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(VERIFY_WRITE, 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);
231 if (src_file.file->f_path.mnt != dst_file->f_path.mnt)
233 ret = vfs_clone_file_range(src_file.file, off, dst_file, destoff, olen);
239 static long ioctl_file_clone_range(struct file *file, void __user *argp)
241 struct file_clone_range args;
243 if (copy_from_user(&args, argp, sizeof(args)))
245 return ioctl_file_clone(file, args.src_fd, args.src_offset,
246 args.src_length, args.dest_offset);
251 static inline sector_t logical_to_blk(struct inode *inode, loff_t offset)
253 return (offset >> inode->i_blkbits);
256 static inline loff_t blk_to_logical(struct inode *inode, sector_t blk)
258 return (blk << inode->i_blkbits);
262 * __generic_block_fiemap - FIEMAP for block based inodes (no locking)
263 * @inode: the inode to map
264 * @fieinfo: the fiemap info struct that will be passed back to userspace
265 * @start: where to start mapping in the inode
266 * @len: how much space to map
267 * @get_block: the fs's get_block function
269 * This does FIEMAP for block based inodes. Basically it will just loop
270 * through get_block until we hit the number of extents we want to map, or we
271 * go past the end of the file and hit a hole.
273 * If it is possible to have data blocks beyond a hole past @inode->i_size, then
274 * please do not use this function, it will stop at the first unmapped block
277 * If you use this function directly, you need to do your own locking. Use
278 * generic_block_fiemap if you want the locking done for you.
281 int __generic_block_fiemap(struct inode *inode,
282 struct fiemap_extent_info *fieinfo, loff_t start,
283 loff_t len, get_block_t *get_block)
285 struct buffer_head map_bh;
286 sector_t start_blk, last_blk;
287 loff_t isize = i_size_read(inode);
288 u64 logical = 0, phys = 0, size = 0;
289 u32 flags = FIEMAP_EXTENT_MERGED;
290 bool past_eof = false, whole_file = false;
293 ret = fiemap_check_flags(fieinfo, FIEMAP_FLAG_SYNC);
298 * Either the i_mutex or other appropriate locking needs to be held
299 * since we expect isize to not change at all through the duration of
308 * Some filesystems can't deal with being asked to map less than
309 * blocksize, so make sure our len is at least block length.
311 if (logical_to_blk(inode, len) == 0)
312 len = blk_to_logical(inode, 1);
314 start_blk = logical_to_blk(inode, start);
315 last_blk = logical_to_blk(inode, start + len - 1);
319 * we set b_size to the total size we want so it will map as
320 * many contiguous blocks as possible at once
322 memset(&map_bh, 0, sizeof(struct buffer_head));
325 ret = get_block(inode, start_blk, &map_bh, 0);
330 if (!buffer_mapped(&map_bh)) {
334 * We want to handle the case where there is an
335 * allocated block at the front of the file, and then
336 * nothing but holes up to the end of the file properly,
337 * to make sure that extent at the front gets properly
338 * marked with FIEMAP_EXTENT_LAST
341 blk_to_logical(inode, start_blk) >= isize)
345 * First hole after going past the EOF, this is our
348 if (past_eof && size) {
349 flags = FIEMAP_EXTENT_MERGED|FIEMAP_EXTENT_LAST;
350 ret = fiemap_fill_next_extent(fieinfo, logical,
354 ret = fiemap_fill_next_extent(fieinfo, logical,
359 /* if we have holes up to/past EOF then we're done */
360 if (start_blk > last_blk || past_eof || ret)
364 * We have gone over the length of what we wanted to
365 * map, and it wasn't the entire file, so add the extent
366 * we got last time and exit.
368 * This is for the case where say we want to map all the
369 * way up to the second to the last block in a file, but
370 * the last block is a hole, making the second to last
371 * block FIEMAP_EXTENT_LAST. In this case we want to
372 * see if there is a hole after the second to last block
373 * so we can mark it properly. If we found data after
374 * we exceeded the length we were requesting, then we
375 * are good to go, just add the extent to the fieinfo
378 if (start_blk > last_blk && !whole_file) {
379 ret = fiemap_fill_next_extent(fieinfo, logical,
386 * if size != 0 then we know we already have an extent
390 ret = fiemap_fill_next_extent(fieinfo, logical,
397 logical = blk_to_logical(inode, start_blk);
398 phys = blk_to_logical(inode, map_bh.b_blocknr);
399 size = map_bh.b_size;
400 flags = FIEMAP_EXTENT_MERGED;
402 start_blk += logical_to_blk(inode, size);
405 * If we are past the EOF, then we need to make sure as
406 * soon as we find a hole that the last extent we found
407 * is marked with FIEMAP_EXTENT_LAST
409 if (!past_eof && logical + size >= isize)
413 if (fatal_signal_pending(current)) {
420 /* If ret is 1 then we just hit the end of the extent array */
426 EXPORT_SYMBOL(__generic_block_fiemap);
429 * generic_block_fiemap - FIEMAP for block based inodes
430 * @inode: The inode to map
431 * @fieinfo: The mapping information
432 * @start: The initial block to map
433 * @len: The length of the extect to attempt to map
434 * @get_block: The block mapping function for the fs
436 * Calls __generic_block_fiemap to map the inode, after taking
437 * the inode's mutex lock.
440 int generic_block_fiemap(struct inode *inode,
441 struct fiemap_extent_info *fieinfo, u64 start,
442 u64 len, get_block_t *get_block)
446 ret = __generic_block_fiemap(inode, fieinfo, start, len, get_block);
450 EXPORT_SYMBOL(generic_block_fiemap);
452 #endif /* CONFIG_BLOCK */
455 * This provides compatibility with legacy XFS pre-allocation ioctls
456 * which predate the fallocate syscall.
458 * Only the l_start, l_len and l_whence fields of the 'struct space_resv'
459 * are used here, rest are ignored.
461 int ioctl_preallocate(struct file *filp, void __user *argp)
463 struct inode *inode = file_inode(filp);
464 struct space_resv sr;
466 if (copy_from_user(&sr, argp, sizeof(sr)))
469 switch (sr.l_whence) {
473 sr.l_start += filp->f_pos;
476 sr.l_start += i_size_read(inode);
482 return vfs_fallocate(filp, FALLOC_FL_KEEP_SIZE, sr.l_start, sr.l_len);
485 static int file_ioctl(struct file *filp, unsigned int cmd,
488 struct inode *inode = file_inode(filp);
489 int __user *p = (int __user *)arg;
493 return ioctl_fibmap(filp, p);
495 return put_user(i_size_read(inode) - filp->f_pos, p);
497 case FS_IOC_RESVSP64:
498 return ioctl_preallocate(filp, p);
501 return vfs_ioctl(filp, cmd, arg);
504 static int ioctl_fionbio(struct file *filp, int __user *argp)
509 error = get_user(on, argp);
514 /* SunOS compatibility item. */
515 if (O_NONBLOCK != O_NDELAY)
518 spin_lock(&filp->f_lock);
520 filp->f_flags |= flag;
522 filp->f_flags &= ~flag;
523 spin_unlock(&filp->f_lock);
527 static int ioctl_fioasync(unsigned int fd, struct file *filp,
533 error = get_user(on, argp);
536 flag = on ? FASYNC : 0;
538 /* Did FASYNC state change ? */
539 if ((flag ^ filp->f_flags) & FASYNC) {
540 if (filp->f_op->fasync)
541 /* fasync() adjusts filp->f_flags */
542 error = filp->f_op->fasync(fd, filp, on);
546 return error < 0 ? error : 0;
549 static int ioctl_fsfreeze(struct file *filp)
551 struct super_block *sb = file_inode(filp)->i_sb;
553 if (!ns_capable(sb->s_user_ns, CAP_SYS_ADMIN))
556 /* If filesystem doesn't support freeze feature, return. */
557 if (sb->s_op->freeze_fs == NULL && sb->s_op->freeze_super == NULL)
561 if (sb->s_op->freeze_super)
562 return sb->s_op->freeze_super(sb);
563 return freeze_super(sb);
566 static int ioctl_fsthaw(struct file *filp)
568 struct super_block *sb = file_inode(filp)->i_sb;
570 if (!ns_capable(sb->s_user_ns, CAP_SYS_ADMIN))
574 if (sb->s_op->thaw_super)
575 return sb->s_op->thaw_super(sb);
576 return thaw_super(sb);
579 static int ioctl_file_dedupe_range(struct file *file, void __user *arg)
581 struct file_dedupe_range __user *argp = arg;
582 struct file_dedupe_range *same = NULL;
587 if (get_user(count, &argp->dest_count)) {
592 size = offsetof(struct file_dedupe_range __user, info[count]);
593 if (size > PAGE_SIZE) {
598 same = memdup_user(argp, size);
605 same->dest_count = count;
606 ret = vfs_dedupe_file_range(file, same);
610 ret = copy_to_user(argp, same, size);
620 * When you add any new common ioctls to the switches above and below
621 * please update compat_sys_ioctl() too.
623 * do_vfs_ioctl() is not for drivers and not intended to be EXPORT_SYMBOL()'d.
624 * It's just a simple helper for sys_ioctl and compat_sys_ioctl.
626 int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
630 int __user *argp = (int __user *)arg;
631 struct inode *inode = file_inode(filp);
635 set_close_on_exec(fd, 1);
639 set_close_on_exec(fd, 0);
643 error = ioctl_fionbio(filp, argp);
647 error = ioctl_fioasync(fd, filp, argp);
651 if (S_ISDIR(inode->i_mode) || S_ISREG(inode->i_mode) ||
652 S_ISLNK(inode->i_mode)) {
653 loff_t res = inode_get_bytes(inode);
654 error = copy_to_user(argp, &res, sizeof(res)) ?
661 error = ioctl_fsfreeze(filp);
665 error = ioctl_fsthaw(filp);
669 return ioctl_fiemap(filp, arg);
672 /* anon_bdev filesystems may not have a block size */
673 if (!inode->i_sb->s_blocksize)
675 return put_user(inode->i_sb->s_blocksize, argp);
678 return ioctl_file_clone(filp, arg, 0, 0, 0);
681 return ioctl_file_clone_range(filp, argp);
684 return ioctl_file_dedupe_range(filp, argp);
687 if (S_ISREG(inode->i_mode))
688 error = file_ioctl(filp, cmd, arg);
690 error = vfs_ioctl(filp, cmd, arg);
696 int ksys_ioctl(unsigned int fd, unsigned int cmd, unsigned long arg)
699 struct fd f = fdget(fd);
703 error = security_file_ioctl(f.file, cmd, arg);
705 error = do_vfs_ioctl(f.file, fd, cmd, arg);
710 SYSCALL_DEFINE3(ioctl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
712 return ksys_ioctl(fd, cmd, arg);