2 Overview of the Linux Virtual File System
4 Original author: Richard Gooch <rgooch@atnf.csiro.au>
6 Last updated on June 24, 2007.
8 Copyright (C) 1999 Richard Gooch
9 Copyright (C) 2005 Pekka Enberg
11 This file is released under the GPLv2.
17 The Virtual File System (also known as the Virtual Filesystem Switch)
18 is the software layer in the kernel that provides the filesystem
19 interface to userspace programs. It also provides an abstraction
20 within the kernel which allows different filesystem implementations to
23 VFS system calls open(2), stat(2), read(2), write(2), chmod(2) and so
24 on are called from a process context. Filesystem locking is described
25 in the document Documentation/filesystems/Locking.
28 Directory Entry Cache (dcache)
29 ------------------------------
31 The VFS implements the open(2), stat(2), chmod(2), and similar system
32 calls. The pathname argument that is passed to them is used by the VFS
33 to search through the directory entry cache (also known as the dentry
34 cache or dcache). This provides a very fast look-up mechanism to
35 translate a pathname (filename) into a specific dentry. Dentries live
36 in RAM and are never saved to disc: they exist only for performance.
38 The dentry cache is meant to be a view into your entire filespace. As
39 most computers cannot fit all dentries in the RAM at the same time,
40 some bits of the cache are missing. In order to resolve your pathname
41 into a dentry, the VFS may have to resort to creating dentries along
42 the way, and then loading the inode. This is done by looking up the
49 An individual dentry usually has a pointer to an inode. Inodes are
50 filesystem objects such as regular files, directories, FIFOs and other
51 beasts. They live either on the disc (for block device filesystems)
52 or in the memory (for pseudo filesystems). Inodes that live on the
53 disc are copied into the memory when required and changes to the inode
54 are written back to disc. A single inode can be pointed to by multiple
55 dentries (hard links, for example, do this).
57 To look up an inode requires that the VFS calls the lookup() method of
58 the parent directory inode. This method is installed by the specific
59 filesystem implementation that the inode lives in. Once the VFS has
60 the required dentry (and hence the inode), we can do all those boring
61 things like open(2) the file, or stat(2) it to peek at the inode
62 data. The stat(2) operation is fairly simple: once the VFS has the
63 dentry, it peeks at the inode data and passes some of it back to
70 Opening a file requires another operation: allocation of a file
71 structure (this is the kernel-side implementation of file
72 descriptors). The freshly allocated file structure is initialized with
73 a pointer to the dentry and a set of file operation member functions.
74 These are taken from the inode data. The open() file method is then
75 called so the specific filesystem implementation can do its work. You
76 can see that this is another switch performed by the VFS. The file
77 structure is placed into the file descriptor table for the process.
79 Reading, writing and closing files (and other assorted VFS operations)
80 is done by using the userspace file descriptor to grab the appropriate
81 file structure, and then calling the required file structure method to
82 do whatever is required. For as long as the file is open, it keeps the
83 dentry in use, which in turn means that the VFS inode is still in use.
86 Registering and Mounting a Filesystem
87 =====================================
89 To register and unregister a filesystem, use the following API
94 extern int register_filesystem(struct file_system_type *);
95 extern int unregister_filesystem(struct file_system_type *);
97 The passed struct file_system_type describes your filesystem. When a
98 request is made to mount a filesystem onto a directory in your namespace,
99 the VFS will call the appropriate mount() method for the specific
100 filesystem. New vfsmount referring to the tree returned by ->mount()
101 will be attached to the mountpoint, so that when pathname resolution
102 reaches the mountpoint it will jump into the root of that vfsmount.
104 You can see all filesystems that are registered to the kernel in the
105 file /proc/filesystems.
108 struct file_system_type
109 -----------------------
111 This describes the filesystem. As of kernel 2.6.39, the following
114 struct file_system_type {
117 struct dentry *(*mount) (struct file_system_type *, int,
118 const char *, void *);
119 void (*kill_sb) (struct super_block *);
120 struct module *owner;
121 struct file_system_type * next;
122 struct list_head fs_supers;
123 struct lock_class_key s_lock_key;
124 struct lock_class_key s_umount_key;
127 name: the name of the filesystem type, such as "ext2", "iso9660",
130 fs_flags: various flags (i.e. FS_REQUIRES_DEV, FS_NO_DCACHE, etc.)
132 mount: the method to call when a new instance of this
133 filesystem should be mounted
135 kill_sb: the method to call when an instance of this filesystem
138 owner: for internal VFS use: you should initialize this to THIS_MODULE in
141 next: for internal VFS use: you should initialize this to NULL
143 s_lock_key, s_umount_key: lockdep-specific
145 The mount() method has the following arguments:
147 struct file_system_type *fs_type: describes the filesystem, partly initialized
148 by the specific filesystem code
150 int flags: mount flags
152 const char *dev_name: the device name we are mounting.
154 void *data: arbitrary mount options, usually comes as an ASCII
155 string (see "Mount Options" section)
157 The mount() method must return the root dentry of the tree requested by
158 caller. An active reference to its superblock must be grabbed and the
159 superblock must be locked. On failure it should return ERR_PTR(error).
161 The arguments match those of mount(2) and their interpretation
162 depends on filesystem type. E.g. for block filesystems, dev_name is
163 interpreted as block device name, that device is opened and if it
164 contains a suitable filesystem image the method creates and initializes
165 struct super_block accordingly, returning its root dentry to caller.
167 ->mount() may choose to return a subtree of existing filesystem - it
168 doesn't have to create a new one. The main result from the caller's
169 point of view is a reference to dentry at the root of (sub)tree to
170 be attached; creation of new superblock is a common side effect.
172 The most interesting member of the superblock structure that the
173 mount() method fills in is the "s_op" field. This is a pointer to
174 a "struct super_operations" which describes the next level of the
175 filesystem implementation.
177 Usually, a filesystem uses one of the generic mount() implementations
178 and provides a fill_super() callback instead. The generic variants are:
180 mount_bdev: mount a filesystem residing on a block device
182 mount_nodev: mount a filesystem that is not backed by a device
184 mount_single: mount a filesystem which shares the instance between
187 A fill_super() callback implementation has the following arguments:
189 struct super_block *sb: the superblock structure. The callback
190 must initialize this properly.
192 void *data: arbitrary mount options, usually comes as an ASCII
193 string (see "Mount Options" section)
195 int silent: whether or not to be silent on error
198 The Superblock Object
199 =====================
201 A superblock object represents a mounted filesystem.
204 struct super_operations
205 -----------------------
207 This describes how the VFS can manipulate the superblock of your
208 filesystem. As of kernel 2.6.22, the following members are defined:
210 struct super_operations {
211 struct inode *(*alloc_inode)(struct super_block *sb);
212 void (*destroy_inode)(struct inode *);
214 void (*dirty_inode) (struct inode *, int flags);
215 int (*write_inode) (struct inode *, int);
216 void (*drop_inode) (struct inode *);
217 void (*delete_inode) (struct inode *);
218 void (*put_super) (struct super_block *);
219 int (*sync_fs)(struct super_block *sb, int wait);
220 int (*freeze_fs) (struct super_block *);
221 int (*unfreeze_fs) (struct super_block *);
222 int (*statfs) (struct dentry *, struct kstatfs *);
223 int (*remount_fs) (struct super_block *, int *, char *);
224 void (*clear_inode) (struct inode *);
225 void (*umount_begin) (struct super_block *);
227 int (*show_options)(struct seq_file *, struct dentry *);
229 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
230 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
231 int (*nr_cached_objects)(struct super_block *);
232 void (*free_cached_objects)(struct super_block *, int);
235 All methods are called without any locks being held, unless otherwise
236 noted. This means that most methods can block safely. All methods are
237 only called from a process context (i.e. not from an interrupt handler
240 alloc_inode: this method is called by inode_alloc() to allocate memory
241 for struct inode and initialize it. If this function is not
242 defined, a simple 'struct inode' is allocated. Normally
243 alloc_inode will be used to allocate a larger structure which
244 contains a 'struct inode' embedded within it.
246 destroy_inode: this method is called by destroy_inode() to release
247 resources allocated for struct inode. It is only required if
248 ->alloc_inode was defined and simply undoes anything done by
251 dirty_inode: this method is called by the VFS to mark an inode dirty.
253 write_inode: this method is called when the VFS needs to write an
254 inode to disc. The second parameter indicates whether the write
255 should be synchronous or not, not all filesystems check this flag.
257 drop_inode: called when the last access to the inode is dropped,
258 with the inode->i_lock spinlock held.
260 This method should be either NULL (normal UNIX filesystem
261 semantics) or "generic_delete_inode" (for filesystems that do not
262 want to cache inodes - causing "delete_inode" to always be
263 called regardless of the value of i_nlink)
265 The "generic_delete_inode()" behavior is equivalent to the
266 old practice of using "force_delete" in the put_inode() case,
267 but does not have the races that the "force_delete()" approach
270 delete_inode: called when the VFS wants to delete an inode
272 put_super: called when the VFS wishes to free the superblock
273 (i.e. unmount). This is called with the superblock lock held
275 sync_fs: called when VFS is writing out all dirty data associated with
276 a superblock. The second parameter indicates whether the method
277 should wait until the write out has been completed. Optional.
279 freeze_fs: called when VFS is locking a filesystem and
280 forcing it into a consistent state. This method is currently
281 used by the Logical Volume Manager (LVM).
283 unfreeze_fs: called when VFS is unlocking a filesystem and making it writable
286 statfs: called when the VFS needs to get filesystem statistics.
288 remount_fs: called when the filesystem is remounted. This is called
289 with the kernel lock held
291 clear_inode: called then the VFS clears the inode. Optional
293 umount_begin: called when the VFS is unmounting a filesystem.
295 show_options: called by the VFS to show mount options for
296 /proc/<pid>/mounts. (see "Mount Options" section)
298 quota_read: called by the VFS to read from filesystem quota file.
300 quota_write: called by the VFS to write to filesystem quota file.
302 nr_cached_objects: called by the sb cache shrinking function for the
303 filesystem to return the number of freeable cached objects it contains.
306 free_cache_objects: called by the sb cache shrinking function for the
307 filesystem to scan the number of objects indicated to try to free them.
308 Optional, but any filesystem implementing this method needs to also
309 implement ->nr_cached_objects for it to be called correctly.
311 We can't do anything with any errors that the filesystem might
312 encountered, hence the void return type. This will never be called if
313 the VM is trying to reclaim under GFP_NOFS conditions, hence this
314 method does not need to handle that situation itself.
316 Implementations must include conditional reschedule calls inside any
317 scanning loop that is done. This allows the VFS to determine
318 appropriate scan batch sizes without having to worry about whether
319 implementations will cause holdoff problems due to large scan batch
322 Whoever sets up the inode is responsible for filling in the "i_op" field. This
323 is a pointer to a "struct inode_operations" which describes the methods that
324 can be performed on individual inodes.
330 An inode object represents an object within the filesystem.
333 struct inode_operations
334 -----------------------
336 This describes how the VFS can manipulate an inode in your
337 filesystem. As of kernel 2.6.22, the following members are defined:
339 struct inode_operations {
340 int (*create) (struct inode *,struct dentry *, umode_t, bool);
341 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
342 int (*link) (struct dentry *,struct inode *,struct dentry *);
343 int (*unlink) (struct inode *,struct dentry *);
344 int (*symlink) (struct inode *,struct dentry *,const char *);
345 int (*mkdir) (struct inode *,struct dentry *,umode_t);
346 int (*rmdir) (struct inode *,struct dentry *);
347 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
348 int (*rename) (struct inode *, struct dentry *,
349 struct inode *, struct dentry *);
350 int (*rename2) (struct inode *, struct dentry *,
351 struct inode *, struct dentry *, unsigned int);
352 int (*readlink) (struct dentry *, char __user *,int);
353 void * (*follow_link) (struct dentry *, struct nameidata *);
354 void (*put_link) (struct dentry *, struct nameidata *, void *);
355 int (*permission) (struct inode *, int);
356 int (*get_acl)(struct inode *, int);
357 int (*setattr) (struct dentry *, struct iattr *);
358 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
359 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
360 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
361 ssize_t (*listxattr) (struct dentry *, char *, size_t);
362 int (*removexattr) (struct dentry *, const char *);
363 void (*update_time)(struct inode *, struct timespec *, int);
364 int (*atomic_open)(struct inode *, struct dentry *, struct file *,
365 unsigned open_flag, umode_t create_mode, int *opened);
366 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
369 Again, all methods are called without any locks being held, unless
372 create: called by the open(2) and creat(2) system calls. Only
373 required if you want to support regular files. The dentry you
374 get should not have an inode (i.e. it should be a negative
375 dentry). Here you will probably call d_instantiate() with the
376 dentry and the newly created inode
378 lookup: called when the VFS needs to look up an inode in a parent
379 directory. The name to look for is found in the dentry. This
380 method must call d_add() to insert the found inode into the
381 dentry. The "i_count" field in the inode structure should be
382 incremented. If the named inode does not exist a NULL inode
383 should be inserted into the dentry (this is called a negative
384 dentry). Returning an error code from this routine must only
385 be done on a real error, otherwise creating inodes with system
386 calls like create(2), mknod(2), mkdir(2) and so on will fail.
387 If you wish to overload the dentry methods then you should
388 initialise the "d_dop" field in the dentry; this is a pointer
389 to a struct "dentry_operations".
390 This method is called with the directory inode semaphore held
392 link: called by the link(2) system call. Only required if you want
393 to support hard links. You will probably need to call
394 d_instantiate() just as you would in the create() method
396 unlink: called by the unlink(2) system call. Only required if you
397 want to support deleting inodes
399 symlink: called by the symlink(2) system call. Only required if you
400 want to support symlinks. You will probably need to call
401 d_instantiate() just as you would in the create() method
403 mkdir: called by the mkdir(2) system call. Only required if you want
404 to support creating subdirectories. You will probably need to
405 call d_instantiate() just as you would in the create() method
407 rmdir: called by the rmdir(2) system call. Only required if you want
408 to support deleting subdirectories
410 mknod: called by the mknod(2) system call to create a device (char,
411 block) inode or a named pipe (FIFO) or socket. Only required
412 if you want to support creating these types of inodes. You
413 will probably need to call d_instantiate() just as you would
414 in the create() method
416 rename: called by the rename(2) system call to rename the object to
417 have the parent and name given by the second inode and dentry.
419 rename2: this has an additional flags argument compared to rename.
420 If no flags are supported by the filesystem then this method
421 need not be implemented. If some flags are supported then the
422 filesystem must return -EINVAL for any unsupported or unknown
423 flags. Currently the following flags are implemented:
424 (1) RENAME_NOREPLACE: this flag indicates that if the target
425 of the rename exists the rename should fail with -EEXIST
426 instead of replacing the target. The VFS already checks for
427 existence, so for local filesystems the RENAME_NOREPLACE
428 implementation is equivalent to plain rename.
429 (2) RENAME_EXCHANGE: exchange source and target. Both must
430 exist; this is checked by the VFS. Unlike plain rename,
431 source and target may be of different type.
433 readlink: called by the readlink(2) system call. Only required if
434 you want to support reading symbolic links
436 follow_link: called by the VFS to follow a symbolic link to the
437 inode it points to. Only required if you want to support
438 symbolic links. This method returns a void pointer cookie
439 that is passed to put_link().
441 put_link: called by the VFS to release resources allocated by
442 follow_link(). The cookie returned by follow_link() is passed
443 to this method as the last parameter. It is used by
444 filesystems such as NFS where page cache is not stable
445 (i.e. page that was installed when the symbolic link walk
446 started might not be in the page cache at the end of the
449 permission: called by the VFS to check for access rights on a POSIX-like
452 May be called in rcu-walk mode (mask & MAY_NOT_BLOCK). If in rcu-walk
453 mode, the filesystem must check the permission without blocking or
454 storing to the inode.
456 If a situation is encountered that rcu-walk cannot handle, return
457 -ECHILD and it will be called again in ref-walk mode.
459 setattr: called by the VFS to set attributes for a file. This method
460 is called by chmod(2) and related system calls.
462 getattr: called by the VFS to get attributes of a file. This method
463 is called by stat(2) and related system calls.
465 setxattr: called by the VFS to set an extended attribute for a file.
466 Extended attribute is a name:value pair associated with an
467 inode. This method is called by setxattr(2) system call.
469 getxattr: called by the VFS to retrieve the value of an extended
470 attribute name. This method is called by getxattr(2) function
473 listxattr: called by the VFS to list all extended attributes for a
474 given file. This method is called by listxattr(2) system call.
476 removexattr: called by the VFS to remove an extended attribute from
477 a file. This method is called by removexattr(2) system call.
479 update_time: called by the VFS to update a specific time or the i_version of
480 an inode. If this is not defined the VFS will update the inode itself
481 and call mark_inode_dirty_sync.
483 atomic_open: called on the last component of an open. Using this optional
484 method the filesystem can look up, possibly create and open the file in
485 one atomic operation. If it cannot perform this (e.g. the file type
486 turned out to be wrong) it may signal this by returning 1 instead of
487 usual 0 or -ve . This method is only called if the last component is
488 negative or needs lookup. Cached positive dentries are still handled by
489 f_op->open(). If the file was created, the FILE_CREATED flag should be
490 set in "opened". In case of O_EXCL the method must only succeed if the
491 file didn't exist and hence FILE_CREATED shall always be set on success.
493 tmpfile: called in the end of O_TMPFILE open(). Optional, equivalent to
494 atomically creating, opening and unlinking a file in given directory.
496 The Address Space Object
497 ========================
499 The address space object is used to group and manage pages in the page
500 cache. It can be used to keep track of the pages in a file (or
501 anything else) and also track the mapping of sections of the file into
502 process address spaces.
504 There are a number of distinct yet related services that an
505 address-space can provide. These include communicating memory
506 pressure, page lookup by address, and keeping track of pages tagged as
509 The first can be used independently to the others. The VM can try to
510 either write dirty pages in order to clean them, or release clean
511 pages in order to reuse them. To do this it can call the ->writepage
512 method on dirty pages, and ->releasepage on clean pages with
513 PagePrivate set. Clean pages without PagePrivate and with no external
514 references will be released without notice being given to the
517 To achieve this functionality, pages need to be placed on an LRU with
518 lru_cache_add and mark_page_active needs to be called whenever the
521 Pages are normally kept in a radix tree index by ->index. This tree
522 maintains information about the PG_Dirty and PG_Writeback status of
523 each page, so that pages with either of these flags can be found
526 The Dirty tag is primarily used by mpage_writepages - the default
527 ->writepages method. It uses the tag to find dirty pages to call
528 ->writepage on. If mpage_writepages is not used (i.e. the address
529 provides its own ->writepages) , the PAGECACHE_TAG_DIRTY tag is
530 almost unused. write_inode_now and sync_inode do use it (through
531 __sync_single_inode) to check if ->writepages has been successful in
532 writing out the whole address_space.
534 The Writeback tag is used by filemap*wait* and sync_page* functions,
535 via filemap_fdatawait_range, to wait for all writeback to
536 complete. While waiting ->sync_page (if defined) will be called on
537 each page that is found to require writeback.
539 An address_space handler may attach extra information to a page,
540 typically using the 'private' field in the 'struct page'. If such
541 information is attached, the PG_Private flag should be set. This will
542 cause various VM routines to make extra calls into the address_space
543 handler to deal with that data.
545 An address space acts as an intermediate between storage and
546 application. Data is read into the address space a whole page at a
547 time, and provided to the application either by copying of the page,
548 or by memory-mapping the page.
549 Data is written into the address space by the application, and then
550 written-back to storage typically in whole pages, however the
551 address_space has finer control of write sizes.
553 The read process essentially only requires 'readpage'. The write
554 process is more complicated and uses write_begin/write_end or
555 set_page_dirty to write data into the address_space, and writepage,
556 sync_page, and writepages to writeback data to storage.
558 Adding and removing pages to/from an address_space is protected by the
561 When data is written to a page, the PG_Dirty flag should be set. It
562 typically remains set until writepage asks for it to be written. This
563 should clear PG_Dirty and set PG_Writeback. It can be actually
564 written at any point after PG_Dirty is clear. Once it is known to be
565 safe, PG_Writeback is cleared.
567 Writeback makes use of a writeback_control structure...
569 struct address_space_operations
570 -------------------------------
572 This describes how the VFS can manipulate mapping of a file to page cache in
573 your filesystem. The following members are defined:
575 struct address_space_operations {
576 int (*writepage)(struct page *page, struct writeback_control *wbc);
577 int (*readpage)(struct file *, struct page *);
578 int (*writepages)(struct address_space *, struct writeback_control *);
579 int (*set_page_dirty)(struct page *page);
580 int (*readpages)(struct file *filp, struct address_space *mapping,
581 struct list_head *pages, unsigned nr_pages);
582 int (*write_begin)(struct file *, struct address_space *mapping,
583 loff_t pos, unsigned len, unsigned flags,
584 struct page **pagep, void **fsdata);
585 int (*write_end)(struct file *, struct address_space *mapping,
586 loff_t pos, unsigned len, unsigned copied,
587 struct page *page, void *fsdata);
588 sector_t (*bmap)(struct address_space *, sector_t);
589 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
590 int (*releasepage) (struct page *, int);
591 void (*freepage)(struct page *);
592 ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset);
593 struct page* (*get_xip_page)(struct address_space *, sector_t,
595 /* migrate the contents of a page to the specified target */
596 int (*migratepage) (struct page *, struct page *);
597 int (*launder_page) (struct page *);
598 int (*is_partially_uptodate) (struct page *, unsigned long,
600 void (*is_dirty_writeback) (struct page *, bool *, bool *);
601 int (*error_remove_page) (struct mapping *mapping, struct page *page);
602 int (*swap_activate)(struct file *);
603 int (*swap_deactivate)(struct file *);
606 writepage: called by the VM to write a dirty page to backing store.
607 This may happen for data integrity reasons (i.e. 'sync'), or
608 to free up memory (flush). The difference can be seen in
610 The PG_Dirty flag has been cleared and PageLocked is true.
611 writepage should start writeout, should set PG_Writeback,
612 and should make sure the page is unlocked, either synchronously
613 or asynchronously when the write operation completes.
615 If wbc->sync_mode is WB_SYNC_NONE, ->writepage doesn't have to
616 try too hard if there are problems, and may choose to write out
617 other pages from the mapping if that is easier (e.g. due to
618 internal dependencies). If it chooses not to start writeout, it
619 should return AOP_WRITEPAGE_ACTIVATE so that the VM will not keep
620 calling ->writepage on that page.
622 See the file "Locking" for more details.
624 readpage: called by the VM to read a page from backing store.
625 The page will be Locked when readpage is called, and should be
626 unlocked and marked uptodate once the read completes.
627 If ->readpage discovers that it needs to unlock the page for
628 some reason, it can do so, and then return AOP_TRUNCATED_PAGE.
629 In this case, the page will be relocated, relocked and if
630 that all succeeds, ->readpage will be called again.
632 writepages: called by the VM to write out pages associated with the
633 address_space object. If wbc->sync_mode is WBC_SYNC_ALL, then
634 the writeback_control will specify a range of pages that must be
635 written out. If it is WBC_SYNC_NONE, then a nr_to_write is given
636 and that many pages should be written if possible.
637 If no ->writepages is given, then mpage_writepages is used
638 instead. This will choose pages from the address space that are
639 tagged as DIRTY and will pass them to ->writepage.
641 set_page_dirty: called by the VM to set a page dirty.
642 This is particularly needed if an address space attaches
643 private data to a page, and that data needs to be updated when
644 a page is dirtied. This is called, for example, when a memory
645 mapped page gets modified.
646 If defined, it should set the PageDirty flag, and the
647 PAGECACHE_TAG_DIRTY tag in the radix tree.
649 readpages: called by the VM to read pages associated with the address_space
650 object. This is essentially just a vector version of
651 readpage. Instead of just one page, several pages are
653 readpages is only used for read-ahead, so read errors are
654 ignored. If anything goes wrong, feel free to give up.
657 Called by the generic buffered write code to ask the filesystem to
658 prepare to write len bytes at the given offset in the file. The
659 address_space should check that the write will be able to complete,
660 by allocating space if necessary and doing any other internal
661 housekeeping. If the write will update parts of any basic-blocks on
662 storage, then those blocks should be pre-read (if they haven't been
663 read already) so that the updated blocks can be written out properly.
665 The filesystem must return the locked pagecache page for the specified
666 offset, in *pagep, for the caller to write into.
668 It must be able to cope with short writes (where the length passed to
669 write_begin is greater than the number of bytes copied into the page).
671 flags is a field for AOP_FLAG_xxx flags, described in
674 A void * may be returned in fsdata, which then gets passed into
677 Returns 0 on success; < 0 on failure (which is the error code), in
678 which case write_end is not called.
680 write_end: After a successful write_begin, and data copy, write_end must
681 be called. len is the original len passed to write_begin, and copied
682 is the amount that was able to be copied (copied == len is always true
683 if write_begin was called with the AOP_FLAG_UNINTERRUPTIBLE flag).
685 The filesystem must take care of unlocking the page and releasing it
686 refcount, and updating i_size.
688 Returns < 0 on failure, otherwise the number of bytes (<= 'copied')
689 that were able to be copied into pagecache.
691 bmap: called by the VFS to map a logical block offset within object to
692 physical block number. This method is used by the FIBMAP
693 ioctl and for working with swap-files. To be able to swap to
694 a file, the file must have a stable mapping to a block
695 device. The swap system does not go through the filesystem
696 but instead uses bmap to find out where the blocks in the file
697 are and uses those addresses directly.
700 invalidatepage: If a page has PagePrivate set, then invalidatepage
701 will be called when part or all of the page is to be removed
702 from the address space. This generally corresponds to either a
703 truncation, punch hole or a complete invalidation of the address
704 space (in the latter case 'offset' will always be 0 and 'length'
705 will be PAGE_CACHE_SIZE). Any private data associated with the page
706 should be updated to reflect this truncation. If offset is 0 and
707 length is PAGE_CACHE_SIZE, then the private data should be released,
708 because the page must be able to be completely discarded. This may
709 be done by calling the ->releasepage function, but in this case the
710 release MUST succeed.
712 releasepage: releasepage is called on PagePrivate pages to indicate
713 that the page should be freed if possible. ->releasepage
714 should remove any private data from the page and clear the
715 PagePrivate flag. If releasepage() fails for some reason, it must
716 indicate failure with a 0 return value.
717 releasepage() is used in two distinct though related cases. The
718 first is when the VM finds a clean page with no active users and
719 wants to make it a free page. If ->releasepage succeeds, the
720 page will be removed from the address_space and become free.
722 The second case is when a request has been made to invalidate
723 some or all pages in an address_space. This can happen
724 through the fadvice(POSIX_FADV_DONTNEED) system call or by the
725 filesystem explicitly requesting it as nfs and 9fs do (when
726 they believe the cache may be out of date with storage) by
727 calling invalidate_inode_pages2().
728 If the filesystem makes such a call, and needs to be certain
729 that all pages are invalidated, then its releasepage will
730 need to ensure this. Possibly it can clear the PageUptodate
731 bit if it cannot free private data yet.
733 freepage: freepage is called once the page is no longer visible in
734 the page cache in order to allow the cleanup of any private
735 data. Since it may be called by the memory reclaimer, it
736 should not assume that the original address_space mapping still
737 exists, and it should not block.
739 direct_IO: called by the generic read/write routines to perform
740 direct_IO - that is IO requests which bypass the page cache
741 and transfer data directly between the storage and the
742 application's address space.
744 get_xip_page: called by the VM to translate a block number to a page.
745 The page is valid until the corresponding filesystem is unmounted.
746 Filesystems that want to use execute-in-place (XIP) need to implement
747 it. An example implementation can be found in fs/ext2/xip.c.
749 migrate_page: This is used to compact the physical memory usage.
750 If the VM wants to relocate a page (maybe off a memory card
751 that is signalling imminent failure) it will pass a new page
752 and an old page to this function. migrate_page should
753 transfer any private data across and update any references
754 that it has to the page.
756 launder_page: Called before freeing a page - it writes back the dirty page. To
757 prevent redirtying the page, it is kept locked during the whole
760 is_partially_uptodate: Called by the VM when reading a file through the
761 pagecache when the underlying blocksize != pagesize. If the required
762 block is up to date then the read can complete without needing the IO
763 to bring the whole page up to date.
765 is_dirty_writeback: Called by the VM when attempting to reclaim a page.
766 The VM uses dirty and writeback information to determine if it needs
767 to stall to allow flushers a chance to complete some IO. Ordinarily
768 it can use PageDirty and PageWriteback but some filesystems have
769 more complex state (unstable pages in NFS prevent reclaim) or
770 do not set those flags due to locking problems (jbd). This callback
771 allows a filesystem to indicate to the VM if a page should be
772 treated as dirty or writeback for the purposes of stalling.
774 error_remove_page: normally set to generic_error_remove_page if truncation
775 is ok for this address space. Used for memory failure handling.
776 Setting this implies you deal with pages going away under you,
777 unless you have them locked or reference counts increased.
779 swap_activate: Called when swapon is used on a file to allocate
780 space if necessary and pin the block lookup information in
781 memory. A return value of zero indicates success,
782 in which case this file can be used to back swapspace. The
783 swapspace operations will be proxied to this address space's
784 ->swap_{out,in} methods.
786 swap_deactivate: Called during swapoff on files where swap_activate
793 A file object represents a file opened by a process.
796 struct file_operations
797 ----------------------
799 This describes how the VFS can manipulate an open file. As of kernel
800 3.12, the following members are defined:
802 struct file_operations {
803 struct module *owner;
804 loff_t (*llseek) (struct file *, loff_t, int);
805 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
806 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
807 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
808 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
809 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
810 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
811 int (*iterate) (struct file *, struct dir_context *);
812 unsigned int (*poll) (struct file *, struct poll_table_struct *);
813 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
814 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
815 int (*mmap) (struct file *, struct vm_area_struct *);
816 int (*open) (struct inode *, struct file *);
817 int (*flush) (struct file *);
818 int (*release) (struct inode *, struct file *);
819 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
820 int (*aio_fsync) (struct kiocb *, int datasync);
821 int (*fasync) (int, struct file *, int);
822 int (*lock) (struct file *, int, struct file_lock *);
823 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
824 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
825 int (*check_flags)(int);
826 int (*flock) (struct file *, int, struct file_lock *);
827 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, size_t, unsigned int);
828 ssize_t (*splice_read)(struct file *, struct pipe_inode_info *, size_t, unsigned int);
829 int (*setlease)(struct file *, long arg, struct file_lock **);
830 long (*fallocate)(struct file *, int mode, loff_t offset, loff_t len);
831 int (*show_fdinfo)(struct seq_file *m, struct file *f);
834 Again, all methods are called without any locks being held, unless
837 llseek: called when the VFS needs to move the file position index
839 read: called by read(2) and related system calls
841 aio_read: vectored, possibly asynchronous read
843 read_iter: possibly asynchronous read with iov_iter as destination
845 write: called by write(2) and related system calls
847 aio_write: vectored, possibly asynchronous write
849 write_iter: possibly asynchronous write with iov_iter as source
851 iterate: called when the VFS needs to read the directory contents
853 poll: called by the VFS when a process wants to check if there is
854 activity on this file and (optionally) go to sleep until there
855 is activity. Called by the select(2) and poll(2) system calls
857 unlocked_ioctl: called by the ioctl(2) system call.
859 compat_ioctl: called by the ioctl(2) system call when 32 bit system calls
860 are used on 64 bit kernels.
862 mmap: called by the mmap(2) system call
864 open: called by the VFS when an inode should be opened. When the VFS
865 opens a file, it creates a new "struct file". It then calls the
866 open method for the newly allocated file structure. You might
867 think that the open method really belongs in
868 "struct inode_operations", and you may be right. I think it's
869 done the way it is because it makes filesystems simpler to
870 implement. The open() method is a good place to initialize the
871 "private_data" member in the file structure if you want to point
872 to a device structure
874 flush: called by the close(2) system call to flush a file
876 release: called when the last reference to an open file is closed
878 fsync: called by the fsync(2) system call
880 fasync: called by the fcntl(2) system call when asynchronous
881 (non-blocking) mode is enabled for a file
883 lock: called by the fcntl(2) system call for F_GETLK, F_SETLK, and F_SETLKW
886 get_unmapped_area: called by the mmap(2) system call
888 check_flags: called by the fcntl(2) system call for F_SETFL command
890 flock: called by the flock(2) system call
892 splice_write: called by the VFS to splice data from a pipe to a file. This
893 method is used by the splice(2) system call
895 splice_read: called by the VFS to splice data from file to a pipe. This
896 method is used by the splice(2) system call
898 setlease: called by the VFS to set or release a file lock lease.
899 setlease has the file_lock_lock held and must not sleep.
901 fallocate: called by the VFS to preallocate blocks or punch a hole.
903 Note that the file operations are implemented by the specific
904 filesystem in which the inode resides. When opening a device node
905 (character or block special) most filesystems will call special
906 support routines in the VFS which will locate the required device
907 driver information. These support routines replace the filesystem file
908 operations with those for the device driver, and then proceed to call
909 the new open() method for the file. This is how opening a device file
910 in the filesystem eventually ends up calling the device driver open()
914 Directory Entry Cache (dcache)
915 ==============================
918 struct dentry_operations
919 ------------------------
921 This describes how a filesystem can overload the standard dentry
922 operations. Dentries and the dcache are the domain of the VFS and the
923 individual filesystem implementations. Device drivers have no business
924 here. These methods may be set to NULL, as they are either optional or
925 the VFS uses a default. As of kernel 2.6.22, the following members are
928 struct dentry_operations {
929 int (*d_revalidate)(struct dentry *, unsigned int);
930 int (*d_weak_revalidate)(struct dentry *, unsigned int);
931 int (*d_hash)(const struct dentry *, struct qstr *);
932 int (*d_compare)(const struct dentry *, const struct dentry *,
933 unsigned int, const char *, const struct qstr *);
934 int (*d_delete)(const struct dentry *);
935 void (*d_release)(struct dentry *);
936 void (*d_iput)(struct dentry *, struct inode *);
937 char *(*d_dname)(struct dentry *, char *, int);
938 struct vfsmount *(*d_automount)(struct path *);
939 int (*d_manage)(struct dentry *, bool);
942 d_revalidate: called when the VFS needs to revalidate a dentry. This
943 is called whenever a name look-up finds a dentry in the
944 dcache. Most local filesystems leave this as NULL, because all their
945 dentries in the dcache are valid. Network filesystems are different
946 since things can change on the server without the client necessarily
949 This function should return a positive value if the dentry is still
950 valid, and zero or a negative error code if it isn't.
952 d_revalidate may be called in rcu-walk mode (flags & LOOKUP_RCU).
953 If in rcu-walk mode, the filesystem must revalidate the dentry without
954 blocking or storing to the dentry, d_parent and d_inode should not be
955 used without care (because they can change and, in d_inode case, even
956 become NULL under us).
958 If a situation is encountered that rcu-walk cannot handle, return
959 -ECHILD and it will be called again in ref-walk mode.
961 d_weak_revalidate: called when the VFS needs to revalidate a "jumped" dentry.
962 This is called when a path-walk ends at dentry that was not acquired by
963 doing a lookup in the parent directory. This includes "/", "." and "..",
964 as well as procfs-style symlinks and mountpoint traversal.
966 In this case, we are less concerned with whether the dentry is still
967 fully correct, but rather that the inode is still valid. As with
968 d_revalidate, most local filesystems will set this to NULL since their
969 dcache entries are always valid.
971 This function has the same return code semantics as d_revalidate.
973 d_weak_revalidate is only called after leaving rcu-walk mode.
975 d_hash: called when the VFS adds a dentry to the hash table. The first
976 dentry passed to d_hash is the parent directory that the name is
979 Same locking and synchronisation rules as d_compare regarding
980 what is safe to dereference etc.
982 d_compare: called to compare a dentry name with a given name. The first
983 dentry is the parent of the dentry to be compared, the second is
984 the child dentry. len and name string are properties of the dentry
985 to be compared. qstr is the name to compare it with.
987 Must be constant and idempotent, and should not take locks if
988 possible, and should not or store into the dentry.
989 Should not dereference pointers outside the dentry without
990 lots of care (eg. d_parent, d_inode, d_name should not be used).
992 However, our vfsmount is pinned, and RCU held, so the dentries and
993 inodes won't disappear, neither will our sb or filesystem module.
996 It is a tricky calling convention because it needs to be called under
997 "rcu-walk", ie. without any locks or references on things.
999 d_delete: called when the last reference to a dentry is dropped and the
1000 dcache is deciding whether or not to cache it. Return 1 to delete
1001 immediately, or 0 to cache the dentry. Default is NULL which means to
1002 always cache a reachable dentry. d_delete must be constant and
1005 d_release: called when a dentry is really deallocated
1007 d_iput: called when a dentry loses its inode (just prior to its
1008 being deallocated). The default when this is NULL is that the
1009 VFS calls iput(). If you define this method, you must call
1012 d_dname: called when the pathname of a dentry should be generated.
1013 Useful for some pseudo filesystems (sockfs, pipefs, ...) to delay
1014 pathname generation. (Instead of doing it when dentry is created,
1015 it's done only when the path is needed.). Real filesystems probably
1016 dont want to use it, because their dentries are present in global
1017 dcache hash, so their hash should be an invariant. As no lock is
1018 held, d_dname() should not try to modify the dentry itself, unless
1019 appropriate SMP safety is used. CAUTION : d_path() logic is quite
1020 tricky. The correct way to return for example "Hello" is to put it
1021 at the end of the buffer, and returns a pointer to the first char.
1022 dynamic_dname() helper function is provided to take care of this.
1024 d_automount: called when an automount dentry is to be traversed (optional).
1025 This should create a new VFS mount record and return the record to the
1026 caller. The caller is supplied with a path parameter giving the
1027 automount directory to describe the automount target and the parent
1028 VFS mount record to provide inheritable mount parameters. NULL should
1029 be returned if someone else managed to make the automount first. If
1030 the vfsmount creation failed, then an error code should be returned.
1031 If -EISDIR is returned, then the directory will be treated as an
1032 ordinary directory and returned to pathwalk to continue walking.
1034 If a vfsmount is returned, the caller will attempt to mount it on the
1035 mountpoint and will remove the vfsmount from its expiration list in
1036 the case of failure. The vfsmount should be returned with 2 refs on
1037 it to prevent automatic expiration - the caller will clean up the
1040 This function is only used if DCACHE_NEED_AUTOMOUNT is set on the
1041 dentry. This is set by __d_instantiate() if S_AUTOMOUNT is set on the
1044 d_manage: called to allow the filesystem to manage the transition from a
1045 dentry (optional). This allows autofs, for example, to hold up clients
1046 waiting to explore behind a 'mountpoint' whilst letting the daemon go
1047 past and construct the subtree there. 0 should be returned to let the
1048 calling process continue. -EISDIR can be returned to tell pathwalk to
1049 use this directory as an ordinary directory and to ignore anything
1050 mounted on it and not to check the automount flag. Any other error
1051 code will abort pathwalk completely.
1053 If the 'rcu_walk' parameter is true, then the caller is doing a
1054 pathwalk in RCU-walk mode. Sleeping is not permitted in this mode,
1055 and the caller can be asked to leave it and call again by returning
1058 This function is only used if DCACHE_MANAGE_TRANSIT is set on the
1059 dentry being transited from.
1063 static char *pipefs_dname(struct dentry *dent, char *buffer, int buflen)
1065 return dynamic_dname(dentry, buffer, buflen, "pipe:[%lu]",
1066 dentry->d_inode->i_ino);
1069 Each dentry has a pointer to its parent dentry, as well as a hash list
1070 of child dentries. Child dentries are basically like files in a
1074 Directory Entry Cache API
1075 --------------------------
1077 There are a number of functions defined which permit a filesystem to
1078 manipulate dentries:
1080 dget: open a new handle for an existing dentry (this just increments
1083 dput: close a handle for a dentry (decrements the usage count). If
1084 the usage count drops to 0, and the dentry is still in its
1085 parent's hash, the "d_delete" method is called to check whether
1086 it should be cached. If it should not be cached, or if the dentry
1087 is not hashed, it is deleted. Otherwise cached dentries are put
1088 into an LRU list to be reclaimed on memory shortage.
1090 d_drop: this unhashes a dentry from its parents hash list. A
1091 subsequent call to dput() will deallocate the dentry if its
1092 usage count drops to 0
1094 d_delete: delete a dentry. If there are no other open references to
1095 the dentry then the dentry is turned into a negative dentry
1096 (the d_iput() method is called). If there are other
1097 references, then d_drop() is called instead
1099 d_add: add a dentry to its parents hash list and then calls
1102 d_instantiate: add a dentry to the alias hash list for the inode and
1103 updates the "d_inode" member. The "i_count" member in the
1104 inode structure should be set/incremented. If the inode
1105 pointer is NULL, the dentry is called a "negative
1106 dentry". This function is commonly called when an inode is
1107 created for an existing negative dentry
1109 d_lookup: look up a dentry given its parent and path name component
1110 It looks up the child of that given name from the dcache
1111 hash table. If it is found, the reference count is incremented
1112 and the dentry is returned. The caller must use dput()
1113 to free the dentry when it finishes using it.
1121 On mount and remount the filesystem is passed a string containing a
1122 comma separated list of mount options. The options can have either of
1128 The <linux/parser.h> header defines an API that helps parse these
1129 options. There are plenty of examples on how to use it in existing
1135 If a filesystem accepts mount options, it must define show_options()
1136 to show all the currently active options. The rules are:
1138 - options MUST be shown which are not default or their values differ
1141 - options MAY be shown which are enabled by default or have their
1144 Options used only internally between a mount helper and the kernel
1145 (such as file descriptors), or which only have an effect during the
1146 mounting (such as ones controlling the creation of a journal) are exempt
1147 from the above rules.
1149 The underlying reason for the above rules is to make sure, that a
1150 mount can be accurately replicated (e.g. umounting and mounting again)
1151 based on the information found in /proc/mounts.
1153 A simple method of saving options at mount/remount time and showing
1154 them is provided with the save_mount_options() and
1155 generic_show_options() helper functions. Please note, that using
1156 these may have drawbacks. For more info see header comments for these
1157 functions in fs/namespace.c.
1162 (Note some of these resources are not up-to-date with the latest kernel
1165 Creating Linux virtual filesystems. 2002
1166 <http://lwn.net/Articles/13325/>
1168 The Linux Virtual File-system Layer by Neil Brown. 1999
1169 <http://www.cse.unsw.edu.au/~neilb/oss/linux-commentary/vfs.html>
1171 A tour of the Linux VFS by Michael K. Johnson. 1996
1172 <http://www.tldp.org/LDP/khg/HyperNews/get/fs/vfstour.html>
1174 A small trail through the Linux kernel by Andries Brouwer. 2001
1175 <http://www.win.tue.nl/~aeb/linux/vfs/trail.html>