open.c 28.4 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
	if (!error)
114
		error = security_path_truncate(&path);
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
	if (!error)
169
		error = security_path_truncate(&file->f_path);
Linus Torvalds's avatar
Linus Torvalds committed
170
	if (!error)
171
		error = do_truncate(dentry, length, ATTR_MTIME|ATTR_CTIME, file);
Linus Torvalds's avatar
Linus Torvalds committed
172 173 174 175 176 177
out_putf:
	fput(file);
out:
	return error;
}

178
SYSCALL_DEFINE2(ftruncate, unsigned int, fd, unsigned long, length)
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
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)
Linus Torvalds's avatar
Linus Torvalds committed
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
Linus Torvalds's avatar
Linus Torvalds committed
199

200
SYSCALL_DEFINE(ftruncate64)(unsigned int fd, loff_t length)
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
213
#endif
214
#endif /* BITS_PER_LONG == 32 */
Linus Torvalds's avatar
Linus Torvalds committed
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 */
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

Linus Torvalds's avatar
Linus Torvalds committed
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)
Linus Torvalds's avatar
Linus Torvalds committed
300
{
301 302
	const struct cred *old_cred;
	struct cred *override_cred;
303
	struct path path;
304
	struct inode *inode;
Linus Torvalds's avatar
Linus Torvalds committed
305 306 307 308 309
	int res;

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

310 311 312
	override_cred = prepare_creds();
	if (!override_cred)
		return -ENOMEM;
Linus Torvalds's avatar
Linus Torvalds committed
313

314 315
	override_cred->fsuid = override_cred->uid;
	override_cred->fsgid = override_cred->gid;
Linus Torvalds's avatar
Linus Torvalds committed
316

317
	if (!issecure(SECURE_NO_SETUID_FIXUP)) {
David Howells's avatar
David Howells committed
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))
321
			cap_clear(override_cred->cap_effective);
322
		else
323 324
			override_cred->cap_effective =
				override_cred->cap_permitted;
325
	}
Linus Torvalds's avatar
Linus Torvalds committed
326

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;
Linus Torvalds's avatar
Linus Torvalds committed
361

362
out_path_release:
363
	path_put(&path);
364
out:
365 366
	revert_creds(old_cred);
	put_cred(override_cred);
Linus Torvalds's avatar
Linus Torvalds committed
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)
Linus Torvalds's avatar
Linus Torvalds committed
376
{
377
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
378 379
	int error;

380
	error = user_path_dir(filename, &path);
Linus Torvalds's avatar
Linus Torvalds committed
381 382 383
	if (error)
		goto out;

Eric Paris's avatar
Eric Paris committed
384
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
Linus Torvalds's avatar
Linus Torvalds committed
385 386 387
	if (error)
		goto dput_and_out;

388
	set_fs_pwd(current->fs, &path);
Linus Torvalds's avatar
Linus Torvalds committed
389 390

dput_and_out:
391
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
392 393 394 395
out:
	return error;
}

396
SYSCALL_DEFINE1(fchdir, unsigned int, fd)
Linus Torvalds's avatar
Linus Torvalds committed
397 398 399
{
	struct file *file;
	struct inode *inode;
400
	int error, fput_needed;
Linus Torvalds's avatar
Linus Torvalds committed
401 402

	error = -EBADF;
403
	file = fget_raw_light(fd, &fput_needed);
Linus Torvalds's avatar
Linus Torvalds committed
404 405 406
	if (!file)
		goto out;

407
	inode = file->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
408 409 410 411 412

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

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

422
SYSCALL_DEFINE1(chroot, const char __user *, filename)
Linus Torvalds's avatar
Linus Torvalds committed
423
{
424
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
425 426
	int error;

427
	error = user_path_dir(filename, &path);
Linus Torvalds's avatar
Linus Torvalds committed
428 429 430
	if (error)
		goto out;

Eric Paris's avatar
Eric Paris committed
431
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
Linus Torvalds's avatar
Linus Torvalds committed
432 433 434 435 436 437
	if (error)
		goto dput_and_out;

	error = -EPERM;
	if (!capable(CAP_SYS_CHROOT))
		goto dput_and_out;
438 439 440
	error = security_path_chroot(&path);
	if (error)
		goto dput_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
441

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

450
static int chmod_common(struct path *path, umode_t mode)
Linus Torvalds's avatar
Linus Torvalds committed
451
{
452
	struct inode *inode = path->dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
453
	struct iattr newattrs;
454
	int error;
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
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);
	}
Linus Torvalds's avatar
Linus Torvalds committed
483 484 485
	return err;
}

486
SYSCALL_DEFINE3(fchmodat, int, dfd, const char __user *, filename, umode_t, mode)
Linus Torvalds's avatar
Linus Torvalds committed
487
{
488
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
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);
	}
Linus Torvalds's avatar
Linus Torvalds committed
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)
Linus Torvalds's avatar
Linus Torvalds committed
505
{
506
	struct inode *inode = path->dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
514 515 516

	newattrs.ia_valid =  ATTR_CTIME;
	if (user != (uid_t) -1) {
517 518
		if (!uid_valid(uid))
			return -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
519
		newattrs.ia_valid |= ATTR_UID;
520
		newattrs.ia_uid = uid;
Linus Torvalds's avatar
Linus Torvalds committed
521 522
	}
	if (group != (gid_t) -1) {
523 524
		if (!gid_valid(gid))
			return -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
525
		newattrs.ia_valid |= ATTR_GID;
526
		newattrs.ia_gid = gid;
Linus Torvalds's avatar
Linus Torvalds committed
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

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
	int error = -EINVAL;
564
	int lookup_flags;
565

566
	if ((flag & ~(AT_SYMLINK_NOFOLLOW | AT_EMPTY_PATH)) != 0)
567 568
		goto out;

569 570 571 572
	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);
573 574
	if (error)
		goto out;
575
	error = mnt_want_write(path.mnt);
576 577
	if (error)
		goto out_release;
578
	error = chown_common(&path, user, group);
579
	mnt_drop_write(path.mnt);
580
out_release:
581
	path_put(&path);
582 583 584 585
out:
	return error;
}

586
SYSCALL_DEFINE3(lchown, const char __user *, filename, uid_t, user, gid_t, group)
Linus Torvalds's avatar
Linus Torvalds committed
587
{
588
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
589 590
	int error;

591
	error = user_lpath(filename, &path);
592 593
	if (error)
		goto out;
594
	error = mnt_want_write(path.mnt);
595 596
	if (error)
		goto out_release;
597
	error = chown_common(&path, user, group);
598
	mnt_drop_write(path.mnt);
599
out_release:
600
	path_put(&path);
601
out:
Linus Torvalds's avatar
Linus Torvalds committed
602 603 604
	return error;
}

605
SYSCALL_DEFINE3(fchown, unsigned int, fd, uid_t, user, gid_t, group)
Linus Torvalds's avatar
Linus Torvalds committed
606 607 608
{
	struct file * file;
	int error = -EBADF;
609
	struct dentry * dentry;
Linus Torvalds's avatar
Linus Torvalds committed
610 611

	file = fget(fd);
612 613 614
	if (!file)
		goto out;

615
	error = mnt_want_write_file(file);
616 617
	if (error)
		goto out_fput;
618
	dentry = file->f_path.dentry;
619
	audit_inode(NULL, dentry);
620
	error = chown_common(&file->f_path, user, group);
Al Viro's avatar
Al Viro committed
621
	mnt_drop_write_file(file);
622
out_fput:
623 624
	fput(file);
out:
Linus Torvalds's avatar
Linus Torvalds committed
625 626 627
	return error;
}

628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656
/*
 * 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;
}

Miklos Szeredi's avatar
Miklos Szeredi committed
657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673
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;
}

static struct file *do_dentry_open(struct dentry *dentry, struct vfsmount *mnt,
				   struct file *f,
				   int (*open)(struct inode *, struct file *),
				   const struct cred *cred)
Linus Torvalds's avatar
Linus Torvalds committed
674
{
675
	static const struct file_operations empty_fops = {};
Linus Torvalds's avatar
Linus Torvalds committed
676 677 678
	struct inode *inode;
	int error;

Al Viro's avatar
Al Viro committed
679
	f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
680
				FMODE_PREAD | FMODE_PWRITE;
681 682 683 684

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

Linus Torvalds's avatar
Linus Torvalds committed
685 686
	inode = dentry->d_inode;
	if (f->f_mode & FMODE_WRITE) {
687
		error = __get_file_write_access(inode, mnt);
Linus Torvalds's avatar
Linus Torvalds committed
688 689
		if (error)
			goto cleanup_file;
690 691
		if (!special_file(inode->i_mode))
			file_take_write(f);
Linus Torvalds's avatar
Linus Torvalds committed
692 693 694
	}

	f->f_mapping = inode->i_mapping;
695 696
	f->f_path.dentry = dentry;
	f->f_path.mnt = mnt;
Linus Torvalds's avatar
Linus Torvalds committed
697
	f->f_pos = 0;
Nick Piggin's avatar
Nick Piggin committed
698
	file_sb_list_add(f, inode->i_sb);
Linus Torvalds's avatar
Linus Torvalds committed
699

700 701 702 703 704 705 706
	if (unlikely(f->f_mode & FMODE_PATH)) {
		f->f_op = &empty_fops;
		return f;
	}

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

707
	error = security_file_open(f, cred);
708 709 710
	if (error)
		goto cleanup_all;

711 712 713 714
	error = break_lease(inode, f->f_flags);
	if (error)
		goto cleanup_all;

715 716 717 718
	if (!open && f->f_op)
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
Linus Torvalds's avatar
Linus Torvalds committed
719 720 721
		if (error)
			goto cleanup_all;
	}
722 723
	if ((f->f_mode & (FMODE_READ | FMODE_WRITE)) == FMODE_READ)
		i_readcount_inc(inode);
724

Linus Torvalds's avatar
Linus Torvalds committed
725 726 727 728 729 730 731 732
	f->f_flags &= ~(O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC);

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

	return f;

cleanup_all:
	fops_put(f->f_op);
733
	if (f->f_mode & FMODE_WRITE) {
Linus Torvalds's avatar
Linus Torvalds committed
734
		put_write_access(inode);
735 736 737 738 739 740 741 742
		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);
743
			mnt_drop_write(mnt);
744
		}
745
	}
Nick Piggin's avatar
Nick Piggin committed
746
	file_sb_list_del(f);
747 748
	f->f_path.dentry = NULL;
	f->f_path.mnt = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
749 750 751 752 753 754
cleanup_file:
	dput(dentry);
	mntput(mnt);
	return ERR_PTR(error);
}

Miklos Szeredi's avatar
Miklos Szeredi committed
755 756 757 758 759 760 761 762 763 764 765 766
static struct file *__dentry_open(struct dentry *dentry, struct vfsmount *mnt,
				struct file *f,
				int (*open)(struct inode *, struct file *),
				const struct cred *cred)
{
	struct file *res = do_dentry_open(dentry, mnt, f, open, cred);
	if (!IS_ERR(res)) {
		int error = open_check_o_direct(f);
		if (error) {
			fput(res);
			res = ERR_PTR(error);
		}
767 768
	} else {
		put_filp(f);
Miklos Szeredi's avatar
Miklos Szeredi committed
769 770 771 772
	}
	return res;
}

773 774 775 776 777 778 779 780 781 782
/**
 * 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.
783 784 785 786
 * 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).
787 788 789 790 791 792 793 794
 * 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 *))
{
795 796
	const struct cred *cred = current_cred();

797 798 799 800
	if (IS_ERR(nd->intent.open.file))
		goto out;
	if (IS_ERR(dentry))
		goto out_err;
801
	nd->intent.open.file = __dentry_open(dget(dentry), mntget(nd->path.mnt),
802
					     nd->intent.open.file,
803
					     open, cred);
804 805 806 807
out:
	return nd->intent.open.file;
out_err:
	release_open_intent(nd);
808
	nd->intent.open.file = ERR_CAST(dentry);
809 810 811 812
	goto out;
}
EXPORT_SYMBOL_GPL(lookup_instantiate_filp);

Miklos Szeredi's avatar
Miklos Szeredi committed
813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854
/**
 * 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.
 */
struct file *finish_open(struct opendata *od, struct dentry *dentry,
			 int (*open)(struct inode *, struct file *))
{
	struct file *res;

	mntget(od->mnt);
	dget(dentry);

	res = do_dentry_open(dentry, od->mnt, *od->filp, open, current_cred());
	if (!IS_ERR(res))
		*od->filp = NULL;

	return res;
}
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.
 */
void finish_no_open(struct opendata *od, struct dentry *dentry)
{
	od->dentry = dentry;
}
EXPORT_SYMBOL(finish_no_open);

855 856 857 858 859 860 861
/**
 * 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
 */
862
struct file *nameidata_to_filp(struct nameidata *nd)
863
{
864
	const struct cred *cred = current_cred();
865 866 867 868
	struct file *filp;

	/* Pick up the filp from the open intent */
	filp = nd->intent.open.file;
869

870
	/* Has the filesystem initialised the file for us? */
871 872 873
	if (filp->f_path.dentry != NULL) {
		nd->intent.open.file = NULL;
	} else {
874 875
		struct file *res;

Al Viro's avatar
Al Viro committed
876
		path_get(&nd->path);
877 878 879 880 881
		res = do_dentry_open(nd->path.dentry, nd->path.mnt,
				     filp, NULL, cred);
		if (!IS_ERR(res)) {
			int error;

882
			nd->intent.open.file = NULL;
883 884 885 886 887 888 889 890
			BUG_ON(res != filp);

			error = open_check_o_direct(filp);
			if (error) {
				fput(filp);
				filp = ERR_PTR(error);
			}
		} else {
891
			/* Allow nd->intent.open.file to be recycled */
892 893
			filp = res;
		}
Al Viro's avatar
Al Viro committed
894
	}
895 896 897
	return filp;
}

898 899 900 901
/*
 * dentry_open() will have done dput(dentry) and mntput(mnt) if it returns an
 * error.
 */
902 903
struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags,
			 const struct cred *cred)
904 905 906 907
{
	int error;
	struct file *f;

908 909
	validate_creds(cred);

910 911
	/* We must always pass in a valid mount pointer. */
	BUG_ON(!mnt);
912

913 914
	error = -ENFILE;
	f = get_empty_filp();
915 916 917
	if (f == NULL) {
		dput(dentry);
		mntput(mnt);
918
		return ERR_PTR(error);
919
	}
920

921 922
	f->f_flags = flags;
	return __dentry_open(dentry, mnt, f, NULL, cred);
923
}
Linus Torvalds's avatar
Linus Torvalds committed
924 925
EXPORT_SYMBOL(dentry_open);

926
static void __put_unused_fd(struct files_struct *files, unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
927
{
928
	struct fdtable *fdt = files_fdtable(files);
929
	__clear_open_fd(fd, fdt);
930 931
	if (fd < files->next_fd)
		files->next_fd = fd;
Linus Torvalds's avatar
Linus Torvalds committed
932 933
}

934
void put_unused_fd(unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
935 936 937 938 939 940 941 942 943 944
{
	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);

/*
945
 * Install a file pointer in the fd array.
Linus Torvalds's avatar
Linus Torvalds committed
946 947 948 949 950 951 952 953 954 955 956
 *
 * 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.
 */

957
void fd_install(unsigned int fd, struct file *file)
Linus Torvalds's avatar
Linus Torvalds committed
958 959
{
	struct files_struct *files = current->files;
960
	struct fdtable *fdt;
Linus Torvalds's avatar
Linus Torvalds committed
961
	spin_lock(&files->file_lock);
962
	fdt = files_fdtable(files);
963 964
	BUG_ON(fdt->fd[fd] != NULL);
	rcu_assign_pointer(fdt->fd[fd], file);
Linus Torvalds's avatar
Linus Torvalds committed
965 966 967 968 969
	spin_unlock(&files->file_lock);
}

EXPORT_SYMBOL(fd_install);