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_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_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 file
->private_data
= d_alloc_cursor(file
->f_path
.dentry
);
76 return file
->private_data
? 0 : -ENOMEM
;
78 EXPORT_SYMBOL(dcache_dir_open
);
80 int dcache_dir_close(struct inode
*inode
, struct file
*file
)
82 dput(file
->private_data
);
85 EXPORT_SYMBOL(dcache_dir_close
);
87 /* parent is locked at least shared */
88 static struct dentry
*next_positive(struct dentry
*parent
,
89 struct list_head
*from
,
92 struct dentry
*res
= NULL
;
95 spin_lock(&parent
->d_lock
);
96 for (p
= from
->next
; p
!= &parent
->d_subdirs
; p
= p
->next
) {
97 struct dentry
*d
= list_entry(p
, struct dentry
, d_child
);
98 if (simple_positive(d
) && !--count
) {
103 spin_unlock(&parent
->d_lock
);
107 static void move_cursor(struct dentry
*cursor
, struct list_head
*after
)
109 struct dentry
*parent
= cursor
->d_parent
;
111 spin_lock(&parent
->d_lock
);
112 __list_del(cursor
->d_child
.prev
, cursor
->d_child
.next
);
114 list_add(&cursor
->d_child
, after
);
116 list_add_tail(&cursor
->d_child
, &parent
->d_subdirs
);
117 spin_unlock(&parent
->d_lock
);
120 loff_t
dcache_dir_lseek(struct file
*file
, loff_t offset
, int whence
)
122 struct dentry
*dentry
= file
->f_path
.dentry
;
125 offset
+= file
->f_pos
;
132 if (offset
!= file
->f_pos
) {
133 file
->f_pos
= offset
;
134 if (file
->f_pos
>= 2) {
135 struct dentry
*cursor
= file
->private_data
;
137 loff_t n
= file
->f_pos
- 2;
139 inode_lock_shared(dentry
->d_inode
);
140 to
= next_positive(dentry
, &dentry
->d_subdirs
, n
);
141 move_cursor(cursor
, to
? &to
->d_child
: NULL
);
142 inode_unlock_shared(dentry
->d_inode
);
147 EXPORT_SYMBOL(dcache_dir_lseek
);
149 /* Relationship between i_mode and the DT_xxx types */
150 static inline unsigned char dt_type(struct inode
*inode
)
152 return (inode
->i_mode
>> 12) & 15;
156 * Directory is locked and all positive dentries in it are safe, since
157 * for ramfs-type trees they can't go away without unlink() or rmdir(),
158 * both impossible due to the lock on directory.
161 int dcache_readdir(struct file
*file
, struct dir_context
*ctx
)
163 struct dentry
*dentry
= file
->f_path
.dentry
;
164 struct dentry
*cursor
= file
->private_data
;
165 struct list_head
*p
= &cursor
->d_child
;
169 if (!dir_emit_dots(file
, ctx
))
173 p
= &dentry
->d_subdirs
;
174 while ((next
= next_positive(dentry
, p
, 1)) != NULL
) {
175 if (!dir_emit(ctx
, next
->d_name
.name
, next
->d_name
.len
,
176 d_inode(next
)->i_ino
, dt_type(d_inode(next
))))
183 move_cursor(cursor
, p
);
186 EXPORT_SYMBOL(dcache_readdir
);
188 ssize_t
generic_read_dir(struct file
*filp
, char __user
*buf
, size_t siz
, loff_t
*ppos
)
192 EXPORT_SYMBOL(generic_read_dir
);
194 const struct file_operations simple_dir_operations
= {
195 .open
= dcache_dir_open
,
196 .release
= dcache_dir_close
,
197 .llseek
= dcache_dir_lseek
,
198 .read
= generic_read_dir
,
199 .iterate_shared
= dcache_readdir
,
202 EXPORT_SYMBOL(simple_dir_operations
);
204 const struct inode_operations simple_dir_inode_operations
= {
205 .lookup
= simple_lookup
,
207 EXPORT_SYMBOL(simple_dir_inode_operations
);
209 static const struct super_operations simple_super_operations
= {
210 .statfs
= simple_statfs
,
214 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
215 * will never be mountable)
217 struct dentry
*mount_pseudo(struct file_system_type
*fs_type
, char *name
,
218 const struct super_operations
*ops
,
219 const struct dentry_operations
*dops
, unsigned long magic
)
221 struct super_block
*s
;
222 struct dentry
*dentry
;
224 struct qstr d_name
= QSTR_INIT(name
, strlen(name
));
226 s
= sget(fs_type
, NULL
, set_anon_super
, MS_NOUSER
, NULL
);
230 s
->s_maxbytes
= MAX_LFS_FILESIZE
;
231 s
->s_blocksize
= PAGE_SIZE
;
232 s
->s_blocksize_bits
= PAGE_SHIFT
;
234 s
->s_op
= ops
? ops
: &simple_super_operations
;
240 * since this is the first inode, make it number 1. New inodes created
241 * after this must take care not to collide with it (by passing
242 * max_reserved of 1 to iunique).
245 root
->i_mode
= S_IFDIR
| S_IRUSR
| S_IWUSR
;
246 root
->i_atime
= root
->i_mtime
= root
->i_ctime
= CURRENT_TIME
;
247 dentry
= __d_alloc(s
, &d_name
);
252 d_instantiate(dentry
, root
);
255 s
->s_flags
|= MS_ACTIVE
;
256 return dget(s
->s_root
);
259 deactivate_locked_super(s
);
260 return ERR_PTR(-ENOMEM
);
262 EXPORT_SYMBOL(mount_pseudo
);
264 int simple_open(struct inode
*inode
, struct file
*file
)
266 if (inode
->i_private
)
267 file
->private_data
= inode
->i_private
;
270 EXPORT_SYMBOL(simple_open
);
272 int simple_link(struct dentry
*old_dentry
, struct inode
*dir
, struct dentry
*dentry
)
274 struct inode
*inode
= d_inode(old_dentry
);
276 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= CURRENT_TIME
;
280 d_instantiate(dentry
, inode
);
283 EXPORT_SYMBOL(simple_link
);
285 int simple_empty(struct dentry
*dentry
)
287 struct dentry
*child
;
290 spin_lock(&dentry
->d_lock
);
291 list_for_each_entry(child
, &dentry
->d_subdirs
, d_child
) {
292 spin_lock_nested(&child
->d_lock
, DENTRY_D_LOCK_NESTED
);
293 if (simple_positive(child
)) {
294 spin_unlock(&child
->d_lock
);
297 spin_unlock(&child
->d_lock
);
301 spin_unlock(&dentry
->d_lock
);
304 EXPORT_SYMBOL(simple_empty
);
306 int simple_unlink(struct inode
*dir
, struct dentry
*dentry
)
308 struct inode
*inode
= d_inode(dentry
);
310 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= CURRENT_TIME
;
315 EXPORT_SYMBOL(simple_unlink
);
317 int simple_rmdir(struct inode
*dir
, struct dentry
*dentry
)
319 if (!simple_empty(dentry
))
322 drop_nlink(d_inode(dentry
));
323 simple_unlink(dir
, dentry
);
327 EXPORT_SYMBOL(simple_rmdir
);
329 int simple_rename(struct inode
*old_dir
, struct dentry
*old_dentry
,
330 struct inode
*new_dir
, struct dentry
*new_dentry
)
332 struct inode
*inode
= d_inode(old_dentry
);
333 int they_are_dirs
= d_is_dir(old_dentry
);
335 if (!simple_empty(new_dentry
))
338 if (d_really_is_positive(new_dentry
)) {
339 simple_unlink(new_dir
, new_dentry
);
341 drop_nlink(d_inode(new_dentry
));
344 } else if (they_are_dirs
) {
349 old_dir
->i_ctime
= old_dir
->i_mtime
= new_dir
->i_ctime
=
350 new_dir
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
354 EXPORT_SYMBOL(simple_rename
);
357 * simple_setattr - setattr for simple filesystem
359 * @iattr: iattr structure
361 * Returns 0 on success, -error on failure.
363 * simple_setattr is a simple ->setattr implementation without a proper
364 * implementation of size changes.
366 * It can either be used for in-memory filesystems or special files
367 * on simple regular filesystems. Anything that needs to change on-disk
368 * or wire state on size changes needs its own setattr method.
370 int simple_setattr(struct dentry
*dentry
, struct iattr
*iattr
)
372 struct inode
*inode
= d_inode(dentry
);
375 error
= inode_change_ok(inode
, iattr
);
379 if (iattr
->ia_valid
& ATTR_SIZE
)
380 truncate_setsize(inode
, iattr
->ia_size
);
381 setattr_copy(inode
, iattr
);
382 mark_inode_dirty(inode
);
385 EXPORT_SYMBOL(simple_setattr
);
387 int simple_readpage(struct file
*file
, struct page
*page
)
389 clear_highpage(page
);
390 flush_dcache_page(page
);
391 SetPageUptodate(page
);
395 EXPORT_SYMBOL(simple_readpage
);
397 int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
398 loff_t pos
, unsigned len
, unsigned flags
,
399 struct page
**pagep
, void **fsdata
)
404 index
= pos
>> PAGE_SHIFT
;
406 page
= grab_cache_page_write_begin(mapping
, index
, flags
);
412 if (!PageUptodate(page
) && (len
!= PAGE_SIZE
)) {
413 unsigned from
= pos
& (PAGE_SIZE
- 1);
415 zero_user_segments(page
, 0, from
, from
+ len
, PAGE_SIZE
);
419 EXPORT_SYMBOL(simple_write_begin
);
422 * simple_write_end - .write_end helper for non-block-device FSes
423 * @available: See .write_end of address_space_operations
432 * simple_write_end does the minimum needed for updating a page after writing is
433 * done. It has the same API signature as the .write_end of
434 * address_space_operations vector. So it can just be set onto .write_end for
435 * FSes that don't need any other processing. i_mutex is assumed to be held.
436 * Block based filesystems should use generic_write_end().
437 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
438 * is not called, so a filesystem that actually does store data in .write_inode
439 * should extend on what's done here with a call to mark_inode_dirty() in the
440 * case that i_size has changed.
442 int simple_write_end(struct file
*file
, struct address_space
*mapping
,
443 loff_t pos
, unsigned len
, unsigned copied
,
444 struct page
*page
, void *fsdata
)
446 struct inode
*inode
= page
->mapping
->host
;
447 loff_t last_pos
= pos
+ copied
;
449 /* zero the stale part of the page if we did a short copy */
451 unsigned from
= pos
& (PAGE_SIZE
- 1);
453 zero_user(page
, from
+ copied
, len
- copied
);
456 if (!PageUptodate(page
))
457 SetPageUptodate(page
);
459 * No need to use i_size_read() here, the i_size
460 * cannot change under us because we hold the i_mutex.
462 if (last_pos
> inode
->i_size
)
463 i_size_write(inode
, last_pos
);
465 set_page_dirty(page
);
471 EXPORT_SYMBOL(simple_write_end
);
474 * the inodes created here are not hashed. If you use iunique to generate
475 * unique inode values later for this filesystem, then you must take care
476 * to pass it an appropriate max_reserved value to avoid collisions.
478 int simple_fill_super(struct super_block
*s
, unsigned long magic
,
479 struct tree_descr
*files
)
483 struct dentry
*dentry
;
486 s
->s_blocksize
= PAGE_SIZE
;
487 s
->s_blocksize_bits
= PAGE_SHIFT
;
489 s
->s_op
= &simple_super_operations
;
492 inode
= new_inode(s
);
496 * because the root inode is 1, the files array must not contain an
500 inode
->i_mode
= S_IFDIR
| 0755;
501 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
502 inode
->i_op
= &simple_dir_inode_operations
;
503 inode
->i_fop
= &simple_dir_operations
;
505 root
= d_make_root(inode
);
508 for (i
= 0; !files
->name
|| files
->name
[0]; i
++, files
++) {
512 /* warn if it tries to conflict with the root inode */
513 if (unlikely(i
== 1))
514 printk(KERN_WARNING
"%s: %s passed in a files array"
515 "with an index of 1!\n", __func__
,
518 dentry
= d_alloc_name(root
, files
->name
);
521 inode
= new_inode(s
);
526 inode
->i_mode
= S_IFREG
| files
->mode
;
527 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
528 inode
->i_fop
= files
->ops
;
530 d_add(dentry
, inode
);
536 shrink_dcache_parent(root
);
540 EXPORT_SYMBOL(simple_fill_super
);
542 static DEFINE_SPINLOCK(pin_fs_lock
);
544 int simple_pin_fs(struct file_system_type
*type
, struct vfsmount
**mount
, int *count
)
546 struct vfsmount
*mnt
= NULL
;
547 spin_lock(&pin_fs_lock
);
548 if (unlikely(!*mount
)) {
549 spin_unlock(&pin_fs_lock
);
550 mnt
= vfs_kern_mount(type
, MS_KERNMOUNT
, type
->name
, NULL
);
553 spin_lock(&pin_fs_lock
);
559 spin_unlock(&pin_fs_lock
);
563 EXPORT_SYMBOL(simple_pin_fs
);
565 void simple_release_fs(struct vfsmount
**mount
, int *count
)
567 struct vfsmount
*mnt
;
568 spin_lock(&pin_fs_lock
);
572 spin_unlock(&pin_fs_lock
);
575 EXPORT_SYMBOL(simple_release_fs
);
578 * simple_read_from_buffer - copy data from the buffer to user space
579 * @to: the user space buffer to read to
580 * @count: the maximum number of bytes to read
581 * @ppos: the current position in the buffer
582 * @from: the buffer to read from
583 * @available: the size of the buffer
585 * The simple_read_from_buffer() function reads up to @count bytes from the
586 * buffer @from at offset @ppos into the user space address starting at @to.
588 * On success, the number of bytes read is returned and the offset @ppos is
589 * advanced by this number, or negative value is returned on error.
591 ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
, loff_t
*ppos
,
592 const void *from
, size_t available
)
599 if (pos
>= available
|| !count
)
601 if (count
> available
- pos
)
602 count
= available
- pos
;
603 ret
= copy_to_user(to
, from
+ pos
, count
);
610 EXPORT_SYMBOL(simple_read_from_buffer
);
613 * simple_write_to_buffer - copy data from user space to the buffer
614 * @to: the buffer to write to
615 * @available: the size of the buffer
616 * @ppos: the current position in the buffer
617 * @from: the user space buffer to read from
618 * @count: the maximum number of bytes to read
620 * The simple_write_to_buffer() function reads up to @count bytes from the user
621 * space address starting at @from into the buffer @to at offset @ppos.
623 * On success, the number of bytes written is returned and the offset @ppos is
624 * advanced by this number, or negative value is returned on error.
626 ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
627 const void __user
*from
, size_t count
)
634 if (pos
>= available
|| !count
)
636 if (count
> available
- pos
)
637 count
= available
- pos
;
638 res
= copy_from_user(to
+ pos
, from
, count
);
645 EXPORT_SYMBOL(simple_write_to_buffer
);
648 * memory_read_from_buffer - copy data from the buffer
649 * @to: the kernel space buffer to read to
650 * @count: the maximum number of bytes to read
651 * @ppos: the current position in the buffer
652 * @from: the buffer to read from
653 * @available: the size of the buffer
655 * The memory_read_from_buffer() function reads up to @count bytes from the
656 * buffer @from at offset @ppos into the kernel space address starting at @to.
658 * On success, the number of bytes read is returned and the offset @ppos is
659 * advanced by this number, or negative value is returned on error.
661 ssize_t
memory_read_from_buffer(void *to
, size_t count
, loff_t
*ppos
,
662 const void *from
, size_t available
)
668 if (pos
>= available
)
670 if (count
> available
- pos
)
671 count
= available
- pos
;
672 memcpy(to
, from
+ pos
, count
);
677 EXPORT_SYMBOL(memory_read_from_buffer
);
680 * Transaction based IO.
681 * The file expects a single write which triggers the transaction, and then
682 * possibly a read which collects the result - which is stored in a
686 void simple_transaction_set(struct file
*file
, size_t n
)
688 struct simple_transaction_argresp
*ar
= file
->private_data
;
690 BUG_ON(n
> SIMPLE_TRANSACTION_LIMIT
);
693 * The barrier ensures that ar->size will really remain zero until
694 * ar->data is ready for reading.
699 EXPORT_SYMBOL(simple_transaction_set
);
701 char *simple_transaction_get(struct file
*file
, const char __user
*buf
, size_t size
)
703 struct simple_transaction_argresp
*ar
;
704 static DEFINE_SPINLOCK(simple_transaction_lock
);
706 if (size
> SIMPLE_TRANSACTION_LIMIT
- 1)
707 return ERR_PTR(-EFBIG
);
709 ar
= (struct simple_transaction_argresp
*)get_zeroed_page(GFP_KERNEL
);
711 return ERR_PTR(-ENOMEM
);
713 spin_lock(&simple_transaction_lock
);
715 /* only one write allowed per open */
716 if (file
->private_data
) {
717 spin_unlock(&simple_transaction_lock
);
718 free_page((unsigned long)ar
);
719 return ERR_PTR(-EBUSY
);
722 file
->private_data
= ar
;
724 spin_unlock(&simple_transaction_lock
);
726 if (copy_from_user(ar
->data
, buf
, size
))
727 return ERR_PTR(-EFAULT
);
731 EXPORT_SYMBOL(simple_transaction_get
);
733 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
, size_t size
, loff_t
*pos
)
735 struct simple_transaction_argresp
*ar
= file
->private_data
;
739 return simple_read_from_buffer(buf
, size
, pos
, ar
->data
, ar
->size
);
741 EXPORT_SYMBOL(simple_transaction_read
);
743 int simple_transaction_release(struct inode
*inode
, struct file
*file
)
745 free_page((unsigned long)file
->private_data
);
748 EXPORT_SYMBOL(simple_transaction_release
);
750 /* Simple attribute files */
753 int (*get
)(void *, u64
*);
754 int (*set
)(void *, u64
);
755 char get_buf
[24]; /* enough to store a u64 and "\n\0" */
758 const char *fmt
; /* format for read operation */
759 struct mutex mutex
; /* protects access to these buffers */
762 /* simple_attr_open is called by an actual attribute open file operation
763 * to set the attribute specific access operations. */
764 int simple_attr_open(struct inode
*inode
, struct file
*file
,
765 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
768 struct simple_attr
*attr
;
770 attr
= kmalloc(sizeof(*attr
), GFP_KERNEL
);
776 attr
->data
= inode
->i_private
;
778 mutex_init(&attr
->mutex
);
780 file
->private_data
= attr
;
782 return nonseekable_open(inode
, file
);
784 EXPORT_SYMBOL_GPL(simple_attr_open
);
786 int simple_attr_release(struct inode
*inode
, struct file
*file
)
788 kfree(file
->private_data
);
791 EXPORT_SYMBOL_GPL(simple_attr_release
); /* GPL-only? This? Really? */
793 /* read from the buffer that is filled with the get function */
794 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
795 size_t len
, loff_t
*ppos
)
797 struct simple_attr
*attr
;
801 attr
= file
->private_data
;
806 ret
= mutex_lock_interruptible(&attr
->mutex
);
810 if (*ppos
) { /* continued read */
811 size
= strlen(attr
->get_buf
);
812 } else { /* first read */
814 ret
= attr
->get(attr
->data
, &val
);
818 size
= scnprintf(attr
->get_buf
, sizeof(attr
->get_buf
),
819 attr
->fmt
, (unsigned long long)val
);
822 ret
= simple_read_from_buffer(buf
, len
, ppos
, attr
->get_buf
, size
);
824 mutex_unlock(&attr
->mutex
);
827 EXPORT_SYMBOL_GPL(simple_attr_read
);
829 /* interpret the buffer as a number to call the set function with */
830 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
831 size_t len
, loff_t
*ppos
)
833 struct simple_attr
*attr
;
838 attr
= file
->private_data
;
842 ret
= mutex_lock_interruptible(&attr
->mutex
);
847 size
= min(sizeof(attr
->set_buf
) - 1, len
);
848 if (copy_from_user(attr
->set_buf
, buf
, size
))
851 attr
->set_buf
[size
] = '\0';
852 val
= simple_strtoll(attr
->set_buf
, NULL
, 0);
853 ret
= attr
->set(attr
->data
, val
);
855 ret
= len
; /* on success, claim we got the whole input */
857 mutex_unlock(&attr
->mutex
);
860 EXPORT_SYMBOL_GPL(simple_attr_write
);
863 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
864 * @sb: filesystem to do the file handle conversion on
865 * @fid: file handle to convert
866 * @fh_len: length of the file handle in bytes
867 * @fh_type: type of file handle
868 * @get_inode: filesystem callback to retrieve inode
870 * This function decodes @fid as long as it has one of the well-known
871 * Linux filehandle types and calls @get_inode on it to retrieve the
872 * inode for the object specified in the file handle.
874 struct dentry
*generic_fh_to_dentry(struct super_block
*sb
, struct fid
*fid
,
875 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
876 (struct super_block
*sb
, u64 ino
, u32 gen
))
878 struct inode
*inode
= NULL
;
884 case FILEID_INO32_GEN
:
885 case FILEID_INO32_GEN_PARENT
:
886 inode
= get_inode(sb
, fid
->i32
.ino
, fid
->i32
.gen
);
890 return d_obtain_alias(inode
);
892 EXPORT_SYMBOL_GPL(generic_fh_to_dentry
);
895 * generic_fh_to_parent - generic helper for the fh_to_parent export operation
896 * @sb: filesystem to do the file handle conversion on
897 * @fid: file handle to convert
898 * @fh_len: length of the file handle in bytes
899 * @fh_type: type of file handle
900 * @get_inode: filesystem callback to retrieve inode
902 * This function decodes @fid as long as it has one of the well-known
903 * Linux filehandle types and calls @get_inode on it to retrieve the
904 * inode for the _parent_ object specified in the file handle if it
905 * is specified in the file handle, or NULL otherwise.
907 struct dentry
*generic_fh_to_parent(struct super_block
*sb
, struct fid
*fid
,
908 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
909 (struct super_block
*sb
, u64 ino
, u32 gen
))
911 struct inode
*inode
= NULL
;
917 case FILEID_INO32_GEN_PARENT
:
918 inode
= get_inode(sb
, fid
->i32
.parent_ino
,
919 (fh_len
> 3 ? fid
->i32
.parent_gen
: 0));
923 return d_obtain_alias(inode
);
925 EXPORT_SYMBOL_GPL(generic_fh_to_parent
);
928 * __generic_file_fsync - generic fsync implementation for simple filesystems
930 * @file: file to synchronize
931 * @start: start offset in bytes
932 * @end: end offset in bytes (inclusive)
933 * @datasync: only synchronize essential metadata if true
935 * This is a generic implementation of the fsync method for simple
936 * filesystems which track all non-inode metadata in the buffers list
937 * hanging off the address_space structure.
939 int __generic_file_fsync(struct file
*file
, loff_t start
, loff_t end
,
942 struct inode
*inode
= file
->f_mapping
->host
;
946 err
= filemap_write_and_wait_range(inode
->i_mapping
, start
, end
);
951 ret
= sync_mapping_buffers(inode
->i_mapping
);
952 if (!(inode
->i_state
& I_DIRTY_ALL
))
954 if (datasync
&& !(inode
->i_state
& I_DIRTY_DATASYNC
))
957 err
= sync_inode_metadata(inode
, 1);
965 EXPORT_SYMBOL(__generic_file_fsync
);
968 * generic_file_fsync - generic fsync implementation for simple filesystems
970 * @file: file to synchronize
971 * @start: start offset in bytes
972 * @end: end offset in bytes (inclusive)
973 * @datasync: only synchronize essential metadata if true
977 int generic_file_fsync(struct file
*file
, loff_t start
, loff_t end
,
980 struct inode
*inode
= file
->f_mapping
->host
;
983 err
= __generic_file_fsync(file
, start
, end
, datasync
);
986 return blkdev_issue_flush(inode
->i_sb
->s_bdev
, GFP_KERNEL
, NULL
);
988 EXPORT_SYMBOL(generic_file_fsync
);
991 * generic_check_addressable - Check addressability of file system
992 * @blocksize_bits: log of file system block size
993 * @num_blocks: number of blocks in file system
995 * Determine whether a file system with @num_blocks blocks (and a
996 * block size of 2**@blocksize_bits) is addressable by the sector_t
997 * and page cache of the system. Return 0 if so and -EFBIG otherwise.
999 int generic_check_addressable(unsigned blocksize_bits
, u64 num_blocks
)
1001 u64 last_fs_block
= num_blocks
- 1;
1003 last_fs_block
>> (PAGE_SHIFT
- blocksize_bits
);
1005 if (unlikely(num_blocks
== 0))
1008 if ((blocksize_bits
< 9) || (blocksize_bits
> PAGE_SHIFT
))
1011 if ((last_fs_block
> (sector_t
)(~0ULL) >> (blocksize_bits
- 9)) ||
1012 (last_fs_page
> (pgoff_t
)(~0ULL))) {
1017 EXPORT_SYMBOL(generic_check_addressable
);
1020 * No-op implementation of ->fsync for in-memory filesystems.
1022 int noop_fsync(struct file
*file
, loff_t start
, loff_t end
, int datasync
)
1026 EXPORT_SYMBOL(noop_fsync
);
1028 /* Because kfree isn't assignment-compatible with void(void*) ;-/ */
1029 void kfree_link(void *p
)
1033 EXPORT_SYMBOL(kfree_link
);
1036 * nop .set_page_dirty method so that people can use .page_mkwrite on
1039 static int anon_set_page_dirty(struct page
*page
)
1045 * A single inode exists for all anon_inode files. Contrary to pipes,
1046 * anon_inode inodes have no associated per-instance data, so we need
1047 * only allocate one of them.
1049 struct inode
*alloc_anon_inode(struct super_block
*s
)
1051 static const struct address_space_operations anon_aops
= {
1052 .set_page_dirty
= anon_set_page_dirty
,
1054 struct inode
*inode
= new_inode_pseudo(s
);
1057 return ERR_PTR(-ENOMEM
);
1059 inode
->i_ino
= get_next_ino();
1060 inode
->i_mapping
->a_ops
= &anon_aops
;
1063 * Mark the inode dirty from the very beginning,
1064 * that way it will never be moved to the dirty
1065 * list because mark_inode_dirty() will think
1066 * that it already _is_ on the dirty list.
1068 inode
->i_state
= I_DIRTY
;
1069 inode
->i_mode
= S_IRUSR
| S_IWUSR
;
1070 inode
->i_uid
= current_fsuid();
1071 inode
->i_gid
= current_fsgid();
1072 inode
->i_flags
|= S_PRIVATE
;
1073 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
1076 EXPORT_SYMBOL(alloc_anon_inode
);
1079 * simple_nosetlease - generic helper for prohibiting leases
1080 * @filp: file pointer
1081 * @arg: type of lease to obtain
1082 * @flp: new lease supplied for insertion
1083 * @priv: private data for lm_setup operation
1085 * Generic helper for filesystems that do not wish to allow leases to be set.
1086 * All arguments are ignored and it just returns -EINVAL.
1089 simple_nosetlease(struct file
*filp
, long arg
, struct file_lock
**flp
,
1094 EXPORT_SYMBOL(simple_nosetlease
);
1096 const char *simple_get_link(struct dentry
*dentry
, struct inode
*inode
,
1097 struct delayed_call
*done
)
1099 return inode
->i_link
;
1101 EXPORT_SYMBOL(simple_get_link
);
1103 const struct inode_operations simple_symlink_inode_operations
= {
1104 .get_link
= simple_get_link
,
1105 .readlink
= generic_readlink
1107 EXPORT_SYMBOL(simple_symlink_inode_operations
);
1110 * Operations for a permanently empty directory.
1112 static struct dentry
*empty_dir_lookup(struct inode
*dir
, struct dentry
*dentry
, unsigned int flags
)
1114 return ERR_PTR(-ENOENT
);
1117 static int empty_dir_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1120 struct inode
*inode
= d_inode(dentry
);
1121 generic_fillattr(inode
, stat
);
1125 static int empty_dir_setattr(struct dentry
*dentry
, struct iattr
*attr
)
1130 static int empty_dir_setxattr(struct dentry
*dentry
, struct inode
*inode
,
1131 const char *name
, const void *value
,
1132 size_t size
, int flags
)
1137 static ssize_t
empty_dir_getxattr(struct dentry
*dentry
, struct inode
*inode
,
1138 const char *name
, void *value
, size_t size
)
1143 static int empty_dir_removexattr(struct dentry
*dentry
, const char *name
)
1148 static ssize_t
empty_dir_listxattr(struct dentry
*dentry
, char *list
, size_t size
)
1153 static const struct inode_operations empty_dir_inode_operations
= {
1154 .lookup
= empty_dir_lookup
,
1155 .permission
= generic_permission
,
1156 .setattr
= empty_dir_setattr
,
1157 .getattr
= empty_dir_getattr
,
1158 .setxattr
= empty_dir_setxattr
,
1159 .getxattr
= empty_dir_getxattr
,
1160 .removexattr
= empty_dir_removexattr
,
1161 .listxattr
= empty_dir_listxattr
,
1164 static loff_t
empty_dir_llseek(struct file
*file
, loff_t offset
, int whence
)
1166 /* An empty directory has two entries . and .. at offsets 0 and 1 */
1167 return generic_file_llseek_size(file
, offset
, whence
, 2, 2);
1170 static int empty_dir_readdir(struct file
*file
, struct dir_context
*ctx
)
1172 dir_emit_dots(file
, ctx
);
1176 static const struct file_operations empty_dir_operations
= {
1177 .llseek
= empty_dir_llseek
,
1178 .read
= generic_read_dir
,
1179 .iterate_shared
= empty_dir_readdir
,
1180 .fsync
= noop_fsync
,
1184 void make_empty_dir_inode(struct inode
*inode
)
1186 set_nlink(inode
, 2);
1187 inode
->i_mode
= S_IFDIR
| S_IRUGO
| S_IXUGO
;
1188 inode
->i_uid
= GLOBAL_ROOT_UID
;
1189 inode
->i_gid
= GLOBAL_ROOT_GID
;
1192 inode
->i_blkbits
= PAGE_SHIFT
;
1193 inode
->i_blocks
= 0;
1195 inode
->i_op
= &empty_dir_inode_operations
;
1196 inode
->i_fop
= &empty_dir_operations
;
1199 bool is_empty_dir_inode(struct inode
*inode
)
1201 return (inode
->i_fop
== &empty_dir_operations
) &&
1202 (inode
->i_op
== &empty_dir_inode_operations
);