1 // SPDX-License-Identifier: GPL-2.0-only
4 * Library for filesystems writers.
7 #include <linux/blkdev.h>
8 #include <linux/export.h>
9 #include <linux/pagemap.h>
10 #include <linux/slab.h>
11 #include <linux/cred.h>
12 #include <linux/mount.h>
13 #include <linux/vfs.h>
14 #include <linux/quotaops.h>
15 #include <linux/mutex.h>
16 #include <linux/namei.h>
17 #include <linux/exportfs.h>
18 #include <linux/writeback.h>
19 #include <linux/buffer_head.h> /* sync_mapping_buffers */
20 #include <linux/fs_context.h>
21 #include <linux/pseudo_fs.h>
23 #include <linux/uaccess.h>
27 int simple_getattr(const struct path *path, struct kstat *stat,
28 u32 request_mask, unsigned int query_flags)
30 struct inode *inode = d_inode(path->dentry);
31 generic_fillattr(inode, stat);
32 stat->blocks = inode->i_mapping->nrpages << (PAGE_SHIFT - 9);
35 EXPORT_SYMBOL(simple_getattr);
37 int simple_statfs(struct dentry *dentry, struct kstatfs *buf)
39 buf->f_type = dentry->d_sb->s_magic;
40 buf->f_bsize = PAGE_SIZE;
41 buf->f_namelen = NAME_MAX;
44 EXPORT_SYMBOL(simple_statfs);
47 * Retaining negative dentries for an in-memory filesystem just wastes
48 * memory and lookup time: arrange for them to be deleted immediately.
50 int always_delete_dentry(const struct dentry *dentry)
54 EXPORT_SYMBOL(always_delete_dentry);
56 const struct dentry_operations simple_dentry_operations = {
57 .d_delete = always_delete_dentry,
59 EXPORT_SYMBOL(simple_dentry_operations);
62 * Lookup the data. This is trivial - if the dentry didn't already
63 * exist, we know it is negative. Set d_op to delete negative dentries.
65 struct dentry *simple_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
67 if (dentry->d_name.len > NAME_MAX)
68 return ERR_PTR(-ENAMETOOLONG);
69 if (!dentry->d_sb->s_d_op)
70 d_set_d_op(dentry, &simple_dentry_operations);
74 EXPORT_SYMBOL(simple_lookup);
76 int dcache_dir_open(struct inode *inode, struct file *file)
78 file->private_data = d_alloc_cursor(file->f_path.dentry);
80 return file->private_data ? 0 : -ENOMEM;
82 EXPORT_SYMBOL(dcache_dir_open);
84 int dcache_dir_close(struct inode *inode, struct file *file)
86 dput(file->private_data);
89 EXPORT_SYMBOL(dcache_dir_close);
91 /* parent is locked at least shared */
93 * Returns an element of siblings' list.
94 * We are looking for <count>th positive after <p>; if
95 * found, dentry is grabbed and returned to caller.
96 * If no such element exists, NULL is returned.
98 static struct dentry *scan_positives(struct dentry *cursor,
103 struct dentry *dentry = cursor->d_parent, *found = NULL;
105 spin_lock(&dentry->d_lock);
106 while ((p = p->next) != &dentry->d_subdirs) {
107 struct dentry *d = list_entry(p, struct dentry, d_child);
108 // we must at least skip cursors, to avoid livelocks
109 if (d->d_flags & DCACHE_DENTRY_CURSOR)
111 if (simple_positive(d) && !--count) {
112 spin_lock_nested(&d->d_lock, DENTRY_D_LOCK_NESTED);
113 if (simple_positive(d))
114 found = dget_dlock(d);
115 spin_unlock(&d->d_lock);
120 if (need_resched()) {
121 list_move(&cursor->d_child, p);
122 p = &cursor->d_child;
123 spin_unlock(&dentry->d_lock);
125 spin_lock(&dentry->d_lock);
128 spin_unlock(&dentry->d_lock);
133 loff_t dcache_dir_lseek(struct file *file, loff_t offset, int whence)
135 struct dentry *dentry = file->f_path.dentry;
138 offset += file->f_pos;
147 if (offset != file->f_pos) {
148 struct dentry *cursor = file->private_data;
149 struct dentry *to = NULL;
151 inode_lock_shared(dentry->d_inode);
154 to = scan_positives(cursor, &dentry->d_subdirs,
156 spin_lock(&dentry->d_lock);
158 list_move(&cursor->d_child, &to->d_child);
160 list_del_init(&cursor->d_child);
161 spin_unlock(&dentry->d_lock);
164 file->f_pos = offset;
166 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 *anchor = &dentry->d_subdirs;
189 struct dentry *next = NULL;
192 if (!dir_emit_dots(file, ctx))
197 else if (!list_empty(&cursor->d_child))
198 p = &cursor->d_child;
202 while ((next = scan_positives(cursor, p, 1, next)) != NULL) {
203 if (!dir_emit(ctx, next->d_name.name, next->d_name.len,
204 d_inode(next)->i_ino, dt_type(d_inode(next))))
209 spin_lock(&dentry->d_lock);
211 list_move_tail(&cursor->d_child, &next->d_child);
213 list_del_init(&cursor->d_child);
214 spin_unlock(&dentry->d_lock);
219 EXPORT_SYMBOL(dcache_readdir);
221 ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos)
225 EXPORT_SYMBOL(generic_read_dir);
227 const struct file_operations simple_dir_operations = {
228 .open = dcache_dir_open,
229 .release = dcache_dir_close,
230 .llseek = dcache_dir_lseek,
231 .read = generic_read_dir,
232 .iterate_shared = dcache_readdir,
235 EXPORT_SYMBOL(simple_dir_operations);
237 const struct inode_operations simple_dir_inode_operations = {
238 .lookup = simple_lookup,
240 EXPORT_SYMBOL(simple_dir_inode_operations);
242 static const struct super_operations simple_super_operations = {
243 .statfs = simple_statfs,
246 static int pseudo_fs_fill_super(struct super_block *s, struct fs_context *fc)
248 struct pseudo_fs_context *ctx = fc->fs_private;
251 s->s_maxbytes = MAX_LFS_FILESIZE;
252 s->s_blocksize = PAGE_SIZE;
253 s->s_blocksize_bits = PAGE_SHIFT;
254 s->s_magic = ctx->magic;
255 s->s_op = ctx->ops ?: &simple_super_operations;
256 s->s_xattr = ctx->xattr;
263 * since this is the first inode, make it number 1. New inodes created
264 * after this must take care not to collide with it (by passing
265 * max_reserved of 1 to iunique).
268 root->i_mode = S_IFDIR | S_IRUSR | S_IWUSR;
269 root->i_atime = root->i_mtime = root->i_ctime = current_time(root);
270 s->s_root = d_make_root(root);
273 s->s_d_op = ctx->dops;
277 static int pseudo_fs_get_tree(struct fs_context *fc)
279 return get_tree_nodev(fc, pseudo_fs_fill_super);
282 static void pseudo_fs_free(struct fs_context *fc)
284 kfree(fc->fs_private);
287 static const struct fs_context_operations pseudo_fs_context_ops = {
288 .free = pseudo_fs_free,
289 .get_tree = pseudo_fs_get_tree,
293 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
294 * will never be mountable)
296 struct pseudo_fs_context *init_pseudo(struct fs_context *fc,
299 struct pseudo_fs_context *ctx;
301 ctx = kzalloc(sizeof(struct pseudo_fs_context), GFP_KERNEL);
304 fc->fs_private = ctx;
305 fc->ops = &pseudo_fs_context_ops;
306 fc->sb_flags |= SB_NOUSER;
311 EXPORT_SYMBOL(init_pseudo);
313 int simple_open(struct inode *inode, struct file *file)
315 if (inode->i_private)
316 file->private_data = inode->i_private;
319 EXPORT_SYMBOL(simple_open);
321 int simple_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
323 struct inode *inode = d_inode(old_dentry);
325 inode->i_ctime = dir->i_ctime = dir->i_mtime = current_time(inode);
329 d_instantiate(dentry, inode);
332 EXPORT_SYMBOL(simple_link);
334 int simple_empty(struct dentry *dentry)
336 struct dentry *child;
339 spin_lock(&dentry->d_lock);
340 list_for_each_entry(child, &dentry->d_subdirs, d_child) {
341 spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED);
342 if (simple_positive(child)) {
343 spin_unlock(&child->d_lock);
346 spin_unlock(&child->d_lock);
350 spin_unlock(&dentry->d_lock);
353 EXPORT_SYMBOL(simple_empty);
355 int simple_unlink(struct inode *dir, struct dentry *dentry)
357 struct inode *inode = d_inode(dentry);
359 inode->i_ctime = dir->i_ctime = dir->i_mtime = current_time(inode);
364 EXPORT_SYMBOL(simple_unlink);
366 int simple_rmdir(struct inode *dir, struct dentry *dentry)
368 if (!simple_empty(dentry))
371 drop_nlink(d_inode(dentry));
372 simple_unlink(dir, dentry);
376 EXPORT_SYMBOL(simple_rmdir);
378 int simple_rename(struct inode *old_dir, struct dentry *old_dentry,
379 struct inode *new_dir, struct dentry *new_dentry,
382 struct inode *inode = d_inode(old_dentry);
383 int they_are_dirs = d_is_dir(old_dentry);
385 if (flags & ~RENAME_NOREPLACE)
388 if (!simple_empty(new_dentry))
391 if (d_really_is_positive(new_dentry)) {
392 simple_unlink(new_dir, new_dentry);
394 drop_nlink(d_inode(new_dentry));
397 } else if (they_are_dirs) {
402 old_dir->i_ctime = old_dir->i_mtime = new_dir->i_ctime =
403 new_dir->i_mtime = inode->i_ctime = current_time(old_dir);
407 EXPORT_SYMBOL(simple_rename);
410 * simple_setattr - setattr for simple filesystem
412 * @iattr: iattr structure
414 * Returns 0 on success, -error on failure.
416 * simple_setattr is a simple ->setattr implementation without a proper
417 * implementation of size changes.
419 * It can either be used for in-memory filesystems or special files
420 * on simple regular filesystems. Anything that needs to change on-disk
421 * or wire state on size changes needs its own setattr method.
423 int simple_setattr(struct dentry *dentry, struct iattr *iattr)
425 struct inode *inode = d_inode(dentry);
428 error = setattr_prepare(dentry, iattr);
432 if (iattr->ia_valid & ATTR_SIZE)
433 truncate_setsize(inode, iattr->ia_size);
434 setattr_copy(inode, iattr);
435 mark_inode_dirty(inode);
438 EXPORT_SYMBOL(simple_setattr);
440 int simple_readpage(struct file *file, struct page *page)
442 clear_highpage(page);
443 flush_dcache_page(page);
444 SetPageUptodate(page);
448 EXPORT_SYMBOL(simple_readpage);
450 int simple_write_begin(struct file *file, struct address_space *mapping,
451 loff_t pos, unsigned len, unsigned flags,
452 struct page **pagep, void **fsdata)
457 index = pos >> PAGE_SHIFT;
459 page = grab_cache_page_write_begin(mapping, index, flags);
465 if (!PageUptodate(page) && (len != PAGE_SIZE)) {
466 unsigned from = pos & (PAGE_SIZE - 1);
468 zero_user_segments(page, 0, from, from + len, PAGE_SIZE);
472 EXPORT_SYMBOL(simple_write_begin);
475 * simple_write_end - .write_end helper for non-block-device FSes
476 * @file: See .write_end of address_space_operations
484 * simple_write_end does the minimum needed for updating a page after writing is
485 * done. It has the same API signature as the .write_end of
486 * address_space_operations vector. So it can just be set onto .write_end for
487 * FSes that don't need any other processing. i_mutex is assumed to be held.
488 * Block based filesystems should use generic_write_end().
489 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
490 * is not called, so a filesystem that actually does store data in .write_inode
491 * should extend on what's done here with a call to mark_inode_dirty() in the
492 * case that i_size has changed.
494 * Use *ONLY* with simple_readpage()
496 int simple_write_end(struct file *file, struct address_space *mapping,
497 loff_t pos, unsigned len, unsigned copied,
498 struct page *page, void *fsdata)
500 struct inode *inode = page->mapping->host;
501 loff_t last_pos = pos + copied;
503 /* zero the stale part of the page if we did a short copy */
504 if (!PageUptodate(page)) {
506 unsigned from = pos & (PAGE_SIZE - 1);
508 zero_user(page, from + copied, len - copied);
510 SetPageUptodate(page);
513 * No need to use i_size_read() here, the i_size
514 * cannot change under us because we hold the i_mutex.
516 if (last_pos > inode->i_size)
517 i_size_write(inode, last_pos);
519 set_page_dirty(page);
525 EXPORT_SYMBOL(simple_write_end);
528 * the inodes created here are not hashed. If you use iunique to generate
529 * unique inode values later for this filesystem, then you must take care
530 * to pass it an appropriate max_reserved value to avoid collisions.
532 int simple_fill_super(struct super_block *s, unsigned long magic,
533 const struct tree_descr *files)
537 struct dentry *dentry;
540 s->s_blocksize = PAGE_SIZE;
541 s->s_blocksize_bits = PAGE_SHIFT;
543 s->s_op = &simple_super_operations;
546 inode = new_inode(s);
550 * because the root inode is 1, the files array must not contain an
554 inode->i_mode = S_IFDIR | 0755;
555 inode->i_atime = inode->i_mtime = inode->i_ctime = current_time(inode);
556 inode->i_op = &simple_dir_inode_operations;
557 inode->i_fop = &simple_dir_operations;
559 root = d_make_root(inode);
562 for (i = 0; !files->name || files->name[0]; i++, files++) {
566 /* warn if it tries to conflict with the root inode */
567 if (unlikely(i == 1))
568 printk(KERN_WARNING "%s: %s passed in a files array"
569 "with an index of 1!\n", __func__,
572 dentry = d_alloc_name(root, files->name);
575 inode = new_inode(s);
580 inode->i_mode = S_IFREG | files->mode;
581 inode->i_atime = inode->i_mtime = inode->i_ctime = current_time(inode);
582 inode->i_fop = files->ops;
584 d_add(dentry, inode);
590 shrink_dcache_parent(root);
594 EXPORT_SYMBOL(simple_fill_super);
596 static DEFINE_SPINLOCK(pin_fs_lock);
598 int simple_pin_fs(struct file_system_type *type, struct vfsmount **mount, int *count)
600 struct vfsmount *mnt = NULL;
601 spin_lock(&pin_fs_lock);
602 if (unlikely(!*mount)) {
603 spin_unlock(&pin_fs_lock);
604 mnt = vfs_kern_mount(type, SB_KERNMOUNT, type->name, NULL);
607 spin_lock(&pin_fs_lock);
613 spin_unlock(&pin_fs_lock);
617 EXPORT_SYMBOL(simple_pin_fs);
619 void simple_release_fs(struct vfsmount **mount, int *count)
621 struct vfsmount *mnt;
622 spin_lock(&pin_fs_lock);
626 spin_unlock(&pin_fs_lock);
629 EXPORT_SYMBOL(simple_release_fs);
632 * simple_read_from_buffer - copy data from the buffer to user space
633 * @to: the user space buffer to read to
634 * @count: the maximum number of bytes to read
635 * @ppos: the current position in the buffer
636 * @from: the buffer to read from
637 * @available: the size of the buffer
639 * The simple_read_from_buffer() function reads up to @count bytes from the
640 * buffer @from at offset @ppos into the user space address starting at @to.
642 * On success, the number of bytes read is returned and the offset @ppos is
643 * advanced by this number, or negative value is returned on error.
645 ssize_t simple_read_from_buffer(void __user *to, size_t count, loff_t *ppos,
646 const void *from, size_t available)
653 if (pos >= available || !count)
655 if (count > available - pos)
656 count = available - pos;
657 ret = copy_to_user(to, from + pos, count);
664 EXPORT_SYMBOL(simple_read_from_buffer);
667 * simple_write_to_buffer - copy data from user space to the buffer
668 * @to: the buffer to write to
669 * @available: the size of the buffer
670 * @ppos: the current position in the buffer
671 * @from: the user space buffer to read from
672 * @count: the maximum number of bytes to read
674 * The simple_write_to_buffer() function reads up to @count bytes from the user
675 * space address starting at @from into the buffer @to at offset @ppos.
677 * On success, the number of bytes written is returned and the offset @ppos is
678 * advanced by this number, or negative value is returned on error.
680 ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
681 const void __user *from, size_t count)
688 if (pos >= available || !count)
690 if (count > available - pos)
691 count = available - pos;
692 res = copy_from_user(to + pos, from, count);
699 EXPORT_SYMBOL(simple_write_to_buffer);
702 * memory_read_from_buffer - copy data from the buffer
703 * @to: the kernel space buffer to read to
704 * @count: the maximum number of bytes to read
705 * @ppos: the current position in the buffer
706 * @from: the buffer to read from
707 * @available: the size of the buffer
709 * The memory_read_from_buffer() function reads up to @count bytes from the
710 * buffer @from at offset @ppos into the kernel space address starting at @to.
712 * On success, the number of bytes read is returned and the offset @ppos is
713 * advanced by this number, or negative value is returned on error.
715 ssize_t memory_read_from_buffer(void *to, size_t count, loff_t *ppos,
716 const void *from, size_t available)
722 if (pos >= available)
724 if (count > available - pos)
725 count = available - pos;
726 memcpy(to, from + pos, count);
731 EXPORT_SYMBOL(memory_read_from_buffer);
734 * Transaction based IO.
735 * The file expects a single write which triggers the transaction, and then
736 * possibly a read which collects the result - which is stored in a
740 void simple_transaction_set(struct file *file, size_t n)
742 struct simple_transaction_argresp *ar = file->private_data;
744 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
747 * The barrier ensures that ar->size will really remain zero until
748 * ar->data is ready for reading.
753 EXPORT_SYMBOL(simple_transaction_set);
755 char *simple_transaction_get(struct file *file, const char __user *buf, size_t size)
757 struct simple_transaction_argresp *ar;
758 static DEFINE_SPINLOCK(simple_transaction_lock);
760 if (size > SIMPLE_TRANSACTION_LIMIT - 1)
761 return ERR_PTR(-EFBIG);
763 ar = (struct simple_transaction_argresp *)get_zeroed_page(GFP_KERNEL);
765 return ERR_PTR(-ENOMEM);
767 spin_lock(&simple_transaction_lock);
769 /* only one write allowed per open */
770 if (file->private_data) {
771 spin_unlock(&simple_transaction_lock);
772 free_page((unsigned long)ar);
773 return ERR_PTR(-EBUSY);
776 file->private_data = ar;
778 spin_unlock(&simple_transaction_lock);
780 if (copy_from_user(ar->data, buf, size))
781 return ERR_PTR(-EFAULT);
785 EXPORT_SYMBOL(simple_transaction_get);
787 ssize_t simple_transaction_read(struct file *file, char __user *buf, size_t size, loff_t *pos)
789 struct simple_transaction_argresp *ar = file->private_data;
793 return simple_read_from_buffer(buf, size, pos, ar->data, ar->size);
795 EXPORT_SYMBOL(simple_transaction_read);
797 int simple_transaction_release(struct inode *inode, struct file *file)
799 free_page((unsigned long)file->private_data);
802 EXPORT_SYMBOL(simple_transaction_release);
804 /* Simple attribute files */
807 int (*get)(void *, u64 *);
808 int (*set)(void *, u64);
809 char get_buf[24]; /* enough to store a u64 and "\n\0" */
812 const char *fmt; /* format for read operation */
813 struct mutex mutex; /* protects access to these buffers */
816 /* simple_attr_open is called by an actual attribute open file operation
817 * to set the attribute specific access operations. */
818 int simple_attr_open(struct inode *inode, struct file *file,
819 int (*get)(void *, u64 *), int (*set)(void *, u64),
822 struct simple_attr *attr;
824 attr = kmalloc(sizeof(*attr), GFP_KERNEL);
830 attr->data = inode->i_private;
832 mutex_init(&attr->mutex);
834 file->private_data = attr;
836 return nonseekable_open(inode, file);
838 EXPORT_SYMBOL_GPL(simple_attr_open);
840 int simple_attr_release(struct inode *inode, struct file *file)
842 kfree(file->private_data);
845 EXPORT_SYMBOL_GPL(simple_attr_release); /* GPL-only? This? Really? */
847 /* read from the buffer that is filled with the get function */
848 ssize_t simple_attr_read(struct file *file, char __user *buf,
849 size_t len, loff_t *ppos)
851 struct simple_attr *attr;
855 attr = file->private_data;
860 ret = mutex_lock_interruptible(&attr->mutex);
864 if (*ppos) { /* continued read */
865 size = strlen(attr->get_buf);
866 } else { /* first read */
868 ret = attr->get(attr->data, &val);
872 size = scnprintf(attr->get_buf, sizeof(attr->get_buf),
873 attr->fmt, (unsigned long long)val);
876 ret = simple_read_from_buffer(buf, len, ppos, attr->get_buf, size);
878 mutex_unlock(&attr->mutex);
881 EXPORT_SYMBOL_GPL(simple_attr_read);
883 /* interpret the buffer as a number to call the set function with */
884 ssize_t simple_attr_write(struct file *file, const char __user *buf,
885 size_t len, loff_t *ppos)
887 struct simple_attr *attr;
892 attr = file->private_data;
896 ret = mutex_lock_interruptible(&attr->mutex);
901 size = min(sizeof(attr->set_buf) - 1, len);
902 if (copy_from_user(attr->set_buf, buf, size))
905 attr->set_buf[size] = '\0';
906 val = simple_strtoll(attr->set_buf, NULL, 0);
907 ret = attr->set(attr->data, val);
909 ret = len; /* on success, claim we got the whole input */
911 mutex_unlock(&attr->mutex);
914 EXPORT_SYMBOL_GPL(simple_attr_write);
917 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
918 * @sb: filesystem to do the file handle conversion on
919 * @fid: file handle to convert
920 * @fh_len: length of the file handle in bytes
921 * @fh_type: type of file handle
922 * @get_inode: filesystem callback to retrieve inode
924 * This function decodes @fid as long as it has one of the well-known
925 * Linux filehandle types and calls @get_inode on it to retrieve the
926 * inode for the object specified in the file handle.
928 struct dentry *generic_fh_to_dentry(struct super_block *sb, struct fid *fid,
929 int fh_len, int fh_type, struct inode *(*get_inode)
930 (struct super_block *sb, u64 ino, u32 gen))
932 struct inode *inode = NULL;
938 case FILEID_INO32_GEN:
939 case FILEID_INO32_GEN_PARENT:
940 inode = get_inode(sb, fid->i32.ino, fid->i32.gen);
944 return d_obtain_alias(inode);
946 EXPORT_SYMBOL_GPL(generic_fh_to_dentry);
949 * generic_fh_to_parent - generic helper for the fh_to_parent export operation
950 * @sb: filesystem to do the file handle conversion on
951 * @fid: file handle to convert
952 * @fh_len: length of the file handle in bytes
953 * @fh_type: type of file handle
954 * @get_inode: filesystem callback to retrieve inode
956 * This function decodes @fid as long as it has one of the well-known
957 * Linux filehandle types and calls @get_inode on it to retrieve the
958 * inode for the _parent_ object specified in the file handle if it
959 * is specified in the file handle, or NULL otherwise.
961 struct dentry *generic_fh_to_parent(struct super_block *sb, struct fid *fid,
962 int fh_len, int fh_type, struct inode *(*get_inode)
963 (struct super_block *sb, u64 ino, u32 gen))
965 struct inode *inode = NULL;
971 case FILEID_INO32_GEN_PARENT:
972 inode = get_inode(sb, fid->i32.parent_ino,
973 (fh_len > 3 ? fid->i32.parent_gen : 0));
977 return d_obtain_alias(inode);
979 EXPORT_SYMBOL_GPL(generic_fh_to_parent);
982 * __generic_file_fsync - generic fsync implementation for simple filesystems
984 * @file: file to synchronize
985 * @start: start offset in bytes
986 * @end: end offset in bytes (inclusive)
987 * @datasync: only synchronize essential metadata if true
989 * This is a generic implementation of the fsync method for simple
990 * filesystems which track all non-inode metadata in the buffers list
991 * hanging off the address_space structure.
993 int __generic_file_fsync(struct file *file, loff_t start, loff_t end,
996 struct inode *inode = file->f_mapping->host;
1000 err = file_write_and_wait_range(file, start, end);
1005 ret = sync_mapping_buffers(inode->i_mapping);
1006 if (!(inode->i_state & I_DIRTY_ALL))
1008 if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))
1011 err = sync_inode_metadata(inode, 1);
1016 inode_unlock(inode);
1017 /* check and advance again to catch errors after syncing out buffers */
1018 err = file_check_and_advance_wb_err(file);
1023 EXPORT_SYMBOL(__generic_file_fsync);
1026 * generic_file_fsync - generic fsync implementation for simple filesystems
1028 * @file: file to synchronize
1029 * @start: start offset in bytes
1030 * @end: end offset in bytes (inclusive)
1031 * @datasync: only synchronize essential metadata if true
1035 int generic_file_fsync(struct file *file, loff_t start, loff_t end,
1038 struct inode *inode = file->f_mapping->host;
1041 err = __generic_file_fsync(file, start, end, datasync);
1044 return blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL, NULL);
1046 EXPORT_SYMBOL(generic_file_fsync);
1049 * generic_check_addressable - Check addressability of file system
1050 * @blocksize_bits: log of file system block size
1051 * @num_blocks: number of blocks in file system
1053 * Determine whether a file system with @num_blocks blocks (and a
1054 * block size of 2**@blocksize_bits) is addressable by the sector_t
1055 * and page cache of the system. Return 0 if so and -EFBIG otherwise.
1057 int generic_check_addressable(unsigned blocksize_bits, u64 num_blocks)
1059 u64 last_fs_block = num_blocks - 1;
1061 last_fs_block >> (PAGE_SHIFT - blocksize_bits);
1063 if (unlikely(num_blocks == 0))
1066 if ((blocksize_bits < 9) || (blocksize_bits > PAGE_SHIFT))
1069 if ((last_fs_block > (sector_t)(~0ULL) >> (blocksize_bits - 9)) ||
1070 (last_fs_page > (pgoff_t)(~0ULL))) {
1075 EXPORT_SYMBOL(generic_check_addressable);
1078 * No-op implementation of ->fsync for in-memory filesystems.
1080 int noop_fsync(struct file *file, loff_t start, loff_t end, int datasync)
1084 EXPORT_SYMBOL(noop_fsync);
1086 int noop_set_page_dirty(struct page *page)
1089 * Unlike __set_page_dirty_no_writeback that handles dirty page
1090 * tracking in the page object, dax does all dirty tracking in
1091 * the inode address_space in response to mkwrite faults. In the
1092 * dax case we only need to worry about potentially dirty CPU
1093 * caches, not dirty page cache pages to write back.
1095 * This callback is defined to prevent fallback to
1096 * __set_page_dirty_buffers() in set_page_dirty().
1100 EXPORT_SYMBOL_GPL(noop_set_page_dirty);
1102 void noop_invalidatepage(struct page *page, unsigned int offset,
1103 unsigned int length)
1106 * There is no page cache to invalidate in the dax case, however
1107 * we need this callback defined to prevent falling back to
1108 * block_invalidatepage() in do_invalidatepage().
1111 EXPORT_SYMBOL_GPL(noop_invalidatepage);
1113 ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
1116 * iomap based filesystems support direct I/O without need for
1117 * this callback. However, it still needs to be set in
1118 * inode->a_ops so that open/fcntl know that direct I/O is
1119 * generally supported.
1123 EXPORT_SYMBOL_GPL(noop_direct_IO);
1125 /* Because kfree isn't assignment-compatible with void(void*) ;-/ */
1126 void kfree_link(void *p)
1130 EXPORT_SYMBOL(kfree_link);
1133 * nop .set_page_dirty method so that people can use .page_mkwrite on
1136 static int anon_set_page_dirty(struct page *page)
1142 * A single inode exists for all anon_inode files. Contrary to pipes,
1143 * anon_inode inodes have no associated per-instance data, so we need
1144 * only allocate one of them.
1146 struct inode *alloc_anon_inode(struct super_block *s)
1148 static const struct address_space_operations anon_aops = {
1149 .set_page_dirty = anon_set_page_dirty,
1151 struct inode *inode = new_inode_pseudo(s);
1154 return ERR_PTR(-ENOMEM);
1156 inode->i_ino = get_next_ino();
1157 inode->i_mapping->a_ops = &anon_aops;
1160 * Mark the inode dirty from the very beginning,
1161 * that way it will never be moved to the dirty
1162 * list because mark_inode_dirty() will think
1163 * that it already _is_ on the dirty list.
1165 inode->i_state = I_DIRTY;
1166 inode->i_mode = S_IRUSR | S_IWUSR;
1167 inode->i_uid = current_fsuid();
1168 inode->i_gid = current_fsgid();
1169 inode->i_flags |= S_PRIVATE;
1170 inode->i_atime = inode->i_mtime = inode->i_ctime = current_time(inode);
1173 EXPORT_SYMBOL(alloc_anon_inode);
1176 * simple_nosetlease - generic helper for prohibiting leases
1177 * @filp: file pointer
1178 * @arg: type of lease to obtain
1179 * @flp: new lease supplied for insertion
1180 * @priv: private data for lm_setup operation
1182 * Generic helper for filesystems that do not wish to allow leases to be set.
1183 * All arguments are ignored and it just returns -EINVAL.
1186 simple_nosetlease(struct file *filp, long arg, struct file_lock **flp,
1191 EXPORT_SYMBOL(simple_nosetlease);
1194 * simple_get_link - generic helper to get the target of "fast" symlinks
1195 * @dentry: not used here
1196 * @inode: the symlink inode
1197 * @done: not used here
1199 * Generic helper for filesystems to use for symlink inodes where a pointer to
1200 * the symlink target is stored in ->i_link. NOTE: this isn't normally called,
1201 * since as an optimization the path lookup code uses any non-NULL ->i_link
1202 * directly, without calling ->get_link(). But ->get_link() still must be set,
1203 * to mark the inode_operations as being for a symlink.
1205 * Return: the symlink target
1207 const char *simple_get_link(struct dentry *dentry, struct inode *inode,
1208 struct delayed_call *done)
1210 return inode->i_link;
1212 EXPORT_SYMBOL(simple_get_link);
1214 const struct inode_operations simple_symlink_inode_operations = {
1215 .get_link = simple_get_link,
1217 EXPORT_SYMBOL(simple_symlink_inode_operations);
1220 * Operations for a permanently empty directory.
1222 static struct dentry *empty_dir_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
1224 return ERR_PTR(-ENOENT);
1227 static int empty_dir_getattr(const struct path *path, struct kstat *stat,
1228 u32 request_mask, unsigned int query_flags)
1230 struct inode *inode = d_inode(path->dentry);
1231 generic_fillattr(inode, stat);
1235 static int empty_dir_setattr(struct dentry *dentry, struct iattr *attr)
1240 static ssize_t empty_dir_listxattr(struct dentry *dentry, char *list, size_t size)
1245 static const struct inode_operations empty_dir_inode_operations = {
1246 .lookup = empty_dir_lookup,
1247 .permission = generic_permission,
1248 .setattr = empty_dir_setattr,
1249 .getattr = empty_dir_getattr,
1250 .listxattr = empty_dir_listxattr,
1253 static loff_t empty_dir_llseek(struct file *file, loff_t offset, int whence)
1255 /* An empty directory has two entries . and .. at offsets 0 and 1 */
1256 return generic_file_llseek_size(file, offset, whence, 2, 2);
1259 static int empty_dir_readdir(struct file *file, struct dir_context *ctx)
1261 dir_emit_dots(file, ctx);
1265 static const struct file_operations empty_dir_operations = {
1266 .llseek = empty_dir_llseek,
1267 .read = generic_read_dir,
1268 .iterate_shared = empty_dir_readdir,
1269 .fsync = noop_fsync,
1273 void make_empty_dir_inode(struct inode *inode)
1275 set_nlink(inode, 2);
1276 inode->i_mode = S_IFDIR | S_IRUGO | S_IXUGO;
1277 inode->i_uid = GLOBAL_ROOT_UID;
1278 inode->i_gid = GLOBAL_ROOT_GID;
1281 inode->i_blkbits = PAGE_SHIFT;
1282 inode->i_blocks = 0;
1284 inode->i_op = &empty_dir_inode_operations;
1285 inode->i_opflags &= ~IOP_XATTR;
1286 inode->i_fop = &empty_dir_operations;
1289 bool is_empty_dir_inode(struct inode *inode)
1291 return (inode->i_fop == &empty_dir_operations) &&
1292 (inode->i_op == &empty_dir_inode_operations);