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
))
110 if (!inode_owner_or_capable(inode
) || !capable(CAP_SYS_ADMIN
))
113 inode_bl
= ext4_iget(sb
, EXT4_BOOT_LOADER_INO
);
114 if (IS_ERR(inode_bl
))
115 return PTR_ERR(inode_bl
);
116 ei_bl
= EXT4_I(inode_bl
);
118 filemap_flush(inode
->i_mapping
);
119 filemap_flush(inode_bl
->i_mapping
);
121 /* Protect orig inodes against a truncate and make sure,
122 * that only 1 swap_inode_boot_loader is running. */
123 lock_two_nondirectories(inode
, inode_bl
);
125 truncate_inode_pages(&inode
->i_data
, 0);
126 truncate_inode_pages(&inode_bl
->i_data
, 0);
128 /* Wait for all existing dio workers */
129 ext4_inode_block_unlocked_dio(inode
);
130 ext4_inode_block_unlocked_dio(inode_bl
);
131 inode_dio_wait(inode
);
132 inode_dio_wait(inode_bl
);
134 handle
= ext4_journal_start(inode_bl
, EXT4_HT_MOVE_EXTENTS
, 2);
135 if (IS_ERR(handle
)) {
137 goto journal_err_out
;
140 /* Protect extent tree against block allocations via delalloc */
141 ext4_double_down_write_data_sem(inode
, inode_bl
);
143 if (inode_bl
->i_nlink
== 0) {
144 /* this inode has never been used as a BOOT_LOADER */
145 set_nlink(inode_bl
, 1);
146 i_uid_write(inode_bl
, 0);
147 i_gid_write(inode_bl
, 0);
148 inode_bl
->i_flags
= 0;
150 inode_bl
->i_version
= 1;
151 i_size_write(inode_bl
, 0);
152 inode_bl
->i_mode
= S_IFREG
;
153 if (EXT4_HAS_INCOMPAT_FEATURE(sb
,
154 EXT4_FEATURE_INCOMPAT_EXTENTS
)) {
155 ext4_set_inode_flag(inode_bl
, EXT4_INODE_EXTENTS
);
156 ext4_ext_tree_init(handle
, inode_bl
);
158 memset(ei_bl
->i_data
, 0, sizeof(ei_bl
->i_data
));
161 swap_inode_data(inode
, inode_bl
);
163 inode
->i_ctime
= inode_bl
->i_ctime
= ext4_current_time(inode
);
165 spin_lock(&sbi
->s_next_gen_lock
);
166 inode
->i_generation
= sbi
->s_next_generation
++;
167 inode_bl
->i_generation
= sbi
->s_next_generation
++;
168 spin_unlock(&sbi
->s_next_gen_lock
);
170 ext4_discard_preallocations(inode
);
172 err
= ext4_mark_inode_dirty(handle
, inode
);
174 ext4_warning(inode
->i_sb
,
175 "couldn't mark inode #%lu dirty (err %d)",
177 /* Revert all changes: */
178 swap_inode_data(inode
, inode_bl
);
180 err
= ext4_mark_inode_dirty(handle
, inode_bl
);
182 ext4_warning(inode_bl
->i_sb
,
183 "couldn't mark inode #%lu dirty (err %d)",
184 inode_bl
->i_ino
, err
);
185 /* Revert all changes: */
186 swap_inode_data(inode
, inode_bl
);
187 ext4_mark_inode_dirty(handle
, inode
);
190 ext4_journal_stop(handle
);
191 ext4_double_up_write_data_sem(inode
, inode_bl
);
194 ext4_inode_resume_unlocked_dio(inode
);
195 ext4_inode_resume_unlocked_dio(inode_bl
);
196 unlock_two_nondirectories(inode
, inode_bl
);
201 long ext4_ioctl(struct file
*filp
, unsigned int cmd
, unsigned long arg
)
203 struct inode
*inode
= file_inode(filp
);
204 struct super_block
*sb
= inode
->i_sb
;
205 struct ext4_inode_info
*ei
= EXT4_I(inode
);
208 ext4_debug("cmd = %u, arg = %lu\n", cmd
, arg
);
211 case EXT4_IOC_GETFLAGS
:
212 ext4_get_inode_flags(ei
);
213 flags
= ei
->i_flags
& EXT4_FL_USER_VISIBLE
;
214 return put_user(flags
, (int __user
*) arg
);
215 case EXT4_IOC_SETFLAGS
: {
216 handle_t
*handle
= NULL
;
217 int err
, migrate
= 0;
218 struct ext4_iloc iloc
;
219 unsigned int oldflags
, mask
, i
;
222 if (!inode_owner_or_capable(inode
))
225 if (get_user(flags
, (int __user
*) arg
))
228 err
= mnt_want_write_file(filp
);
232 flags
= ext4_mask_flags(inode
->i_mode
, flags
);
235 mutex_lock(&inode
->i_mutex
);
236 /* Is it quota file? Do not allow user to mess with it */
237 if (IS_NOQUOTA(inode
))
240 oldflags
= ei
->i_flags
;
242 /* The JOURNAL_DATA flag is modifiable only by root */
243 jflag
= flags
& EXT4_JOURNAL_DATA_FL
;
246 * The IMMUTABLE and APPEND_ONLY flags can only be changed by
247 * the relevant capability.
249 * This test looks nicer. Thanks to Pauline Middelink
251 if ((flags
^ oldflags
) & (EXT4_APPEND_FL
| EXT4_IMMUTABLE_FL
)) {
252 if (!capable(CAP_LINUX_IMMUTABLE
))
257 * The JOURNAL_DATA flag can only be changed by
258 * the relevant capability.
260 if ((jflag
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
)) {
261 if (!capable(CAP_SYS_RESOURCE
))
264 if ((flags
^ oldflags
) & EXT4_EXTENTS_FL
)
267 if (flags
& EXT4_EOFBLOCKS_FL
) {
268 /* we don't support adding EOFBLOCKS flag */
269 if (!(oldflags
& EXT4_EOFBLOCKS_FL
)) {
273 } else if (oldflags
& EXT4_EOFBLOCKS_FL
)
274 ext4_truncate(inode
);
276 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
277 if (IS_ERR(handle
)) {
278 err
= PTR_ERR(handle
);
282 ext4_handle_sync(handle
);
283 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
287 for (i
= 0, mask
= 1; i
< 32; i
++, mask
<<= 1) {
288 if (!(mask
& EXT4_FL_USER_MODIFIABLE
))
291 ext4_set_inode_flag(inode
, i
);
293 ext4_clear_inode_flag(inode
, i
);
296 ext4_set_inode_flags(inode
);
297 inode
->i_ctime
= ext4_current_time(inode
);
299 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
301 ext4_journal_stop(handle
);
305 if ((jflag
^ oldflags
) & (EXT4_JOURNAL_DATA_FL
))
306 err
= ext4_change_inode_journal_flag(inode
, jflag
);
310 if (flags
& EXT4_EXTENTS_FL
)
311 err
= ext4_ext_migrate(inode
);
313 err
= ext4_ind_migrate(inode
);
317 mutex_unlock(&inode
->i_mutex
);
318 mnt_drop_write_file(filp
);
321 case EXT4_IOC_GETVERSION
:
322 case EXT4_IOC_GETVERSION_OLD
:
323 return put_user(inode
->i_generation
, (int __user
*) arg
);
324 case EXT4_IOC_SETVERSION
:
325 case EXT4_IOC_SETVERSION_OLD
: {
327 struct ext4_iloc iloc
;
331 if (!inode_owner_or_capable(inode
))
334 if (ext4_has_metadata_csum(inode
->i_sb
)) {
335 ext4_warning(sb
, "Setting inode version is not "
336 "supported with metadata_csum enabled.");
340 err
= mnt_want_write_file(filp
);
343 if (get_user(generation
, (int __user
*) arg
)) {
348 mutex_lock(&inode
->i_mutex
);
349 handle
= ext4_journal_start(inode
, EXT4_HT_INODE
, 1);
350 if (IS_ERR(handle
)) {
351 err
= PTR_ERR(handle
);
354 err
= ext4_reserve_inode_write(handle
, inode
, &iloc
);
356 inode
->i_ctime
= ext4_current_time(inode
);
357 inode
->i_generation
= generation
;
358 err
= ext4_mark_iloc_dirty(handle
, inode
, &iloc
);
360 ext4_journal_stop(handle
);
363 mutex_unlock(&inode
->i_mutex
);
365 mnt_drop_write_file(filp
);
368 case EXT4_IOC_GROUP_EXTEND
: {
369 ext4_fsblk_t n_blocks_count
;
372 err
= ext4_resize_begin(sb
);
376 if (get_user(n_blocks_count
, (__u32 __user
*)arg
)) {
378 goto group_extend_out
;
381 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
382 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
383 ext4_msg(sb
, KERN_ERR
,
384 "Online resizing not supported with bigalloc");
386 goto group_extend_out
;
389 err
= mnt_want_write_file(filp
);
391 goto group_extend_out
;
393 err
= ext4_group_extend(sb
, EXT4_SB(sb
)->s_es
, n_blocks_count
);
394 if (EXT4_SB(sb
)->s_journal
) {
395 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
396 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
397 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
401 mnt_drop_write_file(filp
);
407 case EXT4_IOC_MOVE_EXT
: {
408 struct move_extent me
;
412 if (!(filp
->f_mode
& FMODE_READ
) ||
413 !(filp
->f_mode
& FMODE_WRITE
))
416 if (copy_from_user(&me
,
417 (struct move_extent __user
*)arg
, sizeof(me
)))
421 donor
= fdget(me
.donor_fd
);
425 if (!(donor
.file
->f_mode
& FMODE_WRITE
)) {
430 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
431 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
432 ext4_msg(sb
, KERN_ERR
,
433 "Online defrag not supported with bigalloc");
438 err
= mnt_want_write_file(filp
);
442 err
= ext4_move_extents(filp
, donor
.file
, me
.orig_start
,
443 me
.donor_start
, me
.len
, &me
.moved_len
);
444 mnt_drop_write_file(filp
);
446 if (copy_to_user((struct move_extent __user
*)arg
,
454 case EXT4_IOC_GROUP_ADD
: {
455 struct ext4_new_group_data input
;
458 err
= ext4_resize_begin(sb
);
462 if (copy_from_user(&input
, (struct ext4_new_group_input __user
*)arg
,
468 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
469 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
470 ext4_msg(sb
, KERN_ERR
,
471 "Online resizing not supported with bigalloc");
476 err
= mnt_want_write_file(filp
);
480 err
= ext4_group_add(sb
, &input
);
481 if (EXT4_SB(sb
)->s_journal
) {
482 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
483 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
484 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
488 mnt_drop_write_file(filp
);
489 if (!err
&& ext4_has_group_desc_csum(sb
) &&
490 test_opt(sb
, INIT_INODE_TABLE
))
491 err
= ext4_register_li_request(sb
, input
.group
);
497 case EXT4_IOC_MIGRATE
:
500 if (!inode_owner_or_capable(inode
))
503 err
= mnt_want_write_file(filp
);
507 * inode_mutex prevent write and truncate on the file.
508 * Read still goes through. We take i_data_sem in
509 * ext4_ext_swap_inode_data before we switch the
510 * inode format to prevent read.
512 mutex_lock(&(inode
->i_mutex
));
513 err
= ext4_ext_migrate(inode
);
514 mutex_unlock(&(inode
->i_mutex
));
515 mnt_drop_write_file(filp
);
519 case EXT4_IOC_ALLOC_DA_BLKS
:
522 if (!inode_owner_or_capable(inode
))
525 err
= mnt_want_write_file(filp
);
528 err
= ext4_alloc_da_blocks(inode
);
529 mnt_drop_write_file(filp
);
533 case EXT4_IOC_SWAP_BOOT
:
536 if (!(filp
->f_mode
& FMODE_WRITE
))
538 err
= mnt_want_write_file(filp
);
541 err
= swap_inode_boot_loader(sb
, inode
);
542 mnt_drop_write_file(filp
);
546 case EXT4_IOC_RESIZE_FS
: {
547 ext4_fsblk_t n_blocks_count
;
548 int err
= 0, err2
= 0;
549 ext4_group_t o_group
= EXT4_SB(sb
)->s_groups_count
;
551 if (EXT4_HAS_RO_COMPAT_FEATURE(sb
,
552 EXT4_FEATURE_RO_COMPAT_BIGALLOC
)) {
553 ext4_msg(sb
, KERN_ERR
,
554 "Online resizing not (yet) supported with bigalloc");
558 if (copy_from_user(&n_blocks_count
, (__u64 __user
*)arg
,
563 err
= ext4_resize_begin(sb
);
567 err
= mnt_want_write_file(filp
);
571 err
= ext4_resize_fs(sb
, n_blocks_count
);
572 if (EXT4_SB(sb
)->s_journal
) {
573 jbd2_journal_lock_updates(EXT4_SB(sb
)->s_journal
);
574 err2
= jbd2_journal_flush(EXT4_SB(sb
)->s_journal
);
575 jbd2_journal_unlock_updates(EXT4_SB(sb
)->s_journal
);
579 mnt_drop_write_file(filp
);
580 if (!err
&& (o_group
> EXT4_SB(sb
)->s_groups_count
) &&
581 ext4_has_group_desc_csum(sb
) &&
582 test_opt(sb
, INIT_INODE_TABLE
))
583 err
= ext4_register_li_request(sb
, o_group
);
592 struct request_queue
*q
= bdev_get_queue(sb
->s_bdev
);
593 struct fstrim_range range
;
596 if (!capable(CAP_SYS_ADMIN
))
599 if (!blk_queue_discard(q
))
602 if (copy_from_user(&range
, (struct fstrim_range __user
*)arg
,
606 range
.minlen
= max((unsigned int)range
.minlen
,
607 q
->limits
.discard_granularity
);
608 ret
= ext4_trim_fs(sb
, &range
);
612 if (copy_to_user((struct fstrim_range __user
*)arg
, &range
,
618 case EXT4_IOC_PRECACHE_EXTENTS
:
619 return ext4_ext_precache(inode
);
627 long ext4_compat_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
)
629 /* These are just misnamed, they actually get/put from/to user an int */
631 case EXT4_IOC32_GETFLAGS
:
632 cmd
= EXT4_IOC_GETFLAGS
;
634 case EXT4_IOC32_SETFLAGS
:
635 cmd
= EXT4_IOC_SETFLAGS
;
637 case EXT4_IOC32_GETVERSION
:
638 cmd
= EXT4_IOC_GETVERSION
;
640 case EXT4_IOC32_SETVERSION
:
641 cmd
= EXT4_IOC_SETVERSION
;
643 case EXT4_IOC32_GROUP_EXTEND
:
644 cmd
= EXT4_IOC_GROUP_EXTEND
;
646 case EXT4_IOC32_GETVERSION_OLD
:
647 cmd
= EXT4_IOC_GETVERSION_OLD
;
649 case EXT4_IOC32_SETVERSION_OLD
:
650 cmd
= EXT4_IOC_SETVERSION_OLD
;
652 case EXT4_IOC32_GETRSVSZ
:
653 cmd
= EXT4_IOC_GETRSVSZ
;
655 case EXT4_IOC32_SETRSVSZ
:
656 cmd
= EXT4_IOC_SETRSVSZ
;
658 case EXT4_IOC32_GROUP_ADD
: {
659 struct compat_ext4_new_group_input __user
*uinput
;
660 struct ext4_new_group_input input
;
664 uinput
= compat_ptr(arg
);
665 err
= get_user(input
.group
, &uinput
->group
);
666 err
|= get_user(input
.block_bitmap
, &uinput
->block_bitmap
);
667 err
|= get_user(input
.inode_bitmap
, &uinput
->inode_bitmap
);
668 err
|= get_user(input
.inode_table
, &uinput
->inode_table
);
669 err
|= get_user(input
.blocks_count
, &uinput
->blocks_count
);
670 err
|= get_user(input
.reserved_blocks
,
671 &uinput
->reserved_blocks
);
676 err
= ext4_ioctl(file
, EXT4_IOC_GROUP_ADD
,
677 (unsigned long) &input
);
681 case EXT4_IOC_MOVE_EXT
:
683 case EXT4_IOC_RESIZE_FS
:
684 case EXT4_IOC_PRECACHE_EXTENTS
:
689 return ext4_ioctl(file
, cmd
, (unsigned long) compat_ptr(arg
));