open.c 23.6 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
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>
Al Viro's avatar
Al Viro committed
10
#include <linux/fdtable.h>
Robert Love's avatar
Robert Love committed
11
#include <linux/fsnotify.h>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
18 19
#include <linux/security.h>
#include <linux/mount.h>
20
#include <linux/fcntl.h>
21
#include <linux/slab.h>
Linus Torvalds's avatar
Linus Torvalds committed
22 23
#include <asm/uaccess.h>
#include <linux/fs.h>
24
#include <linux/personality.h>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
33

34 35
#include "internal.h"

36 37
int do_truncate(struct dentry *dentry, loff_t length, unsigned int time_attrs,
	struct file *filp)
Linus Torvalds's avatar
Linus Torvalds committed
38
{
39
	int ret;
Linus Torvalds's avatar
Linus Torvalds committed
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;
	}
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
62 63
}

64
static long do_sys_truncate(const char __user *pathname, loff_t length)
Linus Torvalds's avatar
Linus Torvalds committed
65
{
66 67
	struct path path;
	struct inode *inode;
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
75 76
	if (error)
		goto out;
77
	inode = path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
95 96

	error = -EPERM;
97
	if (IS_APPEND(inode))
98
		goto mnt_drop_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
99

100
	error = get_write_access(inode);
Linus Torvalds's avatar
Linus Torvalds committed
101
	if (error)
102
		goto mnt_drop_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
109
	if (error)
110
		goto put_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
111 112

	error = locks_verify_truncate(inode, NULL, length);
113 114
	if (!error)
		error = security_path_truncate(&path, length, 0);
115
	if (!error)
116
		error = do_truncate(path.dentry, length, 0, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
117

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

128
SYSCALL_DEFINE2(truncate, const char __user *, path, long, length)
Linus Torvalds's avatar
Linus Torvalds committed
129
{
130
	return do_sys_truncate(path, length);
Linus Torvalds's avatar
Linus Torvalds committed
131 132
}

133
static long do_sys_ftruncate(unsigned int fd, loff_t length, int small)
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
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);
168 169 170
	if (!error)
		error = security_path_truncate(&file->f_path, length,
					       ATTR_MTIME|ATTR_CTIME);
Linus Torvalds's avatar
Linus Torvalds committed
171
	if (!error)
172
		error = do_truncate(dentry, length, ATTR_MTIME|ATTR_CTIME, file);
Linus Torvalds's avatar
Linus Torvalds committed
173 174 175 176 177 178
out_putf:
	fput(file);
out:
	return error;
}

179
SYSCALL_DEFINE2(ftruncate, unsigned int, fd, unsigned long, length)
Linus Torvalds's avatar
Linus Torvalds committed
180
{
181
	long ret = do_sys_ftruncate(fd, length, 1);
182
	/* avoid REGPARM breakage on x86: */
183
	asmlinkage_protect(2, ret, fd, length);
184
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
185 186 187 188
}

/* LFS versions of truncate are only needed on 32 bit machines */
#if BITS_PER_LONG == 32
189
SYSCALL_DEFINE(truncate64)(const char __user * path, loff_t length)
Linus Torvalds's avatar
Linus Torvalds committed
190 191 192
{
	return do_sys_truncate(path, length);
}
193 194 195 196 197 198 199
#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
Linus Torvalds's avatar
Linus Torvalds committed
200

201
SYSCALL_DEFINE(ftruncate64)(unsigned int fd, loff_t length)
Linus Torvalds's avatar
Linus Torvalds committed
202
{
203
	long ret = do_sys_ftruncate(fd, length, 0);
204
	/* avoid REGPARM breakage on x86: */
205
	asmlinkage_protect(2, ret, fd, length);
206
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
207
}
208 209 210 211 212 213
#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);
Linus Torvalds's avatar
Linus Torvalds committed
214
#endif
215
#endif /* BITS_PER_LONG == 32 */
Linus Torvalds's avatar
Linus Torvalds committed
216

217 218

int do_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
219
{
220 221
	struct inode *inode = file->f_path.dentry->d_inode;
	long ret;
222 223

	if (offset < 0 || len <= 0)
224
		return -EINVAL;
225 226 227

	/* Return error if mode is not supported */
	if (mode && !(mode & FALLOC_FL_KEEP_SIZE))
228
		return -EOPNOTSUPP;
229 230

	if (!(file->f_mode & FMODE_WRITE))
231
		return -EBADF;
232 233 234 235 236 237
	/*
	 * Revalidate the write permissions, in case security policy has
	 * changed since the files were opened.
	 */
	ret = security_file_permission(file, MAY_WRITE);
	if (ret)
238
		return ret;
239 240

	if (S_ISFIFO(inode->i_mode))
241
		return -ESPIPE;
242 243 244 245 246 247

	/*
	 * Let individual file system decide if it supports preallocation
	 * for directories or not.
	 */
	if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode))
248
		return -ENODEV;
249 250 251

	/* Check for wrap through zero too */
	if (((offset + len) > inode->i_sb->s_maxbytes) || ((offset + len) < 0))
252
		return -EFBIG;
253

254 255
	if (!inode->i_op->fallocate)
		return -EOPNOTSUPP;
256

257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
	return inode->i_op->fallocate(inode, mode, offset, len);
}

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;
272
}
273

274 275 276 277 278 279 280
#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
281

Linus Torvalds's avatar
Linus Torvalds committed
282 283 284 285 286
/*
 * 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.
 */
287
SYSCALL_DEFINE3(faccessat, int, dfd, const char __user *, filename, int, mode)
Linus Torvalds's avatar
Linus Torvalds committed
288
{
289 290
	const struct cred *old_cred;
	struct cred *override_cred;
291
	struct path path;
292
	struct inode *inode;
Linus Torvalds's avatar
Linus Torvalds committed
293 294 295 296 297
	int res;

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

298 299 300
	override_cred = prepare_creds();
	if (!override_cred)
		return -ENOMEM;
Linus Torvalds's avatar
Linus Torvalds committed
301

302 303
	override_cred->fsuid = override_cred->uid;
	override_cred->fsgid = override_cred->gid;
Linus Torvalds's avatar
Linus Torvalds committed
304

305
	if (!issecure(SECURE_NO_SETUID_FIXUP)) {
David Howells's avatar
David Howells committed
306
		/* Clear the capabilities if we switch to a non-root user */
307 308
		if (override_cred->uid)
			cap_clear(override_cred->cap_effective);
309
		else
310 311
			override_cred->cap_effective =
				override_cred->cap_permitted;
312
	}
Linus Torvalds's avatar
Linus Torvalds committed
313

314 315
	old_cred = override_creds(override_cred);

316
	res = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
317 318 319
	if (res)
		goto out;

320
	inode = path.dentry->d_inode;
321 322

	if ((mode & MAY_EXEC) && S_ISREG(inode->i_mode)) {
323 324 325 326 327
		/*
		 * MAY_EXEC on regular files is denied if the fs is mounted
		 * with the "noexec" flag.
		 */
		res = -EACCES;
328
		if (path.mnt->mnt_flags & MNT_NOEXEC)
329 330 331
			goto out_path_release;
	}

332
	res = inode_permission(inode, mode | MAY_ACCESS);
333
	/* SuS v2 requires we report a read only fs too */
334
	if (res || !(mode & S_IWOTH) || special_file(inode->i_mode))
335
		goto out_path_release;
336 337 338 339 340 341 342 343 344 345
	/*
	 * 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.
	 */
346
	if (__mnt_is_readonly(path.mnt))
347
		res = -EROFS;
Linus Torvalds's avatar
Linus Torvalds committed
348

349
out_path_release:
350
	path_put(&path);
351
out:
352 353
	revert_creds(old_cred);
	put_cred(override_cred);
Linus Torvalds's avatar
Linus Torvalds committed
354 355 356
	return res;
}

357
SYSCALL_DEFINE2(access, const char __user *, filename, int, mode)
358 359 360 361
{
	return sys_faccessat(AT_FDCWD, filename, mode);
}

362
SYSCALL_DEFINE1(chdir, const char __user *, filename)
Linus Torvalds's avatar
Linus Torvalds committed
363
{
364
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
365 366
	int error;

367
	error = user_path_dir(filename, &path);
Linus Torvalds's avatar
Linus Torvalds committed
368 369 370
	if (error)
		goto out;

371
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_ACCESS);
Linus Torvalds's avatar
Linus Torvalds committed
372 373 374
	if (error)
		goto dput_and_out;

375
	set_fs_pwd(current->fs, &path);
Linus Torvalds's avatar
Linus Torvalds committed
376 377

dput_and_out:
378
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
379 380 381 382
out:
	return error;
}

383
SYSCALL_DEFINE1(fchdir, unsigned int, fd)
Linus Torvalds's avatar
Linus Torvalds committed
384 385 386 387 388 389 390 391 392 393
{
	struct file *file;
	struct inode *inode;
	int error;

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

394
	inode = file->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
395 396 397 398 399

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

400
	error = inode_permission(inode, MAY_EXEC | MAY_ACCESS);
Linus Torvalds's avatar
Linus Torvalds committed
401
	if (!error)
402
		set_fs_pwd(current->fs, &file->f_path);
Linus Torvalds's avatar
Linus Torvalds committed
403 404 405 406 407 408
out_putf:
	fput(file);
out:
	return error;
}

409
SYSCALL_DEFINE1(chroot, const char __user *, filename)
Linus Torvalds's avatar
Linus Torvalds committed
410
{
411
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
412 413
	int error;

414
	error = user_path_dir(filename, &path);
Linus Torvalds's avatar
Linus Torvalds committed
415 416 417
	if (error)
		goto out;

418
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_ACCESS);
Linus Torvalds's avatar
Linus Torvalds committed
419 420 421 422 423 424
	if (error)
		goto dput_and_out;

	error = -EPERM;
	if (!capable(CAP_SYS_CHROOT))
		goto dput_and_out;
425 426 427
	error = security_path_chroot(&path);
	if (error)
		goto dput_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
428

429
	set_fs_root(current->fs, &path);
Linus Torvalds's avatar
Linus Torvalds committed
430 431
	error = 0;
dput_and_out:
432
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
433 434 435 436
out:
	return error;
}

437
SYSCALL_DEFINE2(fchmod, unsigned int, fd, mode_t, mode)
Linus Torvalds's avatar
Linus Torvalds committed
438 439 440 441 442 443 444 445 446 447 448
{
	struct inode * inode;
	struct dentry * dentry;
	struct file * file;
	int err = -EBADF;
	struct iattr newattrs;

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

449
	dentry = file->f_path.dentry;
Linus Torvalds's avatar
Linus Torvalds committed
450 451
	inode = dentry->d_inode;

452
	audit_inode(NULL, dentry);
453

454
	err = mnt_want_write_file(file);
455
	if (err)
Linus Torvalds's avatar
Linus Torvalds committed
456
		goto out_putf;
457
	mutex_lock(&inode->i_mutex);
458 459
	err = security_path_chmod(dentry, file->f_vfsmnt, mode);
	if (err)
460
		goto out_unlock;
Linus Torvalds's avatar
Linus Torvalds committed
461 462 463 464 465
	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);
466
out_unlock:
467
	mutex_unlock(&inode->i_mutex);
468
	mnt_drop_write(file->f_path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
469 470 471 472 473 474
out_putf:
	fput(file);
out:
	return err;
}

475
SYSCALL_DEFINE3(fchmodat, int, dfd, const char __user *, filename, mode_t, mode)
Linus Torvalds's avatar
Linus Torvalds committed
476
{
477 478
	struct path path;
	struct inode *inode;
Linus Torvalds's avatar
Linus Torvalds committed
479 480 481
	int error;
	struct iattr newattrs;

482
	error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
Linus Torvalds's avatar
Linus Torvalds committed
483 484
	if (error)
		goto out;
485
	inode = path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
486

487
	error = mnt_want_write(path.mnt);
488
	if (error)
Linus Torvalds's avatar
Linus Torvalds committed
489
		goto dput_and_out;
490
	mutex_lock(&inode->i_mutex);
491 492
	error = security_path_chmod(path.dentry, path.mnt, mode);
	if (error)
493
		goto out_unlock;
Linus Torvalds's avatar
Linus Torvalds committed
494 495 496 497
	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;
498
	error = notify_change(path.dentry, &newattrs);
499
out_unlock:
500
	mutex_unlock(&inode->i_mutex);
501
	mnt_drop_write(path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
502
dput_and_out:
503
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
504 505 506 507
out:
	return error;
}

508
SYSCALL_DEFINE2(chmod, const char __user *, filename, mode_t, mode)
509 510 511 512
{
	return sys_fchmodat(AT_FDCWD, filename, mode);
}

513
static int chown_common(struct path *path, uid_t user, gid_t group)
Linus Torvalds's avatar
Linus Torvalds committed
514
{
515
	struct inode *inode = path->dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
516 517 518 519 520 521 522 523 524 525 526 527 528
	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))
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

Linus Torvalds's avatar
Linus Torvalds committed
537 538 539
	return error;
}

540
SYSCALL_DEFINE3(chown, const char __user *, filename, uid_t, user, gid_t, group)
Linus Torvalds's avatar
Linus Torvalds committed
541
{
542
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
543 544
	int error;

545
	error = user_path(filename, &path);
546 547
	if (error)
		goto out;
548
	error = mnt_want_write(path.mnt);
549 550
	if (error)
		goto out_release;
551
	error = chown_common(&path, user, group);
552
	mnt_drop_write(path.mnt);
553
out_release:
554
	path_put(&path);
555
out:
Linus Torvalds's avatar
Linus Torvalds committed
556 557 558
	return error;
}

559 560
SYSCALL_DEFINE5(fchownat, int, dfd, const char __user *, filename, uid_t, user,
		gid_t, group, int, flag)
561
{
562
	struct path path;
563 564 565 566 567 568 569
	int error = -EINVAL;
	int follow;

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

	follow = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
570
	error = user_path_at(dfd, filename, follow, &path);
571 572
	if (error)
		goto out;
573
	error = mnt_want_write(path.mnt);
574 575
	if (error)
		goto out_release;
576
	error = chown_common(&path, user, group);
577
	mnt_drop_write(path.mnt);
578
out_release:
579
	path_put(&path);
580 581 582 583
out:
	return error;
}

584
SYSCALL_DEFINE3(lchown, const char __user *, filename, uid_t, user, gid_t, group)
Linus Torvalds's avatar
Linus Torvalds committed
585
{
586
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
587 588
	int error;

589
	error = user_lpath(filename, &path);
590 591
	if (error)
		goto out;
592
	error = mnt_want_write(path.mnt);
593 594
	if (error)
		goto out_release;
595
	error = chown_common(&path, user, group);
596
	mnt_drop_write(path.mnt);
597
out_release:
598
	path_put(&path);
599
out:
Linus Torvalds's avatar
Linus Torvalds committed
600 601 602
	return error;
}

603
SYSCALL_DEFINE3(fchown, unsigned int, fd, uid_t, user, gid_t, group)
Linus Torvalds's avatar
Linus Torvalds committed
604 605 606
{
	struct file * file;
	int error = -EBADF;
607
	struct dentry * dentry;
Linus Torvalds's avatar
Linus Torvalds committed
608 609

	file = fget(fd);
610 611 612
	if (!file)
		goto out;

613
	error = mnt_want_write_file(file);
614 615
	if (error)
		goto out_fput;
616
	dentry = file->f_path.dentry;
617
	audit_inode(NULL, dentry);
618
	error = chown_common(&file->f_path, user, group);
619 620
	mnt_drop_write(file->f_path.mnt);
out_fput:
621 622
	fput(file);
out:
Linus Torvalds's avatar
Linus Torvalds committed
623 624 625
	return error;
}

626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654
/*
 * 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;
}

655
static struct file *__dentry_open(struct dentry *dentry, struct vfsmount *mnt,
656
					struct file *f,
657 658
					int (*open)(struct inode *, struct file *),
					const struct cred *cred)
Linus Torvalds's avatar
Linus Torvalds committed
659 660 661 662
{
	struct inode *inode;
	int error;

Al Viro's avatar
Al Viro committed
663
	f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
664
				FMODE_PREAD | FMODE_PWRITE;
Linus Torvalds's avatar
Linus Torvalds committed
665 666
	inode = dentry->d_inode;
	if (f->f_mode & FMODE_WRITE) {
667
		error = __get_file_write_access(inode, mnt);
Linus Torvalds's avatar
Linus Torvalds committed
668 669
		if (error)
			goto cleanup_file;
670 671
		if (!special_file(inode->i_mode))
			file_take_write(f);
Linus Torvalds's avatar
Linus Torvalds committed
672 673 674
	}

	f->f_mapping = inode->i_mapping;
675 676
	f->f_path.dentry = dentry;
	f->f_path.mnt = mnt;
Linus Torvalds's avatar
Linus Torvalds committed
677 678 679 680
	f->f_pos = 0;
	f->f_op = fops_get(inode->i_fop);
	file_move(f, &inode->i_sb->s_files);

681
	error = security_dentry_open(f, cred);
682 683 684
	if (error)
		goto cleanup_all;

685 686 687 688
	if (!open && f->f_op)
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
Linus Torvalds's avatar
Linus Torvalds committed
689 690 691
		if (error)
			goto cleanup_all;
	}
692
	ima_counts_get(f);
693

Linus Torvalds's avatar
Linus Torvalds committed
694 695 696 697 698 699
	f->f_flags &= ~(O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC);

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

	/* NB: we're sure to have correct a_ops only after f_op->open */
	if (f->f_flags & O_DIRECT) {
700 701
		if (!f->f_mapping->a_ops ||
		    ((!f->f_mapping->a_ops->direct_IO) &&
702
		    (!f->f_mapping->a_ops->get_xip_mem))) {
Linus Torvalds's avatar
Linus Torvalds committed
703 704 705 706 707 708 709 710 711
			fput(f);
			f = ERR_PTR(-EINVAL);
		}
	}

	return f;

cleanup_all:
	fops_put(f->f_op);
712
	if (f->f_mode & FMODE_WRITE) {
Linus Torvalds's avatar
Linus Torvalds committed
713
		put_write_access(inode);
714 715 716 717 718 719 720 721
		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);
722
			mnt_drop_write(mnt);
723
		}
724
	}
Linus Torvalds's avatar
Linus Torvalds committed
725
	file_kill(f);
726 727
	f->f_path.dentry = NULL;
	f->f_path.mnt = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
728 729 730 731 732 733 734
cleanup_file:
	put_filp(f);
	dput(dentry);
	mntput(mnt);
	return ERR_PTR(error);
}

735 736 737 738 739 740 741 742 743 744
/**
 * 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.
745 746 747 748
 * 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).
749 750 751 752 753 754 755 756
 * 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 *))
{
757 758
	const struct cred *cred = current_cred();

759 760 761 762
	if (IS_ERR(nd->intent.open.file))
		goto out;
	if (IS_ERR(dentry))
		goto out_err;
763
	nd->intent.open.file = __dentry_open(dget(dentry), mntget(nd->path.mnt),
764
					     nd->intent.open.file,
765
					     open, cred);
766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781
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
 */
782
struct file *nameidata_to_filp(struct nameidata *nd)
783
{
784
	const struct cred *cred = current_cred();
785 786 787 788 789
	struct file *filp;

	/* Pick up the filp from the open intent */
	filp = nd->intent.open.file;
	/* Has the filesystem initialised the file for us? */
790
	if (filp->f_path.dentry == NULL)
791
		filp = __dentry_open(nd->path.dentry, nd->path.mnt, filp,
792
				     NULL, cred);
793
	else
Jan Blunck's avatar
Jan Blunck committed
794
		path_put(&nd->path);
795 796 797
	return filp;
}

798 799 800 801
/*
 * dentry_open() will have done dput(dentry) and mntput(mnt) if it returns an
 * error.
 */
802 803
struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags,
			 const struct cred *cred)
804 805 806 807
{
	int error;
	struct file *f;

808 809
	validate_creds(cred);

810 811 812 813 814 815 816 817 818 819 820 821
	/*
	 * 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);
	}

822 823
	error = -ENFILE;
	f = get_empty_filp();
824 825 826
	if (f == NULL) {
		dput(dentry);
		mntput(mnt);
827
		return ERR_PTR(error);
828
	}
829

830 831
	f->f_flags = flags;
	return __dentry_open(dentry, mnt, f, NULL, cred);
832
}
Linus Torvalds's avatar
Linus Torvalds committed
833 834
EXPORT_SYMBOL(dentry_open);

835
static void __put_unused_fd(struct files_struct *files, unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
836
{
837 838
	struct fdtable *fdt = files_fdtable(files);
	__FD_CLR(fd, fdt->open_fds);
839 840
	if (fd < files->next_fd)
		files->next_fd = fd;
Linus Torvalds's avatar
Linus Torvalds committed
841 842
}

843
void put_unused_fd(unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
844 845 846 847 848 849 850 851 852 853
{
	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);

/*
854
 * Install a file pointer in the fd array.
Linus Torvalds's avatar
Linus Torvalds committed
855 856 857 858 859 860 861 862 863 864 865
 *
 * 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.
 */

866
void fd_install(unsigned int fd, struct file *file)
Linus Torvalds's avatar
Linus Torvalds committed
867 868
{
	struct files_struct *files = current->files;
869
	struct fdtable *fdt;
Linus Torvalds's avatar
Linus Torvalds committed
870
	spin_lock(&files->file_lock);
871
	fdt = files_fdtable(files);
872 873
	BUG_ON(fdt->fd[fd] != NULL);
	rcu_assign_pointer(fdt->fd[fd], file);
Linus Torvalds's avatar
Linus Torvalds committed
874 875 876 877 878
	spin_unlock(&files->file_lock);
}

EXPORT_SYMBOL(fd_install);

879
long do_sys_open(int dfd, const char __user *filename, int flags, int mode)
Linus Torvalds's avatar
Linus Torvalds committed
880
{
881 882
	char *tmp = getname(filename);
	int fd = PTR_ERR(tmp);
Linus Torvalds's avatar
Linus Torvalds committed
883 884

	if (!IS_ERR(tmp)) {
Ulrich Drepper's avatar
Ulrich Drepper committed
885
		fd = get_unused_fd_flags(flags);
Linus Torvalds's avatar
Linus Torvalds committed
886
		if (fd >= 0) {
887
			struct file *f = do_filp_open(dfd, tmp, flags, mode, 0);
888 889 890 891
			if (IS_ERR(f)) {
				put_unused_fd(fd);
				fd = PTR_ERR(f);
			} else {
892
				fsnotify_open(f->f_path.dentry);
893 894
				fd_install(fd, f);
			}
Linus Torvalds's avatar
Linus Torvalds committed
895 896 897 898 899
		}
		putname(tmp);
	}
	return fd;
}
900

901
SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, int, mode)
902
{
903 904
	long ret;

905 906 907
	if (force_o_largefile())
		flags |= O_LARGEFILE;

908 909
	ret = do_sys_open(AT_FDCWD, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
910
	asmlinkage_protect(3, ret, filename, flags, mode);
911
	return ret;
912
}
Linus Torvalds's avatar
Linus Torvalds committed
913

914 915
SYSCALL_DEFINE4(openat, int, dfd, const char __user *, filename, int, flags,
		int, mode)
916
{
917 918
	long ret;

919 920 921
	if (force_o_largefile())
		flags |= O_LARGEFILE;

922 923
	ret = do_sys_open(dfd, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
924
	asmlinkage_protect(4, ret, dfd, filename, flags, mode);
925
	return ret;
926 927
}

Linus Torvalds's avatar
Linus Torvalds committed
928 929 930 931 932 933
#ifndef __alpha__

/*
 * For backward compatibility?  Maybe this should be moved
 * into arch/i386 instead?
 */
934
SYSCALL_DEFINE2(creat, const char __user *, pathname, int, mode)
Linus Torvalds's avatar
Linus Torvalds committed
935 936 937 938 939 940 941 942 943 944 945 946
{
	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)
{
947
	int retval = 0;
Linus Torvalds's avatar
Linus Torvalds committed
948 949 950

	if (!file_count(filp)) {
		printk(KERN_ERR "VFS: Close: file count is 0\n");
951
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
952 953
	}

954
	if (filp->f_op && filp->f_op->flush)