open.c 25.9 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;
236 237 238 239 240 241 242 243

	/* 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;

244 245 246 247 248 249
	/*
	 * Revalidate the write permissions, in case security policy has
	 * changed since the files were opened.
	 */
	ret = security_file_permission(file, MAY_WRITE);
	if (ret)
250
		return ret;
251 252

	if (S_ISFIFO(inode->i_mode))
253
		return -ESPIPE;
254 255 256 257 258 259

	/*
	 * Let individual file system decide if it supports preallocation
	 * for directories or not.
	 */
	if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode))
260
		return -ENODEV;
261 262 263

	/* Check for wrap through zero too */
	if (((offset + len) > inode->i_sb->s_maxbytes) || ((offset + len) < 0))
264
		return -EFBIG;
265

266
	if (!file->f_op->fallocate)
267
		return -EOPNOTSUPP;
268

269
	return file->f_op->fallocate(file, mode, offset, len);
270 271 272 273 274 275 276 277 278 279 280 281 282 283
}

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;
284
}
285

286 287 288 289 290 291 292
#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
293

L
Linus Torvalds 已提交
294 295 296 297 298
/*
 * 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.
 */
299
SYSCALL_DEFINE3(faccessat, int, dfd, const char __user *, filename, int, mode)
L
Linus Torvalds 已提交
300
{
D
David Howells 已提交
301 302
	const struct cred *old_cred;
	struct cred *override_cred;
303
	struct path path;
304
	struct inode *inode;
L
Linus Torvalds 已提交
305 306 307 308 309
	int res;

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

D
David Howells 已提交
310 311 312
	override_cred = prepare_creds();
	if (!override_cred)
		return -ENOMEM;
L
Linus Torvalds 已提交
313

D
David Howells 已提交
314 315
	override_cred->fsuid = override_cred->uid;
	override_cred->fsgid = override_cred->gid;
L
Linus Torvalds 已提交
316

317
	if (!issecure(SECURE_NO_SETUID_FIXUP)) {
D
David Howells 已提交
318
		/* Clear the capabilities if we switch to a non-root user */
D
David Howells 已提交
319 320
		if (override_cred->uid)
			cap_clear(override_cred->cap_effective);
321
		else
D
David Howells 已提交
322 323
			override_cred->cap_effective =
				override_cred->cap_permitted;
324
	}
L
Linus Torvalds 已提交
325

D
David Howells 已提交
326 327
	old_cred = override_creds(override_cred);

328
	res = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
329 330 331
	if (res)
		goto out;

332
	inode = path.dentry->d_inode;
333 334

	if ((mode & MAY_EXEC) && S_ISREG(inode->i_mode)) {
335 336 337 338 339
		/*
		 * MAY_EXEC on regular files is denied if the fs is mounted
		 * with the "noexec" flag.
		 */
		res = -EACCES;
340
		if (path.mnt->mnt_flags & MNT_NOEXEC)
341 342 343
			goto out_path_release;
	}

344
	res = inode_permission(inode, mode | MAY_ACCESS);
345
	/* SuS v2 requires we report a read only fs too */
346
	if (res || !(mode & S_IWOTH) || special_file(inode->i_mode))
347
		goto out_path_release;
348 349 350 351 352 353 354 355 356 357
	/*
	 * 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.
	 */
358
	if (__mnt_is_readonly(path.mnt))
359
		res = -EROFS;
L
Linus Torvalds 已提交
360

361
out_path_release:
362
	path_put(&path);
363
out:
D
David Howells 已提交
364 365
	revert_creds(old_cred);
	put_cred(override_cred);
L
Linus Torvalds 已提交
366 367 368
	return res;
}

369
SYSCALL_DEFINE2(access, const char __user *, filename, int, mode)
370 371 372 373
{
	return sys_faccessat(AT_FDCWD, filename, mode);
}

374
SYSCALL_DEFINE1(chdir, const char __user *, filename)
L
Linus Torvalds 已提交
375
{
376
	struct path path;
L
Linus Torvalds 已提交
377 378
	int error;

379
	error = user_path_dir(filename, &path);
L
Linus Torvalds 已提交
380 381 382
	if (error)
		goto out;

E
Eric Paris 已提交
383
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
384 385 386
	if (error)
		goto dput_and_out;

387
	set_fs_pwd(current->fs, &path);
L
Linus Torvalds 已提交
388 389

dput_and_out:
390
	path_put(&path);
L
Linus Torvalds 已提交
391 392 393 394
out:
	return error;
}

395
SYSCALL_DEFINE1(fchdir, unsigned int, fd)
L
Linus Torvalds 已提交
396 397 398 399 400 401 402 403 404 405
{
	struct file *file;
	struct inode *inode;
	int error;

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

406
	inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
407 408 409 410 411

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

E
Eric Paris 已提交
412
	error = inode_permission(inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
413
	if (!error)
414
		set_fs_pwd(current->fs, &file->f_path);
L
Linus Torvalds 已提交
415 416 417 418 419 420
out_putf:
	fput(file);
out:
	return error;
}

421
SYSCALL_DEFINE1(chroot, const char __user *, filename)
L
Linus Torvalds 已提交
422
{
423
	struct path path;
L
Linus Torvalds 已提交
424 425
	int error;

426
	error = user_path_dir(filename, &path);
L
Linus Torvalds 已提交
427 428 429
	if (error)
		goto out;

E
Eric Paris 已提交
430
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
431 432 433 434 435 436
	if (error)
		goto dput_and_out;

	error = -EPERM;
	if (!capable(CAP_SYS_CHROOT))
		goto dput_and_out;
T
Tetsuo Handa 已提交
437 438 439
	error = security_path_chroot(&path);
	if (error)
		goto dput_and_out;
L
Linus Torvalds 已提交
440

441
	set_fs_root(current->fs, &path);
L
Linus Torvalds 已提交
442 443
	error = 0;
dput_and_out:
444
	path_put(&path);
L
Linus Torvalds 已提交
445 446 447 448
out:
	return error;
}

449
SYSCALL_DEFINE2(fchmod, unsigned int, fd, mode_t, mode)
L
Linus Torvalds 已提交
450 451 452 453 454 455 456 457 458 459 460
{
	struct inode * inode;
	struct dentry * dentry;
	struct file * file;
	int err = -EBADF;
	struct iattr newattrs;

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

461
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
462 463
	inode = dentry->d_inode;

464
	audit_inode(NULL, dentry);
465

N
npiggin@suse.de 已提交
466
	err = mnt_want_write_file(file);
467
	if (err)
L
Linus Torvalds 已提交
468
		goto out_putf;
469
	mutex_lock(&inode->i_mutex);
470 471
	err = security_path_chmod(dentry, file->f_vfsmnt, mode);
	if (err)
472
		goto out_unlock;
L
Linus Torvalds 已提交
473 474 475 476 477
	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);
478
out_unlock:
479
	mutex_unlock(&inode->i_mutex);
480
	mnt_drop_write(file->f_path.mnt);
L
Linus Torvalds 已提交
481 482 483 484 485 486
out_putf:
	fput(file);
out:
	return err;
}

487
SYSCALL_DEFINE3(fchmodat, int, dfd, const char __user *, filename, mode_t, mode)
L
Linus Torvalds 已提交
488
{
489 490
	struct path path;
	struct inode *inode;
L
Linus Torvalds 已提交
491 492 493
	int error;
	struct iattr newattrs;

494
	error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
L
Linus Torvalds 已提交
495 496
	if (error)
		goto out;
497
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
498

499
	error = mnt_want_write(path.mnt);
500
	if (error)
L
Linus Torvalds 已提交
501
		goto dput_and_out;
502
	mutex_lock(&inode->i_mutex);
503 504
	error = security_path_chmod(path.dentry, path.mnt, mode);
	if (error)
505
		goto out_unlock;
L
Linus Torvalds 已提交
506 507 508 509
	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;
510
	error = notify_change(path.dentry, &newattrs);
511
out_unlock:
512
	mutex_unlock(&inode->i_mutex);
513
	mnt_drop_write(path.mnt);
L
Linus Torvalds 已提交
514
dput_and_out:
515
	path_put(&path);
L
Linus Torvalds 已提交
516 517 518 519
out:
	return error;
}

520
SYSCALL_DEFINE2(chmod, const char __user *, filename, mode_t, mode)
521 522 523 524
{
	return sys_fchmodat(AT_FDCWD, filename, mode);
}

525
static int chown_common(struct path *path, uid_t user, gid_t group)
L
Linus Torvalds 已提交
526
{
527
	struct inode *inode = path->dentry->d_inode;
L
Linus Torvalds 已提交
528 529 530 531 532 533 534 535 536 537 538 539 540
	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))
541 542
		newattrs.ia_valid |=
			ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
543
	mutex_lock(&inode->i_mutex);
544 545 546
	error = security_path_chown(path, user, group);
	if (!error)
		error = notify_change(path->dentry, &newattrs);
547
	mutex_unlock(&inode->i_mutex);
548

L
Linus Torvalds 已提交
549 550 551
	return error;
}

552
SYSCALL_DEFINE3(chown, const char __user *, filename, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
553
{
554
	struct path path;
L
Linus Torvalds 已提交
555 556
	int error;

557
	error = user_path(filename, &path);
558 559
	if (error)
		goto out;
560
	error = mnt_want_write(path.mnt);
561 562
	if (error)
		goto out_release;
563
	error = chown_common(&path, user, group);
564
	mnt_drop_write(path.mnt);
565
out_release:
566
	path_put(&path);
567
out:
L
Linus Torvalds 已提交
568 569 570
	return error;
}

571 572
SYSCALL_DEFINE5(fchownat, int, dfd, const char __user *, filename, uid_t, user,
		gid_t, group, int, flag)
573
{
574
	struct path path;
575 576 577 578 579 580 581
	int error = -EINVAL;
	int follow;

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

	follow = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
582
	error = user_path_at(dfd, filename, follow, &path);
583 584
	if (error)
		goto out;
585
	error = mnt_want_write(path.mnt);
586 587
	if (error)
		goto out_release;
588
	error = chown_common(&path, user, group);
589
	mnt_drop_write(path.mnt);
590
out_release:
591
	path_put(&path);
592 593 594 595
out:
	return error;
}

596
SYSCALL_DEFINE3(lchown, const char __user *, filename, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
597
{
598
	struct path path;
L
Linus Torvalds 已提交
599 600
	int error;

601
	error = user_lpath(filename, &path);
602 603
	if (error)
		goto out;
604
	error = mnt_want_write(path.mnt);
605 606
	if (error)
		goto out_release;
607
	error = chown_common(&path, user, group);
608
	mnt_drop_write(path.mnt);
609
out_release:
610
	path_put(&path);
611
out:
L
Linus Torvalds 已提交
612 613 614
	return error;
}

615
SYSCALL_DEFINE3(fchown, unsigned int, fd, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
616 617 618
{
	struct file * file;
	int error = -EBADF;
619
	struct dentry * dentry;
L
Linus Torvalds 已提交
620 621

	file = fget(fd);
622 623 624
	if (!file)
		goto out;

N
npiggin@suse.de 已提交
625
	error = mnt_want_write_file(file);
626 627
	if (error)
		goto out_fput;
628
	dentry = file->f_path.dentry;
629
	audit_inode(NULL, dentry);
630
	error = chown_common(&file->f_path, user, group);
631 632
	mnt_drop_write(file->f_path.mnt);
out_fput:
633 634
	fput(file);
out:
L
Linus Torvalds 已提交
635 636 637
	return error;
}

638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666
/*
 * 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;
}

667
static struct file *__dentry_open(struct dentry *dentry, struct vfsmount *mnt,
668
					struct file *f,
669 670
					int (*open)(struct inode *, struct file *),
					const struct cred *cred)
L
Linus Torvalds 已提交
671 672 673 674
{
	struct inode *inode;
	int error;

A
Al Viro 已提交
675
	f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
676
				FMODE_PREAD | FMODE_PWRITE;
L
Linus Torvalds 已提交
677 678
	inode = dentry->d_inode;
	if (f->f_mode & FMODE_WRITE) {
679
		error = __get_file_write_access(inode, mnt);
L
Linus Torvalds 已提交
680 681
		if (error)
			goto cleanup_file;
682 683
		if (!special_file(inode->i_mode))
			file_take_write(f);
L
Linus Torvalds 已提交
684 685 686
	}

	f->f_mapping = inode->i_mapping;
687 688
	f->f_path.dentry = dentry;
	f->f_path.mnt = mnt;
L
Linus Torvalds 已提交
689 690
	f->f_pos = 0;
	f->f_op = fops_get(inode->i_fop);
N
Nick Piggin 已提交
691
	file_sb_list_add(f, inode->i_sb);
L
Linus Torvalds 已提交
692

693
	error = security_dentry_open(f, cred);
694 695 696
	if (error)
		goto cleanup_all;

697 698 699 700
	if (!open && f->f_op)
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
L
Linus Torvalds 已提交
701 702 703
		if (error)
			goto cleanup_all;
	}
704
	ima_counts_get(f);
705

L
Linus Torvalds 已提交
706 707 708 709 710 711
	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) {
712 713
		if (!f->f_mapping->a_ops ||
		    ((!f->f_mapping->a_ops->direct_IO) &&
714
		    (!f->f_mapping->a_ops->get_xip_mem))) {
L
Linus Torvalds 已提交
715 716 717 718 719 720 721 722 723
			fput(f);
			f = ERR_PTR(-EINVAL);
		}
	}

	return f;

cleanup_all:
	fops_put(f->f_op);
724
	if (f->f_mode & FMODE_WRITE) {
L
Linus Torvalds 已提交
725
		put_write_access(inode);
726 727 728 729 730 731 732 733
		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);
734
			mnt_drop_write(mnt);
735
		}
736
	}
N
Nick Piggin 已提交
737
	file_sb_list_del(f);
738 739
	f->f_path.dentry = NULL;
	f->f_path.mnt = NULL;
L
Linus Torvalds 已提交
740 741 742 743 744 745 746
cleanup_file:
	put_filp(f);
	dput(dentry);
	mntput(mnt);
	return ERR_PTR(error);
}

747 748 749 750 751 752 753 754 755 756
/**
 * 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.
757 758 759 760
 * 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).
761 762 763 764 765 766 767 768
 * 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 *))
{
769 770
	const struct cred *cred = current_cred();

771 772 773 774
	if (IS_ERR(nd->intent.open.file))
		goto out;
	if (IS_ERR(dentry))
		goto out_err;
775
	nd->intent.open.file = __dentry_open(dget(dentry), mntget(nd->path.mnt),
776
					     nd->intent.open.file,
777
					     open, cred);
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793
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
 */
794
struct file *nameidata_to_filp(struct nameidata *nd)
795
{
796
	const struct cred *cred = current_cred();
797 798 799 800
	struct file *filp;

	/* Pick up the filp from the open intent */
	filp = nd->intent.open.file;
801 802
	nd->intent.open.file = NULL;

803
	/* Has the filesystem initialised the file for us? */
A
Al Viro 已提交
804 805
	if (filp->f_path.dentry == NULL) {
		path_get(&nd->path);
806
		filp = __dentry_open(nd->path.dentry, nd->path.mnt, filp,
807
				     NULL, cred);
A
Al Viro 已提交
808
	}
809 810 811
	return filp;
}

812 813 814 815
/*
 * dentry_open() will have done dput(dentry) and mntput(mnt) if it returns an
 * error.
 */
816 817
struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags,
			 const struct cred *cred)
818 819 820 821
{
	int error;
	struct file *f;

822 823
	validate_creds(cred);

824 825 826 827 828 829 830 831 832 833 834 835
	/*
	 * 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);
	}

836 837
	error = -ENFILE;
	f = get_empty_filp();
838 839 840
	if (f == NULL) {
		dput(dentry);
		mntput(mnt);
841
		return ERR_PTR(error);
842
	}
843

844 845
	f->f_flags = flags;
	return __dentry_open(dentry, mnt, f, NULL, cred);
846
}
L
Linus Torvalds 已提交
847 848
EXPORT_SYMBOL(dentry_open);

849
static void __put_unused_fd(struct files_struct *files, unsigned int fd)
L
Linus Torvalds 已提交
850
{
851 852
	struct fdtable *fdt = files_fdtable(files);
	__FD_CLR(fd, fdt->open_fds);
853 854
	if (fd < files->next_fd)
		files->next_fd = fd;
L
Linus Torvalds 已提交
855 856
}

857
void put_unused_fd(unsigned int fd)
L
Linus Torvalds 已提交
858 859 860 861 862 863 864 865 866 867
{
	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);

/*
868
 * Install a file pointer in the fd array.
L
Linus Torvalds 已提交
869 870 871 872 873 874 875 876 877 878 879
 *
 * 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.
 */

880
void fd_install(unsigned int fd, struct file *file)
L
Linus Torvalds 已提交
881 882
{
	struct files_struct *files = current->files;
883
	struct fdtable *fdt;
L
Linus Torvalds 已提交
884
	spin_lock(&files->file_lock);
885
	fdt = files_fdtable(files);
886 887
	BUG_ON(fdt->fd[fd] != NULL);
	rcu_assign_pointer(fdt->fd[fd], file);
L
Linus Torvalds 已提交
888 889 890 891 892
	spin_unlock(&files->file_lock);
}

EXPORT_SYMBOL(fd_install);

893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961
static inline int build_open_flags(int flags, int mode, struct open_flags *op)
{
	int lookup_flags = 0;
	int acc_mode;

	if (!(flags & O_CREAT))
		mode = 0;
	op->mode = mode;

	/* Must never be set by userspace */
	flags &= ~FMODE_NONOTIFY;

	/*
	 * 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;

	op->open_flag = flags;

	acc_mode = MAY_OPEN | ACC_MODE(flags);

	/* 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;

	op->intent = LOOKUP_OPEN;
	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;
	return lookup_flags;
}

/**
 * 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..
 */
struct file *filp_open(const char *filename, int flags, int mode)
{
	struct open_flags op;
	int lookup = build_open_flags(flags, mode, &op);
	return do_filp_open(AT_FDCWD, filename, &op, lookup);
}
EXPORT_SYMBOL(filp_open);

962
long do_sys_open(int dfd, const char __user *filename, int flags, int mode)
L
Linus Torvalds 已提交
963
{
964 965
	struct open_flags op;
	int lookup = build_open_flags(flags, mode, &op);
966 967
	char *tmp = getname(filename);
	int fd = PTR_ERR(tmp);
L
Linus Torvalds 已提交
968 969

	if (!IS_ERR(tmp)) {
U
Ulrich Drepper 已提交
970
		fd = get_unused_fd_flags(flags);
L
Linus Torvalds 已提交
971
		if (fd >= 0) {
972
			struct file *f = do_filp_open(dfd, tmp, &op, lookup);
T
Telemaque Ndizihiwe 已提交
973 974 975 976
			if (IS_ERR(f)) {
				put_unused_fd(fd);
				fd = PTR_ERR(f);
			} else {
977
				fsnotify_open(f);
T
Telemaque Ndizihiwe 已提交
978 979
				fd_install(fd, f);
			}
L
Linus Torvalds 已提交
980 981 982 983 984
		}
		putname(tmp);
	}
	return fd;
}
985

986
SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, int, mode)
987
{
988 989
	long ret;

990 991 992
	if (force_o_largefile())
		flags |= O_LARGEFILE;

993 994
	ret = do_sys_open(AT_FDCWD, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
995
	asmlinkage_protect(3, ret, filename, flags, mode);
996
	return ret;
997
}
L
Linus Torvalds 已提交
998

999 1000
SYSCALL_DEFINE4(openat, int, dfd, const char __user *, filename, int, flags,
		int, mode)
1001
{
1002 1003
	long ret;

1004 1005 1006
	if (force_o_largefile())
		flags |= O_LARGEFILE;

1007 1008
	ret = do_sys_open(dfd, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
1009
	asmlinkage_protect(4, ret, dfd, filename, flags, mode);
1010
	return ret;
1011 1012
}

L
Linus Torvalds 已提交
1013 1014 1015 1016 1017 1018
#ifndef __alpha__

/*
 * For backward compatibility?  Maybe this should be moved
 * into arch/i386 instead?
 */
1019
SYSCALL_DEFINE2(creat, const char __user *, pathname, int, mode)
L
Linus Torvalds 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
{
	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)
{
1032
	int retval = 0;
L
Linus Torvalds 已提交
1033 1034 1035

	if (!file_count(filp)) {
		printk(KERN_ERR "VFS: Close: file count is 0\n");
1036
		return 0;
L
Linus Torvalds 已提交
1037 1038
	}

1039
	if (filp->f_op && filp->f_op->flush)
1040
		retval = filp->f_op->flush(filp, id);
L
Linus Torvalds 已提交
1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054

	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.
 */
1055
SYSCALL_DEFINE1(close, unsigned int, fd)
L
Linus Torvalds 已提交
1056 1057 1058
{
	struct file * filp;
	struct files_struct *files = current->files;
1059
	struct fdtable *fdt;
1060
	int retval;
L
Linus Torvalds 已提交
1061 1062

	spin_lock(&files->file_lock);
1063 1064
	fdt = files_fdtable(files);
	if (fd >= fdt->max_fds)
L
Linus Torvalds 已提交
1065
		goto out_unlock;
1066
	filp = fdt->fd[fd];
L
Linus Torvalds 已提交
1067 1068
	if (!filp)
		goto out_unlock;
1069
	rcu_assign_pointer(fdt->fd[fd], NULL);
1070
	FD_CLR(fd, fdt->close_on_exec);
L
Linus Torvalds 已提交
1071 1072
	__put_unused_fd(files, fd);
	spin_unlock(&files->file_lock);
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
	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
1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093

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.
 */
1094
SYSCALL_DEFINE0(vhangup)
L
Linus Torvalds 已提交
1095 1096
{
	if (capable(CAP_SYS_TTY_CONFIG)) {
A
Alan Cox 已提交
1097
		tty_vhangup_self();
L
Linus Torvalds 已提交
1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
		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)
1112
		return -EOVERFLOW;
L
Linus Torvalds 已提交
1113 1114 1115 1116 1117 1118 1119
	return 0;
}

EXPORT_SYMBOL(generic_file_open);

/*
 * This is used by subsystems that don't want seekable
1120 1121 1122
 * 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
1123 1124 1125 1126 1127 1128 1129 1130
 */
int nonseekable_open(struct inode *inode, struct file *filp)
{
	filp->f_mode &= ~(FMODE_LSEEK | FMODE_PREAD | FMODE_PWRITE);
	return 0;
}

EXPORT_SYMBOL(nonseekable_open);