]>
Commit | Line | Data |
---|---|---|
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 |
21 | static inline int simple_positive(struct dentry *dentry) |
22 | { | |
23 | return dentry->d_inode && !d_unhashed(dentry); | |
24 | } | |
25 | ||
1da177e4 LT |
26 | int 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 | } | |
12f38872 | 34 | EXPORT_SYMBOL(simple_getattr); |
1da177e4 | 35 | |
726c3342 | 36 | int simple_statfs(struct dentry *dentry, struct kstatfs *buf) |
1da177e4 | 37 | { |
726c3342 | 38 | buf->f_type = dentry->d_sb->s_magic; |
1da177e4 LT |
39 | buf->f_bsize = PAGE_CACHE_SIZE; |
40 | buf->f_namelen = NAME_MAX; | |
41 | return 0; | |
42 | } | |
12f38872 | 43 | EXPORT_SYMBOL(simple_statfs); |
1da177e4 LT |
44 | |
45 | /* | |
46 | * Retaining negative dentries for an in-memory filesystem just wastes | |
47 | * memory and lookup time: arrange for them to be deleted immediately. | |
48 | */ | |
fe15ce44 | 49 | static int simple_delete_dentry(const struct dentry *dentry) |
1da177e4 LT |
50 | { |
51 | return 1; | |
52 | } | |
53 | ||
54 | /* | |
55 | * Lookup the data. This is trivial - if the dentry didn't already | |
56 | * exist, we know it is negative. Set d_op to delete negative dentries. | |
57 | */ | |
00cd8dd3 | 58 | struct dentry *simple_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags) |
1da177e4 | 59 | { |
3ba13d17 | 60 | static const struct dentry_operations simple_dentry_operations = { |
1da177e4 LT |
61 | .d_delete = simple_delete_dentry, |
62 | }; | |
63 | ||
64 | if (dentry->d_name.len > NAME_MAX) | |
65 | return ERR_PTR(-ENAMETOOLONG); | |
74931da7 AV |
66 | if (!dentry->d_sb->s_d_op) |
67 | d_set_d_op(dentry, &simple_dentry_operations); | |
1da177e4 LT |
68 | d_add(dentry, NULL); |
69 | return NULL; | |
70 | } | |
12f38872 | 71 | EXPORT_SYMBOL(simple_lookup); |
1da177e4 | 72 | |
1da177e4 LT |
73 | int dcache_dir_open(struct inode *inode, struct file *file) |
74 | { | |
26fe5750 | 75 | static struct qstr cursor_name = QSTR_INIT(".", 1); |
1da177e4 | 76 | |
0f7fc9e4 | 77 | file->private_data = d_alloc(file->f_path.dentry, &cursor_name); |
1da177e4 LT |
78 | |
79 | return file->private_data ? 0 : -ENOMEM; | |
80 | } | |
12f38872 | 81 | EXPORT_SYMBOL(dcache_dir_open); |
1da177e4 LT |
82 | |
83 | int dcache_dir_close(struct inode *inode, struct file *file) | |
84 | { | |
85 | dput(file->private_data); | |
86 | return 0; | |
87 | } | |
12f38872 | 88 | EXPORT_SYMBOL(dcache_dir_close); |
1da177e4 | 89 | |
965c8e59 | 90 | loff_t dcache_dir_lseek(struct file *file, loff_t offset, int whence) |
1da177e4 | 91 | { |
2fd6b7f5 NP |
92 | struct dentry *dentry = file->f_path.dentry; |
93 | mutex_lock(&dentry->d_inode->i_mutex); | |
965c8e59 | 94 | switch (whence) { |
1da177e4 LT |
95 | case 1: |
96 | offset += file->f_pos; | |
97 | case 0: | |
98 | if (offset >= 0) | |
99 | break; | |
100 | default: | |
2fd6b7f5 | 101 | mutex_unlock(&dentry->d_inode->i_mutex); |
1da177e4 LT |
102 | return -EINVAL; |
103 | } | |
104 | if (offset != file->f_pos) { | |
105 | file->f_pos = offset; | |
106 | if (file->f_pos >= 2) { | |
107 | struct list_head *p; | |
108 | struct dentry *cursor = file->private_data; | |
109 | loff_t n = file->f_pos - 2; | |
110 | ||
2fd6b7f5 NP |
111 | spin_lock(&dentry->d_lock); |
112 | /* d_lock not required for cursor */ | |
5160ee6f | 113 | list_del(&cursor->d_u.d_child); |
2fd6b7f5 NP |
114 | p = dentry->d_subdirs.next; |
115 | while (n && p != &dentry->d_subdirs) { | |
1da177e4 | 116 | struct dentry *next; |
5160ee6f | 117 | next = list_entry(p, struct dentry, d_u.d_child); |
2fd6b7f5 | 118 | spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED); |
da502956 | 119 | if (simple_positive(next)) |
1da177e4 | 120 | n--; |
da502956 | 121 | spin_unlock(&next->d_lock); |
1da177e4 LT |
122 | p = p->next; |
123 | } | |
5160ee6f | 124 | list_add_tail(&cursor->d_u.d_child, p); |
2fd6b7f5 | 125 | spin_unlock(&dentry->d_lock); |
1da177e4 LT |
126 | } |
127 | } | |
2fd6b7f5 | 128 | mutex_unlock(&dentry->d_inode->i_mutex); |
1da177e4 LT |
129 | return offset; |
130 | } | |
12f38872 | 131 | EXPORT_SYMBOL(dcache_dir_lseek); |
1da177e4 LT |
132 | |
133 | /* Relationship between i_mode and the DT_xxx types */ | |
134 | static inline unsigned char dt_type(struct inode *inode) | |
135 | { | |
136 | return (inode->i_mode >> 12) & 15; | |
137 | } | |
138 | ||
139 | /* | |
140 | * Directory is locked and all positive dentries in it are safe, since | |
141 | * for ramfs-type trees they can't go away without unlink() or rmdir(), | |
142 | * both impossible due to the lock on directory. | |
143 | */ | |
144 | ||
5f99f4e7 | 145 | int dcache_readdir(struct file *file, struct dir_context *ctx) |
1da177e4 | 146 | { |
5f99f4e7 AV |
147 | struct dentry *dentry = file->f_path.dentry; |
148 | struct dentry *cursor = file->private_data; | |
5160ee6f | 149 | struct list_head *p, *q = &cursor->d_u.d_child; |
1da177e4 | 150 | |
5f99f4e7 AV |
151 | if (!dir_emit_dots(file, ctx)) |
152 | return 0; | |
153 | spin_lock(&dentry->d_lock); | |
154 | if (ctx->pos == 2) | |
155 | list_move(q, &dentry->d_subdirs); | |
156 | ||
157 | for (p = q->next; p != &dentry->d_subdirs; p = p->next) { | |
158 | struct dentry *next = list_entry(p, struct dentry, d_u.d_child); | |
159 | spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED); | |
160 | if (!simple_positive(next)) { | |
161 | spin_unlock(&next->d_lock); | |
162 | continue; | |
163 | } | |
1da177e4 | 164 | |
5f99f4e7 AV |
165 | spin_unlock(&next->d_lock); |
166 | spin_unlock(&dentry->d_lock); | |
167 | if (!dir_emit(ctx, next->d_name.name, next->d_name.len, | |
168 | next->d_inode->i_ino, dt_type(next->d_inode))) | |
169 | return 0; | |
170 | spin_lock(&dentry->d_lock); | |
171 | spin_lock_nested(&next->d_lock, DENTRY_D_LOCK_NESTED); | |
172 | /* next is still alive */ | |
173 | list_move(q, p); | |
174 | spin_unlock(&next->d_lock); | |
175 | p = q; | |
176 | ctx->pos++; | |
1da177e4 | 177 | } |
5f99f4e7 | 178 | spin_unlock(&dentry->d_lock); |
1da177e4 LT |
179 | return 0; |
180 | } | |
12f38872 | 181 | EXPORT_SYMBOL(dcache_readdir); |
1da177e4 LT |
182 | |
183 | ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos) | |
184 | { | |
185 | return -EISDIR; | |
186 | } | |
12f38872 | 187 | EXPORT_SYMBOL(generic_read_dir); |
1da177e4 | 188 | |
4b6f5d20 | 189 | const struct file_operations simple_dir_operations = { |
1da177e4 LT |
190 | .open = dcache_dir_open, |
191 | .release = dcache_dir_close, | |
192 | .llseek = dcache_dir_lseek, | |
193 | .read = generic_read_dir, | |
5f99f4e7 | 194 | .iterate = dcache_readdir, |
1b061d92 | 195 | .fsync = noop_fsync, |
1da177e4 | 196 | }; |
12f38872 | 197 | EXPORT_SYMBOL(simple_dir_operations); |
1da177e4 | 198 | |
92e1d5be | 199 | const struct inode_operations simple_dir_inode_operations = { |
1da177e4 LT |
200 | .lookup = simple_lookup, |
201 | }; | |
12f38872 | 202 | EXPORT_SYMBOL(simple_dir_inode_operations); |
1da177e4 | 203 | |
759b9775 HD |
204 | static const struct super_operations simple_super_operations = { |
205 | .statfs = simple_statfs, | |
206 | }; | |
207 | ||
1da177e4 LT |
208 | /* |
209 | * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that | |
210 | * will never be mountable) | |
211 | */ | |
51139ada | 212 | struct dentry *mount_pseudo(struct file_system_type *fs_type, char *name, |
c74a1cbb AV |
213 | const struct super_operations *ops, |
214 | const struct dentry_operations *dops, unsigned long magic) | |
1da177e4 | 215 | { |
9249e17f | 216 | struct super_block *s; |
1da177e4 LT |
217 | struct dentry *dentry; |
218 | struct inode *root; | |
26fe5750 | 219 | struct qstr d_name = QSTR_INIT(name, strlen(name)); |
1da177e4 | 220 | |
9249e17f | 221 | s = sget(fs_type, NULL, set_anon_super, MS_NOUSER, NULL); |
1da177e4 | 222 | if (IS_ERR(s)) |
51139ada | 223 | return ERR_CAST(s); |
1da177e4 | 224 | |
89a4eb4b | 225 | s->s_maxbytes = MAX_LFS_FILESIZE; |
3971e1a9 AN |
226 | s->s_blocksize = PAGE_SIZE; |
227 | s->s_blocksize_bits = PAGE_SHIFT; | |
1da177e4 | 228 | s->s_magic = magic; |
759b9775 | 229 | s->s_op = ops ? ops : &simple_super_operations; |
1da177e4 LT |
230 | s->s_time_gran = 1; |
231 | root = new_inode(s); | |
232 | if (!root) | |
233 | goto Enomem; | |
1a1c9bb4 JL |
234 | /* |
235 | * since this is the first inode, make it number 1. New inodes created | |
236 | * after this must take care not to collide with it (by passing | |
237 | * max_reserved of 1 to iunique). | |
238 | */ | |
239 | root->i_ino = 1; | |
1da177e4 | 240 | root->i_mode = S_IFDIR | S_IRUSR | S_IWUSR; |
1da177e4 | 241 | root->i_atime = root->i_mtime = root->i_ctime = CURRENT_TIME; |
a4464dbc | 242 | dentry = __d_alloc(s, &d_name); |
1da177e4 LT |
243 | if (!dentry) { |
244 | iput(root); | |
245 | goto Enomem; | |
246 | } | |
1da177e4 LT |
247 | d_instantiate(dentry, root); |
248 | s->s_root = dentry; | |
c74a1cbb | 249 | s->s_d_op = dops; |
1da177e4 | 250 | s->s_flags |= MS_ACTIVE; |
51139ada | 251 | return dget(s->s_root); |
1da177e4 LT |
252 | |
253 | Enomem: | |
6f5bbff9 | 254 | deactivate_locked_super(s); |
51139ada | 255 | return ERR_PTR(-ENOMEM); |
1da177e4 | 256 | } |
12f38872 | 257 | EXPORT_SYMBOL(mount_pseudo); |
1da177e4 | 258 | |
20955e89 SB |
259 | int simple_open(struct inode *inode, struct file *file) |
260 | { | |
261 | if (inode->i_private) | |
262 | file->private_data = inode->i_private; | |
263 | return 0; | |
264 | } | |
12f38872 | 265 | EXPORT_SYMBOL(simple_open); |
20955e89 | 266 | |
1da177e4 LT |
267 | int simple_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry) |
268 | { | |
269 | struct inode *inode = old_dentry->d_inode; | |
270 | ||
271 | inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME; | |
d8c76e6f | 272 | inc_nlink(inode); |
7de9c6ee | 273 | ihold(inode); |
1da177e4 LT |
274 | dget(dentry); |
275 | d_instantiate(dentry, inode); | |
276 | return 0; | |
277 | } | |
12f38872 | 278 | EXPORT_SYMBOL(simple_link); |
1da177e4 | 279 | |
1da177e4 LT |
280 | int simple_empty(struct dentry *dentry) |
281 | { | |
282 | struct dentry *child; | |
283 | int ret = 0; | |
284 | ||
2fd6b7f5 | 285 | spin_lock(&dentry->d_lock); |
da502956 NP |
286 | list_for_each_entry(child, &dentry->d_subdirs, d_u.d_child) { |
287 | spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED); | |
288 | if (simple_positive(child)) { | |
289 | spin_unlock(&child->d_lock); | |
1da177e4 | 290 | goto out; |
da502956 NP |
291 | } |
292 | spin_unlock(&child->d_lock); | |
293 | } | |
1da177e4 LT |
294 | ret = 1; |
295 | out: | |
2fd6b7f5 | 296 | spin_unlock(&dentry->d_lock); |
1da177e4 LT |
297 | return ret; |
298 | } | |
12f38872 | 299 | EXPORT_SYMBOL(simple_empty); |
1da177e4 LT |
300 | |
301 | int simple_unlink(struct inode *dir, struct dentry *dentry) | |
302 | { | |
303 | struct inode *inode = dentry->d_inode; | |
304 | ||
305 | inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME; | |
9a53c3a7 | 306 | drop_nlink(inode); |
1da177e4 LT |
307 | dput(dentry); |
308 | return 0; | |
309 | } | |
12f38872 | 310 | EXPORT_SYMBOL(simple_unlink); |
1da177e4 LT |
311 | |
312 | int simple_rmdir(struct inode *dir, struct dentry *dentry) | |
313 | { | |
314 | if (!simple_empty(dentry)) | |
315 | return -ENOTEMPTY; | |
316 | ||
9a53c3a7 | 317 | drop_nlink(dentry->d_inode); |
1da177e4 | 318 | simple_unlink(dir, dentry); |
9a53c3a7 | 319 | drop_nlink(dir); |
1da177e4 LT |
320 | return 0; |
321 | } | |
12f38872 | 322 | EXPORT_SYMBOL(simple_rmdir); |
1da177e4 LT |
323 | |
324 | int simple_rename(struct inode *old_dir, struct dentry *old_dentry, | |
325 | struct inode *new_dir, struct dentry *new_dentry) | |
326 | { | |
327 | struct inode *inode = old_dentry->d_inode; | |
328 | int they_are_dirs = S_ISDIR(old_dentry->d_inode->i_mode); | |
329 | ||
330 | if (!simple_empty(new_dentry)) | |
331 | return -ENOTEMPTY; | |
332 | ||
333 | if (new_dentry->d_inode) { | |
334 | simple_unlink(new_dir, new_dentry); | |
841590ce AV |
335 | if (they_are_dirs) { |
336 | drop_nlink(new_dentry->d_inode); | |
9a53c3a7 | 337 | drop_nlink(old_dir); |
841590ce | 338 | } |
1da177e4 | 339 | } else if (they_are_dirs) { |
9a53c3a7 | 340 | drop_nlink(old_dir); |
d8c76e6f | 341 | inc_nlink(new_dir); |
1da177e4 LT |
342 | } |
343 | ||
344 | old_dir->i_ctime = old_dir->i_mtime = new_dir->i_ctime = | |
345 | new_dir->i_mtime = inode->i_ctime = CURRENT_TIME; | |
346 | ||
347 | return 0; | |
348 | } | |
12f38872 | 349 | EXPORT_SYMBOL(simple_rename); |
1da177e4 | 350 | |
7bb46a67 | 351 | /** |
eef2380c | 352 | * simple_setattr - setattr for simple filesystem |
7bb46a67 | 353 | * @dentry: dentry |
354 | * @iattr: iattr structure | |
355 | * | |
356 | * Returns 0 on success, -error on failure. | |
357 | * | |
eef2380c CH |
358 | * simple_setattr is a simple ->setattr implementation without a proper |
359 | * implementation of size changes. | |
360 | * | |
361 | * It can either be used for in-memory filesystems or special files | |
362 | * on simple regular filesystems. Anything that needs to change on-disk | |
363 | * or wire state on size changes needs its own setattr method. | |
7bb46a67 | 364 | */ |
365 | int simple_setattr(struct dentry *dentry, struct iattr *iattr) | |
366 | { | |
367 | struct inode *inode = dentry->d_inode; | |
368 | int error; | |
369 | ||
370 | error = inode_change_ok(inode, iattr); | |
371 | if (error) | |
372 | return error; | |
373 | ||
2c27c65e CH |
374 | if (iattr->ia_valid & ATTR_SIZE) |
375 | truncate_setsize(inode, iattr->ia_size); | |
6a1a90ad | 376 | setattr_copy(inode, iattr); |
eef2380c CH |
377 | mark_inode_dirty(inode); |
378 | return 0; | |
7bb46a67 | 379 | } |
380 | EXPORT_SYMBOL(simple_setattr); | |
381 | ||
1da177e4 LT |
382 | int simple_readpage(struct file *file, struct page *page) |
383 | { | |
c0d92cbc | 384 | clear_highpage(page); |
1da177e4 LT |
385 | flush_dcache_page(page); |
386 | SetPageUptodate(page); | |
1da177e4 LT |
387 | unlock_page(page); |
388 | return 0; | |
389 | } | |
12f38872 | 390 | EXPORT_SYMBOL(simple_readpage); |
1da177e4 | 391 | |
afddba49 NP |
392 | int simple_write_begin(struct file *file, struct address_space *mapping, |
393 | loff_t pos, unsigned len, unsigned flags, | |
394 | struct page **pagep, void **fsdata) | |
395 | { | |
396 | struct page *page; | |
397 | pgoff_t index; | |
afddba49 NP |
398 | |
399 | index = pos >> PAGE_CACHE_SHIFT; | |
afddba49 | 400 | |
54566b2c | 401 | page = grab_cache_page_write_begin(mapping, index, flags); |
afddba49 NP |
402 | if (!page) |
403 | return -ENOMEM; | |
404 | ||
405 | *pagep = page; | |
406 | ||
193cf4b9 BH |
407 | if (!PageUptodate(page) && (len != PAGE_CACHE_SIZE)) { |
408 | unsigned from = pos & (PAGE_CACHE_SIZE - 1); | |
409 | ||
410 | zero_user_segments(page, 0, from, from + len, PAGE_CACHE_SIZE); | |
411 | } | |
412 | return 0; | |
afddba49 | 413 | } |
12f38872 | 414 | EXPORT_SYMBOL(simple_write_begin); |
afddba49 | 415 | |
ad2a722f BH |
416 | /** |
417 | * simple_write_end - .write_end helper for non-block-device FSes | |
418 | * @available: See .write_end of address_space_operations | |
419 | * @file: " | |
420 | * @mapping: " | |
421 | * @pos: " | |
422 | * @len: " | |
423 | * @copied: " | |
424 | * @page: " | |
425 | * @fsdata: " | |
426 | * | |
427 | * simple_write_end does the minimum needed for updating a page after writing is | |
428 | * done. It has the same API signature as the .write_end of | |
429 | * address_space_operations vector. So it can just be set onto .write_end for | |
430 | * FSes that don't need any other processing. i_mutex is assumed to be held. | |
431 | * Block based filesystems should use generic_write_end(). | |
432 | * NOTE: Even though i_size might get updated by this function, mark_inode_dirty | |
433 | * is not called, so a filesystem that actually does store data in .write_inode | |
434 | * should extend on what's done here with a call to mark_inode_dirty() in the | |
435 | * case that i_size has changed. | |
436 | */ | |
afddba49 NP |
437 | int simple_write_end(struct file *file, struct address_space *mapping, |
438 | loff_t pos, unsigned len, unsigned copied, | |
439 | struct page *page, void *fsdata) | |
440 | { | |
ad2a722f BH |
441 | struct inode *inode = page->mapping->host; |
442 | loff_t last_pos = pos + copied; | |
afddba49 NP |
443 | |
444 | /* zero the stale part of the page if we did a short copy */ | |
445 | if (copied < len) { | |
ad2a722f BH |
446 | unsigned from = pos & (PAGE_CACHE_SIZE - 1); |
447 | ||
448 | zero_user(page, from + copied, len - copied); | |
afddba49 NP |
449 | } |
450 | ||
ad2a722f BH |
451 | if (!PageUptodate(page)) |
452 | SetPageUptodate(page); | |
453 | /* | |
454 | * No need to use i_size_read() here, the i_size | |
455 | * cannot change under us because we hold the i_mutex. | |
456 | */ | |
457 | if (last_pos > inode->i_size) | |
458 | i_size_write(inode, last_pos); | |
afddba49 | 459 | |
ad2a722f | 460 | set_page_dirty(page); |
afddba49 NP |
461 | unlock_page(page); |
462 | page_cache_release(page); | |
463 | ||
464 | return copied; | |
465 | } | |
12f38872 | 466 | EXPORT_SYMBOL(simple_write_end); |
afddba49 | 467 | |
1a1c9bb4 JL |
468 | /* |
469 | * the inodes created here are not hashed. If you use iunique to generate | |
470 | * unique inode values later for this filesystem, then you must take care | |
471 | * to pass it an appropriate max_reserved value to avoid collisions. | |
472 | */ | |
7d683a09 RS |
473 | int simple_fill_super(struct super_block *s, unsigned long magic, |
474 | struct tree_descr *files) | |
1da177e4 | 475 | { |
1da177e4 LT |
476 | struct inode *inode; |
477 | struct dentry *root; | |
478 | struct dentry *dentry; | |
479 | int i; | |
480 | ||
481 | s->s_blocksize = PAGE_CACHE_SIZE; | |
482 | s->s_blocksize_bits = PAGE_CACHE_SHIFT; | |
483 | s->s_magic = magic; | |
759b9775 | 484 | s->s_op = &simple_super_operations; |
1da177e4 LT |
485 | s->s_time_gran = 1; |
486 | ||
487 | inode = new_inode(s); | |
488 | if (!inode) | |
489 | return -ENOMEM; | |
1a1c9bb4 JL |
490 | /* |
491 | * because the root inode is 1, the files array must not contain an | |
492 | * entry at index 1 | |
493 | */ | |
494 | inode->i_ino = 1; | |
1da177e4 | 495 | inode->i_mode = S_IFDIR | 0755; |
1da177e4 LT |
496 | inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME; |
497 | inode->i_op = &simple_dir_inode_operations; | |
498 | inode->i_fop = &simple_dir_operations; | |
bfe86848 | 499 | set_nlink(inode, 2); |
48fde701 AV |
500 | root = d_make_root(inode); |
501 | if (!root) | |
1da177e4 | 502 | return -ENOMEM; |
1da177e4 LT |
503 | for (i = 0; !files->name || files->name[0]; i++, files++) { |
504 | if (!files->name) | |
505 | continue; | |
1a1c9bb4 JL |
506 | |
507 | /* warn if it tries to conflict with the root inode */ | |
508 | if (unlikely(i == 1)) | |
509 | printk(KERN_WARNING "%s: %s passed in a files array" | |
510 | "with an index of 1!\n", __func__, | |
511 | s->s_type->name); | |
512 | ||
1da177e4 LT |
513 | dentry = d_alloc_name(root, files->name); |
514 | if (!dentry) | |
515 | goto out; | |
516 | inode = new_inode(s); | |
32096ea1 KK |
517 | if (!inode) { |
518 | dput(dentry); | |
1da177e4 | 519 | goto out; |
32096ea1 | 520 | } |
1da177e4 | 521 | inode->i_mode = S_IFREG | files->mode; |
1da177e4 LT |
522 | inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME; |
523 | inode->i_fop = files->ops; | |
524 | inode->i_ino = i; | |
525 | d_add(dentry, inode); | |
526 | } | |
527 | s->s_root = root; | |
528 | return 0; | |
529 | out: | |
530 | d_genocide(root); | |
640946f2 | 531 | shrink_dcache_parent(root); |
1da177e4 LT |
532 | dput(root); |
533 | return -ENOMEM; | |
534 | } | |
12f38872 | 535 | EXPORT_SYMBOL(simple_fill_super); |
1da177e4 LT |
536 | |
537 | static DEFINE_SPINLOCK(pin_fs_lock); | |
538 | ||
1f5ce9e9 | 539 | int 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 | } | |
12f38872 | 558 | EXPORT_SYMBOL(simple_pin_fs); |
1da177e4 LT |
559 | |
560 | void simple_release_fs(struct vfsmount **mount, int *count) | |
561 | { | |
562 | struct vfsmount *mnt; | |
563 | spin_lock(&pin_fs_lock); | |
564 | mnt = *mount; | |
565 | if (!--*count) | |
566 | *mount = NULL; | |
567 | spin_unlock(&pin_fs_lock); | |
568 | mntput(mnt); | |
569 | } | |
12f38872 | 570 | EXPORT_SYMBOL(simple_release_fs); |
1da177e4 | 571 | |
6d1029b5 AM |
572 | /** |
573 | * simple_read_from_buffer - copy data from the buffer to user space | |
574 | * @to: the user space buffer to read to | |
575 | * @count: the maximum number of bytes to read | |
576 | * @ppos: the current position in the buffer | |
577 | * @from: the buffer to read from | |
578 | * @available: the size of the buffer | |
579 | * | |
580 | * The simple_read_from_buffer() function reads up to @count bytes from the | |
581 | * buffer @from at offset @ppos into the user space address starting at @to. | |
582 | * | |
583 | * On success, the number of bytes read is returned and the offset @ppos is | |
584 | * advanced by this number, or negative value is returned on error. | |
585 | **/ | |
1da177e4 LT |
586 | ssize_t simple_read_from_buffer(void __user *to, size_t count, loff_t *ppos, |
587 | const void *from, size_t available) | |
588 | { | |
589 | loff_t pos = *ppos; | |
14be2746 SR |
590 | size_t ret; |
591 | ||
1da177e4 LT |
592 | if (pos < 0) |
593 | return -EINVAL; | |
14be2746 | 594 | if (pos >= available || !count) |
1da177e4 LT |
595 | return 0; |
596 | if (count > available - pos) | |
597 | count = available - pos; | |
14be2746 SR |
598 | ret = copy_to_user(to, from + pos, count); |
599 | if (ret == count) | |
1da177e4 | 600 | return -EFAULT; |
14be2746 | 601 | count -= ret; |
1da177e4 LT |
602 | *ppos = pos + count; |
603 | return count; | |
604 | } | |
12f38872 | 605 | EXPORT_SYMBOL(simple_read_from_buffer); |
1da177e4 | 606 | |
6a727b43 JS |
607 | /** |
608 | * simple_write_to_buffer - copy data from user space to the buffer | |
609 | * @to: the buffer to write to | |
610 | * @available: the size of the buffer | |
611 | * @ppos: the current position in the buffer | |
612 | * @from: the user space buffer to read from | |
613 | * @count: the maximum number of bytes to read | |
614 | * | |
615 | * The simple_write_to_buffer() function reads up to @count bytes from the user | |
616 | * space address starting at @from into the buffer @to at offset @ppos. | |
617 | * | |
618 | * On success, the number of bytes written is returned and the offset @ppos is | |
619 | * advanced by this number, or negative value is returned on error. | |
620 | **/ | |
621 | ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, | |
622 | const void __user *from, size_t count) | |
623 | { | |
624 | loff_t pos = *ppos; | |
625 | size_t res; | |
626 | ||
627 | if (pos < 0) | |
628 | return -EINVAL; | |
629 | if (pos >= available || !count) | |
630 | return 0; | |
631 | if (count > available - pos) | |
632 | count = available - pos; | |
633 | res = copy_from_user(to + pos, from, count); | |
634 | if (res == count) | |
635 | return -EFAULT; | |
636 | count -= res; | |
637 | *ppos = pos + count; | |
638 | return count; | |
639 | } | |
12f38872 | 640 | EXPORT_SYMBOL(simple_write_to_buffer); |
6a727b43 | 641 | |
6d1029b5 AM |
642 | /** |
643 | * memory_read_from_buffer - copy data from the buffer | |
644 | * @to: the kernel space buffer to read to | |
645 | * @count: the maximum number of bytes to read | |
646 | * @ppos: the current position in the buffer | |
647 | * @from: the buffer to read from | |
648 | * @available: the size of the buffer | |
649 | * | |
650 | * The memory_read_from_buffer() function reads up to @count bytes from the | |
651 | * buffer @from at offset @ppos into the kernel space address starting at @to. | |
652 | * | |
653 | * On success, the number of bytes read is returned and the offset @ppos is | |
654 | * advanced by this number, or negative value is returned on error. | |
655 | **/ | |
93b07113 AM |
656 | ssize_t memory_read_from_buffer(void *to, size_t count, loff_t *ppos, |
657 | const void *from, size_t available) | |
658 | { | |
659 | loff_t pos = *ppos; | |
660 | ||
661 | if (pos < 0) | |
662 | return -EINVAL; | |
663 | if (pos >= available) | |
664 | return 0; | |
665 | if (count > available - pos) | |
666 | count = available - pos; | |
667 | memcpy(to, from + pos, count); | |
668 | *ppos = pos + count; | |
669 | ||
670 | return count; | |
671 | } | |
12f38872 | 672 | EXPORT_SYMBOL(memory_read_from_buffer); |
93b07113 | 673 | |
1da177e4 LT |
674 | /* |
675 | * Transaction based IO. | |
676 | * The file expects a single write which triggers the transaction, and then | |
677 | * possibly a read which collects the result - which is stored in a | |
678 | * file-local buffer. | |
679 | */ | |
76791ab2 IM |
680 | |
681 | void simple_transaction_set(struct file *file, size_t n) | |
682 | { | |
683 | struct simple_transaction_argresp *ar = file->private_data; | |
684 | ||
685 | BUG_ON(n > SIMPLE_TRANSACTION_LIMIT); | |
686 | ||
687 | /* | |
688 | * The barrier ensures that ar->size will really remain zero until | |
689 | * ar->data is ready for reading. | |
690 | */ | |
691 | smp_mb(); | |
692 | ar->size = n; | |
693 | } | |
12f38872 | 694 | EXPORT_SYMBOL(simple_transaction_set); |
76791ab2 | 695 | |
1da177e4 LT |
696 | char *simple_transaction_get(struct file *file, const char __user *buf, size_t size) |
697 | { | |
698 | struct simple_transaction_argresp *ar; | |
699 | static DEFINE_SPINLOCK(simple_transaction_lock); | |
700 | ||
701 | if (size > SIMPLE_TRANSACTION_LIMIT - 1) | |
702 | return ERR_PTR(-EFBIG); | |
703 | ||
704 | ar = (struct simple_transaction_argresp *)get_zeroed_page(GFP_KERNEL); | |
705 | if (!ar) | |
706 | return ERR_PTR(-ENOMEM); | |
707 | ||
708 | spin_lock(&simple_transaction_lock); | |
709 | ||
710 | /* only one write allowed per open */ | |
711 | if (file->private_data) { | |
712 | spin_unlock(&simple_transaction_lock); | |
713 | free_page((unsigned long)ar); | |
714 | return ERR_PTR(-EBUSY); | |
715 | } | |
716 | ||
717 | file->private_data = ar; | |
718 | ||
719 | spin_unlock(&simple_transaction_lock); | |
720 | ||
721 | if (copy_from_user(ar->data, buf, size)) | |
722 | return ERR_PTR(-EFAULT); | |
723 | ||
724 | return ar->data; | |
725 | } | |
12f38872 | 726 | EXPORT_SYMBOL(simple_transaction_get); |
1da177e4 LT |
727 | |
728 | ssize_t simple_transaction_read(struct file *file, char __user *buf, size_t size, loff_t *pos) | |
729 | { | |
730 | struct simple_transaction_argresp *ar = file->private_data; | |
731 | ||
732 | if (!ar) | |
733 | return 0; | |
734 | return simple_read_from_buffer(buf, size, pos, ar->data, ar->size); | |
735 | } | |
12f38872 | 736 | EXPORT_SYMBOL(simple_transaction_read); |
1da177e4 LT |
737 | |
738 | int simple_transaction_release(struct inode *inode, struct file *file) | |
739 | { | |
740 | free_page((unsigned long)file->private_data); | |
741 | return 0; | |
742 | } | |
12f38872 | 743 | EXPORT_SYMBOL(simple_transaction_release); |
1da177e4 | 744 | |
acaefc25 AB |
745 | /* Simple attribute files */ |
746 | ||
747 | struct simple_attr { | |
8b88b099 CH |
748 | int (*get)(void *, u64 *); |
749 | int (*set)(void *, u64); | |
acaefc25 AB |
750 | char get_buf[24]; /* enough to store a u64 and "\n\0" */ |
751 | char set_buf[24]; | |
752 | void *data; | |
753 | const char *fmt; /* format for read operation */ | |
7cf34c76 | 754 | struct mutex mutex; /* protects access to these buffers */ |
acaefc25 AB |
755 | }; |
756 | ||
757 | /* simple_attr_open is called by an actual attribute open file operation | |
758 | * to set the attribute specific access operations. */ | |
759 | int simple_attr_open(struct inode *inode, struct file *file, | |
8b88b099 | 760 | int (*get)(void *, u64 *), int (*set)(void *, u64), |
acaefc25 AB |
761 | const char *fmt) |
762 | { | |
763 | struct simple_attr *attr; | |
764 | ||
765 | attr = kmalloc(sizeof(*attr), GFP_KERNEL); | |
766 | if (!attr) | |
767 | return -ENOMEM; | |
768 | ||
769 | attr->get = get; | |
770 | attr->set = set; | |
8e18e294 | 771 | attr->data = inode->i_private; |
acaefc25 | 772 | attr->fmt = fmt; |
7cf34c76 | 773 | mutex_init(&attr->mutex); |
acaefc25 AB |
774 | |
775 | file->private_data = attr; | |
776 | ||
777 | return nonseekable_open(inode, file); | |
778 | } | |
12f38872 | 779 | EXPORT_SYMBOL_GPL(simple_attr_open); |
acaefc25 | 780 | |
74bedc4d | 781 | int simple_attr_release(struct inode *inode, struct file *file) |
acaefc25 AB |
782 | { |
783 | kfree(file->private_data); | |
784 | return 0; | |
785 | } | |
12f38872 | 786 | EXPORT_SYMBOL_GPL(simple_attr_release); /* GPL-only? This? Really? */ |
acaefc25 AB |
787 | |
788 | /* read from the buffer that is filled with the get function */ | |
789 | ssize_t simple_attr_read(struct file *file, char __user *buf, | |
790 | size_t len, loff_t *ppos) | |
791 | { | |
792 | struct simple_attr *attr; | |
793 | size_t size; | |
794 | ssize_t ret; | |
795 | ||
796 | attr = file->private_data; | |
797 | ||
798 | if (!attr->get) | |
799 | return -EACCES; | |
800 | ||
9261303a CH |
801 | ret = mutex_lock_interruptible(&attr->mutex); |
802 | if (ret) | |
803 | return ret; | |
804 | ||
8b88b099 | 805 | if (*ppos) { /* continued read */ |
acaefc25 | 806 | size = strlen(attr->get_buf); |
8b88b099 CH |
807 | } else { /* first read */ |
808 | u64 val; | |
809 | ret = attr->get(attr->data, &val); | |
810 | if (ret) | |
811 | goto out; | |
812 | ||
acaefc25 | 813 | size = scnprintf(attr->get_buf, sizeof(attr->get_buf), |
8b88b099 CH |
814 | attr->fmt, (unsigned long long)val); |
815 | } | |
acaefc25 AB |
816 | |
817 | ret = simple_read_from_buffer(buf, len, ppos, attr->get_buf, size); | |
8b88b099 | 818 | out: |
7cf34c76 | 819 | mutex_unlock(&attr->mutex); |
acaefc25 AB |
820 | return ret; |
821 | } | |
12f38872 | 822 | EXPORT_SYMBOL_GPL(simple_attr_read); |
acaefc25 AB |
823 | |
824 | /* interpret the buffer as a number to call the set function with */ | |
825 | ssize_t simple_attr_write(struct file *file, const char __user *buf, | |
826 | size_t len, loff_t *ppos) | |
827 | { | |
828 | struct simple_attr *attr; | |
829 | u64 val; | |
830 | size_t size; | |
831 | ssize_t ret; | |
832 | ||
833 | attr = file->private_data; | |
acaefc25 AB |
834 | if (!attr->set) |
835 | return -EACCES; | |
836 | ||
9261303a CH |
837 | ret = mutex_lock_interruptible(&attr->mutex); |
838 | if (ret) | |
839 | return ret; | |
840 | ||
acaefc25 AB |
841 | ret = -EFAULT; |
842 | size = min(sizeof(attr->set_buf) - 1, len); | |
843 | if (copy_from_user(attr->set_buf, buf, size)) | |
844 | goto out; | |
845 | ||
acaefc25 | 846 | attr->set_buf[size] = '\0'; |
f7b88631 | 847 | val = simple_strtoll(attr->set_buf, NULL, 0); |
05cc0cee WF |
848 | ret = attr->set(attr->data, val); |
849 | if (ret == 0) | |
850 | ret = len; /* on success, claim we got the whole input */ | |
acaefc25 | 851 | out: |
7cf34c76 | 852 | mutex_unlock(&attr->mutex); |
acaefc25 AB |
853 | return ret; |
854 | } | |
12f38872 | 855 | EXPORT_SYMBOL_GPL(simple_attr_write); |
acaefc25 | 856 | |
2596110a CH |
857 | /** |
858 | * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation | |
859 | * @sb: filesystem to do the file handle conversion on | |
860 | * @fid: file handle to convert | |
861 | * @fh_len: length of the file handle in bytes | |
862 | * @fh_type: type of file handle | |
863 | * @get_inode: filesystem callback to retrieve inode | |
864 | * | |
865 | * This function decodes @fid as long as it has one of the well-known | |
866 | * Linux filehandle types and calls @get_inode on it to retrieve the | |
867 | * inode for the object specified in the file handle. | |
868 | */ | |
869 | struct dentry *generic_fh_to_dentry(struct super_block *sb, struct fid *fid, | |
870 | int fh_len, int fh_type, struct inode *(*get_inode) | |
871 | (struct super_block *sb, u64 ino, u32 gen)) | |
872 | { | |
873 | struct inode *inode = NULL; | |
874 | ||
875 | if (fh_len < 2) | |
876 | return NULL; | |
877 | ||
878 | switch (fh_type) { | |
879 | case FILEID_INO32_GEN: | |
880 | case FILEID_INO32_GEN_PARENT: | |
881 | inode = get_inode(sb, fid->i32.ino, fid->i32.gen); | |
882 | break; | |
883 | } | |
884 | ||
4ea3ada2 | 885 | return d_obtain_alias(inode); |
2596110a CH |
886 | } |
887 | EXPORT_SYMBOL_GPL(generic_fh_to_dentry); | |
888 | ||
889 | /** | |
ca186830 | 890 | * generic_fh_to_parent - generic helper for the fh_to_parent export operation |
2596110a CH |
891 | * @sb: filesystem to do the file handle conversion on |
892 | * @fid: file handle to convert | |
893 | * @fh_len: length of the file handle in bytes | |
894 | * @fh_type: type of file handle | |
895 | * @get_inode: filesystem callback to retrieve inode | |
896 | * | |
897 | * This function decodes @fid as long as it has one of the well-known | |
898 | * Linux filehandle types and calls @get_inode on it to retrieve the | |
899 | * inode for the _parent_ object specified in the file handle if it | |
900 | * is specified in the file handle, or NULL otherwise. | |
901 | */ | |
902 | struct dentry *generic_fh_to_parent(struct super_block *sb, struct fid *fid, | |
903 | int fh_len, int fh_type, struct inode *(*get_inode) | |
904 | (struct super_block *sb, u64 ino, u32 gen)) | |
905 | { | |
906 | struct inode *inode = NULL; | |
907 | ||
908 | if (fh_len <= 2) | |
909 | return NULL; | |
910 | ||
911 | switch (fh_type) { | |
912 | case FILEID_INO32_GEN_PARENT: | |
913 | inode = get_inode(sb, fid->i32.parent_ino, | |
914 | (fh_len > 3 ? fid->i32.parent_gen : 0)); | |
915 | break; | |
916 | } | |
917 | ||
4ea3ada2 | 918 | return d_obtain_alias(inode); |
2596110a CH |
919 | } |
920 | EXPORT_SYMBOL_GPL(generic_fh_to_parent); | |
921 | ||
1b061d92 CH |
922 | /** |
923 | * generic_file_fsync - generic fsync implementation for simple filesystems | |
924 | * @file: file to synchronize | |
925 | * @datasync: only synchronize essential metadata if true | |
926 | * | |
927 | * This is a generic implementation of the fsync method for simple | |
928 | * filesystems which track all non-inode metadata in the buffers list | |
929 | * hanging off the address_space structure. | |
930 | */ | |
02c24a82 JB |
931 | int generic_file_fsync(struct file *file, loff_t start, loff_t end, |
932 | int datasync) | |
d5aacad5 | 933 | { |
7ea80859 | 934 | struct inode *inode = file->f_mapping->host; |
d5aacad5 AV |
935 | int err; |
936 | int ret; | |
937 | ||
02c24a82 JB |
938 | err = filemap_write_and_wait_range(inode->i_mapping, start, end); |
939 | if (err) | |
940 | return err; | |
941 | ||
942 | mutex_lock(&inode->i_mutex); | |
d5aacad5 AV |
943 | ret = sync_mapping_buffers(inode->i_mapping); |
944 | if (!(inode->i_state & I_DIRTY)) | |
02c24a82 | 945 | goto out; |
d5aacad5 | 946 | if (datasync && !(inode->i_state & I_DIRTY_DATASYNC)) |
02c24a82 | 947 | goto out; |
d5aacad5 | 948 | |
c3765016 | 949 | err = sync_inode_metadata(inode, 1); |
d5aacad5 AV |
950 | if (ret == 0) |
951 | ret = err; | |
02c24a82 JB |
952 | out: |
953 | mutex_unlock(&inode->i_mutex); | |
d5aacad5 AV |
954 | return ret; |
955 | } | |
1b061d92 CH |
956 | EXPORT_SYMBOL(generic_file_fsync); |
957 | ||
30ca22c7 PL |
958 | /** |
959 | * generic_check_addressable - Check addressability of file system | |
960 | * @blocksize_bits: log of file system block size | |
961 | * @num_blocks: number of blocks in file system | |
962 | * | |
963 | * Determine whether a file system with @num_blocks blocks (and a | |
964 | * block size of 2**@blocksize_bits) is addressable by the sector_t | |
965 | * and page cache of the system. Return 0 if so and -EFBIG otherwise. | |
966 | */ | |
967 | int generic_check_addressable(unsigned blocksize_bits, u64 num_blocks) | |
968 | { | |
969 | u64 last_fs_block = num_blocks - 1; | |
a33f13ef JB |
970 | u64 last_fs_page = |
971 | last_fs_block >> (PAGE_CACHE_SHIFT - blocksize_bits); | |
30ca22c7 PL |
972 | |
973 | if (unlikely(num_blocks == 0)) | |
974 | return 0; | |
975 | ||
976 | if ((blocksize_bits < 9) || (blocksize_bits > PAGE_CACHE_SHIFT)) | |
977 | return -EINVAL; | |
978 | ||
a33f13ef JB |
979 | if ((last_fs_block > (sector_t)(~0ULL) >> (blocksize_bits - 9)) || |
980 | (last_fs_page > (pgoff_t)(~0ULL))) { | |
30ca22c7 PL |
981 | return -EFBIG; |
982 | } | |
983 | return 0; | |
984 | } | |
985 | EXPORT_SYMBOL(generic_check_addressable); | |
986 | ||
1b061d92 CH |
987 | /* |
988 | * No-op implementation of ->fsync for in-memory filesystems. | |
989 | */ | |
02c24a82 | 990 | int noop_fsync(struct file *file, loff_t start, loff_t end, int datasync) |
1b061d92 CH |
991 | { |
992 | return 0; | |
993 | } | |
1b061d92 | 994 | EXPORT_SYMBOL(noop_fsync); |