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)
53 static int ioctl_fibmap(struct file *filp, int __user *p)
55 struct address_space *mapping = filp->f_mapping;
58 /* do we support this mess? */
59 if (!mapping->a_ops->bmap)
61 if (!capable(CAP_SYS_RAWIO))
63 res = get_user(block, p);
66 res = mapping->a_ops->bmap(mapping, block);
67 return put_user(res, p);
71 * fiemap_fill_next_extent - Fiemap helper function
72 * @fieinfo: Fiemap context passed into ->fiemap
73 * @logical: Extent logical start offset, in bytes
74 * @phys: Extent physical start offset, in bytes
75 * @len: Extent length, in bytes
76 * @flags: FIEMAP_EXTENT flags that describe this extent
78 * Called from file system ->fiemap callback. Will populate extent
79 * info as passed in via arguments and copy to user memory. On
80 * success, extent count on fieinfo is incremented.
82 * Returns 0 on success, -errno on error, 1 if this was the last
83 * extent that will fit in user array.
85 #define SET_UNKNOWN_FLAGS (FIEMAP_EXTENT_DELALLOC)
86 #define SET_NO_UNMOUNTED_IO_FLAGS (FIEMAP_EXTENT_DATA_ENCRYPTED)
87 #define SET_NOT_ALIGNED_FLAGS (FIEMAP_EXTENT_DATA_TAIL|FIEMAP_EXTENT_DATA_INLINE)
88 int fiemap_fill_next_extent(struct fiemap_extent_info *fieinfo, u64 logical,
89 u64 phys, u64 len, u32 flags)
91 struct fiemap_extent extent;
92 struct fiemap_extent __user *dest = fieinfo->fi_extents_start;
94 /* only count the extents */
95 if (fieinfo->fi_extents_max == 0) {
96 fieinfo->fi_extents_mapped++;
97 return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0;
100 if (fieinfo->fi_extents_mapped >= fieinfo->fi_extents_max)
103 if (flags & SET_UNKNOWN_FLAGS)
104 flags |= FIEMAP_EXTENT_UNKNOWN;
105 if (flags & SET_NO_UNMOUNTED_IO_FLAGS)
106 flags |= FIEMAP_EXTENT_ENCODED;
107 if (flags & SET_NOT_ALIGNED_FLAGS)
108 flags |= FIEMAP_EXTENT_NOT_ALIGNED;
110 memset(&extent, 0, sizeof(extent));
111 extent.fe_logical = logical;
112 extent.fe_physical = phys;
113 extent.fe_length = len;
114 extent.fe_flags = flags;
116 dest += fieinfo->fi_extents_mapped;
117 if (copy_to_user(dest, &extent, sizeof(extent)))
120 fieinfo->fi_extents_mapped++;
121 if (fieinfo->fi_extents_mapped == fieinfo->fi_extents_max)
123 return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0;
125 EXPORT_SYMBOL(fiemap_fill_next_extent);
128 * fiemap_check_flags - check validity of requested flags for fiemap
129 * @fieinfo: Fiemap context passed into ->fiemap
130 * @fs_flags: Set of fiemap flags that the file system understands
132 * Called from file system ->fiemap callback. This will compute the
133 * intersection of valid fiemap flags and those that the fs supports. That
134 * value is then compared against the user supplied flags. In case of bad user
135 * flags, the invalid values will be written into the fieinfo structure, and
136 * -EBADR is returned, which tells ioctl_fiemap() to return those values to
137 * userspace. For this reason, a return code of -EBADR should be preserved.
139 * Returns 0 on success, -EBADR on bad flags.
141 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags)
145 incompat_flags = fieinfo->fi_flags & ~(FIEMAP_FLAGS_COMPAT & fs_flags);
146 if (incompat_flags) {
147 fieinfo->fi_flags = incompat_flags;
152 EXPORT_SYMBOL(fiemap_check_flags);
154 static int fiemap_check_ranges(struct super_block *sb,
155 u64 start, u64 len, u64 *new_len)
157 u64 maxbytes = (u64) sb->s_maxbytes;
164 if (start > maxbytes)
168 * Shrink request scope to what the fs can actually handle.
170 if (len > maxbytes || (maxbytes - len) < start)
171 *new_len = maxbytes - start;
176 static int ioctl_fiemap(struct file *filp, unsigned long arg)
178 struct fiemap fiemap;
179 struct fiemap __user *ufiemap = (struct fiemap __user *) arg;
180 struct fiemap_extent_info fieinfo = { 0, };
181 struct inode *inode = file_inode(filp);
182 struct super_block *sb = inode->i_sb;
186 if (!inode->i_op->fiemap)
189 if (copy_from_user(&fiemap, ufiemap, sizeof(fiemap)))
192 if (fiemap.fm_extent_count > FIEMAP_MAX_EXTENTS)
195 error = fiemap_check_ranges(sb, fiemap.fm_start, fiemap.fm_length,
200 fieinfo.fi_flags = fiemap.fm_flags;
201 fieinfo.fi_extents_max = fiemap.fm_extent_count;
202 fieinfo.fi_extents_start = ufiemap->fm_extents;
204 if (fiemap.fm_extent_count != 0 &&
205 !access_ok(VERIFY_WRITE, fieinfo.fi_extents_start,
206 fieinfo.fi_extents_max * sizeof(struct fiemap_extent)))
209 if (fieinfo.fi_flags & FIEMAP_FLAG_SYNC)
210 filemap_write_and_wait(inode->i_mapping);
212 error = inode->i_op->fiemap(inode, &fieinfo, fiemap.fm_start, len);
213 fiemap.fm_flags = fieinfo.fi_flags;
214 fiemap.fm_mapped_extents = fieinfo.fi_extents_mapped;
215 if (copy_to_user(ufiemap, &fiemap, sizeof(fiemap)))
221 static long ioctl_file_clone(struct file *dst_file, unsigned long srcfd,
222 u64 off, u64 olen, u64 destoff)
224 struct fd src_file = fdget(srcfd);
230 if (src_file.file->f_path.mnt != dst_file->f_path.mnt)
232 ret = do_clone_file_range(src_file.file, off, dst_file, destoff, olen);
238 static long ioctl_file_clone_range(struct file *file, void __user *argp)
240 struct file_clone_range args;
242 if (copy_from_user(&args, argp, sizeof(args)))
244 return ioctl_file_clone(file, args.src_fd, args.src_offset,
245 args.src_length, args.dest_offset);
250 static inline sector_t logical_to_blk(struct inode *inode, loff_t offset)
252 return (offset >> inode->i_blkbits);
255 static inline loff_t blk_to_logical(struct inode *inode, sector_t blk)
257 return (blk << inode->i_blkbits);
261 * __generic_block_fiemap - FIEMAP for block based inodes (no locking)
262 * @inode: the inode to map
263 * @fieinfo: the fiemap info struct that will be passed back to userspace
264 * @start: where to start mapping in the inode
265 * @len: how much space to map
266 * @get_block: the fs's get_block function
268 * This does FIEMAP for block based inodes. Basically it will just loop
269 * through get_block until we hit the number of extents we want to map, or we
270 * go past the end of the file and hit a hole.
272 * If it is possible to have data blocks beyond a hole past @inode->i_size, then
273 * please do not use this function, it will stop at the first unmapped block
276 * If you use this function directly, you need to do your own locking. Use
277 * generic_block_fiemap if you want the locking done for you.
280 int __generic_block_fiemap(struct inode *inode,
281 struct fiemap_extent_info *fieinfo, loff_t start,
282 loff_t len, get_block_t *get_block)
284 struct buffer_head map_bh;
285 sector_t start_blk, last_blk;
286 loff_t isize = i_size_read(inode);
287 u64 logical = 0, phys = 0, size = 0;
288 u32 flags = FIEMAP_EXTENT_MERGED;
289 bool past_eof = false, whole_file = false;
292 ret = fiemap_check_flags(fieinfo, FIEMAP_FLAG_SYNC);
297 * Either the i_mutex or other appropriate locking needs to be held
298 * since we expect isize to not change at all through the duration of
307 * Some filesystems can't deal with being asked to map less than
308 * blocksize, so make sure our len is at least block length.
310 if (logical_to_blk(inode, len) == 0)
311 len = blk_to_logical(inode, 1);
313 start_blk = logical_to_blk(inode, start);
314 last_blk = logical_to_blk(inode, start + len - 1);
318 * we set b_size to the total size we want so it will map as
319 * many contiguous blocks as possible at once
321 memset(&map_bh, 0, sizeof(struct buffer_head));
324 ret = get_block(inode, start_blk, &map_bh, 0);
329 if (!buffer_mapped(&map_bh)) {
333 * We want to handle the case where there is an
334 * allocated block at the front of the file, and then
335 * nothing but holes up to the end of the file properly,
336 * to make sure that extent at the front gets properly
337 * marked with FIEMAP_EXTENT_LAST
340 blk_to_logical(inode, start_blk) >= isize)
344 * First hole after going past the EOF, this is our
347 if (past_eof && size) {
348 flags = FIEMAP_EXTENT_MERGED|FIEMAP_EXTENT_LAST;
349 ret = fiemap_fill_next_extent(fieinfo, logical,
353 ret = fiemap_fill_next_extent(fieinfo, logical,
358 /* if we have holes up to/past EOF then we're done */
359 if (start_blk > last_blk || past_eof || ret)
363 * We have gone over the length of what we wanted to
364 * map, and it wasn't the entire file, so add the extent
365 * we got last time and exit.
367 * This is for the case where say we want to map all the
368 * way up to the second to the last block in a file, but
369 * the last block is a hole, making the second to last
370 * block FIEMAP_EXTENT_LAST. In this case we want to
371 * see if there is a hole after the second to last block
372 * so we can mark it properly. If we found data after
373 * we exceeded the length we were requesting, then we
374 * are good to go, just add the extent to the fieinfo
377 if (start_blk > last_blk && !whole_file) {
378 ret = fiemap_fill_next_extent(fieinfo, logical,
385 * if size != 0 then we know we already have an extent
389 ret = fiemap_fill_next_extent(fieinfo, logical,
396 logical = blk_to_logical(inode, start_blk);
397 phys = blk_to_logical(inode, map_bh.b_blocknr);
398 size = map_bh.b_size;
399 flags = FIEMAP_EXTENT_MERGED;
401 start_blk += logical_to_blk(inode, size);
404 * If we are past the EOF, then we need to make sure as
405 * soon as we find a hole that the last extent we found
406 * is marked with FIEMAP_EXTENT_LAST
408 if (!past_eof && logical + size >= isize)
412 if (fatal_signal_pending(current)) {
419 /* If ret is 1 then we just hit the end of the extent array */
425 EXPORT_SYMBOL(__generic_block_fiemap);
428 * generic_block_fiemap - FIEMAP for block based inodes
429 * @inode: The inode to map
430 * @fieinfo: The mapping information
431 * @start: The initial block to map
432 * @len: The length of the extect to attempt to map
433 * @get_block: The block mapping function for the fs
435 * Calls __generic_block_fiemap to map the inode, after taking
436 * the inode's mutex lock.
439 int generic_block_fiemap(struct inode *inode,
440 struct fiemap_extent_info *fieinfo, u64 start,
441 u64 len, get_block_t *get_block)
445 ret = __generic_block_fiemap(inode, fieinfo, start, len, get_block);
449 EXPORT_SYMBOL(generic_block_fiemap);
451 #endif /* CONFIG_BLOCK */
454 * This provides compatibility with legacy XFS pre-allocation ioctls
455 * which predate the fallocate syscall.
457 * Only the l_start, l_len and l_whence fields of the 'struct space_resv'
458 * are used here, rest are ignored.
460 int ioctl_preallocate(struct file *filp, void __user *argp)
462 struct inode *inode = file_inode(filp);
463 struct space_resv sr;
465 if (copy_from_user(&sr, argp, sizeof(sr)))
468 switch (sr.l_whence) {
472 sr.l_start += filp->f_pos;
475 sr.l_start += i_size_read(inode);
481 return vfs_fallocate(filp, FALLOC_FL_KEEP_SIZE, sr.l_start, sr.l_len);
484 static int file_ioctl(struct file *filp, unsigned int cmd,
487 struct inode *inode = file_inode(filp);
488 int __user *p = (int __user *)arg;
492 return ioctl_fibmap(filp, p);
494 return put_user(i_size_read(inode) - filp->f_pos, p);
496 case FS_IOC_RESVSP64:
497 return ioctl_preallocate(filp, p);
500 return vfs_ioctl(filp, cmd, arg);
503 static int ioctl_fionbio(struct file *filp, int __user *argp)
508 error = get_user(on, argp);
513 /* SunOS compatibility item. */
514 if (O_NONBLOCK != O_NDELAY)
517 spin_lock(&filp->f_lock);
519 filp->f_flags |= flag;
521 filp->f_flags &= ~flag;
522 spin_unlock(&filp->f_lock);
526 static int ioctl_fioasync(unsigned int fd, struct file *filp,
532 error = get_user(on, argp);
535 flag = on ? FASYNC : 0;
537 /* Did FASYNC state change ? */
538 if ((flag ^ filp->f_flags) & FASYNC) {
539 if (filp->f_op->fasync)
540 /* fasync() adjusts filp->f_flags */
541 error = filp->f_op->fasync(fd, filp, on);
545 return error < 0 ? error : 0;
548 static int ioctl_fsfreeze(struct file *filp)
550 struct super_block *sb = file_inode(filp)->i_sb;
552 if (!ns_capable(sb->s_user_ns, CAP_SYS_ADMIN))
555 /* If filesystem doesn't support freeze feature, return. */
556 if (sb->s_op->freeze_fs == NULL && sb->s_op->freeze_super == NULL)
560 if (sb->s_op->freeze_super)
561 return sb->s_op->freeze_super(sb);
562 return freeze_super(sb);
565 static int ioctl_fsthaw(struct file *filp)
567 struct super_block *sb = file_inode(filp)->i_sb;
569 if (!ns_capable(sb->s_user_ns, CAP_SYS_ADMIN))
573 if (sb->s_op->thaw_super)
574 return sb->s_op->thaw_super(sb);
575 return thaw_super(sb);
578 static int ioctl_file_dedupe_range(struct file *file, void __user *arg)
580 struct file_dedupe_range __user *argp = arg;
581 struct file_dedupe_range *same = NULL;
586 if (get_user(count, &argp->dest_count)) {
591 size = offsetof(struct file_dedupe_range __user, info[count]);
592 if (size > PAGE_SIZE) {
597 same = memdup_user(argp, size);
604 same->dest_count = count;
605 ret = vfs_dedupe_file_range(file, same);
609 ret = copy_to_user(argp, same, size);
619 * When you add any new common ioctls to the switches above and below
620 * please update compat_sys_ioctl() too.
622 * do_vfs_ioctl() is not for drivers and not intended to be EXPORT_SYMBOL()'d.
623 * It's just a simple helper for sys_ioctl and compat_sys_ioctl.
625 int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
629 int __user *argp = (int __user *)arg;
630 struct inode *inode = file_inode(filp);
634 set_close_on_exec(fd, 1);
638 set_close_on_exec(fd, 0);
642 error = ioctl_fionbio(filp, argp);
646 error = ioctl_fioasync(fd, filp, argp);
650 if (S_ISDIR(inode->i_mode) || S_ISREG(inode->i_mode) ||
651 S_ISLNK(inode->i_mode)) {
652 loff_t res = inode_get_bytes(inode);
653 error = copy_to_user(argp, &res, sizeof(res)) ?
660 error = ioctl_fsfreeze(filp);
664 error = ioctl_fsthaw(filp);
668 return ioctl_fiemap(filp, arg);
671 return put_user(inode->i_sb->s_blocksize, argp);
674 return ioctl_file_clone(filp, arg, 0, 0, 0);
677 return ioctl_file_clone_range(filp, argp);
680 return ioctl_file_dedupe_range(filp, argp);
683 if (S_ISREG(inode->i_mode))
684 error = file_ioctl(filp, cmd, arg);
686 error = vfs_ioctl(filp, cmd, arg);
692 int ksys_ioctl(unsigned int fd, unsigned int cmd, unsigned long arg)
695 struct fd f = fdget(fd);
699 error = security_file_ioctl(f.file, cmd, arg);
701 error = do_vfs_ioctl(f.file, fd, cmd, arg);
706 SYSCALL_DEFINE3(ioctl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
708 return ksys_ioctl(fd, cmd, arg);