]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - fs/libfs.c
switch hostfs
[mirror_ubuntu-artful-kernel.git] / fs / libfs.c
CommitLineData
1da177e4
LT
1/*
2 * fs/libfs.c
3 * Library for filesystems writers.
4 */
5
6#include <linux/module.h>
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
AV
14#include <linux/writeback.h>
15#include <linux/buffer_head.h>
7cf34c76 16
1da177e4
LT
17#include <asm/uaccess.h>
18
da502956
NP
19static inline int simple_positive(struct dentry *dentry)
20{
21 return dentry->d_inode && !d_unhashed(dentry);
22}
23
1da177e4
LT
24int simple_getattr(struct vfsmount *mnt, struct dentry *dentry,
25 struct kstat *stat)
26{
27 struct inode *inode = dentry->d_inode;
28 generic_fillattr(inode, stat);
29 stat->blocks = inode->i_mapping->nrpages << (PAGE_CACHE_SHIFT - 9);
30 return 0;
31}
32
726c3342 33int simple_statfs(struct dentry *dentry, struct kstatfs *buf)
1da177e4 34{
726c3342 35 buf->f_type = dentry->d_sb->s_magic;
1da177e4
LT
36 buf->f_bsize = PAGE_CACHE_SIZE;
37 buf->f_namelen = NAME_MAX;
38 return 0;
39}
40
41/*
42 * Retaining negative dentries for an in-memory filesystem just wastes
43 * memory and lookup time: arrange for them to be deleted immediately.
44 */
fe15ce44 45static int simple_delete_dentry(const struct dentry *dentry)
1da177e4
LT
46{
47 return 1;
48}
49
50/*
51 * Lookup the data. This is trivial - if the dentry didn't already
52 * exist, we know it is negative. Set d_op to delete negative dentries.
53 */
54struct dentry *simple_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
55{
3ba13d17 56 static const struct dentry_operations simple_dentry_operations = {
1da177e4
LT
57 .d_delete = simple_delete_dentry,
58 };
59
60 if (dentry->d_name.len > NAME_MAX)
61 return ERR_PTR(-ENAMETOOLONG);
fb045adb 62 d_set_d_op(dentry, &simple_dentry_operations);
1da177e4
LT
63 d_add(dentry, NULL);
64 return NULL;
65}
66
1da177e4
LT
67int dcache_dir_open(struct inode *inode, struct file *file)
68{
69 static struct qstr cursor_name = {.len = 1, .name = "."};
70
0f7fc9e4 71 file->private_data = d_alloc(file->f_path.dentry, &cursor_name);
1da177e4
LT
72
73 return file->private_data ? 0 : -ENOMEM;
74}
75
76int dcache_dir_close(struct inode *inode, struct file *file)
77{
78 dput(file->private_data);
79 return 0;
80}
81
82loff_t dcache_dir_lseek(struct file *file, loff_t offset, int origin)
83{
2fd6b7f5
NP
84 struct dentry *dentry = file->f_path.dentry;
85 mutex_lock(&dentry->d_inode->i_mutex);
1da177e4
LT
86 switch (origin) {
87 case 1:
88 offset += file->f_pos;
89 case 0:
90 if (offset >= 0)
91 break;
92 default:
2fd6b7f5 93 mutex_unlock(&dentry->d_inode->i_mutex);
1da177e4
LT
94 return -EINVAL;
95 }
96 if (offset != file->f_pos) {
97 file->f_pos = offset;
98 if (file->f_pos >= 2) {
99 struct list_head *p;
100 struct dentry *cursor = file->private_data;
101 loff_t n = file->f_pos - 2;
102
2fd6b7f5
NP
103 spin_lock(&dentry->d_lock);
104 /* d_lock not required for cursor */
5160ee6f 105 list_del(&cursor->d_u.d_child);
2fd6b7f5
NP
106 p = dentry->d_subdirs.next;
107 while (n && p != &dentry->d_subdirs) {
1da177e4 108 struct dentry *next;
5160ee6f 109 next = list_entry(p, struct dentry, d_u.d_child);
2fd6b7f5 110 spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED);
da502956 111 if (simple_positive(next))
1da177e4 112 n--;
da502956 113 spin_unlock(&next->d_lock);
1da177e4
LT
114 p = p->next;
115 }
5160ee6f 116 list_add_tail(&cursor->d_u.d_child, p);
2fd6b7f5 117 spin_unlock(&dentry->d_lock);
1da177e4
LT
118 }
119 }
2fd6b7f5 120 mutex_unlock(&dentry->d_inode->i_mutex);
1da177e4
LT
121 return offset;
122}
123
124/* Relationship between i_mode and the DT_xxx types */
125static inline unsigned char dt_type(struct inode *inode)
126{
127 return (inode->i_mode >> 12) & 15;
128}
129
130/*
131 * Directory is locked and all positive dentries in it are safe, since
132 * for ramfs-type trees they can't go away without unlink() or rmdir(),
133 * both impossible due to the lock on directory.
134 */
135
136int dcache_readdir(struct file * filp, void * dirent, filldir_t filldir)
137{
0f7fc9e4 138 struct dentry *dentry = filp->f_path.dentry;
1da177e4 139 struct dentry *cursor = filp->private_data;
5160ee6f 140 struct list_head *p, *q = &cursor->d_u.d_child;
1da177e4
LT
141 ino_t ino;
142 int i = filp->f_pos;
143
144 switch (i) {
145 case 0:
146 ino = dentry->d_inode->i_ino;
147 if (filldir(dirent, ".", 1, i, ino, DT_DIR) < 0)
148 break;
149 filp->f_pos++;
150 i++;
151 /* fallthrough */
152 case 1:
153 ino = parent_ino(dentry);
154 if (filldir(dirent, "..", 2, i, ino, DT_DIR) < 0)
155 break;
156 filp->f_pos++;
157 i++;
158 /* fallthrough */
159 default:
2fd6b7f5 160 spin_lock(&dentry->d_lock);
1bfba4e8
AM
161 if (filp->f_pos == 2)
162 list_move(q, &dentry->d_subdirs);
163
1da177e4
LT
164 for (p=q->next; p != &dentry->d_subdirs; p=p->next) {
165 struct dentry *next;
5160ee6f 166 next = list_entry(p, struct dentry, d_u.d_child);
da502956
NP
167 spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED);
168 if (!simple_positive(next)) {
169 spin_unlock(&next->d_lock);
1da177e4 170 continue;
da502956 171 }
1da177e4 172
da502956 173 spin_unlock(&next->d_lock);
2fd6b7f5 174 spin_unlock(&dentry->d_lock);
0f8952c2
RN
175 if (filldir(dirent, next->d_name.name,
176 next->d_name.len, filp->f_pos,
177 next->d_inode->i_ino,
178 dt_type(next->d_inode)) < 0)
1da177e4 179 return 0;
2fd6b7f5
NP
180 spin_lock(&dentry->d_lock);
181 spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED);
1da177e4 182 /* next is still alive */
1bfba4e8 183 list_move(q, p);
2fd6b7f5 184 spin_unlock(&next->d_lock);
1da177e4
LT
185 p = q;
186 filp->f_pos++;
187 }
2fd6b7f5 188 spin_unlock(&dentry->d_lock);
1da177e4
LT
189 }
190 return 0;
191}
192
193ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos)
194{
195 return -EISDIR;
196}
197
4b6f5d20 198const struct file_operations simple_dir_operations = {
1da177e4
LT
199 .open = dcache_dir_open,
200 .release = dcache_dir_close,
201 .llseek = dcache_dir_lseek,
202 .read = generic_read_dir,
203 .readdir = dcache_readdir,
1b061d92 204 .fsync = noop_fsync,
1da177e4
LT
205};
206
92e1d5be 207const struct inode_operations simple_dir_inode_operations = {
1da177e4
LT
208 .lookup = simple_lookup,
209};
210
759b9775
HD
211static const struct super_operations simple_super_operations = {
212 .statfs = simple_statfs,
213};
214
1da177e4
LT
215/*
216 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
217 * will never be mountable)
218 */
51139ada
AV
219struct dentry *mount_pseudo(struct file_system_type *fs_type, char *name,
220 const struct super_operations *ops, unsigned long magic)
1da177e4
LT
221{
222 struct super_block *s = sget(fs_type, NULL, set_anon_super, NULL);
1da177e4
LT
223 struct dentry *dentry;
224 struct inode *root;
225 struct qstr d_name = {.name = name, .len = strlen(name)};
226
227 if (IS_ERR(s))
51139ada 228 return ERR_CAST(s);
1da177e4
LT
229
230 s->s_flags = MS_NOUSER;
89a4eb4b 231 s->s_maxbytes = MAX_LFS_FILESIZE;
3971e1a9
AN
232 s->s_blocksize = PAGE_SIZE;
233 s->s_blocksize_bits = PAGE_SHIFT;
1da177e4 234 s->s_magic = magic;
759b9775 235 s->s_op = ops ? ops : &simple_super_operations;
1da177e4
LT
236 s->s_time_gran = 1;
237 root = new_inode(s);
238 if (!root)
239 goto Enomem;
1a1c9bb4
JL
240 /*
241 * since this is the first inode, make it number 1. New inodes created
242 * after this must take care not to collide with it (by passing
243 * max_reserved of 1 to iunique).
244 */
245 root->i_ino = 1;
1da177e4 246 root->i_mode = S_IFDIR | S_IRUSR | S_IWUSR;
1da177e4
LT
247 root->i_atime = root->i_mtime = root->i_ctime = CURRENT_TIME;
248 dentry = d_alloc(NULL, &d_name);
249 if (!dentry) {
250 iput(root);
251 goto Enomem;
252 }
253 dentry->d_sb = s;
254 dentry->d_parent = dentry;
255 d_instantiate(dentry, root);
256 s->s_root = dentry;
257 s->s_flags |= MS_ACTIVE;
51139ada 258 return dget(s->s_root);
1da177e4
LT
259
260Enomem:
6f5bbff9 261 deactivate_locked_super(s);
51139ada 262 return ERR_PTR(-ENOMEM);
1da177e4
LT
263}
264
265int simple_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
266{
267 struct inode *inode = old_dentry->d_inode;
268
269 inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
d8c76e6f 270 inc_nlink(inode);
7de9c6ee 271 ihold(inode);
1da177e4
LT
272 dget(dentry);
273 d_instantiate(dentry, inode);
274 return 0;
275}
276
1da177e4
LT
277int simple_empty(struct dentry *dentry)
278{
279 struct dentry *child;
280 int ret = 0;
281
2fd6b7f5 282 spin_lock(&dentry->d_lock);
da502956
NP
283 list_for_each_entry(child, &dentry->d_subdirs, d_u.d_child) {
284 spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED);
285 if (simple_positive(child)) {
286 spin_unlock(&child->d_lock);
1da177e4 287 goto out;
da502956
NP
288 }
289 spin_unlock(&child->d_lock);
290 }
1da177e4
LT
291 ret = 1;
292out:
2fd6b7f5 293 spin_unlock(&dentry->d_lock);
1da177e4
LT
294 return ret;
295}
296
297int simple_unlink(struct inode *dir, struct dentry *dentry)
298{
299 struct inode *inode = dentry->d_inode;
300
301 inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
9a53c3a7 302 drop_nlink(inode);
1da177e4
LT
303 dput(dentry);
304 return 0;
305}
306
307int simple_rmdir(struct inode *dir, struct dentry *dentry)
308{
309 if (!simple_empty(dentry))
310 return -ENOTEMPTY;
311
9a53c3a7 312 drop_nlink(dentry->d_inode);
1da177e4 313 simple_unlink(dir, dentry);
9a53c3a7 314 drop_nlink(dir);
1da177e4
LT
315 return 0;
316}
317
318int simple_rename(struct inode *old_dir, struct dentry *old_dentry,
319 struct inode *new_dir, struct dentry *new_dentry)
320{
321 struct inode *inode = old_dentry->d_inode;
322 int they_are_dirs = S_ISDIR(old_dentry->d_inode->i_mode);
323
324 if (!simple_empty(new_dentry))
325 return -ENOTEMPTY;
326
327 if (new_dentry->d_inode) {
328 simple_unlink(new_dir, new_dentry);
329 if (they_are_dirs)
9a53c3a7 330 drop_nlink(old_dir);
1da177e4 331 } else if (they_are_dirs) {
9a53c3a7 332 drop_nlink(old_dir);
d8c76e6f 333 inc_nlink(new_dir);
1da177e4
LT
334 }
335
336 old_dir->i_ctime = old_dir->i_mtime = new_dir->i_ctime =
337 new_dir->i_mtime = inode->i_ctime = CURRENT_TIME;
338
339 return 0;
340}
341
7bb46a67 342/**
eef2380c 343 * simple_setattr - setattr for simple filesystem
7bb46a67 344 * @dentry: dentry
345 * @iattr: iattr structure
346 *
347 * Returns 0 on success, -error on failure.
348 *
eef2380c
CH
349 * simple_setattr is a simple ->setattr implementation without a proper
350 * implementation of size changes.
351 *
352 * It can either be used for in-memory filesystems or special files
353 * on simple regular filesystems. Anything that needs to change on-disk
354 * or wire state on size changes needs its own setattr method.
7bb46a67 355 */
356int simple_setattr(struct dentry *dentry, struct iattr *iattr)
357{
358 struct inode *inode = dentry->d_inode;
359 int error;
360
eef2380c
CH
361 WARN_ON_ONCE(inode->i_op->truncate);
362
7bb46a67 363 error = inode_change_ok(inode, iattr);
364 if (error)
365 return error;
366
2c27c65e
CH
367 if (iattr->ia_valid & ATTR_SIZE)
368 truncate_setsize(inode, iattr->ia_size);
6a1a90ad 369 setattr_copy(inode, iattr);
eef2380c
CH
370 mark_inode_dirty(inode);
371 return 0;
7bb46a67 372}
373EXPORT_SYMBOL(simple_setattr);
374
1da177e4
LT
375int simple_readpage(struct file *file, struct page *page)
376{
c0d92cbc 377 clear_highpage(page);
1da177e4
LT
378 flush_dcache_page(page);
379 SetPageUptodate(page);
1da177e4
LT
380 unlock_page(page);
381 return 0;
382}
383
afddba49
NP
384int simple_write_begin(struct file *file, struct address_space *mapping,
385 loff_t pos, unsigned len, unsigned flags,
386 struct page **pagep, void **fsdata)
387{
388 struct page *page;
389 pgoff_t index;
afddba49
NP
390
391 index = pos >> PAGE_CACHE_SHIFT;
afddba49 392
54566b2c 393 page = grab_cache_page_write_begin(mapping, index, flags);
afddba49
NP
394 if (!page)
395 return -ENOMEM;
396
397 *pagep = page;
398
193cf4b9
BH
399 if (!PageUptodate(page) && (len != PAGE_CACHE_SIZE)) {
400 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
401
402 zero_user_segments(page, 0, from, from + len, PAGE_CACHE_SIZE);
403 }
404 return 0;
afddba49
NP
405}
406
ad2a722f
BH
407/**
408 * simple_write_end - .write_end helper for non-block-device FSes
409 * @available: See .write_end of address_space_operations
410 * @file: "
411 * @mapping: "
412 * @pos: "
413 * @len: "
414 * @copied: "
415 * @page: "
416 * @fsdata: "
417 *
418 * simple_write_end does the minimum needed for updating a page after writing is
419 * done. It has the same API signature as the .write_end of
420 * address_space_operations vector. So it can just be set onto .write_end for
421 * FSes that don't need any other processing. i_mutex is assumed to be held.
422 * Block based filesystems should use generic_write_end().
423 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
424 * is not called, so a filesystem that actually does store data in .write_inode
425 * should extend on what's done here with a call to mark_inode_dirty() in the
426 * case that i_size has changed.
427 */
afddba49
NP
428int simple_write_end(struct file *file, struct address_space *mapping,
429 loff_t pos, unsigned len, unsigned copied,
430 struct page *page, void *fsdata)
431{
ad2a722f
BH
432 struct inode *inode = page->mapping->host;
433 loff_t last_pos = pos + copied;
afddba49
NP
434
435 /* zero the stale part of the page if we did a short copy */
436 if (copied < len) {
ad2a722f
BH
437 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
438
439 zero_user(page, from + copied, len - copied);
afddba49
NP
440 }
441
ad2a722f
BH
442 if (!PageUptodate(page))
443 SetPageUptodate(page);
444 /*
445 * No need to use i_size_read() here, the i_size
446 * cannot change under us because we hold the i_mutex.
447 */
448 if (last_pos > inode->i_size)
449 i_size_write(inode, last_pos);
afddba49 450
ad2a722f 451 set_page_dirty(page);
afddba49
NP
452 unlock_page(page);
453 page_cache_release(page);
454
455 return copied;
456}
457
1a1c9bb4
JL
458/*
459 * the inodes created here are not hashed. If you use iunique to generate
460 * unique inode values later for this filesystem, then you must take care
461 * to pass it an appropriate max_reserved value to avoid collisions.
462 */
7d683a09
RS
463int simple_fill_super(struct super_block *s, unsigned long magic,
464 struct tree_descr *files)
1da177e4 465{
1da177e4
LT
466 struct inode *inode;
467 struct dentry *root;
468 struct dentry *dentry;
469 int i;
470
471 s->s_blocksize = PAGE_CACHE_SIZE;
472 s->s_blocksize_bits = PAGE_CACHE_SHIFT;
473 s->s_magic = magic;
759b9775 474 s->s_op = &simple_super_operations;
1da177e4
LT
475 s->s_time_gran = 1;
476
477 inode = new_inode(s);
478 if (!inode)
479 return -ENOMEM;
1a1c9bb4
JL
480 /*
481 * because the root inode is 1, the files array must not contain an
482 * entry at index 1
483 */
484 inode->i_ino = 1;
1da177e4 485 inode->i_mode = S_IFDIR | 0755;
1da177e4
LT
486 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
487 inode->i_op = &simple_dir_inode_operations;
488 inode->i_fop = &simple_dir_operations;
7656f328 489 inode->i_nlink = 2;
1da177e4
LT
490 root = d_alloc_root(inode);
491 if (!root) {
492 iput(inode);
493 return -ENOMEM;
494 }
495 for (i = 0; !files->name || files->name[0]; i++, files++) {
496 if (!files->name)
497 continue;
1a1c9bb4
JL
498
499 /* warn if it tries to conflict with the root inode */
500 if (unlikely(i == 1))
501 printk(KERN_WARNING "%s: %s passed in a files array"
502 "with an index of 1!\n", __func__,
503 s->s_type->name);
504
1da177e4
LT
505 dentry = d_alloc_name(root, files->name);
506 if (!dentry)
507 goto out;
508 inode = new_inode(s);
509 if (!inode)
510 goto out;
511 inode->i_mode = S_IFREG | files->mode;
1da177e4
LT
512 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
513 inode->i_fop = files->ops;
514 inode->i_ino = i;
515 d_add(dentry, inode);
516 }
517 s->s_root = root;
518 return 0;
519out:
520 d_genocide(root);
521 dput(root);
522 return -ENOMEM;
523}
524
525static DEFINE_SPINLOCK(pin_fs_lock);
526
1f5ce9e9 527int simple_pin_fs(struct file_system_type *type, struct vfsmount **mount, int *count)
1da177e4
LT
528{
529 struct vfsmount *mnt = NULL;
530 spin_lock(&pin_fs_lock);
531 if (unlikely(!*mount)) {
532 spin_unlock(&pin_fs_lock);
1f5ce9e9 533 mnt = vfs_kern_mount(type, 0, type->name, NULL);
1da177e4
LT
534 if (IS_ERR(mnt))
535 return PTR_ERR(mnt);
536 spin_lock(&pin_fs_lock);
537 if (!*mount)
538 *mount = mnt;
539 }
540 mntget(*mount);
541 ++*count;
542 spin_unlock(&pin_fs_lock);
543 mntput(mnt);
544 return 0;
545}
546
547void simple_release_fs(struct vfsmount **mount, int *count)
548{
549 struct vfsmount *mnt;
550 spin_lock(&pin_fs_lock);
551 mnt = *mount;
552 if (!--*count)
553 *mount = NULL;
554 spin_unlock(&pin_fs_lock);
555 mntput(mnt);
556}
557
6d1029b5
AM
558/**
559 * simple_read_from_buffer - copy data from the buffer to user space
560 * @to: the user space buffer to read to
561 * @count: the maximum number of bytes to read
562 * @ppos: the current position in the buffer
563 * @from: the buffer to read from
564 * @available: the size of the buffer
565 *
566 * The simple_read_from_buffer() function reads up to @count bytes from the
567 * buffer @from at offset @ppos into the user space address starting at @to.
568 *
569 * On success, the number of bytes read is returned and the offset @ppos is
570 * advanced by this number, or negative value is returned on error.
571 **/
1da177e4
LT
572ssize_t simple_read_from_buffer(void __user *to, size_t count, loff_t *ppos,
573 const void *from, size_t available)
574{
575 loff_t pos = *ppos;
14be2746
SR
576 size_t ret;
577
1da177e4
LT
578 if (pos < 0)
579 return -EINVAL;
14be2746 580 if (pos >= available || !count)
1da177e4
LT
581 return 0;
582 if (count > available - pos)
583 count = available - pos;
14be2746
SR
584 ret = copy_to_user(to, from + pos, count);
585 if (ret == count)
1da177e4 586 return -EFAULT;
14be2746 587 count -= ret;
1da177e4
LT
588 *ppos = pos + count;
589 return count;
590}
591
6a727b43
JS
592/**
593 * simple_write_to_buffer - copy data from user space to the buffer
594 * @to: the buffer to write to
595 * @available: the size of the buffer
596 * @ppos: the current position in the buffer
597 * @from: the user space buffer to read from
598 * @count: the maximum number of bytes to read
599 *
600 * The simple_write_to_buffer() function reads up to @count bytes from the user
601 * space address starting at @from into the buffer @to at offset @ppos.
602 *
603 * On success, the number of bytes written is returned and the offset @ppos is
604 * advanced by this number, or negative value is returned on error.
605 **/
606ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
607 const void __user *from, size_t count)
608{
609 loff_t pos = *ppos;
610 size_t res;
611
612 if (pos < 0)
613 return -EINVAL;
614 if (pos >= available || !count)
615 return 0;
616 if (count > available - pos)
617 count = available - pos;
618 res = copy_from_user(to + pos, from, count);
619 if (res == count)
620 return -EFAULT;
621 count -= res;
622 *ppos = pos + count;
623 return count;
624}
625
6d1029b5
AM
626/**
627 * memory_read_from_buffer - copy data from the buffer
628 * @to: the kernel space buffer to read to
629 * @count: the maximum number of bytes to read
630 * @ppos: the current position in the buffer
631 * @from: the buffer to read from
632 * @available: the size of the buffer
633 *
634 * The memory_read_from_buffer() function reads up to @count bytes from the
635 * buffer @from at offset @ppos into the kernel space address starting at @to.
636 *
637 * On success, the number of bytes read is returned and the offset @ppos is
638 * advanced by this number, or negative value is returned on error.
639 **/
93b07113
AM
640ssize_t memory_read_from_buffer(void *to, size_t count, loff_t *ppos,
641 const void *from, size_t available)
642{
643 loff_t pos = *ppos;
644
645 if (pos < 0)
646 return -EINVAL;
647 if (pos >= available)
648 return 0;
649 if (count > available - pos)
650 count = available - pos;
651 memcpy(to, from + pos, count);
652 *ppos = pos + count;
653
654 return count;
655}
656
1da177e4
LT
657/*
658 * Transaction based IO.
659 * The file expects a single write which triggers the transaction, and then
660 * possibly a read which collects the result - which is stored in a
661 * file-local buffer.
662 */
76791ab2
IM
663
664void simple_transaction_set(struct file *file, size_t n)
665{
666 struct simple_transaction_argresp *ar = file->private_data;
667
668 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
669
670 /*
671 * The barrier ensures that ar->size will really remain zero until
672 * ar->data is ready for reading.
673 */
674 smp_mb();
675 ar->size = n;
676}
677
1da177e4
LT
678char *simple_transaction_get(struct file *file, const char __user *buf, size_t size)
679{
680 struct simple_transaction_argresp *ar;
681 static DEFINE_SPINLOCK(simple_transaction_lock);
682
683 if (size > SIMPLE_TRANSACTION_LIMIT - 1)
684 return ERR_PTR(-EFBIG);
685
686 ar = (struct simple_transaction_argresp *)get_zeroed_page(GFP_KERNEL);
687 if (!ar)
688 return ERR_PTR(-ENOMEM);
689
690 spin_lock(&simple_transaction_lock);
691
692 /* only one write allowed per open */
693 if (file->private_data) {
694 spin_unlock(&simple_transaction_lock);
695 free_page((unsigned long)ar);
696 return ERR_PTR(-EBUSY);
697 }
698
699 file->private_data = ar;
700
701 spin_unlock(&simple_transaction_lock);
702
703 if (copy_from_user(ar->data, buf, size))
704 return ERR_PTR(-EFAULT);
705
706 return ar->data;
707}
708
709ssize_t simple_transaction_read(struct file *file, char __user *buf, size_t size, loff_t *pos)
710{
711 struct simple_transaction_argresp *ar = file->private_data;
712
713 if (!ar)
714 return 0;
715 return simple_read_from_buffer(buf, size, pos, ar->data, ar->size);
716}
717
718int simple_transaction_release(struct inode *inode, struct file *file)
719{
720 free_page((unsigned long)file->private_data);
721 return 0;
722}
723
acaefc25
AB
724/* Simple attribute files */
725
726struct simple_attr {
8b88b099
CH
727 int (*get)(void *, u64 *);
728 int (*set)(void *, u64);
acaefc25
AB
729 char get_buf[24]; /* enough to store a u64 and "\n\0" */
730 char set_buf[24];
731 void *data;
732 const char *fmt; /* format for read operation */
7cf34c76 733 struct mutex mutex; /* protects access to these buffers */
acaefc25
AB
734};
735
736/* simple_attr_open is called by an actual attribute open file operation
737 * to set the attribute specific access operations. */
738int simple_attr_open(struct inode *inode, struct file *file,
8b88b099 739 int (*get)(void *, u64 *), int (*set)(void *, u64),
acaefc25
AB
740 const char *fmt)
741{
742 struct simple_attr *attr;
743
744 attr = kmalloc(sizeof(*attr), GFP_KERNEL);
745 if (!attr)
746 return -ENOMEM;
747
748 attr->get = get;
749 attr->set = set;
8e18e294 750 attr->data = inode->i_private;
acaefc25 751 attr->fmt = fmt;
7cf34c76 752 mutex_init(&attr->mutex);
acaefc25
AB
753
754 file->private_data = attr;
755
756 return nonseekable_open(inode, file);
757}
758
74bedc4d 759int simple_attr_release(struct inode *inode, struct file *file)
acaefc25
AB
760{
761 kfree(file->private_data);
762 return 0;
763}
764
765/* read from the buffer that is filled with the get function */
766ssize_t simple_attr_read(struct file *file, char __user *buf,
767 size_t len, loff_t *ppos)
768{
769 struct simple_attr *attr;
770 size_t size;
771 ssize_t ret;
772
773 attr = file->private_data;
774
775 if (!attr->get)
776 return -EACCES;
777
9261303a
CH
778 ret = mutex_lock_interruptible(&attr->mutex);
779 if (ret)
780 return ret;
781
8b88b099 782 if (*ppos) { /* continued read */
acaefc25 783 size = strlen(attr->get_buf);
8b88b099
CH
784 } else { /* first read */
785 u64 val;
786 ret = attr->get(attr->data, &val);
787 if (ret)
788 goto out;
789
acaefc25 790 size = scnprintf(attr->get_buf, sizeof(attr->get_buf),
8b88b099
CH
791 attr->fmt, (unsigned long long)val);
792 }
acaefc25
AB
793
794 ret = simple_read_from_buffer(buf, len, ppos, attr->get_buf, size);
8b88b099 795out:
7cf34c76 796 mutex_unlock(&attr->mutex);
acaefc25
AB
797 return ret;
798}
799
800/* interpret the buffer as a number to call the set function with */
801ssize_t simple_attr_write(struct file *file, const char __user *buf,
802 size_t len, loff_t *ppos)
803{
804 struct simple_attr *attr;
805 u64 val;
806 size_t size;
807 ssize_t ret;
808
809 attr = file->private_data;
acaefc25
AB
810 if (!attr->set)
811 return -EACCES;
812
9261303a
CH
813 ret = mutex_lock_interruptible(&attr->mutex);
814 if (ret)
815 return ret;
816
acaefc25
AB
817 ret = -EFAULT;
818 size = min(sizeof(attr->set_buf) - 1, len);
819 if (copy_from_user(attr->set_buf, buf, size))
820 goto out;
821
acaefc25
AB
822 attr->set_buf[size] = '\0';
823 val = simple_strtol(attr->set_buf, NULL, 0);
05cc0cee
WF
824 ret = attr->set(attr->data, val);
825 if (ret == 0)
826 ret = len; /* on success, claim we got the whole input */
acaefc25 827out:
7cf34c76 828 mutex_unlock(&attr->mutex);
acaefc25
AB
829 return ret;
830}
831
2596110a
CH
832/**
833 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
834 * @sb: filesystem to do the file handle conversion on
835 * @fid: file handle to convert
836 * @fh_len: length of the file handle in bytes
837 * @fh_type: type of file handle
838 * @get_inode: filesystem callback to retrieve inode
839 *
840 * This function decodes @fid as long as it has one of the well-known
841 * Linux filehandle types and calls @get_inode on it to retrieve the
842 * inode for the object specified in the file handle.
843 */
844struct dentry *generic_fh_to_dentry(struct super_block *sb, struct fid *fid,
845 int fh_len, int fh_type, struct inode *(*get_inode)
846 (struct super_block *sb, u64 ino, u32 gen))
847{
848 struct inode *inode = NULL;
849
850 if (fh_len < 2)
851 return NULL;
852
853 switch (fh_type) {
854 case FILEID_INO32_GEN:
855 case FILEID_INO32_GEN_PARENT:
856 inode = get_inode(sb, fid->i32.ino, fid->i32.gen);
857 break;
858 }
859
4ea3ada2 860 return d_obtain_alias(inode);
2596110a
CH
861}
862EXPORT_SYMBOL_GPL(generic_fh_to_dentry);
863
864/**
865 * generic_fh_to_dentry - generic helper for the fh_to_parent export operation
866 * @sb: filesystem to do the file handle conversion on
867 * @fid: file handle to convert
868 * @fh_len: length of the file handle in bytes
869 * @fh_type: type of file handle
870 * @get_inode: filesystem callback to retrieve inode
871 *
872 * This function decodes @fid as long as it has one of the well-known
873 * Linux filehandle types and calls @get_inode on it to retrieve the
874 * inode for the _parent_ object specified in the file handle if it
875 * is specified in the file handle, or NULL otherwise.
876 */
877struct dentry *generic_fh_to_parent(struct super_block *sb, struct fid *fid,
878 int fh_len, int fh_type, struct inode *(*get_inode)
879 (struct super_block *sb, u64 ino, u32 gen))
880{
881 struct inode *inode = NULL;
882
883 if (fh_len <= 2)
884 return NULL;
885
886 switch (fh_type) {
887 case FILEID_INO32_GEN_PARENT:
888 inode = get_inode(sb, fid->i32.parent_ino,
889 (fh_len > 3 ? fid->i32.parent_gen : 0));
890 break;
891 }
892
4ea3ada2 893 return d_obtain_alias(inode);
2596110a
CH
894}
895EXPORT_SYMBOL_GPL(generic_fh_to_parent);
896
1b061d92
CH
897/**
898 * generic_file_fsync - generic fsync implementation for simple filesystems
899 * @file: file to synchronize
900 * @datasync: only synchronize essential metadata if true
901 *
902 * This is a generic implementation of the fsync method for simple
903 * filesystems which track all non-inode metadata in the buffers list
904 * hanging off the address_space structure.
905 */
906int generic_file_fsync(struct file *file, int datasync)
d5aacad5 907{
7ea80859 908 struct inode *inode = file->f_mapping->host;
d5aacad5
AV
909 int err;
910 int ret;
911
912 ret = sync_mapping_buffers(inode->i_mapping);
913 if (!(inode->i_state & I_DIRTY))
914 return ret;
915 if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))
916 return ret;
917
c3765016 918 err = sync_inode_metadata(inode, 1);
d5aacad5
AV
919 if (ret == 0)
920 ret = err;
921 return ret;
922}
1b061d92
CH
923EXPORT_SYMBOL(generic_file_fsync);
924
30ca22c7
PL
925/**
926 * generic_check_addressable - Check addressability of file system
927 * @blocksize_bits: log of file system block size
928 * @num_blocks: number of blocks in file system
929 *
930 * Determine whether a file system with @num_blocks blocks (and a
931 * block size of 2**@blocksize_bits) is addressable by the sector_t
932 * and page cache of the system. Return 0 if so and -EFBIG otherwise.
933 */
934int generic_check_addressable(unsigned blocksize_bits, u64 num_blocks)
935{
936 u64 last_fs_block = num_blocks - 1;
a33f13ef
JB
937 u64 last_fs_page =
938 last_fs_block >> (PAGE_CACHE_SHIFT - blocksize_bits);
30ca22c7
PL
939
940 if (unlikely(num_blocks == 0))
941 return 0;
942
943 if ((blocksize_bits < 9) || (blocksize_bits > PAGE_CACHE_SHIFT))
944 return -EINVAL;
945
a33f13ef
JB
946 if ((last_fs_block > (sector_t)(~0ULL) >> (blocksize_bits - 9)) ||
947 (last_fs_page > (pgoff_t)(~0ULL))) {
30ca22c7
PL
948 return -EFBIG;
949 }
950 return 0;
951}
952EXPORT_SYMBOL(generic_check_addressable);
953
1b061d92
CH
954/*
955 * No-op implementation of ->fsync for in-memory filesystems.
956 */
957int noop_fsync(struct file *file, int datasync)
958{
959 return 0;
960}
d5aacad5 961
1da177e4
LT
962EXPORT_SYMBOL(dcache_dir_close);
963EXPORT_SYMBOL(dcache_dir_lseek);
964EXPORT_SYMBOL(dcache_dir_open);
965EXPORT_SYMBOL(dcache_readdir);
966EXPORT_SYMBOL(generic_read_dir);
51139ada 967EXPORT_SYMBOL(mount_pseudo);
afddba49
NP
968EXPORT_SYMBOL(simple_write_begin);
969EXPORT_SYMBOL(simple_write_end);
1da177e4
LT
970EXPORT_SYMBOL(simple_dir_inode_operations);
971EXPORT_SYMBOL(simple_dir_operations);
972EXPORT_SYMBOL(simple_empty);
1da177e4
LT
973EXPORT_SYMBOL(simple_fill_super);
974EXPORT_SYMBOL(simple_getattr);
975EXPORT_SYMBOL(simple_link);
976EXPORT_SYMBOL(simple_lookup);
977EXPORT_SYMBOL(simple_pin_fs);
1da177e4
LT
978EXPORT_SYMBOL(simple_readpage);
979EXPORT_SYMBOL(simple_release_fs);
980EXPORT_SYMBOL(simple_rename);
981EXPORT_SYMBOL(simple_rmdir);
982EXPORT_SYMBOL(simple_statfs);
1b061d92 983EXPORT_SYMBOL(noop_fsync);
1da177e4
LT
984EXPORT_SYMBOL(simple_unlink);
985EXPORT_SYMBOL(simple_read_from_buffer);
6a727b43 986EXPORT_SYMBOL(simple_write_to_buffer);
93b07113 987EXPORT_SYMBOL(memory_read_from_buffer);
76791ab2 988EXPORT_SYMBOL(simple_transaction_set);
1da177e4
LT
989EXPORT_SYMBOL(simple_transaction_get);
990EXPORT_SYMBOL(simple_transaction_read);
991EXPORT_SYMBOL(simple_transaction_release);
acaefc25 992EXPORT_SYMBOL_GPL(simple_attr_open);
74bedc4d 993EXPORT_SYMBOL_GPL(simple_attr_release);
acaefc25
AB
994EXPORT_SYMBOL_GPL(simple_attr_read);
995EXPORT_SYMBOL_GPL(simple_attr_write);