4 * Copyright (C) 1991, 1992 Linus Torvalds
7 #include <linux/syscalls.h>
9 #include <linux/smp_lock.h>
10 #include <linux/capability.h>
11 #include <linux/file.h>
13 #include <linux/security.h>
14 #include <linux/module.h>
15 #include <linux/uaccess.h>
16 #include <linux/writeback.h>
17 #include <linux/buffer_head.h>
19 #include <asm/ioctls.h>
21 /* So that the fiemap access checks can't overflow on 32 bit machines. */
22 #define FIEMAP_MAX_EXTENTS (UINT_MAX / sizeof(struct fiemap_extent))
25 * vfs_ioctl - call filesystem specific ioctl methods
26 * @filp: open file to invoke ioctl method on
27 * @cmd: ioctl command to execute
28 * @arg: command-specific argument for ioctl
30 * Invokes filesystem specific ->unlocked_ioctl, if one exists; otherwise
31 * invokes filesystem specific ->ioctl method. If neither method exists,
34 * Returns 0 on success, -errno on error.
36 static long vfs_ioctl(struct file *filp, unsigned int cmd,
44 if (filp->f_op->unlocked_ioctl) {
45 error = filp->f_op->unlocked_ioctl(filp, cmd, arg);
46 if (error == -ENOIOCTLCMD)
49 } else if (filp->f_op->ioctl) {
51 error = filp->f_op->ioctl(filp->f_path.dentry->d_inode,
60 static int ioctl_fibmap(struct file *filp, int __user *p)
62 struct address_space *mapping = filp->f_mapping;
65 /* do we support this mess? */
66 if (!mapping->a_ops->bmap)
68 if (!capable(CAP_SYS_RAWIO))
70 res = get_user(block, p);
73 res = mapping->a_ops->bmap(mapping, block);
74 return put_user(res, p);
78 * fiemap_fill_next_extent - Fiemap helper function
79 * @fieinfo: Fiemap context passed into ->fiemap
80 * @logical: Extent logical start offset, in bytes
81 * @phys: Extent physical start offset, in bytes
82 * @len: Extent length, in bytes
83 * @flags: FIEMAP_EXTENT flags that describe this extent
85 * Called from file system ->fiemap callback. Will populate extent
86 * info as passed in via arguments and copy to user memory. On
87 * success, extent count on fieinfo is incremented.
89 * Returns 0 on success, -errno on error, 1 if this was the last
90 * extent that will fit in user array.
92 #define SET_UNKNOWN_FLAGS (FIEMAP_EXTENT_DELALLOC)
93 #define SET_NO_UNMOUNTED_IO_FLAGS (FIEMAP_EXTENT_DATA_ENCRYPTED)
94 #define SET_NOT_ALIGNED_FLAGS (FIEMAP_EXTENT_DATA_TAIL|FIEMAP_EXTENT_DATA_INLINE)
95 int fiemap_fill_next_extent(struct fiemap_extent_info *fieinfo, u64 logical,
96 u64 phys, u64 len, u32 flags)
98 struct fiemap_extent extent;
99 struct fiemap_extent *dest = fieinfo->fi_extents_start;
101 /* only count the extents */
102 if (fieinfo->fi_extents_max == 0) {
103 fieinfo->fi_extents_mapped++;
104 return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0;
107 if (fieinfo->fi_extents_mapped >= fieinfo->fi_extents_max)
110 if (flags & SET_UNKNOWN_FLAGS)
111 flags |= FIEMAP_EXTENT_UNKNOWN;
112 if (flags & SET_NO_UNMOUNTED_IO_FLAGS)
113 flags |= FIEMAP_EXTENT_ENCODED;
114 if (flags & SET_NOT_ALIGNED_FLAGS)
115 flags |= FIEMAP_EXTENT_NOT_ALIGNED;
117 memset(&extent, 0, sizeof(extent));
118 extent.fe_logical = logical;
119 extent.fe_physical = phys;
120 extent.fe_length = len;
121 extent.fe_flags = flags;
123 dest += fieinfo->fi_extents_mapped;
124 if (copy_to_user(dest, &extent, sizeof(extent)))
127 fieinfo->fi_extents_mapped++;
128 if (fieinfo->fi_extents_mapped == fieinfo->fi_extents_max)
130 return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0;
132 EXPORT_SYMBOL(fiemap_fill_next_extent);
135 * fiemap_check_flags - check validity of requested flags for fiemap
136 * @fieinfo: Fiemap context passed into ->fiemap
137 * @fs_flags: Set of fiemap flags that the file system understands
139 * Called from file system ->fiemap callback. This will compute the
140 * intersection of valid fiemap flags and those that the fs supports. That
141 * value is then compared against the user supplied flags. In case of bad user
142 * flags, the invalid values will be written into the fieinfo structure, and
143 * -EBADR is returned, which tells ioctl_fiemap() to return those values to
144 * userspace. For this reason, a return code of -EBADR should be preserved.
146 * Returns 0 on success, -EBADR on bad flags.
148 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags)
152 incompat_flags = fieinfo->fi_flags & ~(FIEMAP_FLAGS_COMPAT & fs_flags);
153 if (incompat_flags) {
154 fieinfo->fi_flags = incompat_flags;
159 EXPORT_SYMBOL(fiemap_check_flags);
161 static int fiemap_check_ranges(struct super_block *sb,
162 u64 start, u64 len, u64 *new_len)
169 if (start > sb->s_maxbytes)
173 * Shrink request scope to what the fs can actually handle.
175 if ((len > sb->s_maxbytes) ||
176 (sb->s_maxbytes - len) < start)
177 *new_len = sb->s_maxbytes - start;
182 static int ioctl_fiemap(struct file *filp, unsigned long arg)
184 struct fiemap fiemap;
185 struct fiemap_extent_info fieinfo = { 0, };
186 struct inode *inode = filp->f_path.dentry->d_inode;
187 struct super_block *sb = inode->i_sb;
191 if (!inode->i_op->fiemap)
194 if (copy_from_user(&fiemap, (struct fiemap __user *)arg,
195 sizeof(struct fiemap)))
198 if (fiemap.fm_extent_count > FIEMAP_MAX_EXTENTS)
201 error = fiemap_check_ranges(sb, fiemap.fm_start, fiemap.fm_length,
206 fieinfo.fi_flags = fiemap.fm_flags;
207 fieinfo.fi_extents_max = fiemap.fm_extent_count;
208 fieinfo.fi_extents_start = (struct fiemap_extent *)(arg + sizeof(fiemap));
210 if (fiemap.fm_extent_count != 0 &&
211 !access_ok(VERIFY_WRITE, fieinfo.fi_extents_start,
212 fieinfo.fi_extents_max * sizeof(struct fiemap_extent)))
215 if (fieinfo.fi_flags & FIEMAP_FLAG_SYNC)
216 filemap_write_and_wait(inode->i_mapping);
218 error = inode->i_op->fiemap(inode, &fieinfo, fiemap.fm_start, len);
219 fiemap.fm_flags = fieinfo.fi_flags;
220 fiemap.fm_mapped_extents = fieinfo.fi_extents_mapped;
221 if (copy_to_user((char *)arg, &fiemap, sizeof(fiemap)))
229 #define blk_to_logical(inode, blk) (blk << (inode)->i_blkbits)
230 #define logical_to_blk(inode, offset) (offset >> (inode)->i_blkbits);
233 * __generic_block_fiemap - FIEMAP for block based inodes (no locking)
234 * @inode - the inode to map
235 * @arg - the pointer to userspace where we copy everything to
236 * @get_block - the fs's get_block function
238 * This does FIEMAP for block based inodes. Basically it will just loop
239 * through get_block until we hit the number of extents we want to map, or we
240 * go past the end of the file and hit a hole.
242 * If it is possible to have data blocks beyond a hole past @inode->i_size, then
243 * please do not use this function, it will stop at the first unmapped block
246 * If you use this function directly, you need to do your own locking. Use
247 * generic_block_fiemap if you want the locking done for you.
250 int __generic_block_fiemap(struct inode *inode,
251 struct fiemap_extent_info *fieinfo, u64 start,
252 u64 len, get_block_t *get_block)
254 struct buffer_head tmp;
255 unsigned int start_blk;
256 long long length = 0, map_len = 0;
257 u64 logical = 0, phys = 0, size = 0;
258 u32 flags = FIEMAP_EXTENT_MERGED;
259 int ret = 0, past_eof = 0, whole_file = 0;
261 if ((ret = fiemap_check_flags(fieinfo, FIEMAP_FLAG_SYNC)))
264 start_blk = logical_to_blk(inode, start);
266 length = (long long)min_t(u64, len, i_size_read(inode));
274 * we set b_size to the total size we want so it will map as
275 * many contiguous blocks as possible at once
277 memset(&tmp, 0, sizeof(struct buffer_head));
278 tmp.b_size = map_len;
280 ret = get_block(inode, start_blk, &tmp, 0);
285 if (!buffer_mapped(&tmp)) {
286 length -= blk_to_logical(inode, 1);
290 * we want to handle the case where there is an
291 * allocated block at the front of the file, and then
292 * nothing but holes up to the end of the file properly,
293 * to make sure that extent at the front gets properly
294 * marked with FIEMAP_EXTENT_LAST
297 blk_to_logical(inode, start_blk) >=
298 blk_to_logical(inode, 0)+i_size_read(inode))
302 * first hole after going past the EOF, this is our
305 if (past_eof && size) {
306 flags = FIEMAP_EXTENT_MERGED|FIEMAP_EXTENT_LAST;
307 ret = fiemap_fill_next_extent(fieinfo, logical,
313 /* if we have holes up to/past EOF then we're done */
314 if (length <= 0 || past_eof)
318 * we have gone over the length of what we wanted to
319 * map, and it wasn't the entire file, so add the extent
320 * we got last time and exit.
322 * This is for the case where say we want to map all the
323 * way up to the second to the last block in a file, but
324 * the last block is a hole, making the second to last
325 * block FIEMAP_EXTENT_LAST. In this case we want to
326 * see if there is a hole after the second to last block
327 * so we can mark it properly. If we found data after
328 * we exceeded the length we were requesting, then we
329 * are good to go, just add the extent to the fieinfo
332 if (length <= 0 && !whole_file) {
333 ret = fiemap_fill_next_extent(fieinfo, logical,
340 * if size != 0 then we know we already have an extent
344 ret = fiemap_fill_next_extent(fieinfo, logical,
351 logical = blk_to_logical(inode, start_blk);
352 phys = blk_to_logical(inode, tmp.b_blocknr);
354 flags = FIEMAP_EXTENT_MERGED;
356 length -= tmp.b_size;
357 start_blk += logical_to_blk(inode, size);
360 * If we are past the EOF, then we need to make sure as
361 * soon as we find a hole that the last extent we found
362 * is marked with FIEMAP_EXTENT_LAST
366 blk_to_logical(inode, 0)+i_size_read(inode))
372 /* if ret is 1 then we just hit the end of the extent array */
378 EXPORT_SYMBOL(__generic_block_fiemap);
381 * generic_block_fiemap - FIEMAP for block based inodes
382 * @inode: The inode to map
383 * @fieinfo: The mapping information
384 * @start: The initial block to map
385 * @len: The length of the extect to attempt to map
386 * @get_block: The block mapping function for the fs
388 * Calls __generic_block_fiemap to map the inode, after taking
389 * the inode's mutex lock.
392 int generic_block_fiemap(struct inode *inode,
393 struct fiemap_extent_info *fieinfo, u64 start,
394 u64 len, get_block_t *get_block)
397 mutex_lock(&inode->i_mutex);
398 ret = __generic_block_fiemap(inode, fieinfo, start, len, get_block);
399 mutex_unlock(&inode->i_mutex);
402 EXPORT_SYMBOL(generic_block_fiemap);
404 #endif /* CONFIG_BLOCK */
406 static int file_ioctl(struct file *filp, unsigned int cmd,
409 struct inode *inode = filp->f_path.dentry->d_inode;
410 int __user *p = (int __user *)arg;
414 return ioctl_fibmap(filp, p);
416 return put_user(i_size_read(inode) - filp->f_pos, p);
419 return vfs_ioctl(filp, cmd, arg);
422 static int ioctl_fionbio(struct file *filp, int __user *argp)
427 error = get_user(on, argp);
432 /* SunOS compatibility item. */
433 if (O_NONBLOCK != O_NDELAY)
436 spin_lock(&filp->f_lock);
438 filp->f_flags |= flag;
440 filp->f_flags &= ~flag;
441 spin_unlock(&filp->f_lock);
445 static int ioctl_fioasync(unsigned int fd, struct file *filp,
451 error = get_user(on, argp);
454 flag = on ? FASYNC : 0;
456 /* Did FASYNC state change ? */
457 if ((flag ^ filp->f_flags) & FASYNC) {
458 if (filp->f_op && filp->f_op->fasync)
459 /* fasync() adjusts filp->f_flags */
460 error = filp->f_op->fasync(fd, filp, on);
464 return error < 0 ? error : 0;
467 static int ioctl_fsfreeze(struct file *filp)
469 struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
471 if (!capable(CAP_SYS_ADMIN))
474 /* If filesystem doesn't support freeze feature, return. */
475 if (sb->s_op->freeze_fs == NULL)
478 /* If a blockdevice-backed filesystem isn't specified, return. */
479 if (sb->s_bdev == NULL)
483 sb = freeze_bdev(sb->s_bdev);
489 static int ioctl_fsthaw(struct file *filp)
491 struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
493 if (!capable(CAP_SYS_ADMIN))
496 /* If a blockdevice-backed filesystem isn't specified, return EINVAL. */
497 if (sb->s_bdev == NULL)
501 return thaw_bdev(sb->s_bdev, sb);
505 * When you add any new common ioctls to the switches above and below
506 * please update compat_sys_ioctl() too.
508 * do_vfs_ioctl() is not for drivers and not intended to be EXPORT_SYMBOL()'d.
509 * It's just a simple helper for sys_ioctl and compat_sys_ioctl.
511 int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
515 int __user *argp = (int __user *)arg;
519 set_close_on_exec(fd, 1);
523 set_close_on_exec(fd, 0);
527 error = ioctl_fionbio(filp, argp);
531 error = ioctl_fioasync(fd, filp, argp);
535 if (S_ISDIR(filp->f_path.dentry->d_inode->i_mode) ||
536 S_ISREG(filp->f_path.dentry->d_inode->i_mode) ||
537 S_ISLNK(filp->f_path.dentry->d_inode->i_mode)) {
539 inode_get_bytes(filp->f_path.dentry->d_inode);
540 error = copy_to_user((loff_t __user *)arg, &res,
541 sizeof(res)) ? -EFAULT : 0;
547 error = ioctl_fsfreeze(filp);
551 error = ioctl_fsthaw(filp);
555 return ioctl_fiemap(filp, arg);
559 struct inode *inode = filp->f_path.dentry->d_inode;
560 int __user *p = (int __user *)arg;
561 return put_user(inode->i_sb->s_blocksize, p);
565 if (S_ISREG(filp->f_path.dentry->d_inode->i_mode))
566 error = file_ioctl(filp, cmd, arg);
568 error = vfs_ioctl(filp, cmd, arg);
574 SYSCALL_DEFINE3(ioctl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
580 filp = fget_light(fd, &fput_needed);
584 error = security_file_ioctl(filp, cmd, arg);
588 error = do_vfs_ioctl(filp, fd, cmd, arg);
590 fput_light(filp, fput_needed);