1 /* $FreeBSD: src/sys/netinet6/in6_src.c,v 1.1.2.3 2002/02/26 18:02:06 ume Exp $ */ 2 /* $KAME: in6_src.c,v 1.37 2001/03/29 05:34:31 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 "opt_inet.h" 65 #include "opt_inet6.h" 66 67 #include <sys/param.h> 68 #include <sys/systm.h> 69 #include <sys/jail.h> 70 #include <sys/kernel.h> 71 #include <sys/malloc.h> 72 #include <sys/mbuf.h> 73 #include <sys/protosw.h> 74 #include <sys/socket.h> 75 #include <sys/socketvar.h> 76 #include <sys/sockio.h> 77 #include <sys/sysctl.h> 78 #include <sys/errno.h> 79 #include <sys/time.h> 80 #include <sys/proc.h> 81 #include <sys/priv.h> 82 83 #include <net/if.h> 84 #include <net/route.h> 85 86 #include <netinet/in.h> 87 #include <netinet/in_var.h> 88 #include <netinet/in_systm.h> 89 #include <netinet/ip.h> 90 #include <netinet/in_pcb.h> 91 #include <netinet6/in6_var.h> 92 #include <netinet/ip6.h> 93 #include <netinet6/in6_pcb.h> 94 #include <netinet6/ip6_var.h> 95 #include <netinet6/nd6.h> 96 #ifdef ENABLE_DEFAULT_SCOPE 97 #include <netinet6/scope6_var.h> 98 #endif 99 100 #include <net/net_osdep.h> 101 102 #define ADDR_LABEL_NOTAPP (-1) 103 struct in6_addrpolicy defaultaddrpolicy; 104 105 static void init_policy_queue(void); 106 static int add_addrsel_policyent(struct in6_addrpolicy *); 107 static int delete_addrsel_policyent(struct in6_addrpolicy *); 108 static int walk_addrsel_policy(int (*)(struct in6_addrpolicy *, void *), 109 void *); 110 static int dump_addrsel_policyent(struct in6_addrpolicy *, void *); 111 112 113 /* 114 * Return an IPv6 address, which is the most appropriate for a given 115 * destination and user specified options. 116 * If necessary, this function lookups the routing table and returns 117 * an entry to the caller for later use. 118 */ 119 struct in6_addr * 120 in6_selectsrc(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, 121 struct ip6_moptions *mopts, struct route_in6 *ro, 122 struct in6_addr *laddr, int *errorp, struct thread *td) 123 { 124 struct sockaddr_in6 jsin6; 125 struct ucred *cred = NULL; 126 struct in6_addr *dst; 127 struct in6_ifaddr *ia6 = NULL; 128 struct in6_pktinfo *pi = NULL; 129 int jailed = 0; 130 131 if (td && td->td_proc && td->td_proc->p_ucred) 132 cred = td->td_proc->p_ucred; 133 if (cred && cred->cr_prison) 134 jailed = 1; 135 jsin6.sin6_family = AF_INET6; 136 dst = &dstsock->sin6_addr; 137 *errorp = 0; 138 139 /* 140 * If the source address is explicitly specified by the caller, 141 * use it. 142 */ 143 if (opts && (pi = opts->ip6po_pktinfo) && 144 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) { 145 jsin6.sin6_addr = pi->ipi6_addr; 146 if (jailed && !jailed_ip(cred->cr_prison, 147 (struct sockaddr *)&jsin6)) { 148 return(0); 149 } else { 150 return (&pi->ipi6_addr); 151 } 152 } 153 154 /* 155 * If the source address is not specified but the socket(if any) 156 * is already bound, use the bound address. 157 */ 158 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) { 159 jsin6.sin6_addr = *laddr; 160 if (jailed && !jailed_ip(cred->cr_prison, 161 (struct sockaddr *)&jsin6)) { 162 return(0); 163 } else { 164 return (laddr); 165 } 166 } 167 168 /* 169 * If the caller doesn't specify the source address but 170 * the outgoing interface, use an address associated with 171 * the interface. 172 */ 173 if (pi && pi->ipi6_ifindex) { 174 /* XXX boundary check is assumed to be already done. */ 175 ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex], 176 dst, cred); 177 178 if (ia6 && jailed) { 179 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 180 if (!jailed_ip(cred->cr_prison, 181 (struct sockaddr *)&jsin6)) 182 ia6 = NULL; 183 } 184 185 if (ia6 == NULL) { 186 *errorp = EADDRNOTAVAIL; 187 return (0); 188 } 189 return (&satosin6(&ia6->ia_addr)->sin6_addr); 190 } 191 192 /* 193 * If the destination address is a link-local unicast address or 194 * a multicast address, and if the outgoing interface is specified 195 * by the sin6_scope_id filed, use an address associated with the 196 * interface. 197 * XXX: We're now trying to define more specific semantics of 198 * sin6_scope_id field, so this part will be rewritten in 199 * the near future. 200 */ 201 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) && 202 dstsock->sin6_scope_id) { 203 /* 204 * I'm not sure if boundary check for scope_id is done 205 * somewhere... 206 */ 207 if (dstsock->sin6_scope_id < 0 || 208 if_index < dstsock->sin6_scope_id) { 209 *errorp = ENXIO; /* XXX: better error? */ 210 return (0); 211 } 212 ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id], 213 dst, cred); 214 215 if (ia6 && jailed) { 216 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 217 if (!jailed_ip(cred->cr_prison, 218 (struct sockaddr *)&jsin6)) 219 ia6 = NULL; 220 } 221 222 if (ia6 == NULL) { 223 *errorp = EADDRNOTAVAIL; 224 return (0); 225 } 226 return (&satosin6(&ia6->ia_addr)->sin6_addr); 227 } 228 229 /* 230 * If the destination address is a multicast address and 231 * the outgoing interface for the address is specified 232 * by the caller, use an address associated with the interface. 233 * There is a sanity check here; if the destination has node-local 234 * scope, the outgoing interfacde should be a loopback address. 235 * Even if the outgoing interface is not specified, we also 236 * choose a loopback interface as the outgoing interface. 237 */ 238 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) { 239 struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL; 240 241 if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) { 242 ifp = &loif[0]; 243 } 244 245 if (ifp) { 246 ia6 = in6_ifawithscope(ifp, dst, cred); 247 if (ia6 == NULL) { 248 *errorp = EADDRNOTAVAIL; 249 return (0); 250 } 251 return (&satosin6(&ia6->ia_addr)->sin6_addr); 252 } 253 } 254 255 /* 256 * If the next hop address for the packet is specified 257 * by caller, use an address associated with the route 258 * to the next hop. 259 */ 260 { 261 struct sockaddr_in6 *sin6_next; 262 struct rtentry *rt; 263 264 if (opts && opts->ip6po_nexthop) { 265 sin6_next = satosin6(opts->ip6po_nexthop); 266 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL); 267 if (rt) { 268 ia6 = in6_ifawithscope(rt->rt_ifp, dst, cred); 269 if (ia6 == NULL) 270 ia6 = ifatoia6(rt->rt_ifa); 271 } 272 if (ia6 && jailed) { 273 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 274 if (!jailed_ip(cred->cr_prison, 275 (struct sockaddr *)&jsin6)) 276 ia6 = NULL; 277 } 278 279 if (ia6 == NULL) { 280 *errorp = EADDRNOTAVAIL; 281 return (0); 282 } 283 return (&satosin6(&ia6->ia_addr)->sin6_addr); 284 } 285 } 286 287 /* 288 * If route is known or can be allocated now, 289 * our src addr is taken from the i/f, else punt. 290 */ 291 if (ro) { 292 if (ro->ro_rt && 293 (!(ro->ro_rt->rt_flags & RTF_UP) || 294 satosin6(&ro->ro_dst)->sin6_family != AF_INET6 || 295 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, 296 dst))) { 297 RTFREE(ro->ro_rt); 298 ro->ro_rt = NULL; 299 } 300 if (ro->ro_rt == NULL || ro->ro_rt->rt_ifp == NULL) { 301 struct sockaddr_in6 *sa6; 302 303 /* No route yet, so try to acquire one */ 304 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6)); 305 sa6 = &ro->ro_dst; 306 sa6->sin6_family = AF_INET6; 307 sa6->sin6_len = sizeof(struct sockaddr_in6); 308 sa6->sin6_addr = *dst; 309 sa6->sin6_scope_id = dstsock->sin6_scope_id; 310 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) { 311 ro->ro_rt = 312 rtpurelookup((struct sockaddr *)&ro->ro_dst); 313 } else { 314 rtalloc((struct route *)ro); 315 } 316 } 317 318 /* 319 * in_pcbconnect() checks out IFF_LOOPBACK to skip using 320 * the address. But we don't know why it does so. 321 * It is necessary to ensure the scope even for lo0 322 * so doesn't check out IFF_LOOPBACK. 323 */ 324 325 if (ro->ro_rt) { 326 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst, cred); 327 if (ia6 && jailed) { 328 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 329 if (!jailed_ip(cred->cr_prison, 330 (struct sockaddr *)&jsin6)) 331 ia6 = NULL; 332 } 333 334 if (ia6 == NULL) /* xxx scope error ?*/ 335 ia6 = ifatoia6(ro->ro_rt->rt_ifa); 336 337 if (ia6 && jailed) { 338 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr; 339 if (!jailed_ip(cred->cr_prison, 340 (struct sockaddr *)&jsin6)) 341 ia6 = NULL; 342 } 343 } 344 #if 0 345 /* 346 * xxx The followings are necessary? (kazu) 347 * I don't think so. 348 * It's for SO_DONTROUTE option in IPv4.(jinmei) 349 */ 350 if (ia6 == 0) { 351 struct sockaddr_in6 sin6 = {sizeof(sin6), AF_INET6, 0}; 352 353 sin6->sin6_addr = *dst; 354 355 ia6 = ifatoia6(ifa_ifwithdstaddr(sin6tosa(&sin6))); 356 if (ia6 == 0) 357 ia6 = ifatoia6(ifa_ifwithnet(sin6tosa(&sin6))); 358 if (ia6 == 0) 359 return (0); 360 return (&satosin6(&ia6->ia_addr)->sin6_addr); 361 } 362 #endif /* 0 */ 363 if (ia6 == NULL) { 364 *errorp = EHOSTUNREACH; /* no route */ 365 return (0); 366 } 367 return (&satosin6(&ia6->ia_addr)->sin6_addr); 368 } 369 370 *errorp = EADDRNOTAVAIL; 371 return (0); 372 } 373 374 /* 375 * Default hop limit selection. The precedence is as follows: 376 * 1. Hoplimit value specified via ioctl. 377 * 2. (If the outgoing interface is detected) the current 378 * hop limit of the interface specified by router advertisement. 379 * 3. The system default hoplimit. 380 */ 381 int 382 in6_selecthlim(struct in6pcb *in6p, struct ifnet *ifp) 383 { 384 if (in6p && in6p->in6p_hops >= 0) 385 return (in6p->in6p_hops); 386 else if (ifp) 387 return (ND_IFINFO(ifp)->chlim); 388 else 389 return (ip6_defhlim); 390 } 391 392 /* 393 * XXX: this is borrowed from in6_pcbbind(). If possible, we should 394 * share this function by all *bsd*... 395 */ 396 int 397 in6_pcbsetlport(struct in6_addr *laddr, struct inpcb *inp, struct thread *td) 398 { 399 struct socket *so = inp->inp_socket; 400 u_int16_t lport = 0, first, last, *lastport, step; 401 int count, error = 0, wild = 0; 402 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 403 struct inpcbportinfo *portinfo; 404 struct ucred *cred = NULL; 405 int portinfo_first, portinfo_idx; 406 407 /* XXX: this is redundant when called from in6_pcbbind */ 408 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 409 wild = INPLOOKUP_WILDCARD; 410 if (td->td_proc && td->td_proc->p_ucred) 411 cred = td->td_proc->p_ucred; 412 413 inp->inp_flags |= INP_ANONPORT; 414 415 step = pcbinfo->portinfo_mask + 1; 416 portinfo_first = mycpuid & pcbinfo->portinfo_mask; 417 portinfo_idx = portinfo_first; 418 loop: 419 portinfo = &pcbinfo->portinfo[portinfo_idx]; 420 421 if (inp->inp_flags & INP_HIGHPORT) { 422 first = ipport_hifirstauto; /* sysctl */ 423 last = ipport_hilastauto; 424 lastport = &portinfo->lasthi; 425 } else if (inp->inp_flags & INP_LOWPORT) { 426 if ((error = priv_check(td, PRIV_ROOT)) != 0) 427 return error; 428 first = ipport_lowfirstauto; /* 1023 */ 429 last = ipport_lowlastauto; /* 600 */ 430 lastport = &portinfo->lastlow; 431 } else { 432 first = ipport_firstauto; /* sysctl */ 433 last = ipport_lastauto; 434 lastport = &portinfo->lastport; 435 } 436 437 /* 438 * This has to be atomic. If the porthash is shared across multiple 439 * protocol threads (aka tcp) then the token must be held. 440 */ 441 GET_PORT_TOKEN(portinfo); 442 443 /* 444 * Simple check to ensure all ports are not used up causing 445 * a deadlock here. 446 * 447 * We split the two cases (up and down) so that the direction 448 * is not being tested on each round of the loop. 449 */ 450 if (first > last) { 451 /* 452 * counting down 453 */ 454 in_pcbportrange(&first, &last, portinfo->offset, step); 455 count = (first - last) / step; 456 457 do { 458 if (count-- < 0) { /* completely used? */ 459 error = EAGAIN; 460 goto done; 461 } 462 *lastport -= step; 463 if (*lastport > first || *lastport < last) 464 *lastport = first; 465 KKASSERT((*lastport & pcbinfo->portinfo_mask) == 466 portinfo->offset); 467 lport = htons(*lastport); 468 } while (in6_pcblookup_local(portinfo, &inp->in6p_laddr, 469 lport, wild, cred)); 470 } else { 471 /* 472 * counting up 473 */ 474 in_pcbportrange(&last, &first, portinfo->offset, step); 475 count = (last - first) / step; 476 477 do { 478 if (count-- < 0) { /* completely used? */ 479 error = EAGAIN; 480 goto done; 481 } 482 *lastport += step; 483 if (*lastport < first || *lastport > last) 484 *lastport = first; 485 KKASSERT((*lastport & pcbinfo->portinfo_mask) == 486 portinfo->offset); 487 lport = htons(*lastport); 488 } while (in6_pcblookup_local(portinfo, &inp->in6p_laddr, 489 lport, wild, cred)); 490 } 491 492 inp->inp_lport = lport; 493 in_pcbinsporthash(portinfo, inp); 494 error = 0; 495 done: 496 REL_PORT_TOKEN(portinfo); 497 498 if (error) { 499 /* Try next portinfo */ 500 portinfo_idx++; 501 portinfo_idx &= pcbinfo->portinfo_mask; 502 if (portinfo_idx != portinfo_first) 503 goto loop; 504 505 /* Undo any address bind that may have occurred above. */ 506 inp->in6p_laddr = kin6addr_any; 507 } 508 return error; 509 } 510 511 /* 512 * generate kernel-internal form (scopeid embedded into s6_addr16[1]). 513 * If the address scope of is link-local, embed the interface index in the 514 * address. The routine determines our precedence 515 * between advanced API scope/interface specification and basic API 516 * specification. 517 * 518 * this function should be nuked in the future, when we get rid of 519 * embedded scopeid thing. 520 * 521 * XXX actually, it is over-specification to return ifp against sin6_scope_id. 522 * there can be multiple interfaces that belong to a particular scope zone 523 * (in specification, we have 1:N mapping between a scope zone and interfaces). 524 * we may want to change the function to return something other than ifp. 525 */ 526 int 527 in6_embedscope(struct in6_addr *in6, 528 const struct sockaddr_in6 *sin6, 529 #ifdef HAVE_NRL_INPCB 530 struct inpcb *in6p, 531 #define in6p_outputopts inp_outputopts6 532 #define in6p_moptions inp_moptions6 533 #else 534 struct in6pcb *in6p, 535 #endif 536 struct ifnet **ifpp) 537 { 538 struct ifnet *ifp = NULL; 539 u_int32_t scopeid; 540 541 *in6 = sin6->sin6_addr; 542 scopeid = sin6->sin6_scope_id; 543 if (ifpp) 544 *ifpp = NULL; 545 546 /* 547 * don't try to read sin6->sin6_addr beyond here, since the caller may 548 * ask us to overwrite existing sockaddr_in6 549 */ 550 551 #ifdef ENABLE_DEFAULT_SCOPE 552 if (scopeid == 0) 553 scopeid = scope6_addr2default(in6); 554 #endif 555 556 if (IN6_IS_SCOPE_LINKLOCAL(in6)) { 557 struct in6_pktinfo *pi; 558 559 /* 560 * KAME assumption: link id == interface id 561 */ 562 563 if (in6p && in6p->in6p_outputopts && 564 (pi = in6p->in6p_outputopts->ip6po_pktinfo) && 565 pi->ipi6_ifindex) { 566 ifp = ifindex2ifnet[pi->ipi6_ifindex]; 567 in6->s6_addr16[1] = htons(pi->ipi6_ifindex); 568 } else if (in6p && IN6_IS_ADDR_MULTICAST(in6) && 569 in6p->in6p_moptions && 570 in6p->in6p_moptions->im6o_multicast_ifp) { 571 ifp = in6p->in6p_moptions->im6o_multicast_ifp; 572 in6->s6_addr16[1] = htons(ifp->if_index); 573 } else if (scopeid) { 574 /* boundary check */ 575 if (scopeid < 0 || if_index < scopeid) 576 return ENXIO; /* XXX EINVAL? */ 577 ifp = ifindex2ifnet[scopeid]; 578 /*XXX assignment to 16bit from 32bit variable */ 579 in6->s6_addr16[1] = htons(scopeid & 0xffff); 580 } 581 582 if (ifpp) 583 *ifpp = ifp; 584 } 585 586 return 0; 587 } 588 #ifdef HAVE_NRL_INPCB 589 #undef in6p_outputopts 590 #undef in6p_moptions 591 #endif 592 593 /* 594 * generate standard sockaddr_in6 from embedded form. 595 * touches sin6_addr and sin6_scope_id only. 596 * 597 * this function should be nuked in the future, when we get rid of 598 * embedded scopeid thing. 599 */ 600 int 601 in6_recoverscope(struct sockaddr_in6 *sin6, const struct in6_addr *in6, 602 struct ifnet *ifp) 603 { 604 u_int32_t scopeid; 605 606 sin6->sin6_addr = *in6; 607 608 /* 609 * don't try to read *in6 beyond here, since the caller may 610 * ask us to overwrite existing sockaddr_in6 611 */ 612 613 sin6->sin6_scope_id = 0; 614 if (IN6_IS_SCOPE_LINKLOCAL(in6)) { 615 /* 616 * KAME assumption: link id == interface id 617 */ 618 scopeid = ntohs(sin6->sin6_addr.s6_addr16[1]); 619 if (scopeid) { 620 /* sanity check */ 621 if (scopeid < 0 || if_index < scopeid) 622 return ENXIO; 623 if (ifp && ifp->if_index != scopeid) 624 return ENXIO; 625 sin6->sin6_addr.s6_addr16[1] = 0; 626 sin6->sin6_scope_id = scopeid; 627 } 628 } 629 630 return 0; 631 } 632 633 /* 634 * just clear the embedded scope identifer. 635 * XXX: currently used for bsdi4 only as a supplement function. 636 */ 637 void 638 in6_clearscope(struct in6_addr *addr) 639 { 640 if (IN6_IS_SCOPE_LINKLOCAL(addr)) 641 addr->s6_addr16[1] = 0; 642 } 643 644 void 645 addrsel_policy_init(void) 646 { 647 648 init_policy_queue(); 649 650 /* initialize the "last resort" policy */ 651 bzero(&defaultaddrpolicy, sizeof(defaultaddrpolicy)); 652 defaultaddrpolicy.label = ADDR_LABEL_NOTAPP; 653 } 654 655 /* 656 * Subroutines to manage the address selection policy table via sysctl. 657 */ 658 struct walkarg { 659 struct sysctl_req *w_req; 660 }; 661 662 static int in6_src_sysctl(SYSCTL_HANDLER_ARGS); 663 SYSCTL_DECL(_net_inet6_ip6); 664 SYSCTL_NODE(_net_inet6_ip6, IPV6CTL_ADDRCTLPOLICY, addrctlpolicy, 665 CTLFLAG_RD, in6_src_sysctl, "Address selection policy"); 666 667 static int 668 in6_src_sysctl(SYSCTL_HANDLER_ARGS) 669 { 670 struct walkarg w; 671 672 if (req->newptr) 673 return EPERM; 674 675 bzero(&w, sizeof(w)); 676 w.w_req = req; 677 678 return (walk_addrsel_policy(dump_addrsel_policyent, &w)); 679 } 680 681 int 682 in6_src_ioctl(u_long cmd, caddr_t data) 683 { 684 int i; 685 struct in6_addrpolicy ent0; 686 687 if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY) 688 return (EOPNOTSUPP); /* check for safety */ 689 690 ent0 = *(struct in6_addrpolicy *)data; 691 692 if (ent0.label == ADDR_LABEL_NOTAPP) 693 return (EINVAL); 694 /* check if the prefix mask is consecutive. */ 695 if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0) 696 return (EINVAL); 697 /* clear trailing garbages (if any) of the prefix address. */ 698 for (i = 0; i < 4; i++) { 699 ent0.addr.sin6_addr.s6_addr32[i] &= 700 ent0.addrmask.sin6_addr.s6_addr32[i]; 701 } 702 ent0.use = 0; 703 704 switch (cmd) { 705 case SIOCAADDRCTL_POLICY: 706 return (add_addrsel_policyent(&ent0)); 707 case SIOCDADDRCTL_POLICY: 708 return (delete_addrsel_policyent(&ent0)); 709 } 710 711 return (0); /* XXX: compromise compilers */ 712 } 713 714 /* 715 * The followings are implementation of the policy table using a 716 * simple tail queue. 717 * XXX such details should be hidden. 718 * XXX implementation using binary tree should be more efficient. 719 */ 720 struct addrsel_policyent { 721 TAILQ_ENTRY(addrsel_policyent) ape_entry; 722 struct in6_addrpolicy ape_policy; 723 }; 724 725 TAILQ_HEAD(addrsel_policyhead, addrsel_policyent); 726 727 struct addrsel_policyhead addrsel_policytab; 728 729 static void 730 init_policy_queue(void) 731 { 732 TAILQ_INIT(&addrsel_policytab); 733 } 734 735 static int 736 add_addrsel_policyent(struct in6_addrpolicy *newpolicy) 737 { 738 struct addrsel_policyent *new, *pol; 739 740 /* duplication check */ 741 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 742 pol = TAILQ_NEXT(pol, ape_entry)) { 743 if (SA6_ARE_ADDR_EQUAL(&newpolicy->addr, 744 &pol->ape_policy.addr) && 745 SA6_ARE_ADDR_EQUAL(&newpolicy->addrmask, 746 &pol->ape_policy.addrmask)) { 747 return (EEXIST); /* or override it? */ 748 } 749 } 750 751 new = kmalloc(sizeof(*new), M_IFADDR, M_WAITOK | M_ZERO); 752 753 /* XXX: should validate entry */ 754 new->ape_policy = *newpolicy; 755 756 TAILQ_INSERT_TAIL(&addrsel_policytab, new, ape_entry); 757 758 return (0); 759 } 760 761 static int 762 delete_addrsel_policyent(struct in6_addrpolicy *key) 763 { 764 struct addrsel_policyent *pol; 765 766 /* search for the entry in the table */ 767 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 768 pol = TAILQ_NEXT(pol, ape_entry)) { 769 if (SA6_ARE_ADDR_EQUAL(&key->addr, &pol->ape_policy.addr) && 770 SA6_ARE_ADDR_EQUAL(&key->addrmask, 771 &pol->ape_policy.addrmask)) { 772 break; 773 } 774 } 775 if (pol == NULL) 776 return (ESRCH); 777 778 TAILQ_REMOVE(&addrsel_policytab, pol, ape_entry); 779 kfree(pol, M_IFADDR); 780 781 return (0); 782 } 783 784 static int 785 walk_addrsel_policy(int(*callback)(struct in6_addrpolicy *, void *), void *w) 786 { 787 struct addrsel_policyent *pol; 788 int error = 0; 789 790 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 791 pol = TAILQ_NEXT(pol, ape_entry)) { 792 if ((error = (*callback)(&pol->ape_policy, w)) != 0) 793 return (error); 794 } 795 796 return (error); 797 } 798 799 static int 800 dump_addrsel_policyent(struct in6_addrpolicy *pol, void *arg) 801 { 802 int error = 0; 803 struct walkarg *w = arg; 804 805 error = SYSCTL_OUT(w->w_req, pol, sizeof(*pol)); 806 807 return (error); 808 } 809