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>
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 */
319 320
		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
321
			cap_clear(override_cred->cap_effective);
322
		else
D
David Howells 已提交
323 324
			override_cred->cap_effective =
				override_cred->cap_permitted;
325
	}
L
Linus Torvalds 已提交
326

D
David Howells 已提交
327 328
	old_cred = override_creds(override_cred);

329
	res = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
330 331 332
	if (res)
		goto out;

333
	inode = path.dentry->d_inode;
334 335

	if ((mode & MAY_EXEC) && S_ISREG(inode->i_mode)) {
336 337 338 339 340
		/*
		 * MAY_EXEC on regular files is denied if the fs is mounted
		 * with the "noexec" flag.
		 */
		res = -EACCES;
341
		if (path.mnt->mnt_flags & MNT_NOEXEC)
342 343 344
			goto out_path_release;
	}

345
	res = inode_permission(inode, mode | MAY_ACCESS);
346
	/* SuS v2 requires we report a read only fs too */
347
	if (res || !(mode & S_IWOTH) || special_file(inode->i_mode))
348
		goto out_path_release;
349 350 351 352 353 354 355 356 357 358
	/*
	 * 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.
	 */
359
	if (__mnt_is_readonly(path.mnt))
360
		res = -EROFS;
L
Linus Torvalds 已提交
361

362
out_path_release:
363
	path_put(&path);
364
out:
D
David Howells 已提交
365 366
	revert_creds(old_cred);
	put_cred(override_cred);
L
Linus Torvalds 已提交
367 368 369
	return res;
}

370
SYSCALL_DEFINE2(access, const char __user *, filename, int, mode)
371 372 373 374
{
	return sys_faccessat(AT_FDCWD, filename, mode);
}

375
SYSCALL_DEFINE1(chdir, const char __user *, filename)
L
Linus Torvalds 已提交
376
{
377
	struct path path;
L
Linus Torvalds 已提交
378 379
	int error;

380
	error = user_path_dir(filename, &path);
L
Linus Torvalds 已提交
381 382 383
	if (error)
		goto out;

E
Eric Paris 已提交
384
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
385 386 387
	if (error)
		goto dput_and_out;

388
	set_fs_pwd(current->fs, &path);
L
Linus Torvalds 已提交
389 390

dput_and_out:
391
	path_put(&path);
L
Linus Torvalds 已提交
392 393 394 395
out:
	return error;
}

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

	error = -EBADF;
403
	file = fget_raw_light(fd, &fput_needed);
L
Linus Torvalds 已提交
404 405 406
	if (!file)
		goto out;

407
	inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
408 409 410 411 412

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

E
Eric Paris 已提交
413
	error = inode_permission(inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
414
	if (!error)
415
		set_fs_pwd(current->fs, &file->f_path);
L
Linus Torvalds 已提交
416
out_putf:
417
	fput_light(file, fput_needed);
L
Linus Torvalds 已提交
418 419 420 421
out:
	return error;
}

422
SYSCALL_DEFINE1(chroot, const char __user *, filename)
L
Linus Torvalds 已提交
423
{
424
	struct path path;
L
Linus Torvalds 已提交
425 426
	int error;

427
	error = user_path_dir(filename, &path);
L
Linus Torvalds 已提交
428 429 430
	if (error)
		goto out;

E
Eric Paris 已提交
431
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
L
Linus Torvalds 已提交
432 433 434 435 436 437
	if (error)
		goto dput_and_out;

	error = -EPERM;
	if (!capable(CAP_SYS_CHROOT))
		goto dput_and_out;
T
Tetsuo Handa 已提交
438 439 440
	error = security_path_chroot(&path);
	if (error)
		goto dput_and_out;
L
Linus Torvalds 已提交
441

442
	set_fs_root(current->fs, &path);
L
Linus Torvalds 已提交
443 444
	error = 0;
dput_and_out:
445
	path_put(&path);
L
Linus Torvalds 已提交
446 447 448 449
out:
	return error;
}

450
static int chmod_common(struct path *path, umode_t mode)
L
Linus Torvalds 已提交
451
{
452
	struct inode *inode = path->dentry->d_inode;
L
Linus Torvalds 已提交
453
	struct iattr newattrs;
454
	int error;
L
Linus Torvalds 已提交
455

456 457 458
	error = mnt_want_write(path->mnt);
	if (error)
		return error;
459
	mutex_lock(&inode->i_mutex);
460
	error = security_path_chmod(path, mode);
461
	if (error)
462
		goto out_unlock;
L
Linus Torvalds 已提交
463 464
	newattrs.ia_mode = (mode & S_IALLUGO) | (inode->i_mode & ~S_IALLUGO);
	newattrs.ia_valid = ATTR_MODE | ATTR_CTIME;
465
	error = notify_change(path->dentry, &newattrs);
466
out_unlock:
467
	mutex_unlock(&inode->i_mutex);
468 469 470 471
	mnt_drop_write(path->mnt);
	return error;
}

472
SYSCALL_DEFINE2(fchmod, unsigned int, fd, umode_t, mode)
473 474 475 476 477 478 479 480 481 482
{
	struct file * file;
	int err = -EBADF;

	file = fget(fd);
	if (file) {
		audit_inode(NULL, file->f_path.dentry);
		err = chmod_common(&file->f_path, mode);
		fput(file);
	}
L
Linus Torvalds 已提交
483 484 485
	return err;
}

486
SYSCALL_DEFINE3(fchmodat, int, dfd, const char __user *, filename, umode_t, mode)
L
Linus Torvalds 已提交
487
{
488
	struct path path;
L
Linus Torvalds 已提交
489 490
	int error;

491
	error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
492 493 494 495
	if (!error) {
		error = chmod_common(&path, mode);
		path_put(&path);
	}
L
Linus Torvalds 已提交
496 497 498
	return error;
}

499
SYSCALL_DEFINE2(chmod, const char __user *, filename, umode_t, mode)
500 501 502 503
{
	return sys_fchmodat(AT_FDCWD, filename, mode);
}

504
static int chown_common(struct path *path, uid_t user, gid_t group)
L
Linus Torvalds 已提交
505
{
506
	struct inode *inode = path->dentry->d_inode;
L
Linus Torvalds 已提交
507 508
	int error;
	struct iattr newattrs;
509 510 511 512 513
	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
514 515 516

	newattrs.ia_valid =  ATTR_CTIME;
	if (user != (uid_t) -1) {
517 518
		if (!uid_valid(uid))
			return -EINVAL;
L
Linus Torvalds 已提交
519
		newattrs.ia_valid |= ATTR_UID;
520
		newattrs.ia_uid = uid;
L
Linus Torvalds 已提交
521 522
	}
	if (group != (gid_t) -1) {
523 524
		if (!gid_valid(gid))
			return -EINVAL;
L
Linus Torvalds 已提交
525
		newattrs.ia_valid |= ATTR_GID;
526
		newattrs.ia_gid = gid;
L
Linus Torvalds 已提交
527 528
	}
	if (!S_ISDIR(inode->i_mode))
529 530
		newattrs.ia_valid |=
			ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
531
	mutex_lock(&inode->i_mutex);
532 533 534
	error = security_path_chown(path, user, group);
	if (!error)
		error = notify_change(path->dentry, &newattrs);
535
	mutex_unlock(&inode->i_mutex);
536

L
Linus Torvalds 已提交
537 538 539
	return error;
}

540 541
SYSCALL_DEFINE5(fchownat, int, dfd, const char __user *, filename, uid_t, user,
		gid_t, group, int, flag)
542
{
543
	struct path path;
544
	int error = -EINVAL;
545
	int lookup_flags;
546

547
	if ((flag & ~(AT_SYMLINK_NOFOLLOW | AT_EMPTY_PATH)) != 0)
548 549
		goto out;

550 551 552 553
	lookup_flags = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
	if (flag & AT_EMPTY_PATH)
		lookup_flags |= LOOKUP_EMPTY;
	error = user_path_at(dfd, filename, lookup_flags, &path);
554 555
	if (error)
		goto out;
556
	error = mnt_want_write(path.mnt);
557 558
	if (error)
		goto out_release;
559
	error = chown_common(&path, user, group);
560
	mnt_drop_write(path.mnt);
561
out_release:
562
	path_put(&path);
563 564 565 566
out:
	return error;
}

567
SYSCALL_DEFINE3(chown, const char __user *, filename, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
568
{
569 570
	return sys_fchownat(AT_FDCWD, filename, user, group, 0);
}
L
Linus Torvalds 已提交
571

572 573 574 575
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
576 577
}

578
SYSCALL_DEFINE3(fchown, unsigned int, fd, uid_t, user, gid_t, group)
L
Linus Torvalds 已提交
579 580 581
{
	struct file * file;
	int error = -EBADF;
582
	struct dentry * dentry;
L
Linus Torvalds 已提交
583 584

	file = fget(fd);
585 586 587
	if (!file)
		goto out;

N
npiggin@suse.de 已提交
588
	error = mnt_want_write_file(file);
589 590
	if (error)
		goto out_fput;
591
	dentry = file->f_path.dentry;
592
	audit_inode(NULL, dentry);
593
	error = chown_common(&file->f_path, user, group);
A
Al Viro 已提交
594
	mnt_drop_write_file(file);
595
out_fput:
596 597
	fput(file);
out:
L
Linus Torvalds 已提交
598 599 600
	return error;
}

601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622
/*
 * 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()
		 */
623
		error = __mnt_want_write(mnt);
624 625 626 627 628 629
		if (error)
			put_write_access(inode);
	}
	return error;
}

M
Miklos Szeredi 已提交
630 631 632 633 634 635 636 637 638 639 640 641 642
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;
}

643
static int do_dentry_open(struct file *f,
644 645
			  int (*open)(struct inode *, struct file *),
			  const struct cred *cred)
L
Linus Torvalds 已提交
646
{
647
	static const struct file_operations empty_fops = {};
L
Linus Torvalds 已提交
648 649 650
	struct inode *inode;
	int error;

A
Al Viro 已提交
651
	f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
652
				FMODE_PREAD | FMODE_PWRITE;
653 654 655 656

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

657
	path_get(&f->f_path);
658
	inode = f->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
659
	if (f->f_mode & FMODE_WRITE) {
660
		error = __get_file_write_access(inode, f->f_path.mnt);
L
Linus Torvalds 已提交
661 662
		if (error)
			goto cleanup_file;
663 664
		if (!special_file(inode->i_mode))
			file_take_write(f);
L
Linus Torvalds 已提交
665 666 667 668
	}

	f->f_mapping = inode->i_mapping;
	f->f_pos = 0;
N
Nick Piggin 已提交
669
	file_sb_list_add(f, inode->i_sb);
L
Linus Torvalds 已提交
670

671 672
	if (unlikely(f->f_mode & FMODE_PATH)) {
		f->f_op = &empty_fops;
673
		return 0;
674 675 676 677
	}

	f->f_op = fops_get(inode->i_fop);

678
	error = security_file_open(f, cred);
679 680 681
	if (error)
		goto cleanup_all;

682 683 684 685
	error = break_lease(inode, f->f_flags);
	if (error)
		goto cleanup_all;

686 687 688 689
	if (!open && f->f_op)
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
L
Linus Torvalds 已提交
690 691 692
		if (error)
			goto cleanup_all;
	}
693 694
	if ((f->f_mode & (FMODE_READ | FMODE_WRITE)) == FMODE_READ)
		i_readcount_inc(inode);
695

L
Linus Torvalds 已提交
696 697 698 699
	f->f_flags &= ~(O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC);

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

700
	return 0;
L
Linus Torvalds 已提交
701 702 703

cleanup_all:
	fops_put(f->f_op);
704
	file_sb_list_del(f);
705
	if (f->f_mode & FMODE_WRITE) {
L
Linus Torvalds 已提交
706
		put_write_access(inode);
707 708 709 710 711 712 713 714
		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);
715
			mnt_drop_write(f->f_path.mnt);
716
		}
717
	}
L
Linus Torvalds 已提交
718
cleanup_file:
719 720 721
	path_put(&f->f_path);
	f->f_path.mnt = NULL;
	f->f_path.dentry = NULL;
722
	return error;
L
Linus Torvalds 已提交
723 724
}

M
Miklos Szeredi 已提交
725 726 727 728 729 730 731 732 733 734 735
/**
 * finish_open - finish opening a file
 * @od: opaque open data
 * @dentry: pointer to dentry
 * @open: open callback
 *
 * 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.
 */
A
Al Viro 已提交
736 737 738
int finish_open(struct file *file, struct dentry *dentry,
		int (*open)(struct inode *, struct file *),
		int *opened)
M
Miklos Szeredi 已提交
739
{
740
	int error;
A
Al Viro 已提交
741
	BUG_ON(*opened & FILE_OPENED); /* once it's opened, it's opened */
M
Miklos Szeredi 已提交
742

743
	file->f_path.dentry = dentry;
744
	error = do_dentry_open(file, open, current_cred());
745
	if (!error)
746
		*opened |= FILE_OPENED;
M
Miklos Szeredi 已提交
747

748
	return error;
M
Miklos Szeredi 已提交
749 750 751 752 753 754 755 756 757 758 759 760
}
EXPORT_SYMBOL(finish_open);

/**
 * finish_no_open - finish ->atomic_open() without opening the file
 *
 * @od: opaque open data
 * @dentry: dentry or NULL (as returned from ->lookup())
 *
 * This can be used to set the result of a successful lookup in ->atomic_open().
 * The filesystem's atomic_open() method shall return NULL after calling this.
 */
A
Al Viro 已提交
761
int finish_no_open(struct file *file, struct dentry *dentry)
M
Miklos Szeredi 已提交
762
{
A
Al Viro 已提交
763
	file->f_path.dentry = dentry;
A
Al Viro 已提交
764
	return 1;
M
Miklos Szeredi 已提交
765 766 767
}
EXPORT_SYMBOL(finish_no_open);

768
struct file *dentry_open(const struct path *path, int flags,
769
			 const struct cred *cred)
770 771 772 773
{
	int error;
	struct file *f;

774 775
	validate_creds(cred);

776
	/* We must always pass in a valid mount pointer. */
777
	BUG_ON(!path->mnt);
778

779 780
	error = -ENFILE;
	f = get_empty_filp();
781
	if (f == NULL)
782 783
		return ERR_PTR(error);

784
	f->f_flags = flags;
785
	f->f_path = *path;
786
	error = do_dentry_open(f, NULL, cred);
787 788 789 790 791 792 793 794 795 796 797
	if (!error) {
		error = open_check_o_direct(f);
		if (error) {
			fput(f);
			f = ERR_PTR(error);
		}
	} else { 
		put_filp(f);
		f = ERR_PTR(error);
	}
	return f;
798
}
L
Linus Torvalds 已提交
799 800
EXPORT_SYMBOL(dentry_open);

801
static void __put_unused_fd(struct files_struct *files, unsigned int fd)
L
Linus Torvalds 已提交
802
{
803
	struct fdtable *fdt = files_fdtable(files);
804
	__clear_open_fd(fd, fdt);
805 806
	if (fd < files->next_fd)
		files->next_fd = fd;
L
Linus Torvalds 已提交
807 808
}

809
void put_unused_fd(unsigned int fd)
L
Linus Torvalds 已提交
810 811 812 813 814 815 816 817 818 819
{
	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);

/*
820
 * Install a file pointer in the fd array.
L
Linus Torvalds 已提交
821 822 823 824 825 826 827 828 829 830 831
 *
 * 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.
 */

832
void fd_install(unsigned int fd, struct file *file)
L
Linus Torvalds 已提交
833 834
{
	struct files_struct *files = current->files;
835
	struct fdtable *fdt;
L
Linus Torvalds 已提交
836
	spin_lock(&files->file_lock);
837
	fdt = files_fdtable(files);
838 839
	BUG_ON(fdt->fd[fd] != NULL);
	rcu_assign_pointer(fdt->fd[fd], file);
L
Linus Torvalds 已提交
840 841 842 843 844
	spin_unlock(&files->file_lock);
}

EXPORT_SYMBOL(fd_install);

845
static inline int build_open_flags(int flags, umode_t mode, struct open_flags *op)
846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
{
	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;

866 867 868 869 870 871 872 873 874 875
	/*
	 * If we have O_PATH in the open flag. Then we
	 * cannot have anything other than the below set of flags
	 */
	if (flags & O_PATH) {
		flags &= O_DIRECTORY | O_NOFOLLOW | O_PATH;
		acc_mode = 0;
	} else {
		acc_mode = MAY_OPEN | ACC_MODE(flags);
	}
876

877
	op->open_flag = flags;
878 879 880 881 882 883 884 885 886 887 888 889

	/* 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;

890 891
	op->intent = flags & O_PATH ? 0 : LOOKUP_OPEN;

892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915
	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..
 */
916
struct file *filp_open(const char *filename, int flags, umode_t mode)
917 918 919 920 921 922 923
{
	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);

A
Al Viro 已提交
924 925 926 927 928 929 930 931 932 933 934 935 936 937
struct file *file_open_root(struct dentry *dentry, struct vfsmount *mnt,
			    const char *filename, int flags)
{
	struct open_flags op;
	int lookup = build_open_flags(flags, 0, &op);
	if (flags & O_CREAT)
		return ERR_PTR(-EINVAL);
	if (!filename && (flags & O_DIRECTORY))
		if (!dentry->d_inode->i_op->lookup)
			return ERR_PTR(-ENOTDIR);
	return do_file_open_root(dentry, mnt, filename, &op, lookup);
}
EXPORT_SYMBOL(file_open_root);

938
long do_sys_open(int dfd, const char __user *filename, int flags, umode_t mode)
L
Linus Torvalds 已提交
939
{
940 941
	struct open_flags op;
	int lookup = build_open_flags(flags, mode, &op);
942 943
	char *tmp = getname(filename);
	int fd = PTR_ERR(tmp);
L
Linus Torvalds 已提交
944 945

	if (!IS_ERR(tmp)) {
U
Ulrich Drepper 已提交
946
		fd = get_unused_fd_flags(flags);
L
Linus Torvalds 已提交
947
		if (fd >= 0) {
948
			struct file *f = do_filp_open(dfd, tmp, &op, lookup);
T
Telemaque Ndizihiwe 已提交
949 950 951 952
			if (IS_ERR(f)) {
				put_unused_fd(fd);
				fd = PTR_ERR(f);
			} else {
953
				fsnotify_open(f);
T
Telemaque Ndizihiwe 已提交
954 955
				fd_install(fd, f);
			}
L
Linus Torvalds 已提交
956 957 958 959 960
		}
		putname(tmp);
	}
	return fd;
}
961

962
SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, umode_t, mode)
963
{
964 965
	long ret;

966 967 968
	if (force_o_largefile())
		flags |= O_LARGEFILE;

969 970
	ret = do_sys_open(AT_FDCWD, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
971
	asmlinkage_protect(3, ret, filename, flags, mode);
972
	return ret;
973
}
L
Linus Torvalds 已提交
974

975
SYSCALL_DEFINE4(openat, int, dfd, const char __user *, filename, int, flags,
976
		umode_t, mode)
977
{
978 979
	long ret;

980 981 982
	if (force_o_largefile())
		flags |= O_LARGEFILE;

983 984
	ret = do_sys_open(dfd, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
985
	asmlinkage_protect(4, ret, dfd, filename, flags, mode);
986
	return ret;
987 988
}

L
Linus Torvalds 已提交
989 990 991 992 993 994
#ifndef __alpha__

/*
 * For backward compatibility?  Maybe this should be moved
 * into arch/i386 instead?
 */
995
SYSCALL_DEFINE2(creat, const char __user *, pathname, umode_t, mode)
L
Linus Torvalds 已提交
996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
{
	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)
{
1008
	int retval = 0;
L
Linus Torvalds 已提交
1009 1010 1011

	if (!file_count(filp)) {
		printk(KERN_ERR "VFS: Close: file count is 0\n");
1012
		return 0;
L
Linus Torvalds 已提交
1013 1014
	}

1015
	if (filp->f_op && filp->f_op->flush)
1016
		retval = filp->f_op->flush(filp, id);
L
Linus Torvalds 已提交
1017

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

	spin_lock(&files->file_lock);
1041 1042
	fdt = files_fdtable(files);
	if (fd >= fdt->max_fds)
L
Linus Torvalds 已提交
1043
		goto out_unlock;
1044
	filp = fdt->fd[fd];
L
Linus Torvalds 已提交
1045 1046
	if (!filp)
		goto out_unlock;
1047
	rcu_assign_pointer(fdt->fd[fd], NULL);
1048
	__clear_close_on_exec(fd, fdt);
L
Linus Torvalds 已提交
1049 1050
	__put_unused_fd(files, fd);
	spin_unlock(&files->file_lock);
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
	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
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071

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.
 */
1072
SYSCALL_DEFINE0(vhangup)
L
Linus Torvalds 已提交
1073 1074
{
	if (capable(CAP_SYS_TTY_CONFIG)) {
A
Alan Cox 已提交
1075
		tty_vhangup_self();
L
Linus Torvalds 已提交
1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
		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)
1090
		return -EOVERFLOW;
L
Linus Torvalds 已提交
1091 1092 1093 1094 1095 1096 1097
	return 0;
}

EXPORT_SYMBOL(generic_file_open);

/*
 * This is used by subsystems that don't want seekable
1098 1099 1100
 * 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
1101 1102 1103 1104 1105 1106 1107 1108
 */
int nonseekable_open(struct inode *inode, struct file *filp)
{
	filp->f_mode &= ~(FMODE_LSEEK | FMODE_PREAD | FMODE_PWRITE);
	return 0;
}

EXPORT_SYMBOL(nonseekable_open);