3 * Library for filesystems writers.
6 #include <linux/blkdev.h>
7 #include <linux/export.h>
8 #include <linux/pagemap.h>
9 #include <linux/slab.h>
10 #include <linux/mount.h>
11 #include <linux/vfs.h>
12 #include <linux/quotaops.h>
13 #include <linux/mutex.h>
14 #include <linux/namei.h>
15 #include <linux/exportfs.h>
16 #include <linux/writeback.h>
17 #include <linux/buffer_head.h> /* sync_mapping_buffers */
19 #include <asm/uaccess.h>
23 int simple_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
26 struct inode
*inode
= d_inode(dentry
);
27 generic_fillattr(inode
, stat
);
28 stat
->blocks
= inode
->i_mapping
->nrpages
<< (PAGE_CACHE_SHIFT
- 9);
31 EXPORT_SYMBOL(simple_getattr
);
33 int simple_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
35 buf
->f_type
= dentry
->d_sb
->s_magic
;
36 buf
->f_bsize
= PAGE_CACHE_SIZE
;
37 buf
->f_namelen
= NAME_MAX
;
40 EXPORT_SYMBOL(simple_statfs
);
43 * Retaining negative dentries for an in-memory filesystem just wastes
44 * memory and lookup time: arrange for them to be deleted immediately.
46 int always_delete_dentry(const struct dentry
*dentry
)
50 EXPORT_SYMBOL(always_delete_dentry
);
52 const struct dentry_operations simple_dentry_operations
= {
53 .d_delete
= always_delete_dentry
,
55 EXPORT_SYMBOL(simple_dentry_operations
);
58 * Lookup the data. This is trivial - if the dentry didn't already
59 * exist, we know it is negative. Set d_op to delete negative dentries.
61 struct dentry
*simple_lookup(struct inode
*dir
, struct dentry
*dentry
, unsigned int flags
)
63 if (dentry
->d_name
.len
> NAME_MAX
)
64 return ERR_PTR(-ENAMETOOLONG
);
65 if (!dentry
->d_sb
->s_d_op
)
66 d_set_d_op(dentry
, &simple_dentry_operations
);
70 EXPORT_SYMBOL(simple_lookup
);
72 int dcache_dir_open(struct inode
*inode
, struct file
*file
)
74 static struct qstr cursor_name
= QSTR_INIT(".", 1);
76 file
->private_data
= d_alloc(file
->f_path
.dentry
, &cursor_name
);
78 return file
->private_data
? 0 : -ENOMEM
;
80 EXPORT_SYMBOL(dcache_dir_open
);
82 int dcache_dir_close(struct inode
*inode
, struct file
*file
)
84 dput(file
->private_data
);
87 EXPORT_SYMBOL(dcache_dir_close
);
89 loff_t
dcache_dir_lseek(struct file
*file
, loff_t offset
, int whence
)
91 struct dentry
*dentry
= file
->f_path
.dentry
;
92 mutex_lock(&d_inode(dentry
)->i_mutex
);
95 offset
+= file
->f_pos
;
100 mutex_unlock(&d_inode(dentry
)->i_mutex
);
103 if (offset
!= file
->f_pos
) {
104 file
->f_pos
= offset
;
105 if (file
->f_pos
>= 2) {
107 struct dentry
*cursor
= file
->private_data
;
108 loff_t n
= file
->f_pos
- 2;
110 spin_lock(&dentry
->d_lock
);
111 /* d_lock not required for cursor */
112 list_del(&cursor
->d_child
);
113 p
= dentry
->d_subdirs
.next
;
114 while (n
&& p
!= &dentry
->d_subdirs
) {
116 next
= list_entry(p
, struct dentry
, d_child
);
117 spin_lock_nested(&next
->d_lock
, DENTRY_D_LOCK_NESTED
);
118 if (simple_positive(next
))
120 spin_unlock(&next
->d_lock
);
123 list_add_tail(&cursor
->d_child
, p
);
124 spin_unlock(&dentry
->d_lock
);
127 mutex_unlock(&d_inode(dentry
)->i_mutex
);
130 EXPORT_SYMBOL(dcache_dir_lseek
);
132 /* Relationship between i_mode and the DT_xxx types */
133 static inline unsigned char dt_type(struct inode
*inode
)
135 return (inode
->i_mode
>> 12) & 15;
139 * Directory is locked and all positive dentries in it are safe, since
140 * for ramfs-type trees they can't go away without unlink() or rmdir(),
141 * both impossible due to the lock on directory.
144 int dcache_readdir(struct file
*file
, struct dir_context
*ctx
)
146 struct dentry
*dentry
= file
->f_path
.dentry
;
147 struct dentry
*cursor
= file
->private_data
;
148 struct list_head
*p
, *q
= &cursor
->d_child
;
150 if (!dir_emit_dots(file
, ctx
))
152 spin_lock(&dentry
->d_lock
);
154 list_move(q
, &dentry
->d_subdirs
);
156 for (p
= q
->next
; p
!= &dentry
->d_subdirs
; p
= p
->next
) {
157 struct dentry
*next
= list_entry(p
, struct dentry
, d_child
);
158 spin_lock_nested(&next
->d_lock
, DENTRY_D_LOCK_NESTED
);
159 if (!simple_positive(next
)) {
160 spin_unlock(&next
->d_lock
);
164 spin_unlock(&next
->d_lock
);
165 spin_unlock(&dentry
->d_lock
);
166 if (!dir_emit(ctx
, next
->d_name
.name
, next
->d_name
.len
,
167 d_inode(next
)->i_ino
, dt_type(d_inode(next
))))
169 spin_lock(&dentry
->d_lock
);
170 spin_lock_nested(&next
->d_lock
, DENTRY_D_LOCK_NESTED
);
171 /* next is still alive */
173 spin_unlock(&next
->d_lock
);
177 spin_unlock(&dentry
->d_lock
);
180 EXPORT_SYMBOL(dcache_readdir
);
182 ssize_t
generic_read_dir(struct file
*filp
, char __user
*buf
, size_t siz
, loff_t
*ppos
)
186 EXPORT_SYMBOL(generic_read_dir
);
188 const struct file_operations simple_dir_operations
= {
189 .open
= dcache_dir_open
,
190 .release
= dcache_dir_close
,
191 .llseek
= dcache_dir_lseek
,
192 .read
= generic_read_dir
,
193 .iterate
= dcache_readdir
,
196 EXPORT_SYMBOL(simple_dir_operations
);
198 const struct inode_operations simple_dir_inode_operations
= {
199 .lookup
= simple_lookup
,
201 EXPORT_SYMBOL(simple_dir_inode_operations
);
203 static const struct super_operations simple_super_operations
= {
204 .statfs
= simple_statfs
,
208 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
209 * will never be mountable)
211 struct dentry
*mount_pseudo(struct file_system_type
*fs_type
, char *name
,
212 const struct super_operations
*ops
,
213 const struct dentry_operations
*dops
, unsigned long magic
)
215 struct super_block
*s
;
216 struct dentry
*dentry
;
218 struct qstr d_name
= QSTR_INIT(name
, strlen(name
));
220 s
= sget(fs_type
, NULL
, set_anon_super
, MS_NOUSER
, NULL
);
224 s
->s_maxbytes
= MAX_LFS_FILESIZE
;
225 s
->s_blocksize
= PAGE_SIZE
;
226 s
->s_blocksize_bits
= PAGE_SHIFT
;
228 s
->s_op
= ops
? ops
: &simple_super_operations
;
234 * since this is the first inode, make it number 1. New inodes created
235 * after this must take care not to collide with it (by passing
236 * max_reserved of 1 to iunique).
239 root
->i_mode
= S_IFDIR
| S_IRUSR
| S_IWUSR
;
240 root
->i_atime
= root
->i_mtime
= root
->i_ctime
= CURRENT_TIME
;
241 dentry
= __d_alloc(s
, &d_name
);
246 d_instantiate(dentry
, root
);
249 s
->s_flags
|= MS_ACTIVE
;
250 return dget(s
->s_root
);
253 deactivate_locked_super(s
);
254 return ERR_PTR(-ENOMEM
);
256 EXPORT_SYMBOL(mount_pseudo
);
258 int simple_open(struct inode
*inode
, struct file
*file
)
260 if (inode
->i_private
)
261 file
->private_data
= inode
->i_private
;
264 EXPORT_SYMBOL(simple_open
);
266 int simple_link(struct dentry
*old_dentry
, struct inode
*dir
, struct dentry
*dentry
)
268 struct inode
*inode
= d_inode(old_dentry
);
270 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= CURRENT_TIME
;
274 d_instantiate(dentry
, inode
);
277 EXPORT_SYMBOL(simple_link
);
279 int simple_empty(struct dentry
*dentry
)
281 struct dentry
*child
;
284 spin_lock(&dentry
->d_lock
);
285 list_for_each_entry(child
, &dentry
->d_subdirs
, d_child
) {
286 spin_lock_nested(&child
->d_lock
, DENTRY_D_LOCK_NESTED
);
287 if (simple_positive(child
)) {
288 spin_unlock(&child
->d_lock
);
291 spin_unlock(&child
->d_lock
);
295 spin_unlock(&dentry
->d_lock
);
298 EXPORT_SYMBOL(simple_empty
);
300 int simple_unlink(struct inode
*dir
, struct dentry
*dentry
)
302 struct inode
*inode
= d_inode(dentry
);
304 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= CURRENT_TIME
;
309 EXPORT_SYMBOL(simple_unlink
);
311 int simple_rmdir(struct inode
*dir
, struct dentry
*dentry
)
313 if (!simple_empty(dentry
))
316 drop_nlink(d_inode(dentry
));
317 simple_unlink(dir
, dentry
);
321 EXPORT_SYMBOL(simple_rmdir
);
323 int simple_rename(struct inode
*old_dir
, struct dentry
*old_dentry
,
324 struct inode
*new_dir
, struct dentry
*new_dentry
)
326 struct inode
*inode
= d_inode(old_dentry
);
327 int they_are_dirs
= d_is_dir(old_dentry
);
329 if (!simple_empty(new_dentry
))
332 if (d_really_is_positive(new_dentry
)) {
333 simple_unlink(new_dir
, new_dentry
);
335 drop_nlink(d_inode(new_dentry
));
338 } else if (they_are_dirs
) {
343 old_dir
->i_ctime
= old_dir
->i_mtime
= new_dir
->i_ctime
=
344 new_dir
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
348 EXPORT_SYMBOL(simple_rename
);
351 * simple_setattr - setattr for simple filesystem
353 * @iattr: iattr structure
355 * Returns 0 on success, -error on failure.
357 * simple_setattr is a simple ->setattr implementation without a proper
358 * implementation of size changes.
360 * It can either be used for in-memory filesystems or special files
361 * on simple regular filesystems. Anything that needs to change on-disk
362 * or wire state on size changes needs its own setattr method.
364 int simple_setattr(struct dentry
*dentry
, struct iattr
*iattr
)
366 struct inode
*inode
= d_inode(dentry
);
369 error
= inode_change_ok(inode
, iattr
);
373 if (iattr
->ia_valid
& ATTR_SIZE
)
374 truncate_setsize(inode
, iattr
->ia_size
);
375 setattr_copy(inode
, iattr
);
376 mark_inode_dirty(inode
);
379 EXPORT_SYMBOL(simple_setattr
);
381 int simple_readpage(struct file
*file
, struct page
*page
)
383 clear_highpage(page
);
384 flush_dcache_page(page
);
385 SetPageUptodate(page
);
389 EXPORT_SYMBOL(simple_readpage
);
391 int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
392 loff_t pos
, unsigned len
, unsigned flags
,
393 struct page
**pagep
, void **fsdata
)
398 index
= pos
>> PAGE_CACHE_SHIFT
;
400 page
= grab_cache_page_write_begin(mapping
, index
, flags
);
406 if (!PageUptodate(page
) && (len
!= PAGE_CACHE_SIZE
)) {
407 unsigned from
= pos
& (PAGE_CACHE_SIZE
- 1);
409 zero_user_segments(page
, 0, from
, from
+ len
, PAGE_CACHE_SIZE
);
413 EXPORT_SYMBOL(simple_write_begin
);
416 * simple_write_end - .write_end helper for non-block-device FSes
417 * @available: See .write_end of address_space_operations
426 * simple_write_end does the minimum needed for updating a page after writing is
427 * done. It has the same API signature as the .write_end of
428 * address_space_operations vector. So it can just be set onto .write_end for
429 * FSes that don't need any other processing. i_mutex is assumed to be held.
430 * Block based filesystems should use generic_write_end().
431 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
432 * is not called, so a filesystem that actually does store data in .write_inode
433 * should extend on what's done here with a call to mark_inode_dirty() in the
434 * case that i_size has changed.
436 int simple_write_end(struct file
*file
, struct address_space
*mapping
,
437 loff_t pos
, unsigned len
, unsigned copied
,
438 struct page
*page
, void *fsdata
)
440 struct inode
*inode
= page
->mapping
->host
;
441 loff_t last_pos
= pos
+ copied
;
443 /* zero the stale part of the page if we did a short copy */
445 unsigned from
= pos
& (PAGE_CACHE_SIZE
- 1);
447 zero_user(page
, from
+ copied
, len
- copied
);
450 if (!PageUptodate(page
))
451 SetPageUptodate(page
);
453 * No need to use i_size_read() here, the i_size
454 * cannot change under us because we hold the i_mutex.
456 if (last_pos
> inode
->i_size
)
457 i_size_write(inode
, last_pos
);
459 set_page_dirty(page
);
461 page_cache_release(page
);
465 EXPORT_SYMBOL(simple_write_end
);
468 * the inodes created here are not hashed. If you use iunique to generate
469 * unique inode values later for this filesystem, then you must take care
470 * to pass it an appropriate max_reserved value to avoid collisions.
472 int simple_fill_super(struct super_block
*s
, unsigned long magic
,
473 struct tree_descr
*files
)
477 struct dentry
*dentry
;
480 s
->s_blocksize
= PAGE_CACHE_SIZE
;
481 s
->s_blocksize_bits
= PAGE_CACHE_SHIFT
;
483 s
->s_op
= &simple_super_operations
;
486 inode
= new_inode(s
);
490 * because the root inode is 1, the files array must not contain an
494 inode
->i_mode
= S_IFDIR
| 0755;
495 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
496 inode
->i_op
= &simple_dir_inode_operations
;
497 inode
->i_fop
= &simple_dir_operations
;
499 root
= d_make_root(inode
);
502 for (i
= 0; !files
->name
|| files
->name
[0]; i
++, files
++) {
506 /* warn if it tries to conflict with the root inode */
507 if (unlikely(i
== 1))
508 printk(KERN_WARNING
"%s: %s passed in a files array"
509 "with an index of 1!\n", __func__
,
512 dentry
= d_alloc_name(root
, files
->name
);
515 inode
= new_inode(s
);
520 inode
->i_mode
= S_IFREG
| files
->mode
;
521 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
522 inode
->i_fop
= files
->ops
;
524 d_add(dentry
, inode
);
530 shrink_dcache_parent(root
);
534 EXPORT_SYMBOL(simple_fill_super
);
536 static DEFINE_SPINLOCK(pin_fs_lock
);
538 int simple_pin_fs(struct file_system_type
*type
, struct vfsmount
**mount
, int *count
)
540 struct vfsmount
*mnt
= NULL
;
541 spin_lock(&pin_fs_lock
);
542 if (unlikely(!*mount
)) {
543 spin_unlock(&pin_fs_lock
);
544 mnt
= vfs_kern_mount(type
, MS_KERNMOUNT
, type
->name
, NULL
);
547 spin_lock(&pin_fs_lock
);
553 spin_unlock(&pin_fs_lock
);
557 EXPORT_SYMBOL(simple_pin_fs
);
559 void simple_release_fs(struct vfsmount
**mount
, int *count
)
561 struct vfsmount
*mnt
;
562 spin_lock(&pin_fs_lock
);
566 spin_unlock(&pin_fs_lock
);
569 EXPORT_SYMBOL(simple_release_fs
);
572 * simple_read_from_buffer - copy data from the buffer to user space
573 * @to: the user space buffer to read to
574 * @count: the maximum number of bytes to read
575 * @ppos: the current position in the buffer
576 * @from: the buffer to read from
577 * @available: the size of the buffer
579 * The simple_read_from_buffer() function reads up to @count bytes from the
580 * buffer @from at offset @ppos into the user space address starting at @to.
582 * On success, the number of bytes read is returned and the offset @ppos is
583 * advanced by this number, or negative value is returned on error.
585 ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
, loff_t
*ppos
,
586 const void *from
, size_t available
)
593 if (pos
>= available
|| !count
)
595 if (count
> available
- pos
)
596 count
= available
- pos
;
597 ret
= copy_to_user(to
, from
+ pos
, count
);
604 EXPORT_SYMBOL(simple_read_from_buffer
);
607 * simple_write_to_buffer - copy data from user space to the buffer
608 * @to: the buffer to write to
609 * @available: the size of the buffer
610 * @ppos: the current position in the buffer
611 * @from: the user space buffer to read from
612 * @count: the maximum number of bytes to read
614 * The simple_write_to_buffer() function reads up to @count bytes from the user
615 * space address starting at @from into the buffer @to at offset @ppos.
617 * On success, the number of bytes written is returned and the offset @ppos is
618 * advanced by this number, or negative value is returned on error.
620 ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
621 const void __user
*from
, size_t count
)
628 if (pos
>= available
|| !count
)
630 if (count
> available
- pos
)
631 count
= available
- pos
;
632 res
= copy_from_user(to
+ pos
, from
, count
);
639 EXPORT_SYMBOL(simple_write_to_buffer
);
642 * memory_read_from_buffer - copy data from the buffer
643 * @to: the kernel space buffer to read to
644 * @count: the maximum number of bytes to read
645 * @ppos: the current position in the buffer
646 * @from: the buffer to read from
647 * @available: the size of the buffer
649 * The memory_read_from_buffer() function reads up to @count bytes from the
650 * buffer @from at offset @ppos into the kernel space address starting at @to.
652 * On success, the number of bytes read is returned and the offset @ppos is
653 * advanced by this number, or negative value is returned on error.
655 ssize_t
memory_read_from_buffer(void *to
, size_t count
, loff_t
*ppos
,
656 const void *from
, size_t available
)
662 if (pos
>= available
)
664 if (count
> available
- pos
)
665 count
= available
- pos
;
666 memcpy(to
, from
+ pos
, count
);
671 EXPORT_SYMBOL(memory_read_from_buffer
);
674 * Transaction based IO.
675 * The file expects a single write which triggers the transaction, and then
676 * possibly a read which collects the result - which is stored in a
680 void simple_transaction_set(struct file
*file
, size_t n
)
682 struct simple_transaction_argresp
*ar
= file
->private_data
;
684 BUG_ON(n
> SIMPLE_TRANSACTION_LIMIT
);
687 * The barrier ensures that ar->size will really remain zero until
688 * ar->data is ready for reading.
693 EXPORT_SYMBOL(simple_transaction_set
);
695 char *simple_transaction_get(struct file
*file
, const char __user
*buf
, size_t size
)
697 struct simple_transaction_argresp
*ar
;
698 static DEFINE_SPINLOCK(simple_transaction_lock
);
700 if (size
> SIMPLE_TRANSACTION_LIMIT
- 1)
701 return ERR_PTR(-EFBIG
);
703 ar
= (struct simple_transaction_argresp
*)get_zeroed_page(GFP_KERNEL
);
705 return ERR_PTR(-ENOMEM
);
707 spin_lock(&simple_transaction_lock
);
709 /* only one write allowed per open */
710 if (file
->private_data
) {
711 spin_unlock(&simple_transaction_lock
);
712 free_page((unsigned long)ar
);
713 return ERR_PTR(-EBUSY
);
716 file
->private_data
= ar
;
718 spin_unlock(&simple_transaction_lock
);
720 if (copy_from_user(ar
->data
, buf
, size
))
721 return ERR_PTR(-EFAULT
);
725 EXPORT_SYMBOL(simple_transaction_get
);
727 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
, size_t size
, loff_t
*pos
)
729 struct simple_transaction_argresp
*ar
= file
->private_data
;
733 return simple_read_from_buffer(buf
, size
, pos
, ar
->data
, ar
->size
);
735 EXPORT_SYMBOL(simple_transaction_read
);
737 int simple_transaction_release(struct inode
*inode
, struct file
*file
)
739 free_page((unsigned long)file
->private_data
);
742 EXPORT_SYMBOL(simple_transaction_release
);
744 /* Simple attribute files */
747 int (*get
)(void *, u64
*);
748 int (*set
)(void *, u64
);
749 char get_buf
[24]; /* enough to store a u64 and "\n\0" */
752 const char *fmt
; /* format for read operation */
753 struct mutex mutex
; /* protects access to these buffers */
756 /* simple_attr_open is called by an actual attribute open file operation
757 * to set the attribute specific access operations. */
758 int simple_attr_open(struct inode
*inode
, struct file
*file
,
759 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
762 struct simple_attr
*attr
;
764 attr
= kmalloc(sizeof(*attr
), GFP_KERNEL
);
770 attr
->data
= inode
->i_private
;
772 mutex_init(&attr
->mutex
);
774 file
->private_data
= attr
;
776 return nonseekable_open(inode
, file
);
778 EXPORT_SYMBOL_GPL(simple_attr_open
);
780 int simple_attr_release(struct inode
*inode
, struct file
*file
)
782 kfree(file
->private_data
);
785 EXPORT_SYMBOL_GPL(simple_attr_release
); /* GPL-only? This? Really? */
787 /* read from the buffer that is filled with the get function */
788 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
789 size_t len
, loff_t
*ppos
)
791 struct simple_attr
*attr
;
795 attr
= file
->private_data
;
800 ret
= mutex_lock_interruptible(&attr
->mutex
);
804 if (*ppos
) { /* continued read */
805 size
= strlen(attr
->get_buf
);
806 } else { /* first read */
808 ret
= attr
->get(attr
->data
, &val
);
812 size
= scnprintf(attr
->get_buf
, sizeof(attr
->get_buf
),
813 attr
->fmt
, (unsigned long long)val
);
816 ret
= simple_read_from_buffer(buf
, len
, ppos
, attr
->get_buf
, size
);
818 mutex_unlock(&attr
->mutex
);
821 EXPORT_SYMBOL_GPL(simple_attr_read
);
823 /* interpret the buffer as a number to call the set function with */
824 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
825 size_t len
, loff_t
*ppos
)
827 struct simple_attr
*attr
;
832 attr
= file
->private_data
;
836 ret
= mutex_lock_interruptible(&attr
->mutex
);
841 size
= min(sizeof(attr
->set_buf
) - 1, len
);
842 if (copy_from_user(attr
->set_buf
, buf
, size
))
845 attr
->set_buf
[size
] = '\0';
846 val
= simple_strtoll(attr
->set_buf
, NULL
, 0);
847 ret
= attr
->set(attr
->data
, val
);
849 ret
= len
; /* on success, claim we got the whole input */
851 mutex_unlock(&attr
->mutex
);
854 EXPORT_SYMBOL_GPL(simple_attr_write
);
857 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
858 * @sb: filesystem to do the file handle conversion on
859 * @fid: file handle to convert
860 * @fh_len: length of the file handle in bytes
861 * @fh_type: type of file handle
862 * @get_inode: filesystem callback to retrieve inode
864 * This function decodes @fid as long as it has one of the well-known
865 * Linux filehandle types and calls @get_inode on it to retrieve the
866 * inode for the object specified in the file handle.
868 struct dentry
*generic_fh_to_dentry(struct super_block
*sb
, struct fid
*fid
,
869 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
870 (struct super_block
*sb
, u64 ino
, u32 gen
))
872 struct inode
*inode
= NULL
;
878 case FILEID_INO32_GEN
:
879 case FILEID_INO32_GEN_PARENT
:
880 inode
= get_inode(sb
, fid
->i32
.ino
, fid
->i32
.gen
);
884 return d_obtain_alias(inode
);
886 EXPORT_SYMBOL_GPL(generic_fh_to_dentry
);
889 * generic_fh_to_parent - generic helper for the fh_to_parent export operation
890 * @sb: filesystem to do the file handle conversion on
891 * @fid: file handle to convert
892 * @fh_len: length of the file handle in bytes
893 * @fh_type: type of file handle
894 * @get_inode: filesystem callback to retrieve inode
896 * This function decodes @fid as long as it has one of the well-known
897 * Linux filehandle types and calls @get_inode on it to retrieve the
898 * inode for the _parent_ object specified in the file handle if it
899 * is specified in the file handle, or NULL otherwise.
901 struct dentry
*generic_fh_to_parent(struct super_block
*sb
, struct fid
*fid
,
902 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
903 (struct super_block
*sb
, u64 ino
, u32 gen
))
905 struct inode
*inode
= NULL
;
911 case FILEID_INO32_GEN_PARENT
:
912 inode
= get_inode(sb
, fid
->i32
.parent_ino
,
913 (fh_len
> 3 ? fid
->i32
.parent_gen
: 0));
917 return d_obtain_alias(inode
);
919 EXPORT_SYMBOL_GPL(generic_fh_to_parent
);
922 * __generic_file_fsync - generic fsync implementation for simple filesystems
924 * @file: file to synchronize
925 * @start: start offset in bytes
926 * @end: end offset in bytes (inclusive)
927 * @datasync: only synchronize essential metadata if true
929 * This is a generic implementation of the fsync method for simple
930 * filesystems which track all non-inode metadata in the buffers list
931 * hanging off the address_space structure.
933 int __generic_file_fsync(struct file
*file
, loff_t start
, loff_t end
,
936 struct inode
*inode
= file
->f_mapping
->host
;
940 err
= filemap_write_and_wait_range(inode
->i_mapping
, start
, end
);
944 mutex_lock(&inode
->i_mutex
);
945 ret
= sync_mapping_buffers(inode
->i_mapping
);
946 if (!(inode
->i_state
& I_DIRTY_ALL
))
948 if (datasync
&& !(inode
->i_state
& I_DIRTY_DATASYNC
))
951 err
= sync_inode_metadata(inode
, 1);
956 mutex_unlock(&inode
->i_mutex
);
959 EXPORT_SYMBOL(__generic_file_fsync
);
962 * generic_file_fsync - generic fsync implementation for simple filesystems
964 * @file: file to synchronize
965 * @start: start offset in bytes
966 * @end: end offset in bytes (inclusive)
967 * @datasync: only synchronize essential metadata if true
971 int generic_file_fsync(struct file
*file
, loff_t start
, loff_t end
,
974 struct inode
*inode
= file
->f_mapping
->host
;
977 err
= __generic_file_fsync(file
, start
, end
, datasync
);
980 return blkdev_issue_flush(inode
->i_sb
->s_bdev
, GFP_KERNEL
, NULL
);
982 EXPORT_SYMBOL(generic_file_fsync
);
985 * generic_check_addressable - Check addressability of file system
986 * @blocksize_bits: log of file system block size
987 * @num_blocks: number of blocks in file system
989 * Determine whether a file system with @num_blocks blocks (and a
990 * block size of 2**@blocksize_bits) is addressable by the sector_t
991 * and page cache of the system. Return 0 if so and -EFBIG otherwise.
993 int generic_check_addressable(unsigned blocksize_bits
, u64 num_blocks
)
995 u64 last_fs_block
= num_blocks
- 1;
997 last_fs_block
>> (PAGE_CACHE_SHIFT
- blocksize_bits
);
999 if (unlikely(num_blocks
== 0))
1002 if ((blocksize_bits
< 9) || (blocksize_bits
> PAGE_CACHE_SHIFT
))
1005 if ((last_fs_block
> (sector_t
)(~0ULL) >> (blocksize_bits
- 9)) ||
1006 (last_fs_page
> (pgoff_t
)(~0ULL))) {
1011 EXPORT_SYMBOL(generic_check_addressable
);
1014 * No-op implementation of ->fsync for in-memory filesystems.
1016 int noop_fsync(struct file
*file
, loff_t start
, loff_t end
, int datasync
)
1020 EXPORT_SYMBOL(noop_fsync
);
1022 /* Because kfree isn't assignment-compatible with void(void*) ;-/ */
1023 void kfree_link(void *p
)
1027 EXPORT_SYMBOL(kfree_link
);
1030 * nop .set_page_dirty method so that people can use .page_mkwrite on
1033 static int anon_set_page_dirty(struct page
*page
)
1039 * A single inode exists for all anon_inode files. Contrary to pipes,
1040 * anon_inode inodes have no associated per-instance data, so we need
1041 * only allocate one of them.
1043 struct inode
*alloc_anon_inode(struct super_block
*s
)
1045 static const struct address_space_operations anon_aops
= {
1046 .set_page_dirty
= anon_set_page_dirty
,
1048 struct inode
*inode
= new_inode_pseudo(s
);
1051 return ERR_PTR(-ENOMEM
);
1053 inode
->i_ino
= get_next_ino();
1054 inode
->i_mapping
->a_ops
= &anon_aops
;
1057 * Mark the inode dirty from the very beginning,
1058 * that way it will never be moved to the dirty
1059 * list because mark_inode_dirty() will think
1060 * that it already _is_ on the dirty list.
1062 inode
->i_state
= I_DIRTY
;
1063 inode
->i_mode
= S_IRUSR
| S_IWUSR
;
1064 inode
->i_uid
= current_fsuid();
1065 inode
->i_gid
= current_fsgid();
1066 inode
->i_flags
|= S_PRIVATE
;
1067 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
1070 EXPORT_SYMBOL(alloc_anon_inode
);
1073 * simple_nosetlease - generic helper for prohibiting leases
1074 * @filp: file pointer
1075 * @arg: type of lease to obtain
1076 * @flp: new lease supplied for insertion
1077 * @priv: private data for lm_setup operation
1079 * Generic helper for filesystems that do not wish to allow leases to be set.
1080 * All arguments are ignored and it just returns -EINVAL.
1083 simple_nosetlease(struct file
*filp
, long arg
, struct file_lock
**flp
,
1088 EXPORT_SYMBOL(simple_nosetlease
);
1090 const char *simple_get_link(struct dentry
*dentry
, struct inode
*inode
,
1091 struct delayed_call
*done
)
1093 return inode
->i_link
;
1095 EXPORT_SYMBOL(simple_get_link
);
1097 const struct inode_operations simple_symlink_inode_operations
= {
1098 .get_link
= simple_get_link
,
1099 .readlink
= generic_readlink
1101 EXPORT_SYMBOL(simple_symlink_inode_operations
);
1104 * Operations for a permanently empty directory.
1106 static struct dentry
*empty_dir_lookup(struct inode
*dir
, struct dentry
*dentry
, unsigned int flags
)
1108 return ERR_PTR(-ENOENT
);
1111 static int empty_dir_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1114 struct inode
*inode
= d_inode(dentry
);
1115 generic_fillattr(inode
, stat
);
1119 static int empty_dir_setattr(struct dentry
*dentry
, struct iattr
*attr
)
1124 static int empty_dir_setxattr(struct dentry
*dentry
, const char *name
,
1125 const void *value
, size_t size
, int flags
)
1130 static ssize_t
empty_dir_getxattr(struct dentry
*dentry
, const char *name
,
1131 void *value
, size_t size
)
1136 static int empty_dir_removexattr(struct dentry
*dentry
, const char *name
)
1141 static ssize_t
empty_dir_listxattr(struct dentry
*dentry
, char *list
, size_t size
)
1146 static const struct inode_operations empty_dir_inode_operations
= {
1147 .lookup
= empty_dir_lookup
,
1148 .permission
= generic_permission
,
1149 .setattr
= empty_dir_setattr
,
1150 .getattr
= empty_dir_getattr
,
1151 .setxattr
= empty_dir_setxattr
,
1152 .getxattr
= empty_dir_getxattr
,
1153 .removexattr
= empty_dir_removexattr
,
1154 .listxattr
= empty_dir_listxattr
,
1157 static loff_t
empty_dir_llseek(struct file
*file
, loff_t offset
, int whence
)
1159 /* An empty directory has two entries . and .. at offsets 0 and 1 */
1160 return generic_file_llseek_size(file
, offset
, whence
, 2, 2);
1163 static int empty_dir_readdir(struct file
*file
, struct dir_context
*ctx
)
1165 dir_emit_dots(file
, ctx
);
1169 static const struct file_operations empty_dir_operations
= {
1170 .llseek
= empty_dir_llseek
,
1171 .read
= generic_read_dir
,
1172 .iterate
= empty_dir_readdir
,
1173 .fsync
= noop_fsync
,
1177 void make_empty_dir_inode(struct inode
*inode
)
1179 set_nlink(inode
, 2);
1180 inode
->i_mode
= S_IFDIR
| S_IRUGO
| S_IXUGO
;
1181 inode
->i_uid
= GLOBAL_ROOT_UID
;
1182 inode
->i_gid
= GLOBAL_ROOT_GID
;
1185 inode
->i_blkbits
= PAGE_SHIFT
;
1186 inode
->i_blocks
= 0;
1188 inode
->i_op
= &empty_dir_inode_operations
;
1189 inode
->i_fop
= &empty_dir_operations
;
1192 bool is_empty_dir_inode(struct inode
*inode
)
1194 return (inode
->i_fop
== &empty_dir_operations
) &&
1195 (inode
->i_op
== &empty_dir_inode_operations
);