]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - fs/libfs.c
VFS: Comment mount following code
[mirror_ubuntu-artful-kernel.git] / fs / libfs.c
CommitLineData
1da177e4
LT
1/*
2 * fs/libfs.c
3 * Library for filesystems writers.
4 */
5
630d9c47 6#include <linux/export.h>
1da177e4 7#include <linux/pagemap.h>
5a0e3ad6 8#include <linux/slab.h>
1da177e4
LT
9#include <linux/mount.h>
10#include <linux/vfs.h>
7bb46a67 11#include <linux/quotaops.h>
7cf34c76 12#include <linux/mutex.h>
2596110a 13#include <linux/exportfs.h>
d5aacad5 14#include <linux/writeback.h>
ff01bb48 15#include <linux/buffer_head.h> /* sync_mapping_buffers */
7cf34c76 16
1da177e4
LT
17#include <asm/uaccess.h>
18
a4464dbc
AV
19#include "internal.h"
20
da502956
NP
21static inline int simple_positive(struct dentry *dentry)
22{
23 return dentry->d_inode && !d_unhashed(dentry);
24}
25
1da177e4
LT
26int simple_getattr(struct vfsmount *mnt, struct dentry *dentry,
27 struct kstat *stat)
28{
29 struct inode *inode = dentry->d_inode;
30 generic_fillattr(inode, stat);
31 stat->blocks = inode->i_mapping->nrpages << (PAGE_CACHE_SHIFT - 9);
32 return 0;
33}
34
726c3342 35int simple_statfs(struct dentry *dentry, struct kstatfs *buf)
1da177e4 36{
726c3342 37 buf->f_type = dentry->d_sb->s_magic;
1da177e4
LT
38 buf->f_bsize = PAGE_CACHE_SIZE;
39 buf->f_namelen = NAME_MAX;
40 return 0;
41}
42
43/*
44 * Retaining negative dentries for an in-memory filesystem just wastes
45 * memory and lookup time: arrange for them to be deleted immediately.
46 */
fe15ce44 47static int simple_delete_dentry(const struct dentry *dentry)
1da177e4
LT
48{
49 return 1;
50}
51
52/*
53 * Lookup the data. This is trivial - if the dentry didn't already
54 * exist, we know it is negative. Set d_op to delete negative dentries.
55 */
00cd8dd3 56struct dentry *simple_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
1da177e4 57{
3ba13d17 58 static const struct dentry_operations simple_dentry_operations = {
1da177e4
LT
59 .d_delete = simple_delete_dentry,
60 };
61
62 if (dentry->d_name.len > NAME_MAX)
63 return ERR_PTR(-ENAMETOOLONG);
fb045adb 64 d_set_d_op(dentry, &simple_dentry_operations);
1da177e4
LT
65 d_add(dentry, NULL);
66 return NULL;
67}
68
1da177e4
LT
69int dcache_dir_open(struct inode *inode, struct file *file)
70{
26fe5750 71 static struct qstr cursor_name = QSTR_INIT(".", 1);
1da177e4 72
0f7fc9e4 73 file->private_data = d_alloc(file->f_path.dentry, &cursor_name);
1da177e4
LT
74
75 return file->private_data ? 0 : -ENOMEM;
76}
77
78int dcache_dir_close(struct inode *inode, struct file *file)
79{
80 dput(file->private_data);
81 return 0;
82}
83
84loff_t dcache_dir_lseek(struct file *file, loff_t offset, int origin)
85{
2fd6b7f5
NP
86 struct dentry *dentry = file->f_path.dentry;
87 mutex_lock(&dentry->d_inode->i_mutex);
1da177e4
LT
88 switch (origin) {
89 case 1:
90 offset += file->f_pos;
91 case 0:
92 if (offset >= 0)
93 break;
94 default:
2fd6b7f5 95 mutex_unlock(&dentry->d_inode->i_mutex);
1da177e4
LT
96 return -EINVAL;
97 }
98 if (offset != file->f_pos) {
99 file->f_pos = offset;
100 if (file->f_pos >= 2) {
101 struct list_head *p;
102 struct dentry *cursor = file->private_data;
103 loff_t n = file->f_pos - 2;
104
2fd6b7f5
NP
105 spin_lock(&dentry->d_lock);
106 /* d_lock not required for cursor */
5160ee6f 107 list_del(&cursor->d_u.d_child);
2fd6b7f5
NP
108 p = dentry->d_subdirs.next;
109 while (n && p != &dentry->d_subdirs) {
1da177e4 110 struct dentry *next;
5160ee6f 111 next = list_entry(p, struct dentry, d_u.d_child);
2fd6b7f5 112 spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED);
da502956 113 if (simple_positive(next))
1da177e4 114 n--;
da502956 115 spin_unlock(&next->d_lock);
1da177e4
LT
116 p = p->next;
117 }
5160ee6f 118 list_add_tail(&cursor->d_u.d_child, p);
2fd6b7f5 119 spin_unlock(&dentry->d_lock);
1da177e4
LT
120 }
121 }
2fd6b7f5 122 mutex_unlock(&dentry->d_inode->i_mutex);
1da177e4
LT
123 return offset;
124}
125
126/* Relationship between i_mode and the DT_xxx types */
127static inline unsigned char dt_type(struct inode *inode)
128{
129 return (inode->i_mode >> 12) & 15;
130}
131
132/*
133 * Directory is locked and all positive dentries in it are safe, since
134 * for ramfs-type trees they can't go away without unlink() or rmdir(),
135 * both impossible due to the lock on directory.
136 */
137
138int dcache_readdir(struct file * filp, void * dirent, filldir_t filldir)
139{
0f7fc9e4 140 struct dentry *dentry = filp->f_path.dentry;
1da177e4 141 struct dentry *cursor = filp->private_data;
5160ee6f 142 struct list_head *p, *q = &cursor->d_u.d_child;
1da177e4
LT
143 ino_t ino;
144 int i = filp->f_pos;
145
146 switch (i) {
147 case 0:
148 ino = dentry->d_inode->i_ino;
149 if (filldir(dirent, ".", 1, i, ino, DT_DIR) < 0)
150 break;
151 filp->f_pos++;
152 i++;
153 /* fallthrough */
154 case 1:
155 ino = parent_ino(dentry);
156 if (filldir(dirent, "..", 2, i, ino, DT_DIR) < 0)
157 break;
158 filp->f_pos++;
159 i++;
160 /* fallthrough */
161 default:
2fd6b7f5 162 spin_lock(&dentry->d_lock);
1bfba4e8
AM
163 if (filp->f_pos == 2)
164 list_move(q, &dentry->d_subdirs);
165
1da177e4
LT
166 for (p=q->next; p != &dentry->d_subdirs; p=p->next) {
167 struct dentry *next;
5160ee6f 168 next = list_entry(p, struct dentry, d_u.d_child);
da502956
NP
169 spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED);
170 if (!simple_positive(next)) {
171 spin_unlock(&next->d_lock);
1da177e4 172 continue;
da502956 173 }
1da177e4 174
da502956 175 spin_unlock(&next->d_lock);
2fd6b7f5 176 spin_unlock(&dentry->d_lock);
0f8952c2
RN
177 if (filldir(dirent, next->d_name.name,
178 next->d_name.len, filp->f_pos,
179 next->d_inode->i_ino,
180 dt_type(next->d_inode)) < 0)
1da177e4 181 return 0;
2fd6b7f5
NP
182 spin_lock(&dentry->d_lock);
183 spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED);
1da177e4 184 /* next is still alive */
1bfba4e8 185 list_move(q, p);
2fd6b7f5 186 spin_unlock(&next->d_lock);
1da177e4
LT
187 p = q;
188 filp->f_pos++;
189 }
2fd6b7f5 190 spin_unlock(&dentry->d_lock);
1da177e4
LT
191 }
192 return 0;
193}
194
195ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos)
196{
197 return -EISDIR;
198}
199
4b6f5d20 200const struct file_operations simple_dir_operations = {
1da177e4
LT
201 .open = dcache_dir_open,
202 .release = dcache_dir_close,
203 .llseek = dcache_dir_lseek,
204 .read = generic_read_dir,
205 .readdir = dcache_readdir,
1b061d92 206 .fsync = noop_fsync,
1da177e4
LT
207};
208
92e1d5be 209const struct inode_operations simple_dir_inode_operations = {
1da177e4
LT
210 .lookup = simple_lookup,
211};
212
759b9775
HD
213static const struct super_operations simple_super_operations = {
214 .statfs = simple_statfs,
215};
216
1da177e4
LT
217/*
218 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
219 * will never be mountable)
220 */
51139ada 221struct dentry *mount_pseudo(struct file_system_type *fs_type, char *name,
c74a1cbb
AV
222 const struct super_operations *ops,
223 const struct dentry_operations *dops, unsigned long magic)
1da177e4
LT
224{
225 struct super_block *s = sget(fs_type, NULL, set_anon_super, NULL);
1da177e4
LT
226 struct dentry *dentry;
227 struct inode *root;
26fe5750 228 struct qstr d_name = QSTR_INIT(name, strlen(name));
1da177e4
LT
229
230 if (IS_ERR(s))
51139ada 231 return ERR_CAST(s);
1da177e4
LT
232
233 s->s_flags = MS_NOUSER;
89a4eb4b 234 s->s_maxbytes = MAX_LFS_FILESIZE;
3971e1a9
AN
235 s->s_blocksize = PAGE_SIZE;
236 s->s_blocksize_bits = PAGE_SHIFT;
1da177e4 237 s->s_magic = magic;
759b9775 238 s->s_op = ops ? ops : &simple_super_operations;
1da177e4
LT
239 s->s_time_gran = 1;
240 root = new_inode(s);
241 if (!root)
242 goto Enomem;
1a1c9bb4
JL
243 /*
244 * since this is the first inode, make it number 1. New inodes created
245 * after this must take care not to collide with it (by passing
246 * max_reserved of 1 to iunique).
247 */
248 root->i_ino = 1;
1da177e4 249 root->i_mode = S_IFDIR | S_IRUSR | S_IWUSR;
1da177e4 250 root->i_atime = root->i_mtime = root->i_ctime = CURRENT_TIME;
a4464dbc 251 dentry = __d_alloc(s, &d_name);
1da177e4
LT
252 if (!dentry) {
253 iput(root);
254 goto Enomem;
255 }
1da177e4
LT
256 d_instantiate(dentry, root);
257 s->s_root = dentry;
c74a1cbb 258 s->s_d_op = dops;
1da177e4 259 s->s_flags |= MS_ACTIVE;
51139ada 260 return dget(s->s_root);
1da177e4
LT
261
262Enomem:
6f5bbff9 263 deactivate_locked_super(s);
51139ada 264 return ERR_PTR(-ENOMEM);
1da177e4
LT
265}
266
20955e89
SB
267int simple_open(struct inode *inode, struct file *file)
268{
269 if (inode->i_private)
270 file->private_data = inode->i_private;
271 return 0;
272}
273
1da177e4
LT
274int simple_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
275{
276 struct inode *inode = old_dentry->d_inode;
277
278 inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
d8c76e6f 279 inc_nlink(inode);
7de9c6ee 280 ihold(inode);
1da177e4
LT
281 dget(dentry);
282 d_instantiate(dentry, inode);
283 return 0;
284}
285
1da177e4
LT
286int simple_empty(struct dentry *dentry)
287{
288 struct dentry *child;
289 int ret = 0;
290
2fd6b7f5 291 spin_lock(&dentry->d_lock);
da502956
NP
292 list_for_each_entry(child, &dentry->d_subdirs, d_u.d_child) {
293 spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED);
294 if (simple_positive(child)) {
295 spin_unlock(&child->d_lock);
1da177e4 296 goto out;
da502956
NP
297 }
298 spin_unlock(&child->d_lock);
299 }
1da177e4
LT
300 ret = 1;
301out:
2fd6b7f5 302 spin_unlock(&dentry->d_lock);
1da177e4
LT
303 return ret;
304}
305
306int simple_unlink(struct inode *dir, struct dentry *dentry)
307{
308 struct inode *inode = dentry->d_inode;
309
310 inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
9a53c3a7 311 drop_nlink(inode);
1da177e4
LT
312 dput(dentry);
313 return 0;
314}
315
316int simple_rmdir(struct inode *dir, struct dentry *dentry)
317{
318 if (!simple_empty(dentry))
319 return -ENOTEMPTY;
320
9a53c3a7 321 drop_nlink(dentry->d_inode);
1da177e4 322 simple_unlink(dir, dentry);
9a53c3a7 323 drop_nlink(dir);
1da177e4
LT
324 return 0;
325}
326
327int simple_rename(struct inode *old_dir, struct dentry *old_dentry,
328 struct inode *new_dir, struct dentry *new_dentry)
329{
330 struct inode *inode = old_dentry->d_inode;
331 int they_are_dirs = S_ISDIR(old_dentry->d_inode->i_mode);
332
333 if (!simple_empty(new_dentry))
334 return -ENOTEMPTY;
335
336 if (new_dentry->d_inode) {
337 simple_unlink(new_dir, new_dentry);
841590ce
AV
338 if (they_are_dirs) {
339 drop_nlink(new_dentry->d_inode);
9a53c3a7 340 drop_nlink(old_dir);
841590ce 341 }
1da177e4 342 } else if (they_are_dirs) {
9a53c3a7 343 drop_nlink(old_dir);
d8c76e6f 344 inc_nlink(new_dir);
1da177e4
LT
345 }
346
347 old_dir->i_ctime = old_dir->i_mtime = new_dir->i_ctime =
348 new_dir->i_mtime = inode->i_ctime = CURRENT_TIME;
349
350 return 0;
351}
352
7bb46a67 353/**
eef2380c 354 * simple_setattr - setattr for simple filesystem
7bb46a67 355 * @dentry: dentry
356 * @iattr: iattr structure
357 *
358 * Returns 0 on success, -error on failure.
359 *
eef2380c
CH
360 * simple_setattr is a simple ->setattr implementation without a proper
361 * implementation of size changes.
362 *
363 * It can either be used for in-memory filesystems or special files
364 * on simple regular filesystems. Anything that needs to change on-disk
365 * or wire state on size changes needs its own setattr method.
7bb46a67 366 */
367int simple_setattr(struct dentry *dentry, struct iattr *iattr)
368{
369 struct inode *inode = dentry->d_inode;
370 int error;
371
eef2380c
CH
372 WARN_ON_ONCE(inode->i_op->truncate);
373
7bb46a67 374 error = inode_change_ok(inode, iattr);
375 if (error)
376 return error;
377
2c27c65e
CH
378 if (iattr->ia_valid & ATTR_SIZE)
379 truncate_setsize(inode, iattr->ia_size);
6a1a90ad 380 setattr_copy(inode, iattr);
eef2380c
CH
381 mark_inode_dirty(inode);
382 return 0;
7bb46a67 383}
384EXPORT_SYMBOL(simple_setattr);
385
1da177e4
LT
386int simple_readpage(struct file *file, struct page *page)
387{
c0d92cbc 388 clear_highpage(page);
1da177e4
LT
389 flush_dcache_page(page);
390 SetPageUptodate(page);
1da177e4
LT
391 unlock_page(page);
392 return 0;
393}
394
afddba49
NP
395int simple_write_begin(struct file *file, struct address_space *mapping,
396 loff_t pos, unsigned len, unsigned flags,
397 struct page **pagep, void **fsdata)
398{
399 struct page *page;
400 pgoff_t index;
afddba49
NP
401
402 index = pos >> PAGE_CACHE_SHIFT;
afddba49 403
54566b2c 404 page = grab_cache_page_write_begin(mapping, index, flags);
afddba49
NP
405 if (!page)
406 return -ENOMEM;
407
408 *pagep = page;
409
193cf4b9
BH
410 if (!PageUptodate(page) && (len != PAGE_CACHE_SIZE)) {
411 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
412
413 zero_user_segments(page, 0, from, from + len, PAGE_CACHE_SIZE);
414 }
415 return 0;
afddba49
NP
416}
417
ad2a722f
BH
418/**
419 * simple_write_end - .write_end helper for non-block-device FSes
420 * @available: See .write_end of address_space_operations
421 * @file: "
422 * @mapping: "
423 * @pos: "
424 * @len: "
425 * @copied: "
426 * @page: "
427 * @fsdata: "
428 *
429 * simple_write_end does the minimum needed for updating a page after writing is
430 * done. It has the same API signature as the .write_end of
431 * address_space_operations vector. So it can just be set onto .write_end for
432 * FSes that don't need any other processing. i_mutex is assumed to be held.
433 * Block based filesystems should use generic_write_end().
434 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
435 * is not called, so a filesystem that actually does store data in .write_inode
436 * should extend on what's done here with a call to mark_inode_dirty() in the
437 * case that i_size has changed.
438 */
afddba49
NP
439int simple_write_end(struct file *file, struct address_space *mapping,
440 loff_t pos, unsigned len, unsigned copied,
441 struct page *page, void *fsdata)
442{
ad2a722f
BH
443 struct inode *inode = page->mapping->host;
444 loff_t last_pos = pos + copied;
afddba49
NP
445
446 /* zero the stale part of the page if we did a short copy */
447 if (copied < len) {
ad2a722f
BH
448 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
449
450 zero_user(page, from + copied, len - copied);
afddba49
NP
451 }
452
ad2a722f
BH
453 if (!PageUptodate(page))
454 SetPageUptodate(page);
455 /*
456 * No need to use i_size_read() here, the i_size
457 * cannot change under us because we hold the i_mutex.
458 */
459 if (last_pos > inode->i_size)
460 i_size_write(inode, last_pos);
afddba49 461
ad2a722f 462 set_page_dirty(page);
afddba49
NP
463 unlock_page(page);
464 page_cache_release(page);
465
466 return copied;
467}
468
1a1c9bb4
JL
469/*
470 * the inodes created here are not hashed. If you use iunique to generate
471 * unique inode values later for this filesystem, then you must take care
472 * to pass it an appropriate max_reserved value to avoid collisions.
473 */
7d683a09
RS
474int simple_fill_super(struct super_block *s, unsigned long magic,
475 struct tree_descr *files)
1da177e4 476{
1da177e4
LT
477 struct inode *inode;
478 struct dentry *root;
479 struct dentry *dentry;
480 int i;
481
482 s->s_blocksize = PAGE_CACHE_SIZE;
483 s->s_blocksize_bits = PAGE_CACHE_SHIFT;
484 s->s_magic = magic;
759b9775 485 s->s_op = &simple_super_operations;
1da177e4
LT
486 s->s_time_gran = 1;
487
488 inode = new_inode(s);
489 if (!inode)
490 return -ENOMEM;
1a1c9bb4
JL
491 /*
492 * because the root inode is 1, the files array must not contain an
493 * entry at index 1
494 */
495 inode->i_ino = 1;
1da177e4 496 inode->i_mode = S_IFDIR | 0755;
1da177e4
LT
497 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
498 inode->i_op = &simple_dir_inode_operations;
499 inode->i_fop = &simple_dir_operations;
bfe86848 500 set_nlink(inode, 2);
48fde701
AV
501 root = d_make_root(inode);
502 if (!root)
1da177e4 503 return -ENOMEM;
1da177e4
LT
504 for (i = 0; !files->name || files->name[0]; i++, files++) {
505 if (!files->name)
506 continue;
1a1c9bb4
JL
507
508 /* warn if it tries to conflict with the root inode */
509 if (unlikely(i == 1))
510 printk(KERN_WARNING "%s: %s passed in a files array"
511 "with an index of 1!\n", __func__,
512 s->s_type->name);
513
1da177e4
LT
514 dentry = d_alloc_name(root, files->name);
515 if (!dentry)
516 goto out;
517 inode = new_inode(s);
32096ea1
KK
518 if (!inode) {
519 dput(dentry);
1da177e4 520 goto out;
32096ea1 521 }
1da177e4 522 inode->i_mode = S_IFREG | files->mode;
1da177e4
LT
523 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
524 inode->i_fop = files->ops;
525 inode->i_ino = i;
526 d_add(dentry, inode);
527 }
528 s->s_root = root;
529 return 0;
530out:
531 d_genocide(root);
640946f2 532 shrink_dcache_parent(root);
1da177e4
LT
533 dput(root);
534 return -ENOMEM;
535}
536
537static DEFINE_SPINLOCK(pin_fs_lock);
538
1f5ce9e9 539int simple_pin_fs(struct file_system_type *type, struct vfsmount **mount, int *count)
1da177e4
LT
540{
541 struct vfsmount *mnt = NULL;
542 spin_lock(&pin_fs_lock);
543 if (unlikely(!*mount)) {
544 spin_unlock(&pin_fs_lock);
2452992a 545 mnt = vfs_kern_mount(type, MS_KERNMOUNT, type->name, NULL);
1da177e4
LT
546 if (IS_ERR(mnt))
547 return PTR_ERR(mnt);
548 spin_lock(&pin_fs_lock);
549 if (!*mount)
550 *mount = mnt;
551 }
552 mntget(*mount);
553 ++*count;
554 spin_unlock(&pin_fs_lock);
555 mntput(mnt);
556 return 0;
557}
558
559void simple_release_fs(struct vfsmount **mount, int *count)
560{
561 struct vfsmount *mnt;
562 spin_lock(&pin_fs_lock);
563 mnt = *mount;
564 if (!--*count)
565 *mount = NULL;
566 spin_unlock(&pin_fs_lock);
567 mntput(mnt);
568}
569
6d1029b5
AM
570/**
571 * simple_read_from_buffer - copy data from the buffer to user space
572 * @to: the user space buffer to read to
573 * @count: the maximum number of bytes to read
574 * @ppos: the current position in the buffer
575 * @from: the buffer to read from
576 * @available: the size of the buffer
577 *
578 * The simple_read_from_buffer() function reads up to @count bytes from the
579 * buffer @from at offset @ppos into the user space address starting at @to.
580 *
581 * On success, the number of bytes read is returned and the offset @ppos is
582 * advanced by this number, or negative value is returned on error.
583 **/
1da177e4
LT
584ssize_t simple_read_from_buffer(void __user *to, size_t count, loff_t *ppos,
585 const void *from, size_t available)
586{
587 loff_t pos = *ppos;
14be2746
SR
588 size_t ret;
589
1da177e4
LT
590 if (pos < 0)
591 return -EINVAL;
14be2746 592 if (pos >= available || !count)
1da177e4
LT
593 return 0;
594 if (count > available - pos)
595 count = available - pos;
14be2746
SR
596 ret = copy_to_user(to, from + pos, count);
597 if (ret == count)
1da177e4 598 return -EFAULT;
14be2746 599 count -= ret;
1da177e4
LT
600 *ppos = pos + count;
601 return count;
602}
603
6a727b43
JS
604/**
605 * simple_write_to_buffer - copy data from user space to the buffer
606 * @to: the buffer to write to
607 * @available: the size of the buffer
608 * @ppos: the current position in the buffer
609 * @from: the user space buffer to read from
610 * @count: the maximum number of bytes to read
611 *
612 * The simple_write_to_buffer() function reads up to @count bytes from the user
613 * space address starting at @from into the buffer @to at offset @ppos.
614 *
615 * On success, the number of bytes written is returned and the offset @ppos is
616 * advanced by this number, or negative value is returned on error.
617 **/
618ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
619 const void __user *from, size_t count)
620{
621 loff_t pos = *ppos;
622 size_t res;
623
624 if (pos < 0)
625 return -EINVAL;
626 if (pos >= available || !count)
627 return 0;
628 if (count > available - pos)
629 count = available - pos;
630 res = copy_from_user(to + pos, from, count);
631 if (res == count)
632 return -EFAULT;
633 count -= res;
634 *ppos = pos + count;
635 return count;
636}
637
6d1029b5
AM
638/**
639 * memory_read_from_buffer - copy data from the buffer
640 * @to: the kernel space buffer to read to
641 * @count: the maximum number of bytes to read
642 * @ppos: the current position in the buffer
643 * @from: the buffer to read from
644 * @available: the size of the buffer
645 *
646 * The memory_read_from_buffer() function reads up to @count bytes from the
647 * buffer @from at offset @ppos into the kernel space address starting at @to.
648 *
649 * On success, the number of bytes read is returned and the offset @ppos is
650 * advanced by this number, or negative value is returned on error.
651 **/
93b07113
AM
652ssize_t memory_read_from_buffer(void *to, size_t count, loff_t *ppos,
653 const void *from, size_t available)
654{
655 loff_t pos = *ppos;
656
657 if (pos < 0)
658 return -EINVAL;
659 if (pos >= available)
660 return 0;
661 if (count > available - pos)
662 count = available - pos;
663 memcpy(to, from + pos, count);
664 *ppos = pos + count;
665
666 return count;
667}
668
1da177e4
LT
669/*
670 * Transaction based IO.
671 * The file expects a single write which triggers the transaction, and then
672 * possibly a read which collects the result - which is stored in a
673 * file-local buffer.
674 */
76791ab2
IM
675
676void simple_transaction_set(struct file *file, size_t n)
677{
678 struct simple_transaction_argresp *ar = file->private_data;
679
680 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
681
682 /*
683 * The barrier ensures that ar->size will really remain zero until
684 * ar->data is ready for reading.
685 */
686 smp_mb();
687 ar->size = n;
688}
689
1da177e4
LT
690char *simple_transaction_get(struct file *file, const char __user *buf, size_t size)
691{
692 struct simple_transaction_argresp *ar;
693 static DEFINE_SPINLOCK(simple_transaction_lock);
694
695 if (size > SIMPLE_TRANSACTION_LIMIT - 1)
696 return ERR_PTR(-EFBIG);
697
698 ar = (struct simple_transaction_argresp *)get_zeroed_page(GFP_KERNEL);
699 if (!ar)
700 return ERR_PTR(-ENOMEM);
701
702 spin_lock(&simple_transaction_lock);
703
704 /* only one write allowed per open */
705 if (file->private_data) {
706 spin_unlock(&simple_transaction_lock);
707 free_page((unsigned long)ar);
708 return ERR_PTR(-EBUSY);
709 }
710
711 file->private_data = ar;
712
713 spin_unlock(&simple_transaction_lock);
714
715 if (copy_from_user(ar->data, buf, size))
716 return ERR_PTR(-EFAULT);
717
718 return ar->data;
719}
720
721ssize_t simple_transaction_read(struct file *file, char __user *buf, size_t size, loff_t *pos)
722{
723 struct simple_transaction_argresp *ar = file->private_data;
724
725 if (!ar)
726 return 0;
727 return simple_read_from_buffer(buf, size, pos, ar->data, ar->size);
728}
729
730int simple_transaction_release(struct inode *inode, struct file *file)
731{
732 free_page((unsigned long)file->private_data);
733 return 0;
734}
735
acaefc25
AB
736/* Simple attribute files */
737
738struct simple_attr {
8b88b099
CH
739 int (*get)(void *, u64 *);
740 int (*set)(void *, u64);
acaefc25
AB
741 char get_buf[24]; /* enough to store a u64 and "\n\0" */
742 char set_buf[24];
743 void *data;
744 const char *fmt; /* format for read operation */
7cf34c76 745 struct mutex mutex; /* protects access to these buffers */
acaefc25
AB
746};
747
748/* simple_attr_open is called by an actual attribute open file operation
749 * to set the attribute specific access operations. */
750int simple_attr_open(struct inode *inode, struct file *file,
8b88b099 751 int (*get)(void *, u64 *), int (*set)(void *, u64),
acaefc25
AB
752 const char *fmt)
753{
754 struct simple_attr *attr;
755
756 attr = kmalloc(sizeof(*attr), GFP_KERNEL);
757 if (!attr)
758 return -ENOMEM;
759
760 attr->get = get;
761 attr->set = set;
8e18e294 762 attr->data = inode->i_private;
acaefc25 763 attr->fmt = fmt;
7cf34c76 764 mutex_init(&attr->mutex);
acaefc25
AB
765
766 file->private_data = attr;
767
768 return nonseekable_open(inode, file);
769}
770
74bedc4d 771int simple_attr_release(struct inode *inode, struct file *file)
acaefc25
AB
772{
773 kfree(file->private_data);
774 return 0;
775}
776
777/* read from the buffer that is filled with the get function */
778ssize_t simple_attr_read(struct file *file, char __user *buf,
779 size_t len, loff_t *ppos)
780{
781 struct simple_attr *attr;
782 size_t size;
783 ssize_t ret;
784
785 attr = file->private_data;
786
787 if (!attr->get)
788 return -EACCES;
789
9261303a
CH
790 ret = mutex_lock_interruptible(&attr->mutex);
791 if (ret)
792 return ret;
793
8b88b099 794 if (*ppos) { /* continued read */
acaefc25 795 size = strlen(attr->get_buf);
8b88b099
CH
796 } else { /* first read */
797 u64 val;
798 ret = attr->get(attr->data, &val);
799 if (ret)
800 goto out;
801
acaefc25 802 size = scnprintf(attr->get_buf, sizeof(attr->get_buf),
8b88b099
CH
803 attr->fmt, (unsigned long long)val);
804 }
acaefc25
AB
805
806 ret = simple_read_from_buffer(buf, len, ppos, attr->get_buf, size);
8b88b099 807out:
7cf34c76 808 mutex_unlock(&attr->mutex);
acaefc25
AB
809 return ret;
810}
811
812/* interpret the buffer as a number to call the set function with */
813ssize_t simple_attr_write(struct file *file, const char __user *buf,
814 size_t len, loff_t *ppos)
815{
816 struct simple_attr *attr;
817 u64 val;
818 size_t size;
819 ssize_t ret;
820
821 attr = file->private_data;
acaefc25
AB
822 if (!attr->set)
823 return -EACCES;
824
9261303a
CH
825 ret = mutex_lock_interruptible(&attr->mutex);
826 if (ret)
827 return ret;
828
acaefc25
AB
829 ret = -EFAULT;
830 size = min(sizeof(attr->set_buf) - 1, len);
831 if (copy_from_user(attr->set_buf, buf, size))
832 goto out;
833
acaefc25 834 attr->set_buf[size] = '\0';
f7b88631 835 val = simple_strtoll(attr->set_buf, NULL, 0);
05cc0cee
WF
836 ret = attr->set(attr->data, val);
837 if (ret == 0)
838 ret = len; /* on success, claim we got the whole input */
acaefc25 839out:
7cf34c76 840 mutex_unlock(&attr->mutex);
acaefc25
AB
841 return ret;
842}
843
2596110a
CH
844/**
845 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
846 * @sb: filesystem to do the file handle conversion on
847 * @fid: file handle to convert
848 * @fh_len: length of the file handle in bytes
849 * @fh_type: type of file handle
850 * @get_inode: filesystem callback to retrieve inode
851 *
852 * This function decodes @fid as long as it has one of the well-known
853 * Linux filehandle types and calls @get_inode on it to retrieve the
854 * inode for the object specified in the file handle.
855 */
856struct dentry *generic_fh_to_dentry(struct super_block *sb, struct fid *fid,
857 int fh_len, int fh_type, struct inode *(*get_inode)
858 (struct super_block *sb, u64 ino, u32 gen))
859{
860 struct inode *inode = NULL;
861
862 if (fh_len < 2)
863 return NULL;
864
865 switch (fh_type) {
866 case FILEID_INO32_GEN:
867 case FILEID_INO32_GEN_PARENT:
868 inode = get_inode(sb, fid->i32.ino, fid->i32.gen);
869 break;
870 }
871
4ea3ada2 872 return d_obtain_alias(inode);
2596110a
CH
873}
874EXPORT_SYMBOL_GPL(generic_fh_to_dentry);
875
876/**
877 * generic_fh_to_dentry - generic helper for the fh_to_parent export operation
878 * @sb: filesystem to do the file handle conversion on
879 * @fid: file handle to convert
880 * @fh_len: length of the file handle in bytes
881 * @fh_type: type of file handle
882 * @get_inode: filesystem callback to retrieve inode
883 *
884 * This function decodes @fid as long as it has one of the well-known
885 * Linux filehandle types and calls @get_inode on it to retrieve the
886 * inode for the _parent_ object specified in the file handle if it
887 * is specified in the file handle, or NULL otherwise.
888 */
889struct dentry *generic_fh_to_parent(struct super_block *sb, struct fid *fid,
890 int fh_len, int fh_type, struct inode *(*get_inode)
891 (struct super_block *sb, u64 ino, u32 gen))
892{
893 struct inode *inode = NULL;
894
895 if (fh_len <= 2)
896 return NULL;
897
898 switch (fh_type) {
899 case FILEID_INO32_GEN_PARENT:
900 inode = get_inode(sb, fid->i32.parent_ino,
901 (fh_len > 3 ? fid->i32.parent_gen : 0));
902 break;
903 }
904
4ea3ada2 905 return d_obtain_alias(inode);
2596110a
CH
906}
907EXPORT_SYMBOL_GPL(generic_fh_to_parent);
908
1b061d92
CH
909/**
910 * generic_file_fsync - generic fsync implementation for simple filesystems
911 * @file: file to synchronize
912 * @datasync: only synchronize essential metadata if true
913 *
914 * This is a generic implementation of the fsync method for simple
915 * filesystems which track all non-inode metadata in the buffers list
916 * hanging off the address_space structure.
917 */
02c24a82
JB
918int generic_file_fsync(struct file *file, loff_t start, loff_t end,
919 int datasync)
d5aacad5 920{
7ea80859 921 struct inode *inode = file->f_mapping->host;
d5aacad5
AV
922 int err;
923 int ret;
924
02c24a82
JB
925 err = filemap_write_and_wait_range(inode->i_mapping, start, end);
926 if (err)
927 return err;
928
929 mutex_lock(&inode->i_mutex);
d5aacad5
AV
930 ret = sync_mapping_buffers(inode->i_mapping);
931 if (!(inode->i_state & I_DIRTY))
02c24a82 932 goto out;
d5aacad5 933 if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))
02c24a82 934 goto out;
d5aacad5 935
c3765016 936 err = sync_inode_metadata(inode, 1);
d5aacad5
AV
937 if (ret == 0)
938 ret = err;
02c24a82
JB
939out:
940 mutex_unlock(&inode->i_mutex);
d5aacad5
AV
941 return ret;
942}
1b061d92
CH
943EXPORT_SYMBOL(generic_file_fsync);
944
30ca22c7
PL
945/**
946 * generic_check_addressable - Check addressability of file system
947 * @blocksize_bits: log of file system block size
948 * @num_blocks: number of blocks in file system
949 *
950 * Determine whether a file system with @num_blocks blocks (and a
951 * block size of 2**@blocksize_bits) is addressable by the sector_t
952 * and page cache of the system. Return 0 if so and -EFBIG otherwise.
953 */
954int generic_check_addressable(unsigned blocksize_bits, u64 num_blocks)
955{
956 u64 last_fs_block = num_blocks - 1;
a33f13ef
JB
957 u64 last_fs_page =
958 last_fs_block >> (PAGE_CACHE_SHIFT - blocksize_bits);
30ca22c7
PL
959
960 if (unlikely(num_blocks == 0))
961 return 0;
962
963 if ((blocksize_bits < 9) || (blocksize_bits > PAGE_CACHE_SHIFT))
964 return -EINVAL;
965
a33f13ef
JB
966 if ((last_fs_block > (sector_t)(~0ULL) >> (blocksize_bits - 9)) ||
967 (last_fs_page > (pgoff_t)(~0ULL))) {
30ca22c7
PL
968 return -EFBIG;
969 }
970 return 0;
971}
972EXPORT_SYMBOL(generic_check_addressable);
973
1b061d92
CH
974/*
975 * No-op implementation of ->fsync for in-memory filesystems.
976 */
02c24a82 977int noop_fsync(struct file *file, loff_t start, loff_t end, int datasync)
1b061d92
CH
978{
979 return 0;
980}
d5aacad5 981
1da177e4
LT
982EXPORT_SYMBOL(dcache_dir_close);
983EXPORT_SYMBOL(dcache_dir_lseek);
984EXPORT_SYMBOL(dcache_dir_open);
985EXPORT_SYMBOL(dcache_readdir);
986EXPORT_SYMBOL(generic_read_dir);
51139ada 987EXPORT_SYMBOL(mount_pseudo);
afddba49
NP
988EXPORT_SYMBOL(simple_write_begin);
989EXPORT_SYMBOL(simple_write_end);
1da177e4
LT
990EXPORT_SYMBOL(simple_dir_inode_operations);
991EXPORT_SYMBOL(simple_dir_operations);
992EXPORT_SYMBOL(simple_empty);
1da177e4
LT
993EXPORT_SYMBOL(simple_fill_super);
994EXPORT_SYMBOL(simple_getattr);
20955e89 995EXPORT_SYMBOL(simple_open);
1da177e4
LT
996EXPORT_SYMBOL(simple_link);
997EXPORT_SYMBOL(simple_lookup);
998EXPORT_SYMBOL(simple_pin_fs);
1da177e4
LT
999EXPORT_SYMBOL(simple_readpage);
1000EXPORT_SYMBOL(simple_release_fs);
1001EXPORT_SYMBOL(simple_rename);
1002EXPORT_SYMBOL(simple_rmdir);
1003EXPORT_SYMBOL(simple_statfs);
1b061d92 1004EXPORT_SYMBOL(noop_fsync);
1da177e4
LT
1005EXPORT_SYMBOL(simple_unlink);
1006EXPORT_SYMBOL(simple_read_from_buffer);
6a727b43 1007EXPORT_SYMBOL(simple_write_to_buffer);
93b07113 1008EXPORT_SYMBOL(memory_read_from_buffer);
76791ab2 1009EXPORT_SYMBOL(simple_transaction_set);
1da177e4
LT
1010EXPORT_SYMBOL(simple_transaction_get);
1011EXPORT_SYMBOL(simple_transaction_read);
1012EXPORT_SYMBOL(simple_transaction_release);
acaefc25 1013EXPORT_SYMBOL_GPL(simple_attr_open);
74bedc4d 1014EXPORT_SYMBOL_GPL(simple_attr_release);
acaefc25
AB
1015EXPORT_SYMBOL_GPL(simple_attr_read);
1016EXPORT_SYMBOL_GPL(simple_attr_write);