open.c 27.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9
/*
 *  linux/fs/open.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

#include <linux/string.h>
#include <linux/mm.h>
#include <linux/file.h>
A
Al Viro 已提交
10
#include <linux/fdtable.h>
R
Robert Love 已提交
11
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
12 13 14 15
#include <linux/module.h>
#include <linux/tty.h>
#include <linux/namei.h>
#include <linux/backing-dev.h>
16
#include <linux/capability.h>
17
#include <linux/securebits.h>
L
Linus Torvalds 已提交
18 19
#include <linux/security.h>
#include <linux/mount.h>
20
#include <linux/fcntl.h>
21
#include <linux/slab.h>
L
Linus Torvalds 已提交
22 23
#include <asm/uaccess.h>
#include <linux/fs.h>
24
#include <linux/personality.h>
L
Linus Torvalds 已提交
25 26
#include <linux/pagemap.h>
#include <linux/syscalls.h>
27
#include <linux/rcupdate.h>
28
#include <linux/audit.h>
29
#include <linux/falloc.h>
30
#include <linux/fs_struct.h>
31
#include <linux/ima.h>
32
#include <linux/dnotify.h>
A
Al Viro 已提交
33
#include <linux/compat.h>
L
Linus Torvalds 已提交
34

35 36
#include "internal.h"

37 38
int do_truncate(struct dentry *dentry, loff_t length, unsigned int time_attrs,
	struct file *filp)
L
Linus Torvalds 已提交
39
{
40
	int ret;
L
Linus Torvalds 已提交
41 42 43 44 45 46 47
	struct iattr newattrs;

	/* Not pretty: "inode->i_size" shouldn't really be signed. But it is. */
	if (length < 0)
		return -EINVAL;

	newattrs.ia_size = length;
48
	newattrs.ia_valid = ATTR_SIZE | time_attrs;
49 50 51 52
	if (filp) {
		newattrs.ia_file = filp;
		newattrs.ia_valid |= ATTR_FILE;
	}
L
Linus Torvalds 已提交
53

54 55 56 57
	/* Remove suid, sgid, and file capabilities on truncate too */
	ret = dentry_needs_remove_privs(dentry);
	if (ret < 0)
		return ret;
58 59
	if (ret)
		newattrs.ia_valid |= ret | ATTR_FORCE;
60

A
Al Viro 已提交
61
	inode_lock(dentry->d_inode);
62 63
	/* Note any delegations or leases have already been broken: */
	ret = notify_change(dentry, &newattrs, NULL);
A
Al Viro 已提交
64
	inode_unlock(dentry->d_inode);
65
	return ret;
L
Linus Torvalds 已提交
66 67
}

A
Al Viro 已提交
68
long vfs_truncate(const struct path *path, loff_t length)
L
Linus Torvalds 已提交
69
{
70
	struct inode *inode;
71
	long error;
L
Linus Torvalds 已提交
72

73
	inode = path->dentry->d_inode;
L
Linus Torvalds 已提交
74 75 76

	/* For directories it's -EISDIR, for other non-regulars - -EINVAL */
	if (S_ISDIR(inode->i_mode))
77
		return -EISDIR;
L
Linus Torvalds 已提交
78
	if (!S_ISREG(inode->i_mode))
79
		return -EINVAL;
L
Linus Torvalds 已提交
80

81
	error = mnt_want_write(path->mnt);
L
Linus Torvalds 已提交
82
	if (error)
83
		goto out;
L
Linus Torvalds 已提交
84

85
	error = inode_permission(inode, MAY_WRITE);
86 87
	if (error)
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
88 89

	error = -EPERM;
90
	if (IS_APPEND(inode))
91
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
92

93
	error = get_write_access(inode);
L
Linus Torvalds 已提交
94
	if (error)
95
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
96

97 98 99 100
	/*
	 * Make sure that there are no leases.  get_write_access() protects
	 * against the truncate racing with a lease-granting setlease().
	 */
101
	error = break_lease(inode, O_WRONLY);
L
Linus Torvalds 已提交
102
	if (error)
103
		goto put_write_and_out;
L
Linus Torvalds 已提交
104 105

	error = locks_verify_truncate(inode, NULL, length);
106
	if (!error)
107
		error = security_path_truncate(path);
108
	if (!error)
109
		error = do_truncate(path->dentry, length, 0, NULL);
L
Linus Torvalds 已提交
110

111 112
put_write_and_out:
	put_write_access(inode);
113
mnt_drop_write_and_out:
114
	mnt_drop_write(path->mnt);
L
Linus Torvalds 已提交
115 116 117
out:
	return error;
}
118 119 120 121
EXPORT_SYMBOL_GPL(vfs_truncate);

static long do_sys_truncate(const char __user *pathname, loff_t length)
{
122
	unsigned int lookup_flags = LOOKUP_FOLLOW;
123 124 125 126 127 128
	struct path path;
	int error;

	if (length < 0)	/* sorry, but loff_t says... */
		return -EINVAL;

129 130
retry:
	error = user_path_at(AT_FDCWD, pathname, lookup_flags, &path);
131 132 133 134
	if (!error) {
		error = vfs_truncate(&path, length);
		path_put(&path);
	}
135 136 137 138
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
139 140
	return error;
}
L
Linus Torvalds 已提交
141

142
SYSCALL_DEFINE2(truncate, const char __user *, path, long, length)
L
Linus Torvalds 已提交
143
{
144
	return do_sys_truncate(path, length);
L
Linus Torvalds 已提交
145 146
}

A
Al Viro 已提交
147 148 149 150 151 152 153
#ifdef CONFIG_COMPAT
COMPAT_SYSCALL_DEFINE2(truncate, const char __user *, path, compat_off_t, length)
{
	return do_sys_truncate(path, length);
}
#endif

154
static long do_sys_ftruncate(unsigned int fd, loff_t length, int small)
L
Linus Torvalds 已提交
155
{
A
Al Viro 已提交
156
	struct inode *inode;
L
Linus Torvalds 已提交
157
	struct dentry *dentry;
158
	struct fd f;
L
Linus Torvalds 已提交
159 160 161 162 163 164
	int error;

	error = -EINVAL;
	if (length < 0)
		goto out;
	error = -EBADF;
165 166
	f = fdget(fd);
	if (!f.file)
L
Linus Torvalds 已提交
167 168 169
		goto out;

	/* explicitly opened as large or we are on 64-bit box */
170
	if (f.file->f_flags & O_LARGEFILE)
L
Linus Torvalds 已提交
171 172
		small = 0;

173
	dentry = f.file->f_path.dentry;
L
Linus Torvalds 已提交
174 175
	inode = dentry->d_inode;
	error = -EINVAL;
176
	if (!S_ISREG(inode->i_mode) || !(f.file->f_mode & FMODE_WRITE))
L
Linus Torvalds 已提交
177 178 179 180 181 182 183 184 185 186 187
		goto out_putf;

	error = -EINVAL;
	/* Cannot ftruncate over 2^31 bytes without large file support */
	if (small && length > MAX_NON_LFS)
		goto out_putf;

	error = -EPERM;
	if (IS_APPEND(inode))
		goto out_putf;

188
	sb_start_write(inode->i_sb);
189
	error = locks_verify_truncate(inode, f.file, length);
190
	if (!error)
191
		error = security_path_truncate(&f.file->f_path);
L
Linus Torvalds 已提交
192
	if (!error)
193
		error = do_truncate(dentry, length, ATTR_MTIME|ATTR_CTIME, f.file);
194
	sb_end_write(inode->i_sb);
L
Linus Torvalds 已提交
195
out_putf:
196
	fdput(f);
L
Linus Torvalds 已提交
197 198 199 200
out:
	return error;
}

201
SYSCALL_DEFINE2(ftruncate, unsigned int, fd, unsigned long, length)
L
Linus Torvalds 已提交
202
{
203
	return do_sys_ftruncate(fd, length, 1);
L
Linus Torvalds 已提交
204 205
}

A
Al Viro 已提交
206 207 208 209 210 211 212
#ifdef CONFIG_COMPAT
COMPAT_SYSCALL_DEFINE2(ftruncate, unsigned int, fd, compat_ulong_t, length)
{
	return do_sys_ftruncate(fd, length, 1);
}
#endif

L
Linus Torvalds 已提交
213 214
/* LFS versions of truncate are only needed on 32 bit machines */
#if BITS_PER_LONG == 32
215
SYSCALL_DEFINE2(truncate64, const char __user *, path, loff_t, length)
L
Linus Torvalds 已提交
216 217 218 219
{
	return do_sys_truncate(path, length);
}

220
SYSCALL_DEFINE2(ftruncate64, unsigned int, fd, loff_t, length)
L
Linus Torvalds 已提交
221
{
222
	return do_sys_ftruncate(fd, length, 0);
L
Linus Torvalds 已提交
223
}
224
#endif /* BITS_PER_LONG == 32 */
L
Linus Torvalds 已提交
225

226

227
int vfs_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
228
{
A
Al Viro 已提交
229
	struct inode *inode = file_inode(file);
230
	long ret;
231 232

	if (offset < 0 || len <= 0)
233
		return -EINVAL;
234 235

	/* Return error if mode is not supported */
236
	if (mode & ~FALLOC_FL_SUPPORTED_MASK)
237 238 239 240 241
		return -EOPNOTSUPP;

	/* Punch hole and zero range are mutually exclusive */
	if ((mode & (FALLOC_FL_PUNCH_HOLE | FALLOC_FL_ZERO_RANGE)) ==
	    (FALLOC_FL_PUNCH_HOLE | FALLOC_FL_ZERO_RANGE))
J
Josef Bacik 已提交
242 243 244 245 246
		return -EOPNOTSUPP;

	/* Punch hole must have keep size set */
	if ((mode & FALLOC_FL_PUNCH_HOLE) &&
	    !(mode & FALLOC_FL_KEEP_SIZE))
247
		return -EOPNOTSUPP;
248

249 250 251 252 253
	/* Collapse range should only be used exclusively. */
	if ((mode & FALLOC_FL_COLLAPSE_RANGE) &&
	    (mode & ~FALLOC_FL_COLLAPSE_RANGE))
		return -EINVAL;

254 255 256 257 258
	/* Insert range should only be used exclusively. */
	if ((mode & FALLOC_FL_INSERT_RANGE) &&
	    (mode & ~FALLOC_FL_INSERT_RANGE))
		return -EINVAL;

259
	if (!(file->f_mode & FMODE_WRITE))
260
		return -EBADF;
261

262
	/*
263
	 * We can only allow pure fallocate on append only files
264
	 */
265
	if ((mode & ~FALLOC_FL_KEEP_SIZE) && IS_APPEND(inode))
266 267 268 269 270
		return -EPERM;

	if (IS_IMMUTABLE(inode))
		return -EPERM;

271
	/*
272
	 * We cannot allow any fallocate operation on an active swapfile
273 274
	 */
	if (IS_SWAPFILE(inode))
275
		return -ETXTBSY;
276

277 278 279 280 281 282
	/*
	 * Revalidate the write permissions, in case security policy has
	 * changed since the files were opened.
	 */
	ret = security_file_permission(file, MAY_WRITE);
	if (ret)
283
		return ret;
284 285

	if (S_ISFIFO(inode->i_mode))
286
		return -ESPIPE;
287 288 289 290 291 292

	/*
	 * Let individual file system decide if it supports preallocation
	 * for directories or not.
	 */
	if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode))
293
		return -ENODEV;
294 295 296

	/* Check for wrap through zero too */
	if (((offset + len) > inode->i_sb->s_maxbytes) || ((offset + len) < 0))
297
		return -EFBIG;
298

299
	if (!file->f_op->fallocate)
300
		return -EOPNOTSUPP;
301

302 303
	sb_start_write(inode->i_sb);
	ret = file->f_op->fallocate(file, mode, offset, len);
304 305 306 307 308 309 310 311 312 313 314

	/*
	 * Create inotify and fanotify events.
	 *
	 * To keep the logic simple always create events if fallocate succeeds.
	 * This implies that events are even created if the file size remains
	 * unchanged, e.g. when using flag FALLOC_FL_KEEP_SIZE.
	 */
	if (ret == 0)
		fsnotify_modify(file);

315 316
	sb_end_write(inode->i_sb);
	return ret;
317
}
318
EXPORT_SYMBOL_GPL(vfs_fallocate);
319

320
SYSCALL_DEFINE4(fallocate, int, fd, int, mode, loff_t, offset, loff_t, len)
321
{
322
	struct fd f = fdget(fd);
323 324
	int error = -EBADF;

325
	if (f.file) {
326
		error = vfs_fallocate(f.file, mode, offset, len);
327
		fdput(f);
328 329
	}
	return error;
330
}
331

L
Linus Torvalds 已提交
332 333 334 335 336
/*
 * access() needs to use the real uid/gid, not the effective uid/gid.
 * We do this by temporarily clearing all FS-related capabilities and
 * switching the fsuid/fsgid around to the real ones.
 */
337
SYSCALL_DEFINE3(faccessat, int, dfd, const char __user *, filename, int, mode)
L
Linus Torvalds 已提交
338
{
D
David Howells 已提交
339 340
	const struct cred *old_cred;
	struct cred *override_cred;
341
	struct path path;
342
	struct inode *inode;
L
Linus Torvalds 已提交
343
	int res;
344
	unsigned int lookup_flags = LOOKUP_FOLLOW;
L
Linus Torvalds 已提交
345 346 347 348

	if (mode & ~S_IRWXO)	/* where's F_OK, X_OK, W_OK, R_OK? */
		return -EINVAL;

D
David Howells 已提交
349 350 351
	override_cred = prepare_creds();
	if (!override_cred)
		return -ENOMEM;
L
Linus Torvalds 已提交
352

D
David Howells 已提交
353 354
	override_cred->fsuid = override_cred->uid;
	override_cred->fsgid = override_cred->gid;
L
Linus Torvalds 已提交
355

356
	if (!issecure(SECURE_NO_SETUID_FIXUP)) {
D
David Howells 已提交
357
		/* Clear the capabilities if we switch to a non-root user */
358 359
		kuid_t root_uid = make_kuid(override_cred->user_ns, 0);
		if (!uid_eq(override_cred->uid, root_uid))
D
David Howells 已提交
360
			cap_clear(override_cred->cap_effective);
361
		else
D
David Howells 已提交
362 363
			override_cred->cap_effective =
				override_cred->cap_permitted;
364
	}
L
Linus Torvalds 已提交
365

D
David Howells 已提交
366
	old_cred = override_creds(override_cred);
367 368
retry:
	res = user_path_at(dfd, filename, lookup_flags, &path);
369 370 371
	if (res)
		goto out;

372
	inode = d_backing_inode(path.dentry);
373 374

	if ((mode & MAY_EXEC) && S_ISREG(inode->i_mode)) {
375 376 377 378 379
		/*
		 * MAY_EXEC on regular files is denied if the fs is mounted
		 * with the "noexec" flag.
		 */
		res = -EACCES;
380
		if (path_noexec(&path))
381 382 383
			goto out_path_release;
	}

384
	res = inode_permission(inode, mode | MAY_ACCESS);
385
	/* SuS v2 requires we report a read only fs too */
386
	if (res || !(mode & S_IWOTH) || special_file(inode->i_mode))
387
		goto out_path_release;
388 389 390 391 392 393 394 395 396 397
	/*
	 * This is a rare case where using __mnt_is_readonly()
	 * is OK without a mnt_want/drop_write() pair.  Since
	 * no actual write to the fs is performed here, we do
	 * not need to telegraph to that to anyone.
	 *
	 * By doing this, we accept that this access is
	 * inherently racy and know that the fs may change
	 * state before we even see this result.
	 */
398
	if (__mnt_is_readonly(path.mnt))
399
		res = -EROFS;
L
Linus Torvalds 已提交
400

401
out_path_release:
402
	path_put(&path);
403 404 405 406
	if (retry_estale(res, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
407
out:
D
David Howells 已提交
408 409
	revert_creds(old_cred);
	put_cred(override_cred);
L
Linus Torvalds 已提交
410 411 412
	return res;
}

413
SYSCALL_DEFINE2(access, const char __user *, filename, int, mode)
414 415 416 417
{
	return sys_faccessat(AT_FDCWD, filename, mode);
}

418
SYSCALL_DEFINE1(chdir, const char __user *, filename)
L
Linus Torvalds 已提交
419
{
420
	struct path path;
L
Linus Torvalds 已提交
421
	int error;
422 423 424
	unsigned int lookup_flags = LOOKUP_FOLLOW | LOOKUP_DIRECTORY;
retry:
	error = user_path_at(AT_FDCWD, filename, lookup_flags, &path);
L
Linus Torvalds 已提交
425 426 427
	if (error)
		goto out;

E
Eric Paris 已提交
428
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
429 430 431
	if (error)
		goto dput_and_out;

432
	set_fs_pwd(current->fs, &path);
L
Linus Torvalds 已提交
433 434

dput_and_out:
435
	path_put(&path);
436 437 438 439
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
L
Linus Torvalds 已提交
440 441 442 443
out:
	return error;
}

444
SYSCALL_DEFINE1(fchdir, unsigned int, fd)
L
Linus Torvalds 已提交
445
{
446
	struct fd f = fdget_raw(fd);
L
Linus Torvalds 已提交
447
	struct inode *inode;
448
	int error = -EBADF;
L
Linus Torvalds 已提交
449 450

	error = -EBADF;
451
	if (!f.file)
L
Linus Torvalds 已提交
452 453
		goto out;

A
Al Viro 已提交
454
	inode = file_inode(f.file);
L
Linus Torvalds 已提交
455 456 457 458 459

	error = -ENOTDIR;
	if (!S_ISDIR(inode->i_mode))
		goto out_putf;

E
Eric Paris 已提交
460
	error = inode_permission(inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
461
	if (!error)
462
		set_fs_pwd(current->fs, &f.file->f_path);
L
Linus Torvalds 已提交
463
out_putf:
464
	fdput(f);
L
Linus Torvalds 已提交
465 466 467 468
out:
	return error;
}

469
SYSCALL_DEFINE1(chroot, const char __user *, filename)
L
Linus Torvalds 已提交
470
{
471
	struct path path;
L
Linus Torvalds 已提交
472
	int error;
473 474 475
	unsigned int lookup_flags = LOOKUP_FOLLOW | LOOKUP_DIRECTORY;
retry:
	error = user_path_at(AT_FDCWD, filename, lookup_flags, &path);
L
Linus Torvalds 已提交
476 477 478
	if (error)
		goto out;

E
Eric Paris 已提交
479
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
480 481 482 483
	if (error)
		goto dput_and_out;

	error = -EPERM;
484
	if (!ns_capable(current_user_ns(), CAP_SYS_CHROOT))
L
Linus Torvalds 已提交
485
		goto dput_and_out;
T
Tetsuo Handa 已提交
486 487 488
	error = security_path_chroot(&path);
	if (error)
		goto dput_and_out;
L
Linus Torvalds 已提交
489

490
	set_fs_root(current->fs, &path);
L
Linus Torvalds 已提交
491 492
	error = 0;
dput_and_out:
493
	path_put(&path);
494 495 496 497
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
L
Linus Torvalds 已提交
498 499 500 501
out:
	return error;
}

502
static int chmod_common(const struct path *path, umode_t mode)
L
Linus Torvalds 已提交
503
{
504
	struct inode *inode = path->dentry->d_inode;
505
	struct inode *delegated_inode = NULL;
L
Linus Torvalds 已提交
506
	struct iattr newattrs;
507
	int error;
L
Linus Torvalds 已提交
508

509 510 511
	error = mnt_want_write(path->mnt);
	if (error)
		return error;
512
retry_deleg:
A
Al Viro 已提交
513
	inode_lock(inode);
514
	error = security_path_chmod(path, mode);
515
	if (error)
516
		goto out_unlock;
L
Linus Torvalds 已提交
517 518
	newattrs.ia_mode = (mode & S_IALLUGO) | (inode->i_mode & ~S_IALLUGO);
	newattrs.ia_valid = ATTR_MODE | ATTR_CTIME;
519
	error = notify_change(path->dentry, &newattrs, &delegated_inode);
520
out_unlock:
A
Al Viro 已提交
521
	inode_unlock(inode);
522 523 524 525 526
	if (delegated_inode) {
		error = break_deleg_wait(&delegated_inode);
		if (!error)
			goto retry_deleg;
	}
527 528 529 530
	mnt_drop_write(path->mnt);
	return error;
}

531
SYSCALL_DEFINE2(fchmod, unsigned int, fd, umode_t, mode)
532
{
A
Al Viro 已提交
533
	struct fd f = fdget(fd);
534 535
	int err = -EBADF;

A
Al Viro 已提交
536
	if (f.file) {
A
Al Viro 已提交
537
		audit_file(f.file);
A
Al Viro 已提交
538 539
		err = chmod_common(&f.file->f_path, mode);
		fdput(f);
540
	}
L
Linus Torvalds 已提交
541 542 543
	return err;
}

544
SYSCALL_DEFINE3(fchmodat, int, dfd, const char __user *, filename, umode_t, mode)
L
Linus Torvalds 已提交
545
{
546
	struct path path;
L
Linus Torvalds 已提交
547
	int error;
548 549 550
	unsigned int lookup_flags = LOOKUP_FOLLOW;
retry:
	error = user_path_at(dfd, filename, lookup_flags, &path);
551 552 553
	if (!error) {
		error = chmod_common(&path, mode);
		path_put(&path);
554 555 556 557
		if (retry_estale(error, lookup_flags)) {
			lookup_flags |= LOOKUP_REVAL;
			goto retry;
		}
558
	}
L
Linus Torvalds 已提交
559 560 561
	return error;
}

562
SYSCALL_DEFINE2(chmod, const char __user *, filename, umode_t, mode)
563 564 565 566
{
	return sys_fchmodat(AT_FDCWD, filename, mode);
}

567
static int chown_common(const struct path *path, uid_t user, gid_t group)
L
Linus Torvalds 已提交
568
{
569
	struct inode *inode = path->dentry->d_inode;
570
	struct inode *delegated_inode = NULL;
L
Linus Torvalds 已提交
571 572
	int error;
	struct iattr newattrs;
573 574 575 576 577
	kuid_t uid;
	kgid_t gid;

	uid = make_kuid(current_user_ns(), user);
	gid = make_kgid(current_user_ns(), group);
L
Linus Torvalds 已提交
578

579
retry_deleg:
L
Linus Torvalds 已提交
580 581
	newattrs.ia_valid =  ATTR_CTIME;
	if (user != (uid_t) -1) {
582 583
		if (!uid_valid(uid))
			return -EINVAL;
L
Linus Torvalds 已提交
584
		newattrs.ia_valid |= ATTR_UID;
585
		newattrs.ia_uid = uid;
L
Linus Torvalds 已提交
586 587
	}
	if (group != (gid_t) -1) {
588 589
		if (!gid_valid(gid))
			return -EINVAL;
L
Linus Torvalds 已提交
590
		newattrs.ia_valid |= ATTR_GID;
591
		newattrs.ia_gid = gid;
L
Linus Torvalds 已提交
592 593
	}
	if (!S_ISDIR(inode->i_mode))
594 595
		newattrs.ia_valid |=
			ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
A
Al Viro 已提交
596
	inode_lock(inode);
597
	error = security_path_chown(path, uid, gid);
598
	if (!error)
599
		error = notify_change(path->dentry, &newattrs, &delegated_inode);
A
Al Viro 已提交
600
	inode_unlock(inode);
601 602 603 604 605
	if (delegated_inode) {
		error = break_deleg_wait(&delegated_inode);
		if (!error)
			goto retry_deleg;
	}
L
Linus Torvalds 已提交
606 607 608
	return error;
}

609 610
SYSCALL_DEFINE5(fchownat, int, dfd, const char __user *, filename, uid_t, user,
		gid_t, group, int, flag)
611
{
612
	struct path path;
613
	int error = -EINVAL;
614
	int lookup_flags;
615

616
	if ((flag & ~(AT_SYMLINK_NOFOLLOW | AT_EMPTY_PATH)) != 0)
617 618
		goto out;

619 620 621
	lookup_flags = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
	if (flag & AT_EMPTY_PATH)
		lookup_flags |= LOOKUP_EMPTY;
622
retry:
623
	error = user_path_at(dfd, filename, lookup_flags, &path);
624 625
	if (error)
		goto out;
626
	error = mnt_want_write(path.mnt);
627 628
	if (error)
		goto out_release;
629
	error = chown_common(&path, user, group);
630
	mnt_drop_write(path.mnt);
631
out_release:
632
	path_put(&path);
633 634 635 636
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
637 638 639 640
out:
	return error;
}

641
SYSCALL_DEFINE3(chown, const char __user *, filename, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
642
{
643 644
	return sys_fchownat(AT_FDCWD, filename, user, group, 0);
}
L
Linus Torvalds 已提交
645

646 647 648 649
SYSCALL_DEFINE3(lchown, const char __user *, filename, uid_t, user, gid_t, group)
{
	return sys_fchownat(AT_FDCWD, filename, user, group,
			    AT_SYMLINK_NOFOLLOW);
L
Linus Torvalds 已提交
650 651
}

652
SYSCALL_DEFINE3(fchown, unsigned int, fd, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
653
{
654
	struct fd f = fdget(fd);
L
Linus Torvalds 已提交
655 656
	int error = -EBADF;

657
	if (!f.file)
658 659
		goto out;

660
	error = mnt_want_write_file(f.file);
661 662
	if (error)
		goto out_fput;
A
Al Viro 已提交
663
	audit_file(f.file);
664 665
	error = chown_common(&f.file->f_path, user, group);
	mnt_drop_write_file(f.file);
666
out_fput:
667
	fdput(f);
668
out:
L
Linus Torvalds 已提交
669 670 671
	return error;
}

M
Miklos Szeredi 已提交
672 673 674 675
int open_check_o_direct(struct file *f)
{
	/* NB: we're sure to have correct a_ops only after f_op->open */
	if (f->f_flags & O_DIRECT) {
M
Matthew Wilcox 已提交
676
		if (!f->f_mapping->a_ops || !f->f_mapping->a_ops->direct_IO)
M
Miklos Szeredi 已提交
677 678 679 680 681
			return -EINVAL;
	}
	return 0;
}

682
static int do_dentry_open(struct file *f,
683
			  struct inode *inode,
684 685
			  int (*open)(struct inode *, struct file *),
			  const struct cred *cred)
L
Linus Torvalds 已提交
686
{
687
	static const struct file_operations empty_fops = {};
L
Linus Torvalds 已提交
688 689
	int error;

A
Al Viro 已提交
690
	f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
691
				FMODE_PREAD | FMODE_PWRITE;
692

693
	path_get(&f->f_path);
694
	f->f_inode = inode;
L
Linus Torvalds 已提交
695 696
	f->f_mapping = inode->i_mapping;

A
Al Viro 已提交
697 698
	if (unlikely(f->f_flags & O_PATH)) {
		f->f_mode = FMODE_PATH;
699
		f->f_op = &empty_fops;
700
		return 0;
701 702
	}

703
	if (f->f_mode & FMODE_WRITE && !special_file(inode->i_mode)) {
704
		error = get_write_access(inode);
A
Al Viro 已提交
705
		if (unlikely(error))
L
Linus Torvalds 已提交
706
			goto cleanup_file;
707
		error = __mnt_want_write(f->f_path.mnt);
A
Al Viro 已提交
708
		if (unlikely(error)) {
709 710 711
			put_write_access(inode);
			goto cleanup_file;
		}
712
		f->f_mode |= FMODE_WRITER;
L
Linus Torvalds 已提交
713 714
	}

715
	/* POSIX.1-2008/SUSv4 Section XSI 2.9.7 */
716
	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode))
717 718
		f->f_mode |= FMODE_ATOMIC_POS;

719
	f->f_op = fops_get(inode->i_fop);
A
Al Viro 已提交
720 721 722 723
	if (unlikely(WARN_ON(!f->f_op))) {
		error = -ENODEV;
		goto cleanup_all;
	}
724

725
	error = security_file_open(f, cred);
726 727 728
	if (error)
		goto cleanup_all;

729 730 731 732
	error = break_lease(inode, f->f_flags);
	if (error)
		goto cleanup_all;

A
Al Viro 已提交
733
	if (!open)
734 735 736
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
L
Linus Torvalds 已提交
737 738 739
		if (error)
			goto cleanup_all;
	}
740 741
	if ((f->f_mode & (FMODE_READ | FMODE_WRITE)) == FMODE_READ)
		i_readcount_inc(inode);
742
	if ((f->f_mode & FMODE_READ) &&
A
Al Viro 已提交
743
	     likely(f->f_op->read || f->f_op->read_iter))
744
		f->f_mode |= FMODE_CAN_READ;
745
	if ((f->f_mode & FMODE_WRITE) &&
A
Al Viro 已提交
746
	     likely(f->f_op->write || f->f_op->write_iter))
747
		f->f_mode |= FMODE_CAN_WRITE;
748

L
Linus Torvalds 已提交
749 750 751 752
	f->f_flags &= ~(O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC);

	file_ra_state_init(&f->f_ra, f->f_mapping->host->i_mapping);

753
	return 0;
L
Linus Torvalds 已提交
754 755 756

cleanup_all:
	fops_put(f->f_op);
757
	if (f->f_mode & FMODE_WRITER) {
L
Linus Torvalds 已提交
758
		put_write_access(inode);
759
		__mnt_drop_write(f->f_path.mnt);
760
	}
L
Linus Torvalds 已提交
761
cleanup_file:
762 763 764
	path_put(&f->f_path);
	f->f_path.mnt = NULL;
	f->f_path.dentry = NULL;
765
	f->f_inode = NULL;
766
	return error;
L
Linus Torvalds 已提交
767 768
}

M
Miklos Szeredi 已提交
769 770
/**
 * finish_open - finish opening a file
771
 * @file: file pointer
M
Miklos Szeredi 已提交
772 773
 * @dentry: pointer to dentry
 * @open: open callback
774
 * @opened: state of open
M
Miklos Szeredi 已提交
775 776 777 778 779
 *
 * This can be used to finish opening a file passed to i_op->atomic_open().
 *
 * If the open callback is set to NULL, then the standard f_op->open()
 * filesystem callback is substituted.
780 781 782 783 784 785 786 787 788
 *
 * NB: the dentry reference is _not_ consumed.  If, for example, the dentry is
 * the return value of d_splice_alias(), then the caller needs to perform dput()
 * on it after finish_open().
 *
 * On successful return @file is a fully instantiated open file.  After this, if
 * an error occurs in ->atomic_open(), it needs to clean up with fput().
 *
 * Returns zero on success or -errno if the open failed.
M
Miklos Szeredi 已提交
789
 */
A
Al Viro 已提交
790 791 792
int finish_open(struct file *file, struct dentry *dentry,
		int (*open)(struct inode *, struct file *),
		int *opened)
M
Miklos Szeredi 已提交
793
{
794
	int error;
A
Al Viro 已提交
795
	BUG_ON(*opened & FILE_OPENED); /* once it's opened, it's opened */
M
Miklos Szeredi 已提交
796

797
	file->f_path.dentry = dentry;
798 799
	error = do_dentry_open(file, d_backing_inode(dentry), open,
			       current_cred());
800
	if (!error)
801
		*opened |= FILE_OPENED;
M
Miklos Szeredi 已提交
802

803
	return error;
M
Miklos Szeredi 已提交
804 805 806 807 808 809
}
EXPORT_SYMBOL(finish_open);

/**
 * finish_no_open - finish ->atomic_open() without opening the file
 *
810
 * @file: file pointer
M
Miklos Szeredi 已提交
811 812 813
 * @dentry: dentry or NULL (as returned from ->lookup())
 *
 * This can be used to set the result of a successful lookup in ->atomic_open().
814 815 816 817 818 819
 *
 * NB: unlike finish_open() this function does consume the dentry reference and
 * the caller need not dput() it.
 *
 * Returns "1" which must be the return value of ->atomic_open() after having
 * called this function.
M
Miklos Szeredi 已提交
820
 */
A
Al Viro 已提交
821
int finish_no_open(struct file *file, struct dentry *dentry)
M
Miklos Szeredi 已提交
822
{
A
Al Viro 已提交
823
	file->f_path.dentry = dentry;
A
Al Viro 已提交
824
	return 1;
M
Miklos Szeredi 已提交
825 826 827
}
EXPORT_SYMBOL(finish_no_open);

M
Miklos Szeredi 已提交
828 829 830 831 832 833
char *file_path(struct file *filp, char *buf, int buflen)
{
	return d_path(&filp->f_path, buf, buflen);
}
EXPORT_SYMBOL(file_path);

834 835 836 837 838 839 840 841 842
/**
 * vfs_open - open the file at the given path
 * @path: path to open
 * @file: newly allocated file with f_flag initialized
 * @cred: credentials to use
 */
int vfs_open(const struct path *path, struct file *file,
	     const struct cred *cred)
{
843
	struct dentry *dentry = d_real(path->dentry, NULL, file->f_flags);
844

845 846
	if (IS_ERR(dentry))
		return PTR_ERR(dentry);
847

848
	file->f_path = *path;
849
	return do_dentry_open(file, d_backing_inode(dentry), NULL, cred);
850 851
}

852
struct file *dentry_open(const struct path *path, int flags,
853
			 const struct cred *cred)
854 855 856 857
{
	int error;
	struct file *f;

858 859
	validate_creds(cred);

860
	/* We must always pass in a valid mount pointer. */
861
	BUG_ON(!path->mnt);
862

863
	f = get_empty_filp();
864 865
	if (!IS_ERR(f)) {
		f->f_flags = flags;
M
Miklos Szeredi 已提交
866
		error = vfs_open(path, f, cred);
867 868 869 870 871 872 873 874 875
		if (!error) {
			/* from now on we need fput() to dispose of f */
			error = open_check_o_direct(f);
			if (error) {
				fput(f);
				f = ERR_PTR(error);
			}
		} else { 
			put_filp(f);
876 877 878 879
			f = ERR_PTR(error);
		}
	}
	return f;
880
}
L
Linus Torvalds 已提交
881 882
EXPORT_SYMBOL(dentry_open);

883
static inline int build_open_flags(int flags, umode_t mode, struct open_flags *op)
884 885
{
	int lookup_flags = 0;
A
Al Viro 已提交
886
	int acc_mode = ACC_MODE(flags);
887

888
	if (flags & (O_CREAT | __O_TMPFILE))
889 890 891
		op->mode = (mode & S_IALLUGO) | S_IFREG;
	else
		op->mode = 0;
892 893

	/* Must never be set by userspace */
A
Al Viro 已提交
894
	flags &= ~FMODE_NONOTIFY & ~O_CLOEXEC;
895 896 897 898 899 900 901 902 903 904

	/*
	 * O_SYNC is implemented as __O_SYNC|O_DSYNC.  As many places only
	 * check for O_DSYNC if the need any syncing at all we enforce it's
	 * always set instead of having to deal with possibly weird behaviour
	 * for malicious applications setting only __O_SYNC.
	 */
	if (flags & __O_SYNC)
		flags |= O_DSYNC;

A
Al Viro 已提交
905 906
	if (flags & __O_TMPFILE) {
		if ((flags & O_TMPFILE_MASK) != O_TMPFILE)
907
			return -EINVAL;
A
Al Viro 已提交
908 909
		if (!(acc_mode & MAY_WRITE))
			return -EINVAL;
910 911 912 913 914
	} else if (flags & O_PATH) {
		/*
		 * If we have O_PATH in the open flag. Then we
		 * cannot have anything other than the below set of flags
		 */
915 916 917
		flags &= O_DIRECTORY | O_NOFOLLOW | O_PATH;
		acc_mode = 0;
	}
918

919
	op->open_flag = flags;
920 921 922 923 924 925 926 927 928 929 930 931

	/* O_TRUNC implies we need access checks for write permissions */
	if (flags & O_TRUNC)
		acc_mode |= MAY_WRITE;

	/* Allow the LSM permission hook to distinguish append
	   access from general write access. */
	if (flags & O_APPEND)
		acc_mode |= MAY_APPEND;

	op->acc_mode = acc_mode;

932 933
	op->intent = flags & O_PATH ? 0 : LOOKUP_OPEN;

934 935 936 937 938 939 940 941 942 943
	if (flags & O_CREAT) {
		op->intent |= LOOKUP_CREATE;
		if (flags & O_EXCL)
			op->intent |= LOOKUP_EXCL;
	}

	if (flags & O_DIRECTORY)
		lookup_flags |= LOOKUP_DIRECTORY;
	if (!(flags & O_NOFOLLOW))
		lookup_flags |= LOOKUP_FOLLOW;
944 945
	op->lookup_flags = lookup_flags;
	return 0;
946 947
}

948 949 950 951 952 953 954 955 956 957 958 959 960 961
/**
 * file_open_name - open file and return file pointer
 *
 * @name:	struct filename containing path to open
 * @flags:	open flags as per the open(2) second argument
 * @mode:	mode for the new file if O_CREAT is set, else ignored
 *
 * This is the helper to open a file from kernelspace if you really
 * have to.  But in generally you should not do this, so please move
 * along, nothing to see here..
 */
struct file *file_open_name(struct filename *name, int flags, umode_t mode)
{
	struct open_flags op;
962 963
	int err = build_open_flags(flags, mode, &op);
	return err ? ERR_PTR(err) : do_filp_open(AT_FDCWD, name, &op);
964 965
}

966 967 968 969 970 971 972 973 974 975 976
/**
 * filp_open - open file and return file pointer
 *
 * @filename:	path to open
 * @flags:	open flags as per the open(2) second argument
 * @mode:	mode for the new file if O_CREAT is set, else ignored
 *
 * This is the helper to open a file from kernelspace if you really
 * have to.  But in generally you should not do this, so please move
 * along, nothing to see here..
 */
977
struct file *filp_open(const char *filename, int flags, umode_t mode)
978
{
979 980 981 982 983 984 985 986
	struct filename *name = getname_kernel(filename);
	struct file *file = ERR_CAST(name);
	
	if (!IS_ERR(name)) {
		file = file_open_name(name, flags, mode);
		putname(name);
	}
	return file;
987 988 989
}
EXPORT_SYMBOL(filp_open);

A
Al Viro 已提交
990
struct file *file_open_root(struct dentry *dentry, struct vfsmount *mnt,
991
			    const char *filename, int flags, umode_t mode)
A
Al Viro 已提交
992 993
{
	struct open_flags op;
994
	int err = build_open_flags(flags, mode, &op);
995 996 997
	if (err)
		return ERR_PTR(err);
	return do_file_open_root(dentry, mnt, filename, &op);
A
Al Viro 已提交
998 999 1000
}
EXPORT_SYMBOL(file_open_root);

J
James Bottomley 已提交
1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020
struct file *filp_clone_open(struct file *oldfile)
{
	struct file *file;
	int retval;

	file = get_empty_filp();
	if (IS_ERR(file))
		return file;

	file->f_flags = oldfile->f_flags;
	retval = vfs_open(&oldfile->f_path, file, oldfile->f_cred);
	if (retval) {
		put_filp(file);
		return ERR_PTR(retval);
	}

	return file;
}
EXPORT_SYMBOL(filp_clone_open);

1021
long do_sys_open(int dfd, const char __user *filename, int flags, umode_t mode)
L
Linus Torvalds 已提交
1022
{
1023
	struct open_flags op;
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
	int fd = build_open_flags(flags, mode, &op);
	struct filename *tmp;

	if (fd)
		return fd;

	tmp = getname(filename);
	if (IS_ERR(tmp))
		return PTR_ERR(tmp);

	fd = get_unused_fd_flags(flags);
	if (fd >= 0) {
		struct file *f = do_filp_open(dfd, tmp, &op);
		if (IS_ERR(f)) {
			put_unused_fd(fd);
			fd = PTR_ERR(f);
		} else {
			fsnotify_open(f);
			fd_install(fd, f);
L
Linus Torvalds 已提交
1043 1044
		}
	}
1045
	putname(tmp);
L
Linus Torvalds 已提交
1046 1047
	return fd;
}
1048

1049
SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, umode_t, mode)
1050 1051 1052 1053
{
	if (force_o_largefile())
		flags |= O_LARGEFILE;

1054
	return do_sys_open(AT_FDCWD, filename, flags, mode);
1055
}
L
Linus Torvalds 已提交
1056

1057
SYSCALL_DEFINE4(openat, int, dfd, const char __user *, filename, int, flags,
1058
		umode_t, mode)
1059 1060 1061 1062
{
	if (force_o_largefile())
		flags |= O_LARGEFILE;

1063
	return do_sys_open(dfd, filename, flags, mode);
1064 1065
}

L
Linus Torvalds 已提交
1066 1067 1068 1069 1070 1071
#ifndef __alpha__

/*
 * For backward compatibility?  Maybe this should be moved
 * into arch/i386 instead?
 */
1072
SYSCALL_DEFINE2(creat, const char __user *, pathname, umode_t, mode)
L
Linus Torvalds 已提交
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084
{
	return sys_open(pathname, O_CREAT | O_WRONLY | O_TRUNC, mode);
}

#endif

/*
 * "id" is the POSIX thread ID. We use the
 * files pointer for this..
 */
int filp_close(struct file *filp, fl_owner_t id)
{
1085
	int retval = 0;
L
Linus Torvalds 已提交
1086 1087 1088

	if (!file_count(filp)) {
		printk(KERN_ERR "VFS: Close: file count is 0\n");
1089
		return 0;
L
Linus Torvalds 已提交
1090 1091
	}

A
Al Viro 已提交
1092
	if (filp->f_op->flush)
1093
		retval = filp->f_op->flush(filp, id);
L
Linus Torvalds 已提交
1094

1095 1096 1097 1098
	if (likely(!(filp->f_mode & FMODE_PATH))) {
		dnotify_flush(filp, id);
		locks_remove_posix(filp, id);
	}
L
Linus Torvalds 已提交
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109
	fput(filp);
	return retval;
}

EXPORT_SYMBOL(filp_close);

/*
 * Careful here! We test whether the file pointer is NULL before
 * releasing the fd. This ensures that one clone task can't release
 * an fd while another clone is opening it.
 */
1110
SYSCALL_DEFINE1(close, unsigned int, fd)
L
Linus Torvalds 已提交
1111
{
1112
	int retval = __close_fd(current->files, fd);
1113 1114 1115 1116 1117 1118 1119 1120 1121

	/* can't restart close syscall because file table entry was cleared */
	if (unlikely(retval == -ERESTARTSYS ||
		     retval == -ERESTARTNOINTR ||
		     retval == -ERESTARTNOHAND ||
		     retval == -ERESTART_RESTARTBLOCK))
		retval = -EINTR;

	return retval;
L
Linus Torvalds 已提交
1122 1123 1124 1125 1126 1127 1128
}
EXPORT_SYMBOL(sys_close);

/*
 * This routine simulates a hangup on the tty, to arrange that users
 * are given clean terminals at login time.
 */
1129
SYSCALL_DEFINE0(vhangup)
L
Linus Torvalds 已提交
1130 1131
{
	if (capable(CAP_SYS_TTY_CONFIG)) {
A
Alan Cox 已提交
1132
		tty_vhangup_self();
L
Linus Torvalds 已提交
1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
		return 0;
	}
	return -EPERM;
}

/*
 * Called when an inode is about to be open.
 * We use this to disallow opening large files on 32bit systems if
 * the caller didn't specify O_LARGEFILE.  On 64bit systems we force
 * on this flag in sys_open.
 */
int generic_file_open(struct inode * inode, struct file * filp)
{
	if (!(filp->f_flags & O_LARGEFILE) && i_size_read(inode) > MAX_NON_LFS)
1147
		return -EOVERFLOW;
L
Linus Torvalds 已提交
1148 1149 1150 1151 1152 1153 1154
	return 0;
}

EXPORT_SYMBOL(generic_file_open);

/*
 * This is used by subsystems that don't want seekable
1155 1156 1157
 * file descriptors. The function is not supposed to ever fail, the only
 * reason it returns an 'int' and not 'void' is so that it can be plugged
 * directly into file_operations structure.
L
Linus Torvalds 已提交
1158 1159 1160 1161 1162 1163 1164 1165
 */
int nonseekable_open(struct inode *inode, struct file *filp)
{
	filp->f_mode &= ~(FMODE_LSEEK | FMODE_PREAD | FMODE_PWRITE);
	return 0;
}

EXPORT_SYMBOL(nonseekable_open);