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/caps.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_INTFACELOCAL(dst)) { 242 ifp = loif; 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 (ia6 == NULL) { 345 *errorp = EHOSTUNREACH; /* no route */ 346 return (0); 347 } 348 return (&satosin6(&ia6->ia_addr)->sin6_addr); 349 } 350 351 *errorp = EADDRNOTAVAIL; 352 return (0); 353 } 354 355 /* 356 * Default hop limit selection. The precedence is as follows: 357 * 1. Hoplimit value specified via ioctl. 358 * 2. (If the outgoing interface is detected) the current 359 * hop limit of the interface specified by router advertisement. 360 * 3. The system default hoplimit. 361 */ 362 int 363 in6_selecthlim(struct in6pcb *in6p, struct ifnet *ifp) 364 { 365 int hlim; 366 367 if (in6p && in6p->in6p_hops >= 0) { 368 return (in6p->in6p_hops); 369 } else if (ifp) { 370 hlim = ND_IFINFO(ifp)->chlim; 371 if (hlim < ip6_minhlim) 372 hlim = ip6_minhlim; 373 } else { 374 hlim = ip6_defhlim; 375 } 376 return (hlim); 377 } 378 379 static boolean_t 380 in6_pcbporthash_update(struct inpcbportinfo *portinfo, 381 struct inpcb *inp, u_short lport, struct ucred *cred, int wild) 382 { 383 struct inpcbporthead *porthash; 384 385 /* 386 * This has to be atomic. If the porthash is shared across multiple 387 * protocol threads, e.g. tcp and udp, then the token must be held. 388 */ 389 porthash = in_pcbporthash_head(portinfo, lport); 390 GET_PORTHASH_TOKEN(porthash); 391 392 if (in6_pcblookup_local(porthash, &inp->in6p_laddr, lport, 393 wild, cred) != NULL) { 394 REL_PORTHASH_TOKEN(porthash); 395 return FALSE; 396 } 397 inp->inp_lport = lport; 398 in_pcbinsporthash(porthash, inp); 399 400 REL_PORTHASH_TOKEN(porthash); 401 return TRUE; 402 } 403 404 /* 405 * XXX: this is borrowed from in6_pcbbind(). If possible, we should 406 * share this function by all *bsd*... 407 */ 408 int 409 in6_pcbsetlport(struct in6_addr *laddr, struct inpcb *inp, struct thread *td) 410 { 411 #if 0 412 struct socket *so = inp->inp_socket; 413 #endif 414 uint16_t lport, first, last, step, first0, last0; 415 int count, error = 0, wild = 0; 416 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 417 struct inpcbportinfo *portinfo; 418 struct ucred *cred = NULL; 419 int portinfo_first, portinfo_idx; 420 uint32_t cut; 421 422 #if 0 423 /* XXX: this is redundant when called from in6_pcbbind */ 424 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0) 425 wild = INPLOOKUP_WILDCARD; 426 #endif 427 /* 428 * We force matches against wildcard ports in order to 429 * avoid auto-assigning lport to such ports, which would 430 * cause problems for same-machine connect()s. 431 */ 432 wild = INPLOOKUP_WILDCARD; 433 if (td->td_proc && td->td_proc->p_ucred) 434 cred = td->td_proc->p_ucred; 435 436 inp->inp_flags |= INP_ANONPORT; 437 438 step = pcbinfo->portinfo_cnt; 439 portinfo_first = mycpuid % pcbinfo->portinfo_cnt; 440 portinfo_idx = portinfo_first; 441 442 if (inp->inp_flags & INP_HIGHPORT) { 443 first0 = ipport_hifirstauto; /* sysctl */ 444 last0 = ipport_hilastauto; 445 } else if (inp->inp_flags & INP_LOWPORT) { 446 error = caps_priv_check_td(td, SYSCAP_RESTRICTEDROOT); 447 if (error) 448 return error; 449 first0 = ipport_lowfirstauto; /* 1023 */ 450 last0 = ipport_lowlastauto; /* 600 */ 451 } else { 452 first0 = ipport_firstauto; /* sysctl */ 453 last0 = ipport_lastauto; 454 } 455 if (first0 > last0) { 456 lport = last0; 457 last0 = first0; 458 first0 = lport; 459 } 460 KKASSERT(last0 >= first0); 461 462 cut = karc4random(); 463 loop: 464 portinfo = &pcbinfo->portinfo[portinfo_idx]; 465 first = first0; 466 last = last0; 467 468 /* 469 * Simple check to ensure all ports are not used up causing 470 * a deadlock here. 471 */ 472 in_pcbportrange(&last, &first, portinfo->offset, step); 473 lport = last - first; 474 count = lport / step; 475 476 lport = rounddown(cut % lport, step) + first; 477 KKASSERT(lport % step == portinfo->offset); 478 479 for (;;) { 480 if (count-- < 0) { /* completely used? */ 481 error = EAGAIN; 482 break; 483 } 484 485 if (__predict_false(lport < first || lport > last)) { 486 lport = first; 487 KKASSERT(lport % step == portinfo->offset); 488 } 489 490 if (in6_pcbporthash_update(portinfo, inp, htons(lport), 491 cred, wild)) { 492 error = 0; 493 break; 494 } 495 496 lport += step; 497 KKASSERT(lport % step == portinfo->offset); 498 } 499 500 if (error) { 501 /* Try next portinfo */ 502 portinfo_idx++; 503 portinfo_idx %= pcbinfo->portinfo_cnt; 504 if (portinfo_idx != portinfo_first) 505 goto loop; 506 507 /* Undo any address bind that may have occurred above. */ 508 inp->in6p_laddr = kin6addr_any; 509 } 510 return error; 511 } 512 513 /* 514 * Generate kernel-internal form (scopeid embedded into s6_addr16[1]). 515 * If the address scope of is link-local, embed the interface index in the 516 * address. The routine determines our precedence 517 * between advanced API scope/interface specification and basic API 518 * specification. 519 * 520 * This function should be nuked in the future, when we get rid of embedded 521 * scopeid thing. 522 * 523 * XXX actually, it is over-specification to return ifp against sin6_scope_id. 524 * there can be multiple interfaces that belong to a particular scope zone 525 * (in specification, we have 1:N mapping between a scope zone and interfaces). 526 * we may want to change the function to return something other than ifp. 527 */ 528 int 529 in6_embedscope(struct in6_addr *in6, 530 const struct sockaddr_in6 *sin6, 531 #ifdef HAVE_NRL_INPCB 532 struct inpcb *in6p, 533 #define in6p_outputopts inp_outputopts6 534 #define in6p_moptions inp_moptions6 535 #else 536 struct in6pcb *in6p, 537 #endif 538 struct ifnet **ifpp) 539 { 540 struct ifnet *ifp = NULL; 541 u_int32_t scopeid; 542 543 *in6 = sin6->sin6_addr; 544 scopeid = sin6->sin6_scope_id; 545 if (ifpp) 546 *ifpp = NULL; 547 548 /* 549 * don't try to read sin6->sin6_addr beyond here, since the caller may 550 * ask us to overwrite existing sockaddr_in6 551 */ 552 553 #ifdef ENABLE_DEFAULT_SCOPE 554 if (scopeid == 0) 555 scopeid = scope6_addr2default(in6); 556 #endif 557 558 if (IN6_IS_SCOPE_LINKLOCAL(in6)) { 559 struct in6_pktinfo *pi; 560 561 /* 562 * KAME assumption: link id == interface id 563 */ 564 565 if (in6p && in6p->in6p_outputopts && 566 (pi = in6p->in6p_outputopts->ip6po_pktinfo) && 567 pi->ipi6_ifindex) { 568 ifp = ifindex2ifnet[pi->ipi6_ifindex]; 569 in6->s6_addr16[1] = htons(pi->ipi6_ifindex); 570 } else if (in6p && IN6_IS_ADDR_MULTICAST(in6) && 571 in6p->in6p_moptions && 572 in6p->in6p_moptions->im6o_multicast_ifp) { 573 ifp = in6p->in6p_moptions->im6o_multicast_ifp; 574 in6->s6_addr16[1] = htons(ifp->if_index); 575 } else if (scopeid) { 576 /* boundary check */ 577 if (scopeid < 0 || if_index < scopeid) 578 return ENXIO; /* XXX EINVAL? */ 579 ifp = ifindex2ifnet[scopeid]; 580 /* XXX assignment to 16bit from 32bit variable */ 581 in6->s6_addr16[1] = htons(scopeid & 0xffff); 582 } 583 584 if (ifpp) 585 *ifpp = ifp; 586 } 587 588 return 0; 589 } 590 #ifdef HAVE_NRL_INPCB 591 #undef in6p_outputopts 592 #undef in6p_moptions 593 #endif 594 595 /* 596 * generate standard sockaddr_in6 from embedded form. 597 * touches sin6_addr and sin6_scope_id only. 598 * 599 * this function should be nuked in the future, when we get rid of 600 * embedded scopeid thing. 601 */ 602 int 603 in6_recoverscope(struct sockaddr_in6 *sin6, const struct in6_addr *in6, 604 struct ifnet *ifp) 605 { 606 u_int32_t zoneid; 607 608 sin6->sin6_addr = *in6; 609 610 /* 611 * don't try to read *in6 beyond here, since the caller may 612 * ask us to overwrite existing sockaddr_in6 613 */ 614 615 sin6->sin6_scope_id = 0; 616 if (IN6_IS_SCOPE_LINKLOCAL(in6) || IN6_IS_ADDR_MC_INTFACELOCAL(in6)) { 617 /* 618 * KAME assumption: link id == interface id 619 */ 620 zoneid = ntohs(sin6->sin6_addr.s6_addr16[1]); 621 if (zoneid) { 622 /* sanity check */ 623 if (zoneid < 0 || if_index < zoneid) 624 return ENXIO; 625 if (ifp && ifp->if_index != zoneid) 626 return ENXIO; 627 sin6->sin6_addr.s6_addr16[1] = 0; 628 sin6->sin6_scope_id = zoneid; 629 } 630 } 631 632 return 0; 633 } 634 635 /* 636 * just clear the embedded scope identifier. 637 */ 638 void 639 in6_clearscope(struct in6_addr *addr) 640 { 641 if (IN6_IS_SCOPE_LINKLOCAL(addr) || IN6_IS_ADDR_MC_INTFACELOCAL(addr)) 642 addr->s6_addr16[1] = 0; 643 } 644 645 void 646 addrsel_policy_init(void) 647 { 648 649 init_policy_queue(); 650 651 /* initialize the "last resort" policy */ 652 bzero(&defaultaddrpolicy, sizeof(defaultaddrpolicy)); 653 defaultaddrpolicy.label = ADDR_LABEL_NOTAPP; 654 } 655 656 /* 657 * Subroutines to manage the address selection policy table via sysctl. 658 */ 659 struct walkarg { 660 struct sysctl_req *w_req; 661 }; 662 663 static int in6_src_sysctl(SYSCTL_HANDLER_ARGS); 664 SYSCTL_DECL(_net_inet6_ip6); 665 SYSCTL_NODE(_net_inet6_ip6, IPV6CTL_ADDRCTLPOLICY, addrctlpolicy, 666 CTLFLAG_RD, in6_src_sysctl, "Address selection policy"); 667 668 static int 669 in6_src_sysctl(SYSCTL_HANDLER_ARGS) 670 { 671 struct walkarg w; 672 673 if (req->newptr) 674 return EPERM; 675 676 bzero(&w, sizeof(w)); 677 w.w_req = req; 678 679 return (walk_addrsel_policy(dump_addrsel_policyent, &w)); 680 } 681 682 int 683 in6_src_ioctl(u_long cmd, caddr_t data) 684 { 685 int i; 686 struct in6_addrpolicy ent0; 687 688 if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY) 689 return (EOPNOTSUPP); /* check for safety */ 690 691 ent0 = *(struct in6_addrpolicy *)data; 692 693 if (ent0.label == ADDR_LABEL_NOTAPP) 694 return (EINVAL); 695 /* check if the prefix mask is consecutive. */ 696 if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0) 697 return (EINVAL); 698 /* clear trailing garbages (if any) of the prefix address. */ 699 for (i = 0; i < 4; i++) { 700 ent0.addr.sin6_addr.s6_addr32[i] &= 701 ent0.addrmask.sin6_addr.s6_addr32[i]; 702 } 703 ent0.use = 0; 704 705 switch (cmd) { 706 case SIOCAADDRCTL_POLICY: 707 return (add_addrsel_policyent(&ent0)); 708 case SIOCDADDRCTL_POLICY: 709 return (delete_addrsel_policyent(&ent0)); 710 } 711 712 return (0); /* XXX: compromise compilers */ 713 } 714 715 /* 716 * The followings are implementation of the policy table using a 717 * simple tail queue. 718 * XXX such details should be hidden. 719 * XXX implementation using binary tree should be more efficient. 720 */ 721 struct addrsel_policyent { 722 TAILQ_ENTRY(addrsel_policyent) ape_entry; 723 struct in6_addrpolicy ape_policy; 724 }; 725 726 TAILQ_HEAD(addrsel_policyhead, addrsel_policyent); 727 728 struct addrsel_policyhead addrsel_policytab; 729 730 static void 731 init_policy_queue(void) 732 { 733 TAILQ_INIT(&addrsel_policytab); 734 } 735 736 static int 737 add_addrsel_policyent(struct in6_addrpolicy *newpolicy) 738 { 739 struct addrsel_policyent *new, *pol; 740 741 /* duplication check */ 742 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 743 pol = TAILQ_NEXT(pol, ape_entry)) { 744 if (SA6_ARE_ADDR_EQUAL(&newpolicy->addr, 745 &pol->ape_policy.addr) && 746 SA6_ARE_ADDR_EQUAL(&newpolicy->addrmask, 747 &pol->ape_policy.addrmask)) { 748 return (EEXIST); /* or override it? */ 749 } 750 } 751 752 new = kmalloc(sizeof(*new), M_IFADDR, M_WAITOK | M_ZERO); 753 754 /* XXX: should validate entry */ 755 new->ape_policy = *newpolicy; 756 757 TAILQ_INSERT_TAIL(&addrsel_policytab, new, ape_entry); 758 759 return (0); 760 } 761 762 static int 763 delete_addrsel_policyent(struct in6_addrpolicy *key) 764 { 765 struct addrsel_policyent *pol; 766 767 /* search for the entry in the table */ 768 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 769 pol = TAILQ_NEXT(pol, ape_entry)) { 770 if (SA6_ARE_ADDR_EQUAL(&key->addr, &pol->ape_policy.addr) && 771 SA6_ARE_ADDR_EQUAL(&key->addrmask, 772 &pol->ape_policy.addrmask)) { 773 break; 774 } 775 } 776 if (pol == NULL) 777 return (ESRCH); 778 779 TAILQ_REMOVE(&addrsel_policytab, pol, ape_entry); 780 kfree(pol, M_IFADDR); 781 782 return (0); 783 } 784 785 static int 786 walk_addrsel_policy(int(*callback)(struct in6_addrpolicy *, void *), void *w) 787 { 788 struct addrsel_policyent *pol; 789 int error = 0; 790 791 for (pol = TAILQ_FIRST(&addrsel_policytab); pol; 792 pol = TAILQ_NEXT(pol, ape_entry)) { 793 if ((error = (*callback)(&pol->ape_policy, w)) != 0) 794 return (error); 795 } 796 797 return (error); 798 } 799 800 static int 801 dump_addrsel_policyent(struct in6_addrpolicy *pol, void *arg) 802 { 803 int error = 0; 804 struct walkarg *w = arg; 805 806 error = SYSCTL_OUT(w->w_req, pol, sizeof(*pol)); 807 808 return (error); 809 } 810