Root/
1 | /* |
2 | * fs/dcache.c |
3 | * |
4 | * Complete reimplementation |
5 | * (C) 1997 Thomas Schoebel-Theuer, |
6 | * with heavy changes by Linus Torvalds |
7 | */ |
8 | |
9 | /* |
10 | * Notes on the allocation strategy: |
11 | * |
12 | * The dcache is a master of the icache - whenever a dcache entry |
13 | * exists, the inode will always exist. "iput()" is done either when |
14 | * the dcache entry is deleted or garbage collected. |
15 | */ |
16 | |
17 | #include <linux/syscalls.h> |
18 | #include <linux/string.h> |
19 | #include <linux/mm.h> |
20 | #include <linux/fs.h> |
21 | #include <linux/fsnotify.h> |
22 | #include <linux/slab.h> |
23 | #include <linux/init.h> |
24 | #include <linux/hash.h> |
25 | #include <linux/cache.h> |
26 | #include <linux/module.h> |
27 | #include <linux/mount.h> |
28 | #include <linux/file.h> |
29 | #include <asm/uaccess.h> |
30 | #include <linux/security.h> |
31 | #include <linux/seqlock.h> |
32 | #include <linux/swap.h> |
33 | #include <linux/bootmem.h> |
34 | #include <linux/fs_struct.h> |
35 | #include <linux/hardirq.h> |
36 | #include "internal.h" |
37 | |
38 | int sysctl_vfs_cache_pressure __read_mostly = 100; |
39 | EXPORT_SYMBOL_GPL(sysctl_vfs_cache_pressure); |
40 | |
41 | __cacheline_aligned_in_smp DEFINE_SPINLOCK(dcache_lock); |
42 | __cacheline_aligned_in_smp DEFINE_SEQLOCK(rename_lock); |
43 | |
44 | EXPORT_SYMBOL(dcache_lock); |
45 | |
46 | static struct kmem_cache *dentry_cache __read_mostly; |
47 | |
48 | #define DNAME_INLINE_LEN (sizeof(struct dentry)-offsetof(struct dentry,d_iname)) |
49 | |
50 | /* |
51 | * This is the single most critical data structure when it comes |
52 | * to the dcache: the hashtable for lookups. Somebody should try |
53 | * to make this good - I've just made it work. |
54 | * |
55 | * This hash-function tries to avoid losing too many bits of hash |
56 | * information, yet avoid using a prime hash-size or similar. |
57 | */ |
58 | #define D_HASHBITS d_hash_shift |
59 | #define D_HASHMASK d_hash_mask |
60 | |
61 | static unsigned int d_hash_mask __read_mostly; |
62 | static unsigned int d_hash_shift __read_mostly; |
63 | static struct hlist_head *dentry_hashtable __read_mostly; |
64 | |
65 | /* Statistics gathering. */ |
66 | struct dentry_stat_t dentry_stat = { |
67 | .age_limit = 45, |
68 | }; |
69 | |
70 | static void __d_free(struct dentry *dentry) |
71 | { |
72 | WARN_ON(!list_empty(&dentry->d_alias)); |
73 | if (dname_external(dentry)) |
74 | kfree(dentry->d_name.name); |
75 | kmem_cache_free(dentry_cache, dentry); |
76 | } |
77 | |
78 | static void d_callback(struct rcu_head *head) |
79 | { |
80 | struct dentry * dentry = container_of(head, struct dentry, d_u.d_rcu); |
81 | __d_free(dentry); |
82 | } |
83 | |
84 | /* |
85 | * no dcache_lock, please. The caller must decrement dentry_stat.nr_dentry |
86 | * inside dcache_lock. |
87 | */ |
88 | static void d_free(struct dentry *dentry) |
89 | { |
90 | if (dentry->d_op && dentry->d_op->d_release) |
91 | dentry->d_op->d_release(dentry); |
92 | /* if dentry was never inserted into hash, immediate free is OK */ |
93 | if (hlist_unhashed(&dentry->d_hash)) |
94 | __d_free(dentry); |
95 | else |
96 | call_rcu(&dentry->d_u.d_rcu, d_callback); |
97 | } |
98 | |
99 | /* |
100 | * Release the dentry's inode, using the filesystem |
101 | * d_iput() operation if defined. |
102 | */ |
103 | static void dentry_iput(struct dentry * dentry) |
104 | __releases(dentry->d_lock) |
105 | __releases(dcache_lock) |
106 | { |
107 | struct inode *inode = dentry->d_inode; |
108 | if (inode) { |
109 | dentry->d_inode = NULL; |
110 | list_del_init(&dentry->d_alias); |
111 | spin_unlock(&dentry->d_lock); |
112 | spin_unlock(&dcache_lock); |
113 | if (!inode->i_nlink) |
114 | fsnotify_inoderemove(inode); |
115 | if (dentry->d_op && dentry->d_op->d_iput) |
116 | dentry->d_op->d_iput(dentry, inode); |
117 | else |
118 | iput(inode); |
119 | } else { |
120 | spin_unlock(&dentry->d_lock); |
121 | spin_unlock(&dcache_lock); |
122 | } |
123 | } |
124 | |
125 | /* |
126 | * dentry_lru_(add|add_tail|del|del_init) must be called with dcache_lock held. |
127 | */ |
128 | static void dentry_lru_add(struct dentry *dentry) |
129 | { |
130 | list_add(&dentry->d_lru, &dentry->d_sb->s_dentry_lru); |
131 | dentry->d_sb->s_nr_dentry_unused++; |
132 | dentry_stat.nr_unused++; |
133 | } |
134 | |
135 | static void dentry_lru_add_tail(struct dentry *dentry) |
136 | { |
137 | list_add_tail(&dentry->d_lru, &dentry->d_sb->s_dentry_lru); |
138 | dentry->d_sb->s_nr_dentry_unused++; |
139 | dentry_stat.nr_unused++; |
140 | } |
141 | |
142 | static void dentry_lru_del(struct dentry *dentry) |
143 | { |
144 | if (!list_empty(&dentry->d_lru)) { |
145 | list_del(&dentry->d_lru); |
146 | dentry->d_sb->s_nr_dentry_unused--; |
147 | dentry_stat.nr_unused--; |
148 | } |
149 | } |
150 | |
151 | static void dentry_lru_del_init(struct dentry *dentry) |
152 | { |
153 | if (likely(!list_empty(&dentry->d_lru))) { |
154 | list_del_init(&dentry->d_lru); |
155 | dentry->d_sb->s_nr_dentry_unused--; |
156 | dentry_stat.nr_unused--; |
157 | } |
158 | } |
159 | |
160 | /** |
161 | * d_kill - kill dentry and return parent |
162 | * @dentry: dentry to kill |
163 | * |
164 | * The dentry must already be unhashed and removed from the LRU. |
165 | * |
166 | * If this is the root of the dentry tree, return NULL. |
167 | */ |
168 | static struct dentry *d_kill(struct dentry *dentry) |
169 | __releases(dentry->d_lock) |
170 | __releases(dcache_lock) |
171 | { |
172 | struct dentry *parent; |
173 | |
174 | list_del(&dentry->d_u.d_child); |
175 | dentry_stat.nr_dentry--; /* For d_free, below */ |
176 | /*drops the locks, at that point nobody can reach this dentry */ |
177 | dentry_iput(dentry); |
178 | if (IS_ROOT(dentry)) |
179 | parent = NULL; |
180 | else |
181 | parent = dentry->d_parent; |
182 | d_free(dentry); |
183 | return parent; |
184 | } |
185 | |
186 | /* |
187 | * This is dput |
188 | * |
189 | * This is complicated by the fact that we do not want to put |
190 | * dentries that are no longer on any hash chain on the unused |
191 | * list: we'd much rather just get rid of them immediately. |
192 | * |
193 | * However, that implies that we have to traverse the dentry |
194 | * tree upwards to the parents which might _also_ now be |
195 | * scheduled for deletion (it may have been only waiting for |
196 | * its last child to go away). |
197 | * |
198 | * This tail recursion is done by hand as we don't want to depend |
199 | * on the compiler to always get this right (gcc generally doesn't). |
200 | * Real recursion would eat up our stack space. |
201 | */ |
202 | |
203 | /* |
204 | * dput - release a dentry |
205 | * @dentry: dentry to release |
206 | * |
207 | * Release a dentry. This will drop the usage count and if appropriate |
208 | * call the dentry unlink method as well as removing it from the queues and |
209 | * releasing its resources. If the parent dentries were scheduled for release |
210 | * they too may now get deleted. |
211 | * |
212 | * no dcache lock, please. |
213 | */ |
214 | |
215 | void dput(struct dentry *dentry) |
216 | { |
217 | if (!dentry) |
218 | return; |
219 | |
220 | repeat: |
221 | if (atomic_read(&dentry->d_count) == 1) |
222 | might_sleep(); |
223 | if (!atomic_dec_and_lock(&dentry->d_count, &dcache_lock)) |
224 | return; |
225 | |
226 | spin_lock(&dentry->d_lock); |
227 | if (atomic_read(&dentry->d_count)) { |
228 | spin_unlock(&dentry->d_lock); |
229 | spin_unlock(&dcache_lock); |
230 | return; |
231 | } |
232 | |
233 | /* |
234 | * AV: ->d_delete() is _NOT_ allowed to block now. |
235 | */ |
236 | if (dentry->d_op && dentry->d_op->d_delete) { |
237 | if (dentry->d_op->d_delete(dentry)) |
238 | goto unhash_it; |
239 | } |
240 | /* Unreachable? Get rid of it */ |
241 | if (d_unhashed(dentry)) |
242 | goto kill_it; |
243 | if (list_empty(&dentry->d_lru)) { |
244 | dentry->d_flags |= DCACHE_REFERENCED; |
245 | dentry_lru_add(dentry); |
246 | } |
247 | spin_unlock(&dentry->d_lock); |
248 | spin_unlock(&dcache_lock); |
249 | return; |
250 | |
251 | unhash_it: |
252 | __d_drop(dentry); |
253 | kill_it: |
254 | /* if dentry was on the d_lru list delete it from there */ |
255 | dentry_lru_del(dentry); |
256 | dentry = d_kill(dentry); |
257 | if (dentry) |
258 | goto repeat; |
259 | } |
260 | EXPORT_SYMBOL(dput); |
261 | |
262 | /** |
263 | * d_invalidate - invalidate a dentry |
264 | * @dentry: dentry to invalidate |
265 | * |
266 | * Try to invalidate the dentry if it turns out to be |
267 | * possible. If there are other dentries that can be |
268 | * reached through this one we can't delete it and we |
269 | * return -EBUSY. On success we return 0. |
270 | * |
271 | * no dcache lock. |
272 | */ |
273 | |
274 | int d_invalidate(struct dentry * dentry) |
275 | { |
276 | /* |
277 | * If it's already been dropped, return OK. |
278 | */ |
279 | spin_lock(&dcache_lock); |
280 | if (d_unhashed(dentry)) { |
281 | spin_unlock(&dcache_lock); |
282 | return 0; |
283 | } |
284 | /* |
285 | * Check whether to do a partial shrink_dcache |
286 | * to get rid of unused child entries. |
287 | */ |
288 | if (!list_empty(&dentry->d_subdirs)) { |
289 | spin_unlock(&dcache_lock); |
290 | shrink_dcache_parent(dentry); |
291 | spin_lock(&dcache_lock); |
292 | } |
293 | |
294 | /* |
295 | * Somebody else still using it? |
296 | * |
297 | * If it's a directory, we can't drop it |
298 | * for fear of somebody re-populating it |
299 | * with children (even though dropping it |
300 | * would make it unreachable from the root, |
301 | * we might still populate it if it was a |
302 | * working directory or similar). |
303 | */ |
304 | spin_lock(&dentry->d_lock); |
305 | if (atomic_read(&dentry->d_count) > 1) { |
306 | if (dentry->d_inode && S_ISDIR(dentry->d_inode->i_mode)) { |
307 | spin_unlock(&dentry->d_lock); |
308 | spin_unlock(&dcache_lock); |
309 | return -EBUSY; |
310 | } |
311 | } |
312 | |
313 | __d_drop(dentry); |
314 | spin_unlock(&dentry->d_lock); |
315 | spin_unlock(&dcache_lock); |
316 | return 0; |
317 | } |
318 | EXPORT_SYMBOL(d_invalidate); |
319 | |
320 | /* This should be called _only_ with dcache_lock held */ |
321 | |
322 | static inline struct dentry * __dget_locked(struct dentry *dentry) |
323 | { |
324 | atomic_inc(&dentry->d_count); |
325 | dentry_lru_del_init(dentry); |
326 | return dentry; |
327 | } |
328 | |
329 | struct dentry * dget_locked(struct dentry *dentry) |
330 | { |
331 | return __dget_locked(dentry); |
332 | } |
333 | EXPORT_SYMBOL(dget_locked); |
334 | |
335 | /** |
336 | * d_find_alias - grab a hashed alias of inode |
337 | * @inode: inode in question |
338 | * @want_discon: flag, used by d_splice_alias, to request |
339 | * that only a DISCONNECTED alias be returned. |
340 | * |
341 | * If inode has a hashed alias, or is a directory and has any alias, |
342 | * acquire the reference to alias and return it. Otherwise return NULL. |
343 | * Notice that if inode is a directory there can be only one alias and |
344 | * it can be unhashed only if it has no children, or if it is the root |
345 | * of a filesystem. |
346 | * |
347 | * If the inode has an IS_ROOT, DCACHE_DISCONNECTED alias, then prefer |
348 | * any other hashed alias over that one unless @want_discon is set, |
349 | * in which case only return an IS_ROOT, DCACHE_DISCONNECTED alias. |
350 | */ |
351 | |
352 | static struct dentry * __d_find_alias(struct inode *inode, int want_discon) |
353 | { |
354 | struct list_head *head, *next, *tmp; |
355 | struct dentry *alias, *discon_alias=NULL; |
356 | |
357 | head = &inode->i_dentry; |
358 | next = inode->i_dentry.next; |
359 | while (next != head) { |
360 | tmp = next; |
361 | next = tmp->next; |
362 | prefetch(next); |
363 | alias = list_entry(tmp, struct dentry, d_alias); |
364 | if (S_ISDIR(inode->i_mode) || !d_unhashed(alias)) { |
365 | if (IS_ROOT(alias) && |
366 | (alias->d_flags & DCACHE_DISCONNECTED)) |
367 | discon_alias = alias; |
368 | else if (!want_discon) { |
369 | __dget_locked(alias); |
370 | return alias; |
371 | } |
372 | } |
373 | } |
374 | if (discon_alias) |
375 | __dget_locked(discon_alias); |
376 | return discon_alias; |
377 | } |
378 | |
379 | struct dentry * d_find_alias(struct inode *inode) |
380 | { |
381 | struct dentry *de = NULL; |
382 | |
383 | if (!list_empty(&inode->i_dentry)) { |
384 | spin_lock(&dcache_lock); |
385 | de = __d_find_alias(inode, 0); |
386 | spin_unlock(&dcache_lock); |
387 | } |
388 | return de; |
389 | } |
390 | EXPORT_SYMBOL(d_find_alias); |
391 | |
392 | /* |
393 | * Try to kill dentries associated with this inode. |
394 | * WARNING: you must own a reference to inode. |
395 | */ |
396 | void d_prune_aliases(struct inode *inode) |
397 | { |
398 | struct dentry *dentry; |
399 | restart: |
400 | spin_lock(&dcache_lock); |
401 | list_for_each_entry(dentry, &inode->i_dentry, d_alias) { |
402 | spin_lock(&dentry->d_lock); |
403 | if (!atomic_read(&dentry->d_count)) { |
404 | __dget_locked(dentry); |
405 | __d_drop(dentry); |
406 | spin_unlock(&dentry->d_lock); |
407 | spin_unlock(&dcache_lock); |
408 | dput(dentry); |
409 | goto restart; |
410 | } |
411 | spin_unlock(&dentry->d_lock); |
412 | } |
413 | spin_unlock(&dcache_lock); |
414 | } |
415 | EXPORT_SYMBOL(d_prune_aliases); |
416 | |
417 | /* |
418 | * Throw away a dentry - free the inode, dput the parent. This requires that |
419 | * the LRU list has already been removed. |
420 | * |
421 | * Try to prune ancestors as well. This is necessary to prevent |
422 | * quadratic behavior of shrink_dcache_parent(), but is also expected |
423 | * to be beneficial in reducing dentry cache fragmentation. |
424 | */ |
425 | static void prune_one_dentry(struct dentry * dentry) |
426 | __releases(dentry->d_lock) |
427 | __releases(dcache_lock) |
428 | __acquires(dcache_lock) |
429 | { |
430 | __d_drop(dentry); |
431 | dentry = d_kill(dentry); |
432 | |
433 | /* |
434 | * Prune ancestors. Locking is simpler than in dput(), |
435 | * because dcache_lock needs to be taken anyway. |
436 | */ |
437 | spin_lock(&dcache_lock); |
438 | while (dentry) { |
439 | if (!atomic_dec_and_lock(&dentry->d_count, &dentry->d_lock)) |
440 | return; |
441 | |
442 | if (dentry->d_op && dentry->d_op->d_delete) |
443 | dentry->d_op->d_delete(dentry); |
444 | dentry_lru_del_init(dentry); |
445 | __d_drop(dentry); |
446 | dentry = d_kill(dentry); |
447 | spin_lock(&dcache_lock); |
448 | } |
449 | } |
450 | |
451 | /* |
452 | * Shrink the dentry LRU on a given superblock. |
453 | * @sb : superblock to shrink dentry LRU. |
454 | * @count: If count is NULL, we prune all dentries on superblock. |
455 | * @flags: If flags is non-zero, we need to do special processing based on |
456 | * which flags are set. This means we don't need to maintain multiple |
457 | * similar copies of this loop. |
458 | */ |
459 | static void __shrink_dcache_sb(struct super_block *sb, int *count, int flags) |
460 | { |
461 | LIST_HEAD(referenced); |
462 | LIST_HEAD(tmp); |
463 | struct dentry *dentry; |
464 | int cnt = 0; |
465 | |
466 | BUG_ON(!sb); |
467 | BUG_ON((flags & DCACHE_REFERENCED) && count == NULL); |
468 | spin_lock(&dcache_lock); |
469 | if (count != NULL) |
470 | /* called from prune_dcache() and shrink_dcache_parent() */ |
471 | cnt = *count; |
472 | restart: |
473 | if (count == NULL) |
474 | list_splice_init(&sb->s_dentry_lru, &tmp); |
475 | else { |
476 | while (!list_empty(&sb->s_dentry_lru)) { |
477 | dentry = list_entry(sb->s_dentry_lru.prev, |
478 | struct dentry, d_lru); |
479 | BUG_ON(dentry->d_sb != sb); |
480 | |
481 | spin_lock(&dentry->d_lock); |
482 | /* |
483 | * If we are honouring the DCACHE_REFERENCED flag and |
484 | * the dentry has this flag set, don't free it. Clear |
485 | * the flag and put it back on the LRU. |
486 | */ |
487 | if ((flags & DCACHE_REFERENCED) |
488 | && (dentry->d_flags & DCACHE_REFERENCED)) { |
489 | dentry->d_flags &= ~DCACHE_REFERENCED; |
490 | list_move(&dentry->d_lru, &referenced); |
491 | spin_unlock(&dentry->d_lock); |
492 | } else { |
493 | list_move_tail(&dentry->d_lru, &tmp); |
494 | spin_unlock(&dentry->d_lock); |
495 | cnt--; |
496 | if (!cnt) |
497 | break; |
498 | } |
499 | cond_resched_lock(&dcache_lock); |
500 | } |
501 | } |
502 | while (!list_empty(&tmp)) { |
503 | dentry = list_entry(tmp.prev, struct dentry, d_lru); |
504 | dentry_lru_del_init(dentry); |
505 | spin_lock(&dentry->d_lock); |
506 | /* |
507 | * We found an inuse dentry which was not removed from |
508 | * the LRU because of laziness during lookup. Do not free |
509 | * it - just keep it off the LRU list. |
510 | */ |
511 | if (atomic_read(&dentry->d_count)) { |
512 | spin_unlock(&dentry->d_lock); |
513 | continue; |
514 | } |
515 | prune_one_dentry(dentry); |
516 | /* dentry->d_lock was dropped in prune_one_dentry() */ |
517 | cond_resched_lock(&dcache_lock); |
518 | } |
519 | if (count == NULL && !list_empty(&sb->s_dentry_lru)) |
520 | goto restart; |
521 | if (count != NULL) |
522 | *count = cnt; |
523 | if (!list_empty(&referenced)) |
524 | list_splice(&referenced, &sb->s_dentry_lru); |
525 | spin_unlock(&dcache_lock); |
526 | } |
527 | |
528 | /** |
529 | * prune_dcache - shrink the dcache |
530 | * @count: number of entries to try to free |
531 | * |
532 | * Shrink the dcache. This is done when we need more memory, or simply when we |
533 | * need to unmount something (at which point we need to unuse all dentries). |
534 | * |
535 | * This function may fail to free any resources if all the dentries are in use. |
536 | */ |
537 | static void prune_dcache(int count) |
538 | { |
539 | struct super_block *sb; |
540 | int w_count; |
541 | int unused = dentry_stat.nr_unused; |
542 | int prune_ratio; |
543 | int pruned; |
544 | |
545 | if (unused == 0 || count == 0) |
546 | return; |
547 | spin_lock(&dcache_lock); |
548 | restart: |
549 | if (count >= unused) |
550 | prune_ratio = 1; |
551 | else |
552 | prune_ratio = unused / count; |
553 | spin_lock(&sb_lock); |
554 | list_for_each_entry(sb, &super_blocks, s_list) { |
555 | if (sb->s_nr_dentry_unused == 0) |
556 | continue; |
557 | sb->s_count++; |
558 | /* Now, we reclaim unused dentrins with fairness. |
559 | * We reclaim them same percentage from each superblock. |
560 | * We calculate number of dentries to scan on this sb |
561 | * as follows, but the implementation is arranged to avoid |
562 | * overflows: |
563 | * number of dentries to scan on this sb = |
564 | * count * (number of dentries on this sb / |
565 | * number of dentries in the machine) |
566 | */ |
567 | spin_unlock(&sb_lock); |
568 | if (prune_ratio != 1) |
569 | w_count = (sb->s_nr_dentry_unused / prune_ratio) + 1; |
570 | else |
571 | w_count = sb->s_nr_dentry_unused; |
572 | pruned = w_count; |
573 | /* |
574 | * We need to be sure this filesystem isn't being unmounted, |
575 | * otherwise we could race with generic_shutdown_super(), and |
576 | * end up holding a reference to an inode while the filesystem |
577 | * is unmounted. So we try to get s_umount, and make sure |
578 | * s_root isn't NULL. |
579 | */ |
580 | if (down_read_trylock(&sb->s_umount)) { |
581 | if ((sb->s_root != NULL) && |
582 | (!list_empty(&sb->s_dentry_lru))) { |
583 | spin_unlock(&dcache_lock); |
584 | __shrink_dcache_sb(sb, &w_count, |
585 | DCACHE_REFERENCED); |
586 | pruned -= w_count; |
587 | spin_lock(&dcache_lock); |
588 | } |
589 | up_read(&sb->s_umount); |
590 | } |
591 | spin_lock(&sb_lock); |
592 | count -= pruned; |
593 | /* |
594 | * restart only when sb is no longer on the list and |
595 | * we have more work to do. |
596 | */ |
597 | if (__put_super_and_need_restart(sb) && count > 0) { |
598 | spin_unlock(&sb_lock); |
599 | goto restart; |
600 | } |
601 | } |
602 | spin_unlock(&sb_lock); |
603 | spin_unlock(&dcache_lock); |
604 | } |
605 | |
606 | /** |
607 | * shrink_dcache_sb - shrink dcache for a superblock |
608 | * @sb: superblock |
609 | * |
610 | * Shrink the dcache for the specified super block. This |
611 | * is used to free the dcache before unmounting a file |
612 | * system |
613 | */ |
614 | void shrink_dcache_sb(struct super_block * sb) |
615 | { |
616 | __shrink_dcache_sb(sb, NULL, 0); |
617 | } |
618 | EXPORT_SYMBOL(shrink_dcache_sb); |
619 | |
620 | /* |
621 | * destroy a single subtree of dentries for unmount |
622 | * - see the comments on shrink_dcache_for_umount() for a description of the |
623 | * locking |
624 | */ |
625 | static void shrink_dcache_for_umount_subtree(struct dentry *dentry) |
626 | { |
627 | struct dentry *parent; |
628 | unsigned detached = 0; |
629 | |
630 | BUG_ON(!IS_ROOT(dentry)); |
631 | |
632 | /* detach this root from the system */ |
633 | spin_lock(&dcache_lock); |
634 | dentry_lru_del_init(dentry); |
635 | __d_drop(dentry); |
636 | spin_unlock(&dcache_lock); |
637 | |
638 | for (;;) { |
639 | /* descend to the first leaf in the current subtree */ |
640 | while (!list_empty(&dentry->d_subdirs)) { |
641 | struct dentry *loop; |
642 | |
643 | /* this is a branch with children - detach all of them |
644 | * from the system in one go */ |
645 | spin_lock(&dcache_lock); |
646 | list_for_each_entry(loop, &dentry->d_subdirs, |
647 | d_u.d_child) { |
648 | dentry_lru_del_init(loop); |
649 | __d_drop(loop); |
650 | cond_resched_lock(&dcache_lock); |
651 | } |
652 | spin_unlock(&dcache_lock); |
653 | |
654 | /* move to the first child */ |
655 | dentry = list_entry(dentry->d_subdirs.next, |
656 | struct dentry, d_u.d_child); |
657 | } |
658 | |
659 | /* consume the dentries from this leaf up through its parents |
660 | * until we find one with children or run out altogether */ |
661 | do { |
662 | struct inode *inode; |
663 | |
664 | if (atomic_read(&dentry->d_count) != 0) { |
665 | printk(KERN_ERR |
666 | "BUG: Dentry %p{i=%lx,n=%s}" |
667 | " still in use (%d)" |
668 | " [unmount of %s %s]\n", |
669 | dentry, |
670 | dentry->d_inode ? |
671 | dentry->d_inode->i_ino : 0UL, |
672 | dentry->d_name.name, |
673 | atomic_read(&dentry->d_count), |
674 | dentry->d_sb->s_type->name, |
675 | dentry->d_sb->s_id); |
676 | BUG(); |
677 | } |
678 | |
679 | if (IS_ROOT(dentry)) |
680 | parent = NULL; |
681 | else { |
682 | parent = dentry->d_parent; |
683 | atomic_dec(&parent->d_count); |
684 | } |
685 | |
686 | list_del(&dentry->d_u.d_child); |
687 | detached++; |
688 | |
689 | inode = dentry->d_inode; |
690 | if (inode) { |
691 | dentry->d_inode = NULL; |
692 | list_del_init(&dentry->d_alias); |
693 | if (dentry->d_op && dentry->d_op->d_iput) |
694 | dentry->d_op->d_iput(dentry, inode); |
695 | else |
696 | iput(inode); |
697 | } |
698 | |
699 | d_free(dentry); |
700 | |
701 | /* finished when we fall off the top of the tree, |
702 | * otherwise we ascend to the parent and move to the |
703 | * next sibling if there is one */ |
704 | if (!parent) |
705 | goto out; |
706 | |
707 | dentry = parent; |
708 | |
709 | } while (list_empty(&dentry->d_subdirs)); |
710 | |
711 | dentry = list_entry(dentry->d_subdirs.next, |
712 | struct dentry, d_u.d_child); |
713 | } |
714 | out: |
715 | /* several dentries were freed, need to correct nr_dentry */ |
716 | spin_lock(&dcache_lock); |
717 | dentry_stat.nr_dentry -= detached; |
718 | spin_unlock(&dcache_lock); |
719 | } |
720 | |
721 | /* |
722 | * destroy the dentries attached to a superblock on unmounting |
723 | * - we don't need to use dentry->d_lock, and only need dcache_lock when |
724 | * removing the dentry from the system lists and hashes because: |
725 | * - the superblock is detached from all mountings and open files, so the |
726 | * dentry trees will not be rearranged by the VFS |
727 | * - s_umount is write-locked, so the memory pressure shrinker will ignore |
728 | * any dentries belonging to this superblock that it comes across |
729 | * - the filesystem itself is no longer permitted to rearrange the dentries |
730 | * in this superblock |
731 | */ |
732 | void shrink_dcache_for_umount(struct super_block *sb) |
733 | { |
734 | struct dentry *dentry; |
735 | |
736 | if (down_read_trylock(&sb->s_umount)) |
737 | BUG(); |
738 | |
739 | dentry = sb->s_root; |
740 | sb->s_root = NULL; |
741 | atomic_dec(&dentry->d_count); |
742 | shrink_dcache_for_umount_subtree(dentry); |
743 | |
744 | while (!hlist_empty(&sb->s_anon)) { |
745 | dentry = hlist_entry(sb->s_anon.first, struct dentry, d_hash); |
746 | shrink_dcache_for_umount_subtree(dentry); |
747 | } |
748 | } |
749 | |
750 | /* |
751 | * Search for at least 1 mount point in the dentry's subdirs. |
752 | * We descend to the next level whenever the d_subdirs |
753 | * list is non-empty and continue searching. |
754 | */ |
755 | |
756 | /** |
757 | * have_submounts - check for mounts over a dentry |
758 | * @parent: dentry to check. |
759 | * |
760 | * Return true if the parent or its subdirectories contain |
761 | * a mount point |
762 | */ |
763 | |
764 | int have_submounts(struct dentry *parent) |
765 | { |
766 | struct dentry *this_parent = parent; |
767 | struct list_head *next; |
768 | |
769 | spin_lock(&dcache_lock); |
770 | if (d_mountpoint(parent)) |
771 | goto positive; |
772 | repeat: |
773 | next = this_parent->d_subdirs.next; |
774 | resume: |
775 | while (next != &this_parent->d_subdirs) { |
776 | struct list_head *tmp = next; |
777 | struct dentry *dentry = list_entry(tmp, struct dentry, d_u.d_child); |
778 | next = tmp->next; |
779 | /* Have we found a mount point ? */ |
780 | if (d_mountpoint(dentry)) |
781 | goto positive; |
782 | if (!list_empty(&dentry->d_subdirs)) { |
783 | this_parent = dentry; |
784 | goto repeat; |
785 | } |
786 | } |
787 | /* |
788 | * All done at this level ... ascend and resume the search. |
789 | */ |
790 | if (this_parent != parent) { |
791 | next = this_parent->d_u.d_child.next; |
792 | this_parent = this_parent->d_parent; |
793 | goto resume; |
794 | } |
795 | spin_unlock(&dcache_lock); |
796 | return 0; /* No mount points found in tree */ |
797 | positive: |
798 | spin_unlock(&dcache_lock); |
799 | return 1; |
800 | } |
801 | EXPORT_SYMBOL(have_submounts); |
802 | |
803 | /* |
804 | * Search the dentry child list for the specified parent, |
805 | * and move any unused dentries to the end of the unused |
806 | * list for prune_dcache(). We descend to the next level |
807 | * whenever the d_subdirs list is non-empty and continue |
808 | * searching. |
809 | * |
810 | * It returns zero iff there are no unused children, |
811 | * otherwise it returns the number of children moved to |
812 | * the end of the unused list. This may not be the total |
813 | * number of unused children, because select_parent can |
814 | * drop the lock and return early due to latency |
815 | * constraints. |
816 | */ |
817 | static int select_parent(struct dentry * parent) |
818 | { |
819 | struct dentry *this_parent = parent; |
820 | struct list_head *next; |
821 | int found = 0; |
822 | |
823 | spin_lock(&dcache_lock); |
824 | repeat: |
825 | next = this_parent->d_subdirs.next; |
826 | resume: |
827 | while (next != &this_parent->d_subdirs) { |
828 | struct list_head *tmp = next; |
829 | struct dentry *dentry = list_entry(tmp, struct dentry, d_u.d_child); |
830 | next = tmp->next; |
831 | |
832 | dentry_lru_del_init(dentry); |
833 | /* |
834 | * move only zero ref count dentries to the end |
835 | * of the unused list for prune_dcache |
836 | */ |
837 | if (!atomic_read(&dentry->d_count)) { |
838 | dentry_lru_add_tail(dentry); |
839 | found++; |
840 | } |
841 | |
842 | /* |
843 | * We can return to the caller if we have found some (this |
844 | * ensures forward progress). We'll be coming back to find |
845 | * the rest. |
846 | */ |
847 | if (found && need_resched()) |
848 | goto out; |
849 | |
850 | /* |
851 | * Descend a level if the d_subdirs list is non-empty. |
852 | */ |
853 | if (!list_empty(&dentry->d_subdirs)) { |
854 | this_parent = dentry; |
855 | goto repeat; |
856 | } |
857 | } |
858 | /* |
859 | * All done at this level ... ascend and resume the search. |
860 | */ |
861 | if (this_parent != parent) { |
862 | next = this_parent->d_u.d_child.next; |
863 | this_parent = this_parent->d_parent; |
864 | goto resume; |
865 | } |
866 | out: |
867 | spin_unlock(&dcache_lock); |
868 | return found; |
869 | } |
870 | |
871 | /** |
872 | * shrink_dcache_parent - prune dcache |
873 | * @parent: parent of entries to prune |
874 | * |
875 | * Prune the dcache to remove unused children of the parent dentry. |
876 | */ |
877 | |
878 | void shrink_dcache_parent(struct dentry * parent) |
879 | { |
880 | struct super_block *sb = parent->d_sb; |
881 | int found; |
882 | |
883 | while ((found = select_parent(parent)) != 0) |
884 | __shrink_dcache_sb(sb, &found, 0); |
885 | } |
886 | EXPORT_SYMBOL(shrink_dcache_parent); |
887 | |
888 | /* |
889 | * Scan `nr' dentries and return the number which remain. |
890 | * |
891 | * We need to avoid reentering the filesystem if the caller is performing a |
892 | * GFP_NOFS allocation attempt. One example deadlock is: |
893 | * |
894 | * ext2_new_block->getblk->GFP->shrink_dcache_memory->prune_dcache-> |
895 | * prune_one_dentry->dput->dentry_iput->iput->inode->i_sb->s_op->put_inode-> |
896 | * ext2_discard_prealloc->ext2_free_blocks->lock_super->DEADLOCK. |
897 | * |
898 | * In this case we return -1 to tell the caller that we baled. |
899 | */ |
900 | static int shrink_dcache_memory(int nr, gfp_t gfp_mask) |
901 | { |
902 | if (nr) { |
903 | if (!(gfp_mask & __GFP_FS)) |
904 | return -1; |
905 | prune_dcache(nr); |
906 | } |
907 | return (dentry_stat.nr_unused / 100) * sysctl_vfs_cache_pressure; |
908 | } |
909 | |
910 | static struct shrinker dcache_shrinker = { |
911 | .shrink = shrink_dcache_memory, |
912 | .seeks = DEFAULT_SEEKS, |
913 | }; |
914 | |
915 | /** |
916 | * d_alloc - allocate a dcache entry |
917 | * @parent: parent of entry to allocate |
918 | * @name: qstr of the name |
919 | * |
920 | * Allocates a dentry. It returns %NULL if there is insufficient memory |
921 | * available. On a success the dentry is returned. The name passed in is |
922 | * copied and the copy passed in may be reused after this call. |
923 | */ |
924 | |
925 | struct dentry *d_alloc(struct dentry * parent, const struct qstr *name) |
926 | { |
927 | struct dentry *dentry; |
928 | char *dname; |
929 | |
930 | dentry = kmem_cache_alloc(dentry_cache, GFP_KERNEL); |
931 | if (!dentry) |
932 | return NULL; |
933 | |
934 | if (name->len > DNAME_INLINE_LEN-1) { |
935 | dname = kmalloc(name->len + 1, GFP_KERNEL); |
936 | if (!dname) { |
937 | kmem_cache_free(dentry_cache, dentry); |
938 | return NULL; |
939 | } |
940 | } else { |
941 | dname = dentry->d_iname; |
942 | } |
943 | dentry->d_name.name = dname; |
944 | |
945 | dentry->d_name.len = name->len; |
946 | dentry->d_name.hash = name->hash; |
947 | memcpy(dname, name->name, name->len); |
948 | dname[name->len] = 0; |
949 | |
950 | atomic_set(&dentry->d_count, 1); |
951 | dentry->d_flags = DCACHE_UNHASHED; |
952 | spin_lock_init(&dentry->d_lock); |
953 | dentry->d_inode = NULL; |
954 | dentry->d_parent = NULL; |
955 | dentry->d_sb = NULL; |
956 | dentry->d_op = NULL; |
957 | dentry->d_fsdata = NULL; |
958 | dentry->d_mounted = 0; |
959 | INIT_HLIST_NODE(&dentry->d_hash); |
960 | INIT_LIST_HEAD(&dentry->d_lru); |
961 | INIT_LIST_HEAD(&dentry->d_subdirs); |
962 | INIT_LIST_HEAD(&dentry->d_alias); |
963 | |
964 | if (parent) { |
965 | dentry->d_parent = dget(parent); |
966 | dentry->d_sb = parent->d_sb; |
967 | } else { |
968 | INIT_LIST_HEAD(&dentry->d_u.d_child); |
969 | } |
970 | |
971 | spin_lock(&dcache_lock); |
972 | if (parent) |
973 | list_add(&dentry->d_u.d_child, &parent->d_subdirs); |
974 | dentry_stat.nr_dentry++; |
975 | spin_unlock(&dcache_lock); |
976 | |
977 | return dentry; |
978 | } |
979 | EXPORT_SYMBOL(d_alloc); |
980 | |
981 | struct dentry *d_alloc_name(struct dentry *parent, const char *name) |
982 | { |
983 | struct qstr q; |
984 | |
985 | q.name = name; |
986 | q.len = strlen(name); |
987 | q.hash = full_name_hash(q.name, q.len); |
988 | return d_alloc(parent, &q); |
989 | } |
990 | EXPORT_SYMBOL(d_alloc_name); |
991 | |
992 | /* the caller must hold dcache_lock */ |
993 | static void __d_instantiate(struct dentry *dentry, struct inode *inode) |
994 | { |
995 | if (inode) |
996 | list_add(&dentry->d_alias, &inode->i_dentry); |
997 | dentry->d_inode = inode; |
998 | fsnotify_d_instantiate(dentry, inode); |
999 | } |
1000 | |
1001 | /** |
1002 | * d_instantiate - fill in inode information for a dentry |
1003 | * @entry: dentry to complete |
1004 | * @inode: inode to attach to this dentry |
1005 | * |
1006 | * Fill in inode information in the entry. |
1007 | * |
1008 | * This turns negative dentries into productive full members |
1009 | * of society. |
1010 | * |
1011 | * NOTE! This assumes that the inode count has been incremented |
1012 | * (or otherwise set) by the caller to indicate that it is now |
1013 | * in use by the dcache. |
1014 | */ |
1015 | |
1016 | void d_instantiate(struct dentry *entry, struct inode * inode) |
1017 | { |
1018 | BUG_ON(!list_empty(&entry->d_alias)); |
1019 | spin_lock(&dcache_lock); |
1020 | __d_instantiate(entry, inode); |
1021 | spin_unlock(&dcache_lock); |
1022 | security_d_instantiate(entry, inode); |
1023 | } |
1024 | EXPORT_SYMBOL(d_instantiate); |
1025 | |
1026 | /** |
1027 | * d_instantiate_unique - instantiate a non-aliased dentry |
1028 | * @entry: dentry to instantiate |
1029 | * @inode: inode to attach to this dentry |
1030 | * |
1031 | * Fill in inode information in the entry. On success, it returns NULL. |
1032 | * If an unhashed alias of "entry" already exists, then we return the |
1033 | * aliased dentry instead and drop one reference to inode. |
1034 | * |
1035 | * Note that in order to avoid conflicts with rename() etc, the caller |
1036 | * had better be holding the parent directory semaphore. |
1037 | * |
1038 | * This also assumes that the inode count has been incremented |
1039 | * (or otherwise set) by the caller to indicate that it is now |
1040 | * in use by the dcache. |
1041 | */ |
1042 | static struct dentry *__d_instantiate_unique(struct dentry *entry, |
1043 | struct inode *inode) |
1044 | { |
1045 | struct dentry *alias; |
1046 | int len = entry->d_name.len; |
1047 | const char *name = entry->d_name.name; |
1048 | unsigned int hash = entry->d_name.hash; |
1049 | |
1050 | if (!inode) { |
1051 | __d_instantiate(entry, NULL); |
1052 | return NULL; |
1053 | } |
1054 | |
1055 | list_for_each_entry(alias, &inode->i_dentry, d_alias) { |
1056 | struct qstr *qstr = &alias->d_name; |
1057 | |
1058 | if (qstr->hash != hash) |
1059 | continue; |
1060 | if (alias->d_parent != entry->d_parent) |
1061 | continue; |
1062 | if (qstr->len != len) |
1063 | continue; |
1064 | if (memcmp(qstr->name, name, len)) |
1065 | continue; |
1066 | dget_locked(alias); |
1067 | return alias; |
1068 | } |
1069 | |
1070 | __d_instantiate(entry, inode); |
1071 | return NULL; |
1072 | } |
1073 | |
1074 | struct dentry *d_instantiate_unique(struct dentry *entry, struct inode *inode) |
1075 | { |
1076 | struct dentry *result; |
1077 | |
1078 | BUG_ON(!list_empty(&entry->d_alias)); |
1079 | |
1080 | spin_lock(&dcache_lock); |
1081 | result = __d_instantiate_unique(entry, inode); |
1082 | spin_unlock(&dcache_lock); |
1083 | |
1084 | if (!result) { |
1085 | security_d_instantiate(entry, inode); |
1086 | return NULL; |
1087 | } |
1088 | |
1089 | BUG_ON(!d_unhashed(result)); |
1090 | iput(inode); |
1091 | return result; |
1092 | } |
1093 | |
1094 | EXPORT_SYMBOL(d_instantiate_unique); |
1095 | |
1096 | /** |
1097 | * d_alloc_root - allocate root dentry |
1098 | * @root_inode: inode to allocate the root for |
1099 | * |
1100 | * Allocate a root ("/") dentry for the inode given. The inode is |
1101 | * instantiated and returned. %NULL is returned if there is insufficient |
1102 | * memory or the inode passed is %NULL. |
1103 | */ |
1104 | |
1105 | struct dentry * d_alloc_root(struct inode * root_inode) |
1106 | { |
1107 | struct dentry *res = NULL; |
1108 | |
1109 | if (root_inode) { |
1110 | static const struct qstr name = { .name = "/", .len = 1 }; |
1111 | |
1112 | res = d_alloc(NULL, &name); |
1113 | if (res) { |
1114 | res->d_sb = root_inode->i_sb; |
1115 | res->d_parent = res; |
1116 | d_instantiate(res, root_inode); |
1117 | } |
1118 | } |
1119 | return res; |
1120 | } |
1121 | EXPORT_SYMBOL(d_alloc_root); |
1122 | |
1123 | static inline struct hlist_head *d_hash(struct dentry *parent, |
1124 | unsigned long hash) |
1125 | { |
1126 | hash += ((unsigned long) parent ^ GOLDEN_RATIO_PRIME) / L1_CACHE_BYTES; |
1127 | hash = hash ^ ((hash ^ GOLDEN_RATIO_PRIME) >> D_HASHBITS); |
1128 | return dentry_hashtable + (hash & D_HASHMASK); |
1129 | } |
1130 | |
1131 | /** |
1132 | * d_obtain_alias - find or allocate a dentry for a given inode |
1133 | * @inode: inode to allocate the dentry for |
1134 | * |
1135 | * Obtain a dentry for an inode resulting from NFS filehandle conversion or |
1136 | * similar open by handle operations. The returned dentry may be anonymous, |
1137 | * or may have a full name (if the inode was already in the cache). |
1138 | * |
1139 | * When called on a directory inode, we must ensure that the inode only ever |
1140 | * has one dentry. If a dentry is found, that is returned instead of |
1141 | * allocating a new one. |
1142 | * |
1143 | * On successful return, the reference to the inode has been transferred |
1144 | * to the dentry. In case of an error the reference on the inode is released. |
1145 | * To make it easier to use in export operations a %NULL or IS_ERR inode may |
1146 | * be passed in and will be the error will be propagate to the return value, |
1147 | * with a %NULL @inode replaced by ERR_PTR(-ESTALE). |
1148 | */ |
1149 | struct dentry *d_obtain_alias(struct inode *inode) |
1150 | { |
1151 | static const struct qstr anonstring = { .name = "" }; |
1152 | struct dentry *tmp; |
1153 | struct dentry *res; |
1154 | |
1155 | if (!inode) |
1156 | return ERR_PTR(-ESTALE); |
1157 | if (IS_ERR(inode)) |
1158 | return ERR_CAST(inode); |
1159 | |
1160 | res = d_find_alias(inode); |
1161 | if (res) |
1162 | goto out_iput; |
1163 | |
1164 | tmp = d_alloc(NULL, &anonstring); |
1165 | if (!tmp) { |
1166 | res = ERR_PTR(-ENOMEM); |
1167 | goto out_iput; |
1168 | } |
1169 | tmp->d_parent = tmp; /* make sure dput doesn't croak */ |
1170 | |
1171 | spin_lock(&dcache_lock); |
1172 | res = __d_find_alias(inode, 0); |
1173 | if (res) { |
1174 | spin_unlock(&dcache_lock); |
1175 | dput(tmp); |
1176 | goto out_iput; |
1177 | } |
1178 | |
1179 | /* attach a disconnected dentry */ |
1180 | spin_lock(&tmp->d_lock); |
1181 | tmp->d_sb = inode->i_sb; |
1182 | tmp->d_inode = inode; |
1183 | tmp->d_flags |= DCACHE_DISCONNECTED; |
1184 | tmp->d_flags &= ~DCACHE_UNHASHED; |
1185 | list_add(&tmp->d_alias, &inode->i_dentry); |
1186 | hlist_add_head(&tmp->d_hash, &inode->i_sb->s_anon); |
1187 | spin_unlock(&tmp->d_lock); |
1188 | |
1189 | spin_unlock(&dcache_lock); |
1190 | return tmp; |
1191 | |
1192 | out_iput: |
1193 | iput(inode); |
1194 | return res; |
1195 | } |
1196 | EXPORT_SYMBOL(d_obtain_alias); |
1197 | |
1198 | /** |
1199 | * d_splice_alias - splice a disconnected dentry into the tree if one exists |
1200 | * @inode: the inode which may have a disconnected dentry |
1201 | * @dentry: a negative dentry which we want to point to the inode. |
1202 | * |
1203 | * If inode is a directory and has a 'disconnected' dentry (i.e. IS_ROOT and |
1204 | * DCACHE_DISCONNECTED), then d_move that in place of the given dentry |
1205 | * and return it, else simply d_add the inode to the dentry and return NULL. |
1206 | * |
1207 | * This is needed in the lookup routine of any filesystem that is exportable |
1208 | * (via knfsd) so that we can build dcache paths to directories effectively. |
1209 | * |
1210 | * If a dentry was found and moved, then it is returned. Otherwise NULL |
1211 | * is returned. This matches the expected return value of ->lookup. |
1212 | * |
1213 | */ |
1214 | struct dentry *d_splice_alias(struct inode *inode, struct dentry *dentry) |
1215 | { |
1216 | struct dentry *new = NULL; |
1217 | |
1218 | if (inode && S_ISDIR(inode->i_mode)) { |
1219 | spin_lock(&dcache_lock); |
1220 | new = __d_find_alias(inode, 1); |
1221 | if (new) { |
1222 | BUG_ON(!(new->d_flags & DCACHE_DISCONNECTED)); |
1223 | spin_unlock(&dcache_lock); |
1224 | security_d_instantiate(new, inode); |
1225 | d_move(new, dentry); |
1226 | iput(inode); |
1227 | } else { |
1228 | /* already taking dcache_lock, so d_add() by hand */ |
1229 | __d_instantiate(dentry, inode); |
1230 | spin_unlock(&dcache_lock); |
1231 | security_d_instantiate(dentry, inode); |
1232 | d_rehash(dentry); |
1233 | } |
1234 | } else |
1235 | d_add(dentry, inode); |
1236 | return new; |
1237 | } |
1238 | EXPORT_SYMBOL(d_splice_alias); |
1239 | |
1240 | /** |
1241 | * d_add_ci - lookup or allocate new dentry with case-exact name |
1242 | * @inode: the inode case-insensitive lookup has found |
1243 | * @dentry: the negative dentry that was passed to the parent's lookup func |
1244 | * @name: the case-exact name to be associated with the returned dentry |
1245 | * |
1246 | * This is to avoid filling the dcache with case-insensitive names to the |
1247 | * same inode, only the actual correct case is stored in the dcache for |
1248 | * case-insensitive filesystems. |
1249 | * |
1250 | * For a case-insensitive lookup match and if the the case-exact dentry |
1251 | * already exists in in the dcache, use it and return it. |
1252 | * |
1253 | * If no entry exists with the exact case name, allocate new dentry with |
1254 | * the exact case, and return the spliced entry. |
1255 | */ |
1256 | struct dentry *d_add_ci(struct dentry *dentry, struct inode *inode, |
1257 | struct qstr *name) |
1258 | { |
1259 | int error; |
1260 | struct dentry *found; |
1261 | struct dentry *new; |
1262 | |
1263 | /* |
1264 | * First check if a dentry matching the name already exists, |
1265 | * if not go ahead and create it now. |
1266 | */ |
1267 | found = d_hash_and_lookup(dentry->d_parent, name); |
1268 | if (!found) { |
1269 | new = d_alloc(dentry->d_parent, name); |
1270 | if (!new) { |
1271 | error = -ENOMEM; |
1272 | goto err_out; |
1273 | } |
1274 | |
1275 | found = d_splice_alias(inode, new); |
1276 | if (found) { |
1277 | dput(new); |
1278 | return found; |
1279 | } |
1280 | return new; |
1281 | } |
1282 | |
1283 | /* |
1284 | * If a matching dentry exists, and it's not negative use it. |
1285 | * |
1286 | * Decrement the reference count to balance the iget() done |
1287 | * earlier on. |
1288 | */ |
1289 | if (found->d_inode) { |
1290 | if (unlikely(found->d_inode != inode)) { |
1291 | /* This can't happen because bad inodes are unhashed. */ |
1292 | BUG_ON(!is_bad_inode(inode)); |
1293 | BUG_ON(!is_bad_inode(found->d_inode)); |
1294 | } |
1295 | iput(inode); |
1296 | return found; |
1297 | } |
1298 | |
1299 | /* |
1300 | * Negative dentry: instantiate it unless the inode is a directory and |
1301 | * already has a dentry. |
1302 | */ |
1303 | spin_lock(&dcache_lock); |
1304 | if (!S_ISDIR(inode->i_mode) || list_empty(&inode->i_dentry)) { |
1305 | __d_instantiate(found, inode); |
1306 | spin_unlock(&dcache_lock); |
1307 | security_d_instantiate(found, inode); |
1308 | return found; |
1309 | } |
1310 | |
1311 | /* |
1312 | * In case a directory already has a (disconnected) entry grab a |
1313 | * reference to it, move it in place and use it. |
1314 | */ |
1315 | new = list_entry(inode->i_dentry.next, struct dentry, d_alias); |
1316 | dget_locked(new); |
1317 | spin_unlock(&dcache_lock); |
1318 | security_d_instantiate(found, inode); |
1319 | d_move(new, found); |
1320 | iput(inode); |
1321 | dput(found); |
1322 | return new; |
1323 | |
1324 | err_out: |
1325 | iput(inode); |
1326 | return ERR_PTR(error); |
1327 | } |
1328 | EXPORT_SYMBOL(d_add_ci); |
1329 | |
1330 | /** |
1331 | * d_lookup - search for a dentry |
1332 | * @parent: parent dentry |
1333 | * @name: qstr of name we wish to find |
1334 | * |
1335 | * Searches the children of the parent dentry for the name in question. If |
1336 | * the dentry is found its reference count is incremented and the dentry |
1337 | * is returned. The caller must use dput to free the entry when it has |
1338 | * finished using it. %NULL is returned on failure. |
1339 | * |
1340 | * __d_lookup is dcache_lock free. The hash list is protected using RCU. |
1341 | * Memory barriers are used while updating and doing lockless traversal. |
1342 | * To avoid races with d_move while rename is happening, d_lock is used. |
1343 | * |
1344 | * Overflows in memcmp(), while d_move, are avoided by keeping the length |
1345 | * and name pointer in one structure pointed by d_qstr. |
1346 | * |
1347 | * rcu_read_lock() and rcu_read_unlock() are used to disable preemption while |
1348 | * lookup is going on. |
1349 | * |
1350 | * The dentry unused LRU is not updated even if lookup finds the required dentry |
1351 | * in there. It is updated in places such as prune_dcache, shrink_dcache_sb, |
1352 | * select_parent and __dget_locked. This laziness saves lookup from dcache_lock |
1353 | * acquisition. |
1354 | * |
1355 | * d_lookup() is protected against the concurrent renames in some unrelated |
1356 | * directory using the seqlockt_t rename_lock. |
1357 | */ |
1358 | |
1359 | struct dentry * d_lookup(struct dentry * parent, struct qstr * name) |
1360 | { |
1361 | struct dentry * dentry = NULL; |
1362 | unsigned long seq; |
1363 | |
1364 | do { |
1365 | seq = read_seqbegin(&rename_lock); |
1366 | dentry = __d_lookup(parent, name); |
1367 | if (dentry) |
1368 | break; |
1369 | } while (read_seqretry(&rename_lock, seq)); |
1370 | return dentry; |
1371 | } |
1372 | EXPORT_SYMBOL(d_lookup); |
1373 | |
1374 | struct dentry * __d_lookup(struct dentry * parent, struct qstr * name) |
1375 | { |
1376 | unsigned int len = name->len; |
1377 | unsigned int hash = name->hash; |
1378 | const unsigned char *str = name->name; |
1379 | struct hlist_head *head = d_hash(parent,hash); |
1380 | struct dentry *found = NULL; |
1381 | struct hlist_node *node; |
1382 | struct dentry *dentry; |
1383 | |
1384 | rcu_read_lock(); |
1385 | |
1386 | hlist_for_each_entry_rcu(dentry, node, head, d_hash) { |
1387 | struct qstr *qstr; |
1388 | |
1389 | if (dentry->d_name.hash != hash) |
1390 | continue; |
1391 | if (dentry->d_parent != parent) |
1392 | continue; |
1393 | |
1394 | spin_lock(&dentry->d_lock); |
1395 | |
1396 | /* |
1397 | * Recheck the dentry after taking the lock - d_move may have |
1398 | * changed things. Don't bother checking the hash because we're |
1399 | * about to compare the whole name anyway. |
1400 | */ |
1401 | if (dentry->d_parent != parent) |
1402 | goto next; |
1403 | |
1404 | /* non-existing due to RCU? */ |
1405 | if (d_unhashed(dentry)) |
1406 | goto next; |
1407 | |
1408 | /* |
1409 | * It is safe to compare names since d_move() cannot |
1410 | * change the qstr (protected by d_lock). |
1411 | */ |
1412 | qstr = &dentry->d_name; |
1413 | if (parent->d_op && parent->d_op->d_compare) { |
1414 | if (parent->d_op->d_compare(parent, qstr, name)) |
1415 | goto next; |
1416 | } else { |
1417 | if (qstr->len != len) |
1418 | goto next; |
1419 | if (memcmp(qstr->name, str, len)) |
1420 | goto next; |
1421 | } |
1422 | |
1423 | atomic_inc(&dentry->d_count); |
1424 | found = dentry; |
1425 | spin_unlock(&dentry->d_lock); |
1426 | break; |
1427 | next: |
1428 | spin_unlock(&dentry->d_lock); |
1429 | } |
1430 | rcu_read_unlock(); |
1431 | |
1432 | return found; |
1433 | } |
1434 | |
1435 | /** |
1436 | * d_hash_and_lookup - hash the qstr then search for a dentry |
1437 | * @dir: Directory to search in |
1438 | * @name: qstr of name we wish to find |
1439 | * |
1440 | * On hash failure or on lookup failure NULL is returned. |
1441 | */ |
1442 | struct dentry *d_hash_and_lookup(struct dentry *dir, struct qstr *name) |
1443 | { |
1444 | struct dentry *dentry = NULL; |
1445 | |
1446 | /* |
1447 | * Check for a fs-specific hash function. Note that we must |
1448 | * calculate the standard hash first, as the d_op->d_hash() |
1449 | * routine may choose to leave the hash value unchanged. |
1450 | */ |
1451 | name->hash = full_name_hash(name->name, name->len); |
1452 | if (dir->d_op && dir->d_op->d_hash) { |
1453 | if (dir->d_op->d_hash(dir, name) < 0) |
1454 | goto out; |
1455 | } |
1456 | dentry = d_lookup(dir, name); |
1457 | out: |
1458 | return dentry; |
1459 | } |
1460 | |
1461 | /** |
1462 | * d_validate - verify dentry provided from insecure source |
1463 | * @dentry: The dentry alleged to be valid child of @dparent |
1464 | * @dparent: The parent dentry (known to be valid) |
1465 | * |
1466 | * An insecure source has sent us a dentry, here we verify it and dget() it. |
1467 | * This is used by ncpfs in its readdir implementation. |
1468 | * Zero is returned in the dentry is invalid. |
1469 | */ |
1470 | |
1471 | int d_validate(struct dentry *dentry, struct dentry *dparent) |
1472 | { |
1473 | struct hlist_head *base; |
1474 | struct hlist_node *lhp; |
1475 | |
1476 | /* Check whether the ptr might be valid at all.. */ |
1477 | if (!kmem_ptr_validate(dentry_cache, dentry)) |
1478 | goto out; |
1479 | |
1480 | if (dentry->d_parent != dparent) |
1481 | goto out; |
1482 | |
1483 | spin_lock(&dcache_lock); |
1484 | base = d_hash(dparent, dentry->d_name.hash); |
1485 | hlist_for_each(lhp,base) { |
1486 | /* hlist_for_each_entry_rcu() not required for d_hash list |
1487 | * as it is parsed under dcache_lock |
1488 | */ |
1489 | if (dentry == hlist_entry(lhp, struct dentry, d_hash)) { |
1490 | __dget_locked(dentry); |
1491 | spin_unlock(&dcache_lock); |
1492 | return 1; |
1493 | } |
1494 | } |
1495 | spin_unlock(&dcache_lock); |
1496 | out: |
1497 | return 0; |
1498 | } |
1499 | EXPORT_SYMBOL(d_validate); |
1500 | |
1501 | /* |
1502 | * When a file is deleted, we have two options: |
1503 | * - turn this dentry into a negative dentry |
1504 | * - unhash this dentry and free it. |
1505 | * |
1506 | * Usually, we want to just turn this into |
1507 | * a negative dentry, but if anybody else is |
1508 | * currently using the dentry or the inode |
1509 | * we can't do that and we fall back on removing |
1510 | * it from the hash queues and waiting for |
1511 | * it to be deleted later when it has no users |
1512 | */ |
1513 | |
1514 | /** |
1515 | * d_delete - delete a dentry |
1516 | * @dentry: The dentry to delete |
1517 | * |
1518 | * Turn the dentry into a negative dentry if possible, otherwise |
1519 | * remove it from the hash queues so it can be deleted later |
1520 | */ |
1521 | |
1522 | void d_delete(struct dentry * dentry) |
1523 | { |
1524 | int isdir = 0; |
1525 | /* |
1526 | * Are we the only user? |
1527 | */ |
1528 | spin_lock(&dcache_lock); |
1529 | spin_lock(&dentry->d_lock); |
1530 | isdir = S_ISDIR(dentry->d_inode->i_mode); |
1531 | if (atomic_read(&dentry->d_count) == 1) { |
1532 | dentry_iput(dentry); |
1533 | fsnotify_nameremove(dentry, isdir); |
1534 | return; |
1535 | } |
1536 | |
1537 | if (!d_unhashed(dentry)) |
1538 | __d_drop(dentry); |
1539 | |
1540 | spin_unlock(&dentry->d_lock); |
1541 | spin_unlock(&dcache_lock); |
1542 | |
1543 | fsnotify_nameremove(dentry, isdir); |
1544 | } |
1545 | EXPORT_SYMBOL(d_delete); |
1546 | |
1547 | static void __d_rehash(struct dentry * entry, struct hlist_head *list) |
1548 | { |
1549 | |
1550 | entry->d_flags &= ~DCACHE_UNHASHED; |
1551 | hlist_add_head_rcu(&entry->d_hash, list); |
1552 | } |
1553 | |
1554 | static void _d_rehash(struct dentry * entry) |
1555 | { |
1556 | __d_rehash(entry, d_hash(entry->d_parent, entry->d_name.hash)); |
1557 | } |
1558 | |
1559 | /** |
1560 | * d_rehash - add an entry back to the hash |
1561 | * @entry: dentry to add to the hash |
1562 | * |
1563 | * Adds a dentry to the hash according to its name. |
1564 | */ |
1565 | |
1566 | void d_rehash(struct dentry * entry) |
1567 | { |
1568 | spin_lock(&dcache_lock); |
1569 | spin_lock(&entry->d_lock); |
1570 | _d_rehash(entry); |
1571 | spin_unlock(&entry->d_lock); |
1572 | spin_unlock(&dcache_lock); |
1573 | } |
1574 | EXPORT_SYMBOL(d_rehash); |
1575 | |
1576 | /* |
1577 | * When switching names, the actual string doesn't strictly have to |
1578 | * be preserved in the target - because we're dropping the target |
1579 | * anyway. As such, we can just do a simple memcpy() to copy over |
1580 | * the new name before we switch. |
1581 | * |
1582 | * Note that we have to be a lot more careful about getting the hash |
1583 | * switched - we have to switch the hash value properly even if it |
1584 | * then no longer matches the actual (corrupted) string of the target. |
1585 | * The hash value has to match the hash queue that the dentry is on.. |
1586 | */ |
1587 | static void switch_names(struct dentry *dentry, struct dentry *target) |
1588 | { |
1589 | if (dname_external(target)) { |
1590 | if (dname_external(dentry)) { |
1591 | /* |
1592 | * Both external: swap the pointers |
1593 | */ |
1594 | swap(target->d_name.name, dentry->d_name.name); |
1595 | } else { |
1596 | /* |
1597 | * dentry:internal, target:external. Steal target's |
1598 | * storage and make target internal. |
1599 | */ |
1600 | memcpy(target->d_iname, dentry->d_name.name, |
1601 | dentry->d_name.len + 1); |
1602 | dentry->d_name.name = target->d_name.name; |
1603 | target->d_name.name = target->d_iname; |
1604 | } |
1605 | } else { |
1606 | if (dname_external(dentry)) { |
1607 | /* |
1608 | * dentry:external, target:internal. Give dentry's |
1609 | * storage to target and make dentry internal |
1610 | */ |
1611 | memcpy(dentry->d_iname, target->d_name.name, |
1612 | target->d_name.len + 1); |
1613 | target->d_name.name = dentry->d_name.name; |
1614 | dentry->d_name.name = dentry->d_iname; |
1615 | } else { |
1616 | /* |
1617 | * Both are internal. Just copy target to dentry |
1618 | */ |
1619 | memcpy(dentry->d_iname, target->d_name.name, |
1620 | target->d_name.len + 1); |
1621 | dentry->d_name.len = target->d_name.len; |
1622 | return; |
1623 | } |
1624 | } |
1625 | swap(dentry->d_name.len, target->d_name.len); |
1626 | } |
1627 | |
1628 | /* |
1629 | * We cannibalize "target" when moving dentry on top of it, |
1630 | * because it's going to be thrown away anyway. We could be more |
1631 | * polite about it, though. |
1632 | * |
1633 | * This forceful removal will result in ugly /proc output if |
1634 | * somebody holds a file open that got deleted due to a rename. |
1635 | * We could be nicer about the deleted file, and let it show |
1636 | * up under the name it had before it was deleted rather than |
1637 | * under the original name of the file that was moved on top of it. |
1638 | */ |
1639 | |
1640 | /* |
1641 | * d_move_locked - move a dentry |
1642 | * @dentry: entry to move |
1643 | * @target: new dentry |
1644 | * |
1645 | * Update the dcache to reflect the move of a file name. Negative |
1646 | * dcache entries should not be moved in this way. |
1647 | */ |
1648 | static void d_move_locked(struct dentry * dentry, struct dentry * target) |
1649 | { |
1650 | struct hlist_head *list; |
1651 | |
1652 | if (!dentry->d_inode) |
1653 | printk(KERN_WARNING "VFS: moving negative dcache entry\n"); |
1654 | |
1655 | write_seqlock(&rename_lock); |
1656 | /* |
1657 | * XXXX: do we really need to take target->d_lock? |
1658 | */ |
1659 | if (target < dentry) { |
1660 | spin_lock(&target->d_lock); |
1661 | spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED); |
1662 | } else { |
1663 | spin_lock(&dentry->d_lock); |
1664 | spin_lock_nested(&target->d_lock, DENTRY_D_LOCK_NESTED); |
1665 | } |
1666 | |
1667 | /* Move the dentry to the target hash queue, if on different bucket */ |
1668 | if (d_unhashed(dentry)) |
1669 | goto already_unhashed; |
1670 | |
1671 | hlist_del_rcu(&dentry->d_hash); |
1672 | |
1673 | already_unhashed: |
1674 | list = d_hash(target->d_parent, target->d_name.hash); |
1675 | __d_rehash(dentry, list); |
1676 | |
1677 | /* Unhash the target: dput() will then get rid of it */ |
1678 | __d_drop(target); |
1679 | |
1680 | list_del(&dentry->d_u.d_child); |
1681 | list_del(&target->d_u.d_child); |
1682 | |
1683 | /* Switch the names.. */ |
1684 | switch_names(dentry, target); |
1685 | swap(dentry->d_name.hash, target->d_name.hash); |
1686 | |
1687 | /* ... and switch the parents */ |
1688 | if (IS_ROOT(dentry)) { |
1689 | dentry->d_parent = target->d_parent; |
1690 | target->d_parent = target; |
1691 | INIT_LIST_HEAD(&target->d_u.d_child); |
1692 | } else { |
1693 | swap(dentry->d_parent, target->d_parent); |
1694 | |
1695 | /* And add them back to the (new) parent lists */ |
1696 | list_add(&target->d_u.d_child, &target->d_parent->d_subdirs); |
1697 | } |
1698 | |
1699 | list_add(&dentry->d_u.d_child, &dentry->d_parent->d_subdirs); |
1700 | spin_unlock(&target->d_lock); |
1701 | fsnotify_d_move(dentry); |
1702 | spin_unlock(&dentry->d_lock); |
1703 | write_sequnlock(&rename_lock); |
1704 | } |
1705 | |
1706 | /** |
1707 | * d_move - move a dentry |
1708 | * @dentry: entry to move |
1709 | * @target: new dentry |
1710 | * |
1711 | * Update the dcache to reflect the move of a file name. Negative |
1712 | * dcache entries should not be moved in this way. |
1713 | */ |
1714 | |
1715 | void d_move(struct dentry * dentry, struct dentry * target) |
1716 | { |
1717 | spin_lock(&dcache_lock); |
1718 | d_move_locked(dentry, target); |
1719 | spin_unlock(&dcache_lock); |
1720 | } |
1721 | EXPORT_SYMBOL(d_move); |
1722 | |
1723 | /** |
1724 | * d_ancestor - search for an ancestor |
1725 | * @p1: ancestor dentry |
1726 | * @p2: child dentry |
1727 | * |
1728 | * Returns the ancestor dentry of p2 which is a child of p1, if p1 is |
1729 | * an ancestor of p2, else NULL. |
1730 | */ |
1731 | struct dentry *d_ancestor(struct dentry *p1, struct dentry *p2) |
1732 | { |
1733 | struct dentry *p; |
1734 | |
1735 | for (p = p2; !IS_ROOT(p); p = p->d_parent) { |
1736 | if (p->d_parent == p1) |
1737 | return p; |
1738 | } |
1739 | return NULL; |
1740 | } |
1741 | |
1742 | /* |
1743 | * This helper attempts to cope with remotely renamed directories |
1744 | * |
1745 | * It assumes that the caller is already holding |
1746 | * dentry->d_parent->d_inode->i_mutex and the dcache_lock |
1747 | * |
1748 | * Note: If ever the locking in lock_rename() changes, then please |
1749 | * remember to update this too... |
1750 | */ |
1751 | static struct dentry *__d_unalias(struct dentry *dentry, struct dentry *alias) |
1752 | __releases(dcache_lock) |
1753 | { |
1754 | struct mutex *m1 = NULL, *m2 = NULL; |
1755 | struct dentry *ret; |
1756 | |
1757 | /* If alias and dentry share a parent, then no extra locks required */ |
1758 | if (alias->d_parent == dentry->d_parent) |
1759 | goto out_unalias; |
1760 | |
1761 | /* Check for loops */ |
1762 | ret = ERR_PTR(-ELOOP); |
1763 | if (d_ancestor(alias, dentry)) |
1764 | goto out_err; |
1765 | |
1766 | /* See lock_rename() */ |
1767 | ret = ERR_PTR(-EBUSY); |
1768 | if (!mutex_trylock(&dentry->d_sb->s_vfs_rename_mutex)) |
1769 | goto out_err; |
1770 | m1 = &dentry->d_sb->s_vfs_rename_mutex; |
1771 | if (!mutex_trylock(&alias->d_parent->d_inode->i_mutex)) |
1772 | goto out_err; |
1773 | m2 = &alias->d_parent->d_inode->i_mutex; |
1774 | out_unalias: |
1775 | d_move_locked(alias, dentry); |
1776 | ret = alias; |
1777 | out_err: |
1778 | spin_unlock(&dcache_lock); |
1779 | if (m2) |
1780 | mutex_unlock(m2); |
1781 | if (m1) |
1782 | mutex_unlock(m1); |
1783 | return ret; |
1784 | } |
1785 | |
1786 | /* |
1787 | * Prepare an anonymous dentry for life in the superblock's dentry tree as a |
1788 | * named dentry in place of the dentry to be replaced. |
1789 | */ |
1790 | static void __d_materialise_dentry(struct dentry *dentry, struct dentry *anon) |
1791 | { |
1792 | struct dentry *dparent, *aparent; |
1793 | |
1794 | switch_names(dentry, anon); |
1795 | swap(dentry->d_name.hash, anon->d_name.hash); |
1796 | |
1797 | dparent = dentry->d_parent; |
1798 | aparent = anon->d_parent; |
1799 | |
1800 | dentry->d_parent = (aparent == anon) ? dentry : aparent; |
1801 | list_del(&dentry->d_u.d_child); |
1802 | if (!IS_ROOT(dentry)) |
1803 | list_add(&dentry->d_u.d_child, &dentry->d_parent->d_subdirs); |
1804 | else |
1805 | INIT_LIST_HEAD(&dentry->d_u.d_child); |
1806 | |
1807 | anon->d_parent = (dparent == dentry) ? anon : dparent; |
1808 | list_del(&anon->d_u.d_child); |
1809 | if (!IS_ROOT(anon)) |
1810 | list_add(&anon->d_u.d_child, &anon->d_parent->d_subdirs); |
1811 | else |
1812 | INIT_LIST_HEAD(&anon->d_u.d_child); |
1813 | |
1814 | anon->d_flags &= ~DCACHE_DISCONNECTED; |
1815 | } |
1816 | |
1817 | /** |
1818 | * d_materialise_unique - introduce an inode into the tree |
1819 | * @dentry: candidate dentry |
1820 | * @inode: inode to bind to the dentry, to which aliases may be attached |
1821 | * |
1822 | * Introduces an dentry into the tree, substituting an extant disconnected |
1823 | * root directory alias in its place if there is one |
1824 | */ |
1825 | struct dentry *d_materialise_unique(struct dentry *dentry, struct inode *inode) |
1826 | { |
1827 | struct dentry *actual; |
1828 | |
1829 | BUG_ON(!d_unhashed(dentry)); |
1830 | |
1831 | spin_lock(&dcache_lock); |
1832 | |
1833 | if (!inode) { |
1834 | actual = dentry; |
1835 | __d_instantiate(dentry, NULL); |
1836 | goto found_lock; |
1837 | } |
1838 | |
1839 | if (S_ISDIR(inode->i_mode)) { |
1840 | struct dentry *alias; |
1841 | |
1842 | /* Does an aliased dentry already exist? */ |
1843 | alias = __d_find_alias(inode, 0); |
1844 | if (alias) { |
1845 | actual = alias; |
1846 | /* Is this an anonymous mountpoint that we could splice |
1847 | * into our tree? */ |
1848 | if (IS_ROOT(alias)) { |
1849 | spin_lock(&alias->d_lock); |
1850 | __d_materialise_dentry(dentry, alias); |
1851 | __d_drop(alias); |
1852 | goto found; |
1853 | } |
1854 | /* Nope, but we must(!) avoid directory aliasing */ |
1855 | actual = __d_unalias(dentry, alias); |
1856 | if (IS_ERR(actual)) |
1857 | dput(alias); |
1858 | goto out_nolock; |
1859 | } |
1860 | } |
1861 | |
1862 | /* Add a unique reference */ |
1863 | actual = __d_instantiate_unique(dentry, inode); |
1864 | if (!actual) |
1865 | actual = dentry; |
1866 | else if (unlikely(!d_unhashed(actual))) |
1867 | goto shouldnt_be_hashed; |
1868 | |
1869 | found_lock: |
1870 | spin_lock(&actual->d_lock); |
1871 | found: |
1872 | _d_rehash(actual); |
1873 | spin_unlock(&actual->d_lock); |
1874 | spin_unlock(&dcache_lock); |
1875 | out_nolock: |
1876 | if (actual == dentry) { |
1877 | security_d_instantiate(dentry, inode); |
1878 | return NULL; |
1879 | } |
1880 | |
1881 | iput(inode); |
1882 | return actual; |
1883 | |
1884 | shouldnt_be_hashed: |
1885 | spin_unlock(&dcache_lock); |
1886 | BUG(); |
1887 | } |
1888 | EXPORT_SYMBOL_GPL(d_materialise_unique); |
1889 | |
1890 | static int prepend(char **buffer, int *buflen, const char *str, int namelen) |
1891 | { |
1892 | *buflen -= namelen; |
1893 | if (*buflen < 0) |
1894 | return -ENAMETOOLONG; |
1895 | *buffer -= namelen; |
1896 | memcpy(*buffer, str, namelen); |
1897 | return 0; |
1898 | } |
1899 | |
1900 | static int prepend_name(char **buffer, int *buflen, struct qstr *name) |
1901 | { |
1902 | return prepend(buffer, buflen, name->name, name->len); |
1903 | } |
1904 | |
1905 | /** |
1906 | * __d_path - return the path of a dentry |
1907 | * @path: the dentry/vfsmount to report |
1908 | * @root: root vfsmnt/dentry (may be modified by this function) |
1909 | * @buffer: buffer to return value in |
1910 | * @buflen: buffer length |
1911 | * |
1912 | * Convert a dentry into an ASCII path name. If the entry has been deleted |
1913 | * the string " (deleted)" is appended. Note that this is ambiguous. |
1914 | * |
1915 | * Returns a pointer into the buffer or an error code if the |
1916 | * path was too long. |
1917 | * |
1918 | * "buflen" should be positive. Caller holds the dcache_lock. |
1919 | * |
1920 | * If path is not reachable from the supplied root, then the value of |
1921 | * root is changed (without modifying refcounts). |
1922 | */ |
1923 | char *__d_path(const struct path *path, struct path *root, |
1924 | char *buffer, int buflen) |
1925 | { |
1926 | struct dentry *dentry = path->dentry; |
1927 | struct vfsmount *vfsmnt = path->mnt; |
1928 | char *end = buffer + buflen; |
1929 | char *retval; |
1930 | |
1931 | spin_lock(&vfsmount_lock); |
1932 | prepend(&end, &buflen, "\0", 1); |
1933 | if (d_unlinked(dentry) && |
1934 | (prepend(&end, &buflen, " (deleted)", 10) != 0)) |
1935 | goto Elong; |
1936 | |
1937 | if (buflen < 1) |
1938 | goto Elong; |
1939 | /* Get '/' right */ |
1940 | retval = end-1; |
1941 | *retval = '/'; |
1942 | |
1943 | for (;;) { |
1944 | struct dentry * parent; |
1945 | |
1946 | if (dentry == root->dentry && vfsmnt == root->mnt) |
1947 | break; |
1948 | if (dentry == vfsmnt->mnt_root || IS_ROOT(dentry)) { |
1949 | /* Global root? */ |
1950 | if (vfsmnt->mnt_parent == vfsmnt) { |
1951 | goto global_root; |
1952 | } |
1953 | dentry = vfsmnt->mnt_mountpoint; |
1954 | vfsmnt = vfsmnt->mnt_parent; |
1955 | continue; |
1956 | } |
1957 | parent = dentry->d_parent; |
1958 | prefetch(parent); |
1959 | if ((prepend_name(&end, &buflen, &dentry->d_name) != 0) || |
1960 | (prepend(&end, &buflen, "/", 1) != 0)) |
1961 | goto Elong; |
1962 | retval = end; |
1963 | dentry = parent; |
1964 | } |
1965 | |
1966 | out: |
1967 | spin_unlock(&vfsmount_lock); |
1968 | return retval; |
1969 | |
1970 | global_root: |
1971 | retval += 1; /* hit the slash */ |
1972 | if (prepend_name(&retval, &buflen, &dentry->d_name) != 0) |
1973 | goto Elong; |
1974 | root->mnt = vfsmnt; |
1975 | root->dentry = dentry; |
1976 | goto out; |
1977 | |
1978 | Elong: |
1979 | retval = ERR_PTR(-ENAMETOOLONG); |
1980 | goto out; |
1981 | } |
1982 | |
1983 | /** |
1984 | * d_path - return the path of a dentry |
1985 | * @path: path to report |
1986 | * @buf: buffer to return value in |
1987 | * @buflen: buffer length |
1988 | * |
1989 | * Convert a dentry into an ASCII path name. If the entry has been deleted |
1990 | * the string " (deleted)" is appended. Note that this is ambiguous. |
1991 | * |
1992 | * Returns a pointer into the buffer or an error code if the path was |
1993 | * too long. Note: Callers should use the returned pointer, not the passed |
1994 | * in buffer, to use the name! The implementation often starts at an offset |
1995 | * into the buffer, and may leave 0 bytes at the start. |
1996 | * |
1997 | * "buflen" should be positive. |
1998 | */ |
1999 | char *d_path(const struct path *path, char *buf, int buflen) |
2000 | { |
2001 | char *res; |
2002 | struct path root; |
2003 | struct path tmp; |
2004 | |
2005 | /* |
2006 | * We have various synthetic filesystems that never get mounted. On |
2007 | * these filesystems dentries are never used for lookup purposes, and |
2008 | * thus don't need to be hashed. They also don't need a name until a |
2009 | * user wants to identify the object in /proc/pid/fd/. The little hack |
2010 | * below allows us to generate a name for these objects on demand: |
2011 | */ |
2012 | if (path->dentry->d_op && path->dentry->d_op->d_dname) |
2013 | return path->dentry->d_op->d_dname(path->dentry, buf, buflen); |
2014 | |
2015 | read_lock(¤t->fs->lock); |
2016 | root = current->fs->root; |
2017 | path_get(&root); |
2018 | read_unlock(¤t->fs->lock); |
2019 | spin_lock(&dcache_lock); |
2020 | tmp = root; |
2021 | res = __d_path(path, &tmp, buf, buflen); |
2022 | spin_unlock(&dcache_lock); |
2023 | path_put(&root); |
2024 | return res; |
2025 | } |
2026 | EXPORT_SYMBOL(d_path); |
2027 | |
2028 | /* |
2029 | * Helper function for dentry_operations.d_dname() members |
2030 | */ |
2031 | char *dynamic_dname(struct dentry *dentry, char *buffer, int buflen, |
2032 | const char *fmt, ...) |
2033 | { |
2034 | va_list args; |
2035 | char temp[64]; |
2036 | int sz; |
2037 | |
2038 | va_start(args, fmt); |
2039 | sz = vsnprintf(temp, sizeof(temp), fmt, args) + 1; |
2040 | va_end(args); |
2041 | |
2042 | if (sz > sizeof(temp) || sz > buflen) |
2043 | return ERR_PTR(-ENAMETOOLONG); |
2044 | |
2045 | buffer += buflen - sz; |
2046 | return memcpy(buffer, temp, sz); |
2047 | } |
2048 | |
2049 | /* |
2050 | * Write full pathname from the root of the filesystem into the buffer. |
2051 | */ |
2052 | char *dentry_path(struct dentry *dentry, char *buf, int buflen) |
2053 | { |
2054 | char *end = buf + buflen; |
2055 | char *retval; |
2056 | |
2057 | spin_lock(&dcache_lock); |
2058 | prepend(&end, &buflen, "\0", 1); |
2059 | if (d_unlinked(dentry) && |
2060 | (prepend(&end, &buflen, "//deleted", 9) != 0)) |
2061 | goto Elong; |
2062 | if (buflen < 1) |
2063 | goto Elong; |
2064 | /* Get '/' right */ |
2065 | retval = end-1; |
2066 | *retval = '/'; |
2067 | |
2068 | while (!IS_ROOT(dentry)) { |
2069 | struct dentry *parent = dentry->d_parent; |
2070 | |
2071 | prefetch(parent); |
2072 | if ((prepend_name(&end, &buflen, &dentry->d_name) != 0) || |
2073 | (prepend(&end, &buflen, "/", 1) != 0)) |
2074 | goto Elong; |
2075 | |
2076 | retval = end; |
2077 | dentry = parent; |
2078 | } |
2079 | spin_unlock(&dcache_lock); |
2080 | return retval; |
2081 | Elong: |
2082 | spin_unlock(&dcache_lock); |
2083 | return ERR_PTR(-ENAMETOOLONG); |
2084 | } |
2085 | |
2086 | /* |
2087 | * NOTE! The user-level library version returns a |
2088 | * character pointer. The kernel system call just |
2089 | * returns the length of the buffer filled (which |
2090 | * includes the ending '\0' character), or a negative |
2091 | * error value. So libc would do something like |
2092 | * |
2093 | * char *getcwd(char * buf, size_t size) |
2094 | * { |
2095 | * int retval; |
2096 | * |
2097 | * retval = sys_getcwd(buf, size); |
2098 | * if (retval >= 0) |
2099 | * return buf; |
2100 | * errno = -retval; |
2101 | * return NULL; |
2102 | * } |
2103 | */ |
2104 | SYSCALL_DEFINE2(getcwd, char __user *, buf, unsigned long, size) |
2105 | { |
2106 | int error; |
2107 | struct path pwd, root; |
2108 | char *page = (char *) __get_free_page(GFP_USER); |
2109 | |
2110 | if (!page) |
2111 | return -ENOMEM; |
2112 | |
2113 | read_lock(¤t->fs->lock); |
2114 | pwd = current->fs->pwd; |
2115 | path_get(&pwd); |
2116 | root = current->fs->root; |
2117 | path_get(&root); |
2118 | read_unlock(¤t->fs->lock); |
2119 | |
2120 | error = -ENOENT; |
2121 | spin_lock(&dcache_lock); |
2122 | if (!d_unlinked(pwd.dentry)) { |
2123 | unsigned long len; |
2124 | struct path tmp = root; |
2125 | char * cwd; |
2126 | |
2127 | cwd = __d_path(&pwd, &tmp, page, PAGE_SIZE); |
2128 | spin_unlock(&dcache_lock); |
2129 | |
2130 | error = PTR_ERR(cwd); |
2131 | if (IS_ERR(cwd)) |
2132 | goto out; |
2133 | |
2134 | error = -ERANGE; |
2135 | len = PAGE_SIZE + page - cwd; |
2136 | if (len <= size) { |
2137 | error = len; |
2138 | if (copy_to_user(buf, cwd, len)) |
2139 | error = -EFAULT; |
2140 | } |
2141 | } else |
2142 | spin_unlock(&dcache_lock); |
2143 | |
2144 | out: |
2145 | path_put(&pwd); |
2146 | path_put(&root); |
2147 | free_page((unsigned long) page); |
2148 | return error; |
2149 | } |
2150 | |
2151 | /* |
2152 | * Test whether new_dentry is a subdirectory of old_dentry. |
2153 | * |
2154 | * Trivially implemented using the dcache structure |
2155 | */ |
2156 | |
2157 | /** |
2158 | * is_subdir - is new dentry a subdirectory of old_dentry |
2159 | * @new_dentry: new dentry |
2160 | * @old_dentry: old dentry |
2161 | * |
2162 | * Returns 1 if new_dentry is a subdirectory of the parent (at any depth). |
2163 | * Returns 0 otherwise. |
2164 | * Caller must ensure that "new_dentry" is pinned before calling is_subdir() |
2165 | */ |
2166 | |
2167 | int is_subdir(struct dentry *new_dentry, struct dentry *old_dentry) |
2168 | { |
2169 | int result; |
2170 | unsigned long seq; |
2171 | |
2172 | if (new_dentry == old_dentry) |
2173 | return 1; |
2174 | |
2175 | /* |
2176 | * Need rcu_readlock to protect against the d_parent trashing |
2177 | * due to d_move |
2178 | */ |
2179 | rcu_read_lock(); |
2180 | do { |
2181 | /* for restarting inner loop in case of seq retry */ |
2182 | seq = read_seqbegin(&rename_lock); |
2183 | if (d_ancestor(old_dentry, new_dentry)) |
2184 | result = 1; |
2185 | else |
2186 | result = 0; |
2187 | } while (read_seqretry(&rename_lock, seq)); |
2188 | rcu_read_unlock(); |
2189 | |
2190 | return result; |
2191 | } |
2192 | |
2193 | int path_is_under(struct path *path1, struct path *path2) |
2194 | { |
2195 | struct vfsmount *mnt = path1->mnt; |
2196 | struct dentry *dentry = path1->dentry; |
2197 | int res; |
2198 | spin_lock(&vfsmount_lock); |
2199 | if (mnt != path2->mnt) { |
2200 | for (;;) { |
2201 | if (mnt->mnt_parent == mnt) { |
2202 | spin_unlock(&vfsmount_lock); |
2203 | return 0; |
2204 | } |
2205 | if (mnt->mnt_parent == path2->mnt) |
2206 | break; |
2207 | mnt = mnt->mnt_parent; |
2208 | } |
2209 | dentry = mnt->mnt_mountpoint; |
2210 | } |
2211 | res = is_subdir(dentry, path2->dentry); |
2212 | spin_unlock(&vfsmount_lock); |
2213 | return res; |
2214 | } |
2215 | EXPORT_SYMBOL(path_is_under); |
2216 | |
2217 | void d_genocide(struct dentry *root) |
2218 | { |
2219 | struct dentry *this_parent = root; |
2220 | struct list_head *next; |
2221 | |
2222 | spin_lock(&dcache_lock); |
2223 | repeat: |
2224 | next = this_parent->d_subdirs.next; |
2225 | resume: |
2226 | while (next != &this_parent->d_subdirs) { |
2227 | struct list_head *tmp = next; |
2228 | struct dentry *dentry = list_entry(tmp, struct dentry, d_u.d_child); |
2229 | next = tmp->next; |
2230 | if (d_unhashed(dentry)||!dentry->d_inode) |
2231 | continue; |
2232 | if (!list_empty(&dentry->d_subdirs)) { |
2233 | this_parent = dentry; |
2234 | goto repeat; |
2235 | } |
2236 | atomic_dec(&dentry->d_count); |
2237 | } |
2238 | if (this_parent != root) { |
2239 | next = this_parent->d_u.d_child.next; |
2240 | atomic_dec(&this_parent->d_count); |
2241 | this_parent = this_parent->d_parent; |
2242 | goto resume; |
2243 | } |
2244 | spin_unlock(&dcache_lock); |
2245 | } |
2246 | |
2247 | /** |
2248 | * find_inode_number - check for dentry with name |
2249 | * @dir: directory to check |
2250 | * @name: Name to find. |
2251 | * |
2252 | * Check whether a dentry already exists for the given name, |
2253 | * and return the inode number if it has an inode. Otherwise |
2254 | * 0 is returned. |
2255 | * |
2256 | * This routine is used to post-process directory listings for |
2257 | * filesystems using synthetic inode numbers, and is necessary |
2258 | * to keep getcwd() working. |
2259 | */ |
2260 | |
2261 | ino_t find_inode_number(struct dentry *dir, struct qstr *name) |
2262 | { |
2263 | struct dentry * dentry; |
2264 | ino_t ino = 0; |
2265 | |
2266 | dentry = d_hash_and_lookup(dir, name); |
2267 | if (dentry) { |
2268 | if (dentry->d_inode) |
2269 | ino = dentry->d_inode->i_ino; |
2270 | dput(dentry); |
2271 | } |
2272 | return ino; |
2273 | } |
2274 | EXPORT_SYMBOL(find_inode_number); |
2275 | |
2276 | static __initdata unsigned long dhash_entries; |
2277 | static int __init set_dhash_entries(char *str) |
2278 | { |
2279 | if (!str) |
2280 | return 0; |
2281 | dhash_entries = simple_strtoul(str, &str, 0); |
2282 | return 1; |
2283 | } |
2284 | __setup("dhash_entries=", set_dhash_entries); |
2285 | |
2286 | static void __init dcache_init_early(void) |
2287 | { |
2288 | int loop; |
2289 | |
2290 | /* If hashes are distributed across NUMA nodes, defer |
2291 | * hash allocation until vmalloc space is available. |
2292 | */ |
2293 | if (hashdist) |
2294 | return; |
2295 | |
2296 | dentry_hashtable = |
2297 | alloc_large_system_hash("Dentry cache", |
2298 | sizeof(struct hlist_head), |
2299 | dhash_entries, |
2300 | 13, |
2301 | HASH_EARLY, |
2302 | &d_hash_shift, |
2303 | &d_hash_mask, |
2304 | 0); |
2305 | |
2306 | for (loop = 0; loop < (1 << d_hash_shift); loop++) |
2307 | INIT_HLIST_HEAD(&dentry_hashtable[loop]); |
2308 | } |
2309 | |
2310 | static void __init dcache_init(void) |
2311 | { |
2312 | int loop; |
2313 | |
2314 | /* |
2315 | * A constructor could be added for stable state like the lists, |
2316 | * but it is probably not worth it because of the cache nature |
2317 | * of the dcache. |
2318 | */ |
2319 | dentry_cache = KMEM_CACHE(dentry, |
2320 | SLAB_RECLAIM_ACCOUNT|SLAB_PANIC|SLAB_MEM_SPREAD); |
2321 | |
2322 | register_shrinker(&dcache_shrinker); |
2323 | |
2324 | /* Hash may have been set up in dcache_init_early */ |
2325 | if (!hashdist) |
2326 | return; |
2327 | |
2328 | dentry_hashtable = |
2329 | alloc_large_system_hash("Dentry cache", |
2330 | sizeof(struct hlist_head), |
2331 | dhash_entries, |
2332 | 13, |
2333 | 0, |
2334 | &d_hash_shift, |
2335 | &d_hash_mask, |
2336 | 0); |
2337 | |
2338 | for (loop = 0; loop < (1 << d_hash_shift); loop++) |
2339 | INIT_HLIST_HEAD(&dentry_hashtable[loop]); |
2340 | } |
2341 | |
2342 | /* SLAB cache for __getname() consumers */ |
2343 | struct kmem_cache *names_cachep __read_mostly; |
2344 | EXPORT_SYMBOL(names_cachep); |
2345 | |
2346 | EXPORT_SYMBOL(d_genocide); |
2347 | |
2348 | void __init vfs_caches_init_early(void) |
2349 | { |
2350 | dcache_init_early(); |
2351 | inode_init_early(); |
2352 | } |
2353 | |
2354 | void __init vfs_caches_init(unsigned long mempages) |
2355 | { |
2356 | unsigned long reserve; |
2357 | |
2358 | /* Base hash sizes on available memory, with a reserve equal to |
2359 | 150% of current kernel size */ |
2360 | |
2361 | reserve = min((mempages - nr_free_pages()) * 3/2, mempages - 1); |
2362 | mempages -= reserve; |
2363 | |
2364 | names_cachep = kmem_cache_create("names_cache", PATH_MAX, 0, |
2365 | SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); |
2366 | |
2367 | dcache_init(); |
2368 | inode_init(); |
2369 | files_init(mempages); |
2370 | mnt_init(); |
2371 | bdev_cache_init(); |
2372 | chrdev_init(); |
2373 | } |
2374 |
Branches:
ben-wpan
ben-wpan-stefan
javiroman/ks7010
jz-2.6.34
jz-2.6.34-rc5
jz-2.6.34-rc6
jz-2.6.34-rc7
jz-2.6.35
jz-2.6.36
jz-2.6.37
jz-2.6.38
jz-2.6.39
jz-3.0
jz-3.1
jz-3.11
jz-3.12
jz-3.13
jz-3.15
jz-3.16
jz-3.18-dt
jz-3.2
jz-3.3
jz-3.4
jz-3.5
jz-3.6
jz-3.6-rc2-pwm
jz-3.9
jz-3.9-clk
jz-3.9-rc8
jz47xx
jz47xx-2.6.38
master
Tags:
od-2011-09-04
od-2011-09-18
v2.6.34-rc5
v2.6.34-rc6
v2.6.34-rc7
v3.9