1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 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 * $KAME: in6.c,v 1.259 2002/01/21 11:37:50 keiichi Exp $ 32 */ 33 34 /*- 35 * Copyright (c) 1982, 1986, 1991, 1993 36 * The Regents of the University of California. All rights reserved. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. Neither the name of the University nor the names of its contributors 47 * may be used to endorse or promote products derived from this software 48 * without specific prior written permission. 49 * 50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 53 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 60 * SUCH DAMAGE. 61 */ 62 63 #include "opt_inet.h" 64 #include "opt_inet6.h" 65 66 #include <sys/param.h> 67 #include <sys/eventhandler.h> 68 #include <sys/errno.h> 69 #include <sys/jail.h> 70 #include <sys/malloc.h> 71 #include <sys/socket.h> 72 #include <sys/socketvar.h> 73 #include <sys/sockio.h> 74 #include <sys/systm.h> 75 #include <sys/priv.h> 76 #include <sys/proc.h> 77 #include <sys/time.h> 78 #include <sys/kernel.h> 79 #include <sys/lock.h> 80 #include <sys/rmlock.h> 81 #include <sys/sysctl.h> 82 #include <sys/syslog.h> 83 84 #include <net/if.h> 85 #include <net/if_var.h> 86 #include <net/if_private.h> 87 #include <net/if_types.h> 88 #include <net/if_bridgevar.h> 89 #include <net/route.h> 90 #include <net/route/route_ctl.h> 91 #include <net/route/nhop.h> 92 #include <net/if_dl.h> 93 #include <net/vnet.h> 94 95 #include <netinet/in.h> 96 #include <netinet/in_var.h> 97 #include <net/if_llatbl.h> 98 #include <netinet/if_ether.h> 99 #include <netinet/in_systm.h> 100 #include <netinet/ip.h> 101 #include <netinet/in_pcb.h> 102 #include <netinet/ip_carp.h> 103 #include <netinet/icmp6.h> 104 105 #include <netinet/ip6.h> 106 #include <netinet6/ip6_var.h> 107 #include <netinet6/nd6.h> 108 #include <netinet6/mld6_var.h> 109 #include <netinet6/ip6_mroute.h> 110 #include <netinet6/in6_ifattach.h> 111 #include <netinet6/scope6_var.h> 112 #include <netinet6/in6_fib.h> 113 #include <netinet6/in6_pcb.h> 114 115 #ifdef MAC 116 #include <security/mac/mac_framework.h> 117 #endif 118 119 /* 120 * struct in6_ifreq and struct ifreq must be type punnable for common members 121 * of ifr_ifru to allow accessors to be shared. 122 */ 123 _Static_assert(offsetof(struct in6_ifreq, ifr_ifru) == 124 offsetof(struct ifreq, ifr_ifru), 125 "struct in6_ifreq and struct ifreq are not type punnable"); 126 127 VNET_DEFINE_STATIC(int, icmp6_nodeinfo_oldmcprefix) = 1; 128 #define V_icmp6_nodeinfo_oldmcprefix VNET(icmp6_nodeinfo_oldmcprefix) 129 SYSCTL_INT(_net_inet6_icmp6, ICMPV6CTL_NODEINFO_OLDMCPREFIX, 130 nodeinfo_oldmcprefix, CTLFLAG_VNET | CTLFLAG_RW, 131 &VNET_NAME(icmp6_nodeinfo_oldmcprefix), 0, 132 "Join old IPv6 NI group address in draft-ietf-ipngwg-icmp-name-lookup " 133 "for compatibility with KAME implementation"); 134 135 VNET_DEFINE_STATIC(int, nd6_useloopback) = 1; 136 #define V_nd6_useloopback VNET(nd6_useloopback) 137 SYSCTL_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_USELOOPBACK, nd6_useloopback, 138 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(nd6_useloopback), 0, 139 "Create a loopback route when configuring an IPv6 address"); 140 141 /* 142 * Definitions of some costant IP6 addresses. 143 */ 144 const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT; 145 const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT; 146 const struct in6_addr in6addr_nodelocal_allnodes = 147 IN6ADDR_NODELOCAL_ALLNODES_INIT; 148 const struct in6_addr in6addr_linklocal_allnodes = 149 IN6ADDR_LINKLOCAL_ALLNODES_INIT; 150 const struct in6_addr in6addr_linklocal_allrouters = 151 IN6ADDR_LINKLOCAL_ALLROUTERS_INIT; 152 const struct in6_addr in6addr_linklocal_allv2routers = 153 IN6ADDR_LINKLOCAL_ALLV2ROUTERS_INIT; 154 155 const struct in6_addr in6mask0 = IN6MASK0; 156 const struct in6_addr in6mask32 = IN6MASK32; 157 const struct in6_addr in6mask64 = IN6MASK64; 158 const struct in6_addr in6mask96 = IN6MASK96; 159 const struct in6_addr in6mask128 = IN6MASK128; 160 161 const struct sockaddr_in6 sa6_any = 162 { sizeof(sa6_any), AF_INET6, 0, 0, IN6ADDR_ANY_INIT, 0 }; 163 164 static int in6_notify_ifa(struct ifnet *, struct in6_ifaddr *, 165 struct in6_aliasreq *, int); 166 static void in6_unlink_ifa(struct in6_ifaddr *, struct ifnet *); 167 168 static int in6_validate_ifra(struct ifnet *, struct in6_aliasreq *, 169 struct in6_ifaddr *, int); 170 static struct in6_ifaddr *in6_alloc_ifa(struct ifnet *, 171 struct in6_aliasreq *, int flags); 172 static int in6_update_ifa_internal(struct ifnet *, struct in6_aliasreq *, 173 struct in6_ifaddr *, int, int); 174 static int in6_broadcast_ifa(struct ifnet *, struct in6_aliasreq *, 175 struct in6_ifaddr *, int); 176 177 static void in6_join_proxy_ndp_mc(struct ifnet *, const struct in6_addr *); 178 static void in6_leave_proxy_ndp_mc(struct ifnet *, const struct in6_addr *); 179 180 #define ifa2ia6(ifa) ((struct in6_ifaddr *)(ifa)) 181 #define ia62ifa(ia6) (&((ia6)->ia_ifa)) 182 183 static struct sx in6_control_sx; 184 SX_SYSINIT(in6_control_sx, &in6_control_sx, "in6_control"); 185 186 void 187 in6_newaddrmsg(struct in6_ifaddr *ia, int cmd) 188 { 189 struct rt_addrinfo info; 190 struct ifaddr *ifa; 191 struct sockaddr_dl gateway; 192 int fibnum; 193 194 ifa = &ia->ia_ifa; 195 196 /* 197 * Prepare info data for the host route. 198 * This code mimics one from ifa_maintain_loopback_route(). 199 */ 200 bzero(&info, sizeof(struct rt_addrinfo)); 201 info.rti_flags = ifa->ifa_flags | RTF_HOST | RTF_STATIC | RTF_PINNED; 202 info.rti_info[RTAX_DST] = ifa->ifa_addr; 203 info.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&gateway; 204 link_init_sdl(ifa->ifa_ifp, (struct sockaddr *)&gateway, ifa->ifa_ifp->if_type); 205 if (cmd != RTM_DELETE) 206 info.rti_ifp = V_loif; 207 208 fibnum = ia62ifa(ia)->ifa_ifp->if_fib; 209 210 if (cmd == RTM_ADD) { 211 rt_addrmsg(cmd, &ia->ia_ifa, fibnum); 212 rt_routemsg_info(cmd, &info, fibnum); 213 } else if (cmd == RTM_DELETE) { 214 rt_routemsg_info(cmd, &info, fibnum); 215 rt_addrmsg(cmd, &ia->ia_ifa, fibnum); 216 } 217 } 218 219 int 220 in6_mask2len(struct in6_addr *mask, u_char *lim0) 221 { 222 int x = 0, y; 223 u_char *lim = lim0, *p; 224 225 /* ignore the scope_id part */ 226 if (lim0 == NULL || lim0 - (u_char *)mask > sizeof(*mask)) 227 lim = (u_char *)mask + sizeof(*mask); 228 for (p = (u_char *)mask; p < lim; x++, p++) { 229 if (*p != 0xff) 230 break; 231 } 232 y = 0; 233 if (p < lim) { 234 for (y = 0; y < 8; y++) { 235 if ((*p & (0x80 >> y)) == 0) 236 break; 237 } 238 } 239 240 /* 241 * when the limit pointer is given, do a stricter check on the 242 * remaining bits. 243 */ 244 if (p < lim) { 245 if (y != 0 && (*p & (0x00ff >> y)) != 0) 246 return (-1); 247 for (p = p + 1; p < lim; p++) 248 if (*p != 0) 249 return (-1); 250 } 251 252 return x * 8 + y; 253 } 254 255 #ifdef COMPAT_FREEBSD32 256 struct in6_ndifreq32 { 257 char ifname[IFNAMSIZ]; 258 uint32_t ifindex; 259 }; 260 #define SIOCGDEFIFACE32_IN6 _IOWR('i', 86, struct in6_ndifreq32) 261 #endif 262 263 int 264 in6_control_ioctl(u_long cmd, void *data, 265 struct ifnet *ifp, struct ucred *cred) 266 { 267 struct in6_ifreq *ifr = (struct in6_ifreq *)data; 268 struct in6_ifaddr *ia = NULL; 269 struct in6_aliasreq *ifra = (struct in6_aliasreq *)data; 270 struct sockaddr_in6 *sa6; 271 int error; 272 bool control_locked = false; 273 274 /* 275 * Compat to make pre-10.x ifconfig(8) operable. 276 */ 277 if (cmd == OSIOCAIFADDR_IN6) { 278 cmd = SIOCAIFADDR_IN6; 279 ifra->ifra_vhid = 0; 280 } 281 282 switch (cmd) { 283 case SIOCGETSGCNT_IN6: 284 case SIOCGETMIFCNT_IN6: 285 /* 286 * XXX mrt_ioctl has a 3rd, unused, FIB argument in route.c. 287 * We cannot see how that would be needed, so do not adjust the 288 * KPI blindly; more likely should clean up the IPv4 variant. 289 */ 290 return (mrt6_ioctl ? mrt6_ioctl(cmd, data) : EOPNOTSUPP); 291 } 292 293 switch (cmd) { 294 case SIOCAADDRCTL_POLICY: 295 case SIOCDADDRCTL_POLICY: 296 if (cred != NULL) { 297 error = priv_check_cred(cred, PRIV_NETINET_ADDRCTRL6); 298 if (error) 299 return (error); 300 } 301 return (in6_src_ioctl(cmd, data)); 302 } 303 304 if (ifp == NULL) 305 return (EOPNOTSUPP); 306 307 switch (cmd) { 308 case SIOCSNDFLUSH_IN6: 309 case SIOCSPFXFLUSH_IN6: 310 case SIOCSRTRFLUSH_IN6: 311 case SIOCSDEFIFACE_IN6: 312 case SIOCSIFINFO_FLAGS: 313 case SIOCSIFINFO_IN6: 314 if (cred != NULL) { 315 error = priv_check_cred(cred, PRIV_NETINET_ND6); 316 if (error) 317 return (error); 318 } 319 /* FALLTHROUGH */ 320 case SIOCGIFINFO_IN6: 321 case SIOCGNBRINFO_IN6: 322 case SIOCGDEFIFACE_IN6: 323 return (nd6_ioctl(cmd, data, ifp)); 324 325 #ifdef COMPAT_FREEBSD32 326 case SIOCGDEFIFACE32_IN6: 327 { 328 struct in6_ndifreq ndif; 329 struct in6_ndifreq32 *ndif32; 330 331 error = nd6_ioctl(SIOCGDEFIFACE_IN6, (caddr_t)&ndif, 332 ifp); 333 if (error) 334 return (error); 335 ndif32 = (struct in6_ndifreq32 *)data; 336 ndif32->ifindex = ndif.ifindex; 337 return (0); 338 } 339 #endif 340 } 341 342 switch (cmd) { 343 case SIOCSIFPREFIX_IN6: 344 case SIOCDIFPREFIX_IN6: 345 case SIOCAIFPREFIX_IN6: 346 case SIOCCIFPREFIX_IN6: 347 case SIOCSGIFPREFIX_IN6: 348 case SIOCGIFPREFIX_IN6: 349 log(LOG_NOTICE, 350 "prefix ioctls are now invalidated. " 351 "please use ifconfig.\n"); 352 return (EOPNOTSUPP); 353 } 354 355 switch (cmd) { 356 case SIOCSSCOPE6: 357 if (cred != NULL) { 358 error = priv_check_cred(cred, PRIV_NETINET_SCOPE6); 359 if (error) 360 return (error); 361 } 362 /* FALLTHROUGH */ 363 case SIOCGSCOPE6: 364 case SIOCGSCOPE6DEF: 365 return (scope6_ioctl(cmd, data, ifp)); 366 } 367 368 /* 369 * Find address for this interface, if it exists. 370 * 371 * In netinet code, we have checked ifra_addr in SIOCSIF*ADDR operation 372 * only, and used the first interface address as the target of other 373 * operations (without checking ifra_addr). This was because netinet 374 * code/API assumed at most 1 interface address per interface. 375 * Since IPv6 allows a node to assign multiple addresses 376 * on a single interface, we almost always look and check the 377 * presence of ifra_addr, and reject invalid ones here. 378 * It also decreases duplicated code among SIOC*_IN6 operations. 379 */ 380 switch (cmd) { 381 case SIOCAIFADDR_IN6: 382 case SIOCSIFPHYADDR_IN6: 383 sa6 = &ifra->ifra_addr; 384 break; 385 case SIOCSIFADDR_IN6: 386 case SIOCGIFADDR_IN6: 387 case SIOCSIFDSTADDR_IN6: 388 case SIOCSIFNETMASK_IN6: 389 case SIOCGIFDSTADDR_IN6: 390 case SIOCGIFNETMASK_IN6: 391 case SIOCDIFADDR_IN6: 392 case SIOCGIFPSRCADDR_IN6: 393 case SIOCGIFPDSTADDR_IN6: 394 case SIOCGIFAFLAG_IN6: 395 case SIOCSNDFLUSH_IN6: 396 case SIOCSPFXFLUSH_IN6: 397 case SIOCSRTRFLUSH_IN6: 398 case SIOCGIFALIFETIME_IN6: 399 case SIOCGIFSTAT_IN6: 400 case SIOCGIFSTAT_ICMP6: 401 sa6 = &ifr->ifr_addr; 402 break; 403 case SIOCSIFADDR: 404 case SIOCSIFBRDADDR: 405 case SIOCSIFDSTADDR: 406 case SIOCSIFNETMASK: 407 /* 408 * Although we should pass any non-INET6 ioctl requests 409 * down to driver, we filter some legacy INET requests. 410 * Drivers trust SIOCSIFADDR et al to come from an already 411 * privileged layer, and do not perform any credentials 412 * checks or input validation. 413 */ 414 return (EINVAL); 415 default: 416 sa6 = NULL; 417 break; 418 } 419 if (sa6 && sa6->sin6_family == AF_INET6) { 420 if (sa6->sin6_scope_id != 0) 421 error = sa6_embedscope(sa6, 0); 422 else 423 error = in6_setscope(&sa6->sin6_addr, ifp, NULL); 424 if (error != 0) 425 return (error); 426 if (cred != NULL && (error = prison_check_ip6(cred, 427 &sa6->sin6_addr)) != 0) 428 return (error); 429 sx_xlock(&in6_control_sx); 430 control_locked = true; 431 ia = in6ifa_ifpwithaddr(ifp, &sa6->sin6_addr); 432 } else 433 ia = NULL; 434 435 switch (cmd) { 436 case SIOCSIFADDR_IN6: 437 case SIOCSIFDSTADDR_IN6: 438 case SIOCSIFNETMASK_IN6: 439 /* 440 * Since IPv6 allows a node to assign multiple addresses 441 * on a single interface, SIOCSIFxxx ioctls are deprecated. 442 */ 443 /* we decided to obsolete this command (20000704) */ 444 error = EINVAL; 445 goto out; 446 447 case SIOCDIFADDR_IN6: 448 /* 449 * for IPv4, we look for existing in_ifaddr here to allow 450 * "ifconfig if0 delete" to remove the first IPv4 address on 451 * the interface. For IPv6, as the spec allows multiple 452 * interface address from the day one, we consider "remove the 453 * first one" semantics to be not preferable. 454 */ 455 if (ia == NULL) { 456 error = EADDRNOTAVAIL; 457 goto out; 458 } 459 /* FALLTHROUGH */ 460 case SIOCAIFADDR_IN6: 461 /* 462 * We always require users to specify a valid IPv6 address for 463 * the corresponding operation. 464 */ 465 if (ifra->ifra_addr.sin6_family != AF_INET6 || 466 ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6)) { 467 error = EAFNOSUPPORT; 468 goto out; 469 } 470 471 if (cred != NULL) { 472 error = priv_check_cred(cred, (cmd == SIOCDIFADDR_IN6) ? 473 PRIV_NET_DELIFADDR : PRIV_NET_ADDIFADDR); 474 if (error) 475 goto out; 476 } 477 /* FALLTHROUGH */ 478 case SIOCGIFSTAT_IN6: 479 case SIOCGIFSTAT_ICMP6: 480 if (ifp->if_inet6 == NULL) { 481 error = EPFNOSUPPORT; 482 goto out; 483 } 484 break; 485 486 case SIOCGIFADDR_IN6: 487 /* This interface is basically deprecated. use SIOCGIFCONF. */ 488 /* FALLTHROUGH */ 489 case SIOCGIFAFLAG_IN6: 490 case SIOCGIFNETMASK_IN6: 491 case SIOCGIFDSTADDR_IN6: 492 case SIOCGIFALIFETIME_IN6: 493 /* must think again about its semantics */ 494 if (ia == NULL) { 495 error = EADDRNOTAVAIL; 496 goto out; 497 } 498 break; 499 } 500 501 switch (cmd) { 502 case SIOCGIFADDR_IN6: 503 ifr->ifr_addr = ia->ia_addr; 504 if ((error = sa6_recoverscope(&ifr->ifr_addr)) != 0) 505 goto out; 506 break; 507 508 case SIOCGIFDSTADDR_IN6: 509 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) { 510 error = EINVAL; 511 goto out; 512 } 513 ifr->ifr_dstaddr = ia->ia_dstaddr; 514 if ((error = sa6_recoverscope(&ifr->ifr_dstaddr)) != 0) 515 goto out; 516 break; 517 518 case SIOCGIFNETMASK_IN6: 519 ifr->ifr_addr = ia->ia_prefixmask; 520 break; 521 522 case SIOCGIFAFLAG_IN6: 523 ifr->ifr_ifru.ifru_flags6 = ia->ia6_flags; 524 break; 525 526 case SIOCGIFSTAT_IN6: 527 COUNTER_ARRAY_COPY(ifp->if_inet6->in6_ifstat, 528 &ifr->ifr_ifru.ifru_stat, 529 sizeof(struct in6_ifstat) / sizeof(uint64_t)); 530 break; 531 532 case SIOCGIFSTAT_ICMP6: 533 COUNTER_ARRAY_COPY(ifp->if_inet6->icmp6_ifstat, 534 &ifr->ifr_ifru.ifru_icmp6stat, 535 sizeof(struct icmp6_ifstat) / sizeof(uint64_t)); 536 break; 537 538 case SIOCGIFALIFETIME_IN6: 539 ifr->ifr_ifru.ifru_lifetime = ia->ia6_lifetime; 540 if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) { 541 time_t maxexpire; 542 struct in6_addrlifetime *retlt = 543 &ifr->ifr_ifru.ifru_lifetime; 544 545 /* 546 * XXX: adjust expiration time assuming time_t is 547 * signed. 548 */ 549 maxexpire = (-1) & 550 ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1)); 551 if (ia->ia6_lifetime.ia6t_vltime < 552 maxexpire - ia->ia6_updatetime) { 553 retlt->ia6t_expire = ia->ia6_updatetime + 554 ia->ia6_lifetime.ia6t_vltime; 555 } else 556 retlt->ia6t_expire = maxexpire; 557 } 558 if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) { 559 time_t maxexpire; 560 struct in6_addrlifetime *retlt = 561 &ifr->ifr_ifru.ifru_lifetime; 562 563 /* 564 * XXX: adjust expiration time assuming time_t is 565 * signed. 566 */ 567 maxexpire = (-1) & 568 ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1)); 569 if (ia->ia6_lifetime.ia6t_pltime < 570 maxexpire - ia->ia6_updatetime) { 571 retlt->ia6t_preferred = ia->ia6_updatetime + 572 ia->ia6_lifetime.ia6t_pltime; 573 } else 574 retlt->ia6t_preferred = maxexpire; 575 } 576 break; 577 578 case SIOCAIFADDR_IN6: 579 #ifdef MAC 580 /* Check if a MAC policy disallows setting the IPv6 address. */ 581 error = mac_inet6_check_add_addr(cred, &sa6->sin6_addr, ifp); 582 if (error != 0) 583 goto out; 584 #endif 585 error = in6_addifaddr(ifp, ifra, ia); 586 ia = NULL; 587 goto out; 588 589 case SIOCDIFADDR_IN6: 590 in6_purgeifaddr(ia); 591 EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa, 592 IFADDR_EVENT_DEL); 593 break; 594 595 default: 596 if (ifp->if_ioctl == NULL) { 597 error = EOPNOTSUPP; 598 goto out; 599 } 600 error = (*ifp->if_ioctl)(ifp, cmd, data); 601 goto out; 602 } 603 604 error = 0; 605 out: 606 if (control_locked) 607 sx_xunlock(&in6_control_sx); 608 609 if (ia != NULL) 610 ifa_free(&ia->ia_ifa); 611 return (error); 612 } 613 614 int 615 in6_control(struct socket *so, u_long cmd, void *data, 616 struct ifnet *ifp, struct thread *td) 617 { 618 return (in6_control_ioctl(cmd, data, ifp, td ? td->td_ucred : NULL)); 619 } 620 621 static struct in6_multi_mship * 622 in6_joingroup_legacy(struct ifnet *ifp, const struct in6_addr *mcaddr, 623 int *errorp, int delay) 624 { 625 struct in6_multi_mship *imm; 626 int error; 627 628 imm = malloc(sizeof(*imm), M_IP6MADDR, M_NOWAIT); 629 if (imm == NULL) { 630 *errorp = ENOBUFS; 631 return (NULL); 632 } 633 634 delay = (delay * MLD_FASTHZ) / hz; 635 636 error = in6_joingroup(ifp, mcaddr, NULL, &imm->i6mm_maddr, delay); 637 if (error) { 638 *errorp = error; 639 free(imm, M_IP6MADDR); 640 return (NULL); 641 } 642 643 return (imm); 644 } 645 646 static int 647 in6_solicited_node_maddr(struct in6_addr *maddr, 648 struct ifnet *ifp, const struct in6_addr *base) 649 { 650 int error; 651 652 bzero(maddr, sizeof(struct in6_addr)); 653 maddr->s6_addr32[0] = IPV6_ADDR_INT32_MLL; 654 maddr->s6_addr32[2] = htonl(1); 655 maddr->s6_addr32[3] = base->s6_addr32[3]; 656 maddr->s6_addr8[12] = 0xff; 657 if ((error = in6_setscope(maddr, ifp, NULL)) != 0) { 658 /* XXX: should not happen */ 659 log(LOG_ERR, "%s: in6_setscope failed\n", __func__); 660 } 661 662 return error; 663 } 664 665 /* 666 * Join necessary multicast groups. Factored out from in6_update_ifa(). 667 * This entire work should only be done once, for the default FIB. 668 */ 669 static int 670 in6_update_ifa_join_mc(struct ifnet *ifp, struct in6_aliasreq *ifra, 671 struct in6_ifaddr *ia, int flags, struct in6_multi **in6m_sol) 672 { 673 char ip6buf[INET6_ADDRSTRLEN]; 674 struct in6_addr mltaddr; 675 struct in6_multi_mship *imm; 676 int delay, error; 677 678 KASSERT(in6m_sol != NULL, ("%s: in6m_sol is NULL", __func__)); 679 680 /* Join solicited multicast addr for new host id. */ 681 if ((error = in6_solicited_node_maddr(&mltaddr, ifp, 682 &ifra->ifra_addr.sin6_addr)) != 0) 683 goto cleanup; 684 delay = error = 0; 685 if ((flags & IN6_IFAUPDATE_DADDELAY)) { 686 /* 687 * We need a random delay for DAD on the address being 688 * configured. It also means delaying transmission of the 689 * corresponding MLD report to avoid report collision. 690 * [RFC 4861, Section 6.3.7] 691 */ 692 delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz); 693 } 694 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay); 695 if (imm == NULL) { 696 nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s " 697 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr), 698 if_name(ifp), error)); 699 goto cleanup; 700 } 701 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); 702 *in6m_sol = imm->i6mm_maddr; 703 704 /* 705 * Join link-local all-nodes address. 706 */ 707 mltaddr = in6addr_linklocal_allnodes; 708 if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) 709 goto cleanup; /* XXX: should not fail */ 710 711 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0); 712 if (imm == NULL) { 713 nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s " 714 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr), 715 if_name(ifp), error)); 716 goto cleanup; 717 } 718 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); 719 720 /* 721 * Join node information group address. 722 */ 723 delay = 0; 724 if ((flags & IN6_IFAUPDATE_DADDELAY)) { 725 /* 726 * The spec does not say anything about delay for this group, 727 * but the same logic should apply. 728 */ 729 delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz); 730 } 731 if (in6_nigroup(ifp, NULL, -1, &mltaddr) == 0) { 732 /* XXX jinmei */ 733 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay); 734 if (imm == NULL) 735 nd6log((LOG_WARNING, 736 "%s: in6_joingroup failed for %s on %s " 737 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, 738 &mltaddr), if_name(ifp), error)); 739 /* XXX not very fatal, go on... */ 740 else 741 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); 742 } 743 if (V_icmp6_nodeinfo_oldmcprefix && 744 in6_nigroup_oldmcprefix(ifp, NULL, -1, &mltaddr) == 0) { 745 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay); 746 if (imm == NULL) 747 nd6log((LOG_WARNING, 748 "%s: in6_joingroup failed for %s on %s " 749 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, 750 &mltaddr), if_name(ifp), error)); 751 /* XXX not very fatal, go on... */ 752 else 753 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); 754 } 755 756 /* 757 * Join interface-local all-nodes address. 758 * (ff01::1%ifN, and ff01::%ifN/32) 759 */ 760 mltaddr = in6addr_nodelocal_allnodes; 761 if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) 762 goto cleanup; /* XXX: should not fail */ 763 764 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0); 765 if (imm == NULL) { 766 nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s " 767 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, 768 &mltaddr), if_name(ifp), error)); 769 goto cleanup; 770 } 771 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain); 772 773 cleanup: 774 return (error); 775 } 776 777 /* 778 * Update parameters of an IPv6 interface address. 779 * If necessary, a new entry is created and linked into address chains. 780 * This function is separated from in6_control(). 781 */ 782 int 783 in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, 784 struct in6_ifaddr *ia, int flags) 785 { 786 int error, hostIsNew = 0; 787 788 if ((error = in6_validate_ifra(ifp, ifra, ia, flags)) != 0) 789 return (error); 790 791 if (ia == NULL) { 792 hostIsNew = 1; 793 if ((ia = in6_alloc_ifa(ifp, ifra, flags)) == NULL) 794 return (ENOBUFS); 795 } 796 797 error = in6_update_ifa_internal(ifp, ifra, ia, hostIsNew, flags); 798 if (error != 0) { 799 if (hostIsNew != 0) { 800 in6_unlink_ifa(ia, ifp); 801 ifa_free(&ia->ia_ifa); 802 } 803 return (error); 804 } 805 806 if (hostIsNew) 807 error = in6_broadcast_ifa(ifp, ifra, ia, flags); 808 809 return (error); 810 } 811 812 /* 813 * Fill in basic IPv6 address request info. 814 */ 815 void 816 in6_prepare_ifra(struct in6_aliasreq *ifra, const struct in6_addr *addr, 817 const struct in6_addr *mask) 818 { 819 820 memset(ifra, 0, sizeof(struct in6_aliasreq)); 821 822 ifra->ifra_addr.sin6_family = AF_INET6; 823 ifra->ifra_addr.sin6_len = sizeof(struct sockaddr_in6); 824 if (addr != NULL) 825 ifra->ifra_addr.sin6_addr = *addr; 826 827 ifra->ifra_prefixmask.sin6_family = AF_INET6; 828 ifra->ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6); 829 if (mask != NULL) 830 ifra->ifra_prefixmask.sin6_addr = *mask; 831 } 832 833 static int 834 in6_validate_ifra(struct ifnet *ifp, struct in6_aliasreq *ifra, 835 struct in6_ifaddr *ia, int flags) 836 { 837 int plen = -1; 838 struct sockaddr_in6 dst6; 839 struct in6_addrlifetime *lt; 840 char ip6buf[INET6_ADDRSTRLEN]; 841 842 /* Validate parameters */ 843 if (ifp == NULL || ifra == NULL) /* this maybe redundant */ 844 return (EINVAL); 845 846 /* 847 * The destination address for a p2p link must have a family 848 * of AF_UNSPEC or AF_INET6. 849 */ 850 if ((ifp->if_flags & IFF_POINTOPOINT) != 0 && 851 ifra->ifra_dstaddr.sin6_family != AF_INET6 && 852 ifra->ifra_dstaddr.sin6_family != AF_UNSPEC) 853 return (EAFNOSUPPORT); 854 855 /* 856 * Validate address 857 */ 858 if (ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6) || 859 ifra->ifra_addr.sin6_family != AF_INET6) 860 return (EINVAL); 861 862 /* 863 * validate ifra_prefixmask. don't check sin6_family, netmask 864 * does not carry fields other than sin6_len. 865 */ 866 if (ifra->ifra_prefixmask.sin6_len > sizeof(struct sockaddr_in6)) 867 return (EINVAL); 868 /* 869 * Because the IPv6 address architecture is classless, we require 870 * users to specify a (non 0) prefix length (mask) for a new address. 871 * We also require the prefix (when specified) mask is valid, and thus 872 * reject a non-consecutive mask. 873 */ 874 if (ia == NULL && ifra->ifra_prefixmask.sin6_len == 0) 875 return (EINVAL); 876 if (ifra->ifra_prefixmask.sin6_len != 0) { 877 plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr, 878 (u_char *)&ifra->ifra_prefixmask + 879 ifra->ifra_prefixmask.sin6_len); 880 if (plen <= 0) 881 return (EINVAL); 882 } else { 883 /* 884 * In this case, ia must not be NULL. We just use its prefix 885 * length. 886 */ 887 plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); 888 } 889 /* 890 * If the destination address on a p2p interface is specified, 891 * and the address is a scoped one, validate/set the scope 892 * zone identifier. 893 */ 894 dst6 = ifra->ifra_dstaddr; 895 if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) != 0 && 896 (dst6.sin6_family == AF_INET6)) { 897 struct in6_addr in6_tmp; 898 u_int32_t zoneid; 899 900 in6_tmp = dst6.sin6_addr; 901 if (in6_setscope(&in6_tmp, ifp, &zoneid)) 902 return (EINVAL); /* XXX: should be impossible */ 903 904 if (dst6.sin6_scope_id != 0) { 905 if (dst6.sin6_scope_id != zoneid) 906 return (EINVAL); 907 } else /* user omit to specify the ID. */ 908 dst6.sin6_scope_id = zoneid; 909 910 /* convert into the internal form */ 911 if (sa6_embedscope(&dst6, 0)) 912 return (EINVAL); /* XXX: should be impossible */ 913 } 914 /* Modify original ifra_dstaddr to reflect changes */ 915 ifra->ifra_dstaddr = dst6; 916 917 /* 918 * The destination address can be specified only for a p2p or a 919 * loopback interface. If specified, the corresponding prefix length 920 * must be 128. 921 */ 922 if (ifra->ifra_dstaddr.sin6_family == AF_INET6) { 923 if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) == 0) { 924 /* XXX: noisy message */ 925 nd6log((LOG_INFO, "in6_update_ifa: a destination can " 926 "be specified for a p2p or a loopback IF only\n")); 927 return (EINVAL); 928 } 929 if (plen != 128) { 930 nd6log((LOG_INFO, "in6_update_ifa: prefixlen should " 931 "be 128 when dstaddr is specified\n")); 932 return (EINVAL); 933 } 934 } 935 /* lifetime consistency check */ 936 lt = &ifra->ifra_lifetime; 937 if (lt->ia6t_pltime > lt->ia6t_vltime) 938 return (EINVAL); 939 if (lt->ia6t_vltime == 0) { 940 /* 941 * the following log might be noisy, but this is a typical 942 * configuration mistake or a tool's bug. 943 */ 944 nd6log((LOG_INFO, 945 "in6_update_ifa: valid lifetime is 0 for %s\n", 946 ip6_sprintf(ip6buf, &ifra->ifra_addr.sin6_addr))); 947 948 if (ia == NULL) 949 return (0); /* there's nothing to do */ 950 } 951 952 /* Check prefix mask */ 953 if (ia != NULL && ifra->ifra_prefixmask.sin6_len != 0) { 954 /* 955 * We prohibit changing the prefix length of an existing 956 * address, because 957 * + such an operation should be rare in IPv6, and 958 * + the operation would confuse prefix management. 959 */ 960 if (ia->ia_prefixmask.sin6_len != 0 && 961 in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL) != plen) { 962 nd6log((LOG_INFO, "in6_validate_ifa: the prefix length " 963 "of an existing %s address should not be changed\n", 964 ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); 965 966 return (EINVAL); 967 } 968 } 969 970 return (0); 971 } 972 973 /* 974 * Allocate a new ifaddr and link it into chains. 975 */ 976 static struct in6_ifaddr * 977 in6_alloc_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, int flags) 978 { 979 struct in6_ifaddr *ia; 980 981 /* 982 * When in6_alloc_ifa() is called in a process of a received 983 * RA, it is called under an interrupt context. So, we should 984 * call malloc with M_NOWAIT. 985 */ 986 ia = (struct in6_ifaddr *)ifa_alloc(sizeof(*ia), M_NOWAIT); 987 if (ia == NULL) 988 return (NULL); 989 LIST_INIT(&ia->ia6_memberships); 990 /* Initialize the address and masks, and put time stamp */ 991 ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr; 992 ia->ia_addr.sin6_family = AF_INET6; 993 ia->ia_addr.sin6_len = sizeof(ia->ia_addr); 994 /* XXX: Can we assign ,sin6_addr and skip the rest? */ 995 ia->ia_addr = ifra->ifra_addr; 996 ia->ia6_createtime = time_uptime; 997 if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) != 0) { 998 /* 999 * Some functions expect that ifa_dstaddr is not 1000 * NULL for p2p interfaces. 1001 */ 1002 ia->ia_ifa.ifa_dstaddr = 1003 (struct sockaddr *)&ia->ia_dstaddr; 1004 } else { 1005 ia->ia_ifa.ifa_dstaddr = NULL; 1006 } 1007 1008 /* set prefix mask if any */ 1009 ia->ia_ifa.ifa_netmask = (struct sockaddr *)&ia->ia_prefixmask; 1010 if (ifra->ifra_prefixmask.sin6_len != 0) { 1011 ia->ia_prefixmask.sin6_family = AF_INET6; 1012 ia->ia_prefixmask.sin6_len = ifra->ifra_prefixmask.sin6_len; 1013 ia->ia_prefixmask.sin6_addr = ifra->ifra_prefixmask.sin6_addr; 1014 } 1015 1016 ia->ia_ifp = ifp; 1017 ifa_ref(&ia->ia_ifa); /* if_addrhead */ 1018 IF_ADDR_WLOCK(ifp); 1019 CK_STAILQ_INSERT_TAIL(&ifp->if_addrhead, &ia->ia_ifa, ifa_link); 1020 IF_ADDR_WUNLOCK(ifp); 1021 1022 ifa_ref(&ia->ia_ifa); /* in6_ifaddrhead */ 1023 IN6_IFADDR_WLOCK(); 1024 CK_STAILQ_INSERT_TAIL(&V_in6_ifaddrhead, ia, ia_link); 1025 CK_LIST_INSERT_HEAD(IN6ADDR_HASH(&ia->ia_addr.sin6_addr), ia, ia6_hash); 1026 IN6_IFADDR_WUNLOCK(); 1027 1028 return (ia); 1029 } 1030 1031 time_t 1032 in6_expire_time(uint32_t ltime) 1033 { 1034 if (ltime == ND6_INFINITE_LIFETIME) 1035 return (0); 1036 else 1037 return (time_uptime + ltime); 1038 } 1039 1040 /* 1041 * Update/configure interface address parameters: 1042 * 1043 * 1) Update lifetime 1044 * 2) Update interface metric ad flags 1045 * 3) Notify other subsystems 1046 */ 1047 static int 1048 in6_update_ifa_internal(struct ifnet *ifp, struct in6_aliasreq *ifra, 1049 struct in6_ifaddr *ia, int hostIsNew, int flags) 1050 { 1051 int error; 1052 1053 /* update timestamp */ 1054 ia->ia6_updatetime = time_uptime; 1055 1056 /* 1057 * Set lifetimes. We do not refer to ia6t_expire and ia6t_preferred 1058 * to see if the address is deprecated or invalidated, but initialize 1059 * these members for applications. 1060 */ 1061 ia->ia6_lifetime = ifra->ifra_lifetime; 1062 ia->ia6_lifetime.ia6t_expire = 1063 in6_expire_time(ifra->ifra_lifetime.ia6t_vltime); 1064 ia->ia6_lifetime.ia6t_preferred = 1065 in6_expire_time(ifra->ifra_lifetime.ia6t_pltime); 1066 1067 /* 1068 * backward compatibility - if IN6_IFF_DEPRECATED is set from the 1069 * userland, make it deprecated. 1070 */ 1071 if ((ifra->ifra_flags & IN6_IFF_DEPRECATED) != 0) { 1072 ia->ia6_lifetime.ia6t_pltime = 0; 1073 ia->ia6_lifetime.ia6t_preferred = time_uptime; 1074 } 1075 1076 /* 1077 * configure address flags. 1078 */ 1079 ia->ia6_flags = ifra->ifra_flags; 1080 1081 /* 1082 * Make the address tentative before joining multicast addresses, 1083 * so that corresponding MLD responses would not have a tentative 1084 * source address. 1085 */ 1086 ia->ia6_flags &= ~IN6_IFF_DUPLICATED; /* safety */ 1087 1088 /* 1089 * DAD should be performed for an new address or addresses on 1090 * an interface with ND6_IFF_IFDISABLED. 1091 */ 1092 if (in6if_do_dad(ifp) && 1093 (hostIsNew || (ifp->if_inet6->nd_flags & ND6_IFF_IFDISABLED))) 1094 ia->ia6_flags |= IN6_IFF_TENTATIVE; 1095 1096 /* notify other subsystems */ 1097 error = in6_notify_ifa(ifp, ia, ifra, hostIsNew); 1098 1099 return (error); 1100 } 1101 1102 /* 1103 * Do link-level ifa job: 1104 * 1) Add lle entry for added address 1105 * 2) Notifies routing socket users about new address 1106 * 3) join appropriate multicast group 1107 * 4) start DAD if enabled 1108 */ 1109 static int 1110 in6_broadcast_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, 1111 struct in6_ifaddr *ia, int flags) 1112 { 1113 struct in6_multi *in6m_sol; 1114 int error = 0; 1115 1116 /* Add local address to lltable, if necessary (ex. on p2p link). */ 1117 if ((error = nd6_add_ifa_lle(ia)) != 0) { 1118 in6_purgeaddr(&ia->ia_ifa); 1119 ifa_free(&ia->ia_ifa); 1120 return (error); 1121 } 1122 1123 /* Join necessary multicast groups. */ 1124 in6m_sol = NULL; 1125 if ((ifp->if_flags & IFF_MULTICAST) != 0) { 1126 error = in6_update_ifa_join_mc(ifp, ifra, ia, flags, &in6m_sol); 1127 if (error != 0) { 1128 in6_purgeaddr(&ia->ia_ifa); 1129 ifa_free(&ia->ia_ifa); 1130 return (error); 1131 } 1132 } 1133 1134 /* Perform DAD, if the address is TENTATIVE. */ 1135 if ((ia->ia6_flags & IN6_IFF_TENTATIVE)) { 1136 int delay, mindelay, maxdelay; 1137 1138 delay = 0; 1139 if ((flags & IN6_IFAUPDATE_DADDELAY)) { 1140 /* 1141 * We need to impose a delay before sending an NS 1142 * for DAD. Check if we also needed a delay for the 1143 * corresponding MLD message. If we did, the delay 1144 * should be larger than the MLD delay (this could be 1145 * relaxed a bit, but this simple logic is at least 1146 * safe). 1147 * XXX: Break data hiding guidelines and look at 1148 * state for the solicited multicast group. 1149 */ 1150 mindelay = 0; 1151 if (in6m_sol != NULL && 1152 in6m_sol->in6m_state == MLD_REPORTING_MEMBER) { 1153 mindelay = in6m_sol->in6m_timer; 1154 } 1155 maxdelay = MAX_RTR_SOLICITATION_DELAY * hz; 1156 if (maxdelay - mindelay == 0) 1157 delay = 0; 1158 else { 1159 delay = 1160 (arc4random() % (maxdelay - mindelay)) + 1161 mindelay; 1162 } 1163 } 1164 nd6_dad_start((struct ifaddr *)ia, delay); 1165 } 1166 1167 in6_newaddrmsg(ia, RTM_ADD); 1168 ifa_free(&ia->ia_ifa); 1169 return (error); 1170 } 1171 1172 /* 1173 * Adds or deletes interface route for p2p ifa. 1174 * Returns 0 on success or errno. 1175 */ 1176 static int 1177 in6_handle_dstaddr_rtrequest(int cmd, struct in6_ifaddr *ia) 1178 { 1179 struct epoch_tracker et; 1180 struct ifaddr *ifa = &ia->ia_ifa; 1181 int error; 1182 1183 /* Prepare gateway */ 1184 struct sockaddr_dl_short sdl = { 1185 .sdl_family = AF_LINK, 1186 .sdl_len = sizeof(struct sockaddr_dl_short), 1187 .sdl_type = ifa->ifa_ifp->if_type, 1188 .sdl_index = ifa->ifa_ifp->if_index, 1189 }; 1190 1191 struct sockaddr_in6 dst = { 1192 .sin6_family = AF_INET6, 1193 .sin6_len = sizeof(struct sockaddr_in6), 1194 .sin6_addr = ia->ia_dstaddr.sin6_addr, 1195 }; 1196 1197 struct rt_addrinfo info = { 1198 .rti_ifa = ifa, 1199 .rti_ifp = ifa->ifa_ifp, 1200 .rti_flags = RTF_PINNED | RTF_HOST, 1201 .rti_info = { 1202 [RTAX_DST] = (struct sockaddr *)&dst, 1203 [RTAX_GATEWAY] = (struct sockaddr *)&sdl, 1204 }, 1205 }; 1206 /* Don't set additional per-gw filters on removal */ 1207 1208 NET_EPOCH_ENTER(et); 1209 error = rib_handle_ifaddr_info(ifa->ifa_ifp->if_fib, cmd, &info); 1210 NET_EPOCH_EXIT(et); 1211 1212 return (error); 1213 } 1214 1215 static bool 1216 ifa_is_p2p(struct in6_ifaddr *ia) 1217 { 1218 int plen; 1219 1220 plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */ 1221 1222 if ((plen == 128) && (ia->ia_dstaddr.sin6_family == AF_INET6) && 1223 !IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &ia->ia_dstaddr.sin6_addr)) 1224 return (true); 1225 1226 return (false); 1227 } 1228 1229 int 1230 in6_addifaddr(struct ifnet *ifp, struct in6_aliasreq *ifra, struct in6_ifaddr *ia) 1231 { 1232 struct nd_prefixctl pr0; 1233 struct nd_prefix *pr; 1234 int carp_attached = 0; 1235 int error; 1236 1237 /* 1238 * Check if bridge wants to allow adding addrs to member interfaces. 1239 */ 1240 if (ifp->if_bridge != NULL && ifp->if_type != IFT_GIF && 1241 bridge_member_ifaddrs_p != NULL) { 1242 if (bridge_member_ifaddrs_p()) { 1243 if_printf(ifp, "WARNING: Assigning an IP address to " 1244 "an interface which is also a bridge member is " 1245 "deprecated and will be unsupported in a future " 1246 "release.\n"); 1247 } else { 1248 error = EINVAL; 1249 goto out; 1250 } 1251 } 1252 1253 /* 1254 * first, make or update the interface address structure, 1255 * and link it to the list. 1256 */ 1257 if ((error = in6_update_ifa(ifp, ifra, ia, 0)) != 0) 1258 goto out; 1259 if (ia != NULL) { 1260 if (ia->ia_ifa.ifa_carp) 1261 (*carp_detach_p)(&ia->ia_ifa, true); 1262 ifa_free(&ia->ia_ifa); 1263 } 1264 if ((ia = in6ifa_ifpwithaddr(ifp, &ifra->ifra_addr.sin6_addr)) == NULL) { 1265 /* 1266 * this can happen when the user specify the 0 valid 1267 * lifetime. 1268 */ 1269 return (0); 1270 } 1271 1272 if (ifra->ifra_vhid > 0) { 1273 if (carp_attach_p != NULL) 1274 error = (*carp_attach_p)(&ia->ia_ifa, 1275 ifra->ifra_vhid); 1276 else 1277 error = EPROTONOSUPPORT; 1278 if (error) 1279 goto out; 1280 else 1281 carp_attached = 1; 1282 } 1283 1284 /* 1285 * then, make the prefix on-link on the interface. 1286 * XXX: we'd rather create the prefix before the address, but 1287 * we need at least one address to install the corresponding 1288 * interface route, so we configure the address first. 1289 */ 1290 1291 /* 1292 * convert mask to prefix length (prefixmask has already 1293 * been validated in in6_update_ifa(). 1294 */ 1295 bzero(&pr0, sizeof(pr0)); 1296 pr0.ndpr_ifp = ifp; 1297 pr0.ndpr_plen = ia->ia_plen = 1298 in6_mask2len(&ifra->ifra_prefixmask.sin6_addr, NULL); 1299 if (pr0.ndpr_plen == 128) { 1300 /* we don't need to install a host route. */ 1301 goto aifaddr_out; 1302 } 1303 pr0.ndpr_prefix = ifra->ifra_addr; 1304 /* apply the mask for safety. */ 1305 IN6_MASK_ADDR(&pr0.ndpr_prefix.sin6_addr, 1306 &ifra->ifra_prefixmask.sin6_addr); 1307 1308 /* 1309 * XXX: since we don't have an API to set prefix (not address) 1310 * lifetimes, we just use the same lifetimes as addresses. 1311 * The (temporarily) installed lifetimes can be overridden by 1312 * later advertised RAs (when accept_rtadv is non 0), which is 1313 * an intended behavior. 1314 */ 1315 pr0.ndpr_raf_onlink = 1; /* should be configurable? */ 1316 pr0.ndpr_raf_auto = 1317 ((ifra->ifra_flags & IN6_IFF_AUTOCONF) != 0); 1318 pr0.ndpr_vltime = ifra->ifra_lifetime.ia6t_vltime; 1319 pr0.ndpr_pltime = ifra->ifra_lifetime.ia6t_pltime; 1320 1321 /* add the prefix if not yet. */ 1322 if ((pr = nd6_prefix_lookup(&pr0)) == NULL) { 1323 /* 1324 * nd6_prelist_add will install the corresponding 1325 * interface route. 1326 */ 1327 if ((error = nd6_prelist_add(&pr0, NULL, &pr)) != 0) { 1328 if (carp_attached) 1329 (*carp_detach_p)(&ia->ia_ifa, false); 1330 goto out; 1331 } 1332 } else if (pr->ndpr_raf_onlink) { 1333 time_t expiry; 1334 1335 /* 1336 * If the prefix already exists, update lifetimes, but avoid 1337 * shortening them. 1338 */ 1339 ND6_WLOCK(); 1340 expiry = in6_expire_time(pr0.ndpr_pltime); 1341 if (pr->ndpr_preferred != 0 && 1342 (pr->ndpr_preferred < expiry || expiry == 0)) { 1343 pr->ndpr_pltime = pr0.ndpr_pltime; 1344 pr->ndpr_preferred = expiry; 1345 } 1346 expiry = in6_expire_time(pr0.ndpr_vltime); 1347 if (pr->ndpr_expire != 0 && 1348 (pr->ndpr_expire < expiry || expiry == 0)) { 1349 pr->ndpr_vltime = pr0.ndpr_vltime; 1350 pr->ndpr_expire = expiry; 1351 } 1352 pr->ndpr_lastupdate = time_uptime; 1353 ND6_WUNLOCK(); 1354 } 1355 1356 /* relate the address to the prefix */ 1357 if (ia->ia6_ndpr == NULL) { 1358 ia->ia6_ndpr = pr; 1359 pr->ndpr_addrcnt++; 1360 1361 /* 1362 * If this is the first autoconf address from the 1363 * prefix, create a temporary address as well 1364 * (when required). 1365 */ 1366 if ((ia->ia6_flags & IN6_IFF_AUTOCONF) && 1367 V_ip6_use_tempaddr && pr->ndpr_addrcnt == 1) { 1368 int e; 1369 if ((e = in6_tmpifadd(ia, 1, 0)) != 0) { 1370 log(LOG_NOTICE, "in6_control: failed " 1371 "to create a temporary address, " 1372 "errno=%d\n", e); 1373 } 1374 } 1375 } 1376 nd6_prefix_rele(pr); 1377 1378 /* 1379 * this might affect the status of autoconfigured addresses, 1380 * that is, this address might make other addresses detached. 1381 */ 1382 pfxlist_onlink_check(); 1383 1384 aifaddr_out: 1385 /* 1386 * Try to clear the flag when a new IPv6 address is added 1387 * onto an IFDISABLED interface and it succeeds. 1388 */ 1389 if (ifp->if_inet6->nd_flags & ND6_IFF_IFDISABLED) { 1390 struct in6_ndireq nd; 1391 1392 memset(&nd, 0, sizeof(nd)); 1393 nd.ndi.flags = ifp->if_inet6->nd_flags; 1394 nd.ndi.flags &= ~ND6_IFF_IFDISABLED; 1395 if (nd6_ioctl(SIOCSIFINFO_FLAGS, (caddr_t)&nd, ifp) < 0) 1396 log(LOG_NOTICE, "SIOCAIFADDR_IN6: " 1397 "SIOCSIFINFO_FLAGS for -ifdisabled " 1398 "failed."); 1399 /* 1400 * Ignore failure of clearing the flag intentionally. 1401 * The failure means address duplication was detected. 1402 */ 1403 } 1404 error = 0; 1405 1406 out: 1407 if (ia != NULL) 1408 ifa_free(&ia->ia_ifa); 1409 return (error); 1410 } 1411 1412 void 1413 in6_purgeaddr(struct ifaddr *ifa) 1414 { 1415 struct ifnet *ifp = ifa->ifa_ifp; 1416 struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa; 1417 struct in6_multi_mship *imm; 1418 int error; 1419 1420 if (ifa->ifa_carp) 1421 (*carp_detach_p)(ifa, false); 1422 1423 /* 1424 * Remove the loopback route to the interface address. 1425 * The check for the current setting of "nd6_useloopback" 1426 * is not needed. 1427 */ 1428 if (ia->ia_flags & IFA_RTSELF) { 1429 error = ifa_del_loopback_route((struct ifaddr *)ia, 1430 (struct sockaddr *)&ia->ia_addr); 1431 if (error == 0) 1432 ia->ia_flags &= ~IFA_RTSELF; 1433 } 1434 1435 /* make sure there are no queued ND6 */ 1436 nd6_queue_stop(ifa); 1437 1438 /* stop DAD processing */ 1439 nd6_dad_stop(ifa); 1440 1441 /* Leave multicast groups. */ 1442 while ((imm = LIST_FIRST(&ia->ia6_memberships)) != NULL) { 1443 LIST_REMOVE(imm, i6mm_chain); 1444 if (imm->i6mm_maddr != NULL) 1445 in6_leavegroup(imm->i6mm_maddr, NULL); 1446 free(imm, M_IP6MADDR); 1447 } 1448 /* Check if we need to remove p2p route */ 1449 if ((ia->ia_flags & IFA_ROUTE) && ifa_is_p2p(ia)) { 1450 error = in6_handle_dstaddr_rtrequest(RTM_DELETE, ia); 1451 if (error != 0) 1452 log(LOG_INFO, "%s: err=%d, destination address delete " 1453 "failed\n", __func__, error); 1454 ia->ia_flags &= ~IFA_ROUTE; 1455 } 1456 1457 in6_newaddrmsg(ia, RTM_DELETE); 1458 in6_unlink_ifa(ia, ifp); 1459 } 1460 1461 /* 1462 * Removes @ia from the corresponding interfaces and unlinks corresponding 1463 * prefix if no addresses are using it anymore. 1464 */ 1465 void 1466 in6_purgeifaddr(struct in6_ifaddr *ia) 1467 { 1468 struct nd_prefix *pr; 1469 1470 /* 1471 * If the address being deleted is the only one that owns 1472 * the corresponding prefix, expire the prefix as well. 1473 * XXX: theoretically, we don't have to worry about such 1474 * relationship, since we separate the address management 1475 * and the prefix management. We do this, however, to provide 1476 * as much backward compatibility as possible in terms of 1477 * the ioctl operation. 1478 * Note that in6_purgeaddr() will decrement ndpr_addrcnt. 1479 */ 1480 pr = ia->ia6_ndpr; 1481 in6_purgeaddr(&ia->ia_ifa); 1482 if (pr != NULL && pr->ndpr_addrcnt == 0) { 1483 ND6_WLOCK(); 1484 nd6_prefix_unlink(pr, NULL); 1485 ND6_WUNLOCK(); 1486 nd6_prefix_del(pr); 1487 } 1488 } 1489 1490 1491 static void 1492 in6_unlink_ifa(struct in6_ifaddr *ia, struct ifnet *ifp) 1493 { 1494 char ip6buf[INET6_ADDRSTRLEN]; 1495 int remove_lle; 1496 1497 IF_ADDR_WLOCK(ifp); 1498 CK_STAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifaddr, ifa_link); 1499 IF_ADDR_WUNLOCK(ifp); 1500 ifa_free(&ia->ia_ifa); /* if_addrhead */ 1501 1502 /* 1503 * Defer the release of what might be the last reference to the 1504 * in6_ifaddr so that it can't be freed before the remainder of the 1505 * cleanup. 1506 */ 1507 IN6_IFADDR_WLOCK(); 1508 CK_STAILQ_REMOVE(&V_in6_ifaddrhead, ia, in6_ifaddr, ia_link); 1509 CK_LIST_REMOVE(ia, ia6_hash); 1510 IN6_IFADDR_WUNLOCK(); 1511 1512 /* 1513 * Release the reference to the base prefix. There should be a 1514 * positive reference. 1515 */ 1516 remove_lle = 0; 1517 if (ia->ia6_ndpr != NULL) { 1518 ia->ia6_ndpr->ndpr_addrcnt--; 1519 /* Do not delete lles within prefix if refcont != 0 */ 1520 if (ia->ia6_ndpr->ndpr_addrcnt == 0) 1521 remove_lle = 1; 1522 ia->ia6_ndpr = NULL; 1523 } else if (ia->ia_plen < 128) { 1524 nd6log((LOG_NOTICE, 1525 "in6_unlink_ifa: autoconf'ed address " 1526 "%s has no prefix\n", ip6_sprintf(ip6buf, IA6_IN6(ia)))); 1527 } 1528 1529 nd6_rem_ifa_lle(ia, remove_lle); 1530 1531 /* 1532 * Also, if the address being removed is autoconf'ed, call 1533 * pfxlist_onlink_check() since the release might affect the status of 1534 * other (detached) addresses. 1535 */ 1536 if ((ia->ia6_flags & IN6_IFF_AUTOCONF)) { 1537 pfxlist_onlink_check(); 1538 } 1539 ifa_free(&ia->ia_ifa); /* in6_ifaddrhead */ 1540 } 1541 1542 /* 1543 * Notifies other subsystems about address change/arrival: 1544 * 1) Notifies device handler on the first IPv6 address assignment 1545 * 2) Handle routing table changes for P2P links and route 1546 * 3) Handle routing table changes for address host route 1547 */ 1548 static int 1549 in6_notify_ifa(struct ifnet *ifp, struct in6_ifaddr *ia, 1550 struct in6_aliasreq *ifra, int hostIsNew) 1551 { 1552 int error = 0, ifacount = 0; 1553 struct ifaddr *ifa; 1554 struct sockaddr_in6 *pdst; 1555 char ip6buf[INET6_ADDRSTRLEN]; 1556 1557 /* 1558 * Give the interface a chance to initialize 1559 * if this is its first address, 1560 */ 1561 if (hostIsNew != 0) { 1562 struct epoch_tracker et; 1563 1564 NET_EPOCH_ENTER(et); 1565 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 1566 if (ifa->ifa_addr->sa_family != AF_INET6) 1567 continue; 1568 ifacount++; 1569 } 1570 NET_EPOCH_EXIT(et); 1571 } 1572 1573 if (ifacount <= 1 && ifp->if_ioctl) { 1574 error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia); 1575 if (error) 1576 goto done; 1577 } 1578 1579 /* 1580 * If a new destination address is specified, scrub the old one and 1581 * install the new destination. Note that the interface must be 1582 * p2p or loopback. 1583 */ 1584 pdst = &ifra->ifra_dstaddr; 1585 if (pdst->sin6_family == AF_INET6 && 1586 !IN6_ARE_ADDR_EQUAL(&pdst->sin6_addr, &ia->ia_dstaddr.sin6_addr)) { 1587 if ((ia->ia_flags & IFA_ROUTE) != 0 && 1588 (in6_handle_dstaddr_rtrequest(RTM_DELETE, ia) != 0)) { 1589 nd6log((LOG_ERR, "in6_update_ifa_internal: failed to " 1590 "remove a route to the old destination: %s\n", 1591 ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); 1592 /* proceed anyway... */ 1593 } else 1594 ia->ia_flags &= ~IFA_ROUTE; 1595 ia->ia_dstaddr = *pdst; 1596 } 1597 1598 /* 1599 * If a new destination address is specified for a point-to-point 1600 * interface, install a route to the destination as an interface 1601 * direct route. 1602 * XXX: the logic below rejects assigning multiple addresses on a p2p 1603 * interface that share the same destination. 1604 */ 1605 if (!(ia->ia_flags & IFA_ROUTE) && ifa_is_p2p(ia)) { 1606 error = in6_handle_dstaddr_rtrequest(RTM_ADD, ia); 1607 if (error) 1608 goto done; 1609 ia->ia_flags |= IFA_ROUTE; 1610 } 1611 1612 /* 1613 * add a loopback route to self if not exists 1614 */ 1615 if (!(ia->ia_flags & IFA_RTSELF) && V_nd6_useloopback) { 1616 error = ifa_add_loopback_route((struct ifaddr *)ia, 1617 (struct sockaddr *)&ia->ia_addr); 1618 if (error == 0) 1619 ia->ia_flags |= IFA_RTSELF; 1620 } 1621 done: 1622 WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, 1623 "Invoking IPv6 network device address event may sleep"); 1624 1625 ifa_ref(&ia->ia_ifa); 1626 EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa, 1627 IFADDR_EVENT_ADD); 1628 ifa_free(&ia->ia_ifa); 1629 1630 return (error); 1631 } 1632 1633 /* 1634 * Find an IPv6 interface link-local address specific to an interface. 1635 * ifaddr is returned referenced. 1636 */ 1637 struct in6_ifaddr * 1638 in6ifa_ifpforlinklocal(struct ifnet *ifp, int ignoreflags) 1639 { 1640 struct ifaddr *ifa; 1641 1642 NET_EPOCH_ASSERT(); 1643 1644 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 1645 if (ifa->ifa_addr->sa_family != AF_INET6) 1646 continue; 1647 if (IN6_IS_ADDR_LINKLOCAL(IFA_IN6(ifa))) { 1648 if ((((struct in6_ifaddr *)ifa)->ia6_flags & 1649 ignoreflags) != 0) 1650 continue; 1651 ifa_ref(ifa); 1652 break; 1653 } 1654 } 1655 1656 return ((struct in6_ifaddr *)ifa); 1657 } 1658 1659 /* 1660 * find the interface address corresponding to a given IPv6 address. 1661 * ifaddr is returned referenced if @referenced flag is set. 1662 */ 1663 struct in6_ifaddr * 1664 in6ifa_ifwithaddr(const struct in6_addr *addr, uint32_t zoneid, bool referenced) 1665 { 1666 struct rm_priotracker in6_ifa_tracker; 1667 struct in6_ifaddr *ia; 1668 1669 IN6_IFADDR_RLOCK(&in6_ifa_tracker); 1670 CK_LIST_FOREACH(ia, IN6ADDR_HASH(addr), ia6_hash) { 1671 if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), addr)) { 1672 if (zoneid != 0 && 1673 zoneid != ia->ia_addr.sin6_scope_id) 1674 continue; 1675 if (referenced) 1676 ifa_ref(&ia->ia_ifa); 1677 break; 1678 } 1679 } 1680 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1681 return (ia); 1682 } 1683 1684 /* 1685 * find the internet address corresponding to a given interface and address. 1686 * ifaddr is returned referenced. 1687 */ 1688 struct in6_ifaddr * 1689 in6ifa_ifpwithaddr(struct ifnet *ifp, const struct in6_addr *addr) 1690 { 1691 struct epoch_tracker et; 1692 struct ifaddr *ifa; 1693 1694 NET_EPOCH_ENTER(et); 1695 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 1696 if (ifa->ifa_addr->sa_family != AF_INET6) 1697 continue; 1698 if (IN6_ARE_ADDR_EQUAL(addr, IFA_IN6(ifa))) { 1699 ifa_ref(ifa); 1700 break; 1701 } 1702 } 1703 NET_EPOCH_EXIT(et); 1704 1705 return ((struct in6_ifaddr *)ifa); 1706 } 1707 1708 /* 1709 * Find a link-local scoped address on ifp and return it if any. 1710 */ 1711 struct in6_ifaddr * 1712 in6ifa_llaonifp(struct ifnet *ifp) 1713 { 1714 struct epoch_tracker et; 1715 struct sockaddr_in6 *sin6; 1716 struct ifaddr *ifa; 1717 1718 if (ifp->if_inet6->nd_flags & ND6_IFF_IFDISABLED) 1719 return (NULL); 1720 NET_EPOCH_ENTER(et); 1721 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 1722 if (ifa->ifa_addr->sa_family != AF_INET6) 1723 continue; 1724 sin6 = (struct sockaddr_in6 *)ifa->ifa_addr; 1725 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr) || 1726 IN6_IS_ADDR_MC_INTFACELOCAL(&sin6->sin6_addr) || 1727 IN6_IS_ADDR_MC_NODELOCAL(&sin6->sin6_addr)) 1728 break; 1729 } 1730 NET_EPOCH_EXIT(et); 1731 1732 return ((struct in6_ifaddr *)ifa); 1733 } 1734 1735 /* 1736 * Convert IP6 address to printable (loggable) representation. Caller 1737 * has to make sure that ip6buf is at least INET6_ADDRSTRLEN long. 1738 */ 1739 static char digits[] = "0123456789abcdef"; 1740 char * 1741 ip6_sprintf(char *ip6buf, const struct in6_addr *addr) 1742 { 1743 int i, cnt = 0, maxcnt = 0, idx = 0, index = 0; 1744 char *cp; 1745 const u_int16_t *a = (const u_int16_t *)addr; 1746 const u_int8_t *d; 1747 int dcolon = 0, zero = 0; 1748 1749 cp = ip6buf; 1750 1751 for (i = 0; i < 8; i++) { 1752 if (*(a + i) == 0) { 1753 cnt++; 1754 if (cnt == 1) 1755 idx = i; 1756 } 1757 else if (maxcnt < cnt) { 1758 maxcnt = cnt; 1759 index = idx; 1760 cnt = 0; 1761 } 1762 } 1763 if (maxcnt < cnt) { 1764 maxcnt = cnt; 1765 index = idx; 1766 } 1767 1768 for (i = 0; i < 8; i++) { 1769 if (dcolon == 1) { 1770 if (*a == 0) { 1771 if (i == 7) 1772 *cp++ = ':'; 1773 a++; 1774 continue; 1775 } else 1776 dcolon = 2; 1777 } 1778 if (*a == 0) { 1779 if (dcolon == 0 && *(a + 1) == 0 && i == index) { 1780 if (i == 0) 1781 *cp++ = ':'; 1782 *cp++ = ':'; 1783 dcolon = 1; 1784 } else { 1785 *cp++ = '0'; 1786 *cp++ = ':'; 1787 } 1788 a++; 1789 continue; 1790 } 1791 d = (const u_char *)a; 1792 /* Try to eliminate leading zeros in printout like in :0001. */ 1793 zero = 1; 1794 *cp = digits[*d >> 4]; 1795 if (*cp != '0') { 1796 zero = 0; 1797 cp++; 1798 } 1799 *cp = digits[*d++ & 0xf]; 1800 if (zero == 0 || (*cp != '0')) { 1801 zero = 0; 1802 cp++; 1803 } 1804 *cp = digits[*d >> 4]; 1805 if (zero == 0 || (*cp != '0')) { 1806 zero = 0; 1807 cp++; 1808 } 1809 *cp++ = digits[*d & 0xf]; 1810 *cp++ = ':'; 1811 a++; 1812 } 1813 *--cp = '\0'; 1814 return (ip6buf); 1815 } 1816 1817 int 1818 in6_localaddr(struct in6_addr *in6) 1819 { 1820 struct rm_priotracker in6_ifa_tracker; 1821 struct in6_ifaddr *ia; 1822 1823 if (IN6_IS_ADDR_LOOPBACK(in6) || IN6_IS_ADDR_LINKLOCAL(in6)) 1824 return 1; 1825 1826 IN6_IFADDR_RLOCK(&in6_ifa_tracker); 1827 CK_STAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) { 1828 if (IN6_ARE_MASKED_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr, 1829 &ia->ia_prefixmask.sin6_addr)) { 1830 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1831 return 1; 1832 } 1833 } 1834 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1835 1836 return (0); 1837 } 1838 1839 /* 1840 * Return 1 if an internet address is for the local host and configured 1841 * on one of its interfaces. 1842 */ 1843 int 1844 in6_localip(struct in6_addr *in6) 1845 { 1846 struct rm_priotracker in6_ifa_tracker; 1847 struct in6_ifaddr *ia; 1848 1849 IN6_IFADDR_RLOCK(&in6_ifa_tracker); 1850 CK_LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) { 1851 if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr)) { 1852 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1853 return (1); 1854 } 1855 } 1856 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1857 return (0); 1858 } 1859 1860 /* 1861 * Like in6_localip(), but FIB-aware and carp(4)-aware. 1862 */ 1863 bool 1864 in6_localip_fib(struct in6_addr *in6, uint16_t fib) 1865 { 1866 struct rm_priotracker in6_ifa_tracker; 1867 struct in6_ifaddr *ia; 1868 1869 IN6_IFADDR_RLOCK(&in6_ifa_tracker); 1870 CK_LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) { 1871 if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr) && 1872 (ia->ia_ifa.ifa_carp == NULL || 1873 carp_master_p(&ia->ia_ifa)) && 1874 ia->ia_ifa.ifa_ifp->if_fib == fib) { 1875 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1876 return (true); 1877 } 1878 } 1879 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1880 return (false); 1881 } 1882 1883 /* 1884 * Return 1 if an internet address is configured on an interface. 1885 */ 1886 int 1887 in6_ifhasaddr(struct ifnet *ifp, struct in6_addr *addr) 1888 { 1889 struct in6_addr in6; 1890 struct ifaddr *ifa; 1891 struct in6_ifaddr *ia6; 1892 1893 NET_EPOCH_ASSERT(); 1894 1895 in6 = *addr; 1896 if (in6_clearscope(&in6)) 1897 return (0); 1898 in6_setscope(&in6, ifp, NULL); 1899 1900 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 1901 if (ifa->ifa_addr->sa_family != AF_INET6) 1902 continue; 1903 ia6 = (struct in6_ifaddr *)ifa; 1904 if (IN6_ARE_ADDR_EQUAL(&ia6->ia_addr.sin6_addr, &in6)) 1905 return (1); 1906 } 1907 1908 return (0); 1909 } 1910 1911 int 1912 in6_is_addr_deprecated(struct sockaddr_in6 *sa6) 1913 { 1914 struct rm_priotracker in6_ifa_tracker; 1915 struct in6_ifaddr *ia; 1916 1917 IN6_IFADDR_RLOCK(&in6_ifa_tracker); 1918 CK_LIST_FOREACH(ia, IN6ADDR_HASH(&sa6->sin6_addr), ia6_hash) { 1919 if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), &sa6->sin6_addr)) { 1920 if (ia->ia6_flags & IN6_IFF_DEPRECATED) { 1921 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1922 return (1); /* true */ 1923 } 1924 break; 1925 } 1926 } 1927 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); 1928 1929 return (0); /* false */ 1930 } 1931 1932 /* 1933 * return length of part which dst and src are equal 1934 * hard coding... 1935 */ 1936 int 1937 in6_matchlen(struct in6_addr *src, struct in6_addr *dst) 1938 { 1939 int match = 0; 1940 u_char *s = (u_char *)src, *d = (u_char *)dst; 1941 u_char *lim = s + 16, r; 1942 1943 while (s < lim) 1944 if ((r = (*d++ ^ *s++)) != 0) { 1945 while (r < 128) { 1946 match++; 1947 r <<= 1; 1948 } 1949 break; 1950 } else 1951 match += 8; 1952 return match; 1953 } 1954 1955 /* XXX: to be scope conscious */ 1956 int 1957 in6_are_prefix_equal(struct in6_addr *p1, struct in6_addr *p2, int len) 1958 { 1959 int bytelen, bitlen; 1960 1961 /* sanity check */ 1962 if (0 > len || len > 128) { 1963 log(LOG_ERR, "in6_are_prefix_equal: invalid prefix length(%d)\n", 1964 len); 1965 return (0); 1966 } 1967 1968 bytelen = len / 8; 1969 bitlen = len % 8; 1970 1971 if (bcmp(&p1->s6_addr, &p2->s6_addr, bytelen)) 1972 return (0); 1973 if (bitlen != 0 && 1974 p1->s6_addr[bytelen] >> (8 - bitlen) != 1975 p2->s6_addr[bytelen] >> (8 - bitlen)) 1976 return (0); 1977 1978 return (1); 1979 } 1980 1981 void 1982 in6_prefixlen2mask(struct in6_addr *maskp, int len) 1983 { 1984 u_char maskarray[8] = {0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff}; 1985 int bytelen, bitlen, i; 1986 1987 /* sanity check */ 1988 if (0 > len || len > 128) { 1989 log(LOG_ERR, "in6_prefixlen2mask: invalid prefix length(%d)\n", 1990 len); 1991 return; 1992 } 1993 1994 bzero(maskp, sizeof(*maskp)); 1995 bytelen = len / 8; 1996 bitlen = len % 8; 1997 for (i = 0; i < bytelen; i++) 1998 maskp->s6_addr[i] = 0xff; 1999 if (bitlen) 2000 maskp->s6_addr[bytelen] = maskarray[bitlen - 1]; 2001 } 2002 2003 /* 2004 * return the best address out of the same scope. if no address was 2005 * found, return the first valid address from designated IF. 2006 */ 2007 struct in6_ifaddr * 2008 in6_ifawithifp(struct ifnet *ifp, struct in6_addr *dst) 2009 { 2010 int dst_scope = in6_addrscope(dst), blen = -1, tlen; 2011 struct ifaddr *ifa; 2012 struct in6_ifaddr *besta = NULL; 2013 struct in6_ifaddr *dep[2]; /* last-resort: deprecated */ 2014 2015 NET_EPOCH_ASSERT(); 2016 2017 dep[0] = dep[1] = NULL; 2018 2019 /* 2020 * We first look for addresses in the same scope. 2021 * If there is one, return it. 2022 * If two or more, return one which matches the dst longest. 2023 * If none, return one of global addresses assigned other ifs. 2024 */ 2025 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 2026 if (ifa->ifa_addr->sa_family != AF_INET6) 2027 continue; 2028 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST) 2029 continue; /* XXX: is there any case to allow anycast? */ 2030 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY) 2031 continue; /* don't use this interface */ 2032 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED) 2033 continue; 2034 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) { 2035 if (V_ip6_use_deprecated) 2036 dep[0] = (struct in6_ifaddr *)ifa; 2037 continue; 2038 } 2039 2040 if (dst_scope == in6_addrscope(IFA_IN6(ifa))) { 2041 /* 2042 * call in6_matchlen() as few as possible 2043 */ 2044 if (besta) { 2045 if (blen == -1) 2046 blen = in6_matchlen(&besta->ia_addr.sin6_addr, dst); 2047 tlen = in6_matchlen(IFA_IN6(ifa), dst); 2048 if (tlen > blen) { 2049 blen = tlen; 2050 besta = (struct in6_ifaddr *)ifa; 2051 } 2052 } else 2053 besta = (struct in6_ifaddr *)ifa; 2054 } 2055 } 2056 if (besta) 2057 return (besta); 2058 2059 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 2060 if (ifa->ifa_addr->sa_family != AF_INET6) 2061 continue; 2062 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST) 2063 continue; /* XXX: is there any case to allow anycast? */ 2064 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY) 2065 continue; /* don't use this interface */ 2066 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED) 2067 continue; 2068 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) { 2069 if (V_ip6_use_deprecated) 2070 dep[1] = (struct in6_ifaddr *)ifa; 2071 continue; 2072 } 2073 2074 return (struct in6_ifaddr *)ifa; 2075 } 2076 2077 /* use the last-resort values, that are, deprecated addresses */ 2078 if (dep[0]) 2079 return dep[0]; 2080 if (dep[1]) 2081 return dep[1]; 2082 2083 return NULL; 2084 } 2085 2086 /* 2087 * perform DAD when interface becomes IFF_UP. 2088 */ 2089 void 2090 in6_if_up(struct ifnet *ifp) 2091 { 2092 struct epoch_tracker et; 2093 struct ifaddr *ifa; 2094 struct in6_ifaddr *ia; 2095 2096 NET_EPOCH_ENTER(et); 2097 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 2098 if (ifa->ifa_addr->sa_family != AF_INET6) 2099 continue; 2100 ia = (struct in6_ifaddr *)ifa; 2101 if (ia->ia6_flags & IN6_IFF_TENTATIVE) { 2102 /* 2103 * The TENTATIVE flag was likely set by hand 2104 * beforehand, implicitly indicating the need for DAD. 2105 * We may be able to skip the random delay in this 2106 * case, but we impose delays just in case. 2107 */ 2108 nd6_dad_start(ifa, 2109 arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz)); 2110 } 2111 } 2112 NET_EPOCH_EXIT(et); 2113 2114 /* 2115 * special cases, like 6to4, are handled in in6_ifattach 2116 */ 2117 in6_ifattach(ifp, NULL); 2118 } 2119 2120 static void 2121 in6_ifevent(void *arg __unused, struct ifnet *ifp, int event) 2122 { 2123 if (event == IFNET_EVENT_UP) 2124 in6_if_up(ifp); 2125 } 2126 2127 static void 2128 in6_init(void *arg __unused) 2129 { 2130 EVENTHANDLER_REGISTER(ifnet_event, in6_ifevent, NULL, EVENTHANDLER_PRI_ANY); 2131 } 2132 SYSINIT(in6_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, in6_init, NULL); 2133 2134 int 2135 in6if_do_dad(struct ifnet *ifp) 2136 { 2137 2138 if ((ifp->if_flags & IFF_LOOPBACK) != 0) 2139 return (0); 2140 if ((ifp->if_flags & IFF_MULTICAST) == 0) 2141 return (0); 2142 if ((ifp->if_inet6->nd_flags & 2143 (ND6_IFF_IFDISABLED | ND6_IFF_NO_DAD)) != 0) 2144 return (0); 2145 return (1); 2146 } 2147 2148 /* 2149 * Provide the length of interface identifiers to be used for the link attached 2150 * to the given interface. The length should be defined in "IPv6 over 2151 * xxx-link" document. Note that address architecture might also define 2152 * the length for a particular set of address prefixes, regardless of the 2153 * link type. As clarified in rfc2462bis, those two definitions should be 2154 * consistent, and those really are as of August 2004. 2155 */ 2156 int 2157 in6_if2idlen(struct ifnet *ifp) 2158 { 2159 switch (ifp->if_type) { 2160 case IFT_ETHER: /* RFC2464 */ 2161 case IFT_PROPVIRTUAL: /* XXX: no RFC. treat it as ether */ 2162 case IFT_L2VLAN: /* ditto */ 2163 case IFT_BRIDGE: /* bridge(4) only does Ethernet-like links */ 2164 case IFT_INFINIBAND: 2165 return (64); 2166 case IFT_PPP: /* RFC2472 */ 2167 return (64); 2168 case IFT_FRELAY: /* RFC2590 */ 2169 return (64); 2170 case IFT_IEEE1394: /* RFC3146 */ 2171 return (64); 2172 case IFT_GIF: 2173 return (64); /* draft-ietf-v6ops-mech-v2-07 */ 2174 case IFT_LOOP: 2175 return (64); /* XXX: is this really correct? */ 2176 default: 2177 /* 2178 * Unknown link type: 2179 * It might be controversial to use the today's common constant 2180 * of 64 for these cases unconditionally. For full compliance, 2181 * we should return an error in this case. On the other hand, 2182 * if we simply miss the standard for the link type or a new 2183 * standard is defined for a new link type, the IFID length 2184 * is very likely to be the common constant. As a compromise, 2185 * we always use the constant, but make an explicit notice 2186 * indicating the "unknown" case. 2187 */ 2188 printf("in6_if2idlen: unknown link type (%d)\n", ifp->if_type); 2189 return (64); 2190 } 2191 } 2192 2193 struct in6_llentry { 2194 struct llentry base; 2195 }; 2196 2197 #define IN6_LLTBL_DEFAULT_HSIZE 32 2198 #define IN6_LLTBL_HASH(k, h) \ 2199 (((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1)) 2200 2201 /* 2202 * Do actual deallocation of @lle. 2203 */ 2204 static void 2205 in6_lltable_destroy_lle_unlocked(epoch_context_t ctx) 2206 { 2207 struct llentry *lle; 2208 2209 lle = __containerof(ctx, struct llentry, lle_epoch_ctx); 2210 LLE_LOCK_DESTROY(lle); 2211 LLE_REQ_DESTROY(lle); 2212 free(lle, M_LLTABLE); 2213 } 2214 2215 /* 2216 * Called by LLE_FREE_LOCKED when number of references 2217 * drops to zero. 2218 */ 2219 static void 2220 in6_lltable_destroy_lle(struct llentry *lle) 2221 { 2222 2223 LLE_WUNLOCK(lle); 2224 NET_EPOCH_CALL(in6_lltable_destroy_lle_unlocked, &lle->lle_epoch_ctx); 2225 } 2226 2227 static struct llentry * 2228 in6_lltable_new(const struct in6_addr *addr6, u_int flags) 2229 { 2230 struct in6_llentry *lle; 2231 2232 lle = malloc(sizeof(struct in6_llentry), M_LLTABLE, M_NOWAIT | M_ZERO); 2233 if (lle == NULL) /* NB: caller generates msg */ 2234 return NULL; 2235 2236 lle->base.r_l3addr.addr6 = *addr6; 2237 lle->base.lle_refcnt = 1; 2238 lle->base.lle_free = in6_lltable_destroy_lle; 2239 LLE_LOCK_INIT(&lle->base); 2240 LLE_REQ_INIT(&lle->base); 2241 callout_init(&lle->base.lle_timer, 1); 2242 2243 return (&lle->base); 2244 } 2245 2246 static int 2247 in6_lltable_match_prefix(const struct sockaddr *saddr, 2248 const struct sockaddr *smask, u_int flags, struct llentry *lle) 2249 { 2250 const struct in6_addr *addr, *mask, *lle_addr; 2251 2252 addr = &((const struct sockaddr_in6 *)saddr)->sin6_addr; 2253 mask = &((const struct sockaddr_in6 *)smask)->sin6_addr; 2254 lle_addr = &lle->r_l3addr.addr6; 2255 2256 if (IN6_ARE_MASKED_ADDR_EQUAL(lle_addr, addr, mask) == 0) 2257 return (0); 2258 2259 if (lle->la_flags & LLE_IFADDR) { 2260 /* 2261 * Delete LLE_IFADDR records IFF address & flag matches. 2262 * Note that addr is the interface address within prefix 2263 * being matched. 2264 */ 2265 if (IN6_ARE_ADDR_EQUAL(addr, lle_addr) && 2266 (flags & LLE_STATIC) != 0) 2267 return (1); 2268 return (0); 2269 } 2270 2271 /* flags & LLE_STATIC means deleting both dynamic and static entries */ 2272 if ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC)) 2273 return (1); 2274 2275 return (0); 2276 } 2277 2278 static void 2279 in6_lltable_free_entry(struct lltable *llt, struct llentry *lle) 2280 { 2281 2282 LLE_WLOCK_ASSERT(lle); 2283 KASSERT(llt != NULL, ("lltable is NULL")); 2284 2285 /* Unlink entry from table */ 2286 if ((lle->la_flags & LLE_LINKED) != 0) { 2287 LLTABLE_LOCK_ASSERT(llt); 2288 lltable_unlink_entry(llt, lle); 2289 } 2290 2291 llentry_free(lle); 2292 } 2293 2294 static int 2295 in6_lltable_rtcheck(struct ifnet *ifp, 2296 u_int flags, 2297 const struct sockaddr *l3addr) 2298 { 2299 const struct sockaddr_in6 *sin6; 2300 struct nhop_object *nh; 2301 struct in6_addr dst; 2302 uint32_t scopeid; 2303 char ip6buf[INET6_ADDRSTRLEN]; 2304 int fibnum; 2305 2306 NET_EPOCH_ASSERT(); 2307 KASSERT(l3addr->sa_family == AF_INET6, 2308 ("sin_family %d", l3addr->sa_family)); 2309 2310 sin6 = (const struct sockaddr_in6 *)l3addr; 2311 in6_splitscope(&sin6->sin6_addr, &dst, &scopeid); 2312 fibnum = V_rt_add_addr_allfibs ? RT_DEFAULT_FIB : ifp->if_fib; 2313 nh = fib6_lookup(fibnum, &dst, scopeid, NHR_NONE, 0); 2314 if (nh && ((nh->nh_flags & NHF_GATEWAY) || nh->nh_ifp != ifp)) { 2315 struct ifaddr *ifa; 2316 /* 2317 * Create an ND6 cache for an IPv6 neighbor 2318 * that is not covered by our own prefix. 2319 */ 2320 ifa = ifaof_ifpforaddr(l3addr, ifp); 2321 if (ifa != NULL) { 2322 return 0; 2323 } 2324 log(LOG_INFO, "IPv6 address: \"%s\" is not on the network\n", 2325 ip6_sprintf(ip6buf, &sin6->sin6_addr)); 2326 return EINVAL; 2327 } 2328 return 0; 2329 } 2330 2331 static inline uint32_t 2332 in6_lltable_hash_dst(const struct in6_addr *dst, uint32_t hsize) 2333 { 2334 2335 return (IN6_LLTBL_HASH(dst->s6_addr32[3], hsize)); 2336 } 2337 2338 static uint32_t 2339 in6_lltable_hash(const struct llentry *lle, uint32_t hsize) 2340 { 2341 2342 return (in6_lltable_hash_dst(&lle->r_l3addr.addr6, hsize)); 2343 } 2344 2345 static void 2346 in6_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa) 2347 { 2348 struct sockaddr_in6 *sin6; 2349 2350 sin6 = (struct sockaddr_in6 *)sa; 2351 bzero(sin6, sizeof(*sin6)); 2352 sin6->sin6_family = AF_INET6; 2353 sin6->sin6_len = sizeof(*sin6); 2354 sin6->sin6_addr = lle->r_l3addr.addr6; 2355 } 2356 2357 static inline struct llentry * 2358 in6_lltable_find_dst(struct lltable *llt, const struct in6_addr *dst) 2359 { 2360 struct llentry *lle; 2361 struct llentries *lleh; 2362 u_int hashidx; 2363 2364 hashidx = in6_lltable_hash_dst(dst, llt->llt_hsize); 2365 lleh = &llt->lle_head[hashidx]; 2366 CK_LIST_FOREACH(lle, lleh, lle_next) { 2367 if (lle->la_flags & LLE_DELETED) 2368 continue; 2369 if (IN6_ARE_ADDR_EQUAL(&lle->r_l3addr.addr6, dst)) 2370 break; 2371 } 2372 2373 return (lle); 2374 } 2375 2376 static void 2377 in6_lltable_delete_entry(struct lltable *llt, struct llentry *lle) 2378 { 2379 2380 lle->la_flags |= LLE_DELETED; 2381 2382 /* Leave the solicited multicast group. */ 2383 if ((lle->la_flags & LLE_PUB) != 0) 2384 in6_leave_proxy_ndp_mc(llt->llt_ifp, &lle->r_l3addr.addr6); 2385 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED); 2386 #ifdef DIAGNOSTIC 2387 log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle); 2388 #endif 2389 llentry_free(lle); 2390 } 2391 2392 static struct llentry * 2393 in6_lltable_alloc(struct lltable *llt, u_int flags, 2394 const struct sockaddr *l3addr) 2395 { 2396 const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr; 2397 struct ifnet *ifp = llt->llt_ifp; 2398 struct llentry *lle; 2399 char linkhdr[LLE_MAX_LINKHDR]; 2400 size_t linkhdrsize; 2401 int lladdr_off; 2402 2403 KASSERT(l3addr->sa_family == AF_INET6, 2404 ("sin_family %d", l3addr->sa_family)); 2405 2406 /* 2407 * A route that covers the given address must have 2408 * been installed 1st because we are doing a resolution, 2409 * verify this. 2410 */ 2411 if (!(flags & LLE_IFADDR) && 2412 in6_lltable_rtcheck(ifp, flags, l3addr) != 0) 2413 return (NULL); 2414 2415 lle = in6_lltable_new(&sin6->sin6_addr, flags); 2416 if (lle == NULL) { 2417 log(LOG_INFO, "lla_lookup: new lle malloc failed\n"); 2418 return (NULL); 2419 } 2420 lle->la_flags = flags; 2421 if ((flags & LLE_IFADDR) == LLE_IFADDR) { 2422 linkhdrsize = LLE_MAX_LINKHDR; 2423 if (lltable_calc_llheader(ifp, AF_INET6, IF_LLADDR(ifp), 2424 linkhdr, &linkhdrsize, &lladdr_off) != 0) { 2425 in6_lltable_free_entry(llt, lle); 2426 return (NULL); 2427 } 2428 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize, 2429 lladdr_off); 2430 lle->la_flags |= LLE_STATIC; 2431 } 2432 2433 if ((lle->la_flags & LLE_STATIC) != 0) 2434 lle->ln_state = ND6_LLINFO_REACHABLE; 2435 2436 return (lle); 2437 } 2438 2439 static struct llentry * 2440 in6_lltable_lookup(struct lltable *llt, u_int flags, 2441 const struct sockaddr *l3addr) 2442 { 2443 const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr; 2444 int family = flags >> 16; 2445 struct llentry *lle; 2446 2447 LLTABLE_RLOCK_ASSERT(llt); 2448 KASSERT(l3addr->sa_family == AF_INET6, 2449 ("sin_family %d", l3addr->sa_family)); 2450 KASSERT((flags & (LLE_UNLOCKED | LLE_EXCLUSIVE)) != 2451 (LLE_UNLOCKED | LLE_EXCLUSIVE), 2452 ("wrong lle request flags: %#x", flags)); 2453 2454 lle = in6_lltable_find_dst(llt, &sin6->sin6_addr); 2455 2456 if (__predict_false(family != AF_INET6)) 2457 lle = llentry_lookup_family(lle, family); 2458 2459 if (lle == NULL) 2460 return (NULL); 2461 2462 if (flags & LLE_UNLOCKED) 2463 return (lle); 2464 2465 if (flags & LLE_EXCLUSIVE) 2466 LLE_WLOCK(lle); 2467 else 2468 LLE_RLOCK(lle); 2469 2470 /* 2471 * If the lltable lock is not held, the LLE may have been unlinked while 2472 * we were blocked on the LLE lock. Check for this case. 2473 */ 2474 if (__predict_false((lle->la_flags & LLE_LINKED) == 0)) { 2475 if (flags & LLE_EXCLUSIVE) 2476 LLE_WUNLOCK(lle); 2477 else 2478 LLE_RUNLOCK(lle); 2479 return (NULL); 2480 } 2481 return (lle); 2482 } 2483 2484 static int 2485 in6_lltable_dump_entry(struct lltable *llt, struct llentry *lle, 2486 struct sysctl_req *wr) 2487 { 2488 struct ifnet *ifp = llt->llt_ifp; 2489 /* XXX stack use */ 2490 struct { 2491 struct rt_msghdr rtm; 2492 struct sockaddr_in6 sin6; 2493 /* 2494 * ndp.c assumes that sdl is word aligned 2495 */ 2496 #ifdef __LP64__ 2497 uint32_t pad; 2498 #endif 2499 struct sockaddr_dl sdl; 2500 } ndpc; 2501 struct sockaddr_dl *sdl; 2502 int error; 2503 2504 bzero(&ndpc, sizeof(ndpc)); 2505 /* skip deleted entries */ 2506 if ((lle->la_flags & LLE_DELETED) == LLE_DELETED) 2507 return (0); 2508 /* Skip if jailed and not a valid IP of the prison. */ 2509 lltable_fill_sa_entry(lle, (struct sockaddr *)&ndpc.sin6); 2510 if (prison_if(wr->td->td_ucred, (struct sockaddr *)&ndpc.sin6) != 0) 2511 return (0); 2512 /* 2513 * produce a msg made of: 2514 * struct rt_msghdr; 2515 * struct sockaddr_in6 (IPv6) 2516 * struct sockaddr_dl; 2517 */ 2518 ndpc.rtm.rtm_msglen = sizeof(ndpc); 2519 ndpc.rtm.rtm_version = RTM_VERSION; 2520 ndpc.rtm.rtm_type = RTM_GET; 2521 ndpc.rtm.rtm_flags = RTF_UP; 2522 ndpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY; 2523 sa6_recoverscope(&ndpc.sin6); 2524 2525 /* publish */ 2526 if (lle->la_flags & LLE_PUB) 2527 ndpc.rtm.rtm_flags |= RTF_ANNOUNCE; 2528 2529 sdl = &ndpc.sdl; 2530 sdl->sdl_family = AF_LINK; 2531 sdl->sdl_len = sizeof(*sdl); 2532 sdl->sdl_index = ifp->if_index; 2533 sdl->sdl_type = ifp->if_type; 2534 if ((lle->la_flags & LLE_VALID) == LLE_VALID) { 2535 sdl->sdl_alen = ifp->if_addrlen; 2536 bcopy(lle->ll_addr, LLADDR(sdl), ifp->if_addrlen); 2537 } else { 2538 sdl->sdl_alen = 0; 2539 bzero(LLADDR(sdl), ifp->if_addrlen); 2540 } 2541 if (lle->la_expire != 0) 2542 ndpc.rtm.rtm_rmx.rmx_expire = lle->la_expire + 2543 lle->lle_remtime / hz + time_second - time_uptime; 2544 ndpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA); 2545 if (lle->la_flags & LLE_STATIC) 2546 ndpc.rtm.rtm_flags |= RTF_STATIC; 2547 if (lle->la_flags & LLE_IFADDR) 2548 ndpc.rtm.rtm_flags |= RTF_PINNED; 2549 if (lle->ln_router != 0) 2550 ndpc.rtm.rtm_flags |= RTF_GATEWAY; 2551 ndpc.rtm.rtm_rmx.rmx_pksent = lle->la_asked; 2552 /* Store state in rmx_weight value */ 2553 ndpc.rtm.rtm_rmx.rmx_state = lle->ln_state; 2554 ndpc.rtm.rtm_index = ifp->if_index; 2555 error = SYSCTL_OUT(wr, &ndpc, sizeof(ndpc)); 2556 2557 return (error); 2558 } 2559 2560 static void 2561 in6_lltable_post_resolved(struct lltable *llt, struct llentry *lle) 2562 { 2563 /* Join the solicited multicast group for dst. */ 2564 if ((lle->la_flags & LLE_PUB) == LLE_PUB) 2565 in6_join_proxy_ndp_mc(llt->llt_ifp, &lle->r_l3addr.addr6); 2566 } 2567 2568 static struct lltable * 2569 in6_lltattach(struct ifnet *ifp) 2570 { 2571 struct lltable *llt; 2572 2573 llt = lltable_allocate_htbl(IN6_LLTBL_DEFAULT_HSIZE); 2574 llt->llt_af = AF_INET6; 2575 llt->llt_ifp = ifp; 2576 2577 llt->llt_lookup = in6_lltable_lookup; 2578 llt->llt_alloc_entry = in6_lltable_alloc; 2579 llt->llt_delete_entry = in6_lltable_delete_entry; 2580 llt->llt_dump_entry = in6_lltable_dump_entry; 2581 llt->llt_hash = in6_lltable_hash; 2582 llt->llt_fill_sa_entry = in6_lltable_fill_sa_entry; 2583 llt->llt_free_entry = in6_lltable_free_entry; 2584 llt->llt_match_prefix = in6_lltable_match_prefix; 2585 llt->llt_mark_used = llentry_mark_used; 2586 llt->llt_post_resolved = in6_lltable_post_resolved; 2587 lltable_link(llt); 2588 2589 return (llt); 2590 } 2591 2592 struct lltable * 2593 in6_lltable_get(struct ifnet *ifp) 2594 { 2595 if (ifp->if_inet6 == NULL) 2596 return (NULL); 2597 2598 return (ifp->if_inet6->lltable); 2599 } 2600 2601 void 2602 in6_ifarrival(void *arg __unused, struct ifnet *ifp) 2603 { 2604 struct in6_ifextra *ext; 2605 2606 /* There are not IPv6-capable interfaces. */ 2607 switch (ifp->if_type) { 2608 case IFT_PFLOG: 2609 case IFT_PFSYNC: 2610 ifp->if_inet6 = NULL; 2611 return; 2612 } 2613 ext = ifp->if_inet6 = malloc(sizeof(*ext), M_IFADDR, M_WAITOK | M_ZERO); 2614 COUNTER_ARRAY_ALLOC(ext->in6_ifstat, 2615 sizeof(struct in6_ifstat) / sizeof(uint64_t), M_WAITOK); 2616 COUNTER_ARRAY_ALLOC(ext->icmp6_ifstat, 2617 sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_WAITOK); 2618 nd6_ifattach(ifp); 2619 mld_domifattach(ifp); 2620 scope6_ifattach(ifp); 2621 2622 ext->lltable = in6_lltattach(ifp); 2623 } 2624 EVENTHANDLER_DEFINE(ifnet_arrival_event, in6_ifarrival, NULL, 2625 EVENTHANDLER_PRI_ANY); 2626 2627 uint32_t 2628 in6_ifmtu(const struct ifnet *ifp) 2629 { 2630 const uint32_t 2631 linkmtu = ifp->if_inet6->nd_linkmtu, 2632 maxmtu = ifp->if_inet6->nd_maxmtu, 2633 ifmtu = ifp->if_mtu; 2634 2635 if (linkmtu > 0 && linkmtu < ifmtu) 2636 return (linkmtu); 2637 if (maxmtu > 0 && maxmtu < ifmtu) 2638 return (maxmtu); 2639 return (ifmtu); 2640 } 2641 2642 /* 2643 * Convert sockaddr_in6 to sockaddr_in. Original sockaddr_in6 must be 2644 * v4 mapped addr or v4 compat addr 2645 */ 2646 void 2647 in6_sin6_2_sin(struct sockaddr_in *sin, const struct sockaddr_in6 *sin6) 2648 { 2649 2650 bzero(sin, sizeof(*sin)); 2651 sin->sin_len = sizeof(struct sockaddr_in); 2652 sin->sin_family = AF_INET; 2653 sin->sin_port = sin6->sin6_port; 2654 sin->sin_addr.s_addr = sin6->sin6_addr.s6_addr32[3]; 2655 } 2656 2657 /* Convert sockaddr_in to sockaddr_in6 in v4 mapped addr format. */ 2658 void 2659 in6_sin_2_v4mapsin6(const struct sockaddr_in *sin, struct sockaddr_in6 *sin6) 2660 { 2661 bzero(sin6, sizeof(*sin6)); 2662 sin6->sin6_len = sizeof(struct sockaddr_in6); 2663 sin6->sin6_family = AF_INET6; 2664 sin6->sin6_port = sin->sin_port; 2665 sin6->sin6_addr.s6_addr32[0] = 0; 2666 sin6->sin6_addr.s6_addr32[1] = 0; 2667 sin6->sin6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP; 2668 sin6->sin6_addr.s6_addr32[3] = sin->sin_addr.s_addr; 2669 } 2670 2671 /* Convert sockaddr_in6 into sockaddr_in. */ 2672 void 2673 in6_sin6_2_sin_in_sock(struct sockaddr *nam) 2674 { 2675 struct sockaddr_in *sin_p; 2676 struct sockaddr_in6 sin6; 2677 2678 /* 2679 * Save original sockaddr_in6 addr and convert it 2680 * to sockaddr_in. 2681 */ 2682 sin6 = *(struct sockaddr_in6 *)nam; 2683 sin_p = (struct sockaddr_in *)nam; 2684 in6_sin6_2_sin(sin_p, &sin6); 2685 } 2686 2687 /* 2688 * Join/leave the solicited multicast groups for proxy NDP entries. 2689 */ 2690 static void 2691 in6_join_proxy_ndp_mc(struct ifnet *ifp, const struct in6_addr *dst) 2692 { 2693 struct in6_multi *inm; 2694 struct in6_addr mltaddr; 2695 char ip6buf[INET6_ADDRSTRLEN]; 2696 int error; 2697 2698 if (in6_solicited_node_maddr(&mltaddr, ifp, dst) != 0) 2699 return; /* error logged in in6_solicited_node_maddr. */ 2700 2701 error = in6_joingroup(ifp, &mltaddr, NULL, &inm, 0); 2702 if (error != 0) { 2703 nd6log((LOG_WARNING, 2704 "%s: in6_joingroup failed for %s on %s (errno=%d)\n", 2705 __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp), 2706 error)); 2707 } 2708 } 2709 2710 static void 2711 in6_leave_proxy_ndp_mc(struct ifnet *ifp, const struct in6_addr *dst) 2712 { 2713 struct epoch_tracker et; 2714 struct in6_multi *inm; 2715 struct in6_addr mltaddr; 2716 char ip6buf[INET6_ADDRSTRLEN]; 2717 2718 if (in6_solicited_node_maddr(&mltaddr, ifp, dst) != 0) 2719 return; /* error logged in in6_solicited_node_maddr. */ 2720 2721 NET_EPOCH_ENTER(et); 2722 inm = in6m_lookup(ifp, &mltaddr); 2723 NET_EPOCH_EXIT(et); 2724 if (inm != NULL) 2725 in6_leavegroup(inm, NULL); 2726 else 2727 nd6log((LOG_WARNING, "%s: in6m_lookup failed for %s on %s\n", 2728 __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp))); 2729 } 2730 2731 static bool 2732 in6_lle_match_pub(struct lltable *llt, struct llentry *lle, void *farg) 2733 { 2734 return ((lle->la_flags & LLE_PUB) != 0); 2735 } 2736 2737 void 2738 in6_purge_proxy_ndp(struct ifnet *ifp) 2739 { 2740 struct lltable *llt; 2741 bool need_purge; 2742 2743 if (ifp->if_inet6 == NULL) 2744 return; 2745 2746 llt = LLTABLE6(ifp); 2747 LLTABLE_LOCK(llt); 2748 need_purge = ((llt->llt_flags & LLT_ADDEDPROXY) != 0); 2749 LLTABLE_UNLOCK(llt); 2750 2751 /* 2752 * Ever added proxy ndp entries, leave solicited node multicast 2753 * before deleting the llentry. 2754 */ 2755 if (need_purge) 2756 lltable_delete_conditional(llt, in6_lle_match_pub, NULL); 2757 } 2758