1 /* $OpenBSD: vfs_cache.c,v 1.21 2007/04/19 09:25:33 pedro Exp $ */
2 /* $NetBSD: vfs_cache.c,v 1.13 1996/02/04 02:18:09 christos Exp $ */
3
4 /*
5 * Copyright (c) 1989, 1993
6 * The Regents of the University of California. 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 * 3. 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 8.3 (Berkeley) 8/22/94
33 */
34
35 #include <sys/param.h>
36 #include <sys/systm.h>
37 #include <sys/time.h>
38 #include <sys/mount.h>
39 #include <sys/vnode.h>
40 #include <sys/namei.h>
41 #include <sys/errno.h>
42 #include <sys/malloc.h>
43 #include <sys/pool.h>
44 #include <sys/hash.h>
45
46 /*
47 * TODO: namecache access should really be locked.
48 */
49
50 /*
51 * Name caching works as follows:
52 *
53 * Names found by directory scans are retained in a cache
54 * for future reference. It is managed LRU, so frequently
55 * used names will hang around. Cache is indexed by hash value
56 * obtained from (vp, name) where vp refers to the directory
57 * containing name.
58 *
59 * For simplicity (and economy of storage), names longer than
60 * a maximum length of NCHNAMLEN are not cached; they occur
61 * infrequently in any case, and are almost never of interest.
62 *
63 * Upon reaching the last segment of a path, if the reference
64 * is for DELETE, or NOCACHE is set (rewrite), and the
65 * name is located in the cache, it will be dropped.
66 */
67
68 /*
69 * Structures associated with name caching.
70 */
71 LIST_HEAD(nchashhead, namecache) *nchashtbl;
72 u_long nchash; /* size of hash table - 1 */
73 long numcache; /* number of cache entries allocated */
74 TAILQ_HEAD(, namecache) nclruhead; /* LRU chain */
75 struct nchstats nchstats; /* cache effectiveness statistics */
76
77 LIST_HEAD(ncvhashhead, namecache) *ncvhashtbl;
78 u_long ncvhash; /* size of hash table - 1 */
79
80 #define NCHASH(dvp, cnp) \
81 hash32_buf(&(dvp)->v_id, sizeof((dvp)->v_id), (cnp)->cn_hash) & nchash
82
83 #define NCVHASH(vp) (vp)->v_id & ncvhash
84
85 int doingcache = 1; /* 1 => enable the cache */
86
87 struct pool nch_pool;
88
89 u_long nextvnodeid;
90
91 /*
92 * Look for a the name in the cache. We don't do this
93 * if the segment name is long, simply so the cache can avoid
94 * holding long names (which would either waste space, or
95 * add greatly to the complexity).
96 *
97 * Lookup is called with ni_dvp pointing to the directory to search,
98 * ni_ptr pointing to the name of the entry being sought, ni_namelen
99 * tells the length of the name, and ni_hash contains a hash of
100 * the name. If the lookup succeeds, the vnode is returned in ni_vp
101 * and a status of 0 is returned. If the locking fails for whatever
102 * reason, the vnode is unlocked and the error is returned to caller.
103 * If the lookup determines that the name does not exist (negative caching),
104 * a status of ENOENT is returned. If the lookup fails, a status of -1
105 * is returned.
106 */
107 int
cache_lookup(struct vnode * dvp,struct vnode ** vpp,struct componentname * cnp)108 cache_lookup(struct vnode *dvp, struct vnode **vpp, struct componentname *cnp)
109 {
110 struct namecache *ncp;
111 struct nchashhead *ncpp;
112 struct vnode *vp;
113 struct proc *p = curproc;
114 u_long vpid;
115 int error;
116
117 *vpp = NULL;
118
119 if (!doingcache) {
120 cnp->cn_flags &= ~MAKEENTRY;
121 return (-1);
122 }
123 if (cnp->cn_namelen > NCHNAMLEN) {
124 nchstats.ncs_long++;
125 cnp->cn_flags &= ~MAKEENTRY;
126 return (-1);
127 }
128
129 ncpp = &nchashtbl[NCHASH(dvp, cnp)];
130 LIST_FOREACH(ncp, ncpp, nc_hash) {
131 if (ncp->nc_dvp == dvp &&
132 ncp->nc_dvpid == dvp->v_id &&
133 ncp->nc_nlen == cnp->cn_namelen &&
134 !memcmp(ncp->nc_name, cnp->cn_nameptr, (u_int)ncp->nc_nlen))
135 break;
136 }
137 if (ncp == NULL) {
138 nchstats.ncs_miss++;
139 return (-1);
140 }
141 if ((cnp->cn_flags & MAKEENTRY) == 0) {
142 nchstats.ncs_badhits++;
143 goto remove;
144 } else if (ncp->nc_vp == NULL) {
145 if (cnp->cn_nameiop != CREATE ||
146 (cnp->cn_flags & ISLASTCN) == 0) {
147 nchstats.ncs_neghits++;
148 /*
149 * Move this slot to end of LRU chain,
150 * if not already there.
151 */
152 if (TAILQ_NEXT(ncp, nc_lru) != NULL) {
153 TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
154 TAILQ_INSERT_TAIL(&nclruhead, ncp, nc_lru);
155 }
156 return (ENOENT);
157 } else {
158 nchstats.ncs_badhits++;
159 goto remove;
160 }
161 } else if (ncp->nc_vpid != ncp->nc_vp->v_id) {
162 nchstats.ncs_falsehits++;
163 goto remove;
164 }
165
166 vp = ncp->nc_vp;
167 vpid = vp->v_id;
168 if (vp == dvp) { /* lookup on "." */
169 VREF(dvp);
170 error = 0;
171 } else if (cnp->cn_flags & ISDOTDOT) {
172 VOP_UNLOCK(dvp, 0, p);
173 cnp->cn_flags |= PDIRUNLOCK;
174 error = vget(vp, LK_EXCLUSIVE, p);
175 /*
176 * If the above vget() succeeded and both LOCKPARENT and
177 * ISLASTCN is set, lock the directory vnode as well.
178 */
179 if (!error && (~cnp->cn_flags & (LOCKPARENT|ISLASTCN)) == 0) {
180 if ((error = vn_lock(dvp, LK_EXCLUSIVE, p)) != 0) {
181 vput(vp);
182 return (error);
183 }
184 cnp->cn_flags &= ~PDIRUNLOCK;
185 }
186 } else {
187 error = vget(vp, LK_EXCLUSIVE, p);
188 /*
189 * If the above vget() failed or either of LOCKPARENT or
190 * ISLASTCN is set, unlock the directory vnode.
191 */
192 if (error || (~cnp->cn_flags & (LOCKPARENT|ISLASTCN)) != 0) {
193 VOP_UNLOCK(dvp, 0, p);
194 cnp->cn_flags |= PDIRUNLOCK;
195 }
196 }
197
198 /*
199 * Check that the lock succeeded, and that the capability number did
200 * not change while we were waiting for the lock.
201 */
202 if (error || vpid != vp->v_id) {
203 if (!error) {
204 vput(vp);
205 nchstats.ncs_falsehits++;
206 } else
207 nchstats.ncs_badhits++;
208 /*
209 * The parent needs to be locked when we return to VOP_LOOKUP().
210 * The `.' case here should be extremely rare (if it can happen
211 * at all), so we don't bother optimizing out the unlock/relock.
212 */
213 if (vp == dvp || error ||
214 (~cnp->cn_flags & (LOCKPARENT|ISLASTCN)) != 0) {
215 if ((error = vn_lock(dvp, LK_EXCLUSIVE, p)) != 0)
216 return (error);
217 cnp->cn_flags &= ~PDIRUNLOCK;
218 }
219 return (-1);
220 }
221
222 nchstats.ncs_goodhits++;
223 /*
224 * Move this slot to end of LRU chain, if not already there.
225 */
226 if (TAILQ_NEXT(ncp, nc_lru) != NULL) {
227 TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
228 TAILQ_INSERT_TAIL(&nclruhead, ncp, nc_lru);
229 }
230 *vpp = vp;
231 return (0);
232
233 remove:
234 /*
235 * Last component and we are renaming or deleting,
236 * the cache entry is invalid, or otherwise don't
237 * want cache entry to exist.
238 */
239 TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
240 LIST_REMOVE(ncp, nc_hash);
241 ncp->nc_hash.le_prev = NULL;
242
243 if (ncp->nc_vhash.le_prev != NULL) {
244 LIST_REMOVE(ncp, nc_vhash);
245 ncp->nc_vhash.le_prev = NULL;
246 }
247
248 pool_put(&nch_pool, ncp);
249 numcache--;
250 return (-1);
251 }
252
253 /*
254 * Scan cache looking for name of directory entry pointing at vp.
255 *
256 * Fill in dvpp.
257 *
258 * If bufp is non-NULL, also place the name in the buffer which starts
259 * at bufp, immediately before *bpp, and move bpp backwards to point
260 * at the start of it. (Yes, this is a little baroque, but it's done
261 * this way to cater to the whims of getcwd).
262 *
263 * Returns 0 on success, -1 on cache miss, positive errno on failure.
264 *
265 * TODO: should we return *dvpp locked?
266 */
267
268 int
cache_revlookup(struct vnode * vp,struct vnode ** dvpp,char ** bpp,char * bufp)269 cache_revlookup(struct vnode *vp, struct vnode **dvpp, char **bpp, char *bufp)
270 {
271 struct namecache *ncp;
272 struct vnode *dvp;
273 struct ncvhashhead *nvcpp;
274 char *bp;
275
276 if (!doingcache)
277 goto out;
278
279 nvcpp = &ncvhashtbl[NCVHASH(vp)];
280
281 LIST_FOREACH(ncp, nvcpp, nc_vhash) {
282 if (ncp->nc_vp == vp &&
283 ncp->nc_vpid == vp->v_id &&
284 (dvp = ncp->nc_dvp) != NULL &&
285 /* avoid pesky '.' entries.. */
286 dvp != vp && ncp->nc_dvpid == dvp->v_id) {
287
288 #ifdef DIAGNOSTIC
289 if (ncp->nc_nlen == 1 &&
290 ncp->nc_name[0] == '.')
291 panic("cache_revlookup: found entry for .");
292
293 if (ncp->nc_nlen == 2 &&
294 ncp->nc_name[0] == '.' &&
295 ncp->nc_name[1] == '.')
296 panic("cache_revlookup: found entry for ..");
297 #endif
298 nchstats.ncs_revhits++;
299
300 if (bufp != NULL) {
301 bp = *bpp;
302 bp -= ncp->nc_nlen;
303 if (bp <= bufp) {
304 *dvpp = NULL;
305 return (ERANGE);
306 }
307 memcpy(bp, ncp->nc_name, ncp->nc_nlen);
308 *bpp = bp;
309 }
310
311 *dvpp = dvp;
312
313 /*
314 * XXX: Should we vget() here to have more
315 * consistent semantics with cache_lookup()?
316 *
317 * For MP safety it might be necessary to do
318 * this here, while also protecting hash
319 * tables themselves to provide some sort of
320 * sane inter locking.
321 */
322 return (0);
323 }
324 }
325 nchstats.ncs_revmiss++;
326
327 out:
328 *dvpp = NULL;
329 return (-1);
330 }
331
332 /*
333 * Add an entry to the cache
334 */
335 void
cache_enter(struct vnode * dvp,struct vnode * vp,struct componentname * cnp)336 cache_enter(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
337 {
338 struct namecache *ncp;
339 struct nchashhead *ncpp;
340 struct ncvhashhead *nvcpp;
341
342 if (!doingcache || cnp->cn_namelen > NCHNAMLEN)
343 return;
344
345 /*
346 * Free the cache slot at head of lru chain.
347 */
348 if (numcache < desiredvnodes) {
349 ncp = pool_get(&nch_pool, PR_WAITOK);
350 bzero((char *)ncp, sizeof *ncp);
351 numcache++;
352 } else if ((ncp = TAILQ_FIRST(&nclruhead)) != NULL) {
353 TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
354 if (ncp->nc_hash.le_prev != NULL) {
355 LIST_REMOVE(ncp, nc_hash);
356 ncp->nc_hash.le_prev = NULL;
357 }
358 if (ncp->nc_vhash.le_prev != NULL) {
359 LIST_REMOVE(ncp, nc_vhash);
360 ncp->nc_vhash.le_prev = NULL;
361 }
362 } else
363 return;
364 /* grab the vnode we just found */
365 ncp->nc_vp = vp;
366 if (vp)
367 ncp->nc_vpid = vp->v_id;
368 /* fill in cache info */
369 ncp->nc_dvp = dvp;
370 ncp->nc_dvpid = dvp->v_id;
371 ncp->nc_nlen = cnp->cn_namelen;
372 bcopy(cnp->cn_nameptr, ncp->nc_name, (unsigned)ncp->nc_nlen);
373 TAILQ_INSERT_TAIL(&nclruhead, ncp, nc_lru);
374 ncpp = &nchashtbl[NCHASH(dvp, cnp)];
375 LIST_INSERT_HEAD(ncpp, ncp, nc_hash);
376
377 /*
378 * Create reverse-cache entries (used in getcwd) for
379 * directories.
380 */
381
382 ncp->nc_vhash.le_prev = NULL;
383 ncp->nc_vhash.le_next = NULL;
384
385 if (vp && vp != dvp && vp->v_type == VDIR &&
386 (ncp->nc_nlen > 2 ||
387 (ncp->nc_nlen > 1 && ncp->nc_name[1] != '.') ||
388 (ncp->nc_nlen > 0 && ncp->nc_name[0] != '.'))) {
389 nvcpp = &ncvhashtbl[NCVHASH(vp)];
390 LIST_INSERT_HEAD(nvcpp, ncp, nc_vhash);
391 }
392 }
393
394 /*
395 * Name cache initialization, from vfs_init() when we are booting
396 */
397 void
nchinit(void)398 nchinit(void)
399 {
400
401 TAILQ_INIT(&nclruhead);
402 nchashtbl = hashinit(desiredvnodes, M_CACHE, M_WAITOK, &nchash);
403 ncvhashtbl = hashinit(desiredvnodes/8, M_CACHE, M_WAITOK, &ncvhash);
404 pool_init(&nch_pool, sizeof(struct namecache), 0, 0, 0, "nchpl",
405 &pool_allocator_nointr);
406 }
407
408 /*
409 * Cache flush, a particular vnode; called when a vnode is renamed to
410 * hide entries that would now be invalid
411 */
412 void
cache_purge(struct vnode * vp)413 cache_purge(struct vnode *vp)
414 {
415 struct namecache *ncp;
416 struct nchashhead *ncpp;
417
418 vp->v_id = ++nextvnodeid;
419 if (nextvnodeid != 0)
420 return;
421 for (ncpp = &nchashtbl[nchash]; ncpp >= nchashtbl; ncpp--) {
422 LIST_FOREACH(ncp, ncpp, nc_hash) {
423 ncp->nc_vpid = 0;
424 ncp->nc_dvpid = 0;
425 }
426 }
427 vp->v_id = ++nextvnodeid;
428 }
429
430 /*
431 * Cache flush, a whole filesystem; called when filesys is umounted to
432 * remove entries that would now be invalid
433 */
434 void
cache_purgevfs(struct mount * mp)435 cache_purgevfs(struct mount *mp)
436 {
437 struct namecache *ncp, *nxtcp;
438
439 for (ncp = TAILQ_FIRST(&nclruhead); ncp != TAILQ_END(&nclruhead);
440 ncp = nxtcp) {
441 nxtcp = TAILQ_NEXT(ncp, nc_lru);
442 if (ncp->nc_dvp == NULL || ncp->nc_dvp->v_mount != mp)
443 continue;
444 /* free the resources we had */
445 ncp->nc_vp = NULL;
446 ncp->nc_dvp = NULL;
447 TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
448 if (ncp->nc_hash.le_prev != NULL) {
449 LIST_REMOVE(ncp, nc_hash);
450 ncp->nc_hash.le_prev = NULL;
451 }
452 if (ncp->nc_vhash.le_prev != NULL) {
453 LIST_REMOVE(ncp, nc_vhash);
454 ncp->nc_vhash.le_prev = NULL;
455 }
456 pool_put(&nch_pool, ncp);
457 numcache--;
458 }
459 }
460