namei.c 86.1 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
20 21 22 23
#include <linux/slab.h>
#include <linux/fs.h>
#include <linux/namei.h>
#include <linux/pagemap.h>
Robert Love's avatar
Robert Love committed
24
#include <linux/fsnotify.h>
Linus Torvalds's avatar
Linus Torvalds committed
25 26
#include <linux/personality.h>
#include <linux/security.h>
Mimi Zohar's avatar
Mimi Zohar committed
27
#include <linux/ima.h>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
37 38
#include <asm/uaccess.h>

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

Linus Torvalds's avatar
Linus Torvalds committed
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
Lucas De Marchi's avatar
Lucas De Marchi committed
76
 * the name is a symlink pointing to a non-existent name.
Linus Torvalds's avatar
Linus Torvalds committed
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
Linus Torvalds's avatar
Linus Torvalds committed
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
static char *getname_flags(const char __user *filename, int flags, int *empty)
Linus Torvalds's avatar
Linus Torvalds committed
121
{
122 123
	char *result = __getname(), *err;
	int len;
124

125
	if (unlikely(!result))
126 127
		return ERR_PTR(-ENOMEM);

128 129 130 131 132 133 134 135
	len = strncpy_from_user(result, filename, PATH_MAX);
	err = ERR_PTR(len);
	if (unlikely(len < 0))
		goto error;

	/* The empty path is special. */
	if (unlikely(!len)) {
		if (empty)
136
			*empty = 1;
137 138 139
		err = ERR_PTR(-ENOENT);
		if (!(flags & LOOKUP_EMPTY))
			goto error;
Linus Torvalds's avatar
Linus Torvalds committed
140
	}
141 142 143 144 145 146 147 148 149 150

	err = ERR_PTR(-ENAMETOOLONG);
	if (likely(len < PATH_MAX)) {
		audit_getname(result);
		return result;
	}

error:
	__putname(result);
	return err;
Linus Torvalds's avatar
Linus Torvalds committed
151 152
}

Al Viro's avatar
Al Viro committed
153 154
char *getname(const char __user * filename)
{
155
	return getname_flags(filename, 0, NULL);
Al Viro's avatar
Al Viro committed
156 157
}

Linus Torvalds's avatar
Linus Torvalds committed
158 159 160
#ifdef CONFIG_AUDITSYSCALL
void putname(const char *name)
{
161
	if (unlikely(!audit_dummy_context()))
Linus Torvalds's avatar
Linus Torvalds committed
162 163 164 165 166 167 168
		audit_putname(name);
	else
		__putname(name);
}
EXPORT_SYMBOL(putname);
#endif

169 170
static int check_acl(struct inode *inode, int mask)
{
171
#ifdef CONFIG_FS_POSIX_ACL
172 173 174
	struct posix_acl *acl;

	if (mask & MAY_NOT_BLOCK) {
175 176
		acl = get_cached_acl_rcu(inode, ACL_TYPE_ACCESS);
	        if (!acl)
177
	                return -EAGAIN;
178 179 180
		/* no ->get_acl() calls in RCU mode... */
		if (acl == ACL_NOT_CACHED)
			return -ECHILD;
181
	        return posix_acl_permission(inode, acl, mask & ~MAY_NOT_BLOCK);
182 183 184 185 186
	}

	acl = get_cached_acl(inode, ACL_TYPE_ACCESS);

	/*
187 188 189
	 * 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.
190
	 *
191 192
	 * If the filesystem doesn't have a get_acl() function at all, we'll
	 * just create the negative cache entry.
193 194
	 */
	if (acl == ACL_NOT_CACHED) {
195 196 197 198 199 200 201 202
	        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;
		}
203 204 205 206 207 208 209
	}

	if (acl) {
	        int error = posix_acl_permission(inode, acl, mask);
	        posix_acl_release(acl);
	        return error;
	}
210
#endif
211 212 213 214

	return -EAGAIN;
}

215
/*
216
 * This does the basic permission checking
Linus Torvalds's avatar
Linus Torvalds committed
217
 */
218
static int acl_permission_check(struct inode *inode, int mask)
Linus Torvalds's avatar
Linus Torvalds committed
219
{
220
	unsigned int mode = inode->i_mode;
Linus Torvalds's avatar
Linus Torvalds committed
221

222
	if (likely(uid_eq(current_fsuid(), inode->i_uid)))
Linus Torvalds's avatar
Linus Torvalds committed
223 224
		mode >>= 6;
	else {
225
		if (IS_POSIXACL(inode) && (mode & S_IRWXG)) {
226
			int error = check_acl(inode, mask);
227 228
			if (error != -EAGAIN)
				return error;
Linus Torvalds's avatar
Linus Torvalds committed
229 230 231 232 233 234 235 236 237
		}

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

	/*
	 * If the DACs are ok we don't need any capability check.
	 */
238
	if ((mask & ~mode & (MAY_READ | MAY_WRITE | MAY_EXEC)) == 0)
Linus Torvalds's avatar
Linus Torvalds committed
239
		return 0;
240 241 242 243
	return -EACCES;
}

/**
244
 * generic_permission -  check for access rights on a Posix-like filesystem
245
 * @inode:	inode to check access rights for
246
 * @mask:	right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC, ...)
247 248 249 250
 *
 * 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
251 252 253 254 255
 * 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.
256
 */
257
int generic_permission(struct inode *inode, int mask)
258 259 260 261
{
	int ret;

	/*
262
	 * Do the basic permission checks.
263
	 */
264
	ret = acl_permission_check(inode, mask);
265 266
	if (ret != -EACCES)
		return ret;
Linus Torvalds's avatar
Linus Torvalds committed
267

268 269
	if (S_ISDIR(inode->i_mode)) {
		/* DACs are overridable for directories */
270
		if (inode_capable(inode, CAP_DAC_OVERRIDE))
271 272
			return 0;
		if (!(mask & MAY_WRITE))
273
			if (inode_capable(inode, CAP_DAC_READ_SEARCH))
274 275 276
				return 0;
		return -EACCES;
	}
Linus Torvalds's avatar
Linus Torvalds committed
277 278
	/*
	 * Read/write DACs are always overridable.
279 280
	 * Executable DACs are overridable when there is
	 * at least one exec bit set.
Linus Torvalds's avatar
Linus Torvalds committed
281
	 */
282
	if (!(mask & MAY_EXEC) || (inode->i_mode & S_IXUGO))
283
		if (inode_capable(inode, CAP_DAC_OVERRIDE))
Linus Torvalds's avatar
Linus Torvalds committed
284 285 286 287 288
			return 0;

	/*
	 * Searching includes executable on directories, else just read.
	 */
289
	mask &= MAY_READ | MAY_WRITE | MAY_EXEC;
290
	if (mask == MAY_READ)
291
		if (inode_capable(inode, CAP_DAC_READ_SEARCH))
Linus Torvalds's avatar
Linus Torvalds committed
292 293 294 295 296
			return 0;

	return -EACCES;
}

297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316
/*
 * 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);
}

Christoph Hellwig's avatar
Christoph Hellwig committed
317 318 319
/**
 * inode_permission  -  check for access rights to a given inode
 * @inode:	inode to check permission on
320
 * @mask:	right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC, ...)
Christoph Hellwig's avatar
Christoph Hellwig committed
321 322 323 324 325
 *
 * Used to check for read/write/execute permissions on an inode.
 * We use "fsuid" for this, letting us set arbitrary permissions
 * for filesystem access without changing the "normal" uids which
 * are used for other things.
326 327
 *
 * When checking for MAY_APPEND, MAY_WRITE must also be set in @mask.
Christoph Hellwig's avatar
Christoph Hellwig committed
328
 */
329
int inode_permission(struct inode *inode, int mask)
Linus Torvalds's avatar
Linus Torvalds committed
330
{
331
	int retval;
Linus Torvalds's avatar
Linus Torvalds committed
332

333
	if (unlikely(mask & MAY_WRITE)) {
334
		umode_t mode = inode->i_mode;
Linus Torvalds's avatar
Linus Torvalds committed
335 336 337 338 339 340 341 342 343 344 345 346 347 348 349

		/*
		 * Nobody gets write access to a read-only fs.
		 */
		if (IS_RDONLY(inode) &&
		    (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)))
			return -EROFS;

		/*
		 * Nobody gets write access to an immutable file.
		 */
		if (IS_IMMUTABLE(inode))
			return -EACCES;
	}

350
	retval = do_inode_permission(inode, mask);
Linus Torvalds's avatar
Linus Torvalds committed
351 352 353
	if (retval)
		return retval;

354 355 356 357
	retval = devcgroup_inode_permission(inode, mask);
	if (retval)
		return retval;

358
	return security_inode_permission(inode, mask);
Linus Torvalds's avatar
Linus Torvalds committed
359 360
}

Jan Blunck's avatar
Jan Blunck committed
361 362 363 364 365 366 367 368 369 370 371 372 373
/**
 * 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.
 */
void path_get(struct path *path)
{
	mntget(path->mnt);
	dget(path->dentry);
}
EXPORT_SYMBOL(path_get);

Jan Blunck's avatar
Jan Blunck committed
374 375 376 377 378 379 380
/**
 * 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.
 */
void path_put(struct path *path)
Linus Torvalds's avatar
Linus Torvalds committed
381
{
Jan Blunck's avatar
Jan Blunck committed
382 383
	dput(path->dentry);
	mntput(path->mnt);
Linus Torvalds's avatar
Linus Torvalds committed
384
}
Jan Blunck's avatar
Jan Blunck committed
385
EXPORT_SYMBOL(path_put);
Linus Torvalds's avatar
Linus Torvalds committed
386

387
/*
Nick Piggin's avatar
Nick Piggin committed
388
 * Path walking has 2 modes, rcu-walk and ref-walk (see
389 390 391 392 393 394 395
 * 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.
Nick Piggin's avatar
Nick Piggin committed
396 397 398
 */

/**
399 400 401
 * unlazy_walk - try to switch to ref-walk mode.
 * @nd: nameidata pathwalk data
 * @dentry: child of nd->path.dentry or NULL
402
 * Returns: 0 on success, -ECHILD on failure
Nick Piggin's avatar
Nick Piggin committed
403
 *
404 405 406
 * 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.
Nick Piggin's avatar
Nick Piggin committed
407
 */
408
static int unlazy_walk(struct nameidata *nd, struct dentry *dentry)
Nick Piggin's avatar
Nick Piggin committed
409 410 411
{
	struct fs_struct *fs = current->fs;
	struct dentry *parent = nd->path.dentry;
412
	int want_root = 0;
Nick Piggin's avatar
Nick Piggin committed
413 414

	BUG_ON(!(nd->flags & LOOKUP_RCU));
415 416
	if (nd->root.mnt && !(nd->flags & LOOKUP_ROOT)) {
		want_root = 1;
Nick Piggin's avatar
Nick Piggin committed
417 418 419 420 421 422
		spin_lock(&fs->lock);
		if (nd->root.mnt != fs->root.mnt ||
				nd->root.dentry != fs->root.dentry)
			goto err_root;
	}
	spin_lock(&parent->d_lock);
423 424 425 426 427
	if (!dentry) {
		if (!__d_rcu_to_refcount(parent, nd->seq))
			goto err_parent;
		BUG_ON(nd->inode != parent->d_inode);
	} else {
428 429
		if (dentry->d_parent != parent)
			goto err_parent;
430 431 432 433 434 435 436 437 438 439 440 441 442 443
		spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED);
		if (!__d_rcu_to_refcount(dentry, nd->seq))
			goto err_child;
		/*
		 * If the sequence check on the child dentry passed, then
		 * the child has not been removed from its parent. This
		 * means the parent dentry must be valid and able to take
		 * a reference at this point.
		 */
		BUG_ON(!IS_ROOT(dentry) && dentry->d_parent != parent);
		BUG_ON(!parent->d_count);
		parent->d_count++;
		spin_unlock(&dentry->d_lock);
	}
Nick Piggin's avatar
Nick Piggin committed
444
	spin_unlock(&parent->d_lock);
445
	if (want_root) {
Nick Piggin's avatar
Nick Piggin committed
446 447 448 449 450 451
		path_get(&nd->root);
		spin_unlock(&fs->lock);
	}
	mntget(nd->path.mnt);

	rcu_read_unlock();
Andi Kleen's avatar
Andi Kleen committed
452
	br_read_unlock(&vfsmount_lock);
Nick Piggin's avatar
Nick Piggin committed
453 454
	nd->flags &= ~LOOKUP_RCU;
	return 0;
455 456

err_child:
Nick Piggin's avatar
Nick Piggin committed
457
	spin_unlock(&dentry->d_lock);
458
err_parent:
Nick Piggin's avatar
Nick Piggin committed
459 460
	spin_unlock(&parent->d_lock);
err_root:
461
	if (want_root)
Nick Piggin's avatar
Nick Piggin committed
462 463 464 465
		spin_unlock(&fs->lock);
	return -ECHILD;
}

466 467 468 469 470 471
/**
 * release_open_intent - free up open intent resources
 * @nd: pointer to nameidata
 */
void release_open_intent(struct nameidata *nd)
{
472 473 474 475 476 477 478 479
	struct file *file = nd->intent.open.file;

	if (file && !IS_ERR(file)) {
		if (file->f_path.dentry == NULL)
			put_filp(file);
		else
			fput(file);
	}
480 481
}

Al Viro's avatar
Al Viro committed
482
static inline int d_revalidate(struct dentry *dentry, struct nameidata *nd)
483
{
Al Viro's avatar
Al Viro committed
484
	return dentry->d_op->d_revalidate(dentry, nd);
485 486
}

487 488 489
/**
 * complete_walk - successful completion of path walk
 * @nd:  pointer nameidata
490
 *
491 492 493 494 495
 * 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.
496
 */
497
static int complete_walk(struct nameidata *nd)
498
{
Al Viro's avatar
Al Viro committed
499
	struct dentry *dentry = nd->path.dentry;
500 501
	int status;

502 503 504 505 506 507 508 509
	if (nd->flags & LOOKUP_RCU) {
		nd->flags &= ~LOOKUP_RCU;
		if (!(nd->flags & LOOKUP_ROOT))
			nd->root.mnt = NULL;
		spin_lock(&dentry->d_lock);
		if (unlikely(!__d_rcu_to_refcount(dentry, nd->seq))) {
			spin_unlock(&dentry->d_lock);
			rcu_read_unlock();
Andi Kleen's avatar
Andi Kleen committed
510
			br_read_unlock(&vfsmount_lock);
511 512 513 514 515 516
			return -ECHILD;
		}
		BUG_ON(nd->inode != dentry->d_inode);
		spin_unlock(&dentry->d_lock);
		mntget(nd->path.mnt);
		rcu_read_unlock();
Andi Kleen's avatar
Andi Kleen committed
517
		br_read_unlock(&vfsmount_lock);
518 519
	}

Al Viro's avatar
Al Viro committed
520 521 522 523
	if (likely(!(nd->flags & LOOKUP_JUMPED)))
		return 0;

	if (likely(!(dentry->d_flags & DCACHE_OP_REVALIDATE)))
524 525
		return 0;

Al Viro's avatar
Al Viro committed
526 527 528 529
	if (likely(!(dentry->d_sb->s_type->fs_flags & FS_REVAL_DOT)))
		return 0;

	/* Note: we do not d_invalidate() */
530
	status = d_revalidate(dentry, nd);
531 532 533
	if (status > 0)
		return 0;

Al Viro's avatar
Al Viro committed
534
	if (!status)
535
		status = -ESTALE;
Al Viro's avatar
Al Viro committed
536

537
	path_put(&nd->path);
538 539 540
	return status;
}

Al Viro's avatar
Al Viro committed
541 542
static __always_inline void set_root(struct nameidata *nd)
{
543 544
	if (!nd->root.mnt)
		get_fs_root(current->fs, &nd->root);
Al Viro's avatar
Al Viro committed
545 546
}

547 548
static int link_path_walk(const char *, struct nameidata *);

Nick Piggin's avatar
Nick Piggin committed
549 550 551 552
static __always_inline void set_root_rcu(struct nameidata *nd)
{
	if (!nd->root.mnt) {
		struct fs_struct *fs = current->fs;
Nick Piggin's avatar
Nick Piggin committed
553 554 555 556 557
		unsigned seq;

		do {
			seq = read_seqcount_begin(&fs->seq);
			nd->root = fs->root;
558
			nd->seq = __read_seqcount_begin(&nd->root.dentry->d_seq);
Nick Piggin's avatar
Nick Piggin committed
559
		} while (read_seqcount_retry(&fs->seq, seq));
Nick Piggin's avatar
Nick Piggin committed
560 561 562
	}
}

563
static __always_inline int __vfs_follow_link(struct nameidata *nd, const char *link)
Linus Torvalds's avatar
Linus Torvalds committed
564
{
Nick Piggin's avatar
Nick Piggin committed
565 566
	int ret;

Linus Torvalds's avatar
Linus Torvalds committed
567 568 569 570
	if (IS_ERR(link))
		goto fail;

	if (*link == '/') {
Al Viro's avatar
Al Viro committed
571
		set_root(nd);
Jan Blunck's avatar
Jan Blunck committed
572
		path_put(&nd->path);
Al Viro's avatar
Al Viro committed
573 574
		nd->path = nd->root;
		path_get(&nd->root);
Al Viro's avatar
Al Viro committed
575
		nd->flags |= LOOKUP_JUMPED;
Linus Torvalds's avatar
Linus Torvalds committed
576
	}
Nick Piggin's avatar
Nick Piggin committed
577
	nd->inode = nd->path.dentry->d_inode;
Christoph Hellwig's avatar
Christoph Hellwig committed
578

Nick Piggin's avatar
Nick Piggin committed
579 580
	ret = link_path_walk(link, nd);
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
581
fail:
Jan Blunck's avatar
Jan Blunck committed
582
	path_put(&nd->path);
Linus Torvalds's avatar
Linus Torvalds committed
583 584 585
	return PTR_ERR(link);
}

Jan Blunck's avatar
Jan Blunck committed
586
static void path_put_conditional(struct path *path, struct nameidata *nd)
587 588
{
	dput(path->dentry);
589
	if (path->mnt != nd->path.mnt)
590 591 592
		mntput(path->mnt);
}

593 594
static inline void path_to_nameidata(const struct path *path,
					struct nameidata *nd)
595
{
Nick Piggin's avatar
Nick Piggin committed
596 597 598 599
	if (!(nd->flags & LOOKUP_RCU)) {
		dput(nd->path.dentry);
		if (nd->path.mnt != path->mnt)
			mntput(nd->path.mnt);
600
	}
Nick Piggin's avatar
Nick Piggin committed
601
	nd->path.mnt = path->mnt;
602
	nd->path.dentry = path->dentry;
603 604
}

605 606 607 608 609 610 611 612
static inline void put_link(struct nameidata *nd, struct path *link, void *cookie)
{
	struct inode *inode = link->dentry->d_inode;
	if (!IS_ERR(cookie) && inode->i_op->put_link)
		inode->i_op->put_link(link->dentry, nd, cookie);
	path_put(link);
}

Al Viro's avatar
Al Viro committed
613
static __always_inline int
614
follow_link(struct path *link, struct nameidata *nd, void **p)
Linus Torvalds's avatar
Linus Torvalds committed
615 616
{
	int error;
617
	struct dentry *dentry = link->dentry;
Linus Torvalds's avatar
Linus Torvalds committed
618

619 620
	BUG_ON(nd->flags & LOOKUP_RCU);

Al Viro's avatar
Al Viro committed
621 622 623
	if (link->mnt == nd->path.mnt)
		mntget(link->mnt);

624 625 626 627 628 629 630 631
	if (unlikely(current->total_link_count >= 40)) {
		*p = ERR_PTR(-ELOOP); /* no ->put_link(), please */
		path_put(&nd->path);
		return -ELOOP;
	}
	cond_resched();
	current->total_link_count++;

Al Viro's avatar
Al Viro committed
632
	touch_atime(link);
Linus Torvalds's avatar
Linus Torvalds committed
633
	nd_set_link(nd, NULL);
Al Viro's avatar
Al Viro committed
634

635 636 637 638 639 640 641
	error = security_inode_follow_link(link->dentry, nd);
	if (error) {
		*p = ERR_PTR(error); /* no ->put_link(), please */
		path_put(&nd->path);
		return error;
	}

642
	nd->last_type = LAST_BIND;
Al Viro's avatar
Al Viro committed
643 644 645
	*p = dentry->d_inode->i_op->follow_link(dentry, nd);
	error = PTR_ERR(*p);
	if (!IS_ERR(*p)) {
Linus Torvalds's avatar
Linus Torvalds committed
646
		char *s = nd_get_link(nd);
647
		error = 0;
Linus Torvalds's avatar
Linus Torvalds committed
648 649
		if (s)
			error = __vfs_follow_link(nd, s);
Al Viro's avatar
Al Viro committed
650
		else if (nd->last_type == LAST_BIND) {
Al Viro's avatar
Al Viro committed
651
			nd->flags |= LOOKUP_JUMPED;
652 653
			nd->inode = nd->path.dentry->d_inode;
			if (nd->inode->i_op->follow_link) {
Al Viro's avatar
Al Viro committed
654 655 656 657 658
				/* stepped on a _really_ weird one */
				path_put(&nd->path);
				error = -ELOOP;
			}
		}
Linus Torvalds's avatar
Linus Torvalds committed
659 660 661 662
	}
	return error;
}

Nick Piggin's avatar
Nick Piggin committed
663 664
static int follow_up_rcu(struct path *path)
{
665 666
	struct mount *mnt = real_mount(path->mnt);
	struct mount *parent;
Nick Piggin's avatar
Nick Piggin committed
667 668
	struct dentry *mountpoint;

669 670
	parent = mnt->mnt_parent;
	if (&parent->mnt == path->mnt)
Nick Piggin's avatar
Nick Piggin committed
671
		return 0;
672
	mountpoint = mnt->mnt_mountpoint;
Nick Piggin's avatar
Nick Piggin committed
673
	path->dentry = mountpoint;
674
	path->mnt = &parent->mnt;
Nick Piggin's avatar
Nick Piggin committed
675 676 677
	return 1;
}

Al Viro's avatar
Al Viro committed
678
int follow_up(struct path *path)
Linus Torvalds's avatar
Linus Torvalds committed
679
{
680 681
	struct mount *mnt = real_mount(path->mnt);
	struct mount *parent;
Linus Torvalds's avatar
Linus Torvalds committed
682
	struct dentry *mountpoint;
Nick Piggin's avatar
Nick Piggin committed
683

Andi Kleen's avatar
Andi Kleen committed
684
	br_read_lock(&vfsmount_lock);
685 686
	parent = mnt->mnt_parent;
	if (&parent->mnt == path->mnt) {
Andi Kleen's avatar
Andi Kleen committed
687
		br_read_unlock(&vfsmount_lock);
Linus Torvalds's avatar
Linus Torvalds committed
688 689
		return 0;
	}
690
	mntget(&parent->mnt);
691
	mountpoint = dget(mnt->mnt_mountpoint);
Andi Kleen's avatar
Andi Kleen committed
692
	br_read_unlock(&vfsmount_lock);
Al Viro's avatar
Al Viro committed
693 694 695
	dput(path->dentry);
	path->dentry = mountpoint;
	mntput(path->mnt);
696
	path->mnt = &parent->mnt;
Linus Torvalds's avatar
Linus Torvalds committed
697 698 699
	return 1;
}

Nick Piggin's avatar
Nick Piggin committed
700
/*
701 702 703
 * Perform an automount
 * - return -EISDIR to tell follow_managed() to stop and return the path we
 *   were called with.
Linus Torvalds's avatar
Linus Torvalds committed
704
 */
705 706
static int follow_automount(struct path *path, unsigned flags,
			    bool *need_mntput)
Nick Piggin's avatar
Nick Piggin committed
707
{
708
	struct vfsmount *mnt;
709
	int err;
710 711 712 713

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

714 715 716 717 718 719 720 721 722 723
	/* 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.
724
	 */
725
	if (!(flags & (LOOKUP_PARENT | LOOKUP_DIRECTORY |
726
		     LOOKUP_OPEN | LOOKUP_CREATE | LOOKUP_AUTOMOUNT)) &&
727 728 729
	    path->dentry->d_inode)
		return -EISDIR;

730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
	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.
		 */
Al Viro's avatar
Al Viro committed
745
		if (PTR_ERR(mnt) == -EISDIR && (flags & LOOKUP_PARENT))
746 747
			return -EREMOTE;
		return PTR_ERR(mnt);
Nick Piggin's avatar
Nick Piggin committed
748
	}
749

750 751
	if (!mnt) /* mount collision */
		return 0;
Nick Piggin's avatar
Nick Piggin committed
752

753 754 755 756 757
	if (!*need_mntput) {
		/* lock_mount() may release path->mnt on error */
		mntget(path->mnt);
		*need_mntput = true;
	}
758
	err = finish_automount(mnt, path);
759

760 761 762
	switch (err) {
	case -EBUSY:
		/* Someone else made a mount here whilst we were busy */
763
		return 0;
764
	case 0:
765
		path_put(path);
766 767 768
		path->mnt = mnt;
		path->dentry = dget(mnt->mnt_root);
		return 0;
769 770
	default:
		return err;
771
	}
772

Al Viro's avatar
Al Viro committed
773 774
}

775 776
/*
 * Handle a dentry that is managed in some way.
777
 * - Flagged for transit management (autofs)
778 779 780 781 782 783 784 785
 * - 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)
Linus Torvalds's avatar
Linus Torvalds committed
786
{
787
	struct vfsmount *mnt = path->mnt; /* held by caller, must be left alone */
788 789
	unsigned managed;
	bool need_mntput = false;
790
	int ret = 0;
791 792 793 794 795 796 797

	/* 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)) {
798 799 800 801 802
		/* 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);
803
			ret = path->dentry->d_op->d_manage(path->dentry, false);
804
			if (ret < 0)
805
				break;
806 807
		}

808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830
		/* 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
			 * namespace got unmounted before we managed to get the
			 * vfsmount_lock */
		}

		/* Handle an automount point */
		if (managed & DCACHE_NEED_AUTOMOUNT) {
			ret = follow_automount(path, flags, &need_mntput);
			if (ret < 0)
831
				break;
832 833 834 835 836
			continue;
		}

		/* We didn't change the current path point */
		break;
Linus Torvalds's avatar
Linus Torvalds committed
837
	}
838 839 840 841 842

	if (need_mntput && path->mnt == mnt)
		mntput(path->mnt);
	if (ret == -EISDIR)
		ret = 0;
843
	return ret < 0 ? ret : need_mntput;
Linus Torvalds's avatar
Linus Torvalds committed
844 845
}

846
int follow_down_one(struct path *path)
Linus Torvalds's avatar
Linus Torvalds committed
847 848 849
{
	struct vfsmount *mounted;

Al Viro's avatar
Al Viro committed
850
	mounted = lookup_mnt(path);
Linus Torvalds's avatar
Linus Torvalds committed
851
	if (mounted) {
Al Viro's avatar
Al Viro committed
852 853 854 855
		dput(path->dentry);
		mntput(path->mnt);
		path->mnt = mounted;
		path->dentry = dget(mounted->mnt_root);
Linus Torvalds's avatar
Linus Torvalds committed
856 857 858 859 860
		return 1;
	}
	return 0;
}

861 862 863 864 865 866
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);
}

867
/*
868 869
 * Try to skip to top of mountpoint pile in rcuwalk mode.  Fail if
 * we meet a managed dentry that would need blocking.
870 871
 */
static bool __follow_mount_rcu(struct nameidata *nd, struct path *path,
872
			       struct inode **inode)
873
{
874
	for (;;) {
875
		struct mount *mounted;
876 877 878 879
		/*
		 * Don't forget we might have a non-mountpoint managed dentry
		 * that wants to block transit.
		 */
880
		if (unlikely(managed_dentry_might_block(path->dentry)))
881
			return false;
882 883 884 885

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

886 887 888
		mounted = __lookup_mnt(path->mnt, path->dentry, 1);
		if (!mounted)
			break;
889 890
		path->mnt = &mounted->mnt;
		path->dentry = mounted->mnt.mnt_root;
891
		nd->flags |= LOOKUP_JUMPED;
892
		nd->seq = read_seqcount_begin(&path->dentry->d_seq);
893 894 895 896 897 898
		/*
		 * 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;
899 900 901 902
	}
	return true;
}

903
static void follow_mount_rcu(struct nameidata *nd)
904
{
905
	while (d_mountpoint(nd->path.dentry)) {
906
		struct mount *mounted;
907
		mounted = __lookup_mnt(nd->path.mnt, nd->path.dentry, 1);
908 909
		if (!mounted)
			break;
910 911
		nd->path.mnt = &mounted->mnt;
		nd->path.dentry = mounted->mnt.mnt_root;
912
		nd->seq = read_seqcount_begin(&nd->path.dentry->d_seq);
913 914 915
	}
}

Nick Piggin's avatar
Nick Piggin committed
916 917 918 919
static int follow_dotdot_rcu(struct nameidata *nd)
{
	set_root_rcu(nd);

920
	while (1) {
Nick Piggin's avatar
Nick Piggin committed
921 922 923 924 925 926 927 928 929 930 931
		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))
932
				goto failed;
Nick Piggin's avatar
Nick Piggin committed
933 934 935 936 937 938 939 940
			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);
	}
941 942
	follow_mount_rcu(nd);
	nd->inode = nd->path.dentry->d_inode;
Nick Piggin's avatar
Nick Piggin committed
943
	return 0;
944 945 946

failed:
	nd->flags &= ~LOOKUP_RCU;
947 948
	if (!(nd->flags & LOOKUP_ROOT))
		nd->root.mnt = NULL;
949
	rcu_read_unlock();
Andi Kleen's avatar
Andi Kleen committed
950
	br_read_unlock(&vfsmount_lock);
951
	return -ECHILD;
Nick Piggin's avatar
Nick Piggin committed
952 953
}

954 955 956 957 958
/*
 * 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.
 */
959
int follow_down(struct path *path)
960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978
{
	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);
979
			ret = path->dentry->d_op->d_manage(
980
				path->dentry, false);
981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002
			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;
}

1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018
/*
 * 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);
	}
}

Nick Piggin's avatar
Nick Piggin committed
1019
static void follow_dotdot(struct nameidata *nd)
Linus Torvalds's avatar
Linus Torvalds committed
1020
{
Al Viro's avatar
Al Viro committed
1021
	set_root(nd);
1022

Linus Torvalds's avatar
Linus Torvalds committed
1023
	while(1) {
1024
		struct dentry *old = nd->path.dentry;
Linus Torvalds's avatar
Linus Torvalds committed
1025

Al Viro's avatar
Al Viro committed
1026 1027
		if (nd->path.dentry == nd->root.dentry &&
		    nd->path.mnt == nd->root.mnt) {
Linus Torvalds's avatar
Linus Torvalds committed
1028 1029
			break;
		}
1030
		if (nd->path.dentry != nd->path.mnt->mnt_root) {
Al Viro's avatar
Al Viro committed
1031 1032
			/* rare case of legitimate dget_parent()... */
			nd->path.dentry = dget_parent(nd->path.dentry);
Linus Torvalds's avatar
Linus Torvalds committed
1033 1034 1035
			dput(old);
			break;
		}
Al Viro's avatar
Al Viro committed
1036
		if (!follow_up(&nd->path))
Linus Torvalds's avatar
Linus Torvalds committed
1037 1038
			break;
	}
Al Viro's avatar
Al Viro committed
1039
	follow_mount(&nd->path);
Nick Piggin's avatar
Nick Piggin committed
1040
	nd->inode = nd->path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
1041 1042
}

1043
/*
Miklos Szeredi's avatar
Miklos Szeredi committed
1044 1045 1046 1047 1048
 * 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
1049
 */
Miklos Szeredi's avatar
Miklos Szeredi committed
1050 1051
static struct dentry *lookup_dcache(struct qstr *name, struct dentry *dir,
				    struct nameidata *nd, bool *need_lookup)
1052 1053
{
	struct dentry *dentry;
Miklos Szeredi's avatar
Miklos Szeredi committed
1054
	int error;
1055

Miklos Szeredi's avatar
Miklos Szeredi committed
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073
	*need_lookup = false;
	dentry = d_lookup(dir, name);
	if (dentry) {
		if (d_need_lookup(dentry)) {
			*need_lookup = true;
		} else if (dentry->d_flags & DCACHE_OP_REVALIDATE) {
			error = d_revalidate(dentry, nd);
			if (unlikely(error <= 0)) {
				if (error < 0) {
					dput(dentry);
					return ERR_PTR(error);
				} else if (!d_invalidate(dentry)) {
					dput(dentry);
					dentry = NULL;
				}
			}
		}
	}
1074

Miklos Szeredi's avatar
Miklos Szeredi committed
1075 1076 1077 1078
	if (!dentry) {
		dentry = d_alloc(dir, name);
		if (unlikely(!dentry))
			return ERR_PTR(-ENOMEM);
1079

Miklos Szeredi's avatar
Miklos Szeredi committed
1080
		*need_lookup = true;
1081 1082 1083 1084
	}
	return dentry;
}

1085
/*
Miklos Szeredi's avatar
Miklos Szeredi committed
1086 1087 1088 1089
 * Call i_op->lookup on the dentry.  The dentry must be negative but may be
 * hashed if it was pouplated with DCACHE_NEED_LOOKUP.
 *
 * dir->d_inode->i_mutex must be held
1090
 */
Miklos Szeredi's avatar
Miklos Szeredi committed
1091 1092
static struct dentry *lookup_real(struct inode *dir, struct dentry *dentry,
				  struct nameidata *nd)
1093 1094 1095 1096
{
	struct dentry *old;

	/* Don't create child dentry for a dead directory. */
Miklos Szeredi's avatar
Miklos Szeredi committed
1097
	if (unlikely(IS_DEADDIR(dir))) {
1098
		dput(dentry);
1099
		return ERR_PTR(-ENOENT);
1100
	}
1101

Miklos Szeredi's avatar
Miklos Szeredi committed
1102
	old = dir->i_op->lookup(dir, dentry, nd);
1103 1104 1105 1106 1107 1108 1109
	if (unlikely(old)) {
		dput(dentry);
		dentry = old;
	}
	return dentry;
}

1110 1111 1112
static struct dentry *__lookup_hash(struct qstr *name,
		struct dentry *base, struct nameidata *nd)
{
Miklos Szeredi's avatar
Miklos Szeredi committed
1113
	bool need_lookup;
1114 1115
	struct dentry *dentry;

Miklos Szeredi's avatar
Miklos Szeredi committed
1116 1117 1118
	dentry = lookup_dcache(name, base, nd, &need_lookup);
	if (!need_lookup)
		return dentry;
1119

Miklos Szeredi's avatar
Miklos Szeredi committed
1120
	return lookup_real(base->d_inode, dentry, nd);
1121 1122
}

Linus Torvalds's avatar
Linus Torvalds committed
1123 1124 1125 1126 1127
/*
 *  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.
 */
Miklos Szeredi's avatar
Miklos Szeredi committed
1128 1129
static int lookup_fast(struct nameidata *nd, struct qstr *name,
		       struct path *path, struct inode **inode)
Linus Torvalds's avatar
Linus Torvalds committed
1130
{
1131
	struct vfsmount *mnt = nd->path.mnt;
Nick Piggin's avatar
Nick Piggin committed
1132
	struct dentry *dentry, *parent = nd->path.dentry;
Al Viro's avatar
Al Viro committed
1133 1134
	int need_reval = 1;
	int status = 1;
1135 1136
	int err;

1137 1138 1139 1140 1141
	/*
	 * 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.
	 */
Nick Piggin's avatar
Nick Piggin committed
1142 1143
	if (nd->flags & LOOKUP_RCU) {
		unsigned seq;
1144
		dentry = __d_lookup_rcu(parent, name, &seq, nd->inode);
Al Viro's avatar
Al Viro committed
1145 1146 1147
		if (!dentry)
			goto unlazy;

1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162
		/*
		 * 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.
		 */
Nick Piggin's avatar
Nick Piggin committed
1163 1164 1165
		if (__read_seqcount_retry(&parent->d_seq, nd->seq))
			return -ECHILD;
		nd->seq = seq;
Al Viro's avatar
Al Viro committed
1166

1167 1168
		if (unlikely(d_need_lookup(dentry)))
			goto unlazy;
1169
		if (unlikely(dentry->d_flags & DCACHE_OP_REVALIDATE)) {
Al Viro's avatar
Al Viro committed
1170 1171 1172 1173 1174 1175
			status = d_revalidate(dentry, nd);
			if (unlikely(status <= 0)) {
				if (status != -ECHILD)
					need_reval = 0;
				goto unlazy;
			}
1176
		}
Nick Piggin's avatar
Nick Piggin committed
1177 1178
		path->mnt = mnt;
		path->dentry = dentry;
1179 1180 1181 1182 1183
		if (unlikely(!__follow_mount_rcu(nd, path, inode)))
			goto unlazy;
		if (unlikely(path->dentry->d_flags & DCACHE_NEED_AUTOMOUNT))
			goto unlazy;
		return 0;
Al Viro's avatar
Al Viro committed
1184
unlazy:
1185 1186
		if (unlazy_walk(nd, dentry))
			return -ECHILD;
Al Viro's avatar
Al Viro committed
1187 1188
	} else {
		dentry = __d_lookup(parent, name);
1189
	}
Al Viro's avatar
Al Viro committed
1190

1191 1192 1193 1194
	if (unlikely(!dentry))
		goto need_lookup;

	if (unlikely(d_need_lookup(dentry))) {
1195
		dput(dentry);
1196
		goto need_lookup;
Al Viro's avatar
Al Viro committed
1197
	}
1198

Al Viro's avatar
Al Viro committed
1199 1200 1201 1202 1203 1204 1205 1206 1207
	if (unlikely(dentry->d_flags & DCACHE_OP_REVALIDATE) && need_reval)
		status = d_revalidate(dentry, nd);
	if (unlikely(status <= 0)) {
		if (status < 0) {
			dput(dentry);
			return status;
		}
		if (!d_invalidate(dentry)) {
			dput(dentry);
1208
			goto need_lookup;
Al Viro's avatar
Al Viro committed
1209
		}
1210
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
1211

1212 1213 1214
	path->mnt = mnt;
	path->dentry = dentry;
	err = follow_managed(path, nd->flags);
1215 1216
	if (unlikely(err < 0)) {
		path_put_conditional(path, nd);
1217
		return err;
1218
	}
1219 1220
	if (err)
		nd->flags |= LOOKUP_JUMPED;
1221
	*inode = path->dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
1222
	return 0;
1223 1224

need_lookup:
Miklos Szeredi's avatar
Miklos Szeredi committed
1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
	return 1;
}

/* Fast lookup failed, do it the slow way */
static int lookup_slow(struct nameidata *nd, struct qstr *name,
		       struct path *path)
{
	struct dentry *dentry, *parent;
	int err;

	parent = nd->path.dentry;
1236 1237 1238 1239 1240 1241 1242
	BUG_ON(nd->inode != parent->d_inode);

	mutex_lock(&parent->d_inode->i_mutex);
	dentry = __lookup_hash(name, parent, nd);
	mutex_unlock(&parent->d_inode->i_mutex);
	if (IS_ERR(dentry))
		return PTR_ERR(dentry);
Miklos Szeredi's avatar
Miklos Szeredi committed
1243 1244 1245 1246 1247 1248 1249 1250 1251 1252
	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;
Linus Torvalds's avatar
Linus Torvalds committed
1253 1254
}

1255 1256 1257
static inline int may_lookup(struct nameidata *nd)
{
	if (nd->flags & LOOKUP_RCU) {
1258
		int err = inode_permission(nd->inode, MAY_EXEC|MAY_NOT_BLOCK);
1259 1260
		if (err != -ECHILD)
			return err;
1261
		if (unlazy_walk(nd, NULL))
1262 1263
			return -ECHILD;
	}
1264
	return inode_permission(nd->inode, MAY_EXEC);
1265 1266
}

1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278
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;
}

1279 1280 1281 1282 1283 1284
static void terminate_walk(struct nameidata *nd)
{
	if (!(nd->flags & LOOKUP_RCU)) {
		path_put(&nd->path);
	} else {
		nd->flags &= ~LOOKUP_RCU;
1285 1286
		if (!(nd->flags & LOOKUP_ROOT))
			nd->root.mnt = NULL;
1287
		rcu_read_unlock();
Andi Kleen's avatar
Andi Kleen committed
1288
		br_read_unlock(&vfsmount_lock);
1289 1290 1291
	}
}

1292 1293 1294 1295 1296 1297
/*
 * 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.
 */
1298
static inline int should_follow_link(struct inode *inode, int follow)
1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311
{
	if (unlikely(!(inode->i_opflags & IOP_NOFOLLOW))) {
		if (likely(inode->i_op->follow_link))
			return follow;

		/* This gets set once for the inode lifetime */
		spin_lock(&inode->i_lock);
		inode->i_opflags |= IOP_NOFOLLOW;
		spin_unlock(&inode->i_lock);
	}
	return 0;
}

1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323
static inline int walk_component(struct nameidata *nd, struct path *path,
		struct qstr *name, int type, int follow)
{
	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.
	 */
	if (unlikely(type != LAST_NORM))
		return handle_dots(nd, type);
Miklos Szeredi's avatar
Miklos Szeredi committed
1324
	err = lookup_fast(nd, name, path, &inode);
1325
	if (unlikely(err)) {
Miklos Szeredi's avatar
Miklos Szeredi committed
1326 1327 1328 1329 1330 1331 1332 1333
		if (err < 0)
			goto out_err;

		err = lookup_slow(nd, name, path);
		if (err < 0)
			goto out_err;

		inode = path->dentry->d_inode;
1334
	}
Miklos Szeredi's avatar
Miklos Szeredi committed
1335 1336 1337 1338
	err = -ENOENT;
	if (!inode)
		goto out_path_put;

1339
	if (should_follow_link(inode, follow)) {
1340 1341
		if (nd->flags & LOOKUP_RCU) {
			if (unlikely(unlazy_walk(nd, path->dentry))) {
Miklos Szeredi's avatar
Miklos Szeredi committed
1342 1343
				err = -ECHILD;
				goto out_err;
1344 1345
			}
		}
1346 1347 1348 1349 1350 1351
		BUG_ON(inode != path->dentry->d_inode);
		return 1;
	}
	path_to_nameidata(path, nd);
	nd->inode = inode;
	return 0;
Miklos Szeredi's avatar
Miklos Szeredi committed
1352 1353 1354 1355 1356 1357

out_path_put:
	path_to_nameidata(path, nd);
out_err:
	terminate_walk(nd);
	return err;
1358 1359
}

1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
/*
 * 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;
	}
1376
	BUG_ON(nd->depth >= MAX_NESTED_LINKS);
1377 1378 1379 1380 1381 1382 1383

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

	do {
		struct path link = *path;
		void *cookie;
1384 1385

		res = follow_link(&link, nd, &cookie);
1386 1387 1388
		if (!res)
			res = walk_component(nd, path, &nd->last,
					     nd->last_type, LOOKUP_FOLLOW);
1389
		put_link(nd, &link, cookie);
1390 1391 1392 1393 1394 1395 1396
	} while (res > 0);

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

1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416
/*
 * We really don't want to look at inode->i_op->lookup
 * when we don't have to. So we keep a cache bit in
 * the inode ->i_opflags field that says "yes, we can
 * do lookup on this inode".
 */
static inline int can_lookup(struct inode *inode)
{
	if (likely(inode->i_opflags & IOP_LOOKUP))
		return 1;
	if (likely(!inode->i_op->lookup))
		return 0;

	/* We do this once for the lifetime of the inode */
	spin_lock(&inode->i_lock);
	inode->i_opflags |= IOP_LOOKUP;
	spin_unlock(&inode->i_lock);
	return 1;
}

1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440
/*
 * 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

1441
#include <asm/word-at-a-time.h>
1442

1443
#ifdef CONFIG_64BIT
1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462

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 (;;) {
1463
		a = load_unaligned_zeropad(name);
1464 1465 1466
		if (len < sizeof(unsigned long))
			break;
		hash += a;
1467
		hash *= 9;
1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485
		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)
{
1486 1487
	unsigned long a, b, adata, bdata, mask, hash, len;
	const struct word_at_a_time constants = WORD_AT_A_TIME_CONSTANTS;
1488 1489 1490 1491 1492 1493

	hash = a = 0;
	len = -sizeof(unsigned long);
	do {
		hash = (hash + a) * 9;
		len += sizeof(unsigned long);
1494
		a = load_unaligned_zeropad(name+len);
1495 1496 1497 1498 1499 1500 1501 1502 1503
		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);
1504 1505
	*hashp = fold_hash(hash);

1506
	return len + find_zero(mask);
1507 1508 1509 1510
}

#else

1511 1512 1513 1514 1515 1516 1517
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);
}
1518
EXPORT_SYMBOL(full_name_hash);
1519

1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538
/*
 * 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;
}

1539 1540
#endif

Linus Torvalds's avatar
Linus Torvalds committed
1541 1542
/*
 * Name resolution.
1543 1544
 * This is the basic name resolution function, turning a pathname into
 * the final dentry. We expect 'base' to be positive and a directory.
Linus Torvalds's avatar
Linus Torvalds committed
1545
 *
1546 1547
 * Returns 0 and nd will have valid dentry and mnt on success.
 * Returns error and drops reference to input namei data on failure.
Linus Torvalds's avatar
Linus Torvalds committed
1548
 */
1549
static int link_path_walk(const char *name, struct nameidata *nd)
Linus Torvalds's avatar
Linus Torvalds committed
1550 1551 1552 1553 1554 1555 1556
{
	struct path next;
	int err;
	
	while (*name=='/')
		name++;
	if (!*name)
1557
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
1558 1559 1560 1561

	/* At this point we know we have a real path component. */
	for(;;) {
		struct qstr this;
1562
		long len;
Al Viro's avatar
Al Viro committed
1563
		int type;
Linus Torvalds's avatar
Linus Torvalds committed
1564

1565
		err = may_lookup(nd);
Linus Torvalds's avatar
Linus Torvalds committed
1566 1567 1568
 		if (err)
			break;

1569
		len = hash_name(name, &this.hash);
Linus Torvalds's avatar
Linus Torvalds committed
1570
		this.name = name;
1571
		this.len = len;
Linus Torvalds's avatar
Linus Torvalds committed
1572

Al Viro's avatar
Al Viro committed
1573
		type = LAST_NORM;
1574
		if (name[0] == '.') switch (len) {
Al Viro's avatar
Al Viro committed
1575
			case 2:
1576
				if (name[1] == '.') {
Al Viro's avatar
Al Viro committed
1577
					type = LAST_DOTDOT;
Al Viro's avatar
Al Viro committed
1578 1579
					nd->flags |= LOOKUP_JUMPED;
				}
Al Viro's avatar
Al Viro committed
1580 1581 1582 1583
				break;
			case 1:
				type = LAST_DOT;
		}
1584 1585
		if (likely(type == LAST_NORM)) {
			struct dentry *parent = nd->path.dentry;
Al Viro's avatar
Al Viro committed
1586
			nd->flags &= ~LOOKUP_JUMPED;
1587 1588 1589 1590 1591 1592 1593
			if (unlikely(parent->d_flags & DCACHE_OP_HASH)) {
				err = parent->d_op->d_hash(parent, nd->inode,
							   &this);
				if (err < 0)
					break;
			}
		}
Al Viro's avatar
Al Viro committed
1594

1595
		if (!name[len])
Linus Torvalds's avatar
Linus Torvalds committed
1596
			goto last_component;
1597 1598 1599 1600 1601 1602 1603 1604
		/*
		 * 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])
1605
			goto last_component;
1606
		name += len;
Linus Torvalds's avatar
Linus Torvalds committed
1607

1608 1609 1610
		err = walk_component(nd, &next, &this, type, LOOKUP_FOLLOW);
		if (err < 0)
			return err;
Linus Torvalds's avatar
Linus Torvalds committed
1611

1612
		if (err) {
1613
			err = nested_symlink(&next, nd);
Linus Torvalds's avatar
Linus Torvalds committed
1614
			if (err)
1615
				return err;
Nick Piggin's avatar
Nick Piggin committed
1616
		}
1617 1618
		if (can_lookup(nd->inode))
			continue;
Linus Torvalds's avatar
Linus Torvalds committed
1619
		err = -ENOTDIR; 
1620
		break;
Linus Torvalds's avatar
Linus Torvalds committed
1621 1622 1623
		/* here ends the main loop */

last_component:
1624 1625
		nd->last = this;
		nd->last_type = type;
1626
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
1627
	}
1628
	terminate_walk(nd);
Linus Torvalds's avatar
Linus Torvalds committed
1629 1630 1631
	return err;
}

Al Viro's avatar
Al Viro committed
1632 1633
static int path_init(int dfd, const char *name, unsigned int flags,
		     struct nameidata *nd, struct file **fp)
Nick Piggin's avatar
Nick Piggin committed
1634 1635 1636 1637 1638 1639
{
	int retval = 0;
	int fput_needed;
	struct file *file;

	nd->last_type = LAST_ROOT; /* if there are only slashes... */
Al Viro's avatar
Al Viro committed
1640
	nd->flags = flags | LOOKUP_JUMPED;
Nick Piggin's avatar
Nick Piggin committed
1641
	nd->depth = 0;
1642 1643
	if (flags & LOOKUP_ROOT) {
		struct inode *inode = nd->root.dentry->d_inode;
1644 1645 1646 1647 1648 1649 1650
		if (*name) {
			if (!inode->i_op->lookup)
				return -ENOTDIR;
			retval = inode_permission(inode, MAY_EXEC);
			if (retval)
				return retval;
		}
1651 1652 1653
		nd->path = nd->root;
		nd->inode = inode;
		if (flags & LOOKUP_RCU) {
Andi Kleen's avatar
Andi Kleen committed
1654
			br_read_lock(&vfsmount_lock);
1655 1656 1657 1658 1659 1660 1661 1662
			rcu_read_lock();
			nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq);
		} else {
			path_get(&nd->path);
		}
		return 0;
	}

Nick Piggin's avatar
Nick Piggin committed
1663 1664 1665
	nd->root.mnt = NULL;

	if (*name=='/') {
Al Viro's avatar
Al Viro committed
1666
		if (flags & LOOKUP_RCU) {
Andi Kleen's avatar
Andi Kleen committed
1667
			br_read_lock(&vfsmount_lock);
Al Viro's avatar
Al Viro committed
1668 1669 1670 1671 1672 1673 1674
			rcu_read_lock();
			set_root_rcu(nd);
		} else {
			set_root(nd);
			path_get(&nd->root);
		}
		nd->path = nd->root;
Nick Piggin's avatar
Nick Piggin committed
1675
	} else if (dfd == AT_FDCWD) {
Al Viro's avatar
Al Viro committed
1676 1677 1678
		if (flags & LOOKUP_RCU) {
			struct fs_struct *fs = current->fs;
			unsigned seq;
Nick Piggin's avatar
Nick Piggin committed
1679

Andi Kleen's avatar
Andi Kleen committed
1680
			br_read_lock(&vfsmount_lock);
Al Viro's avatar
Al Viro committed
1681
			rcu_read_lock();
Nick Piggin's avatar
Nick Piggin committed
1682

Al Viro's avatar
Al Viro committed
1683 1684 1685 1686 1687 1688 1689 1690
			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);
		}
Nick Piggin's avatar
Nick Piggin committed
1691 1692 1693
	} else {
		struct dentry *dentry;

1694
		file = fget_raw_light(dfd, &fput_needed);
Nick Piggin's avatar
Nick Piggin committed
1695 1696 1697 1698 1699 1700
		retval = -EBADF;
		if (!file)
			goto out_fail;

		dentry = file->f_path.dentry;

Al Viro's avatar
Al Viro committed
1701 1702 1703 1704
		if (*name) {
			retval = -ENOTDIR;
			if (!S_ISDIR(dentry->d_inode->i_mode))
				goto fput_fail;
Nick Piggin's avatar
Nick Piggin committed
1705

1706
			retval = inode_permission(dentry->d_inode, MAY_EXEC);
Al Viro's avatar
Al Viro committed
1707 1708 1709
			if (retval)
				goto fput_fail;
		}
Nick Piggin's avatar
Nick Piggin committed
1710 1711

		nd->path = file->f_path;
Al Viro's avatar
Al Viro committed
1712 1713
		if (flags & LOOKUP_RCU) {
			if (fput_needed)
Al Viro's avatar
Al Viro committed
1714
				*fp = file;
Al Viro's avatar
Al Viro committed
1715
			nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq);
Andi Kleen's avatar
Andi Kleen committed
1716
			br_read_lock(&vfsmount_lock);
Al Viro's avatar
Al Viro committed
1717 1718 1719 1720 1721
			rcu_read_lock();
		} else {
			path_get(&file->f_path);
			fput_light(file, fput_needed);
		}
Nick Piggin's avatar
Nick Piggin committed
1722 1723 1724
	}

	nd->inode = nd->path.dentry->d_inode;
1725
	return 0;
1726

1727 1728 1729 1730 1731 1732
fput_fail:
	fput_light(file, fput_needed);
out_fail:
	return retval;
}

1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
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;
	return walk_component(nd, path, &nd->last, nd->last_type,
					nd->flags & LOOKUP_FOLLOW);
}

1743
/* Returns 0 and nd will be valid on success; Retuns error, otherwise. */
Al Viro's avatar
Al Viro committed
1744
static int path_lookupat(int dfd, const char *name,
1745 1746
				unsigned int flags, struct nameidata *nd)
{
Al Viro's avatar
Al Viro committed
1747
	struct file *base = NULL;
1748 1749
	struct path path;
	int err;
Nick Piggin's avatar
Nick Piggin committed
1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764

	/*
	 * 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).
	 */
1765
	err = path_init(dfd, name, flags | LOOKUP_PARENT, nd, &base);
Al Viro's avatar
Al Viro committed
1766

1767 1768
	if (unlikely(err))
		return err;
Al Viro's avatar
Al Viro committed
1769 1770

	current->total_link_count = 0;
1771 1772 1773 1774 1775 1776 1777 1778
	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;
			nd->flags |= LOOKUP_PARENT;
1779
			err = follow_link(&link, nd, &cookie);
1780 1781
			if (!err)
				err = lookup_last(nd, &path);
1782
			put_link(nd, &link, cookie);
1783 1784
		}
	}
Al Viro's avatar
Al Viro committed
1785

1786 1787
	if (!err)
		err = complete_walk(nd);
1788 1789 1790 1791

	if (!err && nd->flags & LOOKUP_DIRECTORY) {
		if (!nd->inode->i_op->lookup) {
			path_put(&nd->path);
Al Viro's avatar
Al Viro committed
1792
			err = -ENOTDIR;
1793 1794
		}
	}
Al Viro's avatar
Al Viro committed
1795

Al Viro's avatar
Al Viro committed
1796 1797
	if (base)
		fput(base);
Al Viro's avatar
Al Viro committed
1798

1799
	if (nd->root.mnt && !(nd->flags & LOOKUP_ROOT)) {
Al Viro's avatar
Al Viro committed
1800 1801 1802
		path_put(&nd->root);
		nd->root.mnt = NULL;
	}
1803
	return err;
Al Viro's avatar
Al Viro committed
1804
}
Nick Piggin's avatar
Nick Piggin committed
1805

Al Viro's avatar
Al Viro committed
1806 1807 1808 1809 1810 1811 1812 1813
static int do_path_lookup(int dfd, const char *name,
				unsigned int flags, struct nameidata *nd)
{
	int retval = path_lookupat(dfd, name, flags | LOOKUP_RCU, nd);
	if (unlikely(retval == -ECHILD))
		retval = path_lookupat(dfd, name, flags, nd);
	if (unlikely(retval == -ESTALE))
		retval = path_lookupat(dfd, name, flags | LOOKUP_REVAL, nd);
Nick Piggin's avatar
Nick Piggin committed
1814 1815 1816 1817 1818 1819 1820

	if (likely(!retval)) {
		if (unlikely(!audit_dummy_context())) {
			if (nd->path.dentry && nd->inode)
				audit_inode(name, nd->path.dentry);
		}
	}
1821
	return retval;
Linus Torvalds's avatar
Linus Torvalds committed
1822 1823
}

Al Viro's avatar
Al Viro committed
1824
int kern_path_parent(const char *name, struct nameidata *nd)
1825
{
Al Viro's avatar
Al Viro committed
1826
	return do_path_lookup(AT_FDCWD, name, LOOKUP_PARENT, nd);
1827 1828
}

Al Viro's avatar
Al Viro committed
1829 1830 1831 1832 1833 1834 1835 1836 1837
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;
}

1838 1839 1840 1841 1842 1843
/**
 * 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
1844
 * @path: pointer to struct path to fill
1845 1846 1847
 */
int vfs_path_lookup(struct dentry *dentry, struct vfsmount *mnt,
		    const char *name, unsigned int flags,
1848
		    struct path *path)
1849
{
1850 1851 1852 1853 1854
	struct nameidata nd;
	int err;
	nd.root.dentry = dentry;
	nd.root.mnt = mnt;
	BUG_ON(flags & LOOKUP_PARENT);
1855
	/* the first argument of do_path_lookup() is ignored with LOOKUP_ROOT */
1856 1857 1858 1859
	err = do_path_lookup(AT_FDCWD, name, flags | LOOKUP_ROOT, &nd);
	if (!err)
		*path = nd.path;
	return err;
1860 1861
}

1862 1863 1864 1865 1866
/*
 * Restricted form of lookup. Doesn't follow links, single-component only,
 * needs parent already locked. Doesn't follow mounts.
 * SMP-safe.
 */
1867
static struct dentry *lookup_hash(struct nameidata *nd)
1868
{
1869
	return __lookup_hash(&nd->last, nd->path.dentry, nd);
Linus Torvalds's avatar
Linus Torvalds committed
1870 1871
}

1872
/**
1873
 * lookup_one_len - filesystem helper to lookup single pathname component
1874 1875 1876 1877
 * @name:	pathname component to lookup
 * @base:	base directory to lookup from
 * @len:	maximum length @len should be interpreted to
 *
1878 1879
 * 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
1880 1881 1882
 * nameidata argument is passed to the filesystem methods and a filesystem
 * using this helper needs to be prepared for that.
 */
1883 1884 1885
struct dentry *lookup_one_len(const char *name, struct dentry *base, int len)
{
	struct qstr this;
Al Viro's avatar
Al Viro committed
1886
	unsigned int c;
1887
	int err;
1888

1889 1890
	WARN_ON_ONCE(!mutex_is_locked(&base->d_inode->i_mutex));

Al Viro's avatar
Al Viro committed
1891 1892
	this.name = name;
	this.len = len;
1893
	this.hash = full_name_hash(name, len);
Al Viro's avatar
Al Viro committed
1894 1895 1896 1897 1898 1899 1900 1901
	if (!len)
		return ERR_PTR(-EACCES);

	while (len--) {
		c = *(const unsigned char *)name++;
		if (c == '/' || c == '\0')
			return ERR_PTR(-EACCES);
	}
1902 1903 1904 1905 1906 1907 1908 1909 1910
	/*
	 * See if the low-level filesystem might want
	 * to use its own hash..
	 */
	if (base->d_flags & DCACHE_OP_HASH) {
		int err = base->d_op->d_hash(base, base->d_inode, &this);
		if (err < 0)
			return ERR_PTR(err);
	}
1911

1912 1913 1914 1915
	err = inode_permission(base->d_inode, MAY_EXEC);
	if (err)
		return ERR_PTR(err);

1916
	return __lookup_hash(&this, base, NULL);
1917 1918
}

1919 1920
int user_path_at_empty(int dfd, const char __user *name, unsigned flags,
		 struct path *path, int *empty)
Linus Torvalds's avatar
Linus Torvalds committed
1921
{
1922
	struct nameidata nd;
1923
	char *tmp = getname_flags(name, flags, empty);
Linus Torvalds's avatar
Linus Torvalds committed
1924 1925
	int err = PTR_ERR(tmp);
	if (!IS_ERR(tmp)) {
1926 1927 1928 1929

		BUG_ON(flags & LOOKUP_PARENT);

		err = do_path_lookup(dfd, tmp, flags, &nd);
Linus Torvalds's avatar
Linus Torvalds committed
1930
		putname(tmp);
1931 1932
		if (!err)
			*path = nd.path;
Linus Torvalds's avatar
Linus Torvalds committed
1933 1934 1935 1936
	}
	return err;
}

1937 1938 1939
int user_path_at(int dfd, const char __user *name, unsigned flags,
		 struct path *path)
{
1940
	return user_path_at_empty(dfd, name, flags, path, NULL);
1941 1942
}

1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960
static int user_path_parent(int dfd, const char __user *path,
			struct nameidata *nd, char **name)
{
	char *s = getname(path);
	int error;

	if (IS_ERR(s))
		return PTR_ERR(s);

	error = do_path_lookup(dfd, s, LOOKUP_PARENT, nd);
	if (error)
		putname(s);
	else
		*name = s;

	return error;
}

Linus Torvalds's avatar
Linus Torvalds committed
1961 1962 1963 1964 1965 1966
/*
 * 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)
{
1967
	kuid_t fsuid = current_fsuid();
1968

Linus Torvalds's avatar
Linus Torvalds committed
1969 1970
	if (!(dir->i_mode & S_ISVTX))
		return 0;
1971
	if (uid_eq(inode->i_uid, fsuid))
Linus Torvalds's avatar
Linus Torvalds committed
1972
		return 0;
1973
	if (uid_eq(dir->i_uid, fsuid))
Linus Torvalds's avatar
Linus Torvalds committed
1974
		return 0;
1975
	return !inode_capable(inode, CAP_FOWNER);
Linus Torvalds's avatar
Linus Torvalds committed
1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
}

/*
 *	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().
 */
1997
static int may_delete(struct inode *dir,struct dentry *victim,int isdir)
Linus Torvalds's avatar
Linus Torvalds committed
1998 1999 2000 2001 2002 2003 2004
{
	int error;

	if (!victim->d_inode)
		return -ENOENT;

	BUG_ON(victim->d_parent->d_inode != dir);
2005
	audit_inode_child(victim, dir);
Linus Torvalds's avatar
Linus Torvalds committed
2006

2007
	error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
Linus Torvalds's avatar
Linus Torvalds committed
2008 2009 2010 2011 2012
	if (error)
		return error;
	if (IS_APPEND(dir))
		return -EPERM;
	if (check_sticky(dir, victim->d_inode)||IS_APPEND(victim->d_inode)||
2013
	    IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
Linus Torvalds's avatar
Linus Torvalds committed
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036
		return -EPERM;
	if (isdir) {
		if (!S_ISDIR(victim->d_inode->i_mode))
			return -ENOTDIR;
		if (IS_ROOT(victim))
			return -EBUSY;
	} else if (S_ISDIR(victim->d_inode->i_mode))
		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())
 */
2037
static inline int may_create(struct inode *dir, struct dentry *child)
Linus Torvalds's avatar
Linus Torvalds committed
2038 2039 2040 2041 2042
{
	if (child->d_inode)
		return -EEXIST;
	if (IS_DEADDIR(dir))
		return -ENOENT;
2043
	return inode_permission(dir, MAY_WRITE | MAY_EXEC);
Linus Torvalds's avatar
Linus Torvalds committed
2044 2045 2046 2047 2048 2049 2050 2051 2052 2053
}

/*
 * 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) {
2054
		mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_PARENT);
Linus Torvalds's avatar
Linus Torvalds committed
2055 2056 2057
		return NULL;
	}

2058
	mutex_lock(&p1->d_inode->i_sb->s_vfs_rename_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
2059

2060 2061 2062 2063 2064
	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;
Linus Torvalds's avatar
Linus Torvalds committed
2065 2066
	}

2067 2068 2069 2070 2071
	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;
Linus Torvalds's avatar
Linus Torvalds committed
2072 2073
	}

2074 2075
	mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_PARENT);
	mutex_lock_nested(&p2->d_inode->i_mutex, I_MUTEX_CHILD);
Linus Torvalds's avatar
Linus Torvalds committed
2076 2077 2078 2079 2080
	return NULL;
}

void unlock_rename(struct dentry *p1, struct dentry *p2)
{
2081
	mutex_unlock(&p1->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
2082
	if (p1 != p2) {
2083
		mutex_unlock(&p2->d_inode->i_mutex);
2084
		mutex_unlock(&p1->d_inode->i_sb->s_vfs_rename_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
2085 2086 2087
	}
}

Al Viro's avatar
Al Viro committed
2088
int vfs_create(struct inode *dir, struct dentry *dentry, umode_t mode,
Linus Torvalds's avatar
Linus Torvalds committed
2089 2090
		struct nameidata *nd)
{
2091
	int error = may_create(dir, dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2092 2093 2094 2095

	if (error)
		return error;

Al Viro's avatar
Al Viro committed
2096
	if (!dir->i_op->create)
Linus Torvalds's avatar
Linus Torvalds committed
2097 2098 2099 2100 2101 2102 2103
		return -EACCES;	/* shouldn't it be ENOSYS? */
	mode &= S_IALLUGO;
	mode |= S_IFREG;
	error = security_inode_create(dir, dentry, mode);
	if (error)
		return error;
	error = dir->i_op->create(dir, dentry, mode, nd);
2104
	if (!error)
2105
		fsnotify_create(dir, dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2106 2107 2108
	return error;
}

2109
static int may_open(struct path *path, int acc_mode, int flag)
Linus Torvalds's avatar
Linus Torvalds committed
2110
{
2111
	struct dentry *dentry = path->dentry;
Linus Torvalds's avatar
Linus Torvalds committed
2112 2113 2114
	struct inode *inode = dentry->d_inode;
	int error;

Al Viro's avatar
Al Viro committed
2115 2116 2117 2118
	/* O_PATH? */
	if (!acc_mode)
		return 0;

Linus Torvalds's avatar
Linus Torvalds committed
2119 2120 2121
	if (!inode)
		return -ENOENT;

Christoph Hellwig's avatar
Christoph Hellwig committed
2122 2123
	switch (inode->i_mode & S_IFMT) {
	case S_IFLNK:
Linus Torvalds's avatar
Linus Torvalds committed
2124
		return -ELOOP;
Christoph Hellwig's avatar
Christoph Hellwig committed
2125 2126 2127 2128 2129 2130
	case S_IFDIR:
		if (acc_mode & MAY_WRITE)
			return -EISDIR;
		break;
	case S_IFBLK:
	case S_IFCHR:
2131
		if (path->mnt->mnt_flags & MNT_NODEV)
Linus Torvalds's avatar
Linus Torvalds committed
2132
			return -EACCES;
Christoph Hellwig's avatar
Christoph Hellwig committed
2133 2134 2135
		/*FALLTHRU*/
	case S_IFIFO:
	case S_IFSOCK:
Linus Torvalds's avatar
Linus Torvalds committed
2136
		flag &= ~O_TRUNC;
Christoph Hellwig's avatar
Christoph Hellwig committed
2137
		break;
2138
	}
2139

2140
	error = inode_permission(inode, acc_mode);
2141 2142
	if (error)
		return error;
Mimi Zohar's avatar
Mimi Zohar committed
2143

Linus Torvalds's avatar
Linus Torvalds committed
2144 2145 2146 2147
	/*
	 * An append-only file must be opened in append mode for writing.
	 */
	if (IS_APPEND(inode)) {
2148
		if  ((flag & O_ACCMODE) != O_RDONLY && !(flag & O_APPEND))
2149
			return -EPERM;
Linus Torvalds's avatar
Linus Torvalds committed
2150
		if (flag & O_TRUNC)
2151
			return -EPERM;
Linus Torvalds's avatar
Linus Torvalds committed
2152 2153 2154
	}

	/* O_NOATIME can only be set by the owner or superuser */
2155
	if (flag & O_NOATIME && !inode_owner_or_capable(inode))
2156
		return -EPERM;
Linus Torvalds's avatar
Linus Torvalds committed
2157

2158
	return 0;
2159
}
Linus Torvalds's avatar
Linus Torvalds committed
2160

2161
static int handle_truncate(struct file *filp)
2162
{
2163
	struct path *path = &filp->f_path;
2164 2165 2166 2167 2168 2169 2170 2171 2172
	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)
2173
		error = security_path_truncate(path);
2174 2175 2176
	if (!error) {
		error = do_truncate(path->dentry, 0,
				    ATTR_MTIME|ATTR_CTIME|ATTR_OPEN,
2177
				    filp);
2178 2179
	}
	put_write_access(inode);
Mimi Zohar's avatar
Mimi Zohar committed
2180
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
2181 2182
}

2183 2184
static inline int open_to_namei_flags(int flag)
{
2185 2186
	if ((flag & O_ACCMODE) == 3)
		flag--;
2187 2188 2189
	return flag;
}

Nick Piggin's avatar
Nick Piggin committed
2190
/*
2191
 * Handle the last step of open()
Nick Piggin's avatar
Nick Piggin committed
2192
 */
2193
static struct file *do_last(struct nameidata *nd, struct path *path,
2194
			    const struct open_flags *op, const char *pathname)
2195
{
2196
	struct dentry *dir = nd->path.dentry;
2197
	struct dentry *dentry;
2198
	int open_flag = op->open_flag;
2199
	int will_truncate = open_flag & O_TRUNC;
2200
	int want_write = 0;
Al Viro's avatar
Al Viro committed
2201
	int acc_mode = op->acc_mode;
2202
	struct file *filp;
2203
	struct inode *inode;
2204
	int symlink_ok = 0;
2205 2206
	struct path save_parent = { .dentry = NULL, .mnt = NULL };
	bool retried = false;
Al Viro's avatar
Al Viro committed
2207
	int error;
2208

2209 2210 2211
	nd->flags &= ~LOOKUP_PARENT;
	nd->flags |= op->intent;

2212 2213
	switch (nd->last_type) {
	case LAST_DOTDOT:
2214
	case LAST_DOT:
2215 2216 2217
		error = handle_dots(nd, nd->last_type);
		if (error)
			return ERR_PTR(error);
2218 2219
		/* fallthrough */
	case LAST_ROOT:
2220
		error = complete_walk(nd);
Al Viro's avatar
Al Viro committed
2221
		if (error)
2222
			return ERR_PTR(error);
2223
		audit_inode(pathname, nd->path.dentry);
2224
		if (open_flag & O_CREAT) {
2225 2226 2227 2228
			error = -EISDIR;
			goto exit;
		}
		goto ok;
2229
	case LAST_BIND:
2230
		error = complete_walk(nd);
Al Viro's avatar
Al Viro committed
2231
		if (error)
2232
			return ERR_PTR(error);
2233
		audit_inode(pathname, dir);
2234
		goto ok;
2235
	}
2236

2237
	if (!(open_flag & O_CREAT)) {
2238 2239
		if (nd->last.name[nd->last.len])
			nd->flags |= LOOKUP_FOLLOW | LOOKUP_DIRECTORY;
Al Viro's avatar
Al Viro committed
2240 2241
		if (open_flag & O_PATH && !(nd->flags & LOOKUP_FOLLOW))
			symlink_ok = 1;
2242
		/* we _can_ be in RCU mode here */
2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253
		error = lookup_fast(nd, &nd->last, path, &inode);
		if (unlikely(error)) {
			if (error < 0)
				goto exit;

			error = lookup_slow(nd, &nd->last, path);
			if (error < 0)
				goto exit;

			inode = path->dentry->d_inode;
		}
2254
		goto finish_lookup;
2255 2256 2257
	}

	/* create side of things */
2258 2259 2260 2261
	/*
	 * 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
	 */
2262 2263 2264
	error = complete_walk(nd);
	if (error)
		return ERR_PTR(error);
2265 2266

	audit_inode(pathname, dir);
Al Viro's avatar
Al Viro committed
2267
	error = -EISDIR;
2268
	/* trailing slashes? */
Nick Piggin's avatar
Nick Piggin committed
2269 2270
	if (nd->last.name[nd->last.len])
		goto exit;
Al Viro's avatar
Al Viro committed
2271

2272
retry_lookup:
2273 2274
	mutex_lock(&dir->d_inode->i_mutex);

2275 2276 2277
	dentry = lookup_hash(nd);
	error = PTR_ERR(dentry);
	if (IS_ERR(dentry)) {
2278 2279 2280 2281
		mutex_unlock(&dir->d_inode->i_mutex);
		goto exit;
	}

2282 2283 2284
	path->dentry = dentry;
	path->mnt = nd->path.mnt;

2285
	/* Negative dentry, just create the file */
2286
	if (!dentry->d_inode) {
2287
		umode_t mode = op->mode;
2288 2289
		if (!IS_POSIXACL(dir->d_inode))
			mode &= ~current_umask();
2290 2291
		/*
		 * This write is needed to ensure that a
2292
		 * rw->ro transition does not occur between
2293 2294 2295 2296 2297 2298 2299
		 * the time when the file is created and when
		 * a permanent write count is taken through
		 * the 'struct file' in nameidata_to_filp().
		 */
		error = mnt_want_write(nd->path.mnt);
		if (error)
			goto exit_mutex_unlock;
2300
		want_write = 1;
2301
		/* Don't check for write permission, don't truncate */
2302
		open_flag &= ~O_TRUNC;
2303
		will_truncate = 0;
Al Viro's avatar
Al Viro committed
2304
		acc_mode = MAY_OPEN;
2305 2306 2307 2308 2309 2310 2311 2312 2313
		error = security_path_mknod(&nd->path, dentry, mode, 0);
		if (error)
			goto exit_mutex_unlock;
		error = vfs_create(dir->d_inode, dentry, mode, nd);
		if (error)
			goto exit_mutex_unlock;
		mutex_unlock(&dir->d_inode->i_mutex);
		dput(nd->path.dentry);
		nd->path.dentry = dentry;
2314
		goto common;
2315 2316 2317 2318 2319 2320 2321 2322 2323
	}

	/*
	 * It already exists.
	 */
	mutex_unlock(&dir->d_inode->i_mutex);
	audit_inode(pathname, path->dentry);

	error = -EEXIST;
2324
	if (open_flag & O_EXCL)
2325 2326
		goto exit_dput;

2327 2328 2329
	error = follow_managed(path, nd->flags);
	if (error < 0)
		goto exit_dput;
2330

2331 2332 2333
	if (error)
		nd->flags |= LOOKUP_JUMPED;

2334 2335
	BUG_ON(nd->flags & LOOKUP_RCU);
	inode = path->dentry->d_inode;
2336 2337
finish_lookup:
	/* we _can_ be in RCU mode here */
2338
	error = -ENOENT;
2339 2340 2341 2342
	if (!inode) {
		path_to_nameidata(path, nd);
		goto exit;
	}
Al Viro's avatar
Al Viro committed
2343

2344 2345 2346 2347 2348 2349 2350 2351
	if (should_follow_link(inode, !symlink_ok)) {
		if (nd->flags & LOOKUP_RCU) {
			if (unlikely(unlazy_walk(nd, path->dentry))) {
				error = -ECHILD;
				goto exit;
			}
		}
		BUG_ON(inode != path->dentry->d_inode);
2352
		return NULL;
2353
	}
2354

2355 2356 2357 2358 2359 2360 2361 2362
	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;

	}
2363
	nd->inode = inode;
2364 2365
	/* Why this, you ask?  _Now_ we might have grown LOOKUP_JUMPED... */
	error = complete_walk(nd);
2366 2367
	if (error) {
		path_put(&save_parent);
2368
		return ERR_PTR(error);
2369
	}
2370
	error = -EISDIR;
2371
	if ((open_flag & O_CREAT) && S_ISDIR(nd->inode->i_mode))
2372
		goto exit;
2373 2374 2375
	error = -ENOTDIR;
	if ((nd->flags & LOOKUP_DIRECTORY) && !nd->inode->i_op->lookup)
		goto exit;
2376
	audit_inode(pathname, nd->path.dentry);
2377
ok:
2378 2379 2380
	if (!S_ISREG(nd->inode->i_mode))
		will_truncate = 0;

2381 2382 2383 2384
	if (will_truncate) {
		error = mnt_want_write(nd->path.mnt);
		if (error)
			goto exit;
2385
		want_write = 1;
2386
	}
2387
common:
Al Viro's avatar
Al Viro committed
2388
	error = may_open(&nd->path, acc_mode, open_flag);
2389
	if (error)
2390 2391
		goto exit;
	filp = nameidata_to_filp(nd);
2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405
	if (filp == ERR_PTR(-EOPENSTALE) && save_parent.dentry && !retried) {
		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;
		if (want_write) {
			mnt_drop_write(nd->path.mnt);
			want_write = 0;
		}
		retried = true;
		goto retry_lookup;
	}
2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421
	if (!IS_ERR(filp)) {
		error = ima_file_check(filp, op->acc_mode);
		if (error) {
			fput(filp);
			filp = ERR_PTR(error);
		}
	}
	if (!IS_ERR(filp)) {
		if (will_truncate) {
			error = handle_truncate(filp);
			if (error) {
				fput(filp);
				filp = ERR_PTR(error);
			}
		}
	}
2422 2423
out:
	if (want_write)
2424
		mnt_drop_write(nd->path.mnt);
2425
	path_put(&save_parent);
2426
	terminate_walk(nd);
2427 2428 2429 2430 2431 2432 2433
	return filp;

exit_mutex_unlock:
	mutex_unlock(&dir->d_inode->i_mutex);
exit_dput:
	path_put_conditional(path, nd);
exit:
2434 2435
	filp = ERR_PTR(error);
	goto out;
2436 2437
}

2438
static struct file *path_openat(int dfd, const char *pathname,
2439
		struct nameidata *nd, const struct open_flags *op, int flags)
Linus Torvalds's avatar
Linus Torvalds committed
2440
{
2441
	struct file *base = NULL;
2442
	struct file *filp;
2443
	struct path path;
2444
	int error;
Nick Piggin's avatar
Nick Piggin committed
2445 2446 2447 2448 2449

	filp = get_empty_filp();
	if (!filp)
		return ERR_PTR(-ENFILE);

2450
	filp->f_flags = op->open_flag;
2451 2452 2453
	nd->intent.open.file = filp;
	nd->intent.open.flags = open_to_namei_flags(op->open_flag);
	nd->intent.open.create_mode = op->mode;
Nick Piggin's avatar
Nick Piggin committed
2454

2455
	error = path_init(dfd, pathname, flags | LOOKUP_PARENT, nd, &base);
Nick Piggin's avatar
Nick Piggin committed
2456
	if (unlikely(error))
2457
		goto out_filp;
Nick Piggin's avatar
Nick Piggin committed
2458

2459
	current->total_link_count = 0;
2460
	error = link_path_walk(pathname, nd);
Nick Piggin's avatar
Nick Piggin committed
2461 2462
	if (unlikely(error))
		goto out_filp;
Linus Torvalds's avatar
Linus Torvalds committed
2463

2464
	filp = do_last(nd, &path, op, pathname);
2465
	while (unlikely(!filp)) { /* trailing symlink */
2466
		struct path link = path;
Al Viro's avatar
Al Viro committed
2467
		void *cookie;
2468
		if (!(nd->flags & LOOKUP_FOLLOW)) {
2469 2470
			path_put_conditional(&path, nd);
			path_put(&nd->path);
2471 2472 2473
			filp = ERR_PTR(-ELOOP);
			break;
		}
2474 2475
		nd->flags |= LOOKUP_PARENT;
		nd->flags &= ~(LOOKUP_OPEN|LOOKUP_CREATE|LOOKUP_EXCL);
2476
		error = follow_link(&link, nd, &cookie);
2477
		if (unlikely(error))
2478
			filp = ERR_PTR(error);
2479
		else
2480
			filp = do_last(nd, &path, op, pathname);
2481
		put_link(nd, &link, cookie);
2482
	}
Al Viro's avatar
Al Viro committed
2483
out:
2484 2485
	if (nd->root.mnt && !(nd->flags & LOOKUP_ROOT))
		path_put(&nd->root);
2486 2487
	if (base)
		fput(base);
2488
	release_open_intent(nd);
2489 2490 2491 2492 2493 2494
	if (filp == ERR_PTR(-EOPENSTALE)) {
		if (flags & LOOKUP_RCU)
			filp = ERR_PTR(-ECHILD);
		else
			filp = ERR_PTR(-ESTALE);
	}
Al Viro's avatar
Al Viro committed
2495
	return filp;
Linus Torvalds's avatar
Linus Torvalds committed
2496

Nick Piggin's avatar
Nick Piggin committed
2497
out_filp:
2498
	filp = ERR_PTR(error);
Al Viro's avatar
Al Viro committed
2499
	goto out;
Linus Torvalds's avatar
Linus Torvalds committed
2500 2501
}

2502 2503 2504
struct file *do_filp_open(int dfd, const char *pathname,
		const struct open_flags *op, int flags)
{
2505
	struct nameidata nd;
2506 2507
	struct file *filp;

2508
	filp = path_openat(dfd, pathname, &nd, op, flags | LOOKUP_RCU);
2509
	if (unlikely(filp == ERR_PTR(-ECHILD)))
2510
		filp = path_openat(dfd, pathname, &nd, op, flags);
2511
	if (unlikely(filp == ERR_PTR(-ESTALE)))
2512
		filp = path_openat(dfd, pathname, &nd, op, flags | LOOKUP_REVAL);
2513 2514 2515
	return filp;
}

2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526
struct file *do_file_open_root(struct dentry *dentry, struct vfsmount *mnt,
		const char *name, const struct open_flags *op, int flags)
{
	struct nameidata nd;
	struct file *file;

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

	flags |= LOOKUP_ROOT;

Al Viro's avatar
Al Viro committed
2527
	if (dentry->d_inode->i_op->follow_link && op->intent & LOOKUP_OPEN)
2528 2529 2530 2531 2532 2533 2534 2535 2536 2537
		return ERR_PTR(-ELOOP);

	file = path_openat(-1, name, &nd, op, flags | LOOKUP_RCU);
	if (unlikely(file == ERR_PTR(-ECHILD)))
		file = path_openat(-1, name, &nd, op, flags);
	if (unlikely(file == ERR_PTR(-ESTALE)))
		file = path_openat(-1, name, &nd, op, flags | LOOKUP_REVAL);
	return file;
}

Al Viro's avatar
Al Viro committed
2538
struct dentry *kern_path_create(int dfd, const char *pathname, struct path *path, int is_dir)
Linus Torvalds's avatar
Linus Torvalds committed
2539
{
2540
	struct dentry *dentry = ERR_PTR(-EEXIST);
Al Viro's avatar
Al Viro committed
2541 2542 2543 2544
	struct nameidata nd;
	int error = do_path_lookup(dfd, pathname, LOOKUP_PARENT, &nd);
	if (error)
		return ERR_PTR(error);
Linus Torvalds's avatar
Linus Torvalds committed
2545

2546 2547 2548 2549
	/*
	 * Yucky last component or no last component at all?
	 * (foo/., foo/.., /////)
	 */
Al Viro's avatar
Al Viro committed
2550 2551 2552 2553 2554
	if (nd.last_type != LAST_NORM)
		goto out;
	nd.flags &= ~LOOKUP_PARENT;
	nd.flags |= LOOKUP_CREATE | LOOKUP_EXCL;
	nd.intent.open.flags = O_EXCL;
2555 2556 2557 2558

	/*
	 * Do the final lookup.
	 */
Al Viro's avatar
Al Viro committed
2559 2560
	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
	dentry = lookup_hash(&nd);
Linus Torvalds's avatar
Linus Torvalds committed
2561 2562
	if (IS_ERR(dentry))
		goto fail;
2563

2564 2565
	if (dentry->d_inode)
		goto eexist;
2566 2567 2568 2569 2570 2571
	/*
	 * 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.
	 */
Al Viro's avatar
Al Viro committed
2572
	if (unlikely(!is_dir && nd.last.name[nd.last.len])) {
2573 2574
		dput(dentry);
		dentry = ERR_PTR(-ENOENT);
Al Viro's avatar
Al Viro committed
2575
		goto fail;
2576
	}
Al Viro's avatar
Al Viro committed
2577
	*path = nd.path;
Linus Torvalds's avatar
Linus Torvalds committed
2578
	return dentry;
2579
eexist:
Linus Torvalds's avatar
Linus Torvalds committed
2580
	dput(dentry);
2581
	dentry = ERR_PTR(-EEXIST);
Linus Torvalds's avatar
Linus Torvalds committed
2582
fail:
Al Viro's avatar
Al Viro committed
2583 2584 2585
	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
out:
	path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
2586 2587
	return dentry;
}
2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601
EXPORT_SYMBOL(kern_path_create);

struct dentry *user_path_create(int dfd, const char __user *pathname, struct path *path, int is_dir)
{
	char *tmp = getname(pathname);
	struct dentry *res;
	if (IS_ERR(tmp))
		return ERR_CAST(tmp);
	res = kern_path_create(dfd, tmp, path, is_dir);
	putname(tmp);
	return res;
}
EXPORT_SYMBOL(user_path_create);

Al Viro's avatar
Al Viro committed
2602
int vfs_mknod(struct inode *dir, struct dentry *dentry, umode_t mode, dev_t dev)
Linus Torvalds's avatar
Linus Torvalds committed
2603
{
2604
	int error = may_create(dir, dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2605 2606 2607 2608

	if (error)
		return error;

2609
	if ((S_ISCHR(mode) || S_ISBLK(mode)) && !capable(CAP_MKNOD))
Linus Torvalds's avatar
Linus Torvalds committed
2610 2611
		return -EPERM;

Al Viro's avatar
Al Viro committed
2612
	if (!dir->i_op->mknod)
Linus Torvalds's avatar
Linus Torvalds committed
2613 2614
		return -EPERM;

2615 2616 2617 2618
	error = devcgroup_inode_mknod(mode, dev);
	if (error)
		return error;

Linus Torvalds's avatar
Linus Torvalds committed
2619 2620 2621 2622 2623
	error = security_inode_mknod(dir, dentry, mode, dev);
	if (error)
		return error;

	error = dir->i_op->mknod(dir, dentry, mode, dev);
2624
	if (!error)
2625
		fsnotify_create(dir, dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2626 2627 2628
	return error;
}

Al Viro's avatar
Al Viro committed
2629
static int may_mknod(umode_t mode)
2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645
{
	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;
	}
}

Al Viro's avatar
Al Viro committed
2646
SYSCALL_DEFINE4(mknodat, int, dfd, const char __user *, filename, umode_t, mode,
2647
		unsigned, dev)
Linus Torvalds's avatar
Linus Torvalds committed
2648
{
2649
	struct dentry *dentry;
2650 2651
	struct path path;
	int error;
Linus Torvalds's avatar
Linus Torvalds committed
2652 2653 2654 2655

	if (S_ISDIR(mode))
		return -EPERM;

2656 2657 2658
	dentry = user_path_create(dfd, filename, &path, 0);
	if (IS_ERR(dentry))
		return PTR_ERR(dentry);
2659

2660
	if (!IS_POSIXACL(path.dentry->d_inode))
Al Viro's avatar
Al Viro committed
2661
		mode &= ~current_umask();
2662 2663 2664
	error = may_mknod(mode);
	if (error)
		goto out_dput;
2665
	error = mnt_want_write(path.mnt);
2666 2667
	if (error)
		goto out_dput;
2668
	error = security_path_mknod(&path, dentry, mode, dev);
2669 2670
	if (error)
		goto out_drop_write;
2671
	switch (mode & S_IFMT) {
Linus Torvalds's avatar
Linus Torvalds committed
2672
		case 0: case S_IFREG:
2673
			error = vfs_create(path.dentry->d_inode,dentry,mode,NULL);
Linus Torvalds's avatar
Linus Torvalds committed
2674 2675
			break;
		case S_IFCHR: case S_IFBLK:
2676
			error = vfs_mknod(path.dentry->d_inode,dentry,mode,
Linus Torvalds's avatar
Linus Torvalds committed
2677 2678 2679
					new_decode_dev(dev));
			break;
		case S_IFIFO: case S_IFSOCK:
2680
			error = vfs_mknod(path.dentry->d_inode,dentry,mode,0);
Linus Torvalds's avatar
Linus Torvalds committed
2681 2682
			break;
	}
2683
out_drop_write:
2684
	mnt_drop_write(path.mnt);
2685 2686
out_dput:
	dput(dentry);
2687 2688
	mutex_unlock(&path.dentry->d_inode->i_mutex);
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
2689 2690 2691 2692

	return error;
}

Al Viro's avatar
Al Viro committed
2693
SYSCALL_DEFINE3(mknod, const char __user *, filename, umode_t, mode, unsigned, dev)
2694 2695 2696 2697
{
	return sys_mknodat(AT_FDCWD, filename, mode, dev);
}

2698
int vfs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
Linus Torvalds's avatar
Linus Torvalds committed
2699
{
2700
	int error = may_create(dir, dentry);
2701
	unsigned max_links = dir->i_sb->s_max_links;
Linus Torvalds's avatar
Linus Torvalds committed
2702 2703 2704 2705

	if (error)
		return error;

Al Viro's avatar
Al Viro committed
2706
	if (!dir->i_op->mkdir)
Linus Torvalds's avatar
Linus Torvalds committed
2707 2708 2709 2710 2711 2712 2713
		return -EPERM;

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

2714 2715 2716
	if (max_links && dir->i_nlink >= max_links)
		return -EMLINK;

Linus Torvalds's avatar
Linus Torvalds committed
2717
	error = dir->i_op->mkdir(dir, dentry, mode);
2718
	if (!error)
2719
		fsnotify_mkdir(dir, dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2720 2721 2722
	return error;
}

2723
SYSCALL_DEFINE3(mkdirat, int, dfd, const char __user *, pathname, umode_t, mode)
Linus Torvalds's avatar
Linus Torvalds committed
2724
{
2725
	struct dentry *dentry;
2726 2727
	struct path path;
	int error;
Linus Torvalds's avatar
Linus Torvalds committed
2728

2729
	dentry = user_path_create(dfd, pathname, &path, 1);
2730
	if (IS_ERR(dentry))
2731
		return PTR_ERR(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2732

2733
	if (!IS_POSIXACL(path.dentry->d_inode))
Al Viro's avatar
Al Viro committed
2734
		mode &= ~current_umask();
2735
	error = mnt_want_write(path.mnt);
2736 2737
	if (error)
		goto out_dput;
2738
	error = security_path_mkdir(&path, dentry, mode);
2739 2740
	if (error)
		goto out_drop_write;
2741
	error = vfs_mkdir(path.dentry->d_inode, dentry, mode);
2742
out_drop_write:
2743
	mnt_drop_write(path.mnt);
2744
out_dput:
2745
	dput(dentry);
2746 2747
	mutex_unlock(&path.dentry->d_inode->i_mutex);
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
2748 2749 2750
	return error;
}

2751
SYSCALL_DEFINE2(mkdir, const char __user *, pathname, umode_t, mode)
2752 2753 2754 2755
{
	return sys_mkdirat(AT_FDCWD, pathname, mode);
}

Linus Torvalds's avatar
Linus Torvalds committed
2756
/*
2757
 * The dentry_unhash() helper will try to drop the dentry early: we
2758
 * should have a usage count of 1 if we're the only user of this
2759 2760
 * dentry, and if that is true (possibly after pruning the dcache),
 * then we drop the dentry now.
Linus Torvalds's avatar
Linus Torvalds committed
2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772
 *
 * 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)
{
2773
	shrink_dcache_parent(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2774
	spin_lock(&dentry->d_lock);
2775
	if (dentry->d_count == 1)
Linus Torvalds's avatar
Linus Torvalds committed
2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786
		__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;

Al Viro's avatar
Al Viro committed
2787
	if (!dir->i_op->rmdir)
Linus Torvalds's avatar
Linus Torvalds committed
2788 2789
		return -EPERM;

2790
	dget(dentry);
2791
	mutex_lock(&dentry->d_inode->i_mutex);
Sage Weil's avatar
Sage Weil committed
2792 2793

	error = -EBUSY;
Linus Torvalds's avatar
Linus Torvalds committed
2794
	if (d_mountpoint(dentry))
Sage Weil's avatar
Sage Weil committed
2795 2796 2797 2798 2799 2800
		goto out;

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

2801
	shrink_dcache_parent(dentry);
Sage Weil's avatar
Sage Weil committed
2802 2803 2804 2805 2806 2807 2808 2809
	error = dir->i_op->rmdir(dir, dentry);
	if (error)
		goto out;

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

out:
2810
	mutex_unlock(&dentry->d_inode->i_mutex);
2811
	dput(dentry);
Sage Weil's avatar
Sage Weil committed
2812
	if (!error)
Linus Torvalds's avatar
Linus Torvalds committed
2813 2814 2815 2816
		d_delete(dentry);
	return error;
}

2817
static long do_rmdir(int dfd, const char __user *pathname)
Linus Torvalds's avatar
Linus Torvalds committed
2818 2819 2820 2821 2822 2823
{
	int error = 0;
	char * name;
	struct dentry *dentry;
	struct nameidata nd;

2824
	error = user_path_parent(dfd, pathname, &nd, &name);
Linus Torvalds's avatar
Linus Torvalds committed
2825
	if (error)
2826
		return error;
Linus Torvalds's avatar
Linus Torvalds committed
2827 2828

	switch(nd.last_type) {
2829 2830 2831 2832 2833 2834 2835 2836 2837
	case LAST_DOTDOT:
		error = -ENOTEMPTY;
		goto exit1;
	case LAST_DOT:
		error = -EINVAL;
		goto exit1;
	case LAST_ROOT:
		error = -EBUSY;
		goto exit1;
Linus Torvalds's avatar
Linus Torvalds committed
2838
	}
2839 2840 2841

	nd.flags &= ~LOOKUP_PARENT;

2842
	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
2843
	dentry = lookup_hash(&nd);
Linus Torvalds's avatar
Linus Torvalds committed
2844
	error = PTR_ERR(dentry);
2845 2846
	if (IS_ERR(dentry))
		goto exit2;
2847 2848 2849 2850
	if (!dentry->d_inode) {
		error = -ENOENT;
		goto exit3;
	}
2851 2852 2853
	error = mnt_want_write(nd.path.mnt);
	if (error)
		goto exit3;
2854 2855 2856
	error = security_path_rmdir(&nd.path, dentry);
	if (error)
		goto exit4;
2857
	error = vfs_rmdir(nd.path.dentry->d_inode, dentry);
2858
exit4:
2859 2860
	mnt_drop_write(nd.path.mnt);
exit3:
2861 2862
	dput(dentry);
exit2:
2863
	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
2864
exit1:
Jan Blunck's avatar
Jan Blunck committed
2865
	path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
2866 2867 2868 2869
	putname(name);
	return error;
}

2870
SYSCALL_DEFINE1(rmdir, const char __user *, pathname)
2871 2872 2873 2874
{
	return do_rmdir(AT_FDCWD, pathname);
}

Linus Torvalds's avatar
Linus Torvalds committed
2875 2876 2877 2878 2879 2880 2881
int vfs_unlink(struct inode *dir, struct dentry *dentry)
{
	int error = may_delete(dir, dentry, 0);

	if (error)
		return error;

Al Viro's avatar
Al Viro committed
2882
	if (!dir->i_op->unlink)
Linus Torvalds's avatar
Linus Torvalds committed
2883 2884
		return -EPERM;

2885
	mutex_lock(&dentry->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
2886 2887 2888 2889
	if (d_mountpoint(dentry))
		error = -EBUSY;
	else {
		error = security_inode_unlink(dir, dentry);
2890
		if (!error) {
Linus Torvalds's avatar
Linus Torvalds committed
2891
			error = dir->i_op->unlink(dir, dentry);
2892
			if (!error)
2893
				dont_mount(dentry);
2894
		}
Linus Torvalds's avatar
Linus Torvalds committed
2895
	}
2896
	mutex_unlock(&dentry->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
2897 2898 2899

	/* We don't d_delete() NFS sillyrenamed files--they still exist. */
	if (!error && !(dentry->d_flags & DCACHE_NFSFS_RENAMED)) {
2900
		fsnotify_link_count(dentry->d_inode);
2901
		d_delete(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2902
	}
Robert Love's avatar
Robert Love committed
2903

Linus Torvalds's avatar
Linus Torvalds committed
2904 2905 2906 2907 2908
	return error;
}

/*
 * Make sure that the actual truncation of the file will occur outside its
2909
 * directory's i_mutex.  Truncate can take a long time if there is a lot of
Linus Torvalds's avatar
Linus Torvalds committed
2910 2911 2912
 * writeout happening, and we don't want to prevent access to the directory
 * while waiting on the I/O.
 */
2913
static long do_unlinkat(int dfd, const char __user *pathname)
Linus Torvalds's avatar
Linus Torvalds committed
2914
{
2915 2916
	int error;
	char *name;
Linus Torvalds's avatar
Linus Torvalds committed
2917 2918 2919 2920
	struct dentry *dentry;
	struct nameidata nd;
	struct inode *inode = NULL;

2921
	error = user_path_parent(dfd, pathname, &nd, &name);
Linus Torvalds's avatar
Linus Torvalds committed
2922
	if (error)
2923 2924
		return error;

Linus Torvalds's avatar
Linus Torvalds committed
2925 2926 2927
	error = -EISDIR;
	if (nd.last_type != LAST_NORM)
		goto exit1;
2928 2929 2930

	nd.flags &= ~LOOKUP_PARENT;

2931
	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
2932
	dentry = lookup_hash(&nd);
Linus Torvalds's avatar
Linus Torvalds committed
2933 2934 2935
	error = PTR_ERR(dentry);
	if (!IS_ERR(dentry)) {
		/* Why not before? Because we want correct error value */
2936 2937
		if (nd.last.name[nd.last.len])
			goto slashes;
Linus Torvalds's avatar
Linus Torvalds committed
2938
		inode = dentry->d_inode;
2939
		if (!inode)
2940 2941
			goto slashes;
		ihold(inode);
2942 2943 2944
		error = mnt_want_write(nd.path.mnt);
		if (error)
			goto exit2;
2945 2946 2947
		error = security_path_unlink(&nd.path, dentry);
		if (error)
			goto exit3;
2948
		error = vfs_unlink(nd.path.dentry->d_inode, dentry);
2949
exit3:
2950
		mnt_drop_write(nd.path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
2951 2952 2953
	exit2:
		dput(dentry);
	}
2954
	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
2955 2956 2957
	if (inode)
		iput(inode);	/* truncate the inode here */
exit1:
Jan Blunck's avatar
Jan Blunck committed
2958
	path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
2959 2960 2961 2962 2963 2964 2965 2966 2967
	putname(name);
	return error;

slashes:
	error = !dentry->d_inode ? -ENOENT :
		S_ISDIR(dentry->d_inode->i_mode) ? -EISDIR : -ENOTDIR;
	goto exit2;
}

2968
SYSCALL_DEFINE3(unlinkat, int, dfd, const char __user *, pathname, int, flag)
2969 2970 2971 2972 2973 2974 2975 2976 2977 2978
{
	if ((flag & ~AT_REMOVEDIR) != 0)
		return -EINVAL;

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

	return do_unlinkat(dfd, pathname);
}

2979
SYSCALL_DEFINE1(unlink, const char __user *, pathname)
2980 2981 2982 2983
{
	return do_unlinkat(AT_FDCWD, pathname);
}

2984
int vfs_symlink(struct inode *dir, struct dentry *dentry, const char *oldname)
Linus Torvalds's avatar
Linus Torvalds committed
2985
{
2986
	int error = may_create(dir, dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2987 2988 2989 2990

	if (error)
		return error;

Al Viro's avatar
Al Viro committed
2991
	if (!dir->i_op->symlink)
Linus Torvalds's avatar
Linus Torvalds committed
2992 2993 2994 2995 2996 2997 2998
		return -EPERM;

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

	error = dir->i_op->symlink(dir, dentry, oldname);
2999
	if (!error)
3000
		fsnotify_create(dir, dentry);
Linus Torvalds's avatar
Linus Torvalds committed
3001 3002 3003
	return error;
}

3004 3005
SYSCALL_DEFINE3(symlinkat, const char __user *, oldname,
		int, newdfd, const char __user *, newname)
Linus Torvalds's avatar
Linus Torvalds committed
3006
{
3007 3008
	int error;
	char *from;
3009
	struct dentry *dentry;
3010
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
3011 3012

	from = getname(oldname);
3013
	if (IS_ERR(from))
Linus Torvalds's avatar
Linus Torvalds committed
3014 3015
		return PTR_ERR(from);

3016
	dentry = user_path_create(newdfd, newname, &path, 0);
3017 3018
	error = PTR_ERR(dentry);
	if (IS_ERR(dentry))
3019
		goto out_putname;
3020

3021
	error = mnt_want_write(path.mnt);
3022 3023
	if (error)
		goto out_dput;
3024
	error = security_path_symlink(&path, dentry, from);
3025 3026
	if (error)
		goto out_drop_write;
3027
	error = vfs_symlink(path.dentry->d_inode, dentry, from);
3028
out_drop_write:
3029
	mnt_drop_write(path.mnt);
3030
out_dput:
3031
	dput(dentry);
3032 3033
	mutex_unlock(&path.dentry->d_inode->i_mutex);
	path_put(&path);
3034
out_putname:
Linus Torvalds's avatar
Linus Torvalds committed
3035 3036 3037 3038
	putname(from);
	return error;
}

3039
SYSCALL_DEFINE2(symlink, const char __user *, oldname, const char __user *, newname)
3040 3041 3042 3043
{
	return sys_symlinkat(oldname, AT_FDCWD, newname);
}

Linus Torvalds's avatar
Linus Torvalds committed
3044 3045 3046
int vfs_link(struct dentry *old_dentry, struct inode *dir, struct dentry *new_dentry)
{
	struct inode *inode = old_dentry->d_inode;
3047
	unsigned max_links = dir->i_sb->s_max_links;
Linus Torvalds's avatar
Linus Torvalds committed
3048 3049 3050 3051 3052
	int error;

	if (!inode)
		return -ENOENT;

3053
	error = may_create(dir, new_dentry);
Linus Torvalds's avatar
Linus Torvalds committed
3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064
	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;
Al Viro's avatar
Al Viro committed
3065
	if (!dir->i_op->link)
Linus Torvalds's avatar
Linus Torvalds committed
3066
		return -EPERM;
3067
	if (S_ISDIR(inode->i_mode))
Linus Torvalds's avatar
Linus Torvalds committed
3068 3069 3070 3071 3072 3073
		return -EPERM;

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

3074
	mutex_lock(&inode->i_mutex);
3075 3076 3077
	/* Make sure we don't allow creating hardlink to an unlinked file */
	if (inode->i_nlink == 0)
		error =  -ENOENT;
3078 3079
	else if (max_links && inode->i_nlink >= max_links)
		error = -EMLINK;
3080 3081
	else
		error = dir->i_op->link(old_dentry, dir, new_dentry);
3082
	mutex_unlock(&inode->i_mutex);
3083
	if (!error)
3084
		fsnotify_link(dir, inode, new_dentry);
Linus Torvalds's avatar
Linus Torvalds committed
3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096
	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
 */
3097 3098
SYSCALL_DEFINE5(linkat, int, olddfd, const char __user *, oldname,
		int, newdfd, const char __user *, newname, int, flags)
Linus Torvalds's avatar
Linus Torvalds committed
3099 3100
{
	struct dentry *new_dentry;
3101
	struct path old_path, new_path;
3102
	int how = 0;
Linus Torvalds's avatar
Linus Torvalds committed
3103 3104
	int error;

3105
	if ((flags & ~(AT_SYMLINK_FOLLOW | AT_EMPTY_PATH)) != 0)
3106
		return -EINVAL;
3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119
	/*
	 * 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.
	 */
	if (flags & AT_EMPTY_PATH) {
		if (!capable(CAP_DAC_READ_SEARCH))
			return -ENOENT;
		how = LOOKUP_EMPTY;
	}

	if (flags & AT_SYMLINK_FOLLOW)
		how |= LOOKUP_FOLLOW;
3120

3121
	error = user_path_at(olddfd, oldname, how, &old_path);
Linus Torvalds's avatar
Linus Torvalds committed
3122
	if (error)
3123 3124
		return error;

3125
	new_dentry = user_path_create(newdfd, newname, &new_path, 0);
Linus Torvalds's avatar
Linus Torvalds committed
3126
	error = PTR_ERR(new_dentry);
3127
	if (IS_ERR(new_dentry))
3128 3129 3130 3131 3132 3133
		goto out;

	error = -EXDEV;
	if (old_path.mnt != new_path.mnt)
		goto out_dput;
	error = mnt_want_write(new_path.mnt);
3134 3135
	if (error)
		goto out_dput;
3136
	error = security_path_link(old_path.dentry, &new_path, new_dentry);
3137 3138
	if (error)
		goto out_drop_write;
3139
	error = vfs_link(old_path.dentry, new_path.dentry->d_inode, new_dentry);
3140
out_drop_write:
3141
	mnt_drop_write(new_path.mnt);
3142
out_dput:
3143
	dput(new_dentry);
3144 3145
	mutex_unlock(&new_path.dentry->d_inode->i_mutex);
	path_put(&new_path);
Linus Torvalds's avatar
Linus Torvalds committed
3146
out:
3147
	path_put(&old_path);
Linus Torvalds's avatar
Linus Torvalds committed
3148 3149 3150 3151

	return error;
}

3152
SYSCALL_DEFINE2(link, const char __user *, oldname, const char __user *, newname)
3153
{
3154
	return sys_linkat(AT_FDCWD, oldname, AT_FDCWD, newname, 0);
3155 3156
}

Linus Torvalds's avatar
Linus Torvalds committed
3157 3158 3159 3160 3161 3162 3163
/*
 * 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
3164
 *	   sb->s_vfs_rename_mutex. We might be more accurate, but that's another
Linus Torvalds's avatar
Linus Torvalds committed
3165 3166
 *	   story.
 *	c) we have to lock _three_ objects - parents and victim (if it exists).
3167
 *	   And that - after we got ->i_mutex on parents (until then we don't know
Linus Torvalds's avatar
Linus Torvalds committed
3168 3169
 *	   whether the target exists).  Solution: try to be smart with locking
 *	   order for inodes.  We rely on the fact that tree topology may change
3170
 *	   only under ->s_vfs_rename_mutex _and_ that parent of the object we
Linus Torvalds's avatar
Linus Torvalds committed
3171 3172 3173
 *	   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,
3174
 *	   lock child" and rename is under ->s_vfs_rename_mutex.
Linus Torvalds's avatar
Linus Torvalds committed
3175 3176 3177
 *	   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.
3178
 *	d) conversion from fhandle to dentry may come in the wrong moment - when
3179
 *	   we are removing the target. Solution: we will have to grab ->i_mutex
Linus Torvalds's avatar
Linus Torvalds committed
3180
 *	   in the fhandle_to_dentry code. [FIXME - current nfsfh.c relies on
3181
 *	   ->i_mutex on parents, which works but leads to some truly excessive
Linus Torvalds's avatar
Linus Torvalds committed
3182 3183
 *	   locking].
 */
3184 3185
static int vfs_rename_dir(struct inode *old_dir, struct dentry *old_dentry,
			  struct inode *new_dir, struct dentry *new_dentry)
Linus Torvalds's avatar
Linus Torvalds committed
3186 3187
{
	int error = 0;
Sage Weil's avatar
Sage Weil committed
3188
	struct inode *target = new_dentry->d_inode;
3189
	unsigned max_links = new_dir->i_sb->s_max_links;
Linus Torvalds's avatar
Linus Torvalds committed
3190 3191 3192 3193 3194 3195

	/*
	 * If we are going to change the parent - check write permissions,
	 * we'll need to flip '..'.
	 */
	if (new_dir != old_dir) {
3196
		error = inode_permission(old_dentry->d_inode, MAY_WRITE);
Linus Torvalds's avatar
Linus Torvalds committed
3197 3198 3199 3200 3201 3202 3203 3204
		if (error)
			return error;
	}

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

3205
	dget(new_dentry);
3206
	if (target)
3207
		mutex_lock(&target->i_mutex);
Sage Weil's avatar
Sage Weil committed
3208 3209 3210 3211 3212

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

3213 3214 3215 3216 3217
	error = -EMLINK;
	if (max_links && !target && new_dir != old_dir &&
	    new_dir->i_nlink >= max_links)
		goto out;

3218 3219
	if (target)
		shrink_dcache_parent(new_dentry);
Sage Weil's avatar
Sage Weil committed
3220 3221 3222 3223
	error = old_dir->i_op->rename(old_dir, old_dentry, new_dir, new_dentry);
	if (error)
		goto out;

Linus Torvalds's avatar
Linus Torvalds committed
3224
	if (target) {
Sage Weil's avatar
Sage Weil committed
3225 3226
		target->i_flags |= S_DEAD;
		dont_mount(new_dentry);
Linus Torvalds's avatar
Linus Torvalds committed
3227
	}
Sage Weil's avatar
Sage Weil committed
3228 3229 3230
out:
	if (target)
		mutex_unlock(&target->i_mutex);
3231
	dput(new_dentry);
3232
	if (!error)
3233 3234
		if (!(old_dir->i_sb->s_type->fs_flags & FS_RENAME_DOES_D_MOVE))
			d_move(old_dentry,new_dentry);
Linus Torvalds's avatar
Linus Torvalds committed
3235 3236 3237
	return error;
}

3238 3239
static int vfs_rename_other(struct inode *old_dir, struct dentry *old_dentry,
			    struct inode *new_dir, struct dentry *new_dentry)
Linus Torvalds's avatar
Linus Torvalds committed
3240
{
Sage Weil's avatar
Sage Weil committed
3241
	struct inode *target = new_dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
3242 3243 3244 3245 3246 3247 3248 3249
	int error;

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

	dget(new_dentry);
	if (target)
3250
		mutex_lock(&target->i_mutex);
Sage Weil's avatar
Sage Weil committed
3251 3252

	error = -EBUSY;
Linus Torvalds's avatar
Linus Torvalds committed
3253
	if (d_mountpoint(old_dentry)||d_mountpoint(new_dentry))
Sage Weil's avatar
Sage Weil committed
3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264
		goto out;

	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:
Linus Torvalds's avatar
Linus Torvalds committed
3265
	if (target)
3266
		mutex_unlock(&target->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
3267 3268 3269 3270 3271 3272 3273 3274 3275
	dput(new_dentry);
	return error;
}

int vfs_rename(struct inode *old_dir, struct dentry *old_dentry,
	       struct inode *new_dir, struct dentry *new_dentry)
{
	int error;
	int is_dir = S_ISDIR(old_dentry->d_inode->i_mode);
3276
	const unsigned char *old_name;
Linus Torvalds's avatar
Linus Torvalds committed
3277 3278 3279 3280 3281 3282 3283 3284 3285

	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)
3286
		error = may_create(new_dir, new_dentry);
Linus Torvalds's avatar
Linus Torvalds committed
3287 3288 3289 3290 3291
	else
		error = may_delete(new_dir, new_dentry, is_dir);
	if (error)
		return error;

Al Viro's avatar
Al Viro committed
3292
	if (!old_dir->i_op->rename)
Linus Torvalds's avatar
Linus Torvalds committed
3293 3294
		return -EPERM;

Robert Love's avatar
Robert Love committed
3295 3296
	old_name = fsnotify_oldname_init(old_dentry->d_name.name);

Linus Torvalds's avatar
Linus Torvalds committed
3297 3298 3299 3300
	if (is_dir)
		error = vfs_rename_dir(old_dir,old_dentry,new_dir,new_dentry);
	else
		error = vfs_rename_other(old_dir,old_dentry,new_dir,new_dentry);
3301 3302
	if (!error)
		fsnotify_move(old_dir, new_dir, old_name, is_dir,
3303
			      new_dentry->d_inode, old_dentry);
Robert Love's avatar
Robert Love committed
3304 3305
	fsnotify_oldname_free(old_name);

Linus Torvalds's avatar
Linus Torvalds committed
3306 3307 3308
	return error;
}

3309 3310
SYSCALL_DEFINE4(renameat, int, olddfd, const char __user *, oldname,
		int, newdfd, const char __user *, newname)
Linus Torvalds's avatar
Linus Torvalds committed
3311
{
3312 3313 3314
	struct dentry *old_dir, *new_dir;
	struct dentry *old_dentry, *new_dentry;
	struct dentry *trap;
Linus Torvalds's avatar
Linus Torvalds committed
3315
	struct nameidata oldnd, newnd;
3316 3317 3318
	char *from;
	char *to;
	int error;
Linus Torvalds's avatar
Linus Torvalds committed
3319

3320
	error = user_path_parent(olddfd, oldname, &oldnd, &from);
Linus Torvalds's avatar
Linus Torvalds committed
3321 3322 3323
	if (error)
		goto exit;

3324
	error = user_path_parent(newdfd, newname, &newnd, &to);
Linus Torvalds's avatar
Linus Torvalds committed
3325 3326 3327 3328
	if (error)
		goto exit1;

	error = -EXDEV;
3329
	if (oldnd.path.mnt != newnd.path.mnt)
Linus Torvalds's avatar
Linus Torvalds committed
3330 3331
		goto exit2;

3332
	old_dir = oldnd.path.dentry;
Linus Torvalds's avatar
Linus Torvalds committed
3333 3334 3335 3336
	error = -EBUSY;
	if (oldnd.last_type != LAST_NORM)
		goto exit2;

3337
	new_dir = newnd.path.dentry;
Linus Torvalds's avatar
Linus Torvalds committed
3338 3339 3340
	if (newnd.last_type != LAST_NORM)
		goto exit2;

3341 3342
	oldnd.flags &= ~LOOKUP_PARENT;
	newnd.flags &= ~LOOKUP_PARENT;
3343
	newnd.flags |= LOOKUP_RENAME_TARGET;
3344

Linus Torvalds's avatar
Linus Torvalds committed
3345 3346
	trap = lock_rename(new_dir, old_dir);

3347
	old_dentry = lookup_hash(&oldnd);
Linus Torvalds's avatar
Linus Torvalds committed
3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366
	error = PTR_ERR(old_dentry);
	if (IS_ERR(old_dentry))
		goto exit3;
	/* source must exist */
	error = -ENOENT;
	if (!old_dentry->d_inode)
		goto exit4;
	/* unless the source is a directory trailing slashes give -ENOTDIR */
	if (!S_ISDIR(old_dentry->d_inode->i_mode)) {
		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;
3367
	new_dentry = lookup_hash(&newnd);
Linus Torvalds's avatar
Linus Torvalds committed
3368 3369 3370 3371 3372 3373 3374 3375
	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;

3376 3377 3378
	error = mnt_want_write(oldnd.path.mnt);
	if (error)
		goto exit5;
3379 3380 3381 3382
	error = security_path_rename(&oldnd.path, old_dentry,
				     &newnd.path, new_dentry);
	if (error)
		goto exit6;
Linus Torvalds's avatar
Linus Torvalds committed
3383 3384
	error = vfs_rename(old_dir->d_inode, old_dentry,
				   new_dir->d_inode, new_dentry);
3385
exit6:
3386
	mnt_drop_write(oldnd.path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
3387 3388 3389 3390 3391 3392 3393
exit5:
	dput(new_dentry);
exit4:
	dput(old_dentry);
exit3:
	unlock_rename(new_dir, old_dir);
exit2:
Jan Blunck's avatar
Jan Blunck committed
3394
	path_put(&newnd.path);
3395
	putname(to);
Linus Torvalds's avatar
Linus Torvalds committed
3396
exit1:
Jan Blunck's avatar
Jan Blunck committed
3397
	path_put(&oldnd.path);
Linus Torvalds's avatar
Linus Torvalds committed
3398
	putname(from);
3399
exit:
Linus Torvalds's avatar
Linus Torvalds committed
3400 3401 3402
	return error;
}

3403
SYSCALL_DEFINE2(rename, const char __user *, oldname, const char __user *, newname)
3404 3405 3406 3407
{
	return sys_renameat(AT_FDCWD, oldname, AT_FDCWD, newname);
}

Linus Torvalds's avatar
Linus Torvalds committed
3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432
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;
3433
	void *cookie;
3434
	int res;
3435

Linus Torvalds's avatar
Linus Torvalds committed
3436
	nd.depth = 0;
3437
	cookie = dentry->d_inode->i_op->follow_link(dentry, &nd);
3438 3439 3440 3441 3442 3443 3444
	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;
Linus Torvalds's avatar
Linus Torvalds committed
3445 3446 3447 3448 3449 3450 3451 3452 3453 3454
}

int vfs_follow_link(struct nameidata *nd, const char *link)
{
	return __vfs_follow_link(nd, link);
}

/* get the link contents into pagecache */
static char *page_getlink(struct dentry * dentry, struct page **ppage)
{
3455 3456
	char *kaddr;
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
3457
	struct address_space *mapping = dentry->d_inode->i_mapping;
3458
	page = read_mapping_page(mapping, 0, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
3459
	if (IS_ERR(page))
3460
		return (char*)page;
Linus Torvalds's avatar
Linus Torvalds committed
3461
	*ppage = page;
3462 3463 3464
	kaddr = kmap(page);
	nd_terminate_link(kaddr, dentry->d_inode->i_size, PAGE_SIZE - 1);
	return kaddr;
Linus Torvalds's avatar
Linus Torvalds committed
3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478
}

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;
}

3479
void *page_follow_link_light(struct dentry *dentry, struct nameidata *nd)
Linus Torvalds's avatar
Linus Torvalds committed
3480
{
3481
	struct page *page = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
3482
	nd_set_link(nd, page_getlink(dentry, &page));
3483
	return page;
Linus Torvalds's avatar
Linus Torvalds committed
3484 3485
}

3486
void page_put_link(struct dentry *dentry, struct nameidata *nd, void *cookie)
Linus Torvalds's avatar
Linus Torvalds committed
3487
{
3488 3489 3490
	struct page *page = cookie;

	if (page) {
Linus Torvalds's avatar
Linus Torvalds committed
3491 3492 3493 3494 3495
		kunmap(page);
		page_cache_release(page);
	}
}

3496 3497 3498 3499
/*
 * 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)
Linus Torvalds's avatar
Linus Torvalds committed
3500 3501
{
	struct address_space *mapping = inode->i_mapping;
3502
	struct page *page;
3503
	void *fsdata;
3504
	int err;
Linus Torvalds's avatar
Linus Torvalds committed
3505
	char *kaddr;
3506 3507 3508
	unsigned int flags = AOP_FLAG_UNINTERRUPTIBLE;
	if (nofs)
		flags |= AOP_FLAG_NOFS;
Linus Torvalds's avatar
Linus Torvalds committed
3509

3510
retry:
3511
	err = pagecache_write_begin(NULL, mapping, 0, len-1,
3512
				flags, &page, &fsdata);
Linus Torvalds's avatar
Linus Torvalds committed
3513
	if (err)
3514 3515
		goto fail;

3516
	kaddr = kmap_atomic(page);
Linus Torvalds's avatar
Linus Torvalds committed
3517
	memcpy(kaddr, symname, len-1);
3518
	kunmap_atomic(kaddr);
3519 3520 3521

	err = pagecache_write_end(NULL, mapping, 0, len-1, len-1,
							page, fsdata);
Linus Torvalds's avatar
Linus Torvalds committed
3522 3523
	if (err < 0)
		goto fail;
3524 3525 3526
	if (err < len-1)
		goto retry;

Linus Torvalds's avatar
Linus Torvalds committed
3527 3528 3529 3530 3531 3532
	mark_inode_dirty(inode);
	return 0;
fail:
	return err;
}

3533 3534 3535
int page_symlink(struct inode *inode, const char *symname, int len)
{
	return __page_symlink(inode, symname, len,
3536
			!(mapping_gfp_mask(inode->i_mapping) & __GFP_FS));
3537 3538
}

3539
const struct inode_operations page_symlink_inode_operations = {
Linus Torvalds's avatar
Linus Torvalds committed
3540 3541 3542 3543 3544
	.readlink	= generic_readlink,
	.follow_link	= page_follow_link_light,
	.put_link	= page_put_link,
};

3545
EXPORT_SYMBOL(user_path_at);
3546
EXPORT_SYMBOL(follow_down_one);
Linus Torvalds's avatar
Linus Torvalds committed
3547 3548 3549 3550 3551 3552 3553 3554 3555
EXPORT_SYMBOL(follow_down);
EXPORT_SYMBOL(follow_up);
EXPORT_SYMBOL(get_write_access); /* binfmt_aout */
EXPORT_SYMBOL(getname);
EXPORT_SYMBOL(lock_rename);
EXPORT_SYMBOL(lookup_one_len);
EXPORT_SYMBOL(page_follow_link_light);
EXPORT_SYMBOL(page_put_link);
EXPORT_SYMBOL(page_readlink);
3556
EXPORT_SYMBOL(__page_symlink);
Linus Torvalds's avatar
Linus Torvalds committed
3557 3558
EXPORT_SYMBOL(page_symlink);
EXPORT_SYMBOL(page_symlink_inode_operations);
Al Viro's avatar
Al Viro committed
3559
EXPORT_SYMBOL(kern_path);
3560
EXPORT_SYMBOL(vfs_path_lookup);
3561
EXPORT_SYMBOL(inode_permission);
Linus Torvalds's avatar
Linus Torvalds committed
3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575
EXPORT_SYMBOL(unlock_rename);
EXPORT_SYMBOL(vfs_create);
EXPORT_SYMBOL(vfs_follow_link);
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);