in6_pcb.c revision 1.57
1 /* $NetBSD: in6_pcb.c,v 1.57 2003/09/06 03:12:53 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 1.57 2003/09/06 03:12:53 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 #if 0 330 in6p->in6p_flowinfo = 0; /* XXX */ 331 #endif 332 return (0); 333 } 334 335 /* 336 * Connect from a socket to a specified address. 337 * Both address and port must be specified in argument sin6. 338 * If don't have a local address for this socket yet, 339 * then pick one. 340 */ 341 int 342 in6_pcbconnect(v, nam) 343 void *v; 344 struct mbuf *nam; 345 { 346 struct in6pcb *in6p = v; 347 struct in6_addr *in6a = NULL; 348 struct sockaddr_in6 *sin6 = mtod(nam, struct sockaddr_in6 *); 349 struct ifnet *ifp = NULL; /* outgoing interface */ 350 int error = 0; 351 #ifdef INET 352 struct in6_addr mapped; 353 #endif 354 struct sockaddr_in6 tmp; 355 356 (void)&in6a; /* XXX fool gcc */ 357 358 if (in6p->in6p_af != AF_INET6) 359 return (EINVAL); 360 361 if (nam->m_len != sizeof(*sin6)) 362 return (EINVAL); 363 if (sin6->sin6_family != AF_INET6) 364 return (EAFNOSUPPORT); 365 if (sin6->sin6_port == 0) 366 return (EADDRNOTAVAIL); 367 368 /* sanity check for mapped address case */ 369 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 370 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 371 return EINVAL; 372 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 373 in6p->in6p_laddr.s6_addr16[5] = htons(0xffff); 374 if (!IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 375 return EINVAL; 376 } else 377 { 378 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 379 return EINVAL; 380 } 381 382 /* protect *sin6 from overwrites */ 383 tmp = *sin6; 384 sin6 = &tmp; 385 386 /* KAME hack: embed scopeid */ 387 if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, &ifp) != 0) 388 return EINVAL; 389 390 /* Source address selection. */ 391 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) && 392 in6p->in6p_laddr.s6_addr32[3] == 0) { 393 #ifdef INET 394 struct sockaddr_in sin, *sinp; 395 396 bzero(&sin, sizeof(sin)); 397 sin.sin_len = sizeof(sin); 398 sin.sin_family = AF_INET; 399 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr, 400 sizeof(sin.sin_addr)); 401 sinp = in_selectsrc(&sin, (struct route *)&in6p->in6p_route, 402 in6p->in6p_socket->so_options, NULL, &error); 403 if (sinp == 0) { 404 if (error == 0) 405 error = EADDRNOTAVAIL; 406 return (error); 407 } 408 bzero(&mapped, sizeof(mapped)); 409 mapped.s6_addr16[5] = htons(0xffff); 410 bcopy(&sinp->sin_addr, &mapped.s6_addr32[3], sizeof(sinp->sin_addr)); 411 in6a = &mapped; 412 #else 413 return EADDRNOTAVAIL; 414 #endif 415 } else 416 { 417 /* 418 * XXX: in6_selectsrc might replace the bound local address 419 * with the address specified by setsockopt(IPV6_PKTINFO). 420 * Is it the intended behavior? 421 */ 422 in6a = in6_selectsrc(sin6, in6p->in6p_outputopts, 423 in6p->in6p_moptions, 424 &in6p->in6p_route, 425 &in6p->in6p_laddr, &error); 426 if (in6a == 0) { 427 if (error == 0) 428 error = EADDRNOTAVAIL; 429 return (error); 430 } 431 } 432 if (in6p->in6p_route.ro_rt) 433 ifp = in6p->in6p_route.ro_rt->rt_ifp; 434 435 in6p->in6p_ip6.ip6_hlim = (u_int8_t)in6_selecthlim(in6p, ifp); 436 437 if (in6_pcblookup_connect(in6p->in6p_table, &sin6->sin6_addr, 438 sin6->sin6_port, 439 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ? in6a : &in6p->in6p_laddr, 440 in6p->in6p_lport, 0)) 441 return (EADDRINUSE); 442 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) || 443 (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) && 444 in6p->in6p_laddr.s6_addr32[3] == 0)) 445 { 446 if (in6p->in6p_lport == 0) { 447 (void)in6_pcbbind(in6p, (struct mbuf *)0, 448 (struct proc *)0); 449 } 450 in6p->in6p_laddr = *in6a; 451 } 452 in6p->in6p_faddr = sin6->sin6_addr; 453 in6p->in6p_fport = sin6->sin6_port; 454 in6_pcbstate(in6p, IN6P_CONNECTED); 455 in6p->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK; 456 if (ip6_auto_flowlabel) 457 in6p->in6p_flowinfo |= 458 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 459 #ifdef IPSEC 460 if (in6p->in6p_socket->so_type == SOCK_STREAM) 461 ipsec_pcbconn(in6p->in6p_sp); 462 #endif 463 return (0); 464 } 465 466 void 467 in6_pcbdisconnect(in6p) 468 struct in6pcb *in6p; 469 { 470 bzero((caddr_t)&in6p->in6p_faddr, sizeof(in6p->in6p_faddr)); 471 in6p->in6p_fport = 0; 472 in6_pcbstate(in6p, IN6P_BOUND); 473 in6p->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK; 474 if (in6p->in6p_socket->so_state & SS_NOFDREF) 475 in6_pcbdetach(in6p); 476 #ifdef IPSEC 477 ipsec_pcbdisconn(in6p->in6p_sp); 478 #endif 479 } 480 481 void 482 in6_pcbdetach(in6p) 483 struct in6pcb *in6p; 484 { 485 struct socket *so = in6p->in6p_socket; 486 int s; 487 488 if (in6p->in6p_af != AF_INET6) 489 return; 490 491 #ifdef IPSEC 492 ipsec6_delete_pcbpolicy(in6p); 493 #endif /* IPSEC */ 494 sotoin6pcb(so) = 0; 495 sofree(so); 496 if (in6p->in6p_options) 497 m_freem(in6p->in6p_options); 498 if (in6p->in6p_outputopts) { 499 if (in6p->in6p_outputopts->ip6po_rthdr && 500 in6p->in6p_outputopts->ip6po_route.ro_rt) 501 RTFREE(in6p->in6p_outputopts->ip6po_route.ro_rt); 502 if (in6p->in6p_outputopts->ip6po_m) 503 (void)m_free(in6p->in6p_outputopts->ip6po_m); 504 free(in6p->in6p_outputopts, M_IP6OPT); 505 } 506 if (in6p->in6p_route.ro_rt) 507 rtfree(in6p->in6p_route.ro_rt); 508 ip6_freemoptions(in6p->in6p_moptions); 509 s = splnet(); 510 in6_pcbstate(in6p, IN6P_ATTACHED); 511 CIRCLEQ_REMOVE(&in6p->in6p_table->inpt_queue, &in6p->in6p_head, 512 inph_queue); 513 splx(s); 514 pool_put(&in6pcb_pool, in6p); 515 } 516 517 void 518 in6_setsockaddr(in6p, nam) 519 struct in6pcb *in6p; 520 struct mbuf *nam; 521 { 522 struct sockaddr_in6 *sin6; 523 524 if (in6p->in6p_af != AF_INET6) 525 return; 526 527 nam->m_len = sizeof(*sin6); 528 sin6 = mtod(nam, struct sockaddr_in6 *); 529 bzero((caddr_t)sin6, sizeof(*sin6)); 530 sin6->sin6_family = AF_INET6; 531 sin6->sin6_len = sizeof(struct sockaddr_in6); 532 sin6->sin6_port = in6p->in6p_lport; 533 /* KAME hack: recover scopeid */ 534 (void)in6_recoverscope(sin6, &in6p->in6p_laddr, NULL); 535 } 536 537 void 538 in6_setpeeraddr(in6p, nam) 539 struct in6pcb *in6p; 540 struct mbuf *nam; 541 { 542 struct sockaddr_in6 *sin6; 543 544 if (in6p->in6p_af != AF_INET6) 545 return; 546 547 nam->m_len = sizeof(*sin6); 548 sin6 = mtod(nam, struct sockaddr_in6 *); 549 bzero((caddr_t)sin6, sizeof(*sin6)); 550 sin6->sin6_family = AF_INET6; 551 sin6->sin6_len = sizeof(struct sockaddr_in6); 552 sin6->sin6_port = in6p->in6p_fport; 553 /* KAME hack: recover scopeid */ 554 (void)in6_recoverscope(sin6, &in6p->in6p_faddr, NULL); 555 } 556 557 /* 558 * Pass some notification to all connections of a protocol 559 * associated with address dst. The local address and/or port numbers 560 * may be specified to limit the search. The "usual action" will be 561 * taken, depending on the ctlinput cmd. The caller must filter any 562 * cmds that are uninteresting (e.g., no error in the map). 563 * Call the protocol specific routine (if any) to report 564 * any errors for each matching socket. 565 * 566 * Must be called at splsoftnet. 567 * 568 * Note: src (4th arg) carries the flowlabel value on the original IPv6 569 * header, in sin6_flowinfo member. 570 */ 571 int 572 in6_pcbnotify(table, dst, fport_arg, src, lport_arg, cmd, cmdarg, notify) 573 struct inpcbtable *table; 574 struct sockaddr *dst, *src; 575 u_int fport_arg, lport_arg; 576 int cmd; 577 void *cmdarg; 578 void (*notify) __P((struct in6pcb *, int)); 579 { 580 struct in6pcb *in6p, *nin6p; 581 struct sockaddr_in6 sa6_src, *sa6_dst; 582 u_int16_t fport = fport_arg, lport = lport_arg; 583 int errno; 584 int nmatch = 0; 585 u_int32_t flowinfo; 586 587 if ((unsigned)cmd > PRC_NCMDS || dst->sa_family != AF_INET6) 588 return 0; 589 590 sa6_dst = (struct sockaddr_in6 *)dst; 591 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 592 return 0; 593 594 /* 595 * note that src can be NULL when we get notify by local fragmentation. 596 */ 597 sa6_src = (src == NULL) ? sa6_any : *(struct sockaddr_in6 *)src; 598 flowinfo = sa6_src.sin6_flowinfo; 599 600 /* 601 * Redirects go to all references to the destination, 602 * and use in6_rtchange to invalidate the route cache. 603 * Dead host indications: also use in6_rtchange to invalidate 604 * the cache, and deliver the error to all the sockets. 605 * Otherwise, if we have knowledge of the local port and address, 606 * deliver only to that socket. 607 */ 608 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 609 fport = 0; 610 lport = 0; 611 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 612 613 if (cmd != PRC_HOSTDEAD) 614 notify = in6_rtchange; 615 } 616 617 errno = inet6ctlerrmap[cmd]; 618 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue); 619 in6p != (void *)&table->inpt_queue; 620 in6p = nin6p) { 621 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue); 622 623 if (in6p->in6p_af != AF_INET6) 624 continue; 625 626 /* 627 * Under the following condition, notify of redirects 628 * to the pcb, without making address matches against inpcb. 629 * - redirect notification is arrived. 630 * - the inpcb is unconnected. 631 * - the inpcb is caching !RTF_HOST routing entry. 632 * - the ICMPv6 notification is from the gateway cached in the 633 * inpcb. i.e. ICMPv6 notification is from nexthop gateway 634 * the inpcb used very recently. 635 * 636 * This is to improve interaction between netbsd/openbsd 637 * redirect handling code, and inpcb route cache code. 638 * without the clause, !RTF_HOST routing entry (which carries 639 * gateway used by inpcb right before the ICMPv6 redirect) 640 * will be cached forever in unconnected inpcb. 641 * 642 * There still is a question regarding to what is TRT: 643 * - On bsdi/freebsd, RTF_HOST (cloned) routing entry will be 644 * generated on packet output. inpcb will always cache 645 * RTF_HOST routing entry so there's no need for the clause 646 * (ICMPv6 redirect will update RTF_HOST routing entry, 647 * and inpcb is caching it already). 648 * However, bsdi/freebsd are vulnerable to local DoS attacks 649 * due to the cloned routing entries. 650 * - Specwise, "destination cache" is mentioned in RFC2461. 651 * Jinmei says that it implies bsdi/freebsd behavior, itojun 652 * is not really convinced. 653 * - Having hiwat/lowat on # of cloned host route (redirect/ 654 * pmtud) may be a good idea. netbsd/openbsd has it. see 655 * icmp6_mtudisc_update(). 656 */ 657 if ((PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) && 658 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) && 659 in6p->in6p_route.ro_rt && 660 !(in6p->in6p_route.ro_rt->rt_flags & RTF_HOST)) { 661 struct sockaddr_in6 *dst6; 662 663 dst6 = (struct sockaddr_in6 *)&in6p->in6p_route.ro_dst; 664 if (IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr, 665 &sa6_dst->sin6_addr)) 666 goto do_notify; 667 } 668 669 /* 670 * Detect if we should notify the error. If no source and 671 * destination ports are specified, but non-zero flowinfo and 672 * local address match, notify the error. This is the case 673 * when the error is delivered with an encrypted buffer 674 * by ESP. Otherwise, just compare addresses and ports 675 * as usual. 676 */ 677 if (lport == 0 && fport == 0 && flowinfo && 678 in6p->in6p_socket != NULL && 679 flowinfo == (in6p->in6p_flowinfo & IPV6_FLOWLABEL_MASK) && 680 IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &sa6_src.sin6_addr)) 681 goto do_notify; 682 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, 683 &sa6_dst->sin6_addr) || 684 in6p->in6p_socket == 0 || 685 (lport && in6p->in6p_lport != lport) || 686 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 687 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, 688 &sa6_src.sin6_addr)) || 689 (fport && in6p->in6p_fport != fport)) 690 continue; 691 692 do_notify: 693 if (notify) 694 (*notify)(in6p, errno); 695 nmatch++; 696 } 697 return nmatch; 698 } 699 700 void 701 in6_pcbpurgeif0(table, ifp) 702 struct inpcbtable *table; 703 struct ifnet *ifp; 704 { 705 struct in6pcb *in6p, *nin6p; 706 struct ip6_moptions *im6o; 707 struct in6_multi_mship *imm, *nimm; 708 709 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue); 710 in6p != (void *)&table->inpt_queue; 711 in6p = nin6p) { 712 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue); 713 if (in6p->in6p_af != AF_INET6) 714 continue; 715 716 im6o = in6p->in6p_moptions; 717 if (im6o) { 718 /* 719 * Unselect the outgoing interface if it is being 720 * detached. 721 */ 722 if (im6o->im6o_multicast_ifp == ifp) 723 im6o->im6o_multicast_ifp = NULL; 724 725 /* 726 * Drop multicast group membership if we joined 727 * through the interface being detached. 728 * XXX controversial - is it really legal for kernel 729 * to force this? 730 */ 731 for (imm = im6o->im6o_memberships.lh_first; 732 imm != NULL; imm = nimm) { 733 nimm = imm->i6mm_chain.le_next; 734 if (imm->i6mm_maddr->in6m_ifp == ifp) { 735 LIST_REMOVE(imm, i6mm_chain); 736 in6_leavegroup(imm); 737 } 738 } 739 } 740 } 741 } 742 743 void 744 in6_pcbpurgeif(table, ifp) 745 struct inpcbtable *table; 746 struct ifnet *ifp; 747 { 748 struct in6pcb *in6p, *nin6p; 749 750 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue); 751 in6p != (void *)&table->inpt_queue; 752 in6p = nin6p) { 753 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue); 754 if (in6p->in6p_af != AF_INET6) 755 continue; 756 if (in6p->in6p_route.ro_rt != NULL && 757 in6p->in6p_route.ro_rt->rt_ifp == ifp) 758 in6_rtchange(in6p, 0); 759 } 760 } 761 762 /* 763 * Check for alternatives when higher level complains 764 * about service problems. For now, invalidate cached 765 * routing information. If the route was created dynamically 766 * (by a redirect), time to try a default gateway again. 767 */ 768 void 769 in6_losing(in6p) 770 struct in6pcb *in6p; 771 { 772 struct rtentry *rt; 773 struct rt_addrinfo info; 774 775 if (in6p->in6p_af != AF_INET6) 776 return; 777 778 if ((rt = in6p->in6p_route.ro_rt) != NULL) { 779 in6p->in6p_route.ro_rt = 0; 780 bzero((caddr_t)&info, sizeof(info)); 781 info.rti_info[RTAX_DST] = 782 (struct sockaddr *)&in6p->in6p_route.ro_dst; 783 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 784 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 785 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 786 if (rt->rt_flags & RTF_DYNAMIC) { 787 (void)rtrequest(RTM_DELETE, rt_key(rt), 788 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 789 (struct rtentry **)0); 790 } else { 791 /* 792 * A new route can be allocated 793 * the next time output is attempted. 794 */ 795 rtfree(rt); 796 } 797 } 798 } 799 800 /* 801 * After a routing change, flush old routing 802 * and allocate a (hopefully) better one. 803 */ 804 void 805 in6_rtchange(in6p, errno) 806 struct in6pcb *in6p; 807 int errno; 808 { 809 if (in6p->in6p_af != AF_INET6) 810 return; 811 812 if (in6p->in6p_route.ro_rt) { 813 rtfree(in6p->in6p_route.ro_rt); 814 in6p->in6p_route.ro_rt = 0; 815 /* 816 * A new route can be allocated the next time 817 * output is attempted. 818 */ 819 } 820 } 821 822 struct in6pcb * 823 in6_pcblookup_port(table, laddr6, lport_arg, lookup_wildcard) 824 struct inpcbtable *table; 825 struct in6_addr *laddr6; 826 u_int lport_arg; 827 int lookup_wildcard; 828 { 829 struct inpcb_hdr *inph; 830 struct in6pcb *in6p, *match = 0; 831 int matchwild = 3, wildcard; 832 u_int16_t lport = lport_arg; 833 834 CIRCLEQ_FOREACH(inph, &table->inpt_queue, inph_queue) { 835 in6p = (struct in6pcb *)inph; 836 if (in6p->in6p_af != AF_INET6) 837 continue; 838 839 if (in6p->in6p_lport != lport) 840 continue; 841 wildcard = 0; 842 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) { 843 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 844 continue; 845 } 846 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) 847 wildcard++; 848 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) { 849 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 850 continue; 851 if (!IN6_IS_ADDR_V4MAPPED(laddr6)) 852 continue; 853 854 /* duplicate of IPv4 logic */ 855 wildcard = 0; 856 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr) && 857 in6p->in6p_faddr.s6_addr32[3]) 858 wildcard++; 859 if (!in6p->in6p_laddr.s6_addr32[3]) { 860 if (laddr6->s6_addr32[3]) 861 wildcard++; 862 } else { 863 if (!laddr6->s6_addr32[3]) 864 wildcard++; 865 else { 866 if (in6p->in6p_laddr.s6_addr32[3] != 867 laddr6->s6_addr32[3]) 868 continue; 869 } 870 } 871 } else if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { 872 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 873 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 874 continue; 875 } 876 if (!IN6_IS_ADDR_UNSPECIFIED(laddr6)) 877 wildcard++; 878 } else { 879 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 880 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 881 continue; 882 } 883 if (IN6_IS_ADDR_UNSPECIFIED(laddr6)) 884 wildcard++; 885 else { 886 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, 887 laddr6)) 888 continue; 889 } 890 } 891 if (wildcard && !lookup_wildcard) 892 continue; 893 if (wildcard < matchwild) { 894 match = in6p; 895 matchwild = wildcard; 896 if (matchwild == 0) 897 break; 898 } 899 } 900 return (match); 901 } 902 #undef continue 903 904 /* 905 * WARNING: return value (rtentry) could be IPv4 one if in6pcb is connected to 906 * IPv4 mapped address. 907 */ 908 struct rtentry * 909 in6_pcbrtentry(in6p) 910 struct in6pcb *in6p; 911 { 912 struct route_in6 *ro; 913 struct sockaddr_in6 *dst6; 914 915 ro = &in6p->in6p_route; 916 dst6 = (struct sockaddr_in6 *)&ro->ro_dst; 917 918 if (in6p->in6p_af != AF_INET6) 919 return (NULL); 920 921 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 922 !IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr, &in6p->in6p_faddr))) { 923 RTFREE(ro->ro_rt); 924 ro->ro_rt = (struct rtentry *)NULL; 925 } 926 #ifdef INET 927 if (ro->ro_rt == (struct rtentry *)NULL && 928 IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) { 929 struct sockaddr_in *dst = (struct sockaddr_in *)&ro->ro_dst; 930 931 bzero(dst, sizeof(*dst)); 932 dst->sin_family = AF_INET; 933 dst->sin_len = sizeof(struct sockaddr_in); 934 bcopy(&in6p->in6p_faddr.s6_addr32[3], &dst->sin_addr, 935 sizeof(dst->sin_addr)); 936 rtalloc((struct route *)ro); 937 } else 938 #endif 939 if (ro->ro_rt == (struct rtentry *)NULL && 940 !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) { 941 bzero(dst6, sizeof(*dst6)); 942 dst6->sin6_family = AF_INET6; 943 dst6->sin6_len = sizeof(struct sockaddr_in6); 944 dst6->sin6_addr = in6p->in6p_faddr; 945 rtalloc((struct route *)ro); 946 } 947 return (ro->ro_rt); 948 } 949 950 struct in6pcb * 951 in6_pcblookup_connect(table, faddr6, fport_arg, laddr6, lport_arg, faith) 952 struct inpcbtable *table; 953 struct in6_addr *faddr6, *laddr6; 954 u_int fport_arg, lport_arg; 955 int faith; 956 { 957 struct inpcbhead *head; 958 struct inpcb_hdr *inph; 959 struct in6pcb *in6p; 960 u_int16_t fport = fport_arg, lport = lport_arg; 961 962 head = IN6PCBHASH_CONNECT(table, faddr6, fport, laddr6, lport); 963 LIST_FOREACH(inph, head, inph_hash) { 964 in6p = (struct in6pcb *)inph; 965 if (in6p->in6p_af != AF_INET6) 966 continue; 967 968 /* find exact match on both source and dest */ 969 if (in6p->in6p_fport != fport) 970 continue; 971 if (in6p->in6p_lport != lport) 972 continue; 973 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) 974 continue; 975 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6)) 976 continue; 977 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 978 continue; 979 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 980 continue; 981 if ((IN6_IS_ADDR_V4MAPPED(laddr6) || 982 IN6_IS_ADDR_V4MAPPED(faddr6)) && 983 (in6p->in6p_flags & IN6P_IPV6_V6ONLY)) 984 continue; 985 return in6p; 986 } 987 return NULL; 988 } 989 990 struct in6pcb * 991 in6_pcblookup_bind(table, laddr6, lport_arg, faith) 992 struct inpcbtable *table; 993 struct in6_addr *laddr6; 994 u_int lport_arg; 995 int faith; 996 { 997 struct inpcbhead *head; 998 struct inpcb_hdr *inph; 999 struct in6pcb *in6p; 1000 u_int16_t lport = lport_arg; 1001 #ifdef INET6 1002 struct in6_addr zero_mapped; 1003 #endif 1004 1005 head = IN6PCBHASH_BIND(table, laddr6, lport); 1006 LIST_FOREACH(inph, head, inph_hash) { 1007 in6p = (struct in6pcb *)inph; 1008 if (in6p->in6p_af != AF_INET6) 1009 continue; 1010 1011 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0) 1012 continue; 1013 if (in6p->in6p_fport != 0) 1014 continue; 1015 if (in6p->in6p_lport != lport) 1016 continue; 1017 if (IN6_IS_ADDR_V4MAPPED(laddr6) && 1018 (in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 1019 continue; 1020 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 1021 goto out; 1022 } 1023 #ifdef INET 1024 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 1025 memset(&zero_mapped, 0, sizeof(zero_mapped)); 1026 zero_mapped.s6_addr16[5] = 0xffff; 1027 head = IN6PCBHASH_BIND(table, &zero_mapped, lport); 1028 LIST_FOREACH(inph, head, inph_hash) { 1029 in6p = (struct in6pcb *)inph; 1030 if (in6p->in6p_af != AF_INET6) 1031 continue; 1032 1033 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0) 1034 continue; 1035 if (in6p->in6p_fport != 0) 1036 continue; 1037 if (in6p->in6p_lport != lport) 1038 continue; 1039 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 1040 continue; 1041 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &zero_mapped)) 1042 goto out; 1043 } 1044 } 1045 #endif 1046 head = IN6PCBHASH_BIND(table, &zeroin6_addr, lport); 1047 LIST_FOREACH(inph, head, inph_hash) { 1048 in6p = (struct in6pcb *)inph; 1049 if (in6p->in6p_af != AF_INET6) 1050 continue; 1051 1052 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0) 1053 continue; 1054 if (in6p->in6p_fport != 0) 1055 continue; 1056 if (in6p->in6p_lport != lport) 1057 continue; 1058 if (IN6_IS_ADDR_V4MAPPED(laddr6) && 1059 (in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 1060 continue; 1061 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &zeroin6_addr)) 1062 goto out; 1063 } 1064 return (NULL); 1065 1066 out: 1067 inph = &in6p->in6p_head; 1068 if (inph != LIST_FIRST(head)) { 1069 LIST_REMOVE(inph, inph_hash); 1070 LIST_INSERT_HEAD(head, inph, inph_hash); 1071 } 1072 return in6p; 1073 } 1074 1075 void 1076 in6_pcbstate(in6p, state) 1077 struct in6pcb *in6p; 1078 int state; 1079 { 1080 1081 if (in6p->in6p_af != AF_INET6) 1082 return; 1083 1084 if (in6p->in6p_state > IN6P_ATTACHED) 1085 LIST_REMOVE(&in6p->in6p_head, inph_hash); 1086 1087 switch (state) { 1088 case IN6P_BOUND: 1089 LIST_INSERT_HEAD(IN6PCBHASH_BIND(in6p->in6p_table, 1090 &in6p->in6p_laddr, in6p->in6p_lport), &in6p->in6p_head, 1091 inph_hash); 1092 break; 1093 case IN6P_CONNECTED: 1094 LIST_INSERT_HEAD(IN6PCBHASH_CONNECT(in6p->in6p_table, 1095 &in6p->in6p_faddr, in6p->in6p_fport, 1096 &in6p->in6p_laddr, in6p->in6p_lport), &in6p->in6p_head, 1097 inph_hash); 1098 break; 1099 } 1100 1101 in6p->in6p_state = state; 1102 } 1103
Indexes created Wed Sep 24 14:09:57 GMT 2025