1 // SPDX-License-Identifier: GPL-2.0
5 * Copyright (C) 1991, 1992 Linus Torvalds
8 #include <linux/syscalls.h>
10 #include <linux/capability.h>
11 #include <linux/file.h>
13 #include <linux/security.h>
14 #include <linux/export.h>
15 #include <linux/uaccess.h>
16 #include <linux/writeback.h>
17 #include <linux/buffer_head.h>
18 #include <linux/falloc.h>
19 #include <linux/sched/signal.h>
23 #include <asm/ioctls.h>
25 /* So that the fiemap access checks can't overflow on 32 bit machines. */
26 #define FIEMAP_MAX_EXTENTS (UINT_MAX / sizeof(struct fiemap_extent))
29 * vfs_ioctl - call filesystem specific ioctl methods
30 * @filp: open file to invoke ioctl method on
31 * @cmd: ioctl command to execute
32 * @arg: command-specific argument for ioctl
34 * Invokes filesystem specific ->unlocked_ioctl, if one exists; otherwise
37 * Returns 0 on success, -errno on error.
39 long vfs_ioctl(struct file
*filp
, unsigned int cmd
, unsigned long arg
)
43 if (!filp
->f_op
->unlocked_ioctl
)
46 error
= filp
->f_op
->unlocked_ioctl(filp
, cmd
, arg
);
47 if (error
== -ENOIOCTLCMD
)
52 EXPORT_SYMBOL(vfs_ioctl
);
54 static int ioctl_fibmap(struct file
*filp
, int __user
*p
)
56 struct address_space
*mapping
= filp
->f_mapping
;
59 /* do we support this mess? */
60 if (!mapping
->a_ops
->bmap
)
62 if (!capable(CAP_SYS_RAWIO
))
64 res
= get_user(block
, p
);
67 res
= mapping
->a_ops
->bmap(mapping
, block
);
68 return put_user(res
, p
);
72 * fiemap_fill_next_extent - Fiemap helper function
73 * @fieinfo: Fiemap context passed into ->fiemap
74 * @logical: Extent logical start offset, in bytes
75 * @phys: Extent physical start offset, in bytes
76 * @len: Extent length, in bytes
77 * @flags: FIEMAP_EXTENT flags that describe this extent
79 * Called from file system ->fiemap callback. Will populate extent
80 * info as passed in via arguments and copy to user memory. On
81 * success, extent count on fieinfo is incremented.
83 * Returns 0 on success, -errno on error, 1 if this was the last
84 * extent that will fit in user array.
86 #define SET_UNKNOWN_FLAGS (FIEMAP_EXTENT_DELALLOC)
87 #define SET_NO_UNMOUNTED_IO_FLAGS (FIEMAP_EXTENT_DATA_ENCRYPTED)
88 #define SET_NOT_ALIGNED_FLAGS (FIEMAP_EXTENT_DATA_TAIL|FIEMAP_EXTENT_DATA_INLINE)
89 int fiemap_fill_next_extent(struct fiemap_extent_info
*fieinfo
, u64 logical
,
90 u64 phys
, u64 len
, u32 flags
)
92 struct fiemap_extent extent
;
93 struct fiemap_extent __user
*dest
= fieinfo
->fi_extents_start
;
95 /* only count the extents */
96 if (fieinfo
->fi_extents_max
== 0) {
97 fieinfo
->fi_extents_mapped
++;
98 return (flags
& FIEMAP_EXTENT_LAST
) ? 1 : 0;
101 if (fieinfo
->fi_extents_mapped
>= fieinfo
->fi_extents_max
)
104 if (flags
& SET_UNKNOWN_FLAGS
)
105 flags
|= FIEMAP_EXTENT_UNKNOWN
;
106 if (flags
& SET_NO_UNMOUNTED_IO_FLAGS
)
107 flags
|= FIEMAP_EXTENT_ENCODED
;
108 if (flags
& SET_NOT_ALIGNED_FLAGS
)
109 flags
|= FIEMAP_EXTENT_NOT_ALIGNED
;
111 memset(&extent
, 0, sizeof(extent
));
112 extent
.fe_logical
= logical
;
113 extent
.fe_physical
= phys
;
114 extent
.fe_length
= len
;
115 extent
.fe_flags
= flags
;
117 dest
+= fieinfo
->fi_extents_mapped
;
118 if (copy_to_user(dest
, &extent
, sizeof(extent
)))
121 fieinfo
->fi_extents_mapped
++;
122 if (fieinfo
->fi_extents_mapped
== fieinfo
->fi_extents_max
)
124 return (flags
& FIEMAP_EXTENT_LAST
) ? 1 : 0;
126 EXPORT_SYMBOL(fiemap_fill_next_extent
);
129 * fiemap_check_flags - check validity of requested flags for fiemap
130 * @fieinfo: Fiemap context passed into ->fiemap
131 * @fs_flags: Set of fiemap flags that the file system understands
133 * Called from file system ->fiemap callback. This will compute the
134 * intersection of valid fiemap flags and those that the fs supports. That
135 * value is then compared against the user supplied flags. In case of bad user
136 * flags, the invalid values will be written into the fieinfo structure, and
137 * -EBADR is returned, which tells ioctl_fiemap() to return those values to
138 * userspace. For this reason, a return code of -EBADR should be preserved.
140 * Returns 0 on success, -EBADR on bad flags.
142 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
)
146 incompat_flags
= fieinfo
->fi_flags
& ~(FIEMAP_FLAGS_COMPAT
& fs_flags
);
147 if (incompat_flags
) {
148 fieinfo
->fi_flags
= incompat_flags
;
153 EXPORT_SYMBOL(fiemap_check_flags
);
155 static int fiemap_check_ranges(struct super_block
*sb
,
156 u64 start
, u64 len
, u64
*new_len
)
158 u64 maxbytes
= (u64
) sb
->s_maxbytes
;
165 if (start
> maxbytes
)
169 * Shrink request scope to what the fs can actually handle.
171 if (len
> maxbytes
|| (maxbytes
- len
) < start
)
172 *new_len
= maxbytes
- start
;
177 static int ioctl_fiemap(struct file
*filp
, unsigned long arg
)
179 struct fiemap fiemap
;
180 struct fiemap __user
*ufiemap
= (struct fiemap __user
*) arg
;
181 struct fiemap_extent_info fieinfo
= { 0, };
182 struct inode
*inode
= file_inode(filp
);
183 struct super_block
*sb
= inode
->i_sb
;
187 if (!inode
->i_op
->fiemap
)
190 if (copy_from_user(&fiemap
, ufiemap
, sizeof(fiemap
)))
193 if (fiemap
.fm_extent_count
> FIEMAP_MAX_EXTENTS
)
196 error
= fiemap_check_ranges(sb
, fiemap
.fm_start
, fiemap
.fm_length
,
201 fieinfo
.fi_flags
= fiemap
.fm_flags
;
202 fieinfo
.fi_extents_max
= fiemap
.fm_extent_count
;
203 fieinfo
.fi_extents_start
= ufiemap
->fm_extents
;
205 if (fiemap
.fm_extent_count
!= 0 &&
206 !access_ok(fieinfo
.fi_extents_start
,
207 fieinfo
.fi_extents_max
* sizeof(struct fiemap_extent
)))
210 if (fieinfo
.fi_flags
& FIEMAP_FLAG_SYNC
)
211 filemap_write_and_wait(inode
->i_mapping
);
213 error
= inode
->i_op
->fiemap(inode
, &fieinfo
, fiemap
.fm_start
, len
);
214 fiemap
.fm_flags
= fieinfo
.fi_flags
;
215 fiemap
.fm_mapped_extents
= fieinfo
.fi_extents_mapped
;
216 if (copy_to_user(ufiemap
, &fiemap
, sizeof(fiemap
)))
222 static long ioctl_file_clone(struct file
*dst_file
, unsigned long srcfd
,
223 u64 off
, u64 olen
, u64 destoff
)
225 struct fd src_file
= fdget(srcfd
);
232 if (src_file
.file
->f_path
.mnt
!= dst_file
->f_path
.mnt
)
234 cloned
= vfs_clone_file_range(src_file
.file
, off
, dst_file
, destoff
,
238 else if (olen
&& cloned
!= olen
)
247 static long ioctl_file_clone_range(struct file
*file
, void __user
*argp
)
249 struct file_clone_range args
;
251 if (copy_from_user(&args
, argp
, sizeof(args
)))
253 return ioctl_file_clone(file
, args
.src_fd
, args
.src_offset
,
254 args
.src_length
, args
.dest_offset
);
259 static inline sector_t
logical_to_blk(struct inode
*inode
, loff_t offset
)
261 return (offset
>> inode
->i_blkbits
);
264 static inline loff_t
blk_to_logical(struct inode
*inode
, sector_t blk
)
266 return (blk
<< inode
->i_blkbits
);
270 * __generic_block_fiemap - FIEMAP for block based inodes (no locking)
271 * @inode: the inode to map
272 * @fieinfo: the fiemap info struct that will be passed back to userspace
273 * @start: where to start mapping in the inode
274 * @len: how much space to map
275 * @get_block: the fs's get_block function
277 * This does FIEMAP for block based inodes. Basically it will just loop
278 * through get_block until we hit the number of extents we want to map, or we
279 * go past the end of the file and hit a hole.
281 * If it is possible to have data blocks beyond a hole past @inode->i_size, then
282 * please do not use this function, it will stop at the first unmapped block
285 * If you use this function directly, you need to do your own locking. Use
286 * generic_block_fiemap if you want the locking done for you.
289 int __generic_block_fiemap(struct inode
*inode
,
290 struct fiemap_extent_info
*fieinfo
, loff_t start
,
291 loff_t len
, get_block_t
*get_block
)
293 struct buffer_head map_bh
;
294 sector_t start_blk
, last_blk
;
295 loff_t isize
= i_size_read(inode
);
296 u64 logical
= 0, phys
= 0, size
= 0;
297 u32 flags
= FIEMAP_EXTENT_MERGED
;
298 bool past_eof
= false, whole_file
= false;
301 ret
= fiemap_check_flags(fieinfo
, FIEMAP_FLAG_SYNC
);
306 * Either the i_mutex or other appropriate locking needs to be held
307 * since we expect isize to not change at all through the duration of
316 * Some filesystems can't deal with being asked to map less than
317 * blocksize, so make sure our len is at least block length.
319 if (logical_to_blk(inode
, len
) == 0)
320 len
= blk_to_logical(inode
, 1);
322 start_blk
= logical_to_blk(inode
, start
);
323 last_blk
= logical_to_blk(inode
, start
+ len
- 1);
327 * we set b_size to the total size we want so it will map as
328 * many contiguous blocks as possible at once
330 memset(&map_bh
, 0, sizeof(struct buffer_head
));
333 ret
= get_block(inode
, start_blk
, &map_bh
, 0);
338 if (!buffer_mapped(&map_bh
)) {
342 * We want to handle the case where there is an
343 * allocated block at the front of the file, and then
344 * nothing but holes up to the end of the file properly,
345 * to make sure that extent at the front gets properly
346 * marked with FIEMAP_EXTENT_LAST
349 blk_to_logical(inode
, start_blk
) >= isize
)
353 * First hole after going past the EOF, this is our
356 if (past_eof
&& size
) {
357 flags
= FIEMAP_EXTENT_MERGED
|FIEMAP_EXTENT_LAST
;
358 ret
= fiemap_fill_next_extent(fieinfo
, logical
,
362 ret
= fiemap_fill_next_extent(fieinfo
, logical
,
367 /* if we have holes up to/past EOF then we're done */
368 if (start_blk
> last_blk
|| past_eof
|| ret
)
372 * We have gone over the length of what we wanted to
373 * map, and it wasn't the entire file, so add the extent
374 * we got last time and exit.
376 * This is for the case where say we want to map all the
377 * way up to the second to the last block in a file, but
378 * the last block is a hole, making the second to last
379 * block FIEMAP_EXTENT_LAST. In this case we want to
380 * see if there is a hole after the second to last block
381 * so we can mark it properly. If we found data after
382 * we exceeded the length we were requesting, then we
383 * are good to go, just add the extent to the fieinfo
386 if (start_blk
> last_blk
&& !whole_file
) {
387 ret
= fiemap_fill_next_extent(fieinfo
, logical
,
394 * if size != 0 then we know we already have an extent
398 ret
= fiemap_fill_next_extent(fieinfo
, logical
,
405 logical
= blk_to_logical(inode
, start_blk
);
406 phys
= blk_to_logical(inode
, map_bh
.b_blocknr
);
407 size
= map_bh
.b_size
;
408 flags
= FIEMAP_EXTENT_MERGED
;
410 start_blk
+= logical_to_blk(inode
, size
);
413 * If we are past the EOF, then we need to make sure as
414 * soon as we find a hole that the last extent we found
415 * is marked with FIEMAP_EXTENT_LAST
417 if (!past_eof
&& logical
+ size
>= isize
)
421 if (fatal_signal_pending(current
)) {
428 /* If ret is 1 then we just hit the end of the extent array */
434 EXPORT_SYMBOL(__generic_block_fiemap
);
437 * generic_block_fiemap - FIEMAP for block based inodes
438 * @inode: The inode to map
439 * @fieinfo: The mapping information
440 * @start: The initial block to map
441 * @len: The length of the extect to attempt to map
442 * @get_block: The block mapping function for the fs
444 * Calls __generic_block_fiemap to map the inode, after taking
445 * the inode's mutex lock.
448 int generic_block_fiemap(struct inode
*inode
,
449 struct fiemap_extent_info
*fieinfo
, u64 start
,
450 u64 len
, get_block_t
*get_block
)
454 ret
= __generic_block_fiemap(inode
, fieinfo
, start
, len
, get_block
);
458 EXPORT_SYMBOL(generic_block_fiemap
);
460 #endif /* CONFIG_BLOCK */
463 * This provides compatibility with legacy XFS pre-allocation ioctls
464 * which predate the fallocate syscall.
466 * Only the l_start, l_len and l_whence fields of the 'struct space_resv'
467 * are used here, rest are ignored.
469 int ioctl_preallocate(struct file
*filp
, void __user
*argp
)
471 struct inode
*inode
= file_inode(filp
);
472 struct space_resv sr
;
474 if (copy_from_user(&sr
, argp
, sizeof(sr
)))
477 switch (sr
.l_whence
) {
481 sr
.l_start
+= filp
->f_pos
;
484 sr
.l_start
+= i_size_read(inode
);
490 return vfs_fallocate(filp
, FALLOC_FL_KEEP_SIZE
, sr
.l_start
, sr
.l_len
);
493 static int file_ioctl(struct file
*filp
, unsigned int cmd
,
496 struct inode
*inode
= file_inode(filp
);
497 int __user
*p
= (int __user
*)arg
;
501 return ioctl_fibmap(filp
, p
);
503 return put_user(i_size_read(inode
) - filp
->f_pos
, p
);
505 case FS_IOC_RESVSP64
:
506 return ioctl_preallocate(filp
, p
);
509 return vfs_ioctl(filp
, cmd
, arg
);
512 static int ioctl_fionbio(struct file
*filp
, int __user
*argp
)
517 error
= get_user(on
, argp
);
522 /* SunOS compatibility item. */
523 if (O_NONBLOCK
!= O_NDELAY
)
526 spin_lock(&filp
->f_lock
);
528 filp
->f_flags
|= flag
;
530 filp
->f_flags
&= ~flag
;
531 spin_unlock(&filp
->f_lock
);
535 static int ioctl_fioasync(unsigned int fd
, struct file
*filp
,
541 error
= get_user(on
, argp
);
544 flag
= on
? FASYNC
: 0;
546 /* Did FASYNC state change ? */
547 if ((flag
^ filp
->f_flags
) & FASYNC
) {
548 if (filp
->f_op
->fasync
)
549 /* fasync() adjusts filp->f_flags */
550 error
= filp
->f_op
->fasync(fd
, filp
, on
);
554 return error
< 0 ? error
: 0;
557 static int ioctl_fsfreeze(struct file
*filp
)
559 struct super_block
*sb
= file_inode(filp
)->i_sb
;
561 if (!ns_capable(sb
->s_user_ns
, CAP_SYS_ADMIN
))
564 /* If filesystem doesn't support freeze feature, return. */
565 if (sb
->s_op
->freeze_fs
== NULL
&& sb
->s_op
->freeze_super
== NULL
)
569 if (sb
->s_op
->freeze_super
)
570 return sb
->s_op
->freeze_super(sb
);
571 return freeze_super(sb
);
574 static int ioctl_fsthaw(struct file
*filp
)
576 struct super_block
*sb
= file_inode(filp
)->i_sb
;
578 if (!ns_capable(sb
->s_user_ns
, CAP_SYS_ADMIN
))
582 if (sb
->s_op
->thaw_super
)
583 return sb
->s_op
->thaw_super(sb
);
584 return thaw_super(sb
);
587 static int ioctl_file_dedupe_range(struct file
*file
, void __user
*arg
)
589 struct file_dedupe_range __user
*argp
= arg
;
590 struct file_dedupe_range
*same
= NULL
;
595 if (get_user(count
, &argp
->dest_count
)) {
600 size
= offsetof(struct file_dedupe_range __user
, info
[count
]);
601 if (size
> PAGE_SIZE
) {
606 same
= memdup_user(argp
, size
);
613 same
->dest_count
= count
;
614 ret
= vfs_dedupe_file_range(file
, same
);
618 ret
= copy_to_user(argp
, same
, size
);
628 * When you add any new common ioctls to the switches above and below
629 * please update compat_sys_ioctl() too.
631 * do_vfs_ioctl() is not for drivers and not intended to be EXPORT_SYMBOL()'d.
632 * It's just a simple helper for sys_ioctl and compat_sys_ioctl.
634 int do_vfs_ioctl(struct file
*filp
, unsigned int fd
, unsigned int cmd
,
638 int __user
*argp
= (int __user
*)arg
;
639 struct inode
*inode
= file_inode(filp
);
643 set_close_on_exec(fd
, 1);
647 set_close_on_exec(fd
, 0);
651 error
= ioctl_fionbio(filp
, argp
);
655 error
= ioctl_fioasync(fd
, filp
, argp
);
659 if (S_ISDIR(inode
->i_mode
) || S_ISREG(inode
->i_mode
) ||
660 S_ISLNK(inode
->i_mode
)) {
661 loff_t res
= inode_get_bytes(inode
);
662 error
= copy_to_user(argp
, &res
, sizeof(res
)) ?
669 error
= ioctl_fsfreeze(filp
);
673 error
= ioctl_fsthaw(filp
);
677 return ioctl_fiemap(filp
, arg
);
680 /* anon_bdev filesystems may not have a block size */
681 if (!inode
->i_sb
->s_blocksize
)
683 return put_user(inode
->i_sb
->s_blocksize
, argp
);
686 return ioctl_file_clone(filp
, arg
, 0, 0, 0);
689 return ioctl_file_clone_range(filp
, argp
);
692 return ioctl_file_dedupe_range(filp
, argp
);
695 if (S_ISREG(inode
->i_mode
))
696 error
= file_ioctl(filp
, cmd
, arg
);
698 error
= vfs_ioctl(filp
, cmd
, arg
);
704 int ksys_ioctl(unsigned int fd
, unsigned int cmd
, unsigned long arg
)
707 struct fd f
= fdget(fd
);
711 error
= security_file_ioctl(f
.file
, cmd
, arg
);
713 error
= do_vfs_ioctl(f
.file
, fd
, cmd
, arg
);
718 SYSCALL_DEFINE3(ioctl
, unsigned int, fd
, unsigned int, cmd
, unsigned long, arg
)
720 return ksys_ioctl(fd
, cmd
, arg
);