in6_src.c revision 1.23.2.3
1 /* $KAME: in6_src.c,v 1.159 2005/10/19 01:40:32 t-momose Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /* 33 * Copyright (c) 1982, 1986, 1991, 1993 34 * The Regents of the University of California. All rights reserved. 35 * 36 * Redistribution and use in source and binary forms, with or without 37 * modification, are permitted provided that the following conditions 38 * are met: 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 3. All advertising materials mentioning features or use of this software 45 * must display the following acknowledgement: 46 * This product includes software developed by the University of 47 * California, Berkeley and its contributors. 48 * 4. Neither the name of the University nor the names of its contributors 49 * may be used to endorse or promote products derived from this software 50 * without specific prior written permission. 51 * 52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 55 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 62 * SUCH DAMAGE. 63 * 64 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 65 */ 66 67 #include <sys/cdefs.h> 68 __KERNEL_RCSID(0, "$NetBSD: in6_src.c,v 1.23.2.3 2006/02/23 16:11:13 rpaulo Exp $"); 69 70 #include "opt_inet.h" 71 72 #include <sys/param.h> 73 #include <sys/systm.h> 74 #include <sys/malloc.h> 75 #include <sys/mbuf.h> 76 #include <sys/protosw.h> 77 #include <sys/socket.h> 78 #include <sys/socketvar.h> 79 #ifndef __FreeBSD__ 80 #include <sys/ioctl.h> 81 #else 82 #include <sys/sockio.h> 83 #endif 84 #ifdef __FreeBSD__ 85 #include <sys/sysctl.h> 86 #endif 87 #include <sys/errno.h> 88 #include <sys/time.h> 89 #include <sys/kernel.h> 90 #include <sys/proc.h> 91 92 #include <net/if.h> 93 #include <net/if_types.h> 94 #include <net/route.h> 95 #ifdef RADIX_MPATH 96 #include <net/radix_mpath.h> 97 #endif 98 99 #include <netinet/in.h> 100 #include <netinet/in_var.h> 101 #include <netinet/in_systm.h> 102 #include <netinet/ip.h> 103 #include <netinet/in_pcb.h> 104 #include <netinet6/in6_var.h> 105 #include <netinet/ip6.h> 106 #include <netinet6/ip6_var.h> 107 #include <netinet6/nd6.h> 108 #include <netinet6/scope6_var.h> 109 110 #include <net/net_osdep.h> 111 112 #ifdef MIP6 113 #include <netinet6/mip6.h> 114 #include <netinet6/mip6_var.h> 115 #include "mip.h" 116 #if NMIP > 0 117 #include <net/if_mip.h> 118 #endif /* NMIP > 0 */ 119 #endif /* MIP6 */ 120 121 #ifndef __OpenBSD__ 122 #include "loop.h" 123 #endif 124 #ifdef __NetBSD__ 125 extern struct ifnet loif[NLOOP]; 126 #endif 127 128 #define ADDR_LABEL_NOTAPP (-1) 129 struct in6_addrpolicy defaultaddrpolicy; 130 131 #ifdef notyet /* until introducing ND extensions and address selection */ 132 int ip6_prefer_tempaddr = 0; 133 #endif 134 135 static int selectroute __P((struct sockaddr_in6 *, struct ip6_pktopts *, 136 struct ip6_moptions *, struct route_in6 *, struct ifnet **, 137 struct rtentry **, int, int)); 138 static int in6_selectif __P((struct sockaddr_in6 *, struct ip6_pktopts *, 139 struct ip6_moptions *, struct route_in6 *, struct ifnet **)); 140 141 static struct in6_addrpolicy *lookup_addrsel_policy __P((struct sockaddr_in6 *)); 142 143 static void init_policy_queue __P((void)); 144 static int add_addrsel_policyent __P((struct in6_addrpolicy *)); 145 static int delete_addrsel_policyent __P((struct in6_addrpolicy *)); 146 static int walk_addrsel_policy __P((int (*)(struct in6_addrpolicy *, void *), 147 void *)); 148 static int dump_addrsel_policyent __P((struct in6_addrpolicy *, void *)); 149 static struct in6_addrpolicy *match_addrsel_policy __P((struct sockaddr_in6 *)); 150 151 /* 152 * Return an IPv6 address, which is the most appropriate for a given 153 * destination and user specified options. 154 * If necessary, this function lookups the routing table and returns 155 * an entry to the caller for later use. 156 */ 157 #if 0 /* diabled ad-hoc */ 158 #define REPLACE(r) do {\ 159 if ((r) < sizeof(ip6stat.ip6s_sources_rule) / \ 160 sizeof(ip6stat.ip6s_sources_rule[0])) /* check for safety */ \ 161 ip6stat.ip6s_sources_rule[(r)]++; \ 162 /* printf("in6_selectsrc: replace %s with %s by %d\n", ia_best ? ip6_sprintf(&ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(&ia->ia_addr.sin6_addr), (r)); */ \ 163 goto replace; \ 164 } while(0) 165 #define NEXT(r) do {\ 166 if ((r) < sizeof(ip6stat.ip6s_sources_rule) / \ 167 sizeof(ip6stat.ip6s_sources_rule[0])) /* check for safety */ \ 168 ip6stat.ip6s_sources_rule[(r)]++; \ 169 /* printf("in6_selectsrc: keep %s against %s by %d\n", ia_best ? ip6_sprintf(&ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(&ia->ia_addr.sin6_addr), (r)); */ \ 170 goto next; /* XXX: we can't use 'continue' here */ \ 171 } while(0) 172 #define BREAK(r) do { \ 173 if ((r) < sizeof(ip6stat.ip6s_sources_rule) / \ 174 sizeof(ip6stat.ip6s_sources_rule[0])) /* check for safety */ \ 175 ip6stat.ip6s_sources_rule[(r)]++; \ 176 goto out; /* XXX: we can't use 'break' here */ \ 177 } while(0) 178 #else 179 #define REPLACE(r) goto replace 180 #define NEXT(r) goto next 181 #define BREAK(r) goto out 182 #endif 183 184 struct in6_addr * 185 in6_selectsrc(dstsock, opts, mopts, ro, laddr, ifpp, errorp) 186 struct sockaddr_in6 *dstsock; 187 struct ip6_pktopts *opts; 188 struct ip6_moptions *mopts; 189 struct route_in6 *ro; 190 struct in6_addr *laddr; 191 struct ifnet **ifpp; 192 int *errorp; 193 { 194 struct in6_addr dst; 195 struct ifnet *ifp = NULL; 196 struct in6_ifaddr *ia = NULL, *ia_best = NULL; 197 struct in6_pktinfo *pi = NULL; 198 int dst_scope = -1, best_scope = -1, best_matchlen = -1; 199 struct in6_addrpolicy *dst_policy = NULL, *best_policy = NULL; 200 u_int32_t odstzone; 201 #ifdef notyet /* until introducing ND extensions and address selection */ 202 int prefer_tempaddr; 203 #endif 204 #if defined(MIP6) && NMIP > 0 205 u_int8_t ip6po_usecoa = 0; 206 #endif /* MIP6 && NMIP > 0 */ 207 208 dst = dstsock->sin6_addr; /* make a copy for local operation */ 209 *errorp = 0; 210 if (ifpp) 211 *ifpp = NULL; 212 213 /* 214 * If the source address is explicitly specified by the caller, 215 * check if the requested source address is indeed a unicast address 216 * assigned to the node, and can be used as the packet's source 217 * address. If everything is okay, use the address as source. 218 */ 219 if (opts && (pi = opts->ip6po_pktinfo) && 220 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) { 221 struct sockaddr_in6 srcsock; 222 struct in6_ifaddr *ia6; 223 224 /* get the outgoing interface */ 225 if ((*errorp = in6_selectif(dstsock, opts, mopts, ro, &ifp)) 226 != 0) { 227 return (NULL); 228 } 229 230 /* 231 * Determine the appropriate zone id of the source based on 232 * the zone of the destination and the outgoing interface. 233 * If the specified address is ambiguous wrt the scope zone, 234 * the interface must be specified; otherwise, ifa_ifwithaddr() 235 * will fail matching the address. 236 */ 237 bzero(&srcsock, sizeof(srcsock)); 238 srcsock.sin6_family = AF_INET6; 239 srcsock.sin6_len = sizeof(srcsock); 240 srcsock.sin6_addr = pi->ipi6_addr; 241 if (ifp) { 242 *errorp = in6_setscope(&srcsock.sin6_addr, ifp, NULL); 243 if (*errorp != 0) 244 return (NULL); 245 } 246 247 ia6 = (struct in6_ifaddr *)ifa_ifwithaddr((struct sockaddr *)(&srcsock)); 248 if (ia6 == NULL || 249 (ia6->ia6_flags & (IN6_IFF_ANYCAST | IN6_IFF_NOTREADY))) { 250 *errorp = EADDRNOTAVAIL; 251 return (NULL); 252 } 253 pi->ipi6_addr = srcsock.sin6_addr; /* XXX: this overrides pi */ 254 if (ifpp) 255 *ifpp = ifp; 256 return (&ia6->ia_addr.sin6_addr); 257 } 258 259 /* 260 * Otherwise, if the socket has already bound the source, just use it. 261 */ 262 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) 263 return (laddr); 264 265 /* 266 * If the address is not specified, choose the best one based on 267 * the outgoing interface and the destination address. 268 */ 269 /* get the outgoing interface */ 270 if ((*errorp = in6_selectif(dstsock, opts, mopts, ro, &ifp)) != 0) 271 return (NULL); 272 273 #if defined(MIP6) && NMIP > 0 274 /* 275 * a caller can specify IP6PO_USECOA to not to use a home 276 * address. for example, the case that the neighbour 277 * unreachability detection to the global address. 278 */ 279 if (opts != NULL && 280 (opts->ip6po_flags & IP6PO_USECOA) != 0) { 281 ip6po_usecoa = 1; 282 } 283 #endif /* MIP6 && NMIP > 0 */ 284 285 #ifdef DIAGNOSTIC 286 if (ifp == NULL) /* this should not happen */ 287 panic("in6_selectsrc: NULL ifp"); 288 #endif 289 *errorp = in6_setscope(&dst, ifp, &odstzone); 290 if (*errorp != 0) 291 return (NULL); 292 293 for (ia = in6_ifaddr; ia; ia = ia->ia_next) { 294 int new_scope = -1, new_matchlen = -1; 295 struct in6_addrpolicy *new_policy = NULL; 296 u_int32_t srczone, osrczone, dstzone; 297 struct in6_addr src; 298 struct ifnet *ifp1 = ia->ia_ifp; 299 300 /* 301 * We'll never take an address that breaks the scope zone 302 * of the destination. We also skip an address if its zone 303 * does not contain the outgoing interface. 304 * XXX: we should probably use sin6_scope_id here. 305 */ 306 if (in6_setscope(&dst, ifp1, &dstzone) || 307 odstzone != dstzone) { 308 continue; 309 } 310 src = ia->ia_addr.sin6_addr; 311 if (in6_setscope(&src, ifp, &osrczone) || 312 in6_setscope(&src, ifp1, &srczone) || 313 osrczone != srczone) { 314 continue; 315 } 316 317 /* avoid unusable addresses */ 318 if ((ia->ia6_flags & 319 (IN6_IFF_NOTREADY | IN6_IFF_ANYCAST | IN6_IFF_DETACHED))) { 320 continue; 321 } 322 if (!ip6_use_deprecated && IFA6_IS_DEPRECATED(ia)) 323 continue; 324 325 #if defined(MIP6) && NMIP > 0 326 /* avoid unusable home addresses. */ 327 if ((ia->ia6_flags & IN6_IFF_HOME) && 328 !mip6_ifa6_is_addr_valid_hoa(ia)) 329 continue; 330 #endif /* MIP6 && NMIP > 0 */ 331 332 /* Rule 1: Prefer same address */ 333 if (IN6_ARE_ADDR_EQUAL(&dst, &ia->ia_addr.sin6_addr)) { 334 ia_best = ia; 335 BREAK(1); /* there should be no better candidate */ 336 } 337 338 if (ia_best == NULL) 339 REPLACE(0); 340 341 /* Rule 2: Prefer appropriate scope */ 342 if (dst_scope < 0) 343 dst_scope = in6_addrscope(&dst); 344 new_scope = in6_addrscope(&ia->ia_addr.sin6_addr); 345 if (IN6_ARE_SCOPE_CMP(best_scope, new_scope) < 0) { 346 if (IN6_ARE_SCOPE_CMP(best_scope, dst_scope) < 0) 347 REPLACE(2); 348 NEXT(2); 349 } else if (IN6_ARE_SCOPE_CMP(new_scope, best_scope) < 0) { 350 if (IN6_ARE_SCOPE_CMP(new_scope, dst_scope) < 0) 351 NEXT(2); 352 REPLACE(2); 353 } 354 355 /* 356 * Rule 3: Avoid deprecated addresses. Note that the case of 357 * !ip6_use_deprecated is already rejected above. 358 */ 359 if (!IFA6_IS_DEPRECATED(ia_best) && IFA6_IS_DEPRECATED(ia)) 360 NEXT(3); 361 if (IFA6_IS_DEPRECATED(ia_best) && !IFA6_IS_DEPRECATED(ia)) 362 REPLACE(3); 363 364 /* Rule 4: Prefer home addresses */ 365 #if defined(MIP6) && NMIP > 0 366 if (!MIP6_IS_MN) 367 goto skip_rule4; 368 369 if ((ia_best->ia6_flags & IN6_IFF_HOME) == 0 && 370 (ia->ia6_flags & IN6_IFF_HOME) == 0) { 371 /* both address are not home addresses. */ 372 goto skip_rule4; 373 } 374 375 /* 376 * If SA is simultaneously a home address and care-of 377 * address and SB is not, then prefer SA. Similarly, 378 * if SB is simultaneously a home address and care-of 379 * address and SA is not, then prefer SB. 380 */ 381 if (((ia_best->ia6_flags & IN6_IFF_HOME) != 0 && 382 ia_best->ia_ifp->if_type != IFT_MIP) 383 && 384 ((ia->ia6_flags & IN6_IFF_HOME) != 0 && 385 ia->ia_ifp->if_type == IFT_MIP)) 386 NEXT(4); 387 if (((ia_best->ia6_flags & IN6_IFF_HOME) != 0 && 388 ia_best->ia_ifp->if_type == IFT_MIP) 389 && 390 ((ia->ia6_flags & IN6_IFF_HOME) != 0 && 391 ia->ia_ifp->if_type != IFT_MIP)) 392 REPLACE(4); 393 if (ip6po_usecoa == 0) { 394 /* 395 * If SA is just a home address and SB is just 396 * a care-of address, then prefer 397 * SA. Similarly, if SB is just a home address 398 * and SA is just a care-of address, then 399 * prefer SB. 400 */ 401 if ((ia_best->ia6_flags & IN6_IFF_HOME) != 0 && 402 (ia->ia6_flags & IN6_IFF_HOME) == 0) { 403 NEXT(4); 404 } 405 if ((ia_best->ia6_flags & IN6_IFF_HOME) == 0 && 406 (ia->ia6_flags & IN6_IFF_HOME) != 0) { 407 REPLACE(4); 408 } 409 } else { 410 /* 411 * a sender don't want to use a home address 412 * because: 413 * 414 * 1) we cannot use. (ex. NS or NA to global 415 * addresses.) 416 * 417 * 2) a user specified not to use. 418 * (ex. mip6control -u) 419 */ 420 if ((ia_best->ia6_flags & IN6_IFF_HOME) == 0 && 421 (ia->ia6_flags & IN6_IFF_HOME) != 0) { 422 /* XXX breaks stat */ 423 NEXT(0); 424 } 425 if ((ia_best->ia6_flags & IN6_IFF_HOME) != 0 && 426 (ia->ia6_flags & IN6_IFF_HOME) == 0) { 427 /* XXX breaks stat */ 428 REPLACE(0); 429 } 430 } 431 skip_rule4: 432 #endif /* MIP6 && NMIP > 0 */ 433 434 /* Rule 5: Prefer outgoing interface */ 435 if (ia_best->ia_ifp == ifp && ia->ia_ifp != ifp) 436 NEXT(5); 437 if (ia_best->ia_ifp != ifp && ia->ia_ifp == ifp) 438 REPLACE(5); 439 440 /* 441 * Rule 6: Prefer matching label 442 * Note that best_policy should be non-NULL here. 443 */ 444 if (dst_policy == NULL) 445 dst_policy = lookup_addrsel_policy(dstsock); 446 if (dst_policy->label != ADDR_LABEL_NOTAPP) { 447 new_policy = lookup_addrsel_policy(&ia->ia_addr); 448 if (dst_policy->label == best_policy->label && 449 dst_policy->label != new_policy->label) 450 NEXT(6); 451 if (dst_policy->label != best_policy->label && 452 dst_policy->label == new_policy->label) 453 REPLACE(6); 454 } 455 456 /* 457 * Rule 7: Prefer public addresses. 458 * We allow users to reverse the logic by configuring 459 * a sysctl variable, so that privacy conscious users can 460 * always prefer temporary addresses. 461 */ 462 #ifdef notyet /* until introducing ND extensions and address selection */ 463 if (opts == NULL || 464 opts->ip6po_prefer_tempaddr == IP6PO_TEMPADDR_SYSTEM) { 465 prefer_tempaddr = ip6_prefer_tempaddr; 466 } else if (opts->ip6po_prefer_tempaddr == 467 IP6PO_TEMPADDR_NOTPREFER) { 468 prefer_tempaddr = 0; 469 } else 470 prefer_tempaddr = 1; 471 if (!(ia_best->ia6_flags & IN6_IFF_TEMPORARY) && 472 (ia->ia6_flags & IN6_IFF_TEMPORARY)) { 473 if (prefer_tempaddr) 474 REPLACE(7); 475 else 476 NEXT(7); 477 } 478 if ((ia_best->ia6_flags & IN6_IFF_TEMPORARY) && 479 !(ia->ia6_flags & IN6_IFF_TEMPORARY)) { 480 if (prefer_tempaddr) 481 NEXT(7); 482 else 483 REPLACE(7); 484 } 485 #endif 486 487 /* 488 * Rule 8: prefer addresses on alive interfaces. 489 * This is a KAME specific rule. 490 */ 491 if ((ia_best->ia_ifp->if_flags & IFF_UP) && 492 !(ia->ia_ifp->if_flags & IFF_UP)) 493 NEXT(8); 494 if (!(ia_best->ia_ifp->if_flags & IFF_UP) && 495 (ia->ia_ifp->if_flags & IFF_UP)) 496 REPLACE(8); 497 498 /* 499 * Rule 9: prefer addresses on "preferred" interfaces. 500 * This is a KAME specific rule. 501 */ 502 #ifdef notyet /* until introducing address selection */ 503 #define NDI_BEST ND_IFINFO(ia_best->ia_ifp) 504 #define NDI_NEW ND_IFINFO(ia->ia_ifp) 505 if ((NDI_BEST->flags & ND6_IFF_PREFER_SOURCE) && 506 !(NDI_NEW->flags & ND6_IFF_PREFER_SOURCE)) 507 NEXT(9); 508 if (!(NDI_BEST->flags & ND6_IFF_PREFER_SOURCE) && 509 (NDI_NEW->flags & ND6_IFF_PREFER_SOURCE)) 510 REPLACE(9); 511 #undef NDI_BEST 512 #undef NDI_NEW 513 #endif 514 515 /* 516 * Rule 14: Use longest matching prefix. 517 * Note: in the address selection draft, this rule is 518 * documented as "Rule 8". However, since it is also 519 * documented that this rule can be overridden, we assign 520 * a large number so that it is easy to assign smaller numbers 521 * to more preferred rules. 522 */ 523 new_matchlen = in6_matchlen(&ia->ia_addr.sin6_addr, &dst); 524 if (best_matchlen < new_matchlen) 525 REPLACE(14); 526 if (new_matchlen < best_matchlen) 527 NEXT(14); 528 529 /* Rule 15 is reserved. */ 530 531 /* 532 * Last resort: just keep the current candidate. 533 * Or, do we need more rules? 534 */ 535 continue; 536 537 replace: 538 ia_best = ia; 539 best_scope = (new_scope >= 0 ? new_scope : 540 in6_addrscope(&ia_best->ia_addr.sin6_addr)); 541 best_policy = (new_policy ? new_policy : 542 lookup_addrsel_policy(&ia_best->ia_addr)); 543 best_matchlen = (new_matchlen >= 0 ? new_matchlen : 544 in6_matchlen(&ia_best->ia_addr.sin6_addr, 545 &dst)); 546 547 next: 548 continue; 549 550 out: 551 break; 552 } 553 554 if ((ia = ia_best) == NULL) { 555 *errorp = EADDRNOTAVAIL; 556 return (NULL); 557 } 558 559 if (ifpp) 560 *ifpp = ifp; 561 return (&ia->ia_addr.sin6_addr); 562 } 563 #undef REPLACE 564 #undef BREAK 565 #undef NEXT 566 567 static int 568 selectroute(dstsock, opts, mopts, ro, retifp, retrt, clone, norouteok) 569 struct sockaddr_in6 *dstsock; 570 struct ip6_pktopts *opts; 571 struct ip6_moptions *mopts; 572 #ifdef NEW_STRUCT_ROUTE 573 struct route *ro; 574 #else 575 struct route_in6 *ro; 576 #endif 577 struct ifnet **retifp; 578 struct rtentry **retrt; 579 int clone; 580 int norouteok; 581 { 582 int error = 0; 583 struct ifnet *ifp = NULL; 584 struct rtentry *rt = NULL; 585 struct sockaddr_in6 *sin6_next; 586 struct in6_pktinfo *pi = NULL; 587 struct in6_addr *dst; 588 589 dst = &dstsock->sin6_addr; 590 591 #if 0 592 if (dstsock->sin6_addr.s6_addr32[0] == 0 && 593 dstsock->sin6_addr.s6_addr32[1] == 0 && 594 !IN6_IS_ADDR_LOOPBACK(&dstsock->sin6_addr)) { 595 printf("in6_selectroute: strange destination %s\n", 596 ip6_sprintf(&dstsock->sin6_addr)); 597 } else { 598 printf("in6_selectroute: destination = %s%%%d\n", 599 ip6_sprintf(&dstsock->sin6_addr), 600 dstsock->sin6_scope_id); /* for debug */ 601 } 602 #endif 603 604 /* If the caller specify the outgoing interface explicitly, use it. */ 605 if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) { 606 /* XXX boundary check is assumed to be already done. */ 607 #ifdef __FreeBSD__ 608 ifp = ifnet_byindex(pi->ipi6_ifindex); 609 #else 610 ifp = ifindex2ifnet[pi->ipi6_ifindex]; 611 #endif 612 if (ifp != NULL && 613 (norouteok || retrt == NULL || 614 IN6_IS_ADDR_MULTICAST(dst))) { 615 /* 616 * we do not have to check or get the route for 617 * multicast. 618 */ 619 goto done; 620 } else 621 goto getroute; 622 } 623 624 /* 625 * If the destination address is a multicast address and the outgoing 626 * interface for the address is specified by the caller, use it. 627 */ 628 if (IN6_IS_ADDR_MULTICAST(dst) && 629 mopts != NULL && (ifp = mopts->im6o_multicast_ifp) != NULL) { 630 goto done; /* we do not need a route for multicast. */ 631 } 632 633 getroute: 634 /* 635 * If the next hop address for the packet is specified by the caller, 636 * use it as the gateway. 637 */ 638 if (opts && opts->ip6po_nexthop) { 639 #ifdef notyet /* until introducing RFC3542 support */ 640 struct route_in6 *ron; 641 #endif 642 643 sin6_next = satosin6(opts->ip6po_nexthop); 644 645 /* at this moment, we only support AF_INET6 next hops */ 646 if (sin6_next->sin6_family != AF_INET6) { 647 error = EAFNOSUPPORT; /* or should we proceed? */ 648 goto done; 649 } 650 651 /* 652 * If the next hop is an IPv6 address, then the node identified 653 * by that address must be a neighbor of the sending host. 654 */ 655 #ifdef notyet /* see above */ 656 ron = &opts->ip6po_nextroute; 657 if ((ron->ro_rt && 658 (ron->ro_rt->rt_flags & (RTF_UP | RTF_GATEWAY)) != 659 RTF_UP) || 660 !IN6_ARE_ADDR_EQUAL(&satosin6(&ron->ro_dst)->sin6_addr, 661 &sin6_next->sin6_addr)) { 662 if (ron->ro_rt) { 663 RTFREE(ron->ro_rt); 664 ron->ro_rt = NULL; 665 } 666 *satosin6(&ron->ro_dst) = *sin6_next; 667 } 668 if (ron->ro_rt == NULL) { 669 rtalloc((struct route *)ron); /* multi path case? */ 670 if (ron->ro_rt == NULL || 671 (ron->ro_rt->rt_flags & RTF_GATEWAY)) { 672 if (ron->ro_rt) { 673 RTFREE(ron->ro_rt); 674 ron->ro_rt = NULL; 675 } 676 error = EHOSTUNREACH; 677 goto done; 678 } 679 } 680 if (!nd6_is_addr_neighbor(sin6_next, ron->ro_rt->rt_ifp)) { 681 RTFREE(ron->ro_rt); 682 ron->ro_rt = NULL; 683 error = EHOSTUNREACH; 684 goto done; 685 } 686 rt = ron->ro_rt; 687 ifp = rt->rt_ifp; 688 689 /* 690 * When cloning is required, try to allocate a route to the 691 * destination so that the caller can store path MTU 692 * information. 693 */ 694 if (!clone) 695 goto done; 696 #endif 697 } 698 699 /* 700 * Use a cached route if it exists and is valid, else try to allocate 701 * a new one. Note that we should check the address family of the 702 * cached destination, in case of sharing the cache with IPv4. 703 */ 704 if (ro) { 705 if (ro->ro_rt && 706 (!(ro->ro_rt->rt_flags & RTF_UP) || 707 ((struct sockaddr *)(&ro->ro_dst))->sa_family != AF_INET6 || 708 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, 709 dst))) { 710 RTFREE(ro->ro_rt); 711 ro->ro_rt = (struct rtentry *)NULL; 712 } 713 if (ro->ro_rt == (struct rtentry *)NULL) { 714 struct sockaddr_in6 *sa6; 715 716 /* No route yet, so try to acquire one */ 717 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 718 sa6 = (struct sockaddr_in6 *)&ro->ro_dst; 719 *sa6 = *dstsock; 720 sa6->sin6_scope_id = 0; 721 if (clone) { 722 #ifdef RADIX_MPATH 723 rtalloc_mpath((struct route *)ro, 724 ntohl(sa6->sin6_addr.s6_addr32[3])); 725 #else 726 rtalloc((struct route *)ro); 727 #endif /* RADIX_MPATH */ 728 } else { 729 #ifdef RADIX_MPATH 730 rtalloc_mpath((struct route *)ro, 731 ntohl(sa6->sin6_addr.s6_addr32[3])); 732 #else 733 ro->ro_rt = rtalloc1(&((struct route *)ro) 734 ->ro_dst, 0); 735 #endif /* RADIX_MPATH */ 736 } 737 } 738 739 /* 740 * do not care about the result if we have the nexthop 741 * explicitly specified. 742 */ 743 if (opts && opts->ip6po_nexthop) 744 goto done; 745 746 if (ro->ro_rt) { 747 ifp = ro->ro_rt->rt_ifp; 748 749 if (ifp == NULL) { /* can this really happen? */ 750 RTFREE(ro->ro_rt); 751 ro->ro_rt = NULL; 752 } 753 } 754 if (ro->ro_rt == NULL) 755 error = EHOSTUNREACH; 756 rt = ro->ro_rt; 757 758 /* 759 * Check if the outgoing interface conflicts with 760 * the interface specified by ipi6_ifindex (if specified). 761 * Note that loopback interface is always okay. 762 * (this may happen when we are sending a packet to one of 763 * our own addresses.) 764 */ 765 if (opts && opts->ip6po_pktinfo && 766 opts->ip6po_pktinfo->ipi6_ifindex) { 767 if (!(ifp->if_flags & IFF_LOOPBACK) && 768 ifp->if_index != 769 opts->ip6po_pktinfo->ipi6_ifindex) { 770 error = EHOSTUNREACH; 771 goto done; 772 } 773 } 774 } 775 776 done: 777 if (ifp == NULL && rt == NULL) { 778 /* 779 * This can happen if the caller did not pass a cached route 780 * nor any other hints. We treat this case an error. 781 */ 782 error = EHOSTUNREACH; 783 } 784 if (error == EHOSTUNREACH) 785 ip6stat.ip6s_noroute++; 786 787 if (retifp != NULL) 788 *retifp = ifp; 789 if (retrt != NULL) 790 *retrt = rt; /* rt may be NULL */ 791 792 return (error); 793 } 794 795 static int 796 in6_selectif(dstsock, opts, mopts, ro, retifp) 797 struct sockaddr_in6 *dstsock; 798 struct ip6_pktopts *opts; 799 struct ip6_moptions *mopts; 800 struct route_in6 *ro; 801 struct ifnet **retifp; 802 { 803 int error, clone; 804 struct rtentry *rt = NULL; 805 806 clone = IN6_IS_ADDR_MULTICAST(&dstsock->sin6_addr) ? 0 : 1; 807 if ((error = selectroute(dstsock, opts, mopts, ro, retifp, 808 &rt, clone, 1)) != 0) { 809 return (error); 810 } 811 812 /* 813 * do not use a rejected or black hole route. 814 * XXX: this check should be done in the L2 output routine. 815 * However, if we skipped this check here, we'd see the following 816 * scenario: 817 * - install a rejected route for a scoped address prefix 818 * (like fe80::/10) 819 * - send a packet to a destination that matches the scoped prefix, 820 * with ambiguity about the scope zone. 821 * - pick the outgoing interface from the route, and disambiguate the 822 * scope zone with the interface. 823 * - ip6_output() would try to get another route with the "new" 824 * destination, which may be valid. 825 * - we'd see no error on output. 826 * Although this may not be very harmful, it should still be confusing. 827 * We thus reject the case here. 828 */ 829 if (rt && (rt->rt_flags & (RTF_REJECT | RTF_BLACKHOLE))) 830 return (rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH); 831 832 /* 833 * Adjust the "outgoing" interface. If we're going to loop the packet 834 * back to ourselves, the ifp would be the loopback interface. 835 * However, we'd rather know the interface associated to the 836 * destination address (which should probably be one of our own 837 * addresses.) 838 */ 839 if (rt && rt->rt_ifa && rt->rt_ifa->ifa_ifp) 840 *retifp = rt->rt_ifa->ifa_ifp; 841 842 return (0); 843 } 844 845 int 846 in6_selectroute(dstsock, opts, mopts, ro, retifp, retrt, clone) 847 struct sockaddr_in6 *dstsock; 848 struct ip6_pktopts *opts; 849 struct ip6_moptions *mopts; 850 struct route_in6 *ro; 851 struct ifnet **retifp; 852 struct rtentry **retrt; 853 int clone; /* meaningful only for bsdi and freebsd. */ 854 { 855 return (selectroute(dstsock, opts, mopts, ro, retifp, 856 retrt, clone, 0)); 857 } 858 859 /* 860 * Default hop limit selection. The precedence is as follows: 861 * 1. Hoplimit value specified via ioctl. 862 * 2. (If the outgoing interface is detected) the current 863 * hop limit of the interface specified by router advertisement. 864 * 3. The system default hoplimit. 865 */ 866 int 867 in6_selecthlim(inp, ifp) 868 struct inpcb *inp; 869 struct ifnet *ifp; 870 { 871 if (inp && inp->in6p_hops >= 0) 872 return (inp->in6p_hops); 873 else if (ifp) 874 return (ND_IFINFO(ifp)->chlim); 875 else 876 return (ip6_defhlim); 877 } 878 879 /* 880 * Find an empty port and set it to the specified PCB. 881 */ 882 int 883 in6_pcbsetport(laddr, inp, p) 884 struct in6_addr *laddr; 885 struct inpcb *inp; 886 struct proc *p; 887 { 888 struct socket *so = inp->inp_socket; 889 struct inpcbtable *table = inp->inp_table; 890 int cnt; 891 u_int16_t minport, maxport; 892 u_int16_t lport, *lastport; 893 int wild = 0; 894 void *t; 895 896 /* XXX: this is redundant when called from in6_pcbbind */ 897 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 && 898 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 || 899 (so->so_options & SO_ACCEPTCONN) == 0)) 900 wild = 1; 901 902 if (inp->inp_flags & IN6P_LOWPORT) { 903 #ifndef IPNOPRIVPORTS 904 if (p == 0 || (suser(p->p_ucred, &p->p_acflag) != 0)) 905 return (EACCES); 906 #endif 907 minport = ip6_lowportmin; 908 maxport = ip6_lowportmax; 909 lastport = &table->inpt_lastlow; 910 } else { 911 minport = ip6_anonportmin; 912 maxport = ip6_anonportmax; 913 lastport = &table->inpt_lastport; 914 } 915 916 if (minport > maxport) { /* sanity check */ 917 u_int16_t swp; 918 919 swp = minport; 920 minport = maxport; 921 maxport = swp; 922 } 923 924 lport = *lastport - 1; 925 for (cnt = maxport - minport + 1; cnt; cnt--, lport--) { 926 if (lport < minport || lport > maxport) 927 lport = maxport; 928 #ifdef INET 929 if (IN6_IS_ADDR_V4MAPPED(laddr)) { 930 t = in_pcblookup_port(table, 931 *(struct in_addr *)&laddr->s6_addr32[3], 932 lport, wild); 933 } else 934 #endif 935 { 936 t = in6_pcblookup_port(table, laddr, lport, wild); 937 } 938 if (t == 0) 939 goto found; 940 } 941 942 return (EAGAIN); 943 944 found: 945 inp->inp_flags |= IN6P_ANONPORT; 946 *lastport = lport; 947 inp->inp_lport = htons(lport); 948 in6_pcbstate(inp, INP_BOUND); 949 return (0); /* success */ 950 } 951 952 void 953 addrsel_policy_init() 954 { 955 init_policy_queue(); 956 957 /* initialize the "last resort" policy */ 958 bzero(&defaultaddrpolicy, sizeof(defaultaddrpolicy)); 959 defaultaddrpolicy.label = ADDR_LABEL_NOTAPP; 960 } 961 962 static struct in6_addrpolicy * 963 lookup_addrsel_policy(key) 964 struct sockaddr_in6 *key; 965 { 966 struct in6_addrpolicy *match = NULL; 967 968 match = match_addrsel_policy(key); 969 970 if (match == NULL) 971 match = &defaultaddrpolicy; 972 else 973 match->use++; 974 975 return (match); 976 } 977 978 /* 979 * Subroutines to manage the address selection policy table via sysctl. 980 */ 981 struct walkarg { 982 size_t w_total; 983 size_t w_given; 984 caddr_t w_where; 985 caddr_t w_limit; 986 }; 987 988 int 989 in6_src_sysctl(oldp, oldlenp, newp, newlen) 990 void *oldp; 991 size_t *oldlenp; 992 void *newp; 993 size_t newlen; 994 { 995 int error = 0; 996 int s; 997 998 s = splsoftnet(); 999 1000 if (newp) { 1001 error = EPERM; 1002 goto end; 1003 } 1004 if (oldp && oldlenp == NULL) { 1005 error = EINVAL; 1006 goto end; 1007 } 1008 if (oldp || oldlenp) { 1009 struct walkarg w; 1010 size_t oldlen = (oldlenp ? *oldlenp : 0); 1011 1012 bzero(&w, sizeof(w)); 1013 w.w_given = oldlen; 1014 w.w_where = oldp; 1015 if (oldp) 1016 w.w_limit = (caddr_t)oldp + oldlen; 1017 1018 error = walk_addrsel_policy(dump_addrsel_policyent, &w); 1019 1020 *oldlenp = w.w_total; 1021 if (oldp && w.w_total > oldlen && error == 0) 1022 error = ENOMEM; 1023 } 1024 1025 end: 1026 splx(s); 1027 1028 return (error); 1029 } 1030 1031 int 1032 in6_src_ioctl(cmd, data) 1033 u_long cmd; 1034 caddr_t data; 1035 { 1036 int i; 1037 struct in6_addrpolicy ent0; 1038 1039 if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY) 1040 return (EOPNOTSUPP); /* check for safety */ 1041 1042 ent0 = *(struct in6_addrpolicy *)data; 1043 1044 if (ent0.label == ADDR_LABEL_NOTAPP) 1045 return (EINVAL); 1046 /* check if the prefix mask is consecutive. */ 1047 if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0) 1048 return (EINVAL); 1049 /* clear trailing garbages (if any) of the prefix address. */ 1050 for (i = 0; i < 4; i++) { 1051 ent0.addr.sin6_addr.s6_addr32[i] &= 1052 ent0.addrmask.sin6_addr.s6_addr32[i]; 1053 } 1054 ent0.use = 0; 1055 1056 switch (cmd) { 1057 case SIOCAADDRCTL_POLICY: 1058 return (add_addrsel_policyent(&ent0)); 1059 case SIOCDADDRCTL_POLICY: 1060 return (delete_addrsel_policyent(&ent0)); 1061 } 1062 1063 return (0); /* XXX: compromise compilers */ 1064 } 1065 1066 /* 1067 * The followings are implementation of the policy table using a 1068 * simple tail queue. 1069 * XXX such details should be hidden. 1070 * XXX implementation using binary tree should be more efficient. 1071 */ 1072 struct addrsel_policyent { 1073 TAILQ_ENTRY(addrsel_policyent) ape_entry; 1074 struct in6_addrpolicy ape_policy; 1075 }; 1076 1077 TAILQ_HEAD(addrsel_policyhead, addrsel_policyent); 1078 1079 struct addrsel_policyhead addrsel_policytab; 1080 1081 static void 1082 init_policy_queue() 1083 { 1084 TAILQ_INIT(&addrsel_policytab); 1085 } 1086 1087 static int 1088 add_addrsel_policyent(newpolicy) 1089 struct in6_addrpolicy *newpolicy; 1090 { 1091 struct addrsel_policyent *new, *pol; 1092 1093 /* duplication check */ 1094 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 1095 pol = TAILQ_NEXT(pol, ape_entry)) { 1096 if (IN6_ARE_ADDR_EQUAL(&newpolicy->addr.sin6_addr, 1097 &pol->ape_policy.addr.sin6_addr) && 1098 IN6_ARE_ADDR_EQUAL(&newpolicy->addrmask.sin6_addr, 1099 &pol->ape_policy.addrmask.sin6_addr)) { 1100 return (EEXIST); /* or override it? */ 1101 } 1102 } 1103 1104 MALLOC(new, struct addrsel_policyent *, sizeof(*new), M_IFADDR, 1105 M_WAITOK); 1106 bzero(new, sizeof(*new)); 1107 1108 /* XXX: should validate entry */ 1109 new->ape_policy = *newpolicy; 1110 1111 TAILQ_INSERT_TAIL(&addrsel_policytab, new, ape_entry); 1112 1113 return (0); 1114 } 1115 1116 static int 1117 delete_addrsel_policyent(key) 1118 struct in6_addrpolicy *key; 1119 { 1120 struct addrsel_policyent *pol; 1121 1122 /* search for the entry in the table */ 1123 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 1124 pol = TAILQ_NEXT(pol, ape_entry)) { 1125 if (IN6_ARE_ADDR_EQUAL(&key->addr.sin6_addr, 1126 &pol->ape_policy.addr.sin6_addr) && 1127 IN6_ARE_ADDR_EQUAL(&key->addrmask.sin6_addr, 1128 &pol->ape_policy.addrmask.sin6_addr)) { 1129 break; 1130 } 1131 } 1132 if (pol == NULL) { 1133 return (ESRCH); 1134 } 1135 1136 TAILQ_REMOVE(&addrsel_policytab, pol, ape_entry); 1137 1138 return (0); 1139 } 1140 1141 static int 1142 walk_addrsel_policy(callback, w) 1143 int (*callback) __P((struct in6_addrpolicy *, void *)); 1144 void *w; 1145 { 1146 struct addrsel_policyent *pol; 1147 int error = 0; 1148 1149 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 1150 pol = TAILQ_NEXT(pol, ape_entry)) { 1151 if ((error = (*callback)(&pol->ape_policy, w)) != 0) { 1152 return (error); 1153 } 1154 } 1155 1156 return (error); 1157 } 1158 1159 static int 1160 dump_addrsel_policyent(pol, arg) 1161 struct in6_addrpolicy *pol; 1162 void *arg; 1163 { 1164 int error = 0; 1165 struct walkarg *w = arg; 1166 1167 if (w->w_where && w->w_where + sizeof(*pol) <= w->w_limit) { 1168 if ((error = copyout(pol, w->w_where, sizeof(*pol))) != 0) 1169 return (error); 1170 w->w_where += sizeof(*pol); 1171 } 1172 w->w_total += sizeof(*pol); 1173 1174 return (error); 1175 } 1176 1177 static struct in6_addrpolicy * 1178 match_addrsel_policy(key) 1179 struct sockaddr_in6 *key; 1180 { 1181 struct addrsel_policyent *pent; 1182 struct in6_addrpolicy *bestpol = NULL, *pol; 1183 int matchlen, bestmatchlen = -1; 1184 u_char *mp, *ep, *k, *p, m; 1185 1186 for (pent = TAILQ_FIRST(&addrsel_policytab); pent; 1187 pent = TAILQ_NEXT(pent, ape_entry)) { 1188 matchlen = 0; 1189 1190 pol = &pent->ape_policy; 1191 mp = (u_char *)&pol->addrmask.sin6_addr; 1192 ep = mp + 16; /* XXX: scope field? */ 1193 k = (u_char *)&key->sin6_addr; 1194 p = (u_char *)&pol->addr.sin6_addr; 1195 for (; mp < ep && *mp; mp++, k++, p++) { 1196 m = *mp; 1197 if ((*k & m) != *p) 1198 goto next; /* not match */ 1199 if (m == 0xff) /* short cut for a typical case */ 1200 matchlen += 8; 1201 else { 1202 while (m >= 0x80) { 1203 matchlen++; 1204 m <<= 1; 1205 } 1206 } 1207 } 1208 1209 /* matched. check if this is better than the current best. */ 1210 if (bestpol == NULL || 1211 matchlen > bestmatchlen) { 1212 bestpol = pol; 1213 bestmatchlen = matchlen; 1214 } 1215 1216 next: 1217 continue; 1218 } 1219 1220 return (bestpol); 1221 } 1222
Indexes created Sat Oct 11 16:09:52 GMT 2025