in6_src.c revision 1.60
1/* $OpenBSD: in6_src.c,v 1.60 2015/09/11 19:23:00 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); 130 131 ia6 = ifatoia6(ifa_ifwithaddr(sin6tosa(&sa6), rtableid)); 132 if (ia6 == NULL || 133 (ia6->ia6_flags & (IN6_IFF_ANYCAST | IN6_IFF_NOTREADY))) 134 return (EADDRNOTAVAIL); 135 136 pi->ipi6_addr = sa6.sin6_addr; /* XXX: this overrides pi */ 137 138 *in6src = &pi->ipi6_addr; 139 return (0); 140 } 141 142 /* 143 * If the source address is not specified but the socket(if any) 144 * is already bound, use the bound address. 145 */ 146 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) { 147 *in6src = laddr; 148 return (0); 149 } 150 151 /* 152 * If the caller doesn't specify the source address but 153 * the outgoing interface, use an address associated with 154 * the interface. 155 */ 156 if (pi && pi->ipi6_ifindex) { 157 ifp = if_get(pi->ipi6_ifindex); 158 if (ifp == NULL) 159 return (ENXIO); /* XXX: better error? */ 160 161 ia6 = in6_ifawithscope(ifp, dst, rtableid); 162 if (ia6 == NULL) 163 return (EADDRNOTAVAIL); 164 165 *in6src = &ia6->ia_addr.sin6_addr; 166 return (0); 167 } 168 169 /* 170 * If the destination address is a link-local unicast address or 171 * a link/interface-local multicast address, and if the outgoing 172 * interface is specified by the sin6_scope_id filed, use an address 173 * associated with the interface. 174 * XXX: We're now trying to define more specific semantics of 175 * sin6_scope_id field, so this part will be rewritten in 176 * the near future. 177 */ 178 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MC_LINKLOCAL(dst) || 179 IN6_IS_ADDR_MC_INTFACELOCAL(dst)) && dstsock->sin6_scope_id) { 180 ifp = if_get(dstsock->sin6_scope_id); 181 if (ifp == NULL) 182 return (ENXIO); /* XXX: better error? */ 183 184 ia6 = in6_ifawithscope(ifp, dst, rtableid); 185 if (ia6 == NULL) 186 return (EADDRNOTAVAIL); 187 188 *in6src = &ia6->ia_addr.sin6_addr; 189 return (0); 190 } 191 192 /* 193 * If the destination address is a multicast address and 194 * the outgoing interface for the address is specified 195 * by the caller, use an address associated with the interface. 196 * Even if the outgoing interface is not specified, we also 197 * choose a loopback interface as the outgoing interface. 198 */ 199 if (IN6_IS_ADDR_MULTICAST(dst)) { 200 ifp = mopts ? if_get(mopts->im6o_ifidx) : NULL; 201 202 if (!ifp && dstsock->sin6_scope_id) 203 ifp = if_get(htons(dstsock->sin6_scope_id)); 204 205 if (ifp) { 206 ia6 = in6_ifawithscope(ifp, dst, rtableid); 207 if (ia6 == NULL) 208 return (EADDRNOTAVAIL); 209 210 *in6src = &ia6->ia_addr.sin6_addr; 211 return (0); 212 } 213 } 214 215 /* 216 * If the next hop address for the packet is specified 217 * by caller, use an address associated with the route 218 * to the next hop. 219 */ 220 { 221 struct sockaddr_in6 *sin6_next; 222 struct rtentry *rt; 223 224 if (opts && opts->ip6po_nexthop) { 225 sin6_next = satosin6(opts->ip6po_nexthop); 226 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL, 227 rtableid); 228 if (rt) { 229 ia6 = in6_ifawithscope(rt->rt_ifp, dst, 230 rtableid); 231 if (ia6 == NULL) 232 ia6 = ifatoia6(rt->rt_ifa); 233 } 234 if (ia6 == NULL) 235 return (EADDRNOTAVAIL); 236 237 *in6src = &ia6->ia_addr.sin6_addr; 238 return (0); 239 } 240 } 241 242 /* 243 * If route is known or can be allocated now, 244 * our src addr is taken from the i/f, else punt. 245 */ 246 if (ro) { 247 if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 || 248 !IN6_ARE_ADDR_EQUAL(&ro->ro_dst.sin6_addr, dst))) { 249 rtfree(ro->ro_rt); 250 ro->ro_rt = NULL; 251 } 252 if (ro->ro_rt == (struct rtentry *)0 || 253 ro->ro_rt->rt_ifp == (struct ifnet *)0) { 254 struct sockaddr_in6 *sa6; 255 256 /* No route yet, so try to acquire one */ 257 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 258 ro->ro_tableid = rtableid; 259 sa6 = &ro->ro_dst; 260 sa6->sin6_family = AF_INET6; 261 sa6->sin6_len = sizeof(struct sockaddr_in6); 262 sa6->sin6_addr = *dst; 263 sa6->sin6_scope_id = dstsock->sin6_scope_id; 264 if (IN6_IS_ADDR_MULTICAST(dst)) { 265 ro->ro_rt = rtalloc(sin6tosa(&ro->ro_dst), 266 RT_REPORT|RT_RESOLVE, ro->ro_tableid); 267 } else { 268 ro->ro_rt = rtalloc_mpath(sin6tosa(&ro->ro_dst), 269 NULL, ro->ro_tableid); 270 } 271 } 272 273 /* 274 * in_pcbconnect() checks out IFF_LOOPBACK to skip using 275 * the address. But we don't know why it does so. 276 * It is necessary to ensure the scope even for lo0 277 * so doesn't check out IFF_LOOPBACK. 278 */ 279 280 if (ro->ro_rt) { 281 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst, 282 rtableid); 283 if (ia6 == NULL) /* xxx scope error ?*/ 284 ia6 = ifatoia6(ro->ro_rt->rt_ifa); 285 } 286 if (ia6 == NULL) 287 return (EHOSTUNREACH); /* no route */ 288 289 *in6src = &ia6->ia_addr.sin6_addr; 290 return (0); 291 } 292 293 return (EADDRNOTAVAIL); 294} 295 296struct rtentry * 297in6_selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, 298 struct route_in6 *ro, unsigned int rtableid) 299{ 300 struct sockaddr_in6 *sin6_next; 301 struct in6_addr *dst; 302 303 dst = &dstsock->sin6_addr; 304 305 /* 306 * If the next hop address for the packet is specified by the caller, 307 * use it as the gateway. 308 */ 309 if (opts && opts->ip6po_nexthop) { 310 struct route_in6 *ron; 311 312 sin6_next = satosin6(opts->ip6po_nexthop); 313 314 /* We only support AF_INET6 next hops */ 315 if (sin6_next->sin6_family != AF_INET6) 316 return (NULL); 317 318 /* 319 * If the next hop is an IPv6 address, then the node identified 320 * by that address must be a neighbor of the sending host. 321 */ 322 ron = &opts->ip6po_nextroute; 323 if ((ron->ro_rt && 324 (ron->ro_rt->rt_flags & (RTF_UP | RTF_GATEWAY)) != 325 RTF_UP) || 326 !IN6_ARE_ADDR_EQUAL(&ron->ro_dst.sin6_addr, 327 &sin6_next->sin6_addr)) { 328 if (ron->ro_rt) { 329 rtfree(ron->ro_rt); 330 ron->ro_rt = NULL; 331 } 332 ron->ro_dst = *sin6_next; 333 ron->ro_tableid = rtableid; 334 } 335 if (ron->ro_rt == NULL) { 336 /* multi path case? */ 337 ron->ro_rt = rtalloc(sin6tosa(&ron->ro_dst), 338 RT_REPORT|RT_RESOLVE, ron->ro_tableid); 339 if (ron->ro_rt == NULL || 340 (ron->ro_rt->rt_flags & RTF_GATEWAY)) { 341 if (ron->ro_rt) { 342 rtfree(ron->ro_rt); 343 ron->ro_rt = NULL; 344 } 345 return (NULL); 346 } 347 } 348 if (!nd6_is_addr_neighbor(sin6_next, ron->ro_rt->rt_ifp)) { 349 rtfree(ron->ro_rt); 350 ron->ro_rt = NULL; 351 return (NULL); 352 } 353 354 return (ron->ro_rt); 355 } 356 357 /* 358 * Use a cached route if it exists and is valid, else try to allocate 359 * a new one. Note that we should check the address family of the 360 * cached destination, in case of sharing the cache with IPv4. 361 */ 362 if (ro) { 363 if (ro->ro_rt && 364 (!(ro->ro_rt->rt_flags & RTF_UP) || 365 sin6tosa(&ro->ro_dst)->sa_family != AF_INET6 || 366 !IN6_ARE_ADDR_EQUAL(&ro->ro_dst.sin6_addr, dst))) { 367 rtfree(ro->ro_rt); 368 ro->ro_rt = NULL; 369 } 370 if (ro->ro_rt == NULL) { 371 struct sockaddr_in6 *sa6; 372 373 /* No route yet, so try to acquire one */ 374 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 375 ro->ro_tableid = rtableid; 376 sa6 = &ro->ro_dst; 377 *sa6 = *dstsock; 378 sa6->sin6_scope_id = 0; 379 ro->ro_tableid = rtableid; 380 ro->ro_rt = rtalloc_mpath(sin6tosa(&ro->ro_dst), 381 NULL, ro->ro_tableid); 382 } 383 384 /* 385 * Check if the outgoing interface conflicts with 386 * the interface specified by ipi6_ifindex (if specified). 387 * Note that loopback interface is always okay. 388 * (this may happen when we are sending a packet to one of 389 * our own addresses.) 390 */ 391 if (opts && opts->ip6po_pktinfo && 392 opts->ip6po_pktinfo->ipi6_ifindex) { 393 if (ro->ro_rt != NULL && 394 (ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0 && 395 ro->ro_rt->rt_ifp->if_index != 396 opts->ip6po_pktinfo->ipi6_ifindex) { 397 return (NULL); 398 } 399 } 400 401 return (ro->ro_rt); 402 } 403 404 return (NULL); 405} 406 407int 408in6_selectif(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, 409 struct ip6_moptions *mopts, struct route_in6 *ro, struct ifnet **retifp, 410 u_int rtableid) 411{ 412 struct rtentry *rt = NULL; 413 struct in6_pktinfo *pi = NULL; 414 415 /* If the caller specify the outgoing interface explicitly, use it. */ 416 if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) { 417 *retifp = if_get(pi->ipi6_ifindex); 418 if (*retifp != NULL) 419 return (0); 420 } 421 422 /* 423 * If the destination address is a multicast address and the outgoing 424 * interface for the address is specified by the caller, use it. 425 */ 426 if (IN6_IS_ADDR_MULTICAST(&dstsock->sin6_addr) && 427 mopts != NULL && (*retifp = if_get(mopts->im6o_ifidx)) != NULL) 428 return (0); 429 430 rt = in6_selectroute(dstsock, opts, ro, rtableid); 431 if (rt == NULL) 432 return (EHOSTUNREACH); 433 434 /* 435 * do not use a rejected or black hole route. 436 * XXX: this check should be done in the L2 output routine. 437 * However, if we skipped this check here, we'd see the following 438 * scenario: 439 * - install a rejected route for a scoped address prefix 440 * (like fe80::/10) 441 * - send a packet to a destination that matches the scoped prefix, 442 * with ambiguity about the scope zone. 443 * - pick the outgoing interface from the route, and disambiguate the 444 * scope zone with the interface. 445 * - ip6_output() would try to get another route with the "new" 446 * destination, which may be valid. 447 * - we'd see no error on output. 448 * Although this may not be very harmful, it should still be confusing. 449 * We thus reject the case here. 450 */ 451 if (rt && (rt->rt_flags & (RTF_REJECT | RTF_BLACKHOLE))) 452 return (rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH); 453 454 /* 455 * Adjust the "outgoing" interface. If we're going to loop the packet 456 * back to ourselves, the ifp would be the loopback interface. 457 * However, we'd rather know the interface associated to the 458 * destination address (which should probably be one of our own 459 * addresses.) 460 */ 461 if (rt && rt->rt_ifa && rt->rt_ifa->ifa_ifp) 462 *retifp = rt->rt_ifa->ifa_ifp; 463 464 return (0); 465} 466 467/* 468 * Default hop limit selection. The precedence is as follows: 469 * 1. Hoplimit value specified via ioctl. 470 * 2. (If the outgoing interface is detected) the current 471 * hop limit of the interface specified by router advertisement. 472 * 3. The system default hoplimit. 473*/ 474int 475in6_selecthlim(struct inpcb *in6p, struct ifnet *ifp) 476{ 477 if (in6p && in6p->inp_hops >= 0) 478 return (in6p->inp_hops); 479 else if (ifp) 480 return (ND_IFINFO(ifp)->chlim); 481 else 482 return (ip6_defhlim); 483} 484 485/* 486 * generate kernel-internal form (scopeid embedded into s6_addr16[1]). 487 * If the address scope of is link-local, embed the interface index in the 488 * address. The routine determines our precedence 489 * between advanced API scope/interface specification and basic API 490 * specification. 491 * 492 * this function should be nuked in the future, when we get rid of 493 * embedded scopeid thing. 494 * 495 * XXX actually, it is over-specification to return ifp against sin6_scope_id. 496 * there can be multiple interfaces that belong to a particular scope zone 497 * (in specification, we have 1:N mapping between a scope zone and interfaces). 498 * we may want to change the function to return something other than ifp. 499 */ 500int 501in6_embedscope(struct in6_addr *in6, const struct sockaddr_in6 *sin6, 502 struct inpcb *in6p) 503{ 504 struct ifnet *ifp = NULL; 505 u_int32_t scopeid; 506 507 *in6 = sin6->sin6_addr; 508 scopeid = sin6->sin6_scope_id; 509 510 /* 511 * don't try to read sin6->sin6_addr beyond here, since the caller may 512 * ask us to overwrite existing sockaddr_in6 513 */ 514 515 if (IN6_IS_SCOPE_EMBED(in6)) { 516 struct in6_pktinfo *pi; 517 518 /* 519 * KAME assumption: link id == interface id 520 */ 521 522 if (in6p && in6p->inp_outputopts6 && 523 (pi = in6p->inp_outputopts6->ip6po_pktinfo) && 524 pi->ipi6_ifindex) { 525 ifp = if_get(pi->ipi6_ifindex); 526 if (ifp == NULL) 527 return ENXIO; /* XXX EINVAL? */ 528 in6->s6_addr16[1] = htons(pi->ipi6_ifindex); 529 } else if (in6p && IN6_IS_ADDR_MULTICAST(in6) && 530 in6p->inp_moptions6 && 531 (ifp = if_get(in6p->inp_moptions6->im6o_ifidx))) { 532 in6->s6_addr16[1] = htons(ifp->if_index); 533 } else if (scopeid) { 534 ifp = if_get(scopeid); 535 if (ifp == NULL) 536 return ENXIO; /* XXX EINVAL? */ 537 /*XXX assignment to 16bit from 32bit variable */ 538 in6->s6_addr16[1] = htons(scopeid & 0xffff); 539 } 540 if_put(ifp); 541 } 542 543 return 0; 544} 545 546/* 547 * generate standard sockaddr_in6 from embedded form. 548 * touches sin6_addr and sin6_scope_id only. 549 * 550 * this function should be nuked in the future, when we get rid of 551 * embedded scopeid thing. 552 */ 553void 554in6_recoverscope(struct sockaddr_in6 *sin6, const struct in6_addr *in6) 555{ 556 u_int32_t scopeid; 557 558 sin6->sin6_addr = *in6; 559 560 /* 561 * don't try to read *in6 beyond here, since the caller may 562 * ask us to overwrite existing sockaddr_in6 563 */ 564 565 sin6->sin6_scope_id = 0; 566 if (IN6_IS_SCOPE_EMBED(in6)) { 567 /* 568 * KAME assumption: link id == interface id 569 */ 570 scopeid = ntohs(sin6->sin6_addr.s6_addr16[1]); 571 if (scopeid) { 572 sin6->sin6_addr.s6_addr16[1] = 0; 573 sin6->sin6_scope_id = scopeid; 574 } 575 } 576} 577 578/* 579 * just clear the embedded scope identifer. 580 */ 581void 582in6_clearscope(struct in6_addr *addr) 583{ 584 if (IN6_IS_SCOPE_EMBED(addr)) 585 addr->s6_addr16[1] = 0; 586} 587