1 0 stevel /* 2 8485 Peter * Copyright 2009 Sun Microsystems, Inc. All rights reserved. 3 0 stevel * Use is subject to license terms. 4 0 stevel */ 5 0 stevel 6 0 stevel /* 7 0 stevel * Copyright (c) 1988, 1991, 1993 8 0 stevel * The Regents of the University of California. All rights reserved. 9 0 stevel * 10 0 stevel * Redistribution and use in source and binary forms, with or without 11 0 stevel * modification, are permitted provided that the following conditions 12 0 stevel * are met: 13 0 stevel * 1. Redistributions of source code must retain the above copyright 14 0 stevel * notice, this list of conditions and the following disclaimer. 15 0 stevel * 2. Redistributions in binary form must reproduce the above copyright 16 0 stevel * notice, this list of conditions and the following disclaimer in the 17 0 stevel * documentation and/or other materials provided with the distribution. 18 0 stevel * 3. All advertising materials mentioning features or use of this software 19 0 stevel * must display the following acknowledgement: 20 0 stevel * This product includes software developed by the University of 21 0 stevel * California, Berkeley and its contributors. 22 0 stevel * 4. Neither the name of the University nor the names of its contributors 23 0 stevel * may be used to endorse or promote products derived from this software 24 0 stevel * without specific prior written permission. 25 0 stevel * 26 0 stevel * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 27 0 stevel * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 28 0 stevel * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 29 0 stevel * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 30 0 stevel * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 31 0 stevel * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 32 0 stevel * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 0 stevel * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 34 0 stevel * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 35 0 stevel * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 36 0 stevel * SUCH DAMAGE. 37 0 stevel * 38 0 stevel * @(#)rtsock.c 8.6 (Berkeley) 2/11/95 39 0 stevel */ 40 0 stevel 41 0 stevel /* 42 0 stevel * This file contains routines that processes routing socket requests. 43 0 stevel */ 44 0 stevel 45 0 stevel #include <sys/types.h> 46 0 stevel #include <sys/stream.h> 47 0 stevel #include <sys/stropts.h> 48 0 stevel #include <sys/ddi.h> 49 8778 Erik #include <sys/strsubr.h> 50 0 stevel #include <sys/cmn_err.h> 51 0 stevel #include <sys/debug.h> 52 0 stevel #include <sys/policy.h> 53 0 stevel #include <sys/zone.h> 54 0 stevel 55 0 stevel #include <sys/systm.h> 56 0 stevel #include <sys/param.h> 57 0 stevel #include <sys/socket.h> 58 0 stevel #include <sys/strsun.h> 59 0 stevel #include <net/if.h> 60 0 stevel #include <net/route.h> 61 0 stevel #include <netinet/in.h> 62 0 stevel #include <net/if_dl.h> 63 0 stevel #include <netinet/ip6.h> 64 0 stevel 65 0 stevel #include <inet/common.h> 66 0 stevel #include <inet/ip.h> 67 0 stevel #include <inet/ip6.h> 68 0 stevel #include <inet/ip_if.h> 69 0 stevel #include <inet/ip_ire.h> 70 2535 sangeeta #include <inet/ip_ftable.h> 71 0 stevel #include <inet/ip_rts.h> 72 0 stevel 73 0 stevel #include <inet/ipclassifier.h> 74 0 stevel 75 1676 jpk #include <sys/tsol/tndb.h> 76 1676 jpk #include <sys/tsol/tnet.h> 77 1676 jpk 78 1676 jpk #define RTS_MSG_SIZE(type, rtm_addrs, af, sacnt) \ 79 1676 jpk (rts_data_msg_size(rtm_addrs, af, sacnt) + rts_header_msg_size(type)) 80 0 stevel 81 0 stevel static size_t rts_copyfromsockaddr(struct sockaddr *sa, in6_addr_t *addrp); 82 0 stevel static void rts_fill_msg(int type, int rtm_addrs, ipaddr_t dst, 83 0 stevel ipaddr_t mask, ipaddr_t gateway, ipaddr_t src_addr, ipaddr_t brd_addr, 84 11042 Erik ipaddr_t author, ipaddr_t ifaddr, const ill_t *ill, mblk_t *mp, 85 11042 Erik const tsol_gc_t *); 86 0 stevel static int rts_getaddrs(rt_msghdr_t *rtm, in6_addr_t *dst_addrp, 87 0 stevel in6_addr_t *gw_addrp, in6_addr_t *net_maskp, in6_addr_t *authorp, 88 0 stevel in6_addr_t *if_addrp, in6_addr_t *src_addrp, ushort_t *indexp, 89 4823 seb sa_family_t *afp, tsol_rtsecattr_t *rtsecattr, int *error); 90 1676 jpk static void rts_getifdata(if_data_t *if_data, const ipif_t *ipif); 91 0 stevel static int rts_getmetrics(ire_t *ire, rt_metrics_t *metrics); 92 11042 Erik static mblk_t *rts_rtmget(mblk_t *mp, ire_t *ire, ire_t *ifire, 93 11042 Erik const in6_addr_t *setsrc, tsol_ire_gw_secattr_t *attrp, sa_family_t af); 94 0 stevel static void rts_setmetrics(ire_t *ire, uint_t which, rt_metrics_t *metrics); 95 11042 Erik static ire_t *ire_lookup_v4(ipaddr_t dst_addr, ipaddr_t net_mask, 96 11042 Erik ipaddr_t gw_addr, const ill_t *ill, zoneid_t zoneid, 97 11042 Erik const ts_label_t *tsl, int match_flags, ip_stack_t *ipst, ire_t **pifire, 98 11042 Erik ipaddr_t *v4setsrcp, tsol_ire_gw_secattr_t **gwattrp); 99 11042 Erik static ire_t *ire_lookup_v6(const in6_addr_t *dst_addr_v6, 100 11042 Erik const in6_addr_t *net_mask_v6, const in6_addr_t *gw_addr_v6, 101 11042 Erik const ill_t *ill, zoneid_t zoneid, const ts_label_t *tsl, int match_flags, 102 11042 Erik ip_stack_t *ipst, ire_t **pifire, 103 11042 Erik in6_addr_t *v6setsrcp, tsol_ire_gw_secattr_t **gwattrp); 104 0 stevel 105 0 stevel /* 106 8485 Peter * Send `mp' to all eligible routing queues. A queue is ineligible if: 107 0 stevel * 108 8485 Peter * 1. SO_USELOOPBACK is off and it is not the originating queue. 109 11042 Erik * 2. RTA_UNDER_IPMP is on and RTSQ_UNDER_IPMP is not set in `flags'. 110 11042 Erik * 3. RTA_UNDER_IPMP is off and RTSQ_NORMAL is not set in `flags'. 111 8485 Peter * 4. It is not the same address family as `af', and `af' isn't AF_UNSPEC. 112 0 stevel */ 113 0 stevel void 114 8485 Peter rts_queue_input(mblk_t *mp, conn_t *o_connp, sa_family_t af, uint_t flags, 115 8485 Peter ip_stack_t *ipst) 116 0 stevel { 117 0 stevel mblk_t *mp1; 118 0 stevel conn_t *connp, *next_connp; 119 0 stevel 120 8485 Peter /* 121 8485 Peter * Since we don't have an ill_t here, RTSQ_DEFAULT must already be 122 11042 Erik * resolved to one or more of RTSQ_NORMAL|RTSQ_UNDER_IPMP at this point. 123 8485 Peter */ 124 8485 Peter ASSERT(!(flags & RTSQ_DEFAULT)); 125 8485 Peter 126 3448 dh155122 mutex_enter(&ipst->ips_rts_clients->connf_lock); 127 3448 dh155122 connp = ipst->ips_rts_clients->connf_head; 128 0 stevel 129 8485 Peter for (; connp != NULL; connp = next_connp) { 130 8485 Peter next_connp = connp->conn_next; 131 0 stevel /* 132 0 stevel * If there was a family specified when this routing socket was 133 0 stevel * created and it doesn't match the family of the message to 134 0 stevel * copy, then continue. 135 0 stevel */ 136 0 stevel if ((connp->conn_proto != AF_UNSPEC) && 137 8485 Peter (connp->conn_proto != af)) 138 0 stevel continue; 139 8485 Peter 140 8485 Peter /* 141 8485 Peter * Queue the message only if the conn_t and flags match. 142 8485 Peter */ 143 8485 Peter if (connp->conn_rtaware & RTAW_UNDER_IPMP) { 144 8485 Peter if (!(flags & RTSQ_UNDER_IPMP)) 145 8485 Peter continue; 146 8485 Peter } else { 147 8485 Peter if (!(flags & RTSQ_NORMAL)) 148 8485 Peter continue; 149 0 stevel } 150 0 stevel /* 151 0 stevel * For the originating queue, we only copy the message upstream 152 0 stevel * if loopback is set. For others reading on the routing 153 0 stevel * socket, we check if there is room upstream for a copy of the 154 0 stevel * message. 155 0 stevel */ 156 11042 Erik if ((o_connp == connp) && connp->conn_useloopback == 0) { 157 8485 Peter connp = connp->conn_next; 158 8485 Peter continue; 159 0 stevel } 160 0 stevel CONN_INC_REF(connp); 161 3448 dh155122 mutex_exit(&ipst->ips_rts_clients->connf_lock); 162 5240 nordmark /* Pass to rts_input */ 163 11042 Erik if (IPCL_IS_NONSTR(connp) ? !connp->conn_flow_cntrld : 164 11042 Erik canputnext(connp->conn_rq)) { 165 0 stevel mp1 = dupmsg(mp); 166 0 stevel if (mp1 == NULL) 167 0 stevel mp1 = copymsg(mp); 168 11042 Erik /* Note that we pass a NULL ira to rts_input */ 169 0 stevel if (mp1 != NULL) 170 11042 Erik (connp->conn_recv)(connp, mp1, NULL, NULL); 171 0 stevel } 172 0 stevel 173 3448 dh155122 mutex_enter(&ipst->ips_rts_clients->connf_lock); 174 8485 Peter /* reload next_connp since conn_next may have changed */ 175 0 stevel next_connp = connp->conn_next; 176 0 stevel CONN_DEC_REF(connp); 177 0 stevel } 178 3448 dh155122 mutex_exit(&ipst->ips_rts_clients->connf_lock); 179 0 stevel freemsg(mp); 180 0 stevel } 181 0 stevel 182 0 stevel /* 183 0 stevel * Takes an ire and sends an ack to all the routing sockets. This 184 0 stevel * routine is used 185 0 stevel * - when a route is created/deleted through the ioctl interface. 186 11042 Erik * - when a stale redirect is deleted 187 0 stevel */ 188 0 stevel void 189 3448 dh155122 ip_rts_rtmsg(int type, ire_t *ire, int error, ip_stack_t *ipst) 190 0 stevel { 191 0 stevel mblk_t *mp; 192 0 stevel rt_msghdr_t *rtm; 193 0 stevel int rtm_addrs = (RTA_DST | RTA_NETMASK | RTA_GATEWAY); 194 0 stevel sa_family_t af; 195 0 stevel in6_addr_t gw_addr_v6; 196 0 stevel 197 0 stevel if (ire == NULL) 198 0 stevel return; 199 0 stevel ASSERT(ire->ire_ipversion == IPV4_VERSION || 200 0 stevel ire->ire_ipversion == IPV6_VERSION); 201 0 stevel 202 11042 Erik ASSERT(!(ire->ire_type & IRE_IF_CLONE)); 203 11042 Erik 204 0 stevel if (ire->ire_flags & RTF_SETSRC) 205 0 stevel rtm_addrs |= RTA_SRC; 206 0 stevel 207 0 stevel switch (ire->ire_ipversion) { 208 0 stevel case IPV4_VERSION: 209 0 stevel af = AF_INET; 210 1676 jpk mp = rts_alloc_msg(type, rtm_addrs, af, 0); 211 0 stevel if (mp == NULL) 212 0 stevel return; 213 0 stevel rts_fill_msg(type, rtm_addrs, ire->ire_addr, ire->ire_mask, 214 11042 Erik ire->ire_gateway_addr, ire->ire_setsrc_addr, 0, 0, 0, NULL, 215 11042 Erik mp, NULL); 216 0 stevel break; 217 0 stevel case IPV6_VERSION: 218 0 stevel af = AF_INET6; 219 1676 jpk mp = rts_alloc_msg(type, rtm_addrs, af, 0); 220 0 stevel if (mp == NULL) 221 0 stevel return; 222 0 stevel mutex_enter(&ire->ire_lock); 223 0 stevel gw_addr_v6 = ire->ire_gateway_addr_v6; 224 0 stevel mutex_exit(&ire->ire_lock); 225 0 stevel rts_fill_msg_v6(type, rtm_addrs, &ire->ire_addr_v6, 226 0 stevel &ire->ire_mask_v6, &gw_addr_v6, 227 11042 Erik &ire->ire_setsrc_addr_v6, &ipv6_all_zeros, &ipv6_all_zeros, 228 11042 Erik &ipv6_all_zeros, NULL, mp, NULL); 229 0 stevel break; 230 0 stevel } 231 0 stevel rtm = (rt_msghdr_t *)mp->b_rptr; 232 0 stevel mp->b_wptr = (uchar_t *)&mp->b_rptr[rtm->rtm_msglen]; 233 0 stevel rtm->rtm_addrs = rtm_addrs; 234 0 stevel rtm->rtm_flags = ire->ire_flags; 235 0 stevel if (error != 0) 236 0 stevel rtm->rtm_errno = error; 237 0 stevel else 238 0 stevel rtm->rtm_flags |= RTF_DONE; 239 8485 Peter rts_queue_input(mp, NULL, af, RTSQ_ALL, ipst); 240 0 stevel } 241 0 stevel 242 0 stevel /* 243 5240 nordmark * This is a call from the RTS module 244 5240 nordmark * indicating that this is a Routing Socket 245 5240 nordmark * Stream. Insert this conn_t in routing 246 5240 nordmark * socket client list. 247 5240 nordmark */ 248 5240 nordmark void 249 5240 nordmark ip_rts_register(conn_t *connp) 250 5240 nordmark { 251 5240 nordmark ip_stack_t *ipst = connp->conn_netstack->netstack_ip; 252 5240 nordmark 253 11042 Erik connp->conn_useloopback = 1; 254 5240 nordmark ipcl_hash_insert_wildcard(ipst->ips_rts_clients, connp); 255 5240 nordmark } 256 5240 nordmark 257 5240 nordmark /* 258 5240 nordmark * This is a call from the RTS module indicating that it is closing. 259 5240 nordmark */ 260 5240 nordmark void 261 5240 nordmark ip_rts_unregister(conn_t *connp) 262 5240 nordmark { 263 5240 nordmark ipcl_hash_remove(connp); 264 5240 nordmark } 265 5240 nordmark 266 5240 nordmark /* 267 0 stevel * Processes requests received on a routing socket. It extracts all the 268 0 stevel * arguments and calls the appropriate function to process the request. 269 0 stevel * 270 4823 seb * RTA_SRC bit flag requests are sent by 'route -setsrc'. 271 0 stevel * 272 0 stevel * In general, this function does not consume the message supplied but rather 273 0 stevel * sends the message upstream with an appropriate UNIX errno. 274 0 stevel */ 275 0 stevel int 276 11042 Erik ip_rts_request_common(mblk_t *mp, conn_t *connp, cred_t *ioc_cr) 277 0 stevel { 278 0 stevel rt_msghdr_t *rtm = NULL; 279 0 stevel in6_addr_t dst_addr_v6; 280 0 stevel in6_addr_t src_addr_v6; 281 0 stevel in6_addr_t gw_addr_v6; 282 0 stevel in6_addr_t net_mask_v6; 283 0 stevel in6_addr_t author_v6; 284 0 stevel in6_addr_t if_addr_v6; 285 11042 Erik mblk_t *mp1; 286 0 stevel ire_t *ire = NULL; 287 11042 Erik ire_t *ifire = NULL; 288 11042 Erik ipaddr_t v4setsrc; 289 11042 Erik in6_addr_t v6setsrc = ipv6_all_zeros; 290 11042 Erik tsol_ire_gw_secattr_t *gwattr = NULL; 291 0 stevel int error = 0; 292 0 stevel int match_flags = MATCH_IRE_DSTONLY; 293 2304 wy83408 int match_flags_local = MATCH_IRE_TYPE | MATCH_IRE_GW; 294 0 stevel int found_addrs; 295 0 stevel sa_family_t af; 296 0 stevel ipaddr_t dst_addr; 297 0 stevel ipaddr_t gw_addr; 298 0 stevel ipaddr_t src_addr; 299 0 stevel ipaddr_t net_mask; 300 0 stevel ushort_t index; 301 1676 jpk boolean_t gcgrp_xtraref = B_FALSE; 302 1676 jpk tsol_gcgrp_addr_t ga; 303 1676 jpk tsol_rtsecattr_t rtsecattr; 304 1676 jpk struct rtsa_s *rtsap = NULL; 305 1676 jpk tsol_gcgrp_t *gcgrp = NULL; 306 1676 jpk tsol_gc_t *gc = NULL; 307 2601 wy83408 ts_label_t *tsl = NULL; 308 2733 nordmark zoneid_t zoneid; 309 3448 dh155122 ip_stack_t *ipst; 310 11042 Erik ill_t *ill = NULL; 311 0 stevel 312 2733 nordmark zoneid = connp->conn_zoneid; 313 3448 dh155122 ipst = connp->conn_netstack->netstack_ip; 314 5240 nordmark 315 0 stevel if (mp->b_cont != NULL && !pullupmsg(mp, -1)) { 316 0 stevel freemsg(mp); 317 0 stevel error = EINVAL; 318 0 stevel goto done; 319 0 stevel } 320 0 stevel if ((mp->b_wptr - mp->b_rptr) < sizeof (rt_msghdr_t)) { 321 0 stevel freemsg(mp); 322 0 stevel error = EINVAL; 323 0 stevel goto done; 324 0 stevel } 325 0 stevel 326 0 stevel /* 327 0 stevel * Check the routing message for basic consistency including the 328 0 stevel * version number and that the number of octets written is the same 329 0 stevel * as specified by the rtm_msglen field. 330 0 stevel * 331 0 stevel * At this point, an error can be delivered back via rtm_errno. 332 0 stevel */ 333 0 stevel rtm = (rt_msghdr_t *)mp->b_rptr; 334 0 stevel if ((mp->b_wptr - mp->b_rptr) != rtm->rtm_msglen) { 335 0 stevel error = EINVAL; 336 0 stevel goto done; 337 0 stevel } 338 0 stevel if (rtm->rtm_version != RTM_VERSION) { 339 0 stevel error = EPROTONOSUPPORT; 340 0 stevel goto done; 341 0 stevel } 342 0 stevel 343 0 stevel /* Only allow RTM_GET or RTM_RESOLVE for unprivileged process */ 344 0 stevel if (rtm->rtm_type != RTM_GET && 345 0 stevel rtm->rtm_type != RTM_RESOLVE && 346 0 stevel (ioc_cr == NULL || 347 3448 dh155122 secpolicy_ip_config(ioc_cr, B_FALSE) != 0)) { 348 0 stevel error = EPERM; 349 0 stevel goto done; 350 0 stevel } 351 0 stevel 352 0 stevel found_addrs = rts_getaddrs(rtm, &dst_addr_v6, &gw_addr_v6, &net_mask_v6, 353 4823 seb &author_v6, &if_addr_v6, &src_addr_v6, &index, &af, &rtsecattr, 354 4823 seb &error); 355 1676 jpk 356 1676 jpk if (error != 0) 357 1676 jpk goto done; 358 1676 jpk 359 0 stevel if ((found_addrs & RTA_DST) == 0) { 360 0 stevel error = EINVAL; 361 0 stevel goto done; 362 0 stevel } 363 0 stevel 364 0 stevel /* 365 0 stevel * Based on the address family of the destination address, determine 366 0 stevel * the destination, gateway and netmask and return the appropriate error 367 0 stevel * if an unknown address family was specified (following the errno 368 0 stevel * values that 4.4BSD-Lite2 returns.) 369 0 stevel */ 370 0 stevel switch (af) { 371 0 stevel case AF_INET: 372 0 stevel IN6_V4MAPPED_TO_IPADDR(&dst_addr_v6, dst_addr); 373 0 stevel IN6_V4MAPPED_TO_IPADDR(&src_addr_v6, src_addr); 374 0 stevel IN6_V4MAPPED_TO_IPADDR(&gw_addr_v6, gw_addr); 375 0 stevel if (((found_addrs & RTA_NETMASK) == 0) || 376 0 stevel (rtm->rtm_flags & RTF_HOST)) 377 0 stevel net_mask = IP_HOST_MASK; 378 0 stevel else 379 0 stevel IN6_V4MAPPED_TO_IPADDR(&net_mask_v6, net_mask); 380 0 stevel break; 381 0 stevel case AF_INET6: 382 0 stevel if (((found_addrs & RTA_NETMASK) == 0) || 383 0 stevel (rtm->rtm_flags & RTF_HOST)) 384 0 stevel net_mask_v6 = ipv6_all_ones; 385 0 stevel break; 386 0 stevel default: 387 0 stevel /* 388 0 stevel * These errno values are meant to be compatible with 389 0 stevel * 4.4BSD-Lite2 for the given message types. 390 0 stevel */ 391 0 stevel switch (rtm->rtm_type) { 392 0 stevel case RTM_ADD: 393 0 stevel case RTM_DELETE: 394 0 stevel error = ESRCH; 395 0 stevel goto done; 396 0 stevel case RTM_GET: 397 0 stevel case RTM_CHANGE: 398 0 stevel error = EAFNOSUPPORT; 399 0 stevel goto done; 400 0 stevel default: 401 0 stevel error = EOPNOTSUPP; 402 0 stevel goto done; 403 0 stevel } 404 0 stevel } 405 0 stevel 406 0 stevel /* 407 0 stevel * At this point, the address family must be something known. 408 0 stevel */ 409 0 stevel ASSERT(af == AF_INET || af == AF_INET6); 410 0 stevel 411 11042 Erik /* Handle RTA_IFP */ 412 0 stevel if (index != 0) { 413 11042 Erik ipif_t *ipif; 414 8485 Peter lookup: 415 11042 Erik ill = ill_lookup_on_ifindex(index, af == AF_INET6, ipst); 416 0 stevel if (ill == NULL) { 417 11042 Erik error = EINVAL; 418 0 stevel goto done; 419 0 stevel } 420 0 stevel 421 8485 Peter /* 422 8485 Peter * Since all interfaces in an IPMP group must be equivalent, 423 8485 Peter * we prevent changes to a specific underlying interface's 424 8485 Peter * routing configuration. However, for backward compatibility, 425 8485 Peter * we intepret a request to add a route on an underlying 426 8485 Peter * interface as a request to add a route on its IPMP interface. 427 8485 Peter */ 428 8485 Peter if (IS_UNDER_IPMP(ill)) { 429 8485 Peter switch (rtm->rtm_type) { 430 8485 Peter case RTM_CHANGE: 431 8485 Peter case RTM_DELETE: 432 8485 Peter error = EINVAL; 433 8485 Peter goto done; 434 8485 Peter case RTM_ADD: 435 8485 Peter index = ipmp_ill_get_ipmp_ifindex(ill); 436 8485 Peter ill_refrele(ill); 437 8485 Peter if (index == 0) { 438 11042 Erik ill = NULL; /* already refrele'd */ 439 8485 Peter error = EINVAL; 440 8485 Peter goto done; 441 8485 Peter } 442 8485 Peter goto lookup; 443 8485 Peter } 444 8485 Peter } 445 8485 Peter 446 11042 Erik match_flags |= MATCH_IRE_ILL; 447 11042 Erik /* 448 11042 Erik * This provides the same zoneid as in Solaris 10 449 11042 Erik * that -ifp picks the zoneid from the first ipif on the ill. 450 11042 Erik * But it might not be useful since the first ipif will always 451 11042 Erik * have the same zoneid as the ill. 452 11042 Erik */ 453 0 stevel ipif = ipif_get_next_ipif(NULL, ill); 454 11042 Erik if (ipif != NULL) { 455 11042 Erik zoneid = ipif->ipif_zoneid; 456 11042 Erik ipif_refrele(ipif); 457 11042 Erik } 458 0 stevel } 459 0 stevel 460 0 stevel /* 461 0 stevel * If a netmask was supplied in the message, then subsequent route 462 0 stevel * lookups will attempt to match on the netmask as well. 463 0 stevel */ 464 0 stevel if ((found_addrs & RTA_NETMASK) != 0) 465 0 stevel match_flags |= MATCH_IRE_MASK; 466 0 stevel 467 1676 jpk /* 468 1676 jpk * We only process any passed-in route security attributes for 469 2304 wy83408 * either RTM_ADD or RTM_CHANGE message; We overload them 470 2304 wy83408 * to do an RTM_GET as a different label; ignore otherwise. 471 1676 jpk */ 472 2304 wy83408 if (rtm->rtm_type == RTM_ADD || rtm->rtm_type == RTM_CHANGE || 473 2304 wy83408 rtm->rtm_type == RTM_GET) { 474 1676 jpk ASSERT(rtsecattr.rtsa_cnt <= TSOL_RTSA_REQUEST_MAX); 475 1676 jpk if (rtsecattr.rtsa_cnt > 0) 476 1676 jpk rtsap = &rtsecattr.rtsa_attr[0]; 477 1676 jpk } 478 1676 jpk 479 0 stevel switch (rtm->rtm_type) { 480 0 stevel case RTM_ADD: 481 0 stevel /* if we are adding a route, gateway is a must */ 482 0 stevel if ((found_addrs & RTA_GATEWAY) == 0) { 483 0 stevel error = EINVAL; 484 0 stevel goto done; 485 0 stevel } 486 0 stevel 487 0 stevel /* Multirouting does not support net routes. */ 488 0 stevel if ((rtm->rtm_flags & (RTF_MULTIRT | RTF_HOST)) == 489 0 stevel RTF_MULTIRT) { 490 0 stevel error = EADDRNOTAVAIL; 491 0 stevel goto done; 492 0 stevel } 493 0 stevel 494 0 stevel /* 495 0 stevel * Multirouting and user-specified source addresses 496 0 stevel * do not support interface based routing. 497 0 stevel * Assigning a source address to an interface based 498 0 stevel * route is achievable by plumbing a new ipif and 499 0 stevel * setting up the interface route via this ipif, 500 0 stevel * though. 501 0 stevel */ 502 0 stevel if (rtm->rtm_flags & (RTF_MULTIRT | RTF_SETSRC)) { 503 0 stevel if ((rtm->rtm_flags & RTF_GATEWAY) == 0) { 504 0 stevel error = EADDRNOTAVAIL; 505 0 stevel goto done; 506 0 stevel } 507 0 stevel } 508 0 stevel 509 0 stevel switch (af) { 510 0 stevel case AF_INET: 511 0 stevel if (src_addr != INADDR_ANY) { 512 11042 Erik uint_t type; 513 11042 Erik 514 0 stevel /* 515 0 stevel * The RTF_SETSRC flag is present, check that 516 0 stevel * the supplied src address is not the loopback 517 0 stevel * address. This would produce martian packets. 518 0 stevel */ 519 0 stevel if (src_addr == htonl(INADDR_LOOPBACK)) { 520 0 stevel error = EINVAL; 521 0 stevel goto done; 522 0 stevel } 523 0 stevel /* 524 0 stevel * Also check that the supplied address is a 525 11042 Erik * valid, local one. Only allow IFF_UP ones 526 0 stevel */ 527 11042 Erik type = ip_type_v4(src_addr, ipst); 528 11042 Erik if (!(type & (IRE_LOCAL|IRE_LOOPBACK))) { 529 11042 Erik error = EADDRNOTAVAIL; 530 0 stevel goto done; 531 0 stevel } 532 0 stevel } else { 533 0 stevel /* 534 0 stevel * The RTF_SETSRC modifier must be associated 535 0 stevel * to a non-null source address. 536 0 stevel */ 537 0 stevel if (rtm->rtm_flags & RTF_SETSRC) { 538 0 stevel error = EINVAL; 539 0 stevel goto done; 540 0 stevel } 541 0 stevel } 542 0 stevel 543 1676 jpk error = ip_rt_add(dst_addr, net_mask, gw_addr, src_addr, 544 11042 Erik rtm->rtm_flags, ill, &ire, B_FALSE, 545 11042 Erik rtsap, ipst, zoneid); 546 11042 Erik if (ill != NULL) 547 11042 Erik ASSERT(!MUTEX_HELD(&ill->ill_lock)); 548 0 stevel break; 549 0 stevel case AF_INET6: 550 0 stevel if (!IN6_IS_ADDR_UNSPECIFIED(&src_addr_v6)) { 551 11042 Erik uint_t type; 552 11042 Erik 553 0 stevel /* 554 0 stevel * The RTF_SETSRC flag is present, check that 555 0 stevel * the supplied src address is not the loopback 556 0 stevel * address. This would produce martian packets. 557 0 stevel */ 558 0 stevel if (IN6_IS_ADDR_LOOPBACK(&src_addr_v6)) { 559 0 stevel error = EINVAL; 560 0 stevel goto done; 561 0 stevel } 562 0 stevel /* 563 0 stevel * Also check that the supplied address is a 564 11042 Erik * valid, local one. Only allow UP ones. 565 0 stevel */ 566 11042 Erik type = ip_type_v6(&src_addr_v6, ipst); 567 11042 Erik if (!(type & (IRE_LOCAL|IRE_LOOPBACK))) { 568 11042 Erik error = EADDRNOTAVAIL; 569 0 stevel goto done; 570 0 stevel } 571 0 stevel 572 0 stevel error = ip_rt_add_v6(&dst_addr_v6, &net_mask_v6, 573 0 stevel &gw_addr_v6, &src_addr_v6, rtm->rtm_flags, 574 11042 Erik ill, &ire, rtsap, ipst, zoneid); 575 0 stevel break; 576 0 stevel } 577 0 stevel /* 578 0 stevel * The RTF_SETSRC modifier must be associated 579 0 stevel * to a non-null source address. 580 0 stevel */ 581 0 stevel if (rtm->rtm_flags & RTF_SETSRC) { 582 0 stevel error = EINVAL; 583 0 stevel goto done; 584 0 stevel } 585 0 stevel error = ip_rt_add_v6(&dst_addr_v6, &net_mask_v6, 586 0 stevel &gw_addr_v6, NULL, rtm->rtm_flags, 587 11042 Erik ill, &ire, rtsap, ipst, zoneid); 588 11042 Erik if (ill != NULL) 589 11042 Erik ASSERT(!MUTEX_HELD(&ill->ill_lock)); 590 0 stevel break; 591 0 stevel } 592 0 stevel if (error != 0) 593 0 stevel goto done; 594 0 stevel ASSERT(ire != NULL); 595 0 stevel rts_setmetrics(ire, rtm->rtm_inits, &rtm->rtm_rmx); 596 0 stevel break; 597 0 stevel case RTM_DELETE: 598 0 stevel /* if we are deleting a route, gateway is a must */ 599 0 stevel if ((found_addrs & RTA_GATEWAY) == 0) { 600 0 stevel error = EINVAL; 601 0 stevel goto done; 602 0 stevel } 603 0 stevel /* 604 0 stevel * The RTF_SETSRC modifier does not make sense 605 0 stevel * when deleting a route. 606 0 stevel */ 607 0 stevel if (rtm->rtm_flags & RTF_SETSRC) { 608 0 stevel error = EINVAL; 609 0 stevel goto done; 610 0 stevel } 611 0 stevel 612 0 stevel switch (af) { 613 0 stevel case AF_INET: 614 0 stevel error = ip_rt_delete(dst_addr, net_mask, gw_addr, 615 11042 Erik found_addrs, rtm->rtm_flags, ill, B_FALSE, 616 11042 Erik ipst, zoneid); 617 0 stevel break; 618 0 stevel case AF_INET6: 619 0 stevel error = ip_rt_delete_v6(&dst_addr_v6, &net_mask_v6, 620 11042 Erik &gw_addr_v6, found_addrs, rtm->rtm_flags, ill, 621 11042 Erik ipst, zoneid); 622 0 stevel break; 623 0 stevel } 624 0 stevel break; 625 0 stevel case RTM_GET: 626 0 stevel case RTM_CHANGE: 627 0 stevel /* 628 0 stevel * In the case of RTM_GET, the forwarding table should be 629 11042 Erik * searched recursively. Also, if a gateway was 630 0 stevel * specified then the gateway address must also be matched. 631 0 stevel * 632 0 stevel * In the case of RTM_CHANGE, the gateway address (if supplied) 633 0 stevel * is the new gateway address so matching on the gateway address 634 0 stevel * is not done. This can lead to ambiguity when looking up the 635 0 stevel * route to change as usually only the destination (and netmask, 636 0 stevel * if supplied) is used for the lookup. However if a RTA_IFP 637 0 stevel * sockaddr is also supplied, it can disambiguate which route to 638 0 stevel * change provided the ambigous routes are tied to distinct 639 0 stevel * ill's (or interface indices). If the routes are not tied to 640 0 stevel * any particular interfaces (for example, with traditional 641 0 stevel * gateway routes), then a RTA_IFP sockaddr will be of no use as 642 0 stevel * it won't match any such routes. 643 0 stevel * RTA_SRC is not supported for RTM_GET and RTM_CHANGE, 644 0 stevel * except when RTM_CHANGE is combined to RTF_SETSRC. 645 0 stevel */ 646 0 stevel if (((found_addrs & RTA_SRC) != 0) && 647 0 stevel ((rtm->rtm_type == RTM_GET) || 648 0 stevel !(rtm->rtm_flags & RTF_SETSRC))) { 649 0 stevel error = EOPNOTSUPP; 650 0 stevel goto done; 651 0 stevel } 652 0 stevel 653 0 stevel if (rtm->rtm_type == RTM_GET) { 654 11042 Erik match_flags |= MATCH_IRE_SECATTR; 655 2304 wy83408 match_flags_local |= MATCH_IRE_SECATTR; 656 0 stevel if ((found_addrs & RTA_GATEWAY) != 0) 657 0 stevel match_flags |= MATCH_IRE_GW; 658 2601 wy83408 if (ioc_cr) 659 2601 wy83408 tsl = crgetlabel(ioc_cr); 660 2304 wy83408 if (rtsap != NULL) { 661 2304 wy83408 if (rtsa_validate(rtsap) != 0) { 662 2304 wy83408 error = EINVAL; 663 2304 wy83408 goto done; 664 2304 wy83408 } 665 2601 wy83408 if (tsl != NULL && 666 2601 wy83408 crgetzoneid(ioc_cr) != GLOBAL_ZONEID && 667 2304 wy83408 (tsl->tsl_doi != rtsap->rtsa_doi || 668 2304 wy83408 !bldominates(&tsl->tsl_label, 669 2304 wy83408 &rtsap->rtsa_slrange.lower_bound))) { 670 2304 wy83408 error = EPERM; 671 2304 wy83408 goto done; 672 2304 wy83408 } 673 2304 wy83408 tsl = labelalloc( 674 2304 wy83408 &rtsap->rtsa_slrange.lower_bound, 675 2304 wy83408 rtsap->rtsa_doi, KM_NOSLEEP); 676 2304 wy83408 } 677 0 stevel } 678 0 stevel if (rtm->rtm_type == RTM_CHANGE) { 679 0 stevel if ((found_addrs & RTA_GATEWAY) && 680 0 stevel (rtm->rtm_flags & RTF_SETSRC)) { 681 0 stevel /* 682 0 stevel * Do not want to change the gateway, 683 0 stevel * but rather the source address. 684 0 stevel */ 685 0 stevel match_flags |= MATCH_IRE_GW; 686 0 stevel } 687 0 stevel } 688 0 stevel 689 0 stevel /* 690 0 stevel * If the netmask is all ones (either as supplied or as derived 691 0 stevel * above), then first check for an IRE_LOOPBACK or 692 0 stevel * IRE_LOCAL entry. 693 0 stevel * 694 0 stevel * If we didn't check for or find an IRE_LOOPBACK or IRE_LOCAL 695 11042 Erik * entry, then look for any other type of IRE. 696 0 stevel */ 697 0 stevel switch (af) { 698 0 stevel case AF_INET: 699 0 stevel if (net_mask == IP_HOST_MASK) { 700 11042 Erik ire = ire_ftable_lookup_v4(dst_addr, 0, gw_addr, 701 2733 nordmark IRE_LOCAL | IRE_LOOPBACK, NULL, zoneid, 702 11042 Erik tsl, match_flags_local, 0, ipst, NULL); 703 0 stevel } 704 0 stevel if (ire == NULL) { 705 11042 Erik ire = ire_lookup_v4(dst_addr, net_mask, 706 11042 Erik gw_addr, ill, zoneid, tsl, match_flags, 707 11042 Erik ipst, &ifire, &v4setsrc, &gwattr); 708 11042 Erik IN6_IPADDR_TO_V4MAPPED(v4setsrc, &v6setsrc); 709 0 stevel } 710 0 stevel break; 711 0 stevel case AF_INET6: 712 0 stevel if (IN6_ARE_ADDR_EQUAL(&net_mask_v6, &ipv6_all_ones)) { 713 11042 Erik ire = ire_ftable_lookup_v6(&dst_addr_v6, NULL, 714 0 stevel &gw_addr_v6, IRE_LOCAL | IRE_LOOPBACK, NULL, 715 11042 Erik zoneid, tsl, match_flags_local, 0, ipst, 716 11042 Erik NULL); 717 0 stevel } 718 0 stevel if (ire == NULL) { 719 11042 Erik ire = ire_lookup_v6(&dst_addr_v6, 720 11042 Erik &net_mask_v6, &gw_addr_v6, ill, zoneid, 721 11042 Erik tsl, match_flags, ipst, &ifire, &v6setsrc, 722 11042 Erik &gwattr); 723 0 stevel } 724 0 stevel break; 725 0 stevel } 726 2304 wy83408 if (tsl != NULL && tsl != crgetlabel(ioc_cr)) 727 2304 wy83408 label_rele(tsl); 728 0 stevel 729 0 stevel if (ire == NULL) { 730 0 stevel error = ESRCH; 731 0 stevel goto done; 732 0 stevel } 733 11042 Erik /* 734 11042 Erik * Want to return failure if we get an IRE_NOROUTE from 735 11042 Erik * ire_route_recursive 736 11042 Erik */ 737 11042 Erik if (ire->ire_type & IRE_NOROUTE) { 738 11042 Erik ire_refrele(ire); 739 11042 Erik ire = NULL; 740 11042 Erik error = ESRCH; 741 11042 Erik goto done; 742 11042 Erik } 743 11042 Erik 744 0 stevel /* we know the IRE before we come here */ 745 0 stevel switch (rtm->rtm_type) { 746 0 stevel case RTM_GET: 747 11042 Erik mp1 = rts_rtmget(mp, ire, ifire, &v6setsrc, gwattr, af); 748 0 stevel if (mp1 == NULL) { 749 0 stevel error = ENOBUFS; 750 0 stevel goto done; 751 0 stevel } 752 0 stevel freemsg(mp); 753 0 stevel mp = mp1; 754 0 stevel rtm = (rt_msghdr_t *)mp->b_rptr; 755 0 stevel break; 756 0 stevel case RTM_CHANGE: 757 0 stevel /* 758 0 stevel * Do not allow to the multirouting state of a route 759 0 stevel * to be changed. This aims to prevent undesirable 760 0 stevel * stages where both multirt and non-multirt routes 761 0 stevel * for the same destination are declared. 762 0 stevel */ 763 0 stevel if ((ire->ire_flags & RTF_MULTIRT) != 764 0 stevel (rtm->rtm_flags & RTF_MULTIRT)) { 765 0 stevel error = EINVAL; 766 0 stevel goto done; 767 0 stevel } 768 0 stevel /* 769 0 stevel * Note that we do not need to do 770 0 stevel * ire_flush_cache_*(IRE_FLUSH_ADD) as a change 771 0 stevel * in metrics or gateway will not affect existing 772 0 stevel * routes since it does not create a more specific 773 0 stevel * route. 774 0 stevel */ 775 0 stevel switch (af) { 776 0 stevel case AF_INET: 777 0 stevel if ((found_addrs & RTA_GATEWAY) != 0 && 778 0 stevel (ire->ire_gateway_addr != gw_addr)) { 779 0 stevel ire->ire_gateway_addr = gw_addr; 780 0 stevel } 781 1676 jpk 782 1676 jpk if (rtsap != NULL) { 783 1676 jpk ga.ga_af = AF_INET; 784 1676 jpk IN6_IPADDR_TO_V4MAPPED( 785 1676 jpk ire->ire_gateway_addr, &ga.ga_addr); 786 1676 jpk 787 1676 jpk gcgrp = gcgrp_lookup(&ga, B_TRUE); 788 1676 jpk if (gcgrp == NULL) { 789 1676 jpk error = ENOMEM; 790 1676 jpk goto done; 791 1676 jpk } 792 1676 jpk } 793 1676 jpk 794 0 stevel if ((found_addrs & RTA_SRC) != 0 && 795 0 stevel (rtm->rtm_flags & RTF_SETSRC) != 0 && 796 11042 Erik (ire->ire_setsrc_addr != src_addr)) { 797 11042 Erik if (src_addr != INADDR_ANY) { 798 11042 Erik uint_t type; 799 0 stevel 800 0 stevel /* 801 0 stevel * The RTF_SETSRC flag is 802 0 stevel * present, check that the 803 0 stevel * supplied src address is not 804 0 stevel * the loopback address. This 805 0 stevel * would produce martian 806 0 stevel * packets. 807 0 stevel */ 808 0 stevel if (src_addr == 809 0 stevel htonl(INADDR_LOOPBACK)) { 810 0 stevel error = EINVAL; 811 0 stevel goto done; 812 0 stevel } 813 0 stevel /* 814 11042 Erik * Also check that the 815 0 stevel * supplied addr is a valid 816 0 stevel * local address. 817 0 stevel */ 818 11042 Erik type = ip_type_v4(src_addr, 819 11042 Erik ipst); 820 11042 Erik if (!(type & 821 11042 Erik (IRE_LOCAL|IRE_LOOPBACK))) { 822 11042 Erik error = EADDRNOTAVAIL; 823 0 stevel goto done; 824 0 stevel } 825 0 stevel ire->ire_flags |= RTF_SETSRC; 826 11042 Erik ire->ire_setsrc_addr = 827 11042 Erik src_addr; 828 0 stevel } else { 829 0 stevel ire->ire_flags &= ~RTF_SETSRC; 830 11042 Erik ire->ire_setsrc_addr = 831 11042 Erik INADDR_ANY; 832 0 stevel } 833 11042 Erik /* 834 11042 Erik * Let conn_ixa caching know that 835 11042 Erik * source address selection changed 836 11042 Erik */ 837 11042 Erik ip_update_source_selection(ipst); 838 0 stevel } 839 11042 Erik ire_flush_cache_v4(ire, IRE_FLUSH_GWCHANGE); 840 0 stevel break; 841 0 stevel case AF_INET6: 842 0 stevel mutex_enter(&ire->ire_lock); 843 0 stevel if ((found_addrs & RTA_GATEWAY) != 0 && 844 0 stevel !IN6_ARE_ADDR_EQUAL( 845 0 stevel &ire->ire_gateway_addr_v6, &gw_addr_v6)) { 846 0 stevel ire->ire_gateway_addr_v6 = gw_addr_v6; 847 0 stevel } 848 11042 Erik mutex_exit(&ire->ire_lock); 849 1676 jpk 850 1676 jpk if (rtsap != NULL) { 851 1676 jpk ga.ga_af = AF_INET6; 852 11042 Erik mutex_enter(&ire->ire_lock); 853 1676 jpk ga.ga_addr = ire->ire_gateway_addr_v6; 854 11042 Erik mutex_exit(&ire->ire_lock); 855 1676 jpk 856 1676 jpk gcgrp = gcgrp_lookup(&ga, B_TRUE); 857 1676 jpk if (gcgrp == NULL) { 858 1676 jpk error = ENOMEM; 859 1676 jpk goto done; 860 1676 jpk } 861 1676 jpk } 862 1676 jpk 863 0 stevel if ((found_addrs & RTA_SRC) != 0 && 864 0 stevel (rtm->rtm_flags & RTF_SETSRC) != 0 && 865 0 stevel !IN6_ARE_ADDR_EQUAL( 866 11042 Erik &ire->ire_setsrc_addr_v6, &src_addr_v6)) { 867 0 stevel if (!IN6_IS_ADDR_UNSPECIFIED( 868 0 stevel &src_addr_v6)) { 869 11042 Erik uint_t type; 870 11042 Erik 871 0 stevel /* 872 0 stevel * The RTF_SETSRC flag is 873 0 stevel * present, check that the 874 0 stevel * supplied src address is not 875 0 stevel * the loopback address. This 876 0 stevel * would produce martian 877 0 stevel * packets. 878 0 stevel */ 879 0 stevel if (IN6_IS_ADDR_LOOPBACK( 880 0 stevel &src_addr_v6)) { 881 0 stevel error = EINVAL; 882 0 stevel goto done; 883 0 stevel } 884 0 stevel /* 885 11042 Erik * Also check that the 886 0 stevel * supplied addr is a valid 887 0 stevel * local address. 888 0 stevel */ 889 11042 Erik type = ip_type_v6(&src_addr_v6, 890 11042 Erik ipst); 891 11042 Erik if (!(type & 892 11042 Erik (IRE_LOCAL|IRE_LOOPBACK))) { 893 11042 Erik error = EADDRNOTAVAIL; 894 0 stevel goto done; 895 0 stevel } 896 11042 Erik mutex_enter(&ire->ire_lock); 897 0 stevel ire->ire_flags |= RTF_SETSRC; 898 11042 Erik ire->ire_setsrc_addr_v6 = 899 11042 Erik src_addr_v6; 900 11042 Erik mutex_exit(&ire->ire_lock); 901 0 stevel } else { 902 11042 Erik mutex_enter(&ire->ire_lock); 903 0 stevel ire->ire_flags &= ~RTF_SETSRC; 904 11042 Erik ire->ire_setsrc_addr_v6 = 905 11042 Erik ipv6_all_zeros; 906 11042 Erik mutex_exit(&ire->ire_lock); 907 0 stevel } 908 11042 Erik /* 909 11042 Erik * Let conn_ixa caching know that 910 11042 Erik * source address selection changed 911 11042 Erik */ 912 11042 Erik ip_update_source_selection(ipst); 913 0 stevel } 914 11042 Erik ire_flush_cache_v6(ire, IRE_FLUSH_GWCHANGE); 915 0 stevel break; 916 0 stevel } 917 1676 jpk 918 1676 jpk if (rtsap != NULL) { 919 1676 jpk ASSERT(gcgrp != NULL); 920 1676 jpk 921 1676 jpk /* 922 1676 jpk * Create and add the security attribute to 923 1676 jpk * prefix IRE; it will add a reference to the 924 1676 jpk * group upon allocating a new entry. If it 925 1676 jpk * finds an already-existing entry for the 926 1676 jpk * security attribute, it simply returns it 927 1676 jpk * and no new group reference is made. 928 1676 jpk */ 929 1676 jpk gc = gc_create(rtsap, gcgrp, &gcgrp_xtraref); 930 1676 jpk if (gc == NULL || 931 1676 jpk (error = tsol_ire_init_gwattr(ire, 932 11042 Erik ire->ire_ipversion, gc)) != 0) { 933 1676 jpk if (gc != NULL) { 934 1676 jpk GC_REFRELE(gc); 935 1676 jpk } else { 936 1676 jpk /* gc_create failed */ 937 1676 jpk error = ENOMEM; 938 1676 jpk } 939 1676 jpk goto done; 940 1676 jpk } 941 1676 jpk } 942 0 stevel rts_setmetrics(ire, rtm->rtm_inits, &rtm->rtm_rmx); 943 0 stevel break; 944 0 stevel } 945 0 stevel break; 946 0 stevel default: 947 0 stevel error = EOPNOTSUPP; 948 0 stevel break; 949 0 stevel } 950 0 stevel done: 951 0 stevel if (ire != NULL) 952 0 stevel ire_refrele(ire); 953 11042 Erik if (ifire != NULL) 954 11042 Erik ire_refrele(ifire); 955 11042 Erik if (ill != NULL) 956 11042 Erik ill_refrele(ill); 957 1676 jpk 958 1676 jpk if (gcgrp_xtraref) 959 1676 jpk GCGRP_REFRELE(gcgrp); 960 0 stevel 961 0 stevel if (rtm != NULL) { 962 0 stevel ASSERT(mp->b_wptr <= mp->b_datap->db_lim); 963 0 stevel if (error != 0) { 964 0 stevel rtm->rtm_errno = error; 965 0 stevel /* Send error ACK */ 966 0 stevel ip1dbg(("ip_rts_request: error %d\n", error)); 967 0 stevel } else { 968 0 stevel rtm->rtm_flags |= RTF_DONE; 969 0 stevel /* OK ACK already set up by caller except this */ 970 0 stevel ip2dbg(("ip_rts_request: OK ACK\n")); 971 0 stevel } 972 8485 Peter rts_queue_input(mp, connp, af, RTSQ_ALL, ipst); 973 0 stevel } 974 11042 Erik return (error); 975 11042 Erik } 976 8348 Eric 977 11042 Erik /* 978 11042 Erik * Helper function that can do recursive lookups including when 979 11042 Erik * MATCH_IRE_GW and/or MATCH_IRE_MASK is set. 980 11042 Erik */ 981 11042 Erik static ire_t * 982 11042 Erik ire_lookup_v4(ipaddr_t dst_addr, ipaddr_t net_mask, ipaddr_t gw_addr, 983 11042 Erik const ill_t *ill, zoneid_t zoneid, const ts_label_t *tsl, 984 11042 Erik int match_flags, ip_stack_t *ipst, ire_t **pifire, ipaddr_t *v4setsrcp, 985 11042 Erik tsol_ire_gw_secattr_t **gwattrp) 986 11042 Erik { 987 11042 Erik ire_t *ire; 988 11042 Erik ire_t *ifire = NULL; 989 11042 Erik uint_t ire_type; 990 11042 Erik 991 11042 Erik *pifire = NULL; 992 11042 Erik *v4setsrcp = INADDR_ANY; 993 11042 Erik *gwattrp = NULL; 994 11042 Erik 995 11042 Erik /* Skip IRE_IF_CLONE */ 996 11042 Erik match_flags |= MATCH_IRE_TYPE; 997 11042 Erik ire_type = (IRE_ONLINK|IRE_OFFLINK) & ~IRE_IF_CLONE; 998 11042 Erik 999 11042 Erik /* 1000 11042 Erik * ire_route_recursive can't match gateway or mask thus if they are 1001 11042 Erik * set we have to do two steps of lookups 1002 11042 Erik */ 1003 11042 Erik if (match_flags & (MATCH_IRE_GW|MATCH_IRE_MASK)) { 1004 11042 Erik ire = ire_ftable_lookup_v4(dst_addr, net_mask, gw_addr, 1005 11042 Erik ire_type, ill, zoneid, tsl, match_flags, 0, ipst, NULL); 1006 11042 Erik 1007 11042 Erik if (ire == NULL ||(ire->ire_flags & (RTF_REJECT|RTF_BLACKHOLE))) 1008 11042 Erik return (ire); 1009 11042 Erik 1010 11042 Erik if (ire->ire_type & IRE_ONLINK) 1011 11042 Erik return (ire); 1012 11042 Erik 1013 11042 Erik if (ire->ire_flags & RTF_SETSRC) { 1014 11042 Erik ASSERT(ire->ire_setsrc_addr != INADDR_ANY); 1015 11042 Erik *v4setsrcp = ire->ire_setsrc_addr; 1016 11042 Erik v4setsrcp = NULL; 1017 11042 Erik } 1018 11042 Erik 1019 11042 Erik /* The first ire_gw_secattr is passed back */ 1020 11042 Erik if (ire->ire_gw_secattr != NULL) { 1021 11042 Erik *gwattrp = ire->ire_gw_secattr; 1022 11042 Erik gwattrp = NULL; 1023 11042 Erik } 1024 11042 Erik 1025 11042 Erik /* Look for an interface ire recursively based on the gateway */ 1026 11042 Erik dst_addr = ire->ire_gateway_addr; 1027 11042 Erik match_flags &= ~(MATCH_IRE_GW|MATCH_IRE_MASK); 1028 11042 Erik ifire = ire_route_recursive_v4(dst_addr, ire_type, ill, zoneid, 1029 11042 Erik tsl, match_flags, B_FALSE, 0, ipst, v4setsrcp, gwattrp, 1030 11042 Erik NULL); 1031 11042 Erik } else { 1032 11042 Erik ire = ire_route_recursive_v4(dst_addr, ire_type, ill, zoneid, 1033 11042 Erik tsl, match_flags, B_FALSE, 0, ipst, v4setsrcp, gwattrp, 1034 11042 Erik NULL); 1035 11042 Erik } 1036 11042 Erik *pifire = ifire; 1037 11042 Erik return (ire); 1038 11042 Erik } 1039 11042 Erik 1040 11042 Erik static ire_t * 1041 11042 Erik ire_lookup_v6(const in6_addr_t *dst_addr_v6, 1042 11042 Erik const in6_addr_t *net_mask_v6, const in6_addr_t *gw_addr_v6, 1043 11042 Erik const ill_t *ill, zoneid_t zoneid, const ts_label_t *tsl, int match_flags, 1044 11042 Erik ip_stack_t *ipst, ire_t **pifire, 1045 11042 Erik in6_addr_t *v6setsrcp, tsol_ire_gw_secattr_t **gwattrp) 1046 11042 Erik { 1047 11042 Erik ire_t *ire; 1048 11042 Erik ire_t *ifire = NULL; 1049 11042 Erik uint_t ire_type; 1050 11042 Erik 1051 11042 Erik *pifire = NULL; 1052 11042 Erik *v6setsrcp = ipv6_all_zeros; 1053 11042 Erik *gwattrp = NULL; 1054 11042 Erik 1055 11042 Erik /* Skip IRE_IF_CLONE */ 1056 11042 Erik match_flags |= MATCH_IRE_TYPE; 1057 11042 Erik ire_type = (IRE_ONLINK|IRE_OFFLINK) & ~IRE_IF_CLONE; 1058 11042 Erik 1059 11042 Erik /* 1060 11042 Erik * ire_route_recursive can't match gateway or mask thus if they are 1061 11042 Erik * set we have to do two steps of lookups 1062 11042 Erik */ 1063 11042 Erik if (match_flags & (MATCH_IRE_GW|MATCH_IRE_MASK)) { 1064 11042 Erik in6_addr_t dst; 1065 11042 Erik 1066 11042 Erik ire = ire_ftable_lookup_v6(dst_addr_v6, net_mask_v6, 1067 11042 Erik gw_addr_v6, ire_type, ill, zoneid, tsl, match_flags, 0, 1068 11042 Erik ipst, NULL); 1069 11042 Erik 1070 11042 Erik if (ire == NULL ||(ire->ire_flags & (RTF_REJECT|RTF_BLACKHOLE))) 1071 11042 Erik return (ire); 1072 11042 Erik 1073 11042 Erik if (ire->ire_type & IRE_ONLINK) 1074 11042 Erik return (ire); 1075 11042 Erik 1076 11042 Erik if (ire->ire_flags & RTF_SETSRC) { 1077 11042 Erik ASSERT(!IN6_IS_ADDR_UNSPECIFIED( 1078 11042 Erik &ire->ire_setsrc_addr_v6)); 1079 11042 Erik *v6setsrcp = ire->ire_setsrc_addr_v6; 1080 11042 Erik v6setsrcp = NULL; 1081 11042 Erik } 1082 11042 Erik 1083 11042 Erik /* The first ire_gw_secattr is passed back */ 1084 11042 Erik if (ire->ire_gw_secattr != NULL) { 1085 11042 Erik *gwattrp = ire->ire_gw_secattr; 1086 11042 Erik gwattrp = NULL; 1087 11042 Erik } 1088 11042 Erik 1089 11042 Erik mutex_enter(&ire->ire_lock); 1090 11042 Erik dst = ire->ire_gateway_addr_v6; 1091 11042 Erik mutex_exit(&ire->ire_lock); 1092 11042 Erik match_flags &= ~(MATCH_IRE_GW|MATCH_IRE_MASK); 1093 11042 Erik ifire = ire_route_recursive_v6(&dst, ire_type, ill, zoneid, tsl, 1094 11042 Erik match_flags, B_FALSE, 0, ipst, v6setsrcp, gwattrp, NULL); 1095 11042 Erik } else { 1096 11042 Erik ire = ire_route_recursive_v6(dst_addr_v6, ire_type, ill, zoneid, 1097 11042 Erik tsl, match_flags, B_FALSE, 0, ipst, v6setsrcp, gwattrp, 1098 11042 Erik NULL); 1099 11042 Erik } 1100 11042 Erik *pifire = ifire; 1101 11042 Erik return (ire); 1102 11042 Erik } 1103 11042 Erik 1104 11042 Erik 1105 11042 Erik /* 1106 11042 Erik * Handle IP_IOC_RTS_REQUEST ioctls 1107 11042 Erik */ 1108 11042 Erik int 1109 11042 Erik ip_rts_request(queue_t *q, mblk_t *mp, cred_t *ioc_cr) 1110 11042 Erik { 1111 11042 Erik conn_t *connp = Q_TO_CONN(q); 1112 11042 Erik IOCP iocp = (IOCP)mp->b_rptr; 1113 11042 Erik mblk_t *mp1, *ioc_mp = mp; 1114 11042 Erik int error = 0; 1115 11042 Erik ip_stack_t *ipst; 1116 11042 Erik 1117 11042 Erik ipst = connp->conn_netstack->netstack_ip; 1118 11042 Erik 1119 11042 Erik ASSERT(mp->b_cont != NULL); 1120 11042 Erik /* ioc_mp holds mp */ 1121 11042 Erik mp = mp->b_cont; 1122 11042 Erik 1123 11042 Erik /* 1124 11042 Erik * The Routing Socket data starts on 1125 11042 Erik * next block. If there is no next block 1126 11042 Erik * this is an indication from routing module 1127 11042 Erik * that it is a routing socket stream queue. 1128 11042 Erik * We need to support that for compatibility with SDP since 1129 11042 Erik * it has a contract private interface to use IP_IOC_RTS_REQUEST. 1130 11042 Erik * Note: SDP no longer uses IP_IOC_RTS_REQUEST - we can remove this. 1131 11042 Erik */ 1132 11042 Erik if (mp->b_cont == NULL) { 1133 11042 Erik /* 1134 11042 Erik * This is a message from SDP 1135 11042 Erik * indicating that this is a Routing Socket 1136 11042 Erik * Stream. Insert this conn_t in routing 1137 11042 Erik * socket client list. 1138 11042 Erik */ 1139 11042 Erik connp->conn_useloopback = 1; 1140 11042 Erik ipcl_hash_insert_wildcard(ipst->ips_rts_clients, connp); 1141 11042 Erik goto done; 1142 11042 Erik } 1143 11042 Erik mp1 = dupmsg(mp->b_cont); 1144 11042 Erik if (mp1 == NULL) { 1145 11042 Erik error = ENOBUFS; 1146 11042 Erik goto done; 1147 11042 Erik } 1148 11042 Erik mp = mp1; 1149 11042 Erik 1150 11042 Erik error = ip_rts_request_common(mp, connp, ioc_cr); 1151 11042 Erik done: 1152 0 stevel iocp->ioc_error = error; 1153 0 stevel ioc_mp->b_datap->db_type = M_IOCACK; 1154 0 stevel if (iocp->ioc_error != 0) 1155 0 stevel iocp->ioc_count = 0; 1156 11042 Erik /* Note that we pass a NULL ira to rts_input */ 1157 11042 Erik (connp->conn_recv)(connp, ioc_mp, NULL, NULL); 1158 8348 Eric 1159 0 stevel /* conn was refheld in ip_wput_ioctl. */ 1160 0 stevel CONN_OPER_PENDING_DONE(connp); 1161 0 stevel 1162 0 stevel return (error); 1163 0 stevel } 1164 0 stevel 1165 0 stevel /* 1166 0 stevel * Build a reply to the RTM_GET request contained in the given message block 1167 0 stevel * using the retrieved IRE of the destination address, the parent IRE (if it 1168 0 stevel * exists) and the address family. 1169 0 stevel * 1170 0 stevel * Returns a pointer to a message block containing the reply if successful, 1171 0 stevel * otherwise NULL is returned. 1172 0 stevel */ 1173 1676 jpk static mblk_t * 1174 11042 Erik rts_rtmget(mblk_t *mp, ire_t *ire, ire_t *ifire, const in6_addr_t *setsrc, 1175 11042 Erik tsol_ire_gw_secattr_t *attrp, sa_family_t af) 1176 0 stevel { 1177 0 stevel rt_msghdr_t *rtm; 1178 0 stevel rt_msghdr_t *new_rtm; 1179 0 stevel mblk_t *new_mp; 1180 0 stevel int rtm_addrs; 1181 0 stevel int rtm_flags; 1182 1676 jpk tsol_gc_t *gc = NULL; 1183 1676 jpk tsol_gcgrp_t *gcgrp = NULL; 1184 11042 Erik ill_t *ill; 1185 11042 Erik ipif_t *ipif = NULL; 1186 11042 Erik ipaddr_t brdaddr; /* IFF_POINTOPOINT destination */ 1187 11042 Erik ipaddr_t ifaddr; 1188 11042 Erik in6_addr_t brdaddr6; /* IFF_POINTOPOINT destination */ 1189 11042 Erik in6_addr_t ifaddr6; 1190 11042 Erik ipaddr_t v4setsrc; 1191 0 stevel 1192 0 stevel rtm = (rt_msghdr_t *)mp->b_rptr; 1193 1676 jpk 1194 11042 Erik /* 1195 11042 Erik * Find the ill used to send packets. This will be NULL in case 1196 11042 Erik * of a reject or blackhole. 1197 11042 Erik */ 1198 11042 Erik if (ifire != NULL) 1199 11042 Erik ill = ire_nexthop_ill(ifire); 1200 11042 Erik else 1201 11042 Erik ill = ire_nexthop_ill(ire); 1202 1676 jpk 1203 1676 jpk if (attrp != NULL) { 1204 1676 jpk mutex_enter(&attrp->igsa_lock); 1205 1676 jpk if ((gc = attrp->igsa_gc) != NULL) { 1206 1676 jpk gcgrp = gc->gc_grp; 1207 1676 jpk ASSERT(gcgrp != NULL); 1208 1676 jpk rw_enter(&gcgrp->gcgrp_rwlock, RW_READER); 1209 1676 jpk } 1210 1676 jpk mutex_exit(&attrp->igsa_lock); 1211 1676 jpk } 1212 0 stevel 1213 0 stevel /* 1214 0 stevel * Always return RTA_DST, RTA_GATEWAY and RTA_NETMASK. 1215 0 stevel * 1216 0 stevel * The 4.4BSD-Lite2 code (net/rtsock.c) returns both 1217 0 stevel * RTA_IFP and RTA_IFA if either is defined, and also 1218 0 stevel * returns RTA_BRD if the appropriate interface is 1219 0 stevel * point-to-point. 1220 0 stevel */ 1221 0 stevel rtm_addrs = (RTA_DST | RTA_GATEWAY | RTA_NETMASK); 1222 11042 Erik if ((rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) && ill != NULL) { 1223 0 stevel rtm_addrs |= (RTA_IFP | RTA_IFA); 1224 11042 Erik /* 1225 11042 Erik * We associate an IRE with an ILL, hence we don't exactly 1226 11042 Erik * know what might make sense for RTA_IFA and RTA_BRD. We 1227 11042 Erik * pick the first ipif on the ill. 1228 11042 Erik */ 1229 11042 Erik ipif = ipif_get_next_ipif(NULL, ill); 1230 11042 Erik if (ipif != NULL) { 1231 11042 Erik if (ipif->ipif_isv6) 1232 11042 Erik ifaddr6 = ipif->ipif_v6lcl_addr; 1233 11042 Erik else 1234 11042 Erik ifaddr = ipif->ipif_lcl_addr; 1235 11042 Erik if (ipif->ipif_flags & IPIF_POINTOPOINT) { 1236 11042 Erik rtm_addrs |= RTA_BRD; 1237 11042 Erik if (ipif->ipif_isv6) 1238 11042 Erik brdaddr6 = ipif->ipif_v6pp_dst_addr; 1239 11042 Erik else 1240 11042 Erik brdaddr = ipif->ipif_pp_dst_addr; 1241 11042 Erik } 1242 11042 Erik ipif_refrele(ipif); 1243 11042 Erik } 1244 0 stevel } 1245 0 stevel 1246 11042 Erik new_mp = rts_alloc_msg(RTM_GET, rtm_addrs, af, gc != NULL ? 1 : 0); 1247 1676 jpk if (new_mp == NULL) { 1248 1676 jpk if (gcgrp != NULL) 1249 1676 jpk rw_exit(&gcgrp->gcgrp_rwlock); 1250 11042 Erik if (ill != NULL) 1251 11042 Erik ill_refrele(ill); 1252 0 stevel return (NULL); 1253 1676 jpk } 1254 0 stevel 1255 0 stevel /* 1256 0 stevel * We set the destination address, gateway address, 1257 0 stevel * netmask and flags in the RTM_GET response depending 1258 0 stevel * on whether we found a parent IRE or not. 1259 0 stevel * In particular, if we did find a parent IRE during the 1260 0 stevel * recursive search, use that IRE's gateway address. 1261 0 stevel * Otherwise, we use the IRE's source address for the 1262 0 stevel * gateway address. 1263 0 stevel */ 1264 0 stevel ASSERT(af == AF_INET || af == AF_INET6); 1265 0 stevel switch (af) { 1266 0 stevel case AF_INET: 1267 11042 Erik IN6_V4MAPPED_TO_IPADDR(setsrc, v4setsrc); 1268 11042 Erik if (v4setsrc != INADDR_ANY) 1269 11042 Erik rtm_addrs |= RTA_SRC; 1270 0 stevel 1271 11042 Erik rtm_flags = ire->ire_flags; 1272 11042 Erik rts_fill_msg(RTM_GET, rtm_addrs, ire->ire_addr, 1273 11042 Erik ire->ire_mask, ire->ire_gateway_addr, v4setsrc, 1274 11042 Erik brdaddr, 0, ifaddr, ill, new_mp, gc); 1275 0 stevel break; 1276 0 stevel case AF_INET6: 1277 11042 Erik if (!IN6_IS_ADDR_UNSPECIFIED(setsrc)) 1278 11042 Erik rtm_addrs |= RTA_SRC; 1279 0 stevel 1280 11042 Erik rtm_flags = ire->ire_flags; 1281 11042 Erik rts_fill_msg_v6(RTM_GET, rtm_addrs, &ire->ire_addr_v6, 1282 11042 Erik &ire->ire_mask_v6, &ire->ire_gateway_addr_v6, 1283 11042 Erik setsrc, &brdaddr6, &ipv6_all_zeros, 1284 11042 Erik &ifaddr6, ill, new_mp, gc); 1285 0 stevel break; 1286 0 stevel } 1287 1676 jpk 1288 1676 jpk if (gcgrp != NULL) 1289 1676 jpk rw_exit(&gcgrp->gcgrp_rwlock); 1290 1676 jpk 1291 0 stevel new_rtm = (rt_msghdr_t *)new_mp->b_rptr; 1292 0 stevel 1293 0 stevel /* 1294 0 stevel * The rtm_msglen, rtm_version and rtm_type fields in 1295 0 stevel * RTM_GET response are filled in by rts_fill_msg. 1296 0 stevel * 1297 0 stevel * rtm_addrs and rtm_flags are filled in based on what 1298 0 stevel * was requested and the state of the IREs looked up 1299 0 stevel * above. 1300 0 stevel * 1301 0 stevel * rtm_inits and rtm_rmx are filled in with metrics 1302 0 stevel * based on whether a parent IRE was found or not. 1303 0 stevel * 1304 0 stevel * TODO: rtm_index and rtm_use should probably be 1305 0 stevel * filled in with something resonable here and not just 1306 0 stevel * copied from the request. 1307 0 stevel */ 1308 0 stevel new_rtm->rtm_index = rtm->rtm_index; 1309 0 stevel new_rtm->rtm_pid = rtm->rtm_pid; 1310 0 stevel new_rtm->rtm_seq = rtm->rtm_seq; 1311 0 stevel new_rtm->rtm_use = rtm->rtm_use; 1312 0 stevel new_rtm->rtm_addrs = rtm_addrs; 1313 0 stevel new_rtm->rtm_flags = rtm_flags; 1314 11042 Erik new_rtm->rtm_inits = rts_getmetrics(ire, &new_rtm->rtm_rmx); 1315 11042 Erik if (ill != NULL) 1316 11042 Erik ill_refrele(ill); 1317 0 stevel return (new_mp); 1318 0 stevel } 1319 0 stevel 1320 0 stevel /* 1321 0 stevel * Fill the given if_data_t with interface statistics. 1322 0 stevel */ 1323 0 stevel static void 1324 1676 jpk rts_getifdata(if_data_t *if_data, const ipif_t *ipif) 1325 0 stevel { 1326 11042 Erik if_data->ifi_type = ipif->ipif_ill->ill_type; 1327 11042 Erik /* ethernet, tokenring, etc */ 1328 0 stevel if_data->ifi_addrlen = 0; /* media address length */ 1329 0 stevel if_data->ifi_hdrlen = 0; /* media header length */ 1330 11042 Erik if_data->ifi_mtu = ipif->ipif_ill->ill_mtu; /* mtu */ 1331 0 stevel if_data->ifi_metric = ipif->ipif_metric; /* metric (external only) */ 1332 0 stevel if_data->ifi_baudrate = 0; /* linespeed */ 1333 0 stevel 1334 0 stevel if_data->ifi_ipackets = 0; /* packets received on if */ 1335 0 stevel if_data->ifi_ierrors = 0; /* input errors on interface */ 1336 0 stevel if_data->ifi_opackets = 0; /* packets sent on interface */ 1337 0 stevel if_data->ifi_oerrors = 0; /* output errors on if */ 1338 0 stevel if_data->ifi_collisions = 0; /* collisions on csma if */ 1339 0 stevel if_data->ifi_ibytes = 0; /* total number received */ 1340 0 stevel if_data->ifi_obytes = 0; /* total number sent */ 1341 0 stevel if_data->ifi_imcasts = 0; /* multicast packets received */ 1342 0 stevel if_data->ifi_omcasts = 0; /* multicast packets sent */ 1343 0 stevel if_data->ifi_iqdrops = 0; /* dropped on input */ 1344 0 stevel if_data->ifi_noproto = 0; /* destined for unsupported */ 1345 0 stevel /* protocol. */ 1346 0 stevel } 1347 0 stevel 1348 0 stevel /* 1349 0 stevel * Set the metrics on a forwarding table route. 1350 0 stevel */ 1351 0 stevel static void 1352 0 stevel rts_setmetrics(ire_t *ire, uint_t which, rt_metrics_t *metrics) 1353 0 stevel { 1354 0 stevel clock_t rtt; 1355 0 stevel clock_t rtt_sd; 1356 11042 Erik ill_t *ill; 1357 0 stevel ifrt_t *ifrt; 1358 0 stevel mblk_t *mp; 1359 0 stevel in6_addr_t gw_addr_v6; 1360 0 stevel 1361 11042 Erik /* Need to add back some metrics to the IRE? */ 1362 0 stevel /* 1363 11042 Erik * Bypass obtaining the lock and searching ill_saved_ire_mp in the 1364 0 stevel * common case of no metrics. 1365 0 stevel */ 1366 0 stevel if (which == 0) 1367 0 stevel return; 1368 11042 Erik ire->ire_metrics.iulp_set = B_TRUE; 1369 0 stevel 1370 0 stevel /* 1371 0 stevel * iulp_rtt and iulp_rtt_sd are in milliseconds, but 4.4BSD-Lite2's 1372 0 stevel * <net/route.h> says: rmx_rtt and rmx_rttvar are stored as 1373 0 stevel * microseconds. 1374 0 stevel */ 1375 0 stevel if (which & RTV_RTT) 1376 0 stevel rtt = metrics->rmx_rtt / 1000; 1377 0 stevel if (which & RTV_RTTVAR) 1378 0 stevel rtt_sd = metrics->rmx_rttvar / 1000; 1379 0 stevel 1380 0 stevel /* 1381 0 stevel * Update the metrics in the IRE itself. 1382 0 stevel */ 1383 0 stevel mutex_enter(&ire->ire_lock); 1384 0 stevel if (which & RTV_MTU) 1385 11042 Erik ire->ire_metrics.iulp_mtu = metrics->rmx_mtu; 1386 0 stevel if (which & RTV_RTT) 1387 11042 Erik ire->ire_metrics.iulp_rtt = rtt; 1388 0 stevel if (which & RTV_SSTHRESH) 1389 11042 Erik ire->ire_metrics.iulp_ssthresh = metrics->rmx_ssthresh; 1390 0 stevel if (which & RTV_RTTVAR) 1391 11042 Erik ire->ire_metrics.iulp_rtt_sd = rtt_sd; 1392 0 stevel if (which & RTV_SPIPE) 1393 11042 Erik ire->ire_metrics.iulp_spipe = metrics->rmx_sendpipe; 1394 0 stevel if (which & RTV_RPIPE) 1395 11042 Erik ire->ire_metrics.iulp_rpipe = metrics->rmx_recvpipe; 1396 0 stevel mutex_exit(&ire->ire_lock); 1397 0 stevel 1398 0 stevel /* 1399 11042 Erik * Search through the ifrt_t chain hanging off the ILL in order to 1400 0 stevel * reflect the metric change there. 1401 0 stevel */ 1402 11042 Erik ill = ire->ire_ill; 1403 11042 Erik if (ill == NULL) 1404 0 stevel return; 1405 11042 Erik ASSERT((ill->ill_isv6 && ire->ire_ipversion == IPV6_VERSION) || 1406 11042 Erik ((!ill->ill_isv6 && ire->ire_ipversion == IPV4_VERSION))); 1407 11042 Erik if (ill->ill_isv6) { 1408 0 stevel mutex_enter(&ire->ire_lock); 1409 0 stevel gw_addr_v6 = ire->ire_gateway_addr_v6; 1410 0 stevel mutex_exit(&ire->ire_lock); 1411 0 stevel } 1412 11042 Erik mutex_enter(&ill->ill_saved_ire_lock); 1413 11042 Erik for (mp = ill->ill_saved_ire_mp; mp != NULL; mp = mp->b_cont) { 1414 0 stevel /* 1415 11042 Erik * On a given ill, the tuple of address, gateway, mask, 1416 11042 Erik * ire_type and zoneid unique for each saved IRE. 1417 0 stevel */ 1418 0 stevel ifrt = (ifrt_t *)mp->b_rptr; 1419 11042 Erik if (ill->ill_isv6) { 1420 0 stevel if (!IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6addr, 1421 0 stevel &ire->ire_addr_v6) || 1422 0 stevel !IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6gateway_addr, 1423 0 stevel &gw_addr_v6) || 1424 0 stevel !IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6mask, 1425 0 stevel &ire->ire_mask_v6)) 1426 0 stevel continue; 1427 0 stevel } else { 1428 0 stevel if (ifrt->ifrt_addr != ire->ire_addr || 1429 0 stevel ifrt->ifrt_gateway_addr != ire->ire_gateway_addr || 1430 0 stevel ifrt->ifrt_mask != ire->ire_mask) 1431 0 stevel continue; 1432 0 stevel } 1433 11042 Erik if (ifrt->ifrt_zoneid != ire->ire_zoneid || 1434 11042 Erik ifrt->ifrt_type != ire->ire_type) 1435 11042 Erik continue; 1436 11042 Erik 1437 0 stevel if (which & RTV_MTU) 1438 11042 Erik ifrt->ifrt_metrics.iulp_mtu = metrics->rmx_mtu; 1439 0 stevel if (which & RTV_RTT) 1440 11042 Erik ifrt->ifrt_metrics.iulp_rtt = rtt; 1441 0 stevel if (which & RTV_SSTHRESH) { 1442 11042 Erik ifrt->ifrt_metrics.iulp_ssthresh = 1443 0 stevel metrics->rmx_ssthresh; 1444 0 stevel } 1445 0 stevel if (which & RTV_RTTVAR) 1446 11042 Erik ifrt->ifrt_metrics.iulp_rtt_sd = metrics->rmx_rttvar; 1447 0 stevel if (which & RTV_SPIPE) 1448 11042 Erik ifrt->ifrt_metrics.iulp_spipe = metrics->rmx_sendpipe; 1449 0 stevel if (which & RTV_RPIPE) 1450 11042 Erik ifrt->ifrt_metrics.iulp_rpipe = metrics->rmx_recvpipe; 1451 0 stevel break; 1452 0 stevel } 1453 11042 Erik mutex_exit(&ill->ill_saved_ire_lock); 1454 11042 Erik 1455 11042 Erik /* 1456 11042 Erik * Update any IRE_IF_CLONE hanging created from this IRE_IF so they 1457 11042 Erik * get any new iulp_mtu. 1458 11042 Erik * We do that by deleting them; ire_create_if_clone will pick 1459 11042 Erik * up the new metrics. 1460 11042 Erik */ 1461 11042 Erik if ((ire->ire_type & IRE_INTERFACE) && ire->ire_dep_children != 0) 1462 11042 Erik ire_dep_delete_if_clone(ire); 1463 0 stevel } 1464 0 stevel 1465 0 stevel /* 1466 0 stevel * Get the metrics from a forwarding table route. 1467 0 stevel */ 1468 0 stevel static int 1469 0 stevel rts_getmetrics(ire_t *ire, rt_metrics_t *metrics) 1470 0 stevel { 1471 0 stevel int metrics_set = 0; 1472 0 stevel 1473 0 stevel bzero(metrics, sizeof (rt_metrics_t)); 1474 11042 Erik 1475 0 stevel /* 1476 0 stevel * iulp_rtt and iulp_rtt_sd are in milliseconds, but 4.4BSD-Lite2's 1477 0 stevel * <net/route.h> says: rmx_rtt and rmx_rttvar are stored as 1478 0 stevel * microseconds. 1479 0 stevel */ 1480 11042 Erik metrics->rmx_rtt = ire->ire_metrics.iulp_rtt * 1000; 1481 0 stevel metrics_set |= RTV_RTT; 1482 11042 Erik metrics->rmx_mtu = ire->ire_metrics.iulp_mtu; 1483 0 stevel metrics_set |= RTV_MTU; 1484 11042 Erik metrics->rmx_ssthresh = ire->ire_metrics.iulp_ssthresh; 1485 0 stevel metrics_set |= RTV_SSTHRESH; 1486 11042 Erik metrics->rmx_rttvar = ire->ire_metrics.iulp_rtt_sd * 1000; 1487 0 stevel metrics_set |= RTV_RTTVAR; 1488 11042 Erik metrics->rmx_sendpipe = ire->ire_metrics.iulp_spipe; 1489 0 stevel metrics_set |= RTV_SPIPE; 1490 11042 Erik metrics->rmx_recvpipe = ire->ire_metrics.iulp_rpipe; 1491 0 stevel metrics_set |= RTV_RPIPE; 1492 0 stevel return (metrics_set); 1493 0 stevel } 1494 11042 Erik 1495 11042 Erik /* 1496 11042 Erik * Given two sets of metrics (src and dst), use the dst values if they are 1497 11042 Erik * set. If a dst value is not set but the src value is set, then we use 1498 11042 Erik * the src value. 1499 11042 Erik * dst is updated with the new values. 1500 11042 Erik * This is used to merge information from a dce_t and ire_metrics, where the 1501 11042 Erik * dce values takes precedence. 1502 11042 Erik */ 1503 11042 Erik void 1504 11042 Erik rts_merge_metrics(iulp_t *dst, const iulp_t *src) 1505 11042 Erik { 1506 11042 Erik if (!src->iulp_set) 1507 11042 Erik return; 1508 11042 Erik 1509 11042 Erik if (dst->iulp_ssthresh == 0) 1510 11042 Erik dst->iulp_ssthresh = src->iulp_ssthresh; 1511 11042 Erik if (dst->iulp_rtt == 0) 1512 11042 Erik dst->iulp_rtt = src->iulp_rtt; 1513 11042 Erik if (dst->iulp_rtt_sd == 0) 1514 11042 Erik dst->iulp_rtt_sd = src->iulp_rtt_sd; 1515 11042 Erik if (dst->iulp_spipe == 0) 1516 11042 Erik dst->iulp_spipe = src->iulp_spipe; 1517 11042 Erik if (dst->iulp_rpipe == 0) 1518 11042 Erik dst->iulp_rpipe = src->iulp_rpipe; 1519 11042 Erik if (dst->iulp_rtomax == 0) 1520 11042 Erik dst->iulp_rtomax = src->iulp_rtomax; 1521 11042 Erik if (dst->iulp_sack == 0) 1522 11042 Erik dst->iulp_sack = src->iulp_sack; 1523 11042 Erik if (dst->iulp_tstamp_ok == 0) 1524 11042 Erik dst->iulp_tstamp_ok = src->iulp_tstamp_ok; 1525 11042 Erik if (dst->iulp_wscale_ok == 0) 1526 11042 Erik dst->iulp_wscale_ok = src->iulp_wscale_ok; 1527 11042 Erik if (dst->iulp_ecn_ok == 0) 1528 11042 Erik dst->iulp_ecn_ok = src->iulp_ecn_ok; 1529 11042 Erik if (dst->iulp_pmtud_ok == 0) 1530 11042 Erik dst->iulp_pmtud_ok = src->iulp_pmtud_ok; 1531 11042 Erik if (dst->iulp_mtu == 0) 1532 11042 Erik dst->iulp_mtu = src->iulp_mtu; 1533 11042 Erik } 1534 11042 Erik 1535 0 stevel 1536 0 stevel /* 1537 0 stevel * Takes a pointer to a routing message and extracts necessary info by looking 1538 0 stevel * at the rtm->rtm_addrs bits and store the requested sockaddrs in the pointers 1539 0 stevel * passed (all of which must be valid). 1540 0 stevel * 1541 0 stevel * The bitmask of sockaddrs actually found in the message is returned, or zero 1542 0 stevel * is returned in the case of an error. 1543 0 stevel */ 1544 0 stevel static int 1545 0 stevel rts_getaddrs(rt_msghdr_t *rtm, in6_addr_t *dst_addrp, in6_addr_t *gw_addrp, 1546 0 stevel in6_addr_t *net_maskp, in6_addr_t *authorp, in6_addr_t *if_addrp, 1547 4823 seb in6_addr_t *in_src_addrp, ushort_t *indexp, sa_family_t *afp, 1548 4823 seb tsol_rtsecattr_t *rtsecattr, int *error) 1549 0 stevel { 1550 0 stevel struct sockaddr *sa; 1551 0 stevel int i; 1552 0 stevel int addr_bits; 1553 0 stevel int length; 1554 0 stevel int found_addrs = 0; 1555 0 stevel caddr_t cp; 1556 0 stevel size_t size; 1557 0 stevel struct sockaddr_dl *sdl; 1558 0 stevel 1559 0 stevel *dst_addrp = ipv6_all_zeros; 1560 0 stevel *gw_addrp = ipv6_all_zeros; 1561 0 stevel *net_maskp = ipv6_all_zeros; 1562 0 stevel *authorp = ipv6_all_zeros; 1563 0 stevel *if_addrp = ipv6_all_zeros; 1564 0 stevel *in_src_addrp = ipv6_all_zeros; 1565 0 stevel *indexp = 0; 1566 0 stevel *afp = AF_UNSPEC; 1567 1676 jpk rtsecattr->rtsa_cnt = 0; 1568 1676 jpk *error = 0; 1569 0 stevel 1570 0 stevel /* 1571 0 stevel * At present we handle only RTA_DST, RTA_GATEWAY, RTA_NETMASK, RTA_IFP, 1572 0 stevel * RTA_IFA and RTA_AUTHOR. The rest will be added as we need them. 1573 0 stevel */ 1574 0 stevel cp = (caddr_t)&rtm[1]; 1575 0 stevel length = rtm->rtm_msglen; 1576 0 stevel for (i = 0; (i < RTA_NUMBITS) && ((cp - (caddr_t)rtm) < length); i++) { 1577 0 stevel /* 1578 0 stevel * The address family we are working with starts out as 1579 0 stevel * AF_UNSPEC, but is set to the one specified with the 1580 0 stevel * destination address. 1581 0 stevel * 1582 0 stevel * If the "working" address family that has been set to 1583 0 stevel * something other than AF_UNSPEC, then the address family of 1584 0 stevel * subsequent sockaddrs must either be AF_UNSPEC (for 1585 0 stevel * compatibility with older programs) or must be the same as our 1586 0 stevel * "working" one. 1587 0 stevel * 1588 0 stevel * This code assumes that RTA_DST (1) comes first in the loop. 1589 0 stevel */ 1590 0 stevel sa = (struct sockaddr *)cp; 1591 0 stevel addr_bits = (rtm->rtm_addrs & (1 << i)); 1592 0 stevel if (addr_bits == 0) 1593 0 stevel continue; 1594 0 stevel switch (addr_bits) { 1595 0 stevel case RTA_DST: 1596 0 stevel size = rts_copyfromsockaddr(sa, dst_addrp); 1597 0 stevel *afp = sa->sa_family; 1598 0 stevel break; 1599 0 stevel case RTA_GATEWAY: 1600 0 stevel if (sa->sa_family != *afp && sa->sa_family != AF_UNSPEC) 1601 0 stevel return (0); 1602 0 stevel size = rts_copyfromsockaddr(sa, gw_addrp); 1603 0 stevel break; 1604 0 stevel case RTA_NETMASK: 1605 0 stevel if (sa->sa_family != *afp && sa->sa_family != AF_UNSPEC) 1606 0 stevel return (0); 1607 0 stevel size = rts_copyfromsockaddr(sa, net_maskp); 1608 0 stevel break; 1609 0 stevel case RTA_IFP: 1610 0 stevel if (sa->sa_family != AF_LINK && 1611 0 stevel sa->sa_family != AF_UNSPEC) 1612 0 stevel return (0); 1613 0 stevel sdl = (struct sockaddr_dl *)cp; 1614 0 stevel *indexp = sdl->sdl_index; 1615 0 stevel size = sizeof (struct sockaddr_dl); 1616 0 stevel break; 1617 0 stevel case RTA_SRC: 1618 0 stevel /* Source address of the incoming packet */ 1619 0 stevel size = rts_copyfromsockaddr(sa, in_src_addrp); 1620 0 stevel *afp = sa->sa_family; 1621 0 stevel break; 1622 0 stevel case RTA_IFA: 1623 0 stevel if (sa->sa_family != *afp && sa->sa_family != AF_UNSPEC) 1624 0 stevel return (0); 1625 0 stevel size = rts_copyfromsockaddr(sa, if_addrp); 1626 0 stevel break; 1627 0 stevel case RTA_AUTHOR: 1628 0 stevel if (sa->sa_family != *afp && sa->sa_family != AF_UNSPEC) 1629 0 stevel return (0); 1630 0 stevel size = rts_copyfromsockaddr(sa, authorp); 1631 0 stevel break; 1632 0 stevel default: 1633 0 stevel return (0); 1634 0 stevel } 1635 0 stevel if (size == 0) 1636 0 stevel return (0); 1637 0 stevel cp += size; 1638 0 stevel found_addrs |= addr_bits; 1639 0 stevel } 1640 1676 jpk 1641 1676 jpk /* 1642 1676 jpk * Parse the routing message and look for any security- 1643 1676 jpk * related attributes for the route. For each valid 1644 1676 jpk * attribute, allocate/obtain the corresponding kernel 1645 1676 jpk * route security attributes. 1646 1676 jpk */ 1647 10138 hn if (((cp - (caddr_t)rtm) < length) && is_system_labeled()) { 1648 10138 hn *error = tsol_rtsa_init(rtm, rtsecattr, cp); 1649 10138 hn ASSERT(rtsecattr->rtsa_cnt <= TSOL_RTSA_REQUEST_MAX); 1650 10138 hn } 1651 1676 jpk 1652 0 stevel return (found_addrs); 1653 0 stevel } 1654 0 stevel 1655 0 stevel /* 1656 0 stevel * Fills the message with the given info. 1657 0 stevel */ 1658 0 stevel static void 1659 0 stevel rts_fill_msg(int type, int rtm_addrs, ipaddr_t dst, ipaddr_t mask, 1660 0 stevel ipaddr_t gateway, ipaddr_t src_addr, ipaddr_t brd_addr, ipaddr_t author, 1661 11042 Erik ipaddr_t ifaddr, const ill_t *ill, mblk_t *mp, 1662 11042 Erik const tsol_gc_t *gc) 1663 0 stevel { 1664 0 stevel rt_msghdr_t *rtm; 1665 0 stevel sin_t *sin; 1666 0 stevel size_t data_size, header_size; 1667 0 stevel uchar_t *cp; 1668 0 stevel int i; 1669 0 stevel 1670 0 stevel ASSERT(mp != NULL); 1671 0 stevel /* 1672 0 stevel * First find the type of the message 1673 0 stevel * and its length. 1674 0 stevel */ 1675 0 stevel header_size = rts_header_msg_size(type); 1676 0 stevel /* 1677 0 stevel * Now find the size of the data 1678 0 stevel * that follows the message header. 1679 0 stevel */ 1680 11042 Erik data_size = rts_data_msg_size(rtm_addrs, AF_INET, gc != NULL ? 1 : 0); 1681 0 stevel 1682 0 stevel rtm = (rt_msghdr_t *)mp->b_rptr; 1683 0 stevel mp->b_wptr = &mp->b_rptr[header_size]; 1684 0 stevel cp = mp->b_wptr; 1685 0 stevel bzero(cp, data_size); 1686 0 stevel for (i = 0; i < RTA_NUMBITS; i++) { 1687 0 stevel sin = (sin_t *)cp; 1688 0 stevel switch (rtm_addrs & (1 << i)) { 1689 0 stevel case RTA_DST: 1690 0 stevel sin->sin_addr.s_addr = dst; 1691 0 stevel sin->sin_family = AF_INET; 1692 0 stevel cp += sizeof (sin_t); 1693 0 stevel break; 1694 0 stevel case RTA_GATEWAY: 1695 0 stevel sin->sin_addr.s_addr = gateway; 1696 0 stevel sin->sin_family = AF_INET; 1697 0 stevel cp += sizeof (sin_t); 1698 0 stevel break; 1699 0 stevel case RTA_NETMASK: 1700 0 stevel sin->sin_addr.s_addr = mask; 1701 0 stevel sin->sin_family = AF_INET; 1702 0 stevel cp += sizeof (sin_t); 1703 0 stevel break; 1704 0 stevel case RTA_IFP: 1705 11042 Erik cp += ill_dls_info((struct sockaddr_dl *)cp, ill); 1706 0 stevel break; 1707 0 stevel case RTA_IFA: 1708 11042 Erik sin->sin_addr.s_addr = ifaddr; 1709 11042 Erik sin->sin_family = AF_INET; 1710 11042 Erik cp += sizeof (sin_t); 1711 11042 Erik break; 1712 0 stevel case RTA_SRC: 1713 0 stevel sin->sin_addr.s_addr = src_addr; 1714 0 stevel sin->sin_family = AF_INET; 1715 0 stevel cp += sizeof (sin_t); 1716 0 stevel break; 1717 0 stevel case RTA_AUTHOR: 1718 0 stevel sin->sin_addr.s_addr = author; 1719 0 stevel sin->sin_family = AF_INET; 1720 0 stevel cp += sizeof (sin_t); 1721 0 stevel break; 1722 0 stevel case RTA_BRD: 1723 0 stevel /* 1724 0 stevel * RTA_BRD is used typically to specify a point-to-point 1725 0 stevel * destination address. 1726 0 stevel */ 1727 0 stevel sin->sin_addr.s_addr = brd_addr; 1728 0 stevel sin->sin_family = AF_INET; 1729 0 stevel cp += sizeof (sin_t); 1730 0 stevel break; 1731 0 stevel } 1732 0 stevel } 1733 1676 jpk 1734 1676 jpk if (gc != NULL) { 1735 1676 jpk rtm_ext_t *rtm_ext; 1736 1676 jpk struct rtsa_s *rp_dst; 1737 1676 jpk tsol_rtsecattr_t *rsap; 1738 1676 jpk 1739 1676 jpk ASSERT(gc->gc_grp != NULL); 1740 1676 jpk ASSERT(RW_LOCK_HELD(&gc->gc_grp->gcgrp_rwlock)); 1741 1676 jpk 1742 1676 jpk rtm_ext = (rtm_ext_t *)cp; 1743 1676 jpk rtm_ext->rtmex_type = RTMEX_GATEWAY_SECATTR; 1744 11042 Erik rtm_ext->rtmex_len = TSOL_RTSECATTR_SIZE(1); 1745 1676 jpk 1746 1676 jpk rsap = (tsol_rtsecattr_t *)(rtm_ext + 1); 1747 11042 Erik rsap->rtsa_cnt = 1; 1748 1676 jpk rp_dst = rsap->rtsa_attr; 1749 1676 jpk 1750 11042 Erik ASSERT(gc->gc_db != NULL); 1751 11042 Erik bcopy(&gc->gc_db->gcdb_attr, rp_dst, sizeof (*rp_dst)); 1752 1676 jpk cp = (uchar_t *)rp_dst; 1753 1676 jpk } 1754 1676 jpk 1755 0 stevel mp->b_wptr = cp; 1756 0 stevel mp->b_cont = NULL; 1757 0 stevel /* 1758 0 stevel * set the fields that are common to 1759 0 stevel * to different messages. 1760 0 stevel */ 1761 0 stevel rtm->rtm_msglen = (short)(header_size + data_size); 1762 0 stevel rtm->rtm_version = RTM_VERSION; 1763 0 stevel rtm->rtm_type = (uchar_t)type; 1764 0 stevel } 1765 0 stevel 1766 0 stevel /* 1767 0 stevel * Allocates and initializes a routing socket message. 1768 11042 Erik * Note that sacnt is either zero or one. 1769 0 stevel */ 1770 0 stevel mblk_t * 1771 1676 jpk rts_alloc_msg(int type, int rtm_addrs, sa_family_t af, uint_t sacnt) 1772 0 stevel { 1773 0 stevel size_t length; 1774 0 stevel mblk_t *mp; 1775 0 stevel 1776 1676 jpk length = RTS_MSG_SIZE(type, rtm_addrs, af, sacnt); 1777 0 stevel mp = allocb(length, BPRI_MED); 1778 0 stevel if (mp == NULL) 1779 0 stevel return (mp); 1780 0 stevel bzero(mp->b_rptr, length); 1781 0 stevel return (mp); 1782 0 stevel } 1783 0 stevel 1784 0 stevel /* 1785 0 stevel * Returns the size of the routing 1786 0 stevel * socket message header size. 1787 0 stevel */ 1788 0 stevel size_t 1789 0 stevel rts_header_msg_size(int type) 1790 0 stevel { 1791 0 stevel switch (type) { 1792 0 stevel case RTM_DELADDR: 1793 0 stevel case RTM_NEWADDR: 1794 11076 Cathy case RTM_CHGADDR: 1795 11076 Cathy case RTM_FREEADDR: 1796 0 stevel return (sizeof (ifa_msghdr_t)); 1797 0 stevel case RTM_IFINFO: 1798 0 stevel return (sizeof (if_msghdr_t)); 1799 0 stevel default: 1800 0 stevel return (sizeof (rt_msghdr_t)); 1801 0 stevel } 1802 0 stevel } 1803 0 stevel 1804 0 stevel /* 1805 0 stevel * Returns the size of the message needed with the given rtm_addrs and family. 1806 0 stevel * 1807 0 stevel * It is assumed that all of the sockaddrs (with the exception of RTA_IFP) are 1808 0 stevel * of the same family (currently either AF_INET or AF_INET6). 1809 0 stevel */ 1810 0 stevel size_t 1811 1676 jpk rts_data_msg_size(int rtm_addrs, sa_family_t af, uint_t sacnt) 1812 0 stevel { 1813 0 stevel int i; 1814 0 stevel size_t length = 0; 1815 0 stevel 1816 0 stevel for (i = 0; i < RTA_NUMBITS; i++) { 1817 0 stevel switch (rtm_addrs & (1 << i)) { 1818 0 stevel case RTA_IFP: 1819 0 stevel length += sizeof (struct sockaddr_dl); 1820 0 stevel break; 1821 0 stevel case RTA_DST: 1822 0 stevel case RTA_GATEWAY: 1823 0 stevel case RTA_NETMASK: 1824 0 stevel case RTA_SRC: 1825 0 stevel case RTA_IFA: 1826 0 stevel case RTA_AUTHOR: 1827 0 stevel case RTA_BRD: 1828 0 stevel ASSERT(af == AF_INET || af == AF_INET6); 1829 0 stevel switch (af) { 1830 0 stevel case AF_INET: 1831 0 stevel length += sizeof (sin_t); 1832 0 stevel break; 1833 0 stevel case AF_INET6: 1834 0 stevel length += sizeof (sin6_t); 1835 0 stevel break; 1836 0 stevel } 1837 0 stevel break; 1838 0 stevel } 1839 0 stevel } 1840 1676 jpk if (sacnt > 0) 1841 1676 jpk length += sizeof (rtm_ext_t) + TSOL_RTSECATTR_SIZE(sacnt); 1842 1676 jpk 1843 0 stevel return (length); 1844 0 stevel } 1845 0 stevel 1846 0 stevel /* 1847 0 stevel * This routine is called to generate a message to the routing 1848 0 stevel * socket indicating that a redirect has occured, a routing lookup 1849 0 stevel * has failed, or that a protocol has detected timeouts to a particular 1850 0 stevel * destination. This routine is called for message types RTM_LOSING, 1851 0 stevel * RTM_REDIRECT, and RTM_MISS. 1852 0 stevel */ 1853 0 stevel void 1854 0 stevel ip_rts_change(int type, ipaddr_t dst_addr, ipaddr_t gw_addr, ipaddr_t net_mask, 1855 3448 dh155122 ipaddr_t source, ipaddr_t author, int flags, int error, int rtm_addrs, 1856 3448 dh155122 ip_stack_t *ipst) 1857 0 stevel { 1858 0 stevel rt_msghdr_t *rtm; 1859 0 stevel mblk_t *mp; 1860 0 stevel 1861 0 stevel if (rtm_addrs == 0) 1862 0 stevel return; 1863 1676 jpk mp = rts_alloc_msg(type, rtm_addrs, AF_INET, 0); 1864 0 stevel if (mp == NULL) 1865 0 stevel return; 1866 0 stevel rts_fill_msg(type, rtm_addrs, dst_addr, net_mask, gw_addr, source, 0, 1867 11042 Erik author, 0, NULL, mp, NULL); 1868 0 stevel rtm = (rt_msghdr_t *)mp->b_rptr; 1869 0 stevel rtm->rtm_flags = flags; 1870 0 stevel rtm->rtm_errno = error; 1871 0 stevel rtm->rtm_flags |= RTF_DONE; 1872 0 stevel rtm->rtm_addrs = rtm_addrs; 1873 8485 Peter rts_queue_input(mp, NULL, AF_INET, RTSQ_ALL, ipst); 1874 0 stevel } 1875 0 stevel 1876 0 stevel /* 1877 0 stevel * This routine is called to generate a message to the routing 1878 0 stevel * socket indicating that the status of a network interface has changed. 1879 0 stevel * Message type generated RTM_IFINFO. 1880 0 stevel */ 1881 0 stevel void 1882 8485 Peter ip_rts_ifmsg(const ipif_t *ipif, uint_t flags) 1883 8485 Peter { 1884 8485 Peter ip_rts_xifmsg(ipif, 0, 0, flags); 1885 8485 Peter } 1886 8485 Peter 1887 8485 Peter void 1888 8485 Peter ip_rts_xifmsg(const ipif_t *ipif, uint64_t set, uint64_t clear, uint_t flags) 1889 0 stevel { 1890 0 stevel if_msghdr_t *ifm; 1891 0 stevel mblk_t *mp; 1892 0 stevel sa_family_t af; 1893 3448 dh155122 ip_stack_t *ipst = ipif->ipif_ill->ill_ipst; 1894 0 stevel 1895 0 stevel /* 1896 11042 Erik * This message should be generated only 1897 11042 Erik * when the physical device is changing 1898 11042 Erik * state. 1899 0 stevel */ 1900 0 stevel if (ipif->ipif_id != 0) 1901 0 stevel return; 1902 0 stevel if (ipif->ipif_isv6) { 1903 0 stevel af = AF_INET6; 1904 1676 jpk mp = rts_alloc_msg(RTM_IFINFO, RTA_IFP, af, 0); 1905 0 stevel if (mp == NULL) 1906 0 stevel return; 1907 0 stevel rts_fill_msg_v6(RTM_IFINFO, RTA_IFP, &ipv6_all_zeros, 1908 0 stevel &ipv6_all_zeros, &ipv6_all_zeros, &ipv6_all_zeros, 1909 11042 Erik &ipv6_all_zeros, &ipv6_all_zeros, &ipv6_all_zeros, 1910 11042 Erik ipif->ipif_ill, mp, NULL); 1911 0 stevel } else { 1912 0 stevel af = AF_INET; 1913 1676 jpk mp = rts_alloc_msg(RTM_IFINFO, RTA_IFP, af, 0); 1914 0 stevel if (mp == NULL) 1915 0 stevel return; 1916 11042 Erik rts_fill_msg(RTM_IFINFO, RTA_IFP, 0, 0, 0, 0, 0, 0, 0, 1917 11042 Erik ipif->ipif_ill, mp, NULL); 1918 0 stevel } 1919 0 stevel ifm = (if_msghdr_t *)mp->b_rptr; 1920 0 stevel ifm->ifm_index = ipif->ipif_ill->ill_phyint->phyint_ifindex; 1921 8485 Peter ifm->ifm_flags = (ipif->ipif_flags | ipif->ipif_ill->ill_flags | 1922 8485 Peter ipif->ipif_ill->ill_phyint->phyint_flags | set) & ~clear; 1923 0 stevel rts_getifdata(&ifm->ifm_data, ipif); 1924 0 stevel ifm->ifm_addrs = RTA_IFP; 1925 8485 Peter 1926 8485 Peter if (flags & RTSQ_DEFAULT) { 1927 8485 Peter flags = RTSQ_ALL; 1928 8485 Peter /* 1929 8485 Peter * If this message is for an underlying interface, prevent 1930 8485 Peter * "normal" (IPMP-unaware) routing sockets from seeing it. 1931 8485 Peter */ 1932 8485 Peter if (IS_UNDER_IPMP(ipif->ipif_ill)) 1933 8485 Peter flags &= ~RTSQ_NORMAL; 1934 8485 Peter } 1935 8485 Peter 1936 8485 Peter rts_queue_input(mp, NULL, af, flags, ipst); 1937 0 stevel } 1938 0 stevel 1939 0 stevel /* 1940 11076 Cathy * If cmd is RTM_ADD or RTM_DELETE, generate the rt_msghdr_t message; 1941 11076 Cathy * otherwise (RTM_NEWADDR, RTM_DELADDR, RTM_CHGADDR and RTM_FREEADDR) 1942 11076 Cathy * generate the ifa_msghdr_t message. 1943 0 stevel */ 1944 11076 Cathy static void 1945 11076 Cathy rts_new_rtsmsg(int cmd, int error, const ipif_t *ipif, uint_t flags) 1946 0 stevel { 1947 0 stevel int rtm_addrs; 1948 0 stevel mblk_t *mp; 1949 0 stevel ifa_msghdr_t *ifam; 1950 0 stevel rt_msghdr_t *rtm; 1951 0 stevel sa_family_t af; 1952 3448 dh155122 ip_stack_t *ipst = ipif->ipif_ill->ill_ipst; 1953 0 stevel 1954 11042 Erik /* 1955 11076 Cathy * Do not report unspecified address if this is the RTM_CHGADDR or 1956 11076 Cathy * RTM_FREEADDR message. 1957 11042 Erik */ 1958 11076 Cathy if (cmd == RTM_CHGADDR || cmd == RTM_FREEADDR) { 1959 11076 Cathy if (!ipif->ipif_isv6) { 1960 11076 Cathy if (ipif->ipif_lcl_addr == INADDR_ANY) 1961 11076 Cathy return; 1962 11076 Cathy } else if (IN6_IS_ADDR_UNSPECIFIED(&ipif->ipif_v6lcl_addr)) { 1963 11076 Cathy return; 1964 11076 Cathy } 1965 11076 Cathy } 1966 11042 Erik 1967 0 stevel if (ipif->ipif_isv6) 1968 0 stevel af = AF_INET6; 1969 0 stevel else 1970 0 stevel af = AF_INET; 1971 11076 Cathy 1972 11076 Cathy if (cmd == RTM_ADD || cmd == RTM_DELETE) 1973 11076 Cathy rtm_addrs = (RTA_DST | RTA_NETMASK); 1974 11076 Cathy else 1975 11076 Cathy rtm_addrs = (RTA_IFA | RTA_NETMASK | RTA_BRD | RTA_IFP); 1976 11076 Cathy 1977 11076 Cathy mp = rts_alloc_msg(cmd, rtm_addrs, af, 0); 1978 11076 Cathy if (mp == NULL) 1979 11076 Cathy return; 1980 11076 Cathy 1981 11076 Cathy if (cmd != RTM_ADD && cmd != RTM_DELETE) { 1982 11076 Cathy switch (af) { 1983 11076 Cathy case AF_INET: 1984 11076 Cathy rts_fill_msg(cmd, rtm_addrs, 0, 1985 11076 Cathy ipif->ipif_net_mask, 0, ipif->ipif_lcl_addr, 1986 11076 Cathy ipif->ipif_pp_dst_addr, 0, 1987 11076 Cathy ipif->ipif_lcl_addr, ipif->ipif_ill, 1988 11076 Cathy mp, NULL); 1989 11076 Cathy break; 1990 11076 Cathy case AF_INET6: 1991 11076 Cathy rts_fill_msg_v6(cmd, rtm_addrs, 1992 11076 Cathy &ipv6_all_zeros, &ipif->ipif_v6net_mask, 1993 11076 Cathy &ipv6_all_zeros, &ipif->ipif_v6lcl_addr, 1994 11076 Cathy &ipif->ipif_v6pp_dst_addr, &ipv6_all_zeros, 1995 11076 Cathy &ipif->ipif_v6lcl_addr, ipif->ipif_ill, 1996 11076 Cathy mp, NULL); 1997 11076 Cathy break; 1998 11076 Cathy } 1999 11076 Cathy ifam = (ifa_msghdr_t *)mp->b_rptr; 2000 11076 Cathy ifam->ifam_index = 2001 11076 Cathy ipif->ipif_ill->ill_phyint->phyint_ifindex; 2002 11076 Cathy ifam->ifam_metric = ipif->ipif_metric; 2003 11076 Cathy ifam->ifam_flags = ((cmd == RTM_NEWADDR) ? RTF_UP : 0); 2004 11076 Cathy ifam->ifam_addrs = rtm_addrs; 2005 11076 Cathy } else { 2006 11076 Cathy switch (af) { 2007 11076 Cathy case AF_INET: 2008 11076 Cathy rts_fill_msg(cmd, rtm_addrs, 2009 11076 Cathy ipif->ipif_lcl_addr, ipif->ipif_net_mask, 0, 2010 11076 Cathy 0, 0, 0, 0, NULL, mp, NULL); 2011 11076 Cathy break; 2012 11076 Cathy case AF_INET6: 2013 11076 Cathy rts_fill_msg_v6(cmd, rtm_addrs, 2014 11076 Cathy &ipif->ipif_v6lcl_addr, 2015 11076 Cathy &ipif->ipif_v6net_mask, &ipv6_all_zeros, 2016 11076 Cathy &ipv6_all_zeros, &ipv6_all_zeros, 2017 11076 Cathy &ipv6_all_zeros, &ipv6_all_zeros, 2018 11076 Cathy NULL, mp, NULL); 2019 11076 Cathy break; 2020 11076 Cathy } 2021 11076 Cathy rtm = (rt_msghdr_t *)mp->b_rptr; 2022 11076 Cathy rtm->rtm_index = 2023 11076 Cathy ipif->ipif_ill->ill_phyint->phyint_ifindex; 2024 11076 Cathy rtm->rtm_flags = ((cmd == RTM_ADD) ? RTF_UP : 0); 2025 11076 Cathy rtm->rtm_errno = error; 2026 11076 Cathy if (error == 0) 2027 11076 Cathy rtm->rtm_flags |= RTF_DONE; 2028 11076 Cathy rtm->rtm_addrs = rtm_addrs; 2029 11076 Cathy } 2030 11076 Cathy rts_queue_input(mp, NULL, af, flags, ipst); 2031 11076 Cathy } 2032 11076 Cathy 2033 11076 Cathy /* 2034 11076 Cathy * This is called to generate messages to the routing socket 2035 11076 Cathy * indicating a network interface has had addresses associated with it. 2036 11076 Cathy * The structure of the code is based on the 4.4BSD-Lite2 <net/rtsock.c>. 2037 11076 Cathy */ 2038 11076 Cathy void 2039 11076 Cathy ip_rts_newaddrmsg(int cmd, int error, const ipif_t *ipif, uint_t flags) 2040 11076 Cathy { 2041 11076 Cathy ip_stack_t *ipst = ipif->ipif_ill->ill_ipst; 2042 8485 Peter 2043 8485 Peter if (flags & RTSQ_DEFAULT) { 2044 8485 Peter flags = RTSQ_ALL; 2045 8485 Peter /* 2046 8485 Peter * If this message is for an underlying interface, prevent 2047 8485 Peter * "normal" (IPMP-unaware) routing sockets from seeing it. 2048 8485 Peter */ 2049 8485 Peter if (IS_UNDER_IPMP(ipif->ipif_ill)) 2050 8485 Peter flags &= ~RTSQ_NORMAL; 2051 8485 Peter } 2052 8485 Peter 2053 0 stevel /* 2054 11076 Cathy * Let conn_ixa caching know that source address selection 2055 11076 Cathy * changed 2056 11076 Cathy */ 2057 11076 Cathy if (cmd == RTM_ADD || cmd == RTM_DELETE) 2058 11076 Cathy ip_update_source_selection(ipst); 2059 11076 Cathy 2060 11076 Cathy /* 2061 0 stevel * If the request is DELETE, send RTM_DELETE and RTM_DELADDR. 2062 0 stevel * if the request is ADD, send RTM_NEWADDR and RTM_ADD. 2063 11076 Cathy * otherwise simply send the request. 2064 0 stevel */ 2065 11076 Cathy switch (cmd) { 2066 11076 Cathy case RTM_ADD: 2067 11076 Cathy rts_new_rtsmsg(RTM_NEWADDR, error, ipif, flags); 2068 11076 Cathy rts_new_rtsmsg(RTM_ADD, error, ipif, flags); 2069 11076 Cathy break; 2070 11076 Cathy case RTM_DELETE: 2071 11076 Cathy rts_new_rtsmsg(RTM_DELETE, error, ipif, flags); 2072 11076 Cathy rts_new_rtsmsg(RTM_DELADDR, error, ipif, flags); 2073 11076 Cathy break; 2074 11076 Cathy default: 2075 11076 Cathy rts_new_rtsmsg(cmd, error, ipif, flags); 2076 11076 Cathy break; 2077 0 stevel } 2078 0 stevel } 2079 0 stevel 2080 0 stevel /* 2081 0 stevel * Based on the address family specified in a sockaddr, copy the address field 2082 0 stevel * into an in6_addr_t. 2083 0 stevel * 2084 0 stevel * In the case of AF_UNSPEC, we assume the family is actually AF_INET for 2085 0 stevel * compatibility with programs that leave the family cleared in the sockaddr. 2086 0 stevel * Callers of rts_copyfromsockaddr should check the family themselves if they 2087 0 stevel * wish to verify its value. 2088 0 stevel * 2089 0 stevel * In the case of AF_INET6, a check is made to ensure that address is not an 2090 0 stevel * IPv4-mapped address. 2091 0 stevel */ 2092 0 stevel size_t 2093 0 stevel rts_copyfromsockaddr(struct sockaddr *sa, in6_addr_t *addrp) 2094 0 stevel { 2095 0 stevel switch (sa->sa_family) { 2096 0 stevel case AF_INET: 2097 0 stevel case AF_UNSPEC: 2098 0 stevel IN6_IPADDR_TO_V4MAPPED(((sin_t *)sa)->sin_addr.s_addr, addrp); 2099 0 stevel return (sizeof (sin_t)); 2100 0 stevel case AF_INET6: 2101 0 stevel *addrp = ((sin6_t *)sa)->sin6_addr; 2102 0 stevel if (IN6_IS_ADDR_V4MAPPED(addrp)) 2103 0 stevel return (0); 2104 0 stevel return (sizeof (sin6_t)); 2105 0 stevel default: 2106 0 stevel return (0); 2107 0 stevel } 2108 0 stevel } 2109
Indexes created Sat Nov 28 12:39:51 UTC 2009
Terms of Use |
Privacy |
Trademarks |
Copyright Policy |
Site Guidelines |
Help
Your use of this web site or any of its content or software indicates your agreement to be bound by these Terms of Use.
Copyright © 1995-2008 Sun Microsystems, Inc.