blob: 40562224b71899c12377eb6b572f2bcac672f57c [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/**
Christoph Hellwigeef23802010-06-04 11:30:01 +0200373 * simple_setattr - setattr for simple filesystem
npiggin@suse.de7bb46a62010-05-27 01:05:33 +1000374 * @dentry: dentry
375 * @iattr: iattr structure
376 *
377 * Returns 0 on success, -error on failure.
378 *
Christoph Hellwigeef23802010-06-04 11:30:01 +0200379 * simple_setattr is a simple ->setattr implementation without a proper
380 * implementation of size changes.
381 *
382 * It can either be used for in-memory filesystems or special files
383 * on simple regular filesystems. Anything that needs to change on-disk
384 * or wire state on size changes needs its own setattr method.
npiggin@suse.de7bb46a62010-05-27 01:05:33 +1000385 */
386int simple_setattr(struct dentry *dentry, struct iattr *iattr)
387{
388 struct inode *inode = dentry->d_inode;
389 int error;
390
Christoph Hellwigeef23802010-06-04 11:30:01 +0200391 WARN_ON_ONCE(inode->i_op->truncate);
392
npiggin@suse.de7bb46a62010-05-27 01:05:33 +1000393 error = inode_change_ok(inode, iattr);
394 if (error)
395 return error;
396
397 if (iattr->ia_valid & ATTR_SIZE) {
398 error = simple_setsize(inode, iattr->ia_size);
399 if (error)
400 return error;
401 }
402
Christoph Hellwig6a1a90a2010-06-04 11:30:00 +0200403 setattr_copy(inode, iattr);
Christoph Hellwigeef23802010-06-04 11:30:01 +0200404 mark_inode_dirty(inode);
405 return 0;
npiggin@suse.de7bb46a62010-05-27 01:05:33 +1000406}
407EXPORT_SYMBOL(simple_setattr);
408
Linus Torvalds1da177e2005-04-16 15:20:36 -0700409int simple_readpage(struct file *file, struct page *page)
410{
Pekka J Enbergc0d92cb2006-09-29 01:59:09 -0700411 clear_highpage(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700412 flush_dcache_page(page);
413 SetPageUptodate(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414 unlock_page(page);
415 return 0;
416}
417
Nick Pigginafddba42007-10-16 01:25:01 -0700418int simple_write_begin(struct file *file, struct address_space *mapping,
419 loff_t pos, unsigned len, unsigned flags,
420 struct page **pagep, void **fsdata)
421{
422 struct page *page;
423 pgoff_t index;
Nick Pigginafddba42007-10-16 01:25:01 -0700424
425 index = pos >> PAGE_CACHE_SHIFT;
Nick Pigginafddba42007-10-16 01:25:01 -0700426
Nick Piggin54566b22009-01-04 12:00:53 -0800427 page = grab_cache_page_write_begin(mapping, index, flags);
Nick Pigginafddba42007-10-16 01:25:01 -0700428 if (!page)
429 return -ENOMEM;
430
431 *pagep = page;
432
Boaz Harrosh193cf4b2010-01-12 16:18:08 +0200433 if (!PageUptodate(page) && (len != PAGE_CACHE_SIZE)) {
434 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
435
436 zero_user_segments(page, 0, from, from + len, PAGE_CACHE_SIZE);
437 }
438 return 0;
Nick Pigginafddba42007-10-16 01:25:01 -0700439}
440
Boaz Harroshad2a722f2010-01-12 15:13:47 +0200441/**
442 * simple_write_end - .write_end helper for non-block-device FSes
443 * @available: See .write_end of address_space_operations
444 * @file: "
445 * @mapping: "
446 * @pos: "
447 * @len: "
448 * @copied: "
449 * @page: "
450 * @fsdata: "
451 *
452 * simple_write_end does the minimum needed for updating a page after writing is
453 * done. It has the same API signature as the .write_end of
454 * address_space_operations vector. So it can just be set onto .write_end for
455 * FSes that don't need any other processing. i_mutex is assumed to be held.
456 * Block based filesystems should use generic_write_end().
457 * NOTE: Even though i_size might get updated by this function, mark_inode_dirty
458 * is not called, so a filesystem that actually does store data in .write_inode
459 * should extend on what's done here with a call to mark_inode_dirty() in the
460 * case that i_size has changed.
461 */
462int simple_write_end(struct file *file, struct address_space *mapping,
463 loff_t pos, unsigned len, unsigned copied,
464 struct page *page, void *fsdata)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700465{
466 struct inode *inode = page->mapping->host;
Boaz Harroshad2a722f2010-01-12 15:13:47 +0200467 loff_t last_pos = pos + copied;
468
469 /* zero the stale part of the page if we did a short copy */
470 if (copied < len) {
471 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
472
473 zero_user(page, from + copied, len - copied);
474 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700475
Nick Piggin955eff52007-02-20 13:58:08 -0800476 if (!PageUptodate(page))
477 SetPageUptodate(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700478 /*
479 * No need to use i_size_read() here, the i_size
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800480 * cannot change under us because we hold the i_mutex.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700481 */
Boaz Harroshad2a722f2010-01-12 15:13:47 +0200482 if (last_pos > inode->i_size)
483 i_size_write(inode, last_pos);
484
Linus Torvalds1da177e2005-04-16 15:20:36 -0700485 set_page_dirty(page);
Nick Pigginafddba42007-10-16 01:25:01 -0700486 unlock_page(page);
487 page_cache_release(page);
488
489 return copied;
490}
491
Jeff Layton1a1c9bb2007-05-08 00:32:31 -0700492/*
493 * the inodes created here are not hashed. If you use iunique to generate
494 * unique inode values later for this filesystem, then you must take care
495 * to pass it an appropriate max_reserved value to avoid collisions.
496 */
Roberto Sassu7d683a02010-06-03 11:58:28 +0200497int simple_fill_super(struct super_block *s, unsigned long magic,
498 struct tree_descr *files)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700499{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700500 struct inode *inode;
501 struct dentry *root;
502 struct dentry *dentry;
503 int i;
504
505 s->s_blocksize = PAGE_CACHE_SIZE;
506 s->s_blocksize_bits = PAGE_CACHE_SHIFT;
507 s->s_magic = magic;
Hugh Dickins759b9772007-03-05 00:30:28 -0800508 s->s_op = &simple_super_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700509 s->s_time_gran = 1;
510
511 inode = new_inode(s);
512 if (!inode)
513 return -ENOMEM;
Jeff Layton1a1c9bb2007-05-08 00:32:31 -0700514 /*
515 * because the root inode is 1, the files array must not contain an
516 * entry at index 1
517 */
518 inode->i_ino = 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700519 inode->i_mode = S_IFDIR | 0755;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700520 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
521 inode->i_op = &simple_dir_inode_operations;
522 inode->i_fop = &simple_dir_operations;
Vincent Hanquez7656f322006-02-03 03:04:48 -0800523 inode->i_nlink = 2;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700524 root = d_alloc_root(inode);
525 if (!root) {
526 iput(inode);
527 return -ENOMEM;
528 }
529 for (i = 0; !files->name || files->name[0]; i++, files++) {
530 if (!files->name)
531 continue;
Jeff Layton1a1c9bb2007-05-08 00:32:31 -0700532
533 /* warn if it tries to conflict with the root inode */
534 if (unlikely(i == 1))
535 printk(KERN_WARNING "%s: %s passed in a files array"
536 "with an index of 1!\n", __func__,
537 s->s_type->name);
538
Linus Torvalds1da177e2005-04-16 15:20:36 -0700539 dentry = d_alloc_name(root, files->name);
540 if (!dentry)
541 goto out;
542 inode = new_inode(s);
543 if (!inode)
544 goto out;
545 inode->i_mode = S_IFREG | files->mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700546 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
547 inode->i_fop = files->ops;
548 inode->i_ino = i;
549 d_add(dentry, inode);
550 }
551 s->s_root = root;
552 return 0;
553out:
554 d_genocide(root);
555 dput(root);
556 return -ENOMEM;
557}
558
559static DEFINE_SPINLOCK(pin_fs_lock);
560
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -0400561int simple_pin_fs(struct file_system_type *type, struct vfsmount **mount, int *count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700562{
563 struct vfsmount *mnt = NULL;
564 spin_lock(&pin_fs_lock);
565 if (unlikely(!*mount)) {
566 spin_unlock(&pin_fs_lock);
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -0400567 mnt = vfs_kern_mount(type, 0, type->name, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700568 if (IS_ERR(mnt))
569 return PTR_ERR(mnt);
570 spin_lock(&pin_fs_lock);
571 if (!*mount)
572 *mount = mnt;
573 }
574 mntget(*mount);
575 ++*count;
576 spin_unlock(&pin_fs_lock);
577 mntput(mnt);
578 return 0;
579}
580
581void simple_release_fs(struct vfsmount **mount, int *count)
582{
583 struct vfsmount *mnt;
584 spin_lock(&pin_fs_lock);
585 mnt = *mount;
586 if (!--*count)
587 *mount = NULL;
588 spin_unlock(&pin_fs_lock);
589 mntput(mnt);
590}
591
Akinobu Mita6d1029b2008-07-04 09:59:51 -0700592/**
593 * simple_read_from_buffer - copy data from the buffer to user space
594 * @to: the user space buffer to read to
595 * @count: the maximum number of bytes to read
596 * @ppos: the current position in the buffer
597 * @from: the buffer to read from
598 * @available: the size of the buffer
599 *
600 * The simple_read_from_buffer() function reads up to @count bytes from the
601 * buffer @from at offset @ppos into the user space address starting at @to.
602 *
603 * On success, the number of bytes read is returned and the offset @ppos is
604 * advanced by this number, or negative value is returned on error.
605 **/
Linus Torvalds1da177e2005-04-16 15:20:36 -0700606ssize_t simple_read_from_buffer(void __user *to, size_t count, loff_t *ppos,
607 const void *from, size_t available)
608{
609 loff_t pos = *ppos;
Steven Rostedt14be2742009-09-18 13:05:42 -0700610 size_t ret;
611
Linus Torvalds1da177e2005-04-16 15:20:36 -0700612 if (pos < 0)
613 return -EINVAL;
Steven Rostedt14be2742009-09-18 13:05:42 -0700614 if (pos >= available || !count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700615 return 0;
616 if (count > available - pos)
617 count = available - pos;
Steven Rostedt14be2742009-09-18 13:05:42 -0700618 ret = copy_to_user(to, from + pos, count);
619 if (ret == count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700620 return -EFAULT;
Steven Rostedt14be2742009-09-18 13:05:42 -0700621 count -= ret;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700622 *ppos = pos + count;
623 return count;
624}
625
Akinobu Mita6d1029b2008-07-04 09:59:51 -0700626/**
Jiri Slaby6a727b42010-05-01 23:51:22 +0200627 * simple_write_to_buffer - copy data from user space to the buffer
628 * @to: the buffer to write to
629 * @available: the size of the buffer
630 * @ppos: the current position in the buffer
631 * @from: the user space buffer to read from
632 * @count: the maximum number of bytes to read
633 *
634 * The simple_write_to_buffer() function reads up to @count bytes from the user
635 * space address starting at @from into the buffer @to at offset @ppos.
636 *
637 * On success, the number of bytes written is returned and the offset @ppos is
638 * advanced by this number, or negative value is returned on error.
639 **/
640ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
641 const void __user *from, size_t count)
642{
643 loff_t pos = *ppos;
644 size_t res;
645
646 if (pos < 0)
647 return -EINVAL;
648 if (pos >= available || !count)
649 return 0;
650 if (count > available - pos)
651 count = available - pos;
652 res = copy_from_user(to + pos, from, count);
653 if (res == count)
654 return -EFAULT;
655 count -= res;
656 *ppos = pos + count;
657 return count;
658}
659
660/**
Akinobu Mita6d1029b2008-07-04 09:59:51 -0700661 * memory_read_from_buffer - copy data from the buffer
662 * @to: the kernel space buffer to read to
663 * @count: the maximum number of bytes to read
664 * @ppos: the current position in the buffer
665 * @from: the buffer to read from
666 * @available: the size of the buffer
667 *
668 * The memory_read_from_buffer() function reads up to @count bytes from the
669 * buffer @from at offset @ppos into the kernel space address starting at @to.
670 *
671 * On success, the number of bytes read is returned and the offset @ppos is
672 * advanced by this number, or negative value is returned on error.
673 **/
Akinobu Mita93b07112008-06-05 22:46:21 -0700674ssize_t memory_read_from_buffer(void *to, size_t count, loff_t *ppos,
675 const void *from, size_t available)
676{
677 loff_t pos = *ppos;
678
679 if (pos < 0)
680 return -EINVAL;
681 if (pos >= available)
682 return 0;
683 if (count > available - pos)
684 count = available - pos;
685 memcpy(to, from + pos, count);
686 *ppos = pos + count;
687
688 return count;
689}
690
Linus Torvalds1da177e2005-04-16 15:20:36 -0700691/*
692 * Transaction based IO.
693 * The file expects a single write which triggers the transaction, and then
694 * possibly a read which collects the result - which is stored in a
695 * file-local buffer.
696 */
Ingo Molnar76791ab2009-03-25 16:48:35 +0100697
698void simple_transaction_set(struct file *file, size_t n)
699{
700 struct simple_transaction_argresp *ar = file->private_data;
701
702 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
703
704 /*
705 * The barrier ensures that ar->size will really remain zero until
706 * ar->data is ready for reading.
707 */
708 smp_mb();
709 ar->size = n;
710}
711
Linus Torvalds1da177e2005-04-16 15:20:36 -0700712char *simple_transaction_get(struct file *file, const char __user *buf, size_t size)
713{
714 struct simple_transaction_argresp *ar;
715 static DEFINE_SPINLOCK(simple_transaction_lock);
716
717 if (size > SIMPLE_TRANSACTION_LIMIT - 1)
718 return ERR_PTR(-EFBIG);
719
720 ar = (struct simple_transaction_argresp *)get_zeroed_page(GFP_KERNEL);
721 if (!ar)
722 return ERR_PTR(-ENOMEM);
723
724 spin_lock(&simple_transaction_lock);
725
726 /* only one write allowed per open */
727 if (file->private_data) {
728 spin_unlock(&simple_transaction_lock);
729 free_page((unsigned long)ar);
730 return ERR_PTR(-EBUSY);
731 }
732
733 file->private_data = ar;
734
735 spin_unlock(&simple_transaction_lock);
736
737 if (copy_from_user(ar->data, buf, size))
738 return ERR_PTR(-EFAULT);
739
740 return ar->data;
741}
742
743ssize_t simple_transaction_read(struct file *file, char __user *buf, size_t size, loff_t *pos)
744{
745 struct simple_transaction_argresp *ar = file->private_data;
746
747 if (!ar)
748 return 0;
749 return simple_read_from_buffer(buf, size, pos, ar->data, ar->size);
750}
751
752int simple_transaction_release(struct inode *inode, struct file *file)
753{
754 free_page((unsigned long)file->private_data);
755 return 0;
756}
757
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200758/* Simple attribute files */
759
760struct simple_attr {
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800761 int (*get)(void *, u64 *);
762 int (*set)(void *, u64);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200763 char get_buf[24]; /* enough to store a u64 and "\n\0" */
764 char set_buf[24];
765 void *data;
766 const char *fmt; /* format for read operation */
Ingo Molnar7cf34c72006-03-23 03:00:36 -0800767 struct mutex mutex; /* protects access to these buffers */
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200768};
769
770/* simple_attr_open is called by an actual attribute open file operation
771 * to set the attribute specific access operations. */
772int simple_attr_open(struct inode *inode, struct file *file,
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800773 int (*get)(void *, u64 *), int (*set)(void *, u64),
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200774 const char *fmt)
775{
776 struct simple_attr *attr;
777
778 attr = kmalloc(sizeof(*attr), GFP_KERNEL);
779 if (!attr)
780 return -ENOMEM;
781
782 attr->get = get;
783 attr->set = set;
Theodore Ts'o8e18e292006-09-27 01:50:46 -0700784 attr->data = inode->i_private;
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200785 attr->fmt = fmt;
Ingo Molnar7cf34c72006-03-23 03:00:36 -0800786 mutex_init(&attr->mutex);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200787
788 file->private_data = attr;
789
790 return nonseekable_open(inode, file);
791}
792
Christoph Hellwig74bedc42008-02-08 04:20:28 -0800793int simple_attr_release(struct inode *inode, struct file *file)
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200794{
795 kfree(file->private_data);
796 return 0;
797}
798
799/* read from the buffer that is filled with the get function */
800ssize_t simple_attr_read(struct file *file, char __user *buf,
801 size_t len, loff_t *ppos)
802{
803 struct simple_attr *attr;
804 size_t size;
805 ssize_t ret;
806
807 attr = file->private_data;
808
809 if (!attr->get)
810 return -EACCES;
811
Christoph Hellwig92613032008-02-08 04:20:27 -0800812 ret = mutex_lock_interruptible(&attr->mutex);
813 if (ret)
814 return ret;
815
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800816 if (*ppos) { /* continued read */
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200817 size = strlen(attr->get_buf);
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800818 } else { /* first read */
819 u64 val;
820 ret = attr->get(attr->data, &val);
821 if (ret)
822 goto out;
823
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200824 size = scnprintf(attr->get_buf, sizeof(attr->get_buf),
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800825 attr->fmt, (unsigned long long)val);
826 }
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200827
828 ret = simple_read_from_buffer(buf, len, ppos, attr->get_buf, size);
Christoph Hellwig8b88b092008-02-08 04:20:26 -0800829out:
Ingo Molnar7cf34c72006-03-23 03:00:36 -0800830 mutex_unlock(&attr->mutex);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200831 return ret;
832}
833
834/* interpret the buffer as a number to call the set function with */
835ssize_t simple_attr_write(struct file *file, const char __user *buf,
836 size_t len, loff_t *ppos)
837{
838 struct simple_attr *attr;
839 u64 val;
840 size_t size;
841 ssize_t ret;
842
843 attr = file->private_data;
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200844 if (!attr->set)
845 return -EACCES;
846
Christoph Hellwig92613032008-02-08 04:20:27 -0800847 ret = mutex_lock_interruptible(&attr->mutex);
848 if (ret)
849 return ret;
850
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200851 ret = -EFAULT;
852 size = min(sizeof(attr->set_buf) - 1, len);
853 if (copy_from_user(attr->set_buf, buf, size))
854 goto out;
855
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200856 attr->set_buf[size] = '\0';
857 val = simple_strtol(attr->set_buf, NULL, 0);
Wu Fengguang05cc0ce2009-09-18 13:06:03 -0700858 ret = attr->set(attr->data, val);
859 if (ret == 0)
860 ret = len; /* on success, claim we got the whole input */
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200861out:
Ingo Molnar7cf34c72006-03-23 03:00:36 -0800862 mutex_unlock(&attr->mutex);
Arnd Bergmannacaefc22005-05-18 14:40:59 +0200863 return ret;
864}
865
Christoph Hellwig25961102007-10-21 16:42:05 -0700866/**
867 * generic_fh_to_dentry - generic helper for the fh_to_dentry export operation
868 * @sb: filesystem to do the file handle conversion on
869 * @fid: file handle to convert
870 * @fh_len: length of the file handle in bytes
871 * @fh_type: type of file handle
872 * @get_inode: filesystem callback to retrieve inode
873 *
874 * This function decodes @fid as long as it has one of the well-known
875 * Linux filehandle types and calls @get_inode on it to retrieve the
876 * inode for the object specified in the file handle.
877 */
878struct dentry *generic_fh_to_dentry(struct super_block *sb, struct fid *fid,
879 int fh_len, int fh_type, struct inode *(*get_inode)
880 (struct super_block *sb, u64 ino, u32 gen))
881{
882 struct inode *inode = NULL;
883
884 if (fh_len < 2)
885 return NULL;
886
887 switch (fh_type) {
888 case FILEID_INO32_GEN:
889 case FILEID_INO32_GEN_PARENT:
890 inode = get_inode(sb, fid->i32.ino, fid->i32.gen);
891 break;
892 }
893
Christoph Hellwig4ea3ada2008-08-11 15:48:57 +0200894 return d_obtain_alias(inode);
Christoph Hellwig25961102007-10-21 16:42:05 -0700895}
896EXPORT_SYMBOL_GPL(generic_fh_to_dentry);
897
898/**
899 * generic_fh_to_dentry - generic helper for the fh_to_parent export operation
900 * @sb: filesystem to do the file handle conversion on
901 * @fid: file handle to convert
902 * @fh_len: length of the file handle in bytes
903 * @fh_type: type of file handle
904 * @get_inode: filesystem callback to retrieve inode
905 *
906 * This function decodes @fid as long as it has one of the well-known
907 * Linux filehandle types and calls @get_inode on it to retrieve the
908 * inode for the _parent_ object specified in the file handle if it
909 * is specified in the file handle, or NULL otherwise.
910 */
911struct dentry *generic_fh_to_parent(struct super_block *sb, struct fid *fid,
912 int fh_len, int fh_type, struct inode *(*get_inode)
913 (struct super_block *sb, u64 ino, u32 gen))
914{
915 struct inode *inode = NULL;
916
917 if (fh_len <= 2)
918 return NULL;
919
920 switch (fh_type) {
921 case FILEID_INO32_GEN_PARENT:
922 inode = get_inode(sb, fid->i32.parent_ino,
923 (fh_len > 3 ? fid->i32.parent_gen : 0));
924 break;
925 }
926
Christoph Hellwig4ea3ada2008-08-11 15:48:57 +0200927 return d_obtain_alias(inode);
Christoph Hellwig25961102007-10-21 16:42:05 -0700928}
929EXPORT_SYMBOL_GPL(generic_fh_to_parent);
930
Christoph Hellwig1b061d92010-05-26 17:53:41 +0200931/**
932 * generic_file_fsync - generic fsync implementation for simple filesystems
933 * @file: file to synchronize
934 * @datasync: only synchronize essential metadata if true
935 *
936 * This is a generic implementation of the fsync method for simple
937 * filesystems which track all non-inode metadata in the buffers list
938 * hanging off the address_space structure.
939 */
940int generic_file_fsync(struct file *file, int datasync)
Al Virod5aacad2009-06-07 14:56:44 -0400941{
942 struct writeback_control wbc = {
943 .sync_mode = WB_SYNC_ALL,
944 .nr_to_write = 0, /* metadata-only; caller takes care of data */
945 };
Christoph Hellwig7ea80852010-05-26 17:53:25 +0200946 struct inode *inode = file->f_mapping->host;
Al Virod5aacad2009-06-07 14:56:44 -0400947 int err;
948 int ret;
949
950 ret = sync_mapping_buffers(inode->i_mapping);
951 if (!(inode->i_state & I_DIRTY))
952 return ret;
953 if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))
954 return ret;
955
956 err = sync_inode(inode, &wbc);
957 if (ret == 0)
958 ret = err;
959 return ret;
960}
Christoph Hellwig1b061d92010-05-26 17:53:41 +0200961EXPORT_SYMBOL(generic_file_fsync);
962
963/*
964 * No-op implementation of ->fsync for in-memory filesystems.
965 */
966int noop_fsync(struct file *file, int datasync)
967{
968 return 0;
969}
Al Virod5aacad2009-06-07 14:56:44 -0400970
Linus Torvalds1da177e2005-04-16 15:20:36 -0700971EXPORT_SYMBOL(dcache_dir_close);
972EXPORT_SYMBOL(dcache_dir_lseek);
973EXPORT_SYMBOL(dcache_dir_open);
974EXPORT_SYMBOL(dcache_readdir);
975EXPORT_SYMBOL(generic_read_dir);
976EXPORT_SYMBOL(get_sb_pseudo);
Nick Pigginafddba42007-10-16 01:25:01 -0700977EXPORT_SYMBOL(simple_write_begin);
978EXPORT_SYMBOL(simple_write_end);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700979EXPORT_SYMBOL(simple_dir_inode_operations);
980EXPORT_SYMBOL(simple_dir_operations);
981EXPORT_SYMBOL(simple_empty);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700982EXPORT_SYMBOL(simple_fill_super);
983EXPORT_SYMBOL(simple_getattr);
984EXPORT_SYMBOL(simple_link);
985EXPORT_SYMBOL(simple_lookup);
986EXPORT_SYMBOL(simple_pin_fs);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700987EXPORT_SYMBOL(simple_readpage);
988EXPORT_SYMBOL(simple_release_fs);
989EXPORT_SYMBOL(simple_rename);
990EXPORT_SYMBOL(simple_rmdir);
991EXPORT_SYMBOL(simple_statfs);
Christoph Hellwig1b061d92010-05-26 17:53:41 +0200992EXPORT_SYMBOL(noop_fsync);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700993EXPORT_SYMBOL(simple_unlink);
994EXPORT_SYMBOL(simple_read_from_buffer);
Jiri Slaby6a727b42010-05-01 23:51:22 +0200995EXPORT_SYMBOL(simple_write_to_buffer);
Akinobu Mita93b07112008-06-05 22:46:21 -0700996EXPORT_SYMBOL(memory_read_from_buffer);
Ingo Molnar76791ab2009-03-25 16:48:35 +0100997EXPORT_SYMBOL(simple_transaction_set);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700998EXPORT_SYMBOL(simple_transaction_get);
999EXPORT_SYMBOL(simple_transaction_read);
1000EXPORT_SYMBOL(simple_transaction_release);
Arnd Bergmannacaefc22005-05-18 14:40:59 +02001001EXPORT_SYMBOL_GPL(simple_attr_open);
Christoph Hellwig74bedc42008-02-08 04:20:28 -08001002EXPORT_SYMBOL_GPL(simple_attr_release);
Arnd Bergmannacaefc22005-05-18 14:40:59 +02001003EXPORT_SYMBOL_GPL(simple_attr_read);
1004EXPORT_SYMBOL_GPL(simple_attr_write);