4 * Copyright (C) 1991, 1992 Linus Torvalds
7 #include <linux/syscalls.h>
9 #include <linux/capability.h>
10 #include <linux/file.h>
12 #include <linux/security.h>
13 #include <linux/export.h>
14 #include <linux/uaccess.h>
15 #include <linux/writeback.h>
16 #include <linux/buffer_head.h>
17 #include <linux/falloc.h>
20 #include <asm/ioctls.h>
22 /* So that the fiemap access checks can't overflow on 32 bit machines. */
23 #define FIEMAP_MAX_EXTENTS (UINT_MAX / sizeof(struct fiemap_extent))
26 * vfs_ioctl - call filesystem specific ioctl methods
27 * @filp: open file to invoke ioctl method on
28 * @cmd: ioctl command to execute
29 * @arg: command-specific argument for ioctl
31 * Invokes filesystem specific ->unlocked_ioctl, if one exists; otherwise
34 * Returns 0 on success, -errno on error.
36 long vfs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
40 if (!filp->f_op->unlocked_ioctl)
43 error = filp->f_op->unlocked_ioctl(filp, cmd, arg);
44 if (error == -ENOIOCTLCMD)
50 static int ioctl_fibmap(struct file *filp, int __user *p)
52 struct address_space *mapping = filp->f_mapping;
55 /* do we support this mess? */
56 if (!mapping->a_ops->bmap)
58 if (!capable(CAP_SYS_RAWIO))
60 res = get_user(block, p);
63 res = mapping->a_ops->bmap(mapping, block);
64 return put_user(res, p);
68 * fiemap_fill_next_extent - Fiemap helper function
69 * @fieinfo: Fiemap context passed into ->fiemap
70 * @logical: Extent logical start offset, in bytes
71 * @phys: Extent physical start offset, in bytes
72 * @len: Extent length, in bytes
73 * @flags: FIEMAP_EXTENT flags that describe this extent
75 * Called from file system ->fiemap callback. Will populate extent
76 * info as passed in via arguments and copy to user memory. On
77 * success, extent count on fieinfo is incremented.
79 * Returns 0 on success, -errno on error, 1 if this was the last
80 * extent that will fit in user array.
82 #define SET_UNKNOWN_FLAGS (FIEMAP_EXTENT_DELALLOC)
83 #define SET_NO_UNMOUNTED_IO_FLAGS (FIEMAP_EXTENT_DATA_ENCRYPTED)
84 #define SET_NOT_ALIGNED_FLAGS (FIEMAP_EXTENT_DATA_TAIL|FIEMAP_EXTENT_DATA_INLINE)
85 int fiemap_fill_next_extent(struct fiemap_extent_info *fieinfo, u64 logical,
86 u64 phys, u64 len, u32 flags)
88 struct fiemap_extent extent;
89 struct fiemap_extent __user *dest = fieinfo->fi_extents_start;
91 /* only count the extents */
92 if (fieinfo->fi_extents_max == 0) {
93 fieinfo->fi_extents_mapped++;
94 return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0;
97 if (fieinfo->fi_extents_mapped >= fieinfo->fi_extents_max)
100 if (flags & SET_UNKNOWN_FLAGS)
101 flags |= FIEMAP_EXTENT_UNKNOWN;
102 if (flags & SET_NO_UNMOUNTED_IO_FLAGS)
103 flags |= FIEMAP_EXTENT_ENCODED;
104 if (flags & SET_NOT_ALIGNED_FLAGS)
105 flags |= FIEMAP_EXTENT_NOT_ALIGNED;
107 memset(&extent, 0, sizeof(extent));
108 extent.fe_logical = logical;
109 extent.fe_physical = phys;
110 extent.fe_length = len;
111 extent.fe_flags = flags;
113 dest += fieinfo->fi_extents_mapped;
114 if (copy_to_user(dest, &extent, sizeof(extent)))
117 fieinfo->fi_extents_mapped++;
118 if (fieinfo->fi_extents_mapped == fieinfo->fi_extents_max)
120 return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0;
122 EXPORT_SYMBOL(fiemap_fill_next_extent);
125 * fiemap_check_flags - check validity of requested flags for fiemap
126 * @fieinfo: Fiemap context passed into ->fiemap
127 * @fs_flags: Set of fiemap flags that the file system understands
129 * Called from file system ->fiemap callback. This will compute the
130 * intersection of valid fiemap flags and those that the fs supports. That
131 * value is then compared against the user supplied flags. In case of bad user
132 * flags, the invalid values will be written into the fieinfo structure, and
133 * -EBADR is returned, which tells ioctl_fiemap() to return those values to
134 * userspace. For this reason, a return code of -EBADR should be preserved.
136 * Returns 0 on success, -EBADR on bad flags.
138 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags)
142 incompat_flags = fieinfo->fi_flags & ~(FIEMAP_FLAGS_COMPAT & fs_flags);
143 if (incompat_flags) {
144 fieinfo->fi_flags = incompat_flags;
149 EXPORT_SYMBOL(fiemap_check_flags);
151 static int fiemap_check_ranges(struct super_block *sb,
152 u64 start, u64 len, u64 *new_len)
154 u64 maxbytes = (u64) sb->s_maxbytes;
161 if (start > maxbytes)
165 * Shrink request scope to what the fs can actually handle.
167 if (len > maxbytes || (maxbytes - len) < start)
168 *new_len = maxbytes - start;
173 static int ioctl_fiemap(struct file *filp, unsigned long arg)
175 struct fiemap fiemap;
176 struct fiemap __user *ufiemap = (struct fiemap __user *) arg;
177 struct fiemap_extent_info fieinfo = { 0, };
178 struct inode *inode = file_inode(filp);
179 struct super_block *sb = inode->i_sb;
183 if (!inode->i_op->fiemap)
186 if (copy_from_user(&fiemap, ufiemap, sizeof(fiemap)))
189 if (fiemap.fm_extent_count > FIEMAP_MAX_EXTENTS)
192 error = fiemap_check_ranges(sb, fiemap.fm_start, fiemap.fm_length,
197 fieinfo.fi_flags = fiemap.fm_flags;
198 fieinfo.fi_extents_max = fiemap.fm_extent_count;
199 fieinfo.fi_extents_start = ufiemap->fm_extents;
201 if (fiemap.fm_extent_count != 0 &&
202 !access_ok(VERIFY_WRITE, fieinfo.fi_extents_start,
203 fieinfo.fi_extents_max * sizeof(struct fiemap_extent)))
206 if (fieinfo.fi_flags & FIEMAP_FLAG_SYNC)
207 filemap_write_and_wait(inode->i_mapping);
209 error = inode->i_op->fiemap(inode, &fieinfo, fiemap.fm_start, len);
210 fiemap.fm_flags = fieinfo.fi_flags;
211 fiemap.fm_mapped_extents = fieinfo.fi_extents_mapped;
212 if (copy_to_user(ufiemap, &fiemap, sizeof(fiemap)))
218 static long ioctl_file_clone(struct file *dst_file, unsigned long srcfd,
219 u64 off, u64 olen, u64 destoff)
221 struct fd src_file = fdget(srcfd);
226 ret = vfs_clone_file_range(src_file.file, off, dst_file, destoff, olen);
231 static long ioctl_file_clone_range(struct file *file, void __user *argp)
233 struct file_clone_range args;
235 if (copy_from_user(&args, argp, sizeof(args)))
237 return ioctl_file_clone(file, args.src_fd, args.src_offset,
238 args.src_length, args.dest_offset);
243 static inline sector_t logical_to_blk(struct inode *inode, loff_t offset)
245 return (offset >> inode->i_blkbits);
248 static inline loff_t blk_to_logical(struct inode *inode, sector_t blk)
250 return (blk << inode->i_blkbits);
254 * __generic_block_fiemap - FIEMAP for block based inodes (no locking)
255 * @inode: the inode to map
256 * @fieinfo: the fiemap info struct that will be passed back to userspace
257 * @start: where to start mapping in the inode
258 * @len: how much space to map
259 * @get_block: the fs's get_block function
261 * This does FIEMAP for block based inodes. Basically it will just loop
262 * through get_block until we hit the number of extents we want to map, or we
263 * go past the end of the file and hit a hole.
265 * If it is possible to have data blocks beyond a hole past @inode->i_size, then
266 * please do not use this function, it will stop at the first unmapped block
269 * If you use this function directly, you need to do your own locking. Use
270 * generic_block_fiemap if you want the locking done for you.
273 int __generic_block_fiemap(struct inode *inode,
274 struct fiemap_extent_info *fieinfo, loff_t start,
275 loff_t len, get_block_t *get_block)
277 struct buffer_head map_bh;
278 sector_t start_blk, last_blk;
279 loff_t isize = i_size_read(inode);
280 u64 logical = 0, phys = 0, size = 0;
281 u32 flags = FIEMAP_EXTENT_MERGED;
282 bool past_eof = false, whole_file = false;
285 ret = fiemap_check_flags(fieinfo, FIEMAP_FLAG_SYNC);
290 * Either the i_mutex or other appropriate locking needs to be held
291 * since we expect isize to not change at all through the duration of
300 * Some filesystems can't deal with being asked to map less than
301 * blocksize, so make sure our len is at least block length.
303 if (logical_to_blk(inode, len) == 0)
304 len = blk_to_logical(inode, 1);
306 start_blk = logical_to_blk(inode, start);
307 last_blk = logical_to_blk(inode, start + len - 1);
311 * we set b_size to the total size we want so it will map as
312 * many contiguous blocks as possible at once
314 memset(&map_bh, 0, sizeof(struct buffer_head));
317 ret = get_block(inode, start_blk, &map_bh, 0);
322 if (!buffer_mapped(&map_bh)) {
326 * We want to handle the case where there is an
327 * allocated block at the front of the file, and then
328 * nothing but holes up to the end of the file properly,
329 * to make sure that extent at the front gets properly
330 * marked with FIEMAP_EXTENT_LAST
333 blk_to_logical(inode, start_blk) >= isize)
337 * First hole after going past the EOF, this is our
340 if (past_eof && size) {
341 flags = FIEMAP_EXTENT_MERGED|FIEMAP_EXTENT_LAST;
342 ret = fiemap_fill_next_extent(fieinfo, logical,
346 ret = fiemap_fill_next_extent(fieinfo, logical,
351 /* if we have holes up to/past EOF then we're done */
352 if (start_blk > last_blk || past_eof || ret)
356 * We have gone over the length of what we wanted to
357 * map, and it wasn't the entire file, so add the extent
358 * we got last time and exit.
360 * This is for the case where say we want to map all the
361 * way up to the second to the last block in a file, but
362 * the last block is a hole, making the second to last
363 * block FIEMAP_EXTENT_LAST. In this case we want to
364 * see if there is a hole after the second to last block
365 * so we can mark it properly. If we found data after
366 * we exceeded the length we were requesting, then we
367 * are good to go, just add the extent to the fieinfo
370 if (start_blk > last_blk && !whole_file) {
371 ret = fiemap_fill_next_extent(fieinfo, logical,
378 * if size != 0 then we know we already have an extent
382 ret = fiemap_fill_next_extent(fieinfo, logical,
389 logical = blk_to_logical(inode, start_blk);
390 phys = blk_to_logical(inode, map_bh.b_blocknr);
391 size = map_bh.b_size;
392 flags = FIEMAP_EXTENT_MERGED;
394 start_blk += logical_to_blk(inode, size);
397 * If we are past the EOF, then we need to make sure as
398 * soon as we find a hole that the last extent we found
399 * is marked with FIEMAP_EXTENT_LAST
401 if (!past_eof && logical + size >= isize)
405 if (fatal_signal_pending(current)) {
412 /* If ret is 1 then we just hit the end of the extent array */
418 EXPORT_SYMBOL(__generic_block_fiemap);
421 * generic_block_fiemap - FIEMAP for block based inodes
422 * @inode: The inode to map
423 * @fieinfo: The mapping information
424 * @start: The initial block to map
425 * @len: The length of the extect to attempt to map
426 * @get_block: The block mapping function for the fs
428 * Calls __generic_block_fiemap to map the inode, after taking
429 * the inode's mutex lock.
432 int generic_block_fiemap(struct inode *inode,
433 struct fiemap_extent_info *fieinfo, u64 start,
434 u64 len, get_block_t *get_block)
438 ret = __generic_block_fiemap(inode, fieinfo, start, len, get_block);
442 EXPORT_SYMBOL(generic_block_fiemap);
444 #endif /* CONFIG_BLOCK */
447 * This provides compatibility with legacy XFS pre-allocation ioctls
448 * which predate the fallocate syscall.
450 * Only the l_start, l_len and l_whence fields of the 'struct space_resv'
451 * are used here, rest are ignored.
453 int ioctl_preallocate(struct file *filp, void __user *argp)
455 struct inode *inode = file_inode(filp);
456 struct space_resv sr;
458 if (copy_from_user(&sr, argp, sizeof(sr)))
461 switch (sr.l_whence) {
465 sr.l_start += filp->f_pos;
468 sr.l_start += i_size_read(inode);
474 return vfs_fallocate(filp, FALLOC_FL_KEEP_SIZE, sr.l_start, sr.l_len);
477 static int file_ioctl(struct file *filp, unsigned int cmd,
480 struct inode *inode = file_inode(filp);
481 int __user *p = (int __user *)arg;
485 return ioctl_fibmap(filp, p);
487 return put_user(i_size_read(inode) - filp->f_pos, p);
489 case FS_IOC_RESVSP64:
490 return ioctl_preallocate(filp, p);
493 return vfs_ioctl(filp, cmd, arg);
496 static int ioctl_fionbio(struct file *filp, int __user *argp)
501 error = get_user(on, argp);
506 /* SunOS compatibility item. */
507 if (O_NONBLOCK != O_NDELAY)
510 spin_lock(&filp->f_lock);
512 filp->f_flags |= flag;
514 filp->f_flags &= ~flag;
515 spin_unlock(&filp->f_lock);
519 static int ioctl_fioasync(unsigned int fd, struct file *filp,
525 error = get_user(on, argp);
528 flag = on ? FASYNC : 0;
530 /* Did FASYNC state change ? */
531 if ((flag ^ filp->f_flags) & FASYNC) {
532 if (filp->f_op->fasync)
533 /* fasync() adjusts filp->f_flags */
534 error = filp->f_op->fasync(fd, filp, on);
538 return error < 0 ? error : 0;
541 static int ioctl_fsfreeze(struct file *filp)
543 struct super_block *sb = file_inode(filp)->i_sb;
545 if (!capable(CAP_SYS_ADMIN))
548 /* If filesystem doesn't support freeze feature, return. */
549 if (sb->s_op->freeze_fs == NULL && sb->s_op->freeze_super == NULL)
553 if (sb->s_op->freeze_super)
554 return sb->s_op->freeze_super(sb);
555 return freeze_super(sb);
558 static int ioctl_fsthaw(struct file *filp)
560 struct super_block *sb = file_inode(filp)->i_sb;
562 if (!capable(CAP_SYS_ADMIN))
566 if (sb->s_op->thaw_super)
567 return sb->s_op->thaw_super(sb);
568 return thaw_super(sb);
571 static long ioctl_file_dedupe_range(struct file *file, void __user *arg)
573 struct file_dedupe_range __user *argp = arg;
574 struct file_dedupe_range *same = NULL;
579 if (get_user(count, &argp->dest_count)) {
584 size = offsetof(struct file_dedupe_range __user, info[count]);
586 same = memdup_user(argp, size);
593 ret = vfs_dedupe_file_range(file, same);
597 ret = copy_to_user(argp, same, size);
607 * When you add any new common ioctls to the switches above and below
608 * please update compat_sys_ioctl() too.
610 * do_vfs_ioctl() is not for drivers and not intended to be EXPORT_SYMBOL()'d.
611 * It's just a simple helper for sys_ioctl and compat_sys_ioctl.
613 int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
617 int __user *argp = (int __user *)arg;
618 struct inode *inode = file_inode(filp);
622 set_close_on_exec(fd, 1);
626 set_close_on_exec(fd, 0);
630 error = ioctl_fionbio(filp, argp);
634 error = ioctl_fioasync(fd, filp, argp);
638 if (S_ISDIR(inode->i_mode) || S_ISREG(inode->i_mode) ||
639 S_ISLNK(inode->i_mode)) {
640 loff_t res = inode_get_bytes(inode);
641 error = copy_to_user(argp, &res, sizeof(res)) ?
648 error = ioctl_fsfreeze(filp);
652 error = ioctl_fsthaw(filp);
656 return ioctl_fiemap(filp, arg);
659 return put_user(inode->i_sb->s_blocksize, argp);
662 return ioctl_file_clone(filp, arg, 0, 0, 0);
665 return ioctl_file_clone_range(filp, argp);
668 return ioctl_file_dedupe_range(filp, argp);
671 if (S_ISREG(inode->i_mode))
672 error = file_ioctl(filp, cmd, arg);
674 error = vfs_ioctl(filp, cmd, arg);
680 SYSCALL_DEFINE3(ioctl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
683 struct fd f = fdget(fd);
687 error = security_file_ioctl(f.file, cmd, arg);
689 error = do_vfs_ioctl(f.file, fd, cmd, arg);