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/cred.h>
11 #include <linux/mount.h>
12 #include <linux/vfs.h>
13 #include <linux/quotaops.h>
14 #include <linux/mutex.h>
15 #include <linux/namei.h>
16 #include <linux/exportfs.h>
17 #include <linux/writeback.h>
18 #include <linux/buffer_head.h> /* sync_mapping_buffers */
20 #include <linux/uaccess.h>
24 int simple_getattr(const struct path
*path
, struct kstat
*stat
,
25 u32 request_mask
, unsigned int query_flags
)
27 struct inode
*inode
= d_inode(path
->dentry
);
28 generic_fillattr(inode
, stat
);
29 stat
->blocks
= inode
->i_mapping
->nrpages
<< (PAGE_SHIFT
- 9);
32 EXPORT_SYMBOL(simple_getattr
);
34 int simple_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
36 buf
->f_type
= dentry
->d_sb
->s_magic
;
37 buf
->f_bsize
= PAGE_SIZE
;
38 buf
->f_namelen
= NAME_MAX
;
41 EXPORT_SYMBOL(simple_statfs
);
44 * Retaining negative dentries for an in-memory filesystem just wastes
45 * memory and lookup time: arrange for them to be deleted immediately.
47 int always_delete_dentry(const struct dentry
*dentry
)
51 EXPORT_SYMBOL(always_delete_dentry
);
53 const struct dentry_operations simple_dentry_operations
= {
54 .d_delete
= always_delete_dentry
,
56 EXPORT_SYMBOL(simple_dentry_operations
);
59 * Lookup the data. This is trivial - if the dentry didn't already
60 * exist, we know it is negative. Set d_op to delete negative dentries.
62 struct dentry
*simple_lookup(struct inode
*dir
, struct dentry
*dentry
, unsigned int flags
)
64 if (dentry
->d_name
.len
> NAME_MAX
)
65 return ERR_PTR(-ENAMETOOLONG
);
66 if (!dentry
->d_sb
->s_d_op
)
67 d_set_d_op(dentry
, &simple_dentry_operations
);
71 EXPORT_SYMBOL(simple_lookup
);
73 int dcache_dir_open(struct inode
*inode
, struct file
*file
)
75 file
->private_data
= d_alloc_cursor(file
->f_path
.dentry
);
77 return file
->private_data
? 0 : -ENOMEM
;
79 EXPORT_SYMBOL(dcache_dir_open
);
81 int dcache_dir_close(struct inode
*inode
, struct file
*file
)
83 dput(file
->private_data
);
86 EXPORT_SYMBOL(dcache_dir_close
);
88 /* parent is locked at least shared */
89 static struct dentry
*next_positive(struct dentry
*parent
,
90 struct list_head
*from
,
93 unsigned *seq
= &parent
->d_inode
->i_dir_seq
, n
;
102 n
= smp_load_acquire(seq
) & ~1;
105 for (p
= from
->next
; p
!= &parent
->d_subdirs
; p
= p
->next
) {
106 struct dentry
*d
= list_entry(p
, struct dentry
, d_child
);
107 if (!simple_positive(d
)) {
117 if (unlikely(*seq
!= n
))
123 static void move_cursor(struct dentry
*cursor
, struct list_head
*after
)
125 struct dentry
*parent
= cursor
->d_parent
;
126 unsigned n
, *seq
= &parent
->d_inode
->i_dir_seq
;
127 spin_lock(&parent
->d_lock
);
130 if (!(n
& 1) && cmpxchg(seq
, n
, n
+ 1) == n
)
134 __list_del(cursor
->d_child
.prev
, cursor
->d_child
.next
);
136 list_add(&cursor
->d_child
, after
);
138 list_add_tail(&cursor
->d_child
, &parent
->d_subdirs
);
139 smp_store_release(seq
, n
+ 2);
140 spin_unlock(&parent
->d_lock
);
143 loff_t
dcache_dir_lseek(struct file
*file
, loff_t offset
, int whence
)
145 struct dentry
*dentry
= file
->f_path
.dentry
;
148 offset
+= file
->f_pos
;
155 if (offset
!= file
->f_pos
) {
156 file
->f_pos
= offset
;
157 if (file
->f_pos
>= 2) {
158 struct dentry
*cursor
= file
->private_data
;
160 loff_t n
= file
->f_pos
- 2;
162 inode_lock_shared(dentry
->d_inode
);
163 to
= next_positive(dentry
, &dentry
->d_subdirs
, n
);
164 move_cursor(cursor
, to
? &to
->d_child
: NULL
);
165 inode_unlock_shared(dentry
->d_inode
);
170 EXPORT_SYMBOL(dcache_dir_lseek
);
172 /* Relationship between i_mode and the DT_xxx types */
173 static inline unsigned char dt_type(struct inode
*inode
)
175 return (inode
->i_mode
>> 12) & 15;
179 * Directory is locked and all positive dentries in it are safe, since
180 * for ramfs-type trees they can't go away without unlink() or rmdir(),
181 * both impossible due to the lock on directory.
184 int dcache_readdir(struct file
*file
, struct dir_context
*ctx
)
186 struct dentry
*dentry
= file
->f_path
.dentry
;
187 struct dentry
*cursor
= file
->private_data
;
188 struct list_head
*p
= &cursor
->d_child
;
192 if (!dir_emit_dots(file
, ctx
))
196 p
= &dentry
->d_subdirs
;
197 while ((next
= next_positive(dentry
, p
, 1)) != NULL
) {
198 if (!dir_emit(ctx
, next
->d_name
.name
, next
->d_name
.len
,
199 d_inode(next
)->i_ino
, dt_type(d_inode(next
))))
206 move_cursor(cursor
, p
);
209 EXPORT_SYMBOL(dcache_readdir
);
211 ssize_t
generic_read_dir(struct file
*filp
, char __user
*buf
, size_t siz
, loff_t
*ppos
)
215 EXPORT_SYMBOL(generic_read_dir
);
217 const struct file_operations simple_dir_operations
= {
218 .open
= dcache_dir_open
,
219 .release
= dcache_dir_close
,
220 .llseek
= dcache_dir_lseek
,
221 .read
= generic_read_dir
,
222 .iterate_shared
= dcache_readdir
,
225 EXPORT_SYMBOL(simple_dir_operations
);
227 const struct inode_operations simple_dir_inode_operations
= {
228 .lookup
= simple_lookup
,
230 EXPORT_SYMBOL(simple_dir_inode_operations
);
232 static const struct super_operations simple_super_operations
= {
233 .statfs
= simple_statfs
,
237 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
238 * will never be mountable)
240 struct dentry
*mount_pseudo_xattr(struct file_system_type
*fs_type
, char *name
,
241 const struct super_operations
*ops
, const struct xattr_handler
**xattr
,
242 const struct dentry_operations
*dops
, unsigned long magic
)
244 struct super_block
*s
;
245 struct dentry
*dentry
;
247 struct qstr d_name
= QSTR_INIT(name
, strlen(name
));
249 s
= sget_userns(fs_type
, NULL
, set_anon_super
, MS_KERNMOUNT
|MS_NOUSER
,
250 &init_user_ns
, NULL
);
254 s
->s_maxbytes
= MAX_LFS_FILESIZE
;
255 s
->s_blocksize
= PAGE_SIZE
;
256 s
->s_blocksize_bits
= PAGE_SHIFT
;
258 s
->s_op
= ops
? ops
: &simple_super_operations
;
265 * since this is the first inode, make it number 1. New inodes created
266 * after this must take care not to collide with it (by passing
267 * max_reserved of 1 to iunique).
270 root
->i_mode
= S_IFDIR
| S_IRUSR
| S_IWUSR
;
271 root
->i_atime
= root
->i_mtime
= root
->i_ctime
= current_time(root
);
272 dentry
= __d_alloc(s
, &d_name
);
277 d_instantiate(dentry
, root
);
280 s
->s_flags
|= MS_ACTIVE
;
281 return dget(s
->s_root
);
284 deactivate_locked_super(s
);
285 return ERR_PTR(-ENOMEM
);
287 EXPORT_SYMBOL(mount_pseudo_xattr
);
289 int simple_open(struct inode
*inode
, struct file
*file
)
291 if (inode
->i_private
)
292 file
->private_data
= inode
->i_private
;
295 EXPORT_SYMBOL(simple_open
);
297 int simple_link(struct dentry
*old_dentry
, struct inode
*dir
, struct dentry
*dentry
)
299 struct inode
*inode
= d_inode(old_dentry
);
301 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= current_time(inode
);
305 d_instantiate(dentry
, inode
);
308 EXPORT_SYMBOL(simple_link
);
310 int simple_empty(struct dentry
*dentry
)
312 struct dentry
*child
;
315 spin_lock(&dentry
->d_lock
);
316 list_for_each_entry(child
, &dentry
->d_subdirs
, d_child
) {
317 spin_lock_nested(&child
->d_lock
, DENTRY_D_LOCK_NESTED
);
318 if (simple_positive(child
)) {
319 spin_unlock(&child
->d_lock
);
322 spin_unlock(&child
->d_lock
);
326 spin_unlock(&dentry
->d_lock
);
329 EXPORT_SYMBOL(simple_empty
);
331 int simple_unlink(struct inode
*dir
, struct dentry
*dentry
)
333 struct inode
*inode
= d_inode(dentry
);
335 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= current_time(inode
);
340 EXPORT_SYMBOL(simple_unlink
);
342 int simple_rmdir(struct inode
*dir
, struct dentry
*dentry
)
344 if (!simple_empty(dentry
))
347 drop_nlink(d_inode(dentry
));
348 simple_unlink(dir
, dentry
);
352 EXPORT_SYMBOL(simple_rmdir
);
354 int simple_rename(struct inode
*old_dir
, struct dentry
*old_dentry
,
355 struct inode
*new_dir
, struct dentry
*new_dentry
,
358 struct inode
*inode
= d_inode(old_dentry
);
359 int they_are_dirs
= d_is_dir(old_dentry
);
361 if (flags
& ~RENAME_NOREPLACE
)
364 if (!simple_empty(new_dentry
))
367 if (d_really_is_positive(new_dentry
)) {
368 simple_unlink(new_dir
, new_dentry
);
370 drop_nlink(d_inode(new_dentry
));
373 } else if (they_are_dirs
) {
378 old_dir
->i_ctime
= old_dir
->i_mtime
= new_dir
->i_ctime
=
379 new_dir
->i_mtime
= inode
->i_ctime
= current_time(old_dir
);
383 EXPORT_SYMBOL(simple_rename
);
386 * simple_setattr - setattr for simple filesystem
388 * @iattr: iattr structure
390 * Returns 0 on success, -error on failure.
392 * simple_setattr is a simple ->setattr implementation without a proper
393 * implementation of size changes.
395 * It can either be used for in-memory filesystems or special files
396 * on simple regular filesystems. Anything that needs to change on-disk
397 * or wire state on size changes needs its own setattr method.
399 int simple_setattr(struct dentry
*dentry
, struct iattr
*iattr
)
401 struct inode
*inode
= d_inode(dentry
);
404 error
= setattr_prepare(dentry
, iattr
);
408 if (iattr
->ia_valid
& ATTR_SIZE
)
409 truncate_setsize(inode
, iattr
->ia_size
);
410 setattr_copy(inode
, iattr
);
411 mark_inode_dirty(inode
);
414 EXPORT_SYMBOL(simple_setattr
);
416 int simple_readpage(struct file
*file
, struct page
*page
)
418 clear_highpage(page
);
419 flush_dcache_page(page
);
420 SetPageUptodate(page
);
424 EXPORT_SYMBOL(simple_readpage
);
426 int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
427 loff_t pos
, unsigned len
, unsigned flags
,
428 struct page
**pagep
, void **fsdata
)
433 index
= pos
>> PAGE_SHIFT
;
435 page
= grab_cache_page_write_begin(mapping
, index
, flags
);
441 if (!PageUptodate(page
) && (len
!= PAGE_SIZE
)) {
442 unsigned from
= pos
& (PAGE_SIZE
- 1);
444 zero_user_segments(page
, 0, from
, from
+ len
, PAGE_SIZE
);
448 EXPORT_SYMBOL(simple_write_begin
);
451 * simple_write_end - .write_end helper for non-block-device FSes
452 * @available: See .write_end of address_space_operations
461 * simple_write_end does the minimum needed for updating a page after writing is
462 * done. It has the same API signature as the .write_end of
463 * address_space_operations vector. So it can just be set onto .write_end for
464 * FSes that don't need any other processing. i_mutex is assumed to be held.
465 * Block based filesystems should use generic_write_end().
466 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
467 * is not called, so a filesystem that actually does store data in .write_inode
468 * should extend on what's done here with a call to mark_inode_dirty() in the
469 * case that i_size has changed.
471 * Use *ONLY* with simple_readpage()
473 int simple_write_end(struct file
*file
, struct address_space
*mapping
,
474 loff_t pos
, unsigned len
, unsigned copied
,
475 struct page
*page
, void *fsdata
)
477 struct inode
*inode
= page
->mapping
->host
;
478 loff_t last_pos
= pos
+ copied
;
480 /* zero the stale part of the page if we did a short copy */
481 if (!PageUptodate(page
)) {
483 unsigned from
= pos
& (PAGE_SIZE
- 1);
485 zero_user(page
, from
+ copied
, len
- copied
);
487 SetPageUptodate(page
);
490 * No need to use i_size_read() here, the i_size
491 * cannot change under us because we hold the i_mutex.
493 if (last_pos
> inode
->i_size
)
494 i_size_write(inode
, last_pos
);
496 set_page_dirty(page
);
502 EXPORT_SYMBOL(simple_write_end
);
505 * the inodes created here are not hashed. If you use iunique to generate
506 * unique inode values later for this filesystem, then you must take care
507 * to pass it an appropriate max_reserved value to avoid collisions.
509 int simple_fill_super(struct super_block
*s
, unsigned long magic
,
510 struct tree_descr
*files
)
514 struct dentry
*dentry
;
517 s
->s_blocksize
= PAGE_SIZE
;
518 s
->s_blocksize_bits
= PAGE_SHIFT
;
520 s
->s_op
= &simple_super_operations
;
523 inode
= new_inode(s
);
527 * because the root inode is 1, the files array must not contain an
531 inode
->i_mode
= S_IFDIR
| 0755;
532 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= current_time(inode
);
533 inode
->i_op
= &simple_dir_inode_operations
;
534 inode
->i_fop
= &simple_dir_operations
;
536 root
= d_make_root(inode
);
539 for (i
= 0; !files
->name
|| files
->name
[0]; i
++, files
++) {
543 /* warn if it tries to conflict with the root inode */
544 if (unlikely(i
== 1))
545 printk(KERN_WARNING
"%s: %s passed in a files array"
546 "with an index of 1!\n", __func__
,
549 dentry
= d_alloc_name(root
, files
->name
);
552 inode
= new_inode(s
);
557 inode
->i_mode
= S_IFREG
| files
->mode
;
558 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= current_time(inode
);
559 inode
->i_fop
= files
->ops
;
561 d_add(dentry
, inode
);
567 shrink_dcache_parent(root
);
571 EXPORT_SYMBOL(simple_fill_super
);
573 static DEFINE_SPINLOCK(pin_fs_lock
);
575 int simple_pin_fs(struct file_system_type
*type
, struct vfsmount
**mount
, int *count
)
577 struct vfsmount
*mnt
= NULL
;
578 spin_lock(&pin_fs_lock
);
579 if (unlikely(!*mount
)) {
580 spin_unlock(&pin_fs_lock
);
581 mnt
= vfs_kern_mount(type
, MS_KERNMOUNT
, type
->name
, NULL
);
584 spin_lock(&pin_fs_lock
);
590 spin_unlock(&pin_fs_lock
);
594 EXPORT_SYMBOL(simple_pin_fs
);
596 void simple_release_fs(struct vfsmount
**mount
, int *count
)
598 struct vfsmount
*mnt
;
599 spin_lock(&pin_fs_lock
);
603 spin_unlock(&pin_fs_lock
);
606 EXPORT_SYMBOL(simple_release_fs
);
609 * simple_read_from_buffer - copy data from the buffer to user space
610 * @to: the user space buffer to read to
611 * @count: the maximum number of bytes to read
612 * @ppos: the current position in the buffer
613 * @from: the buffer to read from
614 * @available: the size of the buffer
616 * The simple_read_from_buffer() function reads up to @count bytes from the
617 * buffer @from at offset @ppos into the user space address starting at @to.
619 * On success, the number of bytes read is returned and the offset @ppos is
620 * advanced by this number, or negative value is returned on error.
622 ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
, loff_t
*ppos
,
623 const void *from
, size_t available
)
630 if (pos
>= available
|| !count
)
632 if (count
> available
- pos
)
633 count
= available
- pos
;
634 ret
= copy_to_user(to
, from
+ pos
, count
);
641 EXPORT_SYMBOL(simple_read_from_buffer
);
644 * simple_write_to_buffer - copy data from user space to the buffer
645 * @to: the buffer to write to
646 * @available: the size of the buffer
647 * @ppos: the current position in the buffer
648 * @from: the user space buffer to read from
649 * @count: the maximum number of bytes to read
651 * The simple_write_to_buffer() function reads up to @count bytes from the user
652 * space address starting at @from into the buffer @to at offset @ppos.
654 * On success, the number of bytes written is returned and the offset @ppos is
655 * advanced by this number, or negative value is returned on error.
657 ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
658 const void __user
*from
, size_t count
)
665 if (pos
>= available
|| !count
)
667 if (count
> available
- pos
)
668 count
= available
- pos
;
669 res
= copy_from_user(to
+ pos
, from
, count
);
676 EXPORT_SYMBOL(simple_write_to_buffer
);
679 * memory_read_from_buffer - copy data from the buffer
680 * @to: the kernel space buffer to read to
681 * @count: the maximum number of bytes to read
682 * @ppos: the current position in the buffer
683 * @from: the buffer to read from
684 * @available: the size of the buffer
686 * The memory_read_from_buffer() function reads up to @count bytes from the
687 * buffer @from at offset @ppos into the kernel space address starting at @to.
689 * On success, the number of bytes read is returned and the offset @ppos is
690 * advanced by this number, or negative value is returned on error.
692 ssize_t
memory_read_from_buffer(void *to
, size_t count
, loff_t
*ppos
,
693 const void *from
, size_t available
)
699 if (pos
>= available
)
701 if (count
> available
- pos
)
702 count
= available
- pos
;
703 memcpy(to
, from
+ pos
, count
);
708 EXPORT_SYMBOL(memory_read_from_buffer
);
711 * Transaction based IO.
712 * The file expects a single write which triggers the transaction, and then
713 * possibly a read which collects the result - which is stored in a
717 void simple_transaction_set(struct file
*file
, size_t n
)
719 struct simple_transaction_argresp
*ar
= file
->private_data
;
721 BUG_ON(n
> SIMPLE_TRANSACTION_LIMIT
);
724 * The barrier ensures that ar->size will really remain zero until
725 * ar->data is ready for reading.
730 EXPORT_SYMBOL(simple_transaction_set
);
732 char *simple_transaction_get(struct file
*file
, const char __user
*buf
, size_t size
)
734 struct simple_transaction_argresp
*ar
;
735 static DEFINE_SPINLOCK(simple_transaction_lock
);
737 if (size
> SIMPLE_TRANSACTION_LIMIT
- 1)
738 return ERR_PTR(-EFBIG
);
740 ar
= (struct simple_transaction_argresp
*)get_zeroed_page(GFP_KERNEL
);
742 return ERR_PTR(-ENOMEM
);
744 spin_lock(&simple_transaction_lock
);
746 /* only one write allowed per open */
747 if (file
->private_data
) {
748 spin_unlock(&simple_transaction_lock
);
749 free_page((unsigned long)ar
);
750 return ERR_PTR(-EBUSY
);
753 file
->private_data
= ar
;
755 spin_unlock(&simple_transaction_lock
);
757 if (copy_from_user(ar
->data
, buf
, size
))
758 return ERR_PTR(-EFAULT
);
762 EXPORT_SYMBOL(simple_transaction_get
);
764 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
, size_t size
, loff_t
*pos
)
766 struct simple_transaction_argresp
*ar
= file
->private_data
;
770 return simple_read_from_buffer(buf
, size
, pos
, ar
->data
, ar
->size
);
772 EXPORT_SYMBOL(simple_transaction_read
);
774 int simple_transaction_release(struct inode
*inode
, struct file
*file
)
776 free_page((unsigned long)file
->private_data
);
779 EXPORT_SYMBOL(simple_transaction_release
);
781 /* Simple attribute files */
784 int (*get
)(void *, u64
*);
785 int (*set
)(void *, u64
);
786 char get_buf
[24]; /* enough to store a u64 and "\n\0" */
789 const char *fmt
; /* format for read operation */
790 struct mutex mutex
; /* protects access to these buffers */
793 /* simple_attr_open is called by an actual attribute open file operation
794 * to set the attribute specific access operations. */
795 int simple_attr_open(struct inode
*inode
, struct file
*file
,
796 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
799 struct simple_attr
*attr
;
801 attr
= kmalloc(sizeof(*attr
), GFP_KERNEL
);
807 attr
->data
= inode
->i_private
;
809 mutex_init(&attr
->mutex
);
811 file
->private_data
= attr
;
813 return nonseekable_open(inode
, file
);
815 EXPORT_SYMBOL_GPL(simple_attr_open
);
817 int simple_attr_release(struct inode
*inode
, struct file
*file
)
819 kfree(file
->private_data
);
822 EXPORT_SYMBOL_GPL(simple_attr_release
); /* GPL-only? This? Really? */
824 /* read from the buffer that is filled with the get function */
825 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
826 size_t len
, loff_t
*ppos
)
828 struct simple_attr
*attr
;
832 attr
= file
->private_data
;
837 ret
= mutex_lock_interruptible(&attr
->mutex
);
841 if (*ppos
) { /* continued read */
842 size
= strlen(attr
->get_buf
);
843 } else { /* first read */
845 ret
= attr
->get(attr
->data
, &val
);
849 size
= scnprintf(attr
->get_buf
, sizeof(attr
->get_buf
),
850 attr
->fmt
, (unsigned long long)val
);
853 ret
= simple_read_from_buffer(buf
, len
, ppos
, attr
->get_buf
, size
);
855 mutex_unlock(&attr
->mutex
);
858 EXPORT_SYMBOL_GPL(simple_attr_read
);
860 /* interpret the buffer as a number to call the set function with */
861 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
862 size_t len
, loff_t
*ppos
)
864 struct simple_attr
*attr
;
869 attr
= file
->private_data
;
873 ret
= mutex_lock_interruptible(&attr
->mutex
);
878 size
= min(sizeof(attr
->set_buf
) - 1, len
);
879 if (copy_from_user(attr
->set_buf
, buf
, size
))
882 attr
->set_buf
[size
] = '\0';
883 val
= simple_strtoll(attr
->set_buf
, NULL
, 0);
884 ret
= attr
->set(attr
->data
, val
);
886 ret
= len
; /* on success, claim we got the whole input */
888 mutex_unlock(&attr
->mutex
);
891 EXPORT_SYMBOL_GPL(simple_attr_write
);
894 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
895 * @sb: filesystem to do the file handle conversion on
896 * @fid: file handle to convert
897 * @fh_len: length of the file handle in bytes
898 * @fh_type: type of file handle
899 * @get_inode: filesystem callback to retrieve inode
901 * This function decodes @fid as long as it has one of the well-known
902 * Linux filehandle types and calls @get_inode on it to retrieve the
903 * inode for the object specified in the file handle.
905 struct dentry
*generic_fh_to_dentry(struct super_block
*sb
, struct fid
*fid
,
906 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
907 (struct super_block
*sb
, u64 ino
, u32 gen
))
909 struct inode
*inode
= NULL
;
915 case FILEID_INO32_GEN
:
916 case FILEID_INO32_GEN_PARENT
:
917 inode
= get_inode(sb
, fid
->i32
.ino
, fid
->i32
.gen
);
921 return d_obtain_alias(inode
);
923 EXPORT_SYMBOL_GPL(generic_fh_to_dentry
);
926 * generic_fh_to_parent - generic helper for the fh_to_parent export operation
927 * @sb: filesystem to do the file handle conversion on
928 * @fid: file handle to convert
929 * @fh_len: length of the file handle in bytes
930 * @fh_type: type of file handle
931 * @get_inode: filesystem callback to retrieve inode
933 * This function decodes @fid as long as it has one of the well-known
934 * Linux filehandle types and calls @get_inode on it to retrieve the
935 * inode for the _parent_ object specified in the file handle if it
936 * is specified in the file handle, or NULL otherwise.
938 struct dentry
*generic_fh_to_parent(struct super_block
*sb
, struct fid
*fid
,
939 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
940 (struct super_block
*sb
, u64 ino
, u32 gen
))
942 struct inode
*inode
= NULL
;
948 case FILEID_INO32_GEN_PARENT
:
949 inode
= get_inode(sb
, fid
->i32
.parent_ino
,
950 (fh_len
> 3 ? fid
->i32
.parent_gen
: 0));
954 return d_obtain_alias(inode
);
956 EXPORT_SYMBOL_GPL(generic_fh_to_parent
);
959 * __generic_file_fsync - generic fsync implementation for simple filesystems
961 * @file: file to synchronize
962 * @start: start offset in bytes
963 * @end: end offset in bytes (inclusive)
964 * @datasync: only synchronize essential metadata if true
966 * This is a generic implementation of the fsync method for simple
967 * filesystems which track all non-inode metadata in the buffers list
968 * hanging off the address_space structure.
970 int __generic_file_fsync(struct file
*file
, loff_t start
, loff_t end
,
973 struct inode
*inode
= file
->f_mapping
->host
;
977 err
= filemap_write_and_wait_range(inode
->i_mapping
, start
, end
);
982 ret
= sync_mapping_buffers(inode
->i_mapping
);
983 if (!(inode
->i_state
& I_DIRTY_ALL
))
985 if (datasync
&& !(inode
->i_state
& I_DIRTY_DATASYNC
))
988 err
= sync_inode_metadata(inode
, 1);
996 EXPORT_SYMBOL(__generic_file_fsync
);
999 * generic_file_fsync - generic fsync implementation for simple filesystems
1001 * @file: file to synchronize
1002 * @start: start offset in bytes
1003 * @end: end offset in bytes (inclusive)
1004 * @datasync: only synchronize essential metadata if true
1008 int generic_file_fsync(struct file
*file
, loff_t start
, loff_t end
,
1011 struct inode
*inode
= file
->f_mapping
->host
;
1014 err
= __generic_file_fsync(file
, start
, end
, datasync
);
1017 return blkdev_issue_flush(inode
->i_sb
->s_bdev
, GFP_KERNEL
, NULL
);
1019 EXPORT_SYMBOL(generic_file_fsync
);
1022 * generic_check_addressable - Check addressability of file system
1023 * @blocksize_bits: log of file system block size
1024 * @num_blocks: number of blocks in file system
1026 * Determine whether a file system with @num_blocks blocks (and a
1027 * block size of 2**@blocksize_bits) is addressable by the sector_t
1028 * and page cache of the system. Return 0 if so and -EFBIG otherwise.
1030 int generic_check_addressable(unsigned blocksize_bits
, u64 num_blocks
)
1032 u64 last_fs_block
= num_blocks
- 1;
1034 last_fs_block
>> (PAGE_SHIFT
- blocksize_bits
);
1036 if (unlikely(num_blocks
== 0))
1039 if ((blocksize_bits
< 9) || (blocksize_bits
> PAGE_SHIFT
))
1042 if ((last_fs_block
> (sector_t
)(~0ULL) >> (blocksize_bits
- 9)) ||
1043 (last_fs_page
> (pgoff_t
)(~0ULL))) {
1048 EXPORT_SYMBOL(generic_check_addressable
);
1051 * No-op implementation of ->fsync for in-memory filesystems.
1053 int noop_fsync(struct file
*file
, loff_t start
, loff_t end
, int datasync
)
1057 EXPORT_SYMBOL(noop_fsync
);
1059 /* Because kfree isn't assignment-compatible with void(void*) ;-/ */
1060 void kfree_link(void *p
)
1064 EXPORT_SYMBOL(kfree_link
);
1067 * nop .set_page_dirty method so that people can use .page_mkwrite on
1070 static int anon_set_page_dirty(struct page
*page
)
1076 * A single inode exists for all anon_inode files. Contrary to pipes,
1077 * anon_inode inodes have no associated per-instance data, so we need
1078 * only allocate one of them.
1080 struct inode
*alloc_anon_inode(struct super_block
*s
)
1082 static const struct address_space_operations anon_aops
= {
1083 .set_page_dirty
= anon_set_page_dirty
,
1085 struct inode
*inode
= new_inode_pseudo(s
);
1088 return ERR_PTR(-ENOMEM
);
1090 inode
->i_ino
= get_next_ino();
1091 inode
->i_mapping
->a_ops
= &anon_aops
;
1094 * Mark the inode dirty from the very beginning,
1095 * that way it will never be moved to the dirty
1096 * list because mark_inode_dirty() will think
1097 * that it already _is_ on the dirty list.
1099 inode
->i_state
= I_DIRTY
;
1100 inode
->i_mode
= S_IRUSR
| S_IWUSR
;
1101 inode
->i_uid
= current_fsuid();
1102 inode
->i_gid
= current_fsgid();
1103 inode
->i_flags
|= S_PRIVATE
;
1104 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= current_time(inode
);
1107 EXPORT_SYMBOL(alloc_anon_inode
);
1110 * simple_nosetlease - generic helper for prohibiting leases
1111 * @filp: file pointer
1112 * @arg: type of lease to obtain
1113 * @flp: new lease supplied for insertion
1114 * @priv: private data for lm_setup operation
1116 * Generic helper for filesystems that do not wish to allow leases to be set.
1117 * All arguments are ignored and it just returns -EINVAL.
1120 simple_nosetlease(struct file
*filp
, long arg
, struct file_lock
**flp
,
1125 EXPORT_SYMBOL(simple_nosetlease
);
1127 const char *simple_get_link(struct dentry
*dentry
, struct inode
*inode
,
1128 struct delayed_call
*done
)
1130 return inode
->i_link
;
1132 EXPORT_SYMBOL(simple_get_link
);
1134 const struct inode_operations simple_symlink_inode_operations
= {
1135 .get_link
= simple_get_link
,
1137 EXPORT_SYMBOL(simple_symlink_inode_operations
);
1140 * Operations for a permanently empty directory.
1142 static struct dentry
*empty_dir_lookup(struct inode
*dir
, struct dentry
*dentry
, unsigned int flags
)
1144 return ERR_PTR(-ENOENT
);
1147 static int empty_dir_getattr(const struct path
*path
, struct kstat
*stat
,
1148 u32 request_mask
, unsigned int query_flags
)
1150 struct inode
*inode
= d_inode(path
->dentry
);
1151 generic_fillattr(inode
, stat
);
1155 static int empty_dir_setattr(struct dentry
*dentry
, struct iattr
*attr
)
1160 static ssize_t
empty_dir_listxattr(struct dentry
*dentry
, char *list
, size_t size
)
1165 static const struct inode_operations empty_dir_inode_operations
= {
1166 .lookup
= empty_dir_lookup
,
1167 .permission
= generic_permission
,
1168 .setattr
= empty_dir_setattr
,
1169 .getattr
= empty_dir_getattr
,
1170 .listxattr
= empty_dir_listxattr
,
1173 static loff_t
empty_dir_llseek(struct file
*file
, loff_t offset
, int whence
)
1175 /* An empty directory has two entries . and .. at offsets 0 and 1 */
1176 return generic_file_llseek_size(file
, offset
, whence
, 2, 2);
1179 static int empty_dir_readdir(struct file
*file
, struct dir_context
*ctx
)
1181 dir_emit_dots(file
, ctx
);
1185 static const struct file_operations empty_dir_operations
= {
1186 .llseek
= empty_dir_llseek
,
1187 .read
= generic_read_dir
,
1188 .iterate_shared
= empty_dir_readdir
,
1189 .fsync
= noop_fsync
,
1193 void make_empty_dir_inode(struct inode
*inode
)
1195 set_nlink(inode
, 2);
1196 inode
->i_mode
= S_IFDIR
| S_IRUGO
| S_IXUGO
;
1197 inode
->i_uid
= GLOBAL_ROOT_UID
;
1198 inode
->i_gid
= GLOBAL_ROOT_GID
;
1201 inode
->i_blkbits
= PAGE_SHIFT
;
1202 inode
->i_blocks
= 0;
1204 inode
->i_op
= &empty_dir_inode_operations
;
1205 inode
->i_opflags
&= ~IOP_XATTR
;
1206 inode
->i_fop
= &empty_dir_operations
;
1209 bool is_empty_dir_inode(struct inode
*inode
)
1211 return (inode
->i_fop
== &empty_dir_operations
) &&
1212 (inode
->i_op
== &empty_dir_inode_operations
);