open.c 25.2 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
	/* Remove suid/sgid on truncate too */
55 56 57
	ret = should_remove_suid(dentry);
	if (ret)
		newattrs.ia_valid |= ret | ATTR_FORCE;
58

59
	mutex_lock(&dentry->d_inode->i_mutex);
60 61
	/* Note any delegations or leases have already been broken: */
	ret = notify_change(dentry, &newattrs, NULL);
62
	mutex_unlock(&dentry->d_inode->i_mutex);
63
	return ret;
L
Linus Torvalds 已提交
64 65
}

66
long vfs_truncate(struct path *path, loff_t length)
L
Linus Torvalds 已提交
67
{
68
	struct inode *inode;
69
	long error;
L
Linus Torvalds 已提交
70

71
	inode = path->dentry->d_inode;
L
Linus Torvalds 已提交
72 73 74

	/* For directories it's -EISDIR, for other non-regulars - -EINVAL */
	if (S_ISDIR(inode->i_mode))
75
		return -EISDIR;
L
Linus Torvalds 已提交
76
	if (!S_ISREG(inode->i_mode))
77
		return -EINVAL;
L
Linus Torvalds 已提交
78

79
	error = mnt_want_write(path->mnt);
L
Linus Torvalds 已提交
80
	if (error)
81
		goto out;
L
Linus Torvalds 已提交
82

83
	error = inode_permission(inode, MAY_WRITE);
84 85
	if (error)
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
86 87

	error = -EPERM;
88
	if (IS_APPEND(inode))
89
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
90

91
	error = get_write_access(inode);
L
Linus Torvalds 已提交
92
	if (error)
93
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
94

95 96 97 98
	/*
	 * Make sure that there are no leases.  get_write_access() protects
	 * against the truncate racing with a lease-granting setlease().
	 */
99
	error = break_lease(inode, O_WRONLY);
L
Linus Torvalds 已提交
100
	if (error)
101
		goto put_write_and_out;
L
Linus Torvalds 已提交
102 103

	error = locks_verify_truncate(inode, NULL, length);
104
	if (!error)
105
		error = security_path_truncate(path);
106
	if (!error)
107
		error = do_truncate(path->dentry, length, 0, NULL);
L
Linus Torvalds 已提交
108

109 110
put_write_and_out:
	put_write_access(inode);
111
mnt_drop_write_and_out:
112
	mnt_drop_write(path->mnt);
L
Linus Torvalds 已提交
113 114 115
out:
	return error;
}
116 117 118 119
EXPORT_SYMBOL_GPL(vfs_truncate);

static long do_sys_truncate(const char __user *pathname, loff_t length)
{
120
	unsigned int lookup_flags = LOOKUP_FOLLOW;
121 122 123 124 125 126
	struct path path;
	int error;

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

127 128
retry:
	error = user_path_at(AT_FDCWD, pathname, lookup_flags, &path);
129 130 131 132
	if (!error) {
		error = vfs_truncate(&path, length);
		path_put(&path);
	}
133 134 135 136
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
137 138
	return error;
}
L
Linus Torvalds 已提交
139

140
SYSCALL_DEFINE2(truncate, const char __user *, path, long, length)
L
Linus Torvalds 已提交
141
{
142
	return do_sys_truncate(path, length);
L
Linus Torvalds 已提交
143 144
}

A
Al Viro 已提交
145 146 147 148 149 150 151
#ifdef CONFIG_COMPAT
COMPAT_SYSCALL_DEFINE2(truncate, const char __user *, path, compat_off_t, length)
{
	return do_sys_truncate(path, length);
}
#endif

152
static long do_sys_ftruncate(unsigned int fd, loff_t length, int small)
L
Linus Torvalds 已提交
153
{
A
Al Viro 已提交
154
	struct inode *inode;
L
Linus Torvalds 已提交
155
	struct dentry *dentry;
156
	struct fd f;
L
Linus Torvalds 已提交
157 158 159 160 161 162
	int error;

	error = -EINVAL;
	if (length < 0)
		goto out;
	error = -EBADF;
163 164
	f = fdget(fd);
	if (!f.file)
L
Linus Torvalds 已提交
165 166 167
		goto out;

	/* explicitly opened as large or we are on 64-bit box */
168
	if (f.file->f_flags & O_LARGEFILE)
L
Linus Torvalds 已提交
169 170
		small = 0;

171
	dentry = f.file->f_path.dentry;
L
Linus Torvalds 已提交
172 173
	inode = dentry->d_inode;
	error = -EINVAL;
174
	if (!S_ISREG(inode->i_mode) || !(f.file->f_mode & FMODE_WRITE))
L
Linus Torvalds 已提交
175 176 177 178 179 180 181 182 183 184 185
		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;

186
	sb_start_write(inode->i_sb);
187
	error = locks_verify_truncate(inode, f.file, length);
188
	if (!error)
189
		error = security_path_truncate(&f.file->f_path);
L
Linus Torvalds 已提交
190
	if (!error)
191
		error = do_truncate(dentry, length, ATTR_MTIME|ATTR_CTIME, f.file);
192
	sb_end_write(inode->i_sb);
L
Linus Torvalds 已提交
193
out_putf:
194
	fdput(f);
L
Linus Torvalds 已提交
195 196 197 198
out:
	return error;
}

199
SYSCALL_DEFINE2(ftruncate, unsigned int, fd, unsigned long, length)
L
Linus Torvalds 已提交
200
{
201
	return do_sys_ftruncate(fd, length, 1);
L
Linus Torvalds 已提交
202 203
}

A
Al Viro 已提交
204 205 206 207 208 209 210
#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
211 212
/* LFS versions of truncate are only needed on 32 bit machines */
#if BITS_PER_LONG == 32
213
SYSCALL_DEFINE2(truncate64, const char __user *, path, loff_t, length)
L
Linus Torvalds 已提交
214 215 216 217
{
	return do_sys_truncate(path, length);
}

218
SYSCALL_DEFINE2(ftruncate64, unsigned int, fd, loff_t, length)
L
Linus Torvalds 已提交
219
{
220
	return do_sys_ftruncate(fd, length, 0);
L
Linus Torvalds 已提交
221
}
222
#endif /* BITS_PER_LONG == 32 */
L
Linus Torvalds 已提交
223

224 225

int do_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
226
{
A
Al Viro 已提交
227
	struct inode *inode = file_inode(file);
228
	long ret;
229 230

	if (offset < 0 || len <= 0)
231
		return -EINVAL;
232 233

	/* Return error if mode is not supported */
J
Josef Bacik 已提交
234 235 236 237 238 239
	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
		return -EOPNOTSUPP;

	/* Punch hole must have keep size set */
	if ((mode & FALLOC_FL_PUNCH_HOLE) &&
	    !(mode & FALLOC_FL_KEEP_SIZE))
240
		return -EOPNOTSUPP;
241 242

	if (!(file->f_mode & FMODE_WRITE))
243
		return -EBADF;
244 245 246 247 248 249 250 251

	/* It's not possible punch hole on append only file */
	if (mode & FALLOC_FL_PUNCH_HOLE && IS_APPEND(inode))
		return -EPERM;

	if (IS_IMMUTABLE(inode))
		return -EPERM;

252 253 254 255 256 257
	/*
	 * Revalidate the write permissions, in case security policy has
	 * changed since the files were opened.
	 */
	ret = security_file_permission(file, MAY_WRITE);
	if (ret)
258
		return ret;
259 260

	if (S_ISFIFO(inode->i_mode))
261
		return -ESPIPE;
262 263 264 265 266 267

	/*
	 * Let individual file system decide if it supports preallocation
	 * for directories or not.
	 */
	if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode))
268
		return -ENODEV;
269 270 271

	/* Check for wrap through zero too */
	if (((offset + len) > inode->i_sb->s_maxbytes) || ((offset + len) < 0))
272
		return -EFBIG;
273

274
	if (!file->f_op->fallocate)
275
		return -EOPNOTSUPP;
276

277 278 279 280
	sb_start_write(inode->i_sb);
	ret = file->f_op->fallocate(file, mode, offset, len);
	sb_end_write(inode->i_sb);
	return ret;
281 282
}

283
SYSCALL_DEFINE4(fallocate, int, fd, int, mode, loff_t, offset, loff_t, len)
284
{
285
	struct fd f = fdget(fd);
286 287
	int error = -EBADF;

288 289 290
	if (f.file) {
		error = do_fallocate(f.file, mode, offset, len);
		fdput(f);
291 292
	}
	return error;
293
}
294

L
Linus Torvalds 已提交
295 296 297 298 299
/*
 * 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.
 */
300
SYSCALL_DEFINE3(faccessat, int, dfd, const char __user *, filename, int, mode)
L
Linus Torvalds 已提交
301
{
D
David Howells 已提交
302 303
	const struct cred *old_cred;
	struct cred *override_cred;
304
	struct path path;
305
	struct inode *inode;
L
Linus Torvalds 已提交
306
	int res;
307
	unsigned int lookup_flags = LOOKUP_FOLLOW;
L
Linus Torvalds 已提交
308 309 310 311

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

D
David Howells 已提交
312 313 314
	override_cred = prepare_creds();
	if (!override_cred)
		return -ENOMEM;
L
Linus Torvalds 已提交
315

D
David Howells 已提交
316 317
	override_cred->fsuid = override_cred->uid;
	override_cred->fsgid = override_cred->gid;
L
Linus Torvalds 已提交
318

319
	if (!issecure(SECURE_NO_SETUID_FIXUP)) {
D
David Howells 已提交
320
		/* Clear the capabilities if we switch to a non-root user */
321 322
		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
323
			cap_clear(override_cred->cap_effective);
324
		else
D
David Howells 已提交
325 326
			override_cred->cap_effective =
				override_cred->cap_permitted;
327
	}
L
Linus Torvalds 已提交
328

D
David Howells 已提交
329
	old_cred = override_creds(override_cred);
330 331
retry:
	res = user_path_at(dfd, filename, lookup_flags, &path);
332 333 334
	if (res)
		goto out;

335
	inode = path.dentry->d_inode;
336 337

	if ((mode & MAY_EXEC) && S_ISREG(inode->i_mode)) {
338 339 340 341 342
		/*
		 * MAY_EXEC on regular files is denied if the fs is mounted
		 * with the "noexec" flag.
		 */
		res = -EACCES;
343
		if (path.mnt->mnt_flags & MNT_NOEXEC)
344 345 346
			goto out_path_release;
	}

347
	res = inode_permission(inode, mode | MAY_ACCESS);
348
	/* SuS v2 requires we report a read only fs too */
349
	if (res || !(mode & S_IWOTH) || special_file(inode->i_mode))
350
		goto out_path_release;
351 352 353 354 355 356 357 358 359 360
	/*
	 * 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.
	 */
361
	if (__mnt_is_readonly(path.mnt))
362
		res = -EROFS;
L
Linus Torvalds 已提交
363

364
out_path_release:
365
	path_put(&path);
366 367 368 369
	if (retry_estale(res, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
370
out:
D
David Howells 已提交
371 372
	revert_creds(old_cred);
	put_cred(override_cred);
L
Linus Torvalds 已提交
373 374 375
	return res;
}

376
SYSCALL_DEFINE2(access, const char __user *, filename, int, mode)
377 378 379 380
{
	return sys_faccessat(AT_FDCWD, filename, mode);
}

381
SYSCALL_DEFINE1(chdir, const char __user *, filename)
L
Linus Torvalds 已提交
382
{
383
	struct path path;
L
Linus Torvalds 已提交
384
	int error;
385 386 387
	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
388 389 390
	if (error)
		goto out;

E
Eric Paris 已提交
391
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
392 393 394
	if (error)
		goto dput_and_out;

395
	set_fs_pwd(current->fs, &path);
L
Linus Torvalds 已提交
396 397

dput_and_out:
398
	path_put(&path);
399 400 401 402
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
L
Linus Torvalds 已提交
403 404 405 406
out:
	return error;
}

407
SYSCALL_DEFINE1(fchdir, unsigned int, fd)
L
Linus Torvalds 已提交
408
{
409
	struct fd f = fdget_raw(fd);
L
Linus Torvalds 已提交
410
	struct inode *inode;
411
	int error = -EBADF;
L
Linus Torvalds 已提交
412 413

	error = -EBADF;
414
	if (!f.file)
L
Linus Torvalds 已提交
415 416
		goto out;

A
Al Viro 已提交
417
	inode = file_inode(f.file);
L
Linus Torvalds 已提交
418 419 420 421 422

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

E
Eric Paris 已提交
423
	error = inode_permission(inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
424
	if (!error)
425
		set_fs_pwd(current->fs, &f.file->f_path);
L
Linus Torvalds 已提交
426
out_putf:
427
	fdput(f);
L
Linus Torvalds 已提交
428 429 430 431
out:
	return error;
}

432
SYSCALL_DEFINE1(chroot, const char __user *, filename)
L
Linus Torvalds 已提交
433
{
434
	struct path path;
L
Linus Torvalds 已提交
435
	int error;
436 437 438
	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
439 440 441
	if (error)
		goto out;

E
Eric Paris 已提交
442
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
443 444 445 446
	if (error)
		goto dput_and_out;

	error = -EPERM;
447
	if (!ns_capable(current_user_ns(), CAP_SYS_CHROOT))
L
Linus Torvalds 已提交
448
		goto dput_and_out;
T
Tetsuo Handa 已提交
449 450 451
	error = security_path_chroot(&path);
	if (error)
		goto dput_and_out;
L
Linus Torvalds 已提交
452

453
	set_fs_root(current->fs, &path);
L
Linus Torvalds 已提交
454 455
	error = 0;
dput_and_out:
456
	path_put(&path);
457 458 459 460
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
L
Linus Torvalds 已提交
461 462 463 464
out:
	return error;
}

465
static int chmod_common(struct path *path, umode_t mode)
L
Linus Torvalds 已提交
466
{
467
	struct inode *inode = path->dentry->d_inode;
468
	struct inode *delegated_inode = NULL;
L
Linus Torvalds 已提交
469
	struct iattr newattrs;
470
	int error;
L
Linus Torvalds 已提交
471

472 473 474
	error = mnt_want_write(path->mnt);
	if (error)
		return error;
475
retry_deleg:
476
	mutex_lock(&inode->i_mutex);
477
	error = security_path_chmod(path, mode);
478
	if (error)
479
		goto out_unlock;
L
Linus Torvalds 已提交
480 481
	newattrs.ia_mode = (mode & S_IALLUGO) | (inode->i_mode & ~S_IALLUGO);
	newattrs.ia_valid = ATTR_MODE | ATTR_CTIME;
482
	error = notify_change(path->dentry, &newattrs, &delegated_inode);
483
out_unlock:
484
	mutex_unlock(&inode->i_mutex);
485 486 487 488 489
	if (delegated_inode) {
		error = break_deleg_wait(&delegated_inode);
		if (!error)
			goto retry_deleg;
	}
490 491 492 493
	mnt_drop_write(path->mnt);
	return error;
}

494
SYSCALL_DEFINE2(fchmod, unsigned int, fd, umode_t, mode)
495
{
A
Al Viro 已提交
496
	struct fd f = fdget(fd);
497 498
	int err = -EBADF;

A
Al Viro 已提交
499 500 501 502
	if (f.file) {
		audit_inode(NULL, f.file->f_path.dentry, 0);
		err = chmod_common(&f.file->f_path, mode);
		fdput(f);
503
	}
L
Linus Torvalds 已提交
504 505 506
	return err;
}

507
SYSCALL_DEFINE3(fchmodat, int, dfd, const char __user *, filename, umode_t, mode)
L
Linus Torvalds 已提交
508
{
509
	struct path path;
L
Linus Torvalds 已提交
510
	int error;
511 512 513
	unsigned int lookup_flags = LOOKUP_FOLLOW;
retry:
	error = user_path_at(dfd, filename, lookup_flags, &path);
514 515 516
	if (!error) {
		error = chmod_common(&path, mode);
		path_put(&path);
517 518 519 520
		if (retry_estale(error, lookup_flags)) {
			lookup_flags |= LOOKUP_REVAL;
			goto retry;
		}
521
	}
L
Linus Torvalds 已提交
522 523 524
	return error;
}

525
SYSCALL_DEFINE2(chmod, const char __user *, filename, umode_t, mode)
526 527 528 529
{
	return sys_fchmodat(AT_FDCWD, filename, mode);
}

530
static int chown_common(struct path *path, uid_t user, gid_t group)
L
Linus Torvalds 已提交
531
{
532
	struct inode *inode = path->dentry->d_inode;
533
	struct inode *delegated_inode = NULL;
L
Linus Torvalds 已提交
534 535
	int error;
	struct iattr newattrs;
536 537 538 539 540
	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
541 542 543

	newattrs.ia_valid =  ATTR_CTIME;
	if (user != (uid_t) -1) {
544 545
		if (!uid_valid(uid))
			return -EINVAL;
L
Linus Torvalds 已提交
546
		newattrs.ia_valid |= ATTR_UID;
547
		newattrs.ia_uid = uid;
L
Linus Torvalds 已提交
548 549
	}
	if (group != (gid_t) -1) {
550 551
		if (!gid_valid(gid))
			return -EINVAL;
L
Linus Torvalds 已提交
552
		newattrs.ia_valid |= ATTR_GID;
553
		newattrs.ia_gid = gid;
L
Linus Torvalds 已提交
554 555
	}
	if (!S_ISDIR(inode->i_mode))
556 557
		newattrs.ia_valid |=
			ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
558
retry_deleg:
559
	mutex_lock(&inode->i_mutex);
560
	error = security_path_chown(path, uid, gid);
561
	if (!error)
562
		error = notify_change(path->dentry, &newattrs, &delegated_inode);
563
	mutex_unlock(&inode->i_mutex);
564 565 566 567 568
	if (delegated_inode) {
		error = break_deleg_wait(&delegated_inode);
		if (!error)
			goto retry_deleg;
	}
L
Linus Torvalds 已提交
569 570 571
	return error;
}

572 573
SYSCALL_DEFINE5(fchownat, int, dfd, const char __user *, filename, uid_t, user,
		gid_t, group, int, flag)
574
{
575
	struct path path;
576
	int error = -EINVAL;
577
	int lookup_flags;
578

579
	if ((flag & ~(AT_SYMLINK_NOFOLLOW | AT_EMPTY_PATH)) != 0)
580 581
		goto out;

582 583 584
	lookup_flags = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
	if (flag & AT_EMPTY_PATH)
		lookup_flags |= LOOKUP_EMPTY;
585
retry:
586
	error = user_path_at(dfd, filename, lookup_flags, &path);
587 588
	if (error)
		goto out;
589
	error = mnt_want_write(path.mnt);
590 591
	if (error)
		goto out_release;
592
	error = chown_common(&path, user, group);
593
	mnt_drop_write(path.mnt);
594
out_release:
595
	path_put(&path);
596 597 598 599
	if (retry_estale(error, lookup_flags)) {
		lookup_flags |= LOOKUP_REVAL;
		goto retry;
	}
600 601 602 603
out:
	return error;
}

604
SYSCALL_DEFINE3(chown, const char __user *, filename, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
605
{
606 607
	return sys_fchownat(AT_FDCWD, filename, user, group, 0);
}
L
Linus Torvalds 已提交
608

609 610 611 612
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
613 614
}

615
SYSCALL_DEFINE3(fchown, unsigned int, fd, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
616
{
617
	struct fd f = fdget(fd);
L
Linus Torvalds 已提交
618 619
	int error = -EBADF;

620
	if (!f.file)
621 622
		goto out;

623
	error = mnt_want_write_file(f.file);
624 625
	if (error)
		goto out_fput;
626
	audit_inode(NULL, f.file->f_path.dentry, 0);
627 628
	error = chown_common(&f.file->f_path, user, group);
	mnt_drop_write_file(f.file);
629
out_fput:
630
	fdput(f);
631
out:
L
Linus Torvalds 已提交
632 633 634
	return error;
}

M
Miklos Szeredi 已提交
635 636 637 638 639 640 641 642 643 644 645 646 647
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) {
		if (!f->f_mapping->a_ops ||
		    ((!f->f_mapping->a_ops->direct_IO) &&
		    (!f->f_mapping->a_ops->get_xip_mem))) {
			return -EINVAL;
		}
	}
	return 0;
}

648
static int do_dentry_open(struct file *f,
649 650
			  int (*open)(struct inode *, struct file *),
			  const struct cred *cred)
L
Linus Torvalds 已提交
651
{
652
	static const struct file_operations empty_fops = {};
L
Linus Torvalds 已提交
653 654 655
	struct inode *inode;
	int error;

A
Al Viro 已提交
656
	f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
657
				FMODE_PREAD | FMODE_PWRITE;
658 659 660 661

	if (unlikely(f->f_flags & O_PATH))
		f->f_mode = FMODE_PATH;

662
	path_get(&f->f_path);
663
	inode = f->f_inode = f->f_path.dentry->d_inode;
664
	if (f->f_mode & FMODE_WRITE && !special_file(inode->i_mode)) {
665
		error = get_write_access(inode);
L
Linus Torvalds 已提交
666 667
		if (error)
			goto cleanup_file;
668 669 670 671 672
		error = __mnt_want_write(f->f_path.mnt);
		if (error) {
			put_write_access(inode);
			goto cleanup_file;
		}
L
Linus Torvalds 已提交
673 674 675 676
	}

	f->f_mapping = inode->i_mapping;

677 678
	if (unlikely(f->f_mode & FMODE_PATH)) {
		f->f_op = &empty_fops;
679
		return 0;
680 681
	}

682 683 684 685
	/* POSIX.1-2008/SUSv4 Section XSI 2.9.7 */
	if (S_ISREG(inode->i_mode))
		f->f_mode |= FMODE_ATOMIC_POS;

686
	f->f_op = fops_get(inode->i_fop);
A
Al Viro 已提交
687 688 689 690
	if (unlikely(WARN_ON(!f->f_op))) {
		error = -ENODEV;
		goto cleanup_all;
	}
691

692
	error = security_file_open(f, cred);
693 694 695
	if (error)
		goto cleanup_all;

696 697 698 699
	error = break_lease(inode, f->f_flags);
	if (error)
		goto cleanup_all;

A
Al Viro 已提交
700
	if (!open)
701 702 703
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
L
Linus Torvalds 已提交
704 705 706
		if (error)
			goto cleanup_all;
	}
707 708
	if ((f->f_mode & (FMODE_READ | FMODE_WRITE)) == FMODE_READ)
		i_readcount_inc(inode);
709

L
Linus Torvalds 已提交
710 711 712 713
	f->f_flags &= ~(O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC);

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

714
	return 0;
L
Linus Torvalds 已提交
715 716 717

cleanup_all:
	fops_put(f->f_op);
718
	if (f->f_mode & FMODE_WRITE) {
719
		if (!special_file(inode->i_mode)) {
720
			put_write_access(inode);
721
			__mnt_drop_write(f->f_path.mnt);
722
		}
723
	}
L
Linus Torvalds 已提交
724
cleanup_file:
725 726 727
	path_put(&f->f_path);
	f->f_path.mnt = NULL;
	f->f_path.dentry = NULL;
728
	f->f_inode = NULL;
729
	return error;
L
Linus Torvalds 已提交
730 731
}

M
Miklos Szeredi 已提交
732 733
/**
 * finish_open - finish opening a file
734
 * @file: file pointer
M
Miklos Szeredi 已提交
735 736
 * @dentry: pointer to dentry
 * @open: open callback
737
 * @opened: state of open
M
Miklos Szeredi 已提交
738 739 740 741 742
 *
 * 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.
743 744 745 746 747 748 749 750 751
 *
 * 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
752
 */
A
Al Viro 已提交
753 754 755
int finish_open(struct file *file, struct dentry *dentry,
		int (*open)(struct inode *, struct file *),
		int *opened)
M
Miklos Szeredi 已提交
756
{
757
	int error;
A
Al Viro 已提交
758
	BUG_ON(*opened & FILE_OPENED); /* once it's opened, it's opened */
M
Miklos Szeredi 已提交
759

760
	file->f_path.dentry = dentry;
761
	error = do_dentry_open(file, open, current_cred());
762
	if (!error)
763
		*opened |= FILE_OPENED;
M
Miklos Szeredi 已提交
764

765
	return error;
M
Miklos Szeredi 已提交
766 767 768 769 770 771
}
EXPORT_SYMBOL(finish_open);

/**
 * finish_no_open - finish ->atomic_open() without opening the file
 *
772
 * @file: file pointer
M
Miklos Szeredi 已提交
773 774 775
 * @dentry: dentry or NULL (as returned from ->lookup())
 *
 * This can be used to set the result of a successful lookup in ->atomic_open().
776 777 778 779 780 781
 *
 * 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
782
 */
A
Al Viro 已提交
783
int finish_no_open(struct file *file, struct dentry *dentry)
M
Miklos Szeredi 已提交
784
{
A
Al Viro 已提交
785
	file->f_path.dentry = dentry;
A
Al Viro 已提交
786
	return 1;
M
Miklos Szeredi 已提交
787 788 789
}
EXPORT_SYMBOL(finish_no_open);

790
struct file *dentry_open(const struct path *path, int flags,
791
			 const struct cred *cred)
792 793 794 795
{
	int error;
	struct file *f;

796 797
	validate_creds(cred);

798
	/* We must always pass in a valid mount pointer. */
799
	BUG_ON(!path->mnt);
800

801
	f = get_empty_filp();
802 803 804 805 806 807 808 809 810 811 812 813 814
	if (!IS_ERR(f)) {
		f->f_flags = flags;
		f->f_path = *path;
		error = do_dentry_open(f, NULL, cred);
		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);
815 816 817 818
			f = ERR_PTR(error);
		}
	}
	return f;
819
}
L
Linus Torvalds 已提交
820 821
EXPORT_SYMBOL(dentry_open);

822
static inline int build_open_flags(int flags, umode_t mode, struct open_flags *op)
823 824 825 826
{
	int lookup_flags = 0;
	int acc_mode;

827
	if (flags & (O_CREAT | __O_TMPFILE))
828 829 830
		op->mode = (mode & S_IALLUGO) | S_IFREG;
	else
		op->mode = 0;
831 832

	/* Must never be set by userspace */
A
Al Viro 已提交
833
	flags &= ~FMODE_NONOTIFY & ~O_CLOEXEC;
834 835 836 837 838 839 840 841 842 843

	/*
	 * 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
844 845
	if (flags & __O_TMPFILE) {
		if ((flags & O_TMPFILE_MASK) != O_TMPFILE)
846 847
			return -EINVAL;
		acc_mode = MAY_OPEN | ACC_MODE(flags);
A
Al Viro 已提交
848 849
		if (!(acc_mode & MAY_WRITE))
			return -EINVAL;
850 851 852 853 854
	} 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
		 */
855 856 857 858 859
		flags &= O_DIRECTORY | O_NOFOLLOW | O_PATH;
		acc_mode = 0;
	} else {
		acc_mode = MAY_OPEN | ACC_MODE(flags);
	}
860

861
	op->open_flag = flags;
862 863 864 865 866 867 868 869 870 871 872 873

	/* 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;

874 875
	op->intent = flags & O_PATH ? 0 : LOOKUP_OPEN;

876 877 878 879 880 881 882 883 884 885
	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;
886 887
	op->lookup_flags = lookup_flags;
	return 0;
888 889
}

890 891 892 893 894 895 896 897 898 899 900 901 902 903
/**
 * 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;
904 905
	int err = build_open_flags(flags, mode, &op);
	return err ? ERR_PTR(err) : do_filp_open(AT_FDCWD, name, &op);
906 907
}

908 909 910 911 912 913 914 915 916 917 918
/**
 * 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..
 */
919
struct file *filp_open(const char *filename, int flags, umode_t mode)
920
{
921 922
	struct filename name = {.name = filename};
	return file_open_name(&name, flags, mode);
923 924 925
}
EXPORT_SYMBOL(filp_open);

A
Al Viro 已提交
926 927 928 929
struct file *file_open_root(struct dentry *dentry, struct vfsmount *mnt,
			    const char *filename, int flags)
{
	struct open_flags op;
930 931 932
	int err = build_open_flags(flags, 0, &op);
	if (err)
		return ERR_PTR(err);
A
Al Viro 已提交
933 934 935 936 937
	if (flags & O_CREAT)
		return ERR_PTR(-EINVAL);
	if (!filename && (flags & O_DIRECTORY))
		if (!dentry->d_inode->i_op->lookup)
			return ERR_PTR(-ENOTDIR);
938
	return do_file_open_root(dentry, mnt, filename, &op);
A
Al Viro 已提交
939 940 941
}
EXPORT_SYMBOL(file_open_root);

942
long do_sys_open(int dfd, const char __user *filename, int flags, umode_t mode)
L
Linus Torvalds 已提交
943
{
944
	struct open_flags op;
945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963
	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
964 965
		}
	}
966
	putname(tmp);
L
Linus Torvalds 已提交
967 968
	return fd;
}
969

970
SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, umode_t, mode)
971 972 973 974
{
	if (force_o_largefile())
		flags |= O_LARGEFILE;

975
	return do_sys_open(AT_FDCWD, filename, flags, mode);
976
}
L
Linus Torvalds 已提交
977

978
SYSCALL_DEFINE4(openat, int, dfd, const char __user *, filename, int, flags,
979
		umode_t, mode)
980 981 982 983
{
	if (force_o_largefile())
		flags |= O_LARGEFILE;

984
	return do_sys_open(dfd, filename, flags, mode);
985 986
}

L
Linus Torvalds 已提交
987 988 989 990 991 992
#ifndef __alpha__

/*
 * For backward compatibility?  Maybe this should be moved
 * into arch/i386 instead?
 */
993
SYSCALL_DEFINE2(creat, const char __user *, pathname, umode_t, mode)
L
Linus Torvalds 已提交
994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
{
	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)
{
1006
	int retval = 0;
L
Linus Torvalds 已提交
1007 1008 1009

	if (!file_count(filp)) {
		printk(KERN_ERR "VFS: Close: file count is 0\n");
1010
		return 0;
L
Linus Torvalds 已提交
1011 1012
	}

A
Al Viro 已提交
1013
	if (filp->f_op->flush)
1014
		retval = filp->f_op->flush(filp, id);
L
Linus Torvalds 已提交
1015

1016 1017 1018 1019
	if (likely(!(filp->f_mode & FMODE_PATH))) {
		dnotify_flush(filp, id);
		locks_remove_posix(filp, id);
	}
L
Linus Torvalds 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030
	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.
 */
1031
SYSCALL_DEFINE1(close, unsigned int, fd)
L
Linus Torvalds 已提交
1032
{
1033
	int retval = __close_fd(current->files, fd);
1034 1035 1036 1037 1038 1039 1040 1041 1042

	/* 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
1043 1044 1045 1046 1047 1048 1049
}
EXPORT_SYMBOL(sys_close);

/*
 * This routine simulates a hangup on the tty, to arrange that users
 * are given clean terminals at login time.
 */
1050
SYSCALL_DEFINE0(vhangup)
L
Linus Torvalds 已提交
1051 1052
{
	if (capable(CAP_SYS_TTY_CONFIG)) {
A
Alan Cox 已提交
1053
		tty_vhangup_self();
L
Linus Torvalds 已提交
1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
		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)
1068
		return -EOVERFLOW;
L
Linus Torvalds 已提交
1069 1070 1071 1072 1073 1074 1075
	return 0;
}

EXPORT_SYMBOL(generic_file_open);

/*
 * This is used by subsystems that don't want seekable
1076 1077 1078
 * 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
1079 1080 1081 1082 1083 1084 1085 1086
 */
int nonseekable_open(struct inode *inode, struct file *filp)
{
	filp->f_mode &= ~(FMODE_LSEEK | FMODE_PREAD | FMODE_PWRITE);
	return 0;
}

EXPORT_SYMBOL(nonseekable_open);