dir.c 22.0 KB
Newer Older
M
Miklos Szeredi 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/*
 *
 * 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/namei.h>
#include <linux/xattr.h>
#include <linux/security.h>
#include <linux/cred.h>
15 16
#include <linux/posix_acl.h>
#include <linux/posix_acl_xattr.h>
17
#include <linux/atomic.h>
M
Miklos Szeredi 已提交
18 19 20 21 22 23 24
#include "overlayfs.h"

void ovl_cleanup(struct inode *wdir, struct dentry *wdentry)
{
	int err;

	dget(wdentry);
25
	if (d_is_dir(wdentry))
M
Miklos Szeredi 已提交
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
		err = ovl_do_rmdir(wdir, wdentry);
	else
		err = ovl_do_unlink(wdir, wdentry);
	dput(wdentry);

	if (err) {
		pr_err("overlayfs: cleanup of '%pd2' failed (%i)\n",
		       wdentry, err);
	}
}

struct dentry *ovl_lookup_temp(struct dentry *workdir, struct dentry *dentry)
{
	struct dentry *temp;
	char name[20];
41
	static atomic_t temp_id = ATOMIC_INIT(0);
M
Miklos Szeredi 已提交
42

43 44
	/* counter is allowed to wrap, since temp dentries are ephemeral */
	snprintf(name, sizeof(name), "#%x", atomic_inc_return(&temp_id));
M
Miklos Szeredi 已提交
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125

	temp = lookup_one_len(name, workdir, strlen(name));
	if (!IS_ERR(temp) && temp->d_inode) {
		pr_err("overlayfs: workdir/%s already exists\n", name);
		dput(temp);
		temp = ERR_PTR(-EIO);
	}

	return temp;
}

/* caller holds i_mutex on workdir */
static struct dentry *ovl_whiteout(struct dentry *workdir,
				   struct dentry *dentry)
{
	int err;
	struct dentry *whiteout;
	struct inode *wdir = workdir->d_inode;

	whiteout = ovl_lookup_temp(workdir, dentry);
	if (IS_ERR(whiteout))
		return whiteout;

	err = ovl_do_whiteout(wdir, whiteout);
	if (err) {
		dput(whiteout);
		whiteout = ERR_PTR(err);
	}

	return whiteout;
}

int ovl_create_real(struct inode *dir, struct dentry *newdentry,
		    struct kstat *stat, const char *link,
		    struct dentry *hardlink, bool debug)
{
	int err;

	if (newdentry->d_inode)
		return -ESTALE;

	if (hardlink) {
		err = ovl_do_link(hardlink, dir, newdentry, debug);
	} else {
		switch (stat->mode & S_IFMT) {
		case S_IFREG:
			err = ovl_do_create(dir, newdentry, stat->mode, debug);
			break;

		case S_IFDIR:
			err = ovl_do_mkdir(dir, newdentry, stat->mode, debug);
			break;

		case S_IFCHR:
		case S_IFBLK:
		case S_IFIFO:
		case S_IFSOCK:
			err = ovl_do_mknod(dir, newdentry,
					   stat->mode, stat->rdev, debug);
			break;

		case S_IFLNK:
			err = ovl_do_symlink(dir, newdentry, link, debug);
			break;

		default:
			err = -EPERM;
		}
	}
	if (!err && WARN_ON(!newdentry->d_inode)) {
		/*
		 * Not quite sure if non-instantiated dentry is legal or not.
		 * VFS doesn't seem to care so check and warn here.
		 */
		err = -ENOENT;
	}
	return err;
}

static int ovl_set_opaque(struct dentry *upperdentry)
{
H
hujianyang 已提交
126
	return ovl_do_setxattr(upperdentry, OVL_XATTR_OPAQUE, "y", 1, 0);
M
Miklos Szeredi 已提交
127 128 129 130 131 132
}

static void ovl_remove_opaque(struct dentry *upperdentry)
{
	int err;

H
hujianyang 已提交
133
	err = ovl_do_removexattr(upperdentry, OVL_XATTR_OPAQUE);
M
Miklos Szeredi 已提交
134 135 136 137 138 139 140 141 142 143 144 145
	if (err) {
		pr_warn("overlayfs: failed to remove opaque from '%s' (%i)\n",
			upperdentry->d_name.name, err);
	}
}

static int ovl_dir_getattr(struct vfsmount *mnt, struct dentry *dentry,
			 struct kstat *stat)
{
	int err;
	enum ovl_path_type type;
	struct path realpath;
146
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
147 148

	type = ovl_path_real(dentry, &realpath);
149
	old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
150
	err = vfs_getattr(&realpath, stat);
151
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
152 153 154 155 156 157 158 159 160 161 162
	if (err)
		return err;

	stat->dev = dentry->d_sb->s_dev;
	stat->ino = dentry->d_inode->i_ino;

	/*
	 * It's probably not worth it to count subdirs to get the
	 * correct link count.  nlink=1 seems to pacify 'find' and
	 * other utilities.
	 */
M
Miklos Szeredi 已提交
163
	if (OVL_TYPE_MERGE(type))
M
Miklos Szeredi 已提交
164 165 166 167 168
		stat->nlink = 1;

	return 0;
}

169 170
/* Common operations required to be done after creation of file on upper */
static void ovl_instantiate(struct dentry *dentry, struct inode *inode,
M
Miklos Szeredi 已提交
171
			    struct dentry *newdentry, bool hardlink)
172 173 174
{
	ovl_dentry_version_inc(dentry->d_parent);
	ovl_dentry_update(dentry, newdentry);
M
Miklos Szeredi 已提交
175 176 177 178 179 180 181
	if (!hardlink) {
		ovl_inode_update(inode, d_inode(newdentry));
		ovl_copyattr(newdentry->d_inode, inode);
	} else {
		WARN_ON(ovl_inode_real(inode, NULL) != d_inode(newdentry));
		inc_nlink(inode);
	}
182 183 184
	d_instantiate(dentry, inode);
}

M
Miklos Szeredi 已提交
185 186 187 188 189 190 191 192 193
static int ovl_create_upper(struct dentry *dentry, struct inode *inode,
			    struct kstat *stat, const char *link,
			    struct dentry *hardlink)
{
	struct dentry *upperdir = ovl_dentry_upper(dentry->d_parent);
	struct inode *udir = upperdir->d_inode;
	struct dentry *newdentry;
	int err;

194 195 196
	if (!hardlink && !IS_POSIXACL(udir))
		stat->mode &= ~current_umask();

A
Al Viro 已提交
197
	inode_lock_nested(udir, I_MUTEX_PARENT);
M
Miklos Szeredi 已提交
198 199 200 201 202 203 204 205 206
	newdentry = lookup_one_len(dentry->d_name.name, upperdir,
				   dentry->d_name.len);
	err = PTR_ERR(newdentry);
	if (IS_ERR(newdentry))
		goto out_unlock;
	err = ovl_create_real(udir, newdentry, stat, link, hardlink, false);
	if (err)
		goto out_dput;

M
Miklos Szeredi 已提交
207
	ovl_instantiate(dentry, inode, newdentry, !!hardlink);
M
Miklos Szeredi 已提交
208 209 210 211
	newdentry = NULL;
out_dput:
	dput(newdentry);
out_unlock:
A
Al Viro 已提交
212
	inode_unlock(udir);
M
Miklos Szeredi 已提交
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248
	return err;
}

static int ovl_lock_rename_workdir(struct dentry *workdir,
				   struct dentry *upperdir)
{
	/* Workdir should not be the same as upperdir */
	if (workdir == upperdir)
		goto err;

	/* Workdir should not be subdir of upperdir and vice versa */
	if (lock_rename(workdir, upperdir) != NULL)
		goto err_unlock;

	return 0;

err_unlock:
	unlock_rename(workdir, upperdir);
err:
	pr_err("overlayfs: failed to lock workdir+upperdir\n");
	return -EIO;
}

static struct dentry *ovl_clear_empty(struct dentry *dentry,
				      struct list_head *list)
{
	struct dentry *workdir = ovl_workdir(dentry);
	struct inode *wdir = workdir->d_inode;
	struct dentry *upperdir = ovl_dentry_upper(dentry->d_parent);
	struct inode *udir = upperdir->d_inode;
	struct path upperpath;
	struct dentry *upper;
	struct dentry *opaquedir;
	struct kstat stat;
	int err;

249 250 251
	if (WARN_ON(!workdir))
		return ERR_PTR(-EROFS);

M
Miklos Szeredi 已提交
252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
	err = ovl_lock_rename_workdir(workdir, upperdir);
	if (err)
		goto out;

	ovl_path_upper(dentry, &upperpath);
	err = vfs_getattr(&upperpath, &stat);
	if (err)
		goto out_unlock;

	err = -ESTALE;
	if (!S_ISDIR(stat.mode))
		goto out_unlock;
	upper = upperpath.dentry;
	if (upper->d_parent->d_inode != udir)
		goto out_unlock;

	opaquedir = ovl_lookup_temp(workdir, dentry);
	err = PTR_ERR(opaquedir);
	if (IS_ERR(opaquedir))
		goto out_unlock;

	err = ovl_create_real(wdir, opaquedir, &stat, NULL, NULL, true);
	if (err)
		goto out_dput;

	err = ovl_copy_xattr(upper, opaquedir);
	if (err)
		goto out_cleanup;

	err = ovl_set_opaque(opaquedir);
	if (err)
		goto out_cleanup;

A
Al Viro 已提交
285
	inode_lock(opaquedir->d_inode);
M
Miklos Szeredi 已提交
286
	err = ovl_set_attr(opaquedir, &stat);
A
Al Viro 已提交
287
	inode_unlock(opaquedir->d_inode);
M
Miklos Szeredi 已提交
288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313
	if (err)
		goto out_cleanup;

	err = ovl_do_rename(wdir, opaquedir, udir, upper, RENAME_EXCHANGE);
	if (err)
		goto out_cleanup;

	ovl_cleanup_whiteouts(upper, list);
	ovl_cleanup(wdir, upper);
	unlock_rename(workdir, upperdir);

	/* dentry's upper doesn't match now, get rid of it */
	d_drop(dentry);

	return opaquedir;

out_cleanup:
	ovl_cleanup(wdir, opaquedir);
out_dput:
	dput(opaquedir);
out_unlock:
	unlock_rename(workdir, upperdir);
out:
	return ERR_PTR(err);
}

M
Miklos Szeredi 已提交
314
static struct dentry *ovl_check_empty_and_clear(struct dentry *dentry)
M
Miklos Szeredi 已提交
315 316 317
{
	int err;
	struct dentry *ret = NULL;
M
Miklos Szeredi 已提交
318
	enum ovl_path_type type = ovl_path_type(dentry);
M
Miklos Szeredi 已提交
319 320 321
	LIST_HEAD(list);

	err = ovl_check_empty_dir(dentry, &list);
M
Miklos Szeredi 已提交
322
	if (err) {
M
Miklos Szeredi 已提交
323
		ret = ERR_PTR(err);
M
Miklos Szeredi 已提交
324
		goto out_free;
M
Miklos Szeredi 已提交
325
	}
M
Miklos Szeredi 已提交
326

M
Miklos Szeredi 已提交
327 328 329 330 331 332 333 334 335 336 337 338 339 340
	/*
	 * When removing an empty opaque directory, then it makes no sense to
	 * replace it with an exact replica of itself.
	 *
	 * If no upperdentry then skip clearing whiteouts.
	 *
	 * Can race with copy-up, since we don't hold the upperdir mutex.
	 * Doesn't matter, since copy-up can't create a non-empty directory
	 * from an empty one.
	 */
	if (OVL_TYPE_UPPER(type) && OVL_TYPE_MERGE(type))
		ret = ovl_clear_empty(dentry, &list);

out_free:
M
Miklos Szeredi 已提交
341 342 343 344 345
	ovl_cache_free(&list);

	return ret;
}

346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371
static int ovl_set_upper_acl(struct dentry *upperdentry, const char *name,
			     const struct posix_acl *acl)
{
	void *buffer;
	size_t size;
	int err;

	if (!IS_ENABLED(CONFIG_FS_POSIX_ACL) || !acl)
		return 0;

	size = posix_acl_to_xattr(NULL, acl, NULL, 0);
	buffer = kmalloc(size, GFP_KERNEL);
	if (!buffer)
		return -ENOMEM;

	size = posix_acl_to_xattr(&init_user_ns, acl, buffer, size);
	err = size;
	if (err < 0)
		goto out_free;

	err = vfs_setxattr(upperdentry, name, buffer, size, XATTR_CREATE);
out_free:
	kfree(buffer);
	return err;
}

M
Miklos Szeredi 已提交
372 373 374 375 376 377 378 379 380 381 382
static int ovl_create_over_whiteout(struct dentry *dentry, struct inode *inode,
				    struct kstat *stat, const char *link,
				    struct dentry *hardlink)
{
	struct dentry *workdir = ovl_workdir(dentry);
	struct inode *wdir = workdir->d_inode;
	struct dentry *upperdir = ovl_dentry_upper(dentry->d_parent);
	struct inode *udir = upperdir->d_inode;
	struct dentry *upper;
	struct dentry *newdentry;
	int err;
383
	struct posix_acl *acl, *default_acl;
M
Miklos Szeredi 已提交
384

385 386 387
	if (WARN_ON(!workdir))
		return -EROFS;

388 389 390 391 392 393 394
	if (!hardlink) {
		err = posix_acl_create(dentry->d_parent->d_inode,
				       &stat->mode, &default_acl, &acl);
		if (err)
			return err;
	}

M
Miklos Szeredi 已提交
395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
	err = ovl_lock_rename_workdir(workdir, upperdir);
	if (err)
		goto out;

	newdentry = ovl_lookup_temp(workdir, dentry);
	err = PTR_ERR(newdentry);
	if (IS_ERR(newdentry))
		goto out_unlock;

	upper = lookup_one_len(dentry->d_name.name, upperdir,
			       dentry->d_name.len);
	err = PTR_ERR(upper);
	if (IS_ERR(upper))
		goto out_dput;

	err = ovl_create_real(wdir, newdentry, stat, link, hardlink, true);
	if (err)
		goto out_dput2;

M
Miklos Szeredi 已提交
414 415 416
	/*
	 * mode could have been mutilated due to umask (e.g. sgid directory)
	 */
M
Miklos Szeredi 已提交
417 418
	if (!hardlink &&
	    !S_ISLNK(stat->mode) && newdentry->d_inode->i_mode != stat->mode) {
M
Miklos Szeredi 已提交
419 420 421 422 423 424 425 426 427 428
		struct iattr attr = {
			.ia_valid = ATTR_MODE,
			.ia_mode = stat->mode,
		};
		inode_lock(newdentry->d_inode);
		err = notify_change(newdentry, &attr, NULL);
		inode_unlock(newdentry->d_inode);
		if (err)
			goto out_cleanup;
	}
429 430 431 432 433 434 435 436 437 438 439
	if (!hardlink) {
		err = ovl_set_upper_acl(newdentry, XATTR_NAME_POSIX_ACL_ACCESS,
					acl);
		if (err)
			goto out_cleanup;

		err = ovl_set_upper_acl(newdentry, XATTR_NAME_POSIX_ACL_DEFAULT,
					default_acl);
		if (err)
			goto out_cleanup;
	}
M
Miklos Szeredi 已提交
440

M
Miklos Szeredi 已提交
441
	if (!hardlink && S_ISDIR(stat->mode)) {
M
Miklos Szeredi 已提交
442 443 444 445 446 447 448 449 450 451 452 453 454 455 456
		err = ovl_set_opaque(newdentry);
		if (err)
			goto out_cleanup;

		err = ovl_do_rename(wdir, newdentry, udir, upper,
				    RENAME_EXCHANGE);
		if (err)
			goto out_cleanup;

		ovl_cleanup(wdir, upper);
	} else {
		err = ovl_do_rename(wdir, newdentry, udir, upper, 0);
		if (err)
			goto out_cleanup;
	}
M
Miklos Szeredi 已提交
457
	ovl_instantiate(dentry, inode, newdentry, !!hardlink);
M
Miklos Szeredi 已提交
458 459 460 461 462 463 464 465
	newdentry = NULL;
out_dput2:
	dput(upper);
out_dput:
	dput(newdentry);
out_unlock:
	unlock_rename(workdir, upperdir);
out:
466 467 468 469
	if (!hardlink) {
		posix_acl_release(acl);
		posix_acl_release(default_acl);
	}
M
Miklos Szeredi 已提交
470 471 472 473 474 475 476
	return err;

out_cleanup:
	ovl_cleanup(wdir, newdentry);
	goto out_dput2;
}

M
Miklos Szeredi 已提交
477 478 479
static int ovl_create_or_link(struct dentry *dentry, struct inode *inode,
			      struct kstat *stat, const char *link,
			      struct dentry *hardlink)
M
Miklos Szeredi 已提交
480 481
{
	int err;
482 483
	const struct cred *old_cred;
	struct cred *override_cred;
M
Miklos Szeredi 已提交
484 485 486

	err = ovl_copy_up(dentry->d_parent);
	if (err)
M
Miklos Szeredi 已提交
487
		return err;
M
Miklos Szeredi 已提交
488

489 490 491 492
	old_cred = ovl_override_creds(dentry->d_sb);
	err = -ENOMEM;
	override_cred = prepare_creds();
	if (override_cred) {
M
Miklos Szeredi 已提交
493 494
		override_cred->fsuid = inode->i_uid;
		override_cred->fsgid = inode->i_gid;
495 496 497 498 499 500 501 502 503
		if (!hardlink) {
			err = security_dentry_create_files_as(dentry,
					stat->mode, &dentry->d_name, old_cred,
					override_cred);
			if (err) {
				put_cred(override_cred);
				goto out_revert_creds;
			}
		}
504 505 506
		put_cred(override_creds(override_cred));
		put_cred(override_cred);

M
Miklos Szeredi 已提交
507
		if (!ovl_dentry_is_whiteout(dentry))
M
Miklos Szeredi 已提交
508
			err = ovl_create_upper(dentry, inode, stat, link,
509 510
						hardlink);
		else
M
Miklos Szeredi 已提交
511
			err = ovl_create_over_whiteout(dentry, inode, stat,
512
							link, hardlink);
M
Miklos Szeredi 已提交
513
	}
514
out_revert_creds:
515
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
516 517 518 519 520 521 522
	if (!err) {
		struct inode *realinode = d_inode(ovl_dentry_upper(dentry));

		WARN_ON(inode->i_mode != realinode->i_mode);
		WARN_ON(!uid_eq(inode->i_uid, realinode->i_uid));
		WARN_ON(!gid_eq(inode->i_gid, realinode->i_gid));
	}
M
Miklos Szeredi 已提交
523 524 525 526 527 528 529
	return err;
}

static int ovl_create_object(struct dentry *dentry, int mode, dev_t rdev,
			     const char *link)
{
	int err;
M
Miklos Szeredi 已提交
530 531 532 533
	struct inode *inode;
	struct kstat stat = {
		.rdev = rdev,
	};
M
Miklos Szeredi 已提交
534 535

	err = ovl_want_write(dentry);
M
Miklos Szeredi 已提交
536 537 538 539
	if (err)
		goto out;

	err = -ENOMEM;
540
	inode = ovl_new_inode(dentry->d_sb, mode, rdev);
M
Miklos Szeredi 已提交
541 542 543 544 545 546 547 548 549
	if (!inode)
		goto out_drop_write;

	inode_init_owner(inode, dentry->d_parent->d_inode, mode);
	stat.mode = inode->i_mode;

	err = ovl_create_or_link(dentry, inode, &stat, link, NULL);
	if (err)
		iput(inode);
M
Miklos Szeredi 已提交
550

M
Miklos Szeredi 已提交
551 552 553
out_drop_write:
	ovl_drop_write(dentry);
out:
M
Miklos Szeredi 已提交
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
	return err;
}

static int ovl_create(struct inode *dir, struct dentry *dentry, umode_t mode,
		      bool excl)
{
	return ovl_create_object(dentry, (mode & 07777) | S_IFREG, 0, NULL);
}

static int ovl_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
{
	return ovl_create_object(dentry, (mode & 07777) | S_IFDIR, 0, NULL);
}

static int ovl_mknod(struct inode *dir, struct dentry *dentry, umode_t mode,
		     dev_t rdev)
{
	/* Don't allow creation of "whiteout" on overlay */
	if (S_ISCHR(mode) && rdev == WHITEOUT_DEV)
		return -EPERM;

	return ovl_create_object(dentry, mode, rdev, NULL);
}

static int ovl_symlink(struct inode *dir, struct dentry *dentry,
		       const char *link)
{
	return ovl_create_object(dentry, S_IFLNK, 0, link);
}

static int ovl_link(struct dentry *old, struct inode *newdir,
		    struct dentry *new)
{
	int err;
M
Miklos Szeredi 已提交
588
	struct inode *inode;
M
Miklos Szeredi 已提交
589 590 591 592 593 594 595 596 597

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

	err = ovl_copy_up(old);
	if (err)
		goto out_drop_write;

M
Miklos Szeredi 已提交
598 599 600 601 602 603
	inode = d_inode(old);
	ihold(inode);

	err = ovl_create_or_link(new, inode, NULL, NULL, ovl_dentry_upper(old));
	if (err)
		iput(inode);
M
Miklos Szeredi 已提交
604 605 606 607 608 609 610

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

M
Miklos Szeredi 已提交
611
static int ovl_remove_and_whiteout(struct dentry *dentry, bool is_dir)
M
Miklos Szeredi 已提交
612 613 614 615 616 617 618 619 620
{
	struct dentry *workdir = ovl_workdir(dentry);
	struct inode *wdir = workdir->d_inode;
	struct dentry *upperdir = ovl_dentry_upper(dentry->d_parent);
	struct inode *udir = upperdir->d_inode;
	struct dentry *whiteout;
	struct dentry *upper;
	struct dentry *opaquedir = NULL;
	int err;
621
	int flags = 0;
M
Miklos Szeredi 已提交
622

623 624 625
	if (WARN_ON(!workdir))
		return -EROFS;

626
	if (is_dir) {
M
Miklos Szeredi 已提交
627 628 629 630
		opaquedir = ovl_check_empty_and_clear(dentry);
		err = PTR_ERR(opaquedir);
		if (IS_ERR(opaquedir))
			goto out;
M
Miklos Szeredi 已提交
631 632 633 634 635 636
	}

	err = ovl_lock_rename_workdir(workdir, upperdir);
	if (err)
		goto out_dput;

637 638 639 640
	upper = lookup_one_len(dentry->d_name.name, upperdir,
			       dentry->d_name.len);
	err = PTR_ERR(upper);
	if (IS_ERR(upper))
M
Miklos Szeredi 已提交
641 642
		goto out_unlock;

643 644 645 646 647 648
	err = -ESTALE;
	if ((opaquedir && upper != opaquedir) ||
	    (!opaquedir && ovl_dentry_upper(dentry) &&
	     upper != ovl_dentry_upper(dentry))) {
		goto out_dput_upper;
	}
M
Miklos Szeredi 已提交
649

650 651 652 653
	whiteout = ovl_whiteout(workdir, dentry);
	err = PTR_ERR(whiteout);
	if (IS_ERR(whiteout))
		goto out_dput_upper;
M
Miklos Szeredi 已提交
654

655 656
	if (d_is_dir(upper))
		flags = RENAME_EXCHANGE;
M
Miklos Szeredi 已提交
657

658 659 660 661 662
	err = ovl_do_rename(wdir, whiteout, udir, upper, flags);
	if (err)
		goto kill_whiteout;
	if (flags)
		ovl_cleanup(wdir, upper);
M
Miklos Szeredi 已提交
663 664 665 666 667

	ovl_dentry_version_inc(dentry->d_parent);
out_d_drop:
	d_drop(dentry);
	dput(whiteout);
668 669
out_dput_upper:
	dput(upper);
M
Miklos Szeredi 已提交
670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685
out_unlock:
	unlock_rename(workdir, upperdir);
out_dput:
	dput(opaquedir);
out:
	return err;

kill_whiteout:
	ovl_cleanup(wdir, whiteout);
	goto out_d_drop;
}

static int ovl_remove_upper(struct dentry *dentry, bool is_dir)
{
	struct dentry *upperdir = ovl_dentry_upper(dentry->d_parent);
	struct inode *dir = upperdir->d_inode;
686
	struct dentry *upper;
M
Miklos Szeredi 已提交
687 688
	int err;

A
Al Viro 已提交
689
	inode_lock_nested(dir, I_MUTEX_PARENT);
690 691 692 693 694 695
	upper = lookup_one_len(dentry->d_name.name, upperdir,
			       dentry->d_name.len);
	err = PTR_ERR(upper);
	if (IS_ERR(upper))
		goto out_unlock;

M
Miklos Szeredi 已提交
696
	err = -ESTALE;
697
	if (upper == ovl_dentry_upper(dentry)) {
M
Miklos Szeredi 已提交
698 699 700 701 702 703
		if (is_dir)
			err = vfs_rmdir(dir, upper);
		else
			err = vfs_unlink(dir, upper, NULL);
		ovl_dentry_version_inc(dentry->d_parent);
	}
704
	dput(upper);
M
Miklos Szeredi 已提交
705 706 707 708 709 710 711

	/*
	 * Keeping this dentry hashed would mean having to release
	 * upperpath/lowerpath, which could only be done if we are the
	 * sole user of this dentry.  Too tricky...  Just unhash for
	 * now.
	 */
712 713
	if (!err)
		d_drop(dentry);
714
out_unlock:
A
Al Viro 已提交
715
	inode_unlock(dir);
M
Miklos Szeredi 已提交
716 717 718 719 720 721 722 723

	return err;
}

static int ovl_do_remove(struct dentry *dentry, bool is_dir)
{
	enum ovl_path_type type;
	int err;
724 725
	const struct cred *old_cred;

M
Miklos Szeredi 已提交
726 727 728 729 730 731 732 733 734 735
	err = ovl_want_write(dentry);
	if (err)
		goto out;

	err = ovl_copy_up(dentry->d_parent);
	if (err)
		goto out_drop_write;

	type = ovl_path_type(dentry);

736 737 738 739
	old_cred = ovl_override_creds(dentry->d_sb);
	if (OVL_TYPE_PURE_UPPER(type))
		err = ovl_remove_upper(dentry, is_dir);
	else
M
Miklos Szeredi 已提交
740
		err = ovl_remove_and_whiteout(dentry, is_dir);
741
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
742 743 744 745 746 747
	if (!err) {
		if (is_dir)
			clear_nlink(dentry->d_inode);
		else
			drop_nlink(dentry->d_inode);
	}
M
Miklos Szeredi 已提交
748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
out_drop_write:
	ovl_drop_write(dentry);
out:
	return err;
}

static int ovl_unlink(struct inode *dir, struct dentry *dentry)
{
	return ovl_do_remove(dentry, false);
}

static int ovl_rmdir(struct inode *dir, struct dentry *dentry)
{
	return ovl_do_remove(dentry, true);
}

764 765 766
static int ovl_rename(struct inode *olddir, struct dentry *old,
		      struct inode *newdir, struct dentry *new,
		      unsigned int flags)
M
Miklos Szeredi 已提交
767 768 769 770 771 772 773 774 775 776 777 778 779
{
	int err;
	enum ovl_path_type old_type;
	enum ovl_path_type new_type;
	struct dentry *old_upperdir;
	struct dentry *new_upperdir;
	struct dentry *olddentry;
	struct dentry *newdentry;
	struct dentry *trap;
	bool old_opaque;
	bool new_opaque;
	bool cleanup_whiteout = false;
	bool overwrite = !(flags & RENAME_EXCHANGE);
780
	bool is_dir = d_is_dir(old);
M
Miklos Szeredi 已提交
781 782 783 784 785 786 787 788 789 790 791 792 793
	bool new_is_dir = false;
	struct dentry *opaquedir = NULL;
	const struct cred *old_cred = NULL;

	err = -EINVAL;
	if (flags & ~(RENAME_EXCHANGE | RENAME_NOREPLACE))
		goto out;

	flags &= ~RENAME_NOREPLACE;

	/* Don't copy up directory trees */
	old_type = ovl_path_type(old);
	err = -EXDEV;
M
Miklos Szeredi 已提交
794
	if (OVL_TYPE_MERGE_OR_LOWER(old_type) && is_dir)
M
Miklos Szeredi 已提交
795 796 797
		goto out;

	if (new->d_inode) {
798
		if (d_is_dir(new))
M
Miklos Szeredi 已提交
799 800 801 802
			new_is_dir = true;

		new_type = ovl_path_type(new);
		err = -EXDEV;
M
Miklos Szeredi 已提交
803
		if (!overwrite && OVL_TYPE_MERGE_OR_LOWER(new_type) && new_is_dir)
M
Miklos Szeredi 已提交
804 805 806
			goto out;
	} else {
		if (ovl_dentry_is_opaque(new))
M
Miklos Szeredi 已提交
807
			new_type = __OVL_PATH_UPPER;
M
Miklos Szeredi 已提交
808
		else
M
Miklos Szeredi 已提交
809
			new_type = __OVL_PATH_UPPER | __OVL_PATH_PURE;
M
Miklos Szeredi 已提交
810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828
	}

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

	err = ovl_copy_up(old);
	if (err)
		goto out_drop_write;

	err = ovl_copy_up(new->d_parent);
	if (err)
		goto out_drop_write;
	if (!overwrite) {
		err = ovl_copy_up(new);
		if (err)
			goto out_drop_write;
	}

M
Miklos Szeredi 已提交
829 830
	old_opaque = !OVL_TYPE_PURE_UPPER(old_type);
	new_opaque = !OVL_TYPE_PURE_UPPER(new_type);
M
Miklos Szeredi 已提交
831

832
	old_cred = ovl_override_creds(old->d_sb);
M
Miklos Szeredi 已提交
833

M
Miklos Szeredi 已提交
834
	if (overwrite && OVL_TYPE_MERGE_OR_LOWER(new_type) && new_is_dir) {
M
Miklos Szeredi 已提交
835
		opaquedir = ovl_check_empty_and_clear(new);
M
Miklos Szeredi 已提交
836 837 838 839 840 841 842 843 844
		err = PTR_ERR(opaquedir);
		if (IS_ERR(opaquedir)) {
			opaquedir = NULL;
			goto out_revert_creds;
		}
	}

	if (overwrite) {
		if (old_opaque) {
M
Miklos Szeredi 已提交
845
			if (!ovl_dentry_is_whiteout(new)) {
M
Miklos Szeredi 已提交
846 847 848 849 850 851
				/* Whiteout source */
				flags |= RENAME_WHITEOUT;
			} else {
				/* Switch whiteouts */
				flags |= RENAME_EXCHANGE;
			}
M
Miklos Szeredi 已提交
852
		} else if (is_dir && ovl_dentry_is_whiteout(new)) {
M
Miklos Szeredi 已提交
853 854 855 856 857 858 859 860 861 862
			flags |= RENAME_EXCHANGE;
			cleanup_whiteout = true;
		}
	}

	old_upperdir = ovl_dentry_upper(old->d_parent);
	new_upperdir = ovl_dentry_upper(new->d_parent);

	trap = lock_rename(new_upperdir, old_upperdir);

863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881

	olddentry = lookup_one_len(old->d_name.name, old_upperdir,
				   old->d_name.len);
	err = PTR_ERR(olddentry);
	if (IS_ERR(olddentry))
		goto out_unlock;

	err = -ESTALE;
	if (olddentry != ovl_dentry_upper(old))
		goto out_dput_old;

	newdentry = lookup_one_len(new->d_name.name, new_upperdir,
				   new->d_name.len);
	err = PTR_ERR(newdentry);
	if (IS_ERR(newdentry))
		goto out_dput_old;

	err = -ESTALE;
	if (ovl_dentry_upper(new)) {
M
Miklos Szeredi 已提交
882
		if (opaquedir) {
883 884
			if (newdentry != opaquedir)
				goto out_dput;
M
Miklos Szeredi 已提交
885
		} else {
886 887
			if (newdentry != ovl_dentry_upper(new))
				goto out_dput;
M
Miklos Szeredi 已提交
888 889
		}
	} else {
890 891 892
		if (!d_is_negative(newdentry) &&
		    (!new_opaque || !ovl_is_whiteout(newdentry)))
			goto out_dput;
M
Miklos Szeredi 已提交
893 894 895 896 897 898 899
	}

	if (olddentry == trap)
		goto out_dput;
	if (newdentry == trap)
		goto out_dput;

M
Miklos Szeredi 已提交
900 901 902
	if (WARN_ON(olddentry->d_inode == newdentry->d_inode))
		goto out_dput;

M
Miklos Szeredi 已提交
903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938
	if (is_dir && !old_opaque && new_opaque) {
		err = ovl_set_opaque(olddentry);
		if (err)
			goto out_dput;
	}
	if (!overwrite && new_is_dir && old_opaque && !new_opaque) {
		err = ovl_set_opaque(newdentry);
		if (err)
			goto out_dput;
	}

	if (old_opaque || new_opaque) {
		err = ovl_do_rename(old_upperdir->d_inode, olddentry,
				    new_upperdir->d_inode, newdentry,
				    flags);
	} else {
		/* No debug for the plain case */
		BUG_ON(flags & ~RENAME_EXCHANGE);
		err = vfs_rename(old_upperdir->d_inode, olddentry,
				 new_upperdir->d_inode, newdentry,
				 NULL, flags);
	}

	if (err) {
		if (is_dir && !old_opaque && new_opaque)
			ovl_remove_opaque(olddentry);
		if (!overwrite && new_is_dir && old_opaque && !new_opaque)
			ovl_remove_opaque(newdentry);
		goto out_dput;
	}

	if (is_dir && old_opaque && !new_opaque)
		ovl_remove_opaque(olddentry);
	if (!overwrite && new_is_dir && !old_opaque && new_opaque)
		ovl_remove_opaque(newdentry);

939 940 941 942 943 944 945
	/*
	 * Old dentry now lives in different location. Dentries in
	 * lowerstack are stale. We cannot drop them here because
	 * access to them is lockless. This could be only pure upper
	 * or opaque directory - numlower is zero. Or upper non-dir
	 * entry - its pureness is tracked by flag opaque.
	 */
M
Miklos Szeredi 已提交
946 947 948 949 950 951 952 953 954 955 956 957 958 959
	if (old_opaque != new_opaque) {
		ovl_dentry_set_opaque(old, new_opaque);
		if (!overwrite)
			ovl_dentry_set_opaque(new, old_opaque);
	}

	if (cleanup_whiteout)
		ovl_cleanup(old_upperdir->d_inode, newdentry);

	ovl_dentry_version_inc(old->d_parent);
	ovl_dentry_version_inc(new->d_parent);

out_dput:
	dput(newdentry);
960 961
out_dput_old:
	dput(olddentry);
M
Miklos Szeredi 已提交
962 963 964
out_unlock:
	unlock_rename(new_upperdir, old_upperdir);
out_revert_creds:
965
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
966 967 968 969 970 971 972 973 974 975 976 977 978
out_drop_write:
	ovl_drop_write(old);
out:
	dput(opaquedir);
	return err;
}

const struct inode_operations ovl_dir_inode_operations = {
	.lookup		= ovl_lookup,
	.mkdir		= ovl_mkdir,
	.symlink	= ovl_symlink,
	.unlink		= ovl_unlink,
	.rmdir		= ovl_rmdir,
979
	.rename		= ovl_rename,
M
Miklos Szeredi 已提交
980 981 982 983 984 985 986
	.link		= ovl_link,
	.setattr	= ovl_setattr,
	.create		= ovl_create,
	.mknod		= ovl_mknod,
	.permission	= ovl_permission,
	.getattr	= ovl_dir_getattr,
	.listxattr	= ovl_listxattr,
987
	.get_acl	= ovl_get_acl,
M
Miklos Szeredi 已提交
988
	.update_time	= ovl_update_time,
M
Miklos Szeredi 已提交
989
};