1 // SPDX-License-Identifier: GPL-2.0
3 * linux/fs/ext4/ioctl.c
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)
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/uuid.h>
20 #include <linux/uaccess.h>
21 #include <linux/delay.h>
22 #include <linux/iversion.h>
23 #include "ext4_jbd2.h"
25 #include <linux/fsmap.h>
27 #include <trace/events/ext4.h>
30 * Swap memory between @a and @b for @len bytes.
32 * @a: pointer to first memory area
33 * @b: pointer to second memory area
34 * @len: number of bytes to swap
37 static void memswap(void *a
, void *b
, size_t len
)
39 unsigned char *ap
, *bp
;
41 ap
= (unsigned char *)a
;
42 bp
= (unsigned char *)b
;
51 * Swap i_data and associated attributes between @inode1 and @inode2.
52 * This function is used for the primary swap between inode1 and inode2
53 * and also to revert this primary swap in case of errors.
55 * Therefore you have to make sure, that calling this method twice
56 * will revert all changes.
58 * @inode1: pointer to first inode
59 * @inode2: pointer to second inode
61 static void swap_inode_data(struct inode
*inode1
, struct inode
*inode2
)
64 struct ext4_inode_info
*ei1
;
65 struct ext4_inode_info
*ei2
;
71 swap(inode1
->i_version
, inode2
->i_version
);
72 swap(inode1
->i_atime
, inode2
->i_atime
);
73 swap(inode1
->i_mtime
, inode2
->i_mtime
);
75 memswap(ei1
->i_data
, ei2
->i_data
, sizeof(ei1
->i_data
));
76 tmp
= ei1
->i_flags
& EXT4_FL_SHOULD_SWAP
;
77 ei1
->i_flags
= (ei2
->i_flags
& EXT4_FL_SHOULD_SWAP
) |
78 (ei1
->i_flags
& ~EXT4_FL_SHOULD_SWAP
);
79 ei2
->i_flags
= tmp
| (ei2
->i_flags
& ~EXT4_FL_SHOULD_SWAP
);
80 swap(ei1
->i_disksize
, ei2
->i_disksize
);
81 ext4_es_remove_extent(inode1
, 0, EXT_MAX_BLOCKS
);
82 ext4_es_remove_extent(inode2
, 0, EXT_MAX_BLOCKS
);
84 isize
= i_size_read(inode1
);
85 i_size_write(inode1
, i_size_read(inode2
));
86 i_size_write(inode2
, isize
);
89 void ext4_reset_inode_seed(struct inode
*inode
)
91 struct ext4_inode_info
*ei
= EXT4_I(inode
);
92 struct ext4_sb_info
*sbi
= EXT4_SB(inode
->i_sb
);
93 __le32 inum
= cpu_to_le32(inode
->i_ino
);
94 __le32 gen
= cpu_to_le32(inode
->i_generation
);
97 if (!ext4_has_metadata_csum(inode
->i_sb
))
100 csum
= ext4_chksum(sbi
, sbi
->s_csum_seed
, (__u8
*)&inum
, sizeof(inum
));
101 ei
->i_csum_seed
= ext4_chksum(sbi
, csum
, (__u8
*)&gen
, sizeof(gen
));
105 * Swap the information from the given @inode and the inode
106 * EXT4_BOOT_LOADER_INO. It will basically swap i_data and all other
107 * important fields of the inodes.
109 * @sb: the super block of the filesystem
110 * @inode: the inode to swap with EXT4_BOOT_LOADER_INO
113 static long swap_inode_boot_loader(struct super_block
*sb
,
118 struct inode
*inode_bl
;
119 struct ext4_inode_info
*ei_bl
;
120 qsize_t size
, size_bl
, diff
;
122 unsigned short bytes
;
124 inode_bl
= ext4_iget(sb
, EXT4_BOOT_LOADER_INO
, EXT4_IGET_SPECIAL
);
125 if (IS_ERR(inode_bl
))
126 return PTR_ERR(inode_bl
);
127 ei_bl
= EXT4_I(inode_bl
);
129 /* Protect orig inodes against a truncate and make sure,
130 * that only 1 swap_inode_boot_loader is running. */
131 lock_two_nondirectories(inode
, inode_bl
);
133 if (inode
->i_nlink
!= 1 || !S_ISREG(inode
->i_mode
) ||
134 IS_SWAPFILE(inode
) || IS_ENCRYPTED(inode
) ||
135 (EXT4_I(inode
)->i_flags
& EXT4_JOURNAL_DATA_FL
) ||
136 ext4_has_inline_data(inode
)) {
138 goto journal_err_out
;
141 if (IS_RDONLY(inode
) || IS_APPEND(inode
) || IS_IMMUTABLE(inode
) ||
142 !inode_owner_or_capable(inode
) || !capable(CAP_SYS_ADMIN
)) {
144 goto journal_err_out
;
147 down_write(&EXT4_I(inode
)->i_mmap_sem
);
148 err
= filemap_write_and_wait(inode
->i_mapping
);
152 err
= filemap_write_and_wait(inode_bl
->i_mapping
);
156 /* Wait for all existing dio workers */
157 inode_dio_wait(inode
);
158 inode_dio_wait(inode_bl
);
160 truncate_inode_pages(&inode
->i_data
, 0);
161 truncate_inode_pages(&inode_bl
->i_data
, 0);
163 handle
= ext4_journal_start(inode_bl
, EXT4_HT_MOVE_EXTENTS
, 2);
164 if (IS_ERR(handle
)) {
168 ext4_fc_start_ineligible(sb
, EXT4_FC_REASON_SWAP_BOOT
);
170 /* Protect extent tree against block allocations via delalloc */
171 ext4_double_down_write_data_sem(inode
, inode_bl
);
173 if (inode_bl
->i_nlink
== 0) {
174 /* this inode has never been used as a BOOT_LOADER */
175 set_nlink(inode_bl
, 1);
176 i_uid_write(inode_bl
, 0);
177 i_gid_write(inode_bl
, 0);
178 inode_bl
->i_flags
= 0;
180 inode_set_iversion(inode_bl
, 1);
181 i_size_write(inode_bl
, 0);
182 inode_bl
->i_mode
= S_IFREG
;
183 if (ext4_has_feature_extents(sb
)) {
184 ext4_set_inode_flag(inode_bl
, EXT4_INODE_EXTENTS
);
185 ext4_ext_tree_init(handle
, inode_bl
);
187 memset(ei_bl
->i_data
, 0, sizeof(ei_bl
->i_data
));
190 err
= dquot_initialize(inode
);
194 size
= (qsize_t
)(inode
->i_blocks
) * (1 << 9) + inode
->i_bytes
;
195 size_bl
= (qsize_t
)(inode_bl
->i_blocks
) * (1 << 9) + inode_bl
->i_bytes
;
196 diff
= size
- size_bl
;
197 swap_inode_data(inode
, inode_bl
);
199 inode
->i_ctime
= inode_bl
->i_ctime
= current_time(inode
);
201 inode
->i_generation
= prandom_u32();
202 inode_bl
->i_generation
= prandom_u32();
203 ext4_reset_inode_seed(inode
);
204 ext4_reset_inode_seed(inode_bl
);
206 ext4_discard_preallocations(inode
, 0);
208 err
= ext4_mark_inode_dirty(handle
, inode
);
210 /* No need to update quota information. */
211 ext4_warning(inode
->i_sb
,
212 "couldn't mark inode #%lu dirty (err %d)",
214 /* Revert all changes: */
215 swap_inode_data(inode
, inode_bl
);
216 ext4_mark_inode_dirty(handle
, inode
);
220 blocks
= inode_bl
->i_blocks
;
221 bytes
= inode_bl
->i_bytes
;
222 inode_bl
->i_blocks
= inode
->i_blocks
;
223 inode_bl
->i_bytes
= inode
->i_bytes
;
224 err
= ext4_mark_inode_dirty(handle
, inode_bl
);
226 /* No need to update quota information. */
227 ext4_warning(inode_bl
->i_sb
,
228 "couldn't mark inode #%lu dirty (err %d)",
229 inode_bl
->i_ino
, err
);
233 /* Bootloader inode should not be counted into quota information. */
235 dquot_free_space(inode
, diff
);
237 err
= dquot_alloc_space(inode
, -1 * diff
);
241 /* Revert all changes: */
242 inode_bl
->i_blocks
= blocks
;
243 inode_bl
->i_bytes
= bytes
;
244 swap_inode_data(inode
, inode_bl
);
245 ext4_mark_inode_dirty(handle
, inode
);
246 ext4_mark_inode_dirty(handle
, inode_bl
);
250 ext4_journal_stop(handle
);
251 ext4_fc_stop_ineligible(sb
);
252 ext4_double_up_write_data_sem(inode
, inode_bl
);
255 up_write(&EXT4_I(inode
)->i_mmap_sem
);
257 unlock_two_nondirectories(inode
, inode_bl
);
262 #ifdef CONFIG_FS_ENCRYPTION
263 static int uuid_is_zero(__u8 u
[16])
267 for (i
= 0; i
< 16; i
++)
275 * If immutable is set and we are not clearing it, we're not allowed to change
276 * anything else in the inode. Don't error out if we're only trying to set
277 * immutable on an immutable file.
279 static int ext4_ioctl_check_immutable(struct inode
*inode
, __u32 new_projid
,
282 struct ext4_inode_info
*ei
= EXT4_I(inode
);
283 unsigned int oldflags
= ei
->i_flags
;
285 if (!(oldflags
& EXT4_IMMUTABLE_FL
) || !(flags
& EXT4_IMMUTABLE_FL
))
288 if ((oldflags
& ~EXT4_IMMUTABLE_FL
) != (flags
& ~EXT4_IMMUTABLE_FL
))
290 if (ext4_has_feature_project(inode
->i_sb
) &&
291 __kprojid_val(ei
->i_projid
) != new_projid
)
297 static void ext4_dax_dontcache(struct inode
*inode
, unsigned int flags
)
299 struct ext4_inode_info
*ei
= EXT4_I(inode
);
301 if (S_ISDIR(inode
->i_mode
))
304 if (test_opt2(inode
->i_sb
, DAX_NEVER
) ||
305 test_opt(inode
->i_sb
, DAX_ALWAYS
))
308 if ((ei
->i_flags
^ flags
) & EXT4_DAX_FL
)
309 d_mark_dontcache(inode
);
312 static bool dax_compatible(struct inode
*inode
, unsigned int oldflags
,
315 if (flags
& EXT4_DAX_FL
) {
316 if ((oldflags
& EXT4_DAX_MUT_EXCL
) ||
317 ext4_test_inode_state(inode
,
318 EXT4_STATE_VERITY_IN_PROGRESS
)) {
323 if ((flags
& EXT4_DAX_MUT_EXCL
) && (oldflags
& EXT4_DAX_FL
))
329 static int ext4_ioctl_setflags(struct inode
*inode
,
332 struct ext4_inode_info
*ei
= EXT4_I(inode
);
333 handle_t
*handle
= NULL
;
334 int err
= -EPERM
, migrate
= 0;
335 struct ext4_iloc iloc
;
336 unsigned int oldflags
, mask
, i
;
337 struct super_block
*sb
= inode
->i_sb
;
339 /* Is it quota file? Do not allow user to mess with it */
340 if (ext4_is_quota_file(inode
))
343 oldflags
= ei
->i_flags
;
345 err
= vfs_ioc_setflags_prepare(inode
, oldflags
, flags
);
350 * The JOURNAL_DATA flag can only be changed by
351 * the relevant capability.
353 if ((flags
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
)) {
354 if (!capable(CAP_SYS_RESOURCE
))
358 if (!dax_compatible(inode
, oldflags
, flags
)) {
363 if ((flags
^ oldflags
) & EXT4_EXTENTS_FL
)
366 if ((flags
^ oldflags
) & EXT4_CASEFOLD_FL
) {
367 if (!ext4_has_feature_casefold(sb
)) {
372 if (!S_ISDIR(inode
->i_mode
)) {
377 if (!ext4_empty_dir(inode
)) {
384 * Wait for all pending directio and then flush all the dirty pages
385 * for this file. The flush marks all the pages readonly, so any
386 * subsequent attempt to write to the file (particularly mmap pages)
387 * will come through the filesystem and fail.
389 if (S_ISREG(inode
->i_mode
) && !IS_IMMUTABLE(inode
) &&
390 (flags
& EXT4_IMMUTABLE_FL
)) {
391 inode_dio_wait(inode
);
392 err
= filemap_write_and_wait(inode
->i_mapping
);
397 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
398 if (IS_ERR(handle
)) {
399 err
= PTR_ERR(handle
);
403 ext4_handle_sync(handle
);
404 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
408 ext4_dax_dontcache(inode
, flags
);
410 for (i
= 0, mask
= 1; i
< 32; i
++, mask
<<= 1) {
411 if (!(mask
& EXT4_FL_USER_MODIFIABLE
))
413 /* These flags get special treatment later */
414 if (mask
== EXT4_JOURNAL_DATA_FL
|| mask
== EXT4_EXTENTS_FL
)
417 ext4_set_inode_flag(inode
, i
);
419 ext4_clear_inode_flag(inode
, i
);
422 ext4_set_inode_flags(inode
, false);
424 inode
->i_ctime
= current_time(inode
);
426 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
428 ext4_journal_stop(handle
);
432 if ((flags
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
)) {
434 * Changes to the journaling mode can cause unsafe changes to
435 * S_DAX if the inode is DAX
442 err
= ext4_change_inode_journal_flag(inode
,
443 flags
& EXT4_JOURNAL_DATA_FL
);
448 if (flags
& EXT4_EXTENTS_FL
)
449 err
= ext4_ext_migrate(inode
);
451 err
= ext4_ind_migrate(inode
);
459 static int ext4_ioctl_setproject(struct file
*filp
, __u32 projid
)
461 struct inode
*inode
= file_inode(filp
);
462 struct super_block
*sb
= inode
->i_sb
;
463 struct ext4_inode_info
*ei
= EXT4_I(inode
);
467 struct ext4_iloc iloc
;
468 struct ext4_inode
*raw_inode
;
469 struct dquot
*transfer_to
[MAXQUOTAS
] = { };
471 if (!ext4_has_feature_project(sb
)) {
472 if (projid
!= EXT4_DEF_PROJID
)
478 if (EXT4_INODE_SIZE(sb
) <= EXT4_GOOD_OLD_INODE_SIZE
)
481 kprojid
= make_kprojid(&init_user_ns
, (projid_t
)projid
);
483 if (projid_eq(kprojid
, EXT4_I(inode
)->i_projid
))
487 /* Is it quota file? Do not allow user to mess with it */
488 if (ext4_is_quota_file(inode
))
491 err
= ext4_get_inode_loc(inode
, &iloc
);
495 raw_inode
= ext4_raw_inode(&iloc
);
496 if (!EXT4_FITS_IN_INODE(raw_inode
, ei
, i_projid
)) {
497 err
= ext4_expand_extra_isize(inode
,
498 EXT4_SB(sb
)->s_want_extra_isize
,
506 err
= dquot_initialize(inode
);
510 handle
= ext4_journal_start(inode
, EXT4_HT_QUOTA
,
511 EXT4_QUOTA_INIT_BLOCKS(sb
) +
512 EXT4_QUOTA_DEL_BLOCKS(sb
) + 3);
514 return PTR_ERR(handle
);
516 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
520 transfer_to
[PRJQUOTA
] = dqget(sb
, make_kqid_projid(kprojid
));
521 if (!IS_ERR(transfer_to
[PRJQUOTA
])) {
523 /* __dquot_transfer() calls back ext4_get_inode_usage() which
524 * counts xattr inode references.
526 down_read(&EXT4_I(inode
)->xattr_sem
);
527 err
= __dquot_transfer(inode
, transfer_to
);
528 up_read(&EXT4_I(inode
)->xattr_sem
);
529 dqput(transfer_to
[PRJQUOTA
]);
534 EXT4_I(inode
)->i_projid
= kprojid
;
535 inode
->i_ctime
= current_time(inode
);
537 rc
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
541 ext4_journal_stop(handle
);
545 static int ext4_ioctl_setproject(struct file
*filp
, __u32 projid
)
547 if (projid
!= EXT4_DEF_PROJID
)
553 /* Transfer internal flags to xflags */
554 static inline __u32
ext4_iflags_to_xflags(unsigned long iflags
)
558 if (iflags
& EXT4_SYNC_FL
)
559 xflags
|= FS_XFLAG_SYNC
;
560 if (iflags
& EXT4_IMMUTABLE_FL
)
561 xflags
|= FS_XFLAG_IMMUTABLE
;
562 if (iflags
& EXT4_APPEND_FL
)
563 xflags
|= FS_XFLAG_APPEND
;
564 if (iflags
& EXT4_NODUMP_FL
)
565 xflags
|= FS_XFLAG_NODUMP
;
566 if (iflags
& EXT4_NOATIME_FL
)
567 xflags
|= FS_XFLAG_NOATIME
;
568 if (iflags
& EXT4_PROJINHERIT_FL
)
569 xflags
|= FS_XFLAG_PROJINHERIT
;
570 if (iflags
& EXT4_DAX_FL
)
571 xflags
|= FS_XFLAG_DAX
;
575 #define EXT4_SUPPORTED_FS_XFLAGS (FS_XFLAG_SYNC | FS_XFLAG_IMMUTABLE | \
576 FS_XFLAG_APPEND | FS_XFLAG_NODUMP | \
577 FS_XFLAG_NOATIME | FS_XFLAG_PROJINHERIT | \
580 /* Transfer xflags flags to internal */
581 static inline unsigned long ext4_xflags_to_iflags(__u32 xflags
)
583 unsigned long iflags
= 0;
585 if (xflags
& FS_XFLAG_SYNC
)
586 iflags
|= EXT4_SYNC_FL
;
587 if (xflags
& FS_XFLAG_IMMUTABLE
)
588 iflags
|= EXT4_IMMUTABLE_FL
;
589 if (xflags
& FS_XFLAG_APPEND
)
590 iflags
|= EXT4_APPEND_FL
;
591 if (xflags
& FS_XFLAG_NODUMP
)
592 iflags
|= EXT4_NODUMP_FL
;
593 if (xflags
& FS_XFLAG_NOATIME
)
594 iflags
|= EXT4_NOATIME_FL
;
595 if (xflags
& FS_XFLAG_PROJINHERIT
)
596 iflags
|= EXT4_PROJINHERIT_FL
;
597 if (xflags
& FS_XFLAG_DAX
)
598 iflags
|= EXT4_DAX_FL
;
603 static int ext4_shutdown(struct super_block
*sb
, unsigned long arg
)
605 struct ext4_sb_info
*sbi
= EXT4_SB(sb
);
608 if (!capable(CAP_SYS_ADMIN
))
611 if (get_user(flags
, (__u32 __user
*)arg
))
614 if (flags
> EXT4_GOING_FLAGS_NOLOGFLUSH
)
617 if (ext4_forced_shutdown(sbi
))
620 ext4_msg(sb
, KERN_ALERT
, "shut down requested (%d)", flags
);
621 trace_ext4_shutdown(sb
, flags
);
624 case EXT4_GOING_FLAGS_DEFAULT
:
625 freeze_bdev(sb
->s_bdev
);
626 set_bit(EXT4_FLAGS_SHUTDOWN
, &sbi
->s_ext4_flags
);
627 thaw_bdev(sb
->s_bdev
, sb
);
629 case EXT4_GOING_FLAGS_LOGFLUSH
:
630 set_bit(EXT4_FLAGS_SHUTDOWN
, &sbi
->s_ext4_flags
);
631 if (sbi
->s_journal
&& !is_journal_aborted(sbi
->s_journal
)) {
632 (void) ext4_force_commit(sb
);
633 jbd2_journal_abort(sbi
->s_journal
, -ESHUTDOWN
);
636 case EXT4_GOING_FLAGS_NOLOGFLUSH
:
637 set_bit(EXT4_FLAGS_SHUTDOWN
, &sbi
->s_ext4_flags
);
638 if (sbi
->s_journal
&& !is_journal_aborted(sbi
->s_journal
))
639 jbd2_journal_abort(sbi
->s_journal
, -ESHUTDOWN
);
644 clear_opt(sb
, DISCARD
);
648 struct getfsmap_info
{
649 struct super_block
*gi_sb
;
650 struct fsmap_head __user
*gi_data
;
655 static int ext4_getfsmap_format(struct ext4_fsmap
*xfm
, void *priv
)
657 struct getfsmap_info
*info
= priv
;
660 trace_ext4_getfsmap_mapping(info
->gi_sb
, xfm
);
662 info
->gi_last_flags
= xfm
->fmr_flags
;
663 ext4_fsmap_from_internal(info
->gi_sb
, &fm
, xfm
);
664 if (copy_to_user(&info
->gi_data
->fmh_recs
[info
->gi_idx
++], &fm
,
665 sizeof(struct fsmap
)))
671 static int ext4_ioc_getfsmap(struct super_block
*sb
,
672 struct fsmap_head __user
*arg
)
674 struct getfsmap_info info
= { NULL
};
675 struct ext4_fsmap_head xhead
= {0};
676 struct fsmap_head head
;
677 bool aborted
= false;
680 if (copy_from_user(&head
, arg
, sizeof(struct fsmap_head
)))
682 if (memchr_inv(head
.fmh_reserved
, 0, sizeof(head
.fmh_reserved
)) ||
683 memchr_inv(head
.fmh_keys
[0].fmr_reserved
, 0,
684 sizeof(head
.fmh_keys
[0].fmr_reserved
)) ||
685 memchr_inv(head
.fmh_keys
[1].fmr_reserved
, 0,
686 sizeof(head
.fmh_keys
[1].fmr_reserved
)))
689 * ext4 doesn't report file extents at all, so the only valid
690 * file offsets are the magic ones (all zeroes or all ones).
692 if (head
.fmh_keys
[0].fmr_offset
||
693 (head
.fmh_keys
[1].fmr_offset
!= 0 &&
694 head
.fmh_keys
[1].fmr_offset
!= -1ULL))
697 xhead
.fmh_iflags
= head
.fmh_iflags
;
698 xhead
.fmh_count
= head
.fmh_count
;
699 ext4_fsmap_to_internal(sb
, &xhead
.fmh_keys
[0], &head
.fmh_keys
[0]);
700 ext4_fsmap_to_internal(sb
, &xhead
.fmh_keys
[1], &head
.fmh_keys
[1]);
702 trace_ext4_getfsmap_low_key(sb
, &xhead
.fmh_keys
[0]);
703 trace_ext4_getfsmap_high_key(sb
, &xhead
.fmh_keys
[1]);
707 error
= ext4_getfsmap(sb
, &xhead
, ext4_getfsmap_format
, &info
);
708 if (error
== EXT4_QUERY_RANGE_ABORT
) {
714 /* If we didn't abort, set the "last" flag in the last fmx */
715 if (!aborted
&& info
.gi_idx
) {
716 info
.gi_last_flags
|= FMR_OF_LAST
;
717 if (copy_to_user(&info
.gi_data
->fmh_recs
[info
.gi_idx
- 1].fmr_flags
,
719 sizeof(info
.gi_last_flags
)))
723 /* copy back header */
724 head
.fmh_entries
= xhead
.fmh_entries
;
725 head
.fmh_oflags
= xhead
.fmh_oflags
;
726 if (copy_to_user(arg
, &head
, sizeof(struct fsmap_head
)))
732 static long ext4_ioctl_group_add(struct file
*file
,
733 struct ext4_new_group_data
*input
)
735 struct super_block
*sb
= file_inode(file
)->i_sb
;
738 err
= ext4_resize_begin(sb
);
742 if (ext4_has_feature_bigalloc(sb
)) {
743 ext4_msg(sb
, KERN_ERR
,
744 "Online resizing not supported with bigalloc");
749 err
= mnt_want_write_file(file
);
753 err
= ext4_group_add(sb
, input
);
754 if (EXT4_SB(sb
)->s_journal
) {
755 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
756 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
757 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
761 mnt_drop_write_file(file
);
762 if (!err
&& ext4_has_group_desc_csum(sb
) &&
763 test_opt(sb
, INIT_INODE_TABLE
))
764 err
= ext4_register_li_request(sb
, input
->group
);
770 static void ext4_fill_fsxattr(struct inode
*inode
, struct fsxattr
*fa
)
772 struct ext4_inode_info
*ei
= EXT4_I(inode
);
774 simple_fill_fsxattr(fa
, ext4_iflags_to_xflags(ei
->i_flags
&
775 EXT4_FL_USER_VISIBLE
));
777 if (ext4_has_feature_project(inode
->i_sb
))
778 fa
->fsx_projid
= from_kprojid(&init_user_ns
, ei
->i_projid
);
781 /* So that the fiemap access checks can't overflow on 32 bit machines. */
782 #define FIEMAP_MAX_EXTENTS (UINT_MAX / sizeof(struct fiemap_extent))
784 static int ext4_ioctl_get_es_cache(struct file
*filp
, unsigned long arg
)
786 struct fiemap fiemap
;
787 struct fiemap __user
*ufiemap
= (struct fiemap __user
*) arg
;
788 struct fiemap_extent_info fieinfo
= { 0, };
789 struct inode
*inode
= file_inode(filp
);
792 if (copy_from_user(&fiemap
, ufiemap
, sizeof(fiemap
)))
795 if (fiemap
.fm_extent_count
> FIEMAP_MAX_EXTENTS
)
798 fieinfo
.fi_flags
= fiemap
.fm_flags
;
799 fieinfo
.fi_extents_max
= fiemap
.fm_extent_count
;
800 fieinfo
.fi_extents_start
= ufiemap
->fm_extents
;
802 error
= ext4_get_es_cache(inode
, &fieinfo
, fiemap
.fm_start
,
804 fiemap
.fm_flags
= fieinfo
.fi_flags
;
805 fiemap
.fm_mapped_extents
= fieinfo
.fi_extents_mapped
;
806 if (copy_to_user(ufiemap
, &fiemap
, sizeof(fiemap
)))
812 static long __ext4_ioctl(struct file
*filp
, unsigned int cmd
, unsigned long arg
)
814 struct inode
*inode
= file_inode(filp
);
815 struct super_block
*sb
= inode
->i_sb
;
816 struct ext4_inode_info
*ei
= EXT4_I(inode
);
819 ext4_debug("cmd = %u, arg = %lu\n", cmd
, arg
);
822 case FS_IOC_GETFSMAP
:
823 return ext4_ioc_getfsmap(sb
, (void __user
*)arg
);
824 case FS_IOC_GETFLAGS
:
825 flags
= ei
->i_flags
& EXT4_FL_USER_VISIBLE
;
826 if (S_ISREG(inode
->i_mode
))
827 flags
&= ~EXT4_PROJINHERIT_FL
;
828 return put_user(flags
, (int __user
*) arg
);
829 case FS_IOC_SETFLAGS
: {
832 if (!inode_owner_or_capable(inode
))
835 if (get_user(flags
, (int __user
*) arg
))
838 if (flags
& ~EXT4_FL_USER_VISIBLE
)
841 * chattr(1) grabs flags via GETFLAGS, modifies the result and
842 * passes that to SETFLAGS. So we cannot easily make SETFLAGS
843 * more restrictive than just silently masking off visible but
844 * not settable flags as we always did.
846 flags
&= EXT4_FL_USER_MODIFIABLE
;
847 if (ext4_mask_flags(inode
->i_mode
, flags
) != flags
)
850 err
= mnt_want_write_file(filp
);
855 err
= ext4_ioctl_check_immutable(inode
,
856 from_kprojid(&init_user_ns
, ei
->i_projid
),
859 err
= ext4_ioctl_setflags(inode
, flags
);
861 mnt_drop_write_file(filp
);
864 case EXT4_IOC_GETVERSION
:
865 case EXT4_IOC_GETVERSION_OLD
:
866 return put_user(inode
->i_generation
, (int __user
*) arg
);
867 case EXT4_IOC_SETVERSION
:
868 case EXT4_IOC_SETVERSION_OLD
: {
870 struct ext4_iloc iloc
;
874 if (!inode_owner_or_capable(inode
))
877 if (ext4_has_metadata_csum(inode
->i_sb
)) {
878 ext4_warning(sb
, "Setting inode version is not "
879 "supported with metadata_csum enabled.");
883 err
= mnt_want_write_file(filp
);
886 if (get_user(generation
, (int __user
*) arg
)) {
892 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
893 if (IS_ERR(handle
)) {
894 err
= PTR_ERR(handle
);
897 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
899 inode
->i_ctime
= current_time(inode
);
900 inode
->i_generation
= generation
;
901 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
903 ext4_journal_stop(handle
);
908 mnt_drop_write_file(filp
);
911 case EXT4_IOC_GROUP_EXTEND
: {
912 ext4_fsblk_t n_blocks_count
;
915 err
= ext4_resize_begin(sb
);
919 if (get_user(n_blocks_count
, (__u32 __user
*)arg
)) {
921 goto group_extend_out
;
924 if (ext4_has_feature_bigalloc(sb
)) {
925 ext4_msg(sb
, KERN_ERR
,
926 "Online resizing not supported with bigalloc");
928 goto group_extend_out
;
931 err
= mnt_want_write_file(filp
);
933 goto group_extend_out
;
935 err
= ext4_group_extend(sb
, EXT4_SB(sb
)->s_es
, n_blocks_count
);
936 if (EXT4_SB(sb
)->s_journal
) {
937 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
938 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
939 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
943 mnt_drop_write_file(filp
);
949 case EXT4_IOC_MOVE_EXT
: {
950 struct move_extent me
;
954 if (!(filp
->f_mode
& FMODE_READ
) ||
955 !(filp
->f_mode
& FMODE_WRITE
))
958 if (copy_from_user(&me
,
959 (struct move_extent __user
*)arg
, sizeof(me
)))
963 donor
= fdget(me
.donor_fd
);
967 if (!(donor
.file
->f_mode
& FMODE_WRITE
)) {
972 if (ext4_has_feature_bigalloc(sb
)) {
973 ext4_msg(sb
, KERN_ERR
,
974 "Online defrag not supported with bigalloc");
977 } else if (IS_DAX(inode
)) {
978 ext4_msg(sb
, KERN_ERR
,
979 "Online defrag not supported with DAX");
984 err
= mnt_want_write_file(filp
);
988 err
= ext4_move_extents(filp
, donor
.file
, me
.orig_start
,
989 me
.donor_start
, me
.len
, &me
.moved_len
);
990 mnt_drop_write_file(filp
);
992 if (copy_to_user((struct move_extent __user
*)arg
,
1000 case EXT4_IOC_GROUP_ADD
: {
1001 struct ext4_new_group_data input
;
1003 if (copy_from_user(&input
, (struct ext4_new_group_input __user
*)arg
,
1007 return ext4_ioctl_group_add(filp
, &input
);
1010 case EXT4_IOC_MIGRATE
:
1013 if (!inode_owner_or_capable(inode
))
1016 err
= mnt_want_write_file(filp
);
1020 * inode_mutex prevent write and truncate on the file.
1021 * Read still goes through. We take i_data_sem in
1022 * ext4_ext_swap_inode_data before we switch the
1023 * inode format to prevent read.
1025 inode_lock((inode
));
1026 err
= ext4_ext_migrate(inode
);
1027 inode_unlock((inode
));
1028 mnt_drop_write_file(filp
);
1032 case EXT4_IOC_ALLOC_DA_BLKS
:
1035 if (!inode_owner_or_capable(inode
))
1038 err
= mnt_want_write_file(filp
);
1041 err
= ext4_alloc_da_blocks(inode
);
1042 mnt_drop_write_file(filp
);
1046 case EXT4_IOC_SWAP_BOOT
:
1049 if (!(filp
->f_mode
& FMODE_WRITE
))
1051 err
= mnt_want_write_file(filp
);
1054 err
= swap_inode_boot_loader(sb
, inode
);
1055 mnt_drop_write_file(filp
);
1059 case EXT4_IOC_RESIZE_FS
: {
1060 ext4_fsblk_t n_blocks_count
;
1061 int err
= 0, err2
= 0;
1062 ext4_group_t o_group
= EXT4_SB(sb
)->s_groups_count
;
1064 if (copy_from_user(&n_blocks_count
, (__u64 __user
*)arg
,
1069 err
= ext4_resize_begin(sb
);
1073 err
= mnt_want_write_file(filp
);
1077 err
= ext4_resize_fs(sb
, n_blocks_count
);
1078 if (EXT4_SB(sb
)->s_journal
) {
1079 ext4_fc_mark_ineligible(sb
, EXT4_FC_REASON_RESIZE
);
1080 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
1081 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
1082 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
1086 mnt_drop_write_file(filp
);
1087 if (!err
&& (o_group
< EXT4_SB(sb
)->s_groups_count
) &&
1088 ext4_has_group_desc_csum(sb
) &&
1089 test_opt(sb
, INIT_INODE_TABLE
))
1090 err
= ext4_register_li_request(sb
, o_group
);
1093 ext4_resize_end(sb
);
1099 struct request_queue
*q
= bdev_get_queue(sb
->s_bdev
);
1100 struct fstrim_range range
;
1103 if (!capable(CAP_SYS_ADMIN
))
1106 if (!blk_queue_discard(q
))
1110 * We haven't replayed the journal, so we cannot use our
1111 * block-bitmap-guided storage zapping commands.
1113 if (test_opt(sb
, NOLOAD
) && ext4_has_feature_journal(sb
))
1116 if (copy_from_user(&range
, (struct fstrim_range __user
*)arg
,
1120 range
.minlen
= max((unsigned int)range
.minlen
,
1121 q
->limits
.discard_granularity
);
1122 ret
= ext4_trim_fs(sb
, &range
);
1126 if (copy_to_user((struct fstrim_range __user
*)arg
, &range
,
1132 case EXT4_IOC_PRECACHE_EXTENTS
:
1133 return ext4_ext_precache(inode
);
1135 case FS_IOC_SET_ENCRYPTION_POLICY
:
1136 if (!ext4_has_feature_encrypt(sb
))
1138 return fscrypt_ioctl_set_policy(filp
, (const void __user
*)arg
);
1140 case FS_IOC_GET_ENCRYPTION_PWSALT
: {
1141 #ifdef CONFIG_FS_ENCRYPTION
1143 struct ext4_sb_info
*sbi
= EXT4_SB(sb
);
1146 if (!ext4_has_feature_encrypt(sb
))
1148 if (uuid_is_zero(sbi
->s_es
->s_encrypt_pw_salt
)) {
1149 err
= mnt_want_write_file(filp
);
1152 handle
= ext4_journal_start_sb(sb
, EXT4_HT_MISC
, 1);
1153 if (IS_ERR(handle
)) {
1154 err
= PTR_ERR(handle
);
1155 goto pwsalt_err_exit
;
1157 err
= ext4_journal_get_write_access(handle
, sbi
->s_sbh
);
1159 goto pwsalt_err_journal
;
1160 lock_buffer(sbi
->s_sbh
);
1161 generate_random_uuid(sbi
->s_es
->s_encrypt_pw_salt
);
1162 ext4_superblock_csum_set(sb
);
1163 unlock_buffer(sbi
->s_sbh
);
1164 err
= ext4_handle_dirty_metadata(handle
, NULL
,
1167 err2
= ext4_journal_stop(handle
);
1171 mnt_drop_write_file(filp
);
1175 if (copy_to_user((void __user
*) arg
,
1176 sbi
->s_es
->s_encrypt_pw_salt
, 16))
1183 case FS_IOC_GET_ENCRYPTION_POLICY
:
1184 if (!ext4_has_feature_encrypt(sb
))
1186 return fscrypt_ioctl_get_policy(filp
, (void __user
*)arg
);
1188 case FS_IOC_GET_ENCRYPTION_POLICY_EX
:
1189 if (!ext4_has_feature_encrypt(sb
))
1191 return fscrypt_ioctl_get_policy_ex(filp
, (void __user
*)arg
);
1193 case FS_IOC_ADD_ENCRYPTION_KEY
:
1194 if (!ext4_has_feature_encrypt(sb
))
1196 return fscrypt_ioctl_add_key(filp
, (void __user
*)arg
);
1198 case FS_IOC_REMOVE_ENCRYPTION_KEY
:
1199 if (!ext4_has_feature_encrypt(sb
))
1201 return fscrypt_ioctl_remove_key(filp
, (void __user
*)arg
);
1203 case FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS
:
1204 if (!ext4_has_feature_encrypt(sb
))
1206 return fscrypt_ioctl_remove_key_all_users(filp
,
1207 (void __user
*)arg
);
1208 case FS_IOC_GET_ENCRYPTION_KEY_STATUS
:
1209 if (!ext4_has_feature_encrypt(sb
))
1211 return fscrypt_ioctl_get_key_status(filp
, (void __user
*)arg
);
1213 case FS_IOC_GET_ENCRYPTION_NONCE
:
1214 if (!ext4_has_feature_encrypt(sb
))
1216 return fscrypt_ioctl_get_nonce(filp
, (void __user
*)arg
);
1218 case EXT4_IOC_CLEAR_ES_CACHE
:
1220 if (!inode_owner_or_capable(inode
))
1222 ext4_clear_inode_es(inode
);
1226 case EXT4_IOC_GETSTATE
:
1230 if (ext4_test_inode_state(inode
, EXT4_STATE_EXT_PRECACHED
))
1231 state
|= EXT4_STATE_FLAG_EXT_PRECACHED
;
1232 if (ext4_test_inode_state(inode
, EXT4_STATE_NEW
))
1233 state
|= EXT4_STATE_FLAG_NEW
;
1234 if (ext4_test_inode_state(inode
, EXT4_STATE_NEWENTRY
))
1235 state
|= EXT4_STATE_FLAG_NEWENTRY
;
1236 if (ext4_test_inode_state(inode
, EXT4_STATE_DA_ALLOC_CLOSE
))
1237 state
|= EXT4_STATE_FLAG_DA_ALLOC_CLOSE
;
1239 return put_user(state
, (__u32 __user
*) arg
);
1242 case EXT4_IOC_GET_ES_CACHE
:
1243 return ext4_ioctl_get_es_cache(filp
, arg
);
1245 case FS_IOC_FSGETXATTR
:
1249 ext4_fill_fsxattr(inode
, &fa
);
1251 if (copy_to_user((struct fsxattr __user
*)arg
,
1256 case FS_IOC_FSSETXATTR
:
1258 struct fsxattr fa
, old_fa
;
1261 if (copy_from_user(&fa
, (struct fsxattr __user
*)arg
,
1265 /* Make sure caller has proper permission */
1266 if (!inode_owner_or_capable(inode
))
1269 if (fa
.fsx_xflags
& ~EXT4_SUPPORTED_FS_XFLAGS
)
1272 flags
= ext4_xflags_to_iflags(fa
.fsx_xflags
);
1273 if (ext4_mask_flags(inode
->i_mode
, flags
) != flags
)
1276 err
= mnt_want_write_file(filp
);
1281 ext4_fill_fsxattr(inode
, &old_fa
);
1282 err
= vfs_ioc_fssetxattr_check(inode
, &old_fa
, &fa
);
1285 flags
= (ei
->i_flags
& ~EXT4_FL_XFLAG_VISIBLE
) |
1286 (flags
& EXT4_FL_XFLAG_VISIBLE
);
1287 err
= ext4_ioctl_check_immutable(inode
, fa
.fsx_projid
, flags
);
1290 err
= ext4_ioctl_setflags(inode
, flags
);
1293 err
= ext4_ioctl_setproject(filp
, fa
.fsx_projid
);
1295 inode_unlock(inode
);
1296 mnt_drop_write_file(filp
);
1299 case EXT4_IOC_SHUTDOWN
:
1300 return ext4_shutdown(sb
, arg
);
1302 case FS_IOC_ENABLE_VERITY
:
1303 if (!ext4_has_feature_verity(sb
))
1305 return fsverity_ioctl_enable(filp
, (const void __user
*)arg
);
1307 case FS_IOC_MEASURE_VERITY
:
1308 if (!ext4_has_feature_verity(sb
))
1310 return fsverity_ioctl_measure(filp
, (void __user
*)arg
);
1317 long ext4_ioctl(struct file
*filp
, unsigned int cmd
, unsigned long arg
)
1321 ext4_fc_start_update(file_inode(filp
));
1322 ret
= __ext4_ioctl(filp
, cmd
, arg
);
1323 ext4_fc_stop_update(file_inode(filp
));
1328 #ifdef CONFIG_COMPAT
1329 long ext4_compat_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
)
1331 /* These are just misnamed, they actually get/put from/to user an int */
1333 case FS_IOC32_GETFLAGS
:
1334 cmd
= FS_IOC_GETFLAGS
;
1336 case FS_IOC32_SETFLAGS
:
1337 cmd
= FS_IOC_SETFLAGS
;
1339 case EXT4_IOC32_GETVERSION
:
1340 cmd
= EXT4_IOC_GETVERSION
;
1342 case EXT4_IOC32_SETVERSION
:
1343 cmd
= EXT4_IOC_SETVERSION
;
1345 case EXT4_IOC32_GROUP_EXTEND
:
1346 cmd
= EXT4_IOC_GROUP_EXTEND
;
1348 case EXT4_IOC32_GETVERSION_OLD
:
1349 cmd
= EXT4_IOC_GETVERSION_OLD
;
1351 case EXT4_IOC32_SETVERSION_OLD
:
1352 cmd
= EXT4_IOC_SETVERSION_OLD
;
1354 case EXT4_IOC32_GETRSVSZ
:
1355 cmd
= EXT4_IOC_GETRSVSZ
;
1357 case EXT4_IOC32_SETRSVSZ
:
1358 cmd
= EXT4_IOC_SETRSVSZ
;
1360 case EXT4_IOC32_GROUP_ADD
: {
1361 struct compat_ext4_new_group_input __user
*uinput
;
1362 struct ext4_new_group_data input
;
1365 uinput
= compat_ptr(arg
);
1366 err
= get_user(input
.group
, &uinput
->group
);
1367 err
|= get_user(input
.block_bitmap
, &uinput
->block_bitmap
);
1368 err
|= get_user(input
.inode_bitmap
, &uinput
->inode_bitmap
);
1369 err
|= get_user(input
.inode_table
, &uinput
->inode_table
);
1370 err
|= get_user(input
.blocks_count
, &uinput
->blocks_count
);
1371 err
|= get_user(input
.reserved_blocks
,
1372 &uinput
->reserved_blocks
);
1375 return ext4_ioctl_group_add(file
, &input
);
1377 case EXT4_IOC_MOVE_EXT
:
1378 case EXT4_IOC_RESIZE_FS
:
1380 case EXT4_IOC_PRECACHE_EXTENTS
:
1381 case FS_IOC_SET_ENCRYPTION_POLICY
:
1382 case FS_IOC_GET_ENCRYPTION_PWSALT
:
1383 case FS_IOC_GET_ENCRYPTION_POLICY
:
1384 case FS_IOC_GET_ENCRYPTION_POLICY_EX
:
1385 case FS_IOC_ADD_ENCRYPTION_KEY
:
1386 case FS_IOC_REMOVE_ENCRYPTION_KEY
:
1387 case FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS
:
1388 case FS_IOC_GET_ENCRYPTION_KEY_STATUS
:
1389 case FS_IOC_GET_ENCRYPTION_NONCE
:
1390 case EXT4_IOC_SHUTDOWN
:
1391 case FS_IOC_GETFSMAP
:
1392 case FS_IOC_ENABLE_VERITY
:
1393 case FS_IOC_MEASURE_VERITY
:
1394 case EXT4_IOC_CLEAR_ES_CACHE
:
1395 case EXT4_IOC_GETSTATE
:
1396 case EXT4_IOC_GET_ES_CACHE
:
1397 case FS_IOC_FSGETXATTR
:
1398 case FS_IOC_FSSETXATTR
:
1401 return -ENOIOCTLCMD
;
1403 return ext4_ioctl(file
, cmd
, (unsigned long) compat_ptr(arg
));