in6_pcb.c revision 1.56
1 /* $NetBSD: in6_pcb.c,v 1.56 2003/09/04 09:17:06 itojun Exp $ */ 2 /* $KAME: in6_pcb.c,v 1.84 2001/02/08 18:02:08 itojun Exp $ */ 3 4 /* 5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 6 * All rights reserved. 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 project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1982, 1986, 1991, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 62 */ 63 64 #include <sys/cdefs.h> 65 __KERNEL_RCSID(0, "$NetBSD: in6_pcb.c,v 1.56 2003/09/04 09:17:06 itojun Exp $"); 66 67 #include "opt_inet.h" 68 #include "opt_ipsec.h" 69 70 #include <sys/param.h> 71 #include <sys/systm.h> 72 #include <sys/malloc.h> 73 #include <sys/mbuf.h> 74 #include <sys/protosw.h> 75 #include <sys/socket.h> 76 #include <sys/socketvar.h> 77 #include <sys/ioctl.h> 78 #include <sys/errno.h> 79 #include <sys/time.h> 80 #include <sys/proc.h> 81 82 #include <net/if.h> 83 #include <net/route.h> 84 85 #include <netinet/in.h> 86 #include <netinet/in_var.h> 87 #include <netinet/in_systm.h> 88 #include <netinet/ip.h> 89 #include <netinet/in_pcb.h> 90 #include <netinet/ip6.h> 91 #include <netinet6/ip6_var.h> 92 #include <netinet6/in6_pcb.h> 93 #include <netinet6/nd6.h> 94 95 #include "loop.h" 96 extern struct ifnet loif[NLOOP]; 97 #include "faith.h" 98 99 #ifdef IPSEC 100 #include <netinet6/ipsec.h> 101 #include <netkey/key.h> 102 #endif /* IPSEC */ 103 104 struct in6_addr zeroin6_addr; 105 106 #define IN6PCBHASH_BIND(table, laddr, lport) \ 107 &(table)->inpt_bindhashtbl[ \ 108 (((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \ 109 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3]) + ntohs(lport)) & \ 110 (table)->inpt_bindhash] 111 #define IN6PCBHASH_CONNECT(table, faddr, fport, laddr, lport) \ 112 &(table)->inpt_bindhashtbl[ \ 113 ((((faddr)->s6_addr32[0] ^ (faddr)->s6_addr32[1] ^ \ 114 (faddr)->s6_addr32[2] ^ (faddr)->s6_addr32[3]) + ntohs(fport)) + \ 115 (((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \ 116 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3]) + \ 117 ntohs(lport))) & (table)->inpt_bindhash] 118 119 int ip6_anonportmin = IPV6PORT_ANONMIN; 120 int ip6_anonportmax = IPV6PORT_ANONMAX; 121 int ip6_lowportmin = IPV6PORT_RESERVEDMIN; 122 int ip6_lowportmax = IPV6PORT_RESERVEDMAX; 123 124 struct pool in6pcb_pool; 125 126 void 127 in6_pcbinit(table, bindhashsize, connecthashsize) 128 struct inpcbtable *table; 129 int bindhashsize, connecthashsize; 130 { 131 static int in6pcb_pool_initialized; 132 133 if (in6pcb_pool_initialized == 0) { 134 pool_init(&in6pcb_pool, sizeof(struct in6pcb), 0, 0, 0, 135 "in6pcbpl", NULL); 136 in6pcb_pool_initialized = 1; 137 } 138 139 in_pcbinit(table, bindhashsize, connecthashsize); 140 table->inpt_lastport = (u_int16_t)ip6_anonportmax; 141 } 142 143 int 144 in6_pcballoc(so, v) 145 struct socket *so; 146 void *v; 147 { 148 struct inpcbtable *table = v; 149 struct in6pcb *in6p; 150 int s; 151 #ifdef IPSEC 152 int error; 153 #endif 154 155 in6p = pool_get(&in6pcb_pool, PR_NOWAIT); 156 if (in6p == NULL) 157 return (ENOBUFS); 158 bzero((caddr_t)in6p, sizeof(*in6p)); 159 in6p->in6p_af = AF_INET6; 160 in6p->in6p_table = table; 161 in6p->in6p_socket = so; 162 in6p->in6p_hops = -1; /* use kernel default */ 163 in6p->in6p_icmp6filt = NULL; 164 #ifdef IPSEC 165 error = ipsec_init_pcbpolicy(so, &in6p->in6p_sp); 166 if (error != 0) { 167 pool_put(&in6pcb_pool, in6p); 168 return error; 169 } 170 #endif /* IPSEC */ 171 s = splnet(); 172 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, (struct inpcb_hdr*)in6p, 173 inph_queue); 174 in6_pcbstate(in6p, IN6P_ATTACHED); 175 splx(s); 176 if (ip6_v6only) 177 in6p->in6p_flags |= IN6P_IPV6_V6ONLY; 178 so->so_pcb = (caddr_t)in6p; 179 return (0); 180 } 181 182 int 183 in6_pcbbind(v, nam, p) 184 void *v; 185 struct mbuf *nam; 186 struct proc *p; 187 { 188 struct in6pcb *in6p = v; 189 struct socket *so = in6p->in6p_socket; 190 struct inpcbtable *table = in6p->in6p_table; 191 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 192 u_int16_t lport = 0; 193 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 194 195 if (in6p->in6p_af != AF_INET6) 196 return (EINVAL); 197 198 if (in6p->in6p_lport || !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 199 return (EINVAL); 200 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 && 201 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 || 202 (so->so_options & SO_ACCEPTCONN) == 0)) 203 wild = 1; 204 if (nam) { 205 sin6 = mtod(nam, struct sockaddr_in6 *); 206 if (nam->m_len != sizeof(*sin6)) 207 return (EINVAL); 208 /* 209 * We should check the family, but old programs 210 * incorrectly fail to intialize it. 211 */ 212 if (sin6->sin6_family != AF_INET6) 213 return (EAFNOSUPPORT); 214 215 #ifndef INET 216 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) 217 return (EADDRNOTAVAIL); 218 #endif 219 220 /* KAME hack: embed scopeid */ 221 if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, NULL) != 0) 222 return EINVAL; 223 /* this must be cleared for ifa_ifwithaddr() */ 224 sin6->sin6_scope_id = 0; 225 226 lport = sin6->sin6_port; 227 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 228 /* 229 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 230 * allow compepte duplication of binding if 231 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 232 * and a multicast address is bound on both 233 * new and duplicated sockets. 234 */ 235 if (so->so_options & SO_REUSEADDR) 236 reuseport = SO_REUSEADDR|SO_REUSEPORT; 237 } 238 else if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 239 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 240 return (EINVAL); 241 if (sin6->sin6_addr.s6_addr32[3]) { 242 struct sockaddr_in sin; 243 244 bzero(&sin, sizeof(sin)); 245 sin.sin_len = sizeof(sin); 246 sin.sin_family = AF_INET; 247 bcopy(&sin6->sin6_addr.s6_addr32[3], 248 &sin.sin_addr, sizeof(sin.sin_addr)); 249 if (ifa_ifwithaddr((struct sockaddr *)&sin) == 0) 250 return EADDRNOTAVAIL; 251 } 252 } 253 else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 254 struct ifaddr *ia = NULL; 255 256 sin6->sin6_port = 0; /* yech... */ 257 if ((in6p->in6p_flags & IN6P_FAITH) == 0 && 258 (ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0) 259 return (EADDRNOTAVAIL); 260 261 /* 262 * bind to an anycast address might accidentally 263 * cause sending a packet with an anycast source 264 * address, so we forbid it. 265 * 266 * We should allow to bind to a deprecated address, 267 * since the application dare to use it. 268 * But, can we assume that they are careful enough 269 * to check if the address is deprecated or not? 270 * Maybe, as a safeguard, we should have a setsockopt 271 * flag to control the bind(2) behavior against 272 * deprecated addresses (default: forbid bind(2)). 273 */ 274 if (ia && 275 ((struct in6_ifaddr *)ia)->ia6_flags & 276 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) 277 return (EADDRNOTAVAIL); 278 } 279 if (lport) { 280 #ifndef IPNOPRIVPORTS 281 int priv; 282 283 /* 284 * NOTE: all operating systems use suser() for 285 * privilege check! do not rewrite it into SS_PRIV. 286 */ 287 priv = (p && !suser(p->p_ucred, &p->p_acflag)) ? 1 : 0; 288 /* GROSS */ 289 if (ntohs(lport) < IPV6PORT_RESERVED && !priv) 290 return (EACCES); 291 #endif 292 293 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 294 #ifdef INET 295 struct inpcb *t; 296 297 t = in_pcblookup_port(table, 298 *(struct in_addr *)&sin6->sin6_addr.s6_addr32[3], 299 lport, wild); 300 if (t && (reuseport & t->inp_socket->so_options) == 0) 301 return (EADDRINUSE); 302 #else 303 return (EADDRNOTAVAIL); 304 #endif 305 } 306 307 { 308 struct in6pcb *t; 309 310 t = in6_pcblookup_port(table, &sin6->sin6_addr, 311 lport, wild); 312 if (t && (reuseport & t->in6p_socket->so_options) == 0) 313 return (EADDRINUSE); 314 } 315 } 316 in6p->in6p_laddr = sin6->sin6_addr; 317 } 318 319 if (lport == 0) { 320 int e; 321 e = in6_pcbsetport(&in6p->in6p_laddr, in6p, p); 322 if (e != 0) 323 return (e); 324 } else { 325 in6p->in6p_lport = lport; 326 in6_pcbstate(in6p, IN6P_BOUND); 327 } 328 329 in6p->in6p_flowinfo = sin6 ? sin6->sin6_flowinfo : 0; /*XXX*/ 330 return (0); 331 } 332 333 /* 334 * Connect from a socket to a specified address. 335 * Both address and port must be specified in argument sin6. 336 * If don't have a local address for this socket yet, 337 * then pick one. 338 */ 339 int 340 in6_pcbconnect(v, nam) 341 void *v; 342 struct mbuf *nam; 343 { 344 struct in6pcb *in6p = v; 345 struct in6_addr *in6a = NULL; 346 struct sockaddr_in6 *sin6 = mtod(nam, struct sockaddr_in6 *); 347 struct ifnet *ifp = NULL; /* outgoing interface */ 348 int error = 0; 349 #ifdef INET 350 struct in6_addr mapped; 351 #endif 352 struct sockaddr_in6 tmp; 353 354 (void)&in6a; /* XXX fool gcc */ 355 356 if (in6p->in6p_af != AF_INET6) 357 return (EINVAL); 358 359 if (nam->m_len != sizeof(*sin6)) 360 return (EINVAL); 361 if (sin6->sin6_family != AF_INET6) 362 return (EAFNOSUPPORT); 363 if (sin6->sin6_port == 0) 364 return (EADDRNOTAVAIL); 365 366 /* sanity check for mapped address case */ 367 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 368 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 369 return EINVAL; 370 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 371 in6p->in6p_laddr.s6_addr16[5] = htons(0xffff); 372 if (!IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 373 return EINVAL; 374 } else 375 { 376 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 377 return EINVAL; 378 } 379 380 /* protect *sin6 from overwrites */ 381 tmp = *sin6; 382 sin6 = &tmp; 383 384 /* KAME hack: embed scopeid */ 385 if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, &ifp) != 0) 386 return EINVAL; 387 388 /* Source address selection. */ 389 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) && 390 in6p->in6p_laddr.s6_addr32[3] == 0) { 391 #ifdef INET 392 struct sockaddr_in sin, *sinp; 393 394 bzero(&sin, sizeof(sin)); 395 sin.sin_len = sizeof(sin); 396 sin.sin_family = AF_INET; 397 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr, 398 sizeof(sin.sin_addr)); 399 sinp = in_selectsrc(&sin, (struct route *)&in6p->in6p_route, 400 in6p->in6p_socket->so_options, NULL, &error); 401 if (sinp == 0) { 402 if (error == 0) 403 error = EADDRNOTAVAIL; 404 return (error); 405 } 406 bzero(&mapped, sizeof(mapped)); 407 mapped.s6_addr16[5] = htons(0xffff); 408 bcopy(&sinp->sin_addr, &mapped.s6_addr32[3], sizeof(sinp->sin_addr)); 409 in6a = &mapped; 410 #else 411 return EADDRNOTAVAIL; 412 #endif 413 } else 414 { 415 /* 416 * XXX: in6_selectsrc might replace the bound local address 417 * with the address specified by setsockopt(IPV6_PKTINFO). 418 * Is it the intended behavior? 419 */ 420 in6a = in6_selectsrc(sin6, in6p->in6p_outputopts, 421 in6p->in6p_moptions, 422 &in6p->in6p_route, 423 &in6p->in6p_laddr, &error); 424 if (in6a == 0) { 425 if (error == 0) 426 error = EADDRNOTAVAIL; 427 return (error); 428 } 429 } 430 if (in6p->in6p_route.ro_rt) 431 ifp = in6p->in6p_route.ro_rt->rt_ifp; 432 433 in6p->in6p_ip6.ip6_hlim = (u_int8_t)in6_selecthlim(in6p, ifp); 434 435 if (in6_pcblookup_connect(in6p->in6p_table, &sin6->sin6_addr, 436 sin6->sin6_port, 437 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ? in6a : &in6p->in6p_laddr, 438 in6p->in6p_lport, 0)) 439 return (EADDRINUSE); 440 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) || 441 (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) && 442 in6p->in6p_laddr.s6_addr32[3] == 0)) 443 { 444 if (in6p->in6p_lport == 0) { 445 (void)in6_pcbbind(in6p, (struct mbuf *)0, 446 (struct proc *)0); 447 } 448 in6p->in6p_laddr = *in6a; 449 } 450 in6p->in6p_faddr = sin6->sin6_addr; 451 in6p->in6p_fport = sin6->sin6_port; 452 in6_pcbstate(in6p, IN6P_CONNECTED); 453 /* 454 * xxx kazu flowlabel is necessary for connect? 455 * but if this line is missing, the garbage value remains. 456 */ 457 in6p->in6p_flowinfo = sin6->sin6_flowinfo; 458 #ifdef IPSEC 459 if (in6p->in6p_socket->so_type == SOCK_STREAM) 460 ipsec_pcbconn(in6p->in6p_sp); 461 #endif 462 return (0); 463 } 464 465 void 466 in6_pcbdisconnect(in6p) 467 struct in6pcb *in6p; 468 { 469 bzero((caddr_t)&in6p->in6p_faddr, sizeof(in6p->in6p_faddr)); 470 in6p->in6p_fport = 0; 471 in6_pcbstate(in6p, IN6P_BOUND); 472 if (in6p->in6p_socket->so_state & SS_NOFDREF) 473 in6_pcbdetach(in6p); 474 #ifdef IPSEC 475 ipsec_pcbdisconn(in6p->in6p_sp); 476 #endif 477 } 478 479 void 480 in6_pcbdetach(in6p) 481 struct in6pcb *in6p; 482 { 483 struct socket *so = in6p->in6p_socket; 484 int s; 485 486 if (in6p->in6p_af != AF_INET6) 487 return; 488 489 #ifdef IPSEC 490 ipsec6_delete_pcbpolicy(in6p); 491 #endif /* IPSEC */ 492 sotoin6pcb(so) = 0; 493 sofree(so); 494 if (in6p->in6p_options) 495 m_freem(in6p->in6p_options); 496 if (in6p->in6p_outputopts) { 497 if (in6p->in6p_outputopts->ip6po_rthdr && 498 in6p->in6p_outputopts->ip6po_route.ro_rt) 499 RTFREE(in6p->in6p_outputopts->ip6po_route.ro_rt); 500 if (in6p->in6p_outputopts->ip6po_m) 501 (void)m_free(in6p->in6p_outputopts->ip6po_m); 502 free(in6p->in6p_outputopts, M_IP6OPT); 503 } 504 if (in6p->in6p_route.ro_rt) 505 rtfree(in6p->in6p_route.ro_rt); 506 ip6_freemoptions(in6p->in6p_moptions); 507 s = splnet(); 508 in6_pcbstate(in6p, IN6P_ATTACHED); 509 CIRCLEQ_REMOVE(&in6p->in6p_table->inpt_queue, &in6p->in6p_head, 510 inph_queue); 511 splx(s); 512 pool_put(&in6pcb_pool, in6p); 513 } 514 515 void 516 in6_setsockaddr(in6p, nam) 517 struct in6pcb *in6p; 518 struct mbuf *nam; 519 { 520 struct sockaddr_in6 *sin6; 521 522 if (in6p->in6p_af != AF_INET6) 523 return; 524 525 nam->m_len = sizeof(*sin6); 526 sin6 = mtod(nam, struct sockaddr_in6 *); 527 bzero((caddr_t)sin6, sizeof(*sin6)); 528 sin6->sin6_family = AF_INET6; 529 sin6->sin6_len = sizeof(struct sockaddr_in6); 530 sin6->sin6_port = in6p->in6p_lport; 531 /* KAME hack: recover scopeid */ 532 (void)in6_recoverscope(sin6, &in6p->in6p_laddr, NULL); 533 } 534 535 void 536 in6_setpeeraddr(in6p, nam) 537 struct in6pcb *in6p; 538 struct mbuf *nam; 539 { 540 struct sockaddr_in6 *sin6; 541 542 if (in6p->in6p_af != AF_INET6) 543 return; 544 545 nam->m_len = sizeof(*sin6); 546 sin6 = mtod(nam, struct sockaddr_in6 *); 547 bzero((caddr_t)sin6, sizeof(*sin6)); 548 sin6->sin6_family = AF_INET6; 549 sin6->sin6_len = sizeof(struct sockaddr_in6); 550 sin6->sin6_port = in6p->in6p_fport; 551 /* KAME hack: recover scopeid */ 552 (void)in6_recoverscope(sin6, &in6p->in6p_faddr, NULL); 553 } 554 555 /* 556 * Pass some notification to all connections of a protocol 557 * associated with address dst. The local address and/or port numbers 558 * may be specified to limit the search. The "usual action" will be 559 * taken, depending on the ctlinput cmd. The caller must filter any 560 * cmds that are uninteresting (e.g., no error in the map). 561 * Call the protocol specific routine (if any) to report 562 * any errors for each matching socket. 563 * 564 * Must be called at splsoftnet. 565 * 566 * Note: src (4th arg) carries the flowlabel value on the original IPv6 567 * header, in sin6_flowinfo member. 568 */ 569 int 570 in6_pcbnotify(table, dst, fport_arg, src, lport_arg, cmd, cmdarg, notify) 571 struct inpcbtable *table; 572 struct sockaddr *dst, *src; 573 u_int fport_arg, lport_arg; 574 int cmd; 575 void *cmdarg; 576 void (*notify) __P((struct in6pcb *, int)); 577 { 578 struct in6pcb *in6p, *nin6p; 579 struct sockaddr_in6 sa6_src, *sa6_dst; 580 u_int16_t fport = fport_arg, lport = lport_arg; 581 int errno; 582 int nmatch = 0; 583 u_int32_t flowinfo; 584 585 if ((unsigned)cmd > PRC_NCMDS || dst->sa_family != AF_INET6) 586 return 0; 587 588 sa6_dst = (struct sockaddr_in6 *)dst; 589 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 590 return 0; 591 592 /* 593 * note that src can be NULL when we get notify by local fragmentation. 594 */ 595 sa6_src = (src == NULL) ? sa6_any : *(struct sockaddr_in6 *)src; 596 flowinfo = sa6_src.sin6_flowinfo; 597 598 /* 599 * Redirects go to all references to the destination, 600 * and use in6_rtchange to invalidate the route cache. 601 * Dead host indications: also use in6_rtchange to invalidate 602 * the cache, and deliver the error to all the sockets. 603 * Otherwise, if we have knowledge of the local port and address, 604 * deliver only to that socket. 605 */ 606 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 607 fport = 0; 608 lport = 0; 609 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 610 611 if (cmd != PRC_HOSTDEAD) 612 notify = in6_rtchange; 613 } 614 615 errno = inet6ctlerrmap[cmd]; 616 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue); 617 in6p != (void *)&table->inpt_queue; 618 in6p = nin6p) { 619 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue); 620 621 if (in6p->in6p_af != AF_INET6) 622 continue; 623 624 /* 625 * Under the following condition, notify of redirects 626 * to the pcb, without making address matches against inpcb. 627 * - redirect notification is arrived. 628 * - the inpcb is unconnected. 629 * - the inpcb is caching !RTF_HOST routing entry. 630 * - the ICMPv6 notification is from the gateway cached in the 631 * inpcb. i.e. ICMPv6 notification is from nexthop gateway 632 * the inpcb used very recently. 633 * 634 * This is to improve interaction between netbsd/openbsd 635 * redirect handling code, and inpcb route cache code. 636 * without the clause, !RTF_HOST routing entry (which carries 637 * gateway used by inpcb right before the ICMPv6 redirect) 638 * will be cached forever in unconnected inpcb. 639 * 640 * There still is a question regarding to what is TRT: 641 * - On bsdi/freebsd, RTF_HOST (cloned) routing entry will be 642 * generated on packet output. inpcb will always cache 643 * RTF_HOST routing entry so there's no need for the clause 644 * (ICMPv6 redirect will update RTF_HOST routing entry, 645 * and inpcb is caching it already). 646 * However, bsdi/freebsd are vulnerable to local DoS attacks 647 * due to the cloned routing entries. 648 * - Specwise, "destination cache" is mentioned in RFC2461. 649 * Jinmei says that it implies bsdi/freebsd behavior, itojun 650 * is not really convinced. 651 * - Having hiwat/lowat on # of cloned host route (redirect/ 652 * pmtud) may be a good idea. netbsd/openbsd has it. see 653 * icmp6_mtudisc_update(). 654 */ 655 if ((PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) && 656 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) && 657 in6p->in6p_route.ro_rt && 658 !(in6p->in6p_route.ro_rt->rt_flags & RTF_HOST)) { 659 struct sockaddr_in6 *dst6; 660 661 dst6 = (struct sockaddr_in6 *)&in6p->in6p_route.ro_dst; 662 if (IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr, 663 &sa6_dst->sin6_addr)) 664 goto do_notify; 665 } 666 667 /* 668 * Detect if we should notify the error. If no source and 669 * destination ports are specified, but non-zero flowinfo and 670 * local address match, notify the error. This is the case 671 * when the error is delivered with an encrypted buffer 672 * by ESP. Otherwise, just compare addresses and ports 673 * as usual. 674 */ 675 if (lport == 0 && fport == 0 && flowinfo && 676 in6p->in6p_socket != NULL && 677 flowinfo == (in6p->in6p_flowinfo & IPV6_FLOWLABEL_MASK) && 678 IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &sa6_src.sin6_addr)) 679 goto do_notify; 680 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, 681 &sa6_dst->sin6_addr) || 682 in6p->in6p_socket == 0 || 683 (lport && in6p->in6p_lport != lport) || 684 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 685 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, 686 &sa6_src.sin6_addr)) || 687 (fport && in6p->in6p_fport != fport)) 688 continue; 689 690 do_notify: 691 if (notify) 692 (*notify)(in6p, errno); 693 nmatch++; 694 } 695 return nmatch; 696 } 697 698 void 699 in6_pcbpurgeif0(table, ifp) 700 struct inpcbtable *table; 701 struct ifnet *ifp; 702 { 703 struct in6pcb *in6p, *nin6p; 704 struct ip6_moptions *im6o; 705 struct in6_multi_mship *imm, *nimm; 706 707 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue); 708 in6p != (void *)&table->inpt_queue; 709 in6p = nin6p) { 710 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue); 711 if (in6p->in6p_af != AF_INET6) 712 continue; 713 714 im6o = in6p->in6p_moptions; 715 if (im6o) { 716 /* 717 * Unselect the outgoing interface if it is being 718 * detached. 719 */ 720 if (im6o->im6o_multicast_ifp == ifp) 721 im6o->im6o_multicast_ifp = NULL; 722 723 /* 724 * Drop multicast group membership if we joined 725 * through the interface being detached. 726 * XXX controversial - is it really legal for kernel 727 * to force this? 728 */ 729 for (imm = im6o->im6o_memberships.lh_first; 730 imm != NULL; imm = nimm) { 731 nimm = imm->i6mm_chain.le_next; 732 if (imm->i6mm_maddr->in6m_ifp == ifp) { 733 LIST_REMOVE(imm, i6mm_chain); 734 in6_leavegroup(imm); 735 } 736 } 737 } 738 } 739 } 740 741 void 742 in6_pcbpurgeif(table, ifp) 743 struct inpcbtable *table; 744 struct ifnet *ifp; 745 { 746 struct in6pcb *in6p, *nin6p; 747 748 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue); 749 in6p != (void *)&table->inpt_queue; 750 in6p = nin6p) { 751 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue); 752 if (in6p->in6p_af != AF_INET6) 753 continue; 754 if (in6p->in6p_route.ro_rt != NULL && 755 in6p->in6p_route.ro_rt->rt_ifp == ifp) 756 in6_rtchange(in6p, 0); 757 } 758 } 759 760 /* 761 * Check for alternatives when higher level complains 762 * about service problems. For now, invalidate cached 763 * routing information. If the route was created dynamically 764 * (by a redirect), time to try a default gateway again. 765 */ 766 void 767 in6_losing(in6p) 768 struct in6pcb *in6p; 769 { 770 struct rtentry *rt; 771 struct rt_addrinfo info; 772 773 if (in6p->in6p_af != AF_INET6) 774 return; 775 776 if ((rt = in6p->in6p_route.ro_rt) != NULL) { 777 in6p->in6p_route.ro_rt = 0; 778 bzero((caddr_t)&info, sizeof(info)); 779 info.rti_info[RTAX_DST] = 780 (struct sockaddr *)&in6p->in6p_route.ro_dst; 781 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 782 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 783 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 784 if (rt->rt_flags & RTF_DYNAMIC) { 785 (void)rtrequest(RTM_DELETE, rt_key(rt), 786 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 787 (struct rtentry **)0); 788 } else { 789 /* 790 * A new route can be allocated 791 * the next time output is attempted. 792 */ 793 rtfree(rt); 794 } 795 } 796 } 797 798 /* 799 * After a routing change, flush old routing 800 * and allocate a (hopefully) better one. 801 */ 802 void 803 in6_rtchange(in6p, errno) 804 struct in6pcb *in6p; 805 int errno; 806 { 807 if (in6p->in6p_af != AF_INET6) 808 return; 809 810 if (in6p->in6p_route.ro_rt) { 811 rtfree(in6p->in6p_route.ro_rt); 812 in6p->in6p_route.ro_rt = 0; 813 /* 814 * A new route can be allocated the next time 815 * output is attempted. 816 */ 817 } 818 } 819 820 struct in6pcb * 821 in6_pcblookup_port(table, laddr6, lport_arg, lookup_wildcard) 822 struct inpcbtable *table; 823 struct in6_addr *laddr6; 824 u_int lport_arg; 825 int lookup_wildcard; 826 { 827 struct inpcb_hdr *inph; 828 struct in6pcb *in6p, *match = 0; 829 int matchwild = 3, wildcard; 830 u_int16_t lport = lport_arg; 831 832 CIRCLEQ_FOREACH(inph, &table->inpt_queue, inph_queue) { 833 in6p = (struct in6pcb *)inph; 834 if (in6p->in6p_af != AF_INET6) 835 continue; 836 837 if (in6p->in6p_lport != lport) 838 continue; 839 wildcard = 0; 840 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) { 841 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 842 continue; 843 } 844 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) 845 wildcard++; 846 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) { 847 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 848 continue; 849 if (!IN6_IS_ADDR_V4MAPPED(laddr6)) 850 continue; 851 852 /* duplicate of IPv4 logic */ 853 wildcard = 0; 854 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr) && 855 in6p->in6p_faddr.s6_addr32[3]) 856 wildcard++; 857 if (!in6p->in6p_laddr.s6_addr32[3]) { 858 if (laddr6->s6_addr32[3]) 859 wildcard++; 860 } else { 861 if (!laddr6->s6_addr32[3]) 862 wildcard++; 863 else { 864 if (in6p->in6p_laddr.s6_addr32[3] != 865 laddr6->s6_addr32[3]) 866 continue; 867 } 868 } 869 } else if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { 870 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 871 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 872 continue; 873 } 874 if (!IN6_IS_ADDR_UNSPECIFIED(laddr6)) 875 wildcard++; 876 } else { 877 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 878 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 879 continue; 880 } 881 if (IN6_IS_ADDR_UNSPECIFIED(laddr6)) 882 wildcard++; 883 else { 884 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, 885 laddr6)) 886 continue; 887 } 888 } 889 if (wildcard && !lookup_wildcard) 890 continue; 891 if (wildcard < matchwild) { 892 match = in6p; 893 matchwild = wildcard; 894 if (matchwild == 0) 895 break; 896 } 897 } 898 return (match); 899 } 900 #undef continue 901 902 /* 903 * WARNING: return value (rtentry) could be IPv4 one if in6pcb is connected to 904 * IPv4 mapped address. 905 */ 906 struct rtentry * 907 in6_pcbrtentry(in6p) 908 struct in6pcb *in6p; 909 { 910 struct route_in6 *ro; 911 struct sockaddr_in6 *dst6; 912 913 ro = &in6p->in6p_route; 914 dst6 = (struct sockaddr_in6 *)&ro->ro_dst; 915 916 if (in6p->in6p_af != AF_INET6) 917 return (NULL); 918 919 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 920 !IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr, &in6p->in6p_faddr))) { 921 RTFREE(ro->ro_rt); 922 ro->ro_rt = (struct rtentry *)NULL; 923 } 924 #ifdef INET 925 if (ro->ro_rt == (struct rtentry *)NULL && 926 IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) { 927 struct sockaddr_in *dst = (struct sockaddr_in *)&ro->ro_dst; 928 929 bzero(dst, sizeof(*dst)); 930 dst->sin_family = AF_INET; 931 dst->sin_len = sizeof(struct sockaddr_in); 932 bcopy(&in6p->in6p_faddr.s6_addr32[3], &dst->sin_addr, 933 sizeof(dst->sin_addr)); 934 rtalloc((struct route *)ro); 935 } else 936 #endif 937 if (ro->ro_rt == (struct rtentry *)NULL && 938 !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) { 939 bzero(dst6, sizeof(*dst6)); 940 dst6->sin6_family = AF_INET6; 941 dst6->sin6_len = sizeof(struct sockaddr_in6); 942 dst6->sin6_addr = in6p->in6p_faddr; 943 rtalloc((struct route *)ro); 944 } 945 return (ro->ro_rt); 946 } 947 948 struct in6pcb * 949 in6_pcblookup_connect(table, faddr6, fport_arg, laddr6, lport_arg, faith) 950 struct inpcbtable *table; 951 struct in6_addr *faddr6, *laddr6; 952 u_int fport_arg, lport_arg; 953 int faith; 954 { 955 struct inpcbhead *head; 956 struct inpcb_hdr *inph; 957 struct in6pcb *in6p; 958 u_int16_t fport = fport_arg, lport = lport_arg; 959 960 head = IN6PCBHASH_CONNECT(table, faddr6, fport, laddr6, lport); 961 LIST_FOREACH(inph, head, inph_hash) { 962 in6p = (struct in6pcb *)inph; 963 if (in6p->in6p_af != AF_INET6) 964 continue; 965 966 /* find exact match on both source and dest */ 967 if (in6p->in6p_fport != fport) 968 continue; 969 if (in6p->in6p_lport != lport) 970 continue; 971 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) 972 continue; 973 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6)) 974 continue; 975 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 976 continue; 977 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 978 continue; 979 if ((IN6_IS_ADDR_V4MAPPED(laddr6) || 980 IN6_IS_ADDR_V4MAPPED(faddr6)) && 981 (in6p->in6p_flags & IN6P_IPV6_V6ONLY)) 982 continue; 983 return in6p; 984 } 985 return NULL; 986 } 987 988 struct in6pcb * 989 in6_pcblookup_bind(table, laddr6, lport_arg, faith) 990 struct inpcbtable *table; 991 struct in6_addr *laddr6; 992 u_int lport_arg; 993 int faith; 994 { 995 struct inpcbhead *head; 996 struct inpcb_hdr *inph; 997 struct in6pcb *in6p; 998 u_int16_t lport = lport_arg; 999 #ifdef INET6 1000 struct in6_addr zero_mapped; 1001 #endif 1002 1003 head = IN6PCBHASH_BIND(table, laddr6, lport); 1004 LIST_FOREACH(inph, head, inph_hash) { 1005 in6p = (struct in6pcb *)inph; 1006 if (in6p->in6p_af != AF_INET6) 1007 continue; 1008 1009 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0) 1010 continue; 1011 if (in6p->in6p_fport != 0) 1012 continue; 1013 if (in6p->in6p_lport != lport) 1014 continue; 1015 if (IN6_IS_ADDR_V4MAPPED(laddr6) && 1016 (in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 1017 continue; 1018 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 1019 goto out; 1020 } 1021 #ifdef INET 1022 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 1023 memset(&zero_mapped, 0, sizeof(zero_mapped)); 1024 zero_mapped.s6_addr16[5] = 0xffff; 1025 head = IN6PCBHASH_BIND(table, &zero_mapped, lport); 1026 LIST_FOREACH(inph, head, inph_hash) { 1027 in6p = (struct in6pcb *)inph; 1028 if (in6p->in6p_af != AF_INET6) 1029 continue; 1030 1031 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0) 1032 continue; 1033 if (in6p->in6p_fport != 0) 1034 continue; 1035 if (in6p->in6p_lport != lport) 1036 continue; 1037 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 1038 continue; 1039 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &zero_mapped)) 1040 goto out; 1041 } 1042 } 1043 #endif 1044 head = IN6PCBHASH_BIND(table, &zeroin6_addr, lport); 1045 LIST_FOREACH(inph, head, inph_hash) { 1046 in6p = (struct in6pcb *)inph; 1047 if (in6p->in6p_af != AF_INET6) 1048 continue; 1049 1050 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0) 1051 continue; 1052 if (in6p->in6p_fport != 0) 1053 continue; 1054 if (in6p->in6p_lport != lport) 1055 continue; 1056 if (IN6_IS_ADDR_V4MAPPED(laddr6) && 1057 (in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 1058 continue; 1059 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &zeroin6_addr)) 1060 goto out; 1061 } 1062 return (NULL); 1063 1064 out: 1065 inph = &in6p->in6p_head; 1066 if (inph != LIST_FIRST(head)) { 1067 LIST_REMOVE(inph, inph_hash); 1068 LIST_INSERT_HEAD(head, inph, inph_hash); 1069 } 1070 return in6p; 1071 } 1072 1073 void 1074 in6_pcbstate(in6p, state) 1075 struct in6pcb *in6p; 1076 int state; 1077 { 1078 1079 if (in6p->in6p_af != AF_INET6) 1080 return; 1081 1082 if (in6p->in6p_state > IN6P_ATTACHED) 1083 LIST_REMOVE(&in6p->in6p_head, inph_hash); 1084 1085 switch (state) { 1086 case IN6P_BOUND: 1087 LIST_INSERT_HEAD(IN6PCBHASH_BIND(in6p->in6p_table, 1088 &in6p->in6p_laddr, in6p->in6p_lport), &in6p->in6p_head, 1089 inph_hash); 1090 break; 1091 case IN6P_CONNECTED: 1092 LIST_INSERT_HEAD(IN6PCBHASH_CONNECT(in6p->in6p_table, 1093 &in6p->in6p_faddr, in6p->in6p_fport, 1094 &in6p->in6p_laddr, in6p->in6p_lport), &in6p->in6p_head, 1095 inph_hash); 1096 break; 1097 } 1098 1099 in6p->in6p_state = state; 1100 } 1101
Indexes created Tue Sep 23 14:10:03 GMT 2025