namei.c 110.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
/*
 *  linux/fs/namei.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

/*
 * Some corrections by tytso.
 */

/* [Feb 1997 T. Schoebel-Theuer] Complete rewrite of the pathname
 * lookup logic.
 */
/* [Feb-Apr 2000, AV] Rewrite to the new namespace architecture.
 */

#include <linux/init.h>
18
#include <linux/export.h>
19
#include <linux/kernel.h>
L
Linus Torvalds 已提交
20 21 22 23
#include <linux/slab.h>
#include <linux/fs.h>
#include <linux/namei.h>
#include <linux/pagemap.h>
R
Robert Love 已提交
24
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
25 26
#include <linux/personality.h>
#include <linux/security.h>
M
Mimi Zohar 已提交
27
#include <linux/ima.h>
L
Linus Torvalds 已提交
28 29 30
#include <linux/syscalls.h>
#include <linux/mount.h>
#include <linux/audit.h>
31
#include <linux/capability.h>
32
#include <linux/file.h>
33
#include <linux/fcntl.h>
34
#include <linux/device_cgroup.h>
35
#include <linux/fs_struct.h>
36
#include <linux/posix_acl.h>
L
Linus Torvalds 已提交
37 38
#include <asm/uaccess.h>

39
#include "internal.h"
40
#include "mount.h"
41

L
Linus Torvalds 已提交
42 43 44 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
/* [Feb-1997 T. Schoebel-Theuer]
 * Fundamental changes in the pathname lookup mechanisms (namei)
 * were necessary because of omirr.  The reason is that omirr needs
 * to know the _real_ pathname, not the user-supplied one, in case
 * of symlinks (and also when transname replacements occur).
 *
 * The new code replaces the old recursive symlink resolution with
 * an iterative one (in case of non-nested symlink chains).  It does
 * this with calls to <fs>_follow_link().
 * As a side effect, dir_namei(), _namei() and follow_link() are now 
 * replaced with a single function lookup_dentry() that can handle all 
 * the special cases of the former code.
 *
 * With the new dcache, the pathname is stored at each inode, at least as
 * long as the refcount of the inode is positive.  As a side effect, the
 * size of the dcache depends on the inode cache and thus is dynamic.
 *
 * [29-Apr-1998 C. Scott Ananian] Updated above description of symlink
 * resolution to correspond with current state of the code.
 *
 * Note that the symlink resolution is not *completely* iterative.
 * There is still a significant amount of tail- and mid- recursion in
 * the algorithm.  Also, note that <fs>_readlink() is not used in
 * lookup_dentry(): lookup_dentry() on the result of <fs>_readlink()
 * may return different results than <fs>_follow_link().  Many virtual
 * filesystems (including /proc) exhibit this behavior.
 */

/* [24-Feb-97 T. Schoebel-Theuer] Side effects caused by new implementation:
 * New symlink semantics: when open() is called with flags O_CREAT | O_EXCL
 * and the name already exists in form of a symlink, try to create the new
 * name indicated by the symlink. The old code always complained that the
 * name already exists, due to not following the symlink even if its target
 * is nonexistent.  The new semantics affects also mknod() and link() when
L
Lucas De Marchi 已提交
76
 * the name is a symlink pointing to a non-existent name.
L
Linus Torvalds 已提交
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
 *
 * I don't know which semantics is the right one, since I have no access
 * to standards. But I found by trial that HP-UX 9.0 has the full "new"
 * semantics implemented, while SunOS 4.1.1 and Solaris (SunOS 5.4) have the
 * "old" one. Personally, I think the new semantics is much more logical.
 * Note that "ln old new" where "new" is a symlink pointing to a non-existing
 * file does succeed in both HP-UX and SunOs, but not in Solaris
 * and in the old Linux semantics.
 */

/* [16-Dec-97 Kevin Buhr] For security reasons, we change some symlink
 * semantics.  See the comments in "open_namei" and "do_link" below.
 *
 * [10-Sep-98 Alan Modra] Another symlink change.
 */

/* [Feb-Apr 2000 AV] Complete rewrite. Rules for symlinks:
 *	inside the path - always follow.
 *	in the last component in creation/removal/renaming - never follow.
 *	if LOOKUP_FOLLOW passed - follow.
 *	if the pathname has trailing slashes - follow.
 *	otherwise - don't follow.
 * (applied in that order).
 *
 * [Jun 2000 AV] Inconsistent behaviour of open() in case if flags==O_CREAT
 * restored for 2.4. This is the last surviving part of old 4.2BSD bug.
 * During the 2.4 we need to fix the userland stuff depending on it -
 * hopefully we will be able to get rid of that wart in 2.5. So far only
 * XEmacs seems to be relying on it...
 */
/*
 * [Sep 2001 AV] Single-semaphore locking scheme (kudos to David Holland)
109
 * implemented.  Let's see if raised priority of ->s_vfs_rename_mutex gives
L
Linus Torvalds 已提交
110 111 112 113 114 115 116 117 118 119
 * any extra contention...
 */

/* In order to reduce some races, while at the same time doing additional
 * checking and hopefully speeding things up, we copy filenames to the
 * kernel data space before using them..
 *
 * POSIX.1 2.4: an empty pathname is invalid (ENOENT).
 * PATH_MAX includes the nul terminator --RR.
 */
120
void final_putname(struct filename *name)
L
Linus Torvalds 已提交
121
{
122 123 124 125 126 127
	if (name->separate) {
		__putname(name->name);
		kfree(name);
	} else {
		__putname(name);
	}
128 129
}

130 131
#define EMBEDDED_NAME_MAX	(PATH_MAX - sizeof(struct filename))

132 133 134 135
static struct filename *
getname_flags(const char __user *filename, int flags, int *empty)
{
	struct filename *result, *err;
136
	int len;
137 138
	long max;
	char *kname;
139

140 141 142 143
	result = audit_reusename(filename);
	if (result)
		return result;

144
	result = __getname();
145
	if (unlikely(!result))
146 147
		return ERR_PTR(-ENOMEM);

148 149 150 151 152
	/*
	 * First, try to embed the struct filename inside the names_cache
	 * allocation
	 */
	kname = (char *)result + sizeof(*result);
153
	result->name = kname;
154 155 156 157 158
	result->separate = false;
	max = EMBEDDED_NAME_MAX;

recopy:
	len = strncpy_from_user(kname, filename, max);
159 160
	if (unlikely(len < 0)) {
		err = ERR_PTR(len);
161
		goto error;
162
	}
163

164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
	/*
	 * Uh-oh. We have a name that's approaching PATH_MAX. Allocate a
	 * separate struct filename so we can dedicate the entire
	 * names_cache allocation for the pathname, and re-do the copy from
	 * userland.
	 */
	if (len == EMBEDDED_NAME_MAX && max == EMBEDDED_NAME_MAX) {
		kname = (char *)result;

		result = kzalloc(sizeof(*result), GFP_KERNEL);
		if (!result) {
			err = ERR_PTR(-ENOMEM);
			result = (struct filename *)kname;
			goto error;
		}
		result->name = kname;
		result->separate = true;
		max = PATH_MAX;
		goto recopy;
	}

185 186 187
	/* The empty path is special. */
	if (unlikely(!len)) {
		if (empty)
188
			*empty = 1;
189 190 191
		err = ERR_PTR(-ENOENT);
		if (!(flags & LOOKUP_EMPTY))
			goto error;
L
Linus Torvalds 已提交
192
	}
193 194

	err = ERR_PTR(-ENAMETOOLONG);
195 196 197 198 199 200
	if (unlikely(len >= PATH_MAX))
		goto error;

	result->uptr = filename;
	audit_getname(result);
	return result;
201 202

error:
203
	final_putname(result);
204
	return err;
L
Linus Torvalds 已提交
205 206
}

207 208
struct filename *
getname(const char __user * filename)
A
Al Viro 已提交
209
{
210
	return getname_flags(filename, 0, NULL);
A
Al Viro 已提交
211
}
212
EXPORT_SYMBOL(getname);
A
Al Viro 已提交
213

L
Linus Torvalds 已提交
214
#ifdef CONFIG_AUDITSYSCALL
215
void putname(struct filename *name)
L
Linus Torvalds 已提交
216
{
217
	if (unlikely(!audit_dummy_context()))
218 219
		return audit_putname(name);
	final_putname(name);
L
Linus Torvalds 已提交
220 221 222
}
#endif

223 224
static int check_acl(struct inode *inode, int mask)
{
225
#ifdef CONFIG_FS_POSIX_ACL
226 227 228
	struct posix_acl *acl;

	if (mask & MAY_NOT_BLOCK) {
229 230
		acl = get_cached_acl_rcu(inode, ACL_TYPE_ACCESS);
	        if (!acl)
231
	                return -EAGAIN;
232 233 234
		/* no ->get_acl() calls in RCU mode... */
		if (acl == ACL_NOT_CACHED)
			return -ECHILD;
A
Ari Savolainen 已提交
235
	        return posix_acl_permission(inode, acl, mask & ~MAY_NOT_BLOCK);
236 237 238 239 240
	}

	acl = get_cached_acl(inode, ACL_TYPE_ACCESS);

	/*
241 242 243
	 * A filesystem can force a ACL callback by just never filling the
	 * ACL cache. But normally you'd fill the cache either at inode
	 * instantiation time, or on the first ->get_acl call.
244
	 *
245 246
	 * If the filesystem doesn't have a get_acl() function at all, we'll
	 * just create the negative cache entry.
247 248
	 */
	if (acl == ACL_NOT_CACHED) {
249 250 251 252 253 254 255 256
	        if (inode->i_op->get_acl) {
			acl = inode->i_op->get_acl(inode, ACL_TYPE_ACCESS);
			if (IS_ERR(acl))
				return PTR_ERR(acl);
		} else {
		        set_cached_acl(inode, ACL_TYPE_ACCESS, NULL);
		        return -EAGAIN;
		}
257 258 259 260 261 262 263
	}

	if (acl) {
	        int error = posix_acl_permission(inode, acl, mask);
	        posix_acl_release(acl);
	        return error;
	}
264
#endif
265 266 267 268

	return -EAGAIN;
}

269
/*
270
 * This does the basic permission checking
L
Linus Torvalds 已提交
271
 */
272
static int acl_permission_check(struct inode *inode, int mask)
L
Linus Torvalds 已提交
273
{
274
	unsigned int mode = inode->i_mode;
L
Linus Torvalds 已提交
275

276
	if (likely(uid_eq(current_fsuid(), inode->i_uid)))
L
Linus Torvalds 已提交
277 278
		mode >>= 6;
	else {
279
		if (IS_POSIXACL(inode) && (mode & S_IRWXG)) {
280
			int error = check_acl(inode, mask);
281 282
			if (error != -EAGAIN)
				return error;
L
Linus Torvalds 已提交
283 284 285 286 287 288 289 290 291
		}

		if (in_group_p(inode->i_gid))
			mode >>= 3;
	}

	/*
	 * If the DACs are ok we don't need any capability check.
	 */
292
	if ((mask & ~mode & (MAY_READ | MAY_WRITE | MAY_EXEC)) == 0)
L
Linus Torvalds 已提交
293
		return 0;
294 295 296 297
	return -EACCES;
}

/**
298
 * generic_permission -  check for access rights on a Posix-like filesystem
299
 * @inode:	inode to check access rights for
300
 * @mask:	right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC, ...)
301 302 303 304
 *
 * Used to check for read/write/execute permissions on a file.
 * We use "fsuid" for this, letting us set arbitrary permissions
 * for filesystem access without changing the "normal" uids which
305 306 307 308 309
 * are used for other things.
 *
 * generic_permission is rcu-walk aware. It returns -ECHILD in case an rcu-walk
 * request cannot be satisfied (eg. requires blocking or too much complexity).
 * It would then be called again in ref-walk mode.
310
 */
311
int generic_permission(struct inode *inode, int mask)
312 313 314 315
{
	int ret;

	/*
316
	 * Do the basic permission checks.
317
	 */
318
	ret = acl_permission_check(inode, mask);
319 320
	if (ret != -EACCES)
		return ret;
L
Linus Torvalds 已提交
321

322 323
	if (S_ISDIR(inode->i_mode)) {
		/* DACs are overridable for directories */
324
		if (inode_capable(inode, CAP_DAC_OVERRIDE))
325 326
			return 0;
		if (!(mask & MAY_WRITE))
327
			if (inode_capable(inode, CAP_DAC_READ_SEARCH))
328 329 330
				return 0;
		return -EACCES;
	}
L
Linus Torvalds 已提交
331 332
	/*
	 * Read/write DACs are always overridable.
333 334
	 * Executable DACs are overridable when there is
	 * at least one exec bit set.
L
Linus Torvalds 已提交
335
	 */
336
	if (!(mask & MAY_EXEC) || (inode->i_mode & S_IXUGO))
337
		if (inode_capable(inode, CAP_DAC_OVERRIDE))
L
Linus Torvalds 已提交
338 339 340 341 342
			return 0;

	/*
	 * Searching includes executable on directories, else just read.
	 */
343
	mask &= MAY_READ | MAY_WRITE | MAY_EXEC;
344
	if (mask == MAY_READ)
345
		if (inode_capable(inode, CAP_DAC_READ_SEARCH))
L
Linus Torvalds 已提交
346 347 348 349 350
			return 0;

	return -EACCES;
}

351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370
/*
 * We _really_ want to just do "generic_permission()" without
 * even looking at the inode->i_op values. So we keep a cache
 * flag in inode->i_opflags, that says "this has not special
 * permission function, use the fast case".
 */
static inline int do_inode_permission(struct inode *inode, int mask)
{
	if (unlikely(!(inode->i_opflags & IOP_FASTPERM))) {
		if (likely(inode->i_op->permission))
			return inode->i_op->permission(inode, mask);

		/* This gets set once for the inode lifetime */
		spin_lock(&inode->i_lock);
		inode->i_opflags |= IOP_FASTPERM;
		spin_unlock(&inode->i_lock);
	}
	return generic_permission(inode, mask);
}

C
Christoph Hellwig 已提交
371
/**
D
David Howells 已提交
372 373 374
 * __inode_permission - Check for access rights to a given inode
 * @inode: Inode to check permission on
 * @mask: Right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC)
C
Christoph Hellwig 已提交
375
 *
D
David Howells 已提交
376
 * Check for read/write/execute permissions on an inode.
377 378
 *
 * When checking for MAY_APPEND, MAY_WRITE must also be set in @mask.
D
David Howells 已提交
379 380 381
 *
 * This does not check for a read-only file system.  You probably want
 * inode_permission().
C
Christoph Hellwig 已提交
382
 */
D
David Howells 已提交
383
int __inode_permission(struct inode *inode, int mask)
L
Linus Torvalds 已提交
384
{
385
	int retval;
L
Linus Torvalds 已提交
386

387
	if (unlikely(mask & MAY_WRITE)) {
L
Linus Torvalds 已提交
388 389 390 391 392 393 394
		/*
		 * Nobody gets write access to an immutable file.
		 */
		if (IS_IMMUTABLE(inode))
			return -EACCES;
	}

395
	retval = do_inode_permission(inode, mask);
L
Linus Torvalds 已提交
396 397 398
	if (retval)
		return retval;

399 400 401 402
	retval = devcgroup_inode_permission(inode, mask);
	if (retval)
		return retval;

403
	return security_inode_permission(inode, mask);
L
Linus Torvalds 已提交
404 405
}

D
David Howells 已提交
406 407 408
/**
 * sb_permission - Check superblock-level permissions
 * @sb: Superblock of inode to check permission on
409
 * @inode: Inode to check permission on
D
David Howells 已提交
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447
 * @mask: Right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC)
 *
 * Separate out file-system wide checks from inode-specific permission checks.
 */
static int sb_permission(struct super_block *sb, struct inode *inode, int mask)
{
	if (unlikely(mask & MAY_WRITE)) {
		umode_t mode = inode->i_mode;

		/* Nobody gets write access to a read-only fs. */
		if ((sb->s_flags & MS_RDONLY) &&
		    (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)))
			return -EROFS;
	}
	return 0;
}

/**
 * inode_permission - Check for access rights to a given inode
 * @inode: Inode to check permission on
 * @mask: Right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC)
 *
 * Check for read/write/execute permissions on an inode.  We use fs[ug]id for
 * this, letting us set arbitrary permissions for filesystem access without
 * changing the "normal" UIDs which are used for other things.
 *
 * When checking for MAY_APPEND, MAY_WRITE must also be set in @mask.
 */
int inode_permission(struct inode *inode, int mask)
{
	int retval;

	retval = sb_permission(inode->i_sb, inode, mask);
	if (retval)
		return retval;
	return __inode_permission(inode, mask);
}

J
Jan Blunck 已提交
448 449 450 451 452 453
/**
 * path_get - get a reference to a path
 * @path: path to get the reference to
 *
 * Given a path increment the reference count to the dentry and the vfsmount.
 */
454
void path_get(const struct path *path)
J
Jan Blunck 已提交
455 456 457 458 459 460
{
	mntget(path->mnt);
	dget(path->dentry);
}
EXPORT_SYMBOL(path_get);

J
Jan Blunck 已提交
461 462 463 464 465 466
/**
 * path_put - put a reference to a path
 * @path: path to put the reference to
 *
 * Given a path decrement the reference count to the dentry and the vfsmount.
 */
467
void path_put(const struct path *path)
L
Linus Torvalds 已提交
468
{
J
Jan Blunck 已提交
469 470
	dput(path->dentry);
	mntput(path->mnt);
L
Linus Torvalds 已提交
471
}
J
Jan Blunck 已提交
472
EXPORT_SYMBOL(path_put);
L
Linus Torvalds 已提交
473

A
Al Viro 已提交
474
/*
N
Nick Piggin 已提交
475
 * Path walking has 2 modes, rcu-walk and ref-walk (see
A
Al Viro 已提交
476 477 478 479 480 481 482
 * Documentation/filesystems/path-lookup.txt).  In situations when we can't
 * continue in RCU mode, we attempt to drop out of rcu-walk mode and grab
 * normal reference counts on dentries and vfsmounts to transition to rcu-walk
 * mode.  Refcounts are grabbed at the last known good point before rcu-walk
 * got stuck, so ref-walk may continue from there. If this is not successful
 * (eg. a seqcount has changed), then failure is returned and it's up to caller
 * to restart the path walk from the beginning in ref-walk mode.
N
Nick Piggin 已提交
483 484 485
 */

/**
A
Al Viro 已提交
486 487 488
 * unlazy_walk - try to switch to ref-walk mode.
 * @nd: nameidata pathwalk data
 * @dentry: child of nd->path.dentry or NULL
489
 * Returns: 0 on success, -ECHILD on failure
N
Nick Piggin 已提交
490
 *
A
Al Viro 已提交
491 492 493
 * unlazy_walk attempts to legitimize the current nd->path, nd->root and dentry
 * for ref-walk mode.  @dentry must be a path found by a do_lookup call on
 * @nd or NULL.  Must be called from rcu-walk context.
N
Nick Piggin 已提交
494
 */
A
Al Viro 已提交
495
static int unlazy_walk(struct nameidata *nd, struct dentry *dentry)
N
Nick Piggin 已提交
496 497 498 499 500
{
	struct fs_struct *fs = current->fs;
	struct dentry *parent = nd->path.dentry;

	BUG_ON(!(nd->flags & LOOKUP_RCU));
501 502

	/*
A
Al Viro 已提交
503 504 505 506 507 508
	 * After legitimizing the bastards, terminate_walk()
	 * will do the right thing for non-RCU mode, and all our
	 * subsequent exit cases should rcu_read_unlock()
	 * before returning.  Do vfsmount first; if dentry
	 * can't be legitimized, just set nd->path.dentry to NULL
	 * and rely on dput(NULL) being a no-op.
509
	 */
A
Al Viro 已提交
510
	if (!legitimize_mnt(nd->path.mnt, nd->m_seq))
511 512
		return -ECHILD;
	nd->flags &= ~LOOKUP_RCU;
513

A
Al Viro 已提交
514 515
	if (!lockref_get_not_dead(&parent->d_lockref)) {
		nd->path.dentry = NULL;	
A
Al Viro 已提交
516
		rcu_read_unlock();
A
Al Viro 已提交
517 518 519
		return -ECHILD;
	}

520 521 522 523 524 525 526 527 528 529 530
	/*
	 * For a negative lookup, the lookup sequence point is the parents
	 * sequence point, and it only needs to revalidate the parent dentry.
	 *
	 * For a positive lookup, we need to move both the parent and the
	 * dentry from the RCU domain to be properly refcounted. And the
	 * sequence number in the dentry validates *both* dentry counters,
	 * since we checked the sequence number of the parent after we got
	 * the child sequence number. So we know the parent must still
	 * be valid if the child sequence number is still valid.
	 */
A
Al Viro 已提交
531
	if (!dentry) {
532 533
		if (read_seqcount_retry(&parent->d_seq, nd->seq))
			goto out;
A
Al Viro 已提交
534 535
		BUG_ON(nd->inode != parent->d_inode);
	} else {
536 537 538 539
		if (!lockref_get_not_dead(&dentry->d_lockref))
			goto out;
		if (read_seqcount_retry(&dentry->d_seq, nd->seq))
			goto drop_dentry;
A
Al Viro 已提交
540
	}
541 542 543 544 545 546 547 548 549

	/*
	 * Sequence counts matched. Now make sure that the root is
	 * still valid and get it if required.
	 */
	if (nd->root.mnt && !(nd->flags & LOOKUP_ROOT)) {
		spin_lock(&fs->lock);
		if (nd->root.mnt != fs->root.mnt || nd->root.dentry != fs->root.dentry)
			goto unlock_and_drop_dentry;
N
Nick Piggin 已提交
550 551 552 553
		path_get(&nd->root);
		spin_unlock(&fs->lock);
	}

A
Al Viro 已提交
554
	rcu_read_unlock();
N
Nick Piggin 已提交
555
	return 0;
A
Al Viro 已提交
556

557 558 559
unlock_and_drop_dentry:
	spin_unlock(&fs->lock);
drop_dentry:
A
Al Viro 已提交
560
	rcu_read_unlock();
561
	dput(dentry);
562
	goto drop_root_mnt;
563
out:
A
Al Viro 已提交
564
	rcu_read_unlock();
565 566 567
drop_root_mnt:
	if (!(nd->flags & LOOKUP_ROOT))
		nd->root.mnt = NULL;
N
Nick Piggin 已提交
568 569 570
	return -ECHILD;
}

571
static inline int d_revalidate(struct dentry *dentry, unsigned int flags)
572
{
573
	return dentry->d_op->d_revalidate(dentry, flags);
574 575
}

576 577 578
/**
 * complete_walk - successful completion of path walk
 * @nd:  pointer nameidata
579
 *
580 581 582 583 584
 * If we had been in RCU mode, drop out of it and legitimize nd->path.
 * Revalidate the final result, unless we'd already done that during
 * the path walk or the filesystem doesn't ask for it.  Return 0 on
 * success, -error on failure.  In case of failure caller does not
 * need to drop nd->path.
585
 */
586
static int complete_walk(struct nameidata *nd)
587
{
A
Al Viro 已提交
588
	struct dentry *dentry = nd->path.dentry;
589 590
	int status;

591 592 593 594
	if (nd->flags & LOOKUP_RCU) {
		nd->flags &= ~LOOKUP_RCU;
		if (!(nd->flags & LOOKUP_ROOT))
			nd->root.mnt = NULL;
595

A
Al Viro 已提交
596
		if (!legitimize_mnt(nd->path.mnt, nd->m_seq)) {
A
Al Viro 已提交
597
			rcu_read_unlock();
A
Al Viro 已提交
598 599
			return -ECHILD;
		}
600
		if (unlikely(!lockref_get_not_dead(&dentry->d_lockref))) {
A
Al Viro 已提交
601
			rcu_read_unlock();
A
Al Viro 已提交
602
			mntput(nd->path.mnt);
603 604 605
			return -ECHILD;
		}
		if (read_seqcount_retry(&dentry->d_seq, nd->seq)) {
A
Al Viro 已提交
606
			rcu_read_unlock();
607
			dput(dentry);
A
Al Viro 已提交
608
			mntput(nd->path.mnt);
609 610
			return -ECHILD;
		}
A
Al Viro 已提交
611
		rcu_read_unlock();
612 613
	}

A
Al Viro 已提交
614 615 616
	if (likely(!(nd->flags & LOOKUP_JUMPED)))
		return 0;

617
	if (likely(!(dentry->d_flags & DCACHE_OP_WEAK_REVALIDATE)))
618 619
		return 0;

620
	status = dentry->d_op->d_weak_revalidate(dentry, nd->flags);
621 622 623
	if (status > 0)
		return 0;

A
Al Viro 已提交
624
	if (!status)
625
		status = -ESTALE;
A
Al Viro 已提交
626

627
	path_put(&nd->path);
628 629 630
	return status;
}

A
Al Viro 已提交
631 632
static __always_inline void set_root(struct nameidata *nd)
{
633 634
	if (!nd->root.mnt)
		get_fs_root(current->fs, &nd->root);
A
Al Viro 已提交
635 636
}

637 638
static int link_path_walk(const char *, struct nameidata *);

N
Nick Piggin 已提交
639 640 641 642
static __always_inline void set_root_rcu(struct nameidata *nd)
{
	if (!nd->root.mnt) {
		struct fs_struct *fs = current->fs;
N
Nick Piggin 已提交
643 644 645 646 647
		unsigned seq;

		do {
			seq = read_seqcount_begin(&fs->seq);
			nd->root = fs->root;
648
			nd->seq = __read_seqcount_begin(&nd->root.dentry->d_seq);
N
Nick Piggin 已提交
649
		} while (read_seqcount_retry(&fs->seq, seq));
N
Nick Piggin 已提交
650 651 652
	}
}

J
Jan Blunck 已提交
653
static void path_put_conditional(struct path *path, struct nameidata *nd)
654 655
{
	dput(path->dentry);
656
	if (path->mnt != nd->path.mnt)
657 658 659
		mntput(path->mnt);
}

660 661
static inline void path_to_nameidata(const struct path *path,
					struct nameidata *nd)
662
{
N
Nick Piggin 已提交
663 664 665 666
	if (!(nd->flags & LOOKUP_RCU)) {
		dput(nd->path.dentry);
		if (nd->path.mnt != path->mnt)
			mntput(nd->path.mnt);
667
	}
N
Nick Piggin 已提交
668
	nd->path.mnt = path->mnt;
669
	nd->path.dentry = path->dentry;
670 671
}

C
Christoph Hellwig 已提交
672 673 674 675 676 677 678 679 680 681 682 683 684
/*
 * Helper to directly jump to a known parsed path from ->follow_link,
 * caller must have taken a reference to path beforehand.
 */
void nd_jump_link(struct nameidata *nd, struct path *path)
{
	path_put(&nd->path);

	nd->path = *path;
	nd->inode = nd->path.dentry->d_inode;
	nd->flags |= LOOKUP_JUMPED;
}

685 686 687
static inline void put_link(struct nameidata *nd, struct path *link, void *cookie)
{
	struct inode *inode = link->dentry->d_inode;
688
	if (inode->i_op->put_link)
689 690 691 692
		inode->i_op->put_link(link->dentry, nd, cookie);
	path_put(link);
}

693 694
int sysctl_protected_symlinks __read_mostly = 0;
int sysctl_protected_hardlinks __read_mostly = 0;
K
Kees Cook 已提交
695 696 697 698

/**
 * may_follow_link - Check symlink following for unsafe situations
 * @link: The path of the symlink
699
 * @nd: nameidata pathwalk data
K
Kees Cook 已提交
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721
 *
 * In the case of the sysctl_protected_symlinks sysctl being enabled,
 * CAP_DAC_OVERRIDE needs to be specifically ignored if the symlink is
 * in a sticky world-writable directory. This is to protect privileged
 * processes from failing races against path names that may change out
 * from under them by way of other users creating malicious symlinks.
 * It will permit symlinks to be followed only when outside a sticky
 * world-writable directory, or when the uid of the symlink and follower
 * match, or when the directory owner matches the symlink's owner.
 *
 * Returns 0 if following the symlink is allowed, -ve on error.
 */
static inline int may_follow_link(struct path *link, struct nameidata *nd)
{
	const struct inode *inode;
	const struct inode *parent;

	if (!sysctl_protected_symlinks)
		return 0;

	/* Allowed if owner and follower match. */
	inode = link->dentry->d_inode;
722
	if (uid_eq(current_cred()->fsuid, inode->i_uid))
K
Kees Cook 已提交
723 724 725 726 727 728 729 730
		return 0;

	/* Allowed if parent directory not sticky and world-writable. */
	parent = nd->path.dentry->d_inode;
	if ((parent->i_mode & (S_ISVTX|S_IWOTH)) != (S_ISVTX|S_IWOTH))
		return 0;

	/* Allowed if parent directory and link owner match. */
731
	if (uid_eq(parent->i_uid, inode->i_uid))
K
Kees Cook 已提交
732 733
		return 0;

734
	audit_log_link_denied("follow_link", link);
K
Kees Cook 已提交
735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800
	path_put_conditional(link, nd);
	path_put(&nd->path);
	return -EACCES;
}

/**
 * safe_hardlink_source - Check for safe hardlink conditions
 * @inode: the source inode to hardlink from
 *
 * Return false if at least one of the following conditions:
 *    - inode is not a regular file
 *    - inode is setuid
 *    - inode is setgid and group-exec
 *    - access failure for read and write
 *
 * Otherwise returns true.
 */
static bool safe_hardlink_source(struct inode *inode)
{
	umode_t mode = inode->i_mode;

	/* Special files should not get pinned to the filesystem. */
	if (!S_ISREG(mode))
		return false;

	/* Setuid files should not get pinned to the filesystem. */
	if (mode & S_ISUID)
		return false;

	/* Executable setgid files should not get pinned to the filesystem. */
	if ((mode & (S_ISGID | S_IXGRP)) == (S_ISGID | S_IXGRP))
		return false;

	/* Hardlinking to unreadable or unwritable sources is dangerous. */
	if (inode_permission(inode, MAY_READ | MAY_WRITE))
		return false;

	return true;
}

/**
 * may_linkat - Check permissions for creating a hardlink
 * @link: the source to hardlink from
 *
 * Block hardlink when all of:
 *  - sysctl_protected_hardlinks enabled
 *  - fsuid does not match inode
 *  - hardlink source is unsafe (see safe_hardlink_source() above)
 *  - not CAP_FOWNER
 *
 * Returns 0 if successful, -ve on error.
 */
static int may_linkat(struct path *link)
{
	const struct cred *cred;
	struct inode *inode;

	if (!sysctl_protected_hardlinks)
		return 0;

	cred = current_cred();
	inode = link->dentry->d_inode;

	/* Source inode owner (or CAP_FOWNER) can hardlink all they like,
	 * otherwise, it must be a safe source.
	 */
801
	if (uid_eq(cred->fsuid, inode->i_uid) || safe_hardlink_source(inode) ||
K
Kees Cook 已提交
802 803 804
	    capable(CAP_FOWNER))
		return 0;

805
	audit_log_link_denied("linkat", link);
K
Kees Cook 已提交
806 807 808
	return -EPERM;
}

A
Al Viro 已提交
809
static __always_inline int
810
follow_link(struct path *link, struct nameidata *nd, void **p)
L
Linus Torvalds 已提交
811
{
812
	struct dentry *dentry = link->dentry;
813 814
	int error;
	char *s;
L
Linus Torvalds 已提交
815

816 817
	BUG_ON(nd->flags & LOOKUP_RCU);

A
Al Viro 已提交
818 819 820
	if (link->mnt == nd->path.mnt)
		mntget(link->mnt);

821 822 823 824
	error = -ELOOP;
	if (unlikely(current->total_link_count >= 40))
		goto out_put_nd_path;

825 826 827
	cond_resched();
	current->total_link_count++;

A
Al Viro 已提交
828
	touch_atime(link);
L
Linus Torvalds 已提交
829
	nd_set_link(nd, NULL);
A
Al Viro 已提交
830

831
	error = security_inode_follow_link(link->dentry, nd);
832 833
	if (error)
		goto out_put_nd_path;
834

835
	nd->last_type = LAST_BIND;
A
Al Viro 已提交
836 837
	*p = dentry->d_inode->i_op->follow_link(dentry, nd);
	error = PTR_ERR(*p);
838
	if (IS_ERR(*p))
839
		goto out_put_nd_path;
840 841 842 843

	error = 0;
	s = nd_get_link(nd);
	if (s) {
844 845 846 847 848 849 850 851 852 853 854 855 856 857
		if (unlikely(IS_ERR(s))) {
			path_put(&nd->path);
			put_link(nd, link, *p);
			return PTR_ERR(s);
		}
		if (*s == '/') {
			set_root(nd);
			path_put(&nd->path);
			nd->path = nd->root;
			path_get(&nd->root);
			nd->flags |= LOOKUP_JUMPED;
		}
		nd->inode = nd->path.dentry->d_inode;
		error = link_path_walk(s, nd);
C
Christoph Hellwig 已提交
858 859
		if (unlikely(error))
			put_link(nd, link, *p);
L
Linus Torvalds 已提交
860
	}
861 862 863 864

	return error;

out_put_nd_path:
A
Arnd Bergmann 已提交
865
	*p = NULL;
866 867
	path_put(&nd->path);
	path_put(link);
L
Linus Torvalds 已提交
868 869 870
	return error;
}

N
Nick Piggin 已提交
871 872
static int follow_up_rcu(struct path *path)
{
873 874
	struct mount *mnt = real_mount(path->mnt);
	struct mount *parent;
N
Nick Piggin 已提交
875 876
	struct dentry *mountpoint;

877 878
	parent = mnt->mnt_parent;
	if (&parent->mnt == path->mnt)
N
Nick Piggin 已提交
879
		return 0;
880
	mountpoint = mnt->mnt_mountpoint;
N
Nick Piggin 已提交
881
	path->dentry = mountpoint;
882
	path->mnt = &parent->mnt;
N
Nick Piggin 已提交
883 884 885
	return 1;
}

886 887 888 889 890 891 892 893 894 895
/*
 * follow_up - Find the mountpoint of path's vfsmount
 *
 * Given a path, find the mountpoint of its source file system.
 * Replace @path with the path of the mountpoint in the parent mount.
 * Up is towards /.
 *
 * Return 1 if we went up a level and 0 if we were already at the
 * root.
 */
A
Al Viro 已提交
896
int follow_up(struct path *path)
L
Linus Torvalds 已提交
897
{
898 899
	struct mount *mnt = real_mount(path->mnt);
	struct mount *parent;
L
Linus Torvalds 已提交
900
	struct dentry *mountpoint;
N
Nick Piggin 已提交
901

A
Al Viro 已提交
902
	read_seqlock_excl(&mount_lock);
903
	parent = mnt->mnt_parent;
A
Al Viro 已提交
904
	if (parent == mnt) {
A
Al Viro 已提交
905
		read_sequnlock_excl(&mount_lock);
L
Linus Torvalds 已提交
906 907
		return 0;
	}
908
	mntget(&parent->mnt);
909
	mountpoint = dget(mnt->mnt_mountpoint);
A
Al Viro 已提交
910
	read_sequnlock_excl(&mount_lock);
A
Al Viro 已提交
911 912 913
	dput(path->dentry);
	path->dentry = mountpoint;
	mntput(path->mnt);
914
	path->mnt = &parent->mnt;
L
Linus Torvalds 已提交
915 916 917
	return 1;
}

N
Nick Piggin 已提交
918
/*
919 920 921
 * Perform an automount
 * - return -EISDIR to tell follow_managed() to stop and return the path we
 *   were called with.
L
Linus Torvalds 已提交
922
 */
923 924
static int follow_automount(struct path *path, unsigned flags,
			    bool *need_mntput)
N
Nick Piggin 已提交
925
{
926
	struct vfsmount *mnt;
927
	int err;
928 929 930 931

	if (!path->dentry->d_op || !path->dentry->d_op->d_automount)
		return -EREMOTE;

932 933 934 935 936 937 938 939 940 941
	/* We don't want to mount if someone's just doing a stat -
	 * unless they're stat'ing a directory and appended a '/' to
	 * the name.
	 *
	 * We do, however, want to mount if someone wants to open or
	 * create a file of any type under the mountpoint, wants to
	 * traverse through the mountpoint or wants to open the
	 * mounted directory.  Also, autofs may mark negative dentries
	 * as being automount points.  These will need the attentions
	 * of the daemon to instantiate them before they can be used.
942
	 */
943
	if (!(flags & (LOOKUP_PARENT | LOOKUP_DIRECTORY |
944
		     LOOKUP_OPEN | LOOKUP_CREATE | LOOKUP_AUTOMOUNT)) &&
945 946 947
	    path->dentry->d_inode)
		return -EISDIR;

948 949 950 951 952 953 954 955 956 957 958 959 960 961 962
	current->total_link_count++;
	if (current->total_link_count >= 40)
		return -ELOOP;

	mnt = path->dentry->d_op->d_automount(path);
	if (IS_ERR(mnt)) {
		/*
		 * The filesystem is allowed to return -EISDIR here to indicate
		 * it doesn't want to automount.  For instance, autofs would do
		 * this so that its userspace daemon can mount on this dentry.
		 *
		 * However, we can only permit this if it's a terminal point in
		 * the path being looked up; if it wasn't then the remainder of
		 * the path is inaccessible and we should say so.
		 */
A
Al Viro 已提交
963
		if (PTR_ERR(mnt) == -EISDIR && (flags & LOOKUP_PARENT))
964 965
			return -EREMOTE;
		return PTR_ERR(mnt);
N
Nick Piggin 已提交
966
	}
967

968 969
	if (!mnt) /* mount collision */
		return 0;
N
Nick Piggin 已提交
970

971 972 973 974 975
	if (!*need_mntput) {
		/* lock_mount() may release path->mnt on error */
		mntget(path->mnt);
		*need_mntput = true;
	}
976
	err = finish_automount(mnt, path);
977

978 979 980
	switch (err) {
	case -EBUSY:
		/* Someone else made a mount here whilst we were busy */
981
		return 0;
982
	case 0:
983
		path_put(path);
984 985 986
		path->mnt = mnt;
		path->dentry = dget(mnt->mnt_root);
		return 0;
987 988
	default:
		return err;
989
	}
990

A
Al Viro 已提交
991 992
}

993 994
/*
 * Handle a dentry that is managed in some way.
995
 * - Flagged for transit management (autofs)
996 997 998 999 1000 1001 1002 1003
 * - Flagged as mountpoint
 * - Flagged as automount point
 *
 * This may only be called in refwalk mode.
 *
 * Serialization is taken care of in namespace.c
 */
static int follow_managed(struct path *path, unsigned flags)
L
Linus Torvalds 已提交
1004
{
1005
	struct vfsmount *mnt = path->mnt; /* held by caller, must be left alone */
1006 1007
	unsigned managed;
	bool need_mntput = false;
1008
	int ret = 0;
1009 1010 1011 1012 1013 1014 1015

	/* Given that we're not holding a lock here, we retain the value in a
	 * local variable for each dentry as we look at it so that we don't see
	 * the components of that value change under us */
	while (managed = ACCESS_ONCE(path->dentry->d_flags),
	       managed &= DCACHE_MANAGED_DENTRY,
	       unlikely(managed != 0)) {
1016 1017 1018 1019 1020
		/* Allow the filesystem to manage the transit without i_mutex
		 * being held. */
		if (managed & DCACHE_MANAGE_TRANSIT) {
			BUG_ON(!path->dentry->d_op);
			BUG_ON(!path->dentry->d_op->d_manage);
1021
			ret = path->dentry->d_op->d_manage(path->dentry, false);
1022
			if (ret < 0)
1023
				break;
1024 1025
		}

1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
		/* Transit to a mounted filesystem. */
		if (managed & DCACHE_MOUNTED) {
			struct vfsmount *mounted = lookup_mnt(path);
			if (mounted) {
				dput(path->dentry);
				if (need_mntput)
					mntput(path->mnt);
				path->mnt = mounted;
				path->dentry = dget(mounted->mnt_root);
				need_mntput = true;
				continue;
			}

			/* Something is mounted on this dentry in another
			 * namespace and/or whatever was mounted there in this
A
Al Viro 已提交
1041 1042
			 * namespace got unmounted before lookup_mnt() could
			 * get it */
1043 1044 1045 1046 1047 1048
		}

		/* Handle an automount point */
		if (managed & DCACHE_NEED_AUTOMOUNT) {
			ret = follow_automount(path, flags, &need_mntput);
			if (ret < 0)
1049
				break;
1050 1051 1052 1053 1054
			continue;
		}

		/* We didn't change the current path point */
		break;
L
Linus Torvalds 已提交
1055
	}
1056 1057 1058 1059 1060

	if (need_mntput && path->mnt == mnt)
		mntput(path->mnt);
	if (ret == -EISDIR)
		ret = 0;
1061
	return ret < 0 ? ret : need_mntput;
L
Linus Torvalds 已提交
1062 1063
}

1064
int follow_down_one(struct path *path)
L
Linus Torvalds 已提交
1065 1066 1067
{
	struct vfsmount *mounted;

A
Al Viro 已提交
1068
	mounted = lookup_mnt(path);
L
Linus Torvalds 已提交
1069
	if (mounted) {
A
Al Viro 已提交
1070 1071 1072 1073
		dput(path->dentry);
		mntput(path->mnt);
		path->mnt = mounted;
		path->dentry = dget(mounted->mnt_root);
L
Linus Torvalds 已提交
1074 1075 1076 1077 1078
		return 1;
	}
	return 0;
}

1079 1080 1081 1082 1083 1084
static inline bool managed_dentry_might_block(struct dentry *dentry)
{
	return (dentry->d_flags & DCACHE_MANAGE_TRANSIT &&
		dentry->d_op->d_manage(dentry, true) < 0);
}

1085
/*
1086 1087
 * Try to skip to top of mountpoint pile in rcuwalk mode.  Fail if
 * we meet a managed dentry that would need blocking.
1088 1089
 */
static bool __follow_mount_rcu(struct nameidata *nd, struct path *path,
1090
			       struct inode **inode)
1091
{
1092
	for (;;) {
1093
		struct mount *mounted;
1094 1095 1096 1097
		/*
		 * Don't forget we might have a non-mountpoint managed dentry
		 * that wants to block transit.
		 */
1098
		if (unlikely(managed_dentry_might_block(path->dentry)))
1099
			return false;
1100 1101 1102 1103

		if (!d_mountpoint(path->dentry))
			break;

A
Al Viro 已提交
1104
		mounted = __lookup_mnt(path->mnt, path->dentry);
1105 1106
		if (!mounted)
			break;
1107 1108
		path->mnt = &mounted->mnt;
		path->dentry = mounted->mnt.mnt_root;
1109
		nd->flags |= LOOKUP_JUMPED;
1110
		nd->seq = read_seqcount_begin(&path->dentry->d_seq);
1111 1112 1113 1114 1115 1116
		/*
		 * Update the inode too. We don't need to re-check the
		 * dentry sequence number here after this d_inode read,
		 * because a mount-point is always pinned.
		 */
		*inode = path->dentry->d_inode;
1117 1118 1119 1120
	}
	return true;
}

1121
static void follow_mount_rcu(struct nameidata *nd)
1122
{
1123
	while (d_mountpoint(nd->path.dentry)) {
1124
		struct mount *mounted;
A
Al Viro 已提交
1125
		mounted = __lookup_mnt(nd->path.mnt, nd->path.dentry);
1126 1127
		if (!mounted)
			break;
1128 1129
		nd->path.mnt = &mounted->mnt;
		nd->path.dentry = mounted->mnt.mnt_root;
1130
		nd->seq = read_seqcount_begin(&nd->path.dentry->d_seq);
1131 1132 1133
	}
}

N
Nick Piggin 已提交
1134 1135 1136 1137
static int follow_dotdot_rcu(struct nameidata *nd)
{
	set_root_rcu(nd);

1138
	while (1) {
N
Nick Piggin 已提交
1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
		if (nd->path.dentry == nd->root.dentry &&
		    nd->path.mnt == nd->root.mnt) {
			break;
		}
		if (nd->path.dentry != nd->path.mnt->mnt_root) {
			struct dentry *old = nd->path.dentry;
			struct dentry *parent = old->d_parent;
			unsigned seq;

			seq = read_seqcount_begin(&parent->d_seq);
			if (read_seqcount_retry(&old->d_seq, nd->seq))
1150
				goto failed;
N
Nick Piggin 已提交
1151 1152 1153 1154 1155 1156 1157 1158
			nd->path.dentry = parent;
			nd->seq = seq;
			break;
		}
		if (!follow_up_rcu(&nd->path))
			break;
		nd->seq = read_seqcount_begin(&nd->path.dentry->d_seq);
	}
1159 1160
	follow_mount_rcu(nd);
	nd->inode = nd->path.dentry->d_inode;
N
Nick Piggin 已提交
1161
	return 0;
1162 1163 1164

failed:
	nd->flags &= ~LOOKUP_RCU;
1165 1166
	if (!(nd->flags & LOOKUP_ROOT))
		nd->root.mnt = NULL;
A
Al Viro 已提交
1167
	rcu_read_unlock();
1168
	return -ECHILD;
N
Nick Piggin 已提交
1169 1170
}

1171 1172 1173 1174 1175
/*
 * Follow down to the covering mount currently visible to userspace.  At each
 * point, the filesystem owning that dentry may be queried as to whether the
 * caller is permitted to proceed or not.
 */
1176
int follow_down(struct path *path)
1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195
{
	unsigned managed;
	int ret;

	while (managed = ACCESS_ONCE(path->dentry->d_flags),
	       unlikely(managed & DCACHE_MANAGED_DENTRY)) {
		/* Allow the filesystem to manage the transit without i_mutex
		 * being held.
		 *
		 * We indicate to the filesystem if someone is trying to mount
		 * something here.  This gives autofs the chance to deny anyone
		 * other than its daemon the right to mount on its
		 * superstructure.
		 *
		 * The filesystem may sleep at this point.
		 */
		if (managed & DCACHE_MANAGE_TRANSIT) {
			BUG_ON(!path->dentry->d_op);
			BUG_ON(!path->dentry->d_op->d_manage);
1196
			ret = path->dentry->d_op->d_manage(
1197
				path->dentry, false);
1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
			if (ret < 0)
				return ret == -EISDIR ? 0 : ret;
		}

		/* Transit to a mounted filesystem. */
		if (managed & DCACHE_MOUNTED) {
			struct vfsmount *mounted = lookup_mnt(path);
			if (!mounted)
				break;
			dput(path->dentry);
			mntput(path->mnt);
			path->mnt = mounted;
			path->dentry = dget(mounted->mnt_root);
			continue;
		}

		/* Don't handle automount points here */
		break;
	}
	return 0;
}

1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
/*
 * Skip to top of mountpoint pile in refwalk mode for follow_dotdot()
 */
static void follow_mount(struct path *path)
{
	while (d_mountpoint(path->dentry)) {
		struct vfsmount *mounted = lookup_mnt(path);
		if (!mounted)
			break;
		dput(path->dentry);
		mntput(path->mnt);
		path->mnt = mounted;
		path->dentry = dget(mounted->mnt_root);
	}
}

N
Nick Piggin 已提交
1236
static void follow_dotdot(struct nameidata *nd)
L
Linus Torvalds 已提交
1237
{
A
Al Viro 已提交
1238
	set_root(nd);
1239

L
Linus Torvalds 已提交
1240
	while(1) {
1241
		struct dentry *old = nd->path.dentry;
L
Linus Torvalds 已提交
1242

A
Al Viro 已提交
1243 1244
		if (nd->path.dentry == nd->root.dentry &&
		    nd->path.mnt == nd->root.mnt) {
L
Linus Torvalds 已提交
1245 1246
			break;
		}
1247
		if (nd->path.dentry != nd->path.mnt->mnt_root) {
A
Al Viro 已提交
1248 1249
			/* rare case of legitimate dget_parent()... */
			nd->path.dentry = dget_parent(nd->path.dentry);
L
Linus Torvalds 已提交
1250 1251 1252
			dput(old);
			break;
		}
A
Al Viro 已提交
1253
		if (!follow_up(&nd->path))
L
Linus Torvalds 已提交
1254 1255
			break;
	}
A
Al Viro 已提交
1256
	follow_mount(&nd->path);
N
Nick Piggin 已提交
1257
	nd->inode = nd->path.dentry->d_inode;
L
Linus Torvalds 已提交
1258 1259
}

1260
/*
M
Miklos Szeredi 已提交
1261 1262 1263 1264 1265
 * This looks up the name in dcache, possibly revalidates the old dentry and
 * allocates a new one if not found or not valid.  In the need_lookup argument
 * returns whether i_op->lookup is necessary.
 *
 * dir->d_inode->i_mutex must be held
1266
 */
M
Miklos Szeredi 已提交
1267
static struct dentry *lookup_dcache(struct qstr *name, struct dentry *dir,
1268
				    unsigned int flags, bool *need_lookup)
1269 1270
{
	struct dentry *dentry;
M
Miklos Szeredi 已提交
1271
	int error;
1272

M
Miklos Szeredi 已提交
1273 1274 1275
	*need_lookup = false;
	dentry = d_lookup(dir, name);
	if (dentry) {
J
Jeff Layton 已提交
1276
		if (dentry->d_flags & DCACHE_OP_REVALIDATE) {
1277
			error = d_revalidate(dentry, flags);
M
Miklos Szeredi 已提交
1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288
			if (unlikely(error <= 0)) {
				if (error < 0) {
					dput(dentry);
					return ERR_PTR(error);
				} else if (!d_invalidate(dentry)) {
					dput(dentry);
					dentry = NULL;
				}
			}
		}
	}
1289

M
Miklos Szeredi 已提交
1290 1291 1292 1293
	if (!dentry) {
		dentry = d_alloc(dir, name);
		if (unlikely(!dentry))
			return ERR_PTR(-ENOMEM);
1294

M
Miklos Szeredi 已提交
1295
		*need_lookup = true;
1296 1297 1298 1299
	}
	return dentry;
}

1300
/*
1301 1302
 * Call i_op->lookup on the dentry.  The dentry must be negative and
 * unhashed.
M
Miklos Szeredi 已提交
1303 1304
 *
 * dir->d_inode->i_mutex must be held
1305
 */
M
Miklos Szeredi 已提交
1306
static struct dentry *lookup_real(struct inode *dir, struct dentry *dentry,
1307
				  unsigned int flags)
1308 1309 1310 1311
{
	struct dentry *old;

	/* Don't create child dentry for a dead directory. */
M
Miklos Szeredi 已提交
1312
	if (unlikely(IS_DEADDIR(dir))) {
1313
		dput(dentry);
1314
		return ERR_PTR(-ENOENT);
1315
	}
1316

1317
	old = dir->i_op->lookup(dir, dentry, flags);
1318 1319 1320 1321 1322 1323 1324
	if (unlikely(old)) {
		dput(dentry);
		dentry = old;
	}
	return dentry;
}

1325
static struct dentry *__lookup_hash(struct qstr *name,
1326
		struct dentry *base, unsigned int flags)
1327
{
M
Miklos Szeredi 已提交
1328
	bool need_lookup;
1329 1330
	struct dentry *dentry;

1331
	dentry = lookup_dcache(name, base, flags, &need_lookup);
M
Miklos Szeredi 已提交
1332 1333
	if (!need_lookup)
		return dentry;
1334

1335
	return lookup_real(base->d_inode, dentry, flags);
1336 1337
}

L
Linus Torvalds 已提交
1338 1339 1340 1341 1342
/*
 *  It's more convoluted than I'd like it to be, but... it's still fairly
 *  small and for now I'd prefer to have fast path as straight as possible.
 *  It _is_ time-critical.
 */
A
Al Viro 已提交
1343
static int lookup_fast(struct nameidata *nd,
M
Miklos Szeredi 已提交
1344
		       struct path *path, struct inode **inode)
L
Linus Torvalds 已提交
1345
{
1346
	struct vfsmount *mnt = nd->path.mnt;
N
Nick Piggin 已提交
1347
	struct dentry *dentry, *parent = nd->path.dentry;
A
Al Viro 已提交
1348 1349
	int need_reval = 1;
	int status = 1;
1350 1351
	int err;

1352 1353 1354 1355 1356
	/*
	 * Rename seqlock is not required here because in the off chance
	 * of a false negative due to a concurrent rename, we're going to
	 * do the non-racy lookup, below.
	 */
N
Nick Piggin 已提交
1357 1358
	if (nd->flags & LOOKUP_RCU) {
		unsigned seq;
1359
		dentry = __d_lookup_rcu(parent, &nd->last, &seq);
A
Al Viro 已提交
1360 1361 1362
		if (!dentry)
			goto unlazy;

1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377
		/*
		 * This sequence count validates that the inode matches
		 * the dentry name information from lookup.
		 */
		*inode = dentry->d_inode;
		if (read_seqcount_retry(&dentry->d_seq, seq))
			return -ECHILD;

		/*
		 * This sequence count validates that the parent had no
		 * changes while we did the lookup of the dentry above.
		 *
		 * The memory barrier in read_seqcount_begin of child is
		 *  enough, we can use __read_seqcount_retry here.
		 */
N
Nick Piggin 已提交
1378 1379 1380
		if (__read_seqcount_retry(&parent->d_seq, nd->seq))
			return -ECHILD;
		nd->seq = seq;
A
Al Viro 已提交
1381

1382
		if (unlikely(dentry->d_flags & DCACHE_OP_REVALIDATE)) {
1383
			status = d_revalidate(dentry, nd->flags);
A
Al Viro 已提交
1384 1385 1386 1387 1388
			if (unlikely(status <= 0)) {
				if (status != -ECHILD)
					need_reval = 0;
				goto unlazy;
			}
1389
		}
N
Nick Piggin 已提交
1390 1391
		path->mnt = mnt;
		path->dentry = dentry;
1392 1393 1394 1395 1396
		if (unlikely(!__follow_mount_rcu(nd, path, inode)))
			goto unlazy;
		if (unlikely(path->dentry->d_flags & DCACHE_NEED_AUTOMOUNT))
			goto unlazy;
		return 0;
A
Al Viro 已提交
1397
unlazy:
A
Al Viro 已提交
1398 1399
		if (unlazy_walk(nd, dentry))
			return -ECHILD;
A
Al Viro 已提交
1400
	} else {
A
Al Viro 已提交
1401
		dentry = __d_lookup(parent, &nd->last);
1402
	}
A
Al Viro 已提交
1403

1404 1405 1406
	if (unlikely(!dentry))
		goto need_lookup;

A
Al Viro 已提交
1407
	if (unlikely(dentry->d_flags & DCACHE_OP_REVALIDATE) && need_reval)
1408
		status = d_revalidate(dentry, nd->flags);
A
Al Viro 已提交
1409 1410 1411 1412 1413 1414 1415
	if (unlikely(status <= 0)) {
		if (status < 0) {
			dput(dentry);
			return status;
		}
		if (!d_invalidate(dentry)) {
			dput(dentry);
1416
			goto need_lookup;
A
Al Viro 已提交
1417
		}
1418
	}
M
Miklos Szeredi 已提交
1419

1420 1421 1422
	path->mnt = mnt;
	path->dentry = dentry;
	err = follow_managed(path, nd->flags);
1423 1424
	if (unlikely(err < 0)) {
		path_put_conditional(path, nd);
1425
		return err;
1426
	}
1427 1428
	if (err)
		nd->flags |= LOOKUP_JUMPED;
1429
	*inode = path->dentry->d_inode;
L
Linus Torvalds 已提交
1430
	return 0;
1431 1432

need_lookup:
M
Miklos Szeredi 已提交
1433 1434 1435 1436
	return 1;
}

/* Fast lookup failed, do it the slow way */
A
Al Viro 已提交
1437
static int lookup_slow(struct nameidata *nd, struct path *path)
M
Miklos Szeredi 已提交
1438 1439 1440 1441 1442
{
	struct dentry *dentry, *parent;
	int err;

	parent = nd->path.dentry;
1443 1444 1445
	BUG_ON(nd->inode != parent->d_inode);

	mutex_lock(&parent->d_inode->i_mutex);
A
Al Viro 已提交
1446
	dentry = __lookup_hash(&nd->last, parent, nd->flags);
1447 1448 1449
	mutex_unlock(&parent->d_inode->i_mutex);
	if (IS_ERR(dentry))
		return PTR_ERR(dentry);
M
Miklos Szeredi 已提交
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459
	path->mnt = nd->path.mnt;
	path->dentry = dentry;
	err = follow_managed(path, nd->flags);
	if (unlikely(err < 0)) {
		path_put_conditional(path, nd);
		return err;
	}
	if (err)
		nd->flags |= LOOKUP_JUMPED;
	return 0;
L
Linus Torvalds 已提交
1460 1461
}

1462 1463 1464
static inline int may_lookup(struct nameidata *nd)
{
	if (nd->flags & LOOKUP_RCU) {
1465
		int err = inode_permission(nd->inode, MAY_EXEC|MAY_NOT_BLOCK);
1466 1467
		if (err != -ECHILD)
			return err;
A
Al Viro 已提交
1468
		if (unlazy_walk(nd, NULL))
1469 1470
			return -ECHILD;
	}
1471
	return inode_permission(nd->inode, MAY_EXEC);
1472 1473
}

1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485
static inline int handle_dots(struct nameidata *nd, int type)
{
	if (type == LAST_DOTDOT) {
		if (nd->flags & LOOKUP_RCU) {
			if (follow_dotdot_rcu(nd))
				return -ECHILD;
		} else
			follow_dotdot(nd);
	}
	return 0;
}

1486 1487 1488 1489 1490 1491
static void terminate_walk(struct nameidata *nd)
{
	if (!(nd->flags & LOOKUP_RCU)) {
		path_put(&nd->path);
	} else {
		nd->flags &= ~LOOKUP_RCU;
1492 1493
		if (!(nd->flags & LOOKUP_ROOT))
			nd->root.mnt = NULL;
A
Al Viro 已提交
1494
		rcu_read_unlock();
1495 1496 1497
	}
}

1498 1499 1500 1501 1502 1503
/*
 * Do we need to follow links? We _really_ want to be able
 * to do this check without having to look at inode->i_op,
 * so we keep a cache of "no, this doesn't need follow_link"
 * for the common case.
 */
1504
static inline int should_follow_link(struct dentry *dentry, int follow)
1505
{
1506
	return unlikely(d_is_symlink(dentry)) ? follow : 0;
1507 1508
}

1509
static inline int walk_component(struct nameidata *nd, struct path *path,
1510
		int follow)
1511 1512 1513 1514 1515 1516 1517 1518
{
	struct inode *inode;
	int err;
	/*
	 * "." and ".." are special - ".." especially so because it has
	 * to be able to know about the current root directory and
	 * parent relationships.
	 */
1519 1520
	if (unlikely(nd->last_type != LAST_NORM))
		return handle_dots(nd, nd->last_type);
A
Al Viro 已提交
1521
	err = lookup_fast(nd, path, &inode);
1522
	if (unlikely(err)) {
M
Miklos Szeredi 已提交
1523 1524 1525
		if (err < 0)
			goto out_err;

A
Al Viro 已提交
1526
		err = lookup_slow(nd, path);
M
Miklos Szeredi 已提交
1527 1528 1529 1530
		if (err < 0)
			goto out_err;

		inode = path->dentry->d_inode;
1531
	}
M
Miklos Szeredi 已提交
1532 1533 1534 1535
	err = -ENOENT;
	if (!inode)
		goto out_path_put;

1536
	if (should_follow_link(path->dentry, follow)) {
A
Al Viro 已提交
1537 1538
		if (nd->flags & LOOKUP_RCU) {
			if (unlikely(unlazy_walk(nd, path->dentry))) {
M
Miklos Szeredi 已提交
1539 1540
				err = -ECHILD;
				goto out_err;
A
Al Viro 已提交
1541 1542
			}
		}
1543 1544 1545 1546 1547 1548
		BUG_ON(inode != path->dentry->d_inode);
		return 1;
	}
	path_to_nameidata(path, nd);
	nd->inode = inode;
	return 0;
M
Miklos Szeredi 已提交
1549 1550 1551 1552 1553 1554

out_path_put:
	path_to_nameidata(path, nd);
out_err:
	terminate_walk(nd);
	return err;
1555 1556
}

1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572
/*
 * This limits recursive symlink follows to 8, while
 * limiting consecutive symlinks to 40.
 *
 * Without that kind of total limit, nasty chains of consecutive
 * symlinks can cause almost arbitrarily long lookups.
 */
static inline int nested_symlink(struct path *path, struct nameidata *nd)
{
	int res;

	if (unlikely(current->link_count >= MAX_NESTED_LINKS)) {
		path_put_conditional(path, nd);
		path_put(&nd->path);
		return -ELOOP;
	}
1573
	BUG_ON(nd->depth >= MAX_NESTED_LINKS);
1574 1575 1576 1577 1578 1579 1580

	nd->depth++;
	current->link_count++;

	do {
		struct path link = *path;
		void *cookie;
1581 1582

		res = follow_link(&link, nd, &cookie);
1583 1584
		if (res)
			break;
1585
		res = walk_component(nd, path, LOOKUP_FOLLOW);
1586
		put_link(nd, &link, cookie);
1587 1588 1589 1590 1591 1592 1593
	} while (res > 0);

	current->link_count--;
	nd->depth--;
	return res;
}

1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617
/*
 * We can do the critical dentry name comparison and hashing
 * operations one word at a time, but we are limited to:
 *
 * - Architectures with fast unaligned word accesses. We could
 *   do a "get_unaligned()" if this helps and is sufficiently
 *   fast.
 *
 * - Little-endian machines (so that we can generate the mask
 *   of low bytes efficiently). Again, we *could* do a byte
 *   swapping load on big-endian architectures if that is not
 *   expensive enough to make the optimization worthless.
 *
 * - non-CONFIG_DEBUG_PAGEALLOC configurations (so that we
 *   do not trap on the (extremely unlikely) case of a page
 *   crossing operation.
 *
 * - Furthermore, we need an efficient 64-bit compile for the
 *   64-bit case in order to generate the "number of bytes in
 *   the final mask". Again, that could be replaced with a
 *   efficient population count instruction or similar.
 */
#ifdef CONFIG_DCACHE_WORD_ACCESS

1618
#include <asm/word-at-a-time.h>
1619

1620
#ifdef CONFIG_64BIT
1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639

static inline unsigned int fold_hash(unsigned long hash)
{
	hash += hash >> (8*sizeof(int));
	return hash;
}

#else	/* 32-bit case */

#define fold_hash(x) (x)

#endif

unsigned int full_name_hash(const unsigned char *name, unsigned int len)
{
	unsigned long a, mask;
	unsigned long hash = 0;

	for (;;) {
1640
		a = load_unaligned_zeropad(name);
1641 1642 1643
		if (len < sizeof(unsigned long))
			break;
		hash += a;
1644
		hash *= 9;
1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662
		name += sizeof(unsigned long);
		len -= sizeof(unsigned long);
		if (!len)
			goto done;
	}
	mask = ~(~0ul << len*8);
	hash += mask & a;
done:
	return fold_hash(hash);
}
EXPORT_SYMBOL(full_name_hash);

/*
 * Calculate the length and hash of the path component, and
 * return the length of the component;
 */
static inline unsigned long hash_name(const char *name, unsigned int *hashp)
{
1663 1664
	unsigned long a, b, adata, bdata, mask, hash, len;
	const struct word_at_a_time constants = WORD_AT_A_TIME_CONSTANTS;
1665 1666 1667 1668 1669 1670

	hash = a = 0;
	len = -sizeof(unsigned long);
	do {
		hash = (hash + a) * 9;
		len += sizeof(unsigned long);
1671
		a = load_unaligned_zeropad(name+len);
1672 1673 1674 1675 1676 1677 1678 1679 1680
		b = a ^ REPEAT_BYTE('/');
	} while (!(has_zero(a, &adata, &constants) | has_zero(b, &bdata, &constants)));

	adata = prep_zero_mask(a, adata, &constants);
	bdata = prep_zero_mask(b, bdata, &constants);

	mask = create_zero_mask(adata | bdata);

	hash += a & zero_bytemask(mask);
1681 1682
	*hashp = fold_hash(hash);

1683
	return len + find_zero(mask);
1684 1685 1686 1687
}

#else

L
Linus Torvalds 已提交
1688 1689 1690 1691 1692 1693 1694
unsigned int full_name_hash(const unsigned char *name, unsigned int len)
{
	unsigned long hash = init_name_hash();
	while (len--)
		hash = partial_name_hash(*name++, hash);
	return end_name_hash(hash);
}
1695
EXPORT_SYMBOL(full_name_hash);
L
Linus Torvalds 已提交
1696

1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715
/*
 * We know there's a real path component here of at least
 * one character.
 */
static inline unsigned long hash_name(const char *name, unsigned int *hashp)
{
	unsigned long hash = init_name_hash();
	unsigned long len = 0, c;

	c = (unsigned char)*name;
	do {
		len++;
		hash = partial_name_hash(c, hash);
		c = (unsigned char)name[len];
	} while (c && c != '/');
	*hashp = end_name_hash(hash);
	return len;
}

1716 1717
#endif

L
Linus Torvalds 已提交
1718 1719
/*
 * Name resolution.
1720 1721
 * This is the basic name resolution function, turning a pathname into
 * the final dentry. We expect 'base' to be positive and a directory.
L
Linus Torvalds 已提交
1722
 *
1723 1724
 * Returns 0 and nd will have valid dentry and mnt on success.
 * Returns error and drops reference to input namei data on failure.
L
Linus Torvalds 已提交
1725
 */
1726
static int link_path_walk(const char *name, struct nameidata *nd)
L
Linus Torvalds 已提交
1727 1728 1729 1730 1731 1732 1733
{
	struct path next;
	int err;
	
	while (*name=='/')
		name++;
	if (!*name)
1734
		return 0;
L
Linus Torvalds 已提交
1735 1736 1737 1738

	/* At this point we know we have a real path component. */
	for(;;) {
		struct qstr this;
1739
		long len;
A
Al Viro 已提交
1740
		int type;
L
Linus Torvalds 已提交
1741

1742
		err = may_lookup(nd);
L
Linus Torvalds 已提交
1743 1744 1745
 		if (err)
			break;

1746
		len = hash_name(name, &this.hash);
L
Linus Torvalds 已提交
1747
		this.name = name;
1748
		this.len = len;
L
Linus Torvalds 已提交
1749

A
Al Viro 已提交
1750
		type = LAST_NORM;
1751
		if (name[0] == '.') switch (len) {
A
Al Viro 已提交
1752
			case 2:
1753
				if (name[1] == '.') {
A
Al Viro 已提交
1754
					type = LAST_DOTDOT;
A
Al Viro 已提交
1755 1756
					nd->flags |= LOOKUP_JUMPED;
				}
A
Al Viro 已提交
1757 1758 1759 1760
				break;
			case 1:
				type = LAST_DOT;
		}
1761 1762
		if (likely(type == LAST_NORM)) {
			struct dentry *parent = nd->path.dentry;
A
Al Viro 已提交
1763
			nd->flags &= ~LOOKUP_JUMPED;
1764
			if (unlikely(parent->d_flags & DCACHE_OP_HASH)) {
1765
				err = parent->d_op->d_hash(parent, &this);
1766 1767 1768 1769
				if (err < 0)
					break;
			}
		}
A
Al Viro 已提交
1770

1771 1772 1773
		nd->last = this;
		nd->last_type = type;

1774
		if (!name[len])
1775
			return 0;
1776 1777 1778 1779 1780 1781 1782 1783
		/*
		 * If it wasn't NUL, we know it was '/'. Skip that
		 * slash, and continue until no more slashes.
		 */
		do {
			len++;
		} while (unlikely(name[len] == '/'));
		if (!name[len])
1784 1785
			return 0;

1786
		name += len;
L
Linus Torvalds 已提交
1787

1788
		err = walk_component(nd, &next, LOOKUP_FOLLOW);
1789 1790
		if (err < 0)
			return err;
L
Linus Torvalds 已提交
1791

1792
		if (err) {
1793
			err = nested_symlink(&next, nd);
L
Linus Torvalds 已提交
1794
			if (err)
1795
				return err;
N
Nick Piggin 已提交
1796
		}
1797
		if (!d_is_directory(nd->path.dentry)) {
1798 1799 1800
			err = -ENOTDIR; 
			break;
		}
L
Linus Torvalds 已提交
1801
	}
1802
	terminate_walk(nd);
L
Linus Torvalds 已提交
1803 1804 1805
	return err;
}

A
Al Viro 已提交
1806 1807
static int path_init(int dfd, const char *name, unsigned int flags,
		     struct nameidata *nd, struct file **fp)
N
Nick Piggin 已提交
1808 1809 1810 1811
{
	int retval = 0;

	nd->last_type = LAST_ROOT; /* if there are only slashes... */
A
Al Viro 已提交
1812
	nd->flags = flags | LOOKUP_JUMPED;
N
Nick Piggin 已提交
1813
	nd->depth = 0;
1814
	if (flags & LOOKUP_ROOT) {
1815 1816
		struct dentry *root = nd->root.dentry;
		struct inode *inode = root->d_inode;
A
Al Viro 已提交
1817
		if (*name) {
1818
			if (!d_is_directory(root))
A
Al Viro 已提交
1819 1820 1821 1822 1823
				return -ENOTDIR;
			retval = inode_permission(inode, MAY_EXEC);
			if (retval)
				return retval;
		}
1824 1825 1826
		nd->path = nd->root;
		nd->inode = inode;
		if (flags & LOOKUP_RCU) {
A
Al Viro 已提交
1827
			rcu_read_lock();
1828
			nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq);
A
Al Viro 已提交
1829
			nd->m_seq = read_seqbegin(&mount_lock);
1830 1831 1832 1833 1834 1835
		} else {
			path_get(&nd->path);
		}
		return 0;
	}

N
Nick Piggin 已提交
1836 1837
	nd->root.mnt = NULL;

A
Al Viro 已提交
1838
	nd->m_seq = read_seqbegin(&mount_lock);
N
Nick Piggin 已提交
1839
	if (*name=='/') {
A
Al Viro 已提交
1840
		if (flags & LOOKUP_RCU) {
A
Al Viro 已提交
1841
			rcu_read_lock();
A
Al Viro 已提交
1842 1843 1844 1845 1846 1847
			set_root_rcu(nd);
		} else {
			set_root(nd);
			path_get(&nd->root);
		}
		nd->path = nd->root;
N
Nick Piggin 已提交
1848
	} else if (dfd == AT_FDCWD) {
A
Al Viro 已提交
1849 1850 1851
		if (flags & LOOKUP_RCU) {
			struct fs_struct *fs = current->fs;
			unsigned seq;
N
Nick Piggin 已提交
1852

A
Al Viro 已提交
1853
			rcu_read_lock();
N
Nick Piggin 已提交
1854

A
Al Viro 已提交
1855 1856 1857 1858 1859 1860 1861 1862
			do {
				seq = read_seqcount_begin(&fs->seq);
				nd->path = fs->pwd;
				nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq);
			} while (read_seqcount_retry(&fs->seq, seq));
		} else {
			get_fs_pwd(current->fs, &nd->path);
		}
N
Nick Piggin 已提交
1863
	} else {
1864
		/* Caller must check execute permissions on the starting path component */
1865
		struct fd f = fdget_raw(dfd);
N
Nick Piggin 已提交
1866 1867
		struct dentry *dentry;

1868 1869
		if (!f.file)
			return -EBADF;
N
Nick Piggin 已提交
1870

1871
		dentry = f.file->f_path.dentry;
N
Nick Piggin 已提交
1872

A
Al Viro 已提交
1873
		if (*name) {
1874
			if (!d_is_directory(dentry)) {
1875 1876 1877
				fdput(f);
				return -ENOTDIR;
			}
A
Al Viro 已提交
1878
		}
N
Nick Piggin 已提交
1879

1880
		nd->path = f.file->f_path;
A
Al Viro 已提交
1881
		if (flags & LOOKUP_RCU) {
1882 1883
			if (f.need_put)
				*fp = f.file;
A
Al Viro 已提交
1884
			nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq);
A
Al Viro 已提交
1885
			rcu_read_lock();
A
Al Viro 已提交
1886
		} else {
1887 1888
			path_get(&nd->path);
			fdput(f);
A
Al Viro 已提交
1889
		}
N
Nick Piggin 已提交
1890 1891 1892
	}

	nd->inode = nd->path.dentry->d_inode;
1893 1894 1895
	return 0;
}

1896 1897 1898 1899 1900 1901
static inline int lookup_last(struct nameidata *nd, struct path *path)
{
	if (nd->last_type == LAST_NORM && nd->last.name[nd->last.len])
		nd->flags |= LOOKUP_FOLLOW | LOOKUP_DIRECTORY;

	nd->flags &= ~LOOKUP_PARENT;
1902
	return walk_component(nd, path, nd->flags & LOOKUP_FOLLOW);
1903 1904
}

1905
/* Returns 0 and nd will be valid on success; Retuns error, otherwise. */
A
Al Viro 已提交
1906
static int path_lookupat(int dfd, const char *name,
1907 1908
				unsigned int flags, struct nameidata *nd)
{
A
Al Viro 已提交
1909
	struct file *base = NULL;
1910 1911
	struct path path;
	int err;
N
Nick Piggin 已提交
1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926

	/*
	 * Path walking is largely split up into 2 different synchronisation
	 * schemes, rcu-walk and ref-walk (explained in
	 * Documentation/filesystems/path-lookup.txt). These share much of the
	 * path walk code, but some things particularly setup, cleanup, and
	 * following mounts are sufficiently divergent that functions are
	 * duplicated. Typically there is a function foo(), and its RCU
	 * analogue, foo_rcu().
	 *
	 * -ECHILD is the error number of choice (just to avoid clashes) that
	 * is returned if some aspect of an rcu-walk fails. Such an error must
	 * be handled by restarting a traditional ref-walk (which will always
	 * be able to complete).
	 */
1927
	err = path_init(dfd, name, flags | LOOKUP_PARENT, nd, &base);
A
Al Viro 已提交
1928

1929 1930
	if (unlikely(err))
		return err;
A
Al Viro 已提交
1931 1932

	current->total_link_count = 0;
1933 1934 1935 1936 1937 1938 1939
	err = link_path_walk(name, nd);

	if (!err && !(flags & LOOKUP_PARENT)) {
		err = lookup_last(nd, &path);
		while (err > 0) {
			void *cookie;
			struct path link = path;
K
Kees Cook 已提交
1940 1941 1942
			err = may_follow_link(&link, nd);
			if (unlikely(err))
				break;
1943
			nd->flags |= LOOKUP_PARENT;
1944
			err = follow_link(&link, nd, &cookie);
1945 1946 1947
			if (err)
				break;
			err = lookup_last(nd, &path);
1948
			put_link(nd, &link, cookie);
1949 1950
		}
	}
A
Al Viro 已提交
1951

1952 1953
	if (!err)
		err = complete_walk(nd);
1954 1955

	if (!err && nd->flags & LOOKUP_DIRECTORY) {
1956
		if (!d_is_directory(nd->path.dentry)) {
1957
			path_put(&nd->path);
A
Al Viro 已提交
1958
			err = -ENOTDIR;
1959 1960
		}
	}
A
Al Viro 已提交
1961

A
Al Viro 已提交
1962 1963
	if (base)
		fput(base);
A
Al Viro 已提交
1964

1965
	if (nd->root.mnt && !(nd->flags & LOOKUP_ROOT)) {
A
Al Viro 已提交
1966 1967 1968
		path_put(&nd->root);
		nd->root.mnt = NULL;
	}
1969
	return err;
A
Al Viro 已提交
1970
}
N
Nick Piggin 已提交
1971

1972
static int filename_lookup(int dfd, struct filename *name,
A
Al Viro 已提交
1973 1974
				unsigned int flags, struct nameidata *nd)
{
1975
	int retval = path_lookupat(dfd, name->name, flags | LOOKUP_RCU, nd);
A
Al Viro 已提交
1976
	if (unlikely(retval == -ECHILD))
1977
		retval = path_lookupat(dfd, name->name, flags, nd);
A
Al Viro 已提交
1978
	if (unlikely(retval == -ESTALE))
1979 1980
		retval = path_lookupat(dfd, name->name,
						flags | LOOKUP_REVAL, nd);
N
Nick Piggin 已提交
1981

1982
	if (likely(!retval))
1983
		audit_inode(name, nd->path.dentry, flags & LOOKUP_PARENT);
1984
	return retval;
L
Linus Torvalds 已提交
1985 1986
}

1987 1988 1989 1990 1991 1992 1993 1994
static int do_path_lookup(int dfd, const char *name,
				unsigned int flags, struct nameidata *nd)
{
	struct filename filename = { .name = name };

	return filename_lookup(dfd, &filename, flags, nd);
}

A
Al Viro 已提交
1995 1996
/* does lookup, returns the object with parent locked */
struct dentry *kern_path_locked(const char *name, struct path *path)
1997
{
A
Al Viro 已提交
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
	struct nameidata nd;
	struct dentry *d;
	int err = do_path_lookup(AT_FDCWD, name, LOOKUP_PARENT, &nd);
	if (err)
		return ERR_PTR(err);
	if (nd.last_type != LAST_NORM) {
		path_put(&nd.path);
		return ERR_PTR(-EINVAL);
	}
	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
2008
	d = __lookup_hash(&nd.last, nd.path.dentry, 0);
A
Al Viro 已提交
2009 2010 2011 2012 2013 2014 2015
	if (IS_ERR(d)) {
		mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
		path_put(&nd.path);
		return d;
	}
	*path = nd.path;
	return d;
2016 2017
}

A
Al Viro 已提交
2018 2019 2020 2021 2022 2023 2024 2025 2026
int kern_path(const char *name, unsigned int flags, struct path *path)
{
	struct nameidata nd;
	int res = do_path_lookup(AT_FDCWD, name, flags, &nd);
	if (!res)
		*path = nd.path;
	return res;
}

2027 2028 2029 2030 2031 2032
/**
 * vfs_path_lookup - lookup a file path relative to a dentry-vfsmount pair
 * @dentry:  pointer to dentry of the base directory
 * @mnt: pointer to vfs mount of the base directory
 * @name: pointer to file name
 * @flags: lookup flags
2033
 * @path: pointer to struct path to fill
2034 2035 2036
 */
int vfs_path_lookup(struct dentry *dentry, struct vfsmount *mnt,
		    const char *name, unsigned int flags,
2037
		    struct path *path)
2038
{
2039 2040 2041 2042 2043
	struct nameidata nd;
	int err;
	nd.root.dentry = dentry;
	nd.root.mnt = mnt;
	BUG_ON(flags & LOOKUP_PARENT);
2044
	/* the first argument of do_path_lookup() is ignored with LOOKUP_ROOT */
2045 2046 2047 2048
	err = do_path_lookup(AT_FDCWD, name, flags | LOOKUP_ROOT, &nd);
	if (!err)
		*path = nd.path;
	return err;
2049 2050
}

2051 2052 2053 2054 2055
/*
 * Restricted form of lookup. Doesn't follow links, single-component only,
 * needs parent already locked. Doesn't follow mounts.
 * SMP-safe.
 */
2056
static struct dentry *lookup_hash(struct nameidata *nd)
2057
{
2058
	return __lookup_hash(&nd->last, nd->path.dentry, nd->flags);
L
Linus Torvalds 已提交
2059 2060
}

2061
/**
2062
 * lookup_one_len - filesystem helper to lookup single pathname component
2063 2064 2065 2066
 * @name:	pathname component to lookup
 * @base:	base directory to lookup from
 * @len:	maximum length @len should be interpreted to
 *
2067 2068
 * Note that this routine is purely a helper for filesystem usage and should
 * not be called by generic code.  Also note that by using this function the
2069 2070 2071
 * nameidata argument is passed to the filesystem methods and a filesystem
 * using this helper needs to be prepared for that.
 */
2072 2073 2074
struct dentry *lookup_one_len(const char *name, struct dentry *base, int len)
{
	struct qstr this;
A
Al Viro 已提交
2075
	unsigned int c;
2076
	int err;
2077

2078 2079
	WARN_ON_ONCE(!mutex_is_locked(&base->d_inode->i_mutex));

A
Al Viro 已提交
2080 2081
	this.name = name;
	this.len = len;
L
Linus Torvalds 已提交
2082
	this.hash = full_name_hash(name, len);
A
Al Viro 已提交
2083 2084 2085
	if (!len)
		return ERR_PTR(-EACCES);

A
Al Viro 已提交
2086 2087 2088 2089 2090
	if (unlikely(name[0] == '.')) {
		if (len < 2 || (len == 2 && name[1] == '.'))
			return ERR_PTR(-EACCES);
	}

A
Al Viro 已提交
2091 2092 2093 2094 2095
	while (len--) {
		c = *(const unsigned char *)name++;
		if (c == '/' || c == '\0')
			return ERR_PTR(-EACCES);
	}
2096 2097 2098 2099 2100
	/*
	 * See if the low-level filesystem might want
	 * to use its own hash..
	 */
	if (base->d_flags & DCACHE_OP_HASH) {
2101
		int err = base->d_op->d_hash(base, &this);
2102 2103 2104
		if (err < 0)
			return ERR_PTR(err);
	}
2105

2106 2107 2108 2109
	err = inode_permission(base->d_inode, MAY_EXEC);
	if (err)
		return ERR_PTR(err);

2110
	return __lookup_hash(&this, base, 0);
2111 2112
}

2113 2114
int user_path_at_empty(int dfd, const char __user *name, unsigned flags,
		 struct path *path, int *empty)
L
Linus Torvalds 已提交
2115
{
2116
	struct nameidata nd;
2117
	struct filename *tmp = getname_flags(name, flags, empty);
L
Linus Torvalds 已提交
2118 2119
	int err = PTR_ERR(tmp);
	if (!IS_ERR(tmp)) {
2120 2121 2122

		BUG_ON(flags & LOOKUP_PARENT);

2123
		err = filename_lookup(dfd, tmp, flags, &nd);
L
Linus Torvalds 已提交
2124
		putname(tmp);
2125 2126
		if (!err)
			*path = nd.path;
L
Linus Torvalds 已提交
2127 2128 2129 2130
	}
	return err;
}

2131 2132 2133
int user_path_at(int dfd, const char __user *name, unsigned flags,
		 struct path *path)
{
2134
	return user_path_at_empty(dfd, name, flags, path, NULL);
2135 2136
}

2137 2138 2139 2140 2141 2142
/*
 * NB: most callers don't do anything directly with the reference to the
 *     to struct filename, but the nd->last pointer points into the name string
 *     allocated by getname. So we must hold the reference to it until all
 *     path-walking is complete.
 */
2143
static struct filename *
2144 2145
user_path_parent(int dfd, const char __user *path, struct nameidata *nd,
		 unsigned int flags)
2146
{
2147
	struct filename *s = getname(path);
2148 2149
	int error;

2150 2151 2152
	/* only LOOKUP_REVAL is allowed in extra flags */
	flags &= LOOKUP_REVAL;

2153
	if (IS_ERR(s))
2154
		return s;
2155

2156
	error = filename_lookup(dfd, s, flags | LOOKUP_PARENT, nd);
2157
	if (error) {
2158
		putname(s);
2159 2160
		return ERR_PTR(error);
	}
2161

2162
	return s;
2163 2164
}

2165
/**
2166
 * mountpoint_last - look up last component for umount
2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192
 * @nd:   pathwalk nameidata - currently pointing at parent directory of "last"
 * @path: pointer to container for result
 *
 * This is a special lookup_last function just for umount. In this case, we
 * need to resolve the path without doing any revalidation.
 *
 * The nameidata should be the result of doing a LOOKUP_PARENT pathwalk. Since
 * mountpoints are always pinned in the dcache, their ancestors are too. Thus,
 * in almost all cases, this lookup will be served out of the dcache. The only
 * cases where it won't are if nd->last refers to a symlink or the path is
 * bogus and it doesn't exist.
 *
 * Returns:
 * -error: if there was an error during lookup. This includes -ENOENT if the
 *         lookup found a negative dentry. The nd->path reference will also be
 *         put in this case.
 *
 * 0:      if we successfully resolved nd->path and found it to not to be a
 *         symlink that needs to be followed. "path" will also be populated.
 *         The nd->path reference will also be put.
 *
 * 1:      if we successfully resolved nd->last and found it to be a symlink
 *         that needs to be followed. "path" will be populated with the path
 *         to the link, and nd->path will *not* be put.
 */
static int
2193
mountpoint_last(struct nameidata *nd, struct path *path)
2194 2195 2196 2197 2198
{
	int error = 0;
	struct dentry *dentry;
	struct dentry *dir = nd->path.dentry;

2199 2200 2201 2202 2203 2204
	/* If we're in rcuwalk, drop out of it to handle last component */
	if (nd->flags & LOOKUP_RCU) {
		if (unlazy_walk(nd, NULL)) {
			error = -ECHILD;
			goto out;
		}
2205 2206 2207 2208 2209 2210
	}

	nd->flags &= ~LOOKUP_PARENT;

	if (unlikely(nd->last_type != LAST_NORM)) {
		error = handle_dots(nd, nd->last_type);
2211 2212 2213 2214
		if (error)
			goto out;
		dentry = dget(nd->path.dentry);
		goto done;
2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227
	}

	mutex_lock(&dir->d_inode->i_mutex);
	dentry = d_lookup(dir, &nd->last);
	if (!dentry) {
		/*
		 * No cached dentry. Mounted dentries are pinned in the cache,
		 * so that means that this dentry is probably a symlink or the
		 * path doesn't actually point to a mounted dentry.
		 */
		dentry = d_alloc(dir, &nd->last);
		if (!dentry) {
			error = -ENOMEM;
2228
			mutex_unlock(&dir->d_inode->i_mutex);
2229
			goto out;
2230
		}
2231 2232
		dentry = lookup_real(dir->d_inode, dentry, nd->flags);
		error = PTR_ERR(dentry);
2233 2234
		if (IS_ERR(dentry)) {
			mutex_unlock(&dir->d_inode->i_mutex);
2235
			goto out;
2236
		}
2237 2238 2239
	}
	mutex_unlock(&dir->d_inode->i_mutex);

2240 2241 2242 2243 2244
done:
	if (!dentry->d_inode) {
		error = -ENOENT;
		dput(dentry);
		goto out;
2245
	}
2246 2247
	path->dentry = dentry;
	path->mnt = mntget(nd->path.mnt);
2248
	if (should_follow_link(dentry, nd->flags & LOOKUP_FOLLOW))
2249 2250 2251 2252
		return 1;
	follow_mount(path);
	error = 0;
out:
2253 2254 2255 2256 2257
	terminate_walk(nd);
	return error;
}

/**
2258
 * path_mountpoint - look up a path to be umounted
2259 2260
 * @dfd:	directory file descriptor to start walk from
 * @name:	full pathname to walk
2261
 * @path:	pointer to container for result
2262 2263 2264
 * @flags:	lookup flags
 *
 * Look up the given name, but don't attempt to revalidate the last component.
2265
 * Returns 0 and "path" will be valid on success; Returns error otherwise.
2266 2267
 */
static int
2268
path_mountpoint(int dfd, const char *name, struct path *path, unsigned int flags)
2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282
{
	struct file *base = NULL;
	struct nameidata nd;
	int err;

	err = path_init(dfd, name, flags | LOOKUP_PARENT, &nd, &base);
	if (unlikely(err))
		return err;

	current->total_link_count = 0;
	err = link_path_walk(name, &nd);
	if (err)
		goto out;

2283
	err = mountpoint_last(&nd, path);
2284 2285 2286 2287 2288 2289 2290 2291 2292 2293
	while (err > 0) {
		void *cookie;
		struct path link = *path;
		err = may_follow_link(&link, &nd);
		if (unlikely(err))
			break;
		nd.flags |= LOOKUP_PARENT;
		err = follow_link(&link, &nd, &cookie);
		if (err)
			break;
2294
		err = mountpoint_last(&nd, path);
2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306
		put_link(&nd, &link, cookie);
	}
out:
	if (base)
		fput(base);

	if (nd.root.mnt && !(nd.flags & LOOKUP_ROOT))
		path_put(&nd.root);

	return err;
}

A
Al Viro 已提交
2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320
static int
filename_mountpoint(int dfd, struct filename *s, struct path *path,
			unsigned int flags)
{
	int error = path_mountpoint(dfd, s->name, path, flags | LOOKUP_RCU);
	if (unlikely(error == -ECHILD))
		error = path_mountpoint(dfd, s->name, path, flags);
	if (unlikely(error == -ESTALE))
		error = path_mountpoint(dfd, s->name, path, flags | LOOKUP_REVAL);
	if (likely(!error))
		audit_inode(s, path->dentry, 0);
	return error;
}

2321
/**
2322
 * user_path_mountpoint_at - lookup a path from userland in order to umount it
2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335
 * @dfd:	directory file descriptor
 * @name:	pathname from userland
 * @flags:	lookup flags
 * @path:	pointer to container to hold result
 *
 * A umount is a special case for path walking. We're not actually interested
 * in the inode in this situation, and ESTALE errors can be a problem. We
 * simply want track down the dentry and vfsmount attached at the mountpoint
 * and avoid revalidating the last component.
 *
 * Returns 0 and populates "path" on success.
 */
int
2336
user_path_mountpoint_at(int dfd, const char __user *name, unsigned int flags,
2337 2338 2339 2340 2341 2342
			struct path *path)
{
	struct filename *s = getname(name);
	int error;
	if (IS_ERR(s))
		return PTR_ERR(s);
A
Al Viro 已提交
2343
	error = filename_mountpoint(dfd, s, path, flags);
2344 2345 2346 2347
	putname(s);
	return error;
}

A
Al Viro 已提交
2348 2349 2350 2351 2352 2353 2354 2355 2356
int
kern_path_mountpoint(int dfd, const char *name, struct path *path,
			unsigned int flags)
{
	struct filename s = {.name = name};
	return filename_mountpoint(dfd, &s, path, flags);
}
EXPORT_SYMBOL(kern_path_mountpoint);

L
Linus Torvalds 已提交
2357 2358 2359 2360 2361 2362
/*
 * It's inline, so penalty for filesystems that don't use sticky bit is
 * minimal.
 */
static inline int check_sticky(struct inode *dir, struct inode *inode)
{
2363
	kuid_t fsuid = current_fsuid();
2364

L
Linus Torvalds 已提交
2365 2366
	if (!(dir->i_mode & S_ISVTX))
		return 0;
2367
	if (uid_eq(inode->i_uid, fsuid))
L
Linus Torvalds 已提交
2368
		return 0;
2369
	if (uid_eq(dir->i_uid, fsuid))
L
Linus Torvalds 已提交
2370
		return 0;
2371
	return !inode_capable(inode, CAP_FOWNER);
L
Linus Torvalds 已提交
2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392
}

/*
 *	Check whether we can remove a link victim from directory dir, check
 *  whether the type of victim is right.
 *  1. We can't do it if dir is read-only (done in permission())
 *  2. We should have write and exec permissions on dir
 *  3. We can't remove anything from append-only dir
 *  4. We can't do anything with immutable dir (done in permission())
 *  5. If the sticky bit on dir is set we should either
 *	a. be owner of dir, or
 *	b. be owner of victim, or
 *	c. have CAP_FOWNER capability
 *  6. If the victim is append-only or immutable we can't do antyhing with
 *     links pointing to it.
 *  7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
 *  8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
 *  9. We can't remove a root or mountpoint.
 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
 *     nfs_async_unlink().
 */
2393
static int may_delete(struct inode *dir, struct dentry *victim, bool isdir)
L
Linus Torvalds 已提交
2394
{
2395
	struct inode *inode = victim->d_inode;
L
Linus Torvalds 已提交
2396 2397
	int error;

2398
	if (d_is_negative(victim))
L
Linus Torvalds 已提交
2399
		return -ENOENT;
2400
	BUG_ON(!inode);
L
Linus Torvalds 已提交
2401 2402

	BUG_ON(victim->d_parent->d_inode != dir);
2403
	audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE);
L
Linus Torvalds 已提交
2404

2405
	error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
L
Linus Torvalds 已提交
2406 2407 2408 2409
	if (error)
		return error;
	if (IS_APPEND(dir))
		return -EPERM;
2410 2411 2412

	if (check_sticky(dir, inode) || IS_APPEND(inode) ||
	    IS_IMMUTABLE(inode) || IS_SWAPFILE(inode))
L
Linus Torvalds 已提交
2413 2414
		return -EPERM;
	if (isdir) {
2415
		if (!d_is_directory(victim) && !d_is_autodir(victim))
L
Linus Torvalds 已提交
2416 2417 2418
			return -ENOTDIR;
		if (IS_ROOT(victim))
			return -EBUSY;
2419
	} else if (d_is_directory(victim) || d_is_autodir(victim))
L
Linus Torvalds 已提交
2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435
		return -EISDIR;
	if (IS_DEADDIR(dir))
		return -ENOENT;
	if (victim->d_flags & DCACHE_NFSFS_RENAMED)
		return -EBUSY;
	return 0;
}

/*	Check whether we can create an object with dentry child in directory
 *  dir.
 *  1. We can't do it if child already exists (open has special treatment for
 *     this case, but since we are inlined it's OK)
 *  2. We can't do it if dir is read-only (done in permission())
 *  3. We should have write and exec permissions on dir
 *  4. We can't do it if dir is immutable (done in permission())
 */
2436
static inline int may_create(struct inode *dir, struct dentry *child)
L
Linus Torvalds 已提交
2437
{
2438
	audit_inode_child(dir, child, AUDIT_TYPE_CHILD_CREATE);
L
Linus Torvalds 已提交
2439 2440 2441 2442
	if (child->d_inode)
		return -EEXIST;
	if (IS_DEADDIR(dir))
		return -ENOENT;
2443
	return inode_permission(dir, MAY_WRITE | MAY_EXEC);
L
Linus Torvalds 已提交
2444 2445 2446 2447 2448 2449 2450 2451 2452 2453
}

/*
 * p1 and p2 should be directories on the same fs.
 */
struct dentry *lock_rename(struct dentry *p1, struct dentry *p2)
{
	struct dentry *p;

	if (p1 == p2) {
I
Ingo Molnar 已提交
2454
		mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
2455 2456 2457
		return NULL;
	}

2458
	mutex_lock(&p1->d_inode->i_sb->s_vfs_rename_mutex);
L
Linus Torvalds 已提交
2459

2460 2461 2462 2463 2464
	p = d_ancestor(p2, p1);
	if (p) {
		mutex_lock_nested(&p2->d_inode->i_mutex, I_MUTEX_PARENT);
		mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_CHILD);
		return p;
L
Linus Torvalds 已提交
2465 2466
	}

2467 2468 2469 2470 2471
	p = d_ancestor(p1, p2);
	if (p) {
		mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_PARENT);
		mutex_lock_nested(&p2->d_inode->i_mutex, I_MUTEX_CHILD);
		return p;
L
Linus Torvalds 已提交
2472 2473
	}

I
Ingo Molnar 已提交
2474 2475
	mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_PARENT);
	mutex_lock_nested(&p2->d_inode->i_mutex, I_MUTEX_CHILD);
L
Linus Torvalds 已提交
2476 2477 2478 2479 2480
	return NULL;
}

void unlock_rename(struct dentry *p1, struct dentry *p2)
{
2481
	mutex_unlock(&p1->d_inode->i_mutex);
L
Linus Torvalds 已提交
2482
	if (p1 != p2) {
2483
		mutex_unlock(&p2->d_inode->i_mutex);
2484
		mutex_unlock(&p1->d_inode->i_sb->s_vfs_rename_mutex);
L
Linus Torvalds 已提交
2485 2486 2487
	}
}

A
Al Viro 已提交
2488
int vfs_create(struct inode *dir, struct dentry *dentry, umode_t mode,
A
Al Viro 已提交
2489
		bool want_excl)
L
Linus Torvalds 已提交
2490
{
2491
	int error = may_create(dir, dentry);
L
Linus Torvalds 已提交
2492 2493 2494
	if (error)
		return error;

A
Al Viro 已提交
2495
	if (!dir->i_op->create)
L
Linus Torvalds 已提交
2496 2497 2498 2499 2500 2501
		return -EACCES;	/* shouldn't it be ENOSYS? */
	mode &= S_IALLUGO;
	mode |= S_IFREG;
	error = security_inode_create(dir, dentry, mode);
	if (error)
		return error;
A
Al Viro 已提交
2502
	error = dir->i_op->create(dir, dentry, mode, want_excl);
2503
	if (!error)
2504
		fsnotify_create(dir, dentry);
L
Linus Torvalds 已提交
2505 2506 2507
	return error;
}

A
Al Viro 已提交
2508
static int may_open(struct path *path, int acc_mode, int flag)
L
Linus Torvalds 已提交
2509
{
2510
	struct dentry *dentry = path->dentry;
L
Linus Torvalds 已提交
2511 2512 2513
	struct inode *inode = dentry->d_inode;
	int error;

A
Al Viro 已提交
2514 2515 2516 2517
	/* O_PATH? */
	if (!acc_mode)
		return 0;

L
Linus Torvalds 已提交
2518 2519 2520
	if (!inode)
		return -ENOENT;

C
Christoph Hellwig 已提交
2521 2522
	switch (inode->i_mode & S_IFMT) {
	case S_IFLNK:
L
Linus Torvalds 已提交
2523
		return -ELOOP;
C
Christoph Hellwig 已提交
2524 2525 2526 2527 2528 2529
	case S_IFDIR:
		if (acc_mode & MAY_WRITE)
			return -EISDIR;
		break;
	case S_IFBLK:
	case S_IFCHR:
2530
		if (path->mnt->mnt_flags & MNT_NODEV)
L
Linus Torvalds 已提交
2531
			return -EACCES;
C
Christoph Hellwig 已提交
2532 2533 2534
		/*FALLTHRU*/
	case S_IFIFO:
	case S_IFSOCK:
L
Linus Torvalds 已提交
2535
		flag &= ~O_TRUNC;
C
Christoph Hellwig 已提交
2536
		break;
2537
	}
2538

2539
	error = inode_permission(inode, acc_mode);
2540 2541
	if (error)
		return error;
M
Mimi Zohar 已提交
2542

L
Linus Torvalds 已提交
2543 2544 2545 2546
	/*
	 * An append-only file must be opened in append mode for writing.
	 */
	if (IS_APPEND(inode)) {
2547
		if  ((flag & O_ACCMODE) != O_RDONLY && !(flag & O_APPEND))
2548
			return -EPERM;
L
Linus Torvalds 已提交
2549
		if (flag & O_TRUNC)
2550
			return -EPERM;
L
Linus Torvalds 已提交
2551 2552 2553
	}

	/* O_NOATIME can only be set by the owner or superuser */
2554
	if (flag & O_NOATIME && !inode_owner_or_capable(inode))
2555
		return -EPERM;
L
Linus Torvalds 已提交
2556

2557
	return 0;
2558
}
L
Linus Torvalds 已提交
2559

2560
static int handle_truncate(struct file *filp)
2561
{
2562
	struct path *path = &filp->f_path;
2563 2564 2565 2566 2567 2568 2569 2570 2571
	struct inode *inode = path->dentry->d_inode;
	int error = get_write_access(inode);
	if (error)
		return error;
	/*
	 * Refuse to truncate files with mandatory locks held on them.
	 */
	error = locks_verify_locked(inode);
	if (!error)
2572
		error = security_path_truncate(path);
2573 2574 2575
	if (!error) {
		error = do_truncate(path->dentry, 0,
				    ATTR_MTIME|ATTR_CTIME|ATTR_OPEN,
2576
				    filp);
2577 2578
	}
	put_write_access(inode);
M
Mimi Zohar 已提交
2579
	return error;
L
Linus Torvalds 已提交
2580 2581
}

2582 2583
static inline int open_to_namei_flags(int flag)
{
2584 2585
	if ((flag & O_ACCMODE) == 3)
		flag--;
2586 2587 2588
	return flag;
}

M
Miklos Szeredi 已提交
2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601
static int may_o_create(struct path *dir, struct dentry *dentry, umode_t mode)
{
	int error = security_path_mknod(dir, dentry, mode, 0);
	if (error)
		return error;

	error = inode_permission(dir->dentry->d_inode, MAY_WRITE | MAY_EXEC);
	if (error)
		return error;

	return security_inode_create(dir->dentry->d_inode, dentry, mode);
}

2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614
/*
 * Attempt to atomically look up, create and open a file from a negative
 * dentry.
 *
 * Returns 0 if successful.  The file will have been created and attached to
 * @file by the filesystem calling finish_open().
 *
 * Returns 1 if the file was looked up only or didn't need creating.  The
 * caller will need to perform the open themselves.  @path will have been
 * updated to point to the new dentry.  This may be negative.
 *
 * Returns an error code otherwise.
 */
2615 2616 2617
static int atomic_open(struct nameidata *nd, struct dentry *dentry,
			struct path *path, struct file *file,
			const struct open_flags *op,
2618
			bool got_write, bool need_lookup,
2619
			int *opened)
M
Miklos Szeredi 已提交
2620 2621 2622 2623 2624 2625 2626 2627
{
	struct inode *dir =  nd->path.dentry->d_inode;
	unsigned open_flag = open_to_namei_flags(op->open_flag);
	umode_t mode;
	int error;
	int acc_mode;
	int create_error = 0;
	struct dentry *const DENTRY_NOT_SET = (void *) -1UL;
2628
	bool excl;
M
Miklos Szeredi 已提交
2629 2630 2631 2632 2633

	BUG_ON(dentry->d_inode);

	/* Don't create child dentry for a dead directory. */
	if (unlikely(IS_DEADDIR(dir))) {
2634
		error = -ENOENT;
M
Miklos Szeredi 已提交
2635 2636 2637
		goto out;
	}

2638
	mode = op->mode;
M
Miklos Szeredi 已提交
2639 2640 2641
	if ((open_flag & O_CREAT) && !IS_POSIXACL(dir))
		mode &= ~current_umask();

2642 2643
	excl = (open_flag & (O_EXCL | O_CREAT)) == (O_EXCL | O_CREAT);
	if (excl)
M
Miklos Szeredi 已提交
2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654
		open_flag &= ~O_TRUNC;

	/*
	 * Checking write permission is tricky, bacuse we don't know if we are
	 * going to actually need it: O_CREAT opens should work as long as the
	 * file exists.  But checking existence breaks atomicity.  The trick is
	 * to check access and if not granted clear O_CREAT from the flags.
	 *
	 * Another problem is returing the "right" error value (e.g. for an
	 * O_EXCL open we want to return EEXIST not EROFS).
	 */
2655 2656 2657
	if (((open_flag & (O_CREAT | O_TRUNC)) ||
	    (open_flag & O_ACCMODE) != O_RDONLY) && unlikely(!got_write)) {
		if (!(open_flag & O_CREAT)) {
M
Miklos Szeredi 已提交
2658 2659 2660 2661 2662 2663 2664
			/*
			 * No O_CREATE -> atomicity not a requirement -> fall
			 * back to lookup + open
			 */
			goto no_open;
		} else if (open_flag & (O_EXCL | O_TRUNC)) {
			/* Fall back and fail with the right error */
2665
			create_error = -EROFS;
M
Miklos Szeredi 已提交
2666 2667 2668
			goto no_open;
		} else {
			/* No side effects, safe to clear O_CREAT */
2669
			create_error = -EROFS;
M
Miklos Szeredi 已提交
2670 2671 2672 2673 2674
			open_flag &= ~O_CREAT;
		}
	}

	if (open_flag & O_CREAT) {
2675
		error = may_o_create(&nd->path, dentry, mode);
M
Miklos Szeredi 已提交
2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686
		if (error) {
			create_error = error;
			if (open_flag & O_EXCL)
				goto no_open;
			open_flag &= ~O_CREAT;
		}
	}

	if (nd->flags & LOOKUP_DIRECTORY)
		open_flag |= O_DIRECTORY;

A
Al Viro 已提交
2687 2688 2689
	file->f_path.dentry = DENTRY_NOT_SET;
	file->f_path.mnt = nd->path.mnt;
	error = dir->i_op->atomic_open(dir, dentry, file, open_flag, mode,
2690
				      opened);
A
Al Viro 已提交
2691 2692 2693
	if (error < 0) {
		if (create_error && error == -ENOENT)
			error = create_error;
M
Miklos Szeredi 已提交
2694 2695 2696
		goto out;
	}

A
Al Viro 已提交
2697
	if (error) {	/* returned 1, that is */
A
Al Viro 已提交
2698
		if (WARN_ON(file->f_path.dentry == DENTRY_NOT_SET)) {
2699
			error = -EIO;
M
Miklos Szeredi 已提交
2700 2701
			goto out;
		}
A
Al Viro 已提交
2702
		if (file->f_path.dentry) {
M
Miklos Szeredi 已提交
2703
			dput(dentry);
A
Al Viro 已提交
2704
			dentry = file->f_path.dentry;
M
Miklos Szeredi 已提交
2705
		}
2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718
		if (*opened & FILE_CREATED)
			fsnotify_create(dir, dentry);
		if (!dentry->d_inode) {
			WARN_ON(*opened & FILE_CREATED);
			if (create_error) {
				error = create_error;
				goto out;
			}
		} else {
			if (excl && !(*opened & FILE_CREATED)) {
				error = -EEXIST;
				goto out;
			}
2719
		}
M
Miklos Szeredi 已提交
2720 2721 2722 2723 2724 2725 2726
		goto looked_up;
	}

	/*
	 * We didn't have the inode before the open, so check open permission
	 * here.
	 */
2727 2728 2729 2730 2731 2732
	acc_mode = op->acc_mode;
	if (*opened & FILE_CREATED) {
		WARN_ON(!(open_flag & O_CREAT));
		fsnotify_create(dir, dentry);
		acc_mode = MAY_OPEN;
	}
2733 2734 2735
	error = may_open(&file->f_path, acc_mode, open_flag);
	if (error)
		fput(file);
M
Miklos Szeredi 已提交
2736 2737 2738

out:
	dput(dentry);
2739
	return error;
M
Miklos Szeredi 已提交
2740 2741 2742

no_open:
	if (need_lookup) {
2743
		dentry = lookup_real(dir, dentry, nd->flags);
M
Miklos Szeredi 已提交
2744
		if (IS_ERR(dentry))
2745
			return PTR_ERR(dentry);
M
Miklos Szeredi 已提交
2746 2747 2748 2749

		if (create_error) {
			int open_flag = op->open_flag;

2750
			error = create_error;
M
Miklos Szeredi 已提交
2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765
			if ((open_flag & O_EXCL)) {
				if (!dentry->d_inode)
					goto out;
			} else if (!dentry->d_inode) {
				goto out;
			} else if ((open_flag & O_TRUNC) &&
				   S_ISREG(dentry->d_inode->i_mode)) {
				goto out;
			}
			/* will fail later, go on to get the right error */
		}
	}
looked_up:
	path->dentry = dentry;
	path->mnt = nd->path.mnt;
2766
	return 1;
M
Miklos Szeredi 已提交
2767 2768
}

M
Miklos Szeredi 已提交
2769
/*
2770
 * Look up and maybe create and open the last component.
M
Miklos Szeredi 已提交
2771 2772 2773
 *
 * Must be called with i_mutex held on parent.
 *
2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785
 * Returns 0 if the file was successfully atomically created (if necessary) and
 * opened.  In this case the file will be returned attached to @file.
 *
 * Returns 1 if the file was not completely opened at this time, though lookups
 * and creations will have been performed and the dentry returned in @path will
 * be positive upon return if O_CREAT was specified.  If O_CREAT wasn't
 * specified then a negative dentry may be returned.
 *
 * An error code is returned otherwise.
 *
 * FILE_CREATE will be set in @*opened if the dentry was created and will be
 * cleared otherwise prior to returning.
M
Miklos Szeredi 已提交
2786
 */
2787 2788 2789
static int lookup_open(struct nameidata *nd, struct path *path,
			struct file *file,
			const struct open_flags *op,
2790
			bool got_write, int *opened)
M
Miklos Szeredi 已提交
2791 2792
{
	struct dentry *dir = nd->path.dentry;
2793
	struct inode *dir_inode = dir->d_inode;
M
Miklos Szeredi 已提交
2794 2795
	struct dentry *dentry;
	int error;
2796
	bool need_lookup;
M
Miklos Szeredi 已提交
2797

2798
	*opened &= ~FILE_CREATED;
2799
	dentry = lookup_dcache(&nd->last, dir, nd->flags, &need_lookup);
M
Miklos Szeredi 已提交
2800
	if (IS_ERR(dentry))
2801
		return PTR_ERR(dentry);
M
Miklos Szeredi 已提交
2802

M
Miklos Szeredi 已提交
2803 2804 2805 2806 2807
	/* Cached positive dentry: will open in f_op->open */
	if (!need_lookup && dentry->d_inode)
		goto out_no_open;

	if ((nd->flags & LOOKUP_OPEN) && dir_inode->i_op->atomic_open) {
2808
		return atomic_open(nd, dentry, path, file, op, got_write,
2809
				   need_lookup, opened);
M
Miklos Szeredi 已提交
2810 2811
	}

2812 2813 2814
	if (need_lookup) {
		BUG_ON(dentry->d_inode);

2815
		dentry = lookup_real(dir_inode, dentry, nd->flags);
2816
		if (IS_ERR(dentry))
2817
			return PTR_ERR(dentry);
2818 2819
	}

M
Miklos Szeredi 已提交
2820 2821 2822 2823 2824 2825 2826 2827 2828 2829
	/* Negative dentry, just create the file */
	if (!dentry->d_inode && (op->open_flag & O_CREAT)) {
		umode_t mode = op->mode;
		if (!IS_POSIXACL(dir->d_inode))
			mode &= ~current_umask();
		/*
		 * This write is needed to ensure that a
		 * rw->ro transition does not occur between
		 * the time when the file is created and when
		 * a permanent write count is taken through
2830
		 * the 'struct file' in finish_open().
M
Miklos Szeredi 已提交
2831
		 */
2832 2833
		if (!got_write) {
			error = -EROFS;
M
Miklos Szeredi 已提交
2834
			goto out_dput;
2835
		}
2836
		*opened |= FILE_CREATED;
M
Miklos Szeredi 已提交
2837 2838 2839
		error = security_path_mknod(&nd->path, dentry, mode, 0);
		if (error)
			goto out_dput;
A
Al Viro 已提交
2840 2841
		error = vfs_create(dir->d_inode, dentry, mode,
				   nd->flags & LOOKUP_EXCL);
M
Miklos Szeredi 已提交
2842 2843 2844
		if (error)
			goto out_dput;
	}
M
Miklos Szeredi 已提交
2845
out_no_open:
M
Miklos Szeredi 已提交
2846 2847
	path->dentry = dentry;
	path->mnt = nd->path.mnt;
2848
	return 1;
M
Miklos Szeredi 已提交
2849 2850 2851

out_dput:
	dput(dentry);
2852
	return error;
M
Miklos Szeredi 已提交
2853 2854
}

N
Nick Piggin 已提交
2855
/*
2856
 * Handle the last step of open()
N
Nick Piggin 已提交
2857
 */
2858 2859
static int do_last(struct nameidata *nd, struct path *path,
		   struct file *file, const struct open_flags *op,
2860
		   int *opened, struct filename *name)
2861
{
2862
	struct dentry *dir = nd->path.dentry;
2863
	int open_flag = op->open_flag;
M
Miklos Szeredi 已提交
2864
	bool will_truncate = (open_flag & O_TRUNC) != 0;
2865
	bool got_write = false;
A
Al Viro 已提交
2866
	int acc_mode = op->acc_mode;
2867
	struct inode *inode;
M
Miklos Szeredi 已提交
2868
	bool symlink_ok = false;
2869 2870
	struct path save_parent = { .dentry = NULL, .mnt = NULL };
	bool retried = false;
A
Al Viro 已提交
2871
	int error;
2872

2873 2874 2875
	nd->flags &= ~LOOKUP_PARENT;
	nd->flags |= op->intent;

2876
	if (nd->last_type != LAST_NORM) {
2877 2878
		error = handle_dots(nd, nd->last_type);
		if (error)
2879
			return error;
M
Miklos Szeredi 已提交
2880
		goto finish_open;
2881
	}
2882

2883
	if (!(open_flag & O_CREAT)) {
2884 2885
		if (nd->last.name[nd->last.len])
			nd->flags |= LOOKUP_FOLLOW | LOOKUP_DIRECTORY;
A
Al Viro 已提交
2886
		if (open_flag & O_PATH && !(nd->flags & LOOKUP_FOLLOW))
M
Miklos Szeredi 已提交
2887
			symlink_ok = true;
2888
		/* we _can_ be in RCU mode here */
A
Al Viro 已提交
2889
		error = lookup_fast(nd, path, &inode);
2890 2891 2892 2893
		if (likely(!error))
			goto finish_lookup;

		if (error < 0)
2894
			goto out;
2895 2896

		BUG_ON(nd->inode != dir->d_inode);
2897 2898 2899 2900 2901 2902 2903 2904 2905
	} else {
		/* create side of things */
		/*
		 * This will *only* deal with leaving RCU mode - LOOKUP_JUMPED
		 * has been cleared when we got to the last component we are
		 * about to look up
		 */
		error = complete_walk(nd);
		if (error)
2906
			return error;
2907

2908
		audit_inode(name, dir, LOOKUP_PARENT);
2909 2910 2911
		error = -EISDIR;
		/* trailing slashes? */
		if (nd->last.name[nd->last.len])
2912
			goto out;
2913
	}
A
Al Viro 已提交
2914

2915
retry_lookup:
2916 2917 2918 2919 2920 2921 2922 2923 2924 2925
	if (op->open_flag & (O_CREAT | O_TRUNC | O_WRONLY | O_RDWR)) {
		error = mnt_want_write(nd->path.mnt);
		if (!error)
			got_write = true;
		/*
		 * do _not_ fail yet - we might not need that or fail with
		 * a different error; let lookup_open() decide; we'll be
		 * dropping this one anyway.
		 */
	}
2926
	mutex_lock(&dir->d_inode->i_mutex);
2927
	error = lookup_open(nd, path, file, op, got_write, opened);
M
Miklos Szeredi 已提交
2928
	mutex_unlock(&dir->d_inode->i_mutex);
2929

2930 2931
	if (error <= 0) {
		if (error)
M
Miklos Szeredi 已提交
2932 2933
			goto out;

2934
		if ((*opened & FILE_CREATED) ||
A
Al Viro 已提交
2935
		    !S_ISREG(file_inode(file)->i_mode))
M
Miklos Szeredi 已提交
2936
			will_truncate = false;
M
Miklos Szeredi 已提交
2937

2938
		audit_inode(name, file->f_path.dentry, 0);
M
Miklos Szeredi 已提交
2939 2940
		goto opened;
	}
2941

2942
	if (*opened & FILE_CREATED) {
2943
		/* Don't check for write permission, don't truncate */
2944
		open_flag &= ~O_TRUNC;
M
Miklos Szeredi 已提交
2945
		will_truncate = false;
A
Al Viro 已提交
2946
		acc_mode = MAY_OPEN;
M
Miklos Szeredi 已提交
2947
		path_to_nameidata(path, nd);
M
Miklos Szeredi 已提交
2948
		goto finish_open_created;
2949 2950 2951
	}

	/*
2952
	 * create/update audit record if it already exists.
2953
	 */
2954
	if (d_is_positive(path->dentry))
2955
		audit_inode(name, path->dentry, 0);
2956

M
Miklos Szeredi 已提交
2957 2958 2959 2960 2961
	/*
	 * If atomic_open() acquired write access it is dropped now due to
	 * possible mount and symlink following (this might be optimized away if
	 * necessary...)
	 */
2962
	if (got_write) {
M
Miklos Szeredi 已提交
2963
		mnt_drop_write(nd->path.mnt);
2964
		got_write = false;
M
Miklos Szeredi 已提交
2965 2966
	}

2967
	error = -EEXIST;
A
Al Viro 已提交
2968
	if ((open_flag & (O_EXCL | O_CREAT)) == (O_EXCL | O_CREAT))
2969 2970
		goto exit_dput;

2971 2972 2973
	error = follow_managed(path, nd->flags);
	if (error < 0)
		goto exit_dput;
2974

2975 2976 2977
	if (error)
		nd->flags |= LOOKUP_JUMPED;

2978 2979
	BUG_ON(nd->flags & LOOKUP_RCU);
	inode = path->dentry->d_inode;
2980 2981
finish_lookup:
	/* we _can_ be in RCU mode here */
2982
	error = -ENOENT;
2983
	if (d_is_negative(path->dentry)) {
2984
		path_to_nameidata(path, nd);
2985
		goto out;
2986
	}
A
Al Viro 已提交
2987

2988
	if (should_follow_link(path->dentry, !symlink_ok)) {
2989 2990 2991
		if (nd->flags & LOOKUP_RCU) {
			if (unlikely(unlazy_walk(nd, path->dentry))) {
				error = -ECHILD;
2992
				goto out;
2993 2994 2995
			}
		}
		BUG_ON(inode != path->dentry->d_inode);
2996
		return 1;
2997
	}
2998

2999 3000 3001 3002 3003 3004 3005 3006
	if ((nd->flags & LOOKUP_RCU) || nd->path.mnt != path->mnt) {
		path_to_nameidata(path, nd);
	} else {
		save_parent.dentry = nd->path.dentry;
		save_parent.mnt = mntget(path->mnt);
		nd->path.dentry = path->dentry;

	}
3007
	nd->inode = inode;
3008
	/* Why this, you ask?  _Now_ we might have grown LOOKUP_JUMPED... */
3009
finish_open:
3010
	error = complete_walk(nd);
3011 3012
	if (error) {
		path_put(&save_parent);
3013
		return error;
3014
	}
3015
	audit_inode(name, nd->path.dentry, 0);
3016
	error = -EISDIR;
3017 3018
	if ((open_flag & O_CREAT) &&
	    (d_is_directory(nd->path.dentry) || d_is_autodir(nd->path.dentry)))
3019
		goto out;
3020
	error = -ENOTDIR;
3021
	if ((nd->flags & LOOKUP_DIRECTORY) && !d_is_directory(nd->path.dentry))
3022
		goto out;
3023
	if (!S_ISREG(nd->inode->i_mode))
M
Miklos Szeredi 已提交
3024
		will_truncate = false;
3025

3026 3027 3028
	if (will_truncate) {
		error = mnt_want_write(nd->path.mnt);
		if (error)
3029
			goto out;
3030
		got_write = true;
3031
	}
M
Miklos Szeredi 已提交
3032
finish_open_created:
A
Al Viro 已提交
3033
	error = may_open(&nd->path, acc_mode, open_flag);
3034
	if (error)
3035
		goto out;
A
Al Viro 已提交
3036 3037 3038 3039
	file->f_path.mnt = nd->path.mnt;
	error = finish_open(file, nd->path.dentry, NULL, opened);
	if (error) {
		if (error == -EOPENSTALE)
M
Miklos Szeredi 已提交
3040
			goto stale_open;
3041
		goto out;
M
Miklos Szeredi 已提交
3042
	}
3043
opened:
3044
	error = open_check_o_direct(file);
3045 3046
	if (error)
		goto exit_fput;
3047
	error = ima_file_check(file, op->acc_mode);
3048 3049 3050 3051
	if (error)
		goto exit_fput;

	if (will_truncate) {
3052
		error = handle_truncate(file);
3053 3054
		if (error)
			goto exit_fput;
3055
	}
3056
out:
3057
	if (got_write)
3058
		mnt_drop_write(nd->path.mnt);
3059
	path_put(&save_parent);
3060
	terminate_walk(nd);
3061
	return error;
3062 3063 3064

exit_dput:
	path_put_conditional(path, nd);
3065
	goto out;
3066
exit_fput:
3067 3068
	fput(file);
	goto out;
3069

M
Miklos Szeredi 已提交
3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080
stale_open:
	/* If no saved parent or already retried then can't retry */
	if (!save_parent.dentry || retried)
		goto out;

	BUG_ON(save_parent.dentry != dir);
	path_put(&nd->path);
	nd->path = save_parent;
	nd->inode = dir->d_inode;
	save_parent.mnt = NULL;
	save_parent.dentry = NULL;
3081
	if (got_write) {
M
Miklos Szeredi 已提交
3082
		mnt_drop_write(nd->path.mnt);
3083
		got_write = false;
M
Miklos Szeredi 已提交
3084 3085 3086
	}
	retried = true;
	goto retry_lookup;
3087 3088
}

3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134
static int do_tmpfile(int dfd, struct filename *pathname,
		struct nameidata *nd, int flags,
		const struct open_flags *op,
		struct file *file, int *opened)
{
	static const struct qstr name = QSTR_INIT("/", 1);
	struct dentry *dentry, *child;
	struct inode *dir;
	int error = path_lookupat(dfd, pathname->name,
				  flags | LOOKUP_DIRECTORY, nd);
	if (unlikely(error))
		return error;
	error = mnt_want_write(nd->path.mnt);
	if (unlikely(error))
		goto out;
	/* we want directory to be writable */
	error = inode_permission(nd->inode, MAY_WRITE | MAY_EXEC);
	if (error)
		goto out2;
	dentry = nd->path.dentry;
	dir = dentry->d_inode;
	if (!dir->i_op->tmpfile) {
		error = -EOPNOTSUPP;
		goto out2;
	}
	child = d_alloc(dentry, &name);
	if (unlikely(!child)) {
		error = -ENOMEM;
		goto out2;
	}
	nd->flags &= ~LOOKUP_DIRECTORY;
	nd->flags |= op->intent;
	dput(nd->path.dentry);
	nd->path.dentry = child;
	error = dir->i_op->tmpfile(dir, nd->path.dentry, op->mode);
	if (error)
		goto out2;
	audit_inode(pathname, nd->path.dentry, 0);
	error = may_open(&nd->path, op->acc_mode, op->open_flag);
	if (error)
		goto out2;
	file->f_path.mnt = nd->path.mnt;
	error = finish_open(file, nd->path.dentry, NULL, opened);
	if (error)
		goto out2;
	error = open_check_o_direct(file);
3135
	if (error) {
3136
		fput(file);
3137 3138 3139 3140 3141 3142
	} else if (!(op->open_flag & O_EXCL)) {
		struct inode *inode = file_inode(file);
		spin_lock(&inode->i_lock);
		inode->i_state |= I_LINKABLE;
		spin_unlock(&inode->i_lock);
	}
3143 3144 3145 3146 3147 3148 3149
out2:
	mnt_drop_write(nd->path.mnt);
out:
	path_put(&nd->path);
	return error;
}

3150
static struct file *path_openat(int dfd, struct filename *pathname,
A
Al Viro 已提交
3151
		struct nameidata *nd, const struct open_flags *op, int flags)
L
Linus Torvalds 已提交
3152
{
3153
	struct file *base = NULL;
A
Al Viro 已提交
3154
	struct file *file;
3155
	struct path path;
3156
	int opened = 0;
3157
	int error;
N
Nick Piggin 已提交
3158

A
Al Viro 已提交
3159
	file = get_empty_filp();
3160 3161
	if (IS_ERR(file))
		return file;
N
Nick Piggin 已提交
3162

A
Al Viro 已提交
3163
	file->f_flags = op->open_flag;
N
Nick Piggin 已提交
3164

A
Al Viro 已提交
3165
	if (unlikely(file->f_flags & __O_TMPFILE)) {
3166 3167 3168 3169
		error = do_tmpfile(dfd, pathname, nd, flags, op, file, &opened);
		goto out;
	}

3170
	error = path_init(dfd, pathname->name, flags | LOOKUP_PARENT, nd, &base);
N
Nick Piggin 已提交
3171
	if (unlikely(error))
3172
		goto out;
N
Nick Piggin 已提交
3173

3174
	current->total_link_count = 0;
3175
	error = link_path_walk(pathname->name, nd);
N
Nick Piggin 已提交
3176
	if (unlikely(error))
3177
		goto out;
L
Linus Torvalds 已提交
3178

3179 3180
	error = do_last(nd, &path, file, op, &opened, pathname);
	while (unlikely(error > 0)) { /* trailing symlink */
3181
		struct path link = path;
A
Al Viro 已提交
3182
		void *cookie;
3183
		if (!(nd->flags & LOOKUP_FOLLOW)) {
A
Al Viro 已提交
3184 3185
			path_put_conditional(&path, nd);
			path_put(&nd->path);
3186
			error = -ELOOP;
3187 3188
			break;
		}
K
Kees Cook 已提交
3189 3190 3191
		error = may_follow_link(&link, nd);
		if (unlikely(error))
			break;
A
Al Viro 已提交
3192 3193
		nd->flags |= LOOKUP_PARENT;
		nd->flags &= ~(LOOKUP_OPEN|LOOKUP_CREATE|LOOKUP_EXCL);
3194
		error = follow_link(&link, nd, &cookie);
3195
		if (unlikely(error))
3196 3197
			break;
		error = do_last(nd, &path, file, op, &opened, pathname);
3198
		put_link(nd, &link, cookie);
3199
	}
A
Al Viro 已提交
3200
out:
A
Al Viro 已提交
3201 3202
	if (nd->root.mnt && !(nd->flags & LOOKUP_ROOT))
		path_put(&nd->root);
3203 3204
	if (base)
		fput(base);
3205 3206
	if (!(opened & FILE_OPENED)) {
		BUG_ON(!error);
A
Al Viro 已提交
3207
		put_filp(file);
3208
	}
3209 3210 3211 3212 3213 3214 3215 3216 3217 3218
	if (unlikely(error)) {
		if (error == -EOPENSTALE) {
			if (flags & LOOKUP_RCU)
				error = -ECHILD;
			else
				error = -ESTALE;
		}
		file = ERR_PTR(error);
	}
	return file;
L
Linus Torvalds 已提交
3219 3220
}

3221
struct file *do_filp_open(int dfd, struct filename *pathname,
3222
		const struct open_flags *op)
3223
{
A
Al Viro 已提交
3224
	struct nameidata nd;
3225
	int flags = op->lookup_flags;
3226 3227
	struct file *filp;

A
Al Viro 已提交
3228
	filp = path_openat(dfd, pathname, &nd, op, flags | LOOKUP_RCU);
3229
	if (unlikely(filp == ERR_PTR(-ECHILD)))
A
Al Viro 已提交
3230
		filp = path_openat(dfd, pathname, &nd, op, flags);
3231
	if (unlikely(filp == ERR_PTR(-ESTALE)))
A
Al Viro 已提交
3232
		filp = path_openat(dfd, pathname, &nd, op, flags | LOOKUP_REVAL);
3233 3234 3235
	return filp;
}

A
Al Viro 已提交
3236
struct file *do_file_open_root(struct dentry *dentry, struct vfsmount *mnt,
3237
		const char *name, const struct open_flags *op)
A
Al Viro 已提交
3238 3239 3240
{
	struct nameidata nd;
	struct file *file;
3241
	struct filename filename = { .name = name };
3242
	int flags = op->lookup_flags | LOOKUP_ROOT;
A
Al Viro 已提交
3243 3244 3245 3246

	nd.root.mnt = mnt;
	nd.root.dentry = dentry;

3247
	if (d_is_symlink(dentry) && op->intent & LOOKUP_OPEN)
A
Al Viro 已提交
3248 3249
		return ERR_PTR(-ELOOP);

3250
	file = path_openat(-1, &filename, &nd, op, flags | LOOKUP_RCU);
A
Al Viro 已提交
3251
	if (unlikely(file == ERR_PTR(-ECHILD)))
3252
		file = path_openat(-1, &filename, &nd, op, flags);
A
Al Viro 已提交
3253
	if (unlikely(file == ERR_PTR(-ESTALE)))
3254
		file = path_openat(-1, &filename, &nd, op, flags | LOOKUP_REVAL);
A
Al Viro 已提交
3255 3256 3257
	return file;
}

3258 3259
struct dentry *kern_path_create(int dfd, const char *pathname,
				struct path *path, unsigned int lookup_flags)
L
Linus Torvalds 已提交
3260
{
3261
	struct dentry *dentry = ERR_PTR(-EEXIST);
A
Al Viro 已提交
3262
	struct nameidata nd;
3263
	int err2;
3264 3265 3266 3267 3268 3269 3270 3271 3272 3273
	int error;
	bool is_dir = (lookup_flags & LOOKUP_DIRECTORY);

	/*
	 * Note that only LOOKUP_REVAL and LOOKUP_DIRECTORY matter here. Any
	 * other flags passed in are ignored!
	 */
	lookup_flags &= LOOKUP_REVAL;

	error = do_path_lookup(dfd, pathname, LOOKUP_PARENT|lookup_flags, &nd);
A
Al Viro 已提交
3274 3275
	if (error)
		return ERR_PTR(error);
L
Linus Torvalds 已提交
3276

3277 3278 3279 3280
	/*
	 * Yucky last component or no last component at all?
	 * (foo/., foo/.., /////)
	 */
A
Al Viro 已提交
3281 3282 3283 3284
	if (nd.last_type != LAST_NORM)
		goto out;
	nd.flags &= ~LOOKUP_PARENT;
	nd.flags |= LOOKUP_CREATE | LOOKUP_EXCL;
3285

3286 3287
	/* don't fail immediately if it's r/o, at least try to report other errors */
	err2 = mnt_want_write(nd.path.mnt);
3288 3289 3290
	/*
	 * Do the final lookup.
	 */
A
Al Viro 已提交
3291 3292
	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
	dentry = lookup_hash(&nd);
L
Linus Torvalds 已提交
3293
	if (IS_ERR(dentry))
3294
		goto unlock;
3295

3296
	error = -EEXIST;
3297
	if (d_is_positive(dentry))
3298
		goto fail;
3299

3300 3301 3302 3303 3304 3305
	/*
	 * Special case - lookup gave negative, but... we had foo/bar/
	 * From the vfs_mknod() POV we just have a negative dentry -
	 * all is fine. Let's be bastards - you had / on the end, you've
	 * been asking for (non-existent) directory. -ENOENT for you.
	 */
A
Al Viro 已提交
3306
	if (unlikely(!is_dir && nd.last.name[nd.last.len])) {
3307
		error = -ENOENT;
A
Al Viro 已提交
3308
		goto fail;
3309
	}
3310 3311
	if (unlikely(err2)) {
		error = err2;
3312
		goto fail;
3313
	}
A
Al Viro 已提交
3314
	*path = nd.path;
L
Linus Torvalds 已提交
3315 3316
	return dentry;
fail:
3317 3318 3319
	dput(dentry);
	dentry = ERR_PTR(error);
unlock:
A
Al Viro 已提交
3320
	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
3321 3322
	if (!err2)
		mnt_drop_write(nd.path.mnt);
A
Al Viro 已提交
3323 3324
out:
	path_put(&nd.path);
L
Linus Torvalds 已提交
3325 3326
	return dentry;
}
3327 3328
EXPORT_SYMBOL(kern_path_create);

A
Al Viro 已提交
3329 3330 3331 3332
void done_path_create(struct path *path, struct dentry *dentry)
{
	dput(dentry);
	mutex_unlock(&path->dentry->d_inode->i_mutex);
3333
	mnt_drop_write(path->mnt);
A
Al Viro 已提交
3334 3335 3336 3337
	path_put(path);
}
EXPORT_SYMBOL(done_path_create);

3338 3339
struct dentry *user_path_create(int dfd, const char __user *pathname,
				struct path *path, unsigned int lookup_flags)
3340
{
3341
	struct filename *tmp = getname(pathname);
3342 3343 3344
	struct dentry *res;
	if (IS_ERR(tmp))
		return ERR_CAST(tmp);
3345
	res = kern_path_create(dfd, tmp->name, path, lookup_flags);
3346 3347 3348 3349 3350
	putname(tmp);
	return res;
}
EXPORT_SYMBOL(user_path_create);

A
Al Viro 已提交
3351
int vfs_mknod(struct inode *dir, struct dentry *dentry, umode_t mode, dev_t dev)
L
Linus Torvalds 已提交
3352
{
3353
	int error = may_create(dir, dentry);
L
Linus Torvalds 已提交
3354 3355 3356 3357

	if (error)
		return error;

3358
	if ((S_ISCHR(mode) || S_ISBLK(mode)) && !capable(CAP_MKNOD))
L
Linus Torvalds 已提交
3359 3360
		return -EPERM;

A
Al Viro 已提交
3361
	if (!dir->i_op->mknod)
L
Linus Torvalds 已提交
3362 3363
		return -EPERM;

3364 3365 3366 3367
	error = devcgroup_inode_mknod(mode, dev);
	if (error)
		return error;

L
Linus Torvalds 已提交
3368 3369 3370 3371 3372
	error = security_inode_mknod(dir, dentry, mode, dev);
	if (error)
		return error;

	error = dir->i_op->mknod(dir, dentry, mode, dev);
3373
	if (!error)
3374
		fsnotify_create(dir, dentry);
L
Linus Torvalds 已提交
3375 3376 3377
	return error;
}

A
Al Viro 已提交
3378
static int may_mknod(umode_t mode)
3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394
{
	switch (mode & S_IFMT) {
	case S_IFREG:
	case S_IFCHR:
	case S_IFBLK:
	case S_IFIFO:
	case S_IFSOCK:
	case 0: /* zero mode translates to S_IFREG */
		return 0;
	case S_IFDIR:
		return -EPERM;
	default:
		return -EINVAL;
	}
}

A
Al Viro 已提交
3395
SYSCALL_DEFINE4(mknodat, int, dfd, const char __user *, filename, umode_t, mode,
3396
		unsigned, dev)
L
Linus Torvalds 已提交
3397
{
3398
	struct dentry *dentry;
3399 3400
	struct path path;
	int error;
3401
	unsigned int lookup_flags = 0;
L
Linus Torvalds 已提交
3402

3403 3404 3405
	error = may_mknod(mode);
	if (error)
		return error;
3406 3407
retry:
	dentry = user_path_create(dfd, filename, &path, lookup_flags);
3408 3409
	if (IS_ERR(dentry))
		return PTR_ERR(dentry);
3410

3411
	if (!IS_POSIXACL(path.dentry->d_inode))
A
Al Viro 已提交
3412
		mode &= ~current_umask();
3413
	error = security_path_mknod(&path, dentry, mode, dev);
3414
	if (error)
3415
		goto out;
3416
	switch (mode & S_IFMT) {
L
Linus Torvalds 已提交
3417
		case 0: case S_IFREG:
A
Al Viro 已提交
3418
			error = vfs_create(path.dentry->d_inode,dentry,mode,true);
L
Linus Torvalds 已提交
3419 3420
			break;
		case S_IFCHR: case S_IFBLK:
3421
			error = vfs_mknod(path.dentry->d_inode,dentry,mode,
L
Linus Torvalds 已提交
3422 3423 3424
					new_decode_dev(dev));
			break;
		case S_IFIFO: case S_IFSOCK:
3425
			error = vfs_mknod(path.dentry->d_inode,dentry,mode,0);
L
Linus Torvalds 已提交
3426 3427
			break;
	}
3428
out:
A
Al Viro 已提交
3429
	done_path_create(&path, dentry);
3430 3431 3432 3433
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
L
Linus Torvalds 已提交
3434 3435 3436
	return error;
}

A
Al Viro 已提交
3437
SYSCALL_DEFINE3(mknod, const char __user *, filename, umode_t, mode, unsigned, dev)
3438 3439 3440 3441
{
	return sys_mknodat(AT_FDCWD, filename, mode, dev);
}

3442
int vfs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
L
Linus Torvalds 已提交
3443
{
3444
	int error = may_create(dir, dentry);
3445
	unsigned max_links = dir->i_sb->s_max_links;
L
Linus Torvalds 已提交
3446 3447 3448 3449

	if (error)
		return error;

A
Al Viro 已提交
3450
	if (!dir->i_op->mkdir)
L
Linus Torvalds 已提交
3451 3452 3453 3454 3455 3456 3457
		return -EPERM;

	mode &= (S_IRWXUGO|S_ISVTX);
	error = security_inode_mkdir(dir, dentry, mode);
	if (error)
		return error;

3458 3459 3460
	if (max_links && dir->i_nlink >= max_links)
		return -EMLINK;

L
Linus Torvalds 已提交
3461
	error = dir->i_op->mkdir(dir, dentry, mode);
3462
	if (!error)
3463
		fsnotify_mkdir(dir, dentry);
L
Linus Torvalds 已提交
3464 3465 3466
	return error;
}

3467
SYSCALL_DEFINE3(mkdirat, int, dfd, const char __user *, pathname, umode_t, mode)
L
Linus Torvalds 已提交
3468
{
3469
	struct dentry *dentry;
3470 3471
	struct path path;
	int error;
3472
	unsigned int lookup_flags = LOOKUP_DIRECTORY;
L
Linus Torvalds 已提交
3473

3474 3475
retry:
	dentry = user_path_create(dfd, pathname, &path, lookup_flags);
3476
	if (IS_ERR(dentry))
3477
		return PTR_ERR(dentry);
L
Linus Torvalds 已提交
3478

3479
	if (!IS_POSIXACL(path.dentry->d_inode))
A
Al Viro 已提交
3480
		mode &= ~current_umask();
3481
	error = security_path_mkdir(&path, dentry, mode);
3482 3483
	if (!error)
		error = vfs_mkdir(path.dentry->d_inode, dentry, mode);
A
Al Viro 已提交
3484
	done_path_create(&path, dentry);
3485 3486 3487 3488
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
L
Linus Torvalds 已提交
3489 3490 3491
	return error;
}

3492
SYSCALL_DEFINE2(mkdir, const char __user *, pathname, umode_t, mode)
3493 3494 3495 3496
{
	return sys_mkdirat(AT_FDCWD, pathname, mode);
}

L
Linus Torvalds 已提交
3497
/*
S
Sage Weil 已提交
3498
 * The dentry_unhash() helper will try to drop the dentry early: we
3499
 * should have a usage count of 1 if we're the only user of this
S
Sage Weil 已提交
3500 3501
 * dentry, and if that is true (possibly after pruning the dcache),
 * then we drop the dentry now.
L
Linus Torvalds 已提交
3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513
 *
 * A low-level filesystem can, if it choses, legally
 * do a
 *
 *	if (!d_unhashed(dentry))
 *		return -EBUSY;
 *
 * if it cannot handle the case of removing a directory
 * that is still in use by something else..
 */
void dentry_unhash(struct dentry *dentry)
{
3514
	shrink_dcache_parent(dentry);
L
Linus Torvalds 已提交
3515
	spin_lock(&dentry->d_lock);
3516
	if (dentry->d_lockref.count == 1)
L
Linus Torvalds 已提交
3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527
		__d_drop(dentry);
	spin_unlock(&dentry->d_lock);
}

int vfs_rmdir(struct inode *dir, struct dentry *dentry)
{
	int error = may_delete(dir, dentry, 1);

	if (error)
		return error;

A
Al Viro 已提交
3528
	if (!dir->i_op->rmdir)
L
Linus Torvalds 已提交
3529 3530
		return -EPERM;

3531
	dget(dentry);
3532
	mutex_lock(&dentry->d_inode->i_mutex);
S
Sage Weil 已提交
3533 3534

	error = -EBUSY;
L
Linus Torvalds 已提交
3535
	if (d_mountpoint(dentry))
S
Sage Weil 已提交
3536 3537 3538 3539 3540 3541
		goto out;

	error = security_inode_rmdir(dir, dentry);
	if (error)
		goto out;

3542
	shrink_dcache_parent(dentry);
S
Sage Weil 已提交
3543 3544 3545 3546 3547 3548 3549 3550
	error = dir->i_op->rmdir(dir, dentry);
	if (error)
		goto out;

	dentry->d_inode->i_flags |= S_DEAD;
	dont_mount(dentry);

out:
3551
	mutex_unlock(&dentry->d_inode->i_mutex);
3552
	dput(dentry);
S
Sage Weil 已提交
3553
	if (!error)
L
Linus Torvalds 已提交
3554 3555 3556 3557
		d_delete(dentry);
	return error;
}

3558
static long do_rmdir(int dfd, const char __user *pathname)
L
Linus Torvalds 已提交
3559 3560
{
	int error = 0;
3561
	struct filename *name;
L
Linus Torvalds 已提交
3562 3563
	struct dentry *dentry;
	struct nameidata nd;
3564 3565 3566
	unsigned int lookup_flags = 0;
retry:
	name = user_path_parent(dfd, pathname, &nd, lookup_flags);
3567 3568
	if (IS_ERR(name))
		return PTR_ERR(name);
L
Linus Torvalds 已提交
3569 3570

	switch(nd.last_type) {
3571 3572 3573 3574 3575 3576 3577 3578 3579
	case LAST_DOTDOT:
		error = -ENOTEMPTY;
		goto exit1;
	case LAST_DOT:
		error = -EINVAL;
		goto exit1;
	case LAST_ROOT:
		error = -EBUSY;
		goto exit1;
L
Linus Torvalds 已提交
3580
	}
3581 3582

	nd.flags &= ~LOOKUP_PARENT;
3583 3584 3585
	error = mnt_want_write(nd.path.mnt);
	if (error)
		goto exit1;
3586

3587
	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
3588
	dentry = lookup_hash(&nd);
L
Linus Torvalds 已提交
3589
	error = PTR_ERR(dentry);
3590 3591
	if (IS_ERR(dentry))
		goto exit2;
3592 3593 3594 3595
	if (!dentry->d_inode) {
		error = -ENOENT;
		goto exit3;
	}
3596 3597
	error = security_path_rmdir(&nd.path, dentry);
	if (error)
3598
		goto exit3;
3599
	error = vfs_rmdir(nd.path.dentry->d_inode, dentry);
3600
exit3:
3601 3602
	dput(dentry);
exit2:
3603
	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
3604
	mnt_drop_write(nd.path.mnt);
L
Linus Torvalds 已提交
3605
exit1:
J
Jan Blunck 已提交
3606
	path_put(&nd.path);
L
Linus Torvalds 已提交
3607
	putname(name);
3608 3609 3610 3611
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
L
Linus Torvalds 已提交
3612 3613 3614
	return error;
}

3615
SYSCALL_DEFINE1(rmdir, const char __user *, pathname)
3616 3617 3618 3619
{
	return do_rmdir(AT_FDCWD, pathname);
}

3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638
/**
 * vfs_unlink - unlink a filesystem object
 * @dir:	parent directory
 * @dentry:	victim
 * @delegated_inode: returns victim inode, if the inode is delegated.
 *
 * The caller must hold dir->i_mutex.
 *
 * If vfs_unlink discovers a delegation, it will return -EWOULDBLOCK and
 * return a reference to the inode in delegated_inode.  The caller
 * should then break the delegation on that inode and retry.  Because
 * breaking a delegation may take a long time, the caller should drop
 * dir->i_mutex before doing so.
 *
 * Alternatively, a caller may pass NULL for delegated_inode.  This may
 * be appropriate for callers that expect the underlying filesystem not
 * to be NFS exported.
 */
int vfs_unlink(struct inode *dir, struct dentry *dentry, struct inode **delegated_inode)
L
Linus Torvalds 已提交
3639
{
J
J. Bruce Fields 已提交
3640
	struct inode *target = dentry->d_inode;
L
Linus Torvalds 已提交
3641 3642 3643 3644 3645
	int error = may_delete(dir, dentry, 0);

	if (error)
		return error;

A
Al Viro 已提交
3646
	if (!dir->i_op->unlink)
L
Linus Torvalds 已提交
3647 3648
		return -EPERM;

J
J. Bruce Fields 已提交
3649
	mutex_lock(&target->i_mutex);
L
Linus Torvalds 已提交
3650 3651 3652 3653
	if (d_mountpoint(dentry))
		error = -EBUSY;
	else {
		error = security_inode_unlink(dir, dentry);
3654
		if (!error) {
3655 3656
			error = try_break_deleg(target, delegated_inode);
			if (error)
3657
				goto out;
L
Linus Torvalds 已提交
3658
			error = dir->i_op->unlink(dir, dentry);
3659
			if (!error)
3660
				dont_mount(dentry);
3661
		}
L
Linus Torvalds 已提交
3662
	}
3663
out:
J
J. Bruce Fields 已提交
3664
	mutex_unlock(&target->i_mutex);
L
Linus Torvalds 已提交
3665 3666 3667

	/* We don't d_delete() NFS sillyrenamed files--they still exist. */
	if (!error && !(dentry->d_flags & DCACHE_NFSFS_RENAMED)) {
J
J. Bruce Fields 已提交
3668
		fsnotify_link_count(target);
J
John McCutchan 已提交
3669
		d_delete(dentry);
L
Linus Torvalds 已提交
3670
	}
R
Robert Love 已提交
3671

L
Linus Torvalds 已提交
3672 3673 3674 3675 3676
	return error;
}

/*
 * Make sure that the actual truncation of the file will occur outside its
3677
 * directory's i_mutex.  Truncate can take a long time if there is a lot of
L
Linus Torvalds 已提交
3678 3679 3680
 * writeout happening, and we don't want to prevent access to the directory
 * while waiting on the I/O.
 */
3681
static long do_unlinkat(int dfd, const char __user *pathname)
L
Linus Torvalds 已提交
3682
{
3683
	int error;
3684
	struct filename *name;
L
Linus Torvalds 已提交
3685 3686 3687
	struct dentry *dentry;
	struct nameidata nd;
	struct inode *inode = NULL;
3688
	struct inode *delegated_inode = NULL;
3689 3690 3691
	unsigned int lookup_flags = 0;
retry:
	name = user_path_parent(dfd, pathname, &nd, lookup_flags);
3692 3693
	if (IS_ERR(name))
		return PTR_ERR(name);
3694

L
Linus Torvalds 已提交
3695 3696 3697
	error = -EISDIR;
	if (nd.last_type != LAST_NORM)
		goto exit1;
3698 3699

	nd.flags &= ~LOOKUP_PARENT;
3700 3701 3702
	error = mnt_want_write(nd.path.mnt);
	if (error)
		goto exit1;
3703
retry_deleg:
3704
	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
3705
	dentry = lookup_hash(&nd);
L
Linus Torvalds 已提交
3706 3707 3708
	error = PTR_ERR(dentry);
	if (!IS_ERR(dentry)) {
		/* Why not before? Because we want correct error value */
3709 3710
		if (nd.last.name[nd.last.len])
			goto slashes;
L
Linus Torvalds 已提交
3711
		inode = dentry->d_inode;
3712
		if (d_is_negative(dentry))
3713 3714
			goto slashes;
		ihold(inode);
3715 3716
		error = security_path_unlink(&nd.path, dentry);
		if (error)
3717
			goto exit2;
3718
		error = vfs_unlink(nd.path.dentry->d_inode, dentry, &delegated_inode);
3719
exit2:
L
Linus Torvalds 已提交
3720 3721
		dput(dentry);
	}
3722
	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
3723 3724
	if (inode)
		iput(inode);	/* truncate the inode here */
3725 3726
	inode = NULL;
	if (delegated_inode) {
3727
		error = break_deleg_wait(&delegated_inode);
3728 3729 3730
		if (!error)
			goto retry_deleg;
	}
3731
	mnt_drop_write(nd.path.mnt);
L
Linus Torvalds 已提交
3732
exit1:
J
Jan Blunck 已提交
3733
	path_put(&nd.path);
L
Linus Torvalds 已提交
3734
	putname(name);
3735 3736 3737 3738 3739
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		inode = NULL;
		goto retry;
	}
L
Linus Torvalds 已提交
3740 3741 3742
	return error;

slashes:
3743 3744 3745 3746 3747 3748
	if (d_is_negative(dentry))
		error = -ENOENT;
	else if (d_is_directory(dentry) || d_is_autodir(dentry))
		error = -EISDIR;
	else
		error = -ENOTDIR;
L
Linus Torvalds 已提交
3749 3750 3751
	goto exit2;
}

3752
SYSCALL_DEFINE3(unlinkat, int, dfd, const char __user *, pathname, int, flag)
3753 3754 3755 3756 3757 3758 3759 3760 3761 3762
{
	if ((flag & ~AT_REMOVEDIR) != 0)
		return -EINVAL;

	if (flag & AT_REMOVEDIR)
		return do_rmdir(dfd, pathname);

	return do_unlinkat(dfd, pathname);
}

3763
SYSCALL_DEFINE1(unlink, const char __user *, pathname)
3764 3765 3766 3767
{
	return do_unlinkat(AT_FDCWD, pathname);
}

3768
int vfs_symlink(struct inode *dir, struct dentry *dentry, const char *oldname)
L
Linus Torvalds 已提交
3769
{
3770
	int error = may_create(dir, dentry);
L
Linus Torvalds 已提交
3771 3772 3773 3774

	if (error)
		return error;

A
Al Viro 已提交
3775
	if (!dir->i_op->symlink)
L
Linus Torvalds 已提交
3776 3777 3778 3779 3780 3781 3782
		return -EPERM;

	error = security_inode_symlink(dir, dentry, oldname);
	if (error)
		return error;

	error = dir->i_op->symlink(dir, dentry, oldname);
3783
	if (!error)
3784
		fsnotify_create(dir, dentry);
L
Linus Torvalds 已提交
3785 3786 3787
	return error;
}

3788 3789
SYSCALL_DEFINE3(symlinkat, const char __user *, oldname,
		int, newdfd, const char __user *, newname)
L
Linus Torvalds 已提交
3790
{
3791
	int error;
3792
	struct filename *from;
3793
	struct dentry *dentry;
3794
	struct path path;
3795
	unsigned int lookup_flags = 0;
L
Linus Torvalds 已提交
3796 3797

	from = getname(oldname);
3798
	if (IS_ERR(from))
L
Linus Torvalds 已提交
3799
		return PTR_ERR(from);
3800 3801
retry:
	dentry = user_path_create(newdfd, newname, &path, lookup_flags);
3802 3803
	error = PTR_ERR(dentry);
	if (IS_ERR(dentry))
3804
		goto out_putname;
3805

3806
	error = security_path_symlink(&path, dentry, from->name);
3807
	if (!error)
3808
		error = vfs_symlink(path.dentry->d_inode, dentry, from->name);
A
Al Viro 已提交
3809
	done_path_create(&path, dentry);
3810 3811 3812 3813
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
3814
out_putname:
L
Linus Torvalds 已提交
3815 3816 3817 3818
	putname(from);
	return error;
}

3819
SYSCALL_DEFINE2(symlink, const char __user *, oldname, const char __user *, newname)
3820 3821 3822 3823
{
	return sys_symlinkat(oldname, AT_FDCWD, newname);
}

J
J. Bruce Fields 已提交
3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843
/**
 * vfs_link - create a new link
 * @old_dentry:	object to be linked
 * @dir:	new parent
 * @new_dentry:	where to create the new link
 * @delegated_inode: returns inode needing a delegation break
 *
 * The caller must hold dir->i_mutex
 *
 * If vfs_link discovers a delegation on the to-be-linked file in need
 * of breaking, it will return -EWOULDBLOCK and return a reference to the
 * inode in delegated_inode.  The caller should then break the delegation
 * and retry.  Because breaking a delegation may take a long time, the
 * caller should drop the i_mutex before doing so.
 *
 * Alternatively, a caller may pass NULL for delegated_inode.  This may
 * be appropriate for callers that expect the underlying filesystem not
 * to be NFS exported.
 */
int vfs_link(struct dentry *old_dentry, struct inode *dir, struct dentry *new_dentry, struct inode **delegated_inode)
L
Linus Torvalds 已提交
3844 3845
{
	struct inode *inode = old_dentry->d_inode;
3846
	unsigned max_links = dir->i_sb->s_max_links;
L
Linus Torvalds 已提交
3847 3848 3849 3850 3851
	int error;

	if (!inode)
		return -ENOENT;

3852
	error = may_create(dir, new_dentry);
L
Linus Torvalds 已提交
3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863
	if (error)
		return error;

	if (dir->i_sb != inode->i_sb)
		return -EXDEV;

	/*
	 * A link to an append-only or immutable file cannot be created.
	 */
	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
		return -EPERM;
A
Al Viro 已提交
3864
	if (!dir->i_op->link)
L
Linus Torvalds 已提交
3865
		return -EPERM;
3866
	if (S_ISDIR(inode->i_mode))
L
Linus Torvalds 已提交
3867 3868 3869 3870 3871 3872
		return -EPERM;

	error = security_inode_link(old_dentry, dir, new_dentry);
	if (error)
		return error;

3873
	mutex_lock(&inode->i_mutex);
3874
	/* Make sure we don't allow creating hardlink to an unlinked file */
3875
	if (inode->i_nlink == 0 && !(inode->i_state & I_LINKABLE))
3876
		error =  -ENOENT;
3877 3878
	else if (max_links && inode->i_nlink >= max_links)
		error = -EMLINK;
J
J. Bruce Fields 已提交
3879 3880 3881 3882 3883
	else {
		error = try_break_deleg(inode, delegated_inode);
		if (!error)
			error = dir->i_op->link(old_dentry, dir, new_dentry);
	}
3884 3885 3886 3887 3888 3889

	if (!error && (inode->i_state & I_LINKABLE)) {
		spin_lock(&inode->i_lock);
		inode->i_state &= ~I_LINKABLE;
		spin_unlock(&inode->i_lock);
	}
3890
	mutex_unlock(&inode->i_mutex);
3891
	if (!error)
3892
		fsnotify_link(dir, inode, new_dentry);
L
Linus Torvalds 已提交
3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904
	return error;
}

/*
 * Hardlinks are often used in delicate situations.  We avoid
 * security-related surprises by not following symlinks on the
 * newname.  --KAB
 *
 * We don't follow them on the oldname either to be compatible
 * with linux 2.0, and to avoid hard-linking to directories
 * and other special files.  --ADM
 */
3905 3906
SYSCALL_DEFINE5(linkat, int, olddfd, const char __user *, oldname,
		int, newdfd, const char __user *, newname, int, flags)
L
Linus Torvalds 已提交
3907 3908
{
	struct dentry *new_dentry;
3909
	struct path old_path, new_path;
J
J. Bruce Fields 已提交
3910
	struct inode *delegated_inode = NULL;
3911
	int how = 0;
L
Linus Torvalds 已提交
3912 3913
	int error;

3914
	if ((flags & ~(AT_SYMLINK_FOLLOW | AT_EMPTY_PATH)) != 0)
3915
		return -EINVAL;
3916
	/*
3917 3918 3919
	 * To use null names we require CAP_DAC_READ_SEARCH
	 * This ensures that not everyone will be able to create
	 * handlink using the passed filedescriptor.
3920
	 */
3921 3922 3923
	if (flags & AT_EMPTY_PATH) {
		if (!capable(CAP_DAC_READ_SEARCH))
			return -ENOENT;
3924
		how = LOOKUP_EMPTY;
3925
	}
3926 3927 3928

	if (flags & AT_SYMLINK_FOLLOW)
		how |= LOOKUP_FOLLOW;
3929
retry:
3930
	error = user_path_at(olddfd, oldname, how, &old_path);
L
Linus Torvalds 已提交
3931
	if (error)
3932 3933
		return error;

3934 3935
	new_dentry = user_path_create(newdfd, newname, &new_path,
					(how & LOOKUP_REVAL));
L
Linus Torvalds 已提交
3936
	error = PTR_ERR(new_dentry);
3937
	if (IS_ERR(new_dentry))
3938 3939 3940 3941 3942
		goto out;

	error = -EXDEV;
	if (old_path.mnt != new_path.mnt)
		goto out_dput;
K
Kees Cook 已提交
3943 3944 3945
	error = may_linkat(&old_path);
	if (unlikely(error))
		goto out_dput;
3946
	error = security_path_link(old_path.dentry, &new_path, new_dentry);
3947
	if (error)
3948
		goto out_dput;
J
J. Bruce Fields 已提交
3949
	error = vfs_link(old_path.dentry, new_path.dentry->d_inode, new_dentry, &delegated_inode);
3950
out_dput:
A
Al Viro 已提交
3951
	done_path_create(&new_path, new_dentry);
J
J. Bruce Fields 已提交
3952 3953 3954 3955 3956
	if (delegated_inode) {
		error = break_deleg_wait(&delegated_inode);
		if (!error)
			goto retry;
	}
3957 3958 3959 3960
	if (retry_estale(error, how)) {
		how |= LOOKUP_REVAL;
		goto retry;
	}
L
Linus Torvalds 已提交
3961
out:
3962
	path_put(&old_path);
L
Linus Torvalds 已提交
3963 3964 3965 3966

	return error;
}

3967
SYSCALL_DEFINE2(link, const char __user *, oldname, const char __user *, newname)
3968
{
3969
	return sys_linkat(AT_FDCWD, oldname, AT_FDCWD, newname, 0);
3970 3971
}

L
Linus Torvalds 已提交
3972 3973 3974 3975 3976 3977 3978
/*
 * The worst of all namespace operations - renaming directory. "Perverted"
 * doesn't even start to describe it. Somebody in UCB had a heck of a trip...
 * Problems:
 *	a) we can get into loop creation. Check is done in is_subdir().
 *	b) race potential - two innocent renames can create a loop together.
 *	   That's where 4.4 screws up. Current fix: serialization on
3979
 *	   sb->s_vfs_rename_mutex. We might be more accurate, but that's another
L
Linus Torvalds 已提交
3980
 *	   story.
3981 3982
 *	c) we have to lock _four_ objects - parents and victim (if it exists),
 *	   and source (if it is not a directory).
3983
 *	   And that - after we got ->i_mutex on parents (until then we don't know
L
Linus Torvalds 已提交
3984 3985
 *	   whether the target exists).  Solution: try to be smart with locking
 *	   order for inodes.  We rely on the fact that tree topology may change
3986
 *	   only under ->s_vfs_rename_mutex _and_ that parent of the object we
L
Linus Torvalds 已提交
3987 3988 3989
 *	   move will be locked.  Thus we can rank directories by the tree
 *	   (ancestors first) and rank all non-directories after them.
 *	   That works since everybody except rename does "lock parent, lookup,
3990
 *	   lock child" and rename is under ->s_vfs_rename_mutex.
L
Linus Torvalds 已提交
3991 3992 3993
 *	   HOWEVER, it relies on the assumption that any object with ->lookup()
 *	   has no more than 1 dentry.  If "hybrid" objects will ever appear,
 *	   we'd better make sure that there's no link(2) for them.
3994
 *	d) conversion from fhandle to dentry may come in the wrong moment - when
3995
 *	   we are removing the target. Solution: we will have to grab ->i_mutex
L
Linus Torvalds 已提交
3996
 *	   in the fhandle_to_dentry code. [FIXME - current nfsfh.c relies on
3997
 *	   ->i_mutex on parents, which works but leads to some truly excessive
L
Linus Torvalds 已提交
3998 3999
 *	   locking].
 */
A
Adrian Bunk 已提交
4000 4001
static int vfs_rename_dir(struct inode *old_dir, struct dentry *old_dentry,
			  struct inode *new_dir, struct dentry *new_dentry)
L
Linus Torvalds 已提交
4002 4003
{
	int error = 0;
S
Sage Weil 已提交
4004
	struct inode *target = new_dentry->d_inode;
4005
	unsigned max_links = new_dir->i_sb->s_max_links;
L
Linus Torvalds 已提交
4006 4007 4008 4009 4010 4011

	/*
	 * If we are going to change the parent - check write permissions,
	 * we'll need to flip '..'.
	 */
	if (new_dir != old_dir) {
4012
		error = inode_permission(old_dentry->d_inode, MAY_WRITE);
L
Linus Torvalds 已提交
4013 4014 4015 4016 4017 4018 4019 4020
		if (error)
			return error;
	}

	error = security_inode_rename(old_dir, old_dentry, new_dir, new_dentry);
	if (error)
		return error;

4021
	dget(new_dentry);
4022
	if (target)
4023
		mutex_lock(&target->i_mutex);
S
Sage Weil 已提交
4024 4025 4026 4027 4028

	error = -EBUSY;
	if (d_mountpoint(old_dentry) || d_mountpoint(new_dentry))
		goto out;

4029 4030 4031 4032 4033
	error = -EMLINK;
	if (max_links && !target && new_dir != old_dir &&
	    new_dir->i_nlink >= max_links)
		goto out;

4034 4035
	if (target)
		shrink_dcache_parent(new_dentry);
S
Sage Weil 已提交
4036 4037 4038 4039
	error = old_dir->i_op->rename(old_dir, old_dentry, new_dir, new_dentry);
	if (error)
		goto out;

L
Linus Torvalds 已提交
4040
	if (target) {
S
Sage Weil 已提交
4041 4042
		target->i_flags |= S_DEAD;
		dont_mount(new_dentry);
L
Linus Torvalds 已提交
4043
	}
S
Sage Weil 已提交
4044 4045 4046
out:
	if (target)
		mutex_unlock(&target->i_mutex);
4047
	dput(new_dentry);
4048
	if (!error)
4049 4050
		if (!(old_dir->i_sb->s_type->fs_flags & FS_RENAME_DOES_D_MOVE))
			d_move(old_dentry,new_dentry);
L
Linus Torvalds 已提交
4051 4052 4053
	return error;
}

A
Adrian Bunk 已提交
4054
static int vfs_rename_other(struct inode *old_dir, struct dentry *old_dentry,
4055 4056
			    struct inode *new_dir, struct dentry *new_dentry,
			    struct inode **delegated_inode)
L
Linus Torvalds 已提交
4057
{
S
Sage Weil 已提交
4058
	struct inode *target = new_dentry->d_inode;
4059
	struct inode *source = old_dentry->d_inode;
L
Linus Torvalds 已提交
4060 4061 4062 4063 4064 4065 4066
	int error;

	error = security_inode_rename(old_dir, old_dentry, new_dir, new_dentry);
	if (error)
		return error;

	dget(new_dentry);
4067
	lock_two_nondirectories(source, target);
S
Sage Weil 已提交
4068 4069

	error = -EBUSY;
L
Linus Torvalds 已提交
4070
	if (d_mountpoint(old_dentry)||d_mountpoint(new_dentry))
S
Sage Weil 已提交
4071 4072
		goto out;

4073 4074 4075 4076 4077 4078 4079 4080
	error = try_break_deleg(source, delegated_inode);
	if (error)
		goto out;
	if (target) {
		error = try_break_deleg(target, delegated_inode);
		if (error)
			goto out;
	}
S
Sage Weil 已提交
4081 4082 4083 4084 4085 4086 4087 4088 4089
	error = old_dir->i_op->rename(old_dir, old_dentry, new_dir, new_dentry);
	if (error)
		goto out;

	if (target)
		dont_mount(new_dentry);
	if (!(old_dir->i_sb->s_type->fs_flags & FS_RENAME_DOES_D_MOVE))
		d_move(old_dentry, new_dentry);
out:
4090
	unlock_two_nondirectories(source, target);
L
Linus Torvalds 已提交
4091 4092 4093 4094
	dput(new_dentry);
	return error;
}

4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115
/**
 * vfs_rename - rename a filesystem object
 * @old_dir:	parent of source
 * @old_dentry:	source
 * @new_dir:	parent of destination
 * @new_dentry:	destination
 * @delegated_inode: returns an inode needing a delegation break
 *
 * The caller must hold multiple mutexes--see lock_rename()).
 *
 * If vfs_rename discovers a delegation in need of breaking at either
 * the source or destination, it will return -EWOULDBLOCK and return a
 * reference to the inode in delegated_inode.  The caller should then
 * break the delegation and retry.  Because breaking a delegation may
 * take a long time, the caller should drop all locks before doing
 * so.
 *
 * Alternatively, a caller may pass NULL for delegated_inode.  This may
 * be appropriate for callers that expect the underlying filesystem not
 * to be NFS exported.
 */
L
Linus Torvalds 已提交
4116
int vfs_rename(struct inode *old_dir, struct dentry *old_dentry,
4117 4118
	       struct inode *new_dir, struct dentry *new_dentry,
	       struct inode **delegated_inode)
L
Linus Torvalds 已提交
4119 4120
{
	int error;
4121
	int is_dir = d_is_directory(old_dentry) || d_is_autodir(old_dentry);
4122
	const unsigned char *old_name;
L
Linus Torvalds 已提交
4123 4124 4125 4126 4127 4128 4129 4130 4131

	if (old_dentry->d_inode == new_dentry->d_inode)
 		return 0;
 
	error = may_delete(old_dir, old_dentry, is_dir);
	if (error)
		return error;

	if (!new_dentry->d_inode)
4132
		error = may_create(new_dir, new_dentry);
L
Linus Torvalds 已提交
4133 4134 4135 4136 4137
	else
		error = may_delete(new_dir, new_dentry, is_dir);
	if (error)
		return error;

A
Al Viro 已提交
4138
	if (!old_dir->i_op->rename)
L
Linus Torvalds 已提交
4139 4140
		return -EPERM;

R
Robert Love 已提交
4141 4142
	old_name = fsnotify_oldname_init(old_dentry->d_name.name);

L
Linus Torvalds 已提交
4143 4144 4145
	if (is_dir)
		error = vfs_rename_dir(old_dir,old_dentry,new_dir,new_dentry);
	else
4146
		error = vfs_rename_other(old_dir,old_dentry,new_dir,new_dentry,delegated_inode);
4147 4148
	if (!error)
		fsnotify_move(old_dir, new_dir, old_name, is_dir,
4149
			      new_dentry->d_inode, old_dentry);
R
Robert Love 已提交
4150 4151
	fsnotify_oldname_free(old_name);

L
Linus Torvalds 已提交
4152 4153 4154
	return error;
}

4155 4156
SYSCALL_DEFINE4(renameat, int, olddfd, const char __user *, oldname,
		int, newdfd, const char __user *, newname)
L
Linus Torvalds 已提交
4157
{
4158 4159 4160
	struct dentry *old_dir, *new_dir;
	struct dentry *old_dentry, *new_dentry;
	struct dentry *trap;
L
Linus Torvalds 已提交
4161
	struct nameidata oldnd, newnd;
4162
	struct inode *delegated_inode = NULL;
4163 4164
	struct filename *from;
	struct filename *to;
4165 4166
	unsigned int lookup_flags = 0;
	bool should_retry = false;
4167
	int error;
4168 4169
retry:
	from = user_path_parent(olddfd, oldname, &oldnd, lookup_flags);
4170 4171
	if (IS_ERR(from)) {
		error = PTR_ERR(from);
L
Linus Torvalds 已提交
4172
		goto exit;
4173
	}
L
Linus Torvalds 已提交
4174

4175
	to = user_path_parent(newdfd, newname, &newnd, lookup_flags);
4176 4177
	if (IS_ERR(to)) {
		error = PTR_ERR(to);
L
Linus Torvalds 已提交
4178
		goto exit1;
4179
	}
L
Linus Torvalds 已提交
4180 4181

	error = -EXDEV;
4182
	if (oldnd.path.mnt != newnd.path.mnt)
L
Linus Torvalds 已提交
4183 4184
		goto exit2;

4185
	old_dir = oldnd.path.dentry;
L
Linus Torvalds 已提交
4186 4187 4188 4189
	error = -EBUSY;
	if (oldnd.last_type != LAST_NORM)
		goto exit2;

4190
	new_dir = newnd.path.dentry;
L
Linus Torvalds 已提交
4191 4192 4193
	if (newnd.last_type != LAST_NORM)
		goto exit2;

4194 4195 4196 4197
	error = mnt_want_write(oldnd.path.mnt);
	if (error)
		goto exit2;

4198 4199
	oldnd.flags &= ~LOOKUP_PARENT;
	newnd.flags &= ~LOOKUP_PARENT;
4200
	newnd.flags |= LOOKUP_RENAME_TARGET;
4201

4202
retry_deleg:
L
Linus Torvalds 已提交
4203 4204
	trap = lock_rename(new_dir, old_dir);

4205
	old_dentry = lookup_hash(&oldnd);
L
Linus Torvalds 已提交
4206 4207 4208 4209 4210
	error = PTR_ERR(old_dentry);
	if (IS_ERR(old_dentry))
		goto exit3;
	/* source must exist */
	error = -ENOENT;
4211
	if (d_is_negative(old_dentry))
L
Linus Torvalds 已提交
4212 4213
		goto exit4;
	/* unless the source is a directory trailing slashes give -ENOTDIR */
4214
	if (!d_is_directory(old_dentry) && !d_is_autodir(old_dentry)) {
L
Linus Torvalds 已提交
4215 4216 4217 4218 4219 4220 4221 4222 4223 4224
		error = -ENOTDIR;
		if (oldnd.last.name[oldnd.last.len])
			goto exit4;
		if (newnd.last.name[newnd.last.len])
			goto exit4;
	}
	/* source should not be ancestor of target */
	error = -EINVAL;
	if (old_dentry == trap)
		goto exit4;
4225
	new_dentry = lookup_hash(&newnd);
L
Linus Torvalds 已提交
4226 4227 4228 4229 4230 4231 4232 4233
	error = PTR_ERR(new_dentry);
	if (IS_ERR(new_dentry))
		goto exit4;
	/* target should not be an ancestor of source */
	error = -ENOTEMPTY;
	if (new_dentry == trap)
		goto exit5;

4234 4235 4236
	error = security_path_rename(&oldnd.path, old_dentry,
				     &newnd.path, new_dentry);
	if (error)
4237
		goto exit5;
L
Linus Torvalds 已提交
4238
	error = vfs_rename(old_dir->d_inode, old_dentry,
4239 4240
				   new_dir->d_inode, new_dentry,
				   &delegated_inode);
L
Linus Torvalds 已提交
4241 4242 4243 4244 4245 4246
exit5:
	dput(new_dentry);
exit4:
	dput(old_dentry);
exit3:
	unlock_rename(new_dir, old_dir);
4247 4248 4249 4250 4251
	if (delegated_inode) {
		error = break_deleg_wait(&delegated_inode);
		if (!error)
			goto retry_deleg;
	}
4252
	mnt_drop_write(oldnd.path.mnt);
L
Linus Torvalds 已提交
4253
exit2:
4254 4255
	if (retry_estale(error, lookup_flags))
		should_retry = true;
J
Jan Blunck 已提交
4256
	path_put(&newnd.path);
4257
	putname(to);
L
Linus Torvalds 已提交
4258
exit1:
J
Jan Blunck 已提交
4259
	path_put(&oldnd.path);
L
Linus Torvalds 已提交
4260
	putname(from);
4261 4262 4263 4264 4265
	if (should_retry) {
		should_retry = false;
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
4266
exit:
L
Linus Torvalds 已提交
4267 4268 4269
	return error;
}

4270
SYSCALL_DEFINE2(rename, const char __user *, oldname, const char __user *, newname)
4271 4272 4273 4274
{
	return sys_renameat(AT_FDCWD, oldname, AT_FDCWD, newname);
}

L
Linus Torvalds 已提交
4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299
int vfs_readlink(struct dentry *dentry, char __user *buffer, int buflen, const char *link)
{
	int len;

	len = PTR_ERR(link);
	if (IS_ERR(link))
		goto out;

	len = strlen(link);
	if (len > (unsigned) buflen)
		len = buflen;
	if (copy_to_user(buffer, link, len))
		len = -EFAULT;
out:
	return len;
}

/*
 * A helper for ->readlink().  This should be used *ONLY* for symlinks that
 * have ->follow_link() touching nd only in nd_set_link().  Using (or not
 * using) it for any given inode is up to filesystem.
 */
int generic_readlink(struct dentry *dentry, char __user *buffer, int buflen)
{
	struct nameidata nd;
4300
	void *cookie;
4301
	int res;
4302

L
Linus Torvalds 已提交
4303
	nd.depth = 0;
4304
	cookie = dentry->d_inode->i_op->follow_link(dentry, &nd);
4305 4306 4307 4308 4309 4310 4311
	if (IS_ERR(cookie))
		return PTR_ERR(cookie);

	res = vfs_readlink(dentry, buffer, buflen, nd_get_link(&nd));
	if (dentry->d_inode->i_op->put_link)
		dentry->d_inode->i_op->put_link(dentry, &nd, cookie);
	return res;
L
Linus Torvalds 已提交
4312 4313 4314 4315 4316
}

/* get the link contents into pagecache */
static char *page_getlink(struct dentry * dentry, struct page **ppage)
{
4317 4318
	char *kaddr;
	struct page *page;
L
Linus Torvalds 已提交
4319
	struct address_space *mapping = dentry->d_inode->i_mapping;
4320
	page = read_mapping_page(mapping, 0, NULL);
L
Linus Torvalds 已提交
4321
	if (IS_ERR(page))
4322
		return (char*)page;
L
Linus Torvalds 已提交
4323
	*ppage = page;
4324 4325 4326
	kaddr = kmap(page);
	nd_terminate_link(kaddr, dentry->d_inode->i_size, PAGE_SIZE - 1);
	return kaddr;
L
Linus Torvalds 已提交
4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340
}

int page_readlink(struct dentry *dentry, char __user *buffer, int buflen)
{
	struct page *page = NULL;
	char *s = page_getlink(dentry, &page);
	int res = vfs_readlink(dentry,buffer,buflen,s);
	if (page) {
		kunmap(page);
		page_cache_release(page);
	}
	return res;
}

4341
void *page_follow_link_light(struct dentry *dentry, struct nameidata *nd)
L
Linus Torvalds 已提交
4342
{
4343
	struct page *page = NULL;
L
Linus Torvalds 已提交
4344
	nd_set_link(nd, page_getlink(dentry, &page));
4345
	return page;
L
Linus Torvalds 已提交
4346 4347
}

4348
void page_put_link(struct dentry *dentry, struct nameidata *nd, void *cookie)
L
Linus Torvalds 已提交
4349
{
4350 4351 4352
	struct page *page = cookie;

	if (page) {
L
Linus Torvalds 已提交
4353 4354 4355 4356 4357
		kunmap(page);
		page_cache_release(page);
	}
}

4358 4359 4360 4361
/*
 * The nofs argument instructs pagecache_write_begin to pass AOP_FLAG_NOFS
 */
int __page_symlink(struct inode *inode, const char *symname, int len, int nofs)
L
Linus Torvalds 已提交
4362 4363
{
	struct address_space *mapping = inode->i_mapping;
4364
	struct page *page;
4365
	void *fsdata;
4366
	int err;
L
Linus Torvalds 已提交
4367
	char *kaddr;
4368 4369 4370
	unsigned int flags = AOP_FLAG_UNINTERRUPTIBLE;
	if (nofs)
		flags |= AOP_FLAG_NOFS;
L
Linus Torvalds 已提交
4371

4372
retry:
4373
	err = pagecache_write_begin(NULL, mapping, 0, len-1,
4374
				flags, &page, &fsdata);
L
Linus Torvalds 已提交
4375
	if (err)
4376 4377
		goto fail;

4378
	kaddr = kmap_atomic(page);
L
Linus Torvalds 已提交
4379
	memcpy(kaddr, symname, len-1);
4380
	kunmap_atomic(kaddr);
4381 4382 4383

	err = pagecache_write_end(NULL, mapping, 0, len-1, len-1,
							page, fsdata);
L
Linus Torvalds 已提交
4384 4385
	if (err < 0)
		goto fail;
4386 4387 4388
	if (err < len-1)
		goto retry;

L
Linus Torvalds 已提交
4389 4390 4391 4392 4393 4394
	mark_inode_dirty(inode);
	return 0;
fail:
	return err;
}

4395 4396 4397
int page_symlink(struct inode *inode, const char *symname, int len)
{
	return __page_symlink(inode, symname, len,
4398
			!(mapping_gfp_mask(inode->i_mapping) & __GFP_FS));
4399 4400
}

4401
const struct inode_operations page_symlink_inode_operations = {
L
Linus Torvalds 已提交
4402 4403 4404 4405 4406
	.readlink	= generic_readlink,
	.follow_link	= page_follow_link_light,
	.put_link	= page_put_link,
};

4407
EXPORT_SYMBOL(user_path_at);
4408
EXPORT_SYMBOL(follow_down_one);
L
Linus Torvalds 已提交
4409 4410
EXPORT_SYMBOL(follow_down);
EXPORT_SYMBOL(follow_up);
A
Al Viro 已提交
4411
EXPORT_SYMBOL(get_write_access); /* nfsd */
L
Linus Torvalds 已提交
4412 4413 4414 4415 4416
EXPORT_SYMBOL(lock_rename);
EXPORT_SYMBOL(lookup_one_len);
EXPORT_SYMBOL(page_follow_link_light);
EXPORT_SYMBOL(page_put_link);
EXPORT_SYMBOL(page_readlink);
4417
EXPORT_SYMBOL(__page_symlink);
L
Linus Torvalds 已提交
4418 4419
EXPORT_SYMBOL(page_symlink);
EXPORT_SYMBOL(page_symlink_inode_operations);
A
Al Viro 已提交
4420
EXPORT_SYMBOL(kern_path);
4421
EXPORT_SYMBOL(vfs_path_lookup);
4422
EXPORT_SYMBOL(inode_permission);
L
Linus Torvalds 已提交
4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435
EXPORT_SYMBOL(unlock_rename);
EXPORT_SYMBOL(vfs_create);
EXPORT_SYMBOL(vfs_link);
EXPORT_SYMBOL(vfs_mkdir);
EXPORT_SYMBOL(vfs_mknod);
EXPORT_SYMBOL(generic_permission);
EXPORT_SYMBOL(vfs_readlink);
EXPORT_SYMBOL(vfs_rename);
EXPORT_SYMBOL(vfs_rmdir);
EXPORT_SYMBOL(vfs_symlink);
EXPORT_SYMBOL(vfs_unlink);
EXPORT_SYMBOL(dentry_unhash);
EXPORT_SYMBOL(generic_readlink);