open.c 27.6 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10
/*
 *  linux/fs/open.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

#include <linux/string.h>
#include <linux/mm.h>
#include <linux/file.h>
#include <linux/quotaops.h>
Robert Love's avatar
Robert Love committed
11
#include <linux/fsnotify.h>
Linus Torvalds's avatar
Linus Torvalds committed
12 13 14 15 16
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/tty.h>
#include <linux/namei.h>
#include <linux/backing-dev.h>
17
#include <linux/capability.h>
Linus Torvalds's avatar
Linus Torvalds committed
18 19 20
#include <linux/security.h>
#include <linux/mount.h>
#include <linux/vfs.h>
21
#include <linux/fcntl.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>
Linus Torvalds's avatar
Linus Torvalds committed
30

31
int vfs_statfs(struct dentry *dentry, struct kstatfs *buf)
Linus Torvalds's avatar
Linus Torvalds committed
32 33 34
{
	int retval = -ENODEV;

35
	if (dentry) {
Linus Torvalds's avatar
Linus Torvalds committed
36
		retval = -ENOSYS;
37
		if (dentry->d_sb->s_op->statfs) {
Linus Torvalds's avatar
Linus Torvalds committed
38
			memset(buf, 0, sizeof(*buf));
39
			retval = security_sb_statfs(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
40 41
			if (retval)
				return retval;
42
			retval = dentry->d_sb->s_op->statfs(dentry, buf);
Linus Torvalds's avatar
Linus Torvalds committed
43 44 45 46 47 48 49 50 51
			if (retval == 0 && buf->f_frsize == 0)
				buf->f_frsize = buf->f_bsize;
		}
	}
	return retval;
}

EXPORT_SYMBOL(vfs_statfs);

52
static int vfs_statfs_native(struct dentry *dentry, struct statfs *buf)
Linus Torvalds's avatar
Linus Torvalds committed
53 54 55 56
{
	struct kstatfs st;
	int retval;

57
	retval = vfs_statfs(dentry, &st);
Linus Torvalds's avatar
Linus Torvalds committed
58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94
	if (retval)
		return retval;

	if (sizeof(*buf) == sizeof(st))
		memcpy(buf, &st, sizeof(st));
	else {
		if (sizeof buf->f_blocks == 4) {
			if ((st.f_blocks | st.f_bfree | st.f_bavail) &
			    0xffffffff00000000ULL)
				return -EOVERFLOW;
			/*
			 * f_files and f_ffree may be -1; it's okay to stuff
			 * that into 32 bits
			 */
			if (st.f_files != -1 &&
			    (st.f_files & 0xffffffff00000000ULL))
				return -EOVERFLOW;
			if (st.f_ffree != -1 &&
			    (st.f_ffree & 0xffffffff00000000ULL))
				return -EOVERFLOW;
		}

		buf->f_type = st.f_type;
		buf->f_bsize = st.f_bsize;
		buf->f_blocks = st.f_blocks;
		buf->f_bfree = st.f_bfree;
		buf->f_bavail = st.f_bavail;
		buf->f_files = st.f_files;
		buf->f_ffree = st.f_ffree;
		buf->f_fsid = st.f_fsid;
		buf->f_namelen = st.f_namelen;
		buf->f_frsize = st.f_frsize;
		memset(buf->f_spare, 0, sizeof(buf->f_spare));
	}
	return 0;
}

95
static int vfs_statfs64(struct dentry *dentry, struct statfs64 *buf)
Linus Torvalds's avatar
Linus Torvalds committed
96 97 98 99
{
	struct kstatfs st;
	int retval;

100
	retval = vfs_statfs(dentry, &st);
Linus Torvalds's avatar
Linus Torvalds committed
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129
	if (retval)
		return retval;

	if (sizeof(*buf) == sizeof(st))
		memcpy(buf, &st, sizeof(st));
	else {
		buf->f_type = st.f_type;
		buf->f_bsize = st.f_bsize;
		buf->f_blocks = st.f_blocks;
		buf->f_bfree = st.f_bfree;
		buf->f_bavail = st.f_bavail;
		buf->f_files = st.f_files;
		buf->f_ffree = st.f_ffree;
		buf->f_fsid = st.f_fsid;
		buf->f_namelen = st.f_namelen;
		buf->f_frsize = st.f_frsize;
		memset(buf->f_spare, 0, sizeof(buf->f_spare));
	}
	return 0;
}

asmlinkage long sys_statfs(const char __user * path, struct statfs __user * buf)
{
	struct nameidata nd;
	int error;

	error = user_path_walk(path, &nd);
	if (!error) {
		struct statfs tmp;
130
		error = vfs_statfs_native(nd.path.dentry, &tmp);
Linus Torvalds's avatar
Linus Torvalds committed
131 132
		if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
			error = -EFAULT;
Jan Blunck's avatar
Jan Blunck committed
133
		path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
	}
	return error;
}


asmlinkage long sys_statfs64(const char __user *path, size_t sz, struct statfs64 __user *buf)
{
	struct nameidata nd;
	long error;

	if (sz != sizeof(*buf))
		return -EINVAL;
	error = user_path_walk(path, &nd);
	if (!error) {
		struct statfs64 tmp;
149
		error = vfs_statfs64(nd.path.dentry, &tmp);
Linus Torvalds's avatar
Linus Torvalds committed
150 151
		if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
			error = -EFAULT;
Jan Blunck's avatar
Jan Blunck committed
152
		path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
153 154 155 156 157 158 159 160 161 162 163 164 165 166 167
	}
	return error;
}


asmlinkage long sys_fstatfs(unsigned int fd, struct statfs __user * buf)
{
	struct file * file;
	struct statfs tmp;
	int error;

	error = -EBADF;
	file = fget(fd);
	if (!file)
		goto out;
168
	error = vfs_statfs_native(file->f_path.dentry, &tmp);
Linus Torvalds's avatar
Linus Torvalds committed
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188
	if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
		error = -EFAULT;
	fput(file);
out:
	return error;
}

asmlinkage long sys_fstatfs64(unsigned int fd, size_t sz, struct statfs64 __user *buf)
{
	struct file * file;
	struct statfs64 tmp;
	int error;

	if (sz != sizeof(*buf))
		return -EINVAL;

	error = -EBADF;
	file = fget(fd);
	if (!file)
		goto out;
189
	error = vfs_statfs64(file->f_path.dentry, &tmp);
Linus Torvalds's avatar
Linus Torvalds committed
190 191 192 193 194 195 196
	if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
		error = -EFAULT;
	fput(file);
out:
	return error;
}

197 198
int do_truncate(struct dentry *dentry, loff_t length, unsigned int time_attrs,
	struct file *filp)
Linus Torvalds's avatar
Linus Torvalds committed
199 200 201 202 203 204 205 206 207
{
	int err;
	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;
208
	newattrs.ia_valid = ATTR_SIZE | time_attrs;
209 210 211 212
	if (filp) {
		newattrs.ia_file = filp;
		newattrs.ia_valid |= ATTR_FILE;
	}
Linus Torvalds's avatar
Linus Torvalds committed
213

214 215 216
	/* Remove suid/sgid on truncate too */
	newattrs.ia_valid |= should_remove_suid(dentry);

217
	mutex_lock(&dentry->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
218
	err = notify_change(dentry, &newattrs);
219
	mutex_unlock(&dentry->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
220 221 222
	return err;
}

223
static long do_sys_truncate(const char __user * path, loff_t length)
Linus Torvalds's avatar
Linus Torvalds committed
224 225 226 227 228 229 230 231 232 233 234 235
{
	struct nameidata nd;
	struct inode * inode;
	int error;

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

	error = user_path_walk(path, &nd);
	if (error)
		goto out;
236
	inode = nd.path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
237 238 239 240 241 242 243 244 245 246

	/* 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;

247
	error = mnt_want_write(nd.path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
248 249 250
	if (error)
		goto dput_and_out;

251 252 253
	error = vfs_permission(&nd, MAY_WRITE);
	if (error)
		goto mnt_drop_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
254 255 256

	error = -EPERM;
	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
257
		goto mnt_drop_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
258

259
	error = get_write_access(inode);
Linus Torvalds's avatar
Linus Torvalds committed
260
	if (error)
261
		goto mnt_drop_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
262

263 264 265 266 267
	/*
	 * Make sure that there are no leases.  get_write_access() protects
	 * against the truncate racing with a lease-granting setlease().
	 */
	error = break_lease(inode, FMODE_WRITE);
Linus Torvalds's avatar
Linus Torvalds committed
268
	if (error)
269
		goto put_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
270 271 272 273

	error = locks_verify_truncate(inode, NULL, length);
	if (!error) {
		DQUOT_INIT(inode);
274
		error = do_truncate(nd.path.dentry, length, 0, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
275 276
	}

277 278
put_write_and_out:
	put_write_access(inode);
279 280
mnt_drop_write_and_out:
	mnt_drop_write(nd.path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
281
dput_and_out:
Jan Blunck's avatar
Jan Blunck committed
282
	path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
283 284 285 286 287 288 289 290 291 292
out:
	return error;
}

asmlinkage long sys_truncate(const char __user * path, unsigned long length)
{
	/* on 32-bit boxen it will cut the range 2^31--2^32-1 off */
	return do_sys_truncate(path, (long)length);
}

293
static long do_sys_ftruncate(unsigned int fd, loff_t length, int small)
Linus Torvalds's avatar
Linus Torvalds committed
294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
{
	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;

312
	dentry = file->f_path.dentry;
Linus Torvalds's avatar
Linus Torvalds committed
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328
	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);
	if (!error)
329
		error = do_truncate(dentry, length, ATTR_MTIME|ATTR_CTIME, file);
Linus Torvalds's avatar
Linus Torvalds committed
330 331 332 333 334 335 336 337
out_putf:
	fput(file);
out:
	return error;
}

asmlinkage long sys_ftruncate(unsigned int fd, unsigned long length)
{
338
	long ret = do_sys_ftruncate(fd, length, 1);
339
	/* avoid REGPARM breakage on x86: */
340
	asmlinkage_protect(2, ret, fd, length);
341
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
342 343 344 345 346 347 348 349 350 351 352
}

/* LFS versions of truncate are only needed on 32 bit machines */
#if BITS_PER_LONG == 32
asmlinkage long sys_truncate64(const char __user * path, loff_t length)
{
	return do_sys_truncate(path, length);
}

asmlinkage long sys_ftruncate64(unsigned int fd, loff_t length)
{
353
	long ret = do_sys_ftruncate(fd, length, 0);
354
	/* avoid REGPARM breakage on x86: */
355
	asmlinkage_protect(2, ret, fd, length);
356
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
357 358 359
}
#endif

360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
asmlinkage long sys_fallocate(int fd, int mode, loff_t offset, loff_t len)
{
	struct file *file;
	struct inode *inode;
	long ret = -EINVAL;

	if (offset < 0 || len <= 0)
		goto out;

	/* Return error if mode is not supported */
	ret = -EOPNOTSUPP;
	if (mode && !(mode & FALLOC_FL_KEEP_SIZE))
		goto out;

	ret = -EBADF;
	file = fget(fd);
	if (!file)
		goto out;
	if (!(file->f_mode & FMODE_WRITE))
		goto out_fput;
	/*
	 * Revalidate the write permissions, in case security policy has
	 * changed since the files were opened.
	 */
	ret = security_file_permission(file, MAY_WRITE);
	if (ret)
		goto out_fput;

	inode = file->f_path.dentry->d_inode;

	ret = -ESPIPE;
	if (S_ISFIFO(inode->i_mode))
		goto out_fput;

	ret = -ENODEV;
	/*
	 * Let individual file system decide if it supports preallocation
	 * for directories or not.
	 */
	if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode))
		goto out_fput;

	ret = -EFBIG;
	/* Check for wrap through zero too */
	if (((offset + len) > inode->i_sb->s_maxbytes) || ((offset + len) < 0))
		goto out_fput;

	if (inode->i_op && inode->i_op->fallocate)
		ret = inode->i_op->fallocate(inode, mode, offset, len);
	else
410
		ret = -EOPNOTSUPP;
411 412 413 414 415 416 417

out_fput:
	fput(file);
out:
	return ret;
}

Linus Torvalds's avatar
Linus Torvalds committed
418 419 420 421 422
/*
 * 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.
 */
423
asmlinkage long sys_faccessat(int dfd, const char __user *filename, int mode)
Linus Torvalds's avatar
Linus Torvalds committed
424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452
{
	struct nameidata nd;
	int old_fsuid, old_fsgid;
	kernel_cap_t old_cap;
	int res;

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

	old_fsuid = current->fsuid;
	old_fsgid = current->fsgid;
	old_cap = current->cap_effective;

	current->fsuid = current->uid;
	current->fsgid = current->gid;

	/*
	 * Clear the capabilities if we switch to a non-root user
	 *
	 * FIXME: There is a race here against sys_capset.  The
	 * capabilities can change yet we will restore the old
	 * value below.  We should hold task_capabilities_lock,
	 * but we cannot because user_path_walk can sleep.
	 */
	if (current->uid)
		cap_clear(current->cap_effective);
	else
		current->cap_effective = current->cap_permitted;

453
	res = __user_walk_fd(dfd, filename, LOOKUP_FOLLOW|LOOKUP_ACCESS, &nd);
454 455 456 457 458 459
	if (res)
		goto out;

	res = vfs_permission(&nd, mode);
	/* SuS v2 requires we report a read only fs too */
	if(res || !(mode & S_IWOTH) ||
460
	   special_file(nd.path.dentry->d_inode->i_mode))
461
		goto out_path_release;
462 463 464 465 466 467 468 469 470 471 472
	/*
	 * 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.
	 */
	if (__mnt_is_readonly(nd.path.mnt))
473
		res = -EROFS;
Linus Torvalds's avatar
Linus Torvalds committed
474

475
out_path_release:
Jan Blunck's avatar
Jan Blunck committed
476
	path_put(&nd.path);
477
out:
Linus Torvalds's avatar
Linus Torvalds committed
478 479 480 481 482 483 484
	current->fsuid = old_fsuid;
	current->fsgid = old_fsgid;
	current->cap_effective = old_cap;

	return res;
}

485 486 487 488 489
asmlinkage long sys_access(const char __user *filename, int mode)
{
	return sys_faccessat(AT_FDCWD, filename, mode);
}

Linus Torvalds's avatar
Linus Torvalds committed
490 491 492 493 494
asmlinkage long sys_chdir(const char __user * filename)
{
	struct nameidata nd;
	int error;

495 496
	error = __user_walk(filename,
			    LOOKUP_FOLLOW|LOOKUP_DIRECTORY|LOOKUP_CHDIR, &nd);
Linus Torvalds's avatar
Linus Torvalds committed
497 498 499
	if (error)
		goto out;

500
	error = vfs_permission(&nd, MAY_EXEC);
Linus Torvalds's avatar
Linus Torvalds committed
501 502 503
	if (error)
		goto dput_and_out;

504
	set_fs_pwd(current->fs, &nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
505 506

dput_and_out:
Jan Blunck's avatar
Jan Blunck committed
507
	path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
out:
	return error;
}

asmlinkage long sys_fchdir(unsigned int fd)
{
	struct file *file;
	struct inode *inode;
	int error;

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

523
	inode = file->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
524 525 526 527 528

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

529
	error = file_permission(file, MAY_EXEC);
Linus Torvalds's avatar
Linus Torvalds committed
530
	if (!error)
531
		set_fs_pwd(current->fs, &file->f_path);
Linus Torvalds's avatar
Linus Torvalds committed
532 533 534 535 536 537 538 539 540 541 542 543 544 545 546
out_putf:
	fput(file);
out:
	return error;
}

asmlinkage long sys_chroot(const char __user * filename)
{
	struct nameidata nd;
	int error;

	error = __user_walk(filename, LOOKUP_FOLLOW | LOOKUP_DIRECTORY | LOOKUP_NOALT, &nd);
	if (error)
		goto out;

547
	error = vfs_permission(&nd, MAY_EXEC);
Linus Torvalds's avatar
Linus Torvalds committed
548 549 550 551 552 553 554
	if (error)
		goto dput_and_out;

	error = -EPERM;
	if (!capable(CAP_SYS_CHROOT))
		goto dput_and_out;

555
	set_fs_root(current->fs, &nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
556 557 558
	set_fs_altroot();
	error = 0;
dput_and_out:
Jan Blunck's avatar
Jan Blunck committed
559
	path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575
out:
	return error;
}

asmlinkage long sys_fchmod(unsigned int fd, mode_t mode)
{
	struct inode * inode;
	struct dentry * dentry;
	struct file * file;
	int err = -EBADF;
	struct iattr newattrs;

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

576
	dentry = file->f_path.dentry;
Linus Torvalds's avatar
Linus Torvalds committed
577 578
	inode = dentry->d_inode;

579
	audit_inode(NULL, dentry);
580

581 582
	err = mnt_want_write(file->f_path.mnt);
	if (err)
Linus Torvalds's avatar
Linus Torvalds committed
583 584 585
		goto out_putf;
	err = -EPERM;
	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
586
		goto out_drop_write;
587
	mutex_lock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
588 589 590 591 592
	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);
593
	mutex_unlock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
594

595 596
out_drop_write:
	mnt_drop_write(file->f_path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
597 598 599 600 601 602
out_putf:
	fput(file);
out:
	return err;
}

603 604
asmlinkage long sys_fchmodat(int dfd, const char __user *filename,
			     mode_t mode)
Linus Torvalds's avatar
Linus Torvalds committed
605 606 607 608 609 610
{
	struct nameidata nd;
	struct inode * inode;
	int error;
	struct iattr newattrs;

611
	error = __user_walk_fd(dfd, filename, LOOKUP_FOLLOW, &nd);
Linus Torvalds's avatar
Linus Torvalds committed
612 613
	if (error)
		goto out;
614
	inode = nd.path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
615

616 617
	error = mnt_want_write(nd.path.mnt);
	if (error)
Linus Torvalds's avatar
Linus Torvalds committed
618 619 620 621
		goto dput_and_out;

	error = -EPERM;
	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
622
		goto out_drop_write;
Linus Torvalds's avatar
Linus Torvalds committed
623

624
	mutex_lock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
625 626 627 628
	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;
629
	error = notify_change(nd.path.dentry, &newattrs);
630
	mutex_unlock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
631

632 633
out_drop_write:
	mnt_drop_write(nd.path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
634
dput_and_out:
Jan Blunck's avatar
Jan Blunck committed
635
	path_put(&nd.path);
Linus Torvalds's avatar
Linus Torvalds committed
636 637 638 639
out:
	return error;
}

640 641 642 643 644
asmlinkage long sys_chmod(const char __user *filename, mode_t mode)
{
	return sys_fchmodat(AT_FDCWD, filename, mode);
}

Linus Torvalds's avatar
Linus Torvalds committed
645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668
static int chown_common(struct dentry * dentry, uid_t user, gid_t group)
{
	struct inode * inode;
	int error;
	struct iattr newattrs;

	error = -ENOENT;
	if (!(inode = dentry->d_inode)) {
		printk(KERN_ERR "chown_common: NULL inode\n");
		goto out;
	}
	error = -EPERM;
	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
		goto out;
	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))
669 670
		newattrs.ia_valid |=
			ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
671
	mutex_lock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
672
	error = notify_change(dentry, &newattrs);
673
	mutex_unlock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
674 675 676 677 678 679 680 681 682 683
out:
	return error;
}

asmlinkage long sys_chown(const char __user * filename, uid_t user, gid_t group)
{
	struct nameidata nd;
	int error;

	error = user_path_walk(filename, &nd);
684 685
	if (error)
		goto out;
686 687 688
	error = mnt_want_write(nd.path.mnt);
	if (error)
		goto out_release;
689
	error = chown_common(nd.path.dentry, user, group);
690 691
	mnt_drop_write(nd.path.mnt);
out_release:
Jan Blunck's avatar
Jan Blunck committed
692
	path_put(&nd.path);
693
out:
Linus Torvalds's avatar
Linus Torvalds committed
694 695 696
	return error;
}

697 698 699 700 701 702 703 704 705 706 707 708
asmlinkage long sys_fchownat(int dfd, const char __user *filename, uid_t user,
			     gid_t group, int flag)
{
	struct nameidata nd;
	int error = -EINVAL;
	int follow;

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

	follow = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
	error = __user_walk_fd(dfd, filename, follow, &nd);
709 710
	if (error)
		goto out;
711 712 713
	error = mnt_want_write(nd.path.mnt);
	if (error)
		goto out_release;
714
	error = chown_common(nd.path.dentry, user, group);
715 716
	mnt_drop_write(nd.path.mnt);
out_release:
Jan Blunck's avatar
Jan Blunck committed
717
	path_put(&nd.path);
718 719 720 721
out:
	return error;
}

Linus Torvalds's avatar
Linus Torvalds committed
722 723 724 725 726 727
asmlinkage long sys_lchown(const char __user * filename, uid_t user, gid_t group)
{
	struct nameidata nd;
	int error;

	error = user_path_walk_link(filename, &nd);
728 729
	if (error)
		goto out;
730 731 732
	error = mnt_want_write(nd.path.mnt);
	if (error)
		goto out_release;
733
	error = chown_common(nd.path.dentry, user, group);
734 735
	mnt_drop_write(nd.path.mnt);
out_release:
Jan Blunck's avatar
Jan Blunck committed
736
	path_put(&nd.path);
737
out:
Linus Torvalds's avatar
Linus Torvalds committed
738 739 740 741 742 743 744 745
	return error;
}


asmlinkage long sys_fchown(unsigned int fd, uid_t user, gid_t group)
{
	struct file * file;
	int error = -EBADF;
746
	struct dentry * dentry;
Linus Torvalds's avatar
Linus Torvalds committed
747 748

	file = fget(fd);
749 750 751
	if (!file)
		goto out;

752 753 754
	error = mnt_want_write(file->f_path.mnt);
	if (error)
		goto out_fput;
755
	dentry = file->f_path.dentry;
756
	audit_inode(NULL, dentry);
757
	error = chown_common(dentry, user, group);
758 759
	mnt_drop_write(file->f_path.mnt);
out_fput:
760 761
	fput(file);
out:
Linus Torvalds's avatar
Linus Torvalds committed
762 763 764
	return error;
}

765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793
/*
 * 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;
}

794
static struct file *__dentry_open(struct dentry *dentry, struct vfsmount *mnt,
795 796
					int flags, struct file *f,
					int (*open)(struct inode *, struct file *))
Linus Torvalds's avatar
Linus Torvalds committed
797 798 799 800 801
{
	struct inode *inode;
	int error;

	f->f_flags = flags;
802 803
	f->f_mode = ((flags+1) & O_ACCMODE) | FMODE_LSEEK |
				FMODE_PREAD | FMODE_PWRITE;
Linus Torvalds's avatar
Linus Torvalds committed
804 805
	inode = dentry->d_inode;
	if (f->f_mode & FMODE_WRITE) {
806
		error = __get_file_write_access(inode, mnt);
Linus Torvalds's avatar
Linus Torvalds committed
807 808
		if (error)
			goto cleanup_file;
809 810
		if (!special_file(inode->i_mode))
			file_take_write(f);
Linus Torvalds's avatar
Linus Torvalds committed
811 812 813
	}

	f->f_mapping = inode->i_mapping;
814 815
	f->f_path.dentry = dentry;
	f->f_path.mnt = mnt;
Linus Torvalds's avatar
Linus Torvalds committed
816 817 818 819
	f->f_pos = 0;
	f->f_op = fops_get(inode->i_fop);
	file_move(f, &inode->i_sb->s_files);

820 821 822 823
	error = security_dentry_open(f);
	if (error)
		goto cleanup_all;

824 825 826 827
	if (!open && f->f_op)
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
Linus Torvalds's avatar
Linus Torvalds committed
828 829 830
		if (error)
			goto cleanup_all;
	}
831

Linus Torvalds's avatar
Linus Torvalds committed
832 833 834 835 836 837
	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) {
838 839 840
		if (!f->f_mapping->a_ops ||
		    ((!f->f_mapping->a_ops->direct_IO) &&
		    (!f->f_mapping->a_ops->get_xip_page))) {
Linus Torvalds's avatar
Linus Torvalds committed
841 842 843 844 845 846 847 848 849
			fput(f);
			f = ERR_PTR(-EINVAL);
		}
	}

	return f;

cleanup_all:
	fops_put(f->f_op);
850
	if (f->f_mode & FMODE_WRITE) {
Linus Torvalds's avatar
Linus Torvalds committed
851
		put_write_access(inode);
852 853 854 855 856 857 858 859
		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);
860
			mnt_drop_write(mnt);
861
		}
862
	}
Linus Torvalds's avatar
Linus Torvalds committed
863
	file_kill(f);
864 865
	f->f_path.dentry = NULL;
	f->f_path.mnt = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
866 867 868 869 870 871 872
cleanup_file:
	put_filp(f);
	dput(dentry);
	mntput(mnt);
	return ERR_PTR(error);
}

873 874 875 876 877 878 879 880 881 882
/**
 * 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.
883 884 885 886
 * 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).
887 888 889 890 891 892 893 894 895 896 897 898
 * 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 *))
{
	if (IS_ERR(nd->intent.open.file))
		goto out;
	if (IS_ERR(dentry))
		goto out_err;
899
	nd->intent.open.file = __dentry_open(dget(dentry), mntget(nd->path.mnt),
900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925
					     nd->intent.open.flags - 1,
					     nd->intent.open.file,
					     open);
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
 */
struct file *nameidata_to_filp(struct nameidata *nd, int flags)
{
	struct file *filp;

	/* Pick up the filp from the open intent */
	filp = nd->intent.open.file;
	/* Has the filesystem initialised the file for us? */
926
	if (filp->f_path.dentry == NULL)
927 928
		filp = __dentry_open(nd->path.dentry, nd->path.mnt, flags, filp,
				     NULL);
929
	else
Jan Blunck's avatar
Jan Blunck committed
930
		path_put(&nd->path);
931 932 933
	return filp;
}

934 935 936 937
/*
 * dentry_open() will have done dput(dentry) and mntput(mnt) if it returns an
 * error.
 */
938 939 940 941 942
struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags)
{
	int error;
	struct file *f;

943 944 945 946 947 948 949 950 951 952 953 954
	/*
	 * 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);
	}

955 956
	error = -ENFILE;
	f = get_empty_filp();
957 958 959
	if (f == NULL) {
		dput(dentry);
		mntput(mnt);
960
		return ERR_PTR(error);
961
	}
962

963
	return __dentry_open(dentry, mnt, flags, f, NULL);
964
}
Linus Torvalds's avatar
Linus Torvalds committed
965 966 967 968 969
EXPORT_SYMBOL(dentry_open);

/*
 * Find an empty file descriptor entry, and mark it busy.
 */
Ulrich Drepper's avatar
Ulrich Drepper committed
970
int get_unused_fd_flags(int flags)
Linus Torvalds's avatar
Linus Torvalds committed
971 972 973
{
	struct files_struct * files = current->files;
	int fd, error;
974
	struct fdtable *fdt;
Linus Torvalds's avatar
Linus Torvalds committed
975 976 977 978 979

  	error = -EMFILE;
	spin_lock(&files->file_lock);

repeat:
980
	fdt = files_fdtable(files);
981
	fd = find_next_zero_bit(fdt->open_fds->fds_bits, fdt->max_fds,
982
				files->next_fd);
Linus Torvalds's avatar
Linus Torvalds committed
983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004

	/*
	 * N.B. For clone tasks sharing a files structure, this test
	 * will limit the total number of files that can be opened.
	 */
	if (fd >= current->signal->rlim[RLIMIT_NOFILE].rlim_cur)
		goto out;

	/* Do we need to expand the fd array or fd set?  */
	error = expand_files(files, fd);
	if (error < 0)
		goto out;

	if (error) {
		/*
	 	 * If we needed to expand the fs array we
		 * might have blocked - try again.
		 */
		error = -EMFILE;
		goto repeat;
	}

1005
	FD_SET(fd, fdt->open_fds);
Ulrich Drepper's avatar
Ulrich Drepper committed
1006 1007 1008 1009
	if (flags & O_CLOEXEC)
		FD_SET(fd, fdt->close_on_exec);
	else
		FD_CLR(fd, fdt->close_on_exec);
1010
	files->next_fd = fd + 1;
Linus Torvalds's avatar
Linus Torvalds committed
1011 1012
#if 1
	/* Sanity check */
1013
	if (fdt->fd[fd] != NULL) {
Linus Torvalds's avatar
Linus Torvalds committed
1014
		printk(KERN_WARNING "get_unused_fd: slot %d not NULL!\n", fd);
1015
		fdt->fd[fd] = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1016 1017 1018 1019 1020 1021 1022 1023 1024
	}
#endif
	error = fd;

out:
	spin_unlock(&files->file_lock);
	return error;
}

Ulrich Drepper's avatar
Ulrich Drepper committed
1025 1026 1027 1028 1029
int get_unused_fd(void)
{
	return get_unused_fd_flags(0);
}

Linus Torvalds's avatar
Linus Torvalds committed
1030 1031
EXPORT_SYMBOL(get_unused_fd);

1032
static void __put_unused_fd(struct files_struct *files, unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
1033
{
1034 1035
	struct fdtable *fdt = files_fdtable(files);
	__FD_CLR(fd, fdt->open_fds);
1036 1037
	if (fd < files->next_fd)
		files->next_fd = fd;
Linus Torvalds's avatar
Linus Torvalds committed
1038 1039
}

1040
void put_unused_fd(unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
1041 1042 1043 1044 1045 1046 1047 1048 1049 1050
{
	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);

/*
1051
 * Install a file pointer in the fd array.
Linus Torvalds's avatar
Linus Torvalds committed
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
 *
 * 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.
 */

1063
void fd_install(unsigned int fd, struct file *file)
Linus Torvalds's avatar
Linus Torvalds committed
1064 1065
{
	struct files_struct *files = current->files;
1066
	struct fdtable *fdt;
Linus Torvalds's avatar
Linus Torvalds committed
1067
	spin_lock(&files->file_lock);
1068
	fdt = files_fdtable(files);
1069 1070
	BUG_ON(fdt->fd[fd] != NULL);
	rcu_assign_pointer(fdt->fd[fd], file);
Linus Torvalds's avatar
Linus Torvalds committed
1071 1072 1073 1074 1075
	spin_unlock(&files->file_lock);
}

EXPORT_SYMBOL(fd_install);

1076
long do_sys_open(int dfd, const char __user *filename, int flags, int mode)
Linus Torvalds's avatar
Linus Torvalds committed
1077
{
1078 1079
	char *tmp = getname(filename);
	int fd = PTR_ERR(tmp);
Linus Torvalds's avatar
Linus Torvalds committed
1080 1081

	if (!IS_ERR(tmp)) {
Ulrich Drepper's avatar
Ulrich Drepper committed
1082
		fd = get_unused_fd_flags(flags);
Linus Torvalds's avatar
Linus Torvalds committed
1083
		if (fd >= 0) {
1084
			struct file *f = do_filp_open(dfd, tmp, flags, mode);
1085 1086 1087 1088
			if (IS_ERR(f)) {
				put_unused_fd(fd);
				fd = PTR_ERR(f);
			} else {
1089
				fsnotify_open(f->f_path.dentry);
1090 1091
				fd_install(fd, f);
			}
Linus Torvalds's avatar
Linus Torvalds committed
1092 1093 1094 1095 1096
		}
		putname(tmp);
	}
	return fd;
}
1097 1098 1099

asmlinkage long sys_open(const char __user *filename, int flags, int mode)
{
1100 1101
	long ret;

1102 1103 1104
	if (force_o_largefile())
		flags |= O_LARGEFILE;

1105 1106
	ret = do_sys_open(AT_FDCWD, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
1107
	asmlinkage_protect(3, ret, filename, flags, mode);
1108
	return ret;
1109
}
Linus Torvalds's avatar
Linus Torvalds committed
1110

1111 1112 1113
asmlinkage long sys_openat(int dfd, const char __user *filename, int flags,
			   int mode)
{
1114 1115
	long ret;

1116 1117 1118
	if (force_o_largefile())
		flags |= O_LARGEFILE;

1119 1120
	ret = do_sys_open(dfd, filename, flags, mode);
	/* avoid REGPARM breakage on x86: */
1121
	asmlinkage_protect(4, ret, dfd, filename, flags, mode);
1122
	return ret;
1123 1124
}

Linus Torvalds's avatar
Linus Torvalds committed
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143
#ifndef __alpha__

/*
 * For backward compatibility?  Maybe this should be moved
 * into arch/i386 instead?
 */
asmlinkage long sys_creat(const char __user * pathname, int mode)
{
	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)
{
1144
	int retval = 0;
Linus Torvalds's avatar
Linus Torvalds committed
1145 1146 1147

	if (!file_count(filp)) {
		printk(KERN_ERR "VFS: Close: file count is 0\n");
1148
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
1149 1150
	}

1151
	if (filp->f_op && filp->f_op->flush)
1152
		retval = filp->f_op->flush(filp, id);
Linus Torvalds's avatar
Linus Torvalds committed
1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170

	dnotify_flush(filp, id);
	locks_remove_posix(filp, id);
	fput(filp);
	return retval;
}

EXPORT_SYMBOL(filp_close);

/*
 * Careful here! We test whether the file pointer is NULL before
 * releasing the fd. This ensures that one clone task can't release
 * an fd while another clone is opening it.
 */
asmlinkage long sys_close(unsigned int fd)
{
	struct file * filp;
	struct files_struct *files = current->files;
1171
	struct fdtable *fdt;
1172
	int retval;
Linus Torvalds's avatar
Linus Torvalds committed
1173 1174

	spin_lock(&files->file_lock);
1175 1176
	fdt = files_fdtable(files);
	if (fd >= fdt->max_fds)
Linus Torvalds's avatar
Linus Torvalds committed
1177
		goto out_unlock;
1178
	filp = fdt->fd[fd];
Linus Torvalds's avatar
Linus Torvalds committed
1179 1180
	if (!filp)
		goto out_unlock;
1181
	rcu_assign_pointer(fdt->fd[fd], NULL);
1182
	FD_CLR(fd, fdt->close_on_exec);
Linus Torvalds's avatar
Linus Torvalds committed
1183 1184
	__put_unused_fd(files, fd);
	spin_unlock(&files->file_lock);
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
	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;
Linus Torvalds's avatar
Linus Torvalds committed
1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209

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.
 */
asmlinkage long sys_vhangup(void)
{
	if (capable(CAP_SYS_TTY_CONFIG)) {
1210
		/* XXX: this needs locking */
Linus Torvalds's avatar
Linus Torvalds committed
1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
		tty_vhangup(current->signal->tty);
		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)
1226
		return -EOVERFLOW;
Linus Torvalds's avatar
Linus Torvalds committed
1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242
	return 0;
}

EXPORT_SYMBOL(generic_file_open);

/*
 * This is used by subsystems that don't want seekable
 * file descriptors
 */
int nonseekable_open(struct inode *inode, struct file *filp)
{
	filp->f_mode &= ~(FMODE_LSEEK | FMODE_PREAD | FMODE_PWRITE);
	return 0;
}

EXPORT_SYMBOL(nonseekable_open);