in6_pcb.c revision 1.54
1 /* $NetBSD: in6_pcb.c,v 1.54 2003/08/07 16:33:24 agc 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 1.54 2003/08/07 16:33:24 agc 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 int ip6_anonportmin = IPV6PORT_ANONMIN; 107 int ip6_anonportmax = IPV6PORT_ANONMAX; 108 int ip6_lowportmin = IPV6PORT_RESERVEDMIN; 109 int ip6_lowportmax = IPV6PORT_RESERVEDMAX; 110 111 int 112 in6_pcballoc(so, head) 113 struct socket *so; 114 struct in6pcb *head; 115 { 116 struct in6pcb *in6p; 117 int s; 118 #ifdef IPSEC 119 int error; 120 #endif 121 122 MALLOC(in6p, struct in6pcb *, sizeof(*in6p), M_PCB, M_NOWAIT); 123 if (in6p == NULL) 124 return (ENOBUFS); 125 bzero((caddr_t)in6p, sizeof(*in6p)); 126 in6p->in6p_head = head; 127 in6p->in6p_socket = so; 128 in6p->in6p_hops = -1; /* use kernel default */ 129 in6p->in6p_icmp6filt = NULL; 130 #ifdef IPSEC 131 error = ipsec_init_pcbpolicy(so, &in6p->in6p_sp); 132 if (error != 0) { 133 FREE(in6p, M_PCB); 134 return error; 135 } 136 #endif /* IPSEC */ 137 s = splnet(); 138 in6p->in6p_next = head->in6p_next; 139 head->in6p_next = in6p; 140 in6p->in6p_prev = head; 141 in6p->in6p_next->in6p_prev = in6p; 142 splx(s); 143 if (ip6_v6only) 144 in6p->in6p_flags |= IN6P_IPV6_V6ONLY; 145 so->so_pcb = (caddr_t)in6p; 146 return (0); 147 } 148 149 int 150 in6_pcbbind(in6p, nam, p) 151 struct in6pcb *in6p; 152 struct mbuf *nam; 153 struct proc *p; 154 { 155 struct socket *so = in6p->in6p_socket; 156 struct in6pcb *head = in6p->in6p_head; 157 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 158 u_int16_t lport = 0; 159 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 160 161 if (in6p->in6p_lport || !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 162 return (EINVAL); 163 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 && 164 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 || 165 (so->so_options & SO_ACCEPTCONN) == 0)) 166 wild = IN6PLOOKUP_WILDCARD; 167 if (nam) { 168 sin6 = mtod(nam, struct sockaddr_in6 *); 169 if (nam->m_len != sizeof(*sin6)) 170 return (EINVAL); 171 /* 172 * We should check the family, but old programs 173 * incorrectly fail to intialize it. 174 */ 175 if (sin6->sin6_family != AF_INET6) 176 return (EAFNOSUPPORT); 177 178 /* 179 * since we do not check port number duplicate with IPv4 space, 180 * we reject it at this moment. If we leave it, we would 181 * mistakenly allow normal users to hijack tcp/udp ports from 182 * other users. 183 */ 184 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) 185 return (EADDRNOTAVAIL); 186 187 /* KAME hack: embed scopeid */ 188 if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, NULL) != 0) 189 return EINVAL; 190 /* this must be cleared for ifa_ifwithaddr() */ 191 sin6->sin6_scope_id = 0; 192 193 lport = sin6->sin6_port; 194 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 195 /* 196 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 197 * allow compepte duplication of binding if 198 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 199 * and a multicast address is bound on both 200 * new and duplicated sockets. 201 */ 202 if (so->so_options & SO_REUSEADDR) 203 reuseport = SO_REUSEADDR|SO_REUSEPORT; 204 } 205 else if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 206 struct sockaddr_in sin; 207 208 bzero(&sin, sizeof(sin)); 209 sin.sin_len = sizeof(sin); 210 sin.sin_family = AF_INET; 211 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr, 212 sizeof(sin.sin_addr)); 213 if (ifa_ifwithaddr((struct sockaddr *)&sin) == 0) 214 return EADDRNOTAVAIL; 215 } 216 else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 217 struct ifaddr *ia = NULL; 218 219 sin6->sin6_port = 0; /* yech... */ 220 if ((in6p->in6p_flags & IN6P_FAITH) == 0 && 221 (ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0) 222 return (EADDRNOTAVAIL); 223 224 /* 225 * bind to an anycast address might accidentally 226 * cause sending a packet with an anycast source 227 * address, so we forbid it. 228 * 229 * We should allow to bind to a deprecated address, 230 * since the application dare to use it. 231 * But, can we assume that they are careful enough 232 * to check if the address is deprecated or not? 233 * Maybe, as a safeguard, we should have a setsockopt 234 * flag to control the bind(2) behavior against 235 * deprecated addresses (default: forbid bind(2)). 236 */ 237 if (ia && 238 ((struct in6_ifaddr *)ia)->ia6_flags & 239 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) 240 return (EADDRNOTAVAIL); 241 } 242 if (lport) { 243 #ifndef IPNOPRIVPORTS 244 int priv; 245 246 /* 247 * NOTE: all operating systems use suser() for 248 * privilege check! do not rewrite it into SS_PRIV. 249 */ 250 priv = (p && !suser(p->p_ucred, &p->p_acflag)) ? 1 : 0; 251 /* GROSS */ 252 if (ntohs(lport) < IPV6PORT_RESERVED && !priv) 253 return (EACCES); 254 #endif 255 256 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 257 /* should check this but we can't ... */ 258 #if 0 259 struct inpcb *t; 260 261 t = in_pcblookup_bind(&tcbtable, 262 (struct in_addr *)&sin6->sin6_addr.s6_addr32[3], 263 lport); 264 if (t && (reuseport & t->inp_socket->so_options) == 0) 265 return EADDRINUSE; 266 #endif 267 } else 268 { 269 struct in6pcb *t; 270 271 t = in6_pcblookup(head, &zeroin6_addr, 0, 272 &sin6->sin6_addr, lport, wild); 273 if (t && (reuseport & t->in6p_socket->so_options) == 0) 274 return (EADDRINUSE); 275 } 276 } 277 in6p->in6p_laddr = sin6->sin6_addr; 278 } 279 280 if (lport == 0) { 281 int e; 282 if ((e = in6_pcbsetport(&in6p->in6p_laddr, in6p, p)) != 0) 283 return (e); 284 } 285 else 286 in6p->in6p_lport = lport; 287 288 in6p->in6p_flowinfo = sin6 ? sin6->sin6_flowinfo : 0; /*XXX*/ 289 return (0); 290 } 291 292 /* 293 * Connect from a socket to a specified address. 294 * Both address and port must be specified in argument sin6. 295 * If don't have a local address for this socket yet, 296 * then pick one. 297 */ 298 int 299 in6_pcbconnect(in6p, nam) 300 struct in6pcb *in6p; 301 struct mbuf *nam; 302 { 303 struct in6_addr *in6a = NULL; 304 struct sockaddr_in6 *sin6 = mtod(nam, struct sockaddr_in6 *); 305 struct ifnet *ifp = NULL; /* outgoing interface */ 306 int error = 0; 307 #ifdef INET 308 struct in6_addr mapped; 309 #endif 310 struct sockaddr_in6 tmp; 311 312 (void)&in6a; /* XXX fool gcc */ 313 314 if (nam->m_len != sizeof(*sin6)) 315 return (EINVAL); 316 if (sin6->sin6_family != AF_INET6) 317 return (EAFNOSUPPORT); 318 if (sin6->sin6_port == 0) 319 return (EADDRNOTAVAIL); 320 321 /* sanity check for mapped address case */ 322 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 323 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 324 return EINVAL; 325 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 326 in6p->in6p_laddr.s6_addr16[5] = htons(0xffff); 327 if (!IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 328 return EINVAL; 329 } else 330 { 331 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 332 return EINVAL; 333 } 334 335 /* protect *sin6 from overwrites */ 336 tmp = *sin6; 337 sin6 = &tmp; 338 339 /* KAME hack: embed scopeid */ 340 if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, &ifp) != 0) 341 return EINVAL; 342 343 /* Source address selection. */ 344 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 345 && in6p->in6p_laddr.s6_addr32[3] == 0) { 346 #ifdef INET 347 struct sockaddr_in sin, *sinp; 348 349 bzero(&sin, sizeof(sin)); 350 sin.sin_len = sizeof(sin); 351 sin.sin_family = AF_INET; 352 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr, 353 sizeof(sin.sin_addr)); 354 sinp = in_selectsrc(&sin, (struct route *)&in6p->in6p_route, 355 in6p->in6p_socket->so_options, NULL, &error); 356 if (sinp == 0) { 357 if (error == 0) 358 error = EADDRNOTAVAIL; 359 return (error); 360 } 361 bzero(&mapped, sizeof(mapped)); 362 mapped.s6_addr16[5] = htons(0xffff); 363 bcopy(&sinp->sin_addr, &mapped.s6_addr32[3], sizeof(sinp->sin_addr)); 364 in6a = &mapped; 365 #else 366 return EADDRNOTAVAIL; 367 #endif 368 } else 369 { 370 /* 371 * XXX: in6_selectsrc might replace the bound local address 372 * with the address specified by setsockopt(IPV6_PKTINFO). 373 * Is it the intended behavior? 374 */ 375 in6a = in6_selectsrc(sin6, in6p->in6p_outputopts, 376 in6p->in6p_moptions, 377 &in6p->in6p_route, 378 &in6p->in6p_laddr, &error); 379 if (in6a == 0) { 380 if (error == 0) 381 error = EADDRNOTAVAIL; 382 return (error); 383 } 384 } 385 if (in6p->in6p_route.ro_rt) 386 ifp = in6p->in6p_route.ro_rt->rt_ifp; 387 388 in6p->in6p_ip6.ip6_hlim = (u_int8_t)in6_selecthlim(in6p, ifp); 389 390 if (in6_pcblookup(in6p->in6p_head, 391 &sin6->sin6_addr, 392 sin6->sin6_port, 393 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ? 394 in6a : &in6p->in6p_laddr, 395 in6p->in6p_lport, 396 0)) 397 return (EADDRINUSE); 398 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) 399 || (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 400 && in6p->in6p_laddr.s6_addr32[3] == 0)) 401 { 402 /* 403 * XXX in IPv4 mapped address case, we should grab fresh 404 * local port number by in_pcbbind, not in6_pcbbind. 405 * if we are in bad luck, we may assign conflicting port number 406 * between IPv4 and IPv6 unwillingly. 407 */ 408 if (in6p->in6p_lport == 0) { 409 (void)in6_pcbbind(in6p, (struct mbuf *)0, 410 (struct proc *)0); 411 } 412 in6p->in6p_laddr = *in6a; 413 } 414 in6p->in6p_faddr = sin6->sin6_addr; 415 in6p->in6p_fport = sin6->sin6_port; 416 /* 417 * xxx kazu flowlabel is necessary for connect? 418 * but if this line is missing, the garbage value remains. 419 */ 420 in6p->in6p_flowinfo = sin6->sin6_flowinfo; 421 #ifdef IPSEC 422 if (in6p->in6p_socket->so_type == SOCK_STREAM) 423 ipsec_pcbconn(in6p->in6p_sp); 424 #endif 425 return (0); 426 } 427 428 void 429 in6_pcbdisconnect(in6p) 430 struct in6pcb *in6p; 431 { 432 bzero((caddr_t)&in6p->in6p_faddr, sizeof(in6p->in6p_faddr)); 433 in6p->in6p_fport = 0; 434 if (in6p->in6p_socket->so_state & SS_NOFDREF) 435 in6_pcbdetach(in6p); 436 #ifdef IPSEC 437 ipsec_pcbdisconn(in6p->in6p_sp); 438 #endif 439 } 440 441 void 442 in6_pcbdetach(in6p) 443 struct in6pcb *in6p; 444 { 445 struct socket *so = in6p->in6p_socket; 446 int s; 447 448 #ifdef IPSEC 449 ipsec6_delete_pcbpolicy(in6p); 450 #endif /* IPSEC */ 451 sotoin6pcb(so) = 0; 452 sofree(so); 453 if (in6p->in6p_options) 454 m_freem(in6p->in6p_options); 455 if (in6p->in6p_outputopts) { 456 if (in6p->in6p_outputopts->ip6po_rthdr && 457 in6p->in6p_outputopts->ip6po_route.ro_rt) 458 RTFREE(in6p->in6p_outputopts->ip6po_route.ro_rt); 459 if (in6p->in6p_outputopts->ip6po_m) 460 (void)m_free(in6p->in6p_outputopts->ip6po_m); 461 free(in6p->in6p_outputopts, M_IP6OPT); 462 } 463 if (in6p->in6p_route.ro_rt) 464 rtfree(in6p->in6p_route.ro_rt); 465 ip6_freemoptions(in6p->in6p_moptions); 466 s = splnet(); 467 in6p->in6p_next->in6p_prev = in6p->in6p_prev; 468 in6p->in6p_prev->in6p_next = in6p->in6p_next; 469 in6p->in6p_prev = NULL; 470 splx(s); 471 FREE(in6p, M_PCB); 472 } 473 474 void 475 in6_setsockaddr(in6p, nam) 476 struct in6pcb *in6p; 477 struct mbuf *nam; 478 { 479 struct sockaddr_in6 *sin6; 480 481 nam->m_len = sizeof(*sin6); 482 sin6 = mtod(nam, struct sockaddr_in6 *); 483 bzero((caddr_t)sin6, sizeof(*sin6)); 484 sin6->sin6_family = AF_INET6; 485 sin6->sin6_len = sizeof(struct sockaddr_in6); 486 sin6->sin6_port = in6p->in6p_lport; 487 /* KAME hack: recover scopeid */ 488 (void)in6_recoverscope(sin6, &in6p->in6p_laddr, NULL); 489 } 490 491 void 492 in6_setpeeraddr(in6p, nam) 493 struct in6pcb *in6p; 494 struct mbuf *nam; 495 { 496 struct sockaddr_in6 *sin6; 497 498 nam->m_len = sizeof(*sin6); 499 sin6 = mtod(nam, struct sockaddr_in6 *); 500 bzero((caddr_t)sin6, sizeof(*sin6)); 501 sin6->sin6_family = AF_INET6; 502 sin6->sin6_len = sizeof(struct sockaddr_in6); 503 sin6->sin6_port = in6p->in6p_fport; 504 /* KAME hack: recover scopeid */ 505 (void)in6_recoverscope(sin6, &in6p->in6p_faddr, NULL); 506 } 507 508 /* 509 * Pass some notification to all connections of a protocol 510 * associated with address dst. The local address and/or port numbers 511 * may be specified to limit the search. The "usual action" will be 512 * taken, depending on the ctlinput cmd. The caller must filter any 513 * cmds that are uninteresting (e.g., no error in the map). 514 * Call the protocol specific routine (if any) to report 515 * any errors for each matching socket. 516 * 517 * Must be called at splsoftnet. 518 * 519 * Note: src (4th arg) carries the flowlabel value on the original IPv6 520 * header, in sin6_flowinfo member. 521 */ 522 int 523 in6_pcbnotify(head, dst, fport_arg, src, lport_arg, cmd, cmdarg, notify) 524 struct in6pcb *head; 525 struct sockaddr *dst, *src; 526 u_int fport_arg, lport_arg; 527 int cmd; 528 void *cmdarg; 529 void (*notify) __P((struct in6pcb *, int)); 530 { 531 struct in6pcb *in6p, *nin6p; 532 struct sockaddr_in6 sa6_src, *sa6_dst; 533 u_int16_t fport = fport_arg, lport = lport_arg; 534 int errno; 535 int nmatch = 0; 536 u_int32_t flowinfo; 537 538 if ((unsigned)cmd > PRC_NCMDS || dst->sa_family != AF_INET6) 539 return 0; 540 541 sa6_dst = (struct sockaddr_in6 *)dst; 542 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 543 return 0; 544 545 /* 546 * note that src can be NULL when we get notify by local fragmentation. 547 */ 548 sa6_src = (src == NULL) ? sa6_any : *(struct sockaddr_in6 *)src; 549 flowinfo = sa6_src.sin6_flowinfo; 550 551 /* 552 * Redirects go to all references to the destination, 553 * and use in6_rtchange to invalidate the route cache. 554 * Dead host indications: also use in6_rtchange to invalidate 555 * the cache, and deliver the error to all the sockets. 556 * Otherwise, if we have knowledge of the local port and address, 557 * deliver only to that socket. 558 */ 559 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 560 fport = 0; 561 lport = 0; 562 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 563 564 if (cmd != PRC_HOSTDEAD) 565 notify = in6_rtchange; 566 } 567 568 errno = inet6ctlerrmap[cmd]; 569 for (in6p = head->in6p_next; in6p != head; in6p = nin6p) { 570 nin6p = in6p->in6p_next; 571 572 /* 573 * Under the following condition, notify of redirects 574 * to the pcb, without making address matches against inpcb. 575 * - redirect notification is arrived. 576 * - the inpcb is unconnected. 577 * - the inpcb is caching !RTF_HOST routing entry. 578 * - the ICMPv6 notification is from the gateway cached in the 579 * inpcb. i.e. ICMPv6 notification is from nexthop gateway 580 * the inpcb used very recently. 581 * 582 * This is to improve interaction between netbsd/openbsd 583 * redirect handling code, and inpcb route cache code. 584 * without the clause, !RTF_HOST routing entry (which carries 585 * gateway used by inpcb right before the ICMPv6 redirect) 586 * will be cached forever in unconnected inpcb. 587 * 588 * There still is a question regarding to what is TRT: 589 * - On bsdi/freebsd, RTF_HOST (cloned) routing entry will be 590 * generated on packet output. inpcb will always cache 591 * RTF_HOST routing entry so there's no need for the clause 592 * (ICMPv6 redirect will update RTF_HOST routing entry, 593 * and inpcb is caching it already). 594 * However, bsdi/freebsd are vulnerable to local DoS attacks 595 * due to the cloned routing entries. 596 * - Specwise, "destination cache" is mentioned in RFC2461. 597 * Jinmei says that it implies bsdi/freebsd behavior, itojun 598 * is not really convinced. 599 * - Having hiwat/lowat on # of cloned host route (redirect/ 600 * pmtud) may be a good idea. netbsd/openbsd has it. see 601 * icmp6_mtudisc_update(). 602 */ 603 if ((PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) && 604 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) && 605 in6p->in6p_route.ro_rt && 606 !(in6p->in6p_route.ro_rt->rt_flags & RTF_HOST)) { 607 struct sockaddr_in6 *dst6; 608 609 dst6 = (struct sockaddr_in6 *)&in6p->in6p_route.ro_dst; 610 if (IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr, 611 &sa6_dst->sin6_addr)) 612 goto do_notify; 613 } 614 615 /* 616 * Detect if we should notify the error. If no source and 617 * destination ports are specified, but non-zero flowinfo and 618 * local address match, notify the error. This is the case 619 * when the error is delivered with an encrypted buffer 620 * by ESP. Otherwise, just compare addresses and ports 621 * as usual. 622 */ 623 if (lport == 0 && fport == 0 && flowinfo && 624 in6p->in6p_socket != NULL && 625 flowinfo == (in6p->in6p_flowinfo & IPV6_FLOWLABEL_MASK) && 626 IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &sa6_src.sin6_addr)) 627 goto do_notify; 628 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, 629 &sa6_dst->sin6_addr) || 630 in6p->in6p_socket == 0 || 631 (lport && in6p->in6p_lport != lport) || 632 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 633 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, 634 &sa6_src.sin6_addr)) || 635 (fport && in6p->in6p_fport != fport)) 636 continue; 637 638 do_notify: 639 if (notify) 640 (*notify)(in6p, errno); 641 nmatch++; 642 } 643 return nmatch; 644 } 645 646 void 647 in6_pcbpurgeif0(head, ifp) 648 struct in6pcb *head; 649 struct ifnet *ifp; 650 { 651 struct in6pcb *in6p, *nin6p; 652 struct ip6_moptions *im6o; 653 struct in6_multi_mship *imm, *nimm; 654 655 for (in6p = head->in6p_next; in6p != head; in6p = nin6p) { 656 nin6p = in6p->in6p_next; 657 im6o = in6p->in6p_moptions; 658 if (im6o) { 659 /* 660 * Unselect the outgoing interface if it is being 661 * detached. 662 */ 663 if (im6o->im6o_multicast_ifp == ifp) 664 im6o->im6o_multicast_ifp = NULL; 665 666 /* 667 * Drop multicast group membership if we joined 668 * through the interface being detached. 669 * XXX controversial - is it really legal for kernel 670 * to force this? 671 */ 672 for (imm = im6o->im6o_memberships.lh_first; 673 imm != NULL; imm = nimm) { 674 nimm = imm->i6mm_chain.le_next; 675 if (imm->i6mm_maddr->in6m_ifp == ifp) { 676 LIST_REMOVE(imm, i6mm_chain); 677 in6_leavegroup(imm); 678 } 679 } 680 } 681 } 682 } 683 684 void 685 in6_pcbpurgeif(head, ifp) 686 struct in6pcb *head; 687 struct ifnet *ifp; 688 { 689 struct in6pcb *in6p, *nin6p; 690 691 for (in6p = head->in6p_next; in6p != head; in6p = nin6p) { 692 nin6p = in6p->in6p_next; 693 if (in6p->in6p_route.ro_rt != NULL && 694 in6p->in6p_route.ro_rt->rt_ifp == ifp) 695 in6_rtchange(in6p, 0); 696 } 697 } 698 699 /* 700 * Check for alternatives when higher level complains 701 * about service problems. For now, invalidate cached 702 * routing information. If the route was created dynamically 703 * (by a redirect), time to try a default gateway again. 704 */ 705 void 706 in6_losing(in6p) 707 struct in6pcb *in6p; 708 { 709 struct rtentry *rt; 710 struct rt_addrinfo info; 711 712 if ((rt = in6p->in6p_route.ro_rt) != NULL) { 713 in6p->in6p_route.ro_rt = 0; 714 bzero((caddr_t)&info, sizeof(info)); 715 info.rti_info[RTAX_DST] = 716 (struct sockaddr *)&in6p->in6p_route.ro_dst; 717 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 718 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 719 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 720 if (rt->rt_flags & RTF_DYNAMIC) { 721 (void)rtrequest(RTM_DELETE, rt_key(rt), 722 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 723 (struct rtentry **)0); 724 } else { 725 /* 726 * A new route can be allocated 727 * the next time output is attempted. 728 */ 729 rtfree(rt); 730 } 731 } 732 } 733 734 /* 735 * After a routing change, flush old routing 736 * and allocate a (hopefully) better one. 737 */ 738 void 739 in6_rtchange(in6p, errno) 740 struct in6pcb *in6p; 741 int errno; 742 { 743 if (in6p->in6p_route.ro_rt) { 744 rtfree(in6p->in6p_route.ro_rt); 745 in6p->in6p_route.ro_rt = 0; 746 /* 747 * A new route can be allocated the next time 748 * output is attempted. 749 */ 750 } 751 } 752 753 struct in6pcb * 754 in6_pcblookup(head, faddr6, fport_arg, laddr6, lport_arg, flags) 755 struct in6pcb *head; 756 struct in6_addr *faddr6, *laddr6; 757 u_int fport_arg, lport_arg; 758 int flags; 759 { 760 struct in6pcb *in6p, *match = 0; 761 int matchwild = 3, wildcard; 762 u_int16_t fport = fport_arg, lport = lport_arg; 763 764 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) { 765 if (in6p->in6p_lport != lport) 766 continue; 767 wildcard = 0; 768 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { 769 if (IN6_IS_ADDR_UNSPECIFIED(laddr6)) 770 wildcard++; 771 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 772 continue; 773 } 774 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 775 && in6p->in6p_laddr.s6_addr32[3] == 0) { 776 if (!IN6_IS_ADDR_V4MAPPED(laddr6)) 777 continue; 778 if (laddr6->s6_addr32[3] == 0) 779 ; 780 else 781 wildcard++; 782 } 783 else { 784 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 785 if (in6p->in6p_flags & IN6P_IPV6_V6ONLY) 786 continue; 787 else 788 wildcard++; 789 } else if (!IN6_IS_ADDR_UNSPECIFIED(laddr6)) 790 wildcard++; 791 } 792 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) { 793 if (IN6_IS_ADDR_UNSPECIFIED(faddr6)) 794 wildcard++; 795 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6) 796 || in6p->in6p_fport != fport) 797 continue; 798 } 799 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr) 800 && in6p->in6p_faddr.s6_addr32[3] == 0) { 801 if (!IN6_IS_ADDR_V4MAPPED(faddr6)) 802 continue; 803 if (faddr6->s6_addr32[3] == 0) 804 ; 805 else 806 wildcard++; 807 } 808 else { 809 if (IN6_IS_ADDR_V4MAPPED(faddr6)) { 810 if (in6p->in6p_flags & IN6P_IPV6_V6ONLY) 811 continue; 812 else 813 wildcard++; 814 } else if (!IN6_IS_ADDR_UNSPECIFIED(faddr6)) 815 wildcard++; 816 } 817 818 if (wildcard && (flags & IN6PLOOKUP_WILDCARD) == 0) 819 continue; 820 if (wildcard < matchwild) { 821 match = in6p; 822 matchwild = wildcard; 823 if (matchwild == 0) 824 break; 825 } 826 } 827 return (match); 828 } 829 830 struct rtentry * 831 in6_pcbrtentry(in6p) 832 struct in6pcb *in6p; 833 { 834 struct route_in6 *ro; 835 struct sockaddr_in6 *dst6; 836 837 ro = &in6p->in6p_route; 838 dst6 = (struct sockaddr_in6 *)&ro->ro_dst; 839 840 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 841 !IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr, &in6p->in6p_faddr))) { 842 RTFREE(ro->ro_rt); 843 ro->ro_rt = (struct rtentry *)NULL; 844 } 845 if (ro->ro_rt == (struct rtentry *)NULL && 846 !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) { 847 bzero(dst6, sizeof(*dst6)); 848 dst6->sin6_family = AF_INET6; 849 dst6->sin6_len = sizeof(struct sockaddr_in6); 850 dst6->sin6_addr = in6p->in6p_faddr; 851 rtalloc((struct route *)ro); 852 } 853 return (ro->ro_rt); 854 } 855 856 struct in6pcb * 857 in6_pcblookup_connect(head, faddr6, fport_arg, laddr6, lport_arg, faith) 858 struct in6pcb *head; 859 struct in6_addr *faddr6, *laddr6; 860 u_int fport_arg, lport_arg; 861 int faith; 862 { 863 struct in6pcb *in6p; 864 u_int16_t fport = fport_arg, lport = lport_arg; 865 866 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) { 867 /* find exact match on both source and dest */ 868 if (in6p->in6p_fport != fport) 869 continue; 870 if (in6p->in6p_lport != lport) 871 continue; 872 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) 873 continue; 874 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6)) 875 continue; 876 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 877 continue; 878 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 879 continue; 880 if ((IN6_IS_ADDR_V4MAPPED(laddr6) || 881 IN6_IS_ADDR_V4MAPPED(faddr6)) && 882 (in6p->in6p_flags & IN6P_IPV6_V6ONLY)) 883 continue; 884 return in6p; 885 } 886 return NULL; 887 } 888 889 struct in6pcb * 890 in6_pcblookup_bind(head, laddr6, lport_arg, faith) 891 struct in6pcb *head; 892 struct in6_addr *laddr6; 893 u_int lport_arg; 894 int faith; 895 { 896 struct in6pcb *in6p, *match; 897 u_int16_t lport = lport_arg; 898 899 match = NULL; 900 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) { 901 /* 902 * find destination match. exact match is preferred 903 * against wildcard match. 904 */ 905 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0) 906 continue; 907 if (in6p->in6p_fport != 0) 908 continue; 909 if (in6p->in6p_lport != lport) 910 continue; 911 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { 912 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 913 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY)) 914 continue; 915 else 916 match = in6p; 917 } else 918 match = in6p; 919 } 920 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) && 921 !(in6p->in6p_flags & IN6P_IPV6_V6ONLY) && 922 in6p->in6p_laddr.s6_addr32[3] == 0) { 923 if (IN6_IS_ADDR_V4MAPPED(laddr6) && 924 laddr6->s6_addr32[3] != 0) 925 match = in6p; 926 } 927 else if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 928 return in6p; 929 } 930 return match; 931 } 932
Indexes created Mon Sep 22 05:09:51 GMT 2025