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
;
157 if (offset
!= file
->f_pos
) {
158 file
->f_pos
= offset
;
159 if (file
->f_pos
>= 2) {
160 struct dentry
*cursor
= file
->private_data
;
162 loff_t n
= file
->f_pos
- 2;
164 inode_lock_shared(dentry
->d_inode
);
165 to
= next_positive(dentry
, &dentry
->d_subdirs
, n
);
166 move_cursor(cursor
, to
? &to
->d_child
: NULL
);
167 inode_unlock_shared(dentry
->d_inode
);
172 EXPORT_SYMBOL(dcache_dir_lseek
);
174 /* Relationship between i_mode and the DT_xxx types */
175 static inline unsigned char dt_type(struct inode
*inode
)
177 return (inode
->i_mode
>> 12) & 15;
181 * Directory is locked and all positive dentries in it are safe, since
182 * for ramfs-type trees they can't go away without unlink() or rmdir(),
183 * both impossible due to the lock on directory.
186 int dcache_readdir(struct file
*file
, struct dir_context
*ctx
)
188 struct dentry
*dentry
= file
->f_path
.dentry
;
189 struct dentry
*cursor
= file
->private_data
;
190 struct list_head
*p
= &cursor
->d_child
;
194 if (!dir_emit_dots(file
, ctx
))
198 p
= &dentry
->d_subdirs
;
199 while ((next
= next_positive(dentry
, p
, 1)) != NULL
) {
200 if (!dir_emit(ctx
, next
->d_name
.name
, next
->d_name
.len
,
201 d_inode(next
)->i_ino
, dt_type(d_inode(next
))))
208 move_cursor(cursor
, p
);
211 EXPORT_SYMBOL(dcache_readdir
);
213 ssize_t
generic_read_dir(struct file
*filp
, char __user
*buf
, size_t siz
, loff_t
*ppos
)
217 EXPORT_SYMBOL(generic_read_dir
);
219 const struct file_operations simple_dir_operations
= {
220 .open
= dcache_dir_open
,
221 .release
= dcache_dir_close
,
222 .llseek
= dcache_dir_lseek
,
223 .read
= generic_read_dir
,
224 .iterate_shared
= dcache_readdir
,
227 EXPORT_SYMBOL(simple_dir_operations
);
229 const struct inode_operations simple_dir_inode_operations
= {
230 .lookup
= simple_lookup
,
232 EXPORT_SYMBOL(simple_dir_inode_operations
);
234 static const struct super_operations simple_super_operations
= {
235 .statfs
= simple_statfs
,
239 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
240 * will never be mountable)
242 struct dentry
*mount_pseudo_xattr(struct file_system_type
*fs_type
,
243 const struct super_operations
*ops
, const struct xattr_handler
**xattr
,
244 const struct dentry_operations
*dops
, unsigned long magic
)
246 struct super_block
*s
;
249 s
= sget_userns(fs_type
, NULL
, set_anon_super
, SB_KERNMOUNT
|SB_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 s
->s_root
= d_make_root(root
);
276 s
->s_flags
|= SB_ACTIVE
;
277 return dget(s
->s_root
);
280 deactivate_locked_super(s
);
281 return ERR_PTR(-ENOMEM
);
283 EXPORT_SYMBOL(mount_pseudo_xattr
);
285 int simple_open(struct inode
*inode
, struct file
*file
)
287 if (inode
->i_private
)
288 file
->private_data
= inode
->i_private
;
291 EXPORT_SYMBOL(simple_open
);
293 int simple_link(struct dentry
*old_dentry
, struct inode
*dir
, struct dentry
*dentry
)
295 struct inode
*inode
= d_inode(old_dentry
);
297 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= current_time(inode
);
301 d_instantiate(dentry
, inode
);
304 EXPORT_SYMBOL(simple_link
);
306 int simple_empty(struct dentry
*dentry
)
308 struct dentry
*child
;
311 spin_lock(&dentry
->d_lock
);
312 list_for_each_entry(child
, &dentry
->d_subdirs
, d_child
) {
313 spin_lock_nested(&child
->d_lock
, DENTRY_D_LOCK_NESTED
);
314 if (simple_positive(child
)) {
315 spin_unlock(&child
->d_lock
);
318 spin_unlock(&child
->d_lock
);
322 spin_unlock(&dentry
->d_lock
);
325 EXPORT_SYMBOL(simple_empty
);
327 int simple_unlink(struct inode
*dir
, struct dentry
*dentry
)
329 struct inode
*inode
= d_inode(dentry
);
331 inode
->i_ctime
= dir
->i_ctime
= dir
->i_mtime
= current_time(inode
);
336 EXPORT_SYMBOL(simple_unlink
);
338 int simple_rmdir(struct inode
*dir
, struct dentry
*dentry
)
340 if (!simple_empty(dentry
))
343 drop_nlink(d_inode(dentry
));
344 simple_unlink(dir
, dentry
);
348 EXPORT_SYMBOL(simple_rmdir
);
350 int simple_rename(struct inode
*old_dir
, struct dentry
*old_dentry
,
351 struct inode
*new_dir
, struct dentry
*new_dentry
,
354 struct inode
*inode
= d_inode(old_dentry
);
355 int they_are_dirs
= d_is_dir(old_dentry
);
357 if (flags
& ~RENAME_NOREPLACE
)
360 if (!simple_empty(new_dentry
))
363 if (d_really_is_positive(new_dentry
)) {
364 simple_unlink(new_dir
, new_dentry
);
366 drop_nlink(d_inode(new_dentry
));
369 } else if (they_are_dirs
) {
374 old_dir
->i_ctime
= old_dir
->i_mtime
= new_dir
->i_ctime
=
375 new_dir
->i_mtime
= inode
->i_ctime
= current_time(old_dir
);
379 EXPORT_SYMBOL(simple_rename
);
382 * simple_setattr - setattr for simple filesystem
384 * @iattr: iattr structure
386 * Returns 0 on success, -error on failure.
388 * simple_setattr is a simple ->setattr implementation without a proper
389 * implementation of size changes.
391 * It can either be used for in-memory filesystems or special files
392 * on simple regular filesystems. Anything that needs to change on-disk
393 * or wire state on size changes needs its own setattr method.
395 int simple_setattr(struct dentry
*dentry
, struct iattr
*iattr
)
397 struct inode
*inode
= d_inode(dentry
);
400 error
= setattr_prepare(dentry
, iattr
);
404 if (iattr
->ia_valid
& ATTR_SIZE
)
405 truncate_setsize(inode
, iattr
->ia_size
);
406 setattr_copy(inode
, iattr
);
407 mark_inode_dirty(inode
);
410 EXPORT_SYMBOL(simple_setattr
);
412 int simple_readpage(struct file
*file
, struct page
*page
)
414 clear_highpage(page
);
415 flush_dcache_page(page
);
416 SetPageUptodate(page
);
420 EXPORT_SYMBOL(simple_readpage
);
422 int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
423 loff_t pos
, unsigned len
, unsigned flags
,
424 struct page
**pagep
, void **fsdata
)
429 index
= pos
>> PAGE_SHIFT
;
431 page
= grab_cache_page_write_begin(mapping
, index
, flags
);
437 if (!PageUptodate(page
) && (len
!= PAGE_SIZE
)) {
438 unsigned from
= pos
& (PAGE_SIZE
- 1);
440 zero_user_segments(page
, 0, from
, from
+ len
, PAGE_SIZE
);
444 EXPORT_SYMBOL(simple_write_begin
);
447 * simple_write_end - .write_end helper for non-block-device FSes
448 * @available: See .write_end of address_space_operations
457 * simple_write_end does the minimum needed for updating a page after writing is
458 * done. It has the same API signature as the .write_end of
459 * address_space_operations vector. So it can just be set onto .write_end for
460 * FSes that don't need any other processing. i_mutex is assumed to be held.
461 * Block based filesystems should use generic_write_end().
462 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
463 * is not called, so a filesystem that actually does store data in .write_inode
464 * should extend on what's done here with a call to mark_inode_dirty() in the
465 * case that i_size has changed.
467 * Use *ONLY* with simple_readpage()
469 int simple_write_end(struct file
*file
, struct address_space
*mapping
,
470 loff_t pos
, unsigned len
, unsigned copied
,
471 struct page
*page
, void *fsdata
)
473 struct inode
*inode
= page
->mapping
->host
;
474 loff_t last_pos
= pos
+ copied
;
476 /* zero the stale part of the page if we did a short copy */
477 if (!PageUptodate(page
)) {
479 unsigned from
= pos
& (PAGE_SIZE
- 1);
481 zero_user(page
, from
+ copied
, len
- copied
);
483 SetPageUptodate(page
);
486 * No need to use i_size_read() here, the i_size
487 * cannot change under us because we hold the i_mutex.
489 if (last_pos
> inode
->i_size
)
490 i_size_write(inode
, last_pos
);
492 set_page_dirty(page
);
498 EXPORT_SYMBOL(simple_write_end
);
501 * the inodes created here are not hashed. If you use iunique to generate
502 * unique inode values later for this filesystem, then you must take care
503 * to pass it an appropriate max_reserved value to avoid collisions.
505 int simple_fill_super(struct super_block
*s
, unsigned long magic
,
506 const struct tree_descr
*files
)
510 struct dentry
*dentry
;
513 s
->s_blocksize
= PAGE_SIZE
;
514 s
->s_blocksize_bits
= PAGE_SHIFT
;
516 s
->s_op
= &simple_super_operations
;
519 inode
= new_inode(s
);
523 * because the root inode is 1, the files array must not contain an
527 inode
->i_mode
= S_IFDIR
| 0755;
528 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= current_time(inode
);
529 inode
->i_op
= &simple_dir_inode_operations
;
530 inode
->i_fop
= &simple_dir_operations
;
532 root
= d_make_root(inode
);
535 for (i
= 0; !files
->name
|| files
->name
[0]; i
++, files
++) {
539 /* warn if it tries to conflict with the root inode */
540 if (unlikely(i
== 1))
541 printk(KERN_WARNING
"%s: %s passed in a files array"
542 "with an index of 1!\n", __func__
,
545 dentry
= d_alloc_name(root
, files
->name
);
548 inode
= new_inode(s
);
553 inode
->i_mode
= S_IFREG
| files
->mode
;
554 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= current_time(inode
);
555 inode
->i_fop
= files
->ops
;
557 d_add(dentry
, inode
);
563 shrink_dcache_parent(root
);
567 EXPORT_SYMBOL(simple_fill_super
);
569 static DEFINE_SPINLOCK(pin_fs_lock
);
571 int simple_pin_fs(struct file_system_type
*type
, struct vfsmount
**mount
, int *count
)
573 struct vfsmount
*mnt
= NULL
;
574 spin_lock(&pin_fs_lock
);
575 if (unlikely(!*mount
)) {
576 spin_unlock(&pin_fs_lock
);
577 mnt
= vfs_kern_mount(type
, SB_KERNMOUNT
, type
->name
, NULL
);
580 spin_lock(&pin_fs_lock
);
586 spin_unlock(&pin_fs_lock
);
590 EXPORT_SYMBOL(simple_pin_fs
);
592 void simple_release_fs(struct vfsmount
**mount
, int *count
)
594 struct vfsmount
*mnt
;
595 spin_lock(&pin_fs_lock
);
599 spin_unlock(&pin_fs_lock
);
602 EXPORT_SYMBOL(simple_release_fs
);
605 * simple_read_from_buffer - copy data from the buffer to user space
606 * @to: the user space buffer to read to
607 * @count: the maximum number of bytes to read
608 * @ppos: the current position in the buffer
609 * @from: the buffer to read from
610 * @available: the size of the buffer
612 * The simple_read_from_buffer() function reads up to @count bytes from the
613 * buffer @from at offset @ppos into the user space address starting at @to.
615 * On success, the number of bytes read is returned and the offset @ppos is
616 * advanced by this number, or negative value is returned on error.
618 ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
, loff_t
*ppos
,
619 const void *from
, size_t available
)
626 if (pos
>= available
|| !count
)
628 if (count
> available
- pos
)
629 count
= available
- pos
;
630 ret
= copy_to_user(to
, from
+ pos
, count
);
637 EXPORT_SYMBOL(simple_read_from_buffer
);
640 * simple_write_to_buffer - copy data from user space to the buffer
641 * @to: the buffer to write to
642 * @available: the size of the buffer
643 * @ppos: the current position in the buffer
644 * @from: the user space buffer to read from
645 * @count: the maximum number of bytes to read
647 * The simple_write_to_buffer() function reads up to @count bytes from the user
648 * space address starting at @from into the buffer @to at offset @ppos.
650 * On success, the number of bytes written is returned and the offset @ppos is
651 * advanced by this number, or negative value is returned on error.
653 ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
654 const void __user
*from
, size_t count
)
661 if (pos
>= available
|| !count
)
663 if (count
> available
- pos
)
664 count
= available
- pos
;
665 res
= copy_from_user(to
+ pos
, from
, count
);
672 EXPORT_SYMBOL(simple_write_to_buffer
);
675 * memory_read_from_buffer - copy data from the buffer
676 * @to: the kernel space buffer to read to
677 * @count: the maximum number of bytes to read
678 * @ppos: the current position in the buffer
679 * @from: the buffer to read from
680 * @available: the size of the buffer
682 * The memory_read_from_buffer() function reads up to @count bytes from the
683 * buffer @from at offset @ppos into the kernel space address starting at @to.
685 * On success, the number of bytes read is returned and the offset @ppos is
686 * advanced by this number, or negative value is returned on error.
688 ssize_t
memory_read_from_buffer(void *to
, size_t count
, loff_t
*ppos
,
689 const void *from
, size_t available
)
695 if (pos
>= available
)
697 if (count
> available
- pos
)
698 count
= available
- pos
;
699 memcpy(to
, from
+ pos
, count
);
704 EXPORT_SYMBOL(memory_read_from_buffer
);
707 * Transaction based IO.
708 * The file expects a single write which triggers the transaction, and then
709 * possibly a read which collects the result - which is stored in a
713 void simple_transaction_set(struct file
*file
, size_t n
)
715 struct simple_transaction_argresp
*ar
= file
->private_data
;
717 BUG_ON(n
> SIMPLE_TRANSACTION_LIMIT
);
720 * The barrier ensures that ar->size will really remain zero until
721 * ar->data is ready for reading.
726 EXPORT_SYMBOL(simple_transaction_set
);
728 char *simple_transaction_get(struct file
*file
, const char __user
*buf
, size_t size
)
730 struct simple_transaction_argresp
*ar
;
731 static DEFINE_SPINLOCK(simple_transaction_lock
);
733 if (size
> SIMPLE_TRANSACTION_LIMIT
- 1)
734 return ERR_PTR(-EFBIG
);
736 ar
= (struct simple_transaction_argresp
*)get_zeroed_page(GFP_KERNEL
);
738 return ERR_PTR(-ENOMEM
);
740 spin_lock(&simple_transaction_lock
);
742 /* only one write allowed per open */
743 if (file
->private_data
) {
744 spin_unlock(&simple_transaction_lock
);
745 free_page((unsigned long)ar
);
746 return ERR_PTR(-EBUSY
);
749 file
->private_data
= ar
;
751 spin_unlock(&simple_transaction_lock
);
753 if (copy_from_user(ar
->data
, buf
, size
))
754 return ERR_PTR(-EFAULT
);
758 EXPORT_SYMBOL(simple_transaction_get
);
760 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
, size_t size
, loff_t
*pos
)
762 struct simple_transaction_argresp
*ar
= file
->private_data
;
766 return simple_read_from_buffer(buf
, size
, pos
, ar
->data
, ar
->size
);
768 EXPORT_SYMBOL(simple_transaction_read
);
770 int simple_transaction_release(struct inode
*inode
, struct file
*file
)
772 free_page((unsigned long)file
->private_data
);
775 EXPORT_SYMBOL(simple_transaction_release
);
777 /* Simple attribute files */
780 int (*get
)(void *, u64
*);
781 int (*set
)(void *, u64
);
782 char get_buf
[24]; /* enough to store a u64 and "\n\0" */
785 const char *fmt
; /* format for read operation */
786 struct mutex mutex
; /* protects access to these buffers */
789 /* simple_attr_open is called by an actual attribute open file operation
790 * to set the attribute specific access operations. */
791 int simple_attr_open(struct inode
*inode
, struct file
*file
,
792 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
795 struct simple_attr
*attr
;
797 attr
= kmalloc(sizeof(*attr
), GFP_KERNEL
);
803 attr
->data
= inode
->i_private
;
805 mutex_init(&attr
->mutex
);
807 file
->private_data
= attr
;
809 return nonseekable_open(inode
, file
);
811 EXPORT_SYMBOL_GPL(simple_attr_open
);
813 int simple_attr_release(struct inode
*inode
, struct file
*file
)
815 kfree(file
->private_data
);
818 EXPORT_SYMBOL_GPL(simple_attr_release
); /* GPL-only? This? Really? */
820 /* read from the buffer that is filled with the get function */
821 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
822 size_t len
, loff_t
*ppos
)
824 struct simple_attr
*attr
;
828 attr
= file
->private_data
;
833 ret
= mutex_lock_interruptible(&attr
->mutex
);
837 if (*ppos
) { /* continued read */
838 size
= strlen(attr
->get_buf
);
839 } else { /* first read */
841 ret
= attr
->get(attr
->data
, &val
);
845 size
= scnprintf(attr
->get_buf
, sizeof(attr
->get_buf
),
846 attr
->fmt
, (unsigned long long)val
);
849 ret
= simple_read_from_buffer(buf
, len
, ppos
, attr
->get_buf
, size
);
851 mutex_unlock(&attr
->mutex
);
854 EXPORT_SYMBOL_GPL(simple_attr_read
);
856 /* interpret the buffer as a number to call the set function with */
857 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
858 size_t len
, loff_t
*ppos
)
860 struct simple_attr
*attr
;
865 attr
= file
->private_data
;
869 ret
= mutex_lock_interruptible(&attr
->mutex
);
874 size
= min(sizeof(attr
->set_buf
) - 1, len
);
875 if (copy_from_user(attr
->set_buf
, buf
, size
))
878 attr
->set_buf
[size
] = '\0';
879 val
= simple_strtoll(attr
->set_buf
, NULL
, 0);
880 ret
= attr
->set(attr
->data
, val
);
882 ret
= len
; /* on success, claim we got the whole input */
884 mutex_unlock(&attr
->mutex
);
887 EXPORT_SYMBOL_GPL(simple_attr_write
);
890 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
891 * @sb: filesystem to do the file handle conversion on
892 * @fid: file handle to convert
893 * @fh_len: length of the file handle in bytes
894 * @fh_type: type of file handle
895 * @get_inode: filesystem callback to retrieve inode
897 * This function decodes @fid as long as it has one of the well-known
898 * Linux filehandle types and calls @get_inode on it to retrieve the
899 * inode for the object specified in the file handle.
901 struct dentry
*generic_fh_to_dentry(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
:
912 case FILEID_INO32_GEN_PARENT
:
913 inode
= get_inode(sb
, fid
->i32
.ino
, fid
->i32
.gen
);
917 return d_obtain_alias(inode
);
919 EXPORT_SYMBOL_GPL(generic_fh_to_dentry
);
922 * generic_fh_to_parent - generic helper for the fh_to_parent export operation
923 * @sb: filesystem to do the file handle conversion on
924 * @fid: file handle to convert
925 * @fh_len: length of the file handle in bytes
926 * @fh_type: type of file handle
927 * @get_inode: filesystem callback to retrieve inode
929 * This function decodes @fid as long as it has one of the well-known
930 * Linux filehandle types and calls @get_inode on it to retrieve the
931 * inode for the _parent_ object specified in the file handle if it
932 * is specified in the file handle, or NULL otherwise.
934 struct dentry
*generic_fh_to_parent(struct super_block
*sb
, struct fid
*fid
,
935 int fh_len
, int fh_type
, struct inode
*(*get_inode
)
936 (struct super_block
*sb
, u64 ino
, u32 gen
))
938 struct inode
*inode
= NULL
;
944 case FILEID_INO32_GEN_PARENT
:
945 inode
= get_inode(sb
, fid
->i32
.parent_ino
,
946 (fh_len
> 3 ? fid
->i32
.parent_gen
: 0));
950 return d_obtain_alias(inode
);
952 EXPORT_SYMBOL_GPL(generic_fh_to_parent
);
955 * __generic_file_fsync - generic fsync implementation for simple filesystems
957 * @file: file to synchronize
958 * @start: start offset in bytes
959 * @end: end offset in bytes (inclusive)
960 * @datasync: only synchronize essential metadata if true
962 * This is a generic implementation of the fsync method for simple
963 * filesystems which track all non-inode metadata in the buffers list
964 * hanging off the address_space structure.
966 int __generic_file_fsync(struct file
*file
, loff_t start
, loff_t end
,
969 struct inode
*inode
= file
->f_mapping
->host
;
973 err
= file_write_and_wait_range(file
, start
, end
);
978 ret
= sync_mapping_buffers(inode
->i_mapping
);
979 if (!(inode
->i_state
& I_DIRTY_ALL
))
981 if (datasync
&& !(inode
->i_state
& I_DIRTY_DATASYNC
))
984 err
= sync_inode_metadata(inode
, 1);
990 /* check and advance again to catch errors after syncing out buffers */
991 err
= file_check_and_advance_wb_err(file
);
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 int noop_set_page_dirty(struct page
*page
)
1062 * Unlike __set_page_dirty_no_writeback that handles dirty page
1063 * tracking in the page object, dax does all dirty tracking in
1064 * the inode address_space in response to mkwrite faults. In the
1065 * dax case we only need to worry about potentially dirty CPU
1066 * caches, not dirty page cache pages to write back.
1068 * This callback is defined to prevent fallback to
1069 * __set_page_dirty_buffers() in set_page_dirty().
1073 EXPORT_SYMBOL_GPL(noop_set_page_dirty
);
1075 void noop_invalidatepage(struct page
*page
, unsigned int offset
,
1076 unsigned int length
)
1079 * There is no page cache to invalidate in the dax case, however
1080 * we need this callback defined to prevent falling back to
1081 * block_invalidatepage() in do_invalidatepage().
1084 EXPORT_SYMBOL_GPL(noop_invalidatepage
);
1086 ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
)
1089 * iomap based filesystems support direct I/O without need for
1090 * this callback. However, it still needs to be set in
1091 * inode->a_ops so that open/fcntl know that direct I/O is
1092 * generally supported.
1096 EXPORT_SYMBOL_GPL(noop_direct_IO
);
1098 /* Because kfree isn't assignment-compatible with void(void*) ;-/ */
1099 void kfree_link(void *p
)
1103 EXPORT_SYMBOL(kfree_link
);
1106 * nop .set_page_dirty method so that people can use .page_mkwrite on
1109 static int anon_set_page_dirty(struct page
*page
)
1115 * A single inode exists for all anon_inode files. Contrary to pipes,
1116 * anon_inode inodes have no associated per-instance data, so we need
1117 * only allocate one of them.
1119 struct inode
*alloc_anon_inode(struct super_block
*s
)
1121 static const struct address_space_operations anon_aops
= {
1122 .set_page_dirty
= anon_set_page_dirty
,
1124 struct inode
*inode
= new_inode_pseudo(s
);
1127 return ERR_PTR(-ENOMEM
);
1129 inode
->i_ino
= get_next_ino();
1130 inode
->i_mapping
->a_ops
= &anon_aops
;
1133 * Mark the inode dirty from the very beginning,
1134 * that way it will never be moved to the dirty
1135 * list because mark_inode_dirty() will think
1136 * that it already _is_ on the dirty list.
1138 inode
->i_state
= I_DIRTY
;
1139 inode
->i_mode
= S_IRUSR
| S_IWUSR
;
1140 inode
->i_uid
= current_fsuid();
1141 inode
->i_gid
= current_fsgid();
1142 inode
->i_flags
|= S_PRIVATE
;
1143 inode
->i_atime
= inode
->i_mtime
= inode
->i_ctime
= current_time(inode
);
1146 EXPORT_SYMBOL(alloc_anon_inode
);
1149 * simple_nosetlease - generic helper for prohibiting leases
1150 * @filp: file pointer
1151 * @arg: type of lease to obtain
1152 * @flp: new lease supplied for insertion
1153 * @priv: private data for lm_setup operation
1155 * Generic helper for filesystems that do not wish to allow leases to be set.
1156 * All arguments are ignored and it just returns -EINVAL.
1159 simple_nosetlease(struct file
*filp
, long arg
, struct file_lock
**flp
,
1164 EXPORT_SYMBOL(simple_nosetlease
);
1167 * simple_get_link - generic helper to get the target of "fast" symlinks
1168 * @dentry: not used here
1169 * @inode: the symlink inode
1170 * @done: not used here
1172 * Generic helper for filesystems to use for symlink inodes where a pointer to
1173 * the symlink target is stored in ->i_link. NOTE: this isn't normally called,
1174 * since as an optimization the path lookup code uses any non-NULL ->i_link
1175 * directly, without calling ->get_link(). But ->get_link() still must be set,
1176 * to mark the inode_operations as being for a symlink.
1178 * Return: the symlink target
1180 const char *simple_get_link(struct dentry
*dentry
, struct inode
*inode
,
1181 struct delayed_call
*done
)
1183 return inode
->i_link
;
1185 EXPORT_SYMBOL(simple_get_link
);
1187 const struct inode_operations simple_symlink_inode_operations
= {
1188 .get_link
= simple_get_link
,
1190 EXPORT_SYMBOL(simple_symlink_inode_operations
);
1193 * Operations for a permanently empty directory.
1195 static struct dentry
*empty_dir_lookup(struct inode
*dir
, struct dentry
*dentry
, unsigned int flags
)
1197 return ERR_PTR(-ENOENT
);
1200 static int empty_dir_getattr(const struct path
*path
, struct kstat
*stat
,
1201 u32 request_mask
, unsigned int query_flags
)
1203 struct inode
*inode
= d_inode(path
->dentry
);
1204 generic_fillattr(inode
, stat
);
1208 static int empty_dir_setattr(struct dentry
*dentry
, struct iattr
*attr
)
1213 static ssize_t
empty_dir_listxattr(struct dentry
*dentry
, char *list
, size_t size
)
1218 static const struct inode_operations empty_dir_inode_operations
= {
1219 .lookup
= empty_dir_lookup
,
1220 .permission
= generic_permission
,
1221 .setattr
= empty_dir_setattr
,
1222 .getattr
= empty_dir_getattr
,
1223 .listxattr
= empty_dir_listxattr
,
1226 static loff_t
empty_dir_llseek(struct file
*file
, loff_t offset
, int whence
)
1228 /* An empty directory has two entries . and .. at offsets 0 and 1 */
1229 return generic_file_llseek_size(file
, offset
, whence
, 2, 2);
1232 static int empty_dir_readdir(struct file
*file
, struct dir_context
*ctx
)
1234 dir_emit_dots(file
, ctx
);
1238 static const struct file_operations empty_dir_operations
= {
1239 .llseek
= empty_dir_llseek
,
1240 .read
= generic_read_dir
,
1241 .iterate_shared
= empty_dir_readdir
,
1242 .fsync
= noop_fsync
,
1246 void make_empty_dir_inode(struct inode
*inode
)
1248 set_nlink(inode
, 2);
1249 inode
->i_mode
= S_IFDIR
| S_IRUGO
| S_IXUGO
;
1250 inode
->i_uid
= GLOBAL_ROOT_UID
;
1251 inode
->i_gid
= GLOBAL_ROOT_GID
;
1254 inode
->i_blkbits
= PAGE_SHIFT
;
1255 inode
->i_blocks
= 0;
1257 inode
->i_op
= &empty_dir_inode_operations
;
1258 inode
->i_opflags
&= ~IOP_XATTR
;
1259 inode
->i_fop
= &empty_dir_operations
;
1262 bool is_empty_dir_inode(struct inode
*inode
)
1264 return (inode
->i_fop
== &empty_dir_operations
) &&
1265 (inode
->i_op
== &empty_dir_inode_operations
);