blob: 94cf97b901d7fb29f0a75b6afae88027ff3d4adb [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001
Pekka J Enberg5ea626a2005-09-09 13:10:19 -07002 Overview of the Linux Virtual File System
Linus Torvalds1da177e2005-04-16 15:20:36 -07003
Pekka J Enberg5ea626a2005-09-09 13:10:19 -07004 Original author: Richard Gooch <rgooch@atnf.csiro.au>
Linus Torvalds1da177e2005-04-16 15:20:36 -07005
Borislav Petkov0746aec2007-07-15 23:41:19 -07006 Last updated on June 24, 2007.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -07007
8 Copyright (C) 1999 Richard Gooch
9 Copyright (C) 2005 Pekka Enberg
10
11 This file is released under the GPLv2.
Linus Torvalds1da177e2005-04-16 15:20:36 -070012
13
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080014Introduction
15============
Linus Torvalds1da177e2005-04-16 15:20:36 -070016
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080017The Virtual File System (also known as the Virtual Filesystem Switch)
18is the software layer in the kernel that provides the filesystem
19interface to userspace programs. It also provides an abstraction
20within the kernel which allows different filesystem implementations to
21coexist.
22
23VFS system calls open(2), stat(2), read(2), write(2), chmod(2) and so
24on are called from a process context. Filesystem locking is described
25in the document Documentation/filesystems/Locking.
Linus Torvalds1da177e2005-04-16 15:20:36 -070026
27
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080028Directory Entry Cache (dcache)
29------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -070030
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080031The VFS implements the open(2), stat(2), chmod(2), and similar system
32calls. The pathname argument that is passed to them is used by the VFS
33to search through the directory entry cache (also known as the dentry
34cache or dcache). This provides a very fast look-up mechanism to
35translate a pathname (filename) into a specific dentry. Dentries live
36in RAM and are never saved to disc: they exist only for performance.
37
38The dentry cache is meant to be a view into your entire filespace. As
39most computers cannot fit all dentries in the RAM at the same time,
40some bits of the cache are missing. In order to resolve your pathname
41into a dentry, the VFS may have to resort to creating dentries along
42the way, and then loading the inode. This is done by looking up the
43inode.
Linus Torvalds1da177e2005-04-16 15:20:36 -070044
Pekka J Enberg5ea626a2005-09-09 13:10:19 -070045
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080046The Inode Object
47----------------
Linus Torvalds1da177e2005-04-16 15:20:36 -070048
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080049An individual dentry usually has a pointer to an inode. Inodes are
50filesystem objects such as regular files, directories, FIFOs and other
51beasts. They live either on the disc (for block device filesystems)
52or in the memory (for pseudo filesystems). Inodes that live on the
53disc are copied into the memory when required and changes to the inode
54are written back to disc. A single inode can be pointed to by multiple
55dentries (hard links, for example, do this).
Linus Torvalds1da177e2005-04-16 15:20:36 -070056
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080057To look up an inode requires that the VFS calls the lookup() method of
58the parent directory inode. This method is installed by the specific
59filesystem implementation that the inode lives in. Once the VFS has
60the required dentry (and hence the inode), we can do all those boring
61things like open(2) the file, or stat(2) it to peek at the inode
62data. The stat(2) operation is fairly simple: once the VFS has the
63dentry, it peeks at the inode data and passes some of it back to
64userspace.
Linus Torvalds1da177e2005-04-16 15:20:36 -070065
Linus Torvalds1da177e2005-04-16 15:20:36 -070066
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080067The File Object
68---------------
Linus Torvalds1da177e2005-04-16 15:20:36 -070069
70Opening a file requires another operation: allocation of a file
71structure (this is the kernel-side implementation of file
Pekka J Enberg5ea626a2005-09-09 13:10:19 -070072descriptors). The freshly allocated file structure is initialized with
Linus Torvalds1da177e2005-04-16 15:20:36 -070073a pointer to the dentry and a set of file operation member functions.
74These are taken from the inode data. The open() file method is then
Francis Galieguea33f3222010-04-23 00:08:02 +020075called so the specific filesystem implementation can do its work. You
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080076can see that this is another switch performed by the VFS. The file
77structure is placed into the file descriptor table for the process.
Linus Torvalds1da177e2005-04-16 15:20:36 -070078
79Reading, writing and closing files (and other assorted VFS operations)
80is done by using the userspace file descriptor to grab the appropriate
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080081file structure, and then calling the required file structure method to
82do whatever is required. For as long as the file is open, it keeps the
83dentry in use, which in turn means that the VFS inode is still in use.
Linus Torvalds1da177e2005-04-16 15:20:36 -070084
Pekka J Enberg5ea626a2005-09-09 13:10:19 -070085
86Registering and Mounting a Filesystem
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080087=====================================
Linus Torvalds1da177e2005-04-16 15:20:36 -070088
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080089To register and unregister a filesystem, use the following API
90functions:
Linus Torvalds1da177e2005-04-16 15:20:36 -070091
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080092 #include <linux/fs.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070093
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080094 extern int register_filesystem(struct file_system_type *);
95 extern int unregister_filesystem(struct file_system_type *);
Linus Torvalds1da177e2005-04-16 15:20:36 -070096
Pekka Enbergcc7d1f82005-11-07 01:01:08 -080097The passed struct file_system_type describes your filesystem. When a
98request is made to mount a device onto a directory in your filespace,
99the VFS will call the appropriate get_sb() method for the specific
100filesystem. The dentry for the mount point will then be updated to
101point to the root inode for the new filesystem.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700102
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800103You can see all filesystems that are registered to the kernel in the
104file /proc/filesystems.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105
106
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700107struct file_system_type
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800108-----------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -0700109
Borislav Petkov0746aec2007-07-15 23:41:19 -0700110This describes the filesystem. As of kernel 2.6.22, the following
Linus Torvalds1da177e2005-04-16 15:20:36 -0700111members are defined:
112
113struct file_system_type {
114 const char *name;
115 int fs_flags;
Jonathan Corbet5d8b2eb2006-07-10 04:44:07 -0700116 int (*get_sb) (struct file_system_type *, int,
117 const char *, void *, struct vfsmount *);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700118 void (*kill_sb) (struct super_block *);
119 struct module *owner;
120 struct file_system_type * next;
121 struct list_head fs_supers;
Borislav Petkov0746aec2007-07-15 23:41:19 -0700122 struct lock_class_key s_lock_key;
123 struct lock_class_key s_umount_key;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700124};
125
126 name: the name of the filesystem type, such as "ext2", "iso9660",
127 "msdos" and so on
128
129 fs_flags: various flags (i.e. FS_REQUIRES_DEV, FS_NO_DCACHE, etc.)
130
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700131 get_sb: the method to call when a new instance of this
Linus Torvalds1da177e2005-04-16 15:20:36 -0700132 filesystem should be mounted
133
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700134 kill_sb: the method to call when an instance of this filesystem
135 should be unmounted
Linus Torvalds1da177e2005-04-16 15:20:36 -0700136
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700137 owner: for internal VFS use: you should initialize this to THIS_MODULE in
138 most cases.
139
140 next: for internal VFS use: you should initialize this to NULL
141
Borislav Petkov0746aec2007-07-15 23:41:19 -0700142 s_lock_key, s_umount_key: lockdep-specific
143
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700144The get_sb() method has the following arguments:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700145
Matt LaPlanted9195882008-07-25 19:45:33 -0700146 struct file_system_type *fs_type: describes the filesystem, partly initialized
Borislav Petkov0746aec2007-07-15 23:41:19 -0700147 by the specific filesystem code
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700148
149 int flags: mount flags
150
151 const char *dev_name: the device name we are mounting.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700152
153 void *data: arbitrary mount options, usually comes as an ASCII
Miklos Szeredif84e3f52008-02-08 04:21:34 -0800154 string (see "Mount Options" section)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700155
Borislav Petkov0746aec2007-07-15 23:41:19 -0700156 struct vfsmount *mnt: a vfs-internal representation of a mount point
Linus Torvalds1da177e2005-04-16 15:20:36 -0700157
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700158The get_sb() method must determine if the block device specified
Borislav Petkov0746aec2007-07-15 23:41:19 -0700159in the dev_name and fs_type contains a filesystem of the type the method
160supports. If it succeeds in opening the named block device, it initializes a
161struct super_block descriptor for the filesystem contained by the block device.
162On failure it returns an error.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700163
164The most interesting member of the superblock structure that the
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700165get_sb() method fills in is the "s_op" field. This is a pointer to
Linus Torvalds1da177e2005-04-16 15:20:36 -0700166a "struct super_operations" which describes the next level of the
167filesystem implementation.
168
Jim Cromiee3e1bfe2006-01-03 13:35:41 +0100169Usually, a filesystem uses one of the generic get_sb() implementations
170and provides a fill_super() method instead. The generic methods are:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700171
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700172 get_sb_bdev: mount a filesystem residing on a block device
173
174 get_sb_nodev: mount a filesystem that is not backed by a device
175
176 get_sb_single: mount a filesystem which shares the instance between
177 all mounts
178
179A fill_super() method implementation has the following arguments:
180
181 struct super_block *sb: the superblock structure. The method fill_super()
182 must initialize this properly.
183
184 void *data: arbitrary mount options, usually comes as an ASCII
Miklos Szeredif84e3f52008-02-08 04:21:34 -0800185 string (see "Mount Options" section)
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700186
187 int silent: whether or not to be silent on error
188
189
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800190The Superblock Object
191=====================
192
193A superblock object represents a mounted filesystem.
194
195
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700196struct super_operations
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800197-----------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -0700198
199This describes how the VFS can manipulate the superblock of your
Borislav Petkov422b14c2007-07-15 23:41:43 -0700200filesystem. As of kernel 2.6.22, the following members are defined:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700201
202struct super_operations {
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700203 struct inode *(*alloc_inode)(struct super_block *sb);
204 void (*destroy_inode)(struct inode *);
205
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700206 void (*dirty_inode) (struct inode *);
207 int (*write_inode) (struct inode *, int);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700208 void (*drop_inode) (struct inode *);
209 void (*delete_inode) (struct inode *);
210 void (*put_super) (struct super_block *);
211 void (*write_super) (struct super_block *);
212 int (*sync_fs)(struct super_block *sb, int wait);
Takashi Satoc4be0c12009-01-09 16:40:58 -0800213 int (*freeze_fs) (struct super_block *);
214 int (*unfreeze_fs) (struct super_block *);
David Howells726c3342006-06-23 02:02:58 -0700215 int (*statfs) (struct dentry *, struct kstatfs *);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700216 int (*remount_fs) (struct super_block *, int *, char *);
217 void (*clear_inode) (struct inode *);
218 void (*umount_begin) (struct super_block *);
219
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700220 int (*show_options)(struct seq_file *, struct vfsmount *);
221
222 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
223 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700224};
225
226All methods are called without any locks being held, unless otherwise
227noted. This means that most methods can block safely. All methods are
228only called from a process context (i.e. not from an interrupt handler
229or bottom half).
230
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700231 alloc_inode: this method is called by inode_alloc() to allocate memory
NeilBrown341546f2006-03-25 03:07:56 -0800232 for struct inode and initialize it. If this function is not
233 defined, a simple 'struct inode' is allocated. Normally
234 alloc_inode will be used to allocate a larger structure which
235 contains a 'struct inode' embedded within it.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700236
237 destroy_inode: this method is called by destroy_inode() to release
NeilBrown341546f2006-03-25 03:07:56 -0800238 resources allocated for struct inode. It is only required if
239 ->alloc_inode was defined and simply undoes anything done by
240 ->alloc_inode.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700241
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700242 dirty_inode: this method is called by the VFS to mark an inode dirty.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700243
244 write_inode: this method is called when the VFS needs to write an
245 inode to disc. The second parameter indicates whether the write
246 should be synchronous or not, not all filesystems check this flag.
247
Linus Torvalds1da177e2005-04-16 15:20:36 -0700248 drop_inode: called when the last access to the inode is dropped,
249 with the inode_lock spinlock held.
250
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700251 This method should be either NULL (normal UNIX filesystem
Linus Torvalds1da177e2005-04-16 15:20:36 -0700252 semantics) or "generic_delete_inode" (for filesystems that do not
253 want to cache inodes - causing "delete_inode" to always be
254 called regardless of the value of i_nlink)
255
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700256 The "generic_delete_inode()" behavior is equivalent to the
Linus Torvalds1da177e2005-04-16 15:20:36 -0700257 old practice of using "force_delete" in the put_inode() case,
258 but does not have the races that the "force_delete()" approach
259 had.
260
261 delete_inode: called when the VFS wants to delete an inode
262
Linus Torvalds1da177e2005-04-16 15:20:36 -0700263 put_super: called when the VFS wishes to free the superblock
264 (i.e. unmount). This is called with the superblock lock held
265
266 write_super: called when the VFS superblock needs to be written to
267 disc. This method is optional
268
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700269 sync_fs: called when VFS is writing out all dirty data associated with
270 a superblock. The second parameter indicates whether the method
271 should wait until the write out has been completed. Optional.
272
Takashi Satoc4be0c12009-01-09 16:40:58 -0800273 freeze_fs: called when VFS is locking a filesystem and
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800274 forcing it into a consistent state. This method is currently
275 used by the Logical Volume Manager (LVM).
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700276
Takashi Satoc4be0c12009-01-09 16:40:58 -0800277 unfreeze_fs: called when VFS is unlocking a filesystem and making it writable
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700278 again.
279
Adrian McMenamin66672fe2009-04-20 18:38:28 -0700280 statfs: called when the VFS needs to get filesystem statistics.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700281
282 remount_fs: called when the filesystem is remounted. This is called
283 with the kernel lock held
284
285 clear_inode: called then the VFS clears the inode. Optional
286
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700287 umount_begin: called when the VFS is unmounting a filesystem.
288
Miklos Szeredif84e3f52008-02-08 04:21:34 -0800289 show_options: called by the VFS to show mount options for
290 /proc/<pid>/mounts. (see "Mount Options" section)
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700291
292 quota_read: called by the VFS to read from filesystem quota file.
293
294 quota_write: called by the VFS to write to filesystem quota file.
295
David Howells12debc42008-02-07 00:15:52 -0800296Whoever sets up the inode is responsible for filling in the "i_op" field. This
297is a pointer to a "struct inode_operations" which describes the methods that
298can be performed on individual inodes.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700299
300
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800301The Inode Object
302================
303
304An inode object represents an object within the filesystem.
305
306
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700307struct inode_operations
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800308-----------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -0700309
310This describes how the VFS can manipulate an inode in your
Borislav Petkov422b14c2007-07-15 23:41:43 -0700311filesystem. As of kernel 2.6.22, the following members are defined:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700312
313struct inode_operations {
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700314 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
315 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700316 int (*link) (struct dentry *,struct inode *,struct dentry *);
317 int (*unlink) (struct inode *,struct dentry *);
318 int (*symlink) (struct inode *,struct dentry *,const char *);
319 int (*mkdir) (struct inode *,struct dentry *,int);
320 int (*rmdir) (struct inode *,struct dentry *);
321 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
322 int (*rename) (struct inode *, struct dentry *,
323 struct inode *, struct dentry *);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700324 int (*readlink) (struct dentry *, char __user *,int);
325 void * (*follow_link) (struct dentry *, struct nameidata *);
326 void (*put_link) (struct dentry *, struct nameidata *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700327 void (*truncate) (struct inode *);
Nick Pigginb74c79e2011-01-07 17:49:58 +1100328 int (*permission) (struct inode *, int, unsigned int);
329 int (*check_acl)(struct inode *, int, unsigned int);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700330 int (*setattr) (struct dentry *, struct iattr *);
331 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
332 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
333 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
334 ssize_t (*listxattr) (struct dentry *, char *, size_t);
335 int (*removexattr) (struct dentry *, const char *);
Borislav Petkov422b14c2007-07-15 23:41:43 -0700336 void (*truncate_range)(struct inode *, loff_t, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700337};
338
339Again, all methods are called without any locks being held, unless
340otherwise noted.
341
Linus Torvalds1da177e2005-04-16 15:20:36 -0700342 create: called by the open(2) and creat(2) system calls. Only
343 required if you want to support regular files. The dentry you
344 get should not have an inode (i.e. it should be a negative
345 dentry). Here you will probably call d_instantiate() with the
346 dentry and the newly created inode
347
348 lookup: called when the VFS needs to look up an inode in a parent
349 directory. The name to look for is found in the dentry. This
350 method must call d_add() to insert the found inode into the
351 dentry. The "i_count" field in the inode structure should be
352 incremented. If the named inode does not exist a NULL inode
353 should be inserted into the dentry (this is called a negative
354 dentry). Returning an error code from this routine must only
355 be done on a real error, otherwise creating inodes with system
356 calls like create(2), mknod(2), mkdir(2) and so on will fail.
357 If you wish to overload the dentry methods then you should
358 initialise the "d_dop" field in the dentry; this is a pointer
359 to a struct "dentry_operations".
360 This method is called with the directory inode semaphore held
361
362 link: called by the link(2) system call. Only required if you want
363 to support hard links. You will probably need to call
364 d_instantiate() just as you would in the create() method
365
366 unlink: called by the unlink(2) system call. Only required if you
367 want to support deleting inodes
368
369 symlink: called by the symlink(2) system call. Only required if you
370 want to support symlinks. You will probably need to call
371 d_instantiate() just as you would in the create() method
372
373 mkdir: called by the mkdir(2) system call. Only required if you want
374 to support creating subdirectories. You will probably need to
375 call d_instantiate() just as you would in the create() method
376
377 rmdir: called by the rmdir(2) system call. Only required if you want
378 to support deleting subdirectories
379
380 mknod: called by the mknod(2) system call to create a device (char,
381 block) inode or a named pipe (FIFO) or socket. Only required
382 if you want to support creating these types of inodes. You
383 will probably need to call d_instantiate() just as you would
384 in the create() method
385
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800386 rename: called by the rename(2) system call to rename the object to
387 have the parent and name given by the second inode and dentry.
388
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389 readlink: called by the readlink(2) system call. Only required if
390 you want to support reading symbolic links
391
392 follow_link: called by the VFS to follow a symbolic link to the
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700393 inode it points to. Only required if you want to support
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800394 symbolic links. This method returns a void pointer cookie
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700395 that is passed to put_link().
396
397 put_link: called by the VFS to release resources allocated by
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800398 follow_link(). The cookie returned by follow_link() is passed
Paolo Ornati670e9f32006-10-03 22:57:56 +0200399 to this method as the last parameter. It is used by
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800400 filesystems such as NFS where page cache is not stable
401 (i.e. page that was installed when the symbolic link walk
402 started might not be in the page cache at the end of the
403 walk).
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700404
npiggin@suse.de7bb46a62010-05-27 01:05:33 +1000405 truncate: Deprecated. This will not be called if ->setsize is defined.
406 Called by the VFS to change the size of a file. The
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800407 i_size field of the inode is set to the desired size by the
408 VFS before this method is called. This method is called by
409 the truncate(2) system call and related functionality.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700410
npiggin@suse.de7bb46a62010-05-27 01:05:33 +1000411 Note: ->truncate and vmtruncate are deprecated. Do not add new
412 instances/calls of these. Filesystems should be converted to do their
413 truncate sequence via ->setattr().
414
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700415 permission: called by the VFS to check for access rights on a POSIX-like
416 filesystem.
417
Nick Piggina82416d2011-01-14 02:26:53 +0000418 May be called in rcu-walk mode (flags & IPERM_FLAG_RCU). If in rcu-walk
419 mode, the filesystem must check the permission without blocking or
Nick Pigginb74c79e2011-01-07 17:49:58 +1100420 storing to the inode.
421
422 If a situation is encountered that rcu-walk cannot handle, return
423 -ECHILD and it will be called again in ref-walk mode.
424
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800425 setattr: called by the VFS to set attributes for a file. This method
426 is called by chmod(2) and related system calls.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700427
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800428 getattr: called by the VFS to get attributes of a file. This method
429 is called by stat(2) and related system calls.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700430
431 setxattr: called by the VFS to set an extended attribute for a file.
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800432 Extended attribute is a name:value pair associated with an
433 inode. This method is called by setxattr(2) system call.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700434
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800435 getxattr: called by the VFS to retrieve the value of an extended
436 attribute name. This method is called by getxattr(2) function
437 call.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700438
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800439 listxattr: called by the VFS to list all extended attributes for a
440 given file. This method is called by listxattr(2) system call.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700441
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800442 removexattr: called by the VFS to remove an extended attribute from
443 a file. This method is called by removexattr(2) system call.
444
Borislav Petkov422b14c2007-07-15 23:41:43 -0700445 truncate_range: a method provided by the underlying filesystem to truncate a
446 range of blocks , i.e. punch a hole somewhere in a file.
447
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800448
449The Address Space Object
450========================
451
NeilBrown341546f2006-03-25 03:07:56 -0800452The address space object is used to group and manage pages in the page
453cache. It can be used to keep track of the pages in a file (or
454anything else) and also track the mapping of sections of the file into
455process address spaces.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700456
NeilBrown341546f2006-03-25 03:07:56 -0800457There are a number of distinct yet related services that an
458address-space can provide. These include communicating memory
459pressure, page lookup by address, and keeping track of pages tagged as
460Dirty or Writeback.
461
NeilBrowna9e102b2006-03-25 03:08:29 -0800462The first can be used independently to the others. The VM can try to
NeilBrown341546f2006-03-25 03:07:56 -0800463either write dirty pages in order to clean them, or release clean
464pages in order to reuse them. To do this it can call the ->writepage
465method on dirty pages, and ->releasepage on clean pages with
466PagePrivate set. Clean pages without PagePrivate and with no external
467references will be released without notice being given to the
468address_space.
469
NeilBrowna9e102b2006-03-25 03:08:29 -0800470To achieve this functionality, pages need to be placed on an LRU with
NeilBrown341546f2006-03-25 03:07:56 -0800471lru_cache_add and mark_page_active needs to be called whenever the
472page is used.
473
474Pages are normally kept in a radix tree index by ->index. This tree
475maintains information about the PG_Dirty and PG_Writeback status of
476each page, so that pages with either of these flags can be found
477quickly.
478
479The Dirty tag is primarily used by mpage_writepages - the default
480->writepages method. It uses the tag to find dirty pages to call
481->writepage on. If mpage_writepages is not used (i.e. the address
NeilBrowna9e102b2006-03-25 03:08:29 -0800482provides its own ->writepages) , the PAGECACHE_TAG_DIRTY tag is
NeilBrown341546f2006-03-25 03:07:56 -0800483almost unused. write_inode_now and sync_inode do use it (through
484__sync_single_inode) to check if ->writepages has been successful in
485writing out the whole address_space.
486
487The Writeback tag is used by filemap*wait* and sync_page* functions,
Christoph Hellwig94004ed2009-09-30 22:16:33 +0200488via filemap_fdatawait_range, to wait for all writeback to
NeilBrown341546f2006-03-25 03:07:56 -0800489complete. While waiting ->sync_page (if defined) will be called on
NeilBrowna9e102b2006-03-25 03:08:29 -0800490each page that is found to require writeback.
NeilBrown341546f2006-03-25 03:07:56 -0800491
492An address_space handler may attach extra information to a page,
493typically using the 'private' field in the 'struct page'. If such
494information is attached, the PG_Private flag should be set. This will
NeilBrowna9e102b2006-03-25 03:08:29 -0800495cause various VM routines to make extra calls into the address_space
NeilBrown341546f2006-03-25 03:07:56 -0800496handler to deal with that data.
497
498An address space acts as an intermediate between storage and
499application. Data is read into the address space a whole page at a
500time, and provided to the application either by copying of the page,
501or by memory-mapping the page.
502Data is written into the address space by the application, and then
503written-back to storage typically in whole pages, however the
NeilBrowna9e102b2006-03-25 03:08:29 -0800504address_space has finer control of write sizes.
NeilBrown341546f2006-03-25 03:07:56 -0800505
506The read process essentially only requires 'readpage'. The write
Nick Piggin4e02ed42008-10-29 14:00:55 -0700507process is more complicated and uses write_begin/write_end or
NeilBrown341546f2006-03-25 03:07:56 -0800508set_page_dirty to write data into the address_space, and writepage,
509sync_page, and writepages to writeback data to storage.
510
511Adding and removing pages to/from an address_space is protected by the
512inode's i_mutex.
513
514When data is written to a page, the PG_Dirty flag should be set. It
515typically remains set until writepage asks for it to be written. This
516should clear PG_Dirty and set PG_Writeback. It can be actually
517written at any point after PG_Dirty is clear. Once it is known to be
518safe, PG_Writeback is cleared.
519
520Writeback makes use of a writeback_control structure...
Linus Torvalds1da177e2005-04-16 15:20:36 -0700521
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700522struct address_space_operations
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800523-------------------------------
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700524
525This describes how the VFS can manipulate mapping of a file to page cache in
Borislav Petkov422b14c2007-07-15 23:41:43 -0700526your filesystem. As of kernel 2.6.22, the following members are defined:
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700527
528struct address_space_operations {
529 int (*writepage)(struct page *page, struct writeback_control *wbc);
530 int (*readpage)(struct file *, struct page *);
531 int (*sync_page)(struct page *);
532 int (*writepages)(struct address_space *, struct writeback_control *);
533 int (*set_page_dirty)(struct page *page);
534 int (*readpages)(struct file *filp, struct address_space *mapping,
535 struct list_head *pages, unsigned nr_pages);
Nick Pigginafddba42007-10-16 01:25:01 -0700536 int (*write_begin)(struct file *, struct address_space *mapping,
537 loff_t pos, unsigned len, unsigned flags,
538 struct page **pagep, void **fsdata);
539 int (*write_end)(struct file *, struct address_space *mapping,
540 loff_t pos, unsigned len, unsigned copied,
541 struct page *page, void *fsdata);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700542 sector_t (*bmap)(struct address_space *, sector_t);
543 int (*invalidatepage) (struct page *, unsigned long);
544 int (*releasepage) (struct page *, int);
Linus Torvalds6072d132010-12-01 13:35:19 -0500545 void (*freepage)(struct page *);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700546 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
547 loff_t offset, unsigned long nr_segs);
548 struct page* (*get_xip_page)(struct address_space *, sector_t,
549 int);
NeilBrown341546f2006-03-25 03:07:56 -0800550 /* migrate the contents of a page to the specified target */
551 int (*migratepage) (struct page *, struct page *);
Borislav Petkov422b14c2007-07-15 23:41:43 -0700552 int (*launder_page) (struct page *);
Andi Kleen25718732009-09-16 11:50:13 +0200553 int (*error_remove_page) (struct mapping *mapping, struct page *page);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700554};
555
NeilBrown341546f2006-03-25 03:07:56 -0800556 writepage: called by the VM to write a dirty page to backing store.
NeilBrowna9e102b2006-03-25 03:08:29 -0800557 This may happen for data integrity reasons (i.e. 'sync'), or
NeilBrown341546f2006-03-25 03:07:56 -0800558 to free up memory (flush). The difference can be seen in
559 wbc->sync_mode.
560 The PG_Dirty flag has been cleared and PageLocked is true.
561 writepage should start writeout, should set PG_Writeback,
562 and should make sure the page is unlocked, either synchronously
563 or asynchronously when the write operation completes.
564
565 If wbc->sync_mode is WB_SYNC_NONE, ->writepage doesn't have to
NeilBrowna9e102b2006-03-25 03:08:29 -0800566 try too hard if there are problems, and may choose to write out
567 other pages from the mapping if that is easier (e.g. due to
568 internal dependencies). If it chooses not to start writeout, it
569 should return AOP_WRITEPAGE_ACTIVATE so that the VM will not keep
NeilBrown341546f2006-03-25 03:07:56 -0800570 calling ->writepage on that page.
571
572 See the file "Locking" for more details.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700573
574 readpage: called by the VM to read a page from backing store.
NeilBrown341546f2006-03-25 03:07:56 -0800575 The page will be Locked when readpage is called, and should be
576 unlocked and marked uptodate once the read completes.
577 If ->readpage discovers that it needs to unlock the page for
578 some reason, it can do so, and then return AOP_TRUNCATED_PAGE.
NeilBrowna9e102b2006-03-25 03:08:29 -0800579 In this case, the page will be relocated, relocked and if
NeilBrown341546f2006-03-25 03:07:56 -0800580 that all succeeds, ->readpage will be called again.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700581
582 sync_page: called by the VM to notify the backing store to perform all
583 queued I/O operations for a page. I/O operations for other pages
584 associated with this address_space object may also be performed.
585
NeilBrown341546f2006-03-25 03:07:56 -0800586 This function is optional and is called only for pages with
587 PG_Writeback set while waiting for the writeback to complete.
588
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700589 writepages: called by the VM to write out pages associated with the
NeilBrowna9e102b2006-03-25 03:08:29 -0800590 address_space object. If wbc->sync_mode is WBC_SYNC_ALL, then
591 the writeback_control will specify a range of pages that must be
592 written out. If it is WBC_SYNC_NONE, then a nr_to_write is given
NeilBrown341546f2006-03-25 03:07:56 -0800593 and that many pages should be written if possible.
594 If no ->writepages is given, then mpage_writepages is used
NeilBrowna9e102b2006-03-25 03:08:29 -0800595 instead. This will choose pages from the address space that are
NeilBrown341546f2006-03-25 03:07:56 -0800596 tagged as DIRTY and will pass them to ->writepage.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700597
598 set_page_dirty: called by the VM to set a page dirty.
NeilBrown341546f2006-03-25 03:07:56 -0800599 This is particularly needed if an address space attaches
600 private data to a page, and that data needs to be updated when
601 a page is dirtied. This is called, for example, when a memory
602 mapped page gets modified.
603 If defined, it should set the PageDirty flag, and the
604 PAGECACHE_TAG_DIRTY tag in the radix tree.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700605
606 readpages: called by the VM to read pages associated with the address_space
NeilBrown341546f2006-03-25 03:07:56 -0800607 object. This is essentially just a vector version of
608 readpage. Instead of just one page, several pages are
609 requested.
NeilBrowna9e102b2006-03-25 03:08:29 -0800610 readpages is only used for read-ahead, so read errors are
NeilBrown341546f2006-03-25 03:07:56 -0800611 ignored. If anything goes wrong, feel free to give up.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700612
Nick Piggin4e02ed42008-10-29 14:00:55 -0700613 write_begin:
Nick Pigginafddba42007-10-16 01:25:01 -0700614 Called by the generic buffered write code to ask the filesystem to
615 prepare to write len bytes at the given offset in the file. The
616 address_space should check that the write will be able to complete,
617 by allocating space if necessary and doing any other internal
618 housekeeping. If the write will update parts of any basic-blocks on
619 storage, then those blocks should be pre-read (if they haven't been
620 read already) so that the updated blocks can be written out properly.
621
622 The filesystem must return the locked pagecache page for the specified
623 offset, in *pagep, for the caller to write into.
624
Nick Piggin4e02ed42008-10-29 14:00:55 -0700625 It must be able to cope with short writes (where the length passed to
626 write_begin is greater than the number of bytes copied into the page).
627
Nick Pigginafddba42007-10-16 01:25:01 -0700628 flags is a field for AOP_FLAG_xxx flags, described in
629 include/linux/fs.h.
630
631 A void * may be returned in fsdata, which then gets passed into
632 write_end.
633
634 Returns 0 on success; < 0 on failure (which is the error code), in
635 which case write_end is not called.
636
637 write_end: After a successful write_begin, and data copy, write_end must
638 be called. len is the original len passed to write_begin, and copied
639 is the amount that was able to be copied (copied == len is always true
640 if write_begin was called with the AOP_FLAG_UNINTERRUPTIBLE flag).
641
642 The filesystem must take care of unlocking the page and releasing it
643 refcount, and updating i_size.
644
645 Returns < 0 on failure, otherwise the number of bytes (<= 'copied')
646 that were able to be copied into pagecache.
647
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700648 bmap: called by the VFS to map a logical block offset within object to
NeilBrowna9e102b2006-03-25 03:08:29 -0800649 physical block number. This method is used by the FIBMAP
NeilBrown341546f2006-03-25 03:07:56 -0800650 ioctl and for working with swap-files. To be able to swap to
NeilBrowna9e102b2006-03-25 03:08:29 -0800651 a file, the file must have a stable mapping to a block
NeilBrown341546f2006-03-25 03:07:56 -0800652 device. The swap system does not go through the filesystem
653 but instead uses bmap to find out where the blocks in the file
654 are and uses those addresses directly.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700655
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700656
NeilBrown341546f2006-03-25 03:07:56 -0800657 invalidatepage: If a page has PagePrivate set, then invalidatepage
658 will be called when part or all of the page is to be removed
NeilBrowna9e102b2006-03-25 03:08:29 -0800659 from the address space. This generally corresponds to either a
NeilBrown341546f2006-03-25 03:07:56 -0800660 truncation or a complete invalidation of the address space
661 (in the latter case 'offset' will always be 0).
662 Any private data associated with the page should be updated
663 to reflect this truncation. If offset is 0, then
664 the private data should be released, because the page
665 must be able to be completely discarded. This may be done by
666 calling the ->releasepage function, but in this case the
667 release MUST succeed.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700668
NeilBrown341546f2006-03-25 03:07:56 -0800669 releasepage: releasepage is called on PagePrivate pages to indicate
670 that the page should be freed if possible. ->releasepage
671 should remove any private data from the page and clear the
Andrew Morton4fe65ca2010-12-02 14:31:19 -0800672 PagePrivate flag. If releasepage() fails for some reason, it must
673 indicate failure with a 0 return value.
674 releasepage() is used in two distinct though related cases. The
675 first is when the VM finds a clean page with no active users and
NeilBrown341546f2006-03-25 03:07:56 -0800676 wants to make it a free page. If ->releasepage succeeds, the
677 page will be removed from the address_space and become free.
678
Shaun Zinckbc5b1d52007-10-20 02:35:36 +0200679 The second case is when a request has been made to invalidate
NeilBrown341546f2006-03-25 03:07:56 -0800680 some or all pages in an address_space. This can happen
681 through the fadvice(POSIX_FADV_DONTNEED) system call or by the
682 filesystem explicitly requesting it as nfs and 9fs do (when
683 they believe the cache may be out of date with storage) by
684 calling invalidate_inode_pages2().
685 If the filesystem makes such a call, and needs to be certain
NeilBrowna9e102b2006-03-25 03:08:29 -0800686 that all pages are invalidated, then its releasepage will
NeilBrown341546f2006-03-25 03:07:56 -0800687 need to ensure this. Possibly it can clear the PageUptodate
688 bit if it cannot free private data yet.
689
Linus Torvalds6072d132010-12-01 13:35:19 -0500690 freepage: freepage is called once the page is no longer visible in
691 the page cache in order to allow the cleanup of any private
692 data. Since it may be called by the memory reclaimer, it
693 should not assume that the original address_space mapping still
694 exists, and it should not block.
695
NeilBrown341546f2006-03-25 03:07:56 -0800696 direct_IO: called by the generic read/write routines to perform
697 direct_IO - that is IO requests which bypass the page cache
NeilBrowna9e102b2006-03-25 03:08:29 -0800698 and transfer data directly between the storage and the
NeilBrown341546f2006-03-25 03:07:56 -0800699 application's address space.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700700
701 get_xip_page: called by the VM to translate a block number to a page.
702 The page is valid until the corresponding filesystem is unmounted.
703 Filesystems that want to use execute-in-place (XIP) need to implement
704 it. An example implementation can be found in fs/ext2/xip.c.
705
NeilBrown341546f2006-03-25 03:07:56 -0800706 migrate_page: This is used to compact the physical memory usage.
707 If the VM wants to relocate a page (maybe off a memory card
708 that is signalling imminent failure) it will pass a new page
709 and an old page to this function. migrate_page should
710 transfer any private data across and update any references
711 that it has to the page.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700712
Borislav Petkov422b14c2007-07-15 23:41:43 -0700713 launder_page: Called before freeing a page - it writes back the dirty page. To
714 prevent redirtying the page, it is kept locked during the whole
715 operation.
716
Andi Kleen25718732009-09-16 11:50:13 +0200717 error_remove_page: normally set to generic_error_remove_page if truncation
718 is ok for this address space. Used for memory failure handling.
719 Setting this implies you deal with pages going away under you,
720 unless you have them locked or reference counts increased.
721
722
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800723The File Object
724===============
725
726A file object represents a file opened by a process.
727
728
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700729struct file_operations
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800730----------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -0700731
732This describes how the VFS can manipulate an open file. As of kernel
Borislav Petkov422b14c2007-07-15 23:41:43 -07007332.6.22, the following members are defined:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700734
735struct file_operations {
Borislav Petkov422b14c2007-07-15 23:41:43 -0700736 struct module *owner;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700737 loff_t (*llseek) (struct file *, loff_t, int);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700738 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700739 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
Badari Pulavarty027445c2006-09-30 23:28:46 -0700740 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
741 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700742 int (*readdir) (struct file *, void *, filldir_t);
743 unsigned int (*poll) (struct file *, struct poll_table_struct *);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700744 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
745 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700746 int (*mmap) (struct file *, struct vm_area_struct *);
747 int (*open) (struct inode *, struct file *);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700748 int (*flush) (struct file *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700749 int (*release) (struct inode *, struct file *);
Christoph Hellwig7ea80852010-05-26 17:53:25 +0200750 int (*fsync) (struct file *, int datasync);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700751 int (*aio_fsync) (struct kiocb *, int datasync);
752 int (*fasync) (int, struct file *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700753 int (*lock) (struct file *, int, struct file_lock *);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700754 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
755 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
756 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
757 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
758 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
759 int (*check_flags)(int);
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700760 int (*flock) (struct file *, int, struct file_lock *);
Borislav Petkov422b14c2007-07-15 23:41:43 -0700761 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, size_t, unsigned int);
762 ssize_t (*splice_read)(struct file *, struct pipe_inode_info *, size_t, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700763};
764
765Again, all methods are called without any locks being held, unless
766otherwise noted.
767
768 llseek: called when the VFS needs to move the file position index
769
770 read: called by read(2) and related system calls
771
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700772 aio_read: called by io_submit(2) and other asynchronous I/O operations
773
Linus Torvalds1da177e2005-04-16 15:20:36 -0700774 write: called by write(2) and related system calls
775
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700776 aio_write: called by io_submit(2) and other asynchronous I/O operations
777
Linus Torvalds1da177e2005-04-16 15:20:36 -0700778 readdir: called when the VFS needs to read the directory contents
779
780 poll: called by the VFS when a process wants to check if there is
781 activity on this file and (optionally) go to sleep until there
782 is activity. Called by the select(2) and poll(2) system calls
783
Arnd Bergmannb19dd422010-07-04 00:15:10 +0200784 unlocked_ioctl: called by the ioctl(2) system call.
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700785
786 compat_ioctl: called by the ioctl(2) system call when 32 bit system calls
787 are used on 64 bit kernels.
788
Linus Torvalds1da177e2005-04-16 15:20:36 -0700789 mmap: called by the mmap(2) system call
790
791 open: called by the VFS when an inode should be opened. When the VFS
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700792 opens a file, it creates a new "struct file". It then calls the
793 open method for the newly allocated file structure. You might
794 think that the open method really belongs in
795 "struct inode_operations", and you may be right. I think it's
796 done the way it is because it makes filesystems simpler to
797 implement. The open() method is a good place to initialize the
798 "private_data" member in the file structure if you want to point
799 to a device structure
800
801 flush: called by the close(2) system call to flush a file
Linus Torvalds1da177e2005-04-16 15:20:36 -0700802
803 release: called when the last reference to an open file is closed
804
805 fsync: called by the fsync(2) system call
806
807 fasync: called by the fcntl(2) system call when asynchronous
808 (non-blocking) mode is enabled for a file
809
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700810 lock: called by the fcntl(2) system call for F_GETLK, F_SETLK, and F_SETLKW
811 commands
812
813 readv: called by the readv(2) system call
814
815 writev: called by the writev(2) system call
816
817 sendfile: called by the sendfile(2) system call
818
819 get_unmapped_area: called by the mmap(2) system call
820
821 check_flags: called by the fcntl(2) system call for F_SETFL command
822
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700823 flock: called by the flock(2) system call
824
Pekka J Enbergd1195c52006-04-11 14:21:59 +0200825 splice_write: called by the VFS to splice data from a pipe to a file. This
826 method is used by the splice(2) system call
827
828 splice_read: called by the VFS to splice data from file to a pipe. This
829 method is used by the splice(2) system call
830
Linus Torvalds1da177e2005-04-16 15:20:36 -0700831Note that the file operations are implemented by the specific
832filesystem in which the inode resides. When opening a device node
833(character or block special) most filesystems will call special
834support routines in the VFS which will locate the required device
835driver information. These support routines replace the filesystem file
836operations with those for the device driver, and then proceed to call
837the new open() method for the file. This is how opening a device file
838in the filesystem eventually ends up calling the device driver open()
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700839method.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700840
841
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700842Directory Entry Cache (dcache)
843==============================
844
Linus Torvalds1da177e2005-04-16 15:20:36 -0700845
846struct dentry_operations
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700847------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -0700848
849This describes how a filesystem can overload the standard dentry
850operations. Dentries and the dcache are the domain of the VFS and the
851individual filesystem implementations. Device drivers have no business
852here. These methods may be set to NULL, as they are either optional or
Eric Dumazetc23fbb62007-05-08 00:26:18 -0700853the VFS uses a default. As of kernel 2.6.22, the following members are
Linus Torvalds1da177e2005-04-16 15:20:36 -0700854defined:
855
856struct dentry_operations {
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700857 int (*d_revalidate)(struct dentry *, struct nameidata *);
Nick Pigginb1e6a012011-01-07 17:49:28 +1100858 int (*d_hash)(const struct dentry *, const struct inode *,
859 struct qstr *);
Nick Piggin621e1552011-01-07 17:49:27 +1100860 int (*d_compare)(const struct dentry *, const struct inode *,
861 const struct dentry *, const struct inode *,
862 unsigned int, const char *, const struct qstr *);
Nick Pigginfe15ce42011-01-07 17:49:23 +1100863 int (*d_delete)(const struct dentry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700864 void (*d_release)(struct dentry *);
865 void (*d_iput)(struct dentry *, struct inode *);
Eric Dumazetc23fbb62007-05-08 00:26:18 -0700866 char *(*d_dname)(struct dentry *, char *, int);
David Howells9875cf82011-01-14 18:45:21 +0000867 struct vfsmount *(*d_automount)(struct path *);
David Howellsab909112011-01-14 18:46:51 +0000868 int (*d_manage)(struct dentry *, bool, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700869};
870
871 d_revalidate: called when the VFS needs to revalidate a dentry. This
872 is called whenever a name look-up finds a dentry in the
873 dcache. Most filesystems leave this as NULL, because all their
874 dentries in the dcache are valid
875
Nick Piggin34286d62011-01-07 17:49:57 +1100876 d_revalidate may be called in rcu-walk mode (nd->flags & LOOKUP_RCU).
877 If in rcu-walk mode, the filesystem must revalidate the dentry without
878 blocking or storing to the dentry, d_parent and d_inode should not be
879 used without care (because they can go NULL), instead nd->inode should
880 be used.
881
882 If a situation is encountered that rcu-walk cannot handle, return
883 -ECHILD and it will be called again in ref-walk mode.
884
Nick Piggin621e1552011-01-07 17:49:27 +1100885 d_hash: called when the VFS adds a dentry to the hash table. The first
886 dentry passed to d_hash is the parent directory that the name is
Nick Pigginb1e6a012011-01-07 17:49:28 +1100887 to be hashed into. The inode is the dentry's inode.
888
889 Same locking and synchronisation rules as d_compare regarding
890 what is safe to dereference etc.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700891
Nick Piggin621e1552011-01-07 17:49:27 +1100892 d_compare: called to compare a dentry name with a given name. The first
893 dentry is the parent of the dentry to be compared, the second is
894 the parent's inode, then the dentry and inode (may be NULL) of the
895 child dentry. len and name string are properties of the dentry to be
896 compared. qstr is the name to compare it with.
897
898 Must be constant and idempotent, and should not take locks if
899 possible, and should not or store into the dentry or inodes.
900 Should not dereference pointers outside the dentry or inodes without
901 lots of care (eg. d_parent, d_inode, d_name should not be used).
902
903 However, our vfsmount is pinned, and RCU held, so the dentries and
904 inodes won't disappear, neither will our sb or filesystem module.
905 ->i_sb and ->d_sb may be used.
906
907 It is a tricky calling convention because it needs to be called under
908 "rcu-walk", ie. without any locks or references on things.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700909
Nick Pigginfe15ce42011-01-07 17:49:23 +1100910 d_delete: called when the last reference to a dentry is dropped and the
911 dcache is deciding whether or not to cache it. Return 1 to delete
912 immediately, or 0 to cache the dentry. Default is NULL which means to
913 always cache a reachable dentry. d_delete must be constant and
914 idempotent.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700915
916 d_release: called when a dentry is really deallocated
917
918 d_iput: called when a dentry loses its inode (just prior to its
919 being deallocated). The default when this is NULL is that the
920 VFS calls iput(). If you define this method, you must call
921 iput() yourself
922
Eric Dumazetc23fbb62007-05-08 00:26:18 -0700923 d_dname: called when the pathname of a dentry should be generated.
Matt LaPlanted9195882008-07-25 19:45:33 -0700924 Useful for some pseudo filesystems (sockfs, pipefs, ...) to delay
Eric Dumazetc23fbb62007-05-08 00:26:18 -0700925 pathname generation. (Instead of doing it when dentry is created,
Matt LaPlanted9195882008-07-25 19:45:33 -0700926 it's done only when the path is needed.). Real filesystems probably
Eric Dumazetc23fbb62007-05-08 00:26:18 -0700927 dont want to use it, because their dentries are present in global
928 dcache hash, so their hash should be an invariant. As no lock is
929 held, d_dname() should not try to modify the dentry itself, unless
930 appropriate SMP safety is used. CAUTION : d_path() logic is quite
931 tricky. The correct way to return for example "Hello" is to put it
932 at the end of the buffer, and returns a pointer to the first char.
933 dynamic_dname() helper function is provided to take care of this.
934
David Howells9875cf82011-01-14 18:45:21 +0000935 d_automount: called when an automount dentry is to be traversed (optional).
David Howellsea5b7782011-01-14 19:10:03 +0000936 This should create a new VFS mount record and return the record to the
937 caller. The caller is supplied with a path parameter giving the
938 automount directory to describe the automount target and the parent
939 VFS mount record to provide inheritable mount parameters. NULL should
940 be returned if someone else managed to make the automount first. If
941 the vfsmount creation failed, then an error code should be returned.
942 If -EISDIR is returned, then the directory will be treated as an
943 ordinary directory and returned to pathwalk to continue walking.
944
945 If a vfsmount is returned, the caller will attempt to mount it on the
946 mountpoint and will remove the vfsmount from its expiration list in
947 the case of failure. The vfsmount should be returned with 2 refs on
948 it to prevent automatic expiration - the caller will clean up the
949 additional ref.
David Howells9875cf82011-01-14 18:45:21 +0000950
951 This function is only used if DCACHE_NEED_AUTOMOUNT is set on the
952 dentry. This is set by __d_instantiate() if S_AUTOMOUNT is set on the
953 inode being added.
954
David Howellscc53ce52011-01-14 18:45:26 +0000955 d_manage: called to allow the filesystem to manage the transition from a
956 dentry (optional). This allows autofs, for example, to hold up clients
957 waiting to explore behind a 'mountpoint' whilst letting the daemon go
958 past and construct the subtree there. 0 should be returned to let the
959 calling process continue. -EISDIR can be returned to tell pathwalk to
960 use this directory as an ordinary directory and to ignore anything
961 mounted on it and not to check the automount flag. Any other error
962 code will abort pathwalk completely.
963
964 If the 'mounting_here' parameter is true, then namespace_sem is being
965 held by the caller and the function should not initiate any mounts or
966 unmounts that it will then wait for.
967
David Howellsab909112011-01-14 18:46:51 +0000968 If the 'rcu_walk' parameter is true, then the caller is doing a
969 pathwalk in RCU-walk mode. Sleeping is not permitted in this mode,
970 and the caller can be asked to leave it and call again by returing
971 -ECHILD.
972
David Howellscc53ce52011-01-14 18:45:26 +0000973 This function is only used if DCACHE_MANAGE_TRANSIT is set on the
974 dentry being transited from.
975
Eric Dumazetc23fbb62007-05-08 00:26:18 -0700976Example :
977
978static char *pipefs_dname(struct dentry *dent, char *buffer, int buflen)
979{
980 return dynamic_dname(dentry, buffer, buflen, "pipe:[%lu]",
981 dentry->d_inode->i_ino);
982}
983
Linus Torvalds1da177e2005-04-16 15:20:36 -0700984Each dentry has a pointer to its parent dentry, as well as a hash list
985of child dentries. Child dentries are basically like files in a
986directory.
987
Pekka J Enberg5ea626a2005-09-09 13:10:19 -0700988
Pekka Enbergcc7d1f82005-11-07 01:01:08 -0800989Directory Entry Cache API
Linus Torvalds1da177e2005-04-16 15:20:36 -0700990--------------------------
991
992There are a number of functions defined which permit a filesystem to
993manipulate dentries:
994
995 dget: open a new handle for an existing dentry (this just increments
996 the usage count)
997
998 dput: close a handle for a dentry (decrements the usage count). If
Nick Pigginfe15ce42011-01-07 17:49:23 +1100999 the usage count drops to 0, and the dentry is still in its
1000 parent's hash, the "d_delete" method is called to check whether
1001 it should be cached. If it should not be cached, or if the dentry
1002 is not hashed, it is deleted. Otherwise cached dentries are put
1003 into an LRU list to be reclaimed on memory shortage.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001004
1005 d_drop: this unhashes a dentry from its parents hash list. A
Pekka J Enberg5ea626a2005-09-09 13:10:19 -07001006 subsequent call to dput() will deallocate the dentry if its
Linus Torvalds1da177e2005-04-16 15:20:36 -07001007 usage count drops to 0
1008
1009 d_delete: delete a dentry. If there are no other open references to
1010 the dentry then the dentry is turned into a negative dentry
1011 (the d_iput() method is called). If there are other
1012 references, then d_drop() is called instead
1013
1014 d_add: add a dentry to its parents hash list and then calls
1015 d_instantiate()
1016
1017 d_instantiate: add a dentry to the alias hash list for the inode and
1018 updates the "d_inode" member. The "i_count" member in the
1019 inode structure should be set/incremented. If the inode
1020 pointer is NULL, the dentry is called a "negative
1021 dentry". This function is commonly called when an inode is
1022 created for an existing negative dentry
1023
1024 d_lookup: look up a dentry given its parent and path name component
1025 It looks up the child of that given name from the dcache
1026 hash table. If it is found, the reference count is incremented
Zhaoleibe42c4c2008-12-01 14:34:58 -08001027 and the dentry is returned. The caller must use dput()
Linus Torvalds1da177e2005-04-16 15:20:36 -07001028 to free the dentry when it finishes using it.
1029
Pekka Enbergcbf8f0f2005-11-07 01:01:09 -08001030For further information on dentry locking, please refer to the document
1031Documentation/filesystems/dentry-locking.txt.
Pekka Enbergcc7d1f82005-11-07 01:01:08 -08001032
Miklos Szeredif84e3f52008-02-08 04:21:34 -08001033Mount Options
1034=============
1035
1036Parsing options
1037---------------
1038
1039On mount and remount the filesystem is passed a string containing a
1040comma separated list of mount options. The options can have either of
1041these forms:
1042
1043 option
1044 option=value
1045
1046The <linux/parser.h> header defines an API that helps parse these
1047options. There are plenty of examples on how to use it in existing
1048filesystems.
1049
1050Showing options
1051---------------
1052
1053If a filesystem accepts mount options, it must define show_options()
1054to show all the currently active options. The rules are:
1055
1056 - options MUST be shown which are not default or their values differ
1057 from the default
1058
1059 - options MAY be shown which are enabled by default or have their
1060 default value
1061
1062Options used only internally between a mount helper and the kernel
1063(such as file descriptors), or which only have an effect during the
1064mounting (such as ones controlling the creation of a journal) are exempt
1065from the above rules.
1066
1067The underlying reason for the above rules is to make sure, that a
1068mount can be accurately replicated (e.g. umounting and mounting again)
1069based on the information found in /proc/mounts.
1070
1071A simple method of saving options at mount/remount time and showing
1072them is provided with the save_mount_options() and
1073generic_show_options() helper functions. Please note, that using
1074these may have drawbacks. For more info see header comments for these
1075functions in fs/namespace.c.
Pekka Enbergcc7d1f82005-11-07 01:01:08 -08001076
1077Resources
1078=========
1079
1080(Note some of these resources are not up-to-date with the latest kernel
1081 version.)
1082
1083Creating Linux virtual filesystems. 2002
1084 <http://lwn.net/Articles/13325/>
1085
1086The Linux Virtual File-system Layer by Neil Brown. 1999
1087 <http://www.cse.unsw.edu.au/~neilb/oss/linux-commentary/vfs.html>
1088
1089A tour of the Linux VFS by Michael K. Johnson. 1996
1090 <http://www.tldp.org/LDP/khg/HyperNews/get/fs/vfstour.html>
1091
1092A small trail through the Linux kernel by Andries Brouwer. 2001
1093 <http://www.win.tue.nl/~aeb/linux/vfs/trail.html>