1 .. SPDX-License-Identifier: GPL-2.0
3 =========================================
4 Overview of the Linux Virtual File System
5 =========================================
7 Original author: Richard Gooch <rgooch@atnf.csiro.au>
9 - Copyright (C) 1999 Richard Gooch
10 - Copyright (C) 2005 Pekka Enberg
16 The Virtual File System (also known as the Virtual Filesystem Switch) is
17 the software layer in the kernel that provides the filesystem interface
18 to userspace programs. It also provides an abstraction within the
19 kernel which allows different filesystem implementations to coexist.
21 VFS system calls open(2), stat(2), read(2), write(2), chmod(2) and so on
22 are called from a process context. Filesystem locking is described in
23 the document Documentation/filesystems/locking.rst.
26 Directory Entry Cache (dcache)
27 ------------------------------
29 The VFS implements the open(2), stat(2), chmod(2), and similar system
30 calls. The pathname argument that is passed to them is used by the VFS
31 to search through the directory entry cache (also known as the dentry
32 cache or dcache). This provides a very fast look-up mechanism to
33 translate a pathname (filename) into a specific dentry. Dentries live
34 in RAM and are never saved to disc: they exist only for performance.
36 The dentry cache is meant to be a view into your entire filespace. As
37 most computers cannot fit all dentries in the RAM at the same time, some
38 bits of the cache are missing. In order to resolve your pathname into a
39 dentry, the VFS may have to resort to creating dentries along the way,
40 and then loading the inode. This is done by looking up the inode.
46 An individual dentry usually has a pointer to an inode. Inodes are
47 filesystem objects such as regular files, directories, FIFOs and other
48 beasts. They live either on the disc (for block device filesystems) or
49 in the memory (for pseudo filesystems). Inodes that live on the disc
50 are copied into the memory when required and changes to the inode are
51 written back to disc. A single inode can be pointed to by multiple
52 dentries (hard links, for example, do this).
54 To look up an inode requires that the VFS calls the lookup() method of
55 the parent directory inode. This method is installed by the specific
56 filesystem implementation that the inode lives in. Once the VFS has the
57 required dentry (and hence the inode), we can do all those boring things
58 like open(2) the file, or stat(2) it to peek at the inode data. The
59 stat(2) operation is fairly simple: once the VFS has the dentry, it
60 peeks at the inode data and passes some of it back to userspace.
66 Opening a file requires another operation: allocation of a file
67 structure (this is the kernel-side implementation of file descriptors).
68 The freshly allocated file structure is initialized with a pointer to
69 the dentry and a set of file operation member functions. These are
70 taken from the inode data. The open() file method is then called so the
71 specific filesystem implementation can do its work. You can see that
72 this is another switch performed by the VFS. The file structure is
73 placed into the file descriptor table for the process.
75 Reading, writing and closing files (and other assorted VFS operations)
76 is done by using the userspace file descriptor to grab the appropriate
77 file structure, and then calling the required file structure method to
78 do whatever is required. For as long as the file is open, it keeps the
79 dentry in use, which in turn means that the VFS inode is still in use.
82 Registering and Mounting a Filesystem
83 =====================================
85 To register and unregister a filesystem, use the following API
92 extern int register_filesystem(struct file_system_type *);
93 extern int unregister_filesystem(struct file_system_type *);
95 The passed struct file_system_type describes your filesystem. When a
96 request is made to mount a filesystem onto a directory in your
97 namespace, the VFS will call the appropriate mount() method for the
98 specific filesystem. New vfsmount referring to the tree returned by
99 ->mount() will be attached to the mountpoint, so that when pathname
100 resolution reaches the mountpoint it will jump into the root of that
103 You can see all filesystems that are registered to the kernel in the
104 file /proc/filesystems.
107 struct file_system_type
108 -----------------------
110 This describes the filesystem. As of kernel 2.6.39, the following
115 struct file_system_type {
118 struct dentry *(*mount) (struct file_system_type *, int,
119 const char *, void *);
120 void (*kill_sb) (struct super_block *);
121 struct module *owner;
122 struct file_system_type * next;
123 struct list_head fs_supers;
124 struct lock_class_key s_lock_key;
125 struct lock_class_key s_umount_key;
129 the name of the filesystem type, such as "ext2", "iso9660",
133 various flags (i.e. FS_REQUIRES_DEV, FS_NO_DCACHE, etc.)
136 the method to call when a new instance of this filesystem should
140 the method to call when an instance of this filesystem should be
145 for internal VFS use: you should initialize this to THIS_MODULE
149 for internal VFS use: you should initialize this to NULL
151 s_lock_key, s_umount_key: lockdep-specific
153 The mount() method has the following arguments:
155 ``struct file_system_type *fs_type``
156 describes the filesystem, partly initialized by the specific
162 ``const char *dev_name``
163 the device name we are mounting.
166 arbitrary mount options, usually comes as an ASCII string (see
167 "Mount Options" section)
169 The mount() method must return the root dentry of the tree requested by
170 caller. An active reference to its superblock must be grabbed and the
171 superblock must be locked. On failure it should return ERR_PTR(error).
173 The arguments match those of mount(2) and their interpretation depends
174 on filesystem type. E.g. for block filesystems, dev_name is interpreted
175 as block device name, that device is opened and if it contains a
176 suitable filesystem image the method creates and initializes struct
177 super_block accordingly, returning its root dentry to caller.
179 ->mount() may choose to return a subtree of existing filesystem - it
180 doesn't have to create a new one. The main result from the caller's
181 point of view is a reference to dentry at the root of (sub)tree to be
182 attached; creation of new superblock is a common side effect.
184 The most interesting member of the superblock structure that the mount()
185 method fills in is the "s_op" field. This is a pointer to a "struct
186 super_operations" which describes the next level of the filesystem
189 Usually, a filesystem uses one of the generic mount() implementations
190 and provides a fill_super() callback instead. The generic variants are:
193 mount a filesystem residing on a block device
196 mount a filesystem that is not backed by a device
199 mount a filesystem which shares the instance between all mounts
201 A fill_super() callback implementation has the following arguments:
203 ``struct super_block *sb``
204 the superblock structure. The callback must initialize this
208 arbitrary mount options, usually comes as an ASCII string (see
209 "Mount Options" section)
212 whether or not to be silent on error
215 The Superblock Object
216 =====================
218 A superblock object represents a mounted filesystem.
221 struct super_operations
222 -----------------------
224 This describes how the VFS can manipulate the superblock of your
225 filesystem. As of kernel 2.6.22, the following members are defined:
229 struct super_operations {
230 struct inode *(*alloc_inode)(struct super_block *sb);
231 void (*destroy_inode)(struct inode *);
233 void (*dirty_inode) (struct inode *, int flags);
234 int (*write_inode) (struct inode *, int);
235 void (*drop_inode) (struct inode *);
236 void (*delete_inode) (struct inode *);
237 void (*put_super) (struct super_block *);
238 int (*sync_fs)(struct super_block *sb, int wait);
239 int (*freeze_fs) (struct super_block *);
240 int (*unfreeze_fs) (struct super_block *);
241 int (*statfs) (struct dentry *, struct kstatfs *);
242 int (*remount_fs) (struct super_block *, int *, char *);
243 void (*clear_inode) (struct inode *);
244 void (*umount_begin) (struct super_block *);
246 int (*show_options)(struct seq_file *, struct dentry *);
248 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
249 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
250 int (*nr_cached_objects)(struct super_block *);
251 void (*free_cached_objects)(struct super_block *, int);
254 All methods are called without any locks being held, unless otherwise
255 noted. This means that most methods can block safely. All methods are
256 only called from a process context (i.e. not from an interrupt handler
260 this method is called by alloc_inode() to allocate memory for
261 struct inode and initialize it. If this function is not
262 defined, a simple 'struct inode' is allocated. Normally
263 alloc_inode will be used to allocate a larger structure which
264 contains a 'struct inode' embedded within it.
267 this method is called by destroy_inode() to release resources
268 allocated for struct inode. It is only required if
269 ->alloc_inode was defined and simply undoes anything done by
273 this method is called by the VFS when an inode is marked dirty.
274 This is specifically for the inode itself being marked dirty,
275 not its data. If the update needs to be persisted by fdatasync(),
276 then I_DIRTY_DATASYNC will be set in the flags argument.
277 I_DIRTY_TIME will be set in the flags in case lazytime is enabled
278 and struct inode has times updated since the last ->dirty_inode
282 this method is called when the VFS needs to write an inode to
283 disc. The second parameter indicates whether the write should
284 be synchronous or not, not all filesystems check this flag.
287 called when the last access to the inode is dropped, with the
288 inode->i_lock spinlock held.
290 This method should be either NULL (normal UNIX filesystem
291 semantics) or "generic_delete_inode" (for filesystems that do
292 not want to cache inodes - causing "delete_inode" to always be
293 called regardless of the value of i_nlink)
295 The "generic_delete_inode()" behavior is equivalent to the old
296 practice of using "force_delete" in the put_inode() case, but
297 does not have the races that the "force_delete()" approach had.
300 called when the VFS wants to delete an inode
303 called when the VFS wishes to free the superblock
304 (i.e. unmount). This is called with the superblock lock held
307 called when VFS is writing out all dirty data associated with a
308 superblock. The second parameter indicates whether the method
309 should wait until the write out has been completed. Optional.
312 called when VFS is locking a filesystem and forcing it into a
313 consistent state. This method is currently used by the Logical
314 Volume Manager (LVM).
317 called when VFS is unlocking a filesystem and making it writable
321 called when the VFS needs to get filesystem statistics.
324 called when the filesystem is remounted. This is called with
328 called then the VFS clears the inode. Optional
331 called when the VFS is unmounting a filesystem.
334 called by the VFS to show mount options for /proc/<pid>/mounts.
335 (see "Mount Options" section)
338 called by the VFS to read from filesystem quota file.
341 called by the VFS to write to filesystem quota file.
343 ``nr_cached_objects``
344 called by the sb cache shrinking function for the filesystem to
345 return the number of freeable cached objects it contains.
348 ``free_cache_objects``
349 called by the sb cache shrinking function for the filesystem to
350 scan the number of objects indicated to try to free them.
351 Optional, but any filesystem implementing this method needs to
352 also implement ->nr_cached_objects for it to be called
355 We can't do anything with any errors that the filesystem might
356 encountered, hence the void return type. This will never be
357 called if the VM is trying to reclaim under GFP_NOFS conditions,
358 hence this method does not need to handle that situation itself.
360 Implementations must include conditional reschedule calls inside
361 any scanning loop that is done. This allows the VFS to
362 determine appropriate scan batch sizes without having to worry
363 about whether implementations will cause holdoff problems due to
364 large scan batch sizes.
366 Whoever sets up the inode is responsible for filling in the "i_op"
367 field. This is a pointer to a "struct inode_operations" which describes
368 the methods that can be performed on individual inodes.
371 struct xattr_handlers
372 ---------------------
374 On filesystems that support extended attributes (xattrs), the s_xattr
375 superblock field points to a NULL-terminated array of xattr handlers.
376 Extended attributes are name:value pairs.
379 Indicates that the handler matches attributes with the specified
380 name (such as "system.posix_acl_access"); the prefix field must
384 Indicates that the handler matches all attributes with the
385 specified name prefix (such as "user."); the name field must be
389 Determine if attributes matching this xattr handler should be
390 listed for a particular dentry. Used by some listxattr
391 implementations like generic_listxattr.
394 Called by the VFS to get the value of a particular extended
395 attribute. This method is called by the getxattr(2) system
399 Called by the VFS to set the value of a particular extended
400 attribute. When the new value is NULL, called to remove a
401 particular extended attribute. This method is called by the
402 setxattr(2) and removexattr(2) system calls.
404 When none of the xattr handlers of a filesystem match the specified
405 attribute name or when a filesystem doesn't support extended attributes,
406 the various ``*xattr(2)`` system calls return -EOPNOTSUPP.
412 An inode object represents an object within the filesystem.
415 struct inode_operations
416 -----------------------
418 This describes how the VFS can manipulate an inode in your filesystem.
419 As of kernel 2.6.22, the following members are defined:
423 struct inode_operations {
424 int (*create) (struct user_namespace *, struct inode *,struct dentry *, umode_t, bool);
425 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
426 int (*link) (struct dentry *,struct inode *,struct dentry *);
427 int (*unlink) (struct inode *,struct dentry *);
428 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,const char *);
429 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,umode_t);
430 int (*rmdir) (struct inode *,struct dentry *);
431 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,umode_t,dev_t);
432 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
433 struct inode *, struct dentry *, unsigned int);
434 int (*readlink) (struct dentry *, char __user *,int);
435 const char *(*get_link) (struct dentry *, struct inode *,
436 struct delayed_call *);
437 int (*permission) (struct user_namespace *, struct inode *, int);
438 struct posix_acl * (*get_acl)(struct inode *, int, bool);
439 int (*setattr) (struct user_namespace *, struct dentry *, struct iattr *);
440 int (*getattr) (struct user_namespace *, const struct path *, struct kstat *, u32, unsigned int);
441 ssize_t (*listxattr) (struct dentry *, char *, size_t);
442 void (*update_time)(struct inode *, struct timespec *, int);
443 int (*atomic_open)(struct inode *, struct dentry *, struct file *,
444 unsigned open_flag, umode_t create_mode);
445 int (*tmpfile) (struct user_namespace *, struct inode *, struct dentry *, umode_t);
446 int (*set_acl)(struct user_namespace *, struct inode *, struct posix_acl *, int);
447 int (*fileattr_set)(struct user_namespace *mnt_userns,
448 struct dentry *dentry, struct fileattr *fa);
449 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
452 Again, all methods are called without any locks being held, unless
456 called by the open(2) and creat(2) system calls. Only required
457 if you want to support regular files. The dentry you get should
458 not have an inode (i.e. it should be a negative dentry). Here
459 you will probably call d_instantiate() with the dentry and the
463 called when the VFS needs to look up an inode in a parent
464 directory. The name to look for is found in the dentry. This
465 method must call d_add() to insert the found inode into the
466 dentry. The "i_count" field in the inode structure should be
467 incremented. If the named inode does not exist a NULL inode
468 should be inserted into the dentry (this is called a negative
469 dentry). Returning an error code from this routine must only be
470 done on a real error, otherwise creating inodes with system
471 calls like create(2), mknod(2), mkdir(2) and so on will fail.
472 If you wish to overload the dentry methods then you should
473 initialise the "d_dop" field in the dentry; this is a pointer to
474 a struct "dentry_operations". This method is called with the
475 directory inode semaphore held
478 called by the link(2) system call. Only required if you want to
479 support hard links. You will probably need to call
480 d_instantiate() just as you would in the create() method
483 called by the unlink(2) system call. Only required if you want
484 to support deleting inodes
487 called by the symlink(2) system call. Only required if you want
488 to support symlinks. You will probably need to call
489 d_instantiate() just as you would in the create() method
492 called by the mkdir(2) system call. Only required if you want
493 to support creating subdirectories. You will probably need to
494 call d_instantiate() just as you would in the create() method
497 called by the rmdir(2) system call. Only required if you want
498 to support deleting subdirectories
501 called by the mknod(2) system call to create a device (char,
502 block) inode or a named pipe (FIFO) or socket. Only required if
503 you want to support creating these types of inodes. You will
504 probably need to call d_instantiate() just as you would in the
508 called by the rename(2) system call to rename the object to have
509 the parent and name given by the second inode and dentry.
511 The filesystem must return -EINVAL for any unsupported or
512 unknown flags. Currently the following flags are implemented:
513 (1) RENAME_NOREPLACE: this flag indicates that if the target of
514 the rename exists the rename should fail with -EEXIST instead of
515 replacing the target. The VFS already checks for existence, so
516 for local filesystems the RENAME_NOREPLACE implementation is
517 equivalent to plain rename.
518 (2) RENAME_EXCHANGE: exchange source and target. Both must
519 exist; this is checked by the VFS. Unlike plain rename, source
520 and target may be of different type.
523 called by the VFS to follow a symbolic link to the inode it
524 points to. Only required if you want to support symbolic links.
525 This method returns the symlink body to traverse (and possibly
526 resets the current position with nd_jump_link()). If the body
527 won't go away until the inode is gone, nothing else is needed;
528 if it needs to be otherwise pinned, arrange for its release by
529 having get_link(..., ..., done) do set_delayed_call(done,
530 destructor, argument). In that case destructor(argument) will
531 be called once VFS is done with the body you've returned. May
532 be called in RCU mode; that is indicated by NULL dentry
533 argument. If request can't be handled without leaving RCU mode,
534 have it return ERR_PTR(-ECHILD).
536 If the filesystem stores the symlink target in ->i_link, the
537 VFS may use it directly without calling ->get_link(); however,
538 ->get_link() must still be provided. ->i_link must not be
539 freed until after an RCU grace period. Writing to ->i_link
540 post-iget() time requires a 'release' memory barrier.
543 this is now just an override for use by readlink(2) for the
544 cases when ->get_link uses nd_jump_link() or object is not in
545 fact a symlink. Normally filesystems should only implement
546 ->get_link for symlinks and readlink(2) will automatically use
550 called by the VFS to check for access rights on a POSIX-like
553 May be called in rcu-walk mode (mask & MAY_NOT_BLOCK). If in
554 rcu-walk mode, the filesystem must check the permission without
555 blocking or storing to the inode.
557 If a situation is encountered that rcu-walk cannot handle,
559 -ECHILD and it will be called again in ref-walk mode.
562 called by the VFS to set attributes for a file. This method is
563 called by chmod(2) and related system calls.
566 called by the VFS to get attributes of a file. This method is
567 called by stat(2) and related system calls.
570 called by the VFS to list all extended attributes for a given
571 file. This method is called by the listxattr(2) system call.
574 called by the VFS to update a specific time or the i_version of
575 an inode. If this is not defined the VFS will update the inode
576 itself and call mark_inode_dirty_sync.
579 called on the last component of an open. Using this optional
580 method the filesystem can look up, possibly create and open the
581 file in one atomic operation. If it wants to leave actual
582 opening to the caller (e.g. if the file turned out to be a
583 symlink, device, or just something filesystem won't do atomic
584 open for), it may signal this by returning finish_no_open(file,
585 dentry). This method is only called if the last component is
586 negative or needs lookup. Cached positive dentries are still
587 handled by f_op->open(). If the file was created, FMODE_CREATED
588 flag should be set in file->f_mode. In case of O_EXCL the
589 method must only succeed if the file didn't exist and hence
590 FMODE_CREATED shall always be set on success.
593 called in the end of O_TMPFILE open(). Optional, equivalent to
594 atomically creating, opening and unlinking a file in given
598 called on ioctl(FS_IOC_GETFLAGS) and ioctl(FS_IOC_FSGETXATTR) to
599 retrieve miscellaneous file flags and attributes. Also called
600 before the relevant SET operation to check what is being changed
601 (in this case with i_rwsem locked exclusive). If unset, then
602 fall back to f_op->ioctl().
605 called on ioctl(FS_IOC_SETFLAGS) and ioctl(FS_IOC_FSSETXATTR) to
606 change miscellaneous file flags and attributes. Callers hold
607 i_rwsem exclusive. If unset, then fall back to f_op->ioctl().
610 The Address Space Object
611 ========================
613 The address space object is used to group and manage pages in the page
614 cache. It can be used to keep track of the pages in a file (or anything
615 else) and also track the mapping of sections of the file into process
618 There are a number of distinct yet related services that an
619 address-space can provide. These include communicating memory pressure,
620 page lookup by address, and keeping track of pages tagged as Dirty or
623 The first can be used independently to the others. The VM can try to
624 either write dirty pages in order to clean them, or release clean pages
625 in order to reuse them. To do this it can call the ->writepage method
626 on dirty pages, and ->release_folio on clean folios with the private
627 flag set. Clean pages without PagePrivate and with no external references
628 will be released without notice being given to the address_space.
630 To achieve this functionality, pages need to be placed on an LRU with
631 lru_cache_add and mark_page_active needs to be called whenever the page
634 Pages are normally kept in a radix tree index by ->index. This tree
635 maintains information about the PG_Dirty and PG_Writeback status of each
636 page, so that pages with either of these flags can be found quickly.
638 The Dirty tag is primarily used by mpage_writepages - the default
639 ->writepages method. It uses the tag to find dirty pages to call
640 ->writepage on. If mpage_writepages is not used (i.e. the address
641 provides its own ->writepages) , the PAGECACHE_TAG_DIRTY tag is almost
642 unused. write_inode_now and sync_inode do use it (through
643 __sync_single_inode) to check if ->writepages has been successful in
644 writing out the whole address_space.
646 The Writeback tag is used by filemap*wait* and sync_page* functions, via
647 filemap_fdatawait_range, to wait for all writeback to complete.
649 An address_space handler may attach extra information to a page,
650 typically using the 'private' field in the 'struct page'. If such
651 information is attached, the PG_Private flag should be set. This will
652 cause various VM routines to make extra calls into the address_space
653 handler to deal with that data.
655 An address space acts as an intermediate between storage and
656 application. Data is read into the address space a whole page at a
657 time, and provided to the application either by copying of the page, or
658 by memory-mapping the page. Data is written into the address space by
659 the application, and then written-back to storage typically in whole
660 pages, however the address_space has finer control of write sizes.
662 The read process essentially only requires 'read_folio'. The write
663 process is more complicated and uses write_begin/write_end or
664 dirty_folio to write data into the address_space, and writepage and
665 writepages to writeback data to storage.
667 Adding and removing pages to/from an address_space is protected by the
670 When data is written to a page, the PG_Dirty flag should be set. It
671 typically remains set until writepage asks for it to be written. This
672 should clear PG_Dirty and set PG_Writeback. It can be actually written
673 at any point after PG_Dirty is clear. Once it is known to be safe,
674 PG_Writeback is cleared.
676 Writeback makes use of a writeback_control structure to direct the
677 operations. This gives the writepage and writepages operations some
678 information about the nature of and reason for the writeback request,
679 and the constraints under which it is being done. It is also used to
680 return information back to the caller about the result of a writepage or
684 Handling errors during writeback
685 --------------------------------
687 Most applications that do buffered I/O will periodically call a file
688 synchronization call (fsync, fdatasync, msync or sync_file_range) to
689 ensure that data written has made it to the backing store. When there
690 is an error during writeback, they expect that error to be reported when
691 a file sync request is made. After an error has been reported on one
692 request, subsequent requests on the same file descriptor should return
693 0, unless further writeback errors have occurred since the previous file
696 Ideally, the kernel would report errors only on file descriptions on
697 which writes were done that subsequently failed to be written back. The
698 generic pagecache infrastructure does not track the file descriptions
699 that have dirtied each individual page however, so determining which
700 file descriptors should get back an error is not possible.
702 Instead, the generic writeback error tracking infrastructure in the
703 kernel settles for reporting errors to fsync on all file descriptions
704 that were open at the time that the error occurred. In a situation with
705 multiple writers, all of them will get back an error on a subsequent
706 fsync, even if all of the writes done through that particular file
707 descriptor succeeded (or even if there were no writes on that file
710 Filesystems that wish to use this infrastructure should call
711 mapping_set_error to record the error in the address_space when it
712 occurs. Then, after writing back data from the pagecache in their
713 file->fsync operation, they should call file_check_and_advance_wb_err to
714 ensure that the struct file's error cursor has advanced to the correct
715 point in the stream of errors emitted by the backing device(s).
718 struct address_space_operations
719 -------------------------------
721 This describes how the VFS can manipulate mapping of a file to page
722 cache in your filesystem. The following members are defined:
726 struct address_space_operations {
727 int (*writepage)(struct page *page, struct writeback_control *wbc);
728 int (*read_folio)(struct file *, struct folio *);
729 int (*writepages)(struct address_space *, struct writeback_control *);
730 bool (*dirty_folio)(struct address_space *, struct folio *);
731 void (*readahead)(struct readahead_control *);
732 int (*write_begin)(struct file *, struct address_space *mapping,
733 loff_t pos, unsigned len,
734 struct page **pagep, void **fsdata);
735 int (*write_end)(struct file *, struct address_space *mapping,
736 loff_t pos, unsigned len, unsigned copied,
737 struct page *page, void *fsdata);
738 sector_t (*bmap)(struct address_space *, sector_t);
739 void (*invalidate_folio) (struct folio *, size_t start, size_t len);
740 bool (*release_folio)(struct folio *, gfp_t);
741 void (*free_folio)(struct folio *);
742 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
743 int (*migrate_folio)(struct mapping *, struct folio *dst,
744 struct folio *src, enum migrate_mode);
745 int (*launder_folio) (struct folio *);
747 bool (*is_partially_uptodate) (struct folio *, size_t from,
749 void (*is_dirty_writeback)(struct folio *, bool *, bool *);
750 int (*error_remove_page) (struct mapping *mapping, struct page *page);
751 int (*swap_activate)(struct swap_info_struct *sis, struct file *f, sector_t *span)
752 int (*swap_deactivate)(struct file *);
753 int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter);
757 called by the VM to write a dirty page to backing store. This
758 may happen for data integrity reasons (i.e. 'sync'), or to free
759 up memory (flush). The difference can be seen in
760 wbc->sync_mode. The PG_Dirty flag has been cleared and
761 PageLocked is true. writepage should start writeout, should set
762 PG_Writeback, and should make sure the page is unlocked, either
763 synchronously or asynchronously when the write operation
766 If wbc->sync_mode is WB_SYNC_NONE, ->writepage doesn't have to
767 try too hard if there are problems, and may choose to write out
768 other pages from the mapping if that is easier (e.g. due to
769 internal dependencies). If it chooses not to start writeout, it
770 should return AOP_WRITEPAGE_ACTIVATE so that the VM will not
771 keep calling ->writepage on that page.
773 See the file "Locking" for more details.
776 Called by the page cache to read a folio from the backing store.
777 The 'file' argument supplies authentication information to network
778 filesystems, and is generally not used by block based filesystems.
779 It may be NULL if the caller does not have an open file (eg if
780 the kernel is performing a read for itself rather than on behalf
781 of a userspace process with an open file).
783 If the mapping does not support large folios, the folio will
784 contain a single page. The folio will be locked when read_folio
785 is called. If the read completes successfully, the folio should
786 be marked uptodate. The filesystem should unlock the folio
787 once the read has completed, whether it was successful or not.
788 The filesystem does not need to modify the refcount on the folio;
789 the page cache holds a reference count and that will not be
790 released until the folio is unlocked.
792 Filesystems may implement ->read_folio() synchronously.
793 In normal operation, folios are read through the ->readahead()
794 method. Only if this fails, or if the caller needs to wait for
795 the read to complete will the page cache call ->read_folio().
796 Filesystems should not attempt to perform their own readahead
797 in the ->read_folio() operation.
799 If the filesystem cannot perform the read at this time, it can
800 unlock the folio, do whatever action it needs to ensure that the
801 read will succeed in the future and return AOP_TRUNCATED_PAGE.
802 In this case, the caller should look up the folio, lock it,
803 and call ->read_folio again.
805 Callers may invoke the ->read_folio() method directly, but using
806 read_mapping_folio() will take care of locking, waiting for the
807 read to complete and handle cases such as AOP_TRUNCATED_PAGE.
810 called by the VM to write out pages associated with the
811 address_space object. If wbc->sync_mode is WB_SYNC_ALL, then
812 the writeback_control will specify a range of pages that must be
813 written out. If it is WB_SYNC_NONE, then a nr_to_write is
814 given and that many pages should be written if possible. If no
815 ->writepages is given, then mpage_writepages is used instead.
816 This will choose pages from the address space that are tagged as
817 DIRTY and will pass them to ->writepage.
820 called by the VM to mark a folio as dirty. This is particularly
821 needed if an address space attaches private data to a folio, and
822 that data needs to be updated when a folio is dirtied. This is
823 called, for example, when a memory mapped page gets modified.
824 If defined, it should set the folio dirty flag, and the
825 PAGECACHE_TAG_DIRTY search mark in i_pages.
828 Called by the VM to read pages associated with the address_space
829 object. The pages are consecutive in the page cache and are
830 locked. The implementation should decrement the page refcount
831 after starting I/O on each page. Usually the page will be
832 unlocked by the I/O completion handler. The set of pages are
833 divided into some sync pages followed by some async pages,
834 rac->ra->async_size gives the number of async pages. The
835 filesystem should attempt to read all sync pages but may decide
836 to stop once it reaches the async pages. If it does decide to
837 stop attempting I/O, it can simply return. The caller will
838 remove the remaining pages from the address space, unlock them
839 and decrement the page refcount. Set PageUptodate if the I/O
840 completes successfully. Setting PageError on any page will be
841 ignored; simply unlock the page if an I/O error occurs.
844 Called by the generic buffered write code to ask the filesystem
845 to prepare to write len bytes at the given offset in the file.
846 The address_space should check that the write will be able to
847 complete, by allocating space if necessary and doing any other
848 internal housekeeping. If the write will update parts of any
849 basic-blocks on storage, then those blocks should be pre-read
850 (if they haven't been read already) so that the updated blocks
851 can be written out properly.
853 The filesystem must return the locked pagecache page for the
854 specified offset, in ``*pagep``, for the caller to write into.
856 It must be able to cope with short writes (where the length
857 passed to write_begin is greater than the number of bytes copied
860 A void * may be returned in fsdata, which then gets passed into
863 Returns 0 on success; < 0 on failure (which is the error code),
864 in which case write_end is not called.
867 After a successful write_begin, and data copy, write_end must be
868 called. len is the original len passed to write_begin, and
869 copied is the amount that was able to be copied.
871 The filesystem must take care of unlocking the page and
872 releasing it refcount, and updating i_size.
874 Returns < 0 on failure, otherwise the number of bytes (<=
875 'copied') that were able to be copied into pagecache.
878 called by the VFS to map a logical block offset within object to
879 physical block number. This method is used by the FIBMAP ioctl
880 and for working with swap-files. To be able to swap to a file,
881 the file must have a stable mapping to a block device. The swap
882 system does not go through the filesystem but instead uses bmap
883 to find out where the blocks in the file are and uses those
887 If a folio has private data, then invalidate_folio will be
888 called when part or all of the folio is to be removed from the
889 address space. This generally corresponds to either a
890 truncation, punch hole or a complete invalidation of the address
891 space (in the latter case 'offset' will always be 0 and 'length'
892 will be folio_size()). Any private data associated with the folio
893 should be updated to reflect this truncation. If offset is 0
894 and length is folio_size(), then the private data should be
895 released, because the folio must be able to be completely
896 discarded. This may be done by calling the ->release_folio
897 function, but in this case the release MUST succeed.
900 release_folio is called on folios with private data to tell the
901 filesystem that the folio is about to be freed. ->release_folio
902 should remove any private data from the folio and clear the
903 private flag. If release_folio() fails, it should return false.
904 release_folio() is used in two distinct though related cases.
905 The first is when the VM wants to free a clean folio with no
906 active users. If ->release_folio succeeds, the folio will be
907 removed from the address_space and be freed.
909 The second case is when a request has been made to invalidate
910 some or all folios in an address_space. This can happen
911 through the fadvise(POSIX_FADV_DONTNEED) system call or by the
912 filesystem explicitly requesting it as nfs and 9p do (when they
913 believe the cache may be out of date with storage) by calling
914 invalidate_inode_pages2(). If the filesystem makes such a call,
915 and needs to be certain that all folios are invalidated, then
916 its release_folio will need to ensure this. Possibly it can
917 clear the uptodate flag if it cannot free private data yet.
920 free_folio is called once the folio is no longer visible in the
921 page cache in order to allow the cleanup of any private data.
922 Since it may be called by the memory reclaimer, it should not
923 assume that the original address_space mapping still exists, and
927 called by the generic read/write routines to perform direct_IO -
928 that is IO requests which bypass the page cache and transfer
929 data directly between the storage and the application's address
933 This is used to compact the physical memory usage. If the VM
934 wants to relocate a folio (maybe from a memory device that is
935 signalling imminent failure) it will pass a new folio and an old
936 folio to this function. migrate_folio should transfer any private
937 data across and update any references that it has to the folio.
940 Called before freeing a folio - it writes back the dirty folio.
941 To prevent redirtying the folio, it is kept locked during the
944 ``is_partially_uptodate``
945 Called by the VM when reading a file through the pagecache when
946 the underlying blocksize is smaller than the size of the folio.
947 If the required block is up to date then the read can complete
948 without needing I/O to bring the whole page up to date.
950 ``is_dirty_writeback``
951 Called by the VM when attempting to reclaim a folio. The VM uses
952 dirty and writeback information to determine if it needs to
953 stall to allow flushers a chance to complete some IO.
954 Ordinarily it can use folio_test_dirty and folio_test_writeback but
955 some filesystems have more complex state (unstable folios in NFS
956 prevent reclaim) or do not set those flags due to locking
957 problems. This callback allows a filesystem to indicate to the
958 VM if a folio should be treated as dirty or writeback for the
959 purposes of stalling.
961 ``error_remove_page``
962 normally set to generic_error_remove_page if truncation is ok
963 for this address space. Used for memory failure handling.
964 Setting this implies you deal with pages going away under you,
965 unless you have them locked or reference counts increased.
969 Called to prepare the given file for swap. It should perform
970 any validation and preparation necessary to ensure that writes
971 can be performed with minimal memory allocation. It should call
972 add_swap_extent(), or the helper iomap_swapfile_activate(), and
973 return the number of extents added. If IO should be submitted
974 through ->swap_rw(), it should set SWP_FS_OPS, otherwise IO will
975 be submitted directly to the block device ``sis->bdev``.
978 Called during swapoff on files where swap_activate was
982 Called to read or write swap pages when SWP_FS_OPS is set.
987 A file object represents a file opened by a process. This is also known
988 as an "open file description" in POSIX parlance.
991 struct file_operations
992 ----------------------
994 This describes how the VFS can manipulate an open file. As of kernel
995 4.18, the following members are defined:
999 struct file_operations {
1000 struct module *owner;
1001 loff_t (*llseek) (struct file *, loff_t, int);
1002 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1003 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1004 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1005 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1006 int (*iopoll)(struct kiocb *kiocb, bool spin);
1007 int (*iterate) (struct file *, struct dir_context *);
1008 int (*iterate_shared) (struct file *, struct dir_context *);
1009 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1010 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1011 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1012 int (*mmap) (struct file *, struct vm_area_struct *);
1013 int (*open) (struct inode *, struct file *);
1014 int (*flush) (struct file *, fl_owner_t id);
1015 int (*release) (struct inode *, struct file *);
1016 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1017 int (*fasync) (int, struct file *, int);
1018 int (*lock) (struct file *, int, struct file_lock *);
1019 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1020 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1021 int (*check_flags)(int);
1022 int (*flock) (struct file *, int, struct file_lock *);
1023 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1024 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1025 int (*setlease)(struct file *, long, struct file_lock **, void **);
1026 long (*fallocate)(struct file *file, int mode, loff_t offset,
1028 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1030 unsigned (*mmap_capabilities)(struct file *);
1032 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, loff_t, size_t, unsigned int);
1033 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1034 struct file *file_out, loff_t pos_out,
1035 loff_t len, unsigned int remap_flags);
1036 int (*fadvise)(struct file *, loff_t, loff_t, int);
1039 Again, all methods are called without any locks being held, unless
1043 called when the VFS needs to move the file position index
1046 called by read(2) and related system calls
1049 possibly asynchronous read with iov_iter as destination
1052 called by write(2) and related system calls
1055 possibly asynchronous write with iov_iter as source
1058 called when aio wants to poll for completions on HIPRI iocbs
1061 called when the VFS needs to read the directory contents
1064 called when the VFS needs to read the directory contents when
1065 filesystem supports concurrent dir iterators
1068 called by the VFS when a process wants to check if there is
1069 activity on this file and (optionally) go to sleep until there
1070 is activity. Called by the select(2) and poll(2) system calls
1073 called by the ioctl(2) system call.
1076 called by the ioctl(2) system call when 32 bit system calls are
1077 used on 64 bit kernels.
1080 called by the mmap(2) system call
1083 called by the VFS when an inode should be opened. When the VFS
1084 opens a file, it creates a new "struct file". It then calls the
1085 open method for the newly allocated file structure. You might
1086 think that the open method really belongs in "struct
1087 inode_operations", and you may be right. I think it's done the
1088 way it is because it makes filesystems simpler to implement.
1089 The open() method is a good place to initialize the
1090 "private_data" member in the file structure if you want to point
1091 to a device structure
1094 called by the close(2) system call to flush a file
1097 called when the last reference to an open file is closed
1100 called by the fsync(2) system call. Also see the section above
1101 entitled "Handling errors during writeback".
1104 called by the fcntl(2) system call when asynchronous
1105 (non-blocking) mode is enabled for a file
1108 called by the fcntl(2) system call for F_GETLK, F_SETLK, and
1111 ``get_unmapped_area``
1112 called by the mmap(2) system call
1115 called by the fcntl(2) system call for F_SETFL command
1118 called by the flock(2) system call
1121 called by the VFS to splice data from a pipe to a file. This
1122 method is used by the splice(2) system call
1125 called by the VFS to splice data from file to a pipe. This
1126 method is used by the splice(2) system call
1129 called by the VFS to set or release a file lock lease. setlease
1130 implementations should call generic_setlease to record or remove
1131 the lease in the inode after setting it.
1134 called by the VFS to preallocate blocks or punch a hole.
1137 called by the copy_file_range(2) system call.
1139 ``remap_file_range``
1140 called by the ioctl(2) system call for FICLONERANGE and FICLONE
1141 and FIDEDUPERANGE commands to remap file ranges. An
1142 implementation should remap len bytes at pos_in of the source
1143 file into the dest file at pos_out. Implementations must handle
1144 callers passing in len == 0; this means "remap to the end of the
1145 source file". The return value should the number of bytes
1146 remapped, or the usual negative error code if errors occurred
1147 before any bytes were remapped. The remap_flags parameter
1148 accepts REMAP_FILE_* flags. If REMAP_FILE_DEDUP is set then the
1149 implementation must only remap if the requested file ranges have
1150 identical contents. If REMAP_FILE_CAN_SHORTEN is set, the caller is
1151 ok with the implementation shortening the request length to
1152 satisfy alignment or EOF requirements (or any other reason).
1155 possibly called by the fadvise64() system call.
1157 Note that the file operations are implemented by the specific
1158 filesystem in which the inode resides. When opening a device node
1159 (character or block special) most filesystems will call special
1160 support routines in the VFS which will locate the required device
1161 driver information. These support routines replace the filesystem file
1162 operations with those for the device driver, and then proceed to call
1163 the new open() method for the file. This is how opening a device file
1164 in the filesystem eventually ends up calling the device driver open()
1168 Directory Entry Cache (dcache)
1169 ==============================
1172 struct dentry_operations
1173 ------------------------
1175 This describes how a filesystem can overload the standard dentry
1176 operations. Dentries and the dcache are the domain of the VFS and the
1177 individual filesystem implementations. Device drivers have no business
1178 here. These methods may be set to NULL, as they are either optional or
1179 the VFS uses a default. As of kernel 2.6.22, the following members are
1184 struct dentry_operations {
1185 int (*d_revalidate)(struct dentry *, unsigned int);
1186 int (*d_weak_revalidate)(struct dentry *, unsigned int);
1187 int (*d_hash)(const struct dentry *, struct qstr *);
1188 int (*d_compare)(const struct dentry *,
1189 unsigned int, const char *, const struct qstr *);
1190 int (*d_delete)(const struct dentry *);
1191 int (*d_init)(struct dentry *);
1192 void (*d_release)(struct dentry *);
1193 void (*d_iput)(struct dentry *, struct inode *);
1194 char *(*d_dname)(struct dentry *, char *, int);
1195 struct vfsmount *(*d_automount)(struct path *);
1196 int (*d_manage)(const struct path *, bool);
1197 struct dentry *(*d_real)(struct dentry *, const struct inode *);
1201 called when the VFS needs to revalidate a dentry. This is
1202 called whenever a name look-up finds a dentry in the dcache.
1203 Most local filesystems leave this as NULL, because all their
1204 dentries in the dcache are valid. Network filesystems are
1205 different since things can change on the server without the
1206 client necessarily being aware of it.
1208 This function should return a positive value if the dentry is
1209 still valid, and zero or a negative error code if it isn't.
1211 d_revalidate may be called in rcu-walk mode (flags &
1212 LOOKUP_RCU). If in rcu-walk mode, the filesystem must
1213 revalidate the dentry without blocking or storing to the dentry,
1214 d_parent and d_inode should not be used without care (because
1215 they can change and, in d_inode case, even become NULL under
1218 If a situation is encountered that rcu-walk cannot handle,
1220 -ECHILD and it will be called again in ref-walk mode.
1222 ``_weak_revalidate``
1223 called when the VFS needs to revalidate a "jumped" dentry. This
1224 is called when a path-walk ends at dentry that was not acquired
1225 by doing a lookup in the parent directory. This includes "/",
1226 "." and "..", as well as procfs-style symlinks and mountpoint
1229 In this case, we are less concerned with whether the dentry is
1230 still fully correct, but rather that the inode is still valid.
1231 As with d_revalidate, most local filesystems will set this to
1232 NULL since their dcache entries are always valid.
1234 This function has the same return code semantics as
1237 d_weak_revalidate is only called after leaving rcu-walk mode.
1240 called when the VFS adds a dentry to the hash table. The first
1241 dentry passed to d_hash is the parent directory that the name is
1244 Same locking and synchronisation rules as d_compare regarding
1245 what is safe to dereference etc.
1248 called to compare a dentry name with a given name. The first
1249 dentry is the parent of the dentry to be compared, the second is
1250 the child dentry. len and name string are properties of the
1251 dentry to be compared. qstr is the name to compare it with.
1253 Must be constant and idempotent, and should not take locks if
1254 possible, and should not or store into the dentry. Should not
1255 dereference pointers outside the dentry without lots of care
1256 (eg. d_parent, d_inode, d_name should not be used).
1258 However, our vfsmount is pinned, and RCU held, so the dentries
1259 and inodes won't disappear, neither will our sb or filesystem
1260 module. ->d_sb may be used.
1262 It is a tricky calling convention because it needs to be called
1263 under "rcu-walk", ie. without any locks or references on things.
1266 called when the last reference to a dentry is dropped and the
1267 dcache is deciding whether or not to cache it. Return 1 to
1268 delete immediately, or 0 to cache the dentry. Default is NULL
1269 which means to always cache a reachable dentry. d_delete must
1270 be constant and idempotent.
1273 called when a dentry is allocated
1276 called when a dentry is really deallocated
1279 called when a dentry loses its inode (just prior to its being
1280 deallocated). The default when this is NULL is that the VFS
1281 calls iput(). If you define this method, you must call iput()
1285 called when the pathname of a dentry should be generated.
1286 Useful for some pseudo filesystems (sockfs, pipefs, ...) to
1287 delay pathname generation. (Instead of doing it when dentry is
1288 created, it's done only when the path is needed.). Real
1289 filesystems probably dont want to use it, because their dentries
1290 are present in global dcache hash, so their hash should be an
1291 invariant. As no lock is held, d_dname() should not try to
1292 modify the dentry itself, unless appropriate SMP safety is used.
1293 CAUTION : d_path() logic is quite tricky. The correct way to
1294 return for example "Hello" is to put it at the end of the
1295 buffer, and returns a pointer to the first char.
1296 dynamic_dname() helper function is provided to take care of
1303 static char *pipefs_dname(struct dentry *dent, char *buffer, int buflen)
1305 return dynamic_dname(dentry, buffer, buflen, "pipe:[%lu]",
1306 dentry->d_inode->i_ino);
1310 called when an automount dentry is to be traversed (optional).
1311 This should create a new VFS mount record and return the record
1312 to the caller. The caller is supplied with a path parameter
1313 giving the automount directory to describe the automount target
1314 and the parent VFS mount record to provide inheritable mount
1315 parameters. NULL should be returned if someone else managed to
1316 make the automount first. If the vfsmount creation failed, then
1317 an error code should be returned. If -EISDIR is returned, then
1318 the directory will be treated as an ordinary directory and
1319 returned to pathwalk to continue walking.
1321 If a vfsmount is returned, the caller will attempt to mount it
1322 on the mountpoint and will remove the vfsmount from its
1323 expiration list in the case of failure. The vfsmount should be
1324 returned with 2 refs on it to prevent automatic expiration - the
1325 caller will clean up the additional ref.
1327 This function is only used if DCACHE_NEED_AUTOMOUNT is set on
1328 the dentry. This is set by __d_instantiate() if S_AUTOMOUNT is
1329 set on the inode being added.
1332 called to allow the filesystem to manage the transition from a
1333 dentry (optional). This allows autofs, for example, to hold up
1334 clients waiting to explore behind a 'mountpoint' while letting
1335 the daemon go past and construct the subtree there. 0 should be
1336 returned to let the calling process continue. -EISDIR can be
1337 returned to tell pathwalk to use this directory as an ordinary
1338 directory and to ignore anything mounted on it and not to check
1339 the automount flag. Any other error code will abort pathwalk
1342 If the 'rcu_walk' parameter is true, then the caller is doing a
1343 pathwalk in RCU-walk mode. Sleeping is not permitted in this
1344 mode, and the caller can be asked to leave it and call again by
1345 returning -ECHILD. -EISDIR may also be returned to tell
1346 pathwalk to ignore d_automount or any mounts.
1348 This function is only used if DCACHE_MANAGE_TRANSIT is set on
1349 the dentry being transited from.
1352 overlay/union type filesystems implement this method to return
1353 one of the underlying dentries hidden by the overlay. It is
1354 used in two different modes:
1356 Called from file_dentry() it returns the real dentry matching
1357 the inode argument. The real dentry may be from a lower layer
1358 already copied up, but still referenced from the file. This
1359 mode is selected with a non-NULL inode argument.
1361 With NULL inode the topmost real underlying dentry is returned.
1363 Each dentry has a pointer to its parent dentry, as well as a hash list
1364 of child dentries. Child dentries are basically like files in a
1368 Directory Entry Cache API
1369 --------------------------
1371 There are a number of functions defined which permit a filesystem to
1372 manipulate dentries:
1375 open a new handle for an existing dentry (this just increments
1379 close a handle for a dentry (decrements the usage count). If
1380 the usage count drops to 0, and the dentry is still in its
1381 parent's hash, the "d_delete" method is called to check whether
1382 it should be cached. If it should not be cached, or if the
1383 dentry is not hashed, it is deleted. Otherwise cached dentries
1384 are put into an LRU list to be reclaimed on memory shortage.
1387 this unhashes a dentry from its parents hash list. A subsequent
1388 call to dput() will deallocate the dentry if its usage count
1392 delete a dentry. If there are no other open references to the
1393 dentry then the dentry is turned into a negative dentry (the
1394 d_iput() method is called). If there are other references, then
1395 d_drop() is called instead
1398 add a dentry to its parents hash list and then calls
1402 add a dentry to the alias hash list for the inode and updates
1403 the "d_inode" member. The "i_count" member in the inode
1404 structure should be set/incremented. If the inode pointer is
1405 NULL, the dentry is called a "negative dentry". This function
1406 is commonly called when an inode is created for an existing
1410 look up a dentry given its parent and path name component It
1411 looks up the child of that given name from the dcache hash
1412 table. If it is found, the reference count is incremented and
1413 the dentry is returned. The caller must use dput() to free the
1414 dentry when it finishes using it.
1424 On mount and remount the filesystem is passed a string containing a
1425 comma separated list of mount options. The options can have either of
1431 The <linux/parser.h> header defines an API that helps parse these
1432 options. There are plenty of examples on how to use it in existing
1439 If a filesystem accepts mount options, it must define show_options() to
1440 show all the currently active options. The rules are:
1442 - options MUST be shown which are not default or their values differ
1445 - options MAY be shown which are enabled by default or have their
1448 Options used only internally between a mount helper and the kernel (such
1449 as file descriptors), or which only have an effect during the mounting
1450 (such as ones controlling the creation of a journal) are exempt from the
1453 The underlying reason for the above rules is to make sure, that a mount
1454 can be accurately replicated (e.g. umounting and mounting again) based
1455 on the information found in /proc/mounts.
1461 (Note some of these resources are not up-to-date with the latest kernel
1464 Creating Linux virtual filesystems. 2002
1465 <https://lwn.net/Articles/13325/>
1467 The Linux Virtual File-system Layer by Neil Brown. 1999
1468 <http://www.cse.unsw.edu.au/~neilb/oss/linux-commentary/vfs.html>
1470 A tour of the Linux VFS by Michael K. Johnson. 1996
1471 <https://www.tldp.org/LDP/khg/HyperNews/get/fs/vfstour.html>
1473 A small trail through the Linux kernel by Andries Brouwer. 2001
1474 <https://www.win.tue.nl/~aeb/linux/vfs/trail.html>