xref: /kernel/linux/linux-6.6/fs/ext4/ioctl.c (revision 62306a36)
1// SPDX-License-Identifier: GPL-2.0
2/*
3 * linux/fs/ext4/ioctl.c
4 *
5 * Copyright (C) 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 */
10
11#include <linux/fs.h>
12#include <linux/capability.h>
13#include <linux/time.h>
14#include <linux/compat.h>
15#include <linux/mount.h>
16#include <linux/file.h>
17#include <linux/quotaops.h>
18#include <linux/random.h>
19#include <linux/uaccess.h>
20#include <linux/delay.h>
21#include <linux/iversion.h>
22#include <linux/fileattr.h>
23#include <linux/uuid.h>
24#include "ext4_jbd2.h"
25#include "ext4.h"
26#include <linux/fsmap.h>
27#include "fsmap.h"
28#include <trace/events/ext4.h>
29
30typedef void ext4_update_sb_callback(struct ext4_super_block *es,
31				       const void *arg);
32
33/*
34 * Superblock modification callback function for changing file system
35 * label
36 */
37static void ext4_sb_setlabel(struct ext4_super_block *es, const void *arg)
38{
39	/* Sanity check, this should never happen */
40	BUILD_BUG_ON(sizeof(es->s_volume_name) < EXT4_LABEL_MAX);
41
42	memcpy(es->s_volume_name, (char *)arg, EXT4_LABEL_MAX);
43}
44
45/*
46 * Superblock modification callback function for changing file system
47 * UUID.
48 */
49static void ext4_sb_setuuid(struct ext4_super_block *es, const void *arg)
50{
51	memcpy(es->s_uuid, (__u8 *)arg, UUID_SIZE);
52}
53
54static
55int ext4_update_primary_sb(struct super_block *sb, handle_t *handle,
56			   ext4_update_sb_callback func,
57			   const void *arg)
58{
59	int err = 0;
60	struct ext4_sb_info *sbi = EXT4_SB(sb);
61	struct buffer_head *bh = sbi->s_sbh;
62	struct ext4_super_block *es = sbi->s_es;
63
64	trace_ext4_update_sb(sb, bh->b_blocknr, 1);
65
66	BUFFER_TRACE(bh, "get_write_access");
67	err = ext4_journal_get_write_access(handle, sb,
68					    bh,
69					    EXT4_JTR_NONE);
70	if (err)
71		goto out_err;
72
73	lock_buffer(bh);
74	func(es, arg);
75	ext4_superblock_csum_set(sb);
76	unlock_buffer(bh);
77
78	if (buffer_write_io_error(bh) || !buffer_uptodate(bh)) {
79		ext4_msg(sbi->s_sb, KERN_ERR, "previous I/O error to "
80			 "superblock detected");
81		clear_buffer_write_io_error(bh);
82		set_buffer_uptodate(bh);
83	}
84
85	err = ext4_handle_dirty_metadata(handle, NULL, bh);
86	if (err)
87		goto out_err;
88	err = sync_dirty_buffer(bh);
89out_err:
90	ext4_std_error(sb, err);
91	return err;
92}
93
94/*
95 * Update one backup superblock in the group 'grp' using the callback
96 * function 'func' and argument 'arg'. If the handle is NULL the
97 * modification is not journalled.
98 *
99 * Returns: 0 when no modification was done (no superblock in the group)
100 *	    1 when the modification was successful
101 *	   <0 on error
102 */
103static int ext4_update_backup_sb(struct super_block *sb,
104				 handle_t *handle, ext4_group_t grp,
105				 ext4_update_sb_callback func, const void *arg)
106{
107	int err = 0;
108	ext4_fsblk_t sb_block;
109	struct buffer_head *bh;
110	unsigned long offset = 0;
111	struct ext4_super_block *es;
112
113	if (!ext4_bg_has_super(sb, grp))
114		return 0;
115
116	/*
117	 * For the group 0 there is always 1k padding, so we have
118	 * either adjust offset, or sb_block depending on blocksize
119	 */
120	if (grp == 0) {
121		sb_block = 1 * EXT4_MIN_BLOCK_SIZE;
122		offset = do_div(sb_block, sb->s_blocksize);
123	} else {
124		sb_block = ext4_group_first_block_no(sb, grp);
125		offset = 0;
126	}
127
128	trace_ext4_update_sb(sb, sb_block, handle ? 1 : 0);
129
130	bh = ext4_sb_bread(sb, sb_block, 0);
131	if (IS_ERR(bh))
132		return PTR_ERR(bh);
133
134	if (handle) {
135		BUFFER_TRACE(bh, "get_write_access");
136		err = ext4_journal_get_write_access(handle, sb,
137						    bh,
138						    EXT4_JTR_NONE);
139		if (err)
140			goto out_bh;
141	}
142
143	es = (struct ext4_super_block *) (bh->b_data + offset);
144	lock_buffer(bh);
145	if (ext4_has_metadata_csum(sb) &&
146	    es->s_checksum != ext4_superblock_csum(sb, es)) {
147		ext4_msg(sb, KERN_ERR, "Invalid checksum for backup "
148		"superblock %llu", sb_block);
149		unlock_buffer(bh);
150		goto out_bh;
151	}
152	func(es, arg);
153	if (ext4_has_metadata_csum(sb))
154		es->s_checksum = ext4_superblock_csum(sb, es);
155	set_buffer_uptodate(bh);
156	unlock_buffer(bh);
157
158	if (handle) {
159		err = ext4_handle_dirty_metadata(handle, NULL, bh);
160		if (err)
161			goto out_bh;
162	} else {
163		BUFFER_TRACE(bh, "marking dirty");
164		mark_buffer_dirty(bh);
165	}
166	err = sync_dirty_buffer(bh);
167
168out_bh:
169	brelse(bh);
170	ext4_std_error(sb, err);
171	return (err) ? err : 1;
172}
173
174/*
175 * Update primary and backup superblocks using the provided function
176 * func and argument arg.
177 *
178 * Only the primary superblock and at most two backup superblock
179 * modifications are journalled; the rest is modified without journal.
180 * This is safe because e2fsck will re-write them if there is a problem,
181 * and we're very unlikely to ever need more than two backups.
182 */
183static
184int ext4_update_superblocks_fn(struct super_block *sb,
185			       ext4_update_sb_callback func,
186			       const void *arg)
187{
188	handle_t *handle;
189	ext4_group_t ngroups;
190	unsigned int three = 1;
191	unsigned int five = 5;
192	unsigned int seven = 7;
193	int err = 0, ret, i;
194	ext4_group_t grp, primary_grp;
195	struct ext4_sb_info *sbi = EXT4_SB(sb);
196
197	/*
198	 * We can't update superblocks while the online resize is running
199	 */
200	if (test_and_set_bit_lock(EXT4_FLAGS_RESIZING,
201				  &sbi->s_ext4_flags)) {
202		ext4_msg(sb, KERN_ERR, "Can't modify superblock while"
203			 "performing online resize");
204		return -EBUSY;
205	}
206
207	/*
208	 * We're only going to update primary superblock and two
209	 * backup superblocks in this transaction.
210	 */
211	handle = ext4_journal_start_sb(sb, EXT4_HT_MISC, 3);
212	if (IS_ERR(handle)) {
213		err = PTR_ERR(handle);
214		goto out;
215	}
216
217	/* Update primary superblock */
218	err = ext4_update_primary_sb(sb, handle, func, arg);
219	if (err) {
220		ext4_msg(sb, KERN_ERR, "Failed to update primary "
221			 "superblock");
222		goto out_journal;
223	}
224
225	primary_grp = ext4_get_group_number(sb, sbi->s_sbh->b_blocknr);
226	ngroups = ext4_get_groups_count(sb);
227
228	/*
229	 * Update backup superblocks. We have to start from group 0
230	 * because it might not be where the primary superblock is
231	 * if the fs is mounted with -o sb=<backup_sb_block>
232	 */
233	i = 0;
234	grp = 0;
235	while (grp < ngroups) {
236		/* Skip primary superblock */
237		if (grp == primary_grp)
238			goto next_grp;
239
240		ret = ext4_update_backup_sb(sb, handle, grp, func, arg);
241		if (ret < 0) {
242			/* Ignore bad checksum; try to update next sb */
243			if (ret == -EFSBADCRC)
244				goto next_grp;
245			err = ret;
246			goto out_journal;
247		}
248
249		i += ret;
250		if (handle && i > 1) {
251			/*
252			 * We're only journalling primary superblock and
253			 * two backup superblocks; the rest is not
254			 * journalled.
255			 */
256			err = ext4_journal_stop(handle);
257			if (err)
258				goto out;
259			handle = NULL;
260		}
261next_grp:
262		grp = ext4_list_backups(sb, &three, &five, &seven);
263	}
264
265out_journal:
266	if (handle) {
267		ret = ext4_journal_stop(handle);
268		if (ret && !err)
269			err = ret;
270	}
271out:
272	clear_bit_unlock(EXT4_FLAGS_RESIZING, &sbi->s_ext4_flags);
273	smp_mb__after_atomic();
274	return err ? err : 0;
275}
276
277/*
278 * Swap memory between @a and @b for @len bytes.
279 *
280 * @a:          pointer to first memory area
281 * @b:          pointer to second memory area
282 * @len:        number of bytes to swap
283 *
284 */
285static void memswap(void *a, void *b, size_t len)
286{
287	unsigned char *ap, *bp;
288
289	ap = (unsigned char *)a;
290	bp = (unsigned char *)b;
291	while (len-- > 0) {
292		swap(*ap, *bp);
293		ap++;
294		bp++;
295	}
296}
297
298/*
299 * Swap i_data and associated attributes between @inode1 and @inode2.
300 * This function is used for the primary swap between inode1 and inode2
301 * and also to revert this primary swap in case of errors.
302 *
303 * Therefore you have to make sure, that calling this method twice
304 * will revert all changes.
305 *
306 * @inode1:     pointer to first inode
307 * @inode2:     pointer to second inode
308 */
309static void swap_inode_data(struct inode *inode1, struct inode *inode2)
310{
311	loff_t isize;
312	struct ext4_inode_info *ei1;
313	struct ext4_inode_info *ei2;
314	unsigned long tmp;
315
316	ei1 = EXT4_I(inode1);
317	ei2 = EXT4_I(inode2);
318
319	swap(inode1->i_version, inode2->i_version);
320	swap(inode1->i_atime, inode2->i_atime);
321	swap(inode1->i_mtime, inode2->i_mtime);
322
323	memswap(ei1->i_data, ei2->i_data, sizeof(ei1->i_data));
324	tmp = ei1->i_flags & EXT4_FL_SHOULD_SWAP;
325	ei1->i_flags = (ei2->i_flags & EXT4_FL_SHOULD_SWAP) |
326		(ei1->i_flags & ~EXT4_FL_SHOULD_SWAP);
327	ei2->i_flags = tmp | (ei2->i_flags & ~EXT4_FL_SHOULD_SWAP);
328	swap(ei1->i_disksize, ei2->i_disksize);
329	ext4_es_remove_extent(inode1, 0, EXT_MAX_BLOCKS);
330	ext4_es_remove_extent(inode2, 0, EXT_MAX_BLOCKS);
331
332	isize = i_size_read(inode1);
333	i_size_write(inode1, i_size_read(inode2));
334	i_size_write(inode2, isize);
335}
336
337void ext4_reset_inode_seed(struct inode *inode)
338{
339	struct ext4_inode_info *ei = EXT4_I(inode);
340	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
341	__le32 inum = cpu_to_le32(inode->i_ino);
342	__le32 gen = cpu_to_le32(inode->i_generation);
343	__u32 csum;
344
345	if (!ext4_has_metadata_csum(inode->i_sb))
346		return;
347
348	csum = ext4_chksum(sbi, sbi->s_csum_seed, (__u8 *)&inum, sizeof(inum));
349	ei->i_csum_seed = ext4_chksum(sbi, csum, (__u8 *)&gen, sizeof(gen));
350}
351
352/*
353 * Swap the information from the given @inode and the inode
354 * EXT4_BOOT_LOADER_INO. It will basically swap i_data and all other
355 * important fields of the inodes.
356 *
357 * @sb:         the super block of the filesystem
358 * @idmap:	idmap of the mount the inode was found from
359 * @inode:      the inode to swap with EXT4_BOOT_LOADER_INO
360 *
361 */
362static long swap_inode_boot_loader(struct super_block *sb,
363				struct mnt_idmap *idmap,
364				struct inode *inode)
365{
366	handle_t *handle;
367	int err;
368	struct inode *inode_bl;
369	struct ext4_inode_info *ei_bl;
370	qsize_t size, size_bl, diff;
371	blkcnt_t blocks;
372	unsigned short bytes;
373
374	inode_bl = ext4_iget(sb, EXT4_BOOT_LOADER_INO,
375			EXT4_IGET_SPECIAL | EXT4_IGET_BAD);
376	if (IS_ERR(inode_bl))
377		return PTR_ERR(inode_bl);
378	ei_bl = EXT4_I(inode_bl);
379
380	/* Protect orig inodes against a truncate and make sure,
381	 * that only 1 swap_inode_boot_loader is running. */
382	lock_two_nondirectories(inode, inode_bl);
383
384	if (inode->i_nlink != 1 || !S_ISREG(inode->i_mode) ||
385	    IS_SWAPFILE(inode) || IS_ENCRYPTED(inode) ||
386	    (EXT4_I(inode)->i_flags & EXT4_JOURNAL_DATA_FL) ||
387	    ext4_has_inline_data(inode)) {
388		err = -EINVAL;
389		goto journal_err_out;
390	}
391
392	if (IS_RDONLY(inode) || IS_APPEND(inode) || IS_IMMUTABLE(inode) ||
393	    !inode_owner_or_capable(idmap, inode) ||
394	    !capable(CAP_SYS_ADMIN)) {
395		err = -EPERM;
396		goto journal_err_out;
397	}
398
399	filemap_invalidate_lock(inode->i_mapping);
400	err = filemap_write_and_wait(inode->i_mapping);
401	if (err)
402		goto err_out;
403
404	err = filemap_write_and_wait(inode_bl->i_mapping);
405	if (err)
406		goto err_out;
407
408	/* Wait for all existing dio workers */
409	inode_dio_wait(inode);
410	inode_dio_wait(inode_bl);
411
412	truncate_inode_pages(&inode->i_data, 0);
413	truncate_inode_pages(&inode_bl->i_data, 0);
414
415	handle = ext4_journal_start(inode_bl, EXT4_HT_MOVE_EXTENTS, 2);
416	if (IS_ERR(handle)) {
417		err = -EINVAL;
418		goto err_out;
419	}
420	ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_SWAP_BOOT, handle);
421
422	/* Protect extent tree against block allocations via delalloc */
423	ext4_double_down_write_data_sem(inode, inode_bl);
424
425	if (is_bad_inode(inode_bl) || !S_ISREG(inode_bl->i_mode)) {
426		/* this inode has never been used as a BOOT_LOADER */
427		set_nlink(inode_bl, 1);
428		i_uid_write(inode_bl, 0);
429		i_gid_write(inode_bl, 0);
430		inode_bl->i_flags = 0;
431		ei_bl->i_flags = 0;
432		inode_set_iversion(inode_bl, 1);
433		i_size_write(inode_bl, 0);
434		EXT4_I(inode_bl)->i_disksize = inode_bl->i_size;
435		inode_bl->i_mode = S_IFREG;
436		if (ext4_has_feature_extents(sb)) {
437			ext4_set_inode_flag(inode_bl, EXT4_INODE_EXTENTS);
438			ext4_ext_tree_init(handle, inode_bl);
439		} else
440			memset(ei_bl->i_data, 0, sizeof(ei_bl->i_data));
441	}
442
443	err = dquot_initialize(inode);
444	if (err)
445		goto err_out1;
446
447	size = (qsize_t)(inode->i_blocks) * (1 << 9) + inode->i_bytes;
448	size_bl = (qsize_t)(inode_bl->i_blocks) * (1 << 9) + inode_bl->i_bytes;
449	diff = size - size_bl;
450	swap_inode_data(inode, inode_bl);
451
452	inode_set_ctime_current(inode);
453	inode_set_ctime_current(inode_bl);
454	inode_inc_iversion(inode);
455
456	inode->i_generation = get_random_u32();
457	inode_bl->i_generation = get_random_u32();
458	ext4_reset_inode_seed(inode);
459	ext4_reset_inode_seed(inode_bl);
460
461	ext4_discard_preallocations(inode, 0);
462
463	err = ext4_mark_inode_dirty(handle, inode);
464	if (err < 0) {
465		/* No need to update quota information. */
466		ext4_warning(inode->i_sb,
467			"couldn't mark inode #%lu dirty (err %d)",
468			inode->i_ino, err);
469		/* Revert all changes: */
470		swap_inode_data(inode, inode_bl);
471		ext4_mark_inode_dirty(handle, inode);
472		goto err_out1;
473	}
474
475	blocks = inode_bl->i_blocks;
476	bytes = inode_bl->i_bytes;
477	inode_bl->i_blocks = inode->i_blocks;
478	inode_bl->i_bytes = inode->i_bytes;
479	err = ext4_mark_inode_dirty(handle, inode_bl);
480	if (err < 0) {
481		/* No need to update quota information. */
482		ext4_warning(inode_bl->i_sb,
483			"couldn't mark inode #%lu dirty (err %d)",
484			inode_bl->i_ino, err);
485		goto revert;
486	}
487
488	/* Bootloader inode should not be counted into quota information. */
489	if (diff > 0)
490		dquot_free_space(inode, diff);
491	else
492		err = dquot_alloc_space(inode, -1 * diff);
493
494	if (err < 0) {
495revert:
496		/* Revert all changes: */
497		inode_bl->i_blocks = blocks;
498		inode_bl->i_bytes = bytes;
499		swap_inode_data(inode, inode_bl);
500		ext4_mark_inode_dirty(handle, inode);
501		ext4_mark_inode_dirty(handle, inode_bl);
502	}
503
504err_out1:
505	ext4_journal_stop(handle);
506	ext4_double_up_write_data_sem(inode, inode_bl);
507
508err_out:
509	filemap_invalidate_unlock(inode->i_mapping);
510journal_err_out:
511	unlock_two_nondirectories(inode, inode_bl);
512	iput(inode_bl);
513	return err;
514}
515
516/*
517 * If immutable is set and we are not clearing it, we're not allowed to change
518 * anything else in the inode.  Don't error out if we're only trying to set
519 * immutable on an immutable file.
520 */
521static int ext4_ioctl_check_immutable(struct inode *inode, __u32 new_projid,
522				      unsigned int flags)
523{
524	struct ext4_inode_info *ei = EXT4_I(inode);
525	unsigned int oldflags = ei->i_flags;
526
527	if (!(oldflags & EXT4_IMMUTABLE_FL) || !(flags & EXT4_IMMUTABLE_FL))
528		return 0;
529
530	if ((oldflags & ~EXT4_IMMUTABLE_FL) != (flags & ~EXT4_IMMUTABLE_FL))
531		return -EPERM;
532	if (ext4_has_feature_project(inode->i_sb) &&
533	    __kprojid_val(ei->i_projid) != new_projid)
534		return -EPERM;
535
536	return 0;
537}
538
539static void ext4_dax_dontcache(struct inode *inode, unsigned int flags)
540{
541	struct ext4_inode_info *ei = EXT4_I(inode);
542
543	if (S_ISDIR(inode->i_mode))
544		return;
545
546	if (test_opt2(inode->i_sb, DAX_NEVER) ||
547	    test_opt(inode->i_sb, DAX_ALWAYS))
548		return;
549
550	if ((ei->i_flags ^ flags) & EXT4_DAX_FL)
551		d_mark_dontcache(inode);
552}
553
554static bool dax_compatible(struct inode *inode, unsigned int oldflags,
555			   unsigned int flags)
556{
557	/* Allow the DAX flag to be changed on inline directories */
558	if (S_ISDIR(inode->i_mode)) {
559		flags &= ~EXT4_INLINE_DATA_FL;
560		oldflags &= ~EXT4_INLINE_DATA_FL;
561	}
562
563	if (flags & EXT4_DAX_FL) {
564		if ((oldflags & EXT4_DAX_MUT_EXCL) ||
565		     ext4_test_inode_state(inode,
566					  EXT4_STATE_VERITY_IN_PROGRESS)) {
567			return false;
568		}
569	}
570
571	if ((flags & EXT4_DAX_MUT_EXCL) && (oldflags & EXT4_DAX_FL))
572			return false;
573
574	return true;
575}
576
577static int ext4_ioctl_setflags(struct inode *inode,
578			       unsigned int flags)
579{
580	struct ext4_inode_info *ei = EXT4_I(inode);
581	handle_t *handle = NULL;
582	int err = -EPERM, migrate = 0;
583	struct ext4_iloc iloc;
584	unsigned int oldflags, mask, i;
585	struct super_block *sb = inode->i_sb;
586
587	/* Is it quota file? Do not allow user to mess with it */
588	if (ext4_is_quota_file(inode))
589		goto flags_out;
590
591	oldflags = ei->i_flags;
592	/*
593	 * The JOURNAL_DATA flag can only be changed by
594	 * the relevant capability.
595	 */
596	if ((flags ^ oldflags) & (EXT4_JOURNAL_DATA_FL)) {
597		if (!capable(CAP_SYS_RESOURCE))
598			goto flags_out;
599	}
600
601	if (!dax_compatible(inode, oldflags, flags)) {
602		err = -EOPNOTSUPP;
603		goto flags_out;
604	}
605
606	if ((flags ^ oldflags) & EXT4_EXTENTS_FL)
607		migrate = 1;
608
609	if ((flags ^ oldflags) & EXT4_CASEFOLD_FL) {
610		if (!ext4_has_feature_casefold(sb)) {
611			err = -EOPNOTSUPP;
612			goto flags_out;
613		}
614
615		if (!S_ISDIR(inode->i_mode)) {
616			err = -ENOTDIR;
617			goto flags_out;
618		}
619
620		if (!ext4_empty_dir(inode)) {
621			err = -ENOTEMPTY;
622			goto flags_out;
623		}
624	}
625
626	/*
627	 * Wait for all pending directio and then flush all the dirty pages
628	 * for this file.  The flush marks all the pages readonly, so any
629	 * subsequent attempt to write to the file (particularly mmap pages)
630	 * will come through the filesystem and fail.
631	 */
632	if (S_ISREG(inode->i_mode) && !IS_IMMUTABLE(inode) &&
633	    (flags & EXT4_IMMUTABLE_FL)) {
634		inode_dio_wait(inode);
635		err = filemap_write_and_wait(inode->i_mapping);
636		if (err)
637			goto flags_out;
638	}
639
640	handle = ext4_journal_start(inode, EXT4_HT_INODE, 1);
641	if (IS_ERR(handle)) {
642		err = PTR_ERR(handle);
643		goto flags_out;
644	}
645	if (IS_SYNC(inode))
646		ext4_handle_sync(handle);
647	err = ext4_reserve_inode_write(handle, inode, &iloc);
648	if (err)
649		goto flags_err;
650
651	ext4_dax_dontcache(inode, flags);
652
653	for (i = 0, mask = 1; i < 32; i++, mask <<= 1) {
654		if (!(mask & EXT4_FL_USER_MODIFIABLE))
655			continue;
656		/* These flags get special treatment later */
657		if (mask == EXT4_JOURNAL_DATA_FL || mask == EXT4_EXTENTS_FL)
658			continue;
659		if (mask & flags)
660			ext4_set_inode_flag(inode, i);
661		else
662			ext4_clear_inode_flag(inode, i);
663	}
664
665	ext4_set_inode_flags(inode, false);
666
667	inode_set_ctime_current(inode);
668	inode_inc_iversion(inode);
669
670	err = ext4_mark_iloc_dirty(handle, inode, &iloc);
671flags_err:
672	ext4_journal_stop(handle);
673	if (err)
674		goto flags_out;
675
676	if ((flags ^ oldflags) & (EXT4_JOURNAL_DATA_FL)) {
677		/*
678		 * Changes to the journaling mode can cause unsafe changes to
679		 * S_DAX if the inode is DAX
680		 */
681		if (IS_DAX(inode)) {
682			err = -EBUSY;
683			goto flags_out;
684		}
685
686		err = ext4_change_inode_journal_flag(inode,
687						     flags & EXT4_JOURNAL_DATA_FL);
688		if (err)
689			goto flags_out;
690	}
691	if (migrate) {
692		if (flags & EXT4_EXTENTS_FL)
693			err = ext4_ext_migrate(inode);
694		else
695			err = ext4_ind_migrate(inode);
696	}
697
698flags_out:
699	return err;
700}
701
702#ifdef CONFIG_QUOTA
703static int ext4_ioctl_setproject(struct inode *inode, __u32 projid)
704{
705	struct super_block *sb = inode->i_sb;
706	struct ext4_inode_info *ei = EXT4_I(inode);
707	int err, rc;
708	handle_t *handle;
709	kprojid_t kprojid;
710	struct ext4_iloc iloc;
711	struct ext4_inode *raw_inode;
712	struct dquot *transfer_to[MAXQUOTAS] = { };
713
714	if (!ext4_has_feature_project(sb)) {
715		if (projid != EXT4_DEF_PROJID)
716			return -EOPNOTSUPP;
717		else
718			return 0;
719	}
720
721	if (EXT4_INODE_SIZE(sb) <= EXT4_GOOD_OLD_INODE_SIZE)
722		return -EOPNOTSUPP;
723
724	kprojid = make_kprojid(&init_user_ns, (projid_t)projid);
725
726	if (projid_eq(kprojid, EXT4_I(inode)->i_projid))
727		return 0;
728
729	err = -EPERM;
730	/* Is it quota file? Do not allow user to mess with it */
731	if (ext4_is_quota_file(inode))
732		return err;
733
734	err = dquot_initialize(inode);
735	if (err)
736		return err;
737
738	err = ext4_get_inode_loc(inode, &iloc);
739	if (err)
740		return err;
741
742	raw_inode = ext4_raw_inode(&iloc);
743	if (!EXT4_FITS_IN_INODE(raw_inode, ei, i_projid)) {
744		err = ext4_expand_extra_isize(inode,
745					      EXT4_SB(sb)->s_want_extra_isize,
746					      &iloc);
747		if (err)
748			return err;
749	} else {
750		brelse(iloc.bh);
751	}
752
753	handle = ext4_journal_start(inode, EXT4_HT_QUOTA,
754		EXT4_QUOTA_INIT_BLOCKS(sb) +
755		EXT4_QUOTA_DEL_BLOCKS(sb) + 3);
756	if (IS_ERR(handle))
757		return PTR_ERR(handle);
758
759	err = ext4_reserve_inode_write(handle, inode, &iloc);
760	if (err)
761		goto out_stop;
762
763	transfer_to[PRJQUOTA] = dqget(sb, make_kqid_projid(kprojid));
764	if (!IS_ERR(transfer_to[PRJQUOTA])) {
765
766		/* __dquot_transfer() calls back ext4_get_inode_usage() which
767		 * counts xattr inode references.
768		 */
769		down_read(&EXT4_I(inode)->xattr_sem);
770		err = __dquot_transfer(inode, transfer_to);
771		up_read(&EXT4_I(inode)->xattr_sem);
772		dqput(transfer_to[PRJQUOTA]);
773		if (err)
774			goto out_dirty;
775	}
776
777	EXT4_I(inode)->i_projid = kprojid;
778	inode_set_ctime_current(inode);
779	inode_inc_iversion(inode);
780out_dirty:
781	rc = ext4_mark_iloc_dirty(handle, inode, &iloc);
782	if (!err)
783		err = rc;
784out_stop:
785	ext4_journal_stop(handle);
786	return err;
787}
788#else
789static int ext4_ioctl_setproject(struct inode *inode, __u32 projid)
790{
791	if (projid != EXT4_DEF_PROJID)
792		return -EOPNOTSUPP;
793	return 0;
794}
795#endif
796
797int ext4_force_shutdown(struct super_block *sb, u32 flags)
798{
799	struct ext4_sb_info *sbi = EXT4_SB(sb);
800	int ret;
801
802	if (flags > EXT4_GOING_FLAGS_NOLOGFLUSH)
803		return -EINVAL;
804
805	if (ext4_forced_shutdown(sb))
806		return 0;
807
808	ext4_msg(sb, KERN_ALERT, "shut down requested (%d)", flags);
809	trace_ext4_shutdown(sb, flags);
810
811	switch (flags) {
812	case EXT4_GOING_FLAGS_DEFAULT:
813		ret = freeze_bdev(sb->s_bdev);
814		if (ret)
815			return ret;
816		set_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
817		thaw_bdev(sb->s_bdev);
818		break;
819	case EXT4_GOING_FLAGS_LOGFLUSH:
820		set_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
821		if (sbi->s_journal && !is_journal_aborted(sbi->s_journal)) {
822			(void) ext4_force_commit(sb);
823			jbd2_journal_abort(sbi->s_journal, -ESHUTDOWN);
824		}
825		break;
826	case EXT4_GOING_FLAGS_NOLOGFLUSH:
827		set_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
828		if (sbi->s_journal && !is_journal_aborted(sbi->s_journal))
829			jbd2_journal_abort(sbi->s_journal, -ESHUTDOWN);
830		break;
831	default:
832		return -EINVAL;
833	}
834	clear_opt(sb, DISCARD);
835	return 0;
836}
837
838static int ext4_ioctl_shutdown(struct super_block *sb, unsigned long arg)
839{
840	u32 flags;
841
842	if (!capable(CAP_SYS_ADMIN))
843		return -EPERM;
844
845	if (get_user(flags, (__u32 __user *)arg))
846		return -EFAULT;
847
848	return ext4_force_shutdown(sb, flags);
849}
850
851struct getfsmap_info {
852	struct super_block	*gi_sb;
853	struct fsmap_head __user *gi_data;
854	unsigned int		gi_idx;
855	__u32			gi_last_flags;
856};
857
858static int ext4_getfsmap_format(struct ext4_fsmap *xfm, void *priv)
859{
860	struct getfsmap_info *info = priv;
861	struct fsmap fm;
862
863	trace_ext4_getfsmap_mapping(info->gi_sb, xfm);
864
865	info->gi_last_flags = xfm->fmr_flags;
866	ext4_fsmap_from_internal(info->gi_sb, &fm, xfm);
867	if (copy_to_user(&info->gi_data->fmh_recs[info->gi_idx++], &fm,
868			sizeof(struct fsmap)))
869		return -EFAULT;
870
871	return 0;
872}
873
874static int ext4_ioc_getfsmap(struct super_block *sb,
875			     struct fsmap_head __user *arg)
876{
877	struct getfsmap_info info = { NULL };
878	struct ext4_fsmap_head xhead = {0};
879	struct fsmap_head head;
880	bool aborted = false;
881	int error;
882
883	if (copy_from_user(&head, arg, sizeof(struct fsmap_head)))
884		return -EFAULT;
885	if (memchr_inv(head.fmh_reserved, 0, sizeof(head.fmh_reserved)) ||
886	    memchr_inv(head.fmh_keys[0].fmr_reserved, 0,
887		       sizeof(head.fmh_keys[0].fmr_reserved)) ||
888	    memchr_inv(head.fmh_keys[1].fmr_reserved, 0,
889		       sizeof(head.fmh_keys[1].fmr_reserved)))
890		return -EINVAL;
891	/*
892	 * ext4 doesn't report file extents at all, so the only valid
893	 * file offsets are the magic ones (all zeroes or all ones).
894	 */
895	if (head.fmh_keys[0].fmr_offset ||
896	    (head.fmh_keys[1].fmr_offset != 0 &&
897	     head.fmh_keys[1].fmr_offset != -1ULL))
898		return -EINVAL;
899
900	xhead.fmh_iflags = head.fmh_iflags;
901	xhead.fmh_count = head.fmh_count;
902	ext4_fsmap_to_internal(sb, &xhead.fmh_keys[0], &head.fmh_keys[0]);
903	ext4_fsmap_to_internal(sb, &xhead.fmh_keys[1], &head.fmh_keys[1]);
904
905	trace_ext4_getfsmap_low_key(sb, &xhead.fmh_keys[0]);
906	trace_ext4_getfsmap_high_key(sb, &xhead.fmh_keys[1]);
907
908	info.gi_sb = sb;
909	info.gi_data = arg;
910	error = ext4_getfsmap(sb, &xhead, ext4_getfsmap_format, &info);
911	if (error == EXT4_QUERY_RANGE_ABORT)
912		aborted = true;
913	else if (error)
914		return error;
915
916	/* If we didn't abort, set the "last" flag in the last fmx */
917	if (!aborted && info.gi_idx) {
918		info.gi_last_flags |= FMR_OF_LAST;
919		if (copy_to_user(&info.gi_data->fmh_recs[info.gi_idx - 1].fmr_flags,
920				 &info.gi_last_flags,
921				 sizeof(info.gi_last_flags)))
922			return -EFAULT;
923	}
924
925	/* copy back header */
926	head.fmh_entries = xhead.fmh_entries;
927	head.fmh_oflags = xhead.fmh_oflags;
928	if (copy_to_user(arg, &head, sizeof(struct fsmap_head)))
929		return -EFAULT;
930
931	return 0;
932}
933
934static long ext4_ioctl_group_add(struct file *file,
935				 struct ext4_new_group_data *input)
936{
937	struct super_block *sb = file_inode(file)->i_sb;
938	int err, err2=0;
939
940	err = ext4_resize_begin(sb);
941	if (err)
942		return err;
943
944	if (ext4_has_feature_bigalloc(sb)) {
945		ext4_msg(sb, KERN_ERR,
946			 "Online resizing not supported with bigalloc");
947		err = -EOPNOTSUPP;
948		goto group_add_out;
949	}
950
951	err = mnt_want_write_file(file);
952	if (err)
953		goto group_add_out;
954
955	err = ext4_group_add(sb, input);
956	if (EXT4_SB(sb)->s_journal) {
957		jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
958		err2 = jbd2_journal_flush(EXT4_SB(sb)->s_journal, 0);
959		jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
960	}
961	if (err == 0)
962		err = err2;
963	mnt_drop_write_file(file);
964	if (!err && ext4_has_group_desc_csum(sb) &&
965	    test_opt(sb, INIT_INODE_TABLE))
966		err = ext4_register_li_request(sb, input->group);
967group_add_out:
968	err2 = ext4_resize_end(sb, false);
969	if (err == 0)
970		err = err2;
971	return err;
972}
973
974int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa)
975{
976	struct inode *inode = d_inode(dentry);
977	struct ext4_inode_info *ei = EXT4_I(inode);
978	u32 flags = ei->i_flags & EXT4_FL_USER_VISIBLE;
979
980	if (S_ISREG(inode->i_mode))
981		flags &= ~FS_PROJINHERIT_FL;
982
983	fileattr_fill_flags(fa, flags);
984	if (ext4_has_feature_project(inode->i_sb))
985		fa->fsx_projid = from_kprojid(&init_user_ns, ei->i_projid);
986
987	return 0;
988}
989
990int ext4_fileattr_set(struct mnt_idmap *idmap,
991		      struct dentry *dentry, struct fileattr *fa)
992{
993	struct inode *inode = d_inode(dentry);
994	u32 flags = fa->flags;
995	int err = -EOPNOTSUPP;
996
997	if (flags & ~EXT4_FL_USER_VISIBLE)
998		goto out;
999
1000	/*
1001	 * chattr(1) grabs flags via GETFLAGS, modifies the result and
1002	 * passes that to SETFLAGS. So we cannot easily make SETFLAGS
1003	 * more restrictive than just silently masking off visible but
1004	 * not settable flags as we always did.
1005	 */
1006	flags &= EXT4_FL_USER_MODIFIABLE;
1007	if (ext4_mask_flags(inode->i_mode, flags) != flags)
1008		goto out;
1009	err = ext4_ioctl_check_immutable(inode, fa->fsx_projid, flags);
1010	if (err)
1011		goto out;
1012	err = ext4_ioctl_setflags(inode, flags);
1013	if (err)
1014		goto out;
1015	err = ext4_ioctl_setproject(inode, fa->fsx_projid);
1016out:
1017	return err;
1018}
1019
1020/* So that the fiemap access checks can't overflow on 32 bit machines. */
1021#define FIEMAP_MAX_EXTENTS	(UINT_MAX / sizeof(struct fiemap_extent))
1022
1023static int ext4_ioctl_get_es_cache(struct file *filp, unsigned long arg)
1024{
1025	struct fiemap fiemap;
1026	struct fiemap __user *ufiemap = (struct fiemap __user *) arg;
1027	struct fiemap_extent_info fieinfo = { 0, };
1028	struct inode *inode = file_inode(filp);
1029	int error;
1030
1031	if (copy_from_user(&fiemap, ufiemap, sizeof(fiemap)))
1032		return -EFAULT;
1033
1034	if (fiemap.fm_extent_count > FIEMAP_MAX_EXTENTS)
1035		return -EINVAL;
1036
1037	fieinfo.fi_flags = fiemap.fm_flags;
1038	fieinfo.fi_extents_max = fiemap.fm_extent_count;
1039	fieinfo.fi_extents_start = ufiemap->fm_extents;
1040
1041	error = ext4_get_es_cache(inode, &fieinfo, fiemap.fm_start,
1042			fiemap.fm_length);
1043	fiemap.fm_flags = fieinfo.fi_flags;
1044	fiemap.fm_mapped_extents = fieinfo.fi_extents_mapped;
1045	if (copy_to_user(ufiemap, &fiemap, sizeof(fiemap)))
1046		error = -EFAULT;
1047
1048	return error;
1049}
1050
1051static int ext4_ioctl_checkpoint(struct file *filp, unsigned long arg)
1052{
1053	int err = 0;
1054	__u32 flags = 0;
1055	unsigned int flush_flags = 0;
1056	struct super_block *sb = file_inode(filp)->i_sb;
1057
1058	if (copy_from_user(&flags, (__u32 __user *)arg,
1059				sizeof(__u32)))
1060		return -EFAULT;
1061
1062	if (!capable(CAP_SYS_ADMIN))
1063		return -EPERM;
1064
1065	/* check for invalid bits set */
1066	if ((flags & ~EXT4_IOC_CHECKPOINT_FLAG_VALID) ||
1067				((flags & JBD2_JOURNAL_FLUSH_DISCARD) &&
1068				(flags & JBD2_JOURNAL_FLUSH_ZEROOUT)))
1069		return -EINVAL;
1070
1071	if (!EXT4_SB(sb)->s_journal)
1072		return -ENODEV;
1073
1074	if ((flags & JBD2_JOURNAL_FLUSH_DISCARD) &&
1075	    !bdev_max_discard_sectors(EXT4_SB(sb)->s_journal->j_dev))
1076		return -EOPNOTSUPP;
1077
1078	if (flags & EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN)
1079		return 0;
1080
1081	if (flags & EXT4_IOC_CHECKPOINT_FLAG_DISCARD)
1082		flush_flags |= JBD2_JOURNAL_FLUSH_DISCARD;
1083
1084	if (flags & EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT) {
1085		flush_flags |= JBD2_JOURNAL_FLUSH_ZEROOUT;
1086		pr_info_ratelimited("warning: checkpointing journal with EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT can be slow");
1087	}
1088
1089	jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
1090	err = jbd2_journal_flush(EXT4_SB(sb)->s_journal, flush_flags);
1091	jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
1092
1093	return err;
1094}
1095
1096static int ext4_ioctl_setlabel(struct file *filp, const char __user *user_label)
1097{
1098	size_t len;
1099	int ret = 0;
1100	char new_label[EXT4_LABEL_MAX + 1];
1101	struct super_block *sb = file_inode(filp)->i_sb;
1102
1103	if (!capable(CAP_SYS_ADMIN))
1104		return -EPERM;
1105
1106	/*
1107	 * Copy the maximum length allowed for ext4 label with one more to
1108	 * find the required terminating null byte in order to test the
1109	 * label length. The on disk label doesn't need to be null terminated.
1110	 */
1111	if (copy_from_user(new_label, user_label, EXT4_LABEL_MAX + 1))
1112		return -EFAULT;
1113
1114	len = strnlen(new_label, EXT4_LABEL_MAX + 1);
1115	if (len > EXT4_LABEL_MAX)
1116		return -EINVAL;
1117
1118	/*
1119	 * Clear the buffer after the new label
1120	 */
1121	memset(new_label + len, 0, EXT4_LABEL_MAX - len);
1122
1123	ret = mnt_want_write_file(filp);
1124	if (ret)
1125		return ret;
1126
1127	ret = ext4_update_superblocks_fn(sb, ext4_sb_setlabel, new_label);
1128
1129	mnt_drop_write_file(filp);
1130	return ret;
1131}
1132
1133static int ext4_ioctl_getlabel(struct ext4_sb_info *sbi, char __user *user_label)
1134{
1135	char label[EXT4_LABEL_MAX + 1];
1136
1137	/*
1138	 * EXT4_LABEL_MAX must always be smaller than FSLABEL_MAX because
1139	 * FSLABEL_MAX must include terminating null byte, while s_volume_name
1140	 * does not have to.
1141	 */
1142	BUILD_BUG_ON(EXT4_LABEL_MAX >= FSLABEL_MAX);
1143
1144	memset(label, 0, sizeof(label));
1145	lock_buffer(sbi->s_sbh);
1146	strncpy(label, sbi->s_es->s_volume_name, EXT4_LABEL_MAX);
1147	unlock_buffer(sbi->s_sbh);
1148
1149	if (copy_to_user(user_label, label, sizeof(label)))
1150		return -EFAULT;
1151	return 0;
1152}
1153
1154static int ext4_ioctl_getuuid(struct ext4_sb_info *sbi,
1155			struct fsuuid __user *ufsuuid)
1156{
1157	struct fsuuid fsuuid;
1158	__u8 uuid[UUID_SIZE];
1159
1160	if (copy_from_user(&fsuuid, ufsuuid, sizeof(fsuuid)))
1161		return -EFAULT;
1162
1163	if (fsuuid.fsu_len == 0) {
1164		fsuuid.fsu_len = UUID_SIZE;
1165		if (copy_to_user(&ufsuuid->fsu_len, &fsuuid.fsu_len,
1166					sizeof(fsuuid.fsu_len)))
1167			return -EFAULT;
1168		return 0;
1169	}
1170
1171	if (fsuuid.fsu_len < UUID_SIZE || fsuuid.fsu_flags != 0)
1172		return -EINVAL;
1173
1174	lock_buffer(sbi->s_sbh);
1175	memcpy(uuid, sbi->s_es->s_uuid, UUID_SIZE);
1176	unlock_buffer(sbi->s_sbh);
1177
1178	fsuuid.fsu_len = UUID_SIZE;
1179	if (copy_to_user(ufsuuid, &fsuuid, sizeof(fsuuid)) ||
1180	    copy_to_user(&ufsuuid->fsu_uuid[0], uuid, UUID_SIZE))
1181		return -EFAULT;
1182	return 0;
1183}
1184
1185static int ext4_ioctl_setuuid(struct file *filp,
1186			const struct fsuuid __user *ufsuuid)
1187{
1188	int ret = 0;
1189	struct super_block *sb = file_inode(filp)->i_sb;
1190	struct fsuuid fsuuid;
1191	__u8 uuid[UUID_SIZE];
1192
1193	if (!capable(CAP_SYS_ADMIN))
1194		return -EPERM;
1195
1196	/*
1197	 * If any checksums (group descriptors or metadata) are being used
1198	 * then the checksum seed feature is required to change the UUID.
1199	 */
1200	if (((ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb))
1201			&& !ext4_has_feature_csum_seed(sb))
1202		|| ext4_has_feature_stable_inodes(sb))
1203		return -EOPNOTSUPP;
1204
1205	if (copy_from_user(&fsuuid, ufsuuid, sizeof(fsuuid)))
1206		return -EFAULT;
1207
1208	if (fsuuid.fsu_len != UUID_SIZE || fsuuid.fsu_flags != 0)
1209		return -EINVAL;
1210
1211	if (copy_from_user(uuid, &ufsuuid->fsu_uuid[0], UUID_SIZE))
1212		return -EFAULT;
1213
1214	ret = mnt_want_write_file(filp);
1215	if (ret)
1216		return ret;
1217
1218	ret = ext4_update_superblocks_fn(sb, ext4_sb_setuuid, &uuid);
1219	mnt_drop_write_file(filp);
1220
1221	return ret;
1222}
1223
1224static long __ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
1225{
1226	struct inode *inode = file_inode(filp);
1227	struct super_block *sb = inode->i_sb;
1228	struct mnt_idmap *idmap = file_mnt_idmap(filp);
1229
1230	ext4_debug("cmd = %u, arg = %lu\n", cmd, arg);
1231
1232	switch (cmd) {
1233	case FS_IOC_GETFSMAP:
1234		return ext4_ioc_getfsmap(sb, (void __user *)arg);
1235	case EXT4_IOC_GETVERSION:
1236	case EXT4_IOC_GETVERSION_OLD:
1237		return put_user(inode->i_generation, (int __user *) arg);
1238	case EXT4_IOC_SETVERSION:
1239	case EXT4_IOC_SETVERSION_OLD: {
1240		handle_t *handle;
1241		struct ext4_iloc iloc;
1242		__u32 generation;
1243		int err;
1244
1245		if (!inode_owner_or_capable(idmap, inode))
1246			return -EPERM;
1247
1248		if (ext4_has_metadata_csum(inode->i_sb)) {
1249			ext4_warning(sb, "Setting inode version is not "
1250				     "supported with metadata_csum enabled.");
1251			return -ENOTTY;
1252		}
1253
1254		err = mnt_want_write_file(filp);
1255		if (err)
1256			return err;
1257		if (get_user(generation, (int __user *) arg)) {
1258			err = -EFAULT;
1259			goto setversion_out;
1260		}
1261
1262		inode_lock(inode);
1263		handle = ext4_journal_start(inode, EXT4_HT_INODE, 1);
1264		if (IS_ERR(handle)) {
1265			err = PTR_ERR(handle);
1266			goto unlock_out;
1267		}
1268		err = ext4_reserve_inode_write(handle, inode, &iloc);
1269		if (err == 0) {
1270			inode_set_ctime_current(inode);
1271			inode_inc_iversion(inode);
1272			inode->i_generation = generation;
1273			err = ext4_mark_iloc_dirty(handle, inode, &iloc);
1274		}
1275		ext4_journal_stop(handle);
1276
1277unlock_out:
1278		inode_unlock(inode);
1279setversion_out:
1280		mnt_drop_write_file(filp);
1281		return err;
1282	}
1283	case EXT4_IOC_GROUP_EXTEND: {
1284		ext4_fsblk_t n_blocks_count;
1285		int err, err2=0;
1286
1287		err = ext4_resize_begin(sb);
1288		if (err)
1289			return err;
1290
1291		if (get_user(n_blocks_count, (__u32 __user *)arg)) {
1292			err = -EFAULT;
1293			goto group_extend_out;
1294		}
1295
1296		if (ext4_has_feature_bigalloc(sb)) {
1297			ext4_msg(sb, KERN_ERR,
1298				 "Online resizing not supported with bigalloc");
1299			err = -EOPNOTSUPP;
1300			goto group_extend_out;
1301		}
1302
1303		err = mnt_want_write_file(filp);
1304		if (err)
1305			goto group_extend_out;
1306
1307		err = ext4_group_extend(sb, EXT4_SB(sb)->s_es, n_blocks_count);
1308		if (EXT4_SB(sb)->s_journal) {
1309			jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
1310			err2 = jbd2_journal_flush(EXT4_SB(sb)->s_journal, 0);
1311			jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
1312		}
1313		if (err == 0)
1314			err = err2;
1315		mnt_drop_write_file(filp);
1316group_extend_out:
1317		err2 = ext4_resize_end(sb, false);
1318		if (err == 0)
1319			err = err2;
1320		return err;
1321	}
1322
1323	case EXT4_IOC_MOVE_EXT: {
1324		struct move_extent me;
1325		struct fd donor;
1326		int err;
1327
1328		if (!(filp->f_mode & FMODE_READ) ||
1329		    !(filp->f_mode & FMODE_WRITE))
1330			return -EBADF;
1331
1332		if (copy_from_user(&me,
1333			(struct move_extent __user *)arg, sizeof(me)))
1334			return -EFAULT;
1335		me.moved_len = 0;
1336
1337		donor = fdget(me.donor_fd);
1338		if (!donor.file)
1339			return -EBADF;
1340
1341		if (!(donor.file->f_mode & FMODE_WRITE)) {
1342			err = -EBADF;
1343			goto mext_out;
1344		}
1345
1346		if (ext4_has_feature_bigalloc(sb)) {
1347			ext4_msg(sb, KERN_ERR,
1348				 "Online defrag not supported with bigalloc");
1349			err = -EOPNOTSUPP;
1350			goto mext_out;
1351		} else if (IS_DAX(inode)) {
1352			ext4_msg(sb, KERN_ERR,
1353				 "Online defrag not supported with DAX");
1354			err = -EOPNOTSUPP;
1355			goto mext_out;
1356		}
1357
1358		err = mnt_want_write_file(filp);
1359		if (err)
1360			goto mext_out;
1361
1362		err = ext4_move_extents(filp, donor.file, me.orig_start,
1363					me.donor_start, me.len, &me.moved_len);
1364		mnt_drop_write_file(filp);
1365
1366		if (copy_to_user((struct move_extent __user *)arg,
1367				 &me, sizeof(me)))
1368			err = -EFAULT;
1369mext_out:
1370		fdput(donor);
1371		return err;
1372	}
1373
1374	case EXT4_IOC_GROUP_ADD: {
1375		struct ext4_new_group_data input;
1376
1377		if (copy_from_user(&input, (struct ext4_new_group_input __user *)arg,
1378				sizeof(input)))
1379			return -EFAULT;
1380
1381		return ext4_ioctl_group_add(filp, &input);
1382	}
1383
1384	case EXT4_IOC_MIGRATE:
1385	{
1386		int err;
1387		if (!inode_owner_or_capable(idmap, inode))
1388			return -EACCES;
1389
1390		err = mnt_want_write_file(filp);
1391		if (err)
1392			return err;
1393		/*
1394		 * inode_mutex prevent write and truncate on the file.
1395		 * Read still goes through. We take i_data_sem in
1396		 * ext4_ext_swap_inode_data before we switch the
1397		 * inode format to prevent read.
1398		 */
1399		inode_lock((inode));
1400		err = ext4_ext_migrate(inode);
1401		inode_unlock((inode));
1402		mnt_drop_write_file(filp);
1403		return err;
1404	}
1405
1406	case EXT4_IOC_ALLOC_DA_BLKS:
1407	{
1408		int err;
1409		if (!inode_owner_or_capable(idmap, inode))
1410			return -EACCES;
1411
1412		err = mnt_want_write_file(filp);
1413		if (err)
1414			return err;
1415		err = ext4_alloc_da_blocks(inode);
1416		mnt_drop_write_file(filp);
1417		return err;
1418	}
1419
1420	case EXT4_IOC_SWAP_BOOT:
1421	{
1422		int err;
1423		if (!(filp->f_mode & FMODE_WRITE))
1424			return -EBADF;
1425		err = mnt_want_write_file(filp);
1426		if (err)
1427			return err;
1428		err = swap_inode_boot_loader(sb, idmap, inode);
1429		mnt_drop_write_file(filp);
1430		return err;
1431	}
1432
1433	case EXT4_IOC_RESIZE_FS: {
1434		ext4_fsblk_t n_blocks_count;
1435		int err = 0, err2 = 0;
1436		ext4_group_t o_group = EXT4_SB(sb)->s_groups_count;
1437
1438		if (copy_from_user(&n_blocks_count, (__u64 __user *)arg,
1439				   sizeof(__u64))) {
1440			return -EFAULT;
1441		}
1442
1443		err = ext4_resize_begin(sb);
1444		if (err)
1445			return err;
1446
1447		err = mnt_want_write_file(filp);
1448		if (err)
1449			goto resizefs_out;
1450
1451		err = ext4_resize_fs(sb, n_blocks_count);
1452		if (EXT4_SB(sb)->s_journal) {
1453			ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_RESIZE, NULL);
1454			jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
1455			err2 = jbd2_journal_flush(EXT4_SB(sb)->s_journal, 0);
1456			jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
1457		}
1458		if (err == 0)
1459			err = err2;
1460		mnt_drop_write_file(filp);
1461		if (!err && (o_group < EXT4_SB(sb)->s_groups_count) &&
1462		    ext4_has_group_desc_csum(sb) &&
1463		    test_opt(sb, INIT_INODE_TABLE))
1464			err = ext4_register_li_request(sb, o_group);
1465
1466resizefs_out:
1467		err2 = ext4_resize_end(sb, true);
1468		if (err == 0)
1469			err = err2;
1470		return err;
1471	}
1472
1473	case FITRIM:
1474	{
1475		struct fstrim_range range;
1476		int ret = 0;
1477
1478		if (!capable(CAP_SYS_ADMIN))
1479			return -EPERM;
1480
1481		if (!bdev_max_discard_sectors(sb->s_bdev))
1482			return -EOPNOTSUPP;
1483
1484		/*
1485		 * We haven't replayed the journal, so we cannot use our
1486		 * block-bitmap-guided storage zapping commands.
1487		 */
1488		if (test_opt(sb, NOLOAD) && ext4_has_feature_journal(sb))
1489			return -EROFS;
1490
1491		if (copy_from_user(&range, (struct fstrim_range __user *)arg,
1492		    sizeof(range)))
1493			return -EFAULT;
1494
1495		ret = ext4_trim_fs(sb, &range);
1496		if (ret < 0)
1497			return ret;
1498
1499		if (copy_to_user((struct fstrim_range __user *)arg, &range,
1500		    sizeof(range)))
1501			return -EFAULT;
1502
1503		return 0;
1504	}
1505	case EXT4_IOC_PRECACHE_EXTENTS:
1506		return ext4_ext_precache(inode);
1507
1508	case FS_IOC_SET_ENCRYPTION_POLICY:
1509		if (!ext4_has_feature_encrypt(sb))
1510			return -EOPNOTSUPP;
1511		return fscrypt_ioctl_set_policy(filp, (const void __user *)arg);
1512
1513	case FS_IOC_GET_ENCRYPTION_PWSALT:
1514		return ext4_ioctl_get_encryption_pwsalt(filp, (void __user *)arg);
1515
1516	case FS_IOC_GET_ENCRYPTION_POLICY:
1517		if (!ext4_has_feature_encrypt(sb))
1518			return -EOPNOTSUPP;
1519		return fscrypt_ioctl_get_policy(filp, (void __user *)arg);
1520
1521	case FS_IOC_GET_ENCRYPTION_POLICY_EX:
1522		if (!ext4_has_feature_encrypt(sb))
1523			return -EOPNOTSUPP;
1524		return fscrypt_ioctl_get_policy_ex(filp, (void __user *)arg);
1525
1526	case FS_IOC_ADD_ENCRYPTION_KEY:
1527		if (!ext4_has_feature_encrypt(sb))
1528			return -EOPNOTSUPP;
1529		return fscrypt_ioctl_add_key(filp, (void __user *)arg);
1530
1531	case FS_IOC_REMOVE_ENCRYPTION_KEY:
1532		if (!ext4_has_feature_encrypt(sb))
1533			return -EOPNOTSUPP;
1534		return fscrypt_ioctl_remove_key(filp, (void __user *)arg);
1535
1536	case FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS:
1537		if (!ext4_has_feature_encrypt(sb))
1538			return -EOPNOTSUPP;
1539		return fscrypt_ioctl_remove_key_all_users(filp,
1540							  (void __user *)arg);
1541	case FS_IOC_GET_ENCRYPTION_KEY_STATUS:
1542		if (!ext4_has_feature_encrypt(sb))
1543			return -EOPNOTSUPP;
1544		return fscrypt_ioctl_get_key_status(filp, (void __user *)arg);
1545
1546	case FS_IOC_GET_ENCRYPTION_NONCE:
1547		if (!ext4_has_feature_encrypt(sb))
1548			return -EOPNOTSUPP;
1549		return fscrypt_ioctl_get_nonce(filp, (void __user *)arg);
1550
1551	case EXT4_IOC_CLEAR_ES_CACHE:
1552	{
1553		if (!inode_owner_or_capable(idmap, inode))
1554			return -EACCES;
1555		ext4_clear_inode_es(inode);
1556		return 0;
1557	}
1558
1559	case EXT4_IOC_GETSTATE:
1560	{
1561		__u32	state = 0;
1562
1563		if (ext4_test_inode_state(inode, EXT4_STATE_EXT_PRECACHED))
1564			state |= EXT4_STATE_FLAG_EXT_PRECACHED;
1565		if (ext4_test_inode_state(inode, EXT4_STATE_NEW))
1566			state |= EXT4_STATE_FLAG_NEW;
1567		if (ext4_test_inode_state(inode, EXT4_STATE_NEWENTRY))
1568			state |= EXT4_STATE_FLAG_NEWENTRY;
1569		if (ext4_test_inode_state(inode, EXT4_STATE_DA_ALLOC_CLOSE))
1570			state |= EXT4_STATE_FLAG_DA_ALLOC_CLOSE;
1571
1572		return put_user(state, (__u32 __user *) arg);
1573	}
1574
1575	case EXT4_IOC_GET_ES_CACHE:
1576		return ext4_ioctl_get_es_cache(filp, arg);
1577
1578	case EXT4_IOC_SHUTDOWN:
1579		return ext4_ioctl_shutdown(sb, arg);
1580
1581	case FS_IOC_ENABLE_VERITY:
1582		if (!ext4_has_feature_verity(sb))
1583			return -EOPNOTSUPP;
1584		return fsverity_ioctl_enable(filp, (const void __user *)arg);
1585
1586	case FS_IOC_MEASURE_VERITY:
1587		if (!ext4_has_feature_verity(sb))
1588			return -EOPNOTSUPP;
1589		return fsverity_ioctl_measure(filp, (void __user *)arg);
1590
1591	case FS_IOC_READ_VERITY_METADATA:
1592		if (!ext4_has_feature_verity(sb))
1593			return -EOPNOTSUPP;
1594		return fsverity_ioctl_read_metadata(filp,
1595						    (const void __user *)arg);
1596
1597	case EXT4_IOC_CHECKPOINT:
1598		return ext4_ioctl_checkpoint(filp, arg);
1599
1600	case FS_IOC_GETFSLABEL:
1601		return ext4_ioctl_getlabel(EXT4_SB(sb), (void __user *)arg);
1602
1603	case FS_IOC_SETFSLABEL:
1604		return ext4_ioctl_setlabel(filp,
1605					   (const void __user *)arg);
1606
1607	case EXT4_IOC_GETFSUUID:
1608		return ext4_ioctl_getuuid(EXT4_SB(sb), (void __user *)arg);
1609	case EXT4_IOC_SETFSUUID:
1610		return ext4_ioctl_setuuid(filp, (const void __user *)arg);
1611	default:
1612		return -ENOTTY;
1613	}
1614}
1615
1616long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
1617{
1618	return __ext4_ioctl(filp, cmd, arg);
1619}
1620
1621#ifdef CONFIG_COMPAT
1622long ext4_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
1623{
1624	/* These are just misnamed, they actually get/put from/to user an int */
1625	switch (cmd) {
1626	case EXT4_IOC32_GETVERSION:
1627		cmd = EXT4_IOC_GETVERSION;
1628		break;
1629	case EXT4_IOC32_SETVERSION:
1630		cmd = EXT4_IOC_SETVERSION;
1631		break;
1632	case EXT4_IOC32_GROUP_EXTEND:
1633		cmd = EXT4_IOC_GROUP_EXTEND;
1634		break;
1635	case EXT4_IOC32_GETVERSION_OLD:
1636		cmd = EXT4_IOC_GETVERSION_OLD;
1637		break;
1638	case EXT4_IOC32_SETVERSION_OLD:
1639		cmd = EXT4_IOC_SETVERSION_OLD;
1640		break;
1641	case EXT4_IOC32_GETRSVSZ:
1642		cmd = EXT4_IOC_GETRSVSZ;
1643		break;
1644	case EXT4_IOC32_SETRSVSZ:
1645		cmd = EXT4_IOC_SETRSVSZ;
1646		break;
1647	case EXT4_IOC32_GROUP_ADD: {
1648		struct compat_ext4_new_group_input __user *uinput;
1649		struct ext4_new_group_data input;
1650		int err;
1651
1652		uinput = compat_ptr(arg);
1653		err = get_user(input.group, &uinput->group);
1654		err |= get_user(input.block_bitmap, &uinput->block_bitmap);
1655		err |= get_user(input.inode_bitmap, &uinput->inode_bitmap);
1656		err |= get_user(input.inode_table, &uinput->inode_table);
1657		err |= get_user(input.blocks_count, &uinput->blocks_count);
1658		err |= get_user(input.reserved_blocks,
1659				&uinput->reserved_blocks);
1660		if (err)
1661			return -EFAULT;
1662		return ext4_ioctl_group_add(file, &input);
1663	}
1664	case EXT4_IOC_MOVE_EXT:
1665	case EXT4_IOC_RESIZE_FS:
1666	case FITRIM:
1667	case EXT4_IOC_PRECACHE_EXTENTS:
1668	case FS_IOC_SET_ENCRYPTION_POLICY:
1669	case FS_IOC_GET_ENCRYPTION_PWSALT:
1670	case FS_IOC_GET_ENCRYPTION_POLICY:
1671	case FS_IOC_GET_ENCRYPTION_POLICY_EX:
1672	case FS_IOC_ADD_ENCRYPTION_KEY:
1673	case FS_IOC_REMOVE_ENCRYPTION_KEY:
1674	case FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS:
1675	case FS_IOC_GET_ENCRYPTION_KEY_STATUS:
1676	case FS_IOC_GET_ENCRYPTION_NONCE:
1677	case EXT4_IOC_SHUTDOWN:
1678	case FS_IOC_GETFSMAP:
1679	case FS_IOC_ENABLE_VERITY:
1680	case FS_IOC_MEASURE_VERITY:
1681	case FS_IOC_READ_VERITY_METADATA:
1682	case EXT4_IOC_CLEAR_ES_CACHE:
1683	case EXT4_IOC_GETSTATE:
1684	case EXT4_IOC_GET_ES_CACHE:
1685	case EXT4_IOC_CHECKPOINT:
1686	case FS_IOC_GETFSLABEL:
1687	case FS_IOC_SETFSLABEL:
1688	case EXT4_IOC_GETFSUUID:
1689	case EXT4_IOC_SETFSUUID:
1690		break;
1691	default:
1692		return -ENOIOCTLCMD;
1693	}
1694	return ext4_ioctl(file, cmd, (unsigned long) compat_ptr(arg));
1695}
1696#endif
1697
1698static void set_overhead(struct ext4_super_block *es, const void *arg)
1699{
1700	es->s_overhead_clusters = cpu_to_le32(*((unsigned long *) arg));
1701}
1702
1703int ext4_update_overhead(struct super_block *sb, bool force)
1704{
1705	struct ext4_sb_info *sbi = EXT4_SB(sb);
1706
1707	if (sb_rdonly(sb))
1708		return 0;
1709	if (!force &&
1710	    (sbi->s_overhead == 0 ||
1711	     sbi->s_overhead == le32_to_cpu(sbi->s_es->s_overhead_clusters)))
1712		return 0;
1713	return ext4_update_superblocks_fn(sb, set_overhead, &sbi->s_overhead);
1714}
1715