2 * linux/fs/ext4/ioctl.c
4 * Copyright (C) 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
11 #include <linux/jbd2.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 <asm/uaccess.h>
18 #include "ext4_jbd2.h"
21 #define MAX_32_NUM ((((unsigned long long) 1) << 32) - 1)
24 * Swap memory between @a and @b for @len bytes.
26 * @a: pointer to first memory area
27 * @b: pointer to second memory area
28 * @len: number of bytes to swap
31 static void memswap(void *a
, void *b
, size_t len
)
33 unsigned char *ap
, *bp
;
36 ap
= (unsigned char *)a
;
37 bp
= (unsigned char *)b
;
48 * Swap i_data and associated attributes between @inode1 and @inode2.
49 * This function is used for the primary swap between inode1 and inode2
50 * and also to revert this primary swap in case of errors.
52 * Therefore you have to make sure, that calling this method twice
53 * will revert all changes.
55 * @inode1: pointer to first inode
56 * @inode2: pointer to second inode
58 static void swap_inode_data(struct inode
*inode1
, struct inode
*inode2
)
61 struct ext4_inode_info
*ei1
;
62 struct ext4_inode_info
*ei2
;
67 memswap(&inode1
->i_flags
, &inode2
->i_flags
, sizeof(inode1
->i_flags
));
68 memswap(&inode1
->i_version
, &inode2
->i_version
,
69 sizeof(inode1
->i_version
));
70 memswap(&inode1
->i_blocks
, &inode2
->i_blocks
,
71 sizeof(inode1
->i_blocks
));
72 memswap(&inode1
->i_bytes
, &inode2
->i_bytes
, sizeof(inode1
->i_bytes
));
73 memswap(&inode1
->i_atime
, &inode2
->i_atime
, sizeof(inode1
->i_atime
));
74 memswap(&inode1
->i_mtime
, &inode2
->i_mtime
, sizeof(inode1
->i_mtime
));
76 memswap(ei1
->i_data
, ei2
->i_data
, sizeof(ei1
->i_data
));
77 memswap(&ei1
->i_flags
, &ei2
->i_flags
, sizeof(ei1
->i_flags
));
78 memswap(&ei1
->i_disksize
, &ei2
->i_disksize
, sizeof(ei1
->i_disksize
));
79 ext4_es_remove_extent(inode1
, 0, EXT_MAX_BLOCKS
);
80 ext4_es_remove_extent(inode2
, 0, EXT_MAX_BLOCKS
);
82 isize
= i_size_read(inode1
);
83 i_size_write(inode1
, i_size_read(inode2
));
84 i_size_write(inode2
, isize
);
88 * Swap the information from the given @inode and the inode
89 * EXT4_BOOT_LOADER_INO. It will basically swap i_data and all other
90 * important fields of the inodes.
92 * @sb: the super block of the filesystem
93 * @inode: the inode to swap with EXT4_BOOT_LOADER_INO
96 static long swap_inode_boot_loader(struct super_block
*sb
,
101 struct inode
*inode_bl
;
102 struct ext4_inode_info
*ei_bl
;
103 struct ext4_sb_info
*sbi
= EXT4_SB(sb
);
105 if (inode
->i_nlink
!= 1 || !S_ISREG(inode
->i_mode
))
108 if (!inode_owner_or_capable(inode
) || !capable(CAP_SYS_ADMIN
))
111 inode_bl
= ext4_iget(sb
, EXT4_BOOT_LOADER_INO
);
112 if (IS_ERR(inode_bl
))
113 return PTR_ERR(inode_bl
);
114 ei_bl
= EXT4_I(inode_bl
);
116 filemap_flush(inode
->i_mapping
);
117 filemap_flush(inode_bl
->i_mapping
);
119 /* Protect orig inodes against a truncate and make sure,
120 * that only 1 swap_inode_boot_loader is running. */
121 lock_two_nondirectories(inode
, inode_bl
);
123 truncate_inode_pages(&inode
->i_data
, 0);
124 truncate_inode_pages(&inode_bl
->i_data
, 0);
126 /* Wait for all existing dio workers */
127 ext4_inode_block_unlocked_dio(inode
);
128 ext4_inode_block_unlocked_dio(inode_bl
);
129 inode_dio_wait(inode
);
130 inode_dio_wait(inode_bl
);
132 handle
= ext4_journal_start(inode_bl
, EXT4_HT_MOVE_EXTENTS
, 2);
133 if (IS_ERR(handle
)) {
135 goto journal_err_out
;
138 /* Protect extent tree against block allocations via delalloc */
139 ext4_double_down_write_data_sem(inode
, inode_bl
);
141 if (inode_bl
->i_nlink
== 0) {
142 /* this inode has never been used as a BOOT_LOADER */
143 set_nlink(inode_bl
, 1);
144 i_uid_write(inode_bl
, 0);
145 i_gid_write(inode_bl
, 0);
146 inode_bl
->i_flags
= 0;
148 inode_bl
->i_version
= 1;
149 i_size_write(inode_bl
, 0);
150 inode_bl
->i_mode
= S_IFREG
;
151 if (EXT4_HAS_INCOMPAT_FEATURE(sb
,
152 EXT4_FEATURE_INCOMPAT_EXTENTS
)) {
153 ext4_set_inode_flag(inode_bl
, EXT4_INODE_EXTENTS
);
154 ext4_ext_tree_init(handle
, inode_bl
);
156 memset(ei_bl
->i_data
, 0, sizeof(ei_bl
->i_data
));
159 swap_inode_data(inode
, inode_bl
);
161 inode
->i_ctime
= inode_bl
->i_ctime
= ext4_current_time(inode
);
163 spin_lock(&sbi
->s_next_gen_lock
);
164 inode
->i_generation
= sbi
->s_next_generation
++;
165 inode_bl
->i_generation
= sbi
->s_next_generation
++;
166 spin_unlock(&sbi
->s_next_gen_lock
);
168 ext4_discard_preallocations(inode
);
170 err
= ext4_mark_inode_dirty(handle
, inode
);
172 ext4_warning(inode
->i_sb
,
173 "couldn't mark inode #%lu dirty (err %d)",
175 /* Revert all changes: */
176 swap_inode_data(inode
, inode_bl
);
178 err
= ext4_mark_inode_dirty(handle
, inode_bl
);
180 ext4_warning(inode_bl
->i_sb
,
181 "couldn't mark inode #%lu dirty (err %d)",
182 inode_bl
->i_ino
, err
);
183 /* Revert all changes: */
184 swap_inode_data(inode
, inode_bl
);
185 ext4_mark_inode_dirty(handle
, inode
);
188 ext4_journal_stop(handle
);
189 ext4_double_up_write_data_sem(inode
, inode_bl
);
192 ext4_inode_resume_unlocked_dio(inode
);
193 ext4_inode_resume_unlocked_dio(inode_bl
);
194 unlock_two_nondirectories(inode
, inode_bl
);
199 long ext4_ioctl(struct file
*filp
, unsigned int cmd
, unsigned long arg
)
201 struct inode
*inode
= file_inode(filp
);
202 struct super_block
*sb
= inode
->i_sb
;
203 struct ext4_inode_info
*ei
= EXT4_I(inode
);
206 ext4_debug("cmd = %u, arg = %lu\n", cmd
, arg
);
209 case EXT4_IOC_GETFLAGS
:
210 ext4_get_inode_flags(ei
);
211 flags
= ei
->i_flags
& EXT4_FL_USER_VISIBLE
;
212 return put_user(flags
, (int __user
*) arg
);
213 case EXT4_IOC_SETFLAGS
: {
214 handle_t
*handle
= NULL
;
215 int err
, migrate
= 0;
216 struct ext4_iloc iloc
;
217 unsigned int oldflags
, mask
, i
;
220 if (!inode_owner_or_capable(inode
))
223 if (get_user(flags
, (int __user
*) arg
))
226 err
= mnt_want_write_file(filp
);
230 flags
= ext4_mask_flags(inode
->i_mode
, flags
);
233 mutex_lock(&inode
->i_mutex
);
234 /* Is it quota file? Do not allow user to mess with it */
235 if (IS_NOQUOTA(inode
))
238 oldflags
= ei
->i_flags
;
240 /* The JOURNAL_DATA flag is modifiable only by root */
241 jflag
= flags
& EXT4_JOURNAL_DATA_FL
;
244 * The IMMUTABLE and APPEND_ONLY flags can only be changed by
245 * the relevant capability.
247 * This test looks nicer. Thanks to Pauline Middelink
249 if ((flags
^ oldflags
) & (EXT4_APPEND_FL
| EXT4_IMMUTABLE_FL
)) {
250 if (!capable(CAP_LINUX_IMMUTABLE
))
255 * The JOURNAL_DATA flag can only be changed by
256 * the relevant capability.
258 if ((jflag
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
)) {
259 if (!capable(CAP_SYS_RESOURCE
))
262 if ((flags
^ oldflags
) & EXT4_EXTENTS_FL
)
265 if (flags
& EXT4_EOFBLOCKS_FL
) {
266 /* we don't support adding EOFBLOCKS flag */
267 if (!(oldflags
& EXT4_EOFBLOCKS_FL
)) {
271 } else if (oldflags
& EXT4_EOFBLOCKS_FL
)
272 ext4_truncate(inode
);
274 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
275 if (IS_ERR(handle
)) {
276 err
= PTR_ERR(handle
);
280 ext4_handle_sync(handle
);
281 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
285 for (i
= 0, mask
= 1; i
< 32; i
++, mask
<<= 1) {
286 if (!(mask
& EXT4_FL_USER_MODIFIABLE
))
289 ext4_set_inode_flag(inode
, i
);
291 ext4_clear_inode_flag(inode
, i
);
294 ext4_set_inode_flags(inode
);
295 inode
->i_ctime
= ext4_current_time(inode
);
297 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
299 ext4_journal_stop(handle
);
303 if ((jflag
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
))
304 err
= ext4_change_inode_journal_flag(inode
, jflag
);
308 if (flags
& EXT4_EXTENTS_FL
)
309 err
= ext4_ext_migrate(inode
);
311 err
= ext4_ind_migrate(inode
);
315 mutex_unlock(&inode
->i_mutex
);
316 mnt_drop_write_file(filp
);
319 case EXT4_IOC_GETVERSION
:
320 case EXT4_IOC_GETVERSION_OLD
:
321 return put_user(inode
->i_generation
, (int __user
*) arg
);
322 case EXT4_IOC_SETVERSION
:
323 case EXT4_IOC_SETVERSION_OLD
: {
325 struct ext4_iloc iloc
;
329 if (!inode_owner_or_capable(inode
))
332 if (ext4_has_metadata_csum(inode
->i_sb
)) {
333 ext4_warning(sb
, "Setting inode version is not "
334 "supported with metadata_csum enabled.");
338 err
= mnt_want_write_file(filp
);
341 if (get_user(generation
, (int __user
*) arg
)) {
346 mutex_lock(&inode
->i_mutex
);
347 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
348 if (IS_ERR(handle
)) {
349 err
= PTR_ERR(handle
);
352 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
354 inode
->i_ctime
= ext4_current_time(inode
);
355 inode
->i_generation
= generation
;
356 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
358 ext4_journal_stop(handle
);
361 mutex_unlock(&inode
->i_mutex
);
363 mnt_drop_write_file(filp
);
366 case EXT4_IOC_GROUP_EXTEND
: {
367 ext4_fsblk_t n_blocks_count
;
370 err
= ext4_resize_begin(sb
);
374 if (get_user(n_blocks_count
, (__u32 __user
*)arg
)) {
376 goto group_extend_out
;
379 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
380 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
381 ext4_msg(sb
, KERN_ERR
,
382 "Online resizing not supported with bigalloc");
384 goto group_extend_out
;
387 err
= mnt_want_write_file(filp
);
389 goto group_extend_out
;
391 err
= ext4_group_extend(sb
, EXT4_SB(sb
)->s_es
, n_blocks_count
);
392 if (EXT4_SB(sb
)->s_journal
) {
393 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
394 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
395 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
399 mnt_drop_write_file(filp
);
405 case EXT4_IOC_MOVE_EXT
: {
406 struct move_extent me
;
410 if (!(filp
->f_mode
& FMODE_READ
) ||
411 !(filp
->f_mode
& FMODE_WRITE
))
414 if (copy_from_user(&me
,
415 (struct move_extent __user
*)arg
, sizeof(me
)))
419 donor
= fdget(me
.donor_fd
);
423 if (!(donor
.file
->f_mode
& FMODE_WRITE
)) {
428 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
429 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
430 ext4_msg(sb
, KERN_ERR
,
431 "Online defrag not supported with bigalloc");
436 err
= mnt_want_write_file(filp
);
440 err
= ext4_move_extents(filp
, donor
.file
, me
.orig_start
,
441 me
.donor_start
, me
.len
, &me
.moved_len
);
442 mnt_drop_write_file(filp
);
444 if (copy_to_user((struct move_extent __user
*)arg
,
452 case EXT4_IOC_GROUP_ADD
: {
453 struct ext4_new_group_data input
;
456 err
= ext4_resize_begin(sb
);
460 if (copy_from_user(&input
, (struct ext4_new_group_input __user
*)arg
,
466 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
467 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
468 ext4_msg(sb
, KERN_ERR
,
469 "Online resizing not supported with bigalloc");
474 err
= mnt_want_write_file(filp
);
478 err
= ext4_group_add(sb
, &input
);
479 if (EXT4_SB(sb
)->s_journal
) {
480 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
481 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
482 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
486 mnt_drop_write_file(filp
);
487 if (!err
&& ext4_has_group_desc_csum(sb
) &&
488 test_opt(sb
, INIT_INODE_TABLE
))
489 err
= ext4_register_li_request(sb
, input
.group
);
495 case EXT4_IOC_MIGRATE
:
498 if (!inode_owner_or_capable(inode
))
501 err
= mnt_want_write_file(filp
);
505 * inode_mutex prevent write and truncate on the file.
506 * Read still goes through. We take i_data_sem in
507 * ext4_ext_swap_inode_data before we switch the
508 * inode format to prevent read.
510 mutex_lock(&(inode
->i_mutex
));
511 err
= ext4_ext_migrate(inode
);
512 mutex_unlock(&(inode
->i_mutex
));
513 mnt_drop_write_file(filp
);
517 case EXT4_IOC_ALLOC_DA_BLKS
:
520 if (!inode_owner_or_capable(inode
))
523 err
= mnt_want_write_file(filp
);
526 err
= ext4_alloc_da_blocks(inode
);
527 mnt_drop_write_file(filp
);
531 case EXT4_IOC_SWAP_BOOT
:
534 if (!(filp
->f_mode
& FMODE_WRITE
))
536 err
= mnt_want_write_file(filp
);
539 err
= swap_inode_boot_loader(sb
, inode
);
540 mnt_drop_write_file(filp
);
544 case EXT4_IOC_RESIZE_FS
: {
545 ext4_fsblk_t n_blocks_count
;
546 int err
= 0, err2
= 0;
547 ext4_group_t o_group
= EXT4_SB(sb
)->s_groups_count
;
549 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
550 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
551 ext4_msg(sb
, KERN_ERR
,
552 "Online resizing not (yet) supported with bigalloc");
556 if (copy_from_user(&n_blocks_count
, (__u64 __user
*)arg
,
561 err
= ext4_resize_begin(sb
);
565 err
= mnt_want_write_file(filp
);
569 err
= ext4_resize_fs(sb
, n_blocks_count
);
570 if (EXT4_SB(sb
)->s_journal
) {
571 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
572 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
573 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
577 mnt_drop_write_file(filp
);
578 if (!err
&& (o_group
> EXT4_SB(sb
)->s_groups_count
) &&
579 ext4_has_group_desc_csum(sb
) &&
580 test_opt(sb
, INIT_INODE_TABLE
))
581 err
= ext4_register_li_request(sb
, o_group
);
590 struct request_queue
*q
= bdev_get_queue(sb
->s_bdev
);
591 struct fstrim_range range
;
594 if (!capable(CAP_SYS_ADMIN
))
597 if (!blk_queue_discard(q
))
600 if (copy_from_user(&range
, (struct fstrim_range __user
*)arg
,
604 range
.minlen
= max((unsigned int)range
.minlen
,
605 q
->limits
.discard_granularity
);
606 ret
= ext4_trim_fs(sb
, &range
);
610 if (copy_to_user((struct fstrim_range __user
*)arg
, &range
,
616 case EXT4_IOC_PRECACHE_EXTENTS
:
617 return ext4_ext_precache(inode
);
625 long ext4_compat_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
)
627 /* These are just misnamed, they actually get/put from/to user an int */
629 case EXT4_IOC32_GETFLAGS
:
630 cmd
= EXT4_IOC_GETFLAGS
;
632 case EXT4_IOC32_SETFLAGS
:
633 cmd
= EXT4_IOC_SETFLAGS
;
635 case EXT4_IOC32_GETVERSION
:
636 cmd
= EXT4_IOC_GETVERSION
;
638 case EXT4_IOC32_SETVERSION
:
639 cmd
= EXT4_IOC_SETVERSION
;
641 case EXT4_IOC32_GROUP_EXTEND
:
642 cmd
= EXT4_IOC_GROUP_EXTEND
;
644 case EXT4_IOC32_GETVERSION_OLD
:
645 cmd
= EXT4_IOC_GETVERSION_OLD
;
647 case EXT4_IOC32_SETVERSION_OLD
:
648 cmd
= EXT4_IOC_SETVERSION_OLD
;
650 case EXT4_IOC32_GETRSVSZ
:
651 cmd
= EXT4_IOC_GETRSVSZ
;
653 case EXT4_IOC32_SETRSVSZ
:
654 cmd
= EXT4_IOC_SETRSVSZ
;
656 case EXT4_IOC32_GROUP_ADD
: {
657 struct compat_ext4_new_group_input __user
*uinput
;
658 struct ext4_new_group_input input
;
662 uinput
= compat_ptr(arg
);
663 err
= get_user(input
.group
, &uinput
->group
);
664 err
|= get_user(input
.block_bitmap
, &uinput
->block_bitmap
);
665 err
|= get_user(input
.inode_bitmap
, &uinput
->inode_bitmap
);
666 err
|= get_user(input
.inode_table
, &uinput
->inode_table
);
667 err
|= get_user(input
.blocks_count
, &uinput
->blocks_count
);
668 err
|= get_user(input
.reserved_blocks
,
669 &uinput
->reserved_blocks
);
674 err
= ext4_ioctl(file
, EXT4_IOC_GROUP_ADD
,
675 (unsigned long) &input
);
679 case EXT4_IOC_MOVE_EXT
:
681 case EXT4_IOC_RESIZE_FS
:
682 case EXT4_IOC_PRECACHE_EXTENTS
:
687 return ext4_ioctl(file
, cmd
, (unsigned long) compat_ptr(arg
));