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
);
81 ext4_es_lru_del(inode1
);
82 ext4_es_lru_del(inode2
);
84 isize
= i_size_read(inode1
);
85 i_size_write(inode1
, i_size_read(inode2
));
86 i_size_write(inode2
, isize
);
90 * Swap the information from the given @inode and the inode
91 * EXT4_BOOT_LOADER_INO. It will basically swap i_data and all other
92 * important fields of the inodes.
94 * @sb: the super block of the filesystem
95 * @inode: the inode to swap with EXT4_BOOT_LOADER_INO
98 static long swap_inode_boot_loader(struct super_block
*sb
,
103 struct inode
*inode_bl
;
104 struct ext4_inode_info
*ei
;
105 struct ext4_inode_info
*ei_bl
;
106 struct ext4_sb_info
*sbi
;
108 if (inode
->i_nlink
!= 1 || !S_ISREG(inode
->i_mode
)) {
113 if (!inode_owner_or_capable(inode
) || !capable(CAP_SYS_ADMIN
)) {
121 inode_bl
= ext4_iget(sb
, EXT4_BOOT_LOADER_INO
);
122 if (IS_ERR(inode_bl
)) {
123 err
= PTR_ERR(inode_bl
);
126 ei_bl
= EXT4_I(inode_bl
);
128 filemap_flush(inode
->i_mapping
);
129 filemap_flush(inode_bl
->i_mapping
);
131 /* Protect orig inodes against a truncate and make sure,
132 * that only 1 swap_inode_boot_loader is running. */
133 ext4_inode_double_lock(inode
, inode_bl
);
135 truncate_inode_pages(&inode
->i_data
, 0);
136 truncate_inode_pages(&inode_bl
->i_data
, 0);
138 /* Wait for all existing dio workers */
139 ext4_inode_block_unlocked_dio(inode
);
140 ext4_inode_block_unlocked_dio(inode_bl
);
141 inode_dio_wait(inode
);
142 inode_dio_wait(inode_bl
);
144 handle
= ext4_journal_start(inode_bl
, EXT4_HT_MOVE_EXTENTS
, 2);
145 if (IS_ERR(handle
)) {
150 /* Protect extent tree against block allocations via delalloc */
151 ext4_double_down_write_data_sem(inode
, inode_bl
);
153 if (inode_bl
->i_nlink
== 0) {
154 /* this inode has never been used as a BOOT_LOADER */
155 set_nlink(inode_bl
, 1);
156 i_uid_write(inode_bl
, 0);
157 i_gid_write(inode_bl
, 0);
158 inode_bl
->i_flags
= 0;
160 inode_bl
->i_version
= 1;
161 i_size_write(inode_bl
, 0);
162 inode_bl
->i_mode
= S_IFREG
;
163 if (EXT4_HAS_INCOMPAT_FEATURE(sb
,
164 EXT4_FEATURE_INCOMPAT_EXTENTS
)) {
165 ext4_set_inode_flag(inode_bl
, EXT4_INODE_EXTENTS
);
166 ext4_ext_tree_init(handle
, inode_bl
);
168 memset(ei_bl
->i_data
, 0, sizeof(ei_bl
->i_data
));
171 swap_inode_data(inode
, inode_bl
);
173 inode
->i_ctime
= inode_bl
->i_ctime
= ext4_current_time(inode
);
175 spin_lock(&sbi
->s_next_gen_lock
);
176 inode
->i_generation
= sbi
->s_next_generation
++;
177 inode_bl
->i_generation
= sbi
->s_next_generation
++;
178 spin_unlock(&sbi
->s_next_gen_lock
);
180 ext4_discard_preallocations(inode
);
182 err
= ext4_mark_inode_dirty(handle
, inode
);
184 ext4_warning(inode
->i_sb
,
185 "couldn't mark inode #%lu dirty (err %d)",
187 /* Revert all changes: */
188 swap_inode_data(inode
, inode_bl
);
190 err
= ext4_mark_inode_dirty(handle
, inode_bl
);
192 ext4_warning(inode_bl
->i_sb
,
193 "couldn't mark inode #%lu dirty (err %d)",
194 inode_bl
->i_ino
, err
);
195 /* Revert all changes: */
196 swap_inode_data(inode
, inode_bl
);
197 ext4_mark_inode_dirty(handle
, inode
);
201 ext4_journal_stop(handle
);
203 ext4_double_up_write_data_sem(inode
, inode_bl
);
205 ext4_inode_resume_unlocked_dio(inode
);
206 ext4_inode_resume_unlocked_dio(inode_bl
);
208 ext4_inode_double_unlock(inode
, inode_bl
);
216 long ext4_ioctl(struct file
*filp
, unsigned int cmd
, unsigned long arg
)
218 struct inode
*inode
= file_inode(filp
);
219 struct super_block
*sb
= inode
->i_sb
;
220 struct ext4_inode_info
*ei
= EXT4_I(inode
);
223 ext4_debug("cmd = %u, arg = %lu\n", cmd
, arg
);
226 case EXT4_IOC_GETFLAGS
:
227 ext4_get_inode_flags(ei
);
228 flags
= ei
->i_flags
& EXT4_FL_USER_VISIBLE
;
229 return put_user(flags
, (int __user
*) arg
);
230 case EXT4_IOC_SETFLAGS
: {
231 handle_t
*handle
= NULL
;
232 int err
, migrate
= 0;
233 struct ext4_iloc iloc
;
234 unsigned int oldflags
, mask
, i
;
237 if (!inode_owner_or_capable(inode
))
240 if (get_user(flags
, (int __user
*) arg
))
243 err
= mnt_want_write_file(filp
);
247 flags
= ext4_mask_flags(inode
->i_mode
, flags
);
250 mutex_lock(&inode
->i_mutex
);
251 /* Is it quota file? Do not allow user to mess with it */
252 if (IS_NOQUOTA(inode
))
255 oldflags
= ei
->i_flags
;
257 /* The JOURNAL_DATA flag is modifiable only by root */
258 jflag
= flags
& EXT4_JOURNAL_DATA_FL
;
261 * The IMMUTABLE and APPEND_ONLY flags can only be changed by
262 * the relevant capability.
264 * This test looks nicer. Thanks to Pauline Middelink
266 if ((flags
^ oldflags
) & (EXT4_APPEND_FL
| EXT4_IMMUTABLE_FL
)) {
267 if (!capable(CAP_LINUX_IMMUTABLE
))
272 * The JOURNAL_DATA flag can only be changed by
273 * the relevant capability.
275 if ((jflag
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
)) {
276 if (!capable(CAP_SYS_RESOURCE
))
279 if ((flags
^ oldflags
) & EXT4_EXTENTS_FL
)
282 if (flags
& EXT4_EOFBLOCKS_FL
) {
283 /* we don't support adding EOFBLOCKS flag */
284 if (!(oldflags
& EXT4_EOFBLOCKS_FL
)) {
288 } else if (oldflags
& EXT4_EOFBLOCKS_FL
)
289 ext4_truncate(inode
);
291 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
292 if (IS_ERR(handle
)) {
293 err
= PTR_ERR(handle
);
297 ext4_handle_sync(handle
);
298 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
302 for (i
= 0, mask
= 1; i
< 32; i
++, mask
<<= 1) {
303 if (!(mask
& EXT4_FL_USER_MODIFIABLE
))
306 ext4_set_inode_flag(inode
, i
);
308 ext4_clear_inode_flag(inode
, i
);
311 ext4_set_inode_flags(inode
);
312 inode
->i_ctime
= ext4_current_time(inode
);
314 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
316 ext4_journal_stop(handle
);
320 if ((jflag
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
))
321 err
= ext4_change_inode_journal_flag(inode
, jflag
);
325 if (flags
& EXT4_EXTENTS_FL
)
326 err
= ext4_ext_migrate(inode
);
328 err
= ext4_ind_migrate(inode
);
332 mutex_unlock(&inode
->i_mutex
);
333 mnt_drop_write_file(filp
);
336 case EXT4_IOC_GETVERSION
:
337 case EXT4_IOC_GETVERSION_OLD
:
338 return put_user(inode
->i_generation
, (int __user
*) arg
);
339 case EXT4_IOC_SETVERSION
:
340 case EXT4_IOC_SETVERSION_OLD
: {
342 struct ext4_iloc iloc
;
346 if (!inode_owner_or_capable(inode
))
349 if (EXT4_HAS_RO_COMPAT_FEATURE(inode
->i_sb
,
350 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM
)) {
351 ext4_warning(sb
, "Setting inode version is not "
352 "supported with metadata_csum enabled.");
356 err
= mnt_want_write_file(filp
);
359 if (get_user(generation
, (int __user
*) arg
)) {
364 mutex_lock(&inode
->i_mutex
);
365 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
366 if (IS_ERR(handle
)) {
367 err
= PTR_ERR(handle
);
370 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
372 inode
->i_ctime
= ext4_current_time(inode
);
373 inode
->i_generation
= generation
;
374 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
376 ext4_journal_stop(handle
);
379 mutex_unlock(&inode
->i_mutex
);
381 mnt_drop_write_file(filp
);
384 case EXT4_IOC_GROUP_EXTEND
: {
385 ext4_fsblk_t n_blocks_count
;
388 err
= ext4_resize_begin(sb
);
392 if (get_user(n_blocks_count
, (__u32 __user
*)arg
)) {
394 goto group_extend_out
;
397 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
398 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
399 ext4_msg(sb
, KERN_ERR
,
400 "Online resizing not supported with bigalloc");
402 goto group_extend_out
;
405 err
= mnt_want_write_file(filp
);
407 goto group_extend_out
;
409 err
= ext4_group_extend(sb
, EXT4_SB(sb
)->s_es
, n_blocks_count
);
410 if (EXT4_SB(sb
)->s_journal
) {
411 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
412 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
413 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
417 mnt_drop_write_file(filp
);
423 case EXT4_IOC_MOVE_EXT
: {
424 struct move_extent me
;
428 if (!(filp
->f_mode
& FMODE_READ
) ||
429 !(filp
->f_mode
& FMODE_WRITE
))
432 if (copy_from_user(&me
,
433 (struct move_extent __user
*)arg
, sizeof(me
)))
437 donor
= fdget(me
.donor_fd
);
441 if (!(donor
.file
->f_mode
& FMODE_WRITE
)) {
446 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
447 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
448 ext4_msg(sb
, KERN_ERR
,
449 "Online defrag not supported with bigalloc");
454 err
= mnt_want_write_file(filp
);
458 err
= ext4_move_extents(filp
, donor
.file
, me
.orig_start
,
459 me
.donor_start
, me
.len
, &me
.moved_len
);
460 mnt_drop_write_file(filp
);
462 if (copy_to_user((struct move_extent __user
*)arg
,
470 case EXT4_IOC_GROUP_ADD
: {
471 struct ext4_new_group_data input
;
474 err
= ext4_resize_begin(sb
);
478 if (copy_from_user(&input
, (struct ext4_new_group_input __user
*)arg
,
484 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
485 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
486 ext4_msg(sb
, KERN_ERR
,
487 "Online resizing not supported with bigalloc");
492 err
= mnt_want_write_file(filp
);
496 err
= ext4_group_add(sb
, &input
);
497 if (EXT4_SB(sb
)->s_journal
) {
498 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
499 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
500 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
504 mnt_drop_write_file(filp
);
505 if (!err
&& ext4_has_group_desc_csum(sb
) &&
506 test_opt(sb
, INIT_INODE_TABLE
))
507 err
= ext4_register_li_request(sb
, input
.group
);
513 case EXT4_IOC_MIGRATE
:
516 if (!inode_owner_or_capable(inode
))
519 err
= mnt_want_write_file(filp
);
523 * inode_mutex prevent write and truncate on the file.
524 * Read still goes through. We take i_data_sem in
525 * ext4_ext_swap_inode_data before we switch the
526 * inode format to prevent read.
528 mutex_lock(&(inode
->i_mutex
));
529 err
= ext4_ext_migrate(inode
);
530 mutex_unlock(&(inode
->i_mutex
));
531 mnt_drop_write_file(filp
);
535 case EXT4_IOC_ALLOC_DA_BLKS
:
538 if (!inode_owner_or_capable(inode
))
541 err
= mnt_want_write_file(filp
);
544 err
= ext4_alloc_da_blocks(inode
);
545 mnt_drop_write_file(filp
);
549 case EXT4_IOC_SWAP_BOOT
:
550 if (!(filp
->f_mode
& FMODE_WRITE
))
552 return swap_inode_boot_loader(sb
, inode
);
554 case EXT4_IOC_RESIZE_FS
: {
555 ext4_fsblk_t n_blocks_count
;
556 int err
= 0, err2
= 0;
557 ext4_group_t o_group
= EXT4_SB(sb
)->s_groups_count
;
559 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
560 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
561 ext4_msg(sb
, KERN_ERR
,
562 "Online resizing not (yet) supported with bigalloc");
566 if (copy_from_user(&n_blocks_count
, (__u64 __user
*)arg
,
571 err
= ext4_resize_begin(sb
);
575 err
= mnt_want_write_file(filp
);
579 err
= ext4_resize_fs(sb
, n_blocks_count
);
580 if (EXT4_SB(sb
)->s_journal
) {
581 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
582 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
583 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
587 mnt_drop_write_file(filp
);
588 if (!err
&& (o_group
> EXT4_SB(sb
)->s_groups_count
) &&
589 ext4_has_group_desc_csum(sb
) &&
590 test_opt(sb
, INIT_INODE_TABLE
))
591 err
= ext4_register_li_request(sb
, o_group
);
600 struct request_queue
*q
= bdev_get_queue(sb
->s_bdev
);
601 struct fstrim_range range
;
604 if (!capable(CAP_SYS_ADMIN
))
607 if (!blk_queue_discard(q
))
610 if (copy_from_user(&range
, (struct fstrim_range __user
*)arg
,
614 range
.minlen
= max((unsigned int)range
.minlen
,
615 q
->limits
.discard_granularity
);
616 ret
= ext4_trim_fs(sb
, &range
);
620 if (copy_to_user((struct fstrim_range __user
*)arg
, &range
,
626 case EXT4_IOC_PRECACHE_EXTENTS
:
627 return ext4_ext_precache(inode
);
635 long ext4_compat_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
)
637 /* These are just misnamed, they actually get/put from/to user an int */
639 case EXT4_IOC32_GETFLAGS
:
640 cmd
= EXT4_IOC_GETFLAGS
;
642 case EXT4_IOC32_SETFLAGS
:
643 cmd
= EXT4_IOC_SETFLAGS
;
645 case EXT4_IOC32_GETVERSION
:
646 cmd
= EXT4_IOC_GETVERSION
;
648 case EXT4_IOC32_SETVERSION
:
649 cmd
= EXT4_IOC_SETVERSION
;
651 case EXT4_IOC32_GROUP_EXTEND
:
652 cmd
= EXT4_IOC_GROUP_EXTEND
;
654 case EXT4_IOC32_GETVERSION_OLD
:
655 cmd
= EXT4_IOC_GETVERSION_OLD
;
657 case EXT4_IOC32_SETVERSION_OLD
:
658 cmd
= EXT4_IOC_SETVERSION_OLD
;
660 case EXT4_IOC32_GETRSVSZ
:
661 cmd
= EXT4_IOC_GETRSVSZ
;
663 case EXT4_IOC32_SETRSVSZ
:
664 cmd
= EXT4_IOC_SETRSVSZ
;
666 case EXT4_IOC32_GROUP_ADD
: {
667 struct compat_ext4_new_group_input __user
*uinput
;
668 struct ext4_new_group_input input
;
672 uinput
= compat_ptr(arg
);
673 err
= get_user(input
.group
, &uinput
->group
);
674 err
|= get_user(input
.block_bitmap
, &uinput
->block_bitmap
);
675 err
|= get_user(input
.inode_bitmap
, &uinput
->inode_bitmap
);
676 err
|= get_user(input
.inode_table
, &uinput
->inode_table
);
677 err
|= get_user(input
.blocks_count
, &uinput
->blocks_count
);
678 err
|= get_user(input
.reserved_blocks
,
679 &uinput
->reserved_blocks
);
684 err
= ext4_ioctl(file
, EXT4_IOC_GROUP_ADD
,
685 (unsigned long) &input
);
689 case EXT4_IOC_MOVE_EXT
:
691 case EXT4_IOC_RESIZE_FS
:
692 case EXT4_IOC_PRECACHE_EXTENTS
:
697 return ext4_ioctl(file
, cmd
, (unsigned long) compat_ptr(arg
));