1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 2004 Luigi Rizzo, Alessandro Cerri. All rights reserved. 5 * Copyright (c) 2004-2008 Qing Li. All rights reserved. 6 * Copyright (c) 2008 Kip Macy. All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 17 * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 */ 29 #include <sys/cdefs.h> 30 __FBSDID("$FreeBSD$"); 31 32 #include "opt_ddb.h" 33 #include "opt_inet.h" 34 #include "opt_inet6.h" 35 36 #include <sys/param.h> 37 #include <sys/systm.h> 38 #include <sys/eventhandler.h> 39 #include <sys/malloc.h> 40 #include <sys/mbuf.h> 41 #include <sys/syslog.h> 42 #include <sys/sysctl.h> 43 #include <sys/socket.h> 44 #include <sys/kernel.h> 45 #include <sys/lock.h> 46 #include <sys/mutex.h> 47 #include <sys/rwlock.h> 48 49 #ifdef DDB 50 #include <ddb/ddb.h> 51 #endif 52 53 #include <vm/uma.h> 54 55 #include <netinet/in.h> 56 #include <net/if_llatbl.h> 57 #include <net/if.h> 58 #include <net/if_dl.h> 59 #include <net/if_var.h> 60 #include <net/route.h> 61 #include <net/route/route_ctl.h> 62 #include <net/vnet.h> 63 #include <netinet/if_ether.h> 64 #include <netinet6/in6_var.h> 65 #include <netinet6/nd6.h> 66 67 MALLOC_DEFINE(M_LLTABLE, "lltable", "link level address tables"); 68 69 VNET_DEFINE_STATIC(SLIST_HEAD(, lltable), lltables) = 70 SLIST_HEAD_INITIALIZER(lltables); 71 #define V_lltables VNET(lltables) 72 73 static struct rwlock lltable_list_lock; 74 RW_SYSINIT(lltable_list_lock, &lltable_list_lock, "lltable_list_lock"); 75 #define LLTABLE_LIST_RLOCK() rw_rlock(&lltable_list_lock) 76 #define LLTABLE_LIST_RUNLOCK() rw_runlock(&lltable_list_lock) 77 #define LLTABLE_LIST_WLOCK() rw_wlock(&lltable_list_lock) 78 #define LLTABLE_LIST_WUNLOCK() rw_wunlock(&lltable_list_lock) 79 #define LLTABLE_LIST_LOCK_ASSERT() rw_assert(&lltable_list_lock, RA_LOCKED) 80 81 static void lltable_unlink(struct lltable *llt); 82 static void llentries_unlink(struct lltable *llt, struct llentries *head); 83 84 /* 85 * Dump lle state for a specific address family. 86 */ 87 static int 88 lltable_dump_af(struct lltable *llt, struct sysctl_req *wr) 89 { 90 struct epoch_tracker et; 91 int error; 92 93 LLTABLE_LIST_LOCK_ASSERT(); 94 95 if (llt->llt_ifp->if_flags & IFF_LOOPBACK) 96 return (0); 97 error = 0; 98 99 NET_EPOCH_ENTER(et); 100 error = lltable_foreach_lle(llt, 101 (llt_foreach_cb_t *)llt->llt_dump_entry, wr); 102 NET_EPOCH_EXIT(et); 103 104 return (error); 105 } 106 107 /* 108 * Dump arp state for a specific address family. 109 */ 110 int 111 lltable_sysctl_dumparp(int af, struct sysctl_req *wr) 112 { 113 struct lltable *llt; 114 int error = 0; 115 116 LLTABLE_LIST_RLOCK(); 117 SLIST_FOREACH(llt, &V_lltables, llt_link) { 118 if (llt->llt_af == af) { 119 error = lltable_dump_af(llt, wr); 120 if (error != 0) 121 goto done; 122 } 123 } 124 done: 125 LLTABLE_LIST_RUNLOCK(); 126 return (error); 127 } 128 129 /* 130 * Common function helpers for chained hash table. 131 */ 132 133 /* 134 * Runs specified callback for each entry in @llt. 135 * Caller does the locking. 136 * 137 */ 138 static int 139 htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg) 140 { 141 struct llentry *lle, *next; 142 int i, error; 143 144 error = 0; 145 146 for (i = 0; i < llt->llt_hsize; i++) { 147 CK_LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) { 148 error = f(llt, lle, farg); 149 if (error != 0) 150 break; 151 } 152 } 153 154 return (error); 155 } 156 157 /* 158 * The htable_[un]link_entry() functions return: 159 * 0 if the entry was (un)linked already and nothing changed, 160 * 1 if the entry was added/removed to/from the table, and 161 * -1 on error (e.g., not being able to add the entry due to limits reached). 162 * While the "unlink" operation should never error, callers of 163 * lltable_link_entry() need to check for errors and handle them. 164 */ 165 static int 166 htable_link_entry(struct lltable *llt, struct llentry *lle) 167 { 168 struct llentries *lleh; 169 uint32_t hashidx; 170 171 if ((lle->la_flags & LLE_LINKED) != 0) 172 return (0); 173 174 IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp); 175 176 if (llt->llt_maxentries > 0 && 177 llt->llt_entries >= llt->llt_maxentries) 178 return (-1); 179 180 hashidx = llt->llt_hash(lle, llt->llt_hsize); 181 lleh = &llt->lle_head[hashidx]; 182 183 lle->lle_tbl = llt; 184 lle->lle_head = lleh; 185 lle->la_flags |= LLE_LINKED; 186 CK_LIST_INSERT_HEAD(lleh, lle, lle_next); 187 llt->llt_entries++; 188 189 return (1); 190 } 191 192 static int 193 htable_unlink_entry(struct llentry *lle) 194 { 195 struct lltable *llt; 196 197 if ((lle->la_flags & LLE_LINKED) == 0) 198 return (0); 199 200 llt = lle->lle_tbl; 201 IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp); 202 KASSERT(llt->llt_entries > 0, ("%s: lltable %p (%s) entries %d <= 0", 203 __func__, llt, if_name(llt->llt_ifp), llt->llt_entries)); 204 205 CK_LIST_REMOVE(lle, lle_next); 206 lle->la_flags &= ~(LLE_VALID | LLE_LINKED); 207 #if 0 208 lle->lle_tbl = NULL; 209 lle->lle_head = NULL; 210 #endif 211 llt->llt_entries--; 212 213 return (1); 214 } 215 216 struct prefix_match_data { 217 const struct sockaddr *addr; 218 const struct sockaddr *mask; 219 struct llentries dchain; 220 u_int flags; 221 }; 222 223 static int 224 htable_prefix_free_cb(struct lltable *llt, struct llentry *lle, void *farg) 225 { 226 struct prefix_match_data *pmd; 227 228 pmd = (struct prefix_match_data *)farg; 229 230 if (llt->llt_match_prefix(pmd->addr, pmd->mask, pmd->flags, lle)) { 231 LLE_WLOCK(lle); 232 CK_LIST_INSERT_HEAD(&pmd->dchain, lle, lle_chain); 233 } 234 235 return (0); 236 } 237 238 static void 239 htable_prefix_free(struct lltable *llt, const struct sockaddr *addr, 240 const struct sockaddr *mask, u_int flags) 241 { 242 struct llentry *lle, *next; 243 struct prefix_match_data pmd; 244 245 bzero(&pmd, sizeof(pmd)); 246 pmd.addr = addr; 247 pmd.mask = mask; 248 pmd.flags = flags; 249 CK_LIST_INIT(&pmd.dchain); 250 251 IF_AFDATA_WLOCK(llt->llt_ifp); 252 /* Push matching lles to chain */ 253 lltable_foreach_lle(llt, htable_prefix_free_cb, &pmd); 254 255 llentries_unlink(llt, &pmd.dchain); 256 IF_AFDATA_WUNLOCK(llt->llt_ifp); 257 258 CK_LIST_FOREACH_SAFE(lle, &pmd.dchain, lle_chain, next) 259 lltable_free_entry(llt, lle); 260 } 261 262 static void 263 htable_free_tbl(struct lltable *llt) 264 { 265 266 free(llt->lle_head, M_LLTABLE); 267 free(llt, M_LLTABLE); 268 } 269 270 static void 271 llentries_unlink(struct lltable *llt, struct llentries *head) 272 { 273 struct llentry *lle, *next; 274 275 CK_LIST_FOREACH_SAFE(lle, head, lle_chain, next) 276 llt->llt_unlink_entry(lle); 277 } 278 279 /* 280 * Helper function used to drop all mbufs in hold queue. 281 * 282 * Returns the number of held packets, if any, that were dropped. 283 */ 284 size_t 285 lltable_drop_entry_queue(struct llentry *lle) 286 { 287 size_t pkts_dropped; 288 struct mbuf *next; 289 290 LLE_WLOCK_ASSERT(lle); 291 292 pkts_dropped = 0; 293 while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) { 294 next = lle->la_hold->m_nextpkt; 295 m_freem(lle->la_hold); 296 lle->la_hold = next; 297 lle->la_numheld--; 298 pkts_dropped++; 299 } 300 301 KASSERT(lle->la_numheld == 0, 302 ("%s: la_numheld %d > 0, pkts_droped %zd", __func__, 303 lle->la_numheld, pkts_dropped)); 304 305 return (pkts_dropped); 306 } 307 308 void 309 lltable_set_entry_addr(struct ifnet *ifp, struct llentry *lle, 310 const char *linkhdr, size_t linkhdrsize, int lladdr_off) 311 { 312 313 memcpy(lle->r_linkdata, linkhdr, linkhdrsize); 314 lle->r_hdrlen = linkhdrsize; 315 lle->ll_addr = &lle->r_linkdata[lladdr_off]; 316 lle->la_flags |= LLE_VALID; 317 lle->r_flags |= RLLE_VALID; 318 } 319 320 /* 321 * Acquires lltable write lock. 322 * 323 * Returns true on success, with both lltable and lle lock held. 324 * On failure, false is returned and lle wlock is still held. 325 */ 326 bool 327 lltable_acquire_wlock(struct ifnet *ifp, struct llentry *lle) 328 { 329 NET_EPOCH_ASSERT(); 330 331 /* Perform real LLE update */ 332 /* use afdata WLOCK to update fields */ 333 LLE_WUNLOCK(lle); 334 IF_AFDATA_WLOCK(ifp); 335 LLE_WLOCK(lle); 336 337 /* 338 * Since we droppped LLE lock, other thread might have deleted 339 * this lle. Check and return 340 */ 341 if ((lle->la_flags & LLE_DELETED) != 0) { 342 IF_AFDATA_WUNLOCK(ifp); 343 return (false); 344 } 345 346 return (true); 347 } 348 349 /* 350 * Tries to update @lle link-level address. 351 * Since update requires AFDATA WLOCK, function 352 * drops @lle lock, acquires AFDATA lock and then acquires 353 * @lle lock to maintain lock order. 354 * 355 * Returns 1 on success. 356 */ 357 int 358 lltable_try_set_entry_addr(struct ifnet *ifp, struct llentry *lle, 359 const char *linkhdr, size_t linkhdrsize, int lladdr_off) 360 { 361 362 if (!lltable_acquire_wlock(ifp, lle)) 363 return (0); 364 365 /* Update data */ 366 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize, lladdr_off); 367 368 IF_AFDATA_WUNLOCK(ifp); 369 370 return (1); 371 } 372 373 /* 374 * Helper function used to pre-compute full/partial link-layer 375 * header data suitable for feeding into if_output(). 376 */ 377 int 378 lltable_calc_llheader(struct ifnet *ifp, int family, char *lladdr, 379 char *buf, size_t *bufsize, int *lladdr_off) 380 { 381 struct if_encap_req ereq; 382 int error; 383 384 bzero(buf, *bufsize); 385 bzero(&ereq, sizeof(ereq)); 386 ereq.buf = buf; 387 ereq.bufsize = *bufsize; 388 ereq.rtype = IFENCAP_LL; 389 ereq.family = family; 390 ereq.lladdr = lladdr; 391 ereq.lladdr_len = ifp->if_addrlen; 392 error = ifp->if_requestencap(ifp, &ereq); 393 if (error == 0) { 394 *bufsize = ereq.bufsize; 395 *lladdr_off = ereq.lladdr_off; 396 } 397 398 return (error); 399 } 400 401 /* 402 * Searches for the child entry matching @family inside @lle. 403 * Returns the entry or NULL. 404 */ 405 struct llentry * 406 llentry_lookup_family(struct llentry *lle, int family) 407 { 408 struct llentry *child_lle; 409 410 if (lle == NULL) 411 return (NULL); 412 413 CK_SLIST_FOREACH(child_lle, &lle->lle_children, lle_child_next) { 414 if (child_lle->r_family == family) 415 return (child_lle); 416 } 417 418 return (NULL); 419 } 420 421 /* 422 * Requests feedback from the datapath. 423 * First packet using @lle should result in 424 * setting r_skip_req back to 0 and updating 425 * lle_hittime to the current time_uptime. 426 */ 427 void 428 llentry_request_feedback(struct llentry *lle) 429 { 430 struct llentry *child_lle; 431 432 LLE_REQ_LOCK(lle); 433 lle->r_skip_req = 1; 434 LLE_REQ_UNLOCK(lle); 435 436 CK_SLIST_FOREACH(child_lle, &lle->lle_children, lle_child_next) { 437 LLE_REQ_LOCK(child_lle); 438 child_lle->r_skip_req = 1; 439 LLE_REQ_UNLOCK(child_lle); 440 } 441 } 442 443 /* 444 * Updates the lle state to mark it has been used 445 * and record the time. 446 * Used by the llentry_provide_feedback() wrapper. 447 */ 448 void 449 llentry_mark_used(struct llentry *lle) 450 { 451 LLE_REQ_LOCK(lle); 452 lle->r_skip_req = 0; 453 lle->lle_hittime = time_uptime; 454 LLE_REQ_UNLOCK(lle); 455 } 456 457 /* 458 * Fetches the time when lle was used. 459 * Return 0 if the entry was not used, relevant time_uptime 460 * otherwise. 461 */ 462 static time_t 463 llentry_get_hittime_raw(struct llentry *lle) 464 { 465 time_t lle_hittime = 0; 466 467 LLE_REQ_LOCK(lle); 468 if ((lle->r_skip_req == 0) && (lle_hittime < lle->lle_hittime)) 469 lle_hittime = lle->lle_hittime; 470 LLE_REQ_UNLOCK(lle); 471 472 return (lle_hittime); 473 } 474 475 time_t 476 llentry_get_hittime(struct llentry *lle) 477 { 478 time_t lle_hittime = 0; 479 struct llentry *child_lle; 480 481 lle_hittime = llentry_get_hittime_raw(lle); 482 483 CK_SLIST_FOREACH(child_lle, &lle->lle_children, lle_child_next) { 484 time_t hittime = llentry_get_hittime_raw(child_lle); 485 if (hittime > lle_hittime) 486 lle_hittime = hittime; 487 } 488 489 return (lle_hittime); 490 } 491 492 /* 493 * Update link-layer header for given @lle after 494 * interface lladdr was changed. 495 */ 496 static int 497 llentry_update_ifaddr(struct lltable *llt, struct llentry *lle, void *farg) 498 { 499 struct ifnet *ifp; 500 u_char linkhdr[LLE_MAX_LINKHDR]; 501 size_t linkhdrsize; 502 u_char *lladdr; 503 int lladdr_off; 504 505 ifp = (struct ifnet *)farg; 506 507 lladdr = lle->ll_addr; 508 509 LLE_WLOCK(lle); 510 if ((lle->la_flags & LLE_VALID) == 0) { 511 LLE_WUNLOCK(lle); 512 return (0); 513 } 514 515 if ((lle->la_flags & LLE_IFADDR) != 0) 516 lladdr = IF_LLADDR(ifp); 517 518 linkhdrsize = sizeof(linkhdr); 519 lltable_calc_llheader(ifp, llt->llt_af, lladdr, linkhdr, &linkhdrsize, 520 &lladdr_off); 521 memcpy(lle->r_linkdata, linkhdr, linkhdrsize); 522 LLE_WUNLOCK(lle); 523 524 return (0); 525 } 526 527 /* 528 * Update all calculated headers for given @llt 529 */ 530 void 531 lltable_update_ifaddr(struct lltable *llt) 532 { 533 534 if (llt->llt_ifp->if_flags & IFF_LOOPBACK) 535 return; 536 537 IF_AFDATA_WLOCK(llt->llt_ifp); 538 lltable_foreach_lle(llt, llentry_update_ifaddr, llt->llt_ifp); 539 IF_AFDATA_WUNLOCK(llt->llt_ifp); 540 } 541 542 /* 543 * 544 * Performs generic cleanup routines and frees lle. 545 * 546 * Called for non-linked entries, with callouts and 547 * other AF-specific cleanups performed. 548 * 549 * @lle must be passed WLOCK'ed 550 * 551 * Returns the number of held packets, if any, that were dropped. 552 */ 553 size_t 554 llentry_free(struct llentry *lle) 555 { 556 size_t pkts_dropped; 557 558 LLE_WLOCK_ASSERT(lle); 559 560 KASSERT((lle->la_flags & LLE_LINKED) == 0, ("freeing linked lle")); 561 562 pkts_dropped = lltable_drop_entry_queue(lle); 563 564 /* cancel timer */ 565 if (callout_stop(&lle->lle_timer) > 0) 566 LLE_REMREF(lle); 567 LLE_FREE_LOCKED(lle); 568 569 return (pkts_dropped); 570 } 571 572 /* 573 * Free all entries from given table and free itself. 574 */ 575 576 static int 577 lltable_free_cb(struct lltable *llt, struct llentry *lle, void *farg) 578 { 579 struct llentries *dchain; 580 581 dchain = (struct llentries *)farg; 582 583 LLE_WLOCK(lle); 584 CK_LIST_INSERT_HEAD(dchain, lle, lle_chain); 585 586 return (0); 587 } 588 589 /* 590 * Free all entries from given table and free itself. 591 */ 592 void 593 lltable_free(struct lltable *llt) 594 { 595 struct llentry *lle, *next; 596 struct llentries dchain; 597 598 KASSERT(llt != NULL, ("%s: llt is NULL", __func__)); 599 600 lltable_unlink(llt); 601 602 CK_LIST_INIT(&dchain); 603 IF_AFDATA_WLOCK(llt->llt_ifp); 604 /* Push all lles to @dchain */ 605 lltable_foreach_lle(llt, lltable_free_cb, &dchain); 606 llentries_unlink(llt, &dchain); 607 IF_AFDATA_WUNLOCK(llt->llt_ifp); 608 609 CK_LIST_FOREACH_SAFE(lle, &dchain, lle_chain, next) { 610 llentry_free(lle); 611 } 612 613 KASSERT(llt->llt_entries == 0, ("%s: lltable %p (%s) entires not 0: %d", 614 __func__, llt, llt->llt_ifp->if_xname, llt->llt_entries)); 615 616 llt->llt_free_tbl(llt); 617 } 618 619 /* 620 * Deletes an address from given lltable. 621 * Used for userland interaction to remove 622 * individual entries. Skips entries added by OS. 623 */ 624 int 625 lltable_delete_addr(struct lltable *llt, u_int flags, 626 const struct sockaddr *l3addr) 627 { 628 struct llentry *lle; 629 struct ifnet *ifp; 630 631 ifp = llt->llt_ifp; 632 IF_AFDATA_WLOCK(ifp); 633 lle = lla_lookup(llt, LLE_SF(l3addr->sa_family, LLE_EXCLUSIVE), l3addr); 634 635 if (lle == NULL) { 636 IF_AFDATA_WUNLOCK(ifp); 637 return (ENOENT); 638 } 639 if ((lle->la_flags & LLE_IFADDR) != 0 && (flags & LLE_IFADDR) == 0) { 640 IF_AFDATA_WUNLOCK(ifp); 641 LLE_WUNLOCK(lle); 642 return (EPERM); 643 } 644 645 lltable_unlink_entry(llt, lle); 646 IF_AFDATA_WUNLOCK(ifp); 647 648 llt->llt_delete_entry(llt, lle); 649 650 return (0); 651 } 652 653 void 654 lltable_prefix_free(int af, struct sockaddr *addr, struct sockaddr *mask, 655 u_int flags) 656 { 657 struct lltable *llt; 658 659 LLTABLE_LIST_RLOCK(); 660 SLIST_FOREACH(llt, &V_lltables, llt_link) { 661 if (llt->llt_af != af) 662 continue; 663 664 llt->llt_prefix_free(llt, addr, mask, flags); 665 } 666 LLTABLE_LIST_RUNLOCK(); 667 } 668 669 struct lltable * 670 lltable_allocate_htbl(uint32_t hsize) 671 { 672 struct lltable *llt; 673 int i; 674 675 llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK | M_ZERO); 676 llt->llt_hsize = hsize; 677 llt->lle_head = malloc(sizeof(struct llentries) * hsize, 678 M_LLTABLE, M_WAITOK | M_ZERO); 679 680 for (i = 0; i < llt->llt_hsize; i++) 681 CK_LIST_INIT(&llt->lle_head[i]); 682 683 /* Set some default callbacks */ 684 llt->llt_link_entry = htable_link_entry; 685 llt->llt_unlink_entry = htable_unlink_entry; 686 llt->llt_prefix_free = htable_prefix_free; 687 llt->llt_foreach_entry = htable_foreach_lle; 688 llt->llt_free_tbl = htable_free_tbl; 689 690 return (llt); 691 } 692 693 /* 694 * Links lltable to global llt list. 695 */ 696 void 697 lltable_link(struct lltable *llt) 698 { 699 700 LLTABLE_LIST_WLOCK(); 701 SLIST_INSERT_HEAD(&V_lltables, llt, llt_link); 702 LLTABLE_LIST_WUNLOCK(); 703 } 704 705 static void 706 lltable_unlink(struct lltable *llt) 707 { 708 709 LLTABLE_LIST_WLOCK(); 710 SLIST_REMOVE(&V_lltables, llt, lltable, llt_link); 711 LLTABLE_LIST_WUNLOCK(); 712 713 } 714 715 /* 716 * External methods used by lltable consumers 717 */ 718 719 int 720 lltable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg) 721 { 722 723 return (llt->llt_foreach_entry(llt, f, farg)); 724 } 725 726 struct llentry * 727 lltable_alloc_entry(struct lltable *llt, u_int flags, 728 const struct sockaddr *l3addr) 729 { 730 731 return (llt->llt_alloc_entry(llt, flags, l3addr)); 732 } 733 734 void 735 lltable_free_entry(struct lltable *llt, struct llentry *lle) 736 { 737 738 llt->llt_free_entry(llt, lle); 739 } 740 741 int 742 lltable_link_entry(struct lltable *llt, struct llentry *lle) 743 { 744 745 return (llt->llt_link_entry(llt, lle)); 746 } 747 748 void 749 lltable_link_child_entry(struct llentry *lle, struct llentry *child_lle) 750 { 751 child_lle->lle_parent = lle; 752 child_lle->lle_tbl = lle->lle_tbl; 753 child_lle->la_flags |= LLE_LINKED; 754 CK_SLIST_INSERT_HEAD(&lle->lle_children, child_lle, lle_child_next); 755 } 756 757 void 758 lltable_unlink_child_entry(struct llentry *child_lle) 759 { 760 struct llentry *lle = child_lle->lle_parent; 761 762 child_lle->la_flags &= ~LLE_LINKED; 763 child_lle->lle_parent = NULL; 764 CK_SLIST_REMOVE(&lle->lle_children, child_lle, llentry, lle_child_next); 765 } 766 767 int 768 lltable_unlink_entry(struct lltable *llt, struct llentry *lle) 769 { 770 771 return (llt->llt_unlink_entry(lle)); 772 } 773 774 void 775 lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa) 776 { 777 struct lltable *llt; 778 779 llt = lle->lle_tbl; 780 llt->llt_fill_sa_entry(lle, sa); 781 } 782 783 struct ifnet * 784 lltable_get_ifp(const struct lltable *llt) 785 { 786 787 return (llt->llt_ifp); 788 } 789 790 int 791 lltable_get_af(const struct lltable *llt) 792 { 793 794 return (llt->llt_af); 795 } 796 797 /* 798 * Called in route_output when rtm_flags contains RTF_LLDATA. 799 */ 800 int 801 lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info) 802 { 803 struct sockaddr_dl *dl = 804 (struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY]; 805 struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST]; 806 struct ifnet *ifp; 807 struct lltable *llt; 808 struct llentry *lle, *lle_tmp; 809 uint8_t linkhdr[LLE_MAX_LINKHDR]; 810 size_t linkhdrsize; 811 int lladdr_off; 812 u_int laflags = 0; 813 int error; 814 815 if (dl == NULL || dl->sdl_family != AF_LINK) 816 return (EINVAL); 817 818 /* XXX: should be ntohs() */ 819 ifp = ifnet_byindex(dl->sdl_index); 820 if (ifp == NULL) { 821 log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n", 822 __func__, dl->sdl_index); 823 return EINVAL; 824 } 825 826 /* XXX linked list may be too expensive */ 827 LLTABLE_LIST_RLOCK(); 828 SLIST_FOREACH(llt, &V_lltables, llt_link) { 829 if (llt->llt_af == dst->sa_family && 830 llt->llt_ifp == ifp) 831 break; 832 } 833 LLTABLE_LIST_RUNLOCK(); 834 if (llt == NULL) 835 return (ESRCH); 836 837 error = 0; 838 839 switch (rtm->rtm_type) { 840 case RTM_ADD: 841 /* Add static LLE */ 842 laflags = 0; 843 if (rtm->rtm_rmx.rmx_expire == 0) 844 laflags = LLE_STATIC; 845 lle = lltable_alloc_entry(llt, laflags, dst); 846 if (lle == NULL) 847 return (ENOMEM); 848 849 linkhdrsize = sizeof(linkhdr); 850 if (lltable_calc_llheader(ifp, dst->sa_family, LLADDR(dl), 851 linkhdr, &linkhdrsize, &lladdr_off) != 0) 852 return (EINVAL); 853 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize, 854 lladdr_off); 855 if ((rtm->rtm_flags & RTF_ANNOUNCE)) 856 lle->la_flags |= LLE_PUB; 857 lle->la_expire = rtm->rtm_rmx.rmx_expire; 858 859 laflags = lle->la_flags; 860 861 /* Try to link new entry */ 862 lle_tmp = NULL; 863 IF_AFDATA_WLOCK(ifp); 864 LLE_WLOCK(lle); 865 lle_tmp = lla_lookup(llt, LLE_EXCLUSIVE, dst); 866 if (lle_tmp != NULL) { 867 /* Check if we are trying to replace immutable entry */ 868 if ((lle_tmp->la_flags & LLE_IFADDR) != 0) { 869 IF_AFDATA_WUNLOCK(ifp); 870 LLE_WUNLOCK(lle_tmp); 871 lltable_free_entry(llt, lle); 872 return (EPERM); 873 } 874 /* Unlink existing entry from table */ 875 lltable_unlink_entry(llt, lle_tmp); 876 } 877 lltable_link_entry(llt, lle); 878 IF_AFDATA_WUNLOCK(ifp); 879 880 if (lle_tmp != NULL) { 881 EVENTHANDLER_INVOKE(lle_event, lle_tmp,LLENTRY_EXPIRED); 882 lltable_free_entry(llt, lle_tmp); 883 } 884 885 /* 886 * By invoking LLE handler here we might get 887 * two events on static LLE entry insertion 888 * in routing socket. However, since we might have 889 * other subscribers we need to generate this event. 890 */ 891 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_RESOLVED); 892 LLE_WUNLOCK(lle); 893 #ifdef INET 894 /* gratuitous ARP */ 895 if ((laflags & LLE_PUB) && dst->sa_family == AF_INET) 896 arprequest(ifp, 897 &((struct sockaddr_in *)dst)->sin_addr, 898 &((struct sockaddr_in *)dst)->sin_addr, 899 (u_char *)LLADDR(dl)); 900 #endif 901 902 break; 903 904 case RTM_DELETE: 905 return (lltable_delete_addr(llt, 0, dst)); 906 907 default: 908 error = EINVAL; 909 } 910 911 return (error); 912 } 913 914 #ifdef DDB 915 struct llentry_sa { 916 struct llentry base; 917 struct sockaddr l3_addr; 918 }; 919 920 static void 921 llatbl_lle_show(struct llentry_sa *la) 922 { 923 struct llentry *lle; 924 uint8_t octet[6]; 925 926 lle = &la->base; 927 db_printf("lle=%p\n", lle); 928 db_printf(" lle_next=%p\n", lle->lle_next.cle_next); 929 db_printf(" lle_lock=%p\n", &lle->lle_lock); 930 db_printf(" lle_tbl=%p\n", lle->lle_tbl); 931 db_printf(" lle_head=%p\n", lle->lle_head); 932 db_printf(" la_hold=%p\n", lle->la_hold); 933 db_printf(" la_numheld=%d\n", lle->la_numheld); 934 db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire); 935 db_printf(" la_flags=0x%04x\n", lle->la_flags); 936 db_printf(" la_asked=%u\n", lle->la_asked); 937 db_printf(" la_preempt=%u\n", lle->la_preempt); 938 db_printf(" ln_state=%d\n", lle->ln_state); 939 db_printf(" ln_router=%u\n", lle->ln_router); 940 db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick); 941 db_printf(" lle_refcnt=%d\n", lle->lle_refcnt); 942 bcopy(lle->ll_addr, octet, sizeof(octet)); 943 db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n", 944 octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]); 945 db_printf(" lle_timer=%p\n", &lle->lle_timer); 946 947 switch (la->l3_addr.sa_family) { 948 #ifdef INET 949 case AF_INET: 950 { 951 struct sockaddr_in *sin; 952 char l3s[INET_ADDRSTRLEN]; 953 954 sin = (struct sockaddr_in *)&la->l3_addr; 955 inet_ntoa_r(sin->sin_addr, l3s); 956 db_printf(" l3_addr=%s\n", l3s); 957 break; 958 } 959 #endif 960 #ifdef INET6 961 case AF_INET6: 962 { 963 struct sockaddr_in6 *sin6; 964 char l3s[INET6_ADDRSTRLEN]; 965 966 sin6 = (struct sockaddr_in6 *)&la->l3_addr; 967 ip6_sprintf(l3s, &sin6->sin6_addr); 968 db_printf(" l3_addr=%s\n", l3s); 969 break; 970 } 971 #endif 972 default: 973 db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family); 974 break; 975 } 976 } 977 978 DB_SHOW_COMMAND(llentry, db_show_llentry) 979 { 980 981 if (!have_addr) { 982 db_printf("usage: show llentry <struct llentry *>\n"); 983 return; 984 } 985 986 llatbl_lle_show((struct llentry_sa *)addr); 987 } 988 989 static void 990 llatbl_llt_show(struct lltable *llt) 991 { 992 int i; 993 struct llentry *lle; 994 995 db_printf("llt=%p llt_af=%d llt_ifp=%p\n", 996 llt, llt->llt_af, llt->llt_ifp); 997 998 for (i = 0; i < llt->llt_hsize; i++) { 999 CK_LIST_FOREACH(lle, &llt->lle_head[i], lle_next) { 1000 llatbl_lle_show((struct llentry_sa *)lle); 1001 if (db_pager_quit) 1002 return; 1003 } 1004 } 1005 } 1006 1007 DB_SHOW_COMMAND(lltable, db_show_lltable) 1008 { 1009 1010 if (!have_addr) { 1011 db_printf("usage: show lltable <struct lltable *>\n"); 1012 return; 1013 } 1014 1015 llatbl_llt_show((struct lltable *)addr); 1016 } 1017 1018 DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables) 1019 { 1020 VNET_ITERATOR_DECL(vnet_iter); 1021 struct lltable *llt; 1022 1023 VNET_FOREACH(vnet_iter) { 1024 CURVNET_SET_QUIET(vnet_iter); 1025 #ifdef VIMAGE 1026 db_printf("vnet=%p\n", curvnet); 1027 #endif 1028 SLIST_FOREACH(llt, &V_lltables, llt_link) { 1029 db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n", 1030 llt, llt->llt_af, llt->llt_ifp, 1031 (llt->llt_ifp != NULL) ? 1032 llt->llt_ifp->if_xname : "?"); 1033 if (have_addr && addr != 0) /* verbose */ 1034 llatbl_llt_show(llt); 1035 if (db_pager_quit) { 1036 CURVNET_RESTORE(); 1037 return; 1038 } 1039 } 1040 CURVNET_RESTORE(); 1041 } 1042 } 1043 #endif 1044