1 /*- 2 * Copyright (c) 1989, 1993, 1995 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * Poul-Henning Kamp of the FreeBSD Project. 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 * 4. Neither the name of the University nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 * 32 * @(#)vfs_cache.c 8.5 (Berkeley) 3/22/95 33 */ 34 35 #include <sys/cdefs.h> 36 __FBSDID("$FreeBSD$"); 37 38 #include "opt_ktrace.h" 39 40 #include <sys/param.h> 41 #include <sys/filedesc.h> 42 #include <sys/fnv_hash.h> 43 #include <sys/kernel.h> 44 #include <sys/lock.h> 45 #include <sys/malloc.h> 46 #include <sys/mount.h> 47 #include <sys/namei.h> 48 #include <sys/proc.h> 49 #include <sys/rwlock.h> 50 #include <sys/syscallsubr.h> 51 #include <sys/sysctl.h> 52 #include <sys/sysproto.h> 53 #include <sys/systm.h> 54 #include <sys/vnode.h> 55 #ifdef KTRACE 56 #include <sys/ktrace.h> 57 #endif 58 59 #include <vm/uma.h> 60 61 /* 62 * This structure describes the elements in the cache of recent 63 * names looked up by namei. 64 */ 65 66 struct namecache { 67 LIST_ENTRY(namecache) nc_hash; /* hash chain */ 68 LIST_ENTRY(namecache) nc_src; /* source vnode list */ 69 TAILQ_ENTRY(namecache) nc_dst; /* destination vnode list */ 70 struct vnode *nc_dvp; /* vnode of parent of name */ 71 struct vnode *nc_vp; /* vnode the name refers to */ 72 u_char nc_flag; /* flag bits */ 73 u_char nc_nlen; /* length of name */ 74 char nc_name[0]; /* segment name */ 75 }; 76 77 /* 78 * Name caching works as follows: 79 * 80 * Names found by directory scans are retained in a cache 81 * for future reference. It is managed LRU, so frequently 82 * used names will hang around. Cache is indexed by hash value 83 * obtained from (vp, name) where vp refers to the directory 84 * containing name. 85 * 86 * If it is a "negative" entry, (i.e. for a name that is known NOT to 87 * exist) the vnode pointer will be NULL. 88 * 89 * Upon reaching the last segment of a path, if the reference 90 * is for DELETE, or NOCACHE is set (rewrite), and the 91 * name is located in the cache, it will be dropped. 92 */ 93 94 /* 95 * Structures associated with name cacheing. 96 */ 97 #define NCHHASH(hash) \ 98 (&nchashtbl[(hash) & nchash]) 99 static LIST_HEAD(nchashhead, namecache) *nchashtbl; /* Hash Table */ 100 static TAILQ_HEAD(, namecache) ncneg; /* Hash Table */ 101 static u_long nchash; /* size of hash table */ 102 SYSCTL_ULONG(_debug, OID_AUTO, nchash, CTLFLAG_RD, &nchash, 0, ""); 103 static u_long ncnegfactor = 16; /* ratio of negative entries */ 104 SYSCTL_ULONG(_debug, OID_AUTO, ncnegfactor, CTLFLAG_RW, &ncnegfactor, 0, ""); 105 static u_long numneg; /* number of cache entries allocated */ 106 SYSCTL_ULONG(_debug, OID_AUTO, numneg, CTLFLAG_RD, &numneg, 0, ""); 107 static u_long numcache; /* number of cache entries allocated */ 108 SYSCTL_ULONG(_debug, OID_AUTO, numcache, CTLFLAG_RD, &numcache, 0, ""); 109 static u_long numcachehv; /* number of cache entries with vnodes held */ 110 SYSCTL_ULONG(_debug, OID_AUTO, numcachehv, CTLFLAG_RD, &numcachehv, 0, ""); 111 #if 0 112 static u_long numcachepl; /* number of cache purge for leaf entries */ 113 SYSCTL_ULONG(_debug, OID_AUTO, numcachepl, CTLFLAG_RD, &numcachepl, 0, ""); 114 #endif 115 struct nchstats nchstats; /* cache effectiveness statistics */ 116 117 static struct rwlock cache_lock; 118 RW_SYSINIT(vfscache, &cache_lock, "Name Cache"); 119 120 #define CACHE_UPGRADE_LOCK() rw_try_upgrade(&cache_lock) 121 #define CACHE_RLOCK() rw_rlock(&cache_lock) 122 #define CACHE_RUNLOCK() rw_runlock(&cache_lock) 123 #define CACHE_WLOCK() rw_wlock(&cache_lock) 124 #define CACHE_WUNLOCK() rw_wunlock(&cache_lock) 125 126 /* 127 * UMA zones for the VFS cache. 128 * 129 * The small cache is used for entries with short names, which are the 130 * most common. The large cache is used for entries which are too big to 131 * fit in the small cache. 132 */ 133 static uma_zone_t cache_zone_small; 134 static uma_zone_t cache_zone_large; 135 136 #define CACHE_PATH_CUTOFF 32 137 #define CACHE_ZONE_SMALL (sizeof(struct namecache) + CACHE_PATH_CUTOFF) 138 #define CACHE_ZONE_LARGE (sizeof(struct namecache) + NAME_MAX) 139 140 #define cache_alloc(len) uma_zalloc(((len) <= CACHE_PATH_CUTOFF) ? \ 141 cache_zone_small : cache_zone_large, M_WAITOK) 142 #define cache_free(ncp) do { \ 143 if (ncp != NULL) \ 144 uma_zfree(((ncp)->nc_nlen <= CACHE_PATH_CUTOFF) ? \ 145 cache_zone_small : cache_zone_large, (ncp)); \ 146 } while (0) 147 148 static int doingcache = 1; /* 1 => enable the cache */ 149 SYSCTL_INT(_debug, OID_AUTO, vfscache, CTLFLAG_RW, &doingcache, 0, ""); 150 151 /* Export size information to userland */ 152 SYSCTL_INT(_debug_sizeof, OID_AUTO, namecache, CTLFLAG_RD, 0, 153 sizeof(struct namecache), ""); 154 155 /* 156 * The new name cache statistics 157 */ 158 static SYSCTL_NODE(_vfs, OID_AUTO, cache, CTLFLAG_RW, 0, "Name cache statistics"); 159 #define STATNODE(mode, name, var) \ 160 SYSCTL_ULONG(_vfs_cache, OID_AUTO, name, mode, var, 0, ""); 161 STATNODE(CTLFLAG_RD, numneg, &numneg); 162 STATNODE(CTLFLAG_RD, numcache, &numcache); 163 static u_long numcalls; STATNODE(CTLFLAG_RD, numcalls, &numcalls); 164 static u_long dothits; STATNODE(CTLFLAG_RD, dothits, &dothits); 165 static u_long dotdothits; STATNODE(CTLFLAG_RD, dotdothits, &dotdothits); 166 static u_long numchecks; STATNODE(CTLFLAG_RD, numchecks, &numchecks); 167 static u_long nummiss; STATNODE(CTLFLAG_RD, nummiss, &nummiss); 168 static u_long nummisszap; STATNODE(CTLFLAG_RD, nummisszap, &nummisszap); 169 static u_long numposzaps; STATNODE(CTLFLAG_RD, numposzaps, &numposzaps); 170 static u_long numposhits; STATNODE(CTLFLAG_RD, numposhits, &numposhits); 171 static u_long numnegzaps; STATNODE(CTLFLAG_RD, numnegzaps, &numnegzaps); 172 static u_long numneghits; STATNODE(CTLFLAG_RD, numneghits, &numneghits); 173 static u_long numupgrades; STATNODE(CTLFLAG_RD, numupgrades, &numupgrades); 174 175 SYSCTL_OPAQUE(_vfs_cache, OID_AUTO, nchstats, CTLFLAG_RD | CTLFLAG_MPSAFE, 176 &nchstats, sizeof(nchstats), "LU", "VFS cache effectiveness statistics"); 177 178 179 180 static void cache_zap(struct namecache *ncp); 181 static int vn_vptocnp(struct vnode **vp, char **bp, char *buf, u_int *buflen); 182 static int vn_fullpath1(struct thread *td, struct vnode *vp, struct vnode *rdir, 183 char *buf, char **retbuf, u_int buflen); 184 185 static MALLOC_DEFINE(M_VFSCACHE, "vfscache", "VFS name cache entries"); 186 187 /* 188 * Flags in namecache.nc_flag 189 */ 190 #define NCF_WHITE 1 191 192 #ifdef DIAGNOSTIC 193 /* 194 * Grab an atomic snapshot of the name cache hash chain lengths 195 */ 196 SYSCTL_NODE(_debug, OID_AUTO, hashstat, CTLFLAG_RW, NULL, "hash table stats"); 197 198 static int 199 sysctl_debug_hashstat_rawnchash(SYSCTL_HANDLER_ARGS) 200 { 201 int error; 202 struct nchashhead *ncpp; 203 struct namecache *ncp; 204 int n_nchash; 205 int count; 206 207 n_nchash = nchash + 1; /* nchash is max index, not count */ 208 if (!req->oldptr) 209 return SYSCTL_OUT(req, 0, n_nchash * sizeof(int)); 210 211 /* Scan hash tables for applicable entries */ 212 for (ncpp = nchashtbl; n_nchash > 0; n_nchash--, ncpp++) { 213 CACHE_RLOCK(); 214 count = 0; 215 LIST_FOREACH(ncp, ncpp, nc_hash) { 216 count++; 217 } 218 CACHE_RUNLOCK(); 219 error = SYSCTL_OUT(req, &count, sizeof(count)); 220 if (error) 221 return (error); 222 } 223 return (0); 224 } 225 SYSCTL_PROC(_debug_hashstat, OID_AUTO, rawnchash, CTLTYPE_INT|CTLFLAG_RD| 226 CTLFLAG_MPSAFE, 0, 0, sysctl_debug_hashstat_rawnchash, "S,int", 227 "nchash chain lengths"); 228 229 static int 230 sysctl_debug_hashstat_nchash(SYSCTL_HANDLER_ARGS) 231 { 232 int error; 233 struct nchashhead *ncpp; 234 struct namecache *ncp; 235 int n_nchash; 236 int count, maxlength, used, pct; 237 238 if (!req->oldptr) 239 return SYSCTL_OUT(req, 0, 4 * sizeof(int)); 240 241 n_nchash = nchash + 1; /* nchash is max index, not count */ 242 used = 0; 243 maxlength = 0; 244 245 /* Scan hash tables for applicable entries */ 246 for (ncpp = nchashtbl; n_nchash > 0; n_nchash--, ncpp++) { 247 count = 0; 248 CACHE_RLOCK(); 249 LIST_FOREACH(ncp, ncpp, nc_hash) { 250 count++; 251 } 252 CACHE_RUNLOCK(); 253 if (count) 254 used++; 255 if (maxlength < count) 256 maxlength = count; 257 } 258 n_nchash = nchash + 1; 259 pct = (used * 100 * 100) / n_nchash; 260 error = SYSCTL_OUT(req, &n_nchash, sizeof(n_nchash)); 261 if (error) 262 return (error); 263 error = SYSCTL_OUT(req, &used, sizeof(used)); 264 if (error) 265 return (error); 266 error = SYSCTL_OUT(req, &maxlength, sizeof(maxlength)); 267 if (error) 268 return (error); 269 error = SYSCTL_OUT(req, &pct, sizeof(pct)); 270 if (error) 271 return (error); 272 return (0); 273 } 274 SYSCTL_PROC(_debug_hashstat, OID_AUTO, nchash, CTLTYPE_INT|CTLFLAG_RD| 275 CTLFLAG_MPSAFE, 0, 0, sysctl_debug_hashstat_nchash, "I", 276 "nchash chain lengths"); 277 #endif 278 279 /* 280 * cache_zap(): 281 * 282 * Removes a namecache entry from cache, whether it contains an actual 283 * pointer to a vnode or if it is just a negative cache entry. 284 */ 285 static void 286 cache_zap(ncp) 287 struct namecache *ncp; 288 { 289 struct vnode *vp; 290 291 rw_assert(&cache_lock, RA_WLOCKED); 292 CTR2(KTR_VFS, "cache_zap(%p) vp %p", ncp, ncp->nc_vp); 293 vp = NULL; 294 LIST_REMOVE(ncp, nc_hash); 295 LIST_REMOVE(ncp, nc_src); 296 if (LIST_EMPTY(&ncp->nc_dvp->v_cache_src)) { 297 vp = ncp->nc_dvp; 298 numcachehv--; 299 } 300 if (ncp->nc_vp) { 301 TAILQ_REMOVE(&ncp->nc_vp->v_cache_dst, ncp, nc_dst); 302 ncp->nc_vp->v_dd = NULL; 303 } else { 304 TAILQ_REMOVE(&ncneg, ncp, nc_dst); 305 numneg--; 306 } 307 numcache--; 308 cache_free(ncp); 309 if (vp) 310 vdrop(vp); 311 } 312 313 /* 314 * Lookup an entry in the cache 315 * 316 * Lookup is called with dvp pointing to the directory to search, 317 * cnp pointing to the name of the entry being sought. If the lookup 318 * succeeds, the vnode is returned in *vpp, and a status of -1 is 319 * returned. If the lookup determines that the name does not exist 320 * (negative cacheing), a status of ENOENT is returned. If the lookup 321 * fails, a status of zero is returned. If the directory vnode is 322 * recycled out from under us due to a forced unmount, a status of 323 * ENOENT is returned. 324 * 325 * vpp is locked and ref'd on return. If we're looking up DOTDOT, dvp is 326 * unlocked. If we're looking up . an extra ref is taken, but the lock is 327 * not recursively acquired. 328 */ 329 330 int 331 cache_lookup(dvp, vpp, cnp) 332 struct vnode *dvp; 333 struct vnode **vpp; 334 struct componentname *cnp; 335 { 336 struct namecache *ncp; 337 u_int32_t hash; 338 int error, ltype, wlocked; 339 340 if (!doingcache) { 341 cnp->cn_flags &= ~MAKEENTRY; 342 return (0); 343 } 344 retry: 345 CACHE_RLOCK(); 346 wlocked = 0; 347 numcalls++; 348 error = 0; 349 350 retry_wlocked: 351 if (cnp->cn_nameptr[0] == '.') { 352 if (cnp->cn_namelen == 1) { 353 *vpp = dvp; 354 CTR2(KTR_VFS, "cache_lookup(%p, %s) found via .", 355 dvp, cnp->cn_nameptr); 356 dothits++; 357 goto success; 358 } 359 if (cnp->cn_namelen == 2 && cnp->cn_nameptr[1] == '.') { 360 dotdothits++; 361 if (dvp->v_dd == NULL || 362 (cnp->cn_flags & MAKEENTRY) == 0) { 363 goto unlock; 364 } 365 *vpp = dvp->v_dd; 366 CTR3(KTR_VFS, "cache_lookup(%p, %s) found %p via ..", 367 dvp, cnp->cn_nameptr, *vpp); 368 goto success; 369 } 370 } 371 372 hash = fnv_32_buf(cnp->cn_nameptr, cnp->cn_namelen, FNV1_32_INIT); 373 hash = fnv_32_buf(&dvp, sizeof(dvp), hash); 374 LIST_FOREACH(ncp, (NCHHASH(hash)), nc_hash) { 375 numchecks++; 376 if (ncp->nc_dvp == dvp && ncp->nc_nlen == cnp->cn_namelen && 377 !bcmp(ncp->nc_name, cnp->cn_nameptr, ncp->nc_nlen)) 378 break; 379 } 380 381 /* We failed to find an entry */ 382 if (ncp == NULL) { 383 if ((cnp->cn_flags & MAKEENTRY) == 0) { 384 nummisszap++; 385 } else { 386 nummiss++; 387 } 388 nchstats.ncs_miss++; 389 goto unlock; 390 } 391 392 /* We don't want to have an entry, so dump it */ 393 if ((cnp->cn_flags & MAKEENTRY) == 0) { 394 numposzaps++; 395 nchstats.ncs_badhits++; 396 if (!wlocked && !CACHE_UPGRADE_LOCK()) 397 goto wlock; 398 cache_zap(ncp); 399 CACHE_WUNLOCK(); 400 return (0); 401 } 402 403 /* We found a "positive" match, return the vnode */ 404 if (ncp->nc_vp) { 405 numposhits++; 406 nchstats.ncs_goodhits++; 407 *vpp = ncp->nc_vp; 408 CTR4(KTR_VFS, "cache_lookup(%p, %s) found %p via ncp %p", 409 dvp, cnp->cn_nameptr, *vpp, ncp); 410 goto success; 411 } 412 413 /* We found a negative match, and want to create it, so purge */ 414 if (cnp->cn_nameiop == CREATE) { 415 numnegzaps++; 416 nchstats.ncs_badhits++; 417 if (!wlocked && !CACHE_UPGRADE_LOCK()) 418 goto wlock; 419 cache_zap(ncp); 420 CACHE_WUNLOCK(); 421 return (0); 422 } 423 424 if (!wlocked && !CACHE_UPGRADE_LOCK()) 425 goto wlock; 426 numneghits++; 427 /* 428 * We found a "negative" match, so we shift it to the end of 429 * the "negative" cache entries queue to satisfy LRU. Also, 430 * check to see if the entry is a whiteout; indicate this to 431 * the componentname, if so. 432 */ 433 TAILQ_REMOVE(&ncneg, ncp, nc_dst); 434 TAILQ_INSERT_TAIL(&ncneg, ncp, nc_dst); 435 nchstats.ncs_neghits++; 436 if (ncp->nc_flag & NCF_WHITE) 437 cnp->cn_flags |= ISWHITEOUT; 438 CACHE_WUNLOCK(); 439 return (ENOENT); 440 441 wlock: 442 /* 443 * We need to update the cache after our lookup, so upgrade to 444 * a write lock and retry the operation. 445 */ 446 CACHE_RUNLOCK(); 447 CACHE_WLOCK(); 448 numupgrades++; 449 wlocked = 1; 450 goto retry_wlocked; 451 452 success: 453 /* 454 * On success we return a locked and ref'd vnode as per the lookup 455 * protocol. 456 */ 457 if (dvp == *vpp) { /* lookup on "." */ 458 VREF(*vpp); 459 if (wlocked) 460 CACHE_WUNLOCK(); 461 else 462 CACHE_RUNLOCK(); 463 /* 464 * When we lookup "." we still can be asked to lock it 465 * differently... 466 */ 467 ltype = cnp->cn_lkflags & LK_TYPE_MASK; 468 if (ltype != VOP_ISLOCKED(*vpp)) { 469 if (ltype == LK_EXCLUSIVE) { 470 vn_lock(*vpp, LK_UPGRADE | LK_RETRY); 471 if ((*vpp)->v_iflag & VI_DOOMED) { 472 /* forced unmount */ 473 vrele(*vpp); 474 *vpp = NULL; 475 return (ENOENT); 476 } 477 } else 478 vn_lock(*vpp, LK_DOWNGRADE | LK_RETRY); 479 } 480 return (-1); 481 } 482 ltype = 0; /* silence gcc warning */ 483 if (cnp->cn_flags & ISDOTDOT) { 484 ltype = VOP_ISLOCKED(dvp); 485 VOP_UNLOCK(dvp, 0); 486 } 487 VI_LOCK(*vpp); 488 if (wlocked) 489 CACHE_WUNLOCK(); 490 else 491 CACHE_RUNLOCK(); 492 error = vget(*vpp, cnp->cn_lkflags | LK_INTERLOCK, cnp->cn_thread); 493 if (cnp->cn_flags & ISDOTDOT) 494 vn_lock(dvp, ltype | LK_RETRY); 495 if (error) { 496 *vpp = NULL; 497 goto retry; 498 } 499 if ((cnp->cn_flags & ISLASTCN) && 500 (cnp->cn_lkflags & LK_TYPE_MASK) == LK_EXCLUSIVE) { 501 ASSERT_VOP_ELOCKED(*vpp, "cache_lookup"); 502 } 503 return (-1); 504 505 unlock: 506 if (wlocked) 507 CACHE_WUNLOCK(); 508 else 509 CACHE_RUNLOCK(); 510 return (0); 511 } 512 513 /* 514 * Add an entry to the cache. 515 */ 516 void 517 cache_enter(dvp, vp, cnp) 518 struct vnode *dvp; 519 struct vnode *vp; 520 struct componentname *cnp; 521 { 522 struct namecache *ncp, *n2; 523 struct nchashhead *ncpp; 524 u_int32_t hash; 525 int hold; 526 int zap; 527 int len; 528 529 CTR3(KTR_VFS, "cache_enter(%p, %p, %s)", dvp, vp, cnp->cn_nameptr); 530 VNASSERT(vp == NULL || (vp->v_iflag & VI_DOOMED) == 0, vp, 531 ("cahe_enter: Adding a doomed vnode")); 532 533 if (!doingcache) 534 return; 535 536 /* 537 * Avoid blowout in namecache entries. 538 */ 539 if (numcache >= desiredvnodes * 2) 540 return; 541 542 if (cnp->cn_nameptr[0] == '.') { 543 if (cnp->cn_namelen == 1) { 544 return; 545 } 546 /* 547 * For dotdot lookups only cache the v_dd pointer if the 548 * directory has a link back to its parent via v_cache_dst. 549 * Without this an unlinked directory would keep a soft 550 * reference to its parent which could not be NULLd at 551 * cache_purge() time. 552 */ 553 if (cnp->cn_namelen == 2 && cnp->cn_nameptr[1] == '.') { 554 CACHE_WLOCK(); 555 if (!TAILQ_EMPTY(&dvp->v_cache_dst)) 556 dvp->v_dd = vp; 557 CACHE_WUNLOCK(); 558 return; 559 } 560 } 561 562 hold = 0; 563 zap = 0; 564 565 /* 566 * Calculate the hash key and setup as much of the new 567 * namecache entry as possible before acquiring the lock. 568 */ 569 ncp = cache_alloc(cnp->cn_namelen); 570 ncp->nc_vp = vp; 571 ncp->nc_dvp = dvp; 572 len = ncp->nc_nlen = cnp->cn_namelen; 573 hash = fnv_32_buf(cnp->cn_nameptr, len, FNV1_32_INIT); 574 bcopy(cnp->cn_nameptr, ncp->nc_name, len); 575 hash = fnv_32_buf(&dvp, sizeof(dvp), hash); 576 CACHE_WLOCK(); 577 578 /* 579 * See if this vnode or negative entry is already in the cache 580 * with this name. This can happen with concurrent lookups of 581 * the same path name. 582 */ 583 ncpp = NCHHASH(hash); 584 LIST_FOREACH(n2, ncpp, nc_hash) { 585 if (n2->nc_dvp == dvp && 586 n2->nc_nlen == cnp->cn_namelen && 587 !bcmp(n2->nc_name, cnp->cn_nameptr, n2->nc_nlen)) { 588 CACHE_WUNLOCK(); 589 cache_free(ncp); 590 return; 591 } 592 } 593 594 numcache++; 595 if (!vp) { 596 numneg++; 597 ncp->nc_flag = cnp->cn_flags & ISWHITEOUT ? NCF_WHITE : 0; 598 } else if (vp->v_type == VDIR) { 599 vp->v_dd = dvp; 600 } else { 601 vp->v_dd = NULL; 602 } 603 604 /* 605 * Insert the new namecache entry into the appropriate chain 606 * within the cache entries table. 607 */ 608 LIST_INSERT_HEAD(ncpp, ncp, nc_hash); 609 if (LIST_EMPTY(&dvp->v_cache_src)) { 610 hold = 1; 611 numcachehv++; 612 } 613 LIST_INSERT_HEAD(&dvp->v_cache_src, ncp, nc_src); 614 /* 615 * If the entry is "negative", we place it into the 616 * "negative" cache queue, otherwise, we place it into the 617 * destination vnode's cache entries queue. 618 */ 619 if (vp) { 620 TAILQ_INSERT_HEAD(&vp->v_cache_dst, ncp, nc_dst); 621 } else { 622 TAILQ_INSERT_TAIL(&ncneg, ncp, nc_dst); 623 } 624 if (numneg * ncnegfactor > numcache) { 625 ncp = TAILQ_FIRST(&ncneg); 626 zap = 1; 627 } 628 if (hold) 629 vhold(dvp); 630 if (zap) 631 cache_zap(ncp); 632 CACHE_WUNLOCK(); 633 } 634 635 /* 636 * Name cache initialization, from vfs_init() when we are booting 637 */ 638 static void 639 nchinit(void *dummy __unused) 640 { 641 642 TAILQ_INIT(&ncneg); 643 644 cache_zone_small = uma_zcreate("S VFS Cache", CACHE_ZONE_SMALL, NULL, 645 NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_ZINIT); 646 cache_zone_large = uma_zcreate("L VFS Cache", CACHE_ZONE_LARGE, NULL, 647 NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_ZINIT); 648 649 nchashtbl = hashinit(desiredvnodes * 2, M_VFSCACHE, &nchash); 650 } 651 SYSINIT(vfs, SI_SUB_VFS, SI_ORDER_SECOND, nchinit, NULL); 652 653 654 /* 655 * Invalidate all entries to a particular vnode. 656 */ 657 void 658 cache_purge(vp) 659 struct vnode *vp; 660 { 661 662 CTR1(KTR_VFS, "cache_purge(%p)", vp); 663 CACHE_WLOCK(); 664 while (!LIST_EMPTY(&vp->v_cache_src)) 665 cache_zap(LIST_FIRST(&vp->v_cache_src)); 666 while (!TAILQ_EMPTY(&vp->v_cache_dst)) 667 cache_zap(TAILQ_FIRST(&vp->v_cache_dst)); 668 vp->v_dd = NULL; 669 CACHE_WUNLOCK(); 670 } 671 672 /* 673 * Invalidate all negative entries for a particular directory vnode. 674 */ 675 void 676 cache_purge_negative(vp) 677 struct vnode *vp; 678 { 679 struct namecache *cp, *ncp; 680 681 CTR1(KTR_VFS, "cache_purge_negative(%p)", vp); 682 CACHE_WLOCK(); 683 LIST_FOREACH_SAFE(cp, &vp->v_cache_src, nc_src, ncp) { 684 if (cp->nc_vp == NULL) 685 cache_zap(cp); 686 } 687 CACHE_WUNLOCK(); 688 } 689 690 /* 691 * Flush all entries referencing a particular filesystem. 692 */ 693 void 694 cache_purgevfs(mp) 695 struct mount *mp; 696 { 697 struct nchashhead *ncpp; 698 struct namecache *ncp, *nnp; 699 700 /* Scan hash tables for applicable entries */ 701 CACHE_WLOCK(); 702 for (ncpp = &nchashtbl[nchash]; ncpp >= nchashtbl; ncpp--) { 703 LIST_FOREACH_SAFE(ncp, ncpp, nc_hash, nnp) { 704 if (ncp->nc_dvp->v_mount == mp) 705 cache_zap(ncp); 706 } 707 } 708 CACHE_WUNLOCK(); 709 } 710 711 /* 712 * Perform canonical checks and cache lookup and pass on to filesystem 713 * through the vop_cachedlookup only if needed. 714 */ 715 716 int 717 vfs_cache_lookup(ap) 718 struct vop_lookup_args /* { 719 struct vnode *a_dvp; 720 struct vnode **a_vpp; 721 struct componentname *a_cnp; 722 } */ *ap; 723 { 724 struct vnode *dvp; 725 int error; 726 struct vnode **vpp = ap->a_vpp; 727 struct componentname *cnp = ap->a_cnp; 728 struct ucred *cred = cnp->cn_cred; 729 int flags = cnp->cn_flags; 730 struct thread *td = cnp->cn_thread; 731 732 *vpp = NULL; 733 dvp = ap->a_dvp; 734 735 if (dvp->v_type != VDIR) 736 return (ENOTDIR); 737 738 if ((flags & ISLASTCN) && (dvp->v_mount->mnt_flag & MNT_RDONLY) && 739 (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME)) 740 return (EROFS); 741 742 error = VOP_ACCESS(dvp, VEXEC, cred, td); 743 if (error) 744 return (error); 745 746 error = cache_lookup(dvp, vpp, cnp); 747 if (error == 0) 748 return (VOP_CACHEDLOOKUP(dvp, vpp, cnp)); 749 if (error == -1) 750 return (0); 751 return (error); 752 } 753 754 755 #ifndef _SYS_SYSPROTO_H_ 756 struct __getcwd_args { 757 u_char *buf; 758 u_int buflen; 759 }; 760 #endif 761 762 /* 763 * XXX All of these sysctls would probably be more productive dead. 764 */ 765 static int disablecwd; 766 SYSCTL_INT(_debug, OID_AUTO, disablecwd, CTLFLAG_RW, &disablecwd, 0, 767 "Disable the getcwd syscall"); 768 769 /* Implementation of the getcwd syscall. */ 770 int 771 __getcwd(td, uap) 772 struct thread *td; 773 struct __getcwd_args *uap; 774 { 775 776 return (kern___getcwd(td, uap->buf, UIO_USERSPACE, uap->buflen)); 777 } 778 779 int 780 kern___getcwd(struct thread *td, u_char *buf, enum uio_seg bufseg, u_int buflen) 781 { 782 char *bp, *tmpbuf; 783 struct filedesc *fdp; 784 struct vnode *cdir, *rdir; 785 int error, vfslocked; 786 787 if (disablecwd) 788 return (ENODEV); 789 if (buflen < 2) 790 return (EINVAL); 791 if (buflen > MAXPATHLEN) 792 buflen = MAXPATHLEN; 793 794 tmpbuf = malloc(buflen, M_TEMP, M_WAITOK); 795 fdp = td->td_proc->p_fd; 796 FILEDESC_SLOCK(fdp); 797 cdir = fdp->fd_cdir; 798 VREF(cdir); 799 rdir = fdp->fd_rdir; 800 VREF(rdir); 801 FILEDESC_SUNLOCK(fdp); 802 error = vn_fullpath1(td, cdir, rdir, tmpbuf, &bp, buflen); 803 vfslocked = VFS_LOCK_GIANT(rdir->v_mount); 804 vrele(rdir); 805 VFS_UNLOCK_GIANT(vfslocked); 806 vfslocked = VFS_LOCK_GIANT(cdir->v_mount); 807 vrele(cdir); 808 VFS_UNLOCK_GIANT(vfslocked); 809 810 if (!error) { 811 if (bufseg == UIO_SYSSPACE) 812 bcopy(bp, buf, strlen(bp) + 1); 813 else 814 error = copyout(bp, buf, strlen(bp) + 1); 815 #ifdef KTRACE 816 if (KTRPOINT(curthread, KTR_NAMEI)) 817 ktrnamei(bp); 818 #endif 819 } 820 free(tmpbuf, M_TEMP); 821 return (error); 822 } 823 824 /* 825 * Thus begins the fullpath magic. 826 */ 827 828 #undef STATNODE 829 #define STATNODE(name) \ 830 static u_int name; \ 831 SYSCTL_UINT(_vfs_cache, OID_AUTO, name, CTLFLAG_RD, &name, 0, "") 832 833 static int disablefullpath; 834 SYSCTL_INT(_debug, OID_AUTO, disablefullpath, CTLFLAG_RW, &disablefullpath, 0, 835 "Disable the vn_fullpath function"); 836 837 /* These count for kern___getcwd(), too. */ 838 STATNODE(numfullpathcalls); 839 STATNODE(numfullpathfail1); 840 STATNODE(numfullpathfail2); 841 STATNODE(numfullpathfail4); 842 STATNODE(numfullpathfound); 843 844 /* 845 * Retrieve the full filesystem path that correspond to a vnode from the name 846 * cache (if available) 847 */ 848 int 849 vn_fullpath(struct thread *td, struct vnode *vn, char **retbuf, char **freebuf) 850 { 851 char *buf; 852 struct filedesc *fdp; 853 struct vnode *rdir; 854 int error, vfslocked; 855 856 if (disablefullpath) 857 return (ENODEV); 858 if (vn == NULL) 859 return (EINVAL); 860 861 buf = malloc(MAXPATHLEN, M_TEMP, M_WAITOK); 862 fdp = td->td_proc->p_fd; 863 FILEDESC_SLOCK(fdp); 864 rdir = fdp->fd_rdir; 865 VREF(rdir); 866 FILEDESC_SUNLOCK(fdp); 867 error = vn_fullpath1(td, vn, rdir, buf, retbuf, MAXPATHLEN); 868 vfslocked = VFS_LOCK_GIANT(rdir->v_mount); 869 vrele(rdir); 870 VFS_UNLOCK_GIANT(vfslocked); 871 872 if (!error) 873 *freebuf = buf; 874 else 875 free(buf, M_TEMP); 876 return (error); 877 } 878 879 /* 880 * This function is similar to vn_fullpath, but it attempts to lookup the 881 * pathname relative to the global root mount point. This is required for the 882 * auditing sub-system, as audited pathnames must be absolute, relative to the 883 * global root mount point. 884 */ 885 int 886 vn_fullpath_global(struct thread *td, struct vnode *vn, 887 char **retbuf, char **freebuf) 888 { 889 char *buf; 890 int error; 891 892 if (disablefullpath) 893 return (ENODEV); 894 if (vn == NULL) 895 return (EINVAL); 896 buf = malloc(MAXPATHLEN, M_TEMP, M_WAITOK); 897 error = vn_fullpath1(td, vn, rootvnode, buf, retbuf, MAXPATHLEN); 898 if (!error) 899 *freebuf = buf; 900 else 901 free(buf, M_TEMP); 902 return (error); 903 } 904 905 static int 906 vn_vptocnp(struct vnode **vp, char **bp, char *buf, u_int *buflen) 907 { 908 struct vnode *dvp; 909 int error, vfslocked; 910 911 vhold(*vp); 912 CACHE_RUNLOCK(); 913 vfslocked = VFS_LOCK_GIANT((*vp)->v_mount); 914 vn_lock(*vp, LK_SHARED | LK_RETRY); 915 error = VOP_VPTOCNP(*vp, &dvp, buf, buflen); 916 VOP_UNLOCK(*vp, 0); 917 vdrop(*vp); 918 VFS_UNLOCK_GIANT(vfslocked); 919 if (error) { 920 numfullpathfail2++; 921 return (error); 922 } 923 *bp = buf + *buflen; 924 *vp = dvp; 925 CACHE_RLOCK(); 926 if ((*vp)->v_iflag & VI_DOOMED) { 927 /* forced unmount */ 928 CACHE_RUNLOCK(); 929 vdrop(*vp); 930 return (ENOENT); 931 } 932 vdrop(*vp); 933 934 return (0); 935 } 936 937 /* 938 * The magic behind kern___getcwd() and vn_fullpath(). 939 */ 940 static int 941 vn_fullpath1(struct thread *td, struct vnode *vp, struct vnode *rdir, 942 char *buf, char **retbuf, u_int buflen) 943 { 944 char *bp; 945 int error, i, slash_prefixed; 946 struct namecache *ncp; 947 948 buflen--; 949 bp = buf + buflen; 950 *bp = '\0'; 951 error = 0; 952 slash_prefixed = 0; 953 954 CACHE_RLOCK(); 955 numfullpathcalls++; 956 if (vp->v_type != VDIR) { 957 ncp = TAILQ_FIRST(&vp->v_cache_dst); 958 if (ncp != NULL) { 959 buflen -= ncp->nc_nlen; 960 for (i = ncp->nc_nlen - 1; i >= 0 && bp != buf; i--) 961 *--bp = ncp->nc_name[i]; 962 if (bp == buf) { 963 numfullpathfail4++; 964 CACHE_RUNLOCK(); 965 return (ENOMEM); 966 } 967 vp = ncp->nc_dvp; 968 } else { 969 error = vn_vptocnp(&vp, &bp, buf, &buflen); 970 if (error) 971 return (error); 972 } 973 if (buflen <= 0) { 974 numfullpathfail4++; 975 CACHE_RUNLOCK(); 976 return (ENOMEM); 977 } 978 *--bp = '/'; 979 buflen--; 980 slash_prefixed = 1; 981 } 982 while (vp != rdir && vp != rootvnode) { 983 if (vp->v_vflag & VV_ROOT) { 984 if (vp->v_iflag & VI_DOOMED) { /* forced unmount */ 985 CACHE_RUNLOCK(); 986 error = ENOENT; 987 break; 988 } 989 vp = vp->v_mount->mnt_vnodecovered; 990 continue; 991 } 992 if (vp->v_type != VDIR) { 993 numfullpathfail1++; 994 CACHE_RUNLOCK(); 995 error = ENOTDIR; 996 break; 997 } 998 ncp = TAILQ_FIRST(&vp->v_cache_dst); 999 if (ncp != NULL) { 1000 MPASS(vp->v_dd == NULL || ncp->nc_dvp == vp->v_dd); 1001 buflen -= ncp->nc_nlen; 1002 for (i = ncp->nc_nlen - 1; i >= 0 && bp != buf; i--) 1003 *--bp = ncp->nc_name[i]; 1004 if (bp == buf) { 1005 numfullpathfail4++; 1006 CACHE_RUNLOCK(); 1007 error = ENOMEM; 1008 break; 1009 } 1010 vp = ncp->nc_dvp; 1011 } else { 1012 error = vn_vptocnp(&vp, &bp, buf, &buflen); 1013 if (error) 1014 break; 1015 } 1016 if (buflen <= 0) { 1017 numfullpathfail4++; 1018 CACHE_RUNLOCK(); 1019 error = ENOMEM; 1020 break; 1021 } 1022 *--bp = '/'; 1023 buflen--; 1024 slash_prefixed = 1; 1025 } 1026 if (error) 1027 return (error); 1028 if (!slash_prefixed) { 1029 if (bp == buf) { 1030 numfullpathfail4++; 1031 CACHE_RUNLOCK(); 1032 return (ENOMEM); 1033 } else 1034 *--bp = '/'; 1035 } 1036 numfullpathfound++; 1037 CACHE_RUNLOCK(); 1038 1039 *retbuf = bp; 1040 return (0); 1041 } 1042 1043 int 1044 vn_commname(struct vnode *vp, char *buf, u_int buflen) 1045 { 1046 struct namecache *ncp; 1047 int l; 1048 1049 CACHE_RLOCK(); 1050 ncp = TAILQ_FIRST(&vp->v_cache_dst); 1051 if (!ncp) { 1052 CACHE_RUNLOCK(); 1053 return (ENOENT); 1054 } 1055 l = min(ncp->nc_nlen, buflen - 1); 1056 memcpy(buf, ncp->nc_name, l); 1057 CACHE_RUNLOCK(); 1058 buf[l] = '\0'; 1059 return (0); 1060 } 1061