3 * Library for filesystems writers.
6 #include <linux/module.h>
7 #include <linux/pagemap.h>
8 #include <linux/mount.h>
10 #include <linux/mutex.h>
11 #include <linux/exportfs.h>
13 #include <asm/uaccess.h>
15 int simple_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
18 struct inode
*inode
= dentry
->d_inode
;
19 generic_fillattr(inode
, stat
);
20 stat
->blocks
= inode
->i_mapping
->nrpages
<< (PAGE_CACHE_SHIFT
- 9);
24 int simple_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
26 buf
->f_type
= dentry
->d_sb
->s_magic
;
27 buf
->f_bsize
= PAGE_CACHE_SIZE
;
28 buf
->f_namelen
= NAME_MAX
;
33 * Retaining negative dentries for an in-memory filesystem just wastes
34 * memory and lookup time: arrange for them to be deleted immediately.
36 static int simple_delete_dentry(struct dentry
*dentry
)
42 * Lookup the data. This is trivial - if the dentry didn't already
43 * exist, we know it is negative. Set d_op to delete negative dentries.
45 struct dentry
*simple_lookup(struct inode
*dir
, struct dentry
*dentry
, struct nameidata
*nd
)
47 static struct dentry_operations simple_dentry_operations
= {
48 .d_delete
= simple_delete_dentry
,
51 if (dentry
->d_name
.len
> NAME_MAX
)
52 return ERR_PTR(-ENAMETOOLONG
);
53 dentry
->d_op
= &simple_dentry_operations
;
58 int simple_sync_file(struct file
* file
, struct dentry
*dentry
, int datasync
)
63 int dcache_dir_open(struct inode
*inode
, struct file
*file
)
65 static struct qstr cursor_name
= {.len
= 1, .name
= "."};
67 file
->private_data
= d_alloc(file
->f_path
.dentry
, &cursor_name
);
69 return file
->private_data
? 0 : -ENOMEM
;
72 int dcache_dir_close(struct inode
*inode
, struct file
*file
)
74 dput(file
->private_data
);
78 loff_t
dcache_dir_lseek(struct file
*file
, loff_t offset
, int origin
)
80 mutex_lock(&file
->f_path
.dentry
->d_inode
->i_mutex
);
83 offset
+= file
->f_pos
;
88 mutex_unlock(&file
->f_path
.dentry
->d_inode
->i_mutex
);
91 if (offset
!= file
->f_pos
) {
93 if (file
->f_pos
>= 2) {
95 struct dentry
*cursor
= file
->private_data
;
96 loff_t n
= file
->f_pos
- 2;
98 spin_lock(&dcache_lock
);
99 list_del(&cursor
->d_u
.d_child
);
100 p
= file
->f_path
.dentry
->d_subdirs
.next
;
101 while (n
&& p
!= &file
->f_path
.dentry
->d_subdirs
) {
103 next
= list_entry(p
, struct dentry
, d_u
.d_child
);
104 if (!d_unhashed(next
) && next
->d_inode
)
108 list_add_tail(&cursor
->d_u
.d_child
, p
);
109 spin_unlock(&dcache_lock
);
112 mutex_unlock(&file
->f_path
.dentry
->d_inode
->i_mutex
);
116 /* Relationship between i_mode and the DT_xxx types */
117 static inline unsigned char dt_type(struct inode
*inode
)
119 return (inode
->i_mode
>> 12) & 15;
123 * Directory is locked and all positive dentries in it are safe, since
124 * for ramfs-type trees they can't go away without unlink() or rmdir(),
125 * both impossible due to the lock on directory.
128 int dcache_readdir(struct file
* filp
, void * dirent
, filldir_t filldir
)
130 struct dentry
*dentry
= filp
->f_path
.dentry
;
131 struct dentry
*cursor
= filp
->private_data
;
132 struct list_head
*p
, *q
= &cursor
->d_u
.d_child
;
138 ino
= dentry
->d_inode
->i_ino
;
139 if (filldir(dirent
, ".", 1, i
, ino
, DT_DIR
) < 0)
145 ino
= parent_ino(dentry
);
146 if (filldir(dirent
, "..", 2, i
, ino
, DT_DIR
) < 0)
152 spin_lock(&dcache_lock
);
153 if (filp
->f_pos
== 2)
154 list_move(q
, &dentry
->d_subdirs
);
156 for (p
=q
->next
; p
!= &dentry
->d_subdirs
; p
=p
->next
) {
158 next
= list_entry(p
, struct dentry
, d_u
.d_child
);
159 if (d_unhashed(next
) || !next
->d_inode
)
162 spin_unlock(&dcache_lock
);
163 if (filldir(dirent
, next
->d_name
.name
,
164 next
->d_name
.len
, filp
->f_pos
,
165 next
->d_inode
->i_ino
,
166 dt_type(next
->d_inode
)) < 0)
168 spin_lock(&dcache_lock
);
169 /* next is still alive */
174 spin_unlock(&dcache_lock
);
179 ssize_t
generic_read_dir(struct file
*filp
, char __user
*buf
, size_t siz
, loff_t
*ppos
)
184 const struct file_operations simple_dir_operations
= {
185 .open
= dcache_dir_open
,
186 .release
= dcache_dir_close
,
187 .llseek
= dcache_dir_lseek
,
188 .read
= generic_read_dir
,
189 .readdir
= dcache_readdir
,
190 .fsync
= simple_sync_file
,
193 const struct inode_operations simple_dir_inode_operations
= {
194 .lookup
= simple_lookup
,
197 static const struct super_operations simple_super_operations
= {
198 .statfs
= simple_statfs
,
202 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
203 * will never be mountable)
205 int get_sb_pseudo(struct file_system_type
*fs_type
, char *name
,
206 const struct super_operations
*ops
, unsigned long magic
,
207 struct vfsmount
*mnt
)
209 struct super_block
*s
= sget(fs_type
, NULL
, set_anon_super
, NULL
);
210 struct dentry
*dentry
;
212 struct qstr d_name
= {.name
= name
, .len
= strlen(name
)};
217 s
->s_flags
= MS_NOUSER
;
218 s
->s_maxbytes
= ~0ULL;
219 s
->s_blocksize
= PAGE_SIZE
;
220 s
->s_blocksize_bits
= PAGE_SHIFT
;
222 s
->s_op
= ops
? ops
: &simple_super_operations
;
228 * since this is the first inode, make it number 1. New inodes created
229 * after this must take care not to collide with it (by passing
230 * max_reserved of 1 to iunique).
233 root
->i_mode
= S_IFDIR
| S_IRUSR
| S_IWUSR
;
234 root
->i_atime
= root
->i_mtime
= root
->i_ctime
= CURRENT_TIME
;
235 dentry
= d_alloc(NULL
, &d_name
);
241 dentry
->d_parent
= dentry
;
242 d_instantiate(dentry
, root
);
244 s
->s_flags
|= MS_ACTIVE
;
245 return simple_set_mnt(mnt
, s
);
248 up_write(&s
->s_umount
);
253 int simple_link(struct dentry
*old_dentry
, struct inode
*dir
, struct dentry
*dentry
)
255 struct inode
*inode
= old_dentry
->d_inode
;
257 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= CURRENT_TIME
;
259 atomic_inc(&inode
->i_count
);
261 d_instantiate(dentry
, inode
);
265 static inline int simple_positive(struct dentry
*dentry
)
267 return dentry
->d_inode
&& !d_unhashed(dentry
);
270 int simple_empty(struct dentry
*dentry
)
272 struct dentry
*child
;
275 spin_lock(&dcache_lock
);
276 list_for_each_entry(child
, &dentry
->d_subdirs
, d_u
.d_child
)
277 if (simple_positive(child
))
281 spin_unlock(&dcache_lock
);
285 int simple_unlink(struct inode
*dir
, struct dentry
*dentry
)
287 struct inode
*inode
= dentry
->d_inode
;
289 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= CURRENT_TIME
;
295 int simple_rmdir(struct inode
*dir
, struct dentry
*dentry
)
297 if (!simple_empty(dentry
))
300 drop_nlink(dentry
->d_inode
);
301 simple_unlink(dir
, dentry
);
306 int simple_rename(struct inode
*old_dir
, struct dentry
*old_dentry
,
307 struct inode
*new_dir
, struct dentry
*new_dentry
)
309 struct inode
*inode
= old_dentry
->d_inode
;
310 int they_are_dirs
= S_ISDIR(old_dentry
->d_inode
->i_mode
);
312 if (!simple_empty(new_dentry
))
315 if (new_dentry
->d_inode
) {
316 simple_unlink(new_dir
, new_dentry
);
319 } else if (they_are_dirs
) {
324 old_dir
->i_ctime
= old_dir
->i_mtime
= new_dir
->i_ctime
=
325 new_dir
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
330 int simple_readpage(struct file
*file
, struct page
*page
)
332 clear_highpage(page
);
333 flush_dcache_page(page
);
334 SetPageUptodate(page
);
339 int simple_prepare_write(struct file
*file
, struct page
*page
,
340 unsigned from
, unsigned to
)
342 if (!PageUptodate(page
)) {
343 if (to
- from
!= PAGE_CACHE_SIZE
)
344 zero_user_segments(page
,
346 to
, PAGE_CACHE_SIZE
);
351 int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
352 loff_t pos
, unsigned len
, unsigned flags
,
353 struct page
**pagep
, void **fsdata
)
359 index
= pos
>> PAGE_CACHE_SHIFT
;
360 from
= pos
& (PAGE_CACHE_SIZE
- 1);
362 page
= grab_cache_page_write_begin(mapping
, index
, flags
);
368 return simple_prepare_write(file
, page
, from
, from
+len
);
371 static int simple_commit_write(struct file
*file
, struct page
*page
,
372 unsigned from
, unsigned to
)
374 struct inode
*inode
= page
->mapping
->host
;
375 loff_t pos
= ((loff_t
)page
->index
<< PAGE_CACHE_SHIFT
) + to
;
377 if (!PageUptodate(page
))
378 SetPageUptodate(page
);
380 * No need to use i_size_read() here, the i_size
381 * cannot change under us because we hold the i_mutex.
383 if (pos
> inode
->i_size
)
384 i_size_write(inode
, pos
);
385 set_page_dirty(page
);
389 int simple_write_end(struct file
*file
, struct address_space
*mapping
,
390 loff_t pos
, unsigned len
, unsigned copied
,
391 struct page
*page
, void *fsdata
)
393 unsigned from
= pos
& (PAGE_CACHE_SIZE
- 1);
395 /* zero the stale part of the page if we did a short copy */
397 void *kaddr
= kmap_atomic(page
, KM_USER0
);
398 memset(kaddr
+ from
+ copied
, 0, len
- copied
);
399 flush_dcache_page(page
);
400 kunmap_atomic(kaddr
, KM_USER0
);
403 simple_commit_write(file
, page
, from
, from
+copied
);
406 page_cache_release(page
);
412 * the inodes created here are not hashed. If you use iunique to generate
413 * unique inode values later for this filesystem, then you must take care
414 * to pass it an appropriate max_reserved value to avoid collisions.
416 int simple_fill_super(struct super_block
*s
, int magic
, struct tree_descr
*files
)
420 struct dentry
*dentry
;
423 s
->s_blocksize
= PAGE_CACHE_SIZE
;
424 s
->s_blocksize_bits
= PAGE_CACHE_SHIFT
;
426 s
->s_op
= &simple_super_operations
;
429 inode
= new_inode(s
);
433 * because the root inode is 1, the files array must not contain an
437 inode
->i_mode
= S_IFDIR
| 0755;
438 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
439 inode
->i_op
= &simple_dir_inode_operations
;
440 inode
->i_fop
= &simple_dir_operations
;
442 root
= d_alloc_root(inode
);
447 for (i
= 0; !files
->name
|| files
->name
[0]; i
++, files
++) {
451 /* warn if it tries to conflict with the root inode */
452 if (unlikely(i
== 1))
453 printk(KERN_WARNING
"%s: %s passed in a files array"
454 "with an index of 1!\n", __func__
,
457 dentry
= d_alloc_name(root
, files
->name
);
460 inode
= new_inode(s
);
463 inode
->i_mode
= S_IFREG
| files
->mode
;
464 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
465 inode
->i_fop
= files
->ops
;
467 d_add(dentry
, inode
);
477 static DEFINE_SPINLOCK(pin_fs_lock
);
479 int simple_pin_fs(struct file_system_type
*type
, struct vfsmount
**mount
, int *count
)
481 struct vfsmount
*mnt
= NULL
;
482 spin_lock(&pin_fs_lock
);
483 if (unlikely(!*mount
)) {
484 spin_unlock(&pin_fs_lock
);
485 mnt
= vfs_kern_mount(type
, 0, type
->name
, NULL
);
488 spin_lock(&pin_fs_lock
);
494 spin_unlock(&pin_fs_lock
);
499 void simple_release_fs(struct vfsmount
**mount
, int *count
)
501 struct vfsmount
*mnt
;
502 spin_lock(&pin_fs_lock
);
506 spin_unlock(&pin_fs_lock
);
511 * simple_read_from_buffer - copy data from the buffer to user space
512 * @to: the user space buffer to read to
513 * @count: the maximum number of bytes to read
514 * @ppos: the current position in the buffer
515 * @from: the buffer to read from
516 * @available: the size of the buffer
518 * The simple_read_from_buffer() function reads up to @count bytes from the
519 * buffer @from at offset @ppos into the user space address starting at @to.
521 * On success, the number of bytes read is returned and the offset @ppos is
522 * advanced by this number, or negative value is returned on error.
524 ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
, loff_t
*ppos
,
525 const void *from
, size_t available
)
530 if (pos
>= available
)
532 if (count
> available
- pos
)
533 count
= available
- pos
;
534 if (copy_to_user(to
, from
+ pos
, count
))
541 * memory_read_from_buffer - copy data from the buffer
542 * @to: the kernel space buffer to read to
543 * @count: the maximum number of bytes to read
544 * @ppos: the current position in the buffer
545 * @from: the buffer to read from
546 * @available: the size of the buffer
548 * The memory_read_from_buffer() function reads up to @count bytes from the
549 * buffer @from at offset @ppos into the kernel space address starting at @to.
551 * On success, the number of bytes read is returned and the offset @ppos is
552 * advanced by this number, or negative value is returned on error.
554 ssize_t
memory_read_from_buffer(void *to
, size_t count
, loff_t
*ppos
,
555 const void *from
, size_t available
)
561 if (pos
>= available
)
563 if (count
> available
- pos
)
564 count
= available
- pos
;
565 memcpy(to
, from
+ pos
, count
);
572 * Transaction based IO.
573 * The file expects a single write which triggers the transaction, and then
574 * possibly a read which collects the result - which is stored in a
577 char *simple_transaction_get(struct file
*file
, const char __user
*buf
, size_t size
)
579 struct simple_transaction_argresp
*ar
;
580 static DEFINE_SPINLOCK(simple_transaction_lock
);
582 if (size
> SIMPLE_TRANSACTION_LIMIT
- 1)
583 return ERR_PTR(-EFBIG
);
585 ar
= (struct simple_transaction_argresp
*)get_zeroed_page(GFP_KERNEL
);
587 return ERR_PTR(-ENOMEM
);
589 spin_lock(&simple_transaction_lock
);
591 /* only one write allowed per open */
592 if (file
->private_data
) {
593 spin_unlock(&simple_transaction_lock
);
594 free_page((unsigned long)ar
);
595 return ERR_PTR(-EBUSY
);
598 file
->private_data
= ar
;
600 spin_unlock(&simple_transaction_lock
);
602 if (copy_from_user(ar
->data
, buf
, size
))
603 return ERR_PTR(-EFAULT
);
608 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
, size_t size
, loff_t
*pos
)
610 struct simple_transaction_argresp
*ar
= file
->private_data
;
614 return simple_read_from_buffer(buf
, size
, pos
, ar
->data
, ar
->size
);
617 int simple_transaction_release(struct inode
*inode
, struct file
*file
)
619 free_page((unsigned long)file
->private_data
);
623 /* Simple attribute files */
626 int (*get
)(void *, u64
*);
627 int (*set
)(void *, u64
);
628 char get_buf
[24]; /* enough to store a u64 and "\n\0" */
631 const char *fmt
; /* format for read operation */
632 struct mutex mutex
; /* protects access to these buffers */
635 /* simple_attr_open is called by an actual attribute open file operation
636 * to set the attribute specific access operations. */
637 int simple_attr_open(struct inode
*inode
, struct file
*file
,
638 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
641 struct simple_attr
*attr
;
643 attr
= kmalloc(sizeof(*attr
), GFP_KERNEL
);
649 attr
->data
= inode
->i_private
;
651 mutex_init(&attr
->mutex
);
653 file
->private_data
= attr
;
655 return nonseekable_open(inode
, file
);
658 int simple_attr_release(struct inode
*inode
, struct file
*file
)
660 kfree(file
->private_data
);
664 /* read from the buffer that is filled with the get function */
665 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
666 size_t len
, loff_t
*ppos
)
668 struct simple_attr
*attr
;
672 attr
= file
->private_data
;
677 ret
= mutex_lock_interruptible(&attr
->mutex
);
681 if (*ppos
) { /* continued read */
682 size
= strlen(attr
->get_buf
);
683 } else { /* first read */
685 ret
= attr
->get(attr
->data
, &val
);
689 size
= scnprintf(attr
->get_buf
, sizeof(attr
->get_buf
),
690 attr
->fmt
, (unsigned long long)val
);
693 ret
= simple_read_from_buffer(buf
, len
, ppos
, attr
->get_buf
, size
);
695 mutex_unlock(&attr
->mutex
);
699 /* interpret the buffer as a number to call the set function with */
700 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
701 size_t len
, loff_t
*ppos
)
703 struct simple_attr
*attr
;
708 attr
= file
->private_data
;
712 ret
= mutex_lock_interruptible(&attr
->mutex
);
717 size
= min(sizeof(attr
->set_buf
) - 1, len
);
718 if (copy_from_user(attr
->set_buf
, buf
, size
))
721 ret
= len
; /* claim we got the whole input */
722 attr
->set_buf
[size
] = '\0';
723 val
= simple_strtol(attr
->set_buf
, NULL
, 0);
724 attr
->set(attr
->data
, val
);
726 mutex_unlock(&attr
->mutex
);
731 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
732 * @sb: filesystem to do the file handle conversion on
733 * @fid: file handle to convert
734 * @fh_len: length of the file handle in bytes
735 * @fh_type: type of file handle
736 * @get_inode: filesystem callback to retrieve inode
738 * This function decodes @fid as long as it has one of the well-known
739 * Linux filehandle types and calls @get_inode on it to retrieve the
740 * inode for the object specified in the file handle.
742 struct dentry
*generic_fh_to_dentry(struct super_block
*sb
, struct fid
*fid
,
743 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
744 (struct super_block
*sb
, u64 ino
, u32 gen
))
746 struct inode
*inode
= NULL
;
752 case FILEID_INO32_GEN
:
753 case FILEID_INO32_GEN_PARENT
:
754 inode
= get_inode(sb
, fid
->i32
.ino
, fid
->i32
.gen
);
758 return d_obtain_alias(inode
);
760 EXPORT_SYMBOL_GPL(generic_fh_to_dentry
);
763 * generic_fh_to_dentry - generic helper for the fh_to_parent export operation
764 * @sb: filesystem to do the file handle conversion on
765 * @fid: file handle to convert
766 * @fh_len: length of the file handle in bytes
767 * @fh_type: type of file handle
768 * @get_inode: filesystem callback to retrieve inode
770 * This function decodes @fid as long as it has one of the well-known
771 * Linux filehandle types and calls @get_inode on it to retrieve the
772 * inode for the _parent_ object specified in the file handle if it
773 * is specified in the file handle, or NULL otherwise.
775 struct dentry
*generic_fh_to_parent(struct super_block
*sb
, struct fid
*fid
,
776 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
777 (struct super_block
*sb
, u64 ino
, u32 gen
))
779 struct inode
*inode
= NULL
;
785 case FILEID_INO32_GEN_PARENT
:
786 inode
= get_inode(sb
, fid
->i32
.parent_ino
,
787 (fh_len
> 3 ? fid
->i32
.parent_gen
: 0));
791 return d_obtain_alias(inode
);
793 EXPORT_SYMBOL_GPL(generic_fh_to_parent
);
795 EXPORT_SYMBOL(dcache_dir_close
);
796 EXPORT_SYMBOL(dcache_dir_lseek
);
797 EXPORT_SYMBOL(dcache_dir_open
);
798 EXPORT_SYMBOL(dcache_readdir
);
799 EXPORT_SYMBOL(generic_read_dir
);
800 EXPORT_SYMBOL(get_sb_pseudo
);
801 EXPORT_SYMBOL(simple_write_begin
);
802 EXPORT_SYMBOL(simple_write_end
);
803 EXPORT_SYMBOL(simple_dir_inode_operations
);
804 EXPORT_SYMBOL(simple_dir_operations
);
805 EXPORT_SYMBOL(simple_empty
);
806 EXPORT_SYMBOL(d_alloc_name
);
807 EXPORT_SYMBOL(simple_fill_super
);
808 EXPORT_SYMBOL(simple_getattr
);
809 EXPORT_SYMBOL(simple_link
);
810 EXPORT_SYMBOL(simple_lookup
);
811 EXPORT_SYMBOL(simple_pin_fs
);
812 EXPORT_UNUSED_SYMBOL(simple_prepare_write
);
813 EXPORT_SYMBOL(simple_readpage
);
814 EXPORT_SYMBOL(simple_release_fs
);
815 EXPORT_SYMBOL(simple_rename
);
816 EXPORT_SYMBOL(simple_rmdir
);
817 EXPORT_SYMBOL(simple_statfs
);
818 EXPORT_SYMBOL(simple_sync_file
);
819 EXPORT_SYMBOL(simple_unlink
);
820 EXPORT_SYMBOL(simple_read_from_buffer
);
821 EXPORT_SYMBOL(memory_read_from_buffer
);
822 EXPORT_SYMBOL(simple_transaction_get
);
823 EXPORT_SYMBOL(simple_transaction_read
);
824 EXPORT_SYMBOL(simple_transaction_release
);
825 EXPORT_SYMBOL_GPL(simple_attr_open
);
826 EXPORT_SYMBOL_GPL(simple_attr_release
);
827 EXPORT_SYMBOL_GPL(simple_attr_read
);
828 EXPORT_SYMBOL_GPL(simple_attr_write
);