inode.c 21.9 KB
Newer Older
M
Miklos Szeredi 已提交
1 2 3 4 5 6 7 8 9 10 11
/*
 *
 * Copyright (C) 2011 Novell Inc.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 as published by
 * the Free Software Foundation.
 */

#include <linux/fs.h>
#include <linux/slab.h>
12
#include <linux/cred.h>
M
Miklos Szeredi 已提交
13
#include <linux/xattr.h>
14
#include <linux/posix_acl.h>
15
#include <linux/ratelimit.h>
M
Miklos Szeredi 已提交
16 17
#include "overlayfs.h"

18

M
Miklos Szeredi 已提交
19 20 21 22
int ovl_setattr(struct dentry *dentry, struct iattr *attr)
{
	int err;
	struct dentry *upperdentry;
23
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
24

25 26 27 28 29 30 31 32 33
	/*
	 * Check for permissions before trying to copy-up.  This is redundant
	 * since it will be rechecked later by ->setattr() on upper dentry.  But
	 * without this, copy-up can be triggered by just about anybody.
	 *
	 * We don't initialize inode->size, which just means that
	 * inode_newsize_ok() will always check against MAX_LFS_FILESIZE and not
	 * check for a swapfile (which this won't be anyway).
	 */
34
	err = setattr_prepare(dentry, attr);
35 36 37
	if (err)
		return err;

M
Miklos Szeredi 已提交
38 39 40 41
	err = ovl_want_write(dentry);
	if (err)
		goto out;

42 43 44 45
	err = ovl_copy_up(dentry);
	if (!err) {
		upperdentry = ovl_dentry_upper(dentry);

M
Miklos Szeredi 已提交
46 47 48
		if (attr->ia_valid & (ATTR_KILL_SUID|ATTR_KILL_SGID))
			attr->ia_valid &= ~ATTR_MODE;

A
Al Viro 已提交
49
		inode_lock(upperdentry->d_inode);
50
		old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
51
		err = notify_change(upperdentry, attr, NULL);
52
		revert_creds(old_cred);
53 54
		if (!err)
			ovl_copyattr(upperdentry->d_inode, dentry->d_inode);
A
Al Viro 已提交
55
		inode_unlock(upperdentry->d_inode);
M
Miklos Szeredi 已提交
56 57 58 59 60 61
	}
	ovl_drop_write(dentry);
out:
	return err;
}

62 63 64 65
static int ovl_map_dev_ino(struct dentry *dentry, struct kstat *stat,
			   struct ovl_layer *lower_layer)
{
	bool samefs = ovl_same_sb(dentry->d_sb);
66
	unsigned int xinobits = ovl_xino_bits(dentry->d_sb);
67 68 69 70 71 72 73 74

	if (samefs) {
		/*
		 * When all layers are on the same fs, all real inode
		 * number are unique, so we use the overlay st_dev,
		 * which is friendly to du -x.
		 */
		stat->dev = dentry->d_sb->s_dev;
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
		return 0;
	} else if (xinobits) {
		unsigned int shift = 64 - xinobits;
		/*
		 * All inode numbers of underlying fs should not be using the
		 * high xinobits, so we use high xinobits to partition the
		 * overlay st_ino address space. The high bits holds the fsid
		 * (upper fsid is 0). This way overlay inode numbers are unique
		 * and all inodes use overlay st_dev. Inode numbers are also
		 * persistent for a given layer configuration.
		 */
		if (stat->ino >> shift) {
			pr_warn_ratelimited("overlayfs: inode number too big (%pd2, ino=%llu, xinobits=%d)\n",
					    dentry, stat->ino, xinobits);
		} else {
			if (lower_layer)
				stat->ino |= ((u64)lower_layer->fsid) << shift;

			stat->dev = dentry->d_sb->s_dev;
			return 0;
		}
	}

	/* The inode could not be mapped to a unified st_ino address space */
	if (S_ISDIR(dentry->d_inode->i_mode)) {
100 101 102 103 104 105 106 107 108 109 110
		/*
		 * Always use the overlay st_dev for directories, so 'find
		 * -xdev' will scan the entire overlay mount and won't cross the
		 * overlay mount boundaries.
		 *
		 * If not all layers are on the same fs the pair {real st_ino;
		 * overlay st_dev} is not unique, so use the non persistent
		 * overlay st_ino for directories.
		 */
		stat->dev = dentry->d_sb->s_dev;
		stat->ino = dentry->d_inode->i_ino;
111
	} else if (lower_layer && lower_layer->fsid) {
112 113 114
		/*
		 * For non-samefs setup, if we cannot map all layers st_ino
		 * to a unified address space, we need to make sure that st_dev
115 116 117
		 * is unique per lower fs. Upper layer uses real st_dev and
		 * lower layers use the unique anonymous bdev assigned to the
		 * lower fs.
118
		 */
119
		stat->dev = lower_layer->fs->pseudo_dev;
120 121 122 123 124
	}

	return 0;
}

125 126
int ovl_getattr(const struct path *path, struct kstat *stat,
		u32 request_mask, unsigned int flags)
M
Miklos Szeredi 已提交
127
{
128
	struct dentry *dentry = path->dentry;
129
	enum ovl_path_type type;
M
Miklos Szeredi 已提交
130
	struct path realpath;
131
	const struct cred *old_cred;
132
	bool is_dir = S_ISDIR(dentry->d_inode->i_mode);
133
	bool samefs = ovl_same_sb(dentry->d_sb);
134
	struct ovl_layer *lower_layer = NULL;
135
	int err;
M
Miklos Szeredi 已提交
136

137
	type = ovl_path_real(dentry, &realpath);
138
	old_cred = ovl_override_creds(dentry->d_sb);
139
	err = vfs_getattr(&realpath, stat, request_mask, flags);
140 141 142 143
	if (err)
		goto out;

	/*
144 145
	 * For non-dir or same fs, we use st_ino of the copy up origin.
	 * This guaranties constant st_dev/st_ino across copy up.
146 147
	 * With xino feature and non-samefs, we use st_ino of the copy up
	 * origin masked with high bits that represent the layer id.
148
	 *
149
	 * If lower filesystem supports NFS file handles, this also guaranties
150 151
	 * persistent st_ino across mount cycle.
	 */
152
	if (!is_dir || samefs || ovl_xino_bits(dentry->d_sb)) {
153 154 155
		if (!OVL_TYPE_UPPER(type)) {
			lower_layer = ovl_layer_lower(dentry);
		} else if (OVL_TYPE_ORIGIN(type)) {
156
			struct kstat lowerstat;
157
			u32 lowermask = STATX_INO | (!is_dir ? STATX_NLINK : 0);
158 159 160

			ovl_path_lower(dentry, &realpath);
			err = vfs_getattr(&realpath, &lowerstat,
161
					  lowermask, flags);
162 163 164 165
			if (err)
				goto out;

			/*
166
			 * Lower hardlinks may be broken on copy up to different
167 168
			 * upper files, so we cannot use the lower origin st_ino
			 * for those different files, even for the same fs case.
169 170 171 172 173 174
			 *
			 * Similarly, several redirected dirs can point to the
			 * same dir on a lower layer. With the "verify_lower"
			 * feature, we do not use the lower origin st_ino, if
			 * we haven't verified that this redirect is unique.
			 *
175
			 * With inodes index enabled, it is safe to use st_ino
176 177 178
			 * of an indexed origin. The index validates that the
			 * upper hardlink is not broken and that a redirected
			 * dir is the only redirect to that origin.
179
			 */
180 181
			if (ovl_test_flag(OVL_INDEX, d_inode(dentry)) ||
			    (!ovl_verify_lower(dentry->d_sb) &&
182
			     (is_dir || lowerstat.nlink == 1))) {
183
				stat->ino = lowerstat.ino;
184
				lower_layer = ovl_layer_lower(dentry);
185
			}
186 187
		}
	}
188

189 190 191 192
	err = ovl_map_dev_ino(dentry, stat, lower_layer);
	if (err)
		goto out;

193 194 195 196 197 198 199 200
	/*
	 * It's probably not worth it to count subdirs to get the
	 * correct link count.  nlink=1 seems to pacify 'find' and
	 * other utilities.
	 */
	if (is_dir && OVL_TYPE_MERGE(type))
		stat->nlink = 1;

201 202 203 204 205 206 207 208 209
	/*
	 * Return the overlay inode nlinks for indexed upper inodes.
	 * Overlay inode nlink counts the union of the upper hardlinks
	 * and non-covered lower hardlinks. It does not include the upper
	 * index hardlink.
	 */
	if (!is_dir && ovl_test_flag(OVL_INDEX, d_inode(dentry)))
		stat->nlink = dentry->d_inode->i_nlink;

210
out:
211
	revert_creds(old_cred);
212

213
	return err;
M
Miklos Szeredi 已提交
214 215 216 217
}

int ovl_permission(struct inode *inode, int mask)
{
218 219
	struct inode *upperinode = ovl_inode_upper(inode);
	struct inode *realinode = upperinode ?: ovl_inode_lower(inode);
220
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
221 222 223 224 225
	int err;

	/* Careful in RCU walk mode */
	if (!realinode) {
		WARN_ON(!(mask & MAY_NOT_BLOCK));
226
		return -ECHILD;
M
Miklos Szeredi 已提交
227 228
	}

229 230 231 232 233 234 235 236 237
	/*
	 * Check overlay inode with the creds of task and underlying inode
	 * with creds of mounter
	 */
	err = generic_permission(inode, mask);
	if (err)
		return err;

	old_cred = ovl_override_creds(inode->i_sb);
238 239
	if (!upperinode &&
	    !special_file(realinode->i_mode) && mask & MAY_WRITE) {
240
		mask &= ~(MAY_WRITE | MAY_APPEND);
241 242 243
		/* Make sure mounter can read file for copy up later */
		mask |= MAY_READ;
	}
244
	err = inode_permission(realinode, mask);
245 246 247
	revert_creds(old_cred);

	return err;
M
Miklos Szeredi 已提交
248 249
}

250
static const char *ovl_get_link(struct dentry *dentry,
251 252
				struct inode *inode,
				struct delayed_call *done)
M
Miklos Szeredi 已提交
253
{
254 255
	const struct cred *old_cred;
	const char *p;
M
Miklos Szeredi 已提交
256

257 258 259
	if (!dentry)
		return ERR_PTR(-ECHILD);

260
	old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
261
	p = vfs_get_link(ovl_dentry_real(dentry), done);
262 263
	revert_creds(old_cred);
	return p;
M
Miklos Szeredi 已提交
264 265
}

M
Miklos Szeredi 已提交
266
bool ovl_is_private_xattr(const char *name)
M
Miklos Szeredi 已提交
267
{
A
Andreas Gruenbacher 已提交
268 269
	return strncmp(name, OVL_XATTR_PREFIX,
		       sizeof(OVL_XATTR_PREFIX) - 1) == 0;
M
Miklos Szeredi 已提交
270 271
}

272 273
int ovl_xattr_set(struct dentry *dentry, struct inode *inode, const char *name,
		  const void *value, size_t size, int flags)
M
Miklos Szeredi 已提交
274 275
{
	int err;
276 277
	struct dentry *upperdentry = ovl_i_dentry_upper(inode);
	struct dentry *realdentry = upperdentry ?: ovl_dentry_lower(dentry);
278
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
279 280 281 282 283

	err = ovl_want_write(dentry);
	if (err)
		goto out;

284 285
	if (!value && !upperdentry) {
		err = vfs_getxattr(realdentry, name, NULL, 0);
286 287 288 289
		if (err < 0)
			goto out_drop_write;
	}

290 291 292 293
	if (!upperdentry) {
		err = ovl_copy_up(dentry);
		if (err)
			goto out_drop_write;
M
Miklos Szeredi 已提交
294

295 296
		realdentry = ovl_dentry_upper(dentry);
	}
297

298
	old_cred = ovl_override_creds(dentry->d_sb);
299
	if (value)
300
		err = vfs_setxattr(realdentry, name, value, size, flags);
301 302
	else {
		WARN_ON(flags != XATTR_REPLACE);
303
		err = vfs_removexattr(realdentry, name);
304
	}
305
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
306 307 308 309 310 311 312

out_drop_write:
	ovl_drop_write(dentry);
out:
	return err;
}

313
int ovl_xattr_get(struct dentry *dentry, struct inode *inode, const char *name,
314
		  void *value, size_t size)
M
Miklos Szeredi 已提交
315
{
316 317
	ssize_t res;
	const struct cred *old_cred;
318 319
	struct dentry *realdentry =
		ovl_i_dentry_upper(inode) ?: ovl_dentry_lower(dentry);
320

321 322 323 324
	old_cred = ovl_override_creds(dentry->d_sb);
	res = vfs_getxattr(realdentry, name, value, size);
	revert_creds(old_cred);
	return res;
M
Miklos Szeredi 已提交
325 326
}

327 328 329 330 331 332 333 334 335 336
static bool ovl_can_list(const char *s)
{
	/* List all non-trusted xatts */
	if (strncmp(s, XATTR_TRUSTED_PREFIX, XATTR_TRUSTED_PREFIX_LEN) != 0)
		return true;

	/* Never list trusted.overlay, list other trusted for superuser only */
	return !ovl_is_private_xattr(s) && capable(CAP_SYS_ADMIN);
}

M
Miklos Szeredi 已提交
337 338
ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size)
{
M
Miklos Szeredi 已提交
339
	struct dentry *realdentry = ovl_dentry_real(dentry);
M
Miklos Szeredi 已提交
340
	ssize_t res;
M
Miklos Szeredi 已提交
341 342
	size_t len;
	char *s;
343
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
344

345
	old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
346
	res = vfs_listxattr(realdentry, list, size);
347
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
348 349 350 351
	if (res <= 0 || size == 0)
		return res;

	/* filter out private xattrs */
M
Miklos Szeredi 已提交
352 353
	for (s = list, len = res; len;) {
		size_t slen = strnlen(s, len) + 1;
M
Miklos Szeredi 已提交
354

M
Miklos Szeredi 已提交
355 356 357
		/* underlying fs providing us with an broken xattr list? */
		if (WARN_ON(slen > len))
			return -EIO;
M
Miklos Szeredi 已提交
358

M
Miklos Szeredi 已提交
359
		len -= slen;
360
		if (!ovl_can_list(s)) {
M
Miklos Szeredi 已提交
361
			res -= slen;
M
Miklos Szeredi 已提交
362
			memmove(s, s + slen, len);
M
Miklos Szeredi 已提交
363
		} else {
M
Miklos Szeredi 已提交
364
			s += slen;
M
Miklos Szeredi 已提交
365 366 367 368 369 370
		}
	}

	return res;
}

371 372
struct posix_acl *ovl_get_acl(struct inode *inode, int type)
{
373
	struct inode *realinode = ovl_inode_real(inode);
374 375
	const struct cred *old_cred;
	struct posix_acl *acl;
376

377
	if (!IS_ENABLED(CONFIG_FS_POSIX_ACL) || !IS_POSIXACL(realinode))
378 379
		return NULL;

380
	old_cred = ovl_override_creds(inode->i_sb);
381
	acl = get_acl(realinode, type);
382 383 384
	revert_creds(old_cred);

	return acl;
385 386
}

387
static bool ovl_open_need_copy_up(struct dentry *dentry, int flags)
M
Miklos Szeredi 已提交
388
{
389
	/* Copy up of disconnected dentry does not set upper alias */
390
	if (ovl_dentry_upper(dentry) &&
391 392
	    (ovl_dentry_has_upper_alias(dentry) ||
	     (dentry->d_flags & DCACHE_DISCONNECTED)))
M
Miklos Szeredi 已提交
393 394
		return false;

395
	if (special_file(d_inode(dentry)->i_mode))
M
Miklos Szeredi 已提交
396 397 398 399 400 401 402 403
		return false;

	if (!(OPEN_FMODE(flags) & FMODE_WRITE) && !(flags & O_TRUNC))
		return false;

	return true;
}

404
int ovl_open_maybe_copy_up(struct dentry *dentry, unsigned int file_flags)
M
Miklos Szeredi 已提交
405
{
406
	int err = 0;
M
Miklos Szeredi 已提交
407

408
	if (ovl_open_need_copy_up(dentry, file_flags)) {
M
Miklos Szeredi 已提交
409
		err = ovl_want_write(dentry);
410
		if (!err) {
411
			err = ovl_copy_up_flags(dentry, file_flags);
412 413
			ovl_drop_write(dentry);
		}
M
Miklos Szeredi 已提交
414 415
	}

416
	return err;
M
Miklos Szeredi 已提交
417 418
}

M
Miklos Szeredi 已提交
419 420
int ovl_update_time(struct inode *inode, struct timespec *ts, int flags)
{
M
Miklos Szeredi 已提交
421 422 423 424 425 426 427 428 429 430 431
	if (flags & S_ATIME) {
		struct ovl_fs *ofs = inode->i_sb->s_fs_info;
		struct path upperpath = {
			.mnt = ofs->upper_mnt,
			.dentry = ovl_upperdentry_dereference(OVL_I(inode)),
		};

		if (upperpath.dentry) {
			touch_atime(&upperpath);
			inode->i_atime = d_inode(upperpath.dentry)->i_atime;
		}
M
Miklos Szeredi 已提交
432 433 434 435
	}
	return 0;
}

M
Miklos Szeredi 已提交
436 437 438 439 440
static const struct inode_operations ovl_file_inode_operations = {
	.setattr	= ovl_setattr,
	.permission	= ovl_permission,
	.getattr	= ovl_getattr,
	.listxattr	= ovl_listxattr,
441
	.get_acl	= ovl_get_acl,
M
Miklos Szeredi 已提交
442
	.update_time	= ovl_update_time,
M
Miklos Szeredi 已提交
443 444 445 446
};

static const struct inode_operations ovl_symlink_inode_operations = {
	.setattr	= ovl_setattr,
447
	.get_link	= ovl_get_link,
M
Miklos Szeredi 已提交
448 449
	.getattr	= ovl_getattr,
	.listxattr	= ovl_listxattr,
M
Miklos Szeredi 已提交
450
	.update_time	= ovl_update_time,
M
Miklos Szeredi 已提交
451 452
};

453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474
/*
 * It is possible to stack overlayfs instance on top of another
 * overlayfs instance as lower layer. We need to annonate the
 * stackable i_mutex locks according to stack level of the super
 * block instance. An overlayfs instance can never be in stack
 * depth 0 (there is always a real fs below it).  An overlayfs
 * inode lock will use the lockdep annotaion ovl_i_mutex_key[depth].
 *
 * For example, here is a snip from /proc/lockdep_chains after
 * dir_iterate of nested overlayfs:
 *
 * [...] &ovl_i_mutex_dir_key[depth]   (stack_depth=2)
 * [...] &ovl_i_mutex_dir_key[depth]#2 (stack_depth=1)
 * [...] &type->i_mutex_dir_key        (stack_depth=0)
 */
#define OVL_MAX_NESTING FILESYSTEM_MAX_STACK_DEPTH

static inline void ovl_lockdep_annotate_inode_mutex_key(struct inode *inode)
{
#ifdef CONFIG_LOCKDEP
	static struct lock_class_key ovl_i_mutex_key[OVL_MAX_NESTING];
	static struct lock_class_key ovl_i_mutex_dir_key[OVL_MAX_NESTING];
475
	static struct lock_class_key ovl_i_lock_key[OVL_MAX_NESTING];
476 477 478 479 480 481 482 483 484 485

	int depth = inode->i_sb->s_stack_depth - 1;

	if (WARN_ON_ONCE(depth < 0 || depth >= OVL_MAX_NESTING))
		depth = 0;

	if (S_ISDIR(inode->i_mode))
		lockdep_set_class(&inode->i_rwsem, &ovl_i_mutex_dir_key[depth]);
	else
		lockdep_set_class(&inode->i_rwsem, &ovl_i_mutex_key[depth]);
486 487

	lockdep_set_class(&OVL_I(inode)->lock, &ovl_i_lock_key[depth]);
488 489 490
#endif
}

491 492
static void ovl_fill_inode(struct inode *inode, umode_t mode, dev_t rdev,
			   unsigned long ino)
M
Miklos Szeredi 已提交
493
{
494 495 496 497 498 499 500 501 502 503 504
	/*
	 * When NFS export is enabled and d_ino is consistent with st_ino
	 * (samefs), set the same value to i_ino, because nfsd readdirplus
	 * compares d_ino values to i_ino values of child entries. When called
	 * from ovl_new_inode(), ino arg is 0, so i_ino will be updated to real
	 * upper inode i_ino on ovl_inode_init() or ovl_inode_update().
	 */
	if (inode->i_sb->s_export_op && ovl_same_sb(inode->i_sb))
		inode->i_ino = ino;
	else
		inode->i_ino = get_next_ino();
M
Miklos Szeredi 已提交
505
	inode->i_mode = mode;
M
Miklos Szeredi 已提交
506
	inode->i_flags |= S_NOCMTIME;
507 508 509
#ifdef CONFIG_FS_POSIX_ACL
	inode->i_acl = inode->i_default_acl = ACL_DONT_CACHE;
#endif
M
Miklos Szeredi 已提交
510

511 512
	ovl_lockdep_annotate_inode_mutex_key(inode);

513 514 515 516 517
	switch (mode & S_IFMT) {
	case S_IFREG:
		inode->i_op = &ovl_file_inode_operations;
		break;

M
Miklos Szeredi 已提交
518 519 520 521 522 523 524 525 526
	case S_IFDIR:
		inode->i_op = &ovl_dir_inode_operations;
		inode->i_fop = &ovl_dir_operations;
		break;

	case S_IFLNK:
		inode->i_op = &ovl_symlink_inode_operations;
		break;

M
Miklos Szeredi 已提交
527
	default:
M
Miklos Szeredi 已提交
528
		inode->i_op = &ovl_file_inode_operations;
529
		init_special_inode(inode, mode, rdev);
M
Miklos Szeredi 已提交
530
		break;
M
Miklos Szeredi 已提交
531 532
	}
}
M
Miklos Szeredi 已提交
533

534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577
/*
 * With inodes index enabled, an overlay inode nlink counts the union of upper
 * hardlinks and non-covered lower hardlinks. During the lifetime of a non-pure
 * upper inode, the following nlink modifying operations can happen:
 *
 * 1. Lower hardlink copy up
 * 2. Upper hardlink created, unlinked or renamed over
 * 3. Lower hardlink whiteout or renamed over
 *
 * For the first, copy up case, the union nlink does not change, whether the
 * operation succeeds or fails, but the upper inode nlink may change.
 * Therefore, before copy up, we store the union nlink value relative to the
 * lower inode nlink in the index inode xattr trusted.overlay.nlink.
 *
 * For the second, upper hardlink case, the union nlink should be incremented
 * or decremented IFF the operation succeeds, aligned with nlink change of the
 * upper inode. Therefore, before link/unlink/rename, we store the union nlink
 * value relative to the upper inode nlink in the index inode.
 *
 * For the last, lower cover up case, we simplify things by preceding the
 * whiteout or cover up with copy up. This makes sure that there is an index
 * upper inode where the nlink xattr can be stored before the copied up upper
 * entry is unlink.
 */
#define OVL_NLINK_ADD_UPPER	(1 << 0)

/*
 * On-disk format for indexed nlink:
 *
 * nlink relative to the upper inode - "U[+-]NUM"
 * nlink relative to the lower inode - "L[+-]NUM"
 */

static int ovl_set_nlink_common(struct dentry *dentry,
				struct dentry *realdentry, const char *format)
{
	struct inode *inode = d_inode(dentry);
	struct inode *realinode = d_inode(realdentry);
	char buf[13];
	int len;

	len = snprintf(buf, sizeof(buf), format,
		       (int) (inode->i_nlink - realinode->i_nlink));

M
Miklos Szeredi 已提交
578 579 580
	if (WARN_ON(len >= sizeof(buf)))
		return -EIO;

581 582 583 584 585 586 587 588 589 590 591 592 593 594
	return ovl_do_setxattr(ovl_dentry_upper(dentry),
			       OVL_XATTR_NLINK, buf, len, 0);
}

int ovl_set_nlink_upper(struct dentry *dentry)
{
	return ovl_set_nlink_common(dentry, ovl_dentry_upper(dentry), "U%+i");
}

int ovl_set_nlink_lower(struct dentry *dentry)
{
	return ovl_set_nlink_common(dentry, ovl_dentry_lower(dentry), "L%+i");
}

595 596 597
unsigned int ovl_get_nlink(struct dentry *lowerdentry,
			   struct dentry *upperdentry,
			   unsigned int fallback)
598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
{
	int nlink_diff;
	int nlink;
	char buf[13];
	int err;

	if (!lowerdentry || !upperdentry || d_inode(lowerdentry)->i_nlink == 1)
		return fallback;

	err = vfs_getxattr(upperdentry, OVL_XATTR_NLINK, &buf, sizeof(buf) - 1);
	if (err < 0)
		goto fail;

	buf[err] = '\0';
	if ((buf[0] != 'L' && buf[0] != 'U') ||
	    (buf[1] != '+' && buf[1] != '-'))
		goto fail;

	err = kstrtoint(buf + 1, 10, &nlink_diff);
	if (err < 0)
		goto fail;

	nlink = d_inode(buf[0] == 'L' ? lowerdentry : upperdentry)->i_nlink;
	nlink += nlink_diff;

	if (nlink <= 0)
		goto fail;

	return nlink;

fail:
	pr_warn_ratelimited("overlayfs: failed to get index nlink (%pd2, err=%i)\n",
			    upperdentry, err);
	return fallback;
}

634
struct inode *ovl_new_inode(struct super_block *sb, umode_t mode, dev_t rdev)
M
Miklos Szeredi 已提交
635 636 637 638 639
{
	struct inode *inode;

	inode = new_inode(sb);
	if (inode)
640
		ovl_fill_inode(inode, mode, rdev, 0);
M
Miklos Szeredi 已提交
641 642 643 644 645 646

	return inode;
}

static int ovl_inode_test(struct inode *inode, void *data)
{
647
	return inode->i_private == data;
M
Miklos Szeredi 已提交
648 649 650 651
}

static int ovl_inode_set(struct inode *inode, void *data)
{
652
	inode->i_private = data;
M
Miklos Szeredi 已提交
653 654 655
	return 0;
}

656
static bool ovl_verify_inode(struct inode *inode, struct dentry *lowerdentry,
657
			     struct dentry *upperdentry, bool strict)
658
{
659 660 661 662 663 664 665
	/*
	 * For directories, @strict verify from lookup path performs consistency
	 * checks, so NULL lower/upper in dentry must match NULL lower/upper in
	 * inode. Non @strict verify from NFS handle decode path passes NULL for
	 * 'unknown' lower/upper.
	 */
	if (S_ISDIR(inode->i_mode) && strict) {
666 667 668 669 670 671 672 673 674
		/* Real lower dir moved to upper layer under us? */
		if (!lowerdentry && ovl_inode_lower(inode))
			return false;

		/* Lookup of an uncovered redirect origin? */
		if (!upperdentry && ovl_inode_upper(inode))
			return false;
	}

675 676 677 678 679 680 681
	/*
	 * Allow non-NULL lower inode in ovl_inode even if lowerdentry is NULL.
	 * This happens when finding a copied up overlay inode for a renamed
	 * or hardlinked overlay dentry and lower dentry cannot be followed
	 * by origin because lower fs does not support file handles.
	 */
	if (lowerdentry && ovl_inode_lower(inode) != d_inode(lowerdentry))
682 683 684 685 686 687 688 689 690 691 692 693
		return false;

	/*
	 * Allow non-NULL __upperdentry in inode even if upperdentry is NULL.
	 * This happens when finding a lower alias for a copied up hard link.
	 */
	if (upperdentry && ovl_inode_upper(inode) != d_inode(upperdentry))
		return false;

	return true;
}

694 695
struct inode *ovl_lookup_inode(struct super_block *sb, struct dentry *real,
			       bool is_upper)
696
{
697
	struct inode *inode, *key = d_inode(real);
698 699 700 701 702

	inode = ilookup5(sb, (unsigned long) key, ovl_inode_test, key);
	if (!inode)
		return NULL;

703 704
	if (!ovl_verify_inode(inode, is_upper ? NULL : real,
			      is_upper ? real : NULL, false)) {
705 706 707 708 709 710 711
		iput(inode);
		return ERR_PTR(-ESTALE);
	}

	return inode;
}

712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
/*
 * Does overlay inode need to be hashed by lower inode?
 */
static bool ovl_hash_bylower(struct super_block *sb, struct dentry *upper,
			     struct dentry *lower, struct dentry *index)
{
	struct ovl_fs *ofs = sb->s_fs_info;

	/* No, if pure upper */
	if (!lower)
		return false;

	/* Yes, if already indexed */
	if (index)
		return true;

	/* Yes, if won't be copied up */
	if (!ofs->upper_mnt)
		return true;

	/* No, if lower hardlink is or will be broken on copy up */
	if ((upper || !ovl_indexdir(sb)) &&
	    !d_is_dir(lower) && d_inode(lower)->i_nlink > 1)
		return false;

	/* No, if non-indexed upper with NFS export */
	if (sb->s_export_op && upper)
		return false;

	/* Otherwise, hash by lower inode for fsnotify */
	return true;
}

745 746 747
struct inode *ovl_get_inode(struct super_block *sb, struct dentry *upperdentry,
			    struct dentry *lowerdentry, struct dentry *index,
			    unsigned int numlower)
M
Miklos Szeredi 已提交
748
{
749
	struct inode *realinode = upperdentry ? d_inode(upperdentry) : NULL;
M
Miklos Szeredi 已提交
750
	struct inode *inode;
751
	bool bylower = ovl_hash_bylower(sb, upperdentry, lowerdentry, index);
752
	bool is_dir;
753
	unsigned long ino = 0;
754

755 756 757
	if (!realinode)
		realinode = d_inode(lowerdentry);

758
	/*
759 760
	 * Copy up origin (lower) may exist for non-indexed upper, but we must
	 * not use lower as hash key if this is a broken hardlink.
761
	 */
762
	is_dir = S_ISDIR(realinode->i_mode);
763 764 765
	if (upperdentry || bylower) {
		struct inode *key = d_inode(bylower ? lowerdentry :
						      upperdentry);
766
		unsigned int nlink = is_dir ? 1 : realinode->i_nlink;
767

768
		inode = iget5_locked(sb, (unsigned long) key,
769
				     ovl_inode_test, ovl_inode_set, key);
770
		if (!inode)
771
			goto out_nomem;
772
		if (!(inode->i_state & I_NEW)) {
773 774 775 776
			/*
			 * Verify that the underlying files stored in the inode
			 * match those in the dentry.
			 */
777 778
			if (!ovl_verify_inode(inode, lowerdentry, upperdentry,
					      true)) {
779 780 781 782 783
				iput(inode);
				inode = ERR_PTR(-ESTALE);
				goto out;
			}

784 785 786
			dput(upperdentry);
			goto out;
		}
M
Miklos Szeredi 已提交
787

788 789 790
		/* Recalculate nlink for non-dir due to indexing */
		if (!is_dir)
			nlink = ovl_get_nlink(lowerdentry, upperdentry, nlink);
791
		set_nlink(inode, nlink);
792
		ino = key->i_ino;
M
Miklos Szeredi 已提交
793
	} else {
794
		/* Lower hardlink that will be broken on copy up */
795
		inode = new_inode(sb);
M
Miklos Szeredi 已提交
796
		if (!inode)
797
			goto out_nomem;
M
Miklos Szeredi 已提交
798
	}
799
	ovl_fill_inode(inode, realinode->i_mode, realinode->i_rdev, ino);
800
	ovl_inode_init(inode, upperdentry, lowerdentry);
M
Miklos Szeredi 已提交
801 802 803 804

	if (upperdentry && ovl_is_impuredir(upperdentry))
		ovl_set_flag(OVL_IMPURE, inode);

V
Vivek Goyal 已提交
805 806 807
	if (index)
		ovl_set_flag(OVL_INDEX, inode);

808
	/* Check for non-merge dir that may have whiteouts */
809
	if (is_dir) {
810
		if (((upperdentry && lowerdentry) || numlower > 1) ||
811 812 813 814 815
		    ovl_check_origin_xattr(upperdentry ?: lowerdentry)) {
			ovl_set_flag(OVL_WHITEOUTS, inode);
		}
	}

M
Miklos Szeredi 已提交
816 817 818
	if (inode->i_state & I_NEW)
		unlock_new_inode(inode);
out:
M
Miklos Szeredi 已提交
819
	return inode;
820 821 822 823

out_nomem:
	inode = ERR_PTR(-ENOMEM);
	goto out;
M
Miklos Szeredi 已提交
824
}