[GITLAB] - A technical upgrade is planned on Thursday the 1st of July at noon on our GITLAB server.

open.c 26.9 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>
Linus Torvalds's avatar
Linus Torvalds committed
11
#include <linux/quotaops.h>
Robert Love's avatar
Robert Love committed
12
#include <linux/fsnotify.h>
Linus Torvalds's avatar
Linus Torvalds committed
13 14 15 16 17
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/tty.h>
#include <linux/namei.h>
#include <linux/backing-dev.h>
18
#include <linux/capability.h>
19
#include <linux/securebits.h>
Linus Torvalds's avatar
Linus Torvalds committed
20 21 22
#include <linux/security.h>
#include <linux/mount.h>
#include <linux/vfs.h>
23
#include <linux/fcntl.h>
Linus Torvalds's avatar
Linus Torvalds committed
24 25
#include <asm/uaccess.h>
#include <linux/fs.h>
26
#include <linux/personality.h>
Linus Torvalds's avatar
Linus Torvalds committed
27 28
#include <linux/pagemap.h>
#include <linux/syscalls.h>
29
#include <linux/rcupdate.h>
30
#include <linux/audit.h>
31
#include <linux/falloc.h>
Linus Torvalds's avatar
Linus Torvalds committed
32

33
int vfs_statfs(struct dentry *dentry, struct kstatfs *buf)
Linus Torvalds's avatar
Linus Torvalds committed
34 35 36
{
	int retval = -ENODEV;

37
	if (dentry) {
Linus Torvalds's avatar
Linus Torvalds committed
38
		retval = -ENOSYS;
39
		if (dentry->d_sb->s_op->statfs) {
Linus Torvalds's avatar
Linus Torvalds committed
40
			memset(buf, 0, sizeof(*buf));
41
			retval = security_sb_statfs(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
42 43
			if (retval)
				return retval;
44
			retval = dentry->d_sb->s_op->statfs(dentry, buf);
Linus Torvalds's avatar
Linus Torvalds committed
45 46 47 48 49 50 51 52 53
			if (retval == 0 && buf->f_frsize == 0)
				buf->f_frsize = buf->f_bsize;
		}
	}
	return retval;
}

EXPORT_SYMBOL(vfs_statfs);

54
static int vfs_statfs_native(struct dentry *dentry, struct statfs *buf)
Linus Torvalds's avatar
Linus Torvalds committed
55 56 57 58
{
	struct kstatfs st;
	int retval;

59
	retval = vfs_statfs(dentry, &st);
Linus Torvalds's avatar
Linus Torvalds committed
60 61 62 63 64 65 66
	if (retval)
		return retval;

	if (sizeof(*buf) == sizeof(st))
		memcpy(buf, &st, sizeof(st));
	else {
		if (sizeof buf->f_blocks == 4) {
Jon Tollefson's avatar
Jon Tollefson committed
67 68
			if ((st.f_blocks | st.f_bfree | st.f_bavail |
			     st.f_bsize | st.f_frsize) &
Linus Torvalds's avatar
Linus Torvalds committed
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 95 96 97
			    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;
}

98
static int vfs_statfs64(struct dentry *dentry, struct statfs64 *buf)
Linus Torvalds's avatar
Linus Torvalds committed
99 100 101 102
{
	struct kstatfs st;
	int retval;

103
	retval = vfs_statfs(dentry, &st);
Linus Torvalds's avatar
Linus Torvalds committed
104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
	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;
}

125
SYSCALL_DEFINE2(statfs, const char __user *, pathname, struct statfs __user *, buf)
Linus Torvalds's avatar
Linus Torvalds committed
126
{
127
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
128 129
	int error;

130
	error = user_path(pathname, &path);
Linus Torvalds's avatar
Linus Torvalds committed
131 132
	if (!error) {
		struct statfs tmp;
133
		error = vfs_statfs_native(path.dentry, &tmp);
Linus Torvalds's avatar
Linus Torvalds committed
134 135
		if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
			error = -EFAULT;
136
		path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
137 138 139 140
	}
	return error;
}

141
SYSCALL_DEFINE3(statfs64, const char __user *, pathname, size_t, sz, struct statfs64 __user *, buf)
Linus Torvalds's avatar
Linus Torvalds committed
142
{
143
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
144 145 146 147
	long error;

	if (sz != sizeof(*buf))
		return -EINVAL;
148
	error = user_path(pathname, &path);
Linus Torvalds's avatar
Linus Torvalds committed
149 150
	if (!error) {
		struct statfs64 tmp;
151
		error = vfs_statfs64(path.dentry, &tmp);
Linus Torvalds's avatar
Linus Torvalds committed
152 153
		if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
			error = -EFAULT;
154
		path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
155 156 157 158
	}
	return error;
}

159
SYSCALL_DEFINE2(fstatfs, unsigned int, fd, struct statfs __user *, buf)
Linus Torvalds's avatar
Linus Torvalds committed
160 161 162 163 164 165 166 167 168
{
	struct file * file;
	struct statfs tmp;
	int error;

	error = -EBADF;
	file = fget(fd);
	if (!file)
		goto out;
169
	error = vfs_statfs_native(file->f_path.dentry, &tmp);
Linus Torvalds's avatar
Linus Torvalds committed
170 171 172 173 174 175 176
	if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
		error = -EFAULT;
	fput(file);
out:
	return error;
}

177
SYSCALL_DEFINE3(fstatfs64, unsigned int, fd, size_t, sz, struct statfs64 __user *, buf)
Linus Torvalds's avatar
Linus Torvalds committed
178 179 180 181 182 183 184 185 186 187 188 189
{
	struct file * file;
	struct statfs64 tmp;
	int error;

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

	error = -EBADF;
	file = fget(fd);
	if (!file)
		goto out;
190
	error = vfs_statfs64(file->f_path.dentry, &tmp);
Linus Torvalds's avatar
Linus Torvalds committed
191 192 193 194 195 196 197
	if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
		error = -EFAULT;
	fput(file);
out:
	return error;
}

198 199
int do_truncate(struct dentry *dentry, loff_t length, unsigned int time_attrs,
	struct file *filp)
Linus Torvalds's avatar
Linus Torvalds committed
200 201 202 203 204 205 206 207 208
{
	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;
209
	newattrs.ia_valid = ATTR_SIZE | time_attrs;
210 211 212 213
	if (filp) {
		newattrs.ia_file = filp;
		newattrs.ia_valid |= ATTR_FILE;
	}
Linus Torvalds's avatar
Linus Torvalds committed
214

215 216 217
	/* Remove suid/sgid on truncate too */
	newattrs.ia_valid |= should_remove_suid(dentry);

218
	mutex_lock(&dentry->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
219
	err = notify_change(dentry, &newattrs);
220
	mutex_unlock(&dentry->d_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
221 222 223
	return err;
}

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

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

234
	error = user_path(pathname, &path);
Linus Torvalds's avatar
Linus Torvalds committed
235 236
	if (error)
		goto out;
237
	inode = path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
238 239 240 241 242 243 244 245 246 247

	/* 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;

248
	error = mnt_want_write(path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
249 250 251
	if (error)
		goto dput_and_out;

252
	error = inode_permission(inode, MAY_WRITE);
253 254
	if (error)
		goto mnt_drop_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
255 256

	error = -EPERM;
257
	if (IS_APPEND(inode))
258
		goto mnt_drop_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
259

260
	error = get_write_access(inode);
Linus Torvalds's avatar
Linus Torvalds committed
261
	if (error)
262
		goto mnt_drop_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
263

264 265 266 267 268
	/*
	 * 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
269
	if (error)
270
		goto put_write_and_out;
Linus Torvalds's avatar
Linus Torvalds committed
271 272

	error = locks_verify_truncate(inode, NULL, length);
273 274
	if (!error)
		error = security_path_truncate(&path, length, 0);
Linus Torvalds's avatar
Linus Torvalds committed
275 276
	if (!error) {
		DQUOT_INIT(inode);
277
		error = do_truncate(path.dentry, length, 0, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
278 279
	}

280 281
put_write_and_out:
	put_write_access(inode);
282
mnt_drop_write_and_out:
283
	mnt_drop_write(path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
284
dput_and_out:
285
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
286 287 288 289
out:
	return error;
}

290
SYSCALL_DEFINE2(truncate, const char __user *, path, unsigned long, length)
Linus Torvalds's avatar
Linus Torvalds committed
291 292 293 294 295
{
	/* on 32-bit boxen it will cut the range 2^31--2^32-1 off */
	return do_sys_truncate(path, (long)length);
}

296
static long do_sys_ftruncate(unsigned int fd, loff_t length, int small)
Linus Torvalds's avatar
Linus Torvalds committed
297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314
{
	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;

315
	dentry = file->f_path.dentry;
Linus Torvalds's avatar
Linus Torvalds committed
316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
	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);
331 332 333
	if (!error)
		error = security_path_truncate(&file->f_path, length,
					       ATTR_MTIME|ATTR_CTIME);
Linus Torvalds's avatar
Linus Torvalds committed
334
	if (!error)
335
		error = do_truncate(dentry, length, ATTR_MTIME|ATTR_CTIME, file);
Linus Torvalds's avatar
Linus Torvalds committed
336 337 338 339 340 341
out_putf:
	fput(file);
out:
	return error;
}

342
SYSCALL_DEFINE2(ftruncate, unsigned int, fd, unsigned long, length)
Linus Torvalds's avatar
Linus Torvalds committed
343
{
344
	long ret = do_sys_ftruncate(fd, length, 1);
345
	/* avoid REGPARM breakage on x86: */
346
	asmlinkage_protect(2, ret, fd, length);
347
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
348 349 350 351
}

/* LFS versions of truncate are only needed on 32 bit machines */
#if BITS_PER_LONG == 32
352
SYSCALL_DEFINE(truncate64)(const char __user * path, loff_t length)
Linus Torvalds's avatar
Linus Torvalds committed
353 354 355
{
	return do_sys_truncate(path, length);
}
356 357 358 359 360 361 362
#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
363

364
SYSCALL_DEFINE(ftruncate64)(unsigned int fd, loff_t length)
Linus Torvalds's avatar
Linus Torvalds committed
365
{
366
	long ret = do_sys_ftruncate(fd, length, 0);
367
	/* avoid REGPARM breakage on x86: */
368
	asmlinkage_protect(2, ret, fd, length);
369
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
370
}
371 372 373 374 375 376
#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
377
#endif
378
#endif /* BITS_PER_LONG == 32 */
Linus Torvalds's avatar
Linus Torvalds committed
379

380
SYSCALL_DEFINE(fallocate)(int fd, int mode, loff_t offset, loff_t len)
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426
{
	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;

Al Viro's avatar
Al Viro committed
427
	if (inode->i_op->fallocate)
428 429
		ret = inode->i_op->fallocate(inode, mode, offset, len);
	else
430
		ret = -EOPNOTSUPP;
431 432 433 434 435 436

out_fput:
	fput(file);
out:
	return ret;
}
437 438 439 440 441 442 443
#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
444

Linus Torvalds's avatar
Linus Torvalds committed
445 446 447 448 449
/*
 * 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.
 */
450
asmlinkage long sys_faccessat(int dfd, const char __user *filename, int mode)
Linus Torvalds's avatar
Linus Torvalds committed
451
{
452 453
	const struct cred *old_cred;
	struct cred *override_cred;
454
	struct path path;
455
	struct inode *inode;
Linus Torvalds's avatar
Linus Torvalds committed
456 457 458 459 460
	int res;

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

461 462 463
	override_cred = prepare_creds();
	if (!override_cred)
		return -ENOMEM;
Linus Torvalds's avatar
Linus Torvalds committed
464

465 466
	override_cred->fsuid = override_cred->uid;
	override_cred->fsgid = override_cred->gid;
Linus Torvalds's avatar
Linus Torvalds committed
467

468
	if (!issecure(SECURE_NO_SETUID_FIXUP)) {
David Howells's avatar
David Howells committed
469
		/* Clear the capabilities if we switch to a non-root user */
470 471
		if (override_cred->uid)
			cap_clear(override_cred->cap_effective);
472
		else
473 474
			override_cred->cap_effective =
				override_cred->cap_permitted;
475
	}
Linus Torvalds's avatar
Linus Torvalds committed
476

477 478
	old_cred = override_creds(override_cred);

479
	res = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
480 481 482
	if (res)
		goto out;

483
	inode = path.dentry->d_inode;
484 485

	if ((mode & MAY_EXEC) && S_ISREG(inode->i_mode)) {
486 487 488 489 490
		/*
		 * MAY_EXEC on regular files is denied if the fs is mounted
		 * with the "noexec" flag.
		 */
		res = -EACCES;
491
		if (path.mnt->mnt_flags & MNT_NOEXEC)
492 493 494
			goto out_path_release;
	}

495
	res = inode_permission(inode, mode | MAY_ACCESS);
496
	/* SuS v2 requires we report a read only fs too */
497
	if (res || !(mode & S_IWOTH) || special_file(inode->i_mode))
498
		goto out_path_release;
499 500 501 502 503 504 505 506 507 508
	/*
	 * 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.
	 */
509
	if (__mnt_is_readonly(path.mnt))
510
		res = -EROFS;
Linus Torvalds's avatar
Linus Torvalds committed
511

512
out_path_release:
513
	path_put(&path);
514
out:
515 516
	revert_creds(old_cred);
	put_cred(override_cred);
Linus Torvalds's avatar
Linus Torvalds committed
517 518 519
	return res;
}

520
SYSCALL_DEFINE2(access, const char __user *, filename, int, mode)
521 522 523 524
{
	return sys_faccessat(AT_FDCWD, filename, mode);
}

525
SYSCALL_DEFINE1(chdir, const char __user *, filename)
Linus Torvalds's avatar
Linus Torvalds committed
526
{
527
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
528 529
	int error;

530
	error = user_path_dir(filename, &path);
Linus Torvalds's avatar
Linus Torvalds committed
531 532 533
	if (error)
		goto out;

534
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_ACCESS);
Linus Torvalds's avatar
Linus Torvalds committed
535 536 537
	if (error)
		goto dput_and_out;

538
	set_fs_pwd(current->fs, &path);
Linus Torvalds's avatar
Linus Torvalds committed
539 540

dput_and_out:
541
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
542 543 544 545
out:
	return error;
}

546
SYSCALL_DEFINE1(fchdir, unsigned int, fd)
Linus Torvalds's avatar
Linus Torvalds committed
547 548 549 550 551 552 553 554 555 556
{
	struct file *file;
	struct inode *inode;
	int error;

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

557
	inode = file->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
558 559 560 561 562

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

563
	error = inode_permission(inode, MAY_EXEC | MAY_ACCESS);
Linus Torvalds's avatar
Linus Torvalds committed
564
	if (!error)
565
		set_fs_pwd(current->fs, &file->f_path);
Linus Torvalds's avatar
Linus Torvalds committed
566 567 568 569 570 571
out_putf:
	fput(file);
out:
	return error;
}

572
SYSCALL_DEFINE1(chroot, const char __user *, filename)
Linus Torvalds's avatar
Linus Torvalds committed
573
{
574
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
575 576
	int error;

577
	error = user_path_dir(filename, &path);
Linus Torvalds's avatar
Linus Torvalds committed
578 579 580
	if (error)
		goto out;

581
	error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_ACCESS);
Linus Torvalds's avatar
Linus Torvalds committed
582 583 584 585 586 587 588
	if (error)
		goto dput_and_out;

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

589
	set_fs_root(current->fs, &path);
Linus Torvalds's avatar
Linus Torvalds committed
590 591
	error = 0;
dput_and_out:
592
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
593 594 595 596
out:
	return error;
}

597
SYSCALL_DEFINE2(fchmod, unsigned int, fd, mode_t, mode)
Linus Torvalds's avatar
Linus Torvalds committed
598 599 600 601 602 603 604 605 606 607 608
{
	struct inode * inode;
	struct dentry * dentry;
	struct file * file;
	int err = -EBADF;
	struct iattr newattrs;

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

609
	dentry = file->f_path.dentry;
Linus Torvalds's avatar
Linus Torvalds committed
610 611
	inode = dentry->d_inode;

612
	audit_inode(NULL, dentry);
613

614 615
	err = mnt_want_write(file->f_path.mnt);
	if (err)
Linus Torvalds's avatar
Linus Torvalds committed
616
		goto out_putf;
617
	mutex_lock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
618 619 620 621 622
	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);
623
	mutex_unlock(&inode->i_mutex);
624
	mnt_drop_write(file->f_path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
625 626 627 628 629 630
out_putf:
	fput(file);
out:
	return err;
}

631 632
asmlinkage long sys_fchmodat(int dfd, const char __user *filename,
			     mode_t mode)
Linus Torvalds's avatar
Linus Torvalds committed
633
{
634 635
	struct path path;
	struct inode *inode;
Linus Torvalds's avatar
Linus Torvalds committed
636 637 638
	int error;
	struct iattr newattrs;

639
	error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
Linus Torvalds's avatar
Linus Torvalds committed
640 641
	if (error)
		goto out;
642
	inode = path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
643

644
	error = mnt_want_write(path.mnt);
645
	if (error)
Linus Torvalds's avatar
Linus Torvalds committed
646
		goto dput_and_out;
647
	mutex_lock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
648 649 650 651
	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;
652
	error = notify_change(path.dentry, &newattrs);
653
	mutex_unlock(&inode->i_mutex);
654
	mnt_drop_write(path.mnt);
Linus Torvalds's avatar
Linus Torvalds committed
655
dput_and_out:
656
	path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
657 658 659 660
out:
	return error;
}

661
SYSCALL_DEFINE2(chmod, const char __user *, filename, mode_t, mode)
662 663 664 665
{
	return sys_fchmodat(AT_FDCWD, filename, mode);
}

Linus Torvalds's avatar
Linus Torvalds committed
666 667
static int chown_common(struct dentry * dentry, uid_t user, gid_t group)
{
668
	struct inode *inode = dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
669 670 671 672 673 674 675 676 677 678 679 680 681
	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))
682 683
		newattrs.ia_valid |=
			ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
684
	mutex_lock(&inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
685
	error = notify_change(dentry, &newattrs);
686
	mutex_unlock(&inode->i_mutex);
687

Linus Torvalds's avatar
Linus Torvalds committed
688 689 690
	return error;
}

691
SYSCALL_DEFINE3(chown, const char __user *, filename, uid_t, user, gid_t, group)
Linus Torvalds's avatar
Linus Torvalds committed
692
{
693
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
694 695
	int error;

696
	error = user_path(filename, &path);
697 698
	if (error)
		goto out;
699
	error = mnt_want_write(path.mnt);
700 701
	if (error)
		goto out_release;
702 703
	error = chown_common(path.dentry, user, group);
	mnt_drop_write(path.mnt);
704
out_release:
705
	path_put(&path);
706
out:
Linus Torvalds's avatar
Linus Torvalds committed
707 708 709
	return error;
}

710 711 712
asmlinkage long sys_fchownat(int dfd, const char __user *filename, uid_t user,
			     gid_t group, int flag)
{
713
	struct path path;
714 715 716 717 718 719 720
	int error = -EINVAL;
	int follow;

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

	follow = (flag & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
721
	error = user_path_at(dfd, filename, follow, &path);
722 723
	if (error)
		goto out;
724
	error = mnt_want_write(path.mnt);
725 726
	if (error)
		goto out_release;
727 728
	error = chown_common(path.dentry, user, group);
	mnt_drop_write(path.mnt);
729
out_release:
730
	path_put(&path);
731 732 733 734
out:
	return error;
}

735
SYSCALL_DEFINE3(lchown, const char __user *, filename, uid_t, user, gid_t, group)
Linus Torvalds's avatar
Linus Torvalds committed
736
{
737
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
738 739
	int error;

740
	error = user_lpath(filename, &path);
741 742
	if (error)
		goto out;
743
	error = mnt_want_write(path.mnt);
744 745
	if (error)
		goto out_release;
746 747
	error = chown_common(path.dentry, user, group);
	mnt_drop_write(path.mnt);
748
out_release:
749
	path_put(&path);
750
out:
Linus Torvalds's avatar
Linus Torvalds committed
751 752 753
	return error;
}

754
SYSCALL_DEFINE3(fchown, unsigned int, fd, uid_t, user, gid_t, group)
Linus Torvalds's avatar
Linus Torvalds committed
755 756 757
{
	struct file * file;
	int error = -EBADF;
758
	struct dentry * dentry;
Linus Torvalds's avatar
Linus Torvalds committed
759 760

	file = fget(fd);
761 762 763
	if (!file)
		goto out;

764 765 766
	error = mnt_want_write(file->f_path.mnt);
	if (error)
		goto out_fput;
767
	dentry = file->f_path.dentry;
768
	audit_inode(NULL, dentry);
769
	error = chown_common(dentry, user, group);
770 771
	mnt_drop_write(file->f_path.mnt);
out_fput:
772 773
	fput(file);
out:
Linus Torvalds's avatar
Linus Torvalds committed
774 775 776
	return error;
}

777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805
/*
 * 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;
}

806
static struct file *__dentry_open(struct dentry *dentry, struct vfsmount *mnt,
807
					int flags, struct file *f,
808 809
					int (*open)(struct inode *, struct file *),
					const struct cred *cred)
Linus Torvalds's avatar
Linus Torvalds committed
810 811 812 813 814
{
	struct inode *inode;
	int error;

	f->f_flags = flags;
815
	f->f_mode = (__force fmode_t)((flags+1) & O_ACCMODE) | FMODE_LSEEK |
816
				FMODE_PREAD | FMODE_PWRITE;
Linus Torvalds's avatar
Linus Torvalds committed
817 818
	inode = dentry->d_inode;
	if (f->f_mode & FMODE_WRITE) {
819
		error = __get_file_write_access(inode, mnt);
Linus Torvalds's avatar
Linus Torvalds committed
820 821
		if (error)
			goto cleanup_file;
822 823
		if (!special_file(inode->i_mode))
			file_take_write(f);
Linus Torvalds's avatar
Linus Torvalds committed
824 825 826
	}

	f->f_mapping = inode->i_mapping;
827 828
	f->f_path.dentry = dentry;
	f->f_path.mnt = mnt;
Linus Torvalds's avatar
Linus Torvalds committed
829 830 831 832
	f->f_pos = 0;
	f->f_op = fops_get(inode->i_fop);
	file_move(f, &inode->i_sb->s_files);

833
	error = security_dentry_open(f, cred);
834 835 836
	if (error)
		goto cleanup_all;

837 838 839 840
	if (!open && f->f_op)
		open = f->f_op->open;
	if (open) {
		error = open(inode, f);
Linus Torvalds's avatar
Linus Torvalds committed
841 842 843
		if (error)
			goto cleanup_all;
	}
844

Linus Torvalds's avatar
Linus Torvalds committed
845 846 847 848 849 850
	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) {
851 852
		if (!f->f_mapping->a_ops ||
		    ((!f->f_mapping->a_ops->direct_IO) &&
853
		    (!f->f_mapping->a_ops->get_xip_mem))) {
Linus Torvalds's avatar
Linus Torvalds committed
854 855 856 857 858 859 860 861 862
			fput(f);
			f = ERR_PTR(-EINVAL);
		}
	}

	return f;

cleanup_all:
	fops_put(f->f_op);
863
	if (f->f_mode & FMODE_WRITE) {
Linus Torvalds's avatar
Linus Torvalds committed
864
		put_write_access(inode);
865 866 867 868 869 870 871 872
		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);
873
			mnt_drop_write(mnt);
874
		}
875
	}
Linus Torvalds's avatar
Linus Torvalds committed
876
	file_kill(f);
877 878
	f->f_path.dentry = NULL;
	f->f_path.mnt = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
879 880 881 882 883 884 885
cleanup_file:
	put_filp(f);
	dput(dentry);
	mntput(mnt);
	return ERR_PTR(error);
}

886 887 888 889 890 891 892 893 894 895
/**
 * 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.
896 897 898 899
 * 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).
900 901 902 903 904 905 906 907
 * 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 *))
{
908 909
	const struct cred *cred = current_cred();

910 911 912 913
	if (IS_ERR(nd->intent.open.file))
		goto out;
	if (IS_ERR(dentry))
		goto out_err;
914
	nd->intent.open.file = __dentry_open(dget(dentry), mntget(nd->path.mnt),
915 916
					     nd->intent.open.flags - 1,
					     nd->intent.open.file,
917
					     open, cred);
918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935
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)
{
936
	const struct cred *cred = current_cred();
937 938 939 940 941
	struct file *filp;

	/* Pick up the filp from the open intent */
	filp = nd->intent.open.file;
	/* Has the filesystem initialised the file for us? */
942
	if (filp->f_path.dentry == NULL)
943
		filp = __dentry_open(nd->path.dentry, nd->path.mnt, flags, filp,
944
				     NULL, cred);
945
	else
Jan Blunck's avatar
Jan Blunck committed
946
		path_put(&nd->path);
947 948 949
	return filp;
}

950 951 952 953
/*
 * dentry_open() will have done dput(dentry) and mntput(mnt) if it returns an
 * error.
 */
954 955
struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags,
			 const struct cred *cred)
956 957 958 959
{
	int error;
	struct file *f;

960 961 962 963 964 965 966 967 968 969 970 971
	/*
	 * 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);
	}

972 973
	error = -ENFILE;
	f = get_empty_filp();
974 975 976
	if (f == NULL) {
		dput(dentry);
		mntput(mnt);
977
		return ERR_PTR(error);
978
	}
979

980
	return __dentry_open(dentry, mnt, flags, f, NULL, cred);
981
}
Linus Torvalds's avatar
Linus Torvalds committed
982 983
EXPORT_SYMBOL(dentry_open);

984
static void __put_unused_fd(struct files_struct *files, unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
985
{
986 987
	struct fdtable *fdt = files_fdtable(files);
	__FD_CLR(fd, fdt->open_fds);
988 989
	if (fd < files->next_fd)
		files->next_fd = fd;
Linus Torvalds's avatar
Linus Torvalds committed
990 991
}

992
void put_unused_fd(unsigned int fd)
Linus Torvalds's avatar
Linus Torvalds committed
993 994 995 996 997 998 999 1000 1001 1002
{
	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);

/*
1003
 * Install a file pointer in the fd array.
Linus Torvalds's avatar
Linus Torvalds committed
1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014
 *
 * 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.
 */

1015
void fd_install(unsigned int fd, struct file *file)
Linus Torvalds's avatar
Linus Torvalds committed
1016 1017
{
	struct files_struct *files = current->files;
1018
	struct fdtable *fdt;
Linus Torvalds's avatar
Linus Torvalds committed
1019
	spin_lock(&files->file_lock);
1020
	fdt = files_fdtable(files);
1021 1022
	BUG_ON(fdt->fd[fd] != NULL);
	rcu_assign_pointer(fdt->fd[fd], file);
Linus Torvalds's avatar
Linus Torvalds committed
1023 1024 1025 1026 1027
	spin_unlock(&files->file_lock);
}

EXPORT_SYMBOL(fd_install);

1028
long do_sys_open(int dfd, const char __user *filename, int flags, int mode)
Linus Torvalds's avatar
Linus Torvalds committed
1029
{
1030 1031
	char *tmp = getname(filename);
	int fd = PTR_ERR(tmp);
Linus Torvalds's avatar
Linus Torvalds committed
1032 1033

	if (!IS_ERR(tmp)) {
Ulrich Drepper's avatar
Ulrich Drepper committed
1034
		fd = get_unused_fd_flags(flags);
Linus Torvalds's avatar
Linus Torvalds committed
1035
		if (fd >= 0) {
1036
			struct file *f = do_filp_open(dfd, tmp, flags, mode);
1037 1038 1039 1040
			if (IS_ERR(f)) {
				put_unused_fd(fd);
				fd = PTR_ERR(f);
			} else {
1041
				fsnotify_open(f->f_path.dentry);
1042 1043
				fd_install(fd, f);
			}
Linus Torvalds's avatar
Linus Torvalds committed
1044 1045 1046 1047 1048
		}
		putname(tmp);
	}
	return fd;
}
1049

1050
SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, int, mode)
1051
{
1052 1053