open.c 23.9 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>
L
Linus Torvalds 已提交
33

34 35
#include "internal.h"

36 37
int do_truncate(struct dentry *dentry, loff_t length, unsigned int time_attrs,
	struct file *filp)
L
Linus Torvalds 已提交
38
{
39
	int ret;
L
Linus Torvalds 已提交
40 41 42 43 44 45 46
	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;
47
	newattrs.ia_valid = ATTR_SIZE | time_attrs;
48 49 50 51
	if (filp) {
		newattrs.ia_file = filp;
		newattrs.ia_valid |= ATTR_FILE;
	}
L
Linus Torvalds 已提交
52

53
	/* Remove suid/sgid on truncate too */
54 55 56
	ret = should_remove_suid(dentry);
	if (ret)
		newattrs.ia_valid |= ret | ATTR_FORCE;
57

58
	mutex_lock(&dentry->d_inode->i_mutex);
59
	ret = notify_change(dentry, &newattrs);
60
	mutex_unlock(&dentry->d_inode->i_mutex);
61
	return ret;
L
Linus Torvalds 已提交
62 63
}

64
static long do_sys_truncate(const char __user *pathname, loff_t length)
L
Linus Torvalds 已提交
65
{
66 67
	struct path path;
	struct inode *inode;
L
Linus Torvalds 已提交
68 69 70 71 72 73
	int error;

	error = -EINVAL;
	if (length < 0)	/* sorry, but loff_t says... */
		goto out;

74
	error = user_path(pathname, &path);
L
Linus Torvalds 已提交
75 76
	if (error)
		goto out;
77
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
78 79 80 81 82 83 84 85 86 87

	/* For directories it's -EISDIR, for other non-regulars - -EINVAL */
	error = -EISDIR;
	if (S_ISDIR(inode->i_mode))
		goto dput_and_out;

	error = -EINVAL;
	if (!S_ISREG(inode->i_mode))
		goto dput_and_out;

88
	error = mnt_want_write(path.mnt);
L
Linus Torvalds 已提交
89 90 91
	if (error)
		goto dput_and_out;

92
	error = inode_permission(inode, MAY_WRITE);
93 94
	if (error)
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
95 96

	error = -EPERM;
97
	if (IS_APPEND(inode))
98
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
99

100
	error = get_write_access(inode);
L
Linus Torvalds 已提交
101
	if (error)
102
		goto mnt_drop_write_and_out;
L
Linus Torvalds 已提交
103

104 105 106 107
	/*
	 * Make sure that there are no leases.  get_write_access() protects
	 * against the truncate racing with a lease-granting setlease().
	 */
108
	error = break_lease(inode, O_WRONLY);
L
Linus Torvalds 已提交
109
	if (error)
110
		goto put_write_and_out;
L
Linus Torvalds 已提交
111 112

	error = locks_verify_truncate(inode, NULL, length);
113
	if (!error)
114
		error = security_path_truncate(&path);
115
	if (!error)
116
		error = do_truncate(path.dentry, length, 0, NULL);
L
Linus Torvalds 已提交
117

118 119
put_write_and_out:
	put_write_access(inode);
120
mnt_drop_write_and_out:
121
	mnt_drop_write(path.mnt);
L
Linus Torvalds 已提交
122
dput_and_out:
123
	path_put(&path);
L
Linus Torvalds 已提交
124 125 126 127
out:
	return error;
}

128
SYSCALL_DEFINE2(truncate, const char __user *, path, long, length)
L
Linus Torvalds 已提交
129
{
130
	return do_sys_truncate(path, length);
L
Linus Torvalds 已提交
131 132
}

133
static long do_sys_ftruncate(unsigned int fd, loff_t length, int small)
L
Linus Torvalds 已提交
134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151
{
	struct inode * inode;
	struct dentry *dentry;
	struct file * file;
	int error;

	error = -EINVAL;
	if (length < 0)
		goto out;
	error = -EBADF;
	file = fget(fd);
	if (!file)
		goto out;

	/* explicitly opened as large or we are on 64-bit box */
	if (file->f_flags & O_LARGEFILE)
		small = 0;

152
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
153 154 155 156 157 158 159 160 161 162 163 164 165 166 167
	inode = dentry->d_inode;
	error = -EINVAL;
	if (!S_ISREG(inode->i_mode) || !(file->f_mode & FMODE_WRITE))
		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;

	error = locks_verify_truncate(inode, file, length);
168
	if (!error)
169
		error = security_path_truncate(&file->f_path);
L
Linus Torvalds 已提交
170
	if (!error)
171
		error = do_truncate(dentry, length, ATTR_MTIME|ATTR_CTIME, file);
L
Linus Torvalds 已提交
172 173 174 175 176 177
out_putf:
	fput(file);
out:
	return error;
}

178
SYSCALL_DEFINE2(ftruncate, unsigned int, fd, unsigned long, length)
L
Linus Torvalds 已提交
179
{
180
	long ret = do_sys_ftruncate(fd, length, 1);
181
	/* avoid REGPARM breakage on x86: */
182
	asmlinkage_protect(2, ret, fd, length);
183
	return ret;
L
Linus Torvalds 已提交
184 185 186 187
}

/* LFS versions of truncate are only needed on 32 bit machines */
#if BITS_PER_LONG == 32
188
SYSCALL_DEFINE(truncate64)(const char __user * path, loff_t length)
L
Linus Torvalds 已提交
189 190 191
{
	return do_sys_truncate(path, length);
}
192 193 194 195 196 197 198
#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
asmlinkage long SyS_truncate64(long path, loff_t length)
{
	return SYSC_truncate64((const char __user *) path, length);
}
SYSCALL_ALIAS(sys_truncate64, SyS_truncate64);
#endif
L
Linus Torvalds 已提交
199

200
SYSCALL_DEFINE(ftruncate64)(unsigned int fd, loff_t length)
L
Linus Torvalds 已提交
201
{
202
	long ret = do_sys_ftruncate(fd, length, 0);
203
	/* avoid REGPARM breakage on x86: */
204
	asmlinkage_protect(2, ret, fd, length);
205
	return ret;
L
Linus Torvalds 已提交
206
}
207 208 209 210 211 212
#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
asmlinkage long SyS_ftruncate64(long fd, loff_t length)
{
	return SYSC_ftruncate64((unsigned int) fd, length);
}
SYSCALL_ALIAS(sys_ftruncate64, SyS_ftruncate64);
L
Linus Torvalds 已提交
213
#endif
214
#endif /* BITS_PER_LONG == 32 */
L
Linus Torvalds 已提交
215

216 217

int do_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
218
{
219 220
	struct inode *inode = file->f_path.dentry->d_inode;
	long ret;
221 222

	if (offset < 0 || len <= 0)
223
		return -EINVAL;
224 225

	/* Return error if mode is not supported */
J
Josef Bacik 已提交
226 227 228 229 230 231
	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))
232
		return -EOPNOTSUPP;
233 234

	if (!(file->f_mode & FMODE_WRITE))
235
		return -EBADF;
236 237 238 239 240 241
	/*
	 * Revalidate the write permissions, in case security policy has
	 * changed since the files were opened.
	 */
	ret = security_file_permission(file, MAY_WRITE);
	if (ret)
242
		return ret;
243 244

	if (S_ISFIFO(inode->i_mode))
245
		return -ESPIPE;
246 247 248 249 250 251

	/*
	 * Let individual file system decide if it supports preallocation
	 * for directories or not.
	 */
	if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode))
252
		return -ENODEV;
253 254 255

	/* Check for wrap through zero too */
	if (((offset + len) > inode->i_sb->s_maxbytes) || ((offset + len) < 0))
256
		return -EFBIG;
257

258
	if (!file->f_op->fallocate)
259
		return -EOPNOTSUPP;
260

261
	return file->f_op->fallocate(file, mode, offset, len);
262 263 264 265 266 267 268 269 270 271 272 273 274 275
}

SYSCALL_DEFINE(fallocate)(int fd, int mode, loff_t offset, loff_t len)
{
	struct file *file;
	int error = -EBADF;

	file = fget(fd);
	if (file) {
		error = do_fallocate(file, mode, offset, len);
		fput(file);
	}

	return error;
276
}
277

278 279 280 281 282 283 284
#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
asmlinkage long SyS_fallocate(long fd, long mode, loff_t offset, loff_t len)
{
	return SYSC_fallocate((int)fd, (int)mode, offset, len);
}
SYSCALL_ALIAS(sys_fallocate, SyS_fallocate);
#endif
285

L
Linus Torvalds 已提交
286 287 288 289 290
/*
 * 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.
 */
291
SYSCALL_DEFINE3(faccessat, int, dfd, const char __user *, filename, int, mode)
L
Linus Torvalds 已提交
292
{
D
David Howells 已提交
293 294
	const struct cred *old_cred;
	struct cred *override_cred;
295
	struct path path;
296
	struct inode *inode;
L
Linus Torvalds 已提交
297 298 299 300 301
	int res;

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

D
David Howells 已提交
302 303 304
	override_cred = prepare_creds();
	if (!override_cred)
		return -ENOMEM;
L
Linus Torvalds 已提交
305

D
David Howells 已提交
306 307
	override_cred->fsuid = override_cred->uid;
	override_cred->fsgid = override_cred->gid;
L
Linus Torvalds 已提交
308

309
	if (!issecure(SECURE_NO_SETUID_FIXUP)) {
D
David Howells 已提交
310
		/* Clear the capabilities if we switch to a non-root user */
D
David Howells 已提交
311 312
		if (override_cred->uid)
			cap_clear(override_cred->cap_effective);
313
		else
D
David Howells 已提交
314 315
			override_cred->cap_effective =
				override_cred->cap_permitted;
316
	}
L
Linus Torvalds 已提交
317

D
David Howells 已提交
318 319
	old_cred = override_creds(override_cred);

320
	res = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
321 322 323
	if (res)
		goto out;

324
	inode = path.dentry->d_inode;
325 326

	if ((mode & MAY_EXEC) && S_ISREG(inode->i_mode)) {
327 328 329 330 331
		/*
		 * MAY_EXEC on regular files is denied if the fs is mounted
		 * with the "noexec" flag.
		 */
		res = -EACCES;
332
		if (path.mnt->mnt_flags & MNT_NOEXEC)
333 334 335
			goto out_path_release;
	}

336
	res = inode_permission(inode, mode | MAY_ACCESS);
337
	/* SuS v2 requires we report a read only fs too */
338
	if (res || !(mode & S_IWOTH) || special_file(inode->i_mode))
339
		goto out_path_release;
340 341 342 343 344 345 346 347 348 349
	/*
	 * 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.
	 */
350
	if (__mnt_is_readonly(path.mnt))
351
		res = -EROFS;
L
Linus Torvalds 已提交
352

353
out_path_release:
354
	path_put(&path);
355
out:
D
David Howells 已提交
356 357
	revert_creds(old_cred);
	put_cred(override_cred);
L
Linus Torvalds 已提交
358 359 360
	return res;
}

361
SYSCALL_DEFINE2(access, const char __user *, filename, int, mode)
362 363 364 365
{
	return sys_faccessat(AT_FDCWD, filename, mode);
}

366
SYSCALL_DEFINE1(chdir, const char __user *, filename)
L
Linus Torvalds 已提交
367
{
368
	struct path path;
L
Linus Torvalds 已提交
369 370
	int error;

371
	error = user_path_dir(filename, &path);
L
Linus Torvalds 已提交
372 373 374
	if (error)
		goto out;

E
Eric Paris 已提交
375
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
376 377 378
	if (error)
		goto dput_and_out;

379
	set_fs_pwd(current->fs, &path);
L
Linus Torvalds 已提交
380 381

dput_and_out:
382
	path_put(&path);
L
Linus Torvalds 已提交
383 384 385 386
out:
	return error;
}

387
SYSCALL_DEFINE1(fchdir, unsigned int, fd)
L
Linus Torvalds 已提交
388 389 390 391 392 393 394 395 396 397
{
	struct file *file;
	struct inode *inode;
	int error;

	error = -EBADF;
	file = fget(fd);
	if (!file)
		goto out;

398
	inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
399 400 401 402 403

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

E
Eric Paris 已提交
404
	error = inode_permission(inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
405
	if (!error)
406
		set_fs_pwd(current->fs, &file->f_path);
L
Linus Torvalds 已提交
407 408 409 410 411 412
out_putf:
	fput(file);
out:
	return error;
}

413
SYSCALL_DEFINE1(chroot, const char __user *, filename)
L
Linus Torvalds 已提交
414
{
415
	struct path path;
L
Linus Torvalds 已提交
416 417
	int error;

418
	error = user_path_dir(filename, &path);
L
Linus Torvalds 已提交
419 420 421
	if (error)
		goto out;

E
Eric Paris 已提交
422
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
423 424 425 426 427 428
	if (error)
		goto dput_and_out;

	error = -EPERM;
	if (!capable(CAP_SYS_CHROOT))
		goto dput_and_out;
T
Tetsuo Handa 已提交
429 430 431
	error = security_path_chroot(&path);
	if (error)
		goto dput_and_out;
L
Linus Torvalds 已提交
432

433
	set_fs_root(current->fs, &path);
L
Linus Torvalds 已提交
434 435
	error = 0;
dput_and_out:
436
	path_put(&path);
L
Linus Torvalds 已提交
437 438 439 440
out:
	return error;
}

441
SYSCALL_DEFINE2(fchmod, unsigned int, fd, mode_t, mode)
L
Linus Torvalds 已提交
442 443 444 445 446 447 448 449 450 451 452
{
	struct inode * inode;
	struct dentry * dentry;
	struct file * file;
	int err = -EBADF;
	struct iattr newattrs;

	file = fget(fd);
	if (!file)
		goto out;

453
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
454 455
	inode = dentry->d_inode;

456
	audit_inode(NULL, dentry);
457

N
npiggin@suse.de 已提交
458
	err = mnt_want_write_file(file);
459
	if (err)
L
Linus Torvalds 已提交
460
		goto out_putf;
461
	mutex_lock(&inode->i_mutex);
462 463
	err = security_path_chmod(dentry, file->f_vfsmnt, mode);
	if (err)
464
		goto out_unlock;
L
Linus Torvalds 已提交
465 466 467 468 469
	if (mode == (mode_t) -1)
		mode = inode->i_mode;
	newattrs.ia_mode = (mode & S_IALLUGO) | (inode->i_mode & ~S_IALLUGO);
	newattrs.ia_valid = ATTR_MODE | ATTR_CTIME;
	err = notify_change(dentry, &newattrs);
470
out_unlock:
471
	mutex_unlock(&inode->i_mutex);
472
	mnt_drop_write(file->f_path.mnt);
L
Linus Torvalds 已提交
473 474 475 476 477 478
out_putf:
	fput(file);
out:
	return err;
}

479
SYSCALL_DEFINE3(fchmodat, int, dfd, const char __user *, filename, mode_t, mode)
L
Linus Torvalds 已提交
480
{
481 482
	struct path path;
	struct inode *inode;
L
Linus Torvalds 已提交
483 484 485
	int error;
	struct iattr newattrs;

486
	error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
L
Linus Torvalds 已提交
487 488
	if (error)
		goto out;
489
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
490

491
	error = mnt_want_write(path.mnt);
492
	if (error)
L
Linus Torvalds 已提交
493
		goto dput_and_out;
494
	mutex_lock(&inode->i_mutex);
495 496
	error = security_path_chmod(path.dentry, path.mnt, mode);
	if (error)
497
		goto out_unlock;
L
Linus Torvalds 已提交
498 499 500 501
	if (mode == (mode_t) -1)
		mode = inode->i_mode;
	newattrs.ia_mode = (mode & S_IALLUGO) | (inode->i_mode & ~S_IALLUGO);
	newattrs.ia_valid = ATTR_MODE | ATTR_CTIME;
502
	error = notify_change(path.dentry, &newattrs);
503
out_unlock:
504
	mutex_unlock(&inode->i_mutex);
505
	mnt_drop_write(path.mnt);
L
Linus Torvalds 已提交
506
dput_and_out:
507
	path_put(&path);
L
Linus Torvalds 已提交
508 509 510 511
out:
	return error;
}

512
SYSCALL_DEFINE2(chmod, const char __user *, filename, mode_t, mode)
513 514 515 516
{
	return sys_fchmodat(AT_FDCWD, filename, mode);
}

517
static int chown_common(struct path *path, uid_t user, gid_t group)
L
Linus Torvalds 已提交
518
{
519
	struct inode *inode = path->dentry->d_inode;
L
Linus Torvalds 已提交
520 521 522 523 524 525 526 527 528 529 530 531 532
	int error;
	struct iattr newattrs;

	newattrs.ia_valid =  ATTR_CTIME;
	if (user != (uid_t) -1) {
		newattrs.ia_valid |= ATTR_UID;
		newattrs.ia_uid = user;
	}
	if (group != (gid_t) -1) {
		newattrs.ia_valid |= ATTR_GID;
		newattrs.ia_gid = group;
	}
	if (!S_ISDIR(inode->i_mode))
533 534
		newattrs.ia_valid |=
			ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
535
	mutex_lock(&inode->i_mutex);
536 537 538
	error = security_path_chown(path, user, group);
	if (!error)
		error = notify_change(path->dentry, &newattrs);
539
	mutex_unlock(&inode->i_mutex);
540

L
Linus Torvalds 已提交
541 542 543
	return error;
}

544
SYSCALL_DEFINE3(chown, const char __user *, filename, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
545
{
546
	struct path path;
L
Linus Torvalds 已提交
547 548
	int error;

549
	error = user_path(filename, &path);
550 551
	if (error)
		goto out;
552
	error = mnt_want_write(path.mnt);
553 554
	if (error)
		goto out_release;
555
	error = chown_common(&path, user, group);
556
	mnt_drop_write(path.mnt);
557
out_release:
558
	path_put(&path);
559
out:
L
Linus Torvalds 已提交
560 561 562
	return error;
}

563 564
SYSCALL_DEFINE5(fchownat, int, dfd, const char __user *, filename, uid_t, user,
		gid_t, group, int, flag)
565
{
566
	struct path path;
567 568 569 570 571 572 573
	int error = -EINVAL;
	int follow;

	if ((flag & ~AT_SYMLINK_NOFOLLOW) != 0)
		goto out;

	follow = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
574
	error = user_path_at(dfd, filename, follow, &path);
575 576
	if (error)
		goto out;
577
	error = mnt_want_write(path.mnt);
578 579
	if (error)
		goto out_release;
580
	error = chown_common(&path, user, group);
581
	mnt_drop_write(path.mnt);
582
out_release:
583
	path_put(&path);
584 585 586 587
out:
	return error;
}

588
SYSCALL_DEFINE3(lchown, const char __user *, filename, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
589
{
590
	struct path path;
L
Linus Torvalds 已提交
591 592
	int error;

593
	error = user_lpath(filename, &path);
594 595
	if (error)
		goto out;
596
	error = mnt_want_write(path.mnt);
597 598
	if (error)
		goto out_release;
599
	error = chown_common(&path, user, group);
600
	mnt_drop_write(path.mnt);
601
out_release:
602
	path_put(&path);
603
out:
L
Linus Torvalds 已提交
604 605 606
	return error;
}

607
SYSCALL_DEFINE3(fchown, unsigned int, fd, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
608 609 610
{
	struct file * file;
	int error = -EBADF;
611
	struct dentry * dentry;
L
Linus Torvalds 已提交
612 613

	file = fget(fd);
614 615 616
	if (!file)
		goto out;

N
npiggin@suse.de 已提交
617
	error = mnt_want_write_file(file);
618 619
	if (error)
		goto out_fput;
620
	dentry = file->f_path.dentry;
621
	audit_inode(NULL, dentry);
622
	error = chown_common(&file->f_path, user, group);
623 624
	mnt_drop_write(file->f_path.mnt);
out_fput:
625 626
	fput(file);
out:
L
Linus Torvalds 已提交
627 628 629
	return error;
}

630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658
/*
 * You have to be very careful that these write
 * counts get cleaned up in error cases and
 * upon __fput().  This should probably never
 * be called outside of __dentry_open().
 */
static inline int __get_file_write_access(struct inode *inode,
					  struct vfsmount *mnt)
{
	int error;
	error = get_write_access(inode);
	if (error)
		return error;
	/*
	 * Do not take mount writer counts on
	 * special files since no writes to
	 * the mount itself will occur.
	 */
	if (!special_file(inode->i_mode)) {
		/*
		 * Balanced in __fput()
		 */
		error = mnt_want_write(mnt);
		if (error)
			put_write_access(inode);
	}
	return error;
}

659
static struct file *__dentry_open(struct dentry *dentry, struct vfsmount *mnt,
660
					struct file *f,
661 662
					int (*open)(struct inode *, struct file *),
					const struct cred *cred)
L
Linus Torvalds 已提交
663 664 665 666
{
	struct inode *inode;
	int error;

A
Al Viro 已提交
667
	f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
668
				FMODE_PREAD | FMODE_PWRITE;
L
Linus Torvalds 已提交
669 670
	inode = dentry->d_inode;
	if (f->f_mode & FMODE_WRITE) {
671
		error = __get_file_write_access(inode, mnt);
L
Linus Torvalds 已提交
672 673
		if (error)
			goto cleanup_file;
674 675
		if (!special_file(inode->i_mode))
			file_take_write(f);
L
Linus Torvalds 已提交
676 677 678
	}

	f->f_mapping = inode->i_mapping;
679 680
	f->f_path.dentry = dentry;
	f->f_path.mnt = mnt;
L
Linus Torvalds 已提交
681 682
	f->f_pos = 0;
	f->f_op = fops_get(inode->i_fop);
N
Nick Piggin 已提交
683
	file_sb_list_add(f, inode->i_sb);
L
Linus Torvalds 已提交
684

685
	error = security_dentry_open(f, cred);
686 687 688
	if (error)
		goto cleanup_all;

689 690 691 692
	if (!open && f->f_op)
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
L
Linus Torvalds 已提交
693 694 695
		if (error)
			goto cleanup_all;
	}
696
	ima_counts_get(f);
697

L
Linus Torvalds 已提交
698 699 700 701 702 703
	f->f_flags &= ~(O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC);

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

	/* NB: we're sure to have correct a_ops only after f_op->open */
	if (f->f_flags & O_DIRECT) {
704 705
		if (!f->f_mapping->a_ops ||
		    ((!f->f_mapping->a_ops->direct_IO) &&
706
		    (!f->f_mapping->a_ops->get_xip_mem))) {
L
Linus Torvalds 已提交
707 708 709 710 711 712 713 714 715
			fput(f);
			f = ERR_PTR(-EINVAL);
		}
	}

	return f;

cleanup_all:
	fops_put(f->f_op);
716
	if (f->f_mode & FMODE_WRITE) {
L
Linus Torvalds 已提交
717
		put_write_access(inode);
718 719 720 721 722 723 724 725
		if (!special_file(inode->i_mode)) {
			/*
			 * We don't consider this a real
			 * mnt_want/drop_write() pair
			 * because it all happenend right
			 * here, so just reset the state.
			 */
			file_reset_write(f);
726
			mnt_drop_write(mnt);
727
		}
728
	}
N
Nick Piggin 已提交
729
	file_sb_list_del(f);
730 731
	f->f_path.dentry = NULL;
	f->f_path.mnt = NULL;
L
Linus Torvalds 已提交
732 733 734 735 736 737 738
cleanup_file:
	put_filp(f);
	dput(dentry);
	mntput(mnt);
	return ERR_PTR(error);
}

739 740 741 742 743 744 745 746 747 748
/**
 * lookup_instantiate_filp - instantiates the open intent filp
 * @nd: pointer to nameidata
 * @dentry: pointer to dentry
 * @open: open callback
 *
 * Helper for filesystems that want to use lookup open intents and pass back
 * a fully instantiated struct file to the caller.
 * This function is meant to be called from within a filesystem's
 * lookup method.
749 750 751 752
 * Beware of calling it for non-regular files! Those ->open methods might block
 * (e.g. in fifo_open), leaving you with parent locked (and in case of fifo,
 * leading to a deadlock, as nobody can open that fifo anymore, because
 * another process to open fifo will block on locked parent when doing lookup).
753 754 755 756 757 758 759 760
 * Note that in case of error, nd->intent.open.file is destroyed, but the
 * path information remains valid.
 * If the open callback is set to NULL, then the standard f_op->open()
 * filesystem callback is substituted.
 */
struct file *lookup_instantiate_filp(struct nameidata *nd, struct dentry *dentry,
		int (*open)(struct inode *, struct file *))
{
761 762
	const struct cred *cred = current_cred();

763 764 765 766
	if (IS_ERR(nd->intent.open.file))
		goto out;
	if (IS_ERR(dentry))
		goto out_err;
767
	nd->intent.open.file = __dentry_open(dget(dentry), mntget(nd->path.mnt),
768
					     nd->intent.open.file,
769
					     open, cred);
770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
out:
	return nd->intent.open.file;
out_err:
	release_open_intent(nd);
	nd->intent.open.file = (struct file *)dentry;
	goto out;
}
EXPORT_SYMBOL_GPL(lookup_instantiate_filp);

/**
 * nameidata_to_filp - convert a nameidata to an open filp.
 * @nd: pointer to nameidata
 * @flags: open flags
 *
 * Note that this function destroys the original nameidata
 */
786
struct file *nameidata_to_filp(struct nameidata *nd)
787
{
788
	const struct cred *cred = current_cred();
789 790 791 792 793
	struct file *filp;

	/* Pick up the filp from the open intent */
	filp = nd->intent.open.file;
	/* Has the filesystem initialised the file for us? */
A
Al Viro 已提交
794 795
	if (filp->f_path.dentry == NULL) {
		path_get(&nd->path);
796
		filp = __dentry_open(nd->path.dentry, nd->path.mnt, filp,
797
				     NULL, cred);
A
Al Viro 已提交
798
	}
799 800 801
	return filp;
}

802 803 804 805
/*
 * dentry_open() will have done dput(dentry) and mntput(mnt) if it returns an
 * error.
 */
806 807
struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags,
			 const struct cred *cred)
808 809 810 811
{
	int error;
	struct file *f;

812 813
	validate_creds(cred);

814 815 816 817 818 819 820 821 822 823 824 825
	/*
	 * We must always pass in a valid mount pointer.   Historically
	 * callers got away with not passing it, but we must enforce this at
	 * the earliest possible point now to avoid strange problems deep in the
	 * filesystem stack.
	 */
	if (!mnt) {
		printk(KERN_WARNING "%s called with NULL vfsmount\n", __func__);
		dump_stack();
		return ERR_PTR(-EINVAL);
	}

826 827
	error = -ENFILE;
	f = get_empty_filp();
828 829 830
	if (f == NULL) {
		dput(dentry);
		mntput(mnt);
831
		return ERR_PTR(error);
832
	}
833

834 835
	f->f_flags = flags;
	return __dentry_open(dentry, mnt, f, NULL, cred);
836
}
L
Linus Torvalds 已提交
837 838
EXPORT_SYMBOL(dentry_open);

839
static void __put_unused_fd(struct files_struct *files, unsigned int fd)
L
Linus Torvalds 已提交
840
{
841 842
	struct fdtable *fdt = files_fdtable(files);
	__FD_CLR(fd, fdt->open_fds);
843 844
	if (fd < files->next_fd)
		files->next_fd = fd;
L
Linus Torvalds 已提交
845 846
}

847
void put_unused_fd(unsigned int fd)
L
Linus Torvalds 已提交
848 849 850 851 852 853 854 855 856 857
{
	struct files_struct *files = current->files;
	spin_lock(&files->file_lock);
	__put_unused_fd(files, fd);
	spin_unlock(&files->file_lock);
}

EXPORT_SYMBOL(put_unused_fd);

/*
858
 * Install a file pointer in the fd array.
L
Linus Torvalds 已提交
859 860 861 862 863 864 865 866 867 868 869
 *
 * The VFS is full of places where we drop the files lock between
 * setting the open_fds bitmap and installing the file in the file
 * array.  At any such point, we are vulnerable to a dup2() race
 * installing a file in the array before us.  We need to detect this and
 * fput() the struct file we are about to overwrite in this case.
 *
 * It should never happen - if we allow dup2() do it, _really_ bad things
 * will follow.
 */

870
void fd_install(unsigned int fd, struct file *file)
L
Linus Torvalds 已提交
871 872
{
	struct files_struct *files = current->files;
873
	struct fdtable *fdt;
L
Linus Torvalds 已提交
874
	spin_lock(&files->file_lock);
875
	fdt = files_fdtable(files);
876 877
	BUG_ON(fdt->fd[fd] != NULL);
	rcu_assign_pointer(fdt->fd[fd], file);
L
Linus Torvalds 已提交
878 879 880 881 882
	spin_unlock(&files->file_lock);
}

EXPORT_SYMBOL(fd_install);

883
long do_sys_open(int dfd, const char __user *filename, int flags, int mode)
L
Linus Torvalds 已提交
884
{
885 886
	char *tmp = getname(filename);
	int fd = PTR_ERR(tmp);
L
Linus Torvalds 已提交
887 888

	if (!IS_ERR(tmp)) {
U
Ulrich Drepper 已提交
889
		fd = get_unused_fd_flags(flags);
L
Linus Torvalds 已提交
890
		if (fd >= 0) {
891
			struct file *f = do_filp_open(dfd, tmp, flags, mode, 0);
T
Telemaque Ndizihiwe 已提交
892 893 894 895
			if (IS_ERR(f)) {
				put_unused_fd(fd);
				fd = PTR_ERR(f);
			} else {
896
				fsnotify_open(f);
T
Telemaque Ndizihiwe 已提交
897 898
				fd_install(fd, f);
			}
L
Linus Torvalds 已提交
899 900 901 902 903
		}
		putname(tmp);
	}
	return fd;
}
904

905
SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, int, mode)
906
{
907 908
	long ret;

909 910 911
	if (force_o_largefile())
		flags |= O_LARGEFILE;

912 913
	ret = do_sys_open(AT_FDCWD, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
914
	asmlinkage_protect(3, ret, filename, flags, mode);
915
	return ret;
916
}
L
Linus Torvalds 已提交
917

918 919
SYSCALL_DEFINE4(openat, int, dfd, const char __user *, filename, int, flags,
		int, mode)
920
{
921 922
	long ret;

923 924 925
	if (force_o_largefile())
		flags |= O_LARGEFILE;

926 927
	ret = do_sys_open(dfd, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
928
	asmlinkage_protect(4, ret, dfd, filename, flags, mode);
929
	return ret;
930 931
}

L
Linus Torvalds 已提交
932 933 934 935 936 937
#ifndef __alpha__

/*
 * For backward compatibility?  Maybe this should be moved
 * into arch/i386 instead?
 */
938
SYSCALL_DEFINE2(creat, const char __user *, pathname, int, mode)
L
Linus Torvalds 已提交
939 940 941 942 943 944 945 946 947 948 949 950
{
	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)
{
951
	int retval = 0;
L
Linus Torvalds 已提交
952 953 954

	if (!file_count(filp)) {
		printk(KERN_ERR "VFS: Close: file count is 0\n");
955
		return 0;
L
Linus Torvalds 已提交
956 957
	}

958
	if (filp->f_op && filp->f_op->flush)
959
		retval = filp->f_op->flush(filp, id);
L
Linus Torvalds 已提交
960 961 962 963 964 965 966 967 968 969 970 971 972 973

	dnotify_flush(filp, id);
	locks_remove_posix(filp, id);
	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.
 */
974
SYSCALL_DEFINE1(close, unsigned int, fd)
L
Linus Torvalds 已提交
975 976 977
{
	struct file * filp;
	struct files_struct *files = current->files;
978
	struct fdtable *fdt;
979
	int retval;
L
Linus Torvalds 已提交
980 981

	spin_lock(&files->file_lock);
982 983
	fdt = files_fdtable(files);
	if (fd >= fdt->max_fds)
L
Linus Torvalds 已提交
984
		goto out_unlock;
985
	filp = fdt->fd[fd];
L
Linus Torvalds 已提交
986 987
	if (!filp)
		goto out_unlock;
988
	rcu_assign_pointer(fdt->fd[fd], NULL);
989
	FD_CLR(fd, fdt->close_on_exec);
L
Linus Torvalds 已提交
990 991
	__put_unused_fd(files, fd);
	spin_unlock(&files->file_lock);
992 993 994 995 996 997 998 999 1000 1001
	retval = filp_close(filp, files);

	/* 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
1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012

out_unlock:
	spin_unlock(&files->file_lock);
	return -EBADF;
}
EXPORT_SYMBOL(sys_close);

/*
 * This routine simulates a hangup on the tty, to arrange that users
 * are given clean terminals at login time.
 */
1013
SYSCALL_DEFINE0(vhangup)
L
Linus Torvalds 已提交
1014 1015
{
	if (capable(CAP_SYS_TTY_CONFIG)) {
A
Alan Cox 已提交
1016
		tty_vhangup_self();
L
Linus Torvalds 已提交
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030
		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)
1031
		return -EOVERFLOW;
L
Linus Torvalds 已提交
1032 1033 1034 1035 1036 1037 1038
	return 0;
}

EXPORT_SYMBOL(generic_file_open);

/*
 * This is used by subsystems that don't want seekable
1039 1040 1041
 * 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
1042 1043 1044 1045 1046 1047 1048 1049
 */
int nonseekable_open(struct inode *inode, struct file *filp)
{
	filp->f_mode &= ~(FMODE_LSEEK | FMODE_PREAD | FMODE_PWRITE);
	return 0;
}

EXPORT_SYMBOL(nonseekable_open);