inode.c 20.9 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 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85
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);

	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;
	} else if (S_ISDIR(dentry->d_inode->i_mode)) {
		/*
		 * 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;
86
	} else if (lower_layer && lower_layer->fsid) {
87 88 89
		/*
		 * 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
90 91 92
		 * is unique per lower fs. Upper layer uses real st_dev and
		 * lower layers use the unique anonymous bdev assigned to the
		 * lower fs.
93
		 */
94
		stat->dev = lower_layer->fs->pseudo_dev;
95 96 97 98 99
	}

	return 0;
}

100 101
int ovl_getattr(const struct path *path, struct kstat *stat,
		u32 request_mask, unsigned int flags)
M
Miklos Szeredi 已提交
102
{
103
	struct dentry *dentry = path->dentry;
104
	enum ovl_path_type type;
M
Miklos Szeredi 已提交
105
	struct path realpath;
106
	const struct cred *old_cred;
107
	bool is_dir = S_ISDIR(dentry->d_inode->i_mode);
108
	bool samefs = ovl_same_sb(dentry->d_sb);
109
	struct ovl_layer *lower_layer = NULL;
110
	int err;
M
Miklos Szeredi 已提交
111

112
	type = ovl_path_real(dentry, &realpath);
113
	old_cred = ovl_override_creds(dentry->d_sb);
114
	err = vfs_getattr(&realpath, stat, request_mask, flags);
115 116 117 118
	if (err)
		goto out;

	/*
119 120
	 * 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.
121
	 *
122
	 * If lower filesystem supports NFS file handles, this also guaranties
123 124
	 * persistent st_ino across mount cycle.
	 */
125
	if (!is_dir || samefs) {
126 127 128
		if (!OVL_TYPE_UPPER(type)) {
			lower_layer = ovl_layer_lower(dentry);
		} else if (OVL_TYPE_ORIGIN(type)) {
129
			struct kstat lowerstat;
130
			u32 lowermask = STATX_INO | (!is_dir ? STATX_NLINK : 0);
131 132 133

			ovl_path_lower(dentry, &realpath);
			err = vfs_getattr(&realpath, &lowerstat,
134
					  lowermask, flags);
135 136 137 138
			if (err)
				goto out;

			/*
139
			 * Lower hardlinks may be broken on copy up to different
140 141
			 * upper files, so we cannot use the lower origin st_ino
			 * for those different files, even for the same fs case.
142 143 144 145 146 147
			 *
			 * 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.
			 *
148
			 * With inodes index enabled, it is safe to use st_ino
149 150 151
			 * 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.
152
			 */
153 154
			if (ovl_test_flag(OVL_INDEX, d_inode(dentry)) ||
			    (!ovl_verify_lower(dentry->d_sb) &&
155
			     (is_dir || lowerstat.nlink == 1))) {
156
				stat->ino = lowerstat.ino;
157
				lower_layer = ovl_layer_lower(dentry);
158
			}
159 160
		}
	}
161

162 163 164 165
	err = ovl_map_dev_ino(dentry, stat, lower_layer);
	if (err)
		goto out;

166 167 168 169 170 171 172 173
	/*
	 * 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;

174 175 176 177 178 179 180 181 182
	/*
	 * 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;

183
out:
184
	revert_creds(old_cred);
185

186
	return err;
M
Miklos Szeredi 已提交
187 188 189 190
}

int ovl_permission(struct inode *inode, int mask)
{
191 192
	struct inode *upperinode = ovl_inode_upper(inode);
	struct inode *realinode = upperinode ?: ovl_inode_lower(inode);
193
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
194 195 196 197 198
	int err;

	/* Careful in RCU walk mode */
	if (!realinode) {
		WARN_ON(!(mask & MAY_NOT_BLOCK));
199
		return -ECHILD;
M
Miklos Szeredi 已提交
200 201
	}

202 203 204 205 206 207 208 209 210
	/*
	 * 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);
211 212
	if (!upperinode &&
	    !special_file(realinode->i_mode) && mask & MAY_WRITE) {
213
		mask &= ~(MAY_WRITE | MAY_APPEND);
214 215 216
		/* Make sure mounter can read file for copy up later */
		mask |= MAY_READ;
	}
217
	err = inode_permission(realinode, mask);
218 219 220
	revert_creds(old_cred);

	return err;
M
Miklos Szeredi 已提交
221 222
}

223
static const char *ovl_get_link(struct dentry *dentry,
224 225
				struct inode *inode,
				struct delayed_call *done)
M
Miklos Szeredi 已提交
226
{
227 228
	const struct cred *old_cred;
	const char *p;
M
Miklos Szeredi 已提交
229

230 231 232
	if (!dentry)
		return ERR_PTR(-ECHILD);

233
	old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
234
	p = vfs_get_link(ovl_dentry_real(dentry), done);
235 236
	revert_creds(old_cred);
	return p;
M
Miklos Szeredi 已提交
237 238
}

M
Miklos Szeredi 已提交
239
bool ovl_is_private_xattr(const char *name)
M
Miklos Szeredi 已提交
240
{
A
Andreas Gruenbacher 已提交
241 242
	return strncmp(name, OVL_XATTR_PREFIX,
		       sizeof(OVL_XATTR_PREFIX) - 1) == 0;
M
Miklos Szeredi 已提交
243 244
}

245 246
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
247 248
{
	int err;
249 250
	struct dentry *upperdentry = ovl_i_dentry_upper(inode);
	struct dentry *realdentry = upperdentry ?: ovl_dentry_lower(dentry);
251
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
252 253 254 255 256

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

257 258
	if (!value && !upperdentry) {
		err = vfs_getxattr(realdentry, name, NULL, 0);
259 260 261 262
		if (err < 0)
			goto out_drop_write;
	}

263 264 265 266
	if (!upperdentry) {
		err = ovl_copy_up(dentry);
		if (err)
			goto out_drop_write;
M
Miklos Szeredi 已提交
267

268 269
		realdentry = ovl_dentry_upper(dentry);
	}
270

271
	old_cred = ovl_override_creds(dentry->d_sb);
272
	if (value)
273
		err = vfs_setxattr(realdentry, name, value, size, flags);
274 275
	else {
		WARN_ON(flags != XATTR_REPLACE);
276
		err = vfs_removexattr(realdentry, name);
277
	}
278
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
279 280 281 282 283 284 285

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

286
int ovl_xattr_get(struct dentry *dentry, struct inode *inode, const char *name,
287
		  void *value, size_t size)
M
Miklos Szeredi 已提交
288
{
289 290
	ssize_t res;
	const struct cred *old_cred;
291 292
	struct dentry *realdentry =
		ovl_i_dentry_upper(inode) ?: ovl_dentry_lower(dentry);
293

294 295 296 297
	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
298 299
}

300 301 302 303 304 305 306 307 308 309
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
310 311
ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size)
{
M
Miklos Szeredi 已提交
312
	struct dentry *realdentry = ovl_dentry_real(dentry);
M
Miklos Szeredi 已提交
313
	ssize_t res;
M
Miklos Szeredi 已提交
314 315
	size_t len;
	char *s;
316
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
317

318
	old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
319
	res = vfs_listxattr(realdentry, list, size);
320
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
321 322 323 324
	if (res <= 0 || size == 0)
		return res;

	/* filter out private xattrs */
M
Miklos Szeredi 已提交
325 326
	for (s = list, len = res; len;) {
		size_t slen = strnlen(s, len) + 1;
M
Miklos Szeredi 已提交
327

M
Miklos Szeredi 已提交
328 329 330
		/* underlying fs providing us with an broken xattr list? */
		if (WARN_ON(slen > len))
			return -EIO;
M
Miklos Szeredi 已提交
331

M
Miklos Szeredi 已提交
332
		len -= slen;
333
		if (!ovl_can_list(s)) {
M
Miklos Szeredi 已提交
334
			res -= slen;
M
Miklos Szeredi 已提交
335
			memmove(s, s + slen, len);
M
Miklos Szeredi 已提交
336
		} else {
M
Miklos Szeredi 已提交
337
			s += slen;
M
Miklos Szeredi 已提交
338 339 340 341 342 343
		}
	}

	return res;
}

344 345
struct posix_acl *ovl_get_acl(struct inode *inode, int type)
{
346
	struct inode *realinode = ovl_inode_real(inode);
347 348
	const struct cred *old_cred;
	struct posix_acl *acl;
349

350
	if (!IS_ENABLED(CONFIG_FS_POSIX_ACL) || !IS_POSIXACL(realinode))
351 352
		return NULL;

353
	old_cred = ovl_override_creds(inode->i_sb);
354
	acl = get_acl(realinode, type);
355 356 357
	revert_creds(old_cred);

	return acl;
358 359
}

360
static bool ovl_open_need_copy_up(struct dentry *dentry, int flags)
M
Miklos Szeredi 已提交
361
{
362
	/* Copy up of disconnected dentry does not set upper alias */
363
	if (ovl_dentry_upper(dentry) &&
364 365
	    (ovl_dentry_has_upper_alias(dentry) ||
	     (dentry->d_flags & DCACHE_DISCONNECTED)))
M
Miklos Szeredi 已提交
366 367
		return false;

368
	if (special_file(d_inode(dentry)->i_mode))
M
Miklos Szeredi 已提交
369 370 371 372 373 374 375 376
		return false;

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

	return true;
}

377
int ovl_open_maybe_copy_up(struct dentry *dentry, unsigned int file_flags)
M
Miklos Szeredi 已提交
378
{
379
	int err = 0;
M
Miklos Szeredi 已提交
380

381
	if (ovl_open_need_copy_up(dentry, file_flags)) {
M
Miklos Szeredi 已提交
382
		err = ovl_want_write(dentry);
383
		if (!err) {
384
			err = ovl_copy_up_flags(dentry, file_flags);
385 386
			ovl_drop_write(dentry);
		}
M
Miklos Szeredi 已提交
387 388
	}

389
	return err;
M
Miklos Szeredi 已提交
390 391
}

M
Miklos Szeredi 已提交
392 393
int ovl_update_time(struct inode *inode, struct timespec *ts, int flags)
{
M
Miklos Szeredi 已提交
394 395 396 397 398 399 400 401 402 403 404
	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
405 406 407 408
	}
	return 0;
}

M
Miklos Szeredi 已提交
409 410 411 412 413
static const struct inode_operations ovl_file_inode_operations = {
	.setattr	= ovl_setattr,
	.permission	= ovl_permission,
	.getattr	= ovl_getattr,
	.listxattr	= ovl_listxattr,
414
	.get_acl	= ovl_get_acl,
M
Miklos Szeredi 已提交
415
	.update_time	= ovl_update_time,
M
Miklos Szeredi 已提交
416 417 418 419
};

static const struct inode_operations ovl_symlink_inode_operations = {
	.setattr	= ovl_setattr,
420
	.get_link	= ovl_get_link,
M
Miklos Szeredi 已提交
421 422
	.getattr	= ovl_getattr,
	.listxattr	= ovl_listxattr,
M
Miklos Szeredi 已提交
423
	.update_time	= ovl_update_time,
M
Miklos Szeredi 已提交
424 425
};

426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447
/*
 * 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];
448
	static struct lock_class_key ovl_i_lock_key[OVL_MAX_NESTING];
449 450 451 452 453 454 455 456 457 458

	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]);
459 460

	lockdep_set_class(&OVL_I(inode)->lock, &ovl_i_lock_key[depth]);
461 462 463
#endif
}

464 465
static void ovl_fill_inode(struct inode *inode, umode_t mode, dev_t rdev,
			   unsigned long ino)
M
Miklos Szeredi 已提交
466
{
467 468 469 470 471 472 473 474 475 476 477
	/*
	 * 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
478
	inode->i_mode = mode;
M
Miklos Szeredi 已提交
479
	inode->i_flags |= S_NOCMTIME;
480 481 482
#ifdef CONFIG_FS_POSIX_ACL
	inode->i_acl = inode->i_default_acl = ACL_DONT_CACHE;
#endif
M
Miklos Szeredi 已提交
483

484 485
	ovl_lockdep_annotate_inode_mutex_key(inode);

486 487 488 489 490
	switch (mode & S_IFMT) {
	case S_IFREG:
		inode->i_op = &ovl_file_inode_operations;
		break;

M
Miklos Szeredi 已提交
491 492 493 494 495 496 497 498 499
	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
500
	default:
M
Miklos Szeredi 已提交
501
		inode->i_op = &ovl_file_inode_operations;
502
		init_special_inode(inode, mode, rdev);
M
Miklos Szeredi 已提交
503
		break;
M
Miklos Szeredi 已提交
504 505
	}
}
M
Miklos Szeredi 已提交
506

507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550
/*
 * 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
551 552 553
	if (WARN_ON(len >= sizeof(buf)))
		return -EIO;

554 555 556 557 558 559 560 561 562 563 564 565 566 567
	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");
}

568 569 570
unsigned int ovl_get_nlink(struct dentry *lowerdentry,
			   struct dentry *upperdentry,
			   unsigned int fallback)
571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606
{
	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;
}

607
struct inode *ovl_new_inode(struct super_block *sb, umode_t mode, dev_t rdev)
M
Miklos Szeredi 已提交
608 609 610 611 612
{
	struct inode *inode;

	inode = new_inode(sb);
	if (inode)
613
		ovl_fill_inode(inode, mode, rdev, 0);
M
Miklos Szeredi 已提交
614 615 616 617 618 619

	return inode;
}

static int ovl_inode_test(struct inode *inode, void *data)
{
620
	return inode->i_private == data;
M
Miklos Szeredi 已提交
621 622 623 624
}

static int ovl_inode_set(struct inode *inode, void *data)
{
625
	inode->i_private = data;
M
Miklos Szeredi 已提交
626 627 628
	return 0;
}

629
static bool ovl_verify_inode(struct inode *inode, struct dentry *lowerdentry,
630
			     struct dentry *upperdentry, bool strict)
631
{
632 633 634 635 636 637 638
	/*
	 * 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) {
639 640 641 642 643 644 645 646 647
		/* 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;
	}

648 649 650 651 652 653 654
	/*
	 * 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))
655 656 657 658 659 660 661 662 663 664 665 666
		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;
}

667 668
struct inode *ovl_lookup_inode(struct super_block *sb, struct dentry *real,
			       bool is_upper)
669
{
670
	struct inode *inode, *key = d_inode(real);
671 672 673 674 675

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

676 677
	if (!ovl_verify_inode(inode, is_upper ? NULL : real,
			      is_upper ? real : NULL, false)) {
678 679 680 681 682 683 684
		iput(inode);
		return ERR_PTR(-ESTALE);
	}

	return inode;
}

685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
/*
 * 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;
}

718 719 720
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
721
{
722
	struct inode *realinode = upperdentry ? d_inode(upperdentry) : NULL;
M
Miklos Szeredi 已提交
723
	struct inode *inode;
724
	bool bylower = ovl_hash_bylower(sb, upperdentry, lowerdentry, index);
725
	bool is_dir;
726
	unsigned long ino = 0;
727

728 729 730
	if (!realinode)
		realinode = d_inode(lowerdentry);

731
	/*
732 733
	 * 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.
734
	 */
735
	is_dir = S_ISDIR(realinode->i_mode);
736 737 738
	if (upperdentry || bylower) {
		struct inode *key = d_inode(bylower ? lowerdentry :
						      upperdentry);
739
		unsigned int nlink = is_dir ? 1 : realinode->i_nlink;
740

741
		inode = iget5_locked(sb, (unsigned long) key,
742
				     ovl_inode_test, ovl_inode_set, key);
743
		if (!inode)
744
			goto out_nomem;
745
		if (!(inode->i_state & I_NEW)) {
746 747 748 749
			/*
			 * Verify that the underlying files stored in the inode
			 * match those in the dentry.
			 */
750 751
			if (!ovl_verify_inode(inode, lowerdentry, upperdentry,
					      true)) {
752 753 754 755 756
				iput(inode);
				inode = ERR_PTR(-ESTALE);
				goto out;
			}

757 758 759
			dput(upperdentry);
			goto out;
		}
M
Miklos Szeredi 已提交
760

761 762 763
		/* Recalculate nlink for non-dir due to indexing */
		if (!is_dir)
			nlink = ovl_get_nlink(lowerdentry, upperdentry, nlink);
764
		set_nlink(inode, nlink);
765
		ino = key->i_ino;
M
Miklos Szeredi 已提交
766
	} else {
767
		/* Lower hardlink that will be broken on copy up */
768
		inode = new_inode(sb);
M
Miklos Szeredi 已提交
769
		if (!inode)
770
			goto out_nomem;
M
Miklos Szeredi 已提交
771
	}
772
	ovl_fill_inode(inode, realinode->i_mode, realinode->i_rdev, ino);
773
	ovl_inode_init(inode, upperdentry, lowerdentry);
M
Miklos Szeredi 已提交
774 775 776 777

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

V
Vivek Goyal 已提交
778 779 780
	if (index)
		ovl_set_flag(OVL_INDEX, inode);

781
	/* Check for non-merge dir that may have whiteouts */
782
	if (is_dir) {
783
		if (((upperdentry && lowerdentry) || numlower > 1) ||
784 785 786 787 788
		    ovl_check_origin_xattr(upperdentry ?: lowerdentry)) {
			ovl_set_flag(OVL_WHITEOUTS, inode);
		}
	}

M
Miklos Szeredi 已提交
789 790 791
	if (inode->i_state & I_NEW)
		unlock_new_inode(inode);
out:
M
Miklos Szeredi 已提交
792
	return inode;
793 794 795 796

out_nomem:
	inode = ERR_PTR(-ENOMEM);
	goto out;
M
Miklos Szeredi 已提交
797
}