blob: dcaf972cbf1b6d81f29e17d2a1fccd259a794777 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * fs/libfs.c
3 * Library for filesystems writers.
4 */
5
6#include <linux/module.h>
7#include <linux/pagemap.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +09008#include <linux/slab.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07009#include <linux/mount.h>
10#include <linux/vfs.h>
npiggin@suse.de7bb46a62010-05-27 01:05:33 +100011#include <linux/quotaops.h>
Ingo Molnar7cf34c72006-03-23 03:00:36 -080012#include <linux/mutex.h>
Christoph Hellwig25961102007-10-21 16:42:05 -070013#include <linux/exportfs.h>
Al Virod5aacad2009-06-07 14:56:44 -040014#include <linux/writeback.h>
15#include <linux/buffer_head.h>
Ingo Molnar7cf34c72006-03-23 03:00:36 -080016
Linus Torvalds1da177e2005-04-16 15:20:36 -070017#include <asm/uaccess.h>
18
19int simple_getattr(struct vfsmount *mnt, struct dentry *dentry,
20 struct kstat *stat)
21{
22 struct inode *inode = dentry->d_inode;
23 generic_fillattr(inode, stat);
24 stat->blocks = inode->i_mapping->nrpages << (PAGE_CACHE_SHIFT - 9);
25 return 0;
26}
27
David Howells726c3342006-06-23 02:02:58 -070028int simple_statfs(struct dentry *dentry, struct kstatfs *buf)
Linus Torvalds1da177e2005-04-16 15:20:36 -070029{
David Howells726c3342006-06-23 02:02:58 -070030 buf->f_type = dentry->d_sb->s_magic;
Linus Torvalds1da177e2005-04-16 15:20:36 -070031 buf->f_bsize = PAGE_CACHE_SIZE;
32 buf->f_namelen = NAME_MAX;
33 return 0;
34}
35
36/*
37 * Retaining negative dentries for an in-memory filesystem just wastes
38 * memory and lookup time: arrange for them to be deleted immediately.
39 */
40static int simple_delete_dentry(struct dentry *dentry)
41{
42 return 1;
43}
44
45/*
46 * Lookup the data. This is trivial - if the dentry didn't already
47 * exist, we know it is negative. Set d_op to delete negative dentries.
48 */
49struct dentry *simple_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
50{
Al Viro3ba13d12009-02-20 06:02:22 +000051 static const struct dentry_operations simple_dentry_operations = {
Linus Torvalds1da177e2005-04-16 15:20:36 -070052 .d_delete = simple_delete_dentry,
53 };
54
55 if (dentry->d_name.len > NAME_MAX)
56 return ERR_PTR(-ENAMETOOLONG);
57 dentry->d_op = &simple_dentry_operations;
58 d_add(dentry, NULL);
59 return NULL;
60}
61
Linus Torvalds1da177e2005-04-16 15:20:36 -070062int dcache_dir_open(struct inode *inode, struct file *file)
63{
64 static struct qstr cursor_name = {.len = 1, .name = "."};
65
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -080066 file->private_data = d_alloc(file->f_path.dentry, &cursor_name);
Linus Torvalds1da177e2005-04-16 15:20:36 -070067
68 return file->private_data ? 0 : -ENOMEM;
69}
70
71int dcache_dir_close(struct inode *inode, struct file *file)
72{
73 dput(file->private_data);
74 return 0;
75}
76
77loff_t dcache_dir_lseek(struct file *file, loff_t offset, int origin)
78{
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -080079 mutex_lock(&file->f_path.dentry->d_inode->i_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -070080 switch (origin) {
81 case 1:
82 offset += file->f_pos;
83 case 0:
84 if (offset >= 0)
85 break;
86 default:
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -080087 mutex_unlock(&file->f_path.dentry->d_inode->i_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -070088 return -EINVAL;
89 }
90 if (offset != file->f_pos) {
91 file->f_pos = offset;
92 if (file->f_pos >= 2) {
93 struct list_head *p;
94 struct dentry *cursor = file->private_data;
95 loff_t n = file->f_pos - 2;
96
97 spin_lock(&dcache_lock);
Eric Dumazet5160ee62006-01-08 01:03:32 -080098 list_del(&cursor->d_u.d_child);
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -080099 p = file->f_path.dentry->d_subdirs.next;
100 while (n && p != &file->f_path.dentry->d_subdirs) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101 struct dentry *next;
Eric Dumazet5160ee62006-01-08 01:03:32 -0800102 next = list_entry(p, struct dentry, d_u.d_child);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103 if (!d_unhashed(next) && next->d_inode)
104 n--;
105 p = p->next;
106 }
Eric Dumazet5160ee62006-01-08 01:03:32 -0800107 list_add_tail(&cursor->d_u.d_child, p);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700108 spin_unlock(&dcache_lock);
109 }
110 }
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -0800111 mutex_unlock(&file->f_path.dentry->d_inode->i_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700112 return offset;
113}
114
115/* Relationship between i_mode and the DT_xxx types */
116static inline unsigned char dt_type(struct inode *inode)
117{
118 return (inode->i_mode >> 12) & 15;
119}
120
121/*
122 * Directory is locked and all positive dentries in it are safe, since
123 * for ramfs-type trees they can't go away without unlink() or rmdir(),
124 * both impossible due to the lock on directory.
125 */
126
127int dcache_readdir(struct file * filp, void * dirent, filldir_t filldir)
128{
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -0800129 struct dentry *dentry = filp->f_path.dentry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700130 struct dentry *cursor = filp->private_data;
Eric Dumazet5160ee62006-01-08 01:03:32 -0800131 struct list_head *p, *q = &cursor->d_u.d_child;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700132 ino_t ino;
133 int i = filp->f_pos;
134
135 switch (i) {
136 case 0:
137 ino = dentry->d_inode->i_ino;
138 if (filldir(dirent, ".", 1, i, ino, DT_DIR) < 0)
139 break;
140 filp->f_pos++;
141 i++;
142 /* fallthrough */
143 case 1:
144 ino = parent_ino(dentry);
145 if (filldir(dirent, "..", 2, i, ino, DT_DIR) < 0)
146 break;
147 filp->f_pos++;
148 i++;
149 /* fallthrough */
150 default:
151 spin_lock(&dcache_lock);
Akinobu Mita1bfba4e2006-06-26 00:24:40 -0700152 if (filp->f_pos == 2)
153 list_move(q, &dentry->d_subdirs);
154
Linus Torvalds1da177e2005-04-16 15:20:36 -0700155 for (p=q->next; p != &dentry->d_subdirs; p=p->next) {
156 struct dentry *next;
Eric Dumazet5160ee62006-01-08 01:03:32 -0800157 next = list_entry(p, struct dentry, d_u.d_child);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700158 if (d_unhashed(next) || !next->d_inode)
159 continue;
160
161 spin_unlock(&dcache_lock);
Ronni Nielsen0f8952c22007-05-09 06:44:57 +0200162 if (filldir(dirent, next->d_name.name,
163 next->d_name.len, filp->f_pos,
164 next->d_inode->i_ino,
165 dt_type(next->d_inode)) < 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700166 return 0;
167 spin_lock(&dcache_lock);
168 /* next is still alive */
Akinobu Mita1bfba4e2006-06-26 00:24:40 -0700169 list_move(q, p);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700170 p = q;
171 filp->f_pos++;
172 }
173 spin_unlock(&dcache_lock);
174 }
175 return 0;
176}
177
178ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos)
179{
180 return -EISDIR;
181}
182
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -0800183const struct file_operations simple_dir_operations = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700184 .open = dcache_dir_open,
185 .release = dcache_dir_close,
186 .llseek = dcache_dir_lseek,
187 .read = generic_read_dir,
188 .readdir = dcache_readdir,
Christoph Hellwig1b061d92010-05-26 17:53:41 +0200189 .fsync = noop_fsync,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700190};
191
Arjan van de Ven92e1d5b2007-02-12 00:55:39 -0800192const struct inode_operations simple_dir_inode_operations = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700193 .lookup = simple_lookup,
194};
195
Hugh Dickins759b9772007-03-05 00:30:28 -0800196static const struct super_operations simple_super_operations = {
197 .statfs = simple_statfs,
198};
199
Linus Torvalds1da177e2005-04-16 15:20:36 -0700200/*
201 * Common helper for pseudo-filesystems (sockfs, pipefs, bdev - stuff that
202 * will never be mountable)
203 */
David Howells454e2392006-06-23 02:02:57 -0700204int get_sb_pseudo(struct file_system_type *fs_type, char *name,
Josef 'Jeff' Sipekee9b6d62007-02-12 00:55:41 -0800205 const struct super_operations *ops, unsigned long magic,
David Howells454e2392006-06-23 02:02:57 -0700206 struct vfsmount *mnt)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700207{
208 struct super_block *s = sget(fs_type, NULL, set_anon_super, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700209 struct dentry *dentry;
210 struct inode *root;
211 struct qstr d_name = {.name = name, .len = strlen(name)};
212
213 if (IS_ERR(s))
David Howells454e2392006-06-23 02:02:57 -0700214 return PTR_ERR(s);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700215
216 s->s_flags = MS_NOUSER;
Jeff Layton89a4eb42009-08-18 14:11:08 -0700217 s->s_maxbytes = MAX_LFS_FILESIZE;
Alex Nixon3971e1a2008-07-29 22:33:03 -0700218 s->s_blocksize = PAGE_SIZE;
219 s->s_blocksize_bits = PAGE_SHIFT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700220 s->s_magic = magic;
Hugh Dickins759b9772007-03-05 00:30:28 -0800221 s->s_op = ops ? ops : &simple_super_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222 s->s_time_gran = 1;
223 root = new_inode(s);
224 if (!root)
225 goto Enomem;
Jeff Layton1a1c9bb2007-05-08 00:32:31 -0700226 /*
227 * since this is the first inode, make it number 1. New inodes created
228 * after this must take care not to collide with it (by passing
229 * max_reserved of 1 to iunique).
230 */
231 root->i_ino = 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232 root->i_mode = S_IFDIR | S_IRUSR | S_IWUSR;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700233 root->i_atime = root->i_mtime = root->i_ctime = CURRENT_TIME;
234 dentry = d_alloc(NULL, &d_name);
235 if (!dentry) {
236 iput(root);
237 goto Enomem;
238 }
239 dentry->d_sb = s;
240 dentry->d_parent = dentry;
241 d_instantiate(dentry, root);
242 s->s_root = dentry;
243 s->s_flags |= MS_ACTIVE;
Sukadev Bhattiprolua3ec9472009-03-04 12:06:34 -0800244 simple_set_mnt(mnt, s);
245 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700246
247Enomem:
Al Viro6f5bbff2009-05-06 01:34:22 -0400248 deactivate_locked_super(s);
David Howells454e2392006-06-23 02:02:57 -0700249 return -ENOMEM;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700250}
251
252int simple_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
253{
254 struct inode *inode = old_dentry->d_inode;
255
256 inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
Dave Hansend8c76e62006-09-30 23:29:04 -0700257 inc_nlink(inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700258 atomic_inc(&inode->i_count);
259 dget(dentry);
260 d_instantiate(dentry, inode);
261 return 0;
262}
263
264static inline int simple_positive(struct dentry *dentry)
265{
266 return dentry->d_inode && !d_unhashed(dentry);
267}
268
269int simple_empty(struct dentry *dentry)
270{
271 struct dentry *child;
272 int ret = 0;
273
274 spin_lock(&dcache_lock);
Eric Dumazet5160ee62006-01-08 01:03:32 -0800275 list_for_each_entry(child, &dentry->d_subdirs, d_u.d_child)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700276 if (simple_positive(child))
277 goto out;
278 ret = 1;
279out:
280 spin_unlock(&dcache_lock);
281 return ret;
282}
283
284int simple_unlink(struct inode *dir, struct dentry *dentry)
285{
286 struct inode *inode = dentry->d_inode;
287
288 inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
Dave Hansen9a53c3a2006-09-30 23:29:03 -0700289 drop_nlink(inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700290 dput(dentry);
291 return 0;
292}
293
294int simple_rmdir(struct inode *dir, struct dentry *dentry)
295{
296 if (!simple_empty(dentry))
297 return -ENOTEMPTY;
298
Dave Hansen9a53c3a2006-09-30 23:29:03 -0700299 drop_nlink(dentry->d_inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700300 simple_unlink(dir, dentry);
Dave Hansen9a53c3a2006-09-30 23:29:03 -0700301 drop_nlink(dir);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700302 return 0;
303}
304
305int simple_rename(struct inode *old_dir, struct dentry *old_dentry,
306 struct inode *new_dir, struct dentry *new_dentry)
307{
308 struct inode *inode = old_dentry->d_inode;
309 int they_are_dirs = S_ISDIR(old_dentry->d_inode->i_mode);
310
311 if (!simple_empty(new_dentry))
312 return -ENOTEMPTY;
313
314 if (new_dentry->d_inode) {
315 simple_unlink(new_dir, new_dentry);
316 if (they_are_dirs)
Dave Hansen9a53c3a2006-09-30 23:29:03 -0700317 drop_nlink(old_dir);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700318 } else if (they_are_dirs) {
Dave Hansen9a53c3a2006-09-30 23:29:03 -0700319 drop_nlink(old_dir);
Dave Hansend8c76e62006-09-30 23:29:04 -0700320 inc_nlink(new_dir);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321 }
322
323 old_dir->i_ctime = old_dir->i_mtime = new_dir->i_ctime =
324 new_dir->i_mtime = inode->i_ctime = CURRENT_TIME;
325
326 return 0;
327}
328
npiggin@suse.de7bb46a62010-05-27 01:05:33 +1000329/**
330 * simple_setsize - handle core mm and vfs requirements for file size change
331 * @inode: inode
332 * @newsize: new file size
333 *
334 * Returns 0 on success, -error on failure.
335 *
336 * simple_setsize must be called with inode_mutex held.
337 *
338 * simple_setsize will check that the requested new size is OK (see
339 * inode_newsize_ok), and then will perform the necessary i_size update
340 * and pagecache truncation (if necessary). It will be typically be called
341 * from the filesystem's setattr function when ATTR_SIZE is passed in.
342 *
343 * The inode itself must have correct permissions and attributes to allow
344 * i_size to be changed, this function then just checks that the new size
345 * requested is valid.
346 *
347 * In the case of simple in-memory filesystems with inodes stored solely
348 * in the inode cache, and file data in the pagecache, nothing more needs
349 * to be done to satisfy a truncate request. Filesystems with on-disk
350 * blocks for example will need to free them in the case of truncate, in
351 * that case it may be easier not to use simple_setsize (but each of its
352 * components will likely be required at some point to update pagecache
353 * and inode etc).
354 */
355int simple_setsize(struct inode *inode, loff_t newsize)
356{
357 loff_t oldsize;
358 int error;
359
360 error = inode_newsize_ok(inode, newsize);
361 if (error)
362 return error;
363
364 oldsize = inode->i_size;
365 i_size_write(inode, newsize);
366 truncate_pagecache(inode, oldsize, newsize);
367
368 return error;
369}
370EXPORT_SYMBOL(simple_setsize);
371
372/**
373 * simple_setattr - setattr for simple in-memory filesystem
374 * @dentry: dentry
375 * @iattr: iattr structure
376 *
377 * Returns 0 on success, -error on failure.
378 *
379 * simple_setattr implements setattr for an in-memory filesystem which
380 * does not store its own file data or metadata (eg. uses the page cache
381 * and inode cache as its data store).
382 */
383int simple_setattr(struct dentry *dentry, struct iattr *iattr)
384{
385 struct inode *inode = dentry->d_inode;
386 int error;
387
388 error = inode_change_ok(inode, iattr);
389 if (error)
390 return error;
391
392 if (iattr->ia_valid & ATTR_SIZE) {
393 error = simple_setsize(inode, iattr->ia_size);
394 if (error)
395 return error;
396 }
397
398 generic_setattr(inode, iattr);
399
400 return error;
401}
402EXPORT_SYMBOL(simple_setattr);
403
Linus Torvalds1da177e2005-04-16 15:20:36 -0700404int simple_readpage(struct file *file, struct page *page)
405{
Pekka J Enbergc0d92cb2006-09-29 01:59:09 -0700406 clear_highpage(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700407 flush_dcache_page(page);
408 SetPageUptodate(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700409 unlock_page(page);
410 return 0;
411}
412
Nick Pigginafddba42007-10-16 01:25:01 -0700413int simple_write_begin(struct file *file, struct address_space *mapping,
414 loff_t pos, unsigned len, unsigned flags,
415 struct page **pagep, void **fsdata)
416{
417 struct page *page;
418 pgoff_t index;
Nick Pigginafddba42007-10-16 01:25:01 -0700419
420 index = pos >> PAGE_CACHE_SHIFT;
Nick Pigginafddba42007-10-16 01:25:01 -0700421
Nick Piggin54566b22009-01-04 12:00:53 -0800422 page = grab_cache_page_write_begin(mapping, index, flags);
Nick Pigginafddba42007-10-16 01:25:01 -0700423 if (!page)
424 return -ENOMEM;
425
426 *pagep = page;
427
Boaz Harrosh193cf4b2010-01-12 16:18:08 +0200428 if (!PageUptodate(page) && (len != PAGE_CACHE_SIZE)) {
429 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
430
431 zero_user_segments(page, 0, from, from + len, PAGE_CACHE_SIZE);
432 }
433 return 0;
Nick Pigginafddba42007-10-16 01:25:01 -0700434}
435
Boaz Harroshad2a722f2010-01-12 15:13:47 +0200436/**
437 * simple_write_end - .write_end helper for non-block-device FSes
438 * @available: See .write_end of address_space_operations
439 * @file: "
440 * @mapping: "
441 * @pos: "
442 * @len: "
443 * @copied: "
444 * @page: "
445 * @fsdata: "
446 *
447 * simple_write_end does the minimum needed for updating a page after writing is
448 * done. It has the same API signature as the .write_end of
449 * address_space_operations vector. So it can just be set onto .write_end for
450 * FSes that don't need any other processing. i_mutex is assumed to be held.
451 * Block based filesystems should use generic_write_end().
452 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
453 * is not called, so a filesystem that actually does store data in .write_inode
454 * should extend on what's done here with a call to mark_inode_dirty() in the
455 * case that i_size has changed.
456 */
457int simple_write_end(struct file *file, struct address_space *mapping,
458 loff_t pos, unsigned len, unsigned copied,
459 struct page *page, void *fsdata)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700460{
461 struct inode *inode = page->mapping->host;
Boaz Harroshad2a722f2010-01-12 15:13:47 +0200462 loff_t last_pos = pos + copied;
463
464 /* zero the stale part of the page if we did a short copy */
465 if (copied < len) {
466 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
467
468 zero_user(page, from + copied, len - copied);
469 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700470
Nick Piggin955eff52007-02-20 13:58:08 -0800471 if (!PageUptodate(page))
472 SetPageUptodate(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700473 /*
474 * No need to use i_size_read() here, the i_size
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800475 * cannot change under us because we hold the i_mutex.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700476 */
Boaz Harroshad2a722f2010-01-12 15:13:47 +0200477 if (last_pos > inode->i_size)
478 i_size_write(inode, last_pos);
479
Linus Torvalds1da177e2005-04-16 15:20:36 -0700480 set_page_dirty(page);
Nick Pigginafddba42007-10-16 01:25:01 -0700481 unlock_page(page);
482 page_cache_release(page);
483
484 return copied;
485}
486
Jeff Layton1a1c9bb2007-05-08 00:32:31 -0700487/*
488 * the inodes created here are not hashed. If you use iunique to generate
489 * unique inode values later for this filesystem, then you must take care
490 * to pass it an appropriate max_reserved value to avoid collisions.
491 */
Roberto Sassu7d683a02010-06-03 11:58:28 +0200492int simple_fill_super(struct super_block *s, unsigned long magic,
493 struct tree_descr *files)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700494{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700495 struct inode *inode;
496 struct dentry *root;
497 struct dentry *dentry;
498 int i;
499
500 s->s_blocksize = PAGE_CACHE_SIZE;
501 s->s_blocksize_bits = PAGE_CACHE_SHIFT;
502 s->s_magic = magic;
Hugh Dickins759b9772007-03-05 00:30:28 -0800503 s->s_op = &simple_super_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700504 s->s_time_gran = 1;
505
506 inode = new_inode(s);
507 if (!inode)
508 return -ENOMEM;
Jeff Layton1a1c9bb2007-05-08 00:32:31 -0700509 /*
510 * because the root inode is 1, the files array must not contain an
511 * entry at index 1
512 */
513 inode->i_ino = 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700514 inode->i_mode = S_IFDIR | 0755;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700515 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
516 inode->i_op = &simple_dir_inode_operations;
517 inode->i_fop = &simple_dir_operations;
Vincent Hanquez7656f322006-02-03 03:04:48 -0800518 inode->i_nlink = 2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700519 root = d_alloc_root(inode);
520 if (!root) {
521 iput(inode);
522 return -ENOMEM;
523 }
524 for (i = 0; !files->name || files->name[0]; i++, files++) {
525 if (!files->name)
526 continue;
Jeff Layton1a1c9bb2007-05-08 00:32:31 -0700527
528 /* warn if it tries to conflict with the root inode */
529 if (unlikely(i == 1))
530 printk(KERN_WARNING "%s: %s passed in a files array"
531 "with an index of 1!\n", __func__,
532 s->s_type->name);
533
Linus Torvalds1da177e2005-04-16 15:20:36 -0700534 dentry = d_alloc_name(root, files->name);
535 if (!dentry)
536 goto out;
537 inode = new_inode(s);
538 if (!inode)
539 goto out;
540 inode->i_mode = S_IFREG | files->mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700541 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
542 inode->i_fop = files->ops;
543 inode->i_ino = i;
544 d_add(dentry, inode);
545 }
546 s->s_root = root;
547 return 0;
548out:
549 d_genocide(root);
550 dput(root);
551 return -ENOMEM;
552}
553
554static DEFINE_SPINLOCK(pin_fs_lock);
555
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -0400556int simple_pin_fs(struct file_system_type *type, struct vfsmount **mount, int *count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700557{
558 struct vfsmount *mnt = NULL;
559 spin_lock(&pin_fs_lock);
560 if (unlikely(!*mount)) {
561 spin_unlock(&pin_fs_lock);
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -0400562 mnt = vfs_kern_mount(type, 0, type->name, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700563 if (IS_ERR(mnt))
564 return PTR_ERR(mnt);
565 spin_lock(&pin_fs_lock);
566 if (!*mount)
567 *mount = mnt;
568 }
569 mntget(*mount);
570 ++*count;
571 spin_unlock(&pin_fs_lock);
572 mntput(mnt);
573 return 0;
574}
575
576void simple_release_fs(struct vfsmount **mount, int *count)
577{
578 struct vfsmount *mnt;
579 spin_lock(&pin_fs_lock);
580 mnt = *mount;
581 if (!--*count)
582 *mount = NULL;
583 spin_unlock(&pin_fs_lock);
584 mntput(mnt);
585}
586
Akinobu Mita6d1029b2008-07-04 09:59:51 -0700587/**
588 * simple_read_from_buffer - copy data from the buffer to user space
589 * @to: the user space buffer to read to
590 * @count: the maximum number of bytes to read
591 * @ppos: the current position in the buffer
592 * @from: the buffer to read from
593 * @available: the size of the buffer
594 *
595 * The simple_read_from_buffer() function reads up to @count bytes from the
596 * buffer @from at offset @ppos into the user space address starting at @to.
597 *
598 * On success, the number of bytes read is returned and the offset @ppos is
599 * advanced by this number, or negative value is returned on error.
600 **/
Linus Torvalds1da177e2005-04-16 15:20:36 -0700601ssize_t simple_read_from_buffer(void __user *to, size_t count, loff_t *ppos,
602 const void *from, size_t available)
603{
604 loff_t pos = *ppos;
Steven Rostedt14be2742009-09-18 13:05:42 -0700605 size_t ret;
606
Linus Torvalds1da177e2005-04-16 15:20:36 -0700607 if (pos < 0)
608 return -EINVAL;
Steven Rostedt14be2742009-09-18 13:05:42 -0700609 if (pos >= available || !count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610 return 0;
611 if (count > available - pos)
612 count = available - pos;
Steven Rostedt14be2742009-09-18 13:05:42 -0700613 ret = copy_to_user(to, from + pos, count);
614 if (ret == count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700615 return -EFAULT;
Steven Rostedt14be2742009-09-18 13:05:42 -0700616 count -= ret;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700617 *ppos = pos + count;
618 return count;
619}
620
Akinobu Mita6d1029b2008-07-04 09:59:51 -0700621/**
Jiri Slaby6a727b42010-05-01 23:51:22 +0200622 * simple_write_to_buffer - copy data from user space to the buffer
623 * @to: the buffer to write to
624 * @available: the size of the buffer
625 * @ppos: the current position in the buffer
626 * @from: the user space buffer to read from
627 * @count: the maximum number of bytes to read
628 *
629 * The simple_write_to_buffer() function reads up to @count bytes from the user
630 * space address starting at @from into the buffer @to at offset @ppos.
631 *
632 * On success, the number of bytes written is returned and the offset @ppos is
633 * advanced by this number, or negative value is returned on error.
634 **/
635ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
636 const void __user *from, size_t count)
637{
638 loff_t pos = *ppos;
639 size_t res;
640
641 if (pos < 0)
642 return -EINVAL;
643 if (pos >= available || !count)
644 return 0;
645 if (count > available - pos)
646 count = available - pos;
647 res = copy_from_user(to + pos, from, count);
648 if (res == count)
649 return -EFAULT;
650 count -= res;
651 *ppos = pos + count;
652 return count;
653}
654
655/**
Akinobu Mita6d1029b2008-07-04 09:59:51 -0700656 * memory_read_from_buffer - copy data from the buffer
657 * @to: the kernel space buffer to read to
658 * @count: the maximum number of bytes to read
659 * @ppos: the current position in the buffer
660 * @from: the buffer to read from
661 * @available: the size of the buffer
662 *
663 * The memory_read_from_buffer() function reads up to @count bytes from the
664 * buffer @from at offset @ppos into the kernel space address starting at @to.
665 *
666 * On success, the number of bytes read is returned and the offset @ppos is
667 * advanced by this number, or negative value is returned on error.
668 **/
Akinobu Mita93b07112008-06-05 22:46:21 -0700669ssize_t memory_read_from_buffer(void *to, size_t count, loff_t *ppos,
670 const void *from, size_t available)
671{
672 loff_t pos = *ppos;
673
674 if (pos < 0)
675 return -EINVAL;
676 if (pos >= available)
677 return 0;
678 if (count > available - pos)
679 count = available - pos;
680 memcpy(to, from + pos, count);
681 *ppos = pos + count;
682
683 return count;
684}
685
Linus Torvalds1da177e2005-04-16 15:20:36 -0700686/*
687 * Transaction based IO.
688 * The file expects a single write which triggers the transaction, and then
689 * possibly a read which collects the result - which is stored in a
690 * file-local buffer.
691 */
Ingo Molnar76791ab2009-03-25 16:48:35 +0100692
693void simple_transaction_set(struct file *file, size_t n)
694{
695 struct simple_transaction_argresp *ar = file->private_data;
696
697 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
698
699 /*
700 * The barrier ensures that ar->size will really remain zero until
701 * ar->data is ready for reading.
702 */
703 smp_mb();
704 ar->size = n;
705}
706
Linus Torvalds1da177e2005-04-16 15:20:36 -0700707char *simple_transaction_get(struct file *file, const char __user *buf, size_t size)
708{
709 struct simple_transaction_argresp *ar;
710 static DEFINE_SPINLOCK(simple_transaction_lock);
711
712 if (size > SIMPLE_TRANSACTION_LIMIT - 1)
713 return ERR_PTR(-EFBIG);
714
715 ar = (struct simple_transaction_argresp *)get_zeroed_page(GFP_KERNEL);
716 if (!ar)
717 return ERR_PTR(-ENOMEM);
718
719 spin_lock(&simple_transaction_lock);
720
721 /* only one write allowed per open */
722 if (file->private_data) {
723 spin_unlock(&simple_transaction_lock);
724 free_page((unsigned long)ar);
725 return ERR_PTR(-EBUSY);
726 }
727
728 file->private_data = ar;
729
730 spin_unlock(&simple_transaction_lock);
731
732 if (copy_from_user(ar->data, buf, size))
733 return ERR_PTR(-EFAULT);
734
735 return ar->data;
736}
737
738ssize_t simple_transaction_read(struct file *file, char __user *buf, size_t size, loff_t *pos)
739{
740 struct simple_transaction_argresp *ar = file->private_data;
741
742 if (!ar)
743 return 0;
744 return simple_read_from_buffer(buf, size, pos, ar->data, ar->size);
745}
746
747int simple_transaction_release(struct inode *inode, struct file *file)
748{
749 free_page((unsigned long)file->private_data);
750 return 0;
751}
752
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200753/* Simple attribute files */
754
755struct simple_attr {
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800756 int (*get)(void *, u64 *);
757 int (*set)(void *, u64);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200758 char get_buf[24]; /* enough to store a u64 and "\n\0" */
759 char set_buf[24];
760 void *data;
761 const char *fmt; /* format for read operation */
Ingo Molnar7cf34c72006-03-23 03:00:36 -0800762 struct mutex mutex; /* protects access to these buffers */
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200763};
764
765/* simple_attr_open is called by an actual attribute open file operation
766 * to set the attribute specific access operations. */
767int simple_attr_open(struct inode *inode, struct file *file,
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800768 int (*get)(void *, u64 *), int (*set)(void *, u64),
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200769 const char *fmt)
770{
771 struct simple_attr *attr;
772
773 attr = kmalloc(sizeof(*attr), GFP_KERNEL);
774 if (!attr)
775 return -ENOMEM;
776
777 attr->get = get;
778 attr->set = set;
Theodore Ts'o8e18e292006-09-27 01:50:46 -0700779 attr->data = inode->i_private;
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200780 attr->fmt = fmt;
Ingo Molnar7cf34c72006-03-23 03:00:36 -0800781 mutex_init(&attr->mutex);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200782
783 file->private_data = attr;
784
785 return nonseekable_open(inode, file);
786}
787
Christoph Hellwig74bedc42008-02-08 04:20:28 -0800788int simple_attr_release(struct inode *inode, struct file *file)
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200789{
790 kfree(file->private_data);
791 return 0;
792}
793
794/* read from the buffer that is filled with the get function */
795ssize_t simple_attr_read(struct file *file, char __user *buf,
796 size_t len, loff_t *ppos)
797{
798 struct simple_attr *attr;
799 size_t size;
800 ssize_t ret;
801
802 attr = file->private_data;
803
804 if (!attr->get)
805 return -EACCES;
806
Christoph Hellwig92613032008-02-08 04:20:27 -0800807 ret = mutex_lock_interruptible(&attr->mutex);
808 if (ret)
809 return ret;
810
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800811 if (*ppos) { /* continued read */
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200812 size = strlen(attr->get_buf);
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800813 } else { /* first read */
814 u64 val;
815 ret = attr->get(attr->data, &val);
816 if (ret)
817 goto out;
818
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200819 size = scnprintf(attr->get_buf, sizeof(attr->get_buf),
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800820 attr->fmt, (unsigned long long)val);
821 }
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200822
823 ret = simple_read_from_buffer(buf, len, ppos, attr->get_buf, size);
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800824out:
Ingo Molnar7cf34c72006-03-23 03:00:36 -0800825 mutex_unlock(&attr->mutex);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200826 return ret;
827}
828
829/* interpret the buffer as a number to call the set function with */
830ssize_t simple_attr_write(struct file *file, const char __user *buf,
831 size_t len, loff_t *ppos)
832{
833 struct simple_attr *attr;
834 u64 val;
835 size_t size;
836 ssize_t ret;
837
838 attr = file->private_data;
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200839 if (!attr->set)
840 return -EACCES;
841
Christoph Hellwig92613032008-02-08 04:20:27 -0800842 ret = mutex_lock_interruptible(&attr->mutex);
843 if (ret)
844 return ret;
845
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200846 ret = -EFAULT;
847 size = min(sizeof(attr->set_buf) - 1, len);
848 if (copy_from_user(attr->set_buf, buf, size))
849 goto out;
850
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200851 attr->set_buf[size] = '\0';
852 val = simple_strtol(attr->set_buf, NULL, 0);
Wu Fengguang05cc0ce2009-09-18 13:06:03 -0700853 ret = attr->set(attr->data, val);
854 if (ret == 0)
855 ret = len; /* on success, claim we got the whole input */
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200856out:
Ingo Molnar7cf34c72006-03-23 03:00:36 -0800857 mutex_unlock(&attr->mutex);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200858 return ret;
859}
860
Christoph Hellwig25961102007-10-21 16:42:05 -0700861/**
862 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
863 * @sb: filesystem to do the file handle conversion on
864 * @fid: file handle to convert
865 * @fh_len: length of the file handle in bytes
866 * @fh_type: type of file handle
867 * @get_inode: filesystem callback to retrieve inode
868 *
869 * This function decodes @fid as long as it has one of the well-known
870 * Linux filehandle types and calls @get_inode on it to retrieve the
871 * inode for the object specified in the file handle.
872 */
873struct dentry *generic_fh_to_dentry(struct super_block *sb, struct fid *fid,
874 int fh_len, int fh_type, struct inode *(*get_inode)
875 (struct super_block *sb, u64 ino, u32 gen))
876{
877 struct inode *inode = NULL;
878
879 if (fh_len < 2)
880 return NULL;
881
882 switch (fh_type) {
883 case FILEID_INO32_GEN:
884 case FILEID_INO32_GEN_PARENT:
885 inode = get_inode(sb, fid->i32.ino, fid->i32.gen);
886 break;
887 }
888
Christoph Hellwig4ea3ada2008-08-11 15:48:57 +0200889 return d_obtain_alias(inode);
Christoph Hellwig25961102007-10-21 16:42:05 -0700890}
891EXPORT_SYMBOL_GPL(generic_fh_to_dentry);
892
893/**
894 * generic_fh_to_dentry - generic helper for the fh_to_parent export operation
895 * @sb: filesystem to do the file handle conversion on
896 * @fid: file handle to convert
897 * @fh_len: length of the file handle in bytes
898 * @fh_type: type of file handle
899 * @get_inode: filesystem callback to retrieve inode
900 *
901 * This function decodes @fid as long as it has one of the well-known
902 * Linux filehandle types and calls @get_inode on it to retrieve the
903 * inode for the _parent_ object specified in the file handle if it
904 * is specified in the file handle, or NULL otherwise.
905 */
906struct dentry *generic_fh_to_parent(struct super_block *sb, struct fid *fid,
907 int fh_len, int fh_type, struct inode *(*get_inode)
908 (struct super_block *sb, u64 ino, u32 gen))
909{
910 struct inode *inode = NULL;
911
912 if (fh_len <= 2)
913 return NULL;
914
915 switch (fh_type) {
916 case FILEID_INO32_GEN_PARENT:
917 inode = get_inode(sb, fid->i32.parent_ino,
918 (fh_len > 3 ? fid->i32.parent_gen : 0));
919 break;
920 }
921
Christoph Hellwig4ea3ada2008-08-11 15:48:57 +0200922 return d_obtain_alias(inode);
Christoph Hellwig25961102007-10-21 16:42:05 -0700923}
924EXPORT_SYMBOL_GPL(generic_fh_to_parent);
925
Christoph Hellwig1b061d92010-05-26 17:53:41 +0200926/**
927 * generic_file_fsync - generic fsync implementation for simple filesystems
928 * @file: file to synchronize
929 * @datasync: only synchronize essential metadata if true
930 *
931 * This is a generic implementation of the fsync method for simple
932 * filesystems which track all non-inode metadata in the buffers list
933 * hanging off the address_space structure.
934 */
935int generic_file_fsync(struct file *file, int datasync)
Al Virod5aacad2009-06-07 14:56:44 -0400936{
937 struct writeback_control wbc = {
938 .sync_mode = WB_SYNC_ALL,
939 .nr_to_write = 0, /* metadata-only; caller takes care of data */
940 };
Christoph Hellwig7ea80852010-05-26 17:53:25 +0200941 struct inode *inode = file->f_mapping->host;
Al Virod5aacad2009-06-07 14:56:44 -0400942 int err;
943 int ret;
944
945 ret = sync_mapping_buffers(inode->i_mapping);
946 if (!(inode->i_state & I_DIRTY))
947 return ret;
948 if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))
949 return ret;
950
951 err = sync_inode(inode, &wbc);
952 if (ret == 0)
953 ret = err;
954 return ret;
955}
Christoph Hellwig1b061d92010-05-26 17:53:41 +0200956EXPORT_SYMBOL(generic_file_fsync);
957
958/*
959 * No-op implementation of ->fsync for in-memory filesystems.
960 */
961int noop_fsync(struct file *file, int datasync)
962{
963 return 0;
964}
Al Virod5aacad2009-06-07 14:56:44 -0400965
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966EXPORT_SYMBOL(dcache_dir_close);
967EXPORT_SYMBOL(dcache_dir_lseek);
968EXPORT_SYMBOL(dcache_dir_open);
969EXPORT_SYMBOL(dcache_readdir);
970EXPORT_SYMBOL(generic_read_dir);
971EXPORT_SYMBOL(get_sb_pseudo);
Nick Pigginafddba42007-10-16 01:25:01 -0700972EXPORT_SYMBOL(simple_write_begin);
973EXPORT_SYMBOL(simple_write_end);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700974EXPORT_SYMBOL(simple_dir_inode_operations);
975EXPORT_SYMBOL(simple_dir_operations);
976EXPORT_SYMBOL(simple_empty);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700977EXPORT_SYMBOL(simple_fill_super);
978EXPORT_SYMBOL(simple_getattr);
979EXPORT_SYMBOL(simple_link);
980EXPORT_SYMBOL(simple_lookup);
981EXPORT_SYMBOL(simple_pin_fs);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700982EXPORT_SYMBOL(simple_readpage);
983EXPORT_SYMBOL(simple_release_fs);
984EXPORT_SYMBOL(simple_rename);
985EXPORT_SYMBOL(simple_rmdir);
986EXPORT_SYMBOL(simple_statfs);
Christoph Hellwig1b061d92010-05-26 17:53:41 +0200987EXPORT_SYMBOL(noop_fsync);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700988EXPORT_SYMBOL(simple_unlink);
989EXPORT_SYMBOL(simple_read_from_buffer);
Jiri Slaby6a727b42010-05-01 23:51:22 +0200990EXPORT_SYMBOL(simple_write_to_buffer);
Akinobu Mita93b07112008-06-05 22:46:21 -0700991EXPORT_SYMBOL(memory_read_from_buffer);
Ingo Molnar76791ab2009-03-25 16:48:35 +0100992EXPORT_SYMBOL(simple_transaction_set);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700993EXPORT_SYMBOL(simple_transaction_get);
994EXPORT_SYMBOL(simple_transaction_read);
995EXPORT_SYMBOL(simple_transaction_release);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200996EXPORT_SYMBOL_GPL(simple_attr_open);
Christoph Hellwig74bedc42008-02-08 04:20:28 -0800997EXPORT_SYMBOL_GPL(simple_attr_release);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200998EXPORT_SYMBOL_GPL(simple_attr_read);
999EXPORT_SYMBOL_GPL(simple_attr_write);