb; 690 if (!inp) { 691 crit_exit(); 692 free(sin, M_SONAME); 693 return (ECONNRESET); 694 } 695 sin->sin_port = inp->inp_fport; 696 sin->sin_addr = inp->inp_faddr; 697 crit_exit(); 698 699 *nam = (struct sockaddr *)sin; 700 return (0); 701 } 702 703 void 704 in_pcbnotifyall(struct inpcbhead *head, struct in_addr faddr, int errno, 705 void (*notify)(struct inpcb *, int)) 706 { 707 struct inpcb *inp, *ninp; 708 709 /* 710 * note: if INP_PLACEMARKER is set we must ignore the rest of 711 * the structure and skip it. 712 */ 713 crit_enter(); 714 LIST_FOREACH_MUTABLE(inp, head, inp_list, ninp) { 715 if (inp->inp_flags & INP_PLACEMARKER) 716 continue; 717 #ifdef INET6 718 if (!(inp->inp_vflag & INP_IPV4)) 719 continue; 720 #endif 721 if (inp->inp_faddr.s_addr != faddr.s_addr || 722 inp->inp_socket == NULL) 723 continue; 724 (*notify)(inp, errno); /* can remove inp from list! */ 725 } 726 crit_exit(); 727 } 728 729 void 730 in_pcbpurgeif0(struct inpcb *head, struct ifnet *ifp) 731 { 732 struct inpcb *inp; 733 struct ip_moptions *imo; 734 int i, gap; 735 736 for (inp = head; inp != NULL; inp = LIST_NEXT(inp, inp_list)) { 737 if (inp->inp_flags & INP_PLACEMARKER) 738 continue; 739 imo = inp->inp_moptions; 740 if ((inp->inp_vflag & INP_IPV4) && imo != NULL) { 741 /* 742 * Unselect the outgoing interface if it is being 743 * detached. 744 */ 745 if (imo->imo_multicast_ifp == ifp) 746 imo->imo_multicast_ifp = NULL; 747 748 /* 749 * Drop multicast group membership if we joined 750 * through the interface being detached. 751 */ 752 for (i = 0, gap = 0; i < imo->imo_num_memberships; 753 i++) { 754 if (imo->imo_membership[i]->inm_ifp == ifp) { 755 in_delmulti(imo->imo_membership[i]); 756 gap++; 757 } else if (gap != 0) 758 imo->imo_membership[i - gap] = 759 imo->imo_membership[i]; 760 } 761 imo->imo_num_memberships -= gap; 762 } 763 } 764 } 765 766 /* 767 * Check for alternatives when higher level complains 768 * about service problems. For now, invalidate cached 769 * routing information. If the route was created dynamically 770 * (by a redirect), time to try a default gateway again. 771 */ 772 void 773 in_losing(struct inpcb *inp) 774 { 775 struct rtentry *rt; 776 struct rt_addrinfo rtinfo; 777 778 if ((rt = inp->inp_route.ro_rt)) { 779 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 780 rtinfo.rti_info[RTAX_DST] = rt_key(rt); 781 rtinfo.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 782 rtinfo.rti_info[RTAX_NETMASK] = rt_mask(rt); 783 rtinfo.rti_flags = rt->rt_flags; 784 rt_missmsg(RTM_LOSING, &rtinfo, rt->rt_flags, 0); 785 if (rt->rt_flags & RTF_DYNAMIC) 786 rtrequest1_global(RTM_DELETE, &rtinfo, NULL, NULL); 787 inp->inp_route.ro_rt = NULL; 788 rtfree(rt); 789 /* 790 * A new route can be allocated 791 * the next time output is attempted. 792 */ 793 } 794 } 795 796 /* 797 * After a routing change, flush old routing 798 * and allocate a (hopefully) better one. 799 */ 800 void 801 in_rtchange(struct inpcb *inp, int errno) 802 { 803 if (inp->inp_route.ro_rt) { 804 rtfree(inp->inp_route.ro_rt); 805 inp->inp_route.ro_rt = NULL; 806 /* 807 * A new route can be allocated the next time 808 * output is attempted. 809 */ 810 } 811 } 812 813 /* 814 * Lookup a PCB based on the local address and port. 815 */ 816 struct inpcb * 817 in_pcblookup_local(struct inpcbinfo *pcbinfo, struct in_addr laddr, u_int lport_arg, 818 int wild_okay) 819 { 820 struct inpcb *inp; 821 int matchwild = 3, wildcard; 822 u_short lport = lport_arg; 823 824 struct inpcbporthead *porthash; 825 struct inpcbport *phd; 826 struct inpcb *match = NULL; 827 828 /* 829 * Best fit PCB lookup. 830 * 831 * First see if this local port is in use by looking on the 832 * port hash list. 833 */ 834 porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport, 835 pcbinfo->porthashmask)]; 836 LIST_FOREACH(phd, porthash, phd_hash) { 837 if (phd->phd_port == lport) 838 break; 839 } 840 if (phd != NULL) { 841 /* 842 * Port is in use by one or more PCBs. Look for best 843 * fit. 844 */ 845 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 846 wildcard = 0; 847 #ifdef INET6 848 if ((inp->inp_vflag & INP_IPV4) == 0) 849 continue; 850 #endif 851 if (inp->inp_faddr.s_addr != INADDR_ANY) 852 wildcard++; 853 if (inp->inp_laddr.s_addr != INADDR_ANY) { 854 if (laddr.s_addr == INADDR_ANY) 855 wildcard++; 856 else if (inp->inp_laddr.s_addr != laddr.s_addr) 857 continue; 858 } else { 859 if (laddr.s_addr != INADDR_ANY) 860 wildcard++; 861 } 862 if (wildcard && !wild_okay) 863 continue; 864 if (wildcard < matchwild) { 865 match = inp; 866 matchwild = wildcard; 867 if (matchwild == 0) { 868 break; 869 } 870 } 871 } 872 } 873 return (match); 874 } 875 876 /* 877 * Lookup PCB in hash list. 878 */ 879 struct inpcb * 880 in_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in_addr faddr, u_int fport_arg, 881 struct in_addr laddr, u_int lport_arg, boolean_t wildcard, 882 struct ifnet *ifp) 883 { 884 struct inpcbhead *head; 885 struct inpcb *inp; 886 u_short fport = fport_arg, lport = lport_arg; 887 888 /* 889 * First look for an exact match. 890 */ 891 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr.s_addr, fport, 892 laddr.s_addr, lport, pcbinfo->hashmask)]; 893 LIST_FOREACH(inp, head, inp_hash) { 894 #ifdef INET6 895 if (!(inp->inp_vflag & INP_IPV4)) 896 continue; 897 #endif 898 if (in_hosteq(inp->inp_faddr, faddr) && 899 in_hosteq(inp->inp_laddr, laddr) && 900 inp->inp_fport == fport && inp->inp_lport == lport) { 901 /* found */ 902 return (inp); 903 } 904 } 905 906 if (wildcard) { 907 struct inpcb *local_wild = NULL; 908 #ifdef INET6 909 struct inpcb *local_wild_mapped = NULL; 910 #endif 911 struct inpcontainer *ic; 912 struct inpcontainerhead *chead; 913 914 chead = &pcbinfo->wildcardhashbase[ 915 INP_PCBWILDCARDHASH(lport, pcbinfo->wildcardhashmask)]; 916 LIST_FOREACH(ic, chead, ic_list) { 917 inp = ic->ic_inp; 918 #ifdef INET6 919 if (!(inp->inp_vflag & INP_IPV4)) 920 continue; 921 #endif 922 if (inp->inp_lport == lport) { 923 if (ifp && ifp->if_type == IFT_FAITH && 924 !(inp->inp_flags & INP_FAITH)) 925 continue; 926 if (inp->inp_laddr.s_addr == laddr.s_addr) 927 return (inp); 928 if (inp->inp_laddr.s_addr == INADDR_ANY) { 929 #ifdef INET6 930 if (INP_CHECK_SOCKAF(inp->inp_socket, 931 AF_INET6)) 932 local_wild_mapped = inp; 933 else 934 #endif 935 local_wild = inp; 936 } 937 } 938 } 939 #ifdef INET6 940 if (local_wild == NULL) 941 return (local_wild_mapped); 942 #endif 943 return (local_wild); 944 } 945 946 /* 947 * Not found. 948 */ 949 return (NULL); 950 } 951 952 /* 953 * Insert PCB into connection hash table. 954 */ 955 void 956 in_pcbinsconnhash(struct inpcb *inp) 957 { 958 struct inpcbinfo *pcbinfo = inp->inp_cpcbinfo; 959 struct inpcbhead *bucket; 960 u_int32_t hashkey_faddr, hashkey_laddr; 961 962 #ifdef INET6 963 if (inp->inp_vflag & INP_IPV6) { 964 hashkey_faddr = inp->in6p_faddr.s6_addr32[3] /* XXX JH */; 965 hashkey_laddr = inp->in6p_laddr.s6_addr32[3] /* XXX JH */; 966 } else { 967 #endif 968 hashkey_faddr = inp->inp_faddr.s_addr; 969 hashkey_laddr = inp->inp_laddr.s_addr; 970 #ifdef INET6 971 } 972 #endif 973 974 KASSERT(!(inp->inp_flags & INP_CONNECTED), ("already on hash list")); 975 inp->inp_flags |= INP_CONNECTED; 976 977 /* 978 * Insert into the connection hash table. 979 */ 980 bucket = &pcbinfo->hashbase[INP_PCBCONNHASH(hashkey_faddr, 981 inp->inp_fport, hashkey_laddr, inp->inp_lport, pcbinfo->hashmask)]; 982 LIST_INSERT_HEAD(bucket, inp, inp_hash); 983 } 984 985 /* 986 * Remove PCB from connection hash table. 987 */ 988 void 989 in_pcbremconnhash(struct inpcb *inp) 990 { 991 KASSERT(inp->inp_flags & INP_CONNECTED, ("inp not connected")); 992 LIST_REMOVE(inp, inp_hash); 993 inp->inp_flags &= ~INP_CONNECTED; 994 } 995 996 /* 997 * Insert PCB into port hash table. 998 */ 999 int 1000 in_pcbinsporthash(struct inpcb *inp) 1001 { 1002 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1003 struct inpcbporthead *pcbporthash; 1004 struct inpcbport *phd; 1005 1006 /* 1007 * Insert into the port hash table. 1008 */ 1009 pcbporthash = &pcbinfo->porthashbase[ 1010 INP_PCBPORTHASH(inp->inp_lport, pcbinfo->porthashmask)]; 1011 1012 /* Go through port list and look for a head for this lport. */ 1013 LIST_FOREACH(phd, pcbporthash, phd_hash) 1014 if (phd->phd_port == inp->inp_lport) 1015 break; 1016 1017 /* If none exists, malloc one and tack it on. */ 1018 if (phd == NULL) { 1019 MALLOC(phd, struct inpcbport *, sizeof(struct inpcbport), 1020 M_PCB, M_INTWAIT | M_NULLOK); 1021 if (phd == NULL) 1022 return (ENOBUFS); /* XXX */ 1023 phd->phd_port = inp->inp_lport; 1024 LIST_INIT(&phd->phd_pcblist); 1025 LIST_INSERT_HEAD(pcbporthash, phd, phd_hash); 1026 } 1027 1028 inp->inp_phd = phd; 1029 LIST_INSERT_HEAD(&phd->phd_pcblist, inp, inp_portlist); 1030 1031 return (0); 1032 } 1033 1034 void 1035 in_pcbinswildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo) 1036 { 1037 struct inpcontainer *ic; 1038 struct inpcontainerhead *bucket; 1039 1040 bucket = &pcbinfo->wildcardhashbase[ 1041 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)]; 1042 1043 ic = malloc(sizeof(struct inpcontainer), M_TEMP, M_INTWAIT); 1044 ic->ic_inp = inp; 1045 LIST_INSERT_HEAD(bucket, ic, ic_list); 1046 } 1047 1048 /* 1049 * Insert PCB into wildcard hash table. 1050 */ 1051 void 1052 in_pcbinswildcardhash(struct inpcb *inp) 1053 { 1054 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1055 1056 KKASSERT(pcbinfo != NULL); 1057 1058 in_pcbinswildcardhash_oncpu(inp, pcbinfo); 1059 inp->inp_flags |= INP_WILDCARD; 1060 } 1061 1062 void 1063 in_pcbremwildcardhash_oncpu(struct inpcb *inp, struct inpcbinfo *pcbinfo) 1064 { 1065 struct inpcontainer *ic; 1066 struct inpcontainerhead *head; 1067 1068 /* find bucket */ 1069 head = &pcbinfo->wildcardhashbase[ 1070 INP_PCBWILDCARDHASH(inp->inp_lport, pcbinfo->wildcardhashmask)]; 1071 1072 LIST_FOREACH(ic, head, ic_list) { 1073 if (ic->ic_inp == inp) 1074 goto found; 1075 } 1076 return; /* not found! */ 1077 1078 found: 1079 LIST_REMOVE(ic, ic_list); /* remove container from bucket chain */ 1080 free(ic, M_TEMP); /* deallocate container */ 1081 } 1082 1083 /* 1084 * Remove PCB from wildcard hash table. 1085 */ 1086 void 1087 in_pcbremwildcardhash(struct inpcb *inp) 1088 { 1089 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 1090 1091 KASSERT(inp->inp_flags & INP_WILDCARD, ("inp not wildcard")); 1092 in_pcbremwildcardhash_oncpu(inp, pcbinfo); 1093 inp->inp_flags &= ~INP_WILDCARD; 1094 } 1095 1096 /* 1097 * Remove PCB from various lists. 1098 */ 1099 void 1100 in_pcbremlists(struct inpcb *inp) 1101 { 1102 if (inp->inp_lport) { 1103 struct inpcbport *phd = inp->inp_phd; 1104 1105 LIST_REMOVE(inp, inp_portlist); 1106 if (LIST_FIRST(&phd->phd_pcblist) == NULL) { 1107 LIST_REMOVE(phd, phd_hash); 1108 free(phd, M_PCB); 1109 } 1110 } 1111 if (inp->inp_flags & INP_WILDCARD) { 1112 in_pcbremwildcardhash(inp); 1113 } else if (inp->inp_flags & INP_CONNECTED) { 1114 in_pcbremconnhash(inp); 1115 } 1116 LIST_REMOVE(inp, inp_list); 1117 inp->inp_pcbinfo->ipi_count--; 1118 } 1119 1120 int 1121 prison_xinpcb(struct thread *td, struct inpcb *inp) 1122 { 1123 struct ucred *cr; 1124 1125 if (td->td_proc == NULL) 1126 return (0); 1127 cr = td->td_proc->p_ucred; 1128 if (cr->cr_prison == NULL) 1129 return (0); 1130 if (ntohl(inp->inp_laddr.s_addr) == cr->cr_prison->pr_ip) 1131 return (0); 1132 return (1); 1133 } 1134 1135 int 1136 in_pcblist_global(SYSCTL_HANDLER_ARGS) 1137 { 1138 struct inpcbinfo *pcbinfo = arg1; 1139 struct inpcb *inp, *marker; 1140 struct xinpcb xi; 1141 int error, i, n; 1142 inp_gen_t gencnt; 1143 1144 /* 1145 * The process of preparing the TCB list is too time-consuming and 1146 * resource-intensive to repeat twice on every request. 1147 */ 1148 if (req->oldptr == NULL) { 1149 n = pcbinfo->ipi_count; 1150 req->oldidx = (n + n/8 + 10) * sizeof(struct xinpcb); 1151 return 0; 1152 } 1153 1154 if (req->newptr != NULL) 1155 return EPERM; 1156 1157 /* 1158 * OK, now we're committed to doing something. Re-fetch ipi_count 1159 * after obtaining the generation count. 1160 */ 1161 gencnt = pcbinfo->ipi_gencnt; 1162 n = pcbinfo->ipi_count; 1163 1164 marker = malloc(sizeof(struct inpcb), M_TEMP, M_WAITOK|M_ZERO); 1165 marker->inp_flags |= INP_PLACEMARKER; 1166 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list); 1167 1168 i = 0; 1169 error = 0; 1170 1171 while ((inp = LIST_NEXT(marker, inp_list)) != NULL && i < n) { 1172 LIST_REMOVE(marker, inp_list); 1173 LIST_INSERT_AFTER(inp, marker, inp_list); 1174 1175 if (inp->inp_flags & INP_PLACEMARKER) 1176 continue; 1177 if (inp->inp_gencnt > gencnt) 1178 continue; 1179 if (prison_xinpcb(req->td, inp)) 1180 continue; 1181 bzero(&xi, sizeof xi); 1182 xi.xi_len = sizeof xi; 1183 bcopy(inp, &xi.xi_inp, sizeof *inp); 1184 if (inp->inp_socket) 1185 sotoxsocket(inp->inp_socket, &xi.xi_socket); 1186 if ((error = SYSCTL_OUT(req, &xi, sizeof xi)) != 0) 1187 break; 1188 ++i; 1189 } 1190 LIST_REMOVE(marker, inp_list); 1191 if (error == 0 && i < n) { 1192 bzero(&xi, sizeof xi); 1193 xi.xi_len = sizeof xi; 1194 while (i < n) { 1195 error = SYSCTL_OUT(req, &xi, sizeof xi); 1196 ++i; 1197 } 1198 } 1199 free(marker, M_TEMP); 1200 return(error); 1201 } 1202