1 /* 2 * Copyright (c) 2004 Luigi Rizzo, Alessandro Cerri. All rights reserved. 3 * Copyright (c) 2004-2008 Qing Li. All rights reserved. 4 * Copyright (c) 2008 Kip Macy. All rights reserved. 5 * 6 * Redistribution and use in source and binary forms, with or without 7 * modification, are permitted provided that the following conditions 8 * are met: 9 * 1. Redistributions of source code must retain the above copyright 10 * notice, this list of conditions and the following disclaimer. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 15 * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND 16 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 18 * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE 19 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 20 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 21 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 22 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 23 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 24 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 25 * SUCH DAMAGE. 26 */ 27 #include <sys/cdefs.h> 28 __FBSDID("$FreeBSD$"); 29 30 #include "opt_ddb.h" 31 #include "opt_inet.h" 32 #include "opt_inet6.h" 33 34 #include <sys/param.h> 35 #include <sys/systm.h> 36 #include <sys/malloc.h> 37 #include <sys/mbuf.h> 38 #include <sys/syslog.h> 39 #include <sys/sysctl.h> 40 #include <sys/socket.h> 41 #include <sys/kernel.h> 42 #include <sys/lock.h> 43 #include <sys/mutex.h> 44 #include <sys/rwlock.h> 45 46 #ifdef DDB 47 #include <ddb/ddb.h> 48 #endif 49 50 #include <vm/uma.h> 51 52 #include <netinet/in.h> 53 #include <net/if_llatbl.h> 54 #include <net/if.h> 55 #include <net/if_dl.h> 56 #include <net/if_var.h> 57 #include <net/route.h> 58 #include <net/vnet.h> 59 #include <netinet/if_ether.h> 60 #include <netinet6/in6_var.h> 61 #include <netinet6/nd6.h> 62 63 MALLOC_DEFINE(M_LLTABLE, "lltable", "link level address tables"); 64 65 static VNET_DEFINE(SLIST_HEAD(, lltable), lltables); 66 #define V_lltables VNET(lltables) 67 68 static void vnet_lltable_init(void); 69 70 struct rwlock lltable_rwlock; 71 RW_SYSINIT(lltable_rwlock, &lltable_rwlock, "lltable_rwlock"); 72 73 /* 74 * Dump arp state for a specific address family. 75 */ 76 int 77 lltable_sysctl_dumparp(int af, struct sysctl_req *wr) 78 { 79 struct lltable *llt; 80 int error = 0; 81 82 LLTABLE_RLOCK(); 83 SLIST_FOREACH(llt, &V_lltables, llt_link) { 84 if (llt->llt_af == af) { 85 error = llt->llt_dump(llt, wr); 86 if (error != 0) 87 goto done; 88 } 89 } 90 done: 91 LLTABLE_RUNLOCK(); 92 return (error); 93 } 94 95 /* 96 * Deletes an address from the address table. 97 * This function is called by the timer functions 98 * such as arptimer() and nd6_llinfo_timer(), and 99 * the caller does the locking. 100 * 101 * Returns the number of held packets, if any, that were dropped. 102 */ 103 size_t 104 llentry_free(struct llentry *lle) 105 { 106 size_t pkts_dropped; 107 struct mbuf *next; 108 109 pkts_dropped = 0; 110 LLE_WLOCK_ASSERT(lle); 111 LIST_REMOVE(lle, lle_next); 112 113 while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) { 114 next = lle->la_hold->m_nextpkt; 115 m_freem(lle->la_hold); 116 lle->la_hold = next; 117 lle->la_numheld--; 118 pkts_dropped++; 119 } 120 121 KASSERT(lle->la_numheld == 0, 122 ("%s: la_numheld %d > 0, pkts_droped %zd", __func__, 123 lle->la_numheld, pkts_dropped)); 124 125 LLE_FREE_LOCKED(lle); 126 127 return (pkts_dropped); 128 } 129 130 /* 131 * Update an llentry for address dst (equivalent to rtalloc for new-arp) 132 * Caller must pass in a valid struct llentry * (or NULL) 133 * 134 * if found the llentry * is returned referenced and unlocked 135 */ 136 int 137 llentry_update(struct llentry **llep, struct lltable *lt, 138 struct sockaddr_storage *dst, struct ifnet *ifp) 139 { 140 struct llentry *la; 141 142 IF_AFDATA_RLOCK(ifp); 143 la = lla_lookup(lt, LLE_EXCLUSIVE, 144 (struct sockaddr *)dst); 145 IF_AFDATA_RUNLOCK(ifp); 146 if ((la == NULL) && 147 (ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) == 0) { 148 IF_AFDATA_WLOCK(ifp); 149 la = lla_lookup(lt, 150 (LLE_CREATE | LLE_EXCLUSIVE), 151 (struct sockaddr *)dst); 152 IF_AFDATA_WUNLOCK(ifp); 153 } 154 if (la != NULL && (*llep != la)) { 155 if (*llep != NULL) 156 LLE_FREE(*llep); 157 LLE_ADDREF(la); 158 LLE_WUNLOCK(la); 159 *llep = la; 160 } else if (la != NULL) 161 LLE_WUNLOCK(la); 162 163 if (la == NULL) 164 return (ENOENT); 165 166 return (0); 167 } 168 169 /* 170 * Free all entries from given table and free itself. 171 */ 172 void 173 lltable_free(struct lltable *llt) 174 { 175 struct llentry *lle, *next; 176 int i; 177 178 KASSERT(llt != NULL, ("%s: llt is NULL", __func__)); 179 180 LLTABLE_WLOCK(); 181 SLIST_REMOVE(&V_lltables, llt, lltable, llt_link); 182 LLTABLE_WUNLOCK(); 183 184 for (i=0; i < LLTBL_HASHTBL_SIZE; i++) { 185 LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) { 186 int canceled; 187 188 canceled = callout_drain(&lle->la_timer); 189 LLE_WLOCK(lle); 190 if (canceled) 191 LLE_REMREF(lle); 192 llentry_free(lle); 193 } 194 } 195 196 free(llt, M_LLTABLE); 197 } 198 199 #if 0 200 void 201 lltable_drain(int af) 202 { 203 struct lltable *llt; 204 struct llentry *lle; 205 register int i; 206 207 LLTABLE_RLOCK(); 208 SLIST_FOREACH(llt, &V_lltables, llt_link) { 209 if (llt->llt_af != af) 210 continue; 211 212 for (i=0; i < LLTBL_HASHTBL_SIZE; i++) { 213 LIST_FOREACH(lle, &llt->lle_head[i], lle_next) { 214 LLE_WLOCK(lle); 215 if (lle->la_hold) { 216 m_freem(lle->la_hold); 217 lle->la_hold = NULL; 218 } 219 LLE_WUNLOCK(lle); 220 } 221 } 222 } 223 LLTABLE_RUNLOCK(); 224 } 225 #endif 226 227 void 228 lltable_prefix_free(int af, struct sockaddr *prefix, struct sockaddr *mask, 229 u_int flags) 230 { 231 struct lltable *llt; 232 233 LLTABLE_RLOCK(); 234 SLIST_FOREACH(llt, &V_lltables, llt_link) { 235 if (llt->llt_af != af) 236 continue; 237 238 llt->llt_prefix_free(llt, prefix, mask, flags); 239 } 240 LLTABLE_RUNLOCK(); 241 } 242 243 244 245 /* 246 * Create a new lltable. 247 */ 248 struct lltable * 249 lltable_init(struct ifnet *ifp, int af) 250 { 251 struct lltable *llt; 252 register int i; 253 254 llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK); 255 256 llt->llt_af = af; 257 llt->llt_ifp = ifp; 258 for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) 259 LIST_INIT(&llt->lle_head[i]); 260 261 LLTABLE_WLOCK(); 262 SLIST_INSERT_HEAD(&V_lltables, llt, llt_link); 263 LLTABLE_WUNLOCK(); 264 265 return (llt); 266 } 267 268 /* 269 * Called in route_output when adding/deleting a route to an interface. 270 */ 271 int 272 lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info) 273 { 274 struct sockaddr_dl *dl = 275 (struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY]; 276 struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST]; 277 struct ifnet *ifp; 278 struct lltable *llt; 279 struct llentry *lle; 280 u_int laflags = 0, flags = 0; 281 int error = 0; 282 283 if (dl == NULL || dl->sdl_family != AF_LINK) { 284 log(LOG_INFO, "%s: invalid dl\n", __func__); 285 return EINVAL; 286 } 287 ifp = ifnet_byindex(dl->sdl_index); 288 if (ifp == NULL) { 289 log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n", 290 __func__, dl->sdl_index); 291 return EINVAL; 292 } 293 294 switch (rtm->rtm_type) { 295 case RTM_ADD: 296 if (rtm->rtm_flags & RTF_ANNOUNCE) { 297 flags |= LLE_PUB; 298 #ifdef INET 299 if (dst->sa_family == AF_INET && 300 ((struct sockaddr_inarp *)dst)->sin_other != 0) { 301 struct rtentry *rt; 302 ((struct sockaddr_inarp *)dst)->sin_other = 0; 303 rt = rtalloc1(dst, 0, 0); 304 if (rt == NULL || !(rt->rt_flags & RTF_HOST)) { 305 log(LOG_INFO, "%s: RTM_ADD publish " 306 "(proxy only) is invalid\n", 307 __func__); 308 if (rt) 309 RTFREE_LOCKED(rt); 310 return EINVAL; 311 } 312 RTFREE_LOCKED(rt); 313 314 flags |= LLE_PROXY; 315 } 316 #endif 317 } 318 flags |= LLE_CREATE; 319 break; 320 321 case RTM_DELETE: 322 flags |= LLE_DELETE; 323 break; 324 325 case RTM_CHANGE: 326 break; 327 328 default: 329 return EINVAL; /* XXX not implemented yet */ 330 } 331 332 /* XXX linked list may be too expensive */ 333 LLTABLE_RLOCK(); 334 SLIST_FOREACH(llt, &V_lltables, llt_link) { 335 if (llt->llt_af == dst->sa_family && 336 llt->llt_ifp == ifp) 337 break; 338 } 339 LLTABLE_RUNLOCK(); 340 KASSERT(llt != NULL, ("Yep, ugly hacks are bad\n")); 341 342 if (flags & LLE_CREATE) 343 flags |= LLE_EXCLUSIVE; 344 345 IF_AFDATA_LOCK(ifp); 346 lle = lla_lookup(llt, flags, dst); 347 IF_AFDATA_UNLOCK(ifp); 348 if (LLE_IS_VALID(lle)) { 349 if (flags & LLE_CREATE) { 350 /* 351 * If we delay the delete, then a subsequent 352 * "arp add" should look up this entry, reset the 353 * LLE_DELETED flag, and reset the expiration timer 354 */ 355 bcopy(LLADDR(dl), &lle->ll_addr, ifp->if_addrlen); 356 lle->la_flags |= (flags & (LLE_PUB | LLE_PROXY)); 357 lle->la_flags |= LLE_VALID; 358 lle->la_flags &= ~LLE_DELETED; 359 #ifdef INET6 360 /* 361 * ND6 362 */ 363 if (dst->sa_family == AF_INET6) 364 lle->ln_state = ND6_LLINFO_REACHABLE; 365 #endif 366 /* 367 * NB: arp and ndp always set (RTF_STATIC | RTF_HOST) 368 */ 369 370 if (rtm->rtm_rmx.rmx_expire == 0) { 371 lle->la_flags |= LLE_STATIC; 372 lle->la_expire = 0; 373 } else 374 lle->la_expire = rtm->rtm_rmx.rmx_expire; 375 laflags = lle->la_flags; 376 LLE_WUNLOCK(lle); 377 #ifdef INET 378 /* gratuitous ARP */ 379 if ((laflags & LLE_PUB) && dst->sa_family == AF_INET) { 380 arprequest(ifp, 381 &((struct sockaddr_in *)dst)->sin_addr, 382 &((struct sockaddr_in *)dst)->sin_addr, 383 ((laflags & LLE_PROXY) ? 384 (u_char *)IF_LLADDR(ifp) : 385 (u_char *)LLADDR(dl))); 386 } 387 #endif 388 } else { 389 if (flags & LLE_EXCLUSIVE) 390 LLE_WUNLOCK(lle); 391 else 392 LLE_RUNLOCK(lle); 393 } 394 } else if ((lle == NULL) && (flags & LLE_DELETE)) 395 error = EINVAL; 396 397 398 return (error); 399 } 400 401 static void 402 vnet_lltable_init() 403 { 404 405 SLIST_INIT(&V_lltables); 406 } 407 VNET_SYSINIT(vnet_lltable_init, SI_SUB_PSEUDO, SI_ORDER_FIRST, 408 vnet_lltable_init, NULL); 409 410 #ifdef DDB 411 struct llentry_sa { 412 struct llentry base; 413 struct sockaddr l3_addr; 414 }; 415 416 static void 417 llatbl_lle_show(struct llentry_sa *la) 418 { 419 struct llentry *lle; 420 uint8_t octet[6]; 421 422 lle = &la->base; 423 db_printf("lle=%p\n", lle); 424 db_printf(" lle_next=%p\n", lle->lle_next.le_next); 425 db_printf(" lle_lock=%p\n", &lle->lle_lock); 426 db_printf(" lle_tbl=%p\n", lle->lle_tbl); 427 db_printf(" lle_head=%p\n", lle->lle_head); 428 db_printf(" la_hold=%p\n", lle->la_hold); 429 db_printf(" la_numheld=%d\n", lle->la_numheld); 430 db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire); 431 db_printf(" la_flags=0x%04x\n", lle->la_flags); 432 db_printf(" la_asked=%u\n", lle->la_asked); 433 db_printf(" la_preempt=%u\n", lle->la_preempt); 434 db_printf(" ln_byhint=%u\n", lle->ln_byhint); 435 db_printf(" ln_state=%d\n", lle->ln_state); 436 db_printf(" ln_router=%u\n", lle->ln_router); 437 db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick); 438 db_printf(" lle_refcnt=%d\n", lle->lle_refcnt); 439 bcopy(&lle->ll_addr.mac16, octet, sizeof(octet)); 440 db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n", 441 octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]); 442 db_printf(" la_timer=%p\n", &lle->la_timer); 443 444 switch (la->l3_addr.sa_family) { 445 #ifdef INET 446 case AF_INET: 447 { 448 struct sockaddr_in *sin; 449 char l3s[INET_ADDRSTRLEN]; 450 451 sin = (struct sockaddr_in *)&la->l3_addr; 452 inet_ntoa_r(sin->sin_addr, l3s); 453 db_printf(" l3_addr=%s\n", l3s); 454 break; 455 } 456 #endif 457 #ifdef INET6 458 case AF_INET6: 459 { 460 struct sockaddr_in6 *sin6; 461 char l3s[INET6_ADDRSTRLEN]; 462 463 sin6 = (struct sockaddr_in6 *)&la->l3_addr; 464 ip6_sprintf(l3s, &sin6->sin6_addr); 465 db_printf(" l3_addr=%s\n", l3s); 466 break; 467 } 468 #endif 469 default: 470 db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family); 471 break; 472 } 473 } 474 475 DB_SHOW_COMMAND(llentry, db_show_llentry) 476 { 477 478 if (!have_addr) { 479 db_printf("usage: show llentry <struct llentry *>\n"); 480 return; 481 } 482 483 llatbl_lle_show((struct llentry_sa *)addr); 484 } 485 486 static void 487 llatbl_llt_show(struct lltable *llt) 488 { 489 int i; 490 struct llentry *lle; 491 492 db_printf("llt=%p llt_af=%d llt_ifp=%p\n", 493 llt, llt->llt_af, llt->llt_ifp); 494 495 for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) { 496 LIST_FOREACH(lle, &llt->lle_head[i], lle_next) { 497 498 llatbl_lle_show((struct llentry_sa *)lle); 499 if (db_pager_quit) 500 return; 501 } 502 } 503 } 504 505 DB_SHOW_COMMAND(lltable, db_show_lltable) 506 { 507 508 if (!have_addr) { 509 db_printf("usage: show lltable <struct lltable *>\n"); 510 return; 511 } 512 513 llatbl_llt_show((struct lltable *)addr); 514 } 515 516 DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables) 517 { 518 VNET_ITERATOR_DECL(vnet_iter); 519 struct lltable *llt; 520 521 VNET_FOREACH(vnet_iter) { 522 CURVNET_SET_QUIET(vnet_iter); 523 #ifdef VIMAGE 524 db_printf("vnet=%p\n", curvnet); 525 #endif 526 SLIST_FOREACH(llt, &V_lltables, llt_link) { 527 db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n", 528 llt, llt->llt_af, llt->llt_ifp, 529 (llt->llt_ifp != NULL) ? 530 llt->llt_ifp->if_xname : "?"); 531 if (have_addr && addr != 0) /* verbose */ 532 llatbl_llt_show(llt); 533 if (db_pager_quit) { 534 CURVNET_RESTORE(); 535 return; 536 } 537 } 538 CURVNET_RESTORE(); 539 } 540 } 541 #endif 542