xref: /linux/fs/overlayfs/readdir.c (revision 1fb8510cdb5b7befe8a59f533c7fc12ef0dac73e)
1 /*
2  *
3  * Copyright (C) 2011 Novell Inc.
4  *
5  * This program is free software; you can redistribute it and/or modify it
6  * under the terms of the GNU General Public License version 2 as published by
7  * the Free Software Foundation.
8  */
9 
10 #include <linux/fs.h>
11 #include <linux/slab.h>
12 #include <linux/namei.h>
13 #include <linux/file.h>
14 #include <linux/xattr.h>
15 #include <linux/rbtree.h>
16 #include <linux/security.h>
17 #include <linux/cred.h>
18 #include "overlayfs.h"
19 
20 struct ovl_cache_entry {
21 	unsigned int len;
22 	unsigned int type;
23 	u64 ino;
24 	bool is_whiteout;
25 	struct list_head l_node;
26 	struct rb_node node;
27 	char name[];
28 };
29 
30 struct ovl_dir_cache {
31 	long refcount;
32 	u64 version;
33 	struct list_head entries;
34 };
35 
36 struct ovl_readdir_data {
37 	struct dir_context ctx;
38 	bool is_merge;
39 	struct rb_root root;
40 	struct list_head *list;
41 	struct list_head middle;
42 	int count;
43 	int err;
44 };
45 
46 struct ovl_dir_file {
47 	bool is_real;
48 	bool is_upper;
49 	struct ovl_dir_cache *cache;
50 	struct ovl_cache_entry cursor;
51 	struct file *realfile;
52 	struct file *upperfile;
53 };
54 
55 static struct ovl_cache_entry *ovl_cache_entry_from_node(struct rb_node *n)
56 {
57 	return container_of(n, struct ovl_cache_entry, node);
58 }
59 
60 static struct ovl_cache_entry *ovl_cache_entry_find(struct rb_root *root,
61 						    const char *name, int len)
62 {
63 	struct rb_node *node = root->rb_node;
64 	int cmp;
65 
66 	while (node) {
67 		struct ovl_cache_entry *p = ovl_cache_entry_from_node(node);
68 
69 		cmp = strncmp(name, p->name, len);
70 		if (cmp > 0)
71 			node = p->node.rb_right;
72 		else if (cmp < 0 || len < p->len)
73 			node = p->node.rb_left;
74 		else
75 			return p;
76 	}
77 
78 	return NULL;
79 }
80 
81 static struct ovl_cache_entry *ovl_cache_entry_new(const char *name, int len,
82 						   u64 ino, unsigned int d_type)
83 {
84 	struct ovl_cache_entry *p;
85 	size_t size = offsetof(struct ovl_cache_entry, name[len + 1]);
86 
87 	p = kmalloc(size, GFP_KERNEL);
88 	if (p) {
89 		memcpy(p->name, name, len);
90 		p->name[len] = '\0';
91 		p->len = len;
92 		p->type = d_type;
93 		p->ino = ino;
94 		p->is_whiteout = false;
95 	}
96 
97 	return p;
98 }
99 
100 static int ovl_cache_entry_add_rb(struct ovl_readdir_data *rdd,
101 				  const char *name, int len, u64 ino,
102 				  unsigned int d_type)
103 {
104 	struct rb_node **newp = &rdd->root.rb_node;
105 	struct rb_node *parent = NULL;
106 	struct ovl_cache_entry *p;
107 
108 	while (*newp) {
109 		int cmp;
110 		struct ovl_cache_entry *tmp;
111 
112 		parent = *newp;
113 		tmp = ovl_cache_entry_from_node(*newp);
114 		cmp = strncmp(name, tmp->name, len);
115 		if (cmp > 0)
116 			newp = &tmp->node.rb_right;
117 		else if (cmp < 0 || len < tmp->len)
118 			newp = &tmp->node.rb_left;
119 		else
120 			return 0;
121 	}
122 
123 	p = ovl_cache_entry_new(name, len, ino, d_type);
124 	if (p == NULL)
125 		return -ENOMEM;
126 
127 	list_add_tail(&p->l_node, rdd->list);
128 	rb_link_node(&p->node, parent, newp);
129 	rb_insert_color(&p->node, &rdd->root);
130 
131 	return 0;
132 }
133 
134 static int ovl_fill_lower(struct ovl_readdir_data *rdd,
135 			  const char *name, int namelen,
136 			  loff_t offset, u64 ino, unsigned int d_type)
137 {
138 	struct ovl_cache_entry *p;
139 
140 	p = ovl_cache_entry_find(&rdd->root, name, namelen);
141 	if (p) {
142 		list_move_tail(&p->l_node, &rdd->middle);
143 	} else {
144 		p = ovl_cache_entry_new(name, namelen, ino, d_type);
145 		if (p == NULL)
146 			rdd->err = -ENOMEM;
147 		else
148 			list_add_tail(&p->l_node, &rdd->middle);
149 	}
150 
151 	return rdd->err;
152 }
153 
154 void ovl_cache_free(struct list_head *list)
155 {
156 	struct ovl_cache_entry *p;
157 	struct ovl_cache_entry *n;
158 
159 	list_for_each_entry_safe(p, n, list, l_node)
160 		kfree(p);
161 
162 	INIT_LIST_HEAD(list);
163 }
164 
165 static void ovl_cache_put(struct ovl_dir_file *od, struct dentry *dentry)
166 {
167 	struct ovl_dir_cache *cache = od->cache;
168 
169 	list_del(&od->cursor.l_node);
170 	WARN_ON(cache->refcount <= 0);
171 	cache->refcount--;
172 	if (!cache->refcount) {
173 		if (ovl_dir_cache(dentry) == cache)
174 			ovl_set_dir_cache(dentry, NULL);
175 
176 		ovl_cache_free(&cache->entries);
177 		kfree(cache);
178 	}
179 }
180 
181 static int ovl_fill_merge(void *buf, const char *name, int namelen,
182 			  loff_t offset, u64 ino, unsigned int d_type)
183 {
184 	struct ovl_readdir_data *rdd = buf;
185 
186 	rdd->count++;
187 	if (!rdd->is_merge)
188 		return ovl_cache_entry_add_rb(rdd, name, namelen, ino, d_type);
189 	else
190 		return ovl_fill_lower(rdd, name, namelen, offset, ino, d_type);
191 }
192 
193 static inline int ovl_dir_read(struct path *realpath,
194 			       struct ovl_readdir_data *rdd)
195 {
196 	struct file *realfile;
197 	int err;
198 
199 	realfile = ovl_path_open(realpath, O_RDONLY | O_DIRECTORY);
200 	if (IS_ERR(realfile))
201 		return PTR_ERR(realfile);
202 
203 	rdd->ctx.pos = 0;
204 	do {
205 		rdd->count = 0;
206 		rdd->err = 0;
207 		err = iterate_dir(realfile, &rdd->ctx);
208 		if (err >= 0)
209 			err = rdd->err;
210 	} while (!err && rdd->count);
211 	fput(realfile);
212 
213 	return err;
214 }
215 
216 static void ovl_dir_reset(struct file *file)
217 {
218 	struct ovl_dir_file *od = file->private_data;
219 	struct ovl_dir_cache *cache = od->cache;
220 	struct dentry *dentry = file->f_path.dentry;
221 	enum ovl_path_type type = ovl_path_type(dentry);
222 
223 	if (cache && ovl_dentry_version_get(dentry) != cache->version) {
224 		ovl_cache_put(od, dentry);
225 		od->cache = NULL;
226 	}
227 	WARN_ON(!od->is_real && type != OVL_PATH_MERGE);
228 	if (od->is_real && type == OVL_PATH_MERGE)
229 		od->is_real = false;
230 }
231 
232 static int ovl_dir_mark_whiteouts(struct dentry *dir,
233 				  struct ovl_readdir_data *rdd)
234 {
235 	struct ovl_cache_entry *p;
236 	struct dentry *dentry;
237 	const struct cred *old_cred;
238 	struct cred *override_cred;
239 
240 	override_cred = prepare_creds();
241 	if (!override_cred) {
242 		ovl_cache_free(rdd->list);
243 		return -ENOMEM;
244 	}
245 
246 	/*
247 	 * CAP_DAC_OVERRIDE for lookup
248 	 */
249 	cap_raise(override_cred->cap_effective, CAP_DAC_OVERRIDE);
250 	old_cred = override_creds(override_cred);
251 
252 	mutex_lock(&dir->d_inode->i_mutex);
253 	list_for_each_entry(p, rdd->list, l_node) {
254 		if (!p->name)
255 			continue;
256 
257 		if (p->type != DT_CHR)
258 			continue;
259 
260 		dentry = lookup_one_len(p->name, dir, p->len);
261 		if (IS_ERR(dentry))
262 			continue;
263 
264 		p->is_whiteout = ovl_is_whiteout(dentry);
265 		dput(dentry);
266 	}
267 	mutex_unlock(&dir->d_inode->i_mutex);
268 
269 	revert_creds(old_cred);
270 	put_cred(override_cred);
271 
272 	return 0;
273 }
274 
275 static inline int ovl_dir_read_merged(struct path *upperpath,
276 				      struct path *lowerpath,
277 				      struct list_head *list)
278 {
279 	int err;
280 	struct ovl_readdir_data rdd = {
281 		.ctx.actor = ovl_fill_merge,
282 		.list = list,
283 		.root = RB_ROOT,
284 		.is_merge = false,
285 	};
286 
287 	if (upperpath->dentry) {
288 		err = ovl_dir_read(upperpath, &rdd);
289 		if (err)
290 			goto out;
291 
292 		if (lowerpath->dentry) {
293 			err = ovl_dir_mark_whiteouts(upperpath->dentry, &rdd);
294 			if (err)
295 				goto out;
296 		}
297 	}
298 	if (lowerpath->dentry) {
299 		/*
300 		 * Insert lowerpath entries before upperpath ones, this allows
301 		 * offsets to be reasonably constant
302 		 */
303 		list_add(&rdd.middle, rdd.list);
304 		rdd.is_merge = true;
305 		err = ovl_dir_read(lowerpath, &rdd);
306 		list_del(&rdd.middle);
307 	}
308 out:
309 	return err;
310 
311 }
312 
313 static void ovl_seek_cursor(struct ovl_dir_file *od, loff_t pos)
314 {
315 	struct ovl_cache_entry *p;
316 	loff_t off = 0;
317 
318 	list_for_each_entry(p, &od->cache->entries, l_node) {
319 		if (!p->name)
320 			continue;
321 		if (off >= pos)
322 			break;
323 		off++;
324 	}
325 	list_move_tail(&od->cursor.l_node, &p->l_node);
326 }
327 
328 static struct ovl_dir_cache *ovl_cache_get(struct dentry *dentry)
329 {
330 	int res;
331 	struct path lowerpath;
332 	struct path upperpath;
333 	struct ovl_dir_cache *cache;
334 
335 	cache = ovl_dir_cache(dentry);
336 	if (cache && ovl_dentry_version_get(dentry) == cache->version) {
337 		cache->refcount++;
338 		return cache;
339 	}
340 	ovl_set_dir_cache(dentry, NULL);
341 
342 	cache = kzalloc(sizeof(struct ovl_dir_cache), GFP_KERNEL);
343 	if (!cache)
344 		return ERR_PTR(-ENOMEM);
345 
346 	cache->refcount = 1;
347 	INIT_LIST_HEAD(&cache->entries);
348 
349 	ovl_path_lower(dentry, &lowerpath);
350 	ovl_path_upper(dentry, &upperpath);
351 
352 	res = ovl_dir_read_merged(&upperpath, &lowerpath, &cache->entries);
353 	if (res) {
354 		ovl_cache_free(&cache->entries);
355 		kfree(cache);
356 		return ERR_PTR(res);
357 	}
358 
359 	cache->version = ovl_dentry_version_get(dentry);
360 	ovl_set_dir_cache(dentry, cache);
361 
362 	return cache;
363 }
364 
365 static int ovl_iterate(struct file *file, struct dir_context *ctx)
366 {
367 	struct ovl_dir_file *od = file->private_data;
368 	struct dentry *dentry = file->f_path.dentry;
369 
370 	if (!ctx->pos)
371 		ovl_dir_reset(file);
372 
373 	if (od->is_real)
374 		return iterate_dir(od->realfile, ctx);
375 
376 	if (!od->cache) {
377 		struct ovl_dir_cache *cache;
378 
379 		cache = ovl_cache_get(dentry);
380 		if (IS_ERR(cache))
381 			return PTR_ERR(cache);
382 
383 		od->cache = cache;
384 		ovl_seek_cursor(od, ctx->pos);
385 	}
386 
387 	while (od->cursor.l_node.next != &od->cache->entries) {
388 		struct ovl_cache_entry *p;
389 
390 		p = list_entry(od->cursor.l_node.next, struct ovl_cache_entry, l_node);
391 		/* Skip cursors */
392 		if (p->name) {
393 			if (!p->is_whiteout) {
394 				if (!dir_emit(ctx, p->name, p->len, p->ino, p->type))
395 					break;
396 			}
397 			ctx->pos++;
398 		}
399 		list_move(&od->cursor.l_node, &p->l_node);
400 	}
401 	return 0;
402 }
403 
404 static loff_t ovl_dir_llseek(struct file *file, loff_t offset, int origin)
405 {
406 	loff_t res;
407 	struct ovl_dir_file *od = file->private_data;
408 
409 	mutex_lock(&file_inode(file)->i_mutex);
410 	if (!file->f_pos)
411 		ovl_dir_reset(file);
412 
413 	if (od->is_real) {
414 		res = vfs_llseek(od->realfile, offset, origin);
415 		file->f_pos = od->realfile->f_pos;
416 	} else {
417 		res = -EINVAL;
418 
419 		switch (origin) {
420 		case SEEK_CUR:
421 			offset += file->f_pos;
422 			break;
423 		case SEEK_SET:
424 			break;
425 		default:
426 			goto out_unlock;
427 		}
428 		if (offset < 0)
429 			goto out_unlock;
430 
431 		if (offset != file->f_pos) {
432 			file->f_pos = offset;
433 			if (od->cache)
434 				ovl_seek_cursor(od, offset);
435 		}
436 		res = offset;
437 	}
438 out_unlock:
439 	mutex_unlock(&file_inode(file)->i_mutex);
440 
441 	return res;
442 }
443 
444 static int ovl_dir_fsync(struct file *file, loff_t start, loff_t end,
445 			 int datasync)
446 {
447 	struct ovl_dir_file *od = file->private_data;
448 	struct dentry *dentry = file->f_path.dentry;
449 	struct file *realfile = od->realfile;
450 
451 	/*
452 	 * Need to check if we started out being a lower dir, but got copied up
453 	 */
454 	if (!od->is_upper && ovl_path_type(dentry) == OVL_PATH_MERGE) {
455 		struct inode *inode = file_inode(file);
456 
457 		realfile = od->upperfile;
458 		if (!realfile) {
459 			struct path upperpath;
460 
461 			ovl_path_upper(dentry, &upperpath);
462 			realfile = ovl_path_open(&upperpath, O_RDONLY);
463 			mutex_lock(&inode->i_mutex);
464 			if (!od->upperfile) {
465 				if (IS_ERR(realfile)) {
466 					mutex_unlock(&inode->i_mutex);
467 					return PTR_ERR(realfile);
468 				}
469 				od->upperfile = realfile;
470 			} else {
471 				/* somebody has beaten us to it */
472 				if (!IS_ERR(realfile))
473 					fput(realfile);
474 				realfile = od->upperfile;
475 			}
476 			mutex_unlock(&inode->i_mutex);
477 		}
478 	}
479 
480 	return vfs_fsync_range(realfile, start, end, datasync);
481 }
482 
483 static int ovl_dir_release(struct inode *inode, struct file *file)
484 {
485 	struct ovl_dir_file *od = file->private_data;
486 
487 	if (od->cache) {
488 		mutex_lock(&inode->i_mutex);
489 		ovl_cache_put(od, file->f_path.dentry);
490 		mutex_unlock(&inode->i_mutex);
491 	}
492 	fput(od->realfile);
493 	if (od->upperfile)
494 		fput(od->upperfile);
495 	kfree(od);
496 
497 	return 0;
498 }
499 
500 static int ovl_dir_open(struct inode *inode, struct file *file)
501 {
502 	struct path realpath;
503 	struct file *realfile;
504 	struct ovl_dir_file *od;
505 	enum ovl_path_type type;
506 
507 	od = kzalloc(sizeof(struct ovl_dir_file), GFP_KERNEL);
508 	if (!od)
509 		return -ENOMEM;
510 
511 	type = ovl_path_real(file->f_path.dentry, &realpath);
512 	realfile = ovl_path_open(&realpath, file->f_flags);
513 	if (IS_ERR(realfile)) {
514 		kfree(od);
515 		return PTR_ERR(realfile);
516 	}
517 	INIT_LIST_HEAD(&od->cursor.l_node);
518 	od->realfile = realfile;
519 	od->is_real = (type != OVL_PATH_MERGE);
520 	od->is_upper = (type != OVL_PATH_LOWER);
521 	file->private_data = od;
522 
523 	return 0;
524 }
525 
526 const struct file_operations ovl_dir_operations = {
527 	.read		= generic_read_dir,
528 	.open		= ovl_dir_open,
529 	.iterate	= ovl_iterate,
530 	.llseek		= ovl_dir_llseek,
531 	.fsync		= ovl_dir_fsync,
532 	.release	= ovl_dir_release,
533 };
534 
535 int ovl_check_empty_dir(struct dentry *dentry, struct list_head *list)
536 {
537 	int err;
538 	struct path lowerpath;
539 	struct path upperpath;
540 	struct ovl_cache_entry *p;
541 
542 	ovl_path_upper(dentry, &upperpath);
543 	ovl_path_lower(dentry, &lowerpath);
544 
545 	err = ovl_dir_read_merged(&upperpath, &lowerpath, list);
546 	if (err)
547 		return err;
548 
549 	err = 0;
550 
551 	list_for_each_entry(p, list, l_node) {
552 		if (p->is_whiteout)
553 			continue;
554 
555 		if (p->name[0] == '.') {
556 			if (p->len == 1)
557 				continue;
558 			if (p->len == 2 && p->name[1] == '.')
559 				continue;
560 		}
561 		err = -ENOTEMPTY;
562 		break;
563 	}
564 
565 	return err;
566 }
567 
568 void ovl_cleanup_whiteouts(struct dentry *upper, struct list_head *list)
569 {
570 	struct ovl_cache_entry *p;
571 
572 	mutex_lock_nested(&upper->d_inode->i_mutex, I_MUTEX_PARENT);
573 	list_for_each_entry(p, list, l_node) {
574 		struct dentry *dentry;
575 
576 		if (!p->is_whiteout)
577 			continue;
578 
579 		dentry = lookup_one_len(p->name, upper, p->len);
580 		if (IS_ERR(dentry)) {
581 			pr_err("overlayfs: lookup '%s/%.*s' failed (%i)\n",
582 			       upper->d_name.name, p->len, p->name,
583 			       (int) PTR_ERR(dentry));
584 			continue;
585 		}
586 		ovl_cleanup(upper->d_inode, dentry);
587 		dput(dentry);
588 	}
589 	mutex_unlock(&upper->d_inode->i_mutex);
590 }
591