namei.c 27.1 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-only
M
Miklos Szeredi 已提交
2 3 4 5 6 7
/*
 * Copyright (C) 2011 Novell Inc.
 * Copyright (C) 2016 Red Hat, Inc.
 */

#include <linux/fs.h>
8
#include <linux/cred.h>
A
Amir Goldstein 已提交
9
#include <linux/ctype.h>
M
Miklos Szeredi 已提交
10 11
#include <linux/namei.h>
#include <linux/xattr.h>
M
Miklos Szeredi 已提交
12
#include <linux/ratelimit.h>
13 14
#include <linux/mount.h>
#include <linux/exportfs.h>
M
Miklos Szeredi 已提交
15 16
#include "overlayfs.h"

17
struct ovl_lookup_data {
A
Amir Goldstein 已提交
18
	struct super_block *sb;
19 20 21 22 23
	struct qstr name;
	bool is_dir;
	bool opaque;
	bool stop;
	bool last;
M
Miklos Szeredi 已提交
24
	char *redirect;
25
	bool metacopy;
26
};
M
Miklos Szeredi 已提交
27

M
Miklos Szeredi 已提交
28 29 30 31
static int ovl_check_redirect(struct dentry *dentry, struct ovl_lookup_data *d,
			      size_t prelen, const char *post)
{
	int res;
32
	char *buf;
M
Miklos Szeredi 已提交
33

34 35 36
	buf = ovl_get_redirect_xattr(dentry, prelen + strlen(post));
	if (IS_ERR_OR_NULL(buf))
		return PTR_ERR(buf);
M
Miklos Szeredi 已提交
37 38

	if (buf[0] == '/') {
39 40 41 42 43 44 45 46 47
		/*
		 * One of the ancestor path elements in an absolute path
		 * lookup in ovl_lookup_layer() could have been opaque and
		 * that will stop further lookup in lower layers (d->stop=true)
		 * But we have found an absolute redirect in decendant path
		 * element and that should force continue lookup in lower
		 * layers (reset d->stop).
		 */
		d->stop = false;
M
Miklos Szeredi 已提交
48
	} else {
49
		res = strlen(buf) + 1;
M
Miklos Szeredi 已提交
50 51 52 53 54 55 56 57 58 59 60 61 62
		memmove(buf + prelen, buf, res);
		memcpy(buf, d->name.name, prelen);
	}

	strcat(buf, post);
	kfree(d->redirect);
	d->redirect = buf;
	d->name.name = d->redirect;
	d->name.len = strlen(d->redirect);

	return 0;
}

63 64
static int ovl_acceptable(void *ctx, struct dentry *dentry)
{
65 66 67 68 69 70 71 72 73 74 75 76 77
	/*
	 * A non-dir origin may be disconnected, which is fine, because
	 * we only need it for its unique inode number.
	 */
	if (!d_is_dir(dentry))
		return 1;

	/* Don't decode a deleted empty directory */
	if (d_unhashed(dentry))
		return 0;

	/* Check if directory belongs to the layer we are decoding from */
	return is_subdir(dentry, ((struct vfsmount *)ctx)->mnt_root);
78 79
}

80 81 82 83 84 85 86
/*
 * Check validity of an overlay file handle buffer.
 *
 * Return 0 for a valid file handle.
 * Return -ENODATA for "origin unknown".
 * Return <0 for an invalid file handle.
 */
87
int ovl_check_fb_len(struct ovl_fb *fb, int fb_len)
88
{
89
	if (fb_len < sizeof(struct ovl_fb) || fb_len < fb->len)
90 91
		return -EINVAL;

92
	if (fb->magic != OVL_FH_MAGIC)
93 94 95
		return -EINVAL;

	/* Treat larger version and unknown flags as "origin unknown" */
96
	if (fb->version > OVL_FH_VERSION || fb->flags & ~OVL_FH_FLAG_ALL)
97 98 99
		return -ENODATA;

	/* Treat endianness mismatch as "origin unknown" */
100 101
	if (!(fb->flags & OVL_FH_FLAG_ANY_ENDIAN) &&
	    (fb->flags & OVL_FH_FLAG_BIG_ENDIAN) != OVL_FH_FLAG_CPU_ENDIAN)
102 103 104 105 106
		return -ENODATA;

	return 0;
}

107
static struct ovl_fh *ovl_get_fh(struct dentry *dentry, const char *name)
108
{
109
	int res, err;
110 111
	struct ovl_fh *fh = NULL;

112
	res = vfs_getxattr(dentry, name, NULL, 0);
113 114 115 116 117 118 119 120 121
	if (res < 0) {
		if (res == -ENODATA || res == -EOPNOTSUPP)
			return NULL;
		goto fail;
	}
	/* Zero size value means "copied up but origin unknown" */
	if (res == 0)
		return NULL;

122
	fh = kzalloc(res + OVL_FH_WIRE_OFFSET, GFP_KERNEL);
123 124 125
	if (!fh)
		return ERR_PTR(-ENOMEM);

126
	res = vfs_getxattr(dentry, name, fh->buf, res);
127 128 129
	if (res < 0)
		goto fail;

130
	err = ovl_check_fb_len(&fh->fb, res);
131 132 133
	if (err < 0) {
		if (err == -ENODATA)
			goto out;
134
		goto invalid;
135
	}
136

137 138 139 140 141 142 143
	return fh;

out:
	kfree(fh);
	return NULL;

fail:
L
lijiazi 已提交
144
	pr_warn_ratelimited("failed to get origin (%i)\n", res);
145 146
	goto out;
invalid:
L
lijiazi 已提交
147
	pr_warn_ratelimited("invalid origin (%*phN)\n", res, fh);
148 149 150
	goto out;
}

151 152
struct dentry *ovl_decode_real_fh(struct ovl_fh *fh, struct vfsmount *mnt,
				  bool connected)
153
{
154
	struct dentry *real;
155 156
	int bytes;

157 158 159 160
	/*
	 * Make sure that the stored uuid matches the uuid of the lower
	 * layer where file handle will be decoded.
	 */
161
	if (!uuid_equal(&fh->fb.uuid, &mnt->mnt_sb->s_uuid))
162
		return NULL;
163

164 165 166
	bytes = (fh->fb.len - offsetof(struct ovl_fb, fid));
	real = exportfs_decode_fh(mnt, (struct fid *)fh->fb.fid,
				  bytes >> 2, (int)fh->fb.type,
167
				  connected ? ovl_acceptable : NULL, mnt);
168 169 170 171 172 173 174 175
	if (IS_ERR(real)) {
		/*
		 * Treat stale file handle to lower file as "origin unknown".
		 * upper file handle could become stale when upper file is
		 * unlinked and this information is needed to handle stale
		 * index entries correctly.
		 */
		if (real == ERR_PTR(-ESTALE) &&
176
		    !(fh->fb.flags & OVL_FH_FLAG_PATH_UPPER))
177 178
			real = NULL;
		return real;
179 180
	}

181 182
	if (ovl_dentry_weird(real)) {
		dput(real);
183 184
		return NULL;
	}
185

186
	return real;
187 188
}

189 190 191 192 193
static bool ovl_is_opaquedir(struct dentry *dentry)
{
	return ovl_check_dir_xattr(dentry, OVL_XATTR_OPAQUE);
}

194 195
static int ovl_lookup_single(struct dentry *base, struct ovl_lookup_data *d,
			     const char *name, unsigned int namelen,
M
Miklos Szeredi 已提交
196
			     size_t prelen, const char *post,
197 198 199 200
			     struct dentry **ret)
{
	struct dentry *this;
	int err;
201
	bool last_element = !post[0];
202

A
Al Viro 已提交
203
	this = lookup_positive_unlocked(name, base, namelen);
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
	if (IS_ERR(this)) {
		err = PTR_ERR(this);
		this = NULL;
		if (err == -ENOENT || err == -ENAMETOOLONG)
			goto out;
		goto out_err;
	}

	if (ovl_dentry_weird(this)) {
		/* Don't support traversing automounts and other weirdness */
		err = -EREMOTE;
		goto out_err;
	}
	if (ovl_is_whiteout(this)) {
		d->stop = d->opaque = true;
		goto put_and_out;
	}
221 222 223 224 225
	/*
	 * This dentry should be a regular file if previous layer lookup
	 * found a metacopy dentry.
	 */
	if (last_element && d->metacopy && !d_is_reg(this)) {
226
		d->stop = true;
227 228 229 230 231
		goto put_and_out;
	}
	if (!d_can_lookup(this)) {
		if (d->is_dir || !last_element) {
			d->stop = true;
232
			goto put_and_out;
233 234 235 236
		}
		err = ovl_check_metacopy_xattr(this);
		if (err < 0)
			goto out_err;
237

238 239
		d->metacopy = err;
		d->stop = !d->metacopy;
240 241
		if (!d->metacopy || d->last)
			goto out;
242
	} else {
A
Amir Goldstein 已提交
243 244 245 246 247 248
		if (ovl_lookup_trap_inode(d->sb, this)) {
			/* Caught in a trap of overlapping layers */
			err = -ELOOP;
			goto out_err;
		}

249
		if (last_element)
250 251 252 253 254 255 256 257 258 259
			d->is_dir = true;
		if (d->last)
			goto out;

		if (ovl_is_opaquedir(this)) {
			d->stop = true;
			if (last_element)
				d->opaque = true;
			goto out;
		}
260
	}
M
Miklos Szeredi 已提交
261 262 263
	err = ovl_check_redirect(this, d, prelen, post);
	if (err)
		goto out_err;
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
out:
	*ret = this;
	return 0;

put_and_out:
	dput(this);
	this = NULL;
	goto out;

out_err:
	dput(this);
	return err;
}

static int ovl_lookup_layer(struct dentry *base, struct ovl_lookup_data *d,
			    struct dentry **ret)
{
281 282
	/* Counting down from the end, since the prefix can change */
	size_t rem = d->name.len - 1;
M
Miklos Szeredi 已提交
283 284 285
	struct dentry *dentry = NULL;
	int err;

286
	if (d->name.name[0] != '/')
M
Miklos Szeredi 已提交
287 288 289
		return ovl_lookup_single(base, d, d->name.name, d->name.len,
					 0, "", ret);

290 291
	while (!IS_ERR_OR_NULL(base) && d_can_lookup(base)) {
		const char *s = d->name.name + d->name.len - rem;
M
Miklos Szeredi 已提交
292
		const char *next = strchrnul(s, '/');
293 294
		size_t thislen = next - s;
		bool end = !next[0];
M
Miklos Szeredi 已提交
295

296 297
		/* Verify we did not go off the rails */
		if (WARN_ON(s[-1] != '/'))
M
Miklos Szeredi 已提交
298 299
			return -EIO;

300 301
		err = ovl_lookup_single(base, d, s, thislen,
					d->name.len - rem, next, &base);
M
Miklos Szeredi 已提交
302 303 304 305
		dput(dentry);
		if (err)
			return err;
		dentry = base;
306 307 308 309 310 311 312
		if (end)
			break;

		rem -= thislen + 1;

		if (WARN_ON(rem >= d->name.len))
			return -EIO;
M
Miklos Szeredi 已提交
313 314 315
	}
	*ret = dentry;
	return 0;
316 317
}

318

319
int ovl_check_origin_fh(struct ovl_fs *ofs, struct ovl_fh *fh, bool connected,
320
			struct dentry *upperdentry, struct ovl_path **stackp)
321
{
322 323
	struct dentry *origin = NULL;
	int i;
324

325
	for (i = 0; i < ofs->numlower; i++) {
326 327 328 329 330 331 332 333
		/*
		 * If lower fs uuid is not unique among lower fs we cannot match
		 * fh->uuid to layer.
		 */
		if (ofs->lower_layers[i].fsid &&
		    ofs->lower_layers[i].fs->bad_uuid)
			continue;

334 335
		origin = ovl_decode_real_fh(fh, ofs->lower_layers[i].mnt,
					    connected);
336 337 338 339 340
		if (origin)
			break;
	}

	if (!origin)
341 342 343 344
		return -ESTALE;
	else if (IS_ERR(origin))
		return PTR_ERR(origin);

345
	if (upperdentry && !ovl_is_whiteout(upperdentry) &&
346 347
	    ((d_inode(origin)->i_mode ^ d_inode(upperdentry)->i_mode) & S_IFMT))
		goto invalid;
348

349
	if (!*stackp)
350
		*stackp = kmalloc(sizeof(struct ovl_path), GFP_KERNEL);
351 352 353 354
	if (!*stackp) {
		dput(origin);
		return -ENOMEM;
	}
355 356 357 358
	**stackp = (struct ovl_path){
		.dentry = origin,
		.layer = &ofs->lower_layers[i]
	};
359 360

	return 0;
361 362

invalid:
L
lijiazi 已提交
363
	pr_warn_ratelimited("invalid origin (%pd2, ftype=%x, origin ftype=%x).\n",
364 365 366 367 368 369
			    upperdentry, d_inode(upperdentry)->i_mode & S_IFMT,
			    d_inode(origin)->i_mode & S_IFMT);
	dput(origin);
	return -EIO;
}

370
static int ovl_check_origin(struct ovl_fs *ofs, struct dentry *upperdentry,
371 372
			    struct ovl_path **stackp, unsigned int *ctrp)
{
373
	struct ovl_fh *fh = ovl_get_fh(upperdentry, OVL_XATTR_ORIGIN);
374 375 376 377 378
	int err;

	if (IS_ERR_OR_NULL(fh))
		return PTR_ERR(fh);

379
	err = ovl_check_origin_fh(ofs, fh, false, upperdentry, stackp);
380 381 382 383 384 385 386 387 388 389 390 391 392
	kfree(fh);

	if (err) {
		if (err == -ESTALE)
			return 0;
		return err;
	}

	if (WARN_ON(*ctrp))
		return -EIO;

	*ctrp = 1;
	return 0;
393 394
}

395
/*
396
 * Verify that @fh matches the file handle stored in xattr @name.
397 398
 * Return 0 on match, -ESTALE on mismatch, < 0 on error.
 */
399 400
static int ovl_verify_fh(struct dentry *dentry, const char *name,
			 const struct ovl_fh *fh)
401
{
402
	struct ovl_fh *ofh = ovl_get_fh(dentry, name);
403 404 405 406 407 408 409 410
	int err = 0;

	if (!ofh)
		return -ENODATA;

	if (IS_ERR(ofh))
		return PTR_ERR(ofh);

411
	if (fh->fb.len != ofh->fb.len || memcmp(&fh->fb, &ofh->fb, fh->fb.len))
412 413 414 415 416 417 418
		err = -ESTALE;

	kfree(ofh);
	return err;
}

/*
419
 * Verify that @real dentry matches the file handle stored in xattr @name.
420
 *
421 422
 * If @set is true and there is no stored file handle, encode @real and store
 * file handle in xattr @name.
423
 *
424
 * Return 0 on match, -ESTALE on mismatch, -ENODATA on no xattr, < 0 on error.
425
 */
426 427
int ovl_verify_set_fh(struct dentry *dentry, const char *name,
		      struct dentry *real, bool is_upper, bool set)
428 429 430 431 432
{
	struct inode *inode;
	struct ovl_fh *fh;
	int err;

433
	fh = ovl_encode_real_fh(real, is_upper);
434
	err = PTR_ERR(fh);
435 436
	if (IS_ERR(fh)) {
		fh = NULL;
437
		goto fail;
438
	}
439

440
	err = ovl_verify_fh(dentry, name, fh);
441
	if (set && err == -ENODATA)
442
		err = ovl_do_setxattr(dentry, name, fh->buf, fh->fb.len, 0);
443 444 445 446 447 448 449 450
	if (err)
		goto fail;

out:
	kfree(fh);
	return err;

fail:
451
	inode = d_inode(real);
L
lijiazi 已提交
452
	pr_warn_ratelimited("failed to verify %s (%pd2, ino=%lu, err=%i)\n",
453 454
			    is_upper ? "upper" : "origin", real,
			    inode ? inode->i_ino : 0, err);
455 456 457
	goto out;
}

458
/* Get upper dentry from index */
459
struct dentry *ovl_index_upper(struct ovl_fs *ofs, struct dentry *index)
460 461 462 463 464 465 466 467 468 469 470
{
	struct ovl_fh *fh;
	struct dentry *upper;

	if (!d_is_dir(index))
		return dget(index);

	fh = ovl_get_fh(index, OVL_XATTR_UPPER);
	if (IS_ERR_OR_NULL(fh))
		return ERR_CAST(fh);

471
	upper = ovl_decode_real_fh(fh, ofs->upper_mnt, true);
472 473 474 475 476 477
	kfree(fh);

	if (IS_ERR_OR_NULL(upper))
		return upper ?: ERR_PTR(-ESTALE);

	if (!d_is_dir(upper)) {
L
lijiazi 已提交
478
		pr_warn_ratelimited("invalid index upper (%pd2, upper=%pd2).\n",
479 480 481 482 483 484 485 486
				    index, upper);
		dput(upper);
		return ERR_PTR(-EIO);
	}

	return upper;
}

A
Amir Goldstein 已提交
487 488 489 490 491 492
/* Is this a leftover from create/whiteout of directory index entry? */
static bool ovl_is_temp_index(struct dentry *index)
{
	return index->d_name.name[0] == '#';
}

493 494 495 496 497
/*
 * Verify that an index entry name matches the origin file handle stored in
 * OVL_XATTR_ORIGIN and that origin file handle can be decoded to lower path.
 * Return 0 on match, -ESTALE on mismatch or stale origin, < 0 on error.
 */
498
int ovl_verify_index(struct ovl_fs *ofs, struct dentry *index)
499 500 501
{
	struct ovl_fh *fh = NULL;
	size_t len;
502 503
	struct ovl_path origin = { };
	struct ovl_path *stack = &origin;
504
	struct dentry *upper = NULL;
505 506 507 508 509
	int err;

	if (!d_inode(index))
		return 0;

A
Amir Goldstein 已提交
510 511 512 513 514
	/* Cleanup leftover from index create/cleanup attempt */
	err = -ESTALE;
	if (ovl_is_temp_index(index))
		goto fail;

515
	err = -EINVAL;
516
	if (index->d_name.len < sizeof(struct ovl_fb)*2)
517 518 519 520
		goto fail;

	err = -ENOMEM;
	len = index->d_name.len / 2;
521
	fh = kzalloc(len + OVL_FH_WIRE_OFFSET, GFP_KERNEL);
522 523 524 525
	if (!fh)
		goto fail;

	err = -EINVAL;
526
	if (hex2bin(fh->buf, index->d_name.name, len))
527 528
		goto fail;

529
	err = ovl_check_fb_len(&fh->fb, len);
530
	if (err)
531 532
		goto fail;

533 534 535 536 537 538 539 540
	/*
	 * Whiteout index entries are used as an indication that an exported
	 * overlay file handle should be treated as stale (i.e. after unlink
	 * of the overlay inode). These entries contain no origin xattr.
	 */
	if (ovl_is_whiteout(index))
		goto out;

541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
	/*
	 * Verifying directory index entries are not stale is expensive, so
	 * only verify stale dir index if NFS export is enabled.
	 */
	if (d_is_dir(index) && !ofs->config.nfs_export)
		goto out;

	/*
	 * Directory index entries should have 'upper' xattr pointing to the
	 * real upper dir. Non-dir index entries are hardlinks to the upper
	 * real inode. For non-dir index, we can read the copy up origin xattr
	 * directly from the index dentry, but for dir index we first need to
	 * decode the upper directory.
	 */
	upper = ovl_index_upper(ofs, index);
	if (IS_ERR_OR_NULL(upper)) {
		err = PTR_ERR(upper);
558 559 560 561 562 563 564 565 566
		/*
		 * Directory index entries with no 'upper' xattr need to be
		 * removed. When dir index entry has a stale 'upper' xattr,
		 * we assume that upper dir was removed and we treat the dir
		 * index as orphan entry that needs to be whited out.
		 */
		if (err == -ESTALE)
			goto orphan;
		else if (!err)
567
			err = -ESTALE;
568
		goto fail;
569
	}
570

571 572
	err = ovl_verify_fh(upper, OVL_XATTR_ORIGIN, fh);
	dput(upper);
573 574 575
	if (err)
		goto fail;

576 577
	/* Check if non-dir index is orphan and don't warn before cleaning it */
	if (!d_is_dir(index) && d_inode(index)->i_nlink == 1) {
578
		err = ovl_check_origin_fh(ofs, fh, false, index, &stack);
579 580 581 582
		if (err)
			goto fail;

		if (ovl_get_nlink(origin.dentry, index, 0) == 0)
583
			goto orphan;
584
	}
585

586
out:
587
	dput(origin.dentry);
588 589 590 591
	kfree(fh);
	return err;

fail:
L
lijiazi 已提交
592
	pr_warn_ratelimited("failed to verify index (%pd2, ftype=%x, err=%i)\n",
593
			    index, d_inode(index)->i_mode & S_IFMT, err);
594
	goto out;
595 596

orphan:
L
lijiazi 已提交
597
	pr_warn_ratelimited("orphan index entry (%pd2, ftype=%x, nlink=%u)\n",
598 599 600 601
			    index, d_inode(index)->i_mode & S_IFMT,
			    d_inode(index)->i_nlink);
	err = -ENOENT;
	goto out;
602 603
}

604 605 606 607
static int ovl_get_index_name_fh(struct ovl_fh *fh, struct qstr *name)
{
	char *n, *s;

608
	n = kcalloc(fh->fb.len, 2, GFP_KERNEL);
609 610 611
	if (!n)
		return -ENOMEM;

612
	s  = bin2hex(n, fh->buf, fh->fb.len);
613 614 615 616 617 618
	*name = (struct qstr) QSTR_INIT(n, s - n);

	return 0;

}

619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636
/*
 * Lookup in indexdir for the index entry of a lower real inode or a copy up
 * origin inode. The index entry name is the hex representation of the lower
 * inode file handle.
 *
 * If the index dentry in negative, then either no lower aliases have been
 * copied up yet, or aliases have been copied up in older kernels and are
 * not indexed.
 *
 * If the index dentry for a copy up origin inode is positive, but points
 * to an inode different than the upper inode, then either the upper inode
 * has been copied up and not indexed or it was indexed, but since then
 * index dir was cleared. Either way, that index cannot be used to indentify
 * the overlay inode.
 */
int ovl_get_index_name(struct dentry *origin, struct qstr *name)
{
	struct ovl_fh *fh;
637
	int err;
638

639
	fh = ovl_encode_real_fh(origin, false);
640 641 642
	if (IS_ERR(fh))
		return PTR_ERR(fh);

643
	err = ovl_get_index_name_fh(fh, name);
644

645
	kfree(fh);
646
	return err;
647 648 649 650 651 652 653 654 655 656 657 658 659
}

/* Lookup index by file handle for NFS export */
struct dentry *ovl_get_index_fh(struct ovl_fs *ofs, struct ovl_fh *fh)
{
	struct dentry *index;
	struct qstr name;
	int err;

	err = ovl_get_index_name_fh(fh, &name);
	if (err)
		return ERR_PTR(err);

A
Al Viro 已提交
660
	index = lookup_positive_unlocked(name.name, ofs->indexdir, name.len);
661 662 663 664 665 666 667
	kfree(name.name);
	if (IS_ERR(index)) {
		if (PTR_ERR(index) == -ENOENT)
			index = NULL;
		return index;
	}

A
Al Viro 已提交
668
	if (ovl_is_whiteout(index))
669 670 671 672 673
		err = -ESTALE;
	else if (ovl_dentry_weird(index))
		err = -EIO;
	else
		return index;
674

675 676
	dput(index);
	return ERR_PTR(err);
677 678
}

679 680
struct dentry *ovl_lookup_index(struct ovl_fs *ofs, struct dentry *upper,
				struct dentry *origin, bool verify)
681 682 683 684
{
	struct dentry *index;
	struct inode *inode;
	struct qstr name;
685
	bool is_dir = d_is_dir(origin);
686 687 688 689 690 691
	int err;

	err = ovl_get_index_name(origin, &name);
	if (err)
		return ERR_PTR(err);

A
Al Viro 已提交
692
	index = lookup_positive_unlocked(name.name, ofs->indexdir, name.len);
693
	if (IS_ERR(index)) {
694
		err = PTR_ERR(index);
695 696 697 698
		if (err == -ENOENT) {
			index = NULL;
			goto out;
		}
L
lijiazi 已提交
699
		pr_warn_ratelimited("failed inode index lookup (ino=%lu, key=%.*s, err=%i);\n"
700 701 702 703 704 705
				    "overlayfs: mount with '-o index=off' to disable inodes index.\n",
				    d_inode(origin)->i_ino, name.len, name.name,
				    err);
		goto out;
	}

706
	inode = d_inode(index);
A
Al Viro 已提交
707
	if (ovl_is_whiteout(index) && !verify) {
708 709 710 711 712 713 714 715 716
		/*
		 * When index lookup is called with !verify for decoding an
		 * overlay file handle, a whiteout index implies that decode
		 * should treat file handle as stale and no need to print a
		 * warning about it.
		 */
		dput(index);
		index = ERR_PTR(-ESTALE);
		goto out;
717 718 719 720 721 722 723 724 725
	} else if (ovl_dentry_weird(index) || ovl_is_whiteout(index) ||
		   ((inode->i_mode ^ d_inode(origin)->i_mode) & S_IFMT)) {
		/*
		 * Index should always be of the same file type as origin
		 * except for the case of a whiteout index. A whiteout
		 * index should only exist if all lower aliases have been
		 * unlinked, which means that finding a lower origin on lookup
		 * whose index is a whiteout should be treated as an error.
		 */
L
lijiazi 已提交
726
		pr_warn_ratelimited("bad index found (index=%pd2, ftype=%x, origin ftype=%x).\n",
727 728
				    index, d_inode(index)->i_mode & S_IFMT,
				    d_inode(origin)->i_mode & S_IFMT);
729
		goto fail;
730
	} else if (is_dir && verify) {
731
		if (!upper) {
L
lijiazi 已提交
732
			pr_warn_ratelimited("suspected uncovered redirected dir found (origin=%pd2, index=%pd2).\n",
733 734 735
					    origin, index);
			goto fail;
		}
736

737 738 739 740
		/* Verify that dir index 'upper' xattr points to upper dir */
		err = ovl_verify_upper(index, upper, false);
		if (err) {
			if (err == -ESTALE) {
L
lijiazi 已提交
741
				pr_warn_ratelimited("suspected multiply redirected dir found (upper=%pd2, origin=%pd2, index=%pd2).\n",
742 743 744 745 746 747 748
						    upper, origin, index);
			}
			goto fail;
		}
	} else if (upper && d_inode(upper) != inode) {
		goto out_dput;
	}
749 750 751 752
out:
	kfree(name.name);
	return index;

753 754 755 756 757
out_dput:
	dput(index);
	index = NULL;
	goto out;

758 759 760 761 762 763
fail:
	dput(index);
	index = ERR_PTR(-EIO);
	goto out;
}

M
Miklos Szeredi 已提交
764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779
/*
 * Returns next layer in stack starting from top.
 * Returns -1 if this is the last layer.
 */
int ovl_path_next(int idx, struct dentry *dentry, struct path *path)
{
	struct ovl_entry *oe = dentry->d_fsdata;

	BUG_ON(idx < 0);
	if (idx == 0) {
		ovl_path_upper(dentry, path);
		if (path->dentry)
			return oe->numlower ? 1 : -1;
		idx++;
	}
	BUG_ON(idx > oe->numlower);
780 781
	path->dentry = oe->lowerstack[idx - 1].dentry;
	path->mnt = oe->lowerstack[idx - 1].layer->mnt;
M
Miklos Szeredi 已提交
782 783 784 785

	return (idx < oe->numlower) ? idx + 1 : -1;
}

786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806
/* Fix missing 'origin' xattr */
static int ovl_fix_origin(struct dentry *dentry, struct dentry *lower,
			  struct dentry *upper)
{
	int err;

	if (ovl_check_origin_xattr(upper))
		return 0;

	err = ovl_want_write(dentry);
	if (err)
		return err;

	err = ovl_set_origin(dentry, lower, upper);
	if (!err)
		err = ovl_set_impure(dentry->d_parent, upper->d_parent);

	ovl_drop_write(dentry);
	return err;
}

M
Miklos Szeredi 已提交
807 808 809 810 811
struct dentry *ovl_lookup(struct inode *dir, struct dentry *dentry,
			  unsigned int flags)
{
	struct ovl_entry *oe;
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
812
	struct ovl_fs *ofs = dentry->d_sb->s_fs_info;
M
Miklos Szeredi 已提交
813
	struct ovl_entry *poe = dentry->d_parent->d_fsdata;
814
	struct ovl_entry *roe = dentry->d_sb->s_root->d_fsdata;
815
	struct ovl_path *stack = NULL, *origin_path = NULL;
M
Miklos Szeredi 已提交
816
	struct dentry *upperdir, *upperdentry = NULL;
817
	struct dentry *origin = NULL;
818
	struct dentry *index = NULL;
M
Miklos Szeredi 已提交
819 820 821
	unsigned int ctr = 0;
	struct inode *inode = NULL;
	bool upperopaque = false;
M
Miklos Szeredi 已提交
822
	char *upperredirect = NULL;
M
Miklos Szeredi 已提交
823 824 825
	struct dentry *this;
	unsigned int i;
	int err;
826
	bool metacopy = false;
827
	struct ovl_lookup_data d = {
A
Amir Goldstein 已提交
828
		.sb = dentry->d_sb,
829 830 831 832
		.name = dentry->d_name,
		.is_dir = false,
		.opaque = false,
		.stop = false,
833
		.last = ofs->config.redirect_follow ? false : !poe->numlower,
M
Miklos Szeredi 已提交
834
		.redirect = NULL,
835
		.metacopy = false,
836
	};
M
Miklos Szeredi 已提交
837

M
Miklos Szeredi 已提交
838 839 840
	if (dentry->d_name.len > ofs->namelen)
		return ERR_PTR(-ENAMETOOLONG);

M
Miklos Szeredi 已提交
841
	old_cred = ovl_override_creds(dentry->d_sb);
842
	upperdir = ovl_dentry_upper(dentry->d_parent);
M
Miklos Szeredi 已提交
843
	if (upperdir) {
844 845
		err = ovl_lookup_layer(upperdir, &d, &upperdentry);
		if (err)
M
Miklos Szeredi 已提交
846 847
			goto out;

848 849 850 851
		if (upperdentry && unlikely(ovl_dentry_remote(upperdentry))) {
			dput(upperdentry);
			err = -EREMOTE;
			goto out;
M
Miklos Szeredi 已提交
852
		}
853
		if (upperdentry && !d.is_dir) {
854 855
			unsigned int origin_ctr = 0;

856 857 858 859 860 861 862 863 864 865
			/*
			 * Lookup copy up origin by decoding origin file handle.
			 * We may get a disconnected dentry, which is fine,
			 * because we only need to hold the origin inode in
			 * cache and use its inode number.  We may even get a
			 * connected dentry, that is not under any of the lower
			 * layers root.  That is also fine for using it's inode
			 * number - it's the same as if we held a reference
			 * to a dentry in lower layer that was moved under us.
			 */
866 867
			err = ovl_check_origin(ofs, upperdentry, &origin_path,
					       &origin_ctr);
868
			if (err)
869
				goto out_put_upper;
870 871 872

			if (d.metacopy)
				metacopy = true;
873
		}
M
Miklos Szeredi 已提交
874 875

		if (d.redirect) {
876
			err = -ENOMEM;
M
Miklos Szeredi 已提交
877 878 879 880
			upperredirect = kstrdup(d.redirect, GFP_KERNEL);
			if (!upperredirect)
				goto out_put_upper;
			if (d.redirect[0] == '/')
881
				poe = roe;
M
Miklos Szeredi 已提交
882
		}
883
		upperopaque = d.opaque;
M
Miklos Szeredi 已提交
884 885
	}

886
	if (!d.stop && poe->numlower) {
M
Miklos Szeredi 已提交
887
		err = -ENOMEM;
888
		stack = kcalloc(ofs->numlower, sizeof(struct ovl_path),
889
				GFP_KERNEL);
M
Miklos Szeredi 已提交
890 891 892 893
		if (!stack)
			goto out_put_upper;
	}

894
	for (i = 0; !d.stop && i < poe->numlower; i++) {
895
		struct ovl_path lower = poe->lowerstack[i];
M
Miklos Szeredi 已提交
896

897 898 899 900 901
		if (!ofs->config.redirect_follow)
			d.last = i == poe->numlower - 1;
		else
			d.last = lower.layer->idx == roe->numlower;

902
		err = ovl_lookup_layer(lower.dentry, &d, &this);
903
		if (err)
M
Miklos Szeredi 已提交
904
			goto out_put;
M
Miklos Szeredi 已提交
905

M
Miklos Szeredi 已提交
906 907 908
		if (!this)
			continue;

909 910 911 912
		/*
		 * If no origin fh is stored in upper of a merge dir, store fh
		 * of lower dir and set upper parent "impure".
		 */
913
		if (upperdentry && !ctr && !ofs->noxattr && d.is_dir) {
914 915 916 917 918 919 920
			err = ovl_fix_origin(dentry, this, upperdentry);
			if (err) {
				dput(this);
				goto out_put;
			}
		}

921 922
		/*
		 * When "verify_lower" feature is enabled, do not merge with a
923 924
		 * lower dir that does not match a stored origin xattr. In any
		 * case, only verified origin is used for index lookup.
925 926 927 928
		 *
		 * For non-dir dentry, if index=on, then ensure origin
		 * matches the dentry found using path based lookup,
		 * otherwise error out.
929
		 */
930 931 932
		if (upperdentry && !ctr &&
		    ((d.is_dir && ovl_verify_lower(dentry->d_sb)) ||
		     (!d.is_dir && ofs->config.index && origin_path))) {
933 934 935
			err = ovl_verify_origin(upperdentry, this, false);
			if (err) {
				dput(this);
936 937 938
				if (d.is_dir)
					break;
				goto out_put;
939
			}
940
			origin = this;
941 942
		}

943 944 945 946 947 948 949 950 951 952 953
		if (d.metacopy)
			metacopy = true;
		/*
		 * Do not store intermediate metacopy dentries in chain,
		 * except top most lower metacopy dentry
		 */
		if (d.metacopy && ctr) {
			dput(this);
			continue;
		}

M
Miklos Szeredi 已提交
954
		stack[ctr].dentry = this;
955
		stack[ctr].layer = lower.layer;
M
Miklos Szeredi 已提交
956
		ctr++;
M
Miklos Szeredi 已提交
957

958 959 960 961 962 963 964 965 966 967 968 969
		/*
		 * Following redirects can have security consequences: it's like
		 * a symlink into the lower layer without the permission checks.
		 * This is only a problem if the upper layer is untrusted (e.g
		 * comes from an USB drive).  This can allow a non-readable file
		 * or directory to become readable.
		 *
		 * Only following redirects when redirects are enabled disables
		 * this attack vector when not necessary.
		 */
		err = -EPERM;
		if (d.redirect && !ofs->config.redirect_follow) {
L
lijiazi 已提交
970
			pr_warn_ratelimited("refusing to follow redirect for (%pd2)\n",
971
					    dentry);
972 973 974
			goto out_put;
		}

975 976 977
		if (d.stop)
			break;

978 979
		if (d.redirect && d.redirect[0] == '/' && poe != roe) {
			poe = roe;
M
Miklos Szeredi 已提交
980
			/* Find the current layer on the root dentry */
981
			i = lower.layer->idx - 1;
M
Miklos Szeredi 已提交
982
		}
M
Miklos Szeredi 已提交
983 984
	}

985 986 987 988 989 990 991 992 993 994 995 996
	if (metacopy) {
		/*
		 * Found a metacopy dentry but did not find corresponding
		 * data dentry
		 */
		if (d.metacopy) {
			err = -EIO;
			goto out_put;
		}

		err = -EPERM;
		if (!ofs->config.metacopy) {
L
lijiazi 已提交
997
			pr_warn_ratelimited("refusing to follow metacopy origin for (%pd2)\n",
998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010
					    dentry);
			goto out_put;
		}
	} else if (!d.is_dir && upperdentry && !ctr && origin_path) {
		if (WARN_ON(stack != NULL)) {
			err = -EIO;
			goto out_put;
		}
		stack = origin_path;
		ctr = 1;
		origin_path = NULL;
	}

1011 1012 1013 1014
	/*
	 * Lookup index by lower inode and verify it matches upper inode.
	 * We only trust dir index if we verified that lower dir matches
	 * origin, otherwise dir index entries may be inconsistent and we
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
	 * ignore them.
	 *
	 * For non-dir upper metacopy dentry, we already set "origin" if we
	 * verified that lower matched upper origin. If upper origin was
	 * not present (because lower layer did not support fh encode/decode),
	 * or indexing is not enabled, do not set "origin" and skip looking up
	 * index. This case should be handled in same way as a non-dir upper
	 * without ORIGIN is handled.
	 *
	 * Always lookup index of non-dir non-metacopy and non-upper.
1025
	 */
1026
	if (ctr && (!upperdentry || (!d.is_dir && !metacopy)))
1027
		origin = stack[0].dentry;
1028

1029 1030
	if (origin && ovl_indexdir(dentry->d_sb) &&
	    (!d.is_dir || ovl_index_all(dentry->d_sb))) {
1031
		index = ovl_lookup_index(ofs, upperdentry, origin, true);
1032 1033 1034 1035 1036 1037 1038
		if (IS_ERR(index)) {
			err = PTR_ERR(index);
			index = NULL;
			goto out_put;
		}
	}

M
Miklos Szeredi 已提交
1039 1040 1041 1042 1043
	oe = ovl_alloc_entry(ctr);
	err = -ENOMEM;
	if (!oe)
		goto out_put;

1044
	memcpy(oe->lowerstack, stack, sizeof(struct ovl_path) * ctr);
M
Miklos Szeredi 已提交
1045
	dentry->d_fsdata = oe;
M
Miklos Szeredi 已提交
1046

1047 1048 1049
	if (upperopaque)
		ovl_dentry_set_opaque(dentry);

1050 1051
	if (upperdentry)
		ovl_dentry_set_upper_alias(dentry);
1052
	else if (index) {
1053
		upperdentry = dget(index);
1054 1055 1056 1057 1058 1059 1060
		upperredirect = ovl_get_redirect_xattr(upperdentry, 0);
		if (IS_ERR(upperredirect)) {
			err = PTR_ERR(upperredirect);
			upperredirect = NULL;
			goto out_free_oe;
		}
	}
1061

M
Miklos Szeredi 已提交
1062
	if (upperdentry || ctr) {
1063 1064 1065 1066 1067
		struct ovl_inode_params oip = {
			.upperdentry = upperdentry,
			.lowerpath = stack,
			.index = index,
			.numlower = ctr,
1068
			.redirect = upperredirect,
1069 1070
			.lowerdata = (ctr > 1 && !d.is_dir) ?
				      stack[ctr - 1].dentry : NULL,
1071 1072 1073
		};

		inode = ovl_get_inode(dentry->d_sb, &oip);
1074 1075
		err = PTR_ERR(inode);
		if (IS_ERR(inode))
M
Miklos Szeredi 已提交
1076 1077 1078 1079
			goto out_free_oe;
	}

	revert_creds(old_cred);
1080 1081 1082 1083
	if (origin_path) {
		dput(origin_path->dentry);
		kfree(origin_path);
	}
1084
	dput(index);
M
Miklos Szeredi 已提交
1085
	kfree(stack);
M
Miklos Szeredi 已提交
1086
	kfree(d.redirect);
1087
	return d_splice_alias(inode, dentry);
M
Miklos Szeredi 已提交
1088 1089

out_free_oe:
M
Miklos Szeredi 已提交
1090
	dentry->d_fsdata = NULL;
M
Miklos Szeredi 已提交
1091 1092
	kfree(oe);
out_put:
1093
	dput(index);
M
Miklos Szeredi 已提交
1094 1095 1096 1097
	for (i = 0; i < ctr; i++)
		dput(stack[i].dentry);
	kfree(stack);
out_put_upper:
1098 1099 1100 1101
	if (origin_path) {
		dput(origin_path->dentry);
		kfree(origin_path);
	}
M
Miklos Szeredi 已提交
1102
	dput(upperdentry);
M
Miklos Szeredi 已提交
1103
	kfree(upperredirect);
M
Miklos Szeredi 已提交
1104
out:
M
Miklos Szeredi 已提交
1105
	kfree(d.redirect);
M
Miklos Szeredi 已提交
1106 1107 1108 1109 1110 1111 1112 1113
	revert_creds(old_cred);
	return ERR_PTR(err);
}

bool ovl_lower_positive(struct dentry *dentry)
{
	struct ovl_entry *poe = dentry->d_parent->d_fsdata;
	const struct qstr *name = &dentry->d_name;
1114
	const struct cred *old_cred;
M
Miklos Szeredi 已提交
1115 1116 1117 1118 1119 1120 1121 1122 1123
	unsigned int i;
	bool positive = false;
	bool done = false;

	/*
	 * If dentry is negative, then lower is positive iff this is a
	 * whiteout.
	 */
	if (!dentry->d_inode)
1124
		return ovl_dentry_is_opaque(dentry);
M
Miklos Szeredi 已提交
1125 1126

	/* Negative upper -> positive lower */
1127
	if (!ovl_dentry_upper(dentry))
M
Miklos Szeredi 已提交
1128 1129
		return true;

1130
	old_cred = ovl_override_creds(dentry->d_sb);
M
Miklos Szeredi 已提交
1131 1132 1133 1134 1135
	/* Positive upper -> have to look up lower to see whether it exists */
	for (i = 0; !done && !positive && i < poe->numlower; i++) {
		struct dentry *this;
		struct dentry *lowerdir = poe->lowerstack[i].dentry;

A
Al Viro 已提交
1136
		this = lookup_positive_unlocked(name->name, lowerdir,
M
Miklos Szeredi 已提交
1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152
					       name->len);
		if (IS_ERR(this)) {
			switch (PTR_ERR(this)) {
			case -ENOENT:
			case -ENAMETOOLONG:
				break;

			default:
				/*
				 * Assume something is there, we just couldn't
				 * access it.
				 */
				positive = true;
				break;
			}
		} else {
A
Al Viro 已提交
1153 1154
			positive = !ovl_is_whiteout(this);
			done = true;
M
Miklos Szeredi 已提交
1155 1156 1157
			dput(this);
		}
	}
1158
	revert_creds(old_cred);
M
Miklos Szeredi 已提交
1159 1160 1161

	return positive;
}