in6_pcb.c revision 1.42
1 /* $NetBSD: in6_pcb.c,v 1.42 2001/10/16 04:57:38 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. All advertising materials mentioning features or use of this software 46 * must display the following acknowledgement: 47 * This product includes software developed by the University of 48 * California, Berkeley and its contributors. 49 * 4. Neither the name of the University nor the names of its contributors 50 * may be used to endorse or promote products derived from this software 51 * without specific prior written permission. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 63 * SUCH DAMAGE. 64 * 65 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 66 */ 67 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 #ifdef IPSEC 118 int error; 119 #endif 120 121 MALLOC(in6p, struct in6pcb *, sizeof(*in6p), M_PCB, M_NOWAIT); 122 if (in6p == NULL) 123 return(ENOBUFS); 124 bzero((caddr_t)in6p, sizeof(*in6p)); 125 in6p->in6p_head = head; 126 in6p->in6p_socket = so; 127 in6p->in6p_hops = -1; /* use kernel default */ 128 in6p->in6p_icmp6filt = NULL; 129 #ifdef IPSEC 130 error = ipsec_init_policy(so, &in6p->in6p_sp); 131 if (error != 0) { 132 FREE(in6p, M_PCB); 133 return error; 134 } 135 #endif /*IPSEC*/ 136 in6p->in6p_next = head->in6p_next; 137 head->in6p_next = in6p; 138 in6p->in6p_prev = head; 139 in6p->in6p_next->in6p_prev = in6p; 140 if (ip6_v6only) 141 in6p->in6p_flags |= IN6P_IPV6_V6ONLY; 142 so->so_pcb = (caddr_t)in6p; 143 return(0); 144 } 145 146 int 147 in6_pcbbind(in6p, nam, p) 148 struct in6pcb *in6p; 149 struct mbuf *nam; 150 struct proc *p; 151 { 152 struct socket *so = in6p->in6p_socket; 153 struct in6pcb *head = in6p->in6p_head; 154 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 155 u_int16_t lport = 0; 156 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT); 157 158 if (in6p->in6p_lport || !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 159 return(EINVAL); 160 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 && 161 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 || 162 (so->so_options & SO_ACCEPTCONN) == 0)) 163 wild = IN6PLOOKUP_WILDCARD; 164 if (nam) { 165 sin6 = mtod(nam, struct sockaddr_in6 *); 166 if (nam->m_len != sizeof(*sin6)) 167 return(EINVAL); 168 /* 169 * We should check the family, but old programs 170 * incorrectly fail to intialize it. 171 */ 172 if (sin6->sin6_family != AF_INET6) 173 return(EAFNOSUPPORT); 174 175 /* 176 * since we do not check port number duplicate with IPv4 space, 177 * we reject it at this moment. If we leave it, we would 178 * mistakenly allow normal users to hijack tcp/udp ports from 179 * other users. 180 */ 181 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) 182 return(EADDRNOTAVAIL); 183 184 /* KAME hack: embed scopeid */ 185 if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, NULL) != 0) 186 return EINVAL; 187 /* this must be cleared for ifa_ifwithaddr() */ 188 sin6->sin6_scope_id = 0; 189 190 lport = sin6->sin6_port; 191 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 192 /* 193 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 194 * allow compepte duplication of binding if 195 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 196 * and a multicast address is bound on both 197 * new and duplicated sockets. 198 */ 199 if (so->so_options & SO_REUSEADDR) 200 reuseport = SO_REUSEADDR|SO_REUSEPORT; 201 } 202 else if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 203 struct sockaddr_in sin; 204 205 bzero(&sin, sizeof(sin)); 206 sin.sin_len = sizeof(sin); 207 sin.sin_family = AF_INET; 208 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr, 209 sizeof(sin.sin_addr)); 210 if (ifa_ifwithaddr((struct sockaddr *)&sin) == 0) 211 return EADDRNOTAVAIL; 212 } 213 else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 214 struct ifaddr *ia = NULL; 215 216 sin6->sin6_port = 0; /* yech... */ 217 if ((in6p->in6p_flags & IN6P_FAITH) == 0 && 218 (ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0) 219 return(EADDRNOTAVAIL); 220 221 /* 222 * XXX: bind to an anycast address might accidentally 223 * cause sending a packet with anycast source address. 224 */ 225 if (ia && 226 ((struct in6_ifaddr *)ia)->ia6_flags & 227 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY| 228 IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) { 229 return(EADDRNOTAVAIL); 230 } 231 } 232 if (lport) { 233 #ifndef IPNOPRIVPORTS 234 int priv; 235 236 /* 237 * NOTE: all operating systems use suser() for 238 * privilege check! do not rewrite it into SS_PRIV. 239 */ 240 priv = (p && !suser(p->p_ucred, &p->p_acflag)) ? 1 : 0; 241 /* GROSS */ 242 if (ntohs(lport) < IPV6PORT_RESERVED && !priv) 243 return(EACCES); 244 #endif 245 246 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 247 /* should check this but we can't ... */ 248 #if 0 249 struct inpcb *t; 250 251 t = in_pcblookup_bind(&tcbtable, 252 (struct in_addr *)&sin6->sin6_addr.s6_addr32[3], 253 lport); 254 if (t && (reuseport & t->inp_socket->so_options) == 0) 255 return EADDRINUSE; 256 #endif 257 } else 258 { 259 struct in6pcb *t; 260 261 t = in6_pcblookup(head, &zeroin6_addr, 0, 262 &sin6->sin6_addr, lport, wild); 263 if (t && (reuseport & t->in6p_socket->so_options) == 0) 264 return(EADDRINUSE); 265 } 266 } 267 in6p->in6p_laddr = sin6->sin6_addr; 268 } 269 270 if (lport == 0) { 271 int e; 272 if ((e = in6_pcbsetport(&in6p->in6p_laddr, in6p)) != 0) 273 return(e); 274 } 275 else 276 in6p->in6p_lport = lport; 277 278 in6p->in6p_flowinfo = sin6 ? sin6->sin6_flowinfo : 0; /*XXX*/ 279 return(0); 280 } 281 282 /* 283 * Connect from a socket to a specified address. 284 * Both address and port must be specified in argument sin6. 285 * If don't have a local address for this socket yet, 286 * then pick one. 287 */ 288 int 289 in6_pcbconnect(in6p, nam) 290 struct in6pcb *in6p; 291 struct mbuf *nam; 292 { 293 struct in6_addr *in6a = NULL; 294 struct sockaddr_in6 *sin6 = mtod(nam, struct sockaddr_in6 *); 295 struct ifnet *ifp = NULL; /* outgoing interface */ 296 int error = 0; 297 #ifdef INET 298 struct in6_addr mapped; 299 #endif 300 struct sockaddr_in6 tmp; 301 302 (void)&in6a; /* XXX fool gcc */ 303 304 if (nam->m_len != sizeof(*sin6)) 305 return(EINVAL); 306 if (sin6->sin6_family != AF_INET6) 307 return(EAFNOSUPPORT); 308 if (sin6->sin6_port == 0) 309 return(EADDRNOTAVAIL); 310 311 /* sanity check for mapped address case */ 312 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { 313 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0) 314 return EINVAL; 315 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 316 in6p->in6p_laddr.s6_addr16[5] = htons(0xffff); 317 if (!IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 318 return EINVAL; 319 } else 320 { 321 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 322 return EINVAL; 323 } 324 325 /* protect *sin6 from overwrites */ 326 tmp = *sin6; 327 sin6 = &tmp; 328 329 /* KAME hack: embed scopeid */ 330 if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, &ifp) != 0) 331 return EINVAL; 332 333 /* Source address selection. */ 334 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 335 && in6p->in6p_laddr.s6_addr32[3] == 0) { 336 #ifdef INET 337 struct sockaddr_in sin, *sinp; 338 339 bzero(&sin, sizeof(sin)); 340 sin.sin_len = sizeof(sin); 341 sin.sin_family = AF_INET; 342 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr, 343 sizeof(sin.sin_addr)); 344 sinp = in_selectsrc(&sin, (struct route *)&in6p->in6p_route, 345 in6p->in6p_socket->so_options, NULL, &error); 346 if (sinp == 0) { 347 if (error == 0) 348 error = EADDRNOTAVAIL; 349 return(error); 350 } 351 bzero(&mapped, sizeof(mapped)); 352 mapped.s6_addr16[5] = htons(0xffff); 353 bcopy(&sinp->sin_addr, &mapped.s6_addr32[3], sizeof(sinp->sin_addr)); 354 in6a = &mapped; 355 #else 356 return EADDRNOTAVAIL; 357 #endif 358 } else 359 { 360 /* 361 * XXX: in6_selectsrc might replace the bound local address 362 * with the address specified by setsockopt(IPV6_PKTINFO). 363 * Is it the intended behavior? 364 */ 365 in6a = in6_selectsrc(sin6, in6p->in6p_outputopts, 366 in6p->in6p_moptions, 367 &in6p->in6p_route, 368 &in6p->in6p_laddr, &error); 369 if (in6a == 0) { 370 if (error == 0) 371 error = EADDRNOTAVAIL; 372 return(error); 373 } 374 } 375 if (in6p->in6p_route.ro_rt) 376 ifp = in6p->in6p_route.ro_rt->rt_ifp; 377 378 in6p->in6p_ip6.ip6_hlim = (u_int8_t)in6_selecthlim(in6p, ifp); 379 380 if (in6_pcblookup(in6p->in6p_head, 381 &sin6->sin6_addr, 382 sin6->sin6_port, 383 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ? 384 in6a : &in6p->in6p_laddr, 385 in6p->in6p_lport, 386 0)) 387 return(EADDRINUSE); 388 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) 389 || (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 390 && in6p->in6p_laddr.s6_addr32[3] == 0)) 391 { 392 /* 393 * XXX in IPv4 mapped address case, we should grab fresh 394 * local port number by in_pcbbind, not in6_pcbbind. 395 * if we are in bad luck, we may assign conflicting port number 396 * between IPv4 and IPv6 unwillingly. 397 */ 398 if (in6p->in6p_lport == 0) { 399 (void)in6_pcbbind(in6p, (struct mbuf *)0, 400 (struct proc *)0); 401 } 402 in6p->in6p_laddr = *in6a; 403 } 404 in6p->in6p_faddr = sin6->sin6_addr; 405 in6p->in6p_fport = sin6->sin6_port; 406 /* 407 * xxx kazu flowlabel is necessary for connect? 408 * but if this line is missing, the garbage value remains. 409 */ 410 in6p->in6p_flowinfo = sin6->sin6_flowinfo; 411 #ifdef IPSEC 412 if (in6p->in6p_socket->so_type == SOCK_STREAM) 413 ipsec_pcbconn(in6p->in6p_sp); 414 #endif 415 return(0); 416 } 417 418 void 419 in6_pcbdisconnect(in6p) 420 struct in6pcb *in6p; 421 { 422 bzero((caddr_t)&in6p->in6p_faddr, sizeof(in6p->in6p_faddr)); 423 in6p->in6p_fport = 0; 424 if (in6p->in6p_socket->so_state & SS_NOFDREF) 425 in6_pcbdetach(in6p); 426 #ifdef IPSEC 427 ipsec_pcbdisconn(in6p->in6p_sp); 428 #endif 429 } 430 431 void 432 in6_pcbdetach(in6p) 433 struct in6pcb *in6p; 434 { 435 struct socket *so = in6p->in6p_socket; 436 437 #ifdef IPSEC 438 ipsec6_delete_pcbpolicy(in6p); 439 #endif /* IPSEC */ 440 sotoin6pcb(so) = 0; 441 sofree(so); 442 if (in6p->in6p_options) 443 m_freem(in6p->in6p_options); 444 if (in6p->in6p_outputopts) { 445 if (in6p->in6p_outputopts->ip6po_rthdr && 446 in6p->in6p_outputopts->ip6po_route.ro_rt) 447 RTFREE(in6p->in6p_outputopts->ip6po_route.ro_rt); 448 if (in6p->in6p_outputopts->ip6po_m) 449 (void)m_free(in6p->in6p_outputopts->ip6po_m); 450 free(in6p->in6p_outputopts, M_IP6OPT); 451 } 452 if (in6p->in6p_route.ro_rt) 453 rtfree(in6p->in6p_route.ro_rt); 454 ip6_freemoptions(in6p->in6p_moptions); 455 in6p->in6p_next->in6p_prev = in6p->in6p_prev; 456 in6p->in6p_prev->in6p_next = in6p->in6p_next; 457 in6p->in6p_prev = NULL; 458 FREE(in6p, M_PCB); 459 } 460 461 void 462 in6_setsockaddr(in6p, nam) 463 struct in6pcb *in6p; 464 struct mbuf *nam; 465 { 466 struct sockaddr_in6 *sin6; 467 468 nam->m_len = sizeof(*sin6); 469 sin6 = mtod(nam, struct sockaddr_in6 *); 470 bzero((caddr_t)sin6, sizeof(*sin6)); 471 sin6->sin6_family = AF_INET6; 472 sin6->sin6_len = sizeof(struct sockaddr_in6); 473 sin6->sin6_port = in6p->in6p_lport; 474 /* KAME hack: recover scopeid */ 475 (void)in6_recoverscope(sin6, &in6p->in6p_laddr, NULL); 476 } 477 478 void 479 in6_setpeeraddr(in6p, nam) 480 struct in6pcb *in6p; 481 struct mbuf *nam; 482 { 483 struct sockaddr_in6 *sin6; 484 485 nam->m_len = sizeof(*sin6); 486 sin6 = mtod(nam, struct sockaddr_in6 *); 487 bzero((caddr_t)sin6, sizeof(*sin6)); 488 sin6->sin6_family = AF_INET6; 489 sin6->sin6_len = sizeof(struct sockaddr_in6); 490 sin6->sin6_port = in6p->in6p_fport; 491 /* KAME hack: recover scopeid */ 492 (void)in6_recoverscope(sin6, &in6p->in6p_faddr, NULL); 493 } 494 495 /* 496 * Pass some notification to all connections of a protocol 497 * associated with address dst. The local address and/or port numbers 498 * may be specified to limit the search. The "usual action" will be 499 * taken, depending on the ctlinput cmd. The caller must filter any 500 * cmds that are uninteresting (e.g., no error in the map). 501 * Call the protocol specific routine (if any) to report 502 * any errors for each matching socket. 503 * 504 * Must be called at splsoftnet. 505 * 506 * Note: src (4th arg) carries the flowlabel value on the original IPv6 507 * header, in sin6_flowinfo member. 508 */ 509 int 510 in6_pcbnotify(head, dst, fport_arg, src, lport_arg, cmd, cmdarg, notify) 511 struct in6pcb *head; 512 struct sockaddr *dst, *src; 513 u_int fport_arg, lport_arg; 514 int cmd; 515 void *cmdarg; 516 void (*notify) __P((struct in6pcb *, int)); 517 { 518 struct in6pcb *in6p, *nin6p; 519 struct sockaddr_in6 sa6_src, *sa6_dst; 520 u_int16_t fport = fport_arg, lport = lport_arg; 521 int errno; 522 int nmatch = 0; 523 u_int32_t flowinfo; 524 525 if ((unsigned)cmd > PRC_NCMDS || dst->sa_family != AF_INET6) 526 return 0; 527 528 sa6_dst = (struct sockaddr_in6 *)dst; 529 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 530 return 0; 531 532 /* 533 * note that src can be NULL when we get notify by local fragmentation. 534 */ 535 sa6_src = (src == NULL) ? sa6_any : *(struct sockaddr_in6 *)src; 536 flowinfo = sa6_src.sin6_flowinfo; 537 538 /* 539 * Redirects go to all references to the destination, 540 * and use in6_rtchange to invalidate the route cache. 541 * Dead host indications: also use in6_rtchange to invalidate 542 * the cache, and deliver the error to all the sockets. 543 * Otherwise, if we have knowledge of the local port and address, 544 * deliver only to that socket. 545 */ 546 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) { 547 fport = 0; 548 lport = 0; 549 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr)); 550 551 if (cmd != PRC_HOSTDEAD) 552 notify = in6_rtchange; 553 } 554 555 errno = inet6ctlerrmap[cmd]; 556 for (in6p = head->in6p_next; in6p != head; in6p = nin6p) { 557 nin6p = in6p->in6p_next; 558 559 /* 560 * Under the following condition, notify of redirects 561 * to the pcb, without making address matches against inpcb. 562 * - redirect notification is arrived. 563 * - the inpcb is unconnected. 564 * - the inpcb is caching !RTF_HOST routing entry. 565 * - the ICMPv6 notification is from the gateway cached in the 566 * inpcb. i.e. ICMPv6 notification is from nexthop gateway 567 * the inpcb used very recently. 568 * 569 * This is to improve interaction between netbsd/openbsd 570 * redirect handling code, and inpcb route cache code. 571 * without the clause, !RTF_HOST routing entry (which carries 572 * gateway used by inpcb right before the ICMPv6 redirect) 573 * will be cached forever in unconnected inpcb. 574 * 575 * There still is a question regarding to what is TRT: 576 * - On bsdi/freebsd, RTF_HOST (cloned) routing entry will be 577 * generated on packet output. inpcb will always cache 578 * RTF_HOST routing entry so there's no need for the clause 579 * (ICMPv6 redirect will update RTF_HOST routing entry, 580 * and inpcb is caching it already). 581 * However, bsdi/freebsd are vulnerable to local DoS attacks 582 * due to the cloned routing entries. 583 * - Specwise, "destination cache" is mentioned in RFC2461. 584 * Jinmei says that it implies bsdi/freebsd behavior, itojun 585 * is not really convinced. 586 * - Having hiwat/lowat on # of cloned host route (redirect/ 587 * pmtud) may be a good idea. netbsd/openbsd has it. see 588 * icmp6_mtudisc_update(). 589 */ 590 if ((PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) && 591 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) && 592 in6p->in6p_route.ro_rt && 593 !(in6p->in6p_route.ro_rt->rt_flags & RTF_HOST)) { 594 struct sockaddr_in6 *dst6; 595 596 dst6 = (struct sockaddr_in6 *)&in6p->in6p_route.ro_dst; 597 if (IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr, 598 &sa6_dst->sin6_addr)) 599 goto do_notify; 600 } 601 602 /* 603 * Detect if we should notify the error. If no source and 604 * destination ports are specified, but non-zero flowinfo and 605 * local address match, notify the error. This is the case 606 * when the error is delivered with an encrypted buffer 607 * by ESP. Otherwise, just compare addresses and ports 608 * as usual. 609 */ 610 if (lport == 0 && fport == 0 && flowinfo && 611 in6p->in6p_socket != NULL && 612 flowinfo == (in6p->in6p_flowinfo & IPV6_FLOWLABEL_MASK) && 613 IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &sa6_src.sin6_addr)) 614 goto do_notify; 615 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, 616 &sa6_dst->sin6_addr) || 617 in6p->in6p_socket == 0 || 618 (lport && in6p->in6p_lport != lport) || 619 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 620 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, 621 &sa6_src.sin6_addr)) || 622 (fport && in6p->in6p_fport != fport)) 623 continue; 624 625 do_notify: 626 if (notify) 627 (*notify)(in6p, errno); 628 nmatch++; 629 } 630 return nmatch; 631 } 632 633 void 634 in6_pcbpurgeif0(head, ifp) 635 struct in6pcb *head; 636 struct ifnet *ifp; 637 { 638 struct in6pcb *in6p, *nin6p; 639 struct ip6_moptions *im6o; 640 struct in6_multi_mship *imm, *nimm; 641 642 for (in6p = head->in6p_next; in6p != head; in6p = nin6p) { 643 nin6p = in6p->in6p_next; 644 im6o = in6p->in6p_moptions; 645 if (im6o) { 646 /* 647 * Unselect the outgoing interface if it is being 648 * detached. 649 */ 650 if (im6o->im6o_multicast_ifp == ifp) 651 im6o->im6o_multicast_ifp = NULL; 652 653 /* 654 * Drop multicast group membership if we joined 655 * through the interface being detached. 656 * XXX controversial - is it really legal for kernel 657 * to force this? 658 */ 659 for (imm = im6o->im6o_memberships.lh_first; 660 imm != NULL; imm = nimm) { 661 nimm = imm->i6mm_chain.le_next; 662 if (imm->i6mm_maddr->in6m_ifp == ifp) { 663 LIST_REMOVE(imm, i6mm_chain); 664 in6_delmulti(imm->i6mm_maddr); 665 free(imm, M_IPMADDR); 666 } 667 } 668 } 669 } 670 } 671 672 void 673 in6_pcbpurgeif(head, ifp) 674 struct in6pcb *head; 675 struct ifnet *ifp; 676 { 677 struct in6pcb *in6p, *nin6p; 678 679 for (in6p = head->in6p_next; in6p != head; in6p = nin6p) { 680 nin6p = in6p->in6p_next; 681 if (in6p->in6p_route.ro_rt != NULL && 682 in6p->in6p_route.ro_rt->rt_ifp == ifp) 683 in6_rtchange(in6p, 0); 684 } 685 } 686 687 /* 688 * Check for alternatives when higher level complains 689 * about service problems. For now, invalidate cached 690 * routing information. If the route was created dynamically 691 * (by a redirect), time to try a default gateway again. 692 */ 693 void 694 in6_losing(in6p) 695 struct in6pcb *in6p; 696 { 697 struct rtentry *rt; 698 struct rt_addrinfo info; 699 700 if ((rt = in6p->in6p_route.ro_rt) != NULL) { 701 in6p->in6p_route.ro_rt = 0; 702 bzero((caddr_t)&info, sizeof(info)); 703 info.rti_info[RTAX_DST] = 704 (struct sockaddr *)&in6p->in6p_route.ro_dst; 705 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 706 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 707 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0); 708 if (rt->rt_flags & RTF_DYNAMIC) { 709 (void)rtrequest(RTM_DELETE, rt_key(rt), 710 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 711 (struct rtentry **)0); 712 } else { 713 /* 714 * A new route can be allocated 715 * the next time output is attempted. 716 */ 717 rtfree(rt); 718 } 719 } 720 } 721 722 /* 723 * After a routing change, flush old routing 724 * and allocate a (hopefully) better one. 725 */ 726 void 727 in6_rtchange(in6p, errno) 728 struct in6pcb *in6p; 729 int errno; 730 { 731 if (in6p->in6p_route.ro_rt) { 732 rtfree(in6p->in6p_route.ro_rt); 733 in6p->in6p_route.ro_rt = 0; 734 /* 735 * A new route can be allocated the next time 736 * output is attempted. 737 */ 738 } 739 } 740 741 struct in6pcb * 742 in6_pcblookup(head, faddr6, fport_arg, laddr6, lport_arg, flags) 743 struct in6pcb *head; 744 struct in6_addr *faddr6, *laddr6; 745 u_int fport_arg, lport_arg; 746 int flags; 747 { 748 struct in6pcb *in6p, *match = 0; 749 int matchwild = 3, wildcard; 750 u_int16_t fport = fport_arg, lport = lport_arg; 751 752 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) { 753 if (in6p->in6p_lport != lport) 754 continue; 755 wildcard = 0; 756 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { 757 if (IN6_IS_ADDR_UNSPECIFIED(laddr6)) 758 wildcard++; 759 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 760 continue; 761 } 762 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 763 && in6p->in6p_laddr.s6_addr32[3] == 0) { 764 if (!IN6_IS_ADDR_V4MAPPED(laddr6)) 765 continue; 766 if (laddr6->s6_addr32[3] == 0) 767 ; 768 else 769 wildcard++; 770 } 771 else { 772 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 773 if (in6p->in6p_flags & IN6P_IPV6_V6ONLY) 774 continue; 775 else 776 wildcard++; 777 } else if (!IN6_IS_ADDR_UNSPECIFIED(laddr6)) 778 wildcard++; 779 } 780 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) { 781 if (IN6_IS_ADDR_UNSPECIFIED(faddr6)) 782 wildcard++; 783 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6) 784 || in6p->in6p_fport != fport) 785 continue; 786 } 787 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr) 788 && in6p->in6p_faddr.s6_addr32[3] == 0) { 789 if (!IN6_IS_ADDR_V4MAPPED(faddr6)) 790 continue; 791 if (faddr6->s6_addr32[3] == 0) 792 ; 793 else 794 wildcard++; 795 } 796 else { 797 if (IN6_IS_ADDR_V4MAPPED(faddr6)) { 798 if (in6p->in6p_flags & IN6P_IPV6_V6ONLY) 799 continue; 800 else 801 wildcard++; 802 } else if (!IN6_IS_ADDR_UNSPECIFIED(faddr6)) 803 wildcard++; 804 } 805 806 if (wildcard && (flags & IN6PLOOKUP_WILDCARD) == 0) 807 continue; 808 if (wildcard < matchwild) { 809 match = in6p; 810 matchwild = wildcard; 811 if (matchwild == 0) 812 break; 813 } 814 } 815 return(match); 816 } 817 818 struct rtentry * 819 in6_pcbrtentry(in6p) 820 struct in6pcb *in6p; 821 { 822 struct route_in6 *ro; 823 struct sockaddr_in6 *dst6; 824 825 ro = &in6p->in6p_route; 826 dst6 = (struct sockaddr_in6 *)&ro->ro_dst; 827 828 if (ro->ro_rt == NULL) { 829 /* 830 * No route yet, so try to acquire one. 831 */ 832 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) { 833 bzero(dst6, sizeof(*dst6)); 834 dst6->sin6_family = AF_INET6; 835 dst6->sin6_len = sizeof(struct sockaddr_in6); 836 dst6->sin6_addr = in6p->in6p_faddr; 837 rtalloc((struct route *)ro); 838 } 839 } 840 return (ro->ro_rt); 841 } 842 843 struct in6pcb * 844 in6_pcblookup_connect(head, faddr6, fport_arg, laddr6, lport_arg, faith) 845 struct in6pcb *head; 846 struct in6_addr *faddr6, *laddr6; 847 u_int fport_arg, lport_arg; 848 int faith; 849 { 850 struct in6pcb *in6p; 851 u_int16_t fport = fport_arg, lport = lport_arg; 852 853 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) { 854 /* find exact match on both source and dest */ 855 if (in6p->in6p_fport != fport) 856 continue; 857 if (in6p->in6p_lport != lport) 858 continue; 859 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) 860 continue; 861 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6)) 862 continue; 863 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) 864 continue; 865 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 866 continue; 867 if ((IN6_IS_ADDR_V4MAPPED(laddr6) || 868 IN6_IS_ADDR_V4MAPPED(faddr6)) && 869 (in6p->in6p_flags & IN6P_IPV6_V6ONLY)) 870 continue; 871 return in6p; 872 } 873 return NULL; 874 } 875 876 struct in6pcb * 877 in6_pcblookup_bind(head, laddr6, lport_arg, faith) 878 struct in6pcb *head; 879 struct in6_addr *laddr6; 880 u_int lport_arg; 881 int faith; 882 { 883 struct in6pcb *in6p, *match; 884 u_int16_t lport = lport_arg; 885 886 match = NULL; 887 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) { 888 /* 889 * find destination match. exact match is preferred 890 * against wildcard match. 891 */ 892 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0) 893 continue; 894 if (in6p->in6p_fport != 0) 895 continue; 896 if (in6p->in6p_lport != lport) 897 continue; 898 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { 899 if (IN6_IS_ADDR_V4MAPPED(laddr6)) { 900 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY)) 901 continue; 902 else 903 match = in6p; 904 } else 905 match = in6p; 906 } 907 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) && 908 !(in6p->in6p_flags & IN6P_IPV6_V6ONLY) && 909 in6p->in6p_laddr.s6_addr32[3] == 0) { 910 if (IN6_IS_ADDR_V4MAPPED(laddr6) && 911 laddr6->s6_addr32[3] != 0) 912 match = in6p; 913 } 914 else if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) 915 return in6p; 916 } 917 return match; 918 } 919
Indexes created Wed Sep 24 05:09:52 GMT 2025