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_bl
;
105 struct ext4_sb_info
*sbi
= EXT4_SB(sb
);
107 if (inode
->i_nlink
!= 1 || !S_ISREG(inode
->i_mode
)) {
112 if (!inode_owner_or_capable(inode
) || !capable(CAP_SYS_ADMIN
)) {
117 inode_bl
= ext4_iget(sb
, EXT4_BOOT_LOADER_INO
);
118 if (IS_ERR(inode_bl
)) {
119 err
= PTR_ERR(inode_bl
);
122 ei_bl
= EXT4_I(inode_bl
);
124 filemap_flush(inode
->i_mapping
);
125 filemap_flush(inode_bl
->i_mapping
);
127 /* Protect orig inodes against a truncate and make sure,
128 * that only 1 swap_inode_boot_loader is running. */
129 lock_two_nondirectories(inode
, inode_bl
);
131 truncate_inode_pages(&inode
->i_data
, 0);
132 truncate_inode_pages(&inode_bl
->i_data
, 0);
134 /* Wait for all existing dio workers */
135 ext4_inode_block_unlocked_dio(inode
);
136 ext4_inode_block_unlocked_dio(inode_bl
);
137 inode_dio_wait(inode
);
138 inode_dio_wait(inode_bl
);
140 handle
= ext4_journal_start(inode_bl
, EXT4_HT_MOVE_EXTENTS
, 2);
141 if (IS_ERR(handle
)) {
146 /* Protect extent tree against block allocations via delalloc */
147 ext4_double_down_write_data_sem(inode
, inode_bl
);
149 if (inode_bl
->i_nlink
== 0) {
150 /* this inode has never been used as a BOOT_LOADER */
151 set_nlink(inode_bl
, 1);
152 i_uid_write(inode_bl
, 0);
153 i_gid_write(inode_bl
, 0);
154 inode_bl
->i_flags
= 0;
156 inode_bl
->i_version
= 1;
157 i_size_write(inode_bl
, 0);
158 inode_bl
->i_mode
= S_IFREG
;
159 if (EXT4_HAS_INCOMPAT_FEATURE(sb
,
160 EXT4_FEATURE_INCOMPAT_EXTENTS
)) {
161 ext4_set_inode_flag(inode_bl
, EXT4_INODE_EXTENTS
);
162 ext4_ext_tree_init(handle
, inode_bl
);
164 memset(ei_bl
->i_data
, 0, sizeof(ei_bl
->i_data
));
167 swap_inode_data(inode
, inode_bl
);
169 inode
->i_ctime
= inode_bl
->i_ctime
= ext4_current_time(inode
);
171 spin_lock(&sbi
->s_next_gen_lock
);
172 inode
->i_generation
= sbi
->s_next_generation
++;
173 inode_bl
->i_generation
= sbi
->s_next_generation
++;
174 spin_unlock(&sbi
->s_next_gen_lock
);
176 ext4_discard_preallocations(inode
);
178 err
= ext4_mark_inode_dirty(handle
, inode
);
180 ext4_warning(inode
->i_sb
,
181 "couldn't mark inode #%lu dirty (err %d)",
183 /* Revert all changes: */
184 swap_inode_data(inode
, inode_bl
);
186 err
= ext4_mark_inode_dirty(handle
, inode_bl
);
188 ext4_warning(inode_bl
->i_sb
,
189 "couldn't mark inode #%lu dirty (err %d)",
190 inode_bl
->i_ino
, err
);
191 /* Revert all changes: */
192 swap_inode_data(inode
, inode_bl
);
193 ext4_mark_inode_dirty(handle
, inode
);
197 ext4_journal_stop(handle
);
199 ext4_double_up_write_data_sem(inode
, inode_bl
);
201 ext4_inode_resume_unlocked_dio(inode
);
202 ext4_inode_resume_unlocked_dio(inode_bl
);
204 unlock_two_nondirectories(inode
, inode_bl
);
212 long ext4_ioctl(struct file
*filp
, unsigned int cmd
, unsigned long arg
)
214 struct inode
*inode
= file_inode(filp
);
215 struct super_block
*sb
= inode
->i_sb
;
216 struct ext4_inode_info
*ei
= EXT4_I(inode
);
219 ext4_debug("cmd = %u, arg = %lu\n", cmd
, arg
);
222 case EXT4_IOC_GETFLAGS
:
223 ext4_get_inode_flags(ei
);
224 flags
= ei
->i_flags
& EXT4_FL_USER_VISIBLE
;
225 return put_user(flags
, (int __user
*) arg
);
226 case EXT4_IOC_SETFLAGS
: {
227 handle_t
*handle
= NULL
;
228 int err
, migrate
= 0;
229 struct ext4_iloc iloc
;
230 unsigned int oldflags
, mask
, i
;
233 if (!inode_owner_or_capable(inode
))
236 if (get_user(flags
, (int __user
*) arg
))
239 err
= mnt_want_write_file(filp
);
243 flags
= ext4_mask_flags(inode
->i_mode
, flags
);
246 mutex_lock(&inode
->i_mutex
);
247 /* Is it quota file? Do not allow user to mess with it */
248 if (IS_NOQUOTA(inode
))
251 oldflags
= ei
->i_flags
;
253 /* The JOURNAL_DATA flag is modifiable only by root */
254 jflag
= flags
& EXT4_JOURNAL_DATA_FL
;
257 * The IMMUTABLE and APPEND_ONLY flags can only be changed by
258 * the relevant capability.
260 * This test looks nicer. Thanks to Pauline Middelink
262 if ((flags
^ oldflags
) & (EXT4_APPEND_FL
| EXT4_IMMUTABLE_FL
)) {
263 if (!capable(CAP_LINUX_IMMUTABLE
))
268 * The JOURNAL_DATA flag can only be changed by
269 * the relevant capability.
271 if ((jflag
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
)) {
272 if (!capable(CAP_SYS_RESOURCE
))
275 if ((flags
^ oldflags
) & EXT4_EXTENTS_FL
)
278 if (flags
& EXT4_EOFBLOCKS_FL
) {
279 /* we don't support adding EOFBLOCKS flag */
280 if (!(oldflags
& EXT4_EOFBLOCKS_FL
)) {
284 } else if (oldflags
& EXT4_EOFBLOCKS_FL
)
285 ext4_truncate(inode
);
287 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
288 if (IS_ERR(handle
)) {
289 err
= PTR_ERR(handle
);
293 ext4_handle_sync(handle
);
294 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
298 for (i
= 0, mask
= 1; i
< 32; i
++, mask
<<= 1) {
299 if (!(mask
& EXT4_FL_USER_MODIFIABLE
))
302 ext4_set_inode_flag(inode
, i
);
304 ext4_clear_inode_flag(inode
, i
);
307 ext4_set_inode_flags(inode
);
308 inode
->i_ctime
= ext4_current_time(inode
);
310 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
312 ext4_journal_stop(handle
);
316 if ((jflag
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
))
317 err
= ext4_change_inode_journal_flag(inode
, jflag
);
321 if (flags
& EXT4_EXTENTS_FL
)
322 err
= ext4_ext_migrate(inode
);
324 err
= ext4_ind_migrate(inode
);
328 mutex_unlock(&inode
->i_mutex
);
329 mnt_drop_write_file(filp
);
332 case EXT4_IOC_GETVERSION
:
333 case EXT4_IOC_GETVERSION_OLD
:
334 return put_user(inode
->i_generation
, (int __user
*) arg
);
335 case EXT4_IOC_SETVERSION
:
336 case EXT4_IOC_SETVERSION_OLD
: {
338 struct ext4_iloc iloc
;
342 if (!inode_owner_or_capable(inode
))
345 if (EXT4_HAS_RO_COMPAT_FEATURE(inode
->i_sb
,
346 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM
)) {
347 ext4_warning(sb
, "Setting inode version is not "
348 "supported with metadata_csum enabled.");
352 err
= mnt_want_write_file(filp
);
355 if (get_user(generation
, (int __user
*) arg
)) {
360 mutex_lock(&inode
->i_mutex
);
361 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
362 if (IS_ERR(handle
)) {
363 err
= PTR_ERR(handle
);
366 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
368 inode
->i_ctime
= ext4_current_time(inode
);
369 inode
->i_generation
= generation
;
370 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
372 ext4_journal_stop(handle
);
375 mutex_unlock(&inode
->i_mutex
);
377 mnt_drop_write_file(filp
);
380 case EXT4_IOC_GROUP_EXTEND
: {
381 ext4_fsblk_t n_blocks_count
;
384 err
= ext4_resize_begin(sb
);
388 if (get_user(n_blocks_count
, (__u32 __user
*)arg
)) {
390 goto group_extend_out
;
393 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
394 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
395 ext4_msg(sb
, KERN_ERR
,
396 "Online resizing not supported with bigalloc");
398 goto group_extend_out
;
401 err
= mnt_want_write_file(filp
);
403 goto group_extend_out
;
405 err
= ext4_group_extend(sb
, EXT4_SB(sb
)->s_es
, n_blocks_count
);
406 if (EXT4_SB(sb
)->s_journal
) {
407 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
408 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
409 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
413 mnt_drop_write_file(filp
);
419 case EXT4_IOC_MOVE_EXT
: {
420 struct move_extent me
;
424 if (!(filp
->f_mode
& FMODE_READ
) ||
425 !(filp
->f_mode
& FMODE_WRITE
))
428 if (copy_from_user(&me
,
429 (struct move_extent __user
*)arg
, sizeof(me
)))
433 donor
= fdget(me
.donor_fd
);
437 if (!(donor
.file
->f_mode
& FMODE_WRITE
)) {
442 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
443 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
444 ext4_msg(sb
, KERN_ERR
,
445 "Online defrag not supported with bigalloc");
450 err
= mnt_want_write_file(filp
);
454 err
= ext4_move_extents(filp
, donor
.file
, me
.orig_start
,
455 me
.donor_start
, me
.len
, &me
.moved_len
);
456 mnt_drop_write_file(filp
);
458 if (copy_to_user((struct move_extent __user
*)arg
,
466 case EXT4_IOC_GROUP_ADD
: {
467 struct ext4_new_group_data input
;
470 err
= ext4_resize_begin(sb
);
474 if (copy_from_user(&input
, (struct ext4_new_group_input __user
*)arg
,
480 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
481 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
482 ext4_msg(sb
, KERN_ERR
,
483 "Online resizing not supported with bigalloc");
488 err
= mnt_want_write_file(filp
);
492 err
= ext4_group_add(sb
, &input
);
493 if (EXT4_SB(sb
)->s_journal
) {
494 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
495 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
496 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
500 mnt_drop_write_file(filp
);
501 if (!err
&& ext4_has_group_desc_csum(sb
) &&
502 test_opt(sb
, INIT_INODE_TABLE
))
503 err
= ext4_register_li_request(sb
, input
.group
);
509 case EXT4_IOC_MIGRATE
:
512 if (!inode_owner_or_capable(inode
))
515 err
= mnt_want_write_file(filp
);
519 * inode_mutex prevent write and truncate on the file.
520 * Read still goes through. We take i_data_sem in
521 * ext4_ext_swap_inode_data before we switch the
522 * inode format to prevent read.
524 mutex_lock(&(inode
->i_mutex
));
525 err
= ext4_ext_migrate(inode
);
526 mutex_unlock(&(inode
->i_mutex
));
527 mnt_drop_write_file(filp
);
531 case EXT4_IOC_ALLOC_DA_BLKS
:
534 if (!inode_owner_or_capable(inode
))
537 err
= mnt_want_write_file(filp
);
540 err
= ext4_alloc_da_blocks(inode
);
541 mnt_drop_write_file(filp
);
545 case EXT4_IOC_SWAP_BOOT
:
546 if (!(filp
->f_mode
& FMODE_WRITE
))
548 return swap_inode_boot_loader(sb
, inode
);
550 case EXT4_IOC_RESIZE_FS
: {
551 ext4_fsblk_t n_blocks_count
;
552 int err
= 0, err2
= 0;
553 ext4_group_t o_group
= EXT4_SB(sb
)->s_groups_count
;
555 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
556 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
557 ext4_msg(sb
, KERN_ERR
,
558 "Online resizing not (yet) supported with bigalloc");
562 if (copy_from_user(&n_blocks_count
, (__u64 __user
*)arg
,
567 err
= ext4_resize_begin(sb
);
571 err
= mnt_want_write_file(filp
);
575 err
= ext4_resize_fs(sb
, n_blocks_count
);
576 if (EXT4_SB(sb
)->s_journal
) {
577 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
578 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
579 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
583 mnt_drop_write_file(filp
);
584 if (!err
&& (o_group
> EXT4_SB(sb
)->s_groups_count
) &&
585 ext4_has_group_desc_csum(sb
) &&
586 test_opt(sb
, INIT_INODE_TABLE
))
587 err
= ext4_register_li_request(sb
, o_group
);
596 struct request_queue
*q
= bdev_get_queue(sb
->s_bdev
);
597 struct fstrim_range range
;
600 if (!capable(CAP_SYS_ADMIN
))
603 if (!blk_queue_discard(q
))
606 if (copy_from_user(&range
, (struct fstrim_range __user
*)arg
,
610 range
.minlen
= max((unsigned int)range
.minlen
,
611 q
->limits
.discard_granularity
);
612 ret
= ext4_trim_fs(sb
, &range
);
616 if (copy_to_user((struct fstrim_range __user
*)arg
, &range
,
622 case EXT4_IOC_PRECACHE_EXTENTS
:
623 return ext4_ext_precache(inode
);
631 long ext4_compat_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
)
633 /* These are just misnamed, they actually get/put from/to user an int */
635 case EXT4_IOC32_GETFLAGS
:
636 cmd
= EXT4_IOC_GETFLAGS
;
638 case EXT4_IOC32_SETFLAGS
:
639 cmd
= EXT4_IOC_SETFLAGS
;
641 case EXT4_IOC32_GETVERSION
:
642 cmd
= EXT4_IOC_GETVERSION
;
644 case EXT4_IOC32_SETVERSION
:
645 cmd
= EXT4_IOC_SETVERSION
;
647 case EXT4_IOC32_GROUP_EXTEND
:
648 cmd
= EXT4_IOC_GROUP_EXTEND
;
650 case EXT4_IOC32_GETVERSION_OLD
:
651 cmd
= EXT4_IOC_GETVERSION_OLD
;
653 case EXT4_IOC32_SETVERSION_OLD
:
654 cmd
= EXT4_IOC_SETVERSION_OLD
;
656 case EXT4_IOC32_GETRSVSZ
:
657 cmd
= EXT4_IOC_GETRSVSZ
;
659 case EXT4_IOC32_SETRSVSZ
:
660 cmd
= EXT4_IOC_SETRSVSZ
;
662 case EXT4_IOC32_GROUP_ADD
: {
663 struct compat_ext4_new_group_input __user
*uinput
;
664 struct ext4_new_group_input input
;
668 uinput
= compat_ptr(arg
);
669 err
= get_user(input
.group
, &uinput
->group
);
670 err
|= get_user(input
.block_bitmap
, &uinput
->block_bitmap
);
671 err
|= get_user(input
.inode_bitmap
, &uinput
->inode_bitmap
);
672 err
|= get_user(input
.inode_table
, &uinput
->inode_table
);
673 err
|= get_user(input
.blocks_count
, &uinput
->blocks_count
);
674 err
|= get_user(input
.reserved_blocks
,
675 &uinput
->reserved_blocks
);
680 err
= ext4_ioctl(file
, EXT4_IOC_GROUP_ADD
,
681 (unsigned long) &input
);
685 case EXT4_IOC_MOVE_EXT
:
687 case EXT4_IOC_RESIZE_FS
:
688 case EXT4_IOC_PRECACHE_EXTENTS
:
693 return ext4_ioctl(file
, cmd
, (unsigned long) compat_ptr(arg
));