in6_src.c revision 1.68
1/* $OpenBSD: in6_src.c,v 1.68 2015/10/24 16:57:46 mpi Exp $ */ 2/* $KAME: in6_src.c,v 1.36 2001/02/06 04:08:17 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/param.h> 65#include <sys/systm.h> 66#include <sys/mbuf.h> 67#include <sys/protosw.h> 68#include <sys/socket.h> 69#include <sys/socketvar.h> 70#include <sys/ioctl.h> 71#include <sys/errno.h> 72#include <sys/time.h> 73 74#include <net/if.h> 75#include <net/if_var.h> 76#include <net/route.h> 77 78#include <netinet/in.h> 79#include <netinet/ip.h> 80#include <netinet/in_pcb.h> 81#include <netinet6/in6_var.h> 82#include <netinet/ip6.h> 83#include <netinet6/ip6_var.h> 84#include <netinet6/nd6.h> 85 86int in6_selectif(struct sockaddr_in6 *, struct ip6_pktopts *, 87 struct ip6_moptions *, struct route_in6 *, struct ifnet **, u_int); 88 89/* 90 * Return an IPv6 address, which is the most appropriate for a given 91 * destination and user specified options. 92 * If necessary, this function lookups the routing table and returns 93 * an entry to the caller for later use. 94 */ 95int 96in6_selectsrc(struct in6_addr **in6src, struct sockaddr_in6 *dstsock, 97 struct ip6_pktopts *opts, struct ip6_moptions *mopts, 98 struct route_in6 *ro, struct in6_addr *laddr, u_int rtableid) 99{ 100 struct ifnet *ifp = NULL; 101 struct in6_addr *dst; 102 struct in6_ifaddr *ia6 = NULL; 103 struct in6_pktinfo *pi = NULL; 104 int error; 105 106 dst = &dstsock->sin6_addr; 107 108 /* 109 * If the source address is explicitly specified by the caller, 110 * check if the requested source address is indeed a unicast address 111 * assigned to the node, and can be used as the packet's source 112 * address. If everything is okay, use the address as source. 113 */ 114 if (opts && (pi = opts->ip6po_pktinfo) && 115 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) { 116 struct sockaddr_in6 sa6; 117 118 /* get the outgoing interface */ 119 error = in6_selectif(dstsock, opts, mopts, ro, &ifp, rtableid); 120 if (error) 121 return (error); 122 123 bzero(&sa6, sizeof(sa6)); 124 sa6.sin6_family = AF_INET6; 125 sa6.sin6_len = sizeof(sa6); 126 sa6.sin6_addr = pi->ipi6_addr; 127 128 if (ifp && IN6_IS_SCOPE_EMBED(&sa6.sin6_addr)) 129 sa6.sin6_addr.s6_addr16[1] = htons(ifp->if_index); 130 if_put(ifp); /* put reference from in6_selectif */ 131 132 ia6 = ifatoia6(ifa_ifwithaddr(sin6tosa(&sa6), rtableid)); 133 if (ia6 == NULL || 134 (ia6->ia6_flags & (IN6_IFF_ANYCAST | IN6_IFF_NOTREADY))) 135 return (EADDRNOTAVAIL); 136 137 pi->ipi6_addr = sa6.sin6_addr; /* XXX: this overrides pi */ 138 139 *in6src = &pi->ipi6_addr; 140 return (0); 141 } 142 143 /* 144 * If the source address is not specified but the socket(if any) 145 * is already bound, use the bound address. 146 */ 147 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) { 148 *in6src = laddr; 149 return (0); 150 } 151 152 /* 153 * If the caller doesn't specify the source address but 154 * the outgoing interface, use an address associated with 155 * the interface. 156 */ 157 if (pi && pi->ipi6_ifindex) { 158 ifp = if_get(pi->ipi6_ifindex); 159 if (ifp == NULL) 160 return (ENXIO); /* XXX: better error? */ 161 162 ia6 = in6_ifawithscope(ifp, dst, rtableid); 163 if_put(ifp); 164 165 if (ia6 == NULL) 166 return (EADDRNOTAVAIL); 167 168 *in6src = &ia6->ia_addr.sin6_addr; 169 return (0); 170 } 171 172 /* 173 * If the destination address is a link-local unicast address or 174 * a link/interface-local multicast address, and if the outgoing 175 * interface is specified by the sin6_scope_id filed, use an address 176 * associated with the interface. 177 * XXX: We're now trying to define more specific semantics of 178 * sin6_scope_id field, so this part will be rewritten in 179 * the near future. 180 */ 181 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MC_LINKLOCAL(dst) || 182 IN6_IS_ADDR_MC_INTFACELOCAL(dst)) && dstsock->sin6_scope_id) { 183 ifp = if_get(dstsock->sin6_scope_id); 184 if (ifp == NULL) 185 return (ENXIO); /* XXX: better error? */ 186 187 ia6 = in6_ifawithscope(ifp, dst, rtableid); 188 if_put(ifp); 189 190 if (ia6 == NULL) 191 return (EADDRNOTAVAIL); 192 193 *in6src = &ia6->ia_addr.sin6_addr; 194 return (0); 195 } 196 197 /* 198 * If the destination address is a multicast address and 199 * the outgoing interface for the address is specified 200 * by the caller, use an address associated with the interface. 201 * Even if the outgoing interface is not specified, we also 202 * choose a loopback interface as the outgoing interface. 203 */ 204 if (IN6_IS_ADDR_MULTICAST(dst)) { 205 ifp = mopts ? if_get(mopts->im6o_ifidx) : NULL; 206 207 if (!ifp && dstsock->sin6_scope_id) 208 ifp = if_get(htons(dstsock->sin6_scope_id)); 209 210 if (ifp) { 211 ia6 = in6_ifawithscope(ifp, dst, rtableid); 212 if_put(ifp); 213 214 if (ia6 == NULL) 215 return (EADDRNOTAVAIL); 216 217 *in6src = &ia6->ia_addr.sin6_addr; 218 return (0); 219 } 220 } 221 222 /* 223 * If the next hop address for the packet is specified 224 * by caller, use an address associated with the route 225 * to the next hop. 226 */ 227 { 228 struct sockaddr_in6 *sin6_next; 229 struct rtentry *rt; 230 231 if (opts && opts->ip6po_nexthop) { 232 sin6_next = satosin6(opts->ip6po_nexthop); 233 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL, 234 rtableid); 235 if (rt != NULL) { 236 ifp = if_get(rt->rt_ifidx); 237 if (ifp != NULL) { 238 ia6 = in6_ifawithscope(ifp, dst, 239 rtableid); 240 if_put(ifp); 241 } 242 if (ia6 == NULL) 243 ia6 = ifatoia6(rt->rt_ifa); 244 rtfree(rt); 245 } 246 if (ia6 == NULL) 247 return (EADDRNOTAVAIL); 248 249 *in6src = &ia6->ia_addr.sin6_addr; 250 return (0); 251 } 252 } 253 254 /* 255 * If route is known or can be allocated now, 256 * our src addr is taken from the i/f, else punt. 257 */ 258 if (ro) { 259 if (!rtisvalid(ro->ro_rt) || (ro->ro_tableid != rtableid) || 260 !IN6_ARE_ADDR_EQUAL(&ro->ro_dst.sin6_addr, dst)) { 261 rtfree(ro->ro_rt); 262 ro->ro_rt = NULL; 263 } 264 if (ro->ro_rt == NULL) { 265 struct sockaddr_in6 *sa6; 266 267 /* No route yet, so try to acquire one */ 268 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 269 ro->ro_tableid = rtableid; 270 sa6 = &ro->ro_dst; 271 sa6->sin6_family = AF_INET6; 272 sa6->sin6_len = sizeof(struct sockaddr_in6); 273 sa6->sin6_addr = *dst; 274 sa6->sin6_scope_id = dstsock->sin6_scope_id; 275 if (IN6_IS_ADDR_MULTICAST(dst)) { 276 ro->ro_rt = rtalloc(sin6tosa(&ro->ro_dst), 277 RT_REPORT|RT_RESOLVE, ro->ro_tableid); 278 } else { 279 ro->ro_rt = rtalloc_mpath(sin6tosa(&ro->ro_dst), 280 NULL, ro->ro_tableid); 281 } 282 } 283 284 /* 285 * in_pcbconnect() checks out IFF_LOOPBACK to skip using 286 * the address. But we don't know why it does so. 287 * It is necessary to ensure the scope even for lo0 288 * so doesn't check out IFF_LOOPBACK. 289 */ 290 291 if (ro->ro_rt) { 292 ifp = if_get(ro->ro_rt->rt_ifidx); 293 if (ifp != NULL) { 294 ia6 = in6_ifawithscope(ifp, dst, rtableid); 295 if_put(ifp); 296 } 297 if (ia6 == NULL) /* xxx scope error ?*/ 298 ia6 = ifatoia6(ro->ro_rt->rt_ifa); 299 } 300 if (ia6 == NULL) 301 return (EHOSTUNREACH); /* no route */ 302 303 *in6src = &ia6->ia_addr.sin6_addr; 304 return (0); 305 } 306 307 return (EADDRNOTAVAIL); 308} 309 310struct rtentry * 311in6_selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, 312 struct route_in6 *ro, unsigned int rtableid) 313{ 314 struct sockaddr_in6 *sin6_next; 315 struct in6_addr *dst; 316 317 dst = &dstsock->sin6_addr; 318 319 /* 320 * If the next hop address for the packet is specified by the caller, 321 * use it as the gateway. 322 */ 323 if (opts && opts->ip6po_nexthop) { 324 struct route_in6 *ron; 325 326 sin6_next = satosin6(opts->ip6po_nexthop); 327 328 /* We only support AF_INET6 next hops */ 329 if (sin6_next->sin6_family != AF_INET6) 330 return (NULL); 331 332 /* 333 * If the next hop is an IPv6 address, then the node identified 334 * by that address must be a neighbor of the sending host. 335 */ 336 ron = &opts->ip6po_nextroute; 337 if (!rtisvalid(ron->ro_rt) || 338 ISSET(ron->ro_rt->rt_flags, RTF_GATEWAY) || 339 !IN6_ARE_ADDR_EQUAL(&ron->ro_dst.sin6_addr, 340 &sin6_next->sin6_addr)) { 341 if (ron->ro_rt) { 342 rtfree(ron->ro_rt); 343 ron->ro_rt = NULL; 344 } 345 ron->ro_dst = *sin6_next; 346 ron->ro_tableid = rtableid; 347 } 348 if (ron->ro_rt == NULL) { 349 /* multi path case? */ 350 ron->ro_rt = rtalloc(sin6tosa(&ron->ro_dst), 351 RT_REPORT|RT_RESOLVE, ron->ro_tableid); 352 if (ron->ro_rt == NULL || 353 (ron->ro_rt->rt_flags & RTF_GATEWAY)) { 354 if (ron->ro_rt) { 355 rtfree(ron->ro_rt); 356 ron->ro_rt = NULL; 357 } 358 return (NULL); 359 } 360 } 361 if (!nd6_is_addr_neighbor(sin6_next, ron->ro_rt->rt_ifp)) { 362 rtfree(ron->ro_rt); 363 ron->ro_rt = NULL; 364 return (NULL); 365 } 366 367 return (ron->ro_rt); 368 } 369 370 /* 371 * Use a cached route if it exists and is valid, else try to allocate 372 * a new one. Note that we should check the address family of the 373 * cached destination, in case of sharing the cache with IPv4. 374 */ 375 if (ro) { 376 if (!rtisvalid(ro->ro_rt) || 377 sin6tosa(&ro->ro_dst)->sa_family != AF_INET6 || 378 !IN6_ARE_ADDR_EQUAL(&ro->ro_dst.sin6_addr, dst)) { 379 rtfree(ro->ro_rt); 380 ro->ro_rt = NULL; 381 } 382 if (ro->ro_rt == NULL) { 383 struct sockaddr_in6 *sa6; 384 385 /* No route yet, so try to acquire one */ 386 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 387 ro->ro_tableid = rtableid; 388 sa6 = &ro->ro_dst; 389 *sa6 = *dstsock; 390 sa6->sin6_scope_id = 0; 391 ro->ro_tableid = rtableid; 392 ro->ro_rt = rtalloc_mpath(sin6tosa(&ro->ro_dst), 393 NULL, ro->ro_tableid); 394 } 395 396 /* 397 * Check if the outgoing interface conflicts with 398 * the interface specified by ipi6_ifindex (if specified). 399 * Note that loopback interface is always okay. 400 * (this may happen when we are sending a packet to one of 401 * our own addresses.) 402 */ 403 if (opts && opts->ip6po_pktinfo && 404 opts->ip6po_pktinfo->ipi6_ifindex) { 405 if (ro->ro_rt != NULL && 406 (ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0 && 407 ro->ro_rt->rt_ifp->if_index != 408 opts->ip6po_pktinfo->ipi6_ifindex) { 409 return (NULL); 410 } 411 } 412 413 return (ro->ro_rt); 414 } 415 416 return (NULL); 417} 418 419int 420in6_selectif(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, 421 struct ip6_moptions *mopts, struct route_in6 *ro, struct ifnet **retifp, 422 u_int rtableid) 423{ 424 struct rtentry *rt = NULL; 425 struct in6_pktinfo *pi = NULL; 426 427 /* If the caller specify the outgoing interface explicitly, use it. */ 428 if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) { 429 *retifp = if_get(pi->ipi6_ifindex); 430 if (*retifp != NULL) 431 return (0); 432 } 433 434 /* 435 * If the destination address is a multicast address and the outgoing 436 * interface for the address is specified by the caller, use it. 437 */ 438 if (IN6_IS_ADDR_MULTICAST(&dstsock->sin6_addr) && 439 mopts != NULL && (*retifp = if_get(mopts->im6o_ifidx)) != NULL) 440 return (0); 441 442 rt = in6_selectroute(dstsock, opts, ro, rtableid); 443 if (rt == NULL) 444 return (EHOSTUNREACH); 445 446 /* 447 * do not use a rejected or black hole route. 448 * XXX: this check should be done in the L2 output routine. 449 * However, if we skipped this check here, we'd see the following 450 * scenario: 451 * - install a rejected route for a scoped address prefix 452 * (like fe80::/10) 453 * - send a packet to a destination that matches the scoped prefix, 454 * with ambiguity about the scope zone. 455 * - pick the outgoing interface from the route, and disambiguate the 456 * scope zone with the interface. 457 * - ip6_output() would try to get another route with the "new" 458 * destination, which may be valid. 459 * - we'd see no error on output. 460 * Although this may not be very harmful, it should still be confusing. 461 * We thus reject the case here. 462 */ 463 if (rt && (rt->rt_flags & (RTF_REJECT | RTF_BLACKHOLE))) 464 return (rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH); 465 466 if (rt != NULL) 467 *retifp = if_get(rt->rt_ifidx); 468 469 return (0); 470} 471 472/* 473 * Default hop limit selection. The precedence is as follows: 474 * 1. Hoplimit value specified via ioctl. 475 * 2. (If the outgoing interface is detected) the current 476 * hop limit of the interface specified by router advertisement. 477 * 3. The system default hoplimit. 478*/ 479int 480in6_selecthlim(struct inpcb *in6p) 481{ 482 if (in6p && in6p->inp_hops >= 0) 483 return (in6p->inp_hops); 484 485 return (ip6_defhlim); 486} 487 488/* 489 * generate kernel-internal form (scopeid embedded into s6_addr16[1]). 490 * If the address scope of is link-local, embed the interface index in the 491 * address. The routine determines our precedence 492 * between advanced API scope/interface specification and basic API 493 * specification. 494 * 495 * this function should be nuked in the future, when we get rid of 496 * embedded scopeid thing. 497 * 498 * XXX actually, it is over-specification to return ifp against sin6_scope_id. 499 * there can be multiple interfaces that belong to a particular scope zone 500 * (in specification, we have 1:N mapping between a scope zone and interfaces). 501 * we may want to change the function to return something other than ifp. 502 */ 503int 504in6_embedscope(struct in6_addr *in6, const struct sockaddr_in6 *sin6, 505 struct inpcb *in6p) 506{ 507 struct ifnet *ifp = NULL; 508 u_int32_t scopeid; 509 510 *in6 = sin6->sin6_addr; 511 scopeid = sin6->sin6_scope_id; 512 513 /* 514 * don't try to read sin6->sin6_addr beyond here, since the caller may 515 * ask us to overwrite existing sockaddr_in6 516 */ 517 518 if (IN6_IS_SCOPE_EMBED(in6)) { 519 struct in6_pktinfo *pi; 520 521 /* 522 * KAME assumption: link id == interface id 523 */ 524 525 if (in6p && in6p->inp_outputopts6 && 526 (pi = in6p->inp_outputopts6->ip6po_pktinfo) && 527 pi->ipi6_ifindex) { 528 ifp = if_get(pi->ipi6_ifindex); 529 if (ifp == NULL) 530 return ENXIO; /* XXX EINVAL? */ 531 in6->s6_addr16[1] = htons(pi->ipi6_ifindex); 532 } else if (in6p && IN6_IS_ADDR_MULTICAST(in6) && 533 in6p->inp_moptions6 && 534 (ifp = if_get(in6p->inp_moptions6->im6o_ifidx))) { 535 in6->s6_addr16[1] = htons(ifp->if_index); 536 } else if (scopeid) { 537 ifp = if_get(scopeid); 538 if (ifp == NULL) 539 return ENXIO; /* XXX EINVAL? */ 540 /*XXX assignment to 16bit from 32bit variable */ 541 in6->s6_addr16[1] = htons(scopeid & 0xffff); 542 } 543 if_put(ifp); 544 } 545 546 return 0; 547} 548 549/* 550 * generate standard sockaddr_in6 from embedded form. 551 * touches sin6_addr and sin6_scope_id only. 552 * 553 * this function should be nuked in the future, when we get rid of 554 * embedded scopeid thing. 555 */ 556void 557in6_recoverscope(struct sockaddr_in6 *sin6, const struct in6_addr *in6) 558{ 559 u_int32_t scopeid; 560 561 sin6->sin6_addr = *in6; 562 563 /* 564 * don't try to read *in6 beyond here, since the caller may 565 * ask us to overwrite existing sockaddr_in6 566 */ 567 568 sin6->sin6_scope_id = 0; 569 if (IN6_IS_SCOPE_EMBED(in6)) { 570 /* 571 * KAME assumption: link id == interface id 572 */ 573 scopeid = ntohs(sin6->sin6_addr.s6_addr16[1]); 574 if (scopeid) { 575 sin6->sin6_addr.s6_addr16[1] = 0; 576 sin6->sin6_scope_id = scopeid; 577 } 578 } 579} 580 581/* 582 * just clear the embedded scope identifer. 583 */ 584void 585in6_clearscope(struct in6_addr *addr) 586{ 587 if (IN6_IS_SCOPE_EMBED(addr)) 588 addr->s6_addr16[1] = 0; 589} 590