1 /* 2 * Copyright 2007 Sun Microsystems, Inc. All rights reserved. 3 * Use is subject to license terms. 4 */ 5 6 /* 7 * Copyright (c) 1988, 1991, 1993 8 * The Regents of the University of California. All rights reserved. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. All advertising materials mentioning features or use of this software 19 * must display the following acknowledgement: 20 * This product includes software developed by the University of 21 * California, Berkeley and its contributors. 22 * 4. Neither the name of the University nor the names of its contributors 23 * may be used to endorse or promote products derived from this software 24 * without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 36 * SUCH DAMAGE. 37 * 38 * @(#)rtsock.c 8.6 (Berkeley) 2/11/95 39 */ 40 41 #pragma ident "%Z%%M% %I% %E% SMI" 42 43 /* 44 * This file contains routines that processes routing socket requests. 45 */ 46 47 #include <sys/types.h> 48 #include <sys/stream.h> 49 #include <sys/stropts.h> 50 #include <sys/ddi.h> 51 #include <sys/cmn_err.h> 52 #include <sys/debug.h> 53 #include <sys/policy.h> 54 #include <sys/zone.h> 55 56 #include <sys/systm.h> 57 #include <sys/param.h> 58 #include <sys/socket.h> 59 #include <sys/strsun.h> 60 #include <net/if.h> 61 #include <net/route.h> 62 #include <netinet/in.h> 63 #include <net/if_dl.h> 64 #include <netinet/ip6.h> 65 66 #include <inet/common.h> 67 #include <inet/ip.h> 68 #include <inet/ip6.h> 69 #include <inet/ip_if.h> 70 #include <inet/ip_ire.h> 71 #include <inet/ip_ftable.h> 72 #include <inet/ip_rts.h> 73 74 #include <inet/ipclassifier.h> 75 76 #include <sys/tsol/tndb.h> 77 #include <sys/tsol/tnet.h> 78 79 #define RTS_MSG_SIZE(type, rtm_addrs, af, sacnt) \ 80 (rts_data_msg_size(rtm_addrs, af, sacnt) + rts_header_msg_size(type)) 81 82 static size_t rts_copyfromsockaddr(struct sockaddr *sa, in6_addr_t *addrp); 83 static void rts_fill_msg(int type, int rtm_addrs, ipaddr_t dst, 84 ipaddr_t mask, ipaddr_t gateway, ipaddr_t src_addr, ipaddr_t brd_addr, 85 ipaddr_t author, const ipif_t *ipif, mblk_t *mp, uint_t, const tsol_gc_t *); 86 static int rts_getaddrs(rt_msghdr_t *rtm, in6_addr_t *dst_addrp, 87 in6_addr_t *gw_addrp, in6_addr_t *net_maskp, in6_addr_t *authorp, 88 in6_addr_t *if_addrp, in6_addr_t *src_addrp, ushort_t *indexp, 89 sa_family_t *afp, tsol_rtsecattr_t *rtsecattr, int *error); 90 static void rts_getifdata(if_data_t *if_data, const ipif_t *ipif); 91 static int rts_getmetrics(ire_t *ire, rt_metrics_t *metrics); 92 static mblk_t *rts_rtmget(mblk_t *mp, ire_t *ire, ire_t *sire, 93 sa_family_t af); 94 static void rts_setmetrics(ire_t *ire, uint_t which, rt_metrics_t *metrics); 95 static void ip_rts_request_retry(ipsq_t *, queue_t *q, mblk_t *mp, void *); 96 97 /* 98 * Send the ack to all the routing queues. In case of the originating queue, 99 * send it only if the loopback is set. 100 * 101 * Messages are sent upstream only on routing sockets that did not specify an 102 * address family when they were created or when the address family matches the 103 * one specified by the caller. 104 * 105 */ 106 void 107 rts_queue_input(mblk_t *mp, queue_t *q, sa_family_t af, ip_stack_t *ipst) 108 { 109 mblk_t *mp1; 110 int checkqfull; 111 conn_t *connp, *next_connp; 112 113 mutex_enter(&ipst->ips_rts_clients->connf_lock); 114 connp = ipst->ips_rts_clients->connf_head; 115 116 while (connp != NULL) { 117 /* 118 * If there was a family specified when this routing socket was 119 * created and it doesn't match the family of the message to 120 * copy, then continue. 121 */ 122 if ((connp->conn_proto != AF_UNSPEC) && 123 (connp->conn_proto != af)) { 124 connp = connp->conn_next; 125 continue; 126 } 127 /* 128 * For the originating queue, we only copy the message upstream 129 * if loopback is set. For others reading on the routing 130 * socket, we check if there is room upstream for a copy of the 131 * message. 132 */ 133 if ((q != NULL) && (CONNP_TO_RQ(connp) == RD(q))) { 134 if (connp->conn_loopback == 0) { 135 connp = connp->conn_next; 136 continue; 137 } 138 /* 139 * Just because it is the same queue doesn't mean it 140 * will promptly read its acks. Have to avoid using 141 * all of kernel memory. 142 */ 143 checkqfull = B_TRUE; 144 } else { 145 checkqfull = B_TRUE; 146 } 147 CONN_INC_REF(connp); 148 mutex_exit(&ipst->ips_rts_clients->connf_lock); 149 /* Pass to rts_input */ 150 if (!checkqfull || canputnext(CONNP_TO_RQ(connp))) { 151 mp1 = dupmsg(mp); 152 if (mp1 == NULL) 153 mp1 = copymsg(mp); 154 if (mp1 != NULL) 155 (connp->conn_recv)(connp, mp1, NULL); 156 } 157 158 mutex_enter(&ipst->ips_rts_clients->connf_lock); 159 /* Follow the next pointer before releasing the conn. */ 160 next_connp = connp->conn_next; 161 CONN_DEC_REF(connp); 162 connp = next_connp; 163 } 164 mutex_exit(&ipst->ips_rts_clients->connf_lock); 165 freemsg(mp); 166 } 167 168 /* 169 * Takes an ire and sends an ack to all the routing sockets. This 170 * routine is used 171 * - when a route is created/deleted through the ioctl interface. 172 * - when ire_expire deletes a stale redirect 173 */ 174 void 175 ip_rts_rtmsg(int type, ire_t *ire, int error, ip_stack_t *ipst) 176 { 177 mblk_t *mp; 178 rt_msghdr_t *rtm; 179 int rtm_addrs = (RTA_DST | RTA_NETMASK | RTA_GATEWAY); 180 sa_family_t af; 181 in6_addr_t gw_addr_v6; 182 183 if (ire == NULL) 184 return; 185 ASSERT(ire->ire_ipversion == IPV4_VERSION || 186 ire->ire_ipversion == IPV6_VERSION); 187 188 if (ire->ire_flags & RTF_SETSRC) 189 rtm_addrs |= RTA_SRC; 190 191 switch (ire->ire_ipversion) { 192 case IPV4_VERSION: 193 af = AF_INET; 194 mp = rts_alloc_msg(type, rtm_addrs, af, 0); 195 if (mp == NULL) 196 return; 197 rts_fill_msg(type, rtm_addrs, ire->ire_addr, ire->ire_mask, 198 ire->ire_gateway_addr, ire->ire_src_addr, 0, 0, NULL, mp, 199 0, NULL); 200 break; 201 case IPV6_VERSION: 202 af = AF_INET6; 203 mp = rts_alloc_msg(type, rtm_addrs, af, 0); 204 if (mp == NULL) 205 return; 206 mutex_enter(&ire->ire_lock); 207 gw_addr_v6 = ire->ire_gateway_addr_v6; 208 mutex_exit(&ire->ire_lock); 209 rts_fill_msg_v6(type, rtm_addrs, &ire->ire_addr_v6, 210 &ire->ire_mask_v6, &gw_addr_v6, 211 &ire->ire_src_addr_v6, &ipv6_all_zeros, &ipv6_all_zeros, 212 NULL, mp, 0, NULL); 213 break; 214 } 215 rtm = (rt_msghdr_t *)mp->b_rptr; 216 mp->b_wptr = (uchar_t *)&mp->b_rptr[rtm->rtm_msglen]; 217 rtm->rtm_addrs = rtm_addrs; 218 rtm->rtm_flags = ire->ire_flags; 219 if (error != 0) 220 rtm->rtm_errno = error; 221 else 222 rtm->rtm_flags |= RTF_DONE; 223 rts_queue_input(mp, NULL, af, ipst); 224 } 225 226 /* ARGSUSED */ 227 static void 228 ip_rts_request_retry(ipsq_t *dummy_sq, queue_t *q, mblk_t *mp, void *dummy) 229 { 230 (void) ip_rts_request(q, mp, DB_CRED(mp)); 231 } 232 233 /* 234 * This is a call from the RTS module 235 * indicating that this is a Routing Socket 236 * Stream. Insert this conn_t in routing 237 * socket client list. 238 */ 239 void 240 ip_rts_register(conn_t *connp) 241 { 242 ip_stack_t *ipst = connp->conn_netstack->netstack_ip; 243 244 connp->conn_loopback = 1; 245 ipcl_hash_insert_wildcard(ipst->ips_rts_clients, connp); 246 } 247 248 /* 249 * This is a call from the RTS module indicating that it is closing. 250 */ 251 void 252 ip_rts_unregister(conn_t *connp) 253 { 254 ipcl_hash_remove(connp); 255 } 256 257 /* 258 * Processes requests received on a routing socket. It extracts all the 259 * arguments and calls the appropriate function to process the request. 260 * 261 * RTA_SRC bit flag requests are sent by 'route -setsrc'. 262 * 263 * In general, this function does not consume the message supplied but rather 264 * sends the message upstream with an appropriate UNIX errno. 265 * 266 * We may need to restart this operation if the ipif cannot be looked up 267 * due to an exclusive operation that is currently in progress. The restart 268 * entry point is ip_rts_request_retry. While the request is enqueud in the 269 * ipsq the ioctl could be aborted and the conn close. To ensure that we don't 270 * have stale conn pointers, ip_wput_ioctl does a conn refhold. This is 271 * released at the completion of the rts ioctl at the end of this function 272 * by calling CONN_OPER_PENDING_DONE or when the ioctl is aborted and 273 * conn close occurs in conn_ioctl_cleanup. 274 */ 275 int 276 ip_rts_request(queue_t *q, mblk_t *mp, cred_t *ioc_cr) 277 { 278 rt_msghdr_t *rtm = NULL; 279 in6_addr_t dst_addr_v6; 280 in6_addr_t src_addr_v6; 281 in6_addr_t gw_addr_v6; 282 in6_addr_t net_mask_v6; 283 in6_addr_t author_v6; 284 in6_addr_t if_addr_v6; 285 mblk_t *mp1, *ioc_mp = mp; 286 ire_t *ire = NULL; 287 ire_t *sire = NULL; 288 int error = 0; 289 int match_flags = MATCH_IRE_DSTONLY; 290 int match_flags_local = MATCH_IRE_TYPE | MATCH_IRE_GW; 291 int found_addrs; 292 sa_family_t af; 293 ipaddr_t dst_addr; 294 ipaddr_t gw_addr; 295 ipaddr_t src_addr; 296 ipaddr_t net_mask; 297 ushort_t index; 298 ipif_t *ipif = NULL; 299 ipif_t *tmp_ipif = NULL; 300 IOCP iocp = (IOCP)mp->b_rptr; 301 conn_t *connp; 302 boolean_t gcgrp_xtraref = B_FALSE; 303 tsol_gcgrp_addr_t ga; 304 tsol_rtsecattr_t rtsecattr; 305 struct rtsa_s *rtsap = NULL; 306 tsol_gcgrp_t *gcgrp = NULL; 307 tsol_gc_t *gc = NULL; 308 ts_label_t *tsl = NULL; 309 zoneid_t zoneid; 310 ip_stack_t *ipst; 311 312 ip1dbg(("ip_rts_request: mp is %x\n", DB_TYPE(mp))); 313 314 ASSERT(CONN_Q(q)); 315 connp = Q_TO_CONN(q); 316 zoneid = connp->conn_zoneid; 317 ipst = connp->conn_netstack->netstack_ip; 318 319 ASSERT(mp->b_cont != NULL); 320 /* ioc_mp holds mp */ 321 mp = mp->b_cont; 322 323 /* 324 * The Routing Socket data starts on 325 * next block. If there is no next block 326 * this is an indication from routing module 327 * that it is a routing socket stream queue. 328 * We need to support that for compatibility with SDP since 329 * it has a contract private interface to use IP_IOC_RTS_REQUEST. 330 */ 331 if (mp->b_cont == NULL) { 332 /* 333 * This is a message from SDP 334 * indicating that this is a Routing Socket 335 * Stream. Insert this conn_t in routing 336 * socket client list. 337 */ 338 connp->conn_loopback = 1; 339 ipcl_hash_insert_wildcard(ipst->ips_rts_clients, connp); 340 goto done; 341 } 342 mp1 = dupmsg(mp->b_cont); 343 if (mp1 == NULL) { 344 error = ENOBUFS; 345 goto done; 346 } 347 mp = mp1; 348 349 if (mp->b_cont != NULL && !pullupmsg(mp, -1)) { 350 freemsg(mp); 351 error = EINVAL; 352 goto done; 353 } 354 if ((mp->b_wptr - mp->b_rptr) < sizeof (rt_msghdr_t)) { 355 freemsg(mp); 356 error = EINVAL; 357 goto done; 358 } 359 360 /* 361 * Check the routing message for basic consistency including the 362 * version number and that the number of octets written is the same 363 * as specified by the rtm_msglen field. 364 * 365 * At this point, an error can be delivered back via rtm_errno. 366 */ 367 rtm = (rt_msghdr_t *)mp->b_rptr; 368 if ((mp->b_wptr - mp->b_rptr) != rtm->rtm_msglen) { 369 error = EINVAL; 370 goto done; 371 } 372 if (rtm->rtm_version != RTM_VERSION) { 373 error = EPROTONOSUPPORT; 374 goto done; 375 } 376 377 /* Only allow RTM_GET or RTM_RESOLVE for unprivileged process */ 378 if (rtm->rtm_type != RTM_GET && 379 rtm->rtm_type != RTM_RESOLVE && 380 (ioc_cr == NULL || 381 secpolicy_ip_config(ioc_cr, B_FALSE) != 0)) { 382 error = EPERM; 383 goto done; 384 } 385 386 found_addrs = rts_getaddrs(rtm, &dst_addr_v6, &gw_addr_v6, &net_mask_v6, 387 &author_v6, &if_addr_v6, &src_addr_v6, &index, &af, &rtsecattr, 388 &error); 389 390 if (error != 0) 391 goto done; 392 393 if ((found_addrs & RTA_DST) == 0) { 394 error = EINVAL; 395 goto done; 396 } 397 398 /* 399 * Based on the address family of the destination address, determine 400 * the destination, gateway and netmask and return the appropriate error 401 * if an unknown address family was specified (following the errno 402 * values that 4.4BSD-Lite2 returns.) 403 */ 404 switch (af) { 405 case AF_INET: 406 IN6_V4MAPPED_TO_IPADDR(&dst_addr_v6, dst_addr); 407 IN6_V4MAPPED_TO_IPADDR(&src_addr_v6, src_addr); 408 IN6_V4MAPPED_TO_IPADDR(&gw_addr_v6, gw_addr); 409 if (((found_addrs & RTA_NETMASK) == 0) || 410 (rtm->rtm_flags & RTF_HOST)) 411 net_mask = IP_HOST_MASK; 412 else 413 IN6_V4MAPPED_TO_IPADDR(&net_mask_v6, net_mask); 414 break; 415 case AF_INET6: 416 if (((found_addrs & RTA_NETMASK) == 0) || 417 (rtm->rtm_flags & RTF_HOST)) 418 net_mask_v6 = ipv6_all_ones; 419 break; 420 default: 421 /* 422 * These errno values are meant to be compatible with 423 * 4.4BSD-Lite2 for the given message types. 424 */ 425 switch (rtm->rtm_type) { 426 case RTM_ADD: 427 case RTM_DELETE: 428 error = ESRCH; 429 goto done; 430 case RTM_GET: 431 case RTM_CHANGE: 432 error = EAFNOSUPPORT; 433 goto done; 434 default: 435 error = EOPNOTSUPP; 436 goto done; 437 } 438 } 439 440 /* 441 * At this point, the address family must be something known. 442 */ 443 ASSERT(af == AF_INET || af == AF_INET6); 444 445 if (index != 0) { 446 ill_t *ill; 447 448 /* 449 * IPC must be refheld somewhere in ip_wput_nondata or 450 * ip_wput_ioctl etc... and cleaned up if ioctl is killed. 451 * If ILL_CHANGING the request is queued in the ipsq. 452 */ 453 ill = ill_lookup_on_ifindex(index, af == AF_INET6, 454 CONNP_TO_WQ(connp), ioc_mp, ip_rts_request_retry, &error, 455 ipst); 456 if (ill == NULL) { 457 if (error != EINPROGRESS) 458 error = EINVAL; 459 goto done; 460 } 461 462 ipif = ipif_get_next_ipif(NULL, ill); 463 ill_refrele(ill); 464 /* 465 * If this is replacement ipif, prevent a route from 466 * being added. 467 */ 468 if (ipif != NULL && ipif->ipif_replace_zero) { 469 error = ENETDOWN; 470 goto done; 471 } 472 match_flags |= MATCH_IRE_ILL; 473 } 474 475 /* 476 * If a netmask was supplied in the message, then subsequent route 477 * lookups will attempt to match on the netmask as well. 478 */ 479 if ((found_addrs & RTA_NETMASK) != 0) 480 match_flags |= MATCH_IRE_MASK; 481 482 /* 483 * We only process any passed-in route security attributes for 484 * either RTM_ADD or RTM_CHANGE message; We overload them 485 * to do an RTM_GET as a different label; ignore otherwise. 486 */ 487 if (rtm->rtm_type == RTM_ADD || rtm->rtm_type == RTM_CHANGE || 488 rtm->rtm_type == RTM_GET) { 489 ASSERT(rtsecattr.rtsa_cnt <= TSOL_RTSA_REQUEST_MAX); 490 if (rtsecattr.rtsa_cnt > 0) 491 rtsap = &rtsecattr.rtsa_attr[0]; 492 } 493 494 switch (rtm->rtm_type) { 495 case RTM_ADD: 496 /* if we are adding a route, gateway is a must */ 497 if ((found_addrs & RTA_GATEWAY) == 0) { 498 error = EINVAL; 499 goto done; 500 } 501 502 /* Multirouting does not support net routes. */ 503 if ((rtm->rtm_flags & (RTF_MULTIRT | RTF_HOST)) == 504 RTF_MULTIRT) { 505 error = EADDRNOTAVAIL; 506 goto done; 507 } 508 509 /* 510 * Multirouting and user-specified source addresses 511 * do not support interface based routing. 512 * Assigning a source address to an interface based 513 * route is achievable by plumbing a new ipif and 514 * setting up the interface route via this ipif, 515 * though. 516 */ 517 if (rtm->rtm_flags & (RTF_MULTIRT | RTF_SETSRC)) { 518 if ((rtm->rtm_flags & RTF_GATEWAY) == 0) { 519 error = EADDRNOTAVAIL; 520 goto done; 521 } 522 } 523 524 switch (af) { 525 case AF_INET: 526 if (src_addr != INADDR_ANY) { 527 /* 528 * The RTF_SETSRC flag is present, check that 529 * the supplied src address is not the loopback 530 * address. This would produce martian packets. 531 */ 532 if (src_addr == htonl(INADDR_LOOPBACK)) { 533 error = EINVAL; 534 goto done; 535 } 536 /* 537 * Also check that the supplied address is a 538 * valid, local one. 539 */ 540 tmp_ipif = ipif_lookup_addr(src_addr, NULL, 541 ALL_ZONES, CONNP_TO_WQ(connp), ioc_mp, 542 ip_rts_request_retry, &error, ipst); 543 if (tmp_ipif == NULL) { 544 if (error != EINPROGRESS) 545 error = EADDRNOTAVAIL; 546 goto done; 547 } 548 if (!(tmp_ipif->ipif_flags & IPIF_UP) || 549 (tmp_ipif->ipif_flags & 550 (IPIF_NOLOCAL | IPIF_ANYCAST))) { 551 error = EINVAL; 552 goto done; 553 } 554 } else { 555 /* 556 * The RTF_SETSRC modifier must be associated 557 * to a non-null source address. 558 */ 559 if (rtm->rtm_flags & RTF_SETSRC) { 560 error = EINVAL; 561 goto done; 562 } 563 } 564 565 error = ip_rt_add(dst_addr, net_mask, gw_addr, src_addr, 566 rtm->rtm_flags, ipif, &ire, B_FALSE, 567 CONNP_TO_WQ(connp), ioc_mp, ip_rts_request_retry, 568 rtsap, ipst); 569 if (ipif != NULL) 570 ASSERT(!MUTEX_HELD(&ipif->ipif_ill->ill_lock)); 571 break; 572 case AF_INET6: 573 if (!IN6_IS_ADDR_UNSPECIFIED(&src_addr_v6)) { 574 /* 575 * The RTF_SETSRC flag is present, check that 576 * the supplied src address is not the loopback 577 * address. This would produce martian packets. 578 */ 579 if (IN6_IS_ADDR_LOOPBACK(&src_addr_v6)) { 580 error = EINVAL; 581 goto done; 582 } 583 /* 584 * Also check that the supplied address is a 585 * valid, local one. 586 */ 587 tmp_ipif = ipif_lookup_addr_v6(&src_addr_v6, 588 NULL, ALL_ZONES, CONNP_TO_WQ(connp), ioc_mp, 589 ip_rts_request_retry, &error, ipst); 590 if (tmp_ipif == NULL) { 591 if (error != EINPROGRESS) 592 error = EADDRNOTAVAIL; 593 goto done; 594 } 595 596 if (!(tmp_ipif->ipif_flags & IPIF_UP) || 597 (tmp_ipif->ipif_flags & 598 (IPIF_NOLOCAL | IPIF_ANYCAST))) { 599 error = EINVAL; 600 goto done; 601 } 602 603 error = ip_rt_add_v6(&dst_addr_v6, &net_mask_v6, 604 &gw_addr_v6, &src_addr_v6, rtm->rtm_flags, 605 ipif, &ire, CONNP_TO_WQ(connp), ioc_mp, 606 ip_rts_request_retry, rtsap, ipst); 607 break; 608 } 609 /* 610 * The RTF_SETSRC modifier must be associated 611 * to a non-null source address. 612 */ 613 if (rtm->rtm_flags & RTF_SETSRC) { 614 error = EINVAL; 615 goto done; 616 } 617 error = ip_rt_add_v6(&dst_addr_v6, &net_mask_v6, 618 &gw_addr_v6, NULL, rtm->rtm_flags, 619 ipif, &ire, CONNP_TO_WQ(connp), ioc_mp, 620 ip_rts_request_retry, rtsap, ipst); 621 if (ipif != NULL) 622 ASSERT(!MUTEX_HELD(&ipif->ipif_ill->ill_lock)); 623 break; 624 } 625 if (error != 0) 626 goto done; 627 ASSERT(ire != NULL); 628 rts_setmetrics(ire, rtm->rtm_inits, &rtm->rtm_rmx); 629 break; 630 case RTM_DELETE: 631 /* if we are deleting a route, gateway is a must */ 632 if ((found_addrs & RTA_GATEWAY) == 0) { 633 error = EINVAL; 634 goto done; 635 } 636 /* 637 * The RTF_SETSRC modifier does not make sense 638 * when deleting a route. 639 */ 640 if (rtm->rtm_flags & RTF_SETSRC) { 641 error = EINVAL; 642 goto done; 643 } 644 645 switch (af) { 646 case AF_INET: 647 error = ip_rt_delete(dst_addr, net_mask, gw_addr, 648 found_addrs, rtm->rtm_flags, ipif, B_FALSE, 649 CONNP_TO_WQ(connp), ioc_mp, ip_rts_request_retry, 650 ipst); 651 break; 652 case AF_INET6: 653 error = ip_rt_delete_v6(&dst_addr_v6, &net_mask_v6, 654 &gw_addr_v6, found_addrs, rtm->rtm_flags, ipif, 655 CONNP_TO_WQ(connp), ioc_mp, ip_rts_request_retry, 656 ipst); 657 break; 658 } 659 break; 660 case RTM_GET: 661 case RTM_CHANGE: 662 /* 663 * In the case of RTM_GET, the forwarding table should be 664 * searched recursively with default being matched if the 665 * specific route doesn't exist. Also, if a gateway was 666 * specified then the gateway address must also be matched. 667 * 668 * In the case of RTM_CHANGE, the gateway address (if supplied) 669 * is the new gateway address so matching on the gateway address 670 * is not done. This can lead to ambiguity when looking up the 671 * route to change as usually only the destination (and netmask, 672 * if supplied) is used for the lookup. However if a RTA_IFP 673 * sockaddr is also supplied, it can disambiguate which route to 674 * change provided the ambigous routes are tied to distinct 675 * ill's (or interface indices). If the routes are not tied to 676 * any particular interfaces (for example, with traditional 677 * gateway routes), then a RTA_IFP sockaddr will be of no use as 678 * it won't match any such routes. 679 * RTA_SRC is not supported for RTM_GET and RTM_CHANGE, 680 * except when RTM_CHANGE is combined to RTF_SETSRC. 681 */ 682 if (((found_addrs & RTA_SRC) != 0) && 683 ((rtm->rtm_type == RTM_GET) || 684 !(rtm->rtm_flags & RTF_SETSRC))) { 685 error = EOPNOTSUPP; 686 goto done; 687 } 688 689 if (rtm->rtm_type == RTM_GET) { 690 match_flags |= 691 (MATCH_IRE_DEFAULT | MATCH_IRE_RECURSIVE | 692 MATCH_IRE_SECATTR); 693 match_flags_local |= MATCH_IRE_SECATTR; 694 if ((found_addrs & RTA_GATEWAY) != 0) 695 match_flags |= MATCH_IRE_GW; 696 if (ioc_cr) 697 tsl = crgetlabel(ioc_cr); 698 if (rtsap != NULL) { 699 if (rtsa_validate(rtsap) != 0) { 700 error = EINVAL; 701 goto done; 702 } 703 if (tsl != NULL && 704 crgetzoneid(ioc_cr) != GLOBAL_ZONEID && 705 (tsl->tsl_doi != rtsap->rtsa_doi || 706 !bldominates(&tsl->tsl_label, 707 &rtsap->rtsa_slrange.lower_bound))) { 708 error = EPERM; 709 goto done; 710 } 711 tsl = labelalloc( 712 &rtsap->rtsa_slrange.lower_bound, 713 rtsap->rtsa_doi, KM_NOSLEEP); 714 } 715 } 716 if (rtm->rtm_type == RTM_CHANGE) { 717 if ((found_addrs & RTA_GATEWAY) && 718 (rtm->rtm_flags & RTF_SETSRC)) { 719 /* 720 * Do not want to change the gateway, 721 * but rather the source address. 722 */ 723 match_flags |= MATCH_IRE_GW; 724 } 725 } 726 727 /* 728 * If the netmask is all ones (either as supplied or as derived 729 * above), then first check for an IRE_LOOPBACK or 730 * IRE_LOCAL entry. 731 * 732 * If we didn't check for or find an IRE_LOOPBACK or IRE_LOCAL 733 * entry, then look in the forwarding table. 734 */ 735 switch (af) { 736 case AF_INET: 737 if (net_mask == IP_HOST_MASK) { 738 ire = ire_ctable_lookup(dst_addr, gw_addr, 739 IRE_LOCAL | IRE_LOOPBACK, NULL, zoneid, 740 tsl, match_flags_local, ipst); 741 /* 742 * If we found an IRE_LOCAL, make sure 743 * it is one that would be used by this 744 * zone to send packets. 745 */ 746 if (ire != NULL && 747 ire->ire_type == IRE_LOCAL && 748 ipst->ips_ip_restrict_interzone_loopback && 749 !ire_local_ok_across_zones(ire, 750 zoneid, &dst_addr, tsl, ipst)) { 751 ire_refrele(ire); 752 ire = NULL; 753 } 754 } 755 if (ire == NULL) { 756 ire = ire_ftable_lookup(dst_addr, net_mask, 757 gw_addr, 0, ipif, &sire, zoneid, 0, 758 tsl, match_flags, ipst); 759 } 760 break; 761 case AF_INET6: 762 if (IN6_ARE_ADDR_EQUAL(&net_mask_v6, &ipv6_all_ones)) { 763 ire = ire_ctable_lookup_v6(&dst_addr_v6, 764 &gw_addr_v6, IRE_LOCAL | IRE_LOOPBACK, NULL, 765 zoneid, tsl, match_flags_local, ipst); 766 /* 767 * If we found an IRE_LOCAL, make sure 768 * it is one that would be used by this 769 * zone to send packets. 770 */ 771 if (ire != NULL && 772 ire->ire_type == IRE_LOCAL && 773 ipst->ips_ip_restrict_interzone_loopback && 774 !ire_local_ok_across_zones(ire, 775 zoneid, (void *)&dst_addr_v6, tsl, ipst)) { 776 ire_refrele(ire); 777 ire = NULL; 778 } 779 } 780 if (ire == NULL) { 781 ire = ire_ftable_lookup_v6(&dst_addr_v6, 782 &net_mask_v6, &gw_addr_v6, 0, ipif, &sire, 783 zoneid, 0, tsl, match_flags, ipst); 784 } 785 break; 786 } 787 if (tsl != NULL && tsl != crgetlabel(ioc_cr)) 788 label_rele(tsl); 789 790 if (ire == NULL) { 791 error = ESRCH; 792 goto done; 793 } 794 /* we know the IRE before we come here */ 795 switch (rtm->rtm_type) { 796 case RTM_GET: 797 mp1 = rts_rtmget(mp, ire, sire, af); 798 if (mp1 == NULL) { 799 error = ENOBUFS; 800 goto done; 801 } 802 freemsg(mp); 803 mp = mp1; 804 rtm = (rt_msghdr_t *)mp->b_rptr; 805 break; 806 case RTM_CHANGE: 807 /* 808 * Do not allow to the multirouting state of a route 809 * to be changed. This aims to prevent undesirable 810 * stages where both multirt and non-multirt routes 811 * for the same destination are declared. 812 */ 813 if ((ire->ire_flags & RTF_MULTIRT) != 814 (rtm->rtm_flags & RTF_MULTIRT)) { 815 error = EINVAL; 816 goto done; 817 } 818 /* 819 * Note that we do not need to do 820 * ire_flush_cache_*(IRE_FLUSH_ADD) as a change 821 * in metrics or gateway will not affect existing 822 * routes since it does not create a more specific 823 * route. 824 */ 825 switch (af) { 826 case AF_INET: 827 ire_flush_cache_v4(ire, IRE_FLUSH_DELETE); 828 if ((found_addrs & RTA_GATEWAY) != 0 && 829 (ire->ire_gateway_addr != gw_addr)) { 830 ire->ire_gateway_addr = gw_addr; 831 } 832 833 if (rtsap != NULL) { 834 ga.ga_af = AF_INET; 835 IN6_IPADDR_TO_V4MAPPED( 836 ire->ire_gateway_addr, &ga.ga_addr); 837 838 gcgrp = gcgrp_lookup(&ga, B_TRUE); 839 if (gcgrp == NULL) { 840 error = ENOMEM; 841 goto done; 842 } 843 } 844 845 if ((found_addrs & RTA_SRC) != 0 && 846 (rtm->rtm_flags & RTF_SETSRC) != 0 && 847 (ire->ire_src_addr != src_addr)) { 848 849 if (src_addr != INADDR_ANY) { 850 /* 851 * The RTF_SETSRC flag is 852 * present, check that the 853 * supplied src address is not 854 * the loopback address. This 855 * would produce martian 856 * packets. 857 */ 858 if (src_addr == 859 htonl(INADDR_LOOPBACK)) { 860 error = EINVAL; 861 goto done; 862 } 863 /* 864 * Also check that the the 865 * supplied addr is a valid 866 * local address. 867 */ 868 tmp_ipif = ipif_lookup_addr( 869 src_addr, NULL, ALL_ZONES, 870 CONNP_TO_WQ(connp), ioc_mp, 871 ip_rts_request_retry, 872 &error, ipst); 873 if (tmp_ipif == NULL) { 874 error = (error == 875 EINPROGRESS) ? 876 error : 877 EADDRNOTAVAIL; 878 goto done; 879 } 880 881 if (!(tmp_ipif->ipif_flags & 882 IPIF_UP) || 883 (tmp_ipif->ipif_flags & 884 (IPIF_NOLOCAL | 885 IPIF_ANYCAST))) { 886 error = EINVAL; 887 goto done; 888 } 889 ire->ire_flags |= RTF_SETSRC; 890 } else { 891 ire->ire_flags &= ~RTF_SETSRC; 892 } 893 ire->ire_src_addr = src_addr; 894 } 895 break; 896 case AF_INET6: 897 ire_flush_cache_v6(ire, IRE_FLUSH_DELETE); 898 mutex_enter(&ire->ire_lock); 899 if ((found_addrs & RTA_GATEWAY) != 0 && 900 !IN6_ARE_ADDR_EQUAL( 901 &ire->ire_gateway_addr_v6, &gw_addr_v6)) { 902 ire->ire_gateway_addr_v6 = gw_addr_v6; 903 } 904 905 if (rtsap != NULL) { 906 ga.ga_af = AF_INET6; 907 ga.ga_addr = ire->ire_gateway_addr_v6; 908 909 gcgrp = gcgrp_lookup(&ga, B_TRUE); 910 if (gcgrp == NULL) { 911 error = ENOMEM; 912 goto done; 913 } 914 } 915 916 if ((found_addrs & RTA_SRC) != 0 && 917 (rtm->rtm_flags & RTF_SETSRC) != 0 && 918 !IN6_ARE_ADDR_EQUAL( 919 &ire->ire_src_addr_v6, &src_addr_v6)) { 920 921 if (!IN6_IS_ADDR_UNSPECIFIED( 922 &src_addr_v6)) { 923 /* 924 * The RTF_SETSRC flag is 925 * present, check that the 926 * supplied src address is not 927 * the loopback address. This 928 * would produce martian 929 * packets. 930 */ 931 if (IN6_IS_ADDR_LOOPBACK( 932 &src_addr_v6)) { 933 mutex_exit( 934 &ire->ire_lock); 935 error = EINVAL; 936 goto done; 937 } 938 /* 939 * Also check that the the 940 * supplied addr is a valid 941 * local address. 942 */ 943 tmp_ipif = ipif_lookup_addr_v6( 944 &src_addr_v6, NULL, 945 ALL_ZONES, 946 CONNP_TO_WQ(connp), ioc_mp, 947 ip_rts_request_retry, 948 &error, ipst); 949 if (tmp_ipif == NULL) { 950 mutex_exit( 951 &ire->ire_lock); 952 error = (error == 953 EINPROGRESS) ? 954 error : 955 EADDRNOTAVAIL; 956 goto done; 957 } 958 if (!(tmp_ipif->ipif_flags & 959 IPIF_UP) || 960 (tmp_ipif->ipif_flags & 961 (IPIF_NOLOCAL | 962 IPIF_ANYCAST))) { 963 mutex_exit( 964 &ire->ire_lock); 965 error = EINVAL; 966 goto done; 967 } 968 ire->ire_flags |= RTF_SETSRC; 969 } else { 970 ire->ire_flags &= ~RTF_SETSRC; 971 } 972 ire->ire_src_addr_v6 = src_addr_v6; 973 } 974 mutex_exit(&ire->ire_lock); 975 break; 976 } 977 978 if (rtsap != NULL) { 979 in_addr_t ga_addr4; 980 981 ASSERT(gcgrp != NULL); 982 983 /* 984 * Create and add the security attribute to 985 * prefix IRE; it will add a reference to the 986 * group upon allocating a new entry. If it 987 * finds an already-existing entry for the 988 * security attribute, it simply returns it 989 * and no new group reference is made. 990 */ 991 gc = gc_create(rtsap, gcgrp, &gcgrp_xtraref); 992 if (gc == NULL || 993 (error = tsol_ire_init_gwattr(ire, 994 ire->ire_ipversion, gc, NULL)) != 0) { 995 if (gc != NULL) { 996 GC_REFRELE(gc); 997 } else { 998 /* gc_create failed */ 999 error = ENOMEM; 1000 } 1001 goto done; 1002 } 1003 1004 /* 1005 * Now delete any existing gateway IRE caches 1006 * as well as all caches using the gateway, 1007 * and allow them to be created on demand 1008 * through ip_newroute{_v6}. 1009 */ 1010 IN6_V4MAPPED_TO_IPADDR(&ga.ga_addr, ga_addr4); 1011 if (af == AF_INET) { 1012 ire_clookup_delete_cache_gw( 1013 ga_addr4, ALL_ZONES, ipst); 1014 } else { 1015 ire_clookup_delete_cache_gw_v6( 1016 &ga.ga_addr, ALL_ZONES, ipst); 1017 } 1018 } 1019 rts_setmetrics(ire, rtm->rtm_inits, &rtm->rtm_rmx); 1020 break; 1021 } 1022 break; 1023 default: 1024 error = EOPNOTSUPP; 1025 break; 1026 } 1027 done: 1028 if (ire != NULL) 1029 ire_refrele(ire); 1030 if (sire != NULL) 1031 ire_refrele(sire); 1032 if (ipif != NULL) 1033 ipif_refrele(ipif); 1034 if (tmp_ipif != NULL) 1035 ipif_refrele(tmp_ipif); 1036 1037 if (gcgrp_xtraref) 1038 GCGRP_REFRELE(gcgrp); 1039 1040 if (error == EINPROGRESS) { 1041 if (rtm != NULL) 1042 freemsg(mp); 1043 return (error); 1044 } 1045 if (rtm != NULL) { 1046 ASSERT(mp->b_wptr <= mp->b_datap->db_lim); 1047 if (error != 0) { 1048 rtm->rtm_errno = error; 1049 /* Send error ACK */ 1050 ip1dbg(("ip_rts_request: error %d\n", error)); 1051 } else { 1052 rtm->rtm_flags |= RTF_DONE; 1053 /* OK ACK already set up by caller except this */ 1054 ip2dbg(("ip_rts_request: OK ACK\n")); 1055 } 1056 rts_queue_input(mp, q, af, ipst); 1057 } 1058 iocp->ioc_error = error; 1059 ioc_mp->b_datap->db_type = M_IOCACK; 1060 if (iocp->ioc_error != 0) 1061 iocp->ioc_count = 0; 1062 (connp->conn_recv)(connp, ioc_mp, NULL); 1063 /* conn was refheld in ip_wput_ioctl. */ 1064 CONN_OPER_PENDING_DONE(connp); 1065 1066 return (error); 1067 } 1068 1069 /* 1070 * Build a reply to the RTM_GET request contained in the given message block 1071 * using the retrieved IRE of the destination address, the parent IRE (if it 1072 * exists) and the address family. 1073 * 1074 * Returns a pointer to a message block containing the reply if successful, 1075 * otherwise NULL is returned. 1076 */ 1077 static mblk_t * 1078 rts_rtmget(mblk_t *mp, ire_t *ire, ire_t *sire, sa_family_t af) 1079 { 1080 rt_msghdr_t *rtm; 1081 rt_msghdr_t *new_rtm; 1082 mblk_t *new_mp; 1083 int rtm_addrs; 1084 int rtm_flags; 1085 in6_addr_t gw_addr_v6; 1086 tsol_ire_gw_secattr_t *attrp = NULL; 1087 tsol_gc_t *