inode.c 22.7 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
	err = setattr_prepare(dentry, attr);
26 27 28
	if (err)
		return err;

M
Miklos Szeredi 已提交
29 30 31 32
	err = ovl_want_write(dentry);
	if (err)
		goto out;

33 34 35 36 37 38 39 40
	if (attr->ia_valid & ATTR_SIZE) {
		struct inode *realinode = d_inode(ovl_dentry_real(dentry));

		err = -ETXTBSY;
		if (atomic_read(&realinode->i_writecount) < 0)
			goto out_drop_write;
	}

41 42
	err = ovl_copy_up(dentry);
	if (!err) {
43 44
		struct inode *winode = NULL;

45 46
		upperdentry = ovl_dentry_upper(dentry);

47 48 49 50 51 52 53
		if (attr->ia_valid & ATTR_SIZE) {
			winode = d_inode(upperdentry);
			err = get_write_access(winode);
			if (err)
				goto out_drop_write;
		}

M
Miklos Szeredi 已提交
54 55 56
		if (attr->ia_valid & (ATTR_KILL_SUID|ATTR_KILL_SGID))
			attr->ia_valid &= ~ATTR_MODE;

A
Al Viro 已提交
57
		inode_lock(upperdentry->d_inode);
58
		old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
59
		err = notify_change(upperdentry, attr, NULL);
60
		revert_creds(old_cred);
61 62
		if (!err)
			ovl_copyattr(upperdentry->d_inode, dentry->d_inode);
A
Al Viro 已提交
63
		inode_unlock(upperdentry->d_inode);
64 65 66

		if (winode)
			put_write_access(winode);
M
Miklos Szeredi 已提交
67
	}
68
out_drop_write:
M
Miklos Szeredi 已提交
69 70 71 72 73
	ovl_drop_write(dentry);
out:
	return err;
}

74 75 76 77
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);
78
	unsigned int xinobits = ovl_xino_bits(dentry->d_sb);
79 80 81 82 83 84 85 86

	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;
87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
		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)) {
112 113 114 115 116 117 118 119 120 121 122
		/*
		 * 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;
123
	} else if (lower_layer && lower_layer->fsid) {
124 125 126
		/*
		 * 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
127 128 129
		 * is unique per lower fs. Upper layer uses real st_dev and
		 * lower layers use the unique anonymous bdev assigned to the
		 * lower fs.
130
		 */
131
		stat->dev = lower_layer->fs->pseudo_dev;
132 133 134 135 136
	}

	return 0;
}

137 138
int ovl_getattr(const struct path *path, struct kstat *stat,
		u32 request_mask, unsigned int flags)
M
Miklos Szeredi 已提交
139
{
140
	struct dentry *dentry = path->dentry;
141
	enum ovl_path_type type;
M
Miklos Szeredi 已提交
142
	struct path realpath;
143
	const struct cred *old_cred;
144
	bool is_dir = S_ISDIR(dentry->d_inode->i_mode);
145
	bool samefs = ovl_same_sb(dentry->d_sb);
146
	struct ovl_layer *lower_layer = NULL;
147
	int err;
M
Miklos Szeredi 已提交
148

149
	type = ovl_path_real(dentry, &realpath);
150
	old_cred = ovl_override_creds(dentry->d_sb);
151
	err = vfs_getattr(&realpath, stat, request_mask, flags);
152 153 154 155
	if (err)
		goto out;

	/*
156 157
	 * 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.
158 159
	 * With xino feature and non-samefs, we use st_ino of the copy up
	 * origin masked with high bits that represent the layer id.
160
	 *
161
	 * If lower filesystem supports NFS file handles, this also guaranties
162 163
	 * persistent st_ino across mount cycle.
	 */
164
	if (!is_dir || samefs || ovl_xino_bits(dentry->d_sb)) {
165 166 167
		if (!OVL_TYPE_UPPER(type)) {
			lower_layer = ovl_layer_lower(dentry);
		} else if (OVL_TYPE_ORIGIN(type)) {
168
			struct kstat lowerstat;
169
			u32 lowermask = STATX_INO | (!is_dir ? STATX_NLINK : 0);
170 171 172

			ovl_path_lower(dentry, &realpath);
			err = vfs_getattr(&realpath, &lowerstat,
173
					  lowermask, flags);
174 175 176 177
			if (err)
				goto out;

			/*
178
			 * Lower hardlinks may be broken on copy up to different
179 180
			 * upper files, so we cannot use the lower origin st_ino
			 * for those different files, even for the same fs case.
181 182 183 184 185 186
			 *
			 * 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.
			 *
187
			 * With inodes index enabled, it is safe to use st_ino
188 189 190
			 * 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.
191
			 */
192 193
			if (ovl_test_flag(OVL_INDEX, d_inode(dentry)) ||
			    (!ovl_verify_lower(dentry->d_sb) &&
194
			     (is_dir || lowerstat.nlink == 1))) {
195
				stat->ino = lowerstat.ino;
196
				lower_layer = ovl_layer_lower(dentry);
197
			}
198 199
		}
	}
200

201 202 203 204
	err = ovl_map_dev_ino(dentry, stat, lower_layer);
	if (err)
		goto out;

205 206 207 208 209 210 211 212
	/*
	 * 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;

213 214 215 216 217 218 219 220 221
	/*
	 * 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;

222
out:
223
	revert_creds(old_cred);
224

225
	return err;
M
Miklos Szeredi 已提交
226 227 228 229
}

int ovl_permission(struct inode *inode, int mask)
{
230 231
	struct inode *upperinode = ovl_inode_upper(inode);
	struct inode *realinode = upperinode ?: ovl_inode_lower(inode);
232
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
233 234 235 236 237
	int err;

	/* Careful in RCU walk mode */
	if (!realinode) {
		WARN_ON(!(mask & MAY_NOT_BLOCK));
238
		return -ECHILD;
M
Miklos Szeredi 已提交
239 240
	}

241 242 243 244 245 246 247 248 249
	/*
	 * 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);
250 251
	if (!upperinode &&
	    !special_file(realinode->i_mode) && mask & MAY_WRITE) {
252
		mask &= ~(MAY_WRITE | MAY_APPEND);
253 254 255
		/* Make sure mounter can read file for copy up later */
		mask |= MAY_READ;
	}
256
	err = inode_permission(realinode, mask);
257 258 259
	revert_creds(old_cred);

	return err;
M
Miklos Szeredi 已提交
260 261
}

262
static const char *ovl_get_link(struct dentry *dentry,
263 264
				struct inode *inode,
				struct delayed_call *done)
M
Miklos Szeredi 已提交
265
{
266 267
	const struct cred *old_cred;
	const char *p;
M
Miklos Szeredi 已提交
268

269 270 271
	if (!dentry)
		return ERR_PTR(-ECHILD);

272
	old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
273
	p = vfs_get_link(ovl_dentry_real(dentry), done);
274 275
	revert_creds(old_cred);
	return p;
M
Miklos Szeredi 已提交
276 277
}

M
Miklos Szeredi 已提交
278
bool ovl_is_private_xattr(const char *name)
M
Miklos Szeredi 已提交
279
{
A
Andreas Gruenbacher 已提交
280 281
	return strncmp(name, OVL_XATTR_PREFIX,
		       sizeof(OVL_XATTR_PREFIX) - 1) == 0;
M
Miklos Szeredi 已提交
282 283
}

284 285
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
286 287
{
	int err;
288 289
	struct dentry *upperdentry = ovl_i_dentry_upper(inode);
	struct dentry *realdentry = upperdentry ?: ovl_dentry_lower(dentry);
290
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
291 292 293 294 295

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

296 297
	if (!value && !upperdentry) {
		err = vfs_getxattr(realdentry, name, NULL, 0);
298 299 300 301
		if (err < 0)
			goto out_drop_write;
	}

302 303 304 305
	if (!upperdentry) {
		err = ovl_copy_up(dentry);
		if (err)
			goto out_drop_write;
M
Miklos Szeredi 已提交
306

307 308
		realdentry = ovl_dentry_upper(dentry);
	}
309

310
	old_cred = ovl_override_creds(dentry->d_sb);
311
	if (value)
312
		err = vfs_setxattr(realdentry, name, value, size, flags);
313 314
	else {
		WARN_ON(flags != XATTR_REPLACE);
315
		err = vfs_removexattr(realdentry, name);
316
	}
317
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
318

M
Miklos Szeredi 已提交
319 320 321
	/* copy c/mtime */
	ovl_copyattr(d_inode(realdentry), inode);

M
Miklos Szeredi 已提交
322 323 324 325 326 327
out_drop_write:
	ovl_drop_write(dentry);
out:
	return err;
}

328
int ovl_xattr_get(struct dentry *dentry, struct inode *inode, const char *name,
329
		  void *value, size_t size)
M
Miklos Szeredi 已提交
330
{
331 332
	ssize_t res;
	const struct cred *old_cred;
333 334
	struct dentry *realdentry =
		ovl_i_dentry_upper(inode) ?: ovl_dentry_lower(dentry);
335

336 337 338 339
	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
340 341
}

342 343 344 345 346 347 348 349 350 351
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
352 353
ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size)
{
M
Miklos Szeredi 已提交
354
	struct dentry *realdentry = ovl_dentry_real(dentry);
M
Miklos Szeredi 已提交
355
	ssize_t res;
M
Miklos Szeredi 已提交
356 357
	size_t len;
	char *s;
358
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
359

360
	old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
361
	res = vfs_listxattr(realdentry, list, size);
362
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
363 364 365 366
	if (res <= 0 || size == 0)
		return res;

	/* filter out private xattrs */
M
Miklos Szeredi 已提交
367 368
	for (s = list, len = res; len;) {
		size_t slen = strnlen(s, len) + 1;
M
Miklos Szeredi 已提交
369

M
Miklos Szeredi 已提交
370 371 372
		/* underlying fs providing us with an broken xattr list? */
		if (WARN_ON(slen > len))
			return -EIO;
M
Miklos Szeredi 已提交
373

M
Miklos Szeredi 已提交
374
		len -= slen;
375
		if (!ovl_can_list(s)) {
M
Miklos Szeredi 已提交
376
			res -= slen;
M
Miklos Szeredi 已提交
377
			memmove(s, s + slen, len);
M
Miklos Szeredi 已提交
378
		} else {
M
Miklos Szeredi 已提交
379
			s += slen;
M
Miklos Szeredi 已提交
380 381 382 383 384 385
		}
	}

	return res;
}

386 387
struct posix_acl *ovl_get_acl(struct inode *inode, int type)
{
388
	struct inode *realinode = ovl_inode_real(inode);
389 390
	const struct cred *old_cred;
	struct posix_acl *acl;
391

392
	if (!IS_ENABLED(CONFIG_FS_POSIX_ACL) || !IS_POSIXACL(realinode))
393 394
		return NULL;

395
	old_cred = ovl_override_creds(inode->i_sb);
396
	acl = get_acl(realinode, type);
397 398 399
	revert_creds(old_cred);

	return acl;
400 401
}

402
int ovl_update_time(struct inode *inode, struct timespec64 *ts, int flags)
M
Miklos Szeredi 已提交
403
{
M
Miklos Szeredi 已提交
404 405 406 407 408 409 410 411 412 413 414
	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
415 416 417 418
	}
	return 0;
}

M
Miklos Szeredi 已提交
419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
static int ovl_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
		      u64 start, u64 len)
{
	int err;
	struct inode *realinode = ovl_inode_real(inode);
	const struct cred *old_cred;

	if (!realinode->i_op->fiemap)
		return -EOPNOTSUPP;

	old_cred = ovl_override_creds(inode->i_sb);
	err = realinode->i_op->fiemap(realinode, fieinfo, start, len);
	revert_creds(old_cred);

	return err;
}

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
	.fiemap		= ovl_fiemap,
M
Miklos Szeredi 已提交
444 445 446 447
};

static const struct inode_operations ovl_symlink_inode_operations = {
	.setattr	= ovl_setattr,
448
	.get_link	= ovl_get_link,
M
Miklos Szeredi 已提交
449 450
	.getattr	= ovl_getattr,
	.listxattr	= ovl_listxattr,
M
Miklos Szeredi 已提交
451
	.update_time	= ovl_update_time,
M
Miklos Szeredi 已提交
452 453
};

M
Miklos Szeredi 已提交
454 455 456 457 458 459 460 461 462
static const struct inode_operations ovl_special_inode_operations = {
	.setattr	= ovl_setattr,
	.permission	= ovl_permission,
	.getattr	= ovl_getattr,
	.listxattr	= ovl_listxattr,
	.get_acl	= ovl_get_acl,
	.update_time	= ovl_update_time,
};

463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484
/*
 * 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];
485
	static struct lock_class_key ovl_i_lock_key[OVL_MAX_NESTING];
486 487 488 489 490 491 492 493 494 495

	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]);
496 497

	lockdep_set_class(&OVL_I(inode)->lock, &ovl_i_lock_key[depth]);
498 499 500
#endif
}

501
static void ovl_fill_inode(struct inode *inode, umode_t mode, dev_t rdev,
502
			   unsigned long ino, int fsid)
M
Miklos Szeredi 已提交
503
{
504 505
	int xinobits = ovl_xino_bits(inode->i_sb);

506 507
	/*
	 * When NFS export is enabled and d_ino is consistent with st_ino
508 509 510 511
	 * (samefs or i_ino has enough bits to encode layer), set the same
	 * value used for d_ino 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
512 513
	 * upper inode i_ino on ovl_inode_init() or ovl_inode_update().
	 */
514 515
	if (inode->i_sb->s_export_op &&
	    (ovl_same_sb(inode->i_sb) || xinobits)) {
516
		inode->i_ino = ino;
517 518 519
		if (xinobits && fsid && !(ino >> (64 - xinobits)))
			inode->i_ino |= (unsigned long)fsid << (64 - xinobits);
	} else {
520
		inode->i_ino = get_next_ino();
521
	}
M
Miklos Szeredi 已提交
522
	inode->i_mode = mode;
M
Miklos Szeredi 已提交
523
	inode->i_flags |= S_NOCMTIME;
524 525 526
#ifdef CONFIG_FS_POSIX_ACL
	inode->i_acl = inode->i_default_acl = ACL_DONT_CACHE;
#endif
M
Miklos Szeredi 已提交
527

528 529
	ovl_lockdep_annotate_inode_mutex_key(inode);

530 531 532
	switch (mode & S_IFMT) {
	case S_IFREG:
		inode->i_op = &ovl_file_inode_operations;
M
Miklos Szeredi 已提交
533
		inode->i_fop = &ovl_file_operations;
534 535
		break;

M
Miklos Szeredi 已提交
536 537 538 539 540 541 542 543 544
	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
545
	default:
M
Miklos Szeredi 已提交
546
		inode->i_op = &ovl_special_inode_operations;
547
		init_special_inode(inode, mode, rdev);
M
Miklos Szeredi 已提交
548
		break;
M
Miklos Szeredi 已提交
549 550
	}
}
M
Miklos Szeredi 已提交
551

552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
/*
 * 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
596 597 598
	if (WARN_ON(len >= sizeof(buf)))
		return -EIO;

599 600 601 602 603 604 605 606 607 608 609 610 611 612
	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");
}

613 614 615
unsigned int ovl_get_nlink(struct dentry *lowerdentry,
			   struct dentry *upperdentry,
			   unsigned int fallback)
616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651
{
	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;
}

652
struct inode *ovl_new_inode(struct super_block *sb, umode_t mode, dev_t rdev)
M
Miklos Szeredi 已提交
653 654 655 656 657
{
	struct inode *inode;

	inode = new_inode(sb);
	if (inode)
658
		ovl_fill_inode(inode, mode, rdev, 0, 0);
M
Miklos Szeredi 已提交
659 660 661 662 663 664

	return inode;
}

static int ovl_inode_test(struct inode *inode, void *data)
{
665
	return inode->i_private == data;
M
Miklos Szeredi 已提交
666 667 668 669
}

static int ovl_inode_set(struct inode *inode, void *data)
{
670
	inode->i_private = data;
M
Miklos Szeredi 已提交
671 672 673
	return 0;
}

674
static bool ovl_verify_inode(struct inode *inode, struct dentry *lowerdentry,
675
			     struct dentry *upperdentry, bool strict)
676
{
677 678 679 680 681 682 683
	/*
	 * 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) {
684 685 686 687 688 689 690 691 692
		/* 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;
	}

693 694 695 696 697 698 699
	/*
	 * 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))
700 701 702 703 704 705 706 707 708 709 710 711
		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;
}

712 713
struct inode *ovl_lookup_inode(struct super_block *sb, struct dentry *real,
			       bool is_upper)
714
{
715
	struct inode *inode, *key = d_inode(real);
716 717 718 719 720

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

721 722
	if (!ovl_verify_inode(inode, is_upper ? NULL : real,
			      is_upper ? real : NULL, false)) {
723 724 725 726 727 728 729
		iput(inode);
		return ERR_PTR(-ESTALE);
	}

	return inode;
}

730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762
/*
 * 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;
}

763 764 765 766 767 768 769 770 771
static struct inode *ovl_iget5(struct super_block *sb, struct inode *newinode,
			       struct inode *key)
{
	return newinode ? inode_insert5(newinode, (unsigned long) key,
					 ovl_inode_test, ovl_inode_set, key) :
			  iget5_locked(sb, (unsigned long) key,
				       ovl_inode_test, ovl_inode_set, key);
}

772 773
struct inode *ovl_get_inode(struct super_block *sb,
			    struct ovl_inode_params *oip)
M
Miklos Szeredi 已提交
774
{
775 776
	struct dentry *upperdentry = oip->upperdentry;
	struct ovl_path *lowerpath = oip->lowerpath;
777
	struct inode *realinode = upperdentry ? d_inode(upperdentry) : NULL;
M
Miklos Szeredi 已提交
778
	struct inode *inode;
779
	struct dentry *lowerdentry = lowerpath ? lowerpath->dentry : NULL;
780 781 782
	bool bylower = ovl_hash_bylower(sb, upperdentry, lowerdentry,
					oip->index);
	int fsid = bylower ? oip->lowerpath->layer->fsid : 0;
783
	bool is_dir;
784
	unsigned long ino = 0;
785

786 787 788
	if (!realinode)
		realinode = d_inode(lowerdentry);

789
	/*
790 791
	 * 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.
792
	 */
793
	is_dir = S_ISDIR(realinode->i_mode);
794 795 796
	if (upperdentry || bylower) {
		struct inode *key = d_inode(bylower ? lowerdentry :
						      upperdentry);
797
		unsigned int nlink = is_dir ? 1 : realinode->i_nlink;
798

799
		inode = ovl_iget5(sb, oip->newinode, key);
800
		if (!inode)
801
			goto out_nomem;
802
		if (!(inode->i_state & I_NEW)) {
803 804 805 806
			/*
			 * Verify that the underlying files stored in the inode
			 * match those in the dentry.
			 */
807 808
			if (!ovl_verify_inode(inode, lowerdentry, upperdentry,
					      true)) {
809 810 811 812 813
				iput(inode);
				inode = ERR_PTR(-ESTALE);
				goto out;
			}

814
			dput(upperdentry);
815
			kfree(oip->redirect);
816 817
			goto out;
		}
M
Miklos Szeredi 已提交
818

819 820 821
		/* Recalculate nlink for non-dir due to indexing */
		if (!is_dir)
			nlink = ovl_get_nlink(lowerdentry, upperdentry, nlink);
822
		set_nlink(inode, nlink);
823
		ino = key->i_ino;
M
Miklos Szeredi 已提交
824
	} else {
825
		/* Lower hardlink that will be broken on copy up */
826
		inode = new_inode(sb);
M
Miklos Szeredi 已提交
827
		if (!inode)
828
			goto out_nomem;
M
Miklos Szeredi 已提交
829
	}
830
	ovl_fill_inode(inode, realinode->i_mode, realinode->i_rdev, ino, fsid);
831
	ovl_inode_init(inode, upperdentry, lowerdentry);
M
Miklos Szeredi 已提交
832 833 834 835

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

836
	if (oip->index)
V
Vivek Goyal 已提交
837 838
		ovl_set_flag(OVL_INDEX, inode);

839 840
	OVL_I(inode)->redirect = oip->redirect;

841
	/* Check for non-merge dir that may have whiteouts */
842
	if (is_dir) {
843
		if (((upperdentry && lowerdentry) || oip->numlower > 1) ||
844 845 846 847 848
		    ovl_check_origin_xattr(upperdentry ?: lowerdentry)) {
			ovl_set_flag(OVL_WHITEOUTS, inode);
		}
	}

M
Miklos Szeredi 已提交
849 850 851
	if (inode->i_state & I_NEW)
		unlock_new_inode(inode);
out:
M
Miklos Szeredi 已提交
852
	return inode;
853 854 855 856

out_nomem:
	inode = ERR_PTR(-ENOMEM);
	goto out;
M
Miklos Szeredi 已提交
857
}