stat.c

  1// SPDX-License-Identifier: GPL-2.0
  2/*
  3 *  linux/fs/stat.c
  4 *
  5 *  Copyright (C) 1991, 1992  Linus Torvalds
  6 */
  7
  8#include <linux/blkdev.h>
  9#include <linux/export.h>
 10#include <linux/mm.h>
 11#include <linux/errno.h>
 12#include <linux/file.h>
 13#include <linux/highuid.h>
 14#include <linux/fs.h>
 15#include <linux/namei.h>
 16#include <linux/security.h>
 17#include <linux/cred.h>
 18#include <linux/syscalls.h>
 19#include <linux/pagemap.h>
 20#include <linux/compat.h>
 21#include <linux/iversion.h>
 22
 23#include <linux/uaccess.h>
 24#include <asm/unistd.h>
 25
 26#include <trace/events/timestamp.h>
 27
 28#include "internal.h"
 29#include "mount.h"
 30
 31/**
 32 * fill_mg_cmtime - Fill in the mtime and ctime and flag ctime as QUERIED
 33 * @stat: where to store the resulting values
 34 * @request_mask: STATX_* values requested
 35 * @inode: inode from which to grab the c/mtime
 36 *
 37 * Given @inode, grab the ctime and mtime out if it and store the result
 38 * in @stat. When fetching the value, flag it as QUERIED (if not already)
 39 * so the next write will record a distinct timestamp.
 40 *
 41 * NB: The QUERIED flag is tracked in the ctime, but we set it there even
 42 * if only the mtime was requested, as that ensures that the next mtime
 43 * change will be distinct.
 44 */
 45void fill_mg_cmtime(struct kstat *stat, u32 request_mask, struct inode *inode)
 46{
 47	atomic_t *pcn = (atomic_t *)&inode->i_ctime_nsec;
 48
 49	/* If neither time was requested, then don't report them */
 50	if (!(request_mask & (STATX_CTIME|STATX_MTIME))) {
 51		stat->result_mask &= ~(STATX_CTIME|STATX_MTIME);
 52		return;
 53	}
 54
 55	stat->mtime = inode_get_mtime(inode);
 56	stat->ctime.tv_sec = inode->i_ctime_sec;
 57	stat->ctime.tv_nsec = (u32)atomic_read(pcn);
 58	if (!(stat->ctime.tv_nsec & I_CTIME_QUERIED))
 59		stat->ctime.tv_nsec = ((u32)atomic_fetch_or(I_CTIME_QUERIED, pcn));
 60	stat->ctime.tv_nsec &= ~I_CTIME_QUERIED;
 61	trace_fill_mg_cmtime(inode, &stat->ctime, &stat->mtime);
 62}
 63EXPORT_SYMBOL(fill_mg_cmtime);
 64
 65/**
 66 * generic_fillattr - Fill in the basic attributes from the inode struct
 67 * @idmap:		idmap of the mount the inode was found from
 68 * @request_mask:	statx request_mask
 69 * @inode:		Inode to use as the source
 70 * @stat:		Where to fill in the attributes
 71 *
 72 * Fill in the basic attributes in the kstat structure from data that's to be
 73 * found on the VFS inode structure.  This is the default if no getattr inode
 74 * operation is supplied.
 75 *
 76 * If the inode has been found through an idmapped mount the idmap of
 77 * the vfsmount must be passed through @idmap. This function will then
 78 * take care to map the inode according to @idmap before filling in the
 79 * uid and gid filds. On non-idmapped mounts or if permission checking is to be
 80 * performed on the raw inode simply pass @nop_mnt_idmap.
 81 */
 82void generic_fillattr(struct mnt_idmap *idmap, u32 request_mask,
 83		      struct inode *inode, struct kstat *stat)
 84{
 85	vfsuid_t vfsuid = i_uid_into_vfsuid(idmap, inode);
 86	vfsgid_t vfsgid = i_gid_into_vfsgid(idmap, inode);
 87
 88	stat->dev = inode->i_sb->s_dev;
 89	stat->ino = inode->i_ino;
 90	stat->mode = inode->i_mode;
 91	stat->nlink = inode->i_nlink;
 92	stat->uid = vfsuid_into_kuid(vfsuid);
 93	stat->gid = vfsgid_into_kgid(vfsgid);
 94	stat->rdev = inode->i_rdev;
 95	stat->size = i_size_read(inode);
 96	stat->atime = inode_get_atime(inode);
 97
 98	if (is_mgtime(inode)) {
 99		fill_mg_cmtime(stat, request_mask, inode);
100	} else {
101		stat->ctime = inode_get_ctime(inode);
102		stat->mtime = inode_get_mtime(inode);
103	}
104
105	stat->blksize = i_blocksize(inode);
106	stat->blocks = inode->i_blocks;
107
108	if ((request_mask & STATX_CHANGE_COOKIE) && IS_I_VERSION(inode)) {
109		stat->result_mask |= STATX_CHANGE_COOKIE;
110		stat->change_cookie = inode_query_iversion(inode);
111	}
112
113}
114EXPORT_SYMBOL(generic_fillattr);
115
116/**
117 * generic_fill_statx_attr - Fill in the statx attributes from the inode flags
118 * @inode:	Inode to use as the source
119 * @stat:	Where to fill in the attribute flags
120 *
121 * Fill in the STATX_ATTR_* flags in the kstat structure for properties of the
122 * inode that are published on i_flags and enforced by the VFS.
123 */
124void generic_fill_statx_attr(struct inode *inode, struct kstat *stat)
125{
126	if (inode->i_flags & S_IMMUTABLE)
127		stat->attributes |= STATX_ATTR_IMMUTABLE;
128	if (inode->i_flags & S_APPEND)
129		stat->attributes |= STATX_ATTR_APPEND;
130	stat->attributes_mask |= KSTAT_ATTR_VFS_FLAGS;
131}
132EXPORT_SYMBOL(generic_fill_statx_attr);
133
134/**
135 * generic_fill_statx_atomic_writes - Fill in atomic writes statx attributes
136 * @stat:	Where to fill in the attribute flags
137 * @unit_min:	Minimum supported atomic write length in bytes
138 * @unit_max:	Maximum supported atomic write length in bytes
139 * @unit_max_opt: Optimised maximum supported atomic write length in bytes
140 *
141 * Fill in the STATX{_ATTR}_WRITE_ATOMIC flags in the kstat structure from
142 * atomic write unit_min and unit_max values.
143 */
144void generic_fill_statx_atomic_writes(struct kstat *stat,
145				      unsigned int unit_min,
146				      unsigned int unit_max,
147				      unsigned int unit_max_opt)
148{
149	/* Confirm that the request type is known */
150	stat->result_mask |= STATX_WRITE_ATOMIC;
151
152	/* Confirm that the file attribute type is known */
153	stat->attributes_mask |= STATX_ATTR_WRITE_ATOMIC;
154
155	if (unit_min) {
156		stat->atomic_write_unit_min = unit_min;
157		stat->atomic_write_unit_max = unit_max;
158		stat->atomic_write_unit_max_opt = unit_max_opt;
159		/* Initially only allow 1x segment */
160		stat->atomic_write_segments_max = 1;
161
162		/* Confirm atomic writes are actually supported */
163		stat->attributes |= STATX_ATTR_WRITE_ATOMIC;
164	}
165}
166EXPORT_SYMBOL_GPL(generic_fill_statx_atomic_writes);
167
168/**
169 * vfs_getattr_nosec - getattr without security checks
170 * @path: file to get attributes from
171 * @stat: structure to return attributes in
172 * @request_mask: STATX_xxx flags indicating what the caller wants
173 * @query_flags: Query mode (AT_STATX_SYNC_TYPE)
174 *
175 * Get attributes without calling security_inode_getattr.
176 *
177 * Currently the only caller other than vfs_getattr is internal to the
178 * filehandle lookup code, which uses only the inode number and returns no
179 * attributes to any user.  Any other code probably wants vfs_getattr.
180 */
181int vfs_getattr_nosec(const struct path *path, struct kstat *stat,
182		      u32 request_mask, unsigned int query_flags)
183{
184	struct mnt_idmap *idmap;
185	struct inode *inode = d_backing_inode(path->dentry);
186
187	memset(stat, 0, sizeof(*stat));
188	stat->result_mask |= STATX_BASIC_STATS;
189	query_flags &= AT_STATX_SYNC_TYPE;
190
191	/* allow the fs to override these if it really wants to */
192	/* SB_NOATIME means filesystem supplies dummy atime value */
193	if (inode->i_sb->s_flags & SB_NOATIME)
194		stat->result_mask &= ~STATX_ATIME;
195
196	/*
197	 * Note: If you add another clause to set an attribute flag, please
198	 * update attributes_mask below.
199	 */
200	if (IS_AUTOMOUNT(inode))
201		stat->attributes |= STATX_ATTR_AUTOMOUNT;
202
203	if (IS_DAX(inode))
204		stat->attributes |= STATX_ATTR_DAX;
205
206	stat->attributes_mask |= (STATX_ATTR_AUTOMOUNT |
207				  STATX_ATTR_DAX);
208
209	idmap = mnt_idmap(path->mnt);
210	if (inode->i_op->getattr) {
211		int ret;
212
213		ret = inode->i_op->getattr(idmap, path, stat, request_mask,
214				query_flags);
215		if (ret)
216			return ret;
217	} else {
218		generic_fillattr(idmap, request_mask, inode, stat);
219	}
220
221	/*
222	 * If this is a block device inode, override the filesystem attributes
223	 * with the block device specific parameters that need to be obtained
224	 * from the bdev backing inode.
225	 */
226	if (S_ISBLK(stat->mode))
227		bdev_statx(path, stat, request_mask);
228
229	return 0;
230}
231EXPORT_SYMBOL(vfs_getattr_nosec);
232
233/*
234 * vfs_getattr - Get the enhanced basic attributes of a file
235 * @path: The file of interest
236 * @stat: Where to return the statistics
237 * @request_mask: STATX_xxx flags indicating what the caller wants
238 * @query_flags: Query mode (AT_STATX_SYNC_TYPE)
239 *
240 * Ask the filesystem for a file's attributes.  The caller must indicate in
241 * request_mask and query_flags to indicate what they want.
242 *
243 * If the file is remote, the filesystem can be forced to update the attributes
244 * from the backing store by passing AT_STATX_FORCE_SYNC in query_flags or can
245 * suppress the update by passing AT_STATX_DONT_SYNC.
246 *
247 * Bits must have been set in request_mask to indicate which attributes the
248 * caller wants retrieving.  Any such attribute not requested may be returned
249 * anyway, but the value may be approximate, and, if remote, may not have been
250 * synchronised with the server.
251 *
252 * 0 will be returned on success, and a -ve error code if unsuccessful.
253 */
254int vfs_getattr(const struct path *path, struct kstat *stat,
255		u32 request_mask, unsigned int query_flags)
256{
257	int retval;
258
259	retval = security_inode_getattr(path);
260	if (unlikely(retval))
261		return retval;
262	return vfs_getattr_nosec(path, stat, request_mask, query_flags);
263}
264EXPORT_SYMBOL(vfs_getattr);
265
266/**
267 * vfs_fstat - Get the basic attributes by file descriptor
268 * @fd: The file descriptor referring to the file of interest
269 * @stat: The result structure to fill in.
270 *
271 * This function is a wrapper around vfs_getattr().  The main difference is
272 * that it uses a file descriptor to determine the file location.
273 *
274 * 0 will be returned on success, and a -ve error code if unsuccessful.
275 */
276int vfs_fstat(int fd, struct kstat *stat)
277{
278	CLASS(fd_raw, f)(fd);
279	if (fd_empty(f))
280		return -EBADF;
281	return vfs_getattr(&fd_file(f)->f_path, stat, STATX_BASIC_STATS, 0);
282}
283
284static int statx_lookup_flags(int flags)
285{
286	int lookup_flags = 0;
287
288	if (!(flags & AT_SYMLINK_NOFOLLOW))
289		lookup_flags |= LOOKUP_FOLLOW;
290	if (!(flags & AT_NO_AUTOMOUNT))
291		lookup_flags |= LOOKUP_AUTOMOUNT;
292
293	return lookup_flags;
294}
295
296static int vfs_statx_path(const struct path *path, int flags, struct kstat *stat,
297			  u32 request_mask)
298{
299	int error = vfs_getattr(path, stat, request_mask, flags);
300	if (error)
301		return error;
302
303	if (request_mask & STATX_MNT_ID_UNIQUE) {
304		stat->mnt_id = real_mount(path->mnt)->mnt_id_unique;
305		stat->result_mask |= STATX_MNT_ID_UNIQUE;
306	} else {
307		stat->mnt_id = real_mount(path->mnt)->mnt_id;
308		stat->result_mask |= STATX_MNT_ID;
309	}
310
311	if (path_mounted(path))
312		stat->attributes |= STATX_ATTR_MOUNT_ROOT;
313	stat->attributes_mask |= STATX_ATTR_MOUNT_ROOT;
314	return 0;
315}
316
317static int vfs_statx_fd(int fd, int flags, struct kstat *stat,
318			  u32 request_mask)
319{
320	CLASS(fd_raw, f)(fd);
321	if (fd_empty(f))
322		return -EBADF;
323	return vfs_statx_path(&fd_file(f)->f_path, flags, stat, request_mask);
324}
325
326/**
327 * vfs_statx - Get basic and extra attributes by filename
328 * @dfd: A file descriptor representing the base dir for a relative filename
329 * @filename: The name of the file of interest
330 * @flags: Flags to control the query
331 * @stat: The result structure to fill in.
332 * @request_mask: STATX_xxx flags indicating what the caller wants
333 *
334 * This function is a wrapper around vfs_getattr().  The main difference is
335 * that it uses a filename and base directory to determine the file location.
336 * Additionally, the use of AT_SYMLINK_NOFOLLOW in flags will prevent a symlink
337 * at the given name from being referenced.
338 *
339 * 0 will be returned on success, and a -ve error code if unsuccessful.
340 */
341static int vfs_statx(int dfd, struct filename *filename, int flags,
342	      struct kstat *stat, u32 request_mask)
343{
344	struct path path;
345	unsigned int lookup_flags = statx_lookup_flags(flags);
346	int error;
347
348	if (flags & ~(AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT | AT_EMPTY_PATH |
349		      AT_STATX_SYNC_TYPE))
350		return -EINVAL;
351
352retry:
353	error = filename_lookup(dfd, filename, lookup_flags, &path, NULL);
354	if (error)
355		return error;
356	error = vfs_statx_path(&path, flags, stat, request_mask);
357	path_put(&path);
358	if (retry_estale(error, lookup_flags)) {
359		lookup_flags |= LOOKUP_REVAL;
360		goto retry;
361	}
362	return error;
363}
364
365int vfs_fstatat(int dfd, const char __user *filename,
366			      struct kstat *stat, int flags)
367{
368	int ret;
369	int statx_flags = flags | AT_NO_AUTOMOUNT;
370	struct filename *name = getname_maybe_null(filename, flags);
371
372	if (!name && dfd >= 0)
373		return vfs_fstat(dfd, stat);
374
375	ret = vfs_statx(dfd, name, statx_flags, stat, STATX_BASIC_STATS);
376	putname(name);
377
378	return ret;
379}
380
381#ifdef __ARCH_WANT_OLD_STAT
382
383/*
384 * For backward compatibility?  Maybe this should be moved
385 * into arch/i386 instead?
386 */
387static int cp_old_stat(struct kstat *stat, struct __old_kernel_stat __user * statbuf)
388{
389	static int warncount = 5;
390	struct __old_kernel_stat tmp;
391
392	if (warncount > 0) {
393		warncount--;
394		printk(KERN_WARNING "VFS: Warning: %s using old stat() call. Recompile your binary.\n",
395			current->comm);
396	} else if (warncount < 0) {
397		/* it's laughable, but... */
398		warncount = 0;
399	}
400
401	memset(&tmp, 0, sizeof(struct __old_kernel_stat));
402	tmp.st_dev = old_encode_dev(stat->dev);
403	tmp.st_ino = stat->ino;
404	if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
405		return -EOVERFLOW;
406	tmp.st_mode = stat->mode;
407	tmp.st_nlink = stat->nlink;
408	if (tmp.st_nlink != stat->nlink)
409		return -EOVERFLOW;
410	SET_UID(tmp.st_uid, from_kuid_munged(current_user_ns(), stat->uid));
411	SET_GID(tmp.st_gid, from_kgid_munged(current_user_ns(), stat->gid));
412	tmp.st_rdev = old_encode_dev(stat->rdev);
413#if BITS_PER_LONG == 32
414	if (stat->size > MAX_NON_LFS)
415		return -EOVERFLOW;
416#endif
417	tmp.st_size = stat->size;
418	tmp.st_atime = stat->atime.tv_sec;
419	tmp.st_mtime = stat->mtime.tv_sec;
420	tmp.st_ctime = stat->ctime.tv_sec;
421	return copy_to_user(statbuf,&tmp,sizeof(tmp)) ? -EFAULT : 0;
422}
423
424SYSCALL_DEFINE2(stat, const char __user *, filename,
425		struct __old_kernel_stat __user *, statbuf)
426{
427	struct kstat stat;
428	int error;
429
430	error = vfs_stat(filename, &stat);
431	if (unlikely(error))
432		return error;
433
434	return cp_old_stat(&stat, statbuf);
435}
436
437SYSCALL_DEFINE2(lstat, const char __user *, filename,
438		struct __old_kernel_stat __user *, statbuf)
439{
440	struct kstat stat;
441	int error;
442
443	error = vfs_lstat(filename, &stat);
444	if (unlikely(error))
445		return error;
446
447	return cp_old_stat(&stat, statbuf);
448}
449
450SYSCALL_DEFINE2(fstat, unsigned int, fd, struct __old_kernel_stat __user *, statbuf)
451{
452	struct kstat stat;
453	int error;
454
455	error = vfs_fstat(fd, &stat);
456	if (unlikely(error))
457		return error;
458
459	return cp_old_stat(&stat, statbuf);
460}
461
462#endif /* __ARCH_WANT_OLD_STAT */
463
464#ifdef __ARCH_WANT_NEW_STAT
465
466#ifndef INIT_STRUCT_STAT_PADDING
467#  define INIT_STRUCT_STAT_PADDING(st) memset(&st, 0, sizeof(st))
468#endif
469
470static int cp_new_stat(struct kstat *stat, struct stat __user *statbuf)
471{
472	struct stat tmp;
473
474	if (sizeof(tmp.st_dev) < 4 && !old_valid_dev(stat->dev))
475		return -EOVERFLOW;
476	if (sizeof(tmp.st_rdev) < 4 && !old_valid_dev(stat->rdev))
477		return -EOVERFLOW;
478#if BITS_PER_LONG == 32
479	if (stat->size > MAX_NON_LFS)
480		return -EOVERFLOW;
481#endif
482
483	INIT_STRUCT_STAT_PADDING(tmp);
484	tmp.st_dev = new_encode_dev(stat->dev);
485	tmp.st_ino = stat->ino;
486	if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
487		return -EOVERFLOW;
488	tmp.st_mode = stat->mode;
489	tmp.st_nlink = stat->nlink;
490	if (tmp.st_nlink != stat->nlink)
491		return -EOVERFLOW;
492	SET_UID(tmp.st_uid, from_kuid_munged(current_user_ns(), stat->uid));
493	SET_GID(tmp.st_gid, from_kgid_munged(current_user_ns(), stat->gid));
494	tmp.st_rdev = new_encode_dev(stat->rdev);
495	tmp.st_size = stat->size;
496	tmp.st_atime = stat->atime.tv_sec;
497	tmp.st_mtime = stat->mtime.tv_sec;
498	tmp.st_ctime = stat->ctime.tv_sec;
499#ifdef STAT_HAVE_NSEC
500	tmp.st_atime_nsec = stat->atime.tv_nsec;
501	tmp.st_mtime_nsec = stat->mtime.tv_nsec;
502	tmp.st_ctime_nsec = stat->ctime.tv_nsec;
503#endif
504	tmp.st_blocks = stat->blocks;
505	tmp.st_blksize = stat->blksize;
506	return copy_to_user(statbuf,&tmp,sizeof(tmp)) ? -EFAULT : 0;
507}
508
509SYSCALL_DEFINE2(newstat, const char __user *, filename,
510		struct stat __user *, statbuf)
511{
512	struct kstat stat;
513	int error;
514
515	error = vfs_stat(filename, &stat);
516	if (unlikely(error))
517		return error;
518
519	return cp_new_stat(&stat, statbuf);
520}
521
522SYSCALL_DEFINE2(newlstat, const char __user *, filename,
523		struct stat __user *, statbuf)
524{
525	struct kstat stat;
526	int error;
527
528	error = vfs_lstat(filename, &stat);
529	if (unlikely(error))
530		return error;
531
532	return cp_new_stat(&stat, statbuf);
533}
534
535#if !defined(__ARCH_WANT_STAT64) || defined(__ARCH_WANT_SYS_NEWFSTATAT)
536SYSCALL_DEFINE4(newfstatat, int, dfd, const char __user *, filename,
537		struct stat __user *, statbuf, int, flag)
538{
539	struct kstat stat;
540	int error;
541
542	error = vfs_fstatat(dfd, filename, &stat, flag);
543	if (unlikely(error))
544		return error;
545
546	return cp_new_stat(&stat, statbuf);
547}
548#endif
549
550SYSCALL_DEFINE2(newfstat, unsigned int, fd, struct stat __user *, statbuf)
551{
552	struct kstat stat;
553	int error;
554
555	error = vfs_fstat(fd, &stat);
556	if (unlikely(error))
557		return error;
558
559	return cp_new_stat(&stat, statbuf);
560}
561#endif
562
563static int do_readlinkat(int dfd, const char __user *pathname,
564			 char __user *buf, int bufsiz)
565{
566	struct path path;
567	struct filename *name;
568	int error;
569	unsigned int lookup_flags = LOOKUP_EMPTY;
570
571	if (bufsiz <= 0)
572		return -EINVAL;
573
574retry:
575	name = getname_flags(pathname, lookup_flags);
576	error = filename_lookup(dfd, name, lookup_flags, &path, NULL);
577	if (unlikely(error)) {
578		putname(name);
579		return error;
580	}
581
582	/*
583	 * AFS mountpoints allow readlink(2) but are not symlinks
584	 */
585	if (d_is_symlink(path.dentry) ||
586	    d_backing_inode(path.dentry)->i_op->readlink) {
587		error = security_inode_readlink(path.dentry);
588		if (!error) {
589			touch_atime(&path);
590			error = vfs_readlink(path.dentry, buf, bufsiz);
591		}
592	} else {
593		error = (name->name[0] == '\0') ? -ENOENT : -EINVAL;
594	}
595	path_put(&path);
596	putname(name);
597	if (retry_estale(error, lookup_flags)) {
598		lookup_flags |= LOOKUP_REVAL;
599		goto retry;
600	}
601	return error;
602}
603
604SYSCALL_DEFINE4(readlinkat, int, dfd, const char __user *, pathname,
605		char __user *, buf, int, bufsiz)
606{
607	return do_readlinkat(dfd, pathname, buf, bufsiz);
608}
609
610SYSCALL_DEFINE3(readlink, const char __user *, path, char __user *, buf,
611		int, bufsiz)
612{
613	return do_readlinkat(AT_FDCWD, path, buf, bufsiz);
614}
615
616
617/* ---------- LFS-64 ----------- */
618#if defined(__ARCH_WANT_STAT64) || defined(__ARCH_WANT_COMPAT_STAT64)
619
620#ifndef INIT_STRUCT_STAT64_PADDING
621#  define INIT_STRUCT_STAT64_PADDING(st) memset(&st, 0, sizeof(st))
622#endif
623
624static long cp_new_stat64(struct kstat *stat, struct stat64 __user *statbuf)
625{
626	struct stat64 tmp;
627
628	INIT_STRUCT_STAT64_PADDING(tmp);
629#ifdef CONFIG_MIPS
630	/* mips has weird padding, so we don't get 64 bits there */
631	tmp.st_dev = new_encode_dev(stat->dev);
632	tmp.st_rdev = new_encode_dev(stat->rdev);
633#else
634	tmp.st_dev = huge_encode_dev(stat->dev);
635	tmp.st_rdev = huge_encode_dev(stat->rdev);
636#endif
637	tmp.st_ino = stat->ino;
638	if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
639		return -EOVERFLOW;
640#ifdef STAT64_HAS_BROKEN_ST_INO
641	tmp.__st_ino = stat->ino;
642#endif
643	tmp.st_mode = stat->mode;
644	tmp.st_nlink = stat->nlink;
645	tmp.st_uid = from_kuid_munged(current_user_ns(), stat->uid);
646	tmp.st_gid = from_kgid_munged(current_user_ns(), stat->gid);
647	tmp.st_atime = stat->atime.tv_sec;
648	tmp.st_atime_nsec = stat->atime.tv_nsec;
649	tmp.st_mtime = stat->mtime.tv_sec;
650	tmp.st_mtime_nsec = stat->mtime.tv_nsec;
651	tmp.st_ctime = stat->ctime.tv_sec;
652	tmp.st_ctime_nsec = stat->ctime.tv_nsec;
653	tmp.st_size = stat->size;
654	tmp.st_blocks = stat->blocks;
655	tmp.st_blksize = stat->blksize;
656	return copy_to_user(statbuf,&tmp,sizeof(tmp)) ? -EFAULT : 0;
657}
658
659SYSCALL_DEFINE2(stat64, const char __user *, filename,
660		struct stat64 __user *, statbuf)
661{
662	struct kstat stat;
663	int error = vfs_stat(filename, &stat);
664
665	if (!error)
666		error = cp_new_stat64(&stat, statbuf);
667
668	return error;
669}
670
671SYSCALL_DEFINE2(lstat64, const char __user *, filename,
672		struct stat64 __user *, statbuf)
673{
674	struct kstat stat;
675	int error = vfs_lstat(filename, &stat);
676
677	if (!error)
678		error = cp_new_stat64(&stat, statbuf);
679
680	return error;
681}
682
683SYSCALL_DEFINE2(fstat64, unsigned long, fd, struct stat64 __user *, statbuf)
684{
685	struct kstat stat;
686	int error = vfs_fstat(fd, &stat);
687
688	if (!error)
689		error = cp_new_stat64(&stat, statbuf);
690
691	return error;
692}
693
694SYSCALL_DEFINE4(fstatat64, int, dfd, const char __user *, filename,
695		struct stat64 __user *, statbuf, int, flag)
696{
697	struct kstat stat;
698	int error;
699
700	error = vfs_fstatat(dfd, filename, &stat, flag);
701	if (error)
702		return error;
703	return cp_new_stat64(&stat, statbuf);
704}
705#endif /* __ARCH_WANT_STAT64 || __ARCH_WANT_COMPAT_STAT64 */
706
707static noinline_for_stack int
708cp_statx(const struct kstat *stat, struct statx __user *buffer)
709{
710	struct statx tmp;
711
712	memset(&tmp, 0, sizeof(tmp));
713
714	/* STATX_CHANGE_COOKIE is kernel-only for now */
715	tmp.stx_mask = stat->result_mask & ~STATX_CHANGE_COOKIE;
716	tmp.stx_blksize = stat->blksize;
717	/* STATX_ATTR_CHANGE_MONOTONIC is kernel-only for now */
718	tmp.stx_attributes = stat->attributes & ~STATX_ATTR_CHANGE_MONOTONIC;
719	tmp.stx_nlink = stat->nlink;
720	tmp.stx_uid = from_kuid_munged(current_user_ns(), stat->uid);
721	tmp.stx_gid = from_kgid_munged(current_user_ns(), stat->gid);
722	tmp.stx_mode = stat->mode;
723	tmp.stx_ino = stat->ino;
724	tmp.stx_size = stat->size;
725	tmp.stx_blocks = stat->blocks;
726	tmp.stx_attributes_mask = stat->attributes_mask;
727	tmp.stx_atime.tv_sec = stat->atime.tv_sec;
728	tmp.stx_atime.tv_nsec = stat->atime.tv_nsec;
729	tmp.stx_btime.tv_sec = stat->btime.tv_sec;
730	tmp.stx_btime.tv_nsec = stat->btime.tv_nsec;
731	tmp.stx_ctime.tv_sec = stat->ctime.tv_sec;
732	tmp.stx_ctime.tv_nsec = stat->ctime.tv_nsec;
733	tmp.stx_mtime.tv_sec = stat->mtime.tv_sec;
734	tmp.stx_mtime.tv_nsec = stat->mtime.tv_nsec;
735	tmp.stx_rdev_major = MAJOR(stat->rdev);
736	tmp.stx_rdev_minor = MINOR(stat->rdev);
737	tmp.stx_dev_major = MAJOR(stat->dev);
738	tmp.stx_dev_minor = MINOR(stat->dev);
739	tmp.stx_mnt_id = stat->mnt_id;
740	tmp.stx_dio_mem_align = stat->dio_mem_align;
741	tmp.stx_dio_offset_align = stat->dio_offset_align;
742	tmp.stx_dio_read_offset_align = stat->dio_read_offset_align;
743	tmp.stx_subvol = stat->subvol;
744	tmp.stx_atomic_write_unit_min = stat->atomic_write_unit_min;
745	tmp.stx_atomic_write_unit_max = stat->atomic_write_unit_max;
746	tmp.stx_atomic_write_segments_max = stat->atomic_write_segments_max;
747	tmp.stx_atomic_write_unit_max_opt = stat->atomic_write_unit_max_opt;
748
749	return copy_to_user(buffer, &tmp, sizeof(tmp)) ? -EFAULT : 0;
750}
751
752int do_statx(int dfd, struct filename *filename, unsigned int flags,
753	     unsigned int mask, struct statx __user *buffer)
754{
755	struct kstat stat;
756	int error;
757
758	if (mask & STATX__RESERVED)
759		return -EINVAL;
760	if ((flags & AT_STATX_SYNC_TYPE) == AT_STATX_SYNC_TYPE)
761		return -EINVAL;
762
763	/*
764	 * STATX_CHANGE_COOKIE is kernel-only for now. Ignore requests
765	 * from userland.
766	 */
767	mask &= ~STATX_CHANGE_COOKIE;
768
769	error = vfs_statx(dfd, filename, flags, &stat, mask);
770	if (error)
771		return error;
772
773	return cp_statx(&stat, buffer);
774}
775
776int do_statx_fd(int fd, unsigned int flags, unsigned int mask,
777	     struct statx __user *buffer)
778{
779	struct kstat stat;
780	int error;
781
782	if (mask & STATX__RESERVED)
783		return -EINVAL;
784	if ((flags & AT_STATX_SYNC_TYPE) == AT_STATX_SYNC_TYPE)
785		return -EINVAL;
786
787	/*
788	 * STATX_CHANGE_COOKIE is kernel-only for now. Ignore requests
789	 * from userland.
790	 */
791	mask &= ~STATX_CHANGE_COOKIE;
792
793	error = vfs_statx_fd(fd, flags, &stat, mask);
794	if (error)
795		return error;
796
797	return cp_statx(&stat, buffer);
798}
799
800/**
801 * sys_statx - System call to get enhanced stats
802 * @dfd: Base directory to pathwalk from *or* fd to stat.
803 * @filename: File to stat or either NULL or "" with AT_EMPTY_PATH
804 * @flags: AT_* flags to control pathwalk.
805 * @mask: Parts of statx struct actually required.
806 * @buffer: Result buffer.
807 *
808 * Note that fstat() can be emulated by setting dfd to the fd of interest,
809 * supplying "" (or preferably NULL) as the filename and setting AT_EMPTY_PATH
810 * in the flags.
811 */
812SYSCALL_DEFINE5(statx,
813		int, dfd, const char __user *, filename, unsigned, flags,
814		unsigned int, mask,
815		struct statx __user *, buffer)
816{
817	int ret;
818	struct filename *name = getname_maybe_null(filename, flags);
819
820	if (!name && dfd >= 0)
821		return do_statx_fd(dfd, flags & ~AT_NO_AUTOMOUNT, mask, buffer);
822
823	ret = do_statx(dfd, name, flags, mask, buffer);
824	putname(name);
825
826	return ret;
827}
828
829#if defined(CONFIG_COMPAT) && defined(__ARCH_WANT_COMPAT_STAT)
830static int cp_compat_stat(struct kstat *stat, struct compat_stat __user *ubuf)
831{
832	struct compat_stat tmp;
833
834	if (sizeof(tmp.st_dev) < 4 && !old_valid_dev(stat->dev))
835		return -EOVERFLOW;
836	if (sizeof(tmp.st_rdev) < 4 && !old_valid_dev(stat->rdev))
837		return -EOVERFLOW;
838
839	memset(&tmp, 0, sizeof(tmp));
840	tmp.st_dev = new_encode_dev(stat->dev);
841	tmp.st_ino = stat->ino;
842	if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
843		return -EOVERFLOW;
844	tmp.st_mode = stat->mode;
845	tmp.st_nlink = stat->nlink;
846	if (tmp.st_nlink != stat->nlink)
847		return -EOVERFLOW;
848	SET_UID(tmp.st_uid, from_kuid_munged(current_user_ns(), stat->uid));
849	SET_GID(tmp.st_gid, from_kgid_munged(current_user_ns(), stat->gid));
850	tmp.st_rdev = new_encode_dev(stat->rdev);
851	if ((u64) stat->size > MAX_NON_LFS)
852		return -EOVERFLOW;
853	tmp.st_size = stat->size;
854	tmp.st_atime = stat->atime.tv_sec;
855	tmp.st_atime_nsec = stat->atime.tv_nsec;
856	tmp.st_mtime = stat->mtime.tv_sec;
857	tmp.st_mtime_nsec = stat->mtime.tv_nsec;
858	tmp.st_ctime = stat->ctime.tv_sec;
859	tmp.st_ctime_nsec = stat->ctime.tv_nsec;
860	tmp.st_blocks = stat->blocks;
861	tmp.st_blksize = stat->blksize;
862	return copy_to_user(ubuf, &tmp, sizeof(tmp)) ? -EFAULT : 0;
863}
864
865COMPAT_SYSCALL_DEFINE2(newstat, const char __user *, filename,
866		       struct compat_stat __user *, statbuf)
867{
868	struct kstat stat;
869	int error;
870
871	error = vfs_stat(filename, &stat);
872	if (error)
873		return error;
874	return cp_compat_stat(&stat, statbuf);
875}
876
877COMPAT_SYSCALL_DEFINE2(newlstat, const char __user *, filename,
878		       struct compat_stat __user *, statbuf)
879{
880	struct kstat stat;
881	int error;
882
883	error = vfs_lstat(filename, &stat);
884	if (error)
885		return error;
886	return cp_compat_stat(&stat, statbuf);
887}
888
889#ifndef __ARCH_WANT_STAT64
890COMPAT_SYSCALL_DEFINE4(newfstatat, unsigned int, dfd,
891		       const char __user *, filename,
892		       struct compat_stat __user *, statbuf, int, flag)
893{
894	struct kstat stat;
895	int error;
896
897	error = vfs_fstatat(dfd, filename, &stat, flag);
898	if (error)
899		return error;
900	return cp_compat_stat(&stat, statbuf);
901}
902#endif
903
904COMPAT_SYSCALL_DEFINE2(newfstat, unsigned int, fd,
905		       struct compat_stat __user *, statbuf)
906{
907	struct kstat stat;
908	int error = vfs_fstat(fd, &stat);
909
910	if (!error)
911		error = cp_compat_stat(&stat, statbuf);
912	return error;
913}
914#endif
915
916/* Caller is here responsible for sufficient locking (ie. inode->i_lock) */
917void __inode_add_bytes(struct inode *inode, loff_t bytes)
918{
919	inode->i_blocks += bytes >> 9;
920	bytes &= 511;
921	inode->i_bytes += bytes;
922	if (inode->i_bytes >= 512) {
923		inode->i_blocks++;
924		inode->i_bytes -= 512;
925	}
926}
927EXPORT_SYMBOL(__inode_add_bytes);
928
929void inode_add_bytes(struct inode *inode, loff_t bytes)
930{
931	spin_lock(&inode->i_lock);
932	__inode_add_bytes(inode, bytes);
933	spin_unlock(&inode->i_lock);
934}
935
936EXPORT_SYMBOL(inode_add_bytes);
937
938void __inode_sub_bytes(struct inode *inode, loff_t bytes)
939{
940	inode->i_blocks -= bytes >> 9;
941	bytes &= 511;
942	if (inode->i_bytes < bytes) {
943		inode->i_blocks--;
944		inode->i_bytes += 512;
945	}
946	inode->i_bytes -= bytes;
947}
948
949EXPORT_SYMBOL(__inode_sub_bytes);
950
951void inode_sub_bytes(struct inode *inode, loff_t bytes)
952{
953	spin_lock(&inode->i_lock);
954	__inode_sub_bytes(inode, bytes);
955	spin_unlock(&inode->i_lock);
956}
957
958EXPORT_SYMBOL(inode_sub_bytes);
959
960loff_t inode_get_bytes(struct inode *inode)
961{
962	loff_t ret;
963
964	spin_lock(&inode->i_lock);
965	ret = __inode_get_bytes(inode);
966	spin_unlock(&inode->i_lock);
967	return ret;
968}
969
970EXPORT_SYMBOL(inode_get_bytes);
971
972void inode_set_bytes(struct inode *inode, loff_t bytes)
973{
974	/* Caller is here responsible for sufficient locking
975	 * (ie. inode->i_lock) */
976	inode->i_blocks = bytes >> 9;
977	inode->i_bytes = bytes & 511;
978}
979
980EXPORT_SYMBOL(inode_set_bytes);